<%BANNER%>

Functional role of neural oscillations in attention and memory in humans

Permanent Link: http://ufdc.ufl.edu/UFE0022215/00001

Material Information

Title: Functional role of neural oscillations in attention and memory in humans
Physical Description: 1 online resource (153 p.)
Language: english
Creator: Rajagovindan, Rajasimhan
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: alpha, attention, biomarkers, brain, causality, cognitive, cortex, cortical, eeg, erp, granger, memory, neural, neuronal, ongoing, oscillations, spectral, theta, topdown
Biomedical Engineering -- Dissertations, Academic -- UF
Genre: Biomedical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: FUNCTIONAL ROLE OF NEURAL OSCILLATIONS IN ATTENTION AND MEMORY IN HUMANS Synchronized oscillatory neural activity involving widespread neural ensemble is common in the central nervous system, and is hypothesized to enable the integration of various attributes of the sensory input, resulting in a unified percept. While activity evoked to external sensory input has been instantly recognized to reflect stimulus processing, ongoing neural activity preceding stimulus onset is often considered as background noise and irrelevant to subsequent stimulus processing. Our study seeks to advance the idea that ongoing neural oscillations are critical indices of the brain state and, as such, play an important role in stimulus-evoked cognitive processing. Applying advanced multivariate spectral analysis to invasive and non-invasive electrophysiological recordings from animals and humans, we address the mechanisms and the functional roles of synchronized prestimulus oscillatory activity in processes such as anticipatory attention, visual processing and encoding of memory. First, applying computational techniques to invasive local field potential recordings from primates we expound the mechanisms leading to zero-lag synchrony across widely separated cortical areas, a phenomenon critically implicated in effective neuronal communication and thus higher-order perceptual processes. Second, we assessed the functional role of prestimulus theta band oscillation in memory related stimulus encoding in the human cortex using invasive electrocorticogram. Third, a physiological model linking prestimulus oscillations and stimulus-evoked response is proposed and is tested on noninvasive studies of human visual processing. Finally, the neural control of anticipatory attention and stimulus selection using electroencephalogram is investigated. The objective here is to expound the signals and the neural mechanisms that mediate these top-down processes that enable selective sensory integration during attentive behavior in humans. In conclusion, our findings suggest that effective sensory-perceptual processing and thus cognitive and behavioral performance to an impending stimulus, depends on the excitability state of the brain defined by these neural oscillations at its onset. Further, it is shown that the prestimulus activity is not merely a random fluctuation in excitability but one that is actively controlled in a goal-directed manner through neural synchronization across widespread functionally relevant cortical networks. Thus, this study bolsters the functional role of prestimulus ongoing activity in subsequent sensory-perceptual and higher-order cognitive processing.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Rajasimhan Rajagovindan.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Ding, Mingzhou.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0022215:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022215/00001

Material Information

Title: Functional role of neural oscillations in attention and memory in humans
Physical Description: 1 online resource (153 p.)
Language: english
Creator: Rajagovindan, Rajasimhan
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: alpha, attention, biomarkers, brain, causality, cognitive, cortex, cortical, eeg, erp, granger, memory, neural, neuronal, ongoing, oscillations, spectral, theta, topdown
Biomedical Engineering -- Dissertations, Academic -- UF
Genre: Biomedical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: FUNCTIONAL ROLE OF NEURAL OSCILLATIONS IN ATTENTION AND MEMORY IN HUMANS Synchronized oscillatory neural activity involving widespread neural ensemble is common in the central nervous system, and is hypothesized to enable the integration of various attributes of the sensory input, resulting in a unified percept. While activity evoked to external sensory input has been instantly recognized to reflect stimulus processing, ongoing neural activity preceding stimulus onset is often considered as background noise and irrelevant to subsequent stimulus processing. Our study seeks to advance the idea that ongoing neural oscillations are critical indices of the brain state and, as such, play an important role in stimulus-evoked cognitive processing. Applying advanced multivariate spectral analysis to invasive and non-invasive electrophysiological recordings from animals and humans, we address the mechanisms and the functional roles of synchronized prestimulus oscillatory activity in processes such as anticipatory attention, visual processing and encoding of memory. First, applying computational techniques to invasive local field potential recordings from primates we expound the mechanisms leading to zero-lag synchrony across widely separated cortical areas, a phenomenon critically implicated in effective neuronal communication and thus higher-order perceptual processes. Second, we assessed the functional role of prestimulus theta band oscillation in memory related stimulus encoding in the human cortex using invasive electrocorticogram. Third, a physiological model linking prestimulus oscillations and stimulus-evoked response is proposed and is tested on noninvasive studies of human visual processing. Finally, the neural control of anticipatory attention and stimulus selection using electroencephalogram is investigated. The objective here is to expound the signals and the neural mechanisms that mediate these top-down processes that enable selective sensory integration during attentive behavior in humans. In conclusion, our findings suggest that effective sensory-perceptual processing and thus cognitive and behavioral performance to an impending stimulus, depends on the excitability state of the brain defined by these neural oscillations at its onset. Further, it is shown that the prestimulus activity is not merely a random fluctuation in excitability but one that is actively controlled in a goal-directed manner through neural synchronization across widespread functionally relevant cortical networks. Thus, this study bolsters the functional role of prestimulus ongoing activity in subsequent sensory-perceptual and higher-order cognitive processing.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Rajasimhan Rajagovindan.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Ding, Mingzhou.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0022215:00001


This item has the following downloads:


Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20110318_AAAALS INGEST_TIME 2011-03-18T20:44:19Z PACKAGE UFE0022215_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 1051981 DFID F20110318_AACCDA ORIGIN DEPOSITOR PATH rajagovindan_r_Page_018.jp2 GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
10cc42eef81584e392a8e73cedf32703
SHA-1
33c3bce20c86c5840c7dd4b34e5b8c17e7e40ec1
138071 F20110318_AACCCM rajagovindan_r_Page_147.jpg
6aa369788fc8fec0e20572f298498111
75239ec127fda220e177ba860e3608616f47a807
108499 F20110318_AACCBX rajagovindan_r_Page_129.jpg
86b2183b2894d6d3fd3e35df1e443082
c6ee069fb18b549ef312bcf79bbee64de7d249ab
1051978 F20110318_AACAUE rajagovindan_r_Page_148.jp2
ccd987c3f0ed65d043fa511889ba52f9
605ae15cc957f6aace69bcca32c23209c28c5c3b
26587 F20110318_AACBXG rajagovindan_r_Page_032.QC.jpg
29a7d4439c8bb818b2a50c6ced0c61c9
b9a78db24b7ec0f31f670f378c48499e0a17ccb9
35384 F20110318_AACBWR rajagovindan_r_Page_020.QC.jpg
98c699277e60ef9351e6862e25ffa562
227c2ba3b998a1e2d067ca750f433114e49b2393
1051967 F20110318_AACCDB rajagovindan_r_Page_020.jp2
a348ed4d8c5491650cf7d2d61b666aea
09fb0c6bbb6cfe1b46f39590d7126fd68b01fb5a
132854 F20110318_AACCCN rajagovindan_r_Page_148.jpg
763fcb798423259310b099f070080cb6
f4b04a3119a66ad6e11018f7e05592f1834e6274
36309 F20110318_AACCBY rajagovindan_r_Page_132.QC.jpg
4557a605187b96f62c1b7a46cb14e6e1
862085d29d78534fd029ecf429592ed5cfb93350
112596 F20110318_AACBWS rajagovindan_r_Page_021.jpg
54f3c84118dde01af3de3da8331b11e0
e51d2c46bbee90b6a60a37d1dd08ea6acf7e1737
1051986 F20110318_AACCDC rajagovindan_r_Page_021.jp2
91147ee825106c7c72e034a007f99030
21781765925f606e263699ede3518f384f017f60
37735 F20110318_AACCCO rajagovindan_r_Page_148.QC.jpg
addea02205408b647fe49dad9de78f55
e85623a44f6cdb65c60aa1f055f5517e082a94ec
7558 F20110318_AACCBZ rajagovindan_r_Page_133.QC.jpg
c52b8f8d63a967c331ef1b3c305c07ff
b8ac7dacca71394b63ff045b426a99b61e73cfdc
9128 F20110318_AACAUF rajagovindan_r_Page_075thm.jpg
e925a06aa5012ecd8cc1c6973e78614f
42477e5075d1786df7d35792876a644ac56dbdae
74579 F20110318_AACBXH rajagovindan_r_Page_033.jpg
c43aac78ba65df8dc36caac0b6f23dc9
ddd4e17f90e6e228460c1460e0c43177402d07dd
36551 F20110318_AACBWT rajagovindan_r_Page_021.QC.jpg
11ef63b0225174cc336359114d39fba1
5aa8e5a42c16cd07a0fe27a255edab268694a071
36777 F20110318_AACBAA rajagovindan_r_Page_143.QC.jpg
3c02ac97520210aa2c889c9e479153a8
4d8d94b7ed4b821d8e25c8eaad20d6773dac83fe
653517 F20110318_AACCDD rajagovindan_r_Page_025.jp2
52819fc55fe47c80f3a5ca70865678f4
22672c0df18603b771cb1c8d7af81c2256e8f4a1
38130 F20110318_AACCCP rajagovindan_r_Page_149.QC.jpg
ad74539cf28f3ae059de255563d34bbd
8523d0380c142bcfe6f35db1e84cde4fd55af7d1
2148 F20110318_AACAUG rajagovindan_r_Page_096.txt
805555d413663afcdb22f55494809b16
51df1d2d733bdaba23b370ae5050f526d083010c
25354 F20110318_AACBXI rajagovindan_r_Page_033.QC.jpg
62bb026aaedd1928b810abdc176c6b48
28e42996f133c50add9a30829efc3c76737581c3
102505 F20110318_AACBWU rajagovindan_r_Page_022.jpg
d3f8ea3cbc7c55b64615cd33864ba3cd
ffa0b4b77d095ae2e25e3abe5c4fe75fad007a53
60066 F20110318_AACBAB rajagovindan_r_Page_118.pro
4e5d74d1d2137cdd982ae4bea330ca62
8a6c866825b34b3be603289e6f287f931aa26c6a
614793 F20110318_AACCDE rajagovindan_r_Page_028.jp2
09035cd69125b1acf0f1251ecdf7e9d0
0756f388075a538bfd14a4c59963e6f34de5b430
127644 F20110318_AACCCQ rajagovindan_r_Page_150.jpg
293d76ccbca9137f33e22b0ba65aab55
8ba1c5dbeaf8ceca6a59be66d1bc03f3cccb2b0d
613500 F20110318_AACAUH rajagovindan_r_Page_057.jp2
5a196b45fb9126a47d1b608d9c546bc0
668375753e125279f78ab137ac97369d085304f2
108064 F20110318_AACBXJ rajagovindan_r_Page_035.jpg
6edfcfa56170268dd2958cad18d76335
845f0a9b30193327fb7323520da8f0785a692b5f
32581 F20110318_AACBWV rajagovindan_r_Page_022.QC.jpg
adc13e6133d4691a51d02f20fe0ff88a
bb5f262ba709b8995283fba004d7f2a92705e410
1051956 F20110318_AACBAC rajagovindan_r_Page_112.jp2
d0f13b202f66da119f93e4148d67b5a7
187dd4b1a213ba3b44e1713d39d9e53fe787692a
826582 F20110318_AACCDF rajagovindan_r_Page_029.jp2
5e349e2273c1aba6f8936ca8b664e1d8
0a548235585e6a9d84330d336c8cb7247ebc4a94
1380 F20110318_AACAUI rajagovindan_r_Page_127.txt
f4e9044ff53937c715d80efca636b3a2
389fae843e4b65e6ab694b12db55beb79ea066b2
110507 F20110318_AACBXK rajagovindan_r_Page_036.jpg
0e6fdd7c4820f5ae4386f02dc6225690
187109499f614a3d1c03de15af04ce04b9a76949
32454 F20110318_AACBWW rajagovindan_r_Page_023.QC.jpg
68f22a487b5b468f6ac789f20efc1100
b49798b856c70888caea4167fe22637ccc5616cd
78069 F20110318_AACBAD rajagovindan_r_Page_126.jpg
e3a3751d93a7e0a0fca2d2cdb1e8d020
36d8f2f41861dbc5a19f4127ca6bc13ebd4e7fff
539339 F20110318_AACCDG rajagovindan_r_Page_031.jp2
6eaf5e5f6cd374eac43996333a64832a
d942c8dc92a5c32b2f9fe7916790c467f3d65be1
36040 F20110318_AACCCR rajagovindan_r_Page_151.QC.jpg
f8e24a5d4af2edfb9c2c25c01f3b241a
2e1856f9ecaeb3101d3fbfab57ac02a49b8f58d4
101191 F20110318_AACBYA rajagovindan_r_Page_047.jpg
d45ddae67456414b9dd42839614771e4
6159209ef8fe3bea436224636a26ea0abed1afbd
1051984 F20110318_AACAUJ rajagovindan_r_Page_036.jp2
42e25d1874377a0da04eb5cb0ef123f0
bc460889c882db0fbe62d1114a99665bda0d1902
94677 F20110318_AACBXL rajagovindan_r_Page_037.jpg
f02a83e2c08d27b50ae13f2938cd3efe
ceecb989f0c744a153a1a0e514771c3fdf4a3cd3
97950 F20110318_AACBWX rajagovindan_r_Page_024.jpg
4f74a8e3073dc94a4e471bd7b32ea212
a51dc988a614222103cdff5d24bdcdd5b46b64ba
108788 F20110318_AACBAE rajagovindan_r_Page_016.jpg
920d050907a2845ecefe890e3ce6f490
ce907fcea95b93fd67627976f2553bf43a943484
851833 F20110318_AACCDH rajagovindan_r_Page_032.jp2
24d65dbcb6d2bd1b875a65817de8f791
d82d1fa1efb6da21a708b5cd949d7987924479f6
18294 F20110318_AACCCS rajagovindan_r_Page_153.QC.jpg
7a010e39dc1de5815096ec653cb6313f
560ef30e2e05e1386328ae01b4fa3c79bf1479a1
33443 F20110318_AACBYB rajagovindan_r_Page_047.QC.jpg
434bc4db8ab1f582c207116784aa0618
5ef15e82406d63db2d69595128d90cfa79203b30
8423998 F20110318_AACAUK rajagovindan_r_Page_143.tif
c9ee2211c44b9fa7c606637745b600dd
5907c002fb7705cdc4e1935a31fccc9e0a79b4a2
31118 F20110318_AACBXM rajagovindan_r_Page_037.QC.jpg
ca4fa8e66dc9f7fb6ac0f3ba1be04d40
4375d092acbe3bde403fe76beae1c8918841721b
31580 F20110318_AACBWY rajagovindan_r_Page_024.QC.jpg
188cd05afb513a3904bfb3d6ee700f9a
2ad0820df10fe68e31cdbd2b85a9ef56114058f3
9237 F20110318_AACBAF rajagovindan_r_Page_144thm.jpg
b9e8b26061f535410f56870ebd790f76
209741c088d78548b486759afd03d11a529a666d
922566 F20110318_AACCDI rajagovindan_r_Page_034.jp2
8e8e0c5a27608da8dde02f6c015d8b16
62d3d2fd5585fd120ff529a4e3e5192c7d41bc9e
29109 F20110318_AACCCT rajagovindan_r_Page_002.jp2
85f36a75963b152af50213769f15dfd2
3219c7b1431fef32cf8025acd5f568ffd95622af
110329 F20110318_AACBYC rajagovindan_r_Page_048.jpg
3bafa58cfd6c587438e27d3e35fbdb2a
014ccd1b78a27d24ad70d8a8657aa7da92a80451
114716 F20110318_AACAUL rajagovindan_r_Page_017.jpg
7d6b8f21738d47aca10b663c6cb0a44d
0df48a3d093aac8ee525b1ee27289b70f09f37d5
36666 F20110318_AACBXN rajagovindan_r_Page_039.QC.jpg
53926e85c39d2e2461b43ffbb030205a
d9acf9b544185aa854797aa10322437e03001e5c
60333 F20110318_AACBWZ rajagovindan_r_Page_026.jpg
3870cddaf28737af6776265beaf786c3
25e0785085f87661ff1e4995d6c7b4d536f3c1eb
1051985 F20110318_AACBAG rajagovindan_r_Page_019.jp2
89fe54f80d2477ab5b6a10cd78ac9643
96bd0887b14a8da64c1c51b9405f65ccbe770ebd
109607 F20110318_AACAVA rajagovindan_r_Page_020.jpg
b0a4d94c9e08fdfb405df05617aebf87
245923112c755111df49092ec39c8a4039cf8cf5
1051982 F20110318_AACCDJ rajagovindan_r_Page_035.jp2
cb992e8163a4a920a36adc3153ab63b5
ac24d6cf4c4158fe5c2a7dcc6ff165862f1b6390
880664 F20110318_AACCCU rajagovindan_r_Page_005.jp2
c8b2dfe69cefb367b22b88ffc5ff9313
720c929a58e0c67a7fc94c3ff758992ea1bf59f8
109616 F20110318_AACBYD rajagovindan_r_Page_049.jpg
12771da86df7f24a9e168398a9945cf8
4de4645f797a90485040ea72cf63f252e6ccade3
226415 F20110318_AACAUM rajagovindan_r_Page_133.jp2
979b3b3135b65b7a34c38c9d8dff138b
597d46ba56f9b7b658043457cb5219b6d85554ff
110683 F20110318_AACBXO rajagovindan_r_Page_040.jpg
9bb0ddd282dfb5db49c710f540f032f1
5ce61252108814b081a5d838f30de5a6d7a4a4b4
8111 F20110318_AACBAH rajagovindan_r_Page_037thm.jpg
772648b1b6840e98fa55b07b5540f8bb
3893d06c9c7208216bc8b3fbcec8f2f043539b56
133942 F20110318_AACAVB rajagovindan_r_Page_149.jpg
7f25266a9a4044f4374412a1126537a5
d87ec1db31d5eb0dbb3b33465265c6feaeb58117
216106 F20110318_AACCDK rajagovindan_r_Page_038.jp2
1b09cbe6e90712d1a89c4f1045b58577
707ff7192e1bbfb5bafb1c4be714807507e0b1ce
999323 F20110318_AACCCV rajagovindan_r_Page_006.jp2
0f94b7ab8cf2458ed19354ff5dc01875
af3d8255374a53fc7f1fb03f1015aa22481bcb99
110099 F20110318_AACBYE rajagovindan_r_Page_050.jpg
4b0b8ddbdf350db2a3941ef3c2d43a63
8fdc0c02500145f96221a00b6f87d790acb3cd37
8859 F20110318_AACAUN rajagovindan_r_Page_058thm.jpg
2840781a7babc5f6c5a791bdd691b9fc
124a2886ce902fb8dfbfc08e6c36de0fe13d2367
35385 F20110318_AACBXP rajagovindan_r_Page_040.QC.jpg
57560fc3100d8a286c640075944a6e2e
67ad887aaf027ad2195dccc23de902f852151237
F20110318_AACBAI rajagovindan_r_Page_146.tif
10ed22ee0fc0a3db193ce1c7da11ea5e
cd3259ca423a486705c4e588e3740b2dc2038071
9263 F20110318_AACAVC rajagovindan_r_Page_090thm.jpg
1bfa03b71538b4d107b1b5c318a7b1a2
46bae566c5486fe15645919191f935f108e0b0b5
1051983 F20110318_AACCDL rajagovindan_r_Page_040.jp2
183dd5bd941782e1d2486f577d1ae9b4
632fd9bd0ab133b12a46958ee71fcf996b33a939
820040 F20110318_AACCCW rajagovindan_r_Page_008.jp2
720388605939f7fd0810da710e307077
075d78d9c7e8482b720cd895f1be9bd223ebfb84
F20110318_AACAUO rajagovindan_r_Page_048.jp2
bf4bf09f0eaec02082d05ee1e6460d45
b286f5bdb843580e7422f143d0846e026b63cef0
95996 F20110318_AACBXQ rajagovindan_r_Page_041.jpg
13e950047341db6670bf321537954e12
87b0b136464dbe85f3e05e9e373776028fc50993
9107 F20110318_AACBAJ rajagovindan_r_Page_131thm.jpg
6ed2b21162f46f21edf7e57fa1a5c1f7
266c06bae403491b0c445cb8bf3ed1caf753c6e7
1051980 F20110318_AACAVD rajagovindan_r_Page_095.jp2
989e7d670678e363c614d155834a935e
c18f76eb77768dc802da92930df29865a48352fc
111860 F20110318_AACBYF rajagovindan_r_Page_051.jpg
194453434523c26863b6c308d7daf02e
790509c2071fc4bceb4ea615491fcad0912dcd9a
F20110318_AACCEA rajagovindan_r_Page_067.jp2
d99111f8eec73759bbf0e93f605e1e3d
3162b08648dffebf56964ebe2d026935560375df
1039773 F20110318_AACCDM rajagovindan_r_Page_041.jp2
45f016e51ac750999706cbeca8a52e0c
1c437746db9cd1da19a31563a6834219df0ae1b4
1051961 F20110318_AACCCX rajagovindan_r_Page_012.jp2
f64154f99f2d500259770950c90f1122
424b5cdbbe0cd23631a61dc91a4d9a8b3480ac9e
36377 F20110318_AACBYG rajagovindan_r_Page_051.QC.jpg
a11252680e1aacd279ad9edcb74b3368
ff6a9ffc5ab5ad6b48b4eb0425609d5198de1787
85483 F20110318_AACAUP rajagovindan_r_Page_034.jpg
7d3aa78b83ec711259a636b062fe663d
3c3bcb9f649e1cd7bb80e5c8cf9ef4943f9fd1f9
31611 F20110318_AACBXR rajagovindan_r_Page_041.QC.jpg
622447655278dee0a9893488b26acb74
0c4831d7d87dfa4cee99096410cebf45adb2ec0d
111206 F20110318_AACBAK rajagovindan_r_Page_119.jpg
b8d31626b18c41c79d026704cbc56138
84a67a44cde40dcf5fb795157ef47a2cdda853cc
8976 F20110318_AACAVE rajagovindan_r_Page_137thm.jpg
2020ed4c42b2d83ebe66c86b42c45420
4f6a1d881739b5c1cb467a0f547abede76423e23
1051968 F20110318_AACCEB rajagovindan_r_Page_069.jp2
4147150a9571aad5ef3f4b0d07b03ed4
ea6859ae6dea4cb407b02ef41ec2c6cbe90cc994
1051977 F20110318_AACCDN rajagovindan_r_Page_045.jp2
0a1486f9e13c40c868db4680a419e5bb
1c558167e3956a2837606d6a3d14540fb5402fd1
681876 F20110318_AACCCY rajagovindan_r_Page_013.jp2
9257e161b2e919d95f25643aa71f460a
8717ace91ba61ee63de74b70722bb0074cd3dca5
36489 F20110318_AACBYH rajagovindan_r_Page_052.QC.jpg
e875b85f906db61228c721115532cba4
dbbbd268f513eb42bce6175464bb4ae227148cf1
35465 F20110318_AACAUQ rajagovindan_r_Page_120.QC.jpg
6a18e4120f22df548be4ec5f1571e563
1f2dc320aea9625fa566337568e40f9213fe350e
102301 F20110318_AACBXS rajagovindan_r_Page_042.jpg
9311317a801cd9415fa86447f317670a
9ff3ac91af1a46c14d0dc0bcf8f4599635594a1c
F20110318_AACBAL rajagovindan_r_Page_026.tif
e4da94edc8cf015e26e15fc76613a5a3
3a931159ac5fef678b004f96fe9db6213b16066a
F20110318_AACAVF rajagovindan_r_Page_108.tif
852aa476b652f53635331530e73baac1
dcc558bc2057b591a5acfafeba0030258f1bdd0f
663559 F20110318_AACCEC rajagovindan_r_Page_071.jp2
d23358e0956b21aed55263d001ed61b0
c8f3ea6ddf955332c596bc14af83807a9d1efb3f
1051964 F20110318_AACCDO rajagovindan_r_Page_047.jp2
133dfb92d3973c72692e8e9ae30fcc41
156504881702bee86c8994cc0752529215a0c6a1
F20110318_AACCCZ rajagovindan_r_Page_016.jp2
31d2f2d940d72a7e73d3fbbd87bcb6f6
c64b5e9e3d1d902e50cd16dea4c37070eaf8b637
83997 F20110318_AACAUR rajagovindan_r_Page_010.jpg
f6869b4c29bf2450fee8eeca8df92ff0
610a0075cb790c80b65502c4268004635284e508
34829 F20110318_AACBXT rajagovindan_r_Page_042.QC.jpg
45c7e74c97cae1f196b016b0c9d2adb1
a2069953ca7bc752efa3d25adc530b3d27c1daea
37033 F20110318_AACBBA rajagovindan_r_Page_048.QC.jpg
fd6c1af5739ec1a1b72f643a7b61631d
c768191ef3f27f52b0461cdf39e985dcc7370148
5015 F20110318_AACBAM rajagovindan_r_Page_124thm.jpg
276298fe43ca082c8600a5dfa213d0a0
f6abd468fc3fadbf73a84dd728f99fde35cab816
567483 F20110318_AACCED rajagovindan_r_Page_072.jp2
c4bc8979b4d0c454aaa7d7eead1ad0c5
9abaac9c813883beecd6249959421ba4356631dd
1051970 F20110318_AACCDP rajagovindan_r_Page_049.jp2
1d29d3fe8fe8ad372c4824dc0cd8ab2c
8a0a4b8f0a38c63605e2bdd9596e6335d548652d
92055 F20110318_AACBYI rajagovindan_r_Page_053.jpg
6629b39486d5525c08d071768da13fd1
f6d83240a7ce71743659ff0535279349208093d2
F20110318_AACAUS rajagovindan_r_Page_103.tif
c46ec4659df94bc9d16746d69d591742
1e89612159d85996650242c3d0cb17c99b3f6bd1
34877 F20110318_AACBXU rajagovindan_r_Page_043.QC.jpg
80b21c2020caf87091a62ec1c882eaba
fc414572ecdffe6638bcb078ca402032f8e29c37
1994 F20110318_AACBBB rajagovindan_r_Page_023.txt
35aea64113e626de39bfbae1fe5ecb9f
cfdc95e2ffca350ff811b6008854e303d3ac2655
2200 F20110318_AACBAN rajagovindan_r_Page_015.txt
ed3ea46d9b46e9c8bb3b5ba9c667529e
618d790f4a79c5927a8761b4bd411dc4c7e6b78d
58743 F20110318_AACAVG rajagovindan_r_Page_081.pro
8d4017f61e5110892cce995a20d6bec7
0c2f9c77018f8071fc80d005ceae4c40d29b9bbe
905431 F20110318_AACCEE rajagovindan_r_Page_073.jp2
0acf69a84803e64991b9a5ac6584a252
c80d07b800e2c3e5010a875ef615e711fe65c87e
1051971 F20110318_AACCDQ rajagovindan_r_Page_051.jp2
19c3e3763cd25f9f395c2e0008f60235
b3bc619a9a6bc55ee302d40e15d2e411bbee37e1
28308 F20110318_AACBYJ rajagovindan_r_Page_053.QC.jpg
4136491aed29a6687e1b65c630c29a05
5655784913f41b4f6f18e18ac7ea63279a2bee9d
789 F20110318_AACAUT rajagovindan_r_Page_071.txt
64b76e9f88b538ea62e3839420edeee9
ce246224d709c0726da202e26574da7cbc177589
77005 F20110318_AACBXV rajagovindan_r_Page_044.jpg
95d2f54fd0e4bc5e7fac4f67a8d3c52d
ea1a64c2faa62c46446d035b399f65c129efc33e
57400 F20110318_AACBBC rajagovindan_r_Page_080.pro
a4013c3821b296c3d05cecd95e398c75
18812c2fa30f9bc18f6618409c7725865d179d81
1051936 F20110318_AACBAO rajagovindan_r_Page_064.jp2
d6540f423ab97e547f562320857ff38d
07b28fbf633559f55b33d59ff2fdfe8476b21be7
1051935 F20110318_AACAVH rajagovindan_r_Page_007.jp2
4e5629924384ed963542a01f25bc9163
61da0e05edd201c23ee477e15df32470cffca50c
612901 F20110318_AACCEF rajagovindan_r_Page_074.jp2
d80c29525dad49e9caa7094a73c8f77d
df956bbb4bd7339b216716116f81efd158fc3cba
1051974 F20110318_AACCDR rajagovindan_r_Page_052.jp2
18d6ea3932ca59b3513d8784cdd8b611
bb5713ab33580d44aa5ff723ba0d1f1eaf3b5afa
85907 F20110318_AACBYK rajagovindan_r_Page_054.jpg
d5960ac076e8add6c6ac3e052b066b48
1ef578d6aaca9bc24558ee4de2e745d8fef14beb
98586 F20110318_AACAUU rajagovindan_r_Page_023.jpg
9d3493784b0093b9acbfdb162ffbfa3c
d0534d80ff5b301d5c421dff9efb040eee483bb2
24418 F20110318_AACBXW rajagovindan_r_Page_044.QC.jpg
69f64a88d7d55dbc229a09024318520e
bee5bb23e36d3a1340f4474c8005232b2f44f752
121110 F20110318_AACBBD rajagovindan_r_Page_143.jpg
3452f74c0884d2629a8fc0d661144dcd
6fa771a0507b5cc313633b45c025ef03a5071848
52996 F20110318_AACBAP rajagovindan_r_Page_152.pro
30b463ed82359bcb096b3e11a6e0129c
f2a7f10efd3a58b4d265432f601ccaf8588ae402
1051962 F20110318_AACAVI rajagovindan_r_Page_117.jp2
55cb88b07e3a19e71e9971b80774048e
247c2b0b5974d24e12637815a167933c2624ee9c
F20110318_AACCEG rajagovindan_r_Page_075.jp2
cac6169a6b0c598d5ed022b5a388c4f4
e9c8722b61c1a10154503a557a2e831f0a78d3fe
106940 F20110318_AACBZA rajagovindan_r_Page_065.jpg
b706cc21836d05441204820a5a0106d3
e9380ec0085a8c9903e8d0fff86a8484d79122d2
24258 F20110318_AACBYL rajagovindan_r_Page_054.QC.jpg
356460c7a76472df5c4419833b0319e5
16c1f7218c911ec865775ccefeb885d42dc01cf3
F20110318_AACAUV rajagovindan_r_Page_063.jp2
97a33d978892dddc5f7aaf0dec562eb9
7a2a9b59f2016c912175a268b34ce32ee0b6a6d8
36419 F20110318_AACBXX rajagovindan_r_Page_045.QC.jpg
21e3a2c3b320e3f6023b3a6094d165dd
9ea430118fd01447c2c506a621d85a51bcee8447
16992 F20110318_AACBBE rajagovindan_r_Page_056.QC.jpg
8cf507314fb667ca6a1d41ccc49b579b
58b673bae6ddd3ba52498cba9a4d810f5d99a1d9
98653 F20110318_AACAVJ rajagovindan_r_Page_006.jpg
e24fd38204f82a84afa30f856666ffbc
7794762fef43797701c7753a7a2123ef6b04555a
F20110318_AACCEH rajagovindan_r_Page_078.jp2
8ce523e6e46c233b1c183dc36eb1fed6
cc450419761c3f5468feb46c65dcd51e55b39ae7
487512 F20110318_AACCDS rajagovindan_r_Page_055.jp2
9f55bb5007879eb9e264f88e14cae311
38034dcf0906be104c584f6a84fd8897feb86111
117467 F20110318_AACBZB rajagovindan_r_Page_066.jpg
06a5a92e35ab8748694ef5666614b3c2
987ad2179d4311abbaa916cd7a993b1a0ce0bf40
15994 F20110318_AACBYM rajagovindan_r_Page_055.QC.jpg
69382363331bed066baccf886d237eaf
b03545d1768f4dabc8345404e8710a375cc9a81e
2256 F20110318_AACAUW rajagovindan_r_Page_061.txt
b3be5be03eacc1229d41f1896076c19c
cddf8e8ac65ad8a0e271b49b72b19ef67c9e2463
111397 F20110318_AACBXY rajagovindan_r_Page_046.jpg
d56335bfdd143e45b2384ed87cda4984
f90f874a63013553487b4a1403d51b9c30ff252a
8737 F20110318_AACBBF rajagovindan_r_Page_084thm.jpg
d0529e87d62b5213868e0e03c1a77fc4
8d76f9c4b9392db5dcd5720b3d96301e9a656d6c
F20110318_AACBAQ rajagovindan_r_Page_042.tif
98268084d31eb07b26fbb0e6bff59066
dd77ec9c16beedc84a5ee750213caa6aa42dc5e9
38535 F20110318_AACAVK rajagovindan_r_Page_144.QC.jpg
78dc98b0fe73b282890b7cde90377df5
32bda42b288415a3b5334ff67f2959a29b98d070
1051972 F20110318_AACCEI rajagovindan_r_Page_080.jp2
ce9b883a0f5972fbfd196b044519c699
345347a7020dcd966041a4054a8ca47301d320a4
485422 F20110318_AACCDT rajagovindan_r_Page_056.jp2
21118f0c06827e5ee920f80a02f9df5f
dd759241e777bf3a607c7aaeeac382d102d558f7
38243 F20110318_AACBZC rajagovindan_r_Page_068.QC.jpg
e368a4abd06727580573b1605f82910e
5dd4ba9deef2de38a9405180e96a54732188ca83
18883 F20110318_AACBYN rajagovindan_r_Page_057.QC.jpg
26b7c07f4ff5116e5ef4db4cb5ee3b73
ebc3223e5aeb9abb02f9ad07477bbc502e5a1b82
36371 F20110318_AACAUX rajagovindan_r_Page_135.QC.jpg
533186c9d73473c7fb2186fa3457e4ff
d25a02f9cd6ca76bd8a0acfc8f270de181e6b1ce
35836 F20110318_AACBXZ rajagovindan_r_Page_046.QC.jpg
02137880fcd9ae1ec849cf3d0ab725b0
c3b3fe60767fc85f75a8dd1e881d94825a71d81e
F20110318_AACBBG rajagovindan_r_Page_137.jp2
8924c507d3d784ea3a74b22216b61140
fcc11dc55ee368b9b068c34a2c92566636b48c45
66820 F20110318_AACAWA rajagovindan_r_Page_142.pro
04c7b19761de8164a8b7353c6c27f3e8
6fce2573f217c1403e7db8de62ae41cbd5b8f88d
1625 F20110318_AACBAR rajagovindan_r_Page_102.txt
fcd130a5b78c741de8db1dd6c79119d7
467eab5e961760be4839c7d102ffba8d73764219
F20110318_AACAVL rajagovindan_r_Page_051.tif
50c363a271bd30de9d983fa9fd5e4968
ac683cc85c4c210ecdcf6c8bfa7cbfa808bf2e38
1051943 F20110318_AACCEJ rajagovindan_r_Page_081.jp2
bc95098108933d7e151e58932d307d49
ec8feca3f857b266d555ac19dddf301c51b9741a
F20110318_AACCDU rajagovindan_r_Page_058.jp2
6743f116e2d81ea20b5a0c3fac240253
b6f84a4b7937936f0dec6066c8069d873dddfb3c
67676 F20110318_AACBZD rajagovindan_r_Page_071.jpg
845bdcd145e0e65da5820b806a7ef302
641f0923ebfe974833ac3a0d8bcc2acc7fd761a3
110619 F20110318_AACBYO rajagovindan_r_Page_058.jpg
7c228f25006a21c1278d40bc960c53bc
8a1a120e772f3cb216eba0bc63ee9c9eeb90560a
7596 F20110318_AACAUY rajagovindan_r_Page_082thm.jpg
cf2a23a3b20ae5cea5bf231193686232
b0dcbdcb9a238c82544707f0e9fc088dd7864a8c
37052 F20110318_AACBBH rajagovindan_r_Page_131.QC.jpg
836a5e7267fa742ccc3ca2a5e712d539
cb5ca898940cf3dd83cd54c45e12408567d0e9bf
F20110318_AACAWB rajagovindan_r_Page_021.tif
186d3a184d13f0bb7b14599453d94ce9
50f5586c07e8366e361fc9ee1c31793c8bea429a
F20110318_AACBAS rajagovindan_r_Page_055.tif
e8b24d20995b7b077f2acd13c6f340bf
b57e1ce7ab3f99cfec0000711b8a9d7dc4bbf82d
5842 F20110318_AACAVM rajagovindan_r_Page_025thm.jpg
d85c71acd1b338c9ad3077f813e41ef8
f2e23953240615f6ab1252ea87a21cf286cc1348
1025724 F20110318_AACCEK rajagovindan_r_Page_082.jp2
12968187f454a4f22231137f8808e8f4
d6aa8d41fdc51ac6388775d423cd0c607f3c2988
F20110318_AACCDV rajagovindan_r_Page_059.jp2
211844d59fa857d8f9e96c32b712d0dd
2de737037c4825314e20bd5200fbf289d693c6de
22844 F20110318_AACBZE rajagovindan_r_Page_071.QC.jpg
becbe6d8cc049270d35e2d204cb066be
388e03e70c23546b731cb48f317128e5451ffd05
112767 F20110318_AACBYP rajagovindan_r_Page_059.jpg
ccf2ebc7d158e195173688f0e037c5ce
a0a2c3e06a664f3f89b2e6f42527fa9cfda9f1e5
9079 F20110318_AACAUZ rajagovindan_r_Page_121thm.jpg
4ffd2c906f1f4360b41d55fea1c685d2
1c68939d48b515444a9207f82ea5cb320b1332e4
2220 F20110318_AACBBI rajagovindan_r_Page_093.txt
6071efe9bf7a667839402bc38ed85694
856b495837c2656587a4ddc3c46bb10a718214ae
F20110318_AACAWC rajagovindan_r_Page_144.tif
afba8a044f10e8ef1d0cdabe31029641
338148a51dcd1e35624c43318ac7eb5dd5d0e7b8
F20110318_AACBAT rajagovindan_r_Page_032.tif
8f7706ec10898f1ef4af5823f8ca637e
86dc0d787b472142aa0e728ee17a096be0be75d7
1051923 F20110318_AACAVN rajagovindan_r_Page_122.jp2
d72840294a6bf8f7252fa1ff6a4d2b02
c92b2bd6d28cdefb3478b780d238e6b4b182ed07
1051937 F20110318_AACCEL rajagovindan_r_Page_083.jp2
ce178a95222e7edded1a9a89b84e4049
573aaea0f613dd02628b07b96df0767d5dc2d3f6
1051906 F20110318_AACCDW rajagovindan_r_Page_060.jp2
bd5ec7ed41044fde5aed46554ff079c0
f97ffe4f8e8e40fa8e32e09d1dce4a973ccbaa9d
58433 F20110318_AACBZF rajagovindan_r_Page_072.jpg
2335cc8abe5af2bd8af3e7e9f9fa334b
efa59840b4e4253630e36d2d5d83f08e2acddbc0
106577 F20110318_AACBYQ rajagovindan_r_Page_060.jpg
a87e57eaee180179777768ec2a0db048
d8df6fcf2d41366a411d1ea81144d9f0c71cab9e
113768 F20110318_AACBBJ rajagovindan_r_Page_085.jpg
4f7a19cc361eb65d26393a2173978d89
abb90dde7adcb25a48886a2a4186875d5aefd127
2717 F20110318_AACAWD rajagovindan_r_Page_147.txt
8f278681a1c35230516913bf3ae4d5a1
eac5fa82fcd522faa71bcdf81e6d4f7cae6eb95a
F20110318_AACBAU rajagovindan_r_Page_042.jp2
0b36a1e85a8add24474d8dac93b6daec
095d3f1b9e16e09049213e0217c271cd712a0497
54097 F20110318_AACAVO rajagovindan_r_Page_063.pro
945f7e49c3eae3d67cebb4e595a9d217
901d65dd9320dbb2838d3286318932b1fd6ba2b8
F20110318_AACCFA rajagovindan_r_Page_108.jp2
7a18c8cbb24c7bc598bb22eddc0a3b1e
d81292c9a7b062162638e899b559af5749486d9f
F20110318_AACCEM rajagovindan_r_Page_087.jp2
05e3ab04ed6af6cd5a782bfd4f94468e
365bf1892a6b9139c86e0f2767797806f9051058
1051958 F20110318_AACCDX rajagovindan_r_Page_061.jp2
a592d8565b821c322e3951783deeaf3b
9521384d5c3a7a5a5826a6bd9b2424383807a683
17772 F20110318_AACBZG rajagovindan_r_Page_072.QC.jpg
ba982a647e8abf97fbe681455a525041
d3f75830448b40aa8b909d56d05cbae3c6b4bfed
34695 F20110318_AACBYR rajagovindan_r_Page_060.QC.jpg
491f365e6424e846688674260ace6879
9527319e900b60107b2f56bb4575fae3c7283c7c
100509 F20110318_AACBBK rajagovindan_r_Page_012.jpg
af749e37ae6b94d0677a4e3b81e75b34
7e26be69be018b6716b7c5bd164651da05bf508d
39563 F20110318_AACAWE rajagovindan_r_Page_118.QC.jpg
cf4ecbd400590e0255ec23f67e2d2927
6c3878ac450dab244d069f65544a3a8d6e599008
38173 F20110318_AACBAV rajagovindan_r_Page_044.pro
1d962886e679b704eb96b7cd5194ff7e
9d79d14f9d124cb3b578ee62767323e010d93dbd
55669 F20110318_AACAVP rajagovindan_r_Page_107.pro
b24b4a016270dc6a1be32f2202617f71
5a3b957cd5aca345eb49d03e52d4d3dfd97dc55b
1051925 F20110318_AACCFB rajagovindan_r_Page_109.jp2
51d851055e4b9cd90d544f8ae0327055
a9ca95d9fe281ea45bfc084160c85f98a75f0890
F20110318_AACCEN rajagovindan_r_Page_088.jp2
9c41057b83d07a61a8a61a0d63024a9e
c8d7bb6d1e27266b8a1d55e413fb99ec95f119e1
F20110318_AACCDY rajagovindan_r_Page_065.jp2
e0d4b0c7340f64b59bb7c9650dd74861
eb2a328b1f5b4b77fbd5808a971f282ee90c0266
85665 F20110318_AACBZH rajagovindan_r_Page_073.jpg
20ec135fcf1ccb14df6abaf463a43408
f39152b164e55ed74cb9213ee648700d8dce6fb9
115621 F20110318_AACBYS rajagovindan_r_Page_061.jpg
d96e7d57f29d571e2b39039272277e16
4af16ab833952e814f5eefcc31b7262572917660
23056 F20110318_AACBBL rajagovindan_r_Page_133.jpg
a202a9295478cbedc257a19c2008a57c
08068cf36b8e182e6819d3b27c49c97f39fb68cc
F20110318_AACAWF rajagovindan_r_Page_113.tif
ff6501c8cddcf219dff279b07f36823e
a57ac9cc60b3e65d71010bdf043f89a09531785f
6395 F20110318_AACBAW rajagovindan_r_Page_071thm.jpg
bf1e26d60ffa2034345031c39e9a1de1
3b274c99be349b8b0d713bd908e24faaaefbebde
2052 F20110318_AACAVQ rajagovindan_r_Page_110.txt
0ca284a0cd2da54a4bb6d56ebc67b4fe
65183211b72e035756a7a89680528e1f275b20df
F20110318_AACCFC rajagovindan_r_Page_113.jp2
d9227e3598e5c82bdaec57dde49988df
4c2f70000d59f6e9e2add8aeae29c20f1b08ded3
F20110318_AACCEO rajagovindan_r_Page_090.jp2
3fd670cfe3eda5932900a2fe303f5413
6e5b4d520c2d635e36a6a54d836fb5f8de09c57d
1051954 F20110318_AACCDZ rajagovindan_r_Page_066.jp2
46cdb0b75ccf78e1cb331bbd7b0937fd
43fdf26db3b64017fafe0d0c8947aee7327009ae
63080 F20110318_AACBZI rajagovindan_r_Page_074.jpg
707a879f61e5dfc96773500a6acb3761
6495ccbbea5c4eb51ed4151f32b00a4b99529f3e
38007 F20110318_AACBYT rajagovindan_r_Page_061.QC.jpg
a702165c0201f65d1aef1a23e37523cc
7cb2a4a1a3ee82fbc79ed206038efbce85aa191e
2173 F20110318_AACBBM rajagovindan_r_Page_018.txt
bead6a135b85f57b1e4cf46fc4970ef0
2f510fcd81d67dbc7992b1b0e5354f7e30e2fc7e
8689 F20110318_AACAWG rajagovindan_r_Page_004thm.jpg
f6cb3f6bb7ae3bb4ba2e70c63918fcfd
5e4c1cdabe5c2d7d5702288bfbcfa990ce2805fe
9174 F20110318_AACBAX rajagovindan_r_Page_115thm.jpg
676704d20e8086fd23a6b44d1c233df2
714f9849f790ac2fc616c7104a3ea7e5c2783a96
7086 F20110318_AACAVR rajagovindan_r_Page_103thm.jpg
d95e54cb8b9f81b6a0420d32ddf771af
95f334faed22123a48a4ef98ee52117d32dd5b1e
596437 F20110318_AACBCA rajagovindan_r_Page_070.jp2
61ebaf52efe44b8652e3952103dc2ce5
e35dd05ddda60a68f862e779b5f5a93a6ab2f37c
1051966 F20110318_AACCFD rajagovindan_r_Page_114.jp2
261afefe87024af05a3e96a524a22d20
5690d376249d63654628753a9057aaf19ce5e65d
F20110318_AACCEP rajagovindan_r_Page_091.jp2
ff6db5e02944c7878a9669f0f1052ed0
b12516a8fae31dc86d362a414ce9c244fb6ff964
104355 F20110318_AACBYU rajagovindan_r_Page_062.jpg
323724c4cf92cc7e16bc461c68fc39ba
6fae774475d0be9f6a337ff75190012faefe0112
F20110318_AACBBN rajagovindan_r_Page_014.jp2
46890900d603c1c34f2efc8244f7f57c
932b1a3a31b67936c037c82206b11417864cc56c
8947 F20110318_AACBAY rajagovindan_r_Page_096thm.jpg
1d48663e16fae250af84dd7cb5dd1358
be15f1dd72c17d11b9180464d0c5369a5bfde986
F20110318_AACAVS rajagovindan_r_Page_144.jp2
decdb3a7d6f47f3d4229833358e61929
b6c41fc70a4799d47f8c24a0cb6fa18715fde338
F20110318_AACBCB rajagovindan_r_Page_086.jp2
321156ec14018dc924b165f2b5cfd51f
5c66dac1fb634c8b6cae1ae8cca41748c85c6a6d
1051979 F20110318_AACCFE rajagovindan_r_Page_115.jp2
2debd814127bbb40b835e06cc6703cbc
c0c6fceddd59401609b0625339a7e6ac29af1e52
F20110318_AACCEQ rajagovindan_r_Page_092.jp2
03dba04d38e0175475100da4831a808f
8008025aa469cfe066e7bef348ecb31aac8500a0
19862 F20110318_AACBZJ rajagovindan_r_Page_074.QC.jpg
20c3260cfd03f535251bd047d2816675
fd92eec225b38335c2017ad2ac506fc5a47540b2
35133 F20110318_AACBYV rajagovindan_r_Page_062.QC.jpg
1f5848506a5acc491b6d4276120fc963
813e2f36a8e097702517f90ec206cca0d77c2698
F20110318_AACBBO rajagovindan_r_Page_132.jp2
5df4f9236e80afd74dc5199c160c7b10
e649bd4bc19cfdb33e29f117f8324dbd0b30145b
12559 F20110318_AACAWH rajagovindan_r_Page_009.QC.jpg
518dc567fd6fd82e29fe4211c684d7f1
3080b3a848c849d0f3c27a5c2eb98dca0735c63d
29245 F20110318_AACBAZ rajagovindan_r_Page_026.pro
b66cc9fb95e9f92c167fa7a59439481a
ce8f55e88e6be166ee6fe18e5f2ab5ca264ec24a
F20110318_AACAVT rajagovindan_r_Page_141.tif
3d93c94b1a2a2e257bedf303926f3d32
17703196658fffdcb43a62848a7515e442a1888e
109396 F20110318_AACBCC rajagovindan_r_Page_014.jpg
a76c030ed96005e9b70b74ca355aa7c5
a1171a5dcb295f12d1a15b32bcb5025db3412315
1051932 F20110318_AACCFF rajagovindan_r_Page_121.jp2
5acb58829c85ea97e9042406444aa460
9fcfb5803a655b7e671bd773383721a721a376ec
F20110318_AACCER rajagovindan_r_Page_093.jp2
33d9b5ffaa4bd7aa6f04c188479da470
203d99ca948e0d10ed86f80634ff6f72cae7ecb2
116423 F20110318_AACBZK rajagovindan_r_Page_075.jpg
9c844c0adaa828818d6fd9dadf582715
5be1b1d041b8457f4e655e73411de1bf0cac407a
110393 F20110318_AACBYW rajagovindan_r_Page_063.jpg
be7d477c6f43955a3af4a17e7df5bddf
c9a0832ed94728ad6cf72cce3f7196589360e41a
F20110318_AACBBP rajagovindan_r_Page_005.tif
6bcad2d7938aa1280290d7aed3a92822
366ab3fece43b5b4c428e750555977d8cf62de34
36700 F20110318_AACAWI rajagovindan_r_Page_084.QC.jpg
a69343aabdbfa8195d74802d7c1a683f
76d7f7c6e6b59e7a4440076ebdb2ce6a3f3d2a7f
F20110318_AACAVU rajagovindan_r_Page_135.tif
11488fea751f6d6b78307742983af660
a8f147dba6ae8f8d67b9de516c226409c0691ab6
37768 F20110318_AACBCD rajagovindan_r_Page_146.QC.jpg
5e765d1e81645945667891cfa821e056
23ff3dfba1e365e30ee50a0b8e0b35b1c578ebb1
F20110318_AACCFG rajagovindan_r_Page_123.jp2
0305832a2fe00b93fe788b7e0a16afd7
2a60ef289115b5865a4b4d4a134f53e793c0f1ef
F20110318_AACCES rajagovindan_r_Page_094.jp2
a4eda66bfbd4337b84b514b99df3caa7
0f5f99a9181181a7aeacf63d3015676bcfb4d42a
111589 F20110318_AACBZL rajagovindan_r_Page_076.jpg
997557081a99349489f2bf1a7d137075
65f8539db0eb7506a98ff6d7b8f1647ba2977d4c
35904 F20110318_AACBYX rajagovindan_r_Page_063.QC.jpg
9fd43de04e8bd25a2c8fd4d28eb86779
db04382f6406dd298c2bfa8fd471b876a7aa2e85
102640 F20110318_AACBBQ rajagovindan_r_Page_110.jpg
9d248600d4327508aa86d4a9778913a8
7f2e09a66896a31b164a42822dea6541fd816b7a
37100 F20110318_AACAWJ rajagovindan_r_Page_119.QC.jpg
fa5bc5d15586f9f3f1a6c622383e49ca
4c7cc74dc9d29ede5bc4af1ab6b63a90210a3b1b
F20110318_AACAVV rajagovindan_r_Page_089.jp2
3515baedbcc4a092fd9c5319e2420bbf
35a041110027a8b79612dee3b19ed0b8dc1800d2
2688 F20110318_AACBCE rajagovindan_r_Page_142.txt
085c0347115dae05695e4c6ddaf50c90
48d623ee0d0059678746789834b8f5e79d0a6c72
399192 F20110318_AACCFH rajagovindan_r_Page_125.jp2
3f81e0bb36b77dda3711a232d164c776
6034f2af07d9c0e0e755094837f73986cb18e3e9
35745 F20110318_AACBZM rajagovindan_r_Page_076.QC.jpg
475b3015ff31b02f1bce765a33e67688
03a3b9bc91f0c2bba7e2764f9c91c317d49fccce
105215 F20110318_AACBYY rajagovindan_r_Page_064.jpg
d8793c0d4e1d1fa4a22179113749fd33
d52ab2c983a63a742cceedd973758122120a8375
F20110318_AACAWK rajagovindan_r_Page_029.tif
699aea5afca6f1f29aaee40adae464ef
48b70498c7791a657f981f181f407950eb6ff389
52859 F20110318_AACAVW rajagovindan_r_Page_058.pro
b65b16bc4f89b13f7fa79c58294e1afc
7e794542522b3b55f537cebac19e1dd1a4e030d0
55618 F20110318_AACBCF rajagovindan_r_Page_059.pro
fdd95e516ffabb499ecb5d28e4406c7d
566f15c387b2c08f99a4756590370eb6fd353746
625474 F20110318_AACCFI rajagovindan_r_Page_127.jp2
26f53d54c5728643ecc052f47746442d
730c9503be1dc7d38b46a269f8560d8ae268a5d4
F20110318_AACCET rajagovindan_r_Page_096.jp2
2b84a46f1a94bfde01320cfe8257f334
5ffc67ad1b885b2ed9f3dd029d3ed2a6d16eaede
108276 F20110318_AACBZN rajagovindan_r_Page_077.jpg
7773be5b8a701b46106884fabd2cf624
b8e2a753891cfa75997f7fe1f16158ff49bc75fc
34372 F20110318_AACBYZ rajagovindan_r_Page_064.QC.jpg
3a3ca204bace1ddfdcda4a17421378a0
b972506d24a870f4a3490098a25f8ef5ef2b8cad
2199 F20110318_AACAXA rajagovindan_r_Page_050.txt
707ea7232dec584b7c82db2b41e228a3
1db51bb26ad769553aced6314e296b8068f703f6
1051965 F20110318_AACBBR rajagovindan_r_Page_119.jp2
dbf611109e2cdaafd8136bcd61027505
acc524a51d790adad653ca4f6bacbf6ac713e7a2
36111 F20110318_AACAWL rajagovindan_r_Page_033.pro
15c6fc240d321226b05a45db2e32e8f6
dd5d8fd8096eeb4c5a3789a7faaa80a914308879
2122 F20110318_AACAVX rajagovindan_r_Page_063.txt
d01369b3da718fa11e6e92b7a5c5dc9d
88d7f523be64e7abb5133d8efe1d21c5dcb706ee
1375 F20110318_AACBCG rajagovindan_r_Page_026.txt
11dc0703a88828c5a98e1a8c1cdf5fff
c06f0e788f848e9d2d97cd3ef742da46b98fb311
F20110318_AACCFJ rajagovindan_r_Page_129.jp2
82179e6c1308163821de75fe2e69a0bd
1ab78777ac42daa09957bde3a37b2b3f2e7459c5
530027 F20110318_AACCEU rajagovindan_r_Page_099.jp2
ddc4cb212deb0bb4ecd6851f234c29bd
48ce47cb6841f7072357d7e934b7f60c92db1eff
35679 F20110318_AACBZO rajagovindan_r_Page_077.QC.jpg
7b42a1f23d35292b50b5801a4da858db
995b8bf56f0469e02486e00f780e7a305e5a30c0
53186 F20110318_AACAXB rajagovindan_r_Page_004.pro
f7de50b033f0c20063a92a26c612015b
7c1f0b7fdceeba9651e8cafecc83011ec63c035f
55909 F20110318_AACBBS rajagovindan_r_Page_094.pro
9e0ecb617adf187daa11caee3306ef21
588d944bc6fddc63900d47ab03c86e3846a430db
22988 F20110318_AACAWM rajagovindan_r_Page_006.QC.jpg
9aa97e7aa591db3652d76d19732e297e
71253c2605a9bf4670cd993ec4d17adc7a449edd
8798 F20110318_AACAVY rajagovindan_r_Page_146thm.jpg
2c2378d34ab70745bbd60cd5a1155be3
04f7d0afe908a226b8de84558ab6d409821ecc0a
61188 F20110318_AACBCH rajagovindan_r_Page_151.pro
bd4c4ab377b4dbe6e53c7e64c54e9607
3ebdb9a798d9bbe3a682590f20f35520c96e210a
F20110318_AACCFK rajagovindan_r_Page_130.jp2
50b8c8415591338068d22f07015716b4
2379de1b060eea1830bd40a44df68ee3b5fa36f5
1051779 F20110318_AACCEV rajagovindan_r_Page_100.jp2
760abe8ce520bb9f03862a359a8fb1e6
5bc6e56a122aad94e205f7cf44f745eca1ccead1
112337 F20110318_AACBZP rajagovindan_r_Page_078.jpg
bcdd4c121505f89f598c266700a80209
0fa548c884c0a31812a19ec54480490a90ccfe6c
35043 F20110318_AACAXC rajagovindan_r_Page_116.QC.jpg
90e3e97fbe0d02392a0e3d651adc4bf4
2ebdc3ac46285f6685533dced9afaf16c5b7e9b5
79169 F20110318_AACBBT rajagovindan_r_Page_029.jpg
b66997ac0158ec2fac25d2c8429877df
b13227a573816327b9d02e1d844cafe64b7faffa
110282 F20110318_AACAWN rajagovindan_r_Page_004.jpg
e26554a1a6e01ab9c09c80f42634849d
8f1f8472bd73381f535ea44f2fc33a0fdbee44df
118993 F20110318_AACAVZ rajagovindan_r_Page_136.jpg
428d36ecd44c747dea1dfc949bee9755
1070d929536dae9f73c5b16da63441d25553699a
1538 F20110318_AACBCI rajagovindan_r_Page_029.txt
125c89857f2094b4d96363a66ca6c54d
f1cf4a97e6501c24cc5660e1b26b55d672850857
F20110318_AACCFL rajagovindan_r_Page_131.jp2
bb7f4a7a3a9878b414c1c1ff16179bb0
9e5d9ee43860fcfa5b431d0e55f6fc937b00b0c6
622721 F20110318_AACCEW rajagovindan_r_Page_101.jp2
071cfe261e374e24a567b70ab0f59ddc
f65f7f63231516aa4c1b376002040e5639894b79
36508 F20110318_AACBZQ rajagovindan_r_Page_078.QC.jpg
b38760c3a6c9aa31d9b05ecfa3b5aaec
d5fd1c78a17bf6e0dd7394e0283a73aabd10f6af
8876 F20110318_AACAXD rajagovindan_r_Page_136thm.jpg
703f48d7a564d7343f366104cb07f77a
db55c8463faf2239366590cd543631b8193fb740
2085 F20110318_AACBBU rajagovindan_r_Page_082.txt
818e631d2cb971a0c1132ee39f9d0af7
a63bdcbd28619ebe8701c1f931ec27d1f8e89547
36707 F20110318_AACAWO rajagovindan_r_Page_140.QC.jpg
07917281286c1c2eacac131adfb8ef3b
3617c667d3280c3f86e6f2df4139f4e245b2bf04
38359 F20110318_AACBCJ rajagovindan_r_Page_069.QC.jpg
a8a7ed487db6c6285232032c871a474b
335547610133e466e0f040eb5859caf0059f257a
F20110318_AACCGA rajagovindan_r_Page_151.jp2
bafb5dc22cf74aa8c58609db59a9cecb
aae58e6eb0b506a69f5c78db3b0447c6b9f66ddf
F20110318_AACCFM rajagovindan_r_Page_134.jp2
49f84b18ca3612619c015ed6a2d005d3
6a41f8febb6d302c8c459db96f9a179acafad534
457946 F20110318_AACCEX rajagovindan_r_Page_104.jp2
94d0f80e30f614a1a8ffa03951c0d331
bd1dd0871e4caba78e8b84dcf8863b40c3d522fe
111504 F20110318_AACBZR rajagovindan_r_Page_079.jpg
5f82f56ee9ef2a0575615e9516a98593
941ff7ddbacc4b244a2d194be03c8b79c04d632f
F20110318_AACAXE rajagovindan_r_Page_085thm.jpg
1080de1cb104bd19811d3bcafd4461db
f514727a1d18f22bc563a5dde03e286b550e6e9a
1511 F20110318_AACBBV rajagovindan_r_Page_070.txt
f37820f34f1790d61935fe709a6c35cb
4a3f8b59e6a99d5bc8127e037332bc3f28558609
108420 F20110318_AACAWP rajagovindan_r_Page_130.jpg
04a1be4e8fd79e0b166cb40c19716432
47fb12554c05b2b46856eb2d6dd5e675c2cf81e0
F20110318_AACBCK rajagovindan_r_Page_024.tif
a6cba1e248a8a1bdb3ca89f301411855
44aacb7282ff6b1f32edec30e29e033fc93567c5
F20110318_AACCGB rajagovindan_r_Page_152.jp2
0008e001f6504fbaefa2744239aa65ed
79613f41ac7edcab096c4ba62d845bafe6c8f2fe
1051960 F20110318_AACCFN rajagovindan_r_Page_135.jp2
894cf8964846e2044dc67a9c1b2a4358
6d10ed41120febef8386871b82f3c6d8da3774d7
409326 F20110318_AACCEY rajagovindan_r_Page_105.jp2
77773850540805de1bdd76f9b37b01d7
01409ab30296a75f050f67e73c4ab3396b3a6cbe
37062 F20110318_AACBZS rajagovindan_r_Page_079.QC.jpg
c51d8fd1ff0c3716ae421cc11b6766d0
bc8dc56206575f459604cc7378fc3ca53613fa3d
F20110318_AACAXF rajagovindan_r_Page_097.tif
a47ea66beedc3a46754a78221359c247
ceacce88f3042463ea733698930465a1aa6e857d
2609 F20110318_AACBBW rajagovindan_r_Page_138.txt
091e5344a44c620131627dd3845281f0
4a623713b9df6f9fdc5f98581ba2111c5c0b52fc
66159 F20110318_AACAWQ rajagovindan_r_Page_149.pro
b542c8efc17a0b1a6996f06a23f6b339
2d26152ceb124aed08e00c1395d877d1448f2f95
113528 F20110318_AACBCL rajagovindan_r_Page_086.jpg
73ad821b2fcaff1219fa0f6039152c50
a753a37bbe05d1c1ef772b3f287a941aaa342bed
578 F20110318_AACCGC rajagovindan_r_Page_002thm.jpg
60551d5d25fc8a693478a113934c8597
25595166aca90f5bff88d3128046425344e5f31c
1051930 F20110318_AACCFO rajagovindan_r_Page_136.jp2
08a171baa1e6945bf92bc2857f44715d
8963ae06d1e8ce0e0d536ed2096a785ac6d6e041
815268 F20110318_AACCEZ rajagovindan_r_Page_106.jp2
ede119f52fa20114578088a201d8f580
47e6cbc207fa57e99f97f2a8373e8ab237acffca
116092 F20110318_AACBZT rajagovindan_r_Page_080.jpg
30124b7cf71e390308bc4eaab6ac038f
6e35760021fc07fc4f895f908cf175fd743f608d
2245 F20110318_AACAXG rajagovindan_r_Page_085.txt
4447a91d7a8b63e7d1cbcf4218aca397
16102fb42b9d4e3e49710817b6e630aa2c022d96
F20110318_AACBBX rajagovindan_r_Page_069.tif
06a45fa47ae1cbc804a8945a9584036c
38470f713fd6c0d26276264cd55cfac75aae3406
59985 F20110318_AACAWR rajagovindan_r_Page_070.jpg
34b02f241f80f8a02196db5b4e6c522d
c9182f023978a7b79c72cc5cbd82a13f6ffc4f64
54341 F20110318_AACBDA rajagovindan_r_Page_111.pro
850e7c6fbc77ac1fc317af24c27c5f87
1be2fa35bc6347fa1f0b7798ae916fc64a50c87f
2197 F20110318_AACBCM rajagovindan_r_Page_094.txt
7339035ed62733c22a628fe81ed7868a
03d99cb2475b2ffd5e013df11bcc73efc5dd82df
6672 F20110318_AACCGD rajagovindan_r_Page_005thm.jpg
7826bc79432cb24668a0885e4ff2f529
826e4ca19314ae4345670e87e73a20f76a85fc8a
F20110318_AACCFP rajagovindan_r_Page_138.jp2
34fa6c0c671111d96b1483da3fc51d32
f83bdaddccdfcb4a416620829b9abe249acae7c0
38553 F20110318_AACBZU rajagovindan_r_Page_080.QC.jpg
3472eddacc964c999a3726f2726c0d3e
cd6ab2a6e44a1fb3b48338cb29e22d2a89e0249d
6681 F20110318_AACAXH rajagovindan_r_Page_033thm.jpg
e382c0ccd30c7ee410f3fad39913aa2d
b9b0baccd9d7e10270d0c1ea9fed19d098c74b2e
2477 F20110318_AACBBY rajagovindan_r_Page_151.txt
532197b00e99e4107dcf0c3cde7d2399
c625677794b71dcddf592f5cd0b05c4f66dc364d
142853 F20110318_AACAWS rajagovindan_r_Page_139.jpg
01550b2bb4cba6105d4231dc02ff1189
d4f6b2bf13174757811fa849573cb31c1bdffe12
21014 F20110318_AACBDB rajagovindan_r_Page_025.QC.jpg
627b859293f9128aa788d40bde6d53a8
14d9a0354f4015eaecc84c756286fee11818ef9f
54629 F20110318_AACBCN rajagovindan_r_Page_092.pro
b3f4b97a27918b8d817586be41e5c7dc
a548754db8f63c7cf3cd9043b91eaca733a2065d
4759 F20110318_AACCGE rajagovindan_r_Page_008thm.jpg
58cff54dca26b3c41ef98ef3e89ad105
71e4461047d7b37ee9aea9fdfb2082c6c3246ee7
1051948 F20110318_AACCFQ rajagovindan_r_Page_139.jp2
da84fa104d76bd52fadd16a9b4bf76e4
a27d9506528c50964455e8b523a2de36b1ad6bb9
116923 F20110318_AACBZV rajagovindan_r_Page_081.jpg
6350e63c6974a7e1e3e5856e799ed001
7283c9c9a1dd6ce0e6c7e123df28dd1772129a99
22357 F20110318_AACBBZ rajagovindan_r_Page_011.jpg
2750b99f017f961b71e43b7461c1c9cd
40eb59c7a3d740b4d83f716b9cbc6f617be3a980
37019 F20110318_AACAWT rajagovindan_r_Page_015.QC.jpg
902bf1ca6d490a4c6e18eacfcf605f45
1dbd6ac0d4aed2ce755c9e7c8fab12c6a122262b
54937 F20110318_AACBDC rajagovindan_r_Page_020.pro
2fcc54198027846908a026ee923dd43e
a30e92f4bb97345af897fbc1de46b6840202801d
63456 F20110318_AACBCO rajagovindan_r_Page_013.jpg
3f12dff14316301a568c110936af1c4f
08893fd6670e1bc324f00186afbd3de6dec2e06e
6204 F20110318_AACCGF rajagovindan_r_Page_010thm.jpg
62b43b470073c1c716db1be8cc6a0761
390068b62df30deee6fdc825830907fadb4bee99
F20110318_AACCFR rajagovindan_r_Page_140.jp2
aa7936c2301d28e9701c3f07d6e30fb0
758195b8974e953bf8c4ec98d9ef7a43d5bcbb7d
30880 F20110318_AACBZW rajagovindan_r_Page_082.QC.jpg
d569c155f5e801b03fc20b1fc342fa3c
f2c60fa910310baefe7913bf8f1260ddcc7b0437
F20110318_AACAXI rajagovindan_r_Page_030.tif
5473a4bc950cd56b2317ceeb4870d878
1eca683a8adcfc02da5845f06f1d1905a95d2abd
61611 F20110318_AACAWU rajagovindan_r_Page_140.pro
3517d862047725fdd1b9aabfacbeb646
a6e8654fdcd60ef4b34fccd10fb180b1daa60687
F20110318_AACBDD rajagovindan_r_Page_116.jp2
96503563a7ff6f15cf85959cd36882ad
5c4f1da538b15c27a103d006007e6a54ae0aa44a
F20110318_AACBCP rajagovindan_r_Page_015.jp2
5021aa15abae67ced57f8a6bd2ad25f4
104ea993a14aa06c98952a3c154d83c6efcdff3e
1770 F20110318_AACCGG rajagovindan_r_Page_011thm.jpg
782cb0ddf749027efdc0e3f5085cc048
579c63c60c768fcd793240b3e0de64efa58f0f2b
F20110318_AACCFS rajagovindan_r_Page_141.jp2
0c691c1faaf87f6d4164c2b55286b110
6388acd5fc9f4cb0287fd45da8db350e8bf70f92
99211 F20110318_AACBZX rajagovindan_r_Page_083.jpg
f44e7d38e90f4b31de69a8163e1a746a
bfc467c3474695e4db8943c513123478d7139db9
31669 F20110318_AACAXJ rajagovindan_r_Page_152.QC.jpg
7c2fb199f241541463a9da86401cdda1
254254649a7b8d9447c4cbb6150cc14473b3ac14
55829 F20110318_AACAWV rajagovindan_r_Page_051.pro
89a844f6cd7af951eadc931c48edaf27
9d018f8082821703d8ea5eabd9b9da4888ff0d2b
2192 F20110318_AACBDE rajagovindan_r_Page_119.txt
73eaf508a1b885dfd6450f3062d9322c
d4faf9ae208ae099424f6d636e4445a8dea0ef89
623567 F20110318_AACBCQ rajagovindan_r_Page_102.jp2
b15392357a9c8f90cf8f9520b625b235
c90fd70cd9e1fdcced1ab325186a57532fc2f804
7727 F20110318_AACCGH rajagovindan_r_Page_012thm.jpg
179cf4feefc241341b5e9723bbdc2350
4dae0912de80e19a4b1233b24ca35b9736aedb6f
1051942 F20110318_AACCFT rajagovindan_r_Page_142.jp2
23d73f23ea8a9d79f32d71e961b0a3c1
15e5f19755ed5b708f429d3c3cd7f09b04c9760b
37350 F20110318_AACBZY rajagovindan_r_Page_085.QC.jpg
0788c5f3c71d24b9d0d8b5232d2c530c
9ede15d66efd4f96b2fbe8065e761b8ee69df034
2065 F20110318_AACAXK rajagovindan_r_Page_064.txt
7d5d57fc1c4b6c1e35ecc8985e026e3f
bd812444fb03a94401b24eecb0b7b24dddd9fb0f
38904 F20110318_AACAWW rajagovindan_r_Page_081.QC.jpg
7d7bc6febd9083c954e1b8f65f598f4c
813848efb6bb7588b476f39d6f7e0da54feeeb5a
F20110318_AACBDF rajagovindan_r_Page_119.tif
1219ce6f6ef022908055eed6429569fb
4c945d1d939983c8c494a5ff654584d38c8acd0a
2185 F20110318_AACBCR rajagovindan_r_Page_048.txt
d20d9c34bed39375fc5e401afbb7bb1e
8fc165dd40ab332637f098252be8cd38bd4f6f6c
4995 F20110318_AACCGI rajagovindan_r_Page_013thm.jpg
f782b2e46b8e8a8d2564579f1617f872
f80c31010a51cb80bb13c1ba75dec4e0826c5276
37951 F20110318_AACBZZ rajagovindan_r_Page_086.QC.jpg
82c0ad59bada0f72228418b5dc7e35cd
4563bad6b15473b00c4c6ceaf4b4b6e12e625219
8830 F20110318_AACAXL rajagovindan_r_Page_145thm.jpg
6c16d0d38afb15310ab4dc8af3d98366
1d2f11a9bd3568f46e9326f46343f8736a2d1a15
50763 F20110318_AACAWX rajagovindan_r_Page_042.pro
3a1e8dbc5b1f376e912de8eebe639800
42ffee72c5097ceada405acba123a983d875d05f
2167 F20110318_AACBDG rajagovindan_r_Page_152.txt
0dfcb139dc333b40e1b4c4847fe1780e
38d548f3ae25a2f5b4ad3ec89c3551c37ba5fbf4
110571 F20110318_AACAYA rajagovindan_r_Page_052.jpg
301d0dbce231d402f8463abf34c27a57
29d8f980c56ef20a02d52337108941a737e3b2cf
8704 F20110318_AACCGJ rajagovindan_r_Page_014thm.jpg
34dc1d00958622e980167033829a69bd
9721970961f24814ffae27471135688a128b0def
1051973 F20110318_AACCFU rajagovindan_r_Page_143.jp2
719d3357e4f85e845b820b69d63daba3
19d50df9ffb5dd61991c22cda87e1b3b0e539ead
2105 F20110318_AACAXM rajagovindan_r_Page_012.txt
d2e24f95cce2414b6ce4a3ebe17b8536
18bb5b3a0f0385cc1278cff49c6202b3482bf979
8984 F20110318_AACAWY rajagovindan_r_Page_150thm.jpg
5eae8aaec63ebf120b2fd9855e6fdf74
79884ee99c54484c1b9bdfbd8e651e260213572c
54452 F20110318_AACBDH rajagovindan_r_Page_079.pro
f0132d8802f7e3b6932ff83eea51c716
6d0928b64f00ba16b5264a9945a5389d38f0da7c
F20110318_AACAYB rajagovindan_r_Page_067.tif
f721e66bccf59718975bcf6a16c0246b
95bd6b5a0550fc22b7009bd1ba7e1461f587f2fe
59305 F20110318_AACBCS rajagovindan_r_Page_054.pro
91c5778715cac52929a7a74734ee3ff5
965c03c6c28dc879553ddba3dd2b7b9912762d97
9113 F20110318_AACCGK rajagovindan_r_Page_017thm.jpg
774d5983a8a0629a3a1ab113d2ac768a
d4f63741b5780a48670c4cbc14f152d834f5e20c
F20110318_AACCFV rajagovindan_r_Page_145.jp2
22f705dd1235f611f105b752c0a6d583
2b5755694ce7698a095e6e08e62a2f910d68bfde
36798 F20110318_AACAXN rajagovindan_r_Page_134.QC.jpg
3c18bc1d7ad408043a93f5117c33507e
5427d65edff95cd4dfb4007dcd8be93ade2f2f8d
F20110318_AACAWZ rajagovindan_r_Page_094.tif
08d97ade6df706b697252b92f50dc240
9a5ef9b82a7ca193ca45ece75d916eb39446756e
1051934 F20110318_AACBDI rajagovindan_r_Page_022.jp2
44b38a100873c71ee625ed6c0b84bc70
ff31df4f0db0df387235cb4e1d694c6614c9b02d
6874 F20110318_AACAYC rajagovindan_r_Page_007thm.jpg
3c31bd7961bb72841019864834842f99
ed8ac92e8882380d59eed87d2c42eb27c29c902c
60496 F20110318_AACBCT rajagovindan_r_Page_057.jpg
c735c77d9445f37f2e73608146f019bb
ca75b85150b28d581f2e453993448fc57ce2493f
8619 F20110318_AACCGL rajagovindan_r_Page_019thm.jpg
038869f01ee188b923acabbe72a77d40
9e58bf5d411f158f9291c8a4593fa49591f21368
F20110318_AACCFW rajagovindan_r_Page_146.jp2
18fea4654d74637d3f73f412d44459c6
c285dea9efacd758eaac07b918d4f7cd20d02197
113433 F20110318_AACAXO rajagovindan_r_Page_132.jpg
9d2df8c189f6f32be77dc453cd9dea39
f14e19081c8d20e93d34f3e67979f6a9564307e4
34565 F20110318_AACBDJ rajagovindan_r_Page_003.jp2
e7fb204d563580a567a5acb9c3f253e5
016ecd643e4126bd007bae59478cdeb04317f40f
2161 F20110318_AACAYD rajagovindan_r_Page_058.txt
1f1a851955f2fab218697be558e96ad4
756e78a95e9c3240e906fdc31c5de678583c3d27
117973 F20110318_AACBCU rajagovindan_r_Page_095.jpg
c688de098f6a6c192783165293fd70e6
f46595e322d38a476d49a8ab68cd20ac511c7f44
8665 F20110318_AACCHA rajagovindan_r_Page_048thm.jpg
2d4d5f436608d8bbb6cd4ffb52e9b1f3
8123a27941a6cab8e92836d2a52be87a2012aa07
8922 F20110318_AACCGM rajagovindan_r_Page_021thm.jpg
2533c09cc33c291445e8a5cbbcc88ac7
10c67092af432d85de332a2d6449b1e0978b6df8
F20110318_AACCFX rajagovindan_r_Page_147.jp2
10e652bcd1874125091def9996d3503e
3d4631a4c1a0d42b74746bb49a55339f09231dd8
35805 F20110318_AACAXP rajagovindan_r_Page_036.QC.jpg
548e9b012e6feb62d3a308298489d3e8
f37159479f862482a37eb0810efb75e1ff4bc1eb
1008435 F20110318_AACBDK rajagovindan_r_Page_053.jp2
74d6f47b50f4c474dad03afb347bb539
a975af8359496278928aefb0280258877237a3c8
7531 F20110318_AACAYE rajagovindan_r_Page_038.QC.jpg
13e6c7e5db19cb59eb2215e3d26df6fc
9a480fc544e09c63e046392ead441653b0e63e0b
37032 F20110318_AACBCV rajagovindan_r_Page_111.QC.jpg
e651ac69a97e82b86bf4dd0efead32d8
b8191d75d90076eccc2ae3d6d8b4be3ad7b513fc
8535 F20110318_AACCHB rajagovindan_r_Page_049thm.jpg
56e7c92f76a1a5a768c15f7993108836
23eeb7e40abc0ccf2a21308a9c53e81315a0bb59
8415 F20110318_AACCGN rajagovindan_r_Page_022thm.jpg
c9bed6b878dd8ef7d0c95f86fb81347b
0532303ed2cb0af0414bf399e636cfa4b2ef55b5
F20110318_AACCFY rajagovindan_r_Page_149.jp2
8a932d31439e360073d76046dcc9baad
40d427c0c35272fc652e8f31b3c8add444f3c1dd
47836 F20110318_AACAXQ rajagovindan_r_Page_037.pro
12c86b8825ef3aaf87a6031bc600e3d9
e153802e89363391549d7c25c0e51fbc389452f6
F20110318_AACBDL rajagovindan_r_Page_038.tif
2b969a621dad7529724003f32bc76b9e
5953db3444fee81dd25c114098c19717aeb004e5
401864 F20110318_AACAYF rajagovindan_r_Page_009.jp2
912175f544c22bd602c8a1943482d566
c0f967eaf25c5eabf72a2b239972d6aca6d11863
131945 F20110318_AACBCW rajagovindan_r_Page_138.jpg
a7c2e280f1014867f7dbc3baa9313a05
6f6060961203999ac4e91d1e8d892d61457aa082
8617 F20110318_AACCHC rajagovindan_r_Page_050thm.jpg
0f277e01975a6076ecdccb349e0df7ed
65e4ad95f3481d417734cf63241fdf274b4e7aff
8236 F20110318_AACCGO rajagovindan_r_Page_023thm.jpg
a4b4d8f7d42c86b7d153b45a308fd0a3
0b9fd14d8f81d96e08cf6ebf9ed604404613f9b3
F20110318_AACCFZ rajagovindan_r_Page_150.jp2
7e27ac002f6a4c83d2c62d3d6398e4f3
37ec242710c310c08eaf4adb7c48bbc6c0b96753
41593 F20110318_AACAXR rajagovindan_r_Page_034.pro
86d427340dbcde8a79636fc300d72e77
1b3cf577bbc66aea6ff9b8c5c5d0dd7d40719e76
7022 F20110318_AACBEA rajagovindan_r_Page_011.QC.jpg
1448122bc8a7d3fc2172a3174194a061
0e75f39817944f7ab730cd6f3b4078d663d4ed07
F20110318_AACBDM rajagovindan_r_Page_091thm.jpg
3e9fd8b33bb8e54cf897e56bd78a9c3f
3b11edce04bbc138f1dee69cace35304169d6f1a
38986 F20110318_AACAYG rajagovindan_r_Page_087.QC.jpg
1686b9ae27c1ade4e740bf12a654aa4a
4745a4e35a5782386a18f1a850dd932c0c236d9c
36952 F20110318_AACBCX rajagovindan_r_Page_128.pro
59750b2f26919a10581589a17dd6cb95
e53ea247220f8622ac9d291a0f6afd93122a2cb9
8966 F20110318_AACCHD rajagovindan_r_Page_051thm.jpg
372826954c8573ce122ab84875c8d64b
52a146fa40411bb7972ac6efddb5451ae11c1e4f
6024 F20110318_AACCGP rajagovindan_r_Page_026thm.jpg
b9e74a192177fdfe707fd67d93fb234c
686f1acf8a66a618f89312452f20dfdee9b87747
F20110318_AACBEB rajagovindan_r_Page_006.tif
d87f4b435074c2ce3d2303089c595f99
3bd8858f6f31ac99dcb6a025afebf555dc0e93b6
7540 F20110318_AACBDN rajagovindan_r_Page_100thm.jpg
387933ec7123b476d4fdc0a4607016d5
4d834d12cca525f3e6681aac81133268e8e4dc80
80019 F20110318_AACAYH rajagovindan_r_Page_106.jpg
9c9d5f4dfb435eb0c94b4e8b0fd24a6f
c4015af9bd328c8d8aab8888ab5d02e6216fb745
23468 F20110318_AACBCY rajagovindan_r_Page_010.QC.jpg
60087a8da8ea89c897e87523f0ca5e1a
d3dec8ca986ffbbec07d449a0f961e2285910dfc
8869 F20110318_AACAXS rajagovindan_r_Page_092thm.jpg
8f97d8b39f704f3de385ad1eb01ed5f9
d7b1b435fa62f8a15b27cce986dbd792a5de187c
7054 F20110318_AACCHE rajagovindan_r_Page_053thm.jpg
02a7e689fb1a1b5320bb012f186a3fb8
cfd7a44dc11ee9cc2ebe72ea928c0d502205e096
5372 F20110318_AACCGQ rajagovindan_r_Page_028thm.jpg
6d903773a339bb310cd7fb0b5fa236db
b047cd471e0f7aebbd97e91416414f41828f611e
F20110318_AACBEC rajagovindan_r_Page_079thm.jpg
fdec4ab47467b5a6fc67d304bc2f1cf5
8690398331ebd4544b1c8a35534bd6dd4d69d37b
347219 F20110318_AACBDO rajagovindan_r_Page_097.jp2
40c588e91c0e5b5182ff994da9592be7
99e7a4873ace27758deb0b1082f76d0ef670a717
2494 F20110318_AACAYI rajagovindan_r_Page_140.txt
22265c18690e1ca95443711a9ffc92bb
de6955adebc02643d94261bfca4e08639a2a3cce
F20110318_AACBCZ rajagovindan_r_Page_076.jp2
dc7993e768382c09ad5688a3747042b1
71fc0782f123e85bf0f9b3c39db34000ee7fe835
1682 F20110318_AACAXT rajagovindan_r_Page_044.txt
66ad9abd384e4d2712e02a9fd62b0ee3
fb5b33c352b0f46b1c1a27dccf03b5748f42e944
6790 F20110318_AACCHF rajagovindan_r_Page_054thm.jpg
6c2d653403f280ad872489ab9d18583e
448ea88fdfd835c404bbdfde92b5d05cfa9a9e4a
7000 F20110318_AACCGR rajagovindan_r_Page_029thm.jpg
f4da77c8aa2df07e40836d36de8ce987
0dbe194a30f92037d74bd071c142c2a569d2e571
731846 F20110318_AACBED rajagovindan_r_Page_128.jp2
991770bf5f38cdeb149620d07ca6712f
61721c483b65b411ca6f2e835cd02a70ef165822
2180 F20110318_AACBDP rajagovindan_r_Page_046.txt
8147e50c9a058dceccfde1e78d7e7490
8043b5bb44ceda02c64955d4e939c2f401761433
56861 F20110318_AACAXU rajagovindan_r_Page_068.pro
0b0fba92fb3c840f26ac4fc0d7636a50
6f3b030041dd0d697ef7c1099de0f104a13d7f24
4847 F20110318_AACCHG rajagovindan_r_Page_055thm.jpg
022360c73576c11b1ce7c062eea038c1
f9089c6d8c5030577f452a3491c921b391d4bffa
6566 F20110318_AACCGS rajagovindan_r_Page_030thm.jpg
ddf0d921e8042695b539ba5b8fa77150
d6576cd2ed33a6ba8a7627113e1cd76aee24e049
71597 F20110318_AACBEE rajagovindan_r_Page_027.jpg
2262f45d01acb7567130b80031fccab6
af765c38c31e14aba4437f3ab356b64146f22d04
1051928 F20110318_AACBDQ rajagovindan_r_Page_107.jp2
d8c547823e02755eb625b7a7c6197a4c
4e07b8238f7968b0c9c0c2ee49ce072656074b4d
25915 F20110318_AACAYJ rajagovindan_r_Page_029.QC.jpg
925209f1e793bbf931b1925e18471461
b8c622394a173812788ee10d2c170d5346fa0dd6
50418 F20110318_AACAXV rajagovindan_r_Page_047.pro
188e2dc1c884df5885c648112f48a3a9
c61ca6cfa5079061c50e91326557c51d43215dc1
5327 F20110318_AACCHH rajagovindan_r_Page_057thm.jpg
bae7b0a670a55a868230d7b37d7fd0f4
909685916f211197672ce9e0bc803859320837b7
5142 F20110318_AACCGT rajagovindan_r_Page_031thm.jpg
16b58ef4f6849302a3e538bf1c082bfd
a77855b9ca083924d2403b62e48e8a860b48f4b7
40126 F20110318_AACBEF rajagovindan_r_Page_089.QC.jpg
c888b77f93793f95bd0b6ef0ab12f6fd
0c3e7685289333cf7f3595ec3a095bf92fd320d5
27111 F20110318_AACBDR rajagovindan_r_Page_009.pro
33967437272c22489e9c3a6ab6148848
a84969af9da05f1f0737dd6ae01c2f10512a2d4b
2275 F20110318_AACAYK rajagovindan_r_Page_069.txt
1ac8476f1e511b3dc3531047d76e71bc
efa610cdeac18f8c96e5528c2ba1bd6f98385d2e
F20110318_AACAXW rajagovindan_r_Page_124.tif
b4423c52e9cae9120828a857a436379b
9a4ce4540a736b4cb93bcc37f7ffe86589419d76
9046 F20110318_AACCHI rajagovindan_r_Page_059thm.jpg
5ff43a312b1db25eb3519958d25d7543
e11cb6bf895f6639139cdcaa742d232dce6a0605
6705 F20110318_AACCGU rajagovindan_r_Page_032thm.jpg
b1b932f8a6ccad7a1140a78e5f366bd6
1bf02628e091c12bda6fb7f9dfeb5a0b5787c36e
25224 F20110318_AACBEG rajagovindan_r_Page_074.pro
9dfbdeae542299d207dc0f4e82a17ac9
0692078d4d5d2b44a964d652f833e6cd94f405ca
35691 F20110318_AACAZA rajagovindan_r_Page_019.QC.jpg
ef46ed3b63c5d88cdec44870119ce00f
77e730cbaa5ded97838c16ec016692434915e5f6
5676 F20110318_AACBDS rajagovindan_r_Page_056thm.jpg
7eae357c492688e4c58a473245d8aa6e
eab0010bca6182d4f41dd77b0718b67ae8b0f92e
2507 F20110318_AACAYL rajagovindan_r_Page_143.txt
01f9d793ad4c00d04db2f44a0df9ce0e
83913b1f914cc5eb2b263b366f84c838c24c99da
F20110318_AACAXX rajagovindan_r_Page_024.jp2
7a60b38654c71cec1f17bf1794d66072
1db2d8536a764ae09dfd73fe67c6d36e2a519f81
9108 F20110318_AACCHJ rajagovindan_r_Page_061thm.jpg
6b4bbf6d4ce918d327ba122a1cc4a672
d117df14d2b8126f2f2e085aa4736101c0b25622
2598 F20110318_AACBEH rajagovindan_r_Page_150.txt
8c944179b074e35c2bf0f7c4d07fa54c
9efa34b7df3ccdbb060915d73a3b9d9057749534
722442 F20110318_AACAZB rajagovindan_r_Page_030.jp2
0c9b43c3c933b8d02c5e1d5218ddfc9b
f79b880063cebd21929f083bc7381bd2c8aa27b7
1127 F20110318_AACAYM rajagovindan_r_Page_002.pro
0ec2e6f661bff115ed90e50f8d412050
de90ab34566364dda5dfc78ad84fe7bdc5b5bf5a
8588 F20110318_AACAXY rajagovindan_r_Page_060thm.jpg
25b3bc8ca75ffadff50a8d81e9065d59
17b545ba891e205bd82723ea65ef15fe62235172
8579 F20110318_AACCHK rajagovindan_r_Page_062thm.jpg
855803a375a6ebc85fc5a346cb2246ab
350484e6fcc30eeb8fb42c6629e1a6d168cf3e1d
8587 F20110318_AACCGV rajagovindan_r_Page_035thm.jpg
8e332aab4c90105a5f067bb505024b15
ed27e29f5727e38498ff8b96043b98242f9121c2
F20110318_AACBEI rajagovindan_r_Page_107.tif
84eb08e91a9ee7c91a22d9a8b52b0797
5c759cafbcc7ad4870b5a3b137becbe75b3a3c2c
9135 F20110318_AACAZC rajagovindan_r_Page_093thm.jpg
c6724cf932a7dff623e2c561a9c50673
75b7326dd2cc4f4a81f8a3f8789363ada86b1837
92865 F20110318_AACBDT rajagovindan_r_Page_082.jpg
70922aab9e4ab381b1172710d5ca2ca8
1355c10b859c0d66383585c8b66fe25f1100ea3c
1528 F20110318_AACAYN rajagovindan_r_Page_003.QC.jpg
2734edd21bbaec6a98e0d6107111ea5a
285889c8b05707b668525660c67344dea41c9155
19662 F20110318_AACAXZ rajagovindan_r_Page_026.QC.jpg
a805f09b616866b1e59e83807fa00d5a
b545375ea1049825d0c6c8bec325b38448cc8ed8
8693 F20110318_AACCHL rajagovindan_r_Page_063thm.jpg
c73c3a6d354d8024829c8cd43dc64c5d
25e8150b1e418b21fbe12018d092477d9d4d9220
2289 F20110318_AACCGW rajagovindan_r_Page_038thm.jpg
78734bb548faa7c78ea4e37c81aa08b4
506b3ea754b4e893db40ab7e841c5fdd1dadccaa
8738 F20110318_AACBEJ rajagovindan_r_Page_151thm.jpg
ccc43a80cb03ce6bedc7b881a87a53c6
c1a0ffe5d89412dc24e255085a6663d6af53066a
2280 F20110318_AACAZD rajagovindan_r_Page_075.txt
8dd09a6de68c1ce0c96065779061021c
31ffb500c22a7aae6e4ac54a1d8255d808b30d37
2168 F20110318_AACBDU rajagovindan_r_Page_108.txt
48e41bae867e8e88c63b317c0781acd2
b555f51441b1ff2dba78a59af3765590cbfa6033
F20110318_AACAYO rajagovindan_r_Page_071.tif
8c88385d2d53336c6417752d49d91a7a
224a513d604e31bc63b6f52c0bbcf0c2f756c289
9358 F20110318_AACCIA rajagovindan_r_Page_089thm.jpg
6ba28cbc5dd1679acf319cb576dcdabd
b0b698b20bd304854517e2d6da0185b18238263e
8506 F20110318_AACCHM rajagovindan_r_Page_064thm.jpg
d6267debd177269b4fd55abdc4c0f60f
6a5e10680b3a7d1692d2d8ccad8209e0796d6238
8285 F20110318_AACCGX rajagovindan_r_Page_042thm.jpg
344f813c47280b911b9684b91e2b587e
e1f1dd18741508f599ab87ce9e30b0f51034050c
50585 F20110318_AACBEK rajagovindan_r_Page_099.jpg
c9219dfcc38145cbbf99aeaaaf8e9932
7d5d3f546c79b4d710d65695371e9d9a06f64d62
F20110318_AACAZE rajagovindan_r_Page_037.jp2
443ede88a807d4902040a0b257391291
0d8fc94d4fffc1272cc0a205346638d5fdd64e5b
F20110318_AACBDV rajagovindan_r_Page_034.tif
33c744327c03267be64687807ecc6d2a
6543a865ad5a7fe3ec4542ff3eddee053562b67f
F20110318_AACAYP rajagovindan_r_Page_076.tif
fb963df8b6bd3ffce279158245df9c97
1d32f3bffca85fd71354dea93260834a24e9062d
9281 F20110318_AACCIB rajagovindan_r_Page_095thm.jpg
3d5ad3f0a17eed37c76eb8fffe6ae179
fbcdc3d1e863683d07d7a47c71b76b3ec1cb1469
8501 F20110318_AACCHN rajagovindan_r_Page_065thm.jpg
ddc58f2e12b87ca85c92374c416f4dee
7c0d14239cc89b8080fe53e3d3844bcf71b9d6d9
8227 F20110318_AACCGY rajagovindan_r_Page_043thm.jpg
02e9548c302ccc42f451c6a0f61ff525
45a071748bc62033958757d43f59396852aed3ce
59131 F20110318_AACBEL rajagovindan_r_Page_028.jpg
ecf86dbdee0de47996b5c772ef528692
2e9fb762489cd3f78e17d9d37d6d9b711ff33bec
53686 F20110318_AACAZF rajagovindan_r_Page_014.pro
f6ce733eeec97439023bdf603ad49e37
542065d4239d313046ca0810dd6f60ef034e9430
1051976 F20110318_AACBDW rajagovindan_r_Page_120.jp2
517e03aa5e6ec3a9830243a4b9d6110f
4d1b6d4b9552b0c60abeccfc1c9cf9bc3250bf99
2054 F20110318_AACAYQ rajagovindan_r_Page_062.txt
ff68a7c580fb0ddb8f243b3ab2e63ccd
cc06e008c5d88830d0e4627608c941063081782d
7477 F20110318_AACCIC rajagovindan_r_Page_098thm.jpg
c73b6e4d5eb24802f7be9113579768a8
f50e4ccaae3796572e76f089caa086236cf201d9
8071 F20110318_AACCHO rajagovindan_r_Page_067thm.jpg
be5e2689652a1b80f8438d9b1fa9144f
d795ec148c667990569f70a4a788a3c20ec93ce0
6678 F20110318_AACCGZ rajagovindan_r_Page_044thm.jpg
49bad8fed0a355685b02d5d929a6151c
45e629dda70abaa8167b002218183ba42bc97997
110999 F20110318_AACBEM rajagovindan_r_Page_152.jpg
c5427a81ed60ee160820f9a49ca4b96d
ae847206a1d9d41115e5365c8cfadc8d51fb1914
2541 F20110318_AACAZG rajagovindan_r_Page_135.txt
2e96326ad5e9aa4a2cbaecbecab32708
d55a0698e647e767600cf94895c4d44c9b5265b8
682411 F20110318_AACBDX rajagovindan_r_Page_124.jp2
4b934480bcad665277c29679881d730a
ce80041fd105ded2eee7790a38dedeffdd95a403
119278 F20110318_AACAYR rajagovindan_r_Page_118.jpg
c30d4033964b680ed558c13c2c2c4c84
0121f4b811f363165ad3267901b9f56e11be890c
F20110318_AACBFA rajagovindan_r_Page_138.tif
103a60a42795eefccc84587239a2cb66
70c770dd45a2de4cb4fbbb83763ed73c1eff7b4e
3982 F20110318_AACCID rajagovindan_r_Page_099thm.jpg
3c035d391aa096444c820dc89cf55782
ba25e29713e2722be1910e6cb2bd2b3551a53888
5343 F20110318_AACCHP rajagovindan_r_Page_070thm.jpg
978e5b043ecf8f4efa6f3cf658839bf6
83e876d1746a9594ab046affa8bfe78d365d8bb2
1997 F20110318_AACBEN rajagovindan_r_Page_042.txt
c0b59bc7b02ef1ff43ae7c90f81efcbf
559a4e22c99076563f12e1ab8687d8c6baa294d2
37567 F20110318_AACAZH rajagovindan_r_Page_107.QC.jpg
2962aa5f320ded55aae4ef2ec45feff8
926bdc3bcd78a893368464a4bf69911743748e86
F20110318_AACBDY rajagovindan_r_Page_151.tif
498bb118ac5b671b5e6d187a46c09791
5128128248373b3af848492abcf80e4402ca3f15
9095 F20110318_AACAYS rajagovindan_r_Page_066thm.jpg
778eddb636f8536a81314b87fc3ef249
0eeaabf5540a63fb65a16d3cf207d7407d651b13
8591 F20110318_AACBFB rajagovindan_r_Page_020thm.jpg
644897a85b1a7a7ec79eff8bc6bacae3
4fb2abbbea2192d955381a9a42fb4a882aa75730
4822 F20110318_AACCIE rajagovindan_r_Page_102thm.jpg
4dda8aa8709e64ec2812dbc8de17e701
3e43bb1e8b6159256eccb057919251001703e386
4939 F20110318_AACCHQ rajagovindan_r_Page_072thm.jpg
205600ac09528848dfb74f1303d71a19
b1ad6000469bcd2fdb2797d2a6b4ac830c5373c8
28603 F20110318_AACBEO rajagovindan_r_Page_100.QC.jpg
020957a77a544b686063d61cc078fa82
cd98bf3136c1ffe062e258e2cc534eada012ce97
7327 F20110318_AACAZI rajagovindan_r_Page_034thm.jpg
724860e4ae5aee20d15d94ebf7611af5
4c5780c5d155f4b58db0fac78706248a62a34d44
F20110318_AACBDZ rajagovindan_r_Page_111.jp2
4aad15ca404ec3d6d1dc406a40cc24fb
7ea231c0df6a76c04b9938822c2fbc10a1382590
8797 F20110318_AACAYT rajagovindan_r_Page_140thm.jpg
9c7c2b26175373772d35917f1addffe4
731bf6f778e677541f98b4c8f67b235fce66974a
9068 F20110318_AACBFC rajagovindan_r_Page_015thm.jpg
37d946ebadebae42902f7a893d70512e
54170b50b33312091fbf4b2baf9c2e7bb905d14a
5354 F20110318_AACCIF rajagovindan_r_Page_104thm.jpg
9c960bc2fc1bb371ce9ff65a022a8a42
b989e9c981ab55267f074ecc8746303cb06ede1f
7935 F20110318_AACCHR rajagovindan_r_Page_073thm.jpg
343a65af1e6ccb5424e41b6f4de5ed1b
dde575b081a7ba8dd595765adbe9cb83f6e414e6
881242 F20110318_AACBEP rajagovindan_r_Page_010.jp2
8d1545d500cc40f7a7b1561dc0a87b1f
7e3488f7303b854abac967b7069692937de71dbe
42090 F20110318_AACAZJ rajagovindan_r_Page_105.jpg
2bcd19306c4976993610e3b8f1d65d1d
07bfd4c1d21d834769606325662164a9933fbdd0
F20110318_AACAYU rajagovindan_r_Page_033.tif
e08b7a5143692d4109762919b4b23033
f279dc88c252ee11f7698b6115e11b2ea85df83b
38264 F20110318_AACBFD rajagovindan_r_Page_147.QC.jpg
1dd7e3611a104c79bbc7ac0b7b3fb265
6b51ff78f271567c80ff5c896751fa79a29a4f2a
3787 F20110318_AACCIG rajagovindan_r_Page_105thm.jpg
7573872ee37bf0b6569282ceb85fbc20
ce028470b9f2c715e2df161d85a7c1fc4441a68c
6182 F20110318_AACCHS rajagovindan_r_Page_074thm.jpg
4553b142a72fd46382bbef87d9e1d395
b6002d10ff5e10c3f6feab830e71d938d3f6f363
35671 F20110318_AACAYV rajagovindan_r_Page_029.pro
374b3f7081303baa397a2e1ae0dea364
fafafcd34ff1b14c6ce377334c7c13a294476490
36895 F20110318_AACBFE rajagovindan_r_Page_049.QC.jpg
f804cef276fcd44754546cc2e3904bf0
4964226e1b442527dc6bcd5400fc16833e536dc0
F20110318_AACBEQ rajagovindan_r_Page_046.jp2
3d83849df322dd9f5067c3c43f1232b9
5d6b788c4c29e771b1babc85ecff8ff8dc45be09
8972 F20110318_AACCIH rajagovindan_r_Page_108thm.jpg
5811c9beeb9c19fcac56fee16dca26ff
747900aa8ecf6c1c1c9ef53459f208b8aa2bc8df
8832 F20110318_AACCHT rajagovindan_r_Page_076thm.jpg
55e0b27b258c52a1b6f31334fc98f1c7
f9efcf4b1b6de073ef39c4850fc480005aa700a4
F20110318_AACAZK rajagovindan_r_Page_062.tif
99334bb9b583ff26ae1cc1c83951a3a1
3a98a48a8a193ed6c367ba361b7ec84d0a789fc7
19900 F20110318_AACAYW rajagovindan_r_Page_105.pro
b6ddd2eba23a9f430306f1ba1c8015db
6cbb6928bb044c83fd326a8ceb8e6735945d569d
9097 F20110318_AACBFF rajagovindan_r_Page_148thm.jpg
335071efe0072f2800583e58916bac89
c0bbfa37710c759f7ac2496a7619e69508d5fd2a
827933 F20110318_AACBER rajagovindan_r_Page_103.jp2
524405d12788161bda3735ee2c8b7efd
5e3314b354b28bddf28fa28799540ea9a90a8b5f
9242 F20110318_AACCII rajagovindan_r_Page_109thm.jpg
c5ece6d2bb82c9b48da2a7e6d34ee7d9
feafda3fdea510847a77487dd4c04bb5f533d9ad
F20110318_AACCHU rajagovindan_r_Page_078thm.jpg
41b242f48d412ed46eebb96c29ab5d01
9122dad8de870905367a701e8f53329459af3f05
20159 F20110318_AACAZL rajagovindan_r_Page_013.QC.jpg
c517837379685a357478703f3f41538b
6ace917e40371850f7ddd30bdcebaf16c6a73c68
F20110318_AACAYX rajagovindan_r_Page_108.QC.jpg
9eafdf39a6197a93da093cdfcd45e98d
b6886ce6fcd06d36510dbd62d096d8f784024cea
30909 F20110318_AACBFG rajagovindan_r_Page_124.pro
84c2824fa66f05ba15961d73a2129faf
cb5d3944d2b9d2719af184af1a56f1a687b85a8f
110268 F20110318_AACBES rajagovindan_r_Page_096.jpg
c645e6c49b5f995ec0a43c41abb31374
90c4502fb0cde0ebce53bf02518a19ab254316a2
8221 F20110318_AACCIJ rajagovindan_r_Page_110thm.jpg
7f159caf62acb4740329abb176bd7410
2294778790d5de5ad9a3e766e99fb198f188a9ec
8937 F20110318_AACCHV rajagovindan_r_Page_080thm.jpg
f32a45095d74751bc762ceb0a2307bdf
d93ae4dd6709a22b62090ab3a56edce7c220514b
2139 F20110318_AACAZM rajagovindan_r_Page_083.txt
29a7ddaeff8221bb6be3d9f737364fbc
63b147814d6c2a3ae9ee95773c9af3a73d9d73b4
36688 F20110318_AACAYY rajagovindan_r_Page_059.QC.jpg
96d398e09b23e31dbedd65135286739b
b1cbeb759766b23014ac6f9997846d237559b737
54059 F20110318_AACBFH rajagovindan_r_Page_077.pro
80c171f38a1d60e2a6aab30cf6242488
8638b518b1a52411e54ea73ec863049fb2cc064b
F20110318_AACBET rajagovindan_r_Page_074.tif
20bbc00655ed9cc15e7a157de8b72cd1
e9eb1432a17ee4ac528aa7bea9a906d156e65d4c
9047 F20110318_AACCIK rajagovindan_r_Page_111thm.jpg
72c7832c8f951ce44069ab08c115659a
9831884113c88b3bf0b1adacda962d688a8f9365
55852 F20110318_AACAZN rajagovindan_r_Page_021.pro
7a4637ba9c3d833d6bc855264969aaee
1810547faf9da81a97a5cab0ccd565adf878c805
8965 F20110318_AACAYZ rajagovindan_r_Page_039thm.jpg
8dcf9f807eb25059ad671d563e95a5bb
bf538f6ac8675390a2c0fd4c3853f299558fa4b8
8912 F20110318_AACBFI rajagovindan_r_Page_107thm.jpg
201a72002b2a49f649d135825b5ea2ea
1c957a81c0f6ac05e408dc9ca9a925ca8c94ab78
9054 F20110318_AACCIL rajagovindan_r_Page_112thm.jpg
ac5557e332eda928d95da3d2d9c8a192
59dffca035233f71c1dc59bc2caace383acfa206
9225 F20110318_AACCHW rajagovindan_r_Page_081thm.jpg
125c0af9de71dab4056d347d08091d35
ae82cb12d543aa698914a90d760c6ca292c6972e
55003 F20110318_AACAZO rajagovindan_r_Page_045.pro
439f0f10066a87524fe91669037dcf63
e1d76d723a3f008ece0ee857db72c069362d79ac
50769 F20110318_AACBFJ rajagovindan_r_Page_022.pro
cd3414a291ea994b40b31c4d259d2ae1
78176bab0776dc04c7a0c926a5b65e8e8e9efead
F20110318_AACBEU rajagovindan_r_Page_118.jp2
a1f75e03e2e2f9614f5af32cfe5bf3b2
476cb4f60e2d8a0c92bcf0517904456b079b0127
F20110318_AACCJA rajagovindan_r_Page_133thm.jpg
2df92db442b2359ecd663d7dcce1ad3c
b3eddc16a27c40431cfb65fcdeb73a853496ea45
8377 F20110318_AACCIM rajagovindan_r_Page_113thm.jpg
ecb26ce65d323c2cb51c73c481cd1676
1380be23d7f755ec49dadf474ab735ef59c3467c
7922 F20110318_AACCHX rajagovindan_r_Page_083thm.jpg
a97c0ccbc427ba4827624a94fcd3608d
e05e9b71e94f91f48bd7d002c24faefd0199e1e5
26982 F20110318_AACAZP rajagovindan_r_Page_001.jpg
d55db39685d61301d67531253690e466
eae0fcde167126a94828c099d4ceeb9758658e7f
F20110318_AACBFK rajagovindan_r_Page_019.tif
dbe112fde7b1a1aed942eea8f16448a1
94b79b74933936f45757c47ac4819b6e162f2535
101 F20110318_AACBEV rajagovindan_r_Page_002.txt
69885d001d48eeeeb6938184a0fb570a
9158ab103b806413b3ebd36f8a33f1c29bb41671
F20110318_AACCJB rajagovindan_r_Page_134thm.jpg
a0be8978064ecdf63c91c31c9b2eb8af
9ec2f3b3c2750c66b01cb721023844ddf3776dcb
F20110318_AACCIN rajagovindan_r_Page_116thm.jpg
e88772a7bc2a5fcc4dc15d5bf0e102c0
ac7bae35365bd2efcc426e4bc07b524c711d8cd1
9052 F20110318_AACCHY rajagovindan_r_Page_086thm.jpg
e063e6aaa107e03c84a73d117bd74d0c
0a10548502791cbdc233661fb92f3db42285d879
8102 F20110318_AACAZQ rajagovindan_r_Page_047thm.jpg
022a3559e9dcb444d516a2abeb0f895c
e9aab2108700af6b12e1c29226978e4517b0b34a
36841 F20110318_AACBFL rajagovindan_r_Page_122.QC.jpg
c48658ad628a0b2d1a04ea249507bb87
8c5a233b426c6943d472a5f6bb460c30f7499b77
9031 F20110318_AACBEW rajagovindan_r_Page_052thm.jpg
a10eab1be450de925d1240a708e78fdd
4e9b45d6bba651c34e37a8c028a8b4a7b7eed2b3
8775 F20110318_AACCJC rajagovindan_r_Page_135thm.jpg
9bc878e60700362c7994e84b2f499fd6
0243c9fdd572bc81bbed94bf76106e5ba1a12860
8881 F20110318_AACCIO rajagovindan_r_Page_117thm.jpg
fc456e89af07a46166e1e0bac1142fbf
d5db239311ffb518db708c67d87788fb5e44449c
9084 F20110318_AACCHZ rajagovindan_r_Page_087thm.jpg
238b406cf1ce129df08beed73632ee76
aeda931127bbeeead53c0a7b6e5603d4175a03cb
F20110318_AACAZR rajagovindan_r_Page_130.tif
4db611a2b23858eeb1b0d4aa92bd9332
7d6b3c96c739dd48ea3f13b53630d02fa57174df
15794 F20110318_AACBGA rajagovindan_r_Page_099.pro
8713a7b30bf21251f97cc8253e5c7d48
c40f308103a82f784f42790a55a24a2460360a15
9365 F20110318_AACBFM rajagovindan_r_Page_149thm.jpg
516bf9a122dae175485fc5154ab19548
d220efeb10d6fc78e92566a304f910e3f4283b11
116350 F20110318_AACBEX rajagovindan_r_Page_069.jpg
a00c2642f1ae1571fcf3dd2e96ad953a
128aa4dee78148ced5000511886176429f59ad7b
9207 F20110318_AACCJD rajagovindan_r_Page_138thm.jpg
d2262d53c4dbc40b27e7065379033f4e
8192d06e2c67ca9786fbb7dbaf0af5374a0a8041
9210 F20110318_AACCIP rajagovindan_r_Page_118thm.jpg
706ee629e0881ea2eb6e29c61d3a8c45
e59348603d1d6fb2f7c1b9cc6efaebb0e29ce074
32088 F20110318_AACAZS rajagovindan_r_Page_113.QC.jpg
3670964d16a5c03784b6981de212f179
29b65738af976edb59763a352a600436d3fbf638
27438 F20110318_AACBGB rajagovindan_r_Page_005.QC.jpg
ada22a4a886de3648db40161223bf635
2be0af61925d8659e44c1441dd267fee8cf80641
80639 F20110318_AACBFN rajagovindan_r_Page_008.jpg
343c09d229d37b3525d6c78239ac692e
b3e525da7b9707114cde1774bb8898ca49468ba9
52487 F20110318_AACBEY rajagovindan_r_Page_060.pro
e9dfe5cb3a9e1a7741c09a98de6e217f
2bd330911c6c5bd3058ce582cfdaaa10ee28c40d
9349 F20110318_AACCJE rajagovindan_r_Page_139thm.jpg
fc2f0227c73ce522a4aeb8122d417db1
c939da71e4fcd339e3e69c023b1ae6d11abd4816
9168 F20110318_AACCIQ rajagovindan_r_Page_119thm.jpg
4249637d905bafb59c2d348c6be4c81c
de647a2951bac1e0352550d25ac9e6168e815df2
4736 F20110318_AACAZT rajagovindan_r_Page_007.txt
4751ea510f810b2ff03c046775e468b8
58a7ee5be105086ddc1a1f7b4ab7a576a4a8883e
F20110318_AACBGC rajagovindan_r_Page_023.jp2
8234888ffa4be1f304cf671b91e15028
31f867544d620470fac116a5954a0edfe98ece59
F20110318_AACBFO rajagovindan_r_Page_086.tif
d2adcf7f8698ca92b1468abc9d7aeb3f
786af6776c8da123496858d47cc5993ad26f141b
2650 F20110318_AACBEZ rajagovindan_r_Page_010.txt
ad7390f16b906a9fcb972b6aad481ada
2a8f54cf3981a064aec4663c40e256bd51cb872c
9351 F20110318_AACCJF rajagovindan_r_Page_142thm.jpg
10e79408bda4f7fc237d7a1d2bf4abd4
9a946aa244b3cc987e7690b576245a8b69fef28e
8992 F20110318_AACCIR rajagovindan_r_Page_120thm.jpg
92510482b11335863711b4dba6b71e97
8e45dcdeba76d2a9f2e5ff768b0d1f064a818695
60117 F20110318_AACAZU rajagovindan_r_Page_101.jpg
e974a5b25e00b1028fed9a577e6753f5
5bf390cd9f728f32021a929b69b500477ee9bab4
F20110318_AACBGD rajagovindan_r_Page_106.tif
1b791d43a8dc929fa6845e3d9ca97606
a0d95019a73ab6aa3414c69b046770897f1a2afa
F20110318_AACBFP rajagovindan_r_Page_110.jp2
71dbe0a9881788babcecf75733ceb556
9f2ba76d464062bc501bce6ec35b0153b6685ad0
8838 F20110318_AACCJG rajagovindan_r_Page_143thm.jpg
ead16c403686bc5fbb630cdbcd9d003b
89af1b689950b013cde7ebc130ab05dfd9152bab
8934 F20110318_AACCIS rajagovindan_r_Page_122thm.jpg
a3d5d2d192525c81164648526518be81
b89e3f0da2214f566e494b455af51072cf9149b5
25955 F20110318_AACAZV rajagovindan_r_Page_073.QC.jpg
afa8434d3bc8bd2aeaaa4c37ed7d0523
c9d957d27a4e161355882e7b62cebdc0086d6005
810940 F20110318_AACBGE rajagovindan_r_Page_044.jp2
2c1e5935e62ca56eb09bffbebb4959c9
be4983764a1650a5e38d384e13eac15368169622
2131 F20110318_AACBFQ rajagovindan_r_Page_111.txt
2d4d225795ca3d9a2c881a23ee472c51
7be5c87677084d6182ad16f49121f8e26a9aebbb
7430 F20110318_AACCJH rajagovindan_r_Page_152thm.jpg
3078eb6bbfab3d446a4d1f2532c4612a
e24cce3a16232f7d784cf3b44f0f930f4c3e2f61
F20110318_AACCIT rajagovindan_r_Page_123thm.jpg
5856e7ed075fe70c9f70e68fc41ce17d
f35b60f3356685d04870e88c9c2fefa54667b00d
27032 F20110318_AACAZW rajagovindan_r_Page_028.pro
1f1f298e28f9bbe816496fedfac90641
4c9f8064b6e229458acb87e0724ab8c6aa657ead
F20110318_AACBGF rajagovindan_r_Page_065.tif
b3be055fe6f872e3cbaeeba0858ea2af
c77bc794d5953ee762ca13d5c4033da2ace226e2
F20110318_AACBFR rajagovindan_r_Page_025.tif
939f42391fd09b6f645c7f30e356e6e8
41f44b7a5911506bf1324c74cb8efdeae80deac9
5150478 F20110318_AACCJI rajagovindan_r.pdf
0207218c78f6c88d3b81274ef38a583c
8ee5ad1fe3bfe73a7906dfa93af20a36cbf59394
4230 F20110318_AACCIU rajagovindan_r_Page_125thm.jpg
00fb4b24e7ad8ebb068a3e5c7415a208
a5078c10bdb64e0d828834c5c75ac062015b6fa4
1735 F20110318_AACAZX rajagovindan_r_Page_034.txt
2f2b6b23bbdd147c12d99e2a91f7b6ce
e90ddac3f36ff08ef88ea9380f5892d48be5e9f8
F20110318_AACBGG rajagovindan_r_Page_037.tif
6e5fe336ffe9f5f924a588ee3f95b48a
ddd207db7370d61c86cc6ffb535d3ba4465bb982
F20110318_AACBFS rajagovindan_r_Page_063.tif
fb9ed9e228f30174443dd49330682baf
839ff65c5d66eee6cb52033e49e17d3cd69c07c4
183984 F20110318_AACCJJ UFE0022215_00001.mets FULL
3604c1dcb752a17266da1ca01a8ae008
78e30e97c7bfe08d4a71181ee3ade8b0028ddced
6716 F20110318_AACCIV rajagovindan_r_Page_126thm.jpg
f26303a0d3b256b6bde36e1d3c63576d
fea03a2cdea7eb779a572cd90ffad2d362150703
1214 F20110318_AACAZY rajagovindan_r_Page_013.txt
83361038422588e2e3f13fd1daab19a0
d6a3447bf983c2f00c57e2f592f70f4e3e3048c3
F20110318_AACBGH rajagovindan_r_Page_102.tif
4a1021ef2d74a6bc38ca111150f962b4
c81a96e2286d86e27dd3655688bcd6833608068d
9491 F20110318_AACBFT rajagovindan_r_Page_038.pro
109a5719bf23c09b7b4d6d7475ee00cb
b2d52293ce10a464364172820379618621ca7805
6342 F20110318_AACCIW rajagovindan_r_Page_128thm.jpg
42d399e7a90a6a58318ab3d81365f918
65fa0bdad836af3517a61cee02aa9e014197f4cd
1051941 F20110318_AACAZZ rajagovindan_r_Page_062.jp2
3c328df7e70b599a2611d40103ae405a
dd52b56b0efd7d9481d8b7d20124199337a0d787
F20110318_AACBGI rajagovindan_r_Page_104.tif
76d80c7853faf5ff9537b2f88eeffc1c
68cd21afefc949ad5886719c4ee19695ebf172fe
35514 F20110318_AACBFU rajagovindan_r_Page_050.QC.jpg
18fc00489a3e786a67700f754ec4b37b
8a906e5f06f26cbd78a0ccbeb57b9b03cff39a8d
56195 F20110318_AACBGJ rajagovindan_r_Page_075.pro
c30aa80dff42d3050ed24e8e1afc323c
2264258d6f6f5732709b0d4cc1e081141b9d99a0
8542 F20110318_AACCIX rajagovindan_r_Page_129thm.jpg
3514889c59e843734d428c267ee39e36
a45984574ba165062f4d7d69c5d06109bef5923d
F20110318_AACBGK rajagovindan_r_Page_004.jp2
666d5d3f5f64071e6f9506fbf281eb5f
77fb592e5b8b401d377e5ac4df873d5b56ef0e83
2205 F20110318_AACBFV rajagovindan_r_Page_086.txt
8a8f9f7ef2862371295ef7ad4fe284ef
3a0d453988e28ad635b8077b69c37d79acf1ff60
8944 F20110318_AACCIY rajagovindan_r_Page_130thm.jpg
40939a255f8c38f352c1b9a1349a3c77
0e9dab660b21da6fa123f074e134229cb76525d9
55929 F20110318_AACBGL rajagovindan_r_Page_114.pro
1589e7f5ddbcba9262fe20c09eccebdc
14c6c063b65ba063f40924542f3589949ab434ec
55400 F20110318_AACBFW rajagovindan_r_Page_108.pro
7d88eac66a85ede52a337052912da2ba
ae6d1efa13e522ccce88907ae816d0a2d1e17a3f
8963 F20110318_AACCIZ rajagovindan_r_Page_132thm.jpg
daf92a4387554700782f1a9f2e29fb65
16918007a38175df47bc3bcb1e8a97de8cc307f7
114787 F20110318_AACBGM rajagovindan_r_Page_131.jpg
7a116fd63b658408e03b5957f69310fe
1b22fcba33ce8b0f6bb3ac8c9bdb82c0c91c167d
22143 F20110318_AACBFX rajagovindan_r_Page_038.jpg
171bccde3a948cf56ce374eedc93917f
c1f4f7d0e8f4292c819e8ba3de4d5b2ec7ad5116
6937 F20110318_AACBHA rajagovindan_r_Page_027thm.jpg
6b42ea58f92f86d4665667f38226769a
de440a3d82feccf51017ccff9431e8a3e8c55d06
8532 F20110318_AACBGN rajagovindan_r_Page_001.QC.jpg
83e907886840947103b2523de9a7e354
36dfd28ddecb4ffe5d29f038e3b7a099f74dde56
F20110318_AACBFY rajagovindan_r_Page_132.txt
b785b0950b286f1d0d51131a6c719f51
7c5c8b42391f214d94745adc72abcbd2d9034f68
47554 F20110318_AACBHB rajagovindan_r_Page_055.jpg
bec6df39759c9a328ad98302a9f74679
0ce91a51fe79cc3159fbfa43012fe6e1266957c8
F20110318_AACBGO rajagovindan_r_Page_085.tif
ddc4ecc39baac5f24eb8161f20239a2e
5e41edc3c606fdd3cfc262539271c200d96ada32
F20110318_AACBFZ rajagovindan_r_Page_039.jp2
085349bfdc7fe8bf28bbbcf6503fa962
2c0e947bf03b10843deab4447e89a6f734483fe4
32897 F20110318_AACBHC rajagovindan_r_Page_067.QC.jpg
c9bbce1c66ff7a5f39e7781babb398c5
e76a36e8c5477744399b407ff12e32d2e23c8147
38147 F20110318_AACBGP rajagovindan_r_Page_123.QC.jpg
82cc76446eb7f4edee2bb39c79b513ba
542f669aa833349bd792bd4e61ae926924689dd6
248815 F20110318_AACBHD rajagovindan_r_Page_001.jp2
a4c087fa272907c77321b51a03d98fdd
b571c40742c464eb953b4df72e02761eeab2756e
F20110318_AACBGQ rajagovindan_r_Page_084.jp2
59cf42a57ac639e38cdfb2014dfa0864
d4280edb54108559afa11df0efdb41b0abf47be4
7111 F20110318_AACBHE rajagovindan_r_Page_106thm.jpg
ca625ccf8a990c0b662f7aa2f7840062
7a798f01a3d8d7e84ec29edaef05e82296a713e2
55223 F20110318_AACBGR rajagovindan_r_Page_084.pro
012cb518d2d99599918257c6b23d0e9b
1fc23a106c8c1943509022012c24b6e7ea8fd48e
37012 F20110318_AACBHF rajagovindan_r_Page_035.QC.jpg
661a7c743374e424ca68dea05eb7f05c
70006f80a4cfc28f37cadb654b839adeb17125ef
36235 F20110318_AACBGS rajagovindan_r_Page_150.QC.jpg
c967a16a6269db4d72aaeb94d3f8e130
4e778ae8a377b6351c40210980eaaddc81debbde
8086 F20110318_AACBHG rajagovindan_r_Page_024thm.jpg
529e98b8812b3583399b8cf376ceba2d
0df75bd29dec62daf4f202c9175fa74d7fe3bf68
F20110318_AACBGT rajagovindan_r_Page_075.tif
b34e4a9cf2ddc3fbf57db235e694e8bf
a0b99901d96e485c4eeabd8fe1df4638919fe4e9
1939 F20110318_AACBHH rajagovindan_r_Page_067.txt
06e421677f5b4aba29ac0d04ea30684e
f89baeabcadcf56e5e96da636471a7e2c5188157
56155 F20110318_AACBGU rajagovindan_r_Page_153.jpg
7cd984e8d3533723e4bf39e578fec223
cf1b297773ff4204345cd828cc5c428d689be4da
2226 F20110318_AACBHI rajagovindan_r_Page_128.txt
5381f279bd677e0aee499a7bab848d6c
dee041bea70d0075b05ba4a0952f6de388b8014b
2542 F20110318_AACBGV rajagovindan_r_Page_137.txt
0533da8ced30697af61e5afb088c8a18
bd5a962ce23b4b2c95f74c57f250a7008b8a52a4
61664 F20110318_AACBHJ rajagovindan_r_Page_102.jpg
7ff14f939c4c231a2d223d68f13acf44
8df62a23fcbab4f49dbaea3be8a8b0a436d04a0b
46197 F20110318_AACBHK rajagovindan_r_Page_100.pro
df966d817f81d38d750e8d4c57f40e28
f116e523ed2680e772edf5f175be299457b2e353
F20110318_AACBGW rajagovindan_r_Page_127.tif
55cb928446ac427aafe6b4cdeaea2dbc
8c1bd797f8079c76950cc2d06f290d3b9048d1cf
2172 F20110318_AACBHL rajagovindan_r_Page_035.txt
61bef105dcc2eeb7f17540af7e129360
1a2d59a91b10ac9e87f6328947e151bf518ed7eb
4508 F20110318_AACBGX rajagovindan_r_Page_127thm.jpg
7f3076bda138f8131ad40b822c2e0680
1658a4e2c508ca4fd857b5765cda7fbf81c1a597
126212 F20110318_AACBIA rajagovindan_r_Page_007.jpg
78c04eb2df41ca1c05d13d7831cf89b6
1a778c5faf6d5087707c8f02d18dccca2b88a682
60887 F20110318_AACBHM rajagovindan_r_Page_025.jpg
286722bed68eacea3c1b2c4d906dc5e0
fa2de53f434a5bf5c1ef724315a16f98c24ffa94
16499 F20110318_AACBGY rajagovindan_r_Page_104.QC.jpg
2af600a766f54bdec87732fcb20885da
02a906244e2a516fde3165a8f5fa8fdebe49dadb
F20110318_AACBIB rajagovindan_r_Page_128.tif
8eb75f27d5d0dc33b7504156ef8d21ab
e0b80fda2b57579de0373f8552dbc56d4d2d75cf
35358 F20110318_AACBHN rajagovindan_r_Page_130.QC.jpg
bf9516fdfaa0421eed45568d9ba0609d
11b7c9489b7d024fc3f0be16030050144a75c035
61377 F20110318_AACBGZ rajagovindan_r_Page_136.pro
a70690379f8ed5e4ed26b3bbff74ff8c
4ea56e8af29ed989a9e96c695763563e72ba55b4
8891 F20110318_AACBIC rajagovindan_r_Page_018thm.jpg
151a4e590a250fdb912bd2e6d565baaf
dbb30dc84b48f9027ed94e252703c020e5cd1cfa
104767 F20110318_AACBHO rajagovindan_r_Page_043.jpg
9850ffa28289202359dc7e5df9f71ae6
29e8d9b0dbe8eae3e578c3a9b14aa8511d7fc659
111662 F20110318_AACBID rajagovindan_r_Page_084.jpg
b1abcd74d649b2b3d7834e3b32a942cc
dcf52eb1ad77f405204df62dd9a2eb2a75e17c70
F20110318_AACBHP rajagovindan_r_Page_050.jp2
e2a8f210b5d3143fb005deca906d26a4
d0b2da46f1b810ebeec83e208cd29d9dfaed4fba
34012 F20110318_AACBIE rajagovindan_r_Page_103.pro
c85531100b68b3042cd01ea0fb8815ee
1b9d095107e2b7e81028bd2cf5c9388d0c8685e2
5414 F20110318_AACBHQ rajagovindan_r_Page_003.jpg
74c909b6a961a756bad4d6edda543253
a08a19a8d356ec31b779a847011f95cdc9faa9c8
49352 F20110318_AACBIF rajagovindan_r_Page_082.pro
69fbcbc0d712cf4dd6874a4905b7c338
3669dcb34beb034459bed02f8e601a4dbe7e3d1a
109703 F20110318_AACBHR rajagovindan_r_Page_039.jpg
74729d6fa0f401104daff6d994d5243a
4588eb48e7eb52026cfa9de842c4d12b37d1ccc5
F20110318_AACBIG rajagovindan_r_Page_099.tif
3f6eae8daf20a1f03ee117a542eea92a
439e88b6b7cfce970c7cdc990d8175623f6c3a1c
F20110318_AACBHS rajagovindan_r_Page_085.jp2
1c1fecf9f2f9e437b4658819cd05dcf2
f63b36df99be3223a72dc53aca92918a4ba2d65a
52876 F20110318_AACBIH rajagovindan_r_Page_065.pro
a790a130150bee97fd32f156b9bd718c
d6d25d2579624f25e5595e353515d9d9b3bca788
2835 F20110318_AACBHT rajagovindan_r_Page_054.txt
2b0855a93fc76e5aabaa34a8e903c9ff
aa50830f136d3e367e0ee12cf814913c50af42a3
F20110318_AACBII rajagovindan_r_Page_057.tif
891fb4b276c9a2e19a3fdb24e2a8d7e2
50d57791558b4b7ef3a59e0067fc96852553ecc7
649798 F20110318_AACBHU rajagovindan_r_Page_026.jp2
a0532c7f19816fb70cae29fe5119affd
136d629d1e0553df34bd74d05bab6a12cce1d9b9
8816 F20110318_AACBIJ rajagovindan_r_Page_045thm.jpg
bef7af6b4f5a37d668d546e2fc70e34b
9b435a634bd4380cd1857b4bcc6e59d1e49dd674
F20110318_AACBHV rajagovindan_r_Page_083.tif
98330226fddd90026d4f98809abac49b
9f09bb12f6c72edbc4f92ce687609a0bad09c4ba
775797 F20110318_AACBIK rajagovindan_r_Page_027.jp2
55d0dd47794d1a958d51419b14a67808
80c418fb4d51d8bfce30de80233b1ae3bdc936f4
F20110318_AACBHW rajagovindan_r_Page_027.tif
53e884c27ef51a304d35cfe81077c963
b5de529dd2352009414f1840d9fea6d7667acc54
56848 F20110318_AACBIL rajagovindan_r_Page_066.pro
cfe7ae861d7a18b4d875912520c714a7
7044f29047bd766ecea58f4ce372252663be9dff
F20110318_AACBJA rajagovindan_r_Page_040.txt
df80fab2bf0e3dba8a3dc165e66eec3e
485dac3a3a3cd5f409ab60ef2b04237276d6a27c
64931 F20110318_AACBIM rajagovindan_r_Page_010.pro
5e40b836c61c8c875bcd046aed9b5faf
e2334171bcac089e3026f1b57b0c5d4e5d93fd9a
1310 F20110318_AACBHX rajagovindan_r_Page_028.txt
6823346003ea29adb96875093ac65316
3dfe5b56dc430e30e1054398c664347810cf566e
31739 F20110318_AACBJB rajagovindan_r_Page_083.QC.jpg
d4bcce02a03d1434ccfbd72986039966
8acad72507d3479a9304f7296e000156c4aeaf1a
112328 F20110318_AACBIN rajagovindan_r_Page_045.jpg
b5f807a03f6e28d21ded6375fd55e0ab
66f5e0b8019dccb5f9dff021ab1d1ea1d54cd5a5
F20110318_AACBHY rajagovindan_r_Page_123.tif
16aaedcb608ffb346406c72bfa985116
5591e8c1e4a3359ff418d17834765be7ea10e6c3
14885 F20110318_AACBJC rajagovindan_r_Page_099.QC.jpg
48fdb5d10ff4c2c9a5769bf4b50f6709
b8074831bbd4e856d00f50e04ccf22b6795ddc6c
885914 F20110318_AACBIO rajagovindan_r_Page_054.jp2
43a7095bc2105e46d23e165f5c9bddcb
32557378c2863e6c610e704f3ec1a4be5b299367
2184 F20110318_AACBHZ rajagovindan_r_Page_014.txt
0579c839de343530b72ac5f8c8562dfc
b06eec5d4a3a0789ab0a4ed652a649370e8270d1
F20110318_AACBJD rajagovindan_r_Page_101.tif
2ca367926c8ede7433f8c168d11a2f93
04d9b6627cc6cd71aed0de9cf0aa52979b9368e3
F20110318_AACBIP rajagovindan_r_Page_001thm.jpg
c78b5b00d843bcd89ae3469b7c99f274
c9419acf0c75d698e5fc4d348ca324da7602a9cd
9238 F20110318_AACBJE rajagovindan_r_Page_147thm.jpg
4bb0681f2de6f0b893be35eb96f5689a
60da00060454f5421fa22f91b21bd54bd829c34e
1372 F20110318_AACBIQ rajagovindan_r_Page_025.txt
c409beda55ed31b2e837aeffb0a87bd6
21658401b4972de2afa5d533daaaa08f27e53735
9087 F20110318_AACBJF rajagovindan_r_Page_088thm.jpg
9339099e8c0f3a396891ea6c37262bac
ad1f95999f7cc4ac756ef101eb29f3bfe046a2b9
1051946 F20110318_AACBIR rajagovindan_r_Page_068.jp2
22e9ac5b06e513d6013e30c523809104
7af1ecda66b13e1878d50097698005b1cf782cea
19561 F20110318_AACBJG rajagovindan_r_Page_070.QC.jpg
a08cf4881ec2609e2098d11ba74f472d
9d797a6d7451aa21f8627f7567ef58ff0ca0b21d
54783 F20110318_AACBIS rajagovindan_r_Page_116.pro
8374985c7ca297fc75b1501755c91995
610ded7baebf1045ad4621a015f3277f1eae856d
1584 F20110318_AACBJH rajagovindan_r_Page_033.txt
723fa8fc2eabc572a2edd9fad6103d0f
3d34fcf255c2329f5d91590aca2c7d3de3cbc35e
F20110318_AACBIT rajagovindan_r_Page_015.tif
f8d195993b956b7bc8bab0a839c54ce5
c1f92e731b44fa14129af016f577af1a04f949cc
2193 F20110318_AACBJI rajagovindan_r_Page_129.txt
a7826728630c1b517f4da406c54279d7
7898843a77d65fd92a78338cd8e966e61af5b0ec
F20110318_AACBIU rajagovindan_r_Page_109.tif
8b997263333e4435fd239bd1ebbdc802
7a8c1a4e42630c37763077248b9ae0dcdf16b493
F20110318_AACBJJ rajagovindan_r_Page_077.jp2
ac8763427eaf484145d9face0b2fccd8
5ed4d377c935070259ccfcc088abe992cfce5cbf
17833 F20110318_AACBIV rajagovindan_r_Page_101.QC.jpg
9861f04347a4b3d2f8564ec32ca0a48b
b858f4425ef99156fd7688a21aac4dd823adfe91
2187 F20110318_AACBJK rajagovindan_r_Page_036.txt
64880dea5edbb0e904fb6a55ee6f01b7
c3c5ff518c9b697fd58b883ff1697155369e38b5
22043 F20110318_AACBIW rajagovindan_r_Page_072.pro
ef0e76948d9eb92a1961071987f5f860
2950efa28c46d52771a9e31c6580db9a8e2130d0
1537 F20110318_AACBJL rajagovindan_r_Page_101.txt
1047ac8c7503638a988f3ed1ac6be186
cf18ebc5f382c62de8fd273bde28e9f16eecb508
58559 F20110318_AACBIX rajagovindan_r_Page_087.pro
1d1a2c543be553e854e1c9f5e41dc2fb
df517399883eaf394192c8bf0093b21956881995
2714 F20110318_AACBJM rajagovindan_r_Page_141.txt
b0efd84b695e201bb75e2b45e60dc9ad
46cc60c7400bae45122a86e96ce50168fb02ee36
53932 F20110318_AACBKA rajagovindan_r_Page_130.pro
0341a9956e26ca1a0f469a296d274019
f4d8721adf671c372b9476cea73350b9ae817a74
36468 F20110318_AACBIY rajagovindan_r_Page_092.QC.jpg
801a15a8c6e54136ca4487a85aa33af0
d48237d0e3964666dcfd819a3436e9ecf451f550
23930 F20110318_AACBKB rajagovindan_r_Page_031.pro
63f0edb6a1e011e9839f0172295a7aa3
da20b084e87940e982a971563b8e4a471637db1b
F20110318_AACBJN rajagovindan_r_Page_046thm.jpg
00266e4b7b9d447b18dd33d31f7e94ee
b4557281d2487f5772430c959ad229d916e17a6d
56275 F20110318_AACBIZ rajagovindan_r_Page_086.pro
9ee890fe33cb0da75a271fa03f3cfc3f
712f55d4699d51be8820b7084f864d723814b36e
8974 F20110318_AACBKC rajagovindan_r_Page_069thm.jpg
bc7ad4891b278a22e6a4349245b9b922
c54bd77131ae4764ccad898406bf71c56a0adbcd
1902 F20110318_AACBJO rajagovindan_r_Page_041.txt
dc48fcc901fad31097da272463620a8b
3db7158180b6918cfd3a4db479a153dfad4f1ea8
51757 F20110318_AACBKD rajagovindan_r_Page_056.jpg
fc571768aca1dc449f17a4f7bf32cf4b
03f9eb475e10a8d4a7cc7fd73933d2e058a4b3d0
17506 F20110318_AACBJP rajagovindan_r_Page_031.QC.jpg
fd5191bdd5bf4e67d1a7dc3bf8df9103
5a3ed688a4757d12857de68e38d37a38bded89a1
475 F20110318_AACBKE rajagovindan_r_Page_038.txt
eedf565ba00b53b5a7e72176380ebea3
6a7a03bf9808ed29f953dc9bf20d72f93f9f4b00
4342 F20110318_AACBJQ rajagovindan_r_Page_153thm.jpg
4e609af13ad2e2b6a01bdb3eadfdccde
ee9daa544df0d8b0701713bed87dc278ee485816
2228 F20110318_AACBKF rajagovindan_r_Page_112.txt
ae0ce5fbd17563636c0a96bdbc474568
1faf3ebf53d019f6ad0baeca0a8fed49d36a3179
50303 F20110318_AACBJR rajagovindan_r_Page_083.pro
96e9453649e4fac61b3d9e9cae996434
1bee66def310e5873fea465b0d57bd2a2461b3d1
62886 F20110318_AACBKG rajagovindan_r_Page_134.pro
00810f2a3a354d110a50f427f33309e6
ec9fd5965041507aa20d07e5b4a5b3d1d65c8a64
112242 F20110318_AACBJS rajagovindan_r_Page_094.jpg
ff70860f47dc8494e803098c7b3a579f
39006c55d5add42d84c356c3e691a899ac87994c
35694 F20110318_AACBKH rajagovindan_r_Page_136.QC.jpg
dab4b346f7e00e2a5065dd83816f41f3
7e76cfba7111d6ffa7eebb5028c19d83c1df4bb2
1366 F20110318_AACBJT rajagovindan_r_Page_073.txt
fc7dd86093839c9b45fd43b48957eca8
bfca518f321a1da23a97cce3df0367c4e596cddd
13053 F20110318_AACBKI rajagovindan_r_Page_105.QC.jpg
6498e83b21a421f25345cf330193d1db
f954dbbdbbe261c7ac5d0685d63a351b51c98f9e
8818 F20110318_AACBJU rajagovindan_r_Page_114thm.jpg
2078224832cc3d0c5a7f59dc64e1328e
6a4318cd457af22b75d0e80e9154d5ac21f0ed99
F20110318_AACBKJ rajagovindan_r_Page_080.tif
c4cfc57028ae755a30e12f151e21a123
9574349075fe44beba1625ade808fede566a7273
48223 F20110318_AACBJV rajagovindan_r_Page_012.pro
cb54321c421b5996e6272b14fad37c6a
f7ffbd11d1fdb2941e191510c07ab8a58d9f8116
2660 F20110318_AACBKK rajagovindan_r_Page_144.txt
28593b6dfafe9bf6e1e99e39bc1e0865
083c3eab2c08c82858f5a3eaaec6bc8a33bfecf4
58312 F20110318_AACBJW rajagovindan_r_Page_069.pro
e2b5e748e9eb7b9705a86b3c0bd0bd17
113bc8100817fde9c819a6c2c51965b64f90bcf7
34880 F20110318_AACBKL rajagovindan_r_Page_016.QC.jpg
6335edc39855dd1e2042b4a34e1bc08e
07a0802a84db725f56ab768f0cd6453ab1455fae
F20110318_AACBJX rajagovindan_r_Page_044.tif
0ae1009b46f041f291ebcac30e7a1a73
0c67de01a1030a91182f5a6bb647c87d33081741
1037 F20110318_AACBLA rajagovindan_r_Page_153.txt
190c20235eb15aa50c7177d9aadc4313
98a8aba8d006e37f83dd82ba758d9a0f6f993012
111041 F20110318_AACBKM rajagovindan_r_Page_111.jpg
a304cc945a6f617120026a631a6b9a60
1f1832dbb5073ba9d0c9bb910e9445e1332c62a6
8750 F20110318_AACBJY rajagovindan_r_Page_077thm.jpg
0be5ca7dc88619a34e5bd5cc87150ce4
c9942dce1974412f3dab857c631958aeef6ea3f3
F20110318_AACBLB rajagovindan_r_Page_017.QC.jpg
f4be362f7b0a5a45bea1b831db37ea04
d5b0523a07c5f4c81593194d9d0d1952c19a94fa
9282 F20110318_AACBKN rajagovindan_r_Page_068thm.jpg
c5f9369a5bf951d36f89db92133d0f68
41278f8322daf075471437c5f5de937a77d265ea
1463 F20110318_AACBLC rajagovindan_r_Page_003.pro
5b9426ebb5071cc1ed4df2cb5643b625
395343910e2f2ea1fde0b592bfcc69bb9f3e39c2
110286 F20110318_AACBKO rajagovindan_r_Page_092.jpg
63a0e2214b033c17ae7b9bfa51d2aa97
27d4648c8f08df4be7c76be50a58514de8dfeb63
8718 F20110318_AACBJZ rajagovindan_r_Page_016thm.jpg
2b99108e5b91bdcc266f8f7bb3f9b22f
47c45ac2da9189efb53365631560c37e69dd4e85
2070 F20110318_AACBLD rajagovindan_r_Page_022.txt
5b9d42c09f2c4f2725bc769664f8a5d7
51c2960fe28206ef527b7b9e6e7813bf349cf18b
8835 F20110318_AACBKP rajagovindan_r_Page_094thm.jpg
02f46ec7bf2c9a79da7e3132a0800c39
641a5f5b8dcad279422f50d62e35865a021145d9
143487 F20110318_AACBLE rajagovindan_r_Page_141.jpg
bb1cc53562f8975d5c6cb4565e63e390
f2e5285308f5f82ef2ba16d165e04b066a57092f
F20110318_AACBKQ rajagovindan_r_Page_004.tif
e1d8a39d5364cbd5700adaf6c99a87f0
1eedb545e067425f62d5fc11ee007e0b4194a2a3
116346 F20110318_AACBLF rajagovindan_r_Page_068.jpg
f47b152034abdaa3e6a80f9f3c9e43ac
fec7c516af4eb80c20693cb89388a91ff2332010
8929 F20110318_AACBKR rajagovindan_r_Page_036thm.jpg
d3552537e881a1e542ba1e8ede065170
83cf0346556f25b62a2d986fffeab0016e401987
1512 F20110318_AACBLG rajagovindan_r_Page_030.txt
25e22aee9d637c355851e3e05b0d1972
9534c160533523f85865d1b6397b0b5af3c9c3a7
F20110318_AACBKS rajagovindan_r_Page_009.tif
45f71e6e8b234615aa4adf18993b1702
848a1863084db7458550f28cb034d54989ca407f
2523 F20110318_AACBLH rajagovindan_r_Page_145.txt
b3325502f303569187c21b8561335c31
666b4ccdaf638179ba80694902268e9a629a33e9
2258 F20110318_AACBKT rajagovindan_r_Page_091.txt
72ab110bbfd60284becc6d51110b035a
9cf86c1a4537e7fa0540310ef7c540cfacaad4ea
2242 F20110318_AACBLI rajagovindan_r_Page_114.txt
1cefe9350d31e322973888ed1cc5e74e
534f7510d37ab886c224dc8dc777b2f154e0fe84
63973 F20110318_AACBKU rajagovindan_r_Page_150.pro
3f28b0dde3995db6e1905157e44e27c1
ad1a578909eeb305d485649a483c4247d0efab53
F20110318_AACBLJ rajagovindan_r_Page_091.tif
bae08bc1980a3932213ab09aadf78df0
c21e83aa328613e094adf4f3211e3836a27829cb
F20110318_AACBKV rajagovindan_r_Page_134.tif
5b0ed4a7d37b0acaf97c46184f04d3ad
00db30ccccbf1b46dfbe5d87814e23edfe96dbb4
38875 F20110318_AACBLK rajagovindan_r_Page_066.QC.jpg
d77ce59f6a77c363c3bd1e044f78c532
773776580643de8ceae5bd2993e19a66c282409e
787346 F20110318_AACBKW rajagovindan_r_Page_126.jp2
2cfa4431fbe0950ff0289ae074c23335
1f0297cf220903bb00c3976fbc92a42abe76a064
3115 F20110318_AACBLL rajagovindan_r_Page_009thm.jpg
5ef83a5e6ab3a1d459daa18b7e03d9a2
8e7575b88b8ddf39aa88615a793df84590126302
7946 F20110318_AACBKX rajagovindan_r_Page_041thm.jpg
b5bb0828d0465173693edbca30c14bc1
7a76d05850d2065a4887431878183b9e4bf41eae
35003 F20110318_AACBLM rajagovindan_r_Page_129.QC.jpg
e71059bbbb890028f88796c4db5329cd
49feddf5a3073c3c34dac541195e0eb558256cef
799593 F20110318_AACBKY rajagovindan_r_Page_033.jp2
e64a725e2af7f7abbdb8b46c10ec9a38
9e9bea5f9f453e0c0c6b280318a59c5b8eebaedd
1304 F20110318_AACBMA rajagovindan_r_Page_031.txt
848a6557cd7cdf1869cb6154d4818b02
ee6c7cd1583680b2ab7b270cc0c91f1e5edb3ff3
F20110318_AACBLN rajagovindan_r_Page_148.tif
17633b2e05accc29c5303621d25781b4
c1d411e621bafbe5ec16a354bb9f6bfe42f1e6b1
F20110318_AACBKZ rajagovindan_r_Page_079.jp2
b2844629a76dea759f9157a8bb8da643
c185a7b68b475753db80eaac24dd658bb96757a4
72697 F20110318_AACBMB rajagovindan_r_Page_128.jpg
de852c7707fca513d79227e592cec45c
a19fbe026ca492af3976dec073d2e1dc8e25864b
29068 F20110318_AACBLO rajagovindan_r_Page_034.QC.jpg
4a48b36048e5b9a9e84efd98851cfc32
be77e2ccec4bd59b931ef5b25fe16af3b758fbc5
2251 F20110318_AACBMC rajagovindan_r_Page_123.txt
b8217b42ec87e1ea6c46b316717b18e4
93deefb838841bbd8ef837c5217a349047f80411
113251 F20110318_AACBLP rajagovindan_r_Page_122.jpg
f18b9e70ace8976bc92324de21af860b
115a817370299a0ba2e22f631e3a3bf17c59a878
34988 F20110318_AACBMD rajagovindan_r_Page_065.QC.jpg
f1a79159ccda0ea144e79b0c3e7349c6
3c8f2a405aefacf76ea2f999f4e6a5b9140d500f
139169 F20110318_AACBLQ rajagovindan_r_Page_142.jpg
edea9e66dba439e32129442cc8048a49
82aa3118299679f1e55c3dc2f64c061da48273b1
5473 F20110318_AACBME rajagovindan_r_Page_006thm.jpg
9c78a829f2aac2f96dda8db502a51296
4257ef4bfbbff2e46ec49e363fe191d64796b548
F20110318_AACBLR rajagovindan_r_Page_043.tif
673781b9b2d1fe96b842d191e2e42fd4
bb0a3e6e04af264d91374a3eb9b7e2aafc56c08d
874977 F20110318_AACBMF rajagovindan_r_Page_098.jp2
fc0fb5d35624c60e8f470a067cfc00ab
6d5788a40e6ca65cfd95aa49f922512717c851b8
660 F20110318_AACBLS rajagovindan_r_Page_003thm.jpg
2533c959e556ece0b8deadb4fd4ed2d9
e7ccd6965396a5efab6185fdca82883c665df4f2
1966 F20110318_AACBMG rajagovindan_r_Page_024.txt
8f2d485aa790ea4caa8a7ef16adb6bae
676d52514593d5617a0f55062f848e223fe756cb
36455 F20110318_AACBLT rajagovindan_r_Page_058.QC.jpg
fc3f7cdd4d7bb486ac1b53846a8b72d9
7d8939f222aff8db7ba2cf037f39654c685c6227
1650 F20110318_AACBMH rajagovindan_r_Page_057.txt
a324d98b3e20db5b35f1946d403cf4af
2a7241da11367fb9b6015b855a3a6cd9220a0c87
F20110318_AACBLU rajagovindan_r_Page_059.tif
7538e65bc61a8c01bdcc733b36fb170b
f47e32a201533f0052cea18e46407fc2835bc01e
F20110318_AACBMI rajagovindan_r_Page_131.txt
d19473c5870fdf17bc457d5f30a2717c
00df14a96ac2e04130c27ae5553e91cf0c1966f8
53125 F20110318_AACBLV rajagovindan_r_Page_053.pro
cfe5d3cd937b1fe1c05442a6777fee97
e0bbbc566512dfb3da8fe24fe9e4bd9409c6d40c
207151 F20110318_AACBMJ rajagovindan_r_Page_011.jp2
e135adb6a2fbcedf7194aa2c80a2656c
4194cd0b013d1f8c37df85924c0ecc625d4174b3
3930 F20110318_AACBLW rajagovindan_r_Page_097thm.jpg
4b3721b5525d99fa248c62158f2be990
b168b674a1a4918b5a49f084efc44d7e812076e0
129071 F20110318_AACBMK rajagovindan_r_Page_140.jpg
922111d15d34f50ead0e28e113fa561b
402fc069f61c32c0bfbb89418a23c16b91ecc0b2
9203 F20110318_AACBLX rajagovindan_r_Page_141thm.jpg
03b3d9a8a7cd266d2dcb97b96bc4de2d
8549679066dde3ae38ec681e84ee7dd32594cad9
F20110318_AACBML rajagovindan_r_Page_043.jp2
63c67e32ebeadb201860043548d18174
0144537d3391149259fd0325da7d9c0d8ad1974e
100287 F20110318_AACBLY rajagovindan_r_Page_067.jpg
7a6735103025114dfa83569c76d3b95c
461c4ee782602dabcc833e7480358db07d4372b2
F20110318_AACBNA rajagovindan_r_Page_002.tif
cbbac58e670fcb08bfd40703c7afd308
e82d72e0c00a2ecf01f98836b4115d1946e3eb40
59231 F20110318_AACBMM rajagovindan_r_Page_095.pro
3d51ae3e571e59f863621b1b0a33e897
0592973d71d454ec5c52de28ac899aa11ff01e5a
F20110318_AACBLZ rajagovindan_r_Page_060.tif
52227e710ee5cccc9c2682724ad99e89
7cb902bd9844110f2b8631902fda0b5adddebdf1
F20110318_AACBNB rajagovindan_r_Page_003.tif
fead769f2e1efde0905c73ea95cf3d97
29d40b4c75c87a0e45e2928a00345b605f7cce90
37629 F20110318_AACBMN rajagovindan_r_Page_114.QC.jpg
2138f6609e744d480b29ad2d6ac4c5e2
11985a5eee53d461f7831097fcad386936e6a5e8
F20110318_AACBNC rajagovindan_r_Page_007.tif
e58663aa59a7f3a00882c5a0c143af67
154262975f19db57f32cd38a167ed6c12adae5ea
56005 F20110318_AACBMO rajagovindan_r_Page_119.pro
555be2a75ba8da12b782e290afa0dd05
b3d643deb017f75952f84b07c938bc5c02b62945
F20110318_AACBND rajagovindan_r_Page_008.tif
168edd81d09f62ce6f5d4be872e1c156
05e9d9350fbc88e6bea637d982dd9e3baa082779
35689 F20110318_AACBMP rajagovindan_r_Page_004.QC.jpg
572300f5d6f374ce265614a5ec4f9813
8427113d1b2dc4f9c4e907c3350d33f0864f2d08
F20110318_AACBNE rajagovindan_r_Page_010.tif
025f8396499dbdd2cc85b23bcd6a94c8
6838840d4663155ad4e1c2f3ea64c5708a75e9aa
F20110318_AACBMQ rajagovindan_r_Page_017.jp2
54997fd8bdba51c8b931e8d4255df4e9
23ec60786d3b9603ff7452fd9451b18a78cb4c4f
F20110318_AACBNF rajagovindan_r_Page_011.tif
b5cf07668d16e386013745130e12cf72
1d1a6742bbeea51a4d6aaebf41e90a0a508705e6
38406 F20110318_AACBMR rajagovindan_r_Page_075.QC.jpg
f61366effc25ad08cfabbb00558b0e65
df9aaebca5b3c8889ac5502187717e70654880fb
F20110318_AACBNG rajagovindan_r_Page_012.tif
1f5b6b68bafe76e8764a3f7ec86d76a8
dccba7ecef3560336faa650fc1475e4c411c1175
F20110318_AACBMS rajagovindan_r_Page_093.tif
7c09fbe1b3a8092238eb590bc1cf32d2
c3bdb2c63bb1dc3525a3613128536fbe04989954
F20110318_AACBNH rajagovindan_r_Page_013.tif
9f2f1bba2c30251d32646121f6162d51
25ee15df1daf8fd310b3bf62391836e33207a989
62583 F20110318_AACBMT rajagovindan_r_Page_137.pro
140cdd0f393e4f827dc4202ff359c73e
1b7c1bc12caa9e1bfaca03072325eb35c989d939
F20110318_AACBNI rajagovindan_r_Page_014.tif
fb39f074fd55e59da91e8d47aea25a8a
797599dcfff7066662edf2a6e5b74be958d67df1
590742 F20110318_AACBMU rajagovindan_r_Page_153.jp2
e143932fb97334915c881b8351d25592
32879d9067000c34f2fef1a107d4db710c2a7d30
F20110318_AACBNJ rajagovindan_r_Page_016.tif
e0210b4ea021830de0057a8386a24092
a5f11d27cb2ec417040db121cfc6218ef8e0d2c2
2032 F20110318_AACBMV rajagovindan_r_Page_043.txt
bf33361728ccfe13527c4476a027cf5a
b12f30f1805e8319d8eb0cd322b195e6d1be4491
F20110318_AACBNK rajagovindan_r_Page_017.tif
7021601e1488bc03891b1a60a7d84682
cb13abf28be88ca8621768c96519cf9fdd396d37
253258 F20110318_AACBMW UFE0022215_00001.xml
c16b7fe76b42fafbfb8331e18a206e38
5cc92b1f420b21ea950d6286bdc12cc01b96349d
F20110318_AACBNL rajagovindan_r_Page_018.tif
3d14def10bf1957e365ffe504cc25d26
a9662e6f7d3f32867ba6f2654f69903965880e14
F20110318_AACBOA rajagovindan_r_Page_049.tif
76fe8c3e81fb2f9612b8a57bc9489b06
60e5d4b82e7939603ec0fe505b00546d133984df
F20110318_AACBNM rajagovindan_r_Page_020.tif
ad228c046ad08a0fdad10b781234396d
bd26ece12c5646d8e07a958bdb19f7ca0b1f99cd
F20110318_AACBOB rajagovindan_r_Page_050.tif
755e7cffd38c29b675181b943bc36d03
7248b0ffe4e37de083b06e34ef32508dcde0c814
F20110318_AACBNN rajagovindan_r_Page_022.tif
ac87d6645ca471938fa3b8755b3c7090
9a8d9863c49a5d6cec1c8381f4ee43ed61a3f404
F20110318_AACBMZ rajagovindan_r_Page_001.tif
bde7409905a48f3a2c8c30249ee91d02
b17c1c5f2b69706566b42a2a377ee1f7ecba550e
F20110318_AACBOC rajagovindan_r_Page_052.tif
99e7cdef99aa91eaf29c3c839d40a82a
fe1b0b78d844663e8d364495b654925b0e38eb81
F20110318_AACBNO rajagovindan_r_Page_023.tif
d33adc5a1662f8fcc2ce181f05b062b6
9ef784f1158a9a011dce52be4dff29e4219a933c
F20110318_AACBOD rajagovindan_r_Page_053.tif
eccf60aafb8b8b86ba623c491cc7f7ad
39fa6214818dcf72ec03bbc0e8273839c3a0100b
F20110318_AACBNP rajagovindan_r_Page_028.tif
bde0708d6ea271278cc2c988598f5e75
54e7f21ef8c91b8fe7f1cef86841166397b1a94d
F20110318_AACBOE rajagovindan_r_Page_054.tif
2ded88bf597920e563e296f77a4f1434
d52c1064a9bbb45fab405b970f4a786d78b2953f
F20110318_AACBNQ rajagovindan_r_Page_031.tif
976adc738859d798aa9ace03638205ed
f46d2e0a80a3e09e0b50cc7d8cdf04cc7a0de91f
F20110318_AACBOF rajagovindan_r_Page_056.tif
b7b25a8ee57e1c02bd28840c2f4f9de0
e06dee19f0868b4bba20db98a229b4277182ae75
F20110318_AACBNR rajagovindan_r_Page_035.tif
7aa44455c0513cb3da8c9e58588dc584
04f5ab49f94017dcf23b614fd59f809b766275d7
F20110318_AACBOG rajagovindan_r_Page_058.tif
cde26041a097689b96702a75af9d48a7
c4d8e42cbfb25f98764b57fba41e7f5f5c5005d1
F20110318_AACBNS rajagovindan_r_Page_036.tif
35009d4bf15a61943d693446de82be25
95e9d83af9bc5cdbe70ab65ef5d954853edcfe42
F20110318_AACBOH rajagovindan_r_Page_061.tif
319ffbc0876c7affb4c6ae48e3b2e30f
f70d7bedff65c8c6630d6f1f1c264890e6c8822d
F20110318_AACBNT rajagovindan_r_Page_039.tif
308e035ab9a813927c44daadd9e92fdf
c3d809dce922e9dcf21262c31bde04f3d2569b5f
F20110318_AACBOI rajagovindan_r_Page_064.tif
ba51b5271903b51b84e24a9be750c847
728e5c6c8ad6f56fce309b1cc645eb83768f52f1
F20110318_AACBNU rajagovindan_r_Page_040.tif
0f5d1d46440b5e5d72e4688a5a89d06e
93163458af8a5e42063296aaad57c08323b0b745
F20110318_AACBOJ rajagovindan_r_Page_066.tif
ad82867a200acd875af32c879e4622cb
e49ecb047cb4fe6b5ab121035a7cc02e8ddea3e2
F20110318_AACBNV rajagovindan_r_Page_041.tif
928cfc5a911c1f4008962c8565fa6152
5f045fa910a6cae1d995cbfbeee790e1dc175ff4
F20110318_AACBNW rajagovindan_r_Page_045.tif
014337069363cb5de76ccfaefa645b1d
a364187ef83b81d9fd09d60992010a7723ebc627
F20110318_AACBOK rajagovindan_r_Page_068.tif
f37cd34bfdcdcf9f8ef630689144f983
8f9e8486713d2ce26854b2de5230d29c20d72495
F20110318_AACBNX rajagovindan_r_Page_046.tif
8810647827a014f30a2f1d82282a4a56
6e72faff1de65a21ca64d08c247dc9f237daf0e5
F20110318_AACBOL rajagovindan_r_Page_070.tif
6fe48c2234157cddcd091f60ffc88562
e3191befba24c2c3105da830855d791d2abfe7f2
F20110318_AACBNY rajagovindan_r_Page_047.tif
b3f72d8d4115903ab9d5cff45e79e558
aa5eb646fdb645d9a9acb3c378b34f223e41e111
F20110318_AACBPA rajagovindan_r_Page_096.tif
6ae77fe4a060de9804b295887df8ade3
6538aee4a1c28be469d0cf2012311efab467f578
F20110318_AACBOM rajagovindan_r_Page_072.tif
2e1051d80e6c45594cad8734a9fc62d9
f3b7be2e67a7bf042b74bbb6224cf8ba3b1fb847
F20110318_AACBNZ rajagovindan_r_Page_048.tif
a218b85c74844e142052bdfb98aa81e3
34a7d7ef72f274f7a3197adc13a2b56c5c2c7af6
F20110318_AACBPB rajagovindan_r_Page_098.tif
06706dbafbc73ab431ece7c0b70f1fce
73b7c0bd3f0e85200bcf5cb04eb6f09f7e48950e
F20110318_AACBON rajagovindan_r_Page_073.tif
8fd80a96ac62a10b9295c05e96b99434
e250f0282e06b3ddcacd296d51b5a2a5bf30fc2c
F20110318_AACBPC rajagovindan_r_Page_100.tif
a57aa96284ad9f0b688c8d9e62148c43
854f05c7f51873708394b5c1495cab0fa16a924f
F20110318_AACBOO rajagovindan_r_Page_077.tif
4fe097426497e964f798e815995aa4e3
671151809d477cda7471089de8a1299b2eecfd21
F20110318_AACBPD rajagovindan_r_Page_105.tif
aad99ae3ff39cde6b6cfdafebc9482b7
ec320b15163f6fbb2210556fb83d095960b756ac
F20110318_AACBOP rajagovindan_r_Page_078.tif
b85514b814c903f14b740d28cff2f3c6
1cab18c1f03e3b68e31d73c55bba221f8b9e6bdd
F20110318_AACBPE rajagovindan_r_Page_110.tif
c8e44b04ebce30b008b5c1bd1b64749a
e36adfde4238b4919e4f558e877d4b1e77079f38
F20110318_AACBOQ rajagovindan_r_Page_079.tif
8b82e9635f5afaaa62411332b4caf286
2da7b2c22bbe40aa04de9f9a0da079111212b219
F20110318_AACBPF rajagovindan_r_Page_111.tif
b60adf2cc5e8ebd41b66ce933c1d37dd
c81f0eddff10c6fa66198f7c0b574062d487e932
F20110318_AACBOR rajagovindan_r_Page_081.tif
4a114c24a5e107ebb3a4ba6f33deb47a
b554733e9d254983bc4c0c36988da07358f9ccc5
F20110318_AACBPG rajagovindan_r_Page_112.tif
5ea30300f787e7d9487be45564a58327
dc046f6f4b4b97fa1d2a2b4b3816ce633c492df3
F20110318_AACBOS rajagovindan_r_Page_082.tif
a0510f2dcf24ba9cc1cf7024e1bb247e
c3cac3e38fba1667caaa409a9b905b669e3e728d
F20110318_AACBPH rajagovindan_r_Page_114.tif
84805c69bf0eb57819664781dad6eabf
1bb53a5af23521e9ba784eb851992ee87a614792
F20110318_AACBOT rajagovindan_r_Page_084.tif
2bb8a4f4ca3b76d58b800ea9115acb61
79c1f54e70a95fc53cbf4ca19724d4a1fd4bbdfb
F20110318_AACBPI rajagovindan_r_Page_115.tif
de9d03efda42b8c293d128264535a4d5
054fc1ce4686b2e6727619e84c746b6ff23a4bb8
F20110318_AACBOU rajagovindan_r_Page_087.tif
4dab93b765c12d3277369157d9d1a199
0fd46a529eccf5cbe8eeb51ce183b3ed686be174
F20110318_AACBPJ rajagovindan_r_Page_116.tif
a47f6eb26398c012df5d550f254849d6
c911faab7e60d07e6bb0ffb394425c6dd3aef7f6
F20110318_AACBOV rajagovindan_r_Page_088.tif
2b0e645f065b3929d078960e119a6417
721a5d98e0c0483c77e276317dd0957eacf9bb43
F20110318_AACBPK rajagovindan_r_Page_117.tif
38c1f760a539546c6d3fd7bbe14af237
2be8cfceaf51b2227ac831585beef6491d97bcb1
F20110318_AACBOW rajagovindan_r_Page_089.tif
be99b0121d0e6c96fe35541e559034f8
f90659b4546a5cb7058cf8b0bdcf8729c11c19a8
F20110318_AACBPL rajagovindan_r_Page_118.tif
88338a5c2a51d9c2b79565f464c4454c
086291f06e03d00335494ded01d5300fa1fda924
F20110318_AACBOX rajagovindan_r_Page_090.tif
fbf46c87b18bec3811a06760cde9d7ca
52ed8b9bc3122f26c05bfa96b36932f6f7178d19
F20110318_AACBQA rajagovindan_r_Page_145.tif
edcf586fc41b4f2b2c6efa07d7383c0c
19f4e97014ba6efd8f01dec9930d966c5de3ac0c
F20110318_AACBPM rajagovindan_r_Page_120.tif
529dfede0d3a2c8be048a816361fc65e
3b675e9e67ddfb01414b06d6972cc0736bdaa3ed
F20110318_AACBOY rajagovindan_r_Page_092.tif
85287d985c4be2385444779250cb3af1
37085f50581f83abb277306ca3b5e854784649da
F20110318_AACBQB rajagovindan_r_Page_147.tif
8956db35a77316c71602eb4a43627553
e697b91e11ab54922e575efc376ef8aa36cbd8ac
F20110318_AACBPN rajagovindan_r_Page_121.tif
aeb71b35c5ea95214744cc19e3c561d1
9677a2df900a3f3067eaf247210183752debabf3
F20110318_AACBOZ rajagovindan_r_Page_095.tif
e3ff22918e0350e12a15fcfd5eda1c24
96ff574979fd75781f68b7476a550fd8eb1924bb
F20110318_AACBQC rajagovindan_r_Page_149.tif
36294e3627ce87720e84356f03ea92ac
43a092db609809dae043cababd54a5846348f572
F20110318_AACBPO rajagovindan_r_Page_122.tif
9a454d1a3aee7e3840f329d8cb037adc
135e0fc20c86655d751f323d836e663b34f84cb6
F20110318_AACBQD rajagovindan_r_Page_150.tif
fb55f90dc49ec6fba06de61acb1ebcdc
6a7a79895bbf6c6e21b8d69d760596b543b8f9b7
F20110318_AACBPP rajagovindan_r_Page_125.tif
b443e1a4b27bff286a846fa14ced6aed
19e2f4e3a21295f92183cd3c78ba721cd9665e74
F20110318_AACBQE rajagovindan_r_Page_152.tif
7b71ac30c1e25727fdcfbf80bcbd78c3
66272b8e10a5df2a2764a7adeb39b4c37b2b131b
F20110318_AACBPQ rajagovindan_r_Page_126.tif
aad09af580665197f115242e8f0ca1f2
dcae265a90fdfc8385aec0b92b7562ed4b57a7e6
F20110318_AACBQF rajagovindan_r_Page_153.tif
2494fe9005d57cbde3c5dc234a4a59f0
61a0fc956adce0cbd4068912bfaabde28d0558c5
F20110318_AACBPR rajagovindan_r_Page_129.tif
600bdab47bca196384bba67cbb763a42
e426457f35269813a76cf51e9104b2c0ccafc08c
487 F20110318_AACBQG rajagovindan_r_Page_001.txt
395abf7c76fd768f5d765efa82019918
7a13792b9f9aa236704efdb0e30593b4b348dfd0
F20110318_AACBPS rajagovindan_r_Page_131.tif
75a09b2dca21c47f0b1ebf30c73f3675
3ca1a93f04e9b865d1ebb19bf225ab178683cd37
112 F20110318_AACBQH rajagovindan_r_Page_003.txt
f32a4957f0f855ae2f12860a1ff9c2f1
a9ef668727c2ccf4f3e48f8420efe56a06fcb376
F20110318_AACBPT rajagovindan_r_Page_132.tif
25f1ae357ac1b885aaa110961d31edcb
4b345a942e706a812e632de621dee8f046df35c4
2125 F20110318_AACBQI rajagovindan_r_Page_004.txt
9c72bc1704286e26a92ffeb6e9e47c51
470fad9fcfe6a77656ebc48d581b3cb048ed482d
F20110318_AACBPU rajagovindan_r_Page_133.tif
a7858bf3014fdb295eb20a3c32fc5473
515c7a4535e82e22aa38c1adc98c07d1c99e7d69
1544 F20110318_AACBQJ rajagovindan_r_Page_005.txt
9584828bc70d9cdd00e1ddcf02a80ec0
23626c2ff9880503f124bf532dc29ef5508d0fc4
F20110318_AACBPV rajagovindan_r_Page_136.tif
e6ad6105601e99da9b6b0027a5f1c1d0
125abb37ed0d76d702c3c5f401504c181d46c6d4
4152 F20110318_AACBQK rajagovindan_r_Page_006.txt
f7de946928f98720af2b0ec2959ad37a
113d854dd5f3fc02f700b96e7426f682bfc609df
F20110318_AACBPW rajagovindan_r_Page_137.tif
add4edb911bc3cd90d4c77b3d14ca386
3f0db91e85a4f607f191df1b02064618ecc3aee7
3169 F20110318_AACBQL rajagovindan_r_Page_008.txt
05f4c4c72179d0ab1e794c08fb7b1102
ddeda7b95a6a87a083f17d412574adee271c3b99
F20110318_AACBPX rajagovindan_r_Page_139.tif
47c6eecfea74b60656261503db9e35d2
1d19e04ec1906eeed0521f48a5f75c0e677d53dc
2189 F20110318_AACBRA rajagovindan_r_Page_051.txt
233b2ef6d430c5c517e98cc479a4076b
772fc301fa8612d74db178a579a117709954ec5d
1123 F20110318_AACBQM rajagovindan_r_Page_009.txt
1ed2cba01ffedbce91b6979e25f8535c
553490ad114462e3573307d2f736607fa121d05c
F20110318_AACBPY rajagovindan_r_Page_140.tif
0e6a8496524cf0c648e9f2ab8bbd53f6
bbd0cc6fb78a3fa52161b5997d1a550f06f7e18d
566 F20110318_AACBQN rajagovindan_r_Page_011.txt
ba076e7bb3a19df0340fa5728972c18c
a341f9d6520375f28cb17970e719fe764252e5c4
F20110318_AACBPZ rajagovindan_r_Page_142.tif
3726b710fb658669b686ede421e560b4
f648ed006295fdefdd8af75e95727b210f2b4688
2207 F20110318_AACBRB rajagovindan_r_Page_052.txt
2ea18fc912a16a301d210eee30dae955
6fe8f7c0e643e73c503714cb823b665175d5a71a
F20110318_AACBQO rajagovindan_r_Page_016.txt
dc9dd8ab076ae4978d77ea2ca1cd846c
58be4fe0048439dfacd05916849809ff02758a07
2259 F20110318_AACBRC rajagovindan_r_Page_053.txt
2ccf22a26578155e662f6701877192b0
b03425054e2b47485400412c73044ef9e1ad8330
2232 F20110318_AACBQP rajagovindan_r_Page_017.txt
a1988840a793d07ff7dd3619175a8d4e
032640d3082d1a7383ca49752c6c65314b01bd1c
928 F20110318_AACBRD rajagovindan_r_Page_055.txt
1f69fc7f11ff021a10c10840f5e24f30
a10ef7dedbdbcf712cfc25941448fb1aea40f8da
2145 F20110318_AACBQQ rajagovindan_r_Page_019.txt
5f949a89344b5ac793896fe304cbb121
1bdc0030a676bad50422aaadcc6d4dc4623903fc
605 F20110318_AACBRE rajagovindan_r_Page_056.txt
0d539897d0171b42fcdd229de5cc525b
22109d658cba16b8569f3037d3c24d7eaef35036
2155 F20110318_AACBQR rajagovindan_r_Page_020.txt
dd9cf7823d7f60c313242cdc063d4e77
fcd5aad0b4efde90e1d41ffeb780480efce8d08e
2183 F20110318_AACBRF rajagovindan_r_Page_059.txt
3f9b4d818f11d50e8866f2bfbdcbf801
891a997ea36a79eab17ca714190d99ebc8c6828d
2257 F20110318_AACBQS rajagovindan_r_Page_021.txt
6d13bbdbe2e23084dc66f148db1ae579
94a2bdcbf9302ae86cbe2ddba02079a80920bae6
2092 F20110318_AACBRG rajagovindan_r_Page_060.txt
1ca7173e378156b93017b8ba7e77b4a2
cb0f3b6e76536c83bb299d765b2fb5f9f53064c3
1494 F20110318_AACBQT rajagovindan_r_Page_027.txt
3f4724e2006dbd57bac9e3e1c57f953e
7426e5b562f7d9592ae38707bee174717ba35702
2114 F20110318_AACBRH rajagovindan_r_Page_065.txt
2c440208535b3d97ec8d89d091aab73b
6da3541fad33c311b4f9ca86fc948a9a5af9f397
1545 F20110318_AACBQU rajagovindan_r_Page_032.txt
cdf74a506da7ccb0e546e329e36a01b9
6d3db1651c16111ef232b7b568d6ac928bc3066c
2218 F20110318_AACBRI rajagovindan_r_Page_066.txt
d171c84de2f55f4ab21f7f4bf14a6994
8202846ea7b3c883e01db141da770a7165067ec7
1950 F20110318_AACBQV rajagovindan_r_Page_037.txt
d2aa81f00b857af466c4a98e5cb6e787
1285d1681b682c2c8f18657ef7468cb5ea2f3f76
2225 F20110318_AACBRJ rajagovindan_r_Page_068.txt
7624f31164434b3cc7c945a2bc1eb592
127b91ca5668bed841085cb465bbd21654910f01
2194 F20110318_AACBQW rajagovindan_r_Page_039.txt
0875343de4781c00179a3836427cb6f0
f7fadae87a112fa076df60c4dde7a313ebe34c22
1016 F20110318_AACBRK rajagovindan_r_Page_072.txt
e8afcb257bf6e705cfb717e0c8167d68
e58e27bc55d7c6ec1efb23313b68e9cddd370f48
2153 F20110318_AACBQX rajagovindan_r_Page_045.txt
d4238d3c6e43fd711510417535eabfb1
16fab897a8253c789e59b617a01e189784e3e76b
2783 F20110318_AACBSA rajagovindan_r_Page_098.txt
1d1bf660675683dd8efdd82064080714
04a2c1812ec775820495867648d538a80b6da1e7
1144 F20110318_AACBRL rajagovindan_r_Page_074.txt
8eed1b3fc95b8b4d1b48c328c6409410
aa331bafa19bbf632b3530538fed41fda06e9dc3
2023 F20110318_AACBQY rajagovindan_r_Page_047.txt
fc9ff2df151d8a3e037eaf7caee70dc2
7edbb317321de87b0fe48b715ccd4ffcd35ae66c
1530 F20110318_AACBSB rajagovindan_r_Page_099.txt
772d3a0c7f2d51f89899c0e4c1a02694
bf212b708869ad7f1e5dcbf09e16ca96e574eef9
2186 F20110318_AACBRM rajagovindan_r_Page_076.txt
cacb6ee2c99eccffd74c0c306e443da8
501622f0833cb49280bdb7dc53deb9a29e89f4e0
2160 F20110318_AACBQZ rajagovindan_r_Page_049.txt
cd1eb90f4f55f394662e884dbbeaff1c
820c33012a3dfa24ead37660324d19d150b3e0c2
F20110318_AACBRN rajagovindan_r_Page_077.txt
d19208b952e7206faf321260f2722a3a
cab3410e358341a36bfa2051baac06a2a4fe5276
2308 F20110318_AACBSC rajagovindan_r_Page_100.txt
e1c49db994e1054c4d9612b33c845b8f
122b7b5a8a895371569be29528abc50f3a84e77f
F20110318_AACBRO rajagovindan_r_Page_078.txt
15e2f5b46f478adc66b2dc6ab899e517
99cc301c2767627e835b16562520a4850d914885
1555 F20110318_AACBSD rajagovindan_r_Page_103.txt
90f0d62688ca0dcf593c79202cefd19f
1201b71166b1f560a6ccb3fef910dac6385d7ed2
2136 F20110318_AACBRP rajagovindan_r_Page_079.txt
b634e0de9da790d13350639c8603c71c
0d942bae37b39c6e3d87bcd4419388078c0629ad
934 F20110318_AACBSE rajagovindan_r_Page_104.txt
f5c2e698106c53693b4475b1820e97db
caff91806782ebed1b40b25f4d6bfe0474275e0f
2249 F20110318_AACBRQ rajagovindan_r_Page_080.txt
8b9ac772b3deee1cbf18af6a47f698f6
79c9b1e9e396ac491ab2a1d5f325ce50a8aeb4a0
1190 F20110318_AACBSF rajagovindan_r_Page_105.txt
44e72d42ea526ec7d3245b42eeca3796
40eddfe4ef7853fe7a9ac75e77ca48005922a35c
2291 F20110318_AACBRR rajagovindan_r_Page_081.txt
37176838c61f2cc6680b235de2ad6990
fc5748ba937d6fa152962133a3bddc11a26a663e
1095 F20110318_AACBSG rajagovindan_r_Page_106.txt
c736889658831a3293ba3941d8ba1b24
d41af8315a4dbda878da1c55a6183cedf8bb530c
2170 F20110318_AACBRS rajagovindan_r_Page_084.txt
f1aa0d7d41bdf5dde45b7c700eb12ece
1d3403688dea11696b596b3bb7bf216759777390
2290 F20110318_AACBSH rajagovindan_r_Page_107.txt
e7d11f1af26e4843c9da0756520672bd
53775753c9746339d12973955219db07ea4e23df
2284 F20110318_AACBRT rajagovindan_r_Page_087.txt
cc3cdd0d3a1682a1c87b7463f5921a14
48e336230fc209a14310d7546e2a2ff47ff863d7
2279 F20110318_AACBSI rajagovindan_r_Page_109.txt
a28e4105107d48c2185e8802db6e83c0
50d6c7e15115301b59d6ddd48a080e8825215550
F20110318_AACBRU rajagovindan_r_Page_088.txt
551de6f2b73603e1d10e6e7302c3e222
add5d7b70815235128a5e66047568c7e8bb85af5
1919 F20110318_AACBSJ rajagovindan_r_Page_113.txt
f4ac0c33858c2794f1bc964746793525
c1e7cc11a10415991a14d8c7e0c47e4867cd0727
2312 F20110318_AACBRV rajagovindan_r_Page_089.txt
958238576356d104228a1f156eb76007
8f56bd3d9b8e4de78c4bfaf030e1c339615600ba
2188 F20110318_AACBSK rajagovindan_r_Page_115.txt
bb372eba4cd07c5676d7d3ebbe146802
c6bc3569d5bcd5fe2fdd9ff92ecf3c782d29242d
2266 F20110318_AACBRW rajagovindan_r_Page_090.txt
ab10dedf0a0e344ebe4d801eb7a4215a
cd54091c85804246b75f1f2ac868fbc061db4ead
2712 F20110318_AACBTA rajagovindan_r_Page_148.txt
121c18bf023845b03c814638cced2bf4
aa478f975d7554bbfa58d3b413f867f5010a26d2
F20110318_AACBSL rajagovindan_r_Page_116.txt
0256df8b4b90b6a0d88c0d0d278bab3d
2ded71a5170f5649168490c566a256d198a0fbc2
2178 F20110318_AACBRX rajagovindan_r_Page_092.txt
463e18456401b8b6b3b63294536fa65c
7bc16e5b75658face44fcc3546261aba4aeb0139
2664 F20110318_AACBTB rajagovindan_r_Page_149.txt
d75a0e7ab9886c239b57b23b521fb021
9ec95c34b68ae1ac3926912546b92087c6ea3927
2234 F20110318_AACBSM rajagovindan_r_Page_117.txt
83d654ac66db69c57fbec7639886a88d
ddb8b89728bc259148316d30be53e1743f14d1ce
2310 F20110318_AACBRY rajagovindan_r_Page_095.txt
847f57555ff47ea2cdbf237bff09f86c
02d550b46a3687aa7d19efbafdee6a42be6bee2b
8275 F20110318_AACBTC rajagovindan_r_Page_001.pro
aa65bd5834c39100d508f5f35efe70ff
f2ff984bccbcbec0313ab72220045bb0672e6502
2342 F20110318_AACBSN rajagovindan_r_Page_118.txt
aa8021b78bddf8c5f9f50a09b6ec6858
c53db1b90978af48580f02fb1c55d30791f5029c
681 F20110318_AACBRZ rajagovindan_r_Page_097.txt
8a410b7aef70276faeea9016890d34de
02c7a11dd89c17b9465bb02ab9b4d5028cff0ff6
F20110318_AACBSO rajagovindan_r_Page_120.txt
af41e8440ad841c3365d5745578adf4b
1ac7343e55536a5cb0812f561e2cf8ab5e7c0cca
38665 F20110318_AACBTD rajagovindan_r_Page_005.pro
72d552faa81b0296fc63d29916984b83
74b75549c93f8dc48fa6d1a2dbe2faf3c3198a13
2270 F20110318_AACBSP rajagovindan_r_Page_121.txt
01f3fe6a1b8b8850fc30da14ef91b5d9
5fc15532f177b12e4fb7adc338abb1b84bda66df
98293 F20110318_AACBTE rajagovindan_r_Page_006.pro
a90498fb1ffb5ba40b7f73ad352aec02
1683269ff49696a172dd2fa7c45a8e7c52aae355
F20110318_AACBSQ rajagovindan_r_Page_122.txt
9b5ee8a753d0449d21e7141aa0b01411
7ee19be815e97a52bb427723b939265058c4a45b
113810 F20110318_AACBTF rajagovindan_r_Page_007.pro
f2b1beab357378f5a619813e72ebed21
f6b77fd52bd9fa93ccde35fcde5b26af2ba23f38
1391 F20110318_AACBSR rajagovindan_r_Page_124.txt
613e046a13877164906a9e4ade26a314
57ce5251c1c70c4a4c34bd7ead4c529e72c5488b
74338 F20110318_AACBTG rajagovindan_r_Page_008.pro
0858220c6f6f749ce9c006f5fce2fda7
5d2c64ba423140657b970b32c23a8d967482443c
805 F20110318_AACBSS rajagovindan_r_Page_125.txt
d331e5372b571dec413ec18eb5fbe0e6
16f6c45f6543d79a0541add9031d1244b515467a
1210 F20110318_AACBST rajagovindan_r_Page_126.txt
94e3b55aa08f8eb08da172f962331551
357a89c72112040bbaae1a8ba8119dd75a28761f
14099 F20110318_AACBTH rajagovindan_r_Page_011.pro
b58eff8f4d36530ca0e83b9db07b8033
7544061fcb91f20efc8b133d902c7e4f19f0a912
2118 F20110318_AACBSU rajagovindan_r_Page_130.txt
345a4a2182b5cc6c8f767c8a71a289df
27dd31797f66b4c8997074808a698c606939ab16
30441 F20110318_AACBTI rajagovindan_r_Page_013.pro
8415b2cb945b0717de39ecab3da72ed0
1f5911ce0e632fb662fbe46315865f510e58dbfc
391 F20110318_AACBSV rajagovindan_r_Page_133.txt
45aa5829681be110554ed369bd97cb35
757aca6bc07e4c28edc59a893da457f469639527
56252 F20110318_AACBTJ rajagovindan_r_Page_015.pro
146bb3c927c8cc0bc67d55ff5df927f6
b3645adfe2406be5217f6f3f38eff40e4f0dd10f
2517 F20110318_AACBSW rajagovindan_r_Page_134.txt
42f87aa5a77e12bee18e9efe46e6892a
14c7f3a92e7f40a6f50ed679a42b3bfb5ad3a0c2
54906 F20110318_AACBTK rajagovindan_r_Page_016.pro
d498213db3f33574728461c7ade46736
162e1c04985acb91a81f6dd2c11d2be9000af649
2470 F20110318_AACBSX rajagovindan_r_Page_136.txt
deab29b86396025729829fcf234f364e
67e824f708486b473a037a45b817a6714f4177eb
55567 F20110318_AACBUA rajagovindan_r_Page_046.pro
2912f8f029c6d3cdb646923f551cd97a
3880bea6780c6096cfa3a19fda67e0b5324b0872
57069 F20110318_AACBTL rajagovindan_r_Page_017.pro
6d3952efdbedab96f30a136f0289cdb3
9f1d80522c56590aeb047aa9e670342e1db0753d
2648 F20110318_AACBSY rajagovindan_r_Page_139.txt
64f75d61283b1794ab00c021adf4204b
464c000f2f07f475d368ccd98b02fd4733eb7062
55634 F20110318_AACBUB rajagovindan_r_Page_048.pro
755ea109a22e63bcdd85be71265b832d
d3e8b698b3c1d9282340684b824c9b09cd785a83
55230 F20110318_AACBTM rajagovindan_r_Page_018.pro
4091d5e999c2fdc970b27dd4fdfc9212
1eb3c94beb8b8eb103fb3a2c1553253ce483b2bc
2708 F20110318_AACBSZ rajagovindan_r_Page_146.txt
1d6e125ab03a3127d09ffee7e503893e
9e9b3fb7111961714ed930b121daa5b160776241
55069 F20110318_AACBUC rajagovindan_r_Page_049.pro
150a8233fb58d96a95f885980d517f49
09fc265d84e50c0743ec4b8b5bf3b818c2b07a52
54627 F20110318_AACBTN rajagovindan_r_Page_019.pro
b460c8ea0fd5d2b1587900ffd456ad43
33cfa757b0cef61bc8bc60addd6e6ab6a46163a8
55111 F20110318_AACBUD rajagovindan_r_Page_050.pro
b94ecc50f4f3d251f5e45e058c91904d
506ee5c04e13447a3fc63d7c5730c91748816e8d
49594 F20110318_AACBTO rajagovindan_r_Page_023.pro
98dc8767c711a0005b76c6b3526a67c2
27039136a66a5d5087bad4432023ab0e2b5ee7c0
47890 F20110318_AACBTP rajagovindan_r_Page_024.pro
828cf8d0430c52450540a23e85de17b9
cefa1aaf53d30dbd0f3e288b4f7efd748799cf81
56222 F20110318_AACBUE rajagovindan_r_Page_052.pro
0e806f5efd83aa59aa9f46c42b3bf448
91d654afb3b31c189aa9f779935d66635553ad8a
28038 F20110318_AACBTQ rajagovindan_r_Page_025.pro
57845dcd14a01f91e70bd135d52b14e1
90013b7da0d440c239388b48b08a10593f3483b9
117346 F20110318_AACCAA rajagovindan_r_Page_087.jpg
884f3fc83fb066c55bf8d07b05907fb9
038af6bf944a8f8e42e3adecb9d8e4ddc36c989a
17082 F20110318_AACBUF rajagovindan_r_Page_055.pro
fb662572238261584767ed34fe8aca1d
d46301af46498e9b3423883113775337083af6ea
34994 F20110318_AACBTR rajagovindan_r_Page_027.pro
8202994b617c14fd0ca1b98537c5a1fb
66d835b9542f03aa573ad0a704190e838c192664
115693 F20110318_AACCAB rajagovindan_r_Page_088.jpg
e215aea16df85f129047b875f7db65a7
f190b01f1e84dd9e43593c36401f8dcf8b91b83a
13548 F20110318_AACBUG rajagovindan_r_Page_056.pro
db75f17efccdd6a1663748684d775dfe
7602bf7af903a5796ed7c10d6b5277ac9dbd018f
31618 F20110318_AACBTS rajagovindan_r_Page_030.pro
abf70b1ed1719331e687bdbec427ae8f
dfd323587209df78aba594762fbce8b17c6542ad
37949 F20110318_AACCAC rajagovindan_r_Page_088.QC.jpg
c02998c4ac3bd3c2d754bfa8be4581e9
5328f3243a0e092449be5bebd374bb41a2ed4a83
29637 F20110318_AACBUH rajagovindan_r_Page_057.pro
01eb1e41eb515a4c6a5ceaf5c1d12522
a399397f91f731ac6d1c76c6a6d5ec32c5cd1cc2
35520 F20110318_AACBTT rajagovindan_r_Page_032.pro
af3cf1e24649b594ff5c0871466d863c
78754815adcc1243fe110e88234ac9de4dc2bfde
119619 F20110318_AACCAD rajagovindan_r_Page_089.jpg
a6a50d90d2e3521b7338ce7c5be36582
6637a984ed7394a383982c25bf9748504b8e1efe
57789 F20110318_AACBUI rajagovindan_r_Page_061.pro
63efe51565c529d88a7788cf0dee8a2d
d42e4e82ce9b2aa5452d2f26c596d3c63cb2df4c
54468 F20110318_AACBTU rajagovindan_r_Page_035.pro
69060264d1b664b10397e5ac7cc6b047
acad346128b7358e4457f05914148c9dde550fa8
116014 F20110318_AACCAE rajagovindan_r_Page_090.jpg
0fb08b55d1504eee277ab51fd254060c
9149a32c362229306f1b48b5f9a5cd5300d233b2
52378 F20110318_AACBUJ rajagovindan_r_Page_062.pro
0577728e9568415c06c0f5ea5efc816b
d35890efd818cb4ab26cf21376db106998723bcf
55897 F20110318_AACBTV rajagovindan_r_Page_036.pro
8910bc7ceb763cd53cc6d1b845c550d8
370c0665b7ef06efa01bba421b239ca731df23f9
37723 F20110318_AACCAF rajagovindan_r_Page_090.QC.jpg
b09d7a3a4fc9bb861a73e85093651065
7c57f9dd36d197b96b85319a0fc3c948487fcb16
51581 F20110318_AACBUK rajagovindan_r_Page_064.pro
a072fff4f35ed0cfacc0f22e2ce23a1d
0e6e068a728dd0678256ed2c9622c20d16785c97
53784 F20110318_AACBTW rajagovindan_r_Page_039.pro
8943f2692fe0b774de4778d648bddf7e
8c71aef21408420c9bb167e65df47479435b96c8
114981 F20110318_AACCAG rajagovindan_r_Page_091.jpg
3d1a2591bb14eec966772e84e8307584
21180531af37d53f2b5aa91d3b1e0dea2afe4fbe
28348 F20110318_AACBVA rajagovindan_r_Page_101.pro
68887e3238e9f5069c9145b3bd1347c9
838f220d0187d42b8db812a39ca9803823088b33
48157 F20110318_AACBUL rajagovindan_r_Page_067.pro
35799ef2b4dec7509e659dc6e5ca3145
fe41e2206bbb09bbb64f2a54f533bd2a10b34332
55049 F20110318_AACBTX rajagovindan_r_Page_040.pro
e50d4eb889ae29f7f5a54bc594b26692
46da1bf6ce612febac32d7815dac99637a8c08ec
38137 F20110318_AACCAH rajagovindan_r_Page_091.QC.jpg
8cf7401e051bdfd09e03578573bcd403
79e2c6fdfe7c62e351718832cd247e975523779a
27313 F20110318_AACBVB rajagovindan_r_Page_102.pro
7b1f0652848ab274640f7e8c97815ce5
ecacb103a70adabb8b9c295ea0c573d96f69a993
30969 F20110318_AACBUM rajagovindan_r_Page_070.pro
33ae2a5d1e19a84aa13b14e2fa7300f2
8feb28f772a9703c2798b38a5959ad75d75c9b53
47457 F20110318_AACBTY rajagovindan_r_Page_041.pro
1bad80e493264c663b64ea9525dbec4a
d6cb3144eb226efdb1aa9095707001fa05a4870e
114576 F20110318_AACCAI rajagovindan_r_Page_093.jpg
a39a9aa8cd3c53c9a48f8673f842f7c0
414f34eb725535c3978c34c5790d621a9c7d2090
18833 F20110318_AACBVC rajagovindan_r_Page_104.pro
0a2897d4d0a40c7f17670ba2ce6eda48
0f64c80a987c6c346faf2a08ea47a26552a4a642
17671 F20110318_AACBUN rajagovindan_r_Page_071.pro
c9f29f44d89ebf1087104b64f731aa4b
e0e8f0816234ca274b563be3a505c26dc23e00d4
51322 F20110318_AACBTZ rajagovindan_r_Page_043.pro
9fc54c42625df9e25e1007bafe9074bc
96e0268814b0a2d9a30923e96a417f1cf2111723
37312 F20110318_AACCAJ rajagovindan_r_Page_093.QC.jpg
57685a4905576042e8c04a50b4432604
78f0bf5968840254eb292152fb675056ee2a9cfa
22649 F20110318_AACBVD rajagovindan_r_Page_106.pro
d3553380fa5221873f55ddcfcdb4e1fe
8757bd49b2b697f25f52a8d2c167aa76d01e2ab3
31057 F20110318_AACBUO rajagovindan_r_Page_073.pro
2c7d03f8c1880430ec2d7d3d17afb868
2200e77be85ee781e8bfefca3d7d8c648256d2a1
37367 F20110318_AACCAK rajagovindan_r_Page_094.QC.jpg
bfef0695d5ef66801033b85921d4554d
55ea91744d780d2e4835cd975e96f29f03bef3a0
58464 F20110318_AACBVE rajagovindan_r_Page_109.pro
72f5c7fa02c1e25a36693b4417f6387c
ca1329ada5e416879d0c48c5f4ad288bb34b1591
55884 F20110318_AACBUP rajagovindan_r_Page_076.pro
f52f2ce8fd1ab64c2cd833bbe11127c9
58c92141e6569d8c32302c6420b00b57549daec7
39387 F20110318_AACCAL rajagovindan_r_Page_095.QC.jpg
1920fd52c56ace7ead864e6855efa9bc
522d89d716b09b73a20b6e4c4f5e4a1f55c5f0e7
55322 F20110318_AACBUQ rajagovindan_r_Page_078.pro
cba4b38f5a0690faf62c852ca54ed564
73d550d16bfb83cd26f8e31b6998d0c1b7b78a96
37842 F20110318_AACCBA rajagovindan_r_Page_109.QC.jpg
985d3f5c080da0ba71a7d95cb247f962
b21bbc21c0ec92f2a502ee403f54a608569f4544
35854 F20110318_AACCAM rajagovindan_r_Page_096.QC.jpg
edee521c038277aff37e2dcef4b50fc9
1b3090b6c14edeb65bf11dbb4951f4c68e863db0
51412 F20110318_AACBVF rajagovindan_r_Page_110.pro
c01620f593a4f5a974c801d08210a7f8
89d2dfd50b42eadbf27333f61acad06dde2d80e3
57564 F20110318_AACBUR rajagovindan_r_Page_085.pro
26919ee79828bf7b687d5fa5790f7f50
3bba7405733b35538eb93bfa30b65e732f6887e3
33866 F20110318_AACCBB rajagovindan_r_Page_110.QC.jpg
11e340898b1bdc50ad0eb99bb0385fb3
a4e25a3e69de6d1092b4eecc54d88a81a37e501c
38367 F20110318_AACCAN rajagovindan_r_Page_097.jpg
7a9087816e5e0d545df0d1c0b7e59dab
276ebf2b1a6196ac298007f5a9e4828b4a9f96bc
57050 F20110318_AACBVG rajagovindan_r_Page_112.pro
b871c1ed3402031fae7b1ba96a488768
178dc056903f81de4a910bdd694ac818eca7e51a
57871 F20110318_AACBUS rajagovindan_r_Page_088.pro
45647548884de1f1938c0bc2935ec915
01d10cffb574c4c264e608d4f9fb95f2c79dcb84
114870 F20110318_AACCBC rajagovindan_r_Page_112.jpg
8735c71f4810efccadacb9d921f7b719
6eca1cf5b1a58ff97293f66e093e8278bfa15a8a
12471 F20110318_AACCAO rajagovindan_r_Page_097.QC.jpg
2c31eb1b6e41bebc662141e75288fbd6
beca4e4ea6a54cc51099bc41cb975e44e82fb31d
48408 F20110318_AACBVH rajagovindan_r_Page_113.pro
0197a0944edf66abb7c166fec4e90bd2
ace2b675ce5a0b3faaf84527b5c4b9ee08190c71
59071 F20110318_AACBUT rajagovindan_r_Page_089.pro
d92758facee78156a7fc05345a1e0dc4
4e37a76fa2dd3313ba874b01c2e55f317f1ceeb1
37829 F20110318_AACCBD rajagovindan_r_Page_112.QC.jpg
e709338b70844d05b1bc72d8a9d7a249
07a236fbd6a3d821ad183a78f63e292d37afcb88
55920 F20110318_AACBVI rajagovindan_r_Page_115.pro
f965c8ce4b72246a00a61834607920cb
d047424938b7043d099624990c476f5bda8c2f83
57972 F20110318_AACBUU rajagovindan_r_Page_090.pro
7611143393eed2015c34f98224465744
5daca690a11a5fe57904919b84da076cfa2e6e58
98332 F20110318_AACCBE rajagovindan_r_Page_113.jpg
04257b6b35b0d08b1cc3f5564d1db675
30a1b21c1790d30a71ef4abd0e156c7791f4d349
85746 F20110318_AACCAP rajagovindan_r_Page_098.jpg
9201489c0c1072d7db1f9617fc98059d
a789fa0ec1f75c4fd69d3aa8b3a648c8d16831fa
55941 F20110318_AACBVJ rajagovindan_r_Page_117.pro
7755b0442e1d39287dedf71e808bc111
303be0cbe06f4c6d892569f661757b6857f7349e
57874 F20110318_AACBUV rajagovindan_r_Page_091.pro
d40d16b9e4099ff417f435caec750032
2c598f471cb2306435fd7cf9c855d099e51b2c4d
113578 F20110318_AACCBF rajagovindan_r_Page_114.jpg
e4820c88f2dd7c2c18753c8238fbfd19
61b29bedd27e1842964120aeaad38f2438c14042
25189 F20110318_AACCAQ rajagovindan_r_Page_098.QC.jpg
a5a4d7bc44ccc969d5e1735462515155
c9b274e845393defdeb1293d1c9e68306dba7f88
55711 F20110318_AACBVK rajagovindan_r_Page_120.pro
e1c77a52df01c7f35093a763d9bf5504
dcbea47f496ef84f43203148c632974c70c68421
56726 F20110318_AACBUW rajagovindan_r_Page_093.pro
a75ca2008c3a5de7df96e24488840f8a
f1458e3c3a3a6b7d9ee4770495afa20b1c050d01
112796 F20110318_AACCBG rajagovindan_r_Page_115.jpg
5f437a9ae2b867ea75ecc92960d3abe2
dcdc4b9071cc3b039895a3fca5f25a87b6387ca3
104446 F20110318_AACCAR rajagovindan_r_Page_100.jpg
2092cffa71ed44435a10a05f411d69ef
c2ff5cb024f7bcd96e072fba01387ea5e56df33f
58064 F20110318_AACBVL rajagovindan_r_Page_121.pro
73dc16f254441a2e0aabf980dc39b72a
d32756c07a1a31859f686be8ec0feaa3cca05225
54815 F20110318_AACBUX rajagovindan_r_Page_096.pro
d01e560b6e7ded7c047647b2e9e7d367
e06ab32880905e77e5ed3f78d73b0a47e4c5c1a0
62247 F20110318_AACBWA rajagovindan_r_Page_145.pro
1b8dbb4e4199914398f590c643ddc8da
2db3d6362e6b4ebdb5fae4c86a09252ef5f30e92
36864 F20110318_AACCBH rajagovindan_r_Page_115.QC.jpg
127a2d6bae1f6ba3682e684a1cefb0f2
6ea8a7182d7d3dee8a1febcd018665a70f8cc983
18715 F20110318_AACCAS rajagovindan_r_Page_102.QC.jpg
ede66b673163c077dd86cb8229dbb688
0bdf6d9a70a0fae1916cb09804cf3bd5ce6c42c3
57138 F20110318_AACBVM rajagovindan_r_Page_122.pro
6a5a2d08912a3e48528f47576867fd8e
40b9350b0ffc0cfa578309c0c21ca27d40bfa274
14417 F20110318_AACBUY rajagovindan_r_Page_097.pro
61e40a944a5b2ba12af06417fe742c3c
bfb2f1e24466d880d8ab46212dbd3d9a8355fdb8
67106 F20110318_AACBWB rajagovindan_r_Page_146.pro
03c0eddb98a002a9096e42037afd6a93
f14f4841afdb368b6b5dee12521b669727e471d6
109207 F20110318_AACCBI rajagovindan_r_Page_116.jpg
6866f42e6a3b678dc2c95ac7844aff5f
065e529833cc969cba3fee9484973dd2a493ba5e
81162 F20110318_AACCAT rajagovindan_r_Page_103.jpg
89238d7c5b5db35f0a808b7a7c56659d
ce3d658d65daef1cef7d795a5c5fe2c6ac569543
57704 F20110318_AACBVN rajagovindan_r_Page_123.pro
aee966647327dc9dcb5e852cd5f97c19
b44126c52bf029e09c8a2fc1ac6cf81bfc84d23a
47108 F20110318_AACBUZ rajagovindan_r_Page_098.pro
5aefc0f2db58f8039ebc34854c13d3c7
ae9f6a4fe46012bca1cf072107769751139f43fe
67398 F20110318_AACBWC rajagovindan_r_Page_147.pro
5aadc372d0587304c75be65501258bfe
68c135274724d3f3adc6490509385ff68d10b7a7
110835 F20110318_AACCBJ rajagovindan_r_Page_117.jpg
b19062de876f4301442748ba6b283481
e159ab3bdcd275f47250e0ef5f976be896eebd8d
23879 F20110318_AACCAU rajagovindan_r_Page_103.QC.jpg
3fddacb9aecdb975cbdd1122fcdbd41f
4414e8db2b29c850d14a45a2d23f08b7c7c843e5
17061 F20110318_AACBVO rajagovindan_r_Page_125.pro
dedfeaffb9d36fbc651d2c0be89bab54
3a9c436c09b8f40b887174474b47f6bd3e624752
67214 F20110318_AACBWD rajagovindan_r_Page_148.pro
11b5791830d77796a9f3c320833d2652
b86af3a96fc880b4616303c39bf40c4e177a49d9
37271 F20110318_AACCBK rajagovindan_r_Page_117.QC.jpg
326384c76f0369d1745265ea43aa253e
3d48977d6435e4b1b92f5c46dccf39114b01f563
47317 F20110318_AACCAV rajagovindan_r_Page_104.jpg
1abfd6102fbe2f4178482071e3783e99
7dc55ed42da4106e477341e2da1923647f5daa21
27269 F20110318_AACBVP rajagovindan_r_Page_126.pro
0413ee9aa32c31821c55eda5db502f37
b397001a24c24fbaf40d365bac0078f5e9401540
25185 F20110318_AACBWE rajagovindan_r_Page_153.pro
c934b2fcb1da80b825cb2f013033e071
566851051623ec4221225a609563f2368855f3a0
111278 F20110318_AACCBL rajagovindan_r_Page_120.jpg
f9cbdf45a1c712245e66ff87e9ba1f8b
fe5bd7150cf2afb997e1273e5704940ad39016af
25088 F20110318_AACCAW rajagovindan_r_Page_106.QC.jpg
e0e3610010923dc890166ac71bb037a8
e3d60217c6f1deb364aea13da7342a91b277f4bf
27663 F20110318_AACBVQ rajagovindan_r_Page_127.pro
ab3b243481616358cfe48610bf812340
0268a0c4a5fb60155e970f6be6b09e6d6f3fd66c
4563 F20110318_AACBWF rajagovindan_r_Page_002.jpg
0063598cbd92bd5817e784ed50aec8e1
6dc8e2ccaf3fe56022afa1f9da03e23384b0fba4
127382 F20110318_AACCCA rajagovindan_r_Page_134.jpg
3823c06c08d4d19a3e8d9c96fe7dd8a8
0d2d693ee6006b599f70ccd08eaf48003d374234
116219 F20110318_AACCBM rajagovindan_r_Page_121.jpg
cd6783844efc5ce72b002d54f61cf4e7
6ac8e1e5034ddaea022f11f5472c8a91c4abfc48
112313 F20110318_AACCAX rajagovindan_r_Page_107.jpg
99b0f16c5591bfe0e45543f1ac783c1e
a276421ad76f443fe5138d7ee0f0d3f81d9acfbd
53840 F20110318_AACBVR rajagovindan_r_Page_129.pro
853b8cb1b7f92238d42c5652161f4e70
026c6f987f24dcba4dbe3d833ef101cc44f3478a
123997 F20110318_AACCCB rajagovindan_r_Page_135.jpg
cb82bc5d4bd93cd37558ab77e14cd1fc
897c63392fed6c95b72ec0095f23d1979c5c0fd0
37833 F20110318_AACCBN rajagovindan_r_Page_121.QC.jpg
d3e92e577d3b78a7438f1cbd1f4b40d2
99799711586a10ef74b75de6abc32e7468077d5a
111406 F20110318_AACCAY rajagovindan_r_Page_108.jpg
3ab81d4aa9ef4af9ede602e81dbbc039
0344f78bd1bf0d1c4cff8a31ebda32d2e440ce72
56754 F20110318_AACBVS rajagovindan_r_Page_131.pro
dbb5e78f5fa29e369ec9a34567f3c80e
b69c8b770e53341d79d0ef038e0f899f83a475d4
1281 F20110318_AACBWG rajagovindan_r_Page_002.QC.jpg
4938d6be4c142fda86f1adb8c3727129
0505e9ef78349682572668b467e22b92e16fca6c
130275 F20110318_AACCCC rajagovindan_r_Page_137.jpg
f7a49d83ca67185dd0fbed389ab601e4
f9d1e6abbddf2cdac741baf52ca31a9d9ae96d10
115639 F20110318_AACCBO rajagovindan_r_Page_123.jpg
be3bd22a15b721d14c2294a373b51741
c2d7dc98b0791ac43618337b6bb10ddab56e362d
117254 F20110318_AACCAZ rajagovindan_r_Page_109.jpg
1c0afdf87ea251427a96e81e29dd241e
17b973d77b93c1ec4e152f144182b98a18227504
56763 F20110318_AACBVT rajagovindan_r_Page_132.pro
87678d90ed63500646e03d93c1777f5a
21c9b63247b122abb2b643f9161d627a03a25929
81658 F20110318_AACBWH rajagovindan_r_Page_005.jpg
f947dea2f1fc7dfcf14f8477f5091cbe
64a76706014314f1b71685cfee2578873a2408ea
37240 F20110318_AACCCD rajagovindan_r_Page_137.QC.jpg
c50c91fb477648c140c7eb7432a2fabc
2c6137a5783b4eff9245dfa5a1b3797decd8b777
64859 F20110318_AACCBP rajagovindan_r_Page_124.jpg
6eb468dd2898beeef8c89af7e9a8986f
6f867b75a3a27033690df1706a7bfb6d86d94f4f
9828 F20110318_AACBVU rajagovindan_r_Page_133.pro
ab1f2586e6b78286d5fb7d9f85bc6615
a50932c82c059844b2a73de0f221a7466e595f85
28541 F20110318_AACBWI rajagovindan_r_Page_007.QC.jpg
b8250942517d35ca2e0d746d123ed137
836b903460ea9b7b8b6d70ab0ef94628ab919432
37887 F20110318_AACCCE rajagovindan_r_Page_138.QC.jpg
809ceb367113efd92a0fa0f38ea04b27
093a6020fe2ab7857945f62661fe0dcd6765e669
63192 F20110318_AACBVV rajagovindan_r_Page_135.pro
cc7d165b31dc327b39b39af27fa157f8
e1df16ed5edd0a956389d945f21491f47c8bb7e8
19942 F20110318_AACBWJ rajagovindan_r_Page_008.QC.jpg
18b9b54e1750fd43a7b3daccffe17841
056d3c786d27f5357f449b2762eadf6f47dcb825
39343 F20110318_AACCCF rajagovindan_r_Page_139.QC.jpg
4c23922e07a46cf4066f949e7aff010a
f1c76d84e3d9af8f9cdbf7850976bae3cf26b7b9
19732 F20110318_AACCBQ rajagovindan_r_Page_124.QC.jpg
63a50e4aa9ca0cdfc5eddccf45907666
fc7c9f970dab63250b0aa581972bf5fbb89306eb
64767 F20110318_AACBVW rajagovindan_r_Page_138.pro
5bb0c97706a2130d04de0faf801eb6c7
452369995d2cd0d94af665eda75add6d22470bbc
39987 F20110318_AACBWK rajagovindan_r_Page_009.jpg
375f12dc2963af7f07a4b3652e3943b1
6493f109da47a64df84876e84422ee054bbb7b01
39473 F20110318_AACCCG rajagovindan_r_Page_141.QC.jpg
2ad1337444063189e3190a292a6aeaf4
8feb33e70be73a19bd6e5fdda08de1732697adc0
42884 F20110318_AACCBR rajagovindan_r_Page_125.jpg
cb6a3e9eaa7696bedca187f682c5dd6f
4e13e96f95fc5cdb05437b564ff843aa8a37710c
67342 F20110318_AACBVX rajagovindan_r_Page_141.pro
2874cd45c1b8a25e89024dcbc8318404
4525b02db931f58284580b91fc6004278ee042a2
24862 F20110318_AACBXA rajagovindan_r_Page_027.QC.jpg
9b49454b892ab8f28d3ab0135f0ec5e5
de178090f19777815f5267d636de673812f4e469
32133 F20110318_AACBWL rajagovindan_r_Page_012.QC.jpg
a77c558de105b25d2af7c7d2eaa518df
57c4b8adb7360dc0c41c0b33646035daaaf91ff4
39359 F20110318_AACCCH rajagovindan_r_Page_142.QC.jpg
b0afc538375b5bb3734ee1aa7f34ed50
3c05602abe1fafcfca7a3f49f554012df0f095be
13487 F20110318_AACCBS rajagovindan_r_Page_125.QC.jpg
cfa2b757fac9c89745dc7f47cf857ce8
fd244417ba0a47d5d1e9ce0427ce878493dfa035
62294 F20110318_AACBVY rajagovindan_r_Page_143.pro
71909a3831a76662d182acb6492ad94b
d6508c5438cc7068f84530e1f7bcefc9300ed728
19101 F20110318_AACBXB rajagovindan_r_Page_028.QC.jpg
ef7ce7973b5bb9799af7ca54d7342b61
2e1dd56b8516ce35395b437ca65cf0e99656816f
34980 F20110318_AACBWM rajagovindan_r_Page_014.QC.jpg
2c9e5de66a680656de367578cc421109
85314719a50c16528cb03a00a8600c9bde8e26de
130235 F20110318_AACCCI rajagovindan_r_Page_144.jpg
b54d778705bb9cddbdfbf97317061796
9dfde939ce6fd989888a0b568dc65311bc677444
23380 F20110318_AACCBT rajagovindan_r_Page_126.QC.jpg
1ff4d387104fc44820162fa0592fb175
e6835cf400b1f3739b3df9cca6dba7a5a66f720e
66221 F20110318_AACBVZ rajagovindan_r_Page_144.pro
57ae08496a18d5760b570681f73da326
61abb6672bcd61cc93394ecd3293e177d5214806
65753 F20110318_AACAUA rajagovindan_r_Page_139.pro
c8f2fb69ca800cb6db8497292c52b625
e38556a62d73b0b04b46f204eb244fa68de923fc
69911 F20110318_AACBXC rajagovindan_r_Page_030.jpg
3746836cbf15ff1370e39ec29fb2c6a0
fe96ac95058c315ea76ceccdb98915dac267fb2b
113597 F20110318_AACBWN rajagovindan_r_Page_015.jpg
ed5e57cd1a32c22017f45a79f8832781
a16b3bd3fed472ec504023012c9b0b31b170142e
124586 F20110318_AACCCJ rajagovindan_r_Page_145.jpg
57c17d7504c1e5dce8162f7aa8071124
201855dd969b506d7aebae706772e75209bc1627
60726 F20110318_AACCBU rajagovindan_r_Page_127.jpg
03c07647d50e5ff6bd935a60634f3886
0eb55ec0a11032c89ee628d5988327c95c0ef07b
8790 F20110318_AACAUB rajagovindan_r_Page_040thm.jpg
2af7339baf1b65c04324902c2372c178
66943a62a7ce494eec863bea49e7cad71e225516
23826 F20110318_AACBXD rajagovindan_r_Page_030.QC.jpg
50f33514fdeee03f9f6ead2a3e462322
e62d198b25c481f2baf2a28d383830e9016130c5
110429 F20110318_AACBWO rajagovindan_r_Page_018.jpg
52a5db8be4b3360243db3aaf4c3874b9
5d1c0583df543520c6e363348f58777fe309f8c0
36110 F20110318_AACCCK rajagovindan_r_Page_145.QC.jpg
94e9dfd37f3ec5ce6b848856c0a16a9b
64feaeb63836e6fcadc622813d97a599425b05ad
17096 F20110318_AACCBV rajagovindan_r_Page_127.QC.jpg
d9da203754c15a2c1d70ba5546e758bf
6763e327684d6e6bfc0b3d6af2e43708074339f6
4606 F20110318_AACAUC rajagovindan_r_Page_101thm.jpg
98dc61e5115dc014d42a0a0ab3cc4541
d1859e3f8e2862e0fc980441bc41a733d8886e99
52484 F20110318_AACBXE rajagovindan_r_Page_031.jpg
6e5785db3155e234cb9af049d99bedf2
123063d3b2e0eeb4b749ad92877a6016e6a03ed9
35294 F20110318_AACBWP rajagovindan_r_Page_018.QC.jpg
f6da24bd53fbaacd12ec7f0b0323c3fa
bee7f26f9c48be90bc468c1d7fc6a574b490bf5b
135543 F20110318_AACCCL rajagovindan_r_Page_146.jpg
2da3a8ca624abc4a2ab83a639e01e201
c20af0a192aa434a3f6915c721168f2aee79b9d3
22939 F20110318_AACCBW rajagovindan_r_Page_128.QC.jpg
2593a6460583a214de00531b163fc44b
fe2777040a8d471f616f500a785739d269e26a0a
125316 F20110318_AACAUD rajagovindan_r_Page_151.jpg
5ffebbc97c17ee8b4af5e03b0694e5ef
e109a34e11bb338f3fe04f04ae0a831dd7daea04
80408 F20110318_AACBXF rajagovindan_r_Page_032.jpg
dab8b340aa2c7198c8adb0dfc3597bba
d57ab6fa8ca5561cd2b513c9aa1afe2d50fc1a9d
110007 F20110318_AACBWQ rajagovindan_r_Page_019.jpg
062e6ef80e9cc53085a39c62375a5a4a
9130167115dc0242571ef1115dee29493b2b2462







FUNCTIONAL ROLE OF NEURAL OSCILLATIONS IN ATTENTION AND MEMORY IN
HUMANS



















By

RAJASIMHAN RAJAGOVINDAN


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2009


































2009 Raj asimhan Raj agovindan


































To my Mom and Dad, for their love and support









ACKNOWLEDGMENTS

This project could be accomplished only with the outstanding guidance and patience and

generous support of my mentor Dr.Mingzhou Ding to whom I am deeply indebted for imbibing

in me an exceptional perspective on highest quality research, and also for his constant

encouragement and constructive role throughout my stay at UF. I cannot thank him enough for

the numerous opportunities he has presented me with through all these years that have helped me

gain invaluable experience and confidence.

I sincerely thank Prof.Venkateswaran Nagarajan of WARFT, for nurturing my early

interest to study the brain and with whose motivation I began to pursue studies in neuroscience.

I also owe a deep debt of gratitude to Dr.Kimford Meador, Dr.Benzi Kluger, Dr.Linda Hermer,

Dr. Thomas DeMarse, Dr.Georges Ghacibeh, Dr.Andreas Keil, Dr.Hans van Oostrom, Dr.Paul

Carney and Dr. Govindan Rangarajan for their cheerful enthusiasm, constant support and with

whom I have the had the pleasure of collaborating and learn on numerous occasions that has

contributed immensely to my scientific development. I am also grateful to Dr.Clifford Saron and

Dr.George Mangun at UC Davis for their kind reception, the many insightful discussions that we

had during my short visit to Davis and for sharing a ton of their excellent experimental data.

A special thanks to Kristopher Anderson for an extremely fruitful collaboration over the

past years and for making the project with the epilepsy patients happen. I am also thankful to

Dr.Kenneth Heilman, Dr. Susan Marino and Dr.David Loring for all the insightful discussions

during our work with the patient population. I would like to thank all the epilepsy monitoring

unit staff, Cathy, Bonnie, Becky, Andrea, and Augusta at the Shands Hospital at the University

of Florida, for their kind assistance throughout that study. I would also like to thank Tasmeer

Hassan and Joshua Proemsey for their support with data acquisition for the 'PD-fatigue' project.









I am deeply grateful to all the volunteers who graciously endured our studies, without whose

participation none of the studies would have been possible.

I am also extremely thankful to all my colleagues and friends including Xue Wang,

Yonghong Chen, Mukesh Dhamala, Alex Cadotte, Xianzhi Shao, Yan Zhang, Anil Bollimunta,

Sahng Min Han, Hariharan Nalatore, Mo Jue, Yuelu Lui, Xiaotong Wen, Chao Wang, Chien Jui-

Hong, II Park, Sanjeev Malkani and Krishna Bharath for the good times and for the many fruitful

discussions that we have had not just limited to science that has been an integral part of my

experience at UF.

I owe a tremendous amount of gratitude to the past and present BME family including

Dr.Hans van Oostrom (once again!), Katherine Whitesides, Tifiny Dyer, Anide Pierre-Louis,

April Derfinyak, Laura Studstill, Michelle Griffin, Mary Wall, Danielle Wise and Art

Bautista-Hardman who handled all the financial, administrative, computing related resources and

so many other nightmares that they made sure I did not even have to worry about and their

immense patience and delightful support while I repeatedly goofed up on my registration, all my

paper work and not to mention the computer mishaps! Above all, my heartfelt thanks to my mom

and dad and all my teachers through all these years who enriched me and my experience, for

their constant encouragement and support, their infinite patience, their affection and their

unwavering confidence in me.









TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ..............................................................................................................4

L IST O F T A B L E S ......... .. .... ............... ............................................................... . 9

LIST OF FIGURES .................................. .. .... ..... ................. 10

ABSTRAC T ................................................... ............... 12

CHAPTER

1 IN TR OD U CTION ......................................................... ................. .. ........ 14

2 G E N E R A L M E T H O D S .............................................................................. .....................2 1

2.1 Parametric M ultivariate Spectral Estimation ..................................... ............... 22
2.2 G ranger C ausality A nalysis................................................. .............................. 25
2.2.1 Tim e D om ain Form ulation.......................................... ........... ............... 25
2.2.2 Frequency Domain Formulation ..................................... .................27
2.2.3 Mathematical Formulation of Block Granger Causality ..............................32
2 .2 .4 In terp retatio n .................. .................................................... .................... .... 3 4
2.2.5 A ssessm ent of Statistical Significance ......................................... .................35
2.3 M ultitaper Spectral Estimation .................................. .....................................35

3 DECOMPOSING NEURAL SYNCHRONY: TOWARD AN EXPLANATION FOR
NEAR-ZERO PHASE-LAG IN CORTICAL OSCILLATORY NETWORKS ....................39

3 .1 In tro d u ctio n ............................................................................................................. 3 9
3 .2 M materials and M ethods................................................................................ ...... ...4 1
3 .2 .1 S im u latio n ................................................................................................... 4 1
3.2.1.1 Setup .................................................. 41
3.2.1.2 Positively correlated common input............................................41
3.2.1.3 B directional interaction ........................................ .....................42
3.2.2 Experim ent .................. .................. ............. .. ......................... 42
3.2.2.1 B ehavioral paradigm ...................................... ......................... 42
3.2.2.2 D ata acquisition ............................... ..... .......... .. .. ........ .... 43
3.2.2.3 D ata set selection ................................................... ............... ... 43
3.2.2.4 Tim e series decom position ...................................... ............... 43
3.2.2.5 D ata analysis protocol..................................................................... 45
3.2.3 Logic of A analysis Protocol .................................. ........................................... 46
3.3 R results ............. .. ............ ..................... .............................. 47
3 .3 .1 S im u latio n ................................................................................................... 4 7
3.3.2 Experiment ............... ............................... ... .... .. ...... .. 47
3.3.2.1 Network identification ..... .................... ...............47
3.3.2.2 Phase-lag distribution ........................................ ...... ............... 48


6









3.3.2.3 Synchrony decomposition and near-zero phase-lag .........................48
3.3.2.4 Instantaneous causality and phase-lag ..............................................49
3.3.2.5 Bidirectional interaction and phase-lag ........................................... 49
3 .4 D iscu ssio n ........................................................................... 5 0

4 PRESTIMULUS THETA OSCILLATIONS IN FRONTO-TEMPORAL NETWORK
PREDICT SUBSEQUENT MEMORY RECALL IN HUMANS..............................58

4 .1 In tro du ctio n ...................................... ................................................. 5 8
4.2 M materials and M ethods............................................................................ ............. 60
4.2.1 Subjects and Electrode Placem ent ............. ................. ................. .............. 60
4 .2 .2 P arad ig m ...................................................... ................ 6 0
4.2.3 D ata Preprocessing...... ....... ..... .. ...... .... ........ ...... .. ....... ..... 61
4 .2 .4 Spectral A naly sis........... ........................................................ .. .... ...... .62
4.2.4.1 Pow er estim ation ......................................................................... 62
4.2.4.2 Significance test for difference in power between conditions............62
4.2.4.3 G roup pow er analysis ........................................ ...... ............... 63
4.2.4.4 Inter-areal synchrony analysis ....................................................... 63
4.2.4.5 Random permutation test for statistical significance........................64
4.3 R results ............... ..... .... ............ .. ............................................65
4 .3 .1 P o w er ............... ..........................................................6 5
4.3.2 Inter-A real Phase Synchrony ............................... ... ............................... 66
4 .4 D iscu ssion ....................... ......... ............. .. ... ..... ........ .............. ............ .67
4.4.1 Prestimulus Ongoing Activity, Theta Oscillations and Memory Encoding......68
4.4.2 Inter-Areal Synchrony and Network Excitability ............................................69

5 FROM PRESTIMULUS ALPHA OSCILLATION TO VISUAL EVOKED
RESPONSE: AN INVERTED-U FUNCTION AND ITS ATTENTIONAL
M O D U L A T IO N .............................................................................. ........................ .. 7 5

5.1 Introduction ...................................... ............. ................. .......... 75
5.2 M materials and M ethods.................................................................... ............... 77
5 .2 .1 Subjects ................... ............... .......... ......................................77
5 .2 .2 P arad ig m ...................................................................... 7 7
5.2.3 D ata A acquisition ................................................ ............ 78
5.2.4 D ata Preprocessing ...................... ......... .................................. ............... 79
5.2.5 Behavior and Event-related Potential Analysis...................... ...............80
5.2.6 Correlation Between Prestimulus Alpha Power and P1 Amplitude ..................80
5.2.7 Characterization of Ongoing Alpha Activity .................................... .........81
5.2.8 Com putational M odeling ............................................................................ 82
5 .3 R e su lts ...................................... ..................................................... 8 3
5 .3 .1 B eh av ior ...................................... .............................. ................ 84
5 .3 .2 E R P A n aly sis ................... ...................................................... .. .... .. .. .. 84
5.3.3 Prestimulus Alpha Oscillations ..................................................... ................ 85
5.3.4 From Ongoing Activity to Evoked Response: A Model............... ...............86
5.3.4.1 The sigmoidal input-output function ..................................86
5.3.4.2 Model reformulation in terms of electroencephalogram variables.....87









5.3.4.3 Experimental test of model predictions ............................................88
5.3.5 Attentional Modulation of Baseline Ongoing Activity: Possible
M mechanism s ........................................ .... .................................. 89
5.3.6 Role of Excitability: Validation in a Single Neuron Model..............................91
5 .4 D isc u ssio n ............................................................................................................... 9 2
5.4 .1 T theoretical M odel .............................. ........................ .. ........ .... ............93
5.4.2 M odel Testing and V alidation...................................... ......................... 94
5.4.3 P1 Genesis and Attentional M odulation .................. ................................. 94
5.4.4 Alpha Oscillation: Neuronal Mechanisms and Attentional Modulation ..........95

6 TOP-DOWN CONTROL OF HUMAN VISUAL CORTEX IN ANTICIPATORY
SPA TIA L V ISU A L A TTEN TION ............................................................ .....................107

6 .1 Intro du action ................................................................................ 10 7
6.2 M materials and M ethods............................................................................... ...... .. 110
6 .2 .1 S u b je cts ............................................................................................1 1 0
6 .2 .2 P a ra d ig m ................................................................. ................................1 1 0
6.2.3 D ata A acquisition ..................................... .................. .............. .. 111
6.2.4 D ata Preprocessing .................................................................................... 111
6.2.5 D ata A analysis ..................................... .................. ................ 112
6 .3 R e su lts ................................................................................ 1 14
6.4 D discussion ................................... .. .............................. ... ....... ..............117
6.4.1 Alpha Band Oscillation: A Neural Correlate of Top-Down Attentional
C control ...................................... ..... ....... ................. . . ............ 118
6.4.2 Top-Down Control of Visual Alpha: Excitation or Disinhibition?...............19
6.4.3 Top-Down Control of Visual Alpha: Transient or Sustained?........................ 123

7 CON CLU SION .......... ............................................... .. .................. .. 129

R E F E R E N C E S .........................................................................134

B IO G R A PH IC A L SK E T C H ......................................................................... ........................ 153




















8









LIST OF TABLES


Table page

3-1 Influence of increased instantaneous causality on phase-lag..............................53

3-2 Influence of increased bidirectional interaction on phase-lag .......................................53

3-3 Mean beta band phase-lag (in radians) for monkeys GE and LU...................................54

3-4 Contribution of intrinsic and exogenous components to total interdependence ..............54

3-5 Correlation between instantaneous causality and phase-lag......................................54

3-6 Partial correlation between instantaneous causality and phase-lag .................................54

4-1 E C oG electrode inform ation...................................................................... ..................70

4-2 ECoG bipolar electrode derivation summary.. ..................... ...... .......... .............. 70

6-1 Attention modulation of oscillatory activity. ...........................................124









LIST OF FIGURES

Figure page

3-1 G ranger causality spectra....... .................... ............ ........................... ............... 55

3-2 Schem atic Granger causality graph ............................................................................55

3-3 Phase distribution......... ............ .............................. .......... 56

3-4 Influence of comm on input on phase-lag...................................... ......................... 56

3-5 Coherence and phase spectra for site pair (S1,7b) in monkey GE..................................57

3-6 Phase locking modes.................................... ... ........... ......... .... 57

4-1 Approximate placement of electrode grids for each of the five subjects.........................71

4-2 Schematic of the experimental paradigm ...............................................71

4-3 Prestimulus differences in spectral power between the recalled and unrecalled
ensembles of words ............. ................................. .. ..... ........ .. 72

4-4 Com prison of theta band pow er ........................ .. ........... ..................... ............... 73

4-5 Com prison of inter-areal synchrony........................ ... .......................... ... ............ 73

4-6 Comparison of percentage of bipolar signal pairs with significant PLVs .........................74

5-1 Experimental task ................................. ........... ................... ...... 97

5-2 ERP attention effect .................. ........ ............... .. .. .. ............ 98

5-3 Scalp distribution of attentional enhancement....................... .............99

5-4 Attentional modulation of alpha band activity....... ...............................100

5-5 Schematic illustration of theoretical model ............................................101

5-6 Model reformulation in terms of EEG variables.................................102

5-7 Relationship between prestimulus alpha power and stimulus processing .......................103

5-8 Schematic illustration of attentional modulation mechanisms of ongoing activity......... 103

5-9 Gain ratio as a function of prestimulus alpha ........... .......................... ................104

5-10 D distribution of alpha pow er ......... ......... .. ....................................... ................ ..... ... 105









5-11 Computational study of relation between excitability and gain....................................106

6-1 Experimental task ............... ................ .......... ................ ............ 125

6-2 Top-down attentional control of visual areas........................................... ..........126

6-3 Pre-target alpha band activity predicts performance ..................................... ........127

6-4 Temporal evolution of regional and network-level alpha band activity ........................128









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

FUNCTIONAL ROLE OF NEURAL OSCILLATIONS IN ATTENTION AND MEMORY IN
HUMANS

By

Raj asimhan Raj agovindan

December 2009

Chair: Mingzhou Ding
Major: Biomedical Engineering

Synchronized oscillatory neural activity involving widespread neural ensemble is common

in the central nervous system, and is hypothesized to enable the integration of various attributes

of the sensory input, resulting in a unified percept. While activity evoked to external sensory

input has been instantly recognized to reflect stimulus processing, ongoing neural activity

preceding stimulus onset is often considered as background noise and irrelevant to subsequent

stimulus processing. Our study seeks to advance the idea that ongoing neural oscillations are

critical indices of the brain state and, as such, play an important role in stimulus-evoked

cognitive processing. Applying advanced multivariate spectral analysis to invasive and non-

invasive electrophysiological recordings from animals and humans, we address the mechanisms

and the functional roles of synchronized prestimulus oscillatory activity in processes such as

anticipatory attention, visual processing and encoding of memory.

First, applying computational techniques to invasive local field potential recordings from

primates we expound the mechanisms leading to zero-lag synchrony across widely separated

cortical areas, a phenomenon critically implicated in effective neuronal communication and thus

higher-order perceptual processes. Second, we assessed the functional role of prestimulus theta

band oscillation in memory related stimulus encoding in the human cortex using invasive









electrocorticogram. Third, a physiological model linking prestimulus oscillations and stimulus-

evoked response is proposed and is tested on noninvasive studies of human visual processing.

Finally, the neural control of anticipatory attention and stimulus selection using

electroencephalogram is investigated. The objective here is to expound the signals and the neural

mechanisms that mediate these top-down processes that enable selective sensory integration

during attentive behavior in humans.

In conclusion, our findings suggest that effective sensory-perceptual processing and thus

cognitive and behavioral performance to an impending stimulus, depends on the excitability state

of the brain defined by these neural oscillations at its onset. Further, it is shown that the

prestimulus activity is not merely a random fluctuation in excitability but one that is actively

controlled in a goal-directed manner through neural synchronization across widespread

functionally relevant cortical networks. Thus, this study bolsters the functional role of

prestimulus ongoing activity in subsequent sensory-perceptual and higher-order cognitive

processing.









CHAPTER 1
INTRODUCTION

The ubiquitous nature of oscillatory neural activity in the human and animal brain is well

established. In 1989, Gray, Singer and colleagues (1989), showed that neurons with distinct

receptive fields in the primary visual cortex when jointly stimulated with one 'perceptual' object,

produced synchronized oscillatory responses. On the other hand, when stimulated by two distinct

perceptual objects in their respective receptive fields, the firing rates remained unaltered, and the

synchrony was abolished. Ever since this early study, it has been shown consistently that all

cognitive processes involve transient integration of several widely distributed cortical areas

(Singer and Gray, 1995; Malsburg, 1995; Roskies, 1999; von Stein et al., 2000). These

synchronous oscillations, resulting from dynamic interaction between distant yet functionally

relevant cortical regions, have been suggested as a putative mechanism for the large-scale

integration of sensory inputs, known as the 'binding' problem (Eckhorn et al., 1988; Sarnthein et

al., 1998; Weiss and Rappelsberger, 2000; Engel et al., 2001; Varela et al., 2001; Fries, 2005;

Jones and Wilson, 2005). Numerous studies have recently reported synchrony with zero time lag

across widely separated cortical regions, and have interpreted the observation as precise timing

of neural activation within sub-millisecond range across functionally relevant cortical regions

facilitating binding and functional integration (Roelfsema et al., 1997; Chawla et al., 2001).

However the mechanisms leading to such zero-lag synchrony across spatially distant areas are

yet unclear.

The field potential of neuronal ensembles as measured by invasive macro-electrodes or by

non-invasive EEG / MEG, exhibits a broad range of oscillatory activity, usually described in

terms of frequency bands, delta (1-3 Hz), theta (4-9 Hz), alpha (8-12 Hz), beta (13-25 Hz) and

gamma (25-90 Hz). The oscillatory neural activity both in humans and animals exhibit task









related modulation in their strength and synchrony. Theta and alpha range oscillations have been

long shown to reflect memory processes (Kahana et al., 1999; Klimesch, 1999; Weiss and

Rappelsberger, 2000; Raghavachari et al., 2001; Jensen and Tesche, 2002; Sederberg et al.,

2003; Summerfield and Mangels, 2005; Vertes, 2005). Alpha oscillations have been associated

with alertness, arousal and attentional demands (Ray and Cole, 1985; LaBerge, 1997; Foxe et al.,

1998; Bastiaansen and Brunia, 2001; Babiloni et al., 2006; Keil et al., 2006; Kelly et al., 2006;

Thut et al., 2006; Fan et al., 2007; Klimesch et al., 2007). The beta rhythm has been shown to

reduce in strength preceding movement and subsequently increase in strength after the

movement is complete (Pfurtscheller and da Silva, 1999). The gamma range oscillations are

often found in early sensory areas and have been suggested to be involved in the binding of

sensory information (Miltner et al., 1999; Muller et al., 2000; von Stein and Sarnthein, 2000; Fell

et al., 2003; Herrmann et al., 2004; Palva et al., 2005; Womelsdorf et al., 2006; Hermer-Vazquez

et al., 2007; Landau et al., 2007; Doesburg et al., 2008).

Increasing number of studies suggest that neural oscillations play significant roles in

cognitive functions such as attention (Worden et al., 2000; Fries et al., 2001; Womelsdorf and

Fries, 2007; Doesburg et al., 2008), anticipation (Foxe et al., 1998; Worden et al., 2000;

Bastiaansen and Brunia, 2001; Liang et al., 2002), associative learning (Miltner et al., 1999),

memory (Burgess and Gruzelier, 1997; Gevins et al., 1997; Klimesch, 1999; Jensen and Tesche,

2002; Herrmann et al., 2004; Jones and Wilson, 2005), motor functions (Roelfsema et al., 1997;

Farmer, 1998; Brovelli et al., 2004; Hermer-Vazquez et al., 2007) and perception (Eckhom et al.,

1988; Rodriguez et al., 1999; Babiloni et al., 2006; Melloni et al., 2007). In particular, the

neuronal oscillatory activity is considered the substrate for communication between functionally

relevant cortical areas (Roelfsema et al., 1997; Miltner et al., 1999; Fries, 2005).









Although neural oscillations have been studied extensively, emphasis has traditionally

been on stimulus-evoked responses and stimulus-related changes in oscillatory activity while the

ubiquitous spontaneous oscillations were regarded as 'noise' and irrelevant for subsequent

stimulus related processing. A landmark study by Arieli et al., (1996) using optical recordings

showed for the first time that the state of the brain at the onset of an external stimulus predicts

the effectiveness of the ensuing stimulus processing, providing the early impetus for a systematic

investigation of the role of spontaneous brain activity in stimulus processing. Recent work has

begun to explore this issue. Computational modeling as well as in-vitro and in-vivo single unit

recordings in animal preparations have explored mechanisms by which background synaptic

activity can influence the responsiveness of cortical neurons to afferent input (Ho and Destexhe,

2000; Chance et al., 2002; McCormick et al., 2003; Wolfart et al., 2005; Haider et al., 2007).

How these mechanisms manifest in humans is not yet clearly established.

This dissertation explores the role of prestimulus ongoing oscillations as well as their

relationship to subsequent stimulus processing and higher-order cognitive processes along four

specific aims.

Aim 1: To identify factors contributing to near zero time lags in synchronous oscillatory

cortical networks. Given that the conduction delay between two brain areas is only a small

fraction of the oscillation cycle, a near-zero phase-lag relation could stem either from reciprocal

communication between the two areas (bidirectional interaction) or from the two areas being

readied to communicate by a third set of areas (common input). The testing of these possibilities

has not been carried out empirically. The main reason is that the commonly used methods such

as cross correlation and coherence lack the ability to decompose neural interactions into their

constituent components. We expound the mechanisms leading to zero-lag synchrony across









widely separated cortical areas by introducing Geweke's time series decomposition theorem into

the analysis of multivariate neural data. Employing Geweke's decomposition theorem, we

decompose the neural synchrony between two areas (A and B) into its constituent terms (A-4B),

(B-4A) describing the direction of neural interaction and a term representing the instantaneous

correlation (instantaneous common input). We then characterized the influence of each of the

constituent interaction terms on the phase-lag to address the factors contributing to near-zero lags

across widely separated cortical areas using simulation examples and local field potential data

recorded from behaving monkeys performing a visuomotor pattern discrimination task.

Aim 2: To investigate the functional role of prestimulus theta band oscillation in memory-

related sensory encoding. It is well known that when instructed to memorize a set of items, we

typically are capable of recollecting only a fraction of the items previously presented. Toward

addressing the question of why some words are remembered better while others are not, it was

shown that the level of physiological responses evoked by the visual word presentation in the

prefrontal cortex (PFC) and the medial temporal lobe (MTL) is predictive of successful

subsequent recall and can thus can be considered an indicator of encoding efficacy. However, the

question that still remains to be addressed is that, why are some words encoded better while

others are not? We hypothesize that the efficacy of memory encoding in humans depends on the

level of excitability over the frontal-medial temporal memory network and that the strength of

frontal-medial temporal interaction immediately prior to stimulus presentation, possibly

reflecting the strength of top-down attentional control originating in the frontal cortex (Furey et

al., 2000; Kimura, 2000; Hasselmo and McGaughy, 2004), is positively correlated with the

efficacy of stimulus encoding and the likelihood of its subsequent recall. Considering that

noninvasive electroencephalogram (EEG) lacks the spatial resolution and the ability to sample









areas proximal to the medial temporal lobe, the hypotheses were instead tested by recording

electrocorticogram (ECoG) from multiple implanted girds in five patients undergoing presurgical

evaluation for intractable epilepsy who performed a subsequent memory task. In this study we

characterize how ongoing oscillatory brain activities in the fronto-temporal network immediately

preceding stimulus presentation affect memory encoding and thus later recall.

Aim 3: To investigate the functional role of ongoing oscillations during anticipatory

attention and their relation to evoked responses in a spatial visual attention task. While numerous

EEG/MEG studies have considered the role of prestimulus oscillations in the genesis of sensory

evoked potentials, no consensus has emerged, and disparate reports remain unreconciled (Brandt

and Jansen, 1991; Foxe et al., 1998; Marrufo et al., 2001; Makeig et al., 2002; Ergenoglu et al.,

2004; Linkenkaer-Hansen et al., 2004; Sauseng et al., 2005; Babiloni et al., 2006; Mazaheri and

Jensen, 2006; Thut et al., 2006; Hanslmayr et al., 2007a; Hanslmayr et al., 2007b; Paul et al.,

2007; Romei et al., 2008b; Romei et al., 2008a; Van Dijk et al., 2008; Zhang and Ding, in press).

Further, sensory information processing depends to a great extent on to what and where an

individual deploys attention. It is well known from early studies that responses to stimuli at

attended locations and attended sensory modalities are enhanced relative to unattended stimuli. It

has also been shown in recent years that attention modulates baseline ongoing brain activity in

advance of sensory input (Kastner et al., 1999; Dehaene and Changeux, 2005; Bestmann et al.,

2007). However, the mechanisms linking the pre and poststimulus attentional effects remain not

well understood.

Our approach to investigating the relation between prestimulus activity and stimulus

evoked response starts by considering a theoretical model in which synaptic input to a sensory

neuronal ensemble comes in two forms: external stimulus (exogenous) or other brain structures









endogenouss). Supposing that, in the absence of sensory input, the firing rate of the neuronal

ensemble is related to the magnitude of endogenous synaptic fluctuations through a sigmoidal

function, and sensory input has an additive synaptic effect, it follows that the stimulus-evoked

response corresponds to the derivative of the sigmoidal function, referred to as the gain, which is

an inverted-U function of the prestimulus level of background activity. Reformulated in terms of

EEG variables where field oscillations and the early component of event-related potential are

treated as indices of background synaptic activity and stimulus-evoked response, the model

makes a number of predictions, which were then tested by recording EEG data from human

volunteers performing a trial-by-trial cued covert spatial visual attention task. The systematic

model based characterization of the pattern of relationship between the early visual-evoked

responses and the prestimulus oscillatory activity also offers an added benefit in that it may yield

insights into the fundamental question of how cognitive operations such as attention modulate

baseline ongoing neural activity to achieve enhanced stimulus information processing.

Based on the proposed model linking prestimulus oscillatory activity and subsequent

stimulus-evoked response, a mechanism on how changes in the global state of the brain, such as

selective attention, influenced the aforementioned sigmoidal curve was proposed. Through

experimental data and computational modeling we present a mechanism of how attention

modulates baseline ongoing activity over early sensory areas to enable enhanced

sensory-perceptual processing.

Aim 4: To investigate the functional role of ongoing oscillations in top-down control of

human visual cortex during anticipatory spatial visual attention. The theoretical framework

adopted thus far emphasizes the role of ongoing EEG oscillations over early sensory areas, in

this case the visual cortex, and addresses the mechanisms linking the attention induced









prestimulus modulations to stimulus-evoked attentional enhancements. However, such

goal-directed prestimulus modulation of sensory cortical activity is not an inherent property of

the early sensory areas. Instead, it is believed that the higher-order areas such as the

frontal-parietal network facilitates an enhancement of the excitability of the task-relevant and

feature-specific visual areas and perhaps a suppression of activity in task-irrelevant visual areas

in advance of visual stimulation by modulating its baseline activity (Chelazzi et al., 1993; Luck

et al., 1997; Hillyard et al., 1998; Kastner et al., 1999; Hopfinger et al., 2000; Kastner and

Ungerleider, 2001; Corbetta and Shulman, 2002; Gazzaley et al., 2005; Giesbrecht et al., 2006;

Grent-'T-Jong and Woldorff, 2007; Klimesch et al., 2007). In line with this notion, it has been

identified that deployment of covert spatial visual attention is associated with increased

activation in a network involving dorsal frontal and parietal cortices with concomitant increases

in activation over retinotopic early visual areas (Desimone, 1996; Gitelman et al., 1999; Kastner

et al., 1999; Hopfinger et al., 2000; Corbetta and Shulman, 2002; Giesbrecht et al., 2003).

While prior neuroimaging and lesion studies of the top-down control of visual attention

have offered insights into the areas in the brain that exert control over stimulus processing, the

neural mechanisms of how this increased excitability over the visual sensory areas is brought

about has not been clearly established. What is the neural signal that mediates the top-down

biasing influence? Is the increased excitability achieved by increasing the excitatory drive on

local pyramidal neurons (excitation) or by decreasing excitatory drive on local inhibitory

interneurons disinhibitionn)? Is the top-down influence transient or sustained? These questions

are addressed by applying time-frequency analysis of power, coherence and Granger causality to

scalp EEG recordings from healthy volunteers performing a spatial visual attention task.









CHAPTER 2
GENERAL METHODS

The primary goals of this dissertation involve the investigation of prestimulus ongoing

activity and its relation to stimulus evoked responses and hence cognitive performance. Neural

ensemble activity exhibit a wide range of oscillations and the role of these oscillations in

cognitive information processing have been increasingly studied in recent years. Physiologically,

cortical field oscillations recorded by scalp EEG/MEG reflect the rhythmic fluctuations in the

correlated background synaptic activity and provide an index of the excitability of the underlying

neural population (Dasilva, 1991). EEG/MEG signals consist of several simultaneous

oscillations, which have traditionally been subdivided into frequency bands such as delta

(1-3 Hz), theta (4-9 Hz), alpha (8-12 Hz), beta (about 13-25 Hz) and gamma (25-90 Hz) each of

which are distinctly related to cognitive information processing and hence behavior. The

oscillatory nature of the signal necessitates the need for a spectral domain approach to study this

problem. Further, it is well agreed upon that complex cognitive processes involve complex

interaction between disparate regions of the brain. Synchrony (functional connectivity) and

Granger causality (effective connectivity) analysis have emerged as principled approach of

identifying and quantifying inter-areal interaction and infer causal influence among distributed

networks (Eckhorn et al., 1988; Schack et al., 2000; von Stein and Samthein, 2000; Baccala and

Sameshima, 2001; Kaminski et al., 2001; Varela et al., 2001; Liang et al., 2002; Brovelli et al.,

2004; Tallon-Baudry et al., 2004; Palva et al., 2005; Seth, 2005; Ding et al., 2006; Wang et al.,

2007b; Bollimunta et al., 2008; Cadotte et al., 2008; Hermer-Vazquez, 2008; Rajagovindan and

Ding, 2008b; Zhang et al., 2008a; Zhang et al., 2008b; Cadotte et al., 2009). Granger causality is

a statistical approach that decomposes neural synchrony into causal directional interactions,

which provides a non-invasive alternative to identify the direction of neural interaction. Below,









some of the spectral estimation techniques employed in this study to quantify neural activity and

also the mathematical framework of Granger causality with special emphasis on its spectral

decomposition is reviewed. Practical issues concerning how to estimate such measures from time

series data are then discussed. The description in the following sections follow that of (Granger,

1969; Geweke, 1982; Ding et al., 2000; Chen et al., 2006; Ding et al., 2006; Wang et al., 2007b;

Rajagovindan and Ding, 2008b)

2.1 Parametric Multivariate Spectral Estimation

Multivariate autoregressive (MVAR) modeling is a parametric spectral analysis method in

which time series models are estimated from datasets having either one long realization or a

number of realizations. In the latter case, the fundamental assumption of this algorithm is that the

short-window time series can be treated as realizations of an underlying stationary stochastic

process. As cognitive information processing involves transient changes in neural activity,

MVAR can be used to investigate time series in short window size (<100 ms). The following is

the procedure of MVAR (Ding et al., 2000).

Let X(t) = [X(1, t), X(2, t),..., X(m, t)]' be a m-dimensional jointly stationary random

process. T denotes matrix transposition. In the case of neural recordings, m refers to the total

number of recording channels to be analyzed. Assuming X(t)is a zero mean stationary process,

X(t) can be modeled by the following pth order autoregressive equations:

X(t) + A(1)X(t -1)+ A(2)X(t- 2) + + A(p)X(t p) = E(t) (2-1)

where p is the model order, and A(i) are the unknown m xm coefficient matrices and E(t) is

the uncorrelated noise term with covariance matrix E Multiplying X' (t- k) to Eq. 2-1 and

taking expectation on both sides we arrive at the Yule-Walker equations.

R(-k)+ A(1)R(-k +1)+ + A(p)R(-k + p)= 0 (2-2)









where R(j) is covariance matrices of X(t)XT (t + j) with lag j. Also note that

< E(t).XT (t -k) >= 0, since E(t) is an uncorrelated noise process. The unbiased estimator of the

covariance matrix in Eq. 2-2 for a single realization of the Xprocess is given by

1 Nn
R(n) = x(i)XT (i + n)
N-n ,1,

In the case of multiple realizations of the process as is usually the case, the covariance matrix is

computed for each individual realization and then averaged across all the realizations to obtain

the most robust estimate. In the case of short window of data, i.e. short sample length where Nis

small, it is apparent that the estimation from a single realization is poor. However, with

increasing number of realizations this problem can be offset for short sample length time series.

In the limiting case with number of realizations approaching infinity, the sample length of each

realization may be as short as the model order plus one (N=p+l).

Coefficient matrices A(i) and covariance matrix I of noise term E(t) is obtained by

solving Eq. 2-2 through the Levinson, Wiggins and Robinson (LWR) algorithm. The noise

covariance matrix I is obtained as part of the LWR algorithm. One may also obtain I as


S= R(0)+ A(i)R(i)


It is instructive to note that the Eq. 2-2 contain a total ofpm2 unknown model coefficients to be

estimated from the same number of simultaneous linear equations.

Although the primary objective is to fit the best model to the data, which is to minimize the

residual noise variance, this constraint alone may result in over-parameterization. The reason for

this is because the variance of the residual term decreases monotonically with incorporating

more and more past values of the process. Over-parameterization leads to inaccuracies in the

estimation of the model coefficient besides the obvious increase in the computational









complexity. Thus a penalization scheme to avoid over-parameterization is employed. Criterions

that incorporate both minimization of the variance of the residual term and penalize excessive

coefficients are the Akiake Information Criterion (AIC) and Bayesian Information Criterion

(BIC) among several other similar criterions. The AIC is defined as


AIC(p)= -2log[det(X)]+ 2mp
Total

where N0otat is the total number of data points from all the trials. The first term on the right hand

side accounts for the variance minimization and the second term correspond to the cost

associated with increasing parameters. Plotted as a function ofp, the optimum model order

corresponds to the minimum of this function. For typical neurobiological data N0otal is very large

and hence for practical ranges ofp, the AIC function does not achieve a minimum due to

inadequate penalization as the second term vanishes with large values of Nottl. An alternative

criterion is the Bayesian Information Criterion (BIC), which is defined as


BIC(p)= -2log[det(E)]+ 2m2plogN0
Total

This criterion can compensate for the large number of data points and may perform better in

neuroscientific applications. A final step and critical step necessary for determining whether the

autoregressive time series model is suited for a given data set, is to check whether the residual

term is white. Here the residual term is obtained by computing the difference between the

model's predicted values and the actually measured values.

Once an autoregressive model is adequately estimated, it becomes the basis for both time

domain and spectral domain causality analysis. Spectral features are derived from MVAR

models after acquiring A(i) and I estimates. Taking the Fourier transform, Eq. 2-1 can be

rewritten in the spectral domain as









A(f)X(f) = E(f),


(m 2
Defining H(f)= A(i)e- ,z2f ,then X(f)= H(f)E(f)
_=0

where H(f) is the transfer function. The spectral matrix can be readily derived from


S(f) = lim -E[X(f)X*(f)]= H(f) H*(f)
N-o N

where means both transpose and complex conjugate.

Spectral power is contained in the diagonal terms of the spectral matrix and the off-

diagonal terms represent the cross spectra. Coherence spectra between two random process

X(i,t) and X(j,t) is defined as:


( S,, (f)12
S,, (f)Sii (f)

If the coherence value is equal to 1 or 0, the two processes are maximally interdependent or

independent, respectively. Once the transfer function, the noise covariance and the spectral

matrix are estimated, Granger causality may be derived according to the procedures outlined in

the following sections.

2.2 Granger Causality Analysis

2.2.1 Time Domain Formulation

Consider two jointly stationary stochastic processes X, and Y;, each having an

autoregressive representation given by


Xt = al X,, + ,, var(,) = x
(2-3)
Yt= dl Yt-, + q7t, var(t) =
J=1









Jointly, they can be represented by the following bivariate autoregressive model


X, = a2,X,- + Cb2Yt- + 2t
J=1 J=1
(2-4)
Y c2 X, + d2 t- + 2t,
J=1 -=1

where, E2t and 72t are uncorrelated over time, and their contemporaneous covariance matrix is


Syx r yy


If bi is not uniformly zero for j =1,2,...,00 then Y is said to have a causal influence on

X,. Likewise, X, is said to have a causal influence on Y, if c, is not uniformly zero for


j = 1, 2,...,00 The total interdependence between X, and Y, is defined as (Gelfand and Yaglom,

1959)

xx
Fx,r = In


where *| denotes the determinant of the enclosed matrix. It can be easily verified that according

to this definition Fx,, = 0 when the two time series are independent, and Fx,, > 0 when they are

not. Considering Eq. 2-3, 2-4, 2x represents the prediction error of X, based on its past values

and xx represents the prediction error, when predicting the present value of X, based on past

values of X andY,. If xx is less than x, then Y, is said to have causal influence on X,

quantified by (Granger, 1969; Geweke, 1982)


Fyx = ln
Y,









It may be noticed that when there is no causal influence from Yto X, Fr,, = 0 and F,, > 0

otherwise. Similarly the causal drive from Xto Y may be defined as (Granger, 1969; Geweke,

1982)


Fxy =In '
yy

If = YX 0, indicating that the noise terms 2t, and 72, are correlated at time t, then the

interdependence between X, and Y, has another contributor that is not explained by the

interaction between X, and Y. This component possibly representing common input from a third

system, will be referred to as the instantaneous causality, defined by (Geweke, 1982)


Fx.Y = In


Notice that when -, = Y- = 0 the instantaneous causality is zero and greater than zero

otherwise.

It follows from definition that the total interdependence is the sum of mutual interaction

between Xand Y, and a component exogenous to the system given by (Geweke, 1982)

F = F +F + F

2.2.2 Frequency Domain Formulation

To derive Geweke's formulation of Granger causality in the frequency domain it is

convenient to repose the autoregressive equations in terms of lag operator. The lag operator L is

defined asLX, = X, Rewriting Eq. 2-1in terms of lag operator

a2(L) b2(L)) Xt>f'X (2-5)
c, (L) d2(L)) Yj, 2

where, a2 (0) = 1, b (0) = 0, c2 (0) = 0, d (0) = 1 Fourier transforming both sides of Eq. 2-5








a, () b2 (o) X(c) A(Xo) X() Ex (o)
C2() d2(0)J) (o)J Y( E (0[))

where,
a2()=l a2je -J, b2(o)=- b2-e -", c2(W) 1 e-"J, d2( = o -) dJe
J1 J]1 J]1 J=1
Further simplifying

X(O)=( H, (o) H, (m) E (
rY(o)J) H o(0) H (m)) J( (m))

where, the transfer function H(o))= A 1 (o) and

1 1
H (o) = -d2 (), H, (o) = b2())
A Al
1 1
H, (m) = C2 (), H, ()) -a a)

Performing proper ensemble averaging, the spectral matrix can be derived as


S( S)= = y '(o) =H(o)EH'(o)
s(O) S, () S, (0)) '

where denotes complex conjugate and matrix transpose.

The diagonal terms of the spectral matrix denote the auto spectra (power spectrum) and the

off-diagonal terms denote the cross spectra. If X and Y are independent, the off-diagonal

elements are zero. The spectral domain representation of the total interdependence between X,

and Y, at frequency co (=27tf) is given by (Geweke, 1982)


/f, (0) = In S= -In(1- C(-Io))
S(O))










where C(O) = SY(-) is the coherence function. The phase-lag between X, andY, at a
.Im 0Sx)() 0)

given frequency is given by tan .1 Sx()
Re{^S(})

Further, IS(w)l = S)(0o))S (o)- S,(o)S,(o) and S, (o) = S~(w) Notice that when X

and Y are independent, IS(w)l reduces to the product of the auto spectra, hence it can be verified

that the total inter dependence is zero. Expanding the entries of the spectral matrix,

S, (o) = H, (o)CHL ( 2o) + 2Y, Re(Hx (m)H (o0)) + H Y, H*3 (o) (2-6)

consider the case when 0,y = 0 that is when there is no instantaneous causality. Then, Sxx(0)

reduces to the sum of the first and the third term in the right hand side of Eq. 2-6. Intuitively, this

can be understood as decomposing the auto spectrum into an intrinsic part and a component

representing the causal drive from the other time series. However if # 0 this form of

interpretation may not be obvious. Consider a transformation introduced by Geweke (Geweke,

1982) referred to as normalization, that removes the effects of the cross terms and enables the

identification of a similar intrinsic power terms and a causal power term possible. Left

multiplying on both sides of Eq. 2-5 by the transformation matrix P given by (Geweke, 1982)

1 0O
P I


we obtain

a,(mo) b3(m)J Y(0)) X(m ) EY (o)
c, () d() d ) Y ()) Y (mO) ET (0))










c3(co)=c2(co)- l-a2(o), d3(co)=d2(co)--b2(o), Ey(o))E (o))- E (o)
2XX XX, XX


The new transfer function H(wo) is the inverse of the A(co)


ft(w))) Tl,(O))
H_ ? (0) H(O)} R -c,3(0) a (0))b


Also, 4A = A thus we have


HT(o) =H (o) + -H (o), H (c) = H, (o)
xx

Hy( o)= H (co) + H(o), HT, ()o)H, (o)


Note that Ex and Ey are uncorrelated. The variance of the residual term for the normalized

Y2
equation for Y, is ',y = Thus after normalization the auto spectrum of X, can be
Sxx

written as

S,(o)= H,(m)) XH,()+ + H,,yH*(m)

As earlier, the auto spectra can now be interpreted as the sum of intrinsic power and causal

power component of X, due to Y, Based on this formulation we now define the frequency

domain causal influence from Y, to X, at frequency co as (Geweke, 1982; Ding et al., 2006)


f (co) = In S (co (2-7)
Hx(co)z~xx(co)

Notice that this definition of causal influence is expressed in terms of the intrinsic power

rather than the causal power. It is expressed in this way so that the causal influence is

zero when the causal power is zero (i.e., the intrinsic power equals the total power), and









increases as the causal power increases (i.e., the intrinsic power decreases).

f1
Now by taking the transformation matrix as Zy and performing the same
0 1

operations as earlier, we get the influence of X, to Y, given by


fy (co) = In Y(co) (2-8)
Hyyy (o)D Hy- (R )


where, Hyy(ow)= H (wo)+ + -HY(o)
yy

Having identified the total interdependence and both the directional influences, the spectral

decomposition of the instantaneous causality can now be readily determined as


f (o) = In ()()( (o )H yY-,y H yy (co)
fr () = In (2-9)
|S(w)|

The above set of variables are related through (Geweke, 1982; Ding et al., 2006)

fx,Y (0) = fxY (a)) + fyx (a) + fx.Y (a))

Under general conditions, the spectral measures derived above are related to the time domain

measures as


Fxy f ifxrdo),
2;r -1 dc

1

Fy x 1- f~x (o))do,
2;








1 )
FX.Y =- ffx.Y (o))do),


The invertible nature of Geweke's formulation between the time domain and frequency

domain representation is a desirable characteristic that is not possessed by other current causality

measures. For practical purposes, the same quantities listed above (Eq. 2-7, 2-8, 2-9) with simple

algebraic modification may be expressed as (Brovelli et al., 2004)


zx- H~i(co)
S (0))1




x(o)) = -Sn(l .)




f.(o)) = ln( I
fX., (a)) = In(--------------------------

2.2.3 Mathematical Formulation of Block Granger Causality

Application of Granger causality to the investigation of causal influences between different

brain structures have so far mainly utilized pairwise analysis (Brovelli et al., 2004) in which two

time series are analyzed at a time. However, pairwise analysis may be considered inadequate

given large sets of multivariate neural time series data. A more pragmatic approach for many

purposes is to combine the time series recorded from different brain regions of interest into

blocks, and then to analyze the relations between the blocks. Although it is possible to achieve

blockwise analysis by combining the results of repeated pairwise analyses, a more effective, and

theoretically more elegant, approach is to address the relation between two blocks of time series









directly (Wang et al., 2007b). This blockwise approach is contained in the general

methodological framework developed by Geweke (1982).

Let the block of signals from two recording sites be denoted by Xt andYt with dimensions

k and 1. In the case of pairwise Granger causality, Xt and Y are two one-dimension time series

(k=l, 1=1); while in the case of block Granger causality, Xt and Y are two sets of time series of

dimensions k and 1 typically not equal to 1 and have no overlap. Jointly, they can be represented

by the following multivariate autoregressive model

P P
X, =Y a. X-, + bY-,t + ,
j=1 j=1
SP (2-10)
Yt = c Xt-j + di Yt- + I,
j=l j=l

where st and 7, are uncorrelated over time, and their contemporaneous covariance matrix is





If bi is not uniformly zero for j = 1, 2,..., p, then Yt is said to have a causal influence on Xt.

Likewise, Xt is said to have a causal influence on Yt if c, is not uniformly zero for

j = 1,2,..., p .

The model orderp is determined using either the Akaike Information Criterion (AIC) or

the Bayesian Information Criterion (BIC) which reflects a tradeoff between sufficient spectral

resolution and over-parameterization. Fourier transforming Eq. 2-10 and performing proper

ensemble average, we obtain the spectral matrix


S(w S ()) Sy()) = (c YH*(c)
syL (0))Syy(0)









where denotes complex conjugate and matrix transpose, and H (m) is the transfer function

matrix.

Similar to the procedure described in the previous section, the Granger causality spectrum

of the two blocks of time series, for Xt driving Yt is given by,


Yxx -E 12()
fx (o) = -ln(1- )


which can be interpreted as the proportion of Xt's causal contribution to the power of the Yt

series at frequency co The logarithm is taken to preserve certain favorable statistical properties.

Similarly, the causality spectrum for Yt driving Xt is given by


Y- H ( 2
fX(co) = -ln(1 )


To obtain the magnitude of causality for a particular frequency band, the above spectrum may be

averaged over the desired frequency band.

2.2.4 Interpretation

Statistically, for two simultaneously measured time series, one series can be called causal

to the other if we can better predict the second series by incorporating past knowledge of the first

one (Wiener, 1956). This concept was formalized by (Granger, 1969) in the context of linear

regression models of stochastic processes. Specifically, if the variance of the prediction error for

the second time series at the present time is reduced by including past measurements from the

first time series in the linear regression model, then the first time series can be said to have a

causal (directional or driving) influence on the second time series. Reversing the roles of the two

time series, one repeats the process to address the question of causal influence in the opposite









direction. From this definition it is clear that the flow of time plays an essential role in allowing

inferences to be made about directional causal influences from time series data. Alternatively,

improvement in prediction can also be viewed from the perspective of comparing relative

estimates of conditional probability. Recent work has modeled the relation between multiple

point processes along this view (Okatan et al., 2005; Truccolo et al., 2005). In our analysis, the

EEG data constitute the time series, and Granger causal influence is equated with the direction of

synaptic transmission between neuronal ensembles.

2.2.5 Assessment of Statistical Significance

For coherence and Granger causality spectra, we have adopted a random permutation

approach (Brovelli et al., 2004) to build a baseline null-hypothesis distribution from which

statistical significance can be derived. Consider two channels of recordings with many repeated

trials We can reasonably assume that the data from different trials are approximately

independent of one another. Randomly pairing data for channel 1 with data for channel 2 from a

different trial leads to the creation of a synthetic ensemble of trials for which there is no

interdependence between the two channels based on construction, while preserving the temporal

structure within a channel. Performing such random pairing with many different permutations

will result in a distribution of coherence or causality spectra corresponding to the null hypothesis

of no statistical interdependence. The calculated value for a given statistic from the actual data is

compared with this baseline null hypothesis distribution for the assessment of significance levels.

Bootstrap resampling techniques (Efron and Gong, 1983; Efron, 1985, 2000) were used to

estimate confidence intervals.

2.3 Multitaper Spectral Estimation

Multitaper spectral estimation (Thomson, 1982) refers to a set of non parametric methods

for estimating power spectra, coherences and related spectral quantities using an orthogonal set









of data tapers, in specific, the discrete prolate spheroidal a.k.a Slepian sequences and their

approximate minimum bias sine tapers.

Before proceeding to the mathematical formulation of the Multitaper spectra estimation

techniques it is worth noting the several favorable properties that motivated the use of this

approach from among several other available estimation techniques. The problem of optimum

spectral estimation is confronted with several key issues including that of minimizing the (a) bias

of the estimators (b) variance of the estimators and (c) the spectral leakage among several other

considerations. For further details on each of these considerations the readers may refer to

(Lfeachor and Jarvis; Thomson, 1982; Mitra and Pesaran, 1999; Mitra and Bokil, 2007). For

example consider one of the conventional nonparametric spectral analysis techniques, the

Welch's method. In order to reduce the variance in the estimation, the approach involves

splitting the data into overlapping segments followed by the estimation of the power or the cross

spectrum for each segment and then averaging over all the segments. For data of short length,

such approaches suffer from severe bias in the estimation. In comparison, multitaper approach

does not suffer from these limitations. Both reduction in bias and variance of the estimation is

achieved since averaging the spectral estimates over different tapers enable reduction in the

variance and since each taper is applied to the entire data instead of short segments of the data as

in Welch's method the resulting bias in the estimation is smaller in comparison. Further the

choice of the orthonormal Slepian tapers enables maximum sidelobe suppression hence offering

the least spectral leakage in comparison to all other spectral estimation techniques. Thus

multitaper approach offers the optimum spectral estimation in the face of short sample length

data or short sample length and limited realizations of data as often encountered in

neuroscientific applications as is the case in this study. It is apparent from the aims set out to be









addressed that it is crucial to be able to achieve single-trial estimation of the spectra. In light of

this crucial requirement, this study utilizes the full advantage of this technique to enable

single-trial estimation of spectral features during short prestimulus time periods which otherwise

would not have been possible to estimate with sufficient accuracy using either the parametric

framework based on autoregressive models presented earlier or through other common non

parametric approaches.

A brief exposition of the mathematical formulation of the multitaper spectral estimation

technique for univariate case is as follows. The extension to multivariate case is straightforward.

Consider time series x(t) (t = 1,2...N) a zero mean second order stationary random process and

let w (t) denote the kth Slepian taper (window sequence) and k = 1,2...K.

Step 1: Let us define the time-bandwidth product given by C = NW, where Nis the sample

length and Wthe desired spectral bandwidth. If C is too small, the estimate will be unstable and

may not have sufficient dynamic range and if C is too large, the estimate may not have adequate

frequency resolution. For a given choice of C there are K = 2C data taper sequences. Since the

energy concentration of higher order tapers is poorer than the low order tapers, in practice it is

common to choose K = 2C- or K = 2C-2.

Step 2: The Slepian tapers having the maximum energy concentration in the bandwidth W

is determined by solving the following eigenvalue problem

N1 sin[2KW(t j)]
Ak(t) = '. (j) fort= 1, 2...N.
j0 ;(t j)

Step 3: The eigen coefficients of this problem reduces to the Fourier transform of the data

sequence multiplied by the tapers (similar to periodogram). The windowed Fourier transform of

the data x(t) is given by









N-1
Ak = Xk(o) = x(t0k (t)e-
t=0

Step 4: The spectrum then estimated as the average of the spectra obtained through each

taper is given by

1 K-1
S( -) = Xk()Xk )
K k=0

In the case of multiple realizations, the above estimate is further averaged across all realizations.

For further treatment of this topic refer (Thomson, 1982; Thomson, 2007).









CHAPTER 3
DECOMPOSING NEURAL SYNCHRONY: TOWARD AN EXPLANATION FOR
NEAR-ZERO PHASE-LAG IN CORTICAL OSCILLATORY NETWORKS

3.1 Introduction

Cortical information processing involves the coordinated activity among many distinct

regions of the brain. Statistically, this coordinated activity manifests as correlated or

synchronized co-variations in the recorded multivariate data. Early studies in animals have

shown that stimulus-evoked short-range synchrony between neurons in the primary visual area

subserves perceptual binding of sensory information (Singer and Gray, 1995; Tallon-Baudry et

al., 2004). Simultaneous action potentials fired by lower order neurons (Malsburg, 1995;

Roskies, 1999) provide an effective drive on higher order neurons, the activations of which

enable pattern discrimination and conscious awareness. In humans, similar observations have

been made (Slotnick et al., 2002) where highly synchronized EEG activity occurs in response to

stimulus input. In all these cases, a near-zero phase-lag relation between different data series is

observed, reflecting the millisecond or even sub-millisecond precision required for feature

integration (Singer and Gray, 1995; Munk et al., 1996; Roelfsema et al., 1997; Witham et al.,

2007).

Increasingly, long-range synchronization of oscillatory field activity with near-zero phase-

lag, often in the absence of stimulus input, has been reported (Bressler et al., 1993; Roelfsema et

al., 1997; von Stein et al., 2000; von Stein and Sarnthein, 2000; Brovelli et al., 2004). Depending

on the task, the strength of the synchronization can influence the efficiency of both sensory and

motor processing (von Stein et al., 2000; Engel et al., 2001; Liang et al., 2002; Womelsdorf et

al., 2006), suggesting that it has a functional role. To date, however, an explanation of the

observed near-zero phase-lag relation in these large-scale networks has not been forthcoming.

Recent work have considered this issue from the perspective of neuronal communication.









Physiologically, field oscillations provide an index of the excitability level of a neuronal

ensemble (Lakatos et al., 2007; Lakatos et al., 2008). During the excitable phase of the

oscillation cycle, presynaptic neurons are more likely to fire action potentials, whereas for the

postsynaptic neuron, action potentials received during the excitable phase are more effective at

integrating the synaptic input and triggering a response (Volgushev et al., 1998; Fries, 2005;

Lakatos et al., 2007). This suggests that long-range synchrony could serve as a gating

mechanism of information flow in cortical circuits (Salinas and Sejnowski, 2001; Averbeck and

Lee, 2004; Fries, 2005; Siapas et al., 2005). Given that the conduction delay between two brain

areas is only a small fraction of the oscillation cycle, a near-zero phase-lag relation could stem

either from reciprocal communication between the two areas (bidirectional interaction) or from

the two areas being readied to communicate by a third set of areas (common input).

Mathematically, it is intuitively clear that a positively correlated common input, if strong

enough, can drive the two areas into near-zero phase-lag synchrony. Alternatively, a recent

computational model based on the anatomical connectivity pattern in the visual system proposes

that the emergence of near-zero phase-lag synchrony can also result from two or more cortical

areas experiencing bidirectional interaction (Chawla et al., 2001). The testing of these

possibilities has not been carried out. The main reason is that the commonly used methods such

as cross correlation and coherence lack the ability to decompose neural interactions into their

constituent components.

In this paper we attempt to address this problem by introducing Geweke's time series

decomposition theorem into the analysis of multivariate neural data. Let the two brain areas be

denoted by A and B. The interaction between these two areas may be mediated by A influencing

B (A-4B), B influencing A (B-4A), and/or A and B both receiving a common input. Geweke's









theorem states that the total interdependence (synchrony) between two stochastic processes from

A and B can be expressed as the sum of the three components: (A-B)+(B-4A)+(common

input). Here the arrow is understood in the sense of Granger causality and the common input is

represented by the instantaneous causality (Geweke, 1982; Chen et al., 2006; Ding et al., 2006).

In this framework it is hypothesized that bidirectional interaction or common input or a

combination of both contributes to the establishment of near-zero phase-lag synchrony. In

particular, when the interaction is clearly unidirectional (e.g. A-4B equals zero but B-1A does

not), the phenomenon of near-zero phase-lag can only occur as a result of strong positively

correlated common input, arising exogenously to A and B. We tested these ideas by analyzing

local field potential data recorded from behaving monkeys performing a visuomotor pattern

discrimination task.

3.2 Materials and Methods

3.2.1 Simulation

3.2.1.1 Setup

Auto-regressive model of the form in Eq. 3-1 was used to generate all the simulated time

series. Two representative types of interaction pattern, namely (a) unidirectional interaction with

positively correlated instantaneous common input and (b) bidirectional interaction, were

considered. The phase-lag as a function of appropriate model parameters for both cases was

studied.

3.2.1.2 Positively correlated common input

A bivariate AR(3) process [p = 3 in Eq. 3-1] in which X drives Y was used. The

coefficients of the model were al = 0.4428, a2 = -0.5134, a3 = 0, di = 0.506, d2 = -0.6703, d3 = 0,

bl = b2= b3 = 0, ci = c2 = 0, c3=0.1, and Exx=yy= 1. The cross terms in the noise covariance

matrix, yx=jxy, reflecting the strength of positively correlated common input, was









systematically varied. The parameter choice above enabled the model to oscillate at 40 Hz for

which the phase-lag was computed. The dataset consisted of 100 epochs of 200 sample points

each. The sampling rate was assumed to be 200 Hz.

3.2.1.3 Bidirectional interaction

A bivariate AR(4) process [p = 4 in Eq. 3-1] was used. The coefficients of the model were

al = 0.9, a2 = -0.5, a3 = 0, a4 = 0, d = 0.8, d2 = -0.5, d3 = d4=0, ,x= yy = 1, and xy= Zyx = 0.

The coefficients of the interaction terms b1,2,3,4 and 01,2,3,4 were varied in tandem to achieve

simultaneous increase in the strength of both feed-forward and feed-back interaction. The model

exhibited narrow frequency band oscillations with a frequency peak at 32 Hz for which the

phase-lag was computed. The dataset consisted of 100 epochs of 200 sample points each. The

sampling rate was assumed to be 200 Hz.

3.2.2 Experiment

3.2.2.1 Behavioral paradigm

Two monkeys (GE and LU) were trained to perform a GO/NO-GO visual pattern

discrimination task in the Laboratory of Neuropsychology at the National Institute of Mental

Health (Bressler et al., 1993; Ledberg et al., 2007). Animal care was in accordance with

institutional guidelines at the time. The monkey initiated each trial by depressing a hand lever

and maintained its depression while anticipating the onset of a visual stimulus. Four squares

arranged in either a line (left-slanting and right-slanting) or a diamond (left-slanting and right-

slanting) shaped formation appeared on a visual display after a random time interval triggered by

the lever depression. The monkey made either a GO (lever release) or a NO-GO (maintaining

lever depression) response upon discriminating the input pattern. For the GO trials, the time

between stimulus onset and the lever release is defined as the response time (RT). The

experiment was conducted in sessions of approximately 1000 trials.









3.2.2.2 Data acquisition

Local field potentials (LFPs) were recorded with bipolar teflon-coated platinum

microelectrodes (51-pm diameter and 2.5-mm tip separation) from up to 15 distributed sites

located in the hemisphere contralateral to the dominant hand (right hemisphere in monkey GE

and left hemisphere in monkey LU). The data collection period started 90 ms before the stimulus

onset and ended approximately 500 ms after (Bressler et al., 1993). LFPs were amplified by

Grass P511J amplifiers (-6 dB at 1 and 100 Hz, 6 dB/octave falloff) and digitized at 200 Hz. As

this study is concerned with the phase between two signals, the bipolar derivation carries certain

arbitrariness. Reversing the order of the two subtracting recording leads can change the phase

from 0 to 7t or vice versa. This can affect the formulation of the hypothesis to be tested. See

Results and Discussion sections for more details.

3.2.2.3 Data set selection

Previous analysis of the same experiment (Brovelli et al., 2004) has identified a coherent

beta (14 to 30 Hz) oscillatory network in the sensorimotor cortex involving both pre- and post-

central sites during the prestimulus time period. For this work the three recording sites that are

common to both monkeys were selected for further analysis: primary somatosensory area (S1),

primary motor area (Ml) area, and posterior parietal area 7b. Trials contaminated with artifacts

or associated with incorrect behavioral responses were rejected. To achieve a sufficient number

of trials, different sessions having similar RT distributions were combined to yield a data set of

approximately 2400 and 1400 trials for monkeys GE and LU, respectively. The prestimulus time

interval from -90 ms to 20 ms was considered.

3.2.2.4 Time series decomposition

Let the LFP data from two recording sites be denoted by X, and Y,. Jointly, they can be

represented by the following bivariate autoregressive model









P P
X, = a X, + b Y,- + c,
X =1 Z=1
Sp (3-1)
Yt = cX,+Zd +
J=l J=l

where s, and 7, are uncorrelated over time, and their contemporaneous covariance matrix is






If b is not uniformly zero for j = 1,2,..., p, then Yjis said to have a causal influence on

X,. Likewise, X, is said to have a causal influence on Y, if c, is not uniformly zero for

j = 1,2,..., p. If = Yyx # 0, indicating that the noise terms E, and r, are correlated at time t,

then the interdependence between X, and Y, has another contributor that is not explained by the

interaction between X, and Y,. This contributor, possibly representing influences exogenous to

the (X,Y) system such as a common input from a third system, will be referred to as the

instantaneous causality (Geweke, 1982; Ding et al., 2006). It can be shown that the above set of

variables are related through (see section 2.2.2)


fxy()) = fxy()) + fy3x ()) + fx.y (W) (3-2)

Intuitively, the decomposition in Eq. 3-2 means that the total amount of statistical

synchrony between two LFP signals is the sum of their causal drives on one another and a

common input component. The expression in Eq. 3-2 can be integrated over the entire frequency

domain to yield the time-domain counterpart. While the frequency-domain formulation is

convenient for estimation, the above decomposition is more readily interpretable in the time-

domain.









3.2.2.5 Data analysis protocol

For each monkey there are 3 distinct pairs of recording sites: (Ml, Si), (Ml, 7b) and (S1,

7b). For each pair, previous work (Brovelli et al., 2004) has identified a prominent coherence

peak in the beta frequency range (14 to 30 Hz). Except for (Ml, S1) in monkey LU the

coherence in other five channel pairs is concentrated in the beta frequency range. These five

channel pairs are further analyzed as the time-domain quantities are equivalent to that in the beta

range. The relative phase, referred to as the phase-lag in this study, is well-defined for the peak

frequency. The dependence of this phase-lag on the factors in Eq. 3-2 was investigated by

carrying out the following analysis. (1) The phase-lag in the beta frequency range was estimated

for each single trial by Fast Fourier Transform (FFT) and sorted according to its value. The

sorted trials were grouped into subensembles with 30% overlap, resulting in 33 and 19 groups in

monkeys GE and LU, respectively. Each subensemble contained 100 trials. (2) For each

subensemble the ensemble mean was estimated and removed from the individual trials within the

subensemble. This is to ensure that the data may be treated as coming from a zero-mean

stochastic process. An AR model of orderp=9 was fit to the mean-removed data. (3) For each

subensemble, power, coherence, causality spectra as well as phase-lag at the peak frequency

were derived from the AR model (Chen et al., 2006; Ding et al., 2006). (4) Spearman rank

correlation (SRCC) and Spearman rank partial correlation (Macklin, 1982) (SRPCC) were

computed between the instantaneous causality measure and the estimated phase-lag to assess the

prediction that the two variables are negatively correlated. To remove the effect of directional

influences a partial correlation analysis was performed. Significance was determined through one

tail t-test with p<.05. (5) For channel pairs determined to have bidirectional interactions,

Spearman rank correlation (SRCC) and Spearman rank partial correlation (Macklin, 1982)









(SRPCC) were computed between the magnitude of the sum of the two directional influences

and the estimated phase-lag to assess whether they exhibit negative correlation.

Pairwise Granger analysis has been adopted since phase-lag is a bivariate phenomenon.

However, evaluating causal interaction patterns in a bivariate framework may ignore potential

influences from other areas that may or may not be sampled. In order to assess more reliable

causal interaction patterns, a multivariate approach applied to all the relevant simultaneously

sampled data may be resorted to and one such approach is the conditional Granger causality

(Chen et al., 2006; Ding et al., 2006). In the broader context of conditional Granger causality

analysis applied to multivariate data, it is only reliable to submit channel pairs that exhibit

significant nonzero conditional Granger causality to the data analysis protocol described above.

3.2.3 Logic of Analysis Protocol

The goal of this work is to test that (a) positively correlated common input and (b)

bidirectional interaction contributes to near-zero phase-lag. In the simulation examples, this is

accomplished by changing the strength of input correlation and bidirectional interaction and

observing the corresponding change in the phase-lag. For actual data, while Geweke's theorem

allows the extraction of various causal influences through the decomposition of synchrony, the

strength of input correlation and bidirectional interaction is not easily manipulated. The sorting

of trials according to their phase-lag is the strategy to deal with this problem. Each subensemble

of trials gives rise to a different phase-lag value. Performing Geweke's decomposition for each

subensemble provides the avenue to observe the correlation between phase-lag and common

input/bidirectional interaction. It is important to note that, for a given pair of recording sites, both

instantaneous causality and the two directional influences may change as functions of the sorted

phase-lag. A simple pairwise correlation analysis may thus become confounded. Partial









correlation is used here to make possible the examination of one factor's contribution to

near-zero phase-lag while the contribution of other factors is statistically removed.

3.3 Results

3.3.1 Simulation

The impact of (a) positively correlated common input and (b) directional interaction on

phase-lag is examined using simulated time series and summarized in Table 3-1 and Table 3-2.

In the case of unidirectional interaction, increased common input to the network, measured by

instantaneous causality, reduces the magnitude of phase-lag from 2.19 radians to a near-zero

value of 0.16 radians (Table 3-1). Similar effect of instantaneous common input was observed in

networks with bidirectional interaction as well (not shown). In the case of the system with

bidirectional interaction and no instantaneous common input, the parametric increase in the

model coefficients b1,2,3,4 and 01,2,3,4 (Eq.3-1) resulted in corresponding increase in the estimated

feed-forward and feed-back interaction patterns and a decrease in the magnitude of phase-lag to

near-zero values (Table 3-2). However, not all reciprocally interacting systems exhibit near-zero

phase-lag synchrony. The actual phase-lag in a network depends on many factors including

delays involved in the feed-forward and feed-back pathways and the strength/type of coupling.

The approach outlined in this study enables one to empirically test the influences of each of the

factors on phase-lag in any given data.

3.3.2 Experiment

3.3.2.1 Network identification

Granger causality spectra are shown in Figure 3-1 for a pair of sites experiencing

unidirectional interaction (A) and another pair of sites undergoing bidirectional interaction.

Figure 3-2 shows the interaction patterns for all five sites pairs in both monkeys in the beta

frequency range where the same significance threshold criterion described in (Brovelli et al.,









2004) were used. Except for S1 and 7b in monkey GE and Ml and S1 in monkey LU, the

remaining site pairs in both monkeys exhibited unidirectional interaction. An earlier study

examining the causal patterns on the same dataset in a multivariate context using conditional

Granger causality has shown that the network analyzed here is robust and free of any significant

mediated influences from other simultaneously recorded areas that were not included in the

pairwise analysis pursued in this study. Unlike the other five site pairs where the interaction is

concentrated in the beta range (see Figure 3-2), the (M1,S1) pair in monkey LU also exhibited

significant interaction in the gamma frequency range, in addition to that in the beta range. For

this pair, the causal influences in the time-domain where instantaneous causality is most readily

interpreted, are confounded and is thus excluded from further analysis. Functionally, the

observation that S1 and 7b play a pivotal role in the organization of the network has led to the

hypothesis that the beta network supports the maintenance of lever depression by facilitating

sensorimotor integration (Brovelli et al., 2004; Ding et al., 2006).

3.3.2.2 Phase-lag distribution

For a given site pair, the phase-lag at the peak beta frequency was estimated for each trial.

Figure 3-3 shows the phase-lag distribution for two different pairs of sites in monkey LU and

GE. Both distributions are unimodal. In particular, despite a unidirectional interaction pattern

between Ml and 7b (Figure 3-2B), the phase-lag is approximately centered around zero with a

mean of 0.04 radians. This suggests that for such pairs the instantaneous causality may contribute

significantly to the overall degree of synchrony. Table 3-3 summarizes the mean phase lag in the

beta band for all five pairs of recording sites.

3.3.2.3 Synchrony decomposition and near-zero phase-lag

For recording sites A and B, according to Eq. 3-2, the total synchrony, which is derived

from the coherence function can be written as the sum of two directional influences (A-4B) and









(B-4A) and instantaneous causality (A.B) (Table 3-4). Intuitively, positively correlated common

input measured by the instantaneous causality has the effect of bringing phase-lag close to zero.

This is particularly so for pairs experiencing unidirectional causal influence. In monkey LU, the

phase-lag between Ml and 7b is near zero (Figure 3-2B and Figure 3-3B), and not surprisingly,

the instantaneous causality in this case makes up 72% percent of total interdependence, a

substantial percentage (Table 3-4). Below we tested the idea by carrying out an analysis for the

site pairs characterized by unidirectional interaction with the analysis protocol outlined in section

3.2.2.5.

3.3.2.4 Instantaneous causality and phase-lag

For each site pair the phase lag was estimated for each trial and the estimated value was

used to sort all trials into subensembles. The phase-lag and the instantaneous causality measure

for each subensemble constituted a point on a scatter plot. Figure 3-4 shows the result for (S1,7b)

in monkey GE and LU. Clearly, the two quantities are negatively correlated, indicating that as

the instantaneous causality increases, the phase-lag decreases and, in fact, approaches zero.

Spearman's rank correlation and Spearman's rank partial correlation coefficients were computed

for all the site pairs and listed in Table 3-5 and 3-6. All correlation coefficients were negative.

Except for (M1,7b) in monkey GE, they were statistically significant at p=0.05 level (one tail t-

test) (Table 3-5). By partialing out the effects of the directional influences, the correlation for

(M1,7b) also became significant (Table 3-6).

3.3.2.5 Bidirectional interaction and phase-lag

Inspection of the causality spectrum between the site pair (S1,7b) in monkey GE revealed

the presence of bidirectional interaction in the beta band (Figure. 3-1B and Figure 3-2B and

Table 3-4). The coherence function and the associated phase spectrum are shown in Figure 3-5.

This channel pair was further analyzed to identify the effect of reciprocal interaction on the









phase-lag. The strength of directional interaction is expressed as the sum of feedforward and

feedback influences. After partialing out the influence of the common input the phase-lag was

negatively correlated with the mutual interaction (r = -0.455 p = 0.0045). This result supports

our early assertion that in addition to instantaneous causality, bidirectional interaction may also

lead to near-zero phase-lag.

3.4 Discussion

The relative phase between two neural signals A and B, referred to as phase-lag here, can

be calculated from the cross-spectrum. For decades, the sign and magnitude of phase-lag have

been used to infer direction of information transmission and delay (Cassidy and Brown, 2003).

Increasingly, phase-lag is found to be near zero in synchronous cortical networks, sometimes

involving distant sites. Such phenomenon renders the use of phase-lag as a measure to identify

directional influences ambiguous. The phenomenon of zero phase-lag has been examined from

the point of view of neuronal communication and is considered a manifestation of the brain

integrating information from diverse sources (Singer and Gray, 1995; Roelfsema et al., 1997;

Engel et al., 2001; Brovelli et al., 2004). Two factors can be identified that contribute to

near-zero phase-lag: (a) positively correlated common input and (b) bidirectional interaction.

Recent reports of near-zero phase-lag arising in networks with a predominantly unidirectional

interaction pattern further highlights the importance of the first factor.

The influence of the two factors on phase-lag was tested on simulated datasets generated

by bivariate autoregressive models. First, it was observed that, for a linear system with

unidirectional interaction, a near-zero phase-lag was unlikely in the absence of an

instantaneously positively correlated input. As such an input is introduced and increased,

phase-lag is seen to decrease and approach zero. A similar influence on phase-lag was also

observed in bidirectionally interacting networks. Second, for the case of a bidirectionally









interacting system with no common input, increase in the strength of both feed-forward and

feed-back interaction leads to the reduction in the magnitude of phase-lag.

The empirical testing of the above ideas faces considerable challenge as a standard

correlation or coherence analysis do not offer sufficient information on the relation between the

two signals A and B. Recently, advanced connectivity tools have been proposed (Baccala and

Sameshima, 2001; Kaminski et al., 2001; Seth, 2005; Chen et al., 2006; Lungarella and Sporns,

2006; Wang et al., 2007b; Guo et al., 2008; Wu et al., 2008) which aim at parsing the synchrony

into directional interaction. Mathematically, a theorem proven by Geweke promises deeper

insights. It states that the total interdependence between A and B can be written as the sum of

three contributing factors: (A-4B), (B-4A) and (A.B). The arrow is understood in the sense of

Granger causality and (A.B) signifies instantaneous causality interpreted as reflecting a common

input. In the present study Geweke's theorem was applied to study the contribution of the two

factors identified above to near-zero phase-lag.

Local field potentials from primary somatosensory (S ), primary motor (Ml), and

posterior parietal (7b) areas were recorded in two monkeys performing a sensorimotor

integration task. A beta oscillatory network involving all three sites was identified by coherence.

The total interdependence was then decomposed into its directional components. Out of five

distinct pairs of recording sites studied, four exhibited predominantly unidirectional interaction

in the beta band. The phase-lag was nearly zero for one of the five pairs and relatively small for

another three. Unlike simulated models, neither the strength of input correlation nor the strength

of feed-forward / feed-back interaction can be manipulated to infer their influences on phase-lag.

The sorting of trials according to their phase-lag is a strategy to deal with this problem. By

sorting the trials according to single trial estimated phase-lag, a negative correlation was found









between the phase-lag and instantaneous causality for all pairs, implying that the stronger is the

common input the closer to zero is the phase-lag. Despite this tendency, the actual value of the

phase-lag for a given pair depends on the relative contribution of the each of the factors in

Eq. 3-2, and may vary broadly [from 0.04 (near-zero) to -1.34 (far-from-zero), see Table 3-6].

If the common input is negatively correlated, then the stronger is this input the close to +7

is the phase-lag (Figure 3-6). A careful inspection of the five pairs of recording sites revealed

that the noise terms in their respective autoregressive models (see Eq. 3-1) were all positively

correlated with the exception of the channel pair (S1,7b) in GE where the noise terms was

negatively correlated. Since the order of the recording leads used for the bipolar derivation was

arbitrary, the signal from S1 was reversed in polarity, which is equivalent to a depth-to-surface

recording. This correction enables the data from all five pairs to be considered under the same

hypothesis. Channel pair (Sl, 7b) in monkey GE also has another differing characteristic: the

interaction is bidirectional in the beta range. In light of the earlier discussion, the reciprocal

interaction in addition to common input could also contribute to the observed near-zero

phase-lag. This prediction was confirmed by a partial correlation analysis between the phase-lag

and the strength of feedforward and feedback interaction after removing the influence of the

instantaneous common input.

In sum, based on our simulated as well as experimental data, for two cortical regions

engaged in unidirectional interaction, instantaneous causality/common input is likely a main

contributor for the near-zero phase-lag between the sites. On the other hand, for two cortical

areas engaged in bidirectional interaction, near-zero phase-lag synchrony can emerge as a result

of reciprocal interaction or instantaneously correlated influence or a combination of both.

Geweke's decomposition theorem, combined with the analysis protocol outlined in the Methods









section can help to ascertain the exact network mechanism for a given problem. Each measure

obtained through this decomposition technique has the desirable feature that they all have clear

physiological correspondence. For example, bidirectional interaction is highly interpretable in

terms of the anatomical connectivity principle of reciprocity in the cortex (Felleman and Van

Essen, 1991). Instantaneous causality/common input may be taken to collectively reflect

activation of a distinct cortical or subcortical region that project to the sampled sites. Volume

conduction, while a possible contributor to instantaneous causality, is an unlikely factor in the

present study as bipolar derivation localizes neural activity to its generator. However, for scalp

EEG, the influence of volume conduction is likely to be significant and must be carefully

considered.



Table 3-1. Influence of increased instantaneous causality on phase-lag
Instantaneous causality Phase-lag (radians)
0.00 2.19
0.04 1.16
0.29 0.44
1.02 0.16
Simulated data were generated by a bivariate model with unidirectional interaction. Here instantaneous causality
characterizes the strength of common input. Magnitude of phase-lag is seen to decrease with increase in
instantaneous causality.




Table 3-2. Influence of increased bidirectional interaction on phase-lag
Fx-y / Fy-x Fx-y + Fy-x Phase-lag (radians)
0.038/0.081 0.119 1.31
0.085/0.117 0.202 0.94
0.267/ 0.131 0.398 0.36
Simulated data were generated by a bivariate model with bidirectional interaction. Fx-y and Fy-x quantify the feed-
forward and feed-back components estimated by Granger causality analysis. Magnitude of phase-lag is seen to
decrease with increase in strength of bidirectional interaction measured by feed-forward and feed-back Granger
causality.










Table 3-3. Mean beta band phase-lag (in radians) for monkeys GE and LU
Site pairs GE LU
M1,S1 -1.34
M1,7b 0.9 0.04
S1,7b -0.42 -0.4


Table 3-4. Contribution of intrinsic and exogenous components to total interdependence
Causality GE (%) LU (%)
M1-S1 13 29
S1-M1 75 27
M1-S1 12 44
M1-7b 1 72
7b-M1 86 25
M1-7b 13 3
S1-7b 17 48
7b-S1 17 9
S1-7b 66 43
-' indicates instantaneous causality, the exogenous component
'-' indicates directed interaction, the intrinsic components.


Table 3-5. Correlation between instantaneous causality and phase-lag
Site pairs GE LU
M1,S1 -0.79
M1,7b -0.33 (.058) -0.70
S1,7b -0.68* -0.73
Strength of Spearman's rank correlation coefficient between instantaneous causality and magnitude of phase-lag in
the beta band between site pairs specified in column 1 for monkeys GE and LU. The results that did not satisfy
significance at p<0.05, are included with their corresponding p-value. denotes the channel pair that exhibits
bidirectional interaction in the beta frequency band.


Table 3-6. Partial correlation between instantaneous causality and phase-lag
Site pairs GE LU
M1,S1 -0.79
M1,7b -0.31 -0.44
S1,7b -0.41* -0.81
Strength of Spearman's partial rank correlation coefficient between instantaneous causality and magnitude of phase-
lag in the beta band between site pairs specified in column 1. The partialing is against the directional influences for
monkeys GE and LU. The results that did not satisfy significance at p<0.05, are included with their corresponding p-
value. denotes the channel pair that exhibits bidirectional interaction in the beta frequency band.




























Frequency (Hz)


0 10 2O 30 40 s0 0o 70 80
Frequency (Hz)


90 100


Figure 3-1. Granger causality spectra. (A) a pair of sites experiencing unidirectional interaction
and (B) a site pair experiencing bidirectional interaction in the beta frequency band.
The threshold level for significance at p =,0.005 is overlaid as a flat line.








A B



7b M1( 7b M1





SI S1




Figure 3-2. Schematic Granger causality graph. (A) monkey GE and (B) monkey LU. Solid
arrows indicate directions of causal influence in the beta frequency band that were
significant at p = 0.005.


-.. -7b
S1


-



























Phase-lag (Radians)


Figure 3-3. Phase distribution. Histogram of single trial phase-lag values at peak beta frequency
for site pairs (A) (S1, 7b) in monkey GE and (B) (M1,7b) in monkey LU.


0.005 0.01 0.015 0.02
Instantaneous Causality


0.025


Figure 3-4. Influence of common input on phase-lag. Scatter plot showing strong negative
correlation between instantaneous causality (IC) and magnitude of phase-lag between
site pairs (S1,7b) in monkeys (A) GE and (B) LU.


Phase-lag (Radians)














0.35 2


I 35 0 *

I I |.1
0.3
0.25 V





0.05

0 10 20 30 40 50 60 70 80 90 100 0 5 10 15 20 25 30 35 40 45 50
Frequency (Hz) Frequency (Hz)


Figure 3-5. Coherence and phase spectra for site pair (S1,7b) in monkey GE. (A) Coherence
spectra for site pair (S1,7b) in monkey GE indicating strong beta band synchrony. (B)
The relative phase spectra for the same site pair. The near-zero phase-lag in the beta
band is the result of both positively correlated common input and reciprocal
interaction.








Uncorrelcorrelated
inputs A B inputs



1 2
Common inputs

If inputs 1 and 2 are positively correlated
As correlation of 1 and 2 increases, 9AB -- 0
If input 1 and 2 are negatively correlated
As correlation of 1 and 2 increases, pAB -X


Figure 3-6. Phase locking modes. Schematic illustration of different modes of phase locking for
a network with unidirectional interaction pattern and instantaneously correlated
inputs. (PAB denotes the phase-lag between sites A and B.


_ 1 11 1 1









CHAPTER 4
PRESTIMULUS THETA OSCILLATIONS IN FRONTO-TEMPORAL NETWORK PREDICT
SUBSEQUENT MEMORY RECALL IN HUMANS

4.1 Introduction

Items presented under identical environmental conditions are sometimes remembered and

sometimes not. Research into the origin of this variability has linked the efficacy of memory

encoding in the prefrontal cortex (PFC) and the medial temporal lobe (MTL) with memory

performance. In particular, it has been shown that the level of physiological responses in PFC

and MTL to presented items reliably predicts the likelihood of these items subsequently being

remembered (Fernandez et al., 1999; Friedman and Johnson, 2000; Tendolkar et al., 2000; Weiss

and Rappelsberger, 2000; Sederberg et al., 2003; Takashima et al., 2006; Sederberg et al., 2007).

While most studies emphasized the stimulus-evoked activity and its variability as a neural

correlate of memory performance, research has gradually expanded to include the

characterization of the state of the brain prior to stimulus input and its impact on memory

encoding (Seager et al., 2002; Guderian et al., 2009). Ongoing brain activity is replete with

oscillations of various frequencies (Niedermeyer et al., 1999; Buzsaki, 2006). Summarizing the

extensive literature on the theta rhythm (4-9 Hz) in the rodent hippocampus, Vertes (2005)

proposed that theta oscillations represent a depolarizing drive on NMDA containing cells, and a

stimulus presented during strong theta activity leads to enhanced response and facilitated

long-term potentiation (LTP) (Buzsaki, 1989; Huerta and Lisman, 1993; Staubli and Xu, 1995;

Buzsaki, 1996; Kocsis et al., 1999; Berry and Seager, 2001; Buzsaki, 2002; Seager et al., 2002;

Vertes et al., 2004; Hasselmo, 2005; Kocsis and Kaminski, 2006). How this hypothesis manifests

in human memory is not well understood. The first goal of the present investigation is to test the

hypothesis that the efficacy of memory encoding in humans depends on the level of excitability









indexed by the strength of theta activity in PFC and MTL immediately preceding the stimulus

onset, with stronger theta predicting better subsequent recall.

Memory is a network phenomenon. Formation of memory traces depends on the

coordinated activity in many brain areas. In particular, the importance of PFC-MTL interaction

in memory encoding has been supported by evidence from a variety of studies using lesion

(Petrides and Milner, 1982; Gershberg and Shimamura, 1995; Shimamura, 1995; Shimamura et

al., 1995; Bechara et al., 1998), neuroimaging (Kapur et al., 1996; Wagner et al., 1998; Cohen et

al., 1999; Grady et al., 2003; Gazzaley et al., 2004), and electrophysiological techniques

(Samthein et al., 1998; Weiss and Rappelsberger, 2000; Anderson et al., in press). Similar to the

earlier hypothesis where theta oscillations serve to index the level of tissue excitability in single

brain areas, it is expected that the level of neuronal interaction and synchrony prior to stimulus

onset acts as a gating mechanism and provides an index of increased processing capacity at a

network level (Ohara et al., 2006; Ohara et al., 2008). The second goal of this investigation is to

test the hypothesis that the strength of PFC-MTL interaction immediately prior to stimulus

presentation, possibly reflecting the strength of top-down attentional control originating in the

PFC (Furey et al., 2000; Kimura, 2000; Hasselmo and McGaughy, 2004), is positively correlated

with the efficacy of stimulus encoding and the likelihood of its subsequent recall.

The above goals are accomplished by recording electrocorticogram (ECoG) also known as

intracranial electroencephalogram (iEEG), from multiple implanted electrode grids in five

patients undergoing presurgical evaluation for intractable epilepsy who performed a subsequent

memory task. The subjects were given a series of words to remember. Then, following a

distraction period, the subjects were asked to recall the words from memory. Data from the

period of initial presentation of words (encoding period) was analyzed where the level of









regional prestimulus oscillatory activity and interregional interaction immediately prior to the

presentation of the words was compared between those that were subsequently recalled and those

that were not.

4.2 Materials and Methods

4.2.1 Subjects and Electrode Placement

Five patients (4 female and 1 male, mean age 38 years) with medically intractable epilepsy,

undergoing invasive video-EEG monitoring for localization of the epileptic focus and subsequent

epilepsy surgery, participated in the study. All participants gave informed consent. The

experimental and recording protocol was approved by the institutional review board of the

University of Florida. Figure 4-1 illustrates the approximate positions of the implanted electrode

grids in each of the five subjects. Subjects 1, 2 and 3 had subdural grids covering the lateral

prefrontal cortex (PFC), Subjects 1, 2 and 4 had subdural grids covering the lateral temporal lobe

(LTL), Subjects 1, 2 and 4 had subdural subtemporal strips on the ventral surface of the temporal

lobe, and Subject 5 had bilateral depth electrodes inserted through the occipital lobes into the

hippocampi. Data from the electrodes of the subtemporal strips and hippocampal depth

electrodes in Subjects 1, 2, 4 and 5 were treated as proxies of medial temporal lobe (MTL)

activity. For Subjects 1 and 2 the frontal and the temporal regions were simultaneously sampled.

4.2.2 Paradigm

The subjects were asked to perform a subsequent memory task. An LCD monitor placed

approximately 90 cms in front of the subjects was used to present the stimuli. A fixation cross

remained in the center of the screen throughout the experiment. The entire experiment was

divided into blocks. Each block, as illustrated in Figure 4-2, consisted of a presentation

(encoding) period, followed by a distraction period, then a free-recall period, and finally a

recognition period. The encoding period consisted of the sequential presentation of 20 words









chosen from the Kucera and Francis word pool (Kucera and Francis 1967). Each word was

displayed on the screen for 2 seconds with a random delay between words of either 1 or 2

seconds (mean 1.5 seconds). Subjects were instructed to memorize as many of the words as

possible. The distraction period lasted 30 seconds and consisted of counting aloud backwards by

threes, starting at a random number. This distraction period served to attenuate both the recency

effect and further encoding by rehearsing. During the free-recall period, subjects were given 50

seconds to recall aloud as many of the previously studied words as they could. This was followed

by a second presentation period of a different set of 20 words, followed by another distraction

period, and a free recall period. Finally, a longer distraction period which lasted 60 seconds was

given followed by the recognition period that consisted of the presentation of 80 words

sequentially on the screen, including the 40 previously studied words along with 40 foil words

not seen previously. The subjects were asked to respond by button presses to indicate whether or

not they recognized the word as being a previously presented word or a new word. The word

stayed on the screen until the subject responded. Each response was followed by a random delay

of either 1 or 2 seconds (mean 1.5 seconds) before the appearance of the next word. Subjects

performed either two or three blocks depending on their tolerance. The focus of this investigation

is how brain activity preceding each word to be memorized (shaded interval in the expanded

encoding period in Figure 4-2) influences subsequent encoding and later recall.

4.2.3 Data Preprocessing

Data were sampled at 400 Hz by a Nicolet amplifier system, band pass filtered from 0.1 to

100 Hz, which was subsequently subject to an offline lowpass filter with a cutoff at 30 Hz and

then downsampled to 250 Hz. Only channels (electrodes) free of pathological abnormalities and

epileptic spikes were included for analysis. Table 4-1 summarizes the number of the studied

electrodes in each cortical area for each of the five subjects. After rejecting data segments









contaminated by movement and other artifacts, the remaining artifact-free data recorded during

the encoding period was epoched from -500 ms to 0 ms (length of the shaded interval in the

expanded encoding period in Figure 4-2), with 0 ms denoting the onset of word presentation.

Epochs where the words were subsequently free-recalled were combined into an ensemble of

trials, referred to as recalled ensemble, and epochs where the words were not recalled were

combined into another ensemble of trials, referred to unrecalled ensemble. Power and synchrony

estimates (see below) were obtained for each ensemble for the prestimulus time period -500 ms

to 0 ms (see Figure 4-2). The statistical differences in these variables between the recalled and

unrecalled conditions were assessed.

4.2.4 Spectral Analysis

4.2.4.1 Power estimation

A multitaper approach was used to estimate power spectra for each of the two ensembles

of trials (recalled versus unrecalled) from frontal, lateral temporal and medial temporal channels

in each of the five subjects (Thomson, 1982; Mitra and Pesaran, 1999; Mitra and Bokil, 2007).

Three tapers were found to give the optimal balance between spectral smoothness and frequency

resolution.

4.2.4.2 Significance test for difference in power between conditions

For a given electrode, the difference between spectral power for the recalled ensemble and

that for the unrecalled ensemble was tested with a nonparametric Jackknife procedure (Bokil et

al., 2007; Thomson, 2007). Figure 4-3A illustrates the behavior of the test statistic for the

difference in power across all frequencies between the recalled and unrecalled ensembles (dark

line) for a single electrode from the frontal grid of Subject 3. The horizontal gray line denotes the

90% confidence interval based on Gaussian assumption of the log spectra. If the test statistic lies

within the 90% confidence band, then the null hypothesis that no difference in power between









the two conditions is not rejected. On the other hand, if the test statistic lies outside the

confidence band, then the null hypothesis of equal power is rejected at the significance level of

a=0.1. To reduce false positives as a result of multiple comparisons, the rejection of the null

hypothesis for any particular frequency (f) was carried out only if the test statistic lay outside the

confidence band for a contiguous frequency range between (f-2) and (f+2) Hz (Figure 4-3A). For

the channel illustrated in Figure 4-3A significant differences in power were observed in the theta

band (4-9 Hz). The aforementioned procedure was carried out for the frequency range between 3

and 28 Hz. Figure 4-3B summarizes the total number of electrodes over all five subjects that

exhibited significant difference in power between the recalled and unrecalled ensembles for each

frequency in the 3-28 Hz range.

4.2.4.3 Group power analysis

The channels having significant differences in power between the recalled and unrecalled

conditions were subjected to group analyses. Band averaged power for each condition was

normalized by dividing by the sum of band averaged power from both conditions and averaged

across electrodes within each of the three brain regions (PFC, LTL, MTL) which was further

averaged across subjects. The group average was compared between the two conditions and all

statistical comparisons were performed using Student's t test.

4.2.4.4 Inter-areal synchrony analysis

The raw data from each implanted electrode were recorded against a common reference

which is typically not free of neural activity. Volume conduction presented a further

complicating factor for connectivity analysis. To overcome these problems, bipolar derivations

or signals, which were computed as the differences between horizontally, vertically, and

diagonally neighboring electrodes, were used for synchrony analysis. Table 4-2 summarizes the

number of bipolar derivations for the two subjects (Subject 1 and 2) who had simultaneous PFC









and temporal lobe recordings. For each pair of inter-grid bipolar signals, a frequency domain

interdependence measure called the phase locking value (PLV) (Jean-Philippe et al., 1999) was

used to evaluate the degree of phase synchrony between the two brain regions in a given

frequency band, with larger PLV values indicating higher levels of phase synchrony, which is

taken to imply greater functional connectivity.

To calculate PLVs, the data were bandpass-filtered between 4 and 9 Hz for theta band, 10

and 14 Hz for the alpha/low beta band and 21 and 25 Hz for the high beta/low gamma band. The

instantaneous phase of the signal was extracted using the Hilbert transform on a trial-by-trial

basis and the difference in phase between the bipolar pairs of interest was computed. Following

an ensemble averaging of the estimate, the instantaneous phase locking measure was derived

according to

1N
PL V(t) = exp(j((, (t, n)- ( (t, n)))
N n=1

where O (t,n) is the phase of the i-th signal at time t and trial n (i=l or 2). This PLV measure was

calculated for the recalled ensemble and unrecalled ensemble separately for each subject and

averaged over the 500 ms period immediately prior to word presentation (i.e. -500 ms to 0 ms).

The values of PLV range from 0 to 1, where 1 indicates perfect synchrony across trials, and 0

indicates that the phase relationship is completely random from trial to trial.

4.2.4.5 Random permutation test for statistical significance

To test the significance of inter-grid PLV, the following procedure was followed (Jean-

Philippe et al., 1999; Brovelli et al., 2004), the aim of which is to create a null hypothesis

distribution for the PLV measure in the frequency band of interest. (1) For each subject, the

500 ms prestimulus epochs were numbered from 1 to N where N is the total number of epochs

across both recalled and unrecalled ensembles. (2) For a specific inter-grid comparison, the









epoch index of one of the grids was permuted randomly to create a synthetic data set where it is

reasonable to assume that no interdependence exists between any pair of inter-grid bipolar

derivations. (3) All pairwise PLV was calculated for the synthetic data set and the single largest

PLV for the frequency range of interest from all bipolar signal pairs was selected. (4) Steps (2)

and (3) were repeated 400 times. (5) The null hypothesis distribution was estimated by fitting an

extreme-value distribution to the histogram of the largest PLV measures (Wang et al., 2007a).

Estimated PLV from the experimental data were considered significant if they exceeded the

99.9th percentile value of the null hypothesis distribution (p<0.001). Note that testing the many

combinations of bipolar signal pairs between grids presents a multiple comparison problem.

Choosing the maximum value in Step (3) is a way to account for this problem (Nichols and

Holmes, 2002).

4.3 Results

Preliminary analysis of this experiment has appeared previously in an abstract form

(Rajagovindan et al., 2007). The subjects were attentive throughout the experiment and

completed the task according to instructions. On average, 31% of the words were free-recalled,

and 88% of the words were correctly identified during the recognition period. These behavioral

results are in line with published reports (Sederberg et al., 2003).

4.3.1 Power

Spectral power over the interval (-500 ms to 0 ms) preceding word presentation in the

encoding period (shaded interval in the expanded encoding period in Figure 4-2) was estimated

for the recalled ensemble of trials and for the unrecalled ensemble of trials. The number of

recording channels exhibiting significant differences between the two conditions across all three

brain areas and subjects is summarized in Figure 4-3B. Strongest modulation is seen in the theta

band (4-9 Hz) with smaller effects also observed in the alpha/low beta band (10-14 Hz) and the









high beta/low gamma band (21-25 Hz). Figure 4-4 shows the group-averaged theta band power

from the significantly modulated channels in Figure 4-3B for each of the three brain regions.

Over the PFC region, theta band power was significantly higher prior to the words that were

subsequently free recalled relative to words that were not (p=0.026). Results in the MTL and

LTL regions are mixed. While the theta band power is generally higher for the recalled ensemble

compared to the unrecalled ensemble, the difference failed to reach the statistical significance

level of p=0.05. However, by pooling LTL and MTL channels together, the overall temporal lobe

theta band power was found to be significantly higher for the recalled words relative to

unrecalled words (p=0.046). Similar analysis for the alpha/ low beta band power (10-14 Hz) and

high beta/low gamma band power (21-25 Hz) revealed no statistically significant difference in

any of the three brain regions (p>0.1).

4.3.2 Inter-Areal Phase Synchrony

For the two subjects (Subjects 1 and 2 in Figure 4-1) with simultaneous coverage of PFC,

MTL and LTL, the pattern of interaction between these areas over the prestimulus interval

(-500 ms to 0 ms) was evaluated by computing the PLV measure in theta, alpha/ low beta, and

high beta/low gamma bands between all possible pairwise combinations of (a) PFC-LTL bipolar

derivations, (b) PFC-MTL bipolar derivations, and (c) MTL-LTL bipolar derivations. Each black

dot in Figure 4-5A (Subject 1) and Figure 4-5B (Subject 2), represents the PLV value for a

distinct pair of bipolar signals, and the height of the rectangular bars is the mean of all black dots

for a given condition. Significantly higher prestimulus theta band synchrony is present for the

recalled ensemble relative to the unrecalled ensemble in all three pairwise combinations of brain

areas (PFC-MTL, PFC-LTL, MTL-LTL) (p<7e-6). Another way of quantifying the task-related

modulation of inter-areal theta synchrony is the number of pairwise combinations whose theta

band PLV values exceed the threshold corresponding to the significance of p=0.001 (horizontal









lines in Figure 4-5A and 4-5B; see Methods). Since the number of bipolar signals in each grid

varied between subjects, the percentages of pairwise bipolar combinations above threshold were

calculated and averaged across subjects. As seen in Figure 4-6A, the percentage of pairs showing

significant theta synchrony is significantly higher for the recalled ensemble than for the

unrecalled ensemble (p=0.0028) in all pairwise combinations of the three brain areas: PFC-MTL,

PFC-LTL and LTL-MTL. The same synchrony analysis in the alpha/low beta band (Figure 4-6B)

and high beta/low gamma band (Figure 4-6C) revealed a similar pattern of higher prestimulus

synchrony between frontal and temporal lobe structures for words that were subsequently

recalled than words that were not (p=5e-5 and p=0.0002, respectively). A schematic illustrating

the relation between the level of prestimulus interregional synchrony and subsequent memory

performance is given in Figure 4-6D.

4.4 Discussion

In this study, we investigated how neuronal activity preceding the events to be memorized

affects their subsequent encoding and recall. ECoG signals from the lateral prefrontal cortex

(PFC), medial temporal lobe (MTL), and lateral temporal lobe (LTL) were recorded from

implanted subdural electrodes in five epilepsy patients who performed a subsequent memory

task. Applying nonparametric spectral methods to the data we obtained two results. First,

free-recalled words were associated with higher theta oscillation prior to their presentation over

the frontal and temporal lobe structures compared to unrecalled words. Second, inter-areal

synchrony in theta as well in alpha/low beta and high beta/low gamma bands preceding the

presentation of free-recalled words was higher than that preceding the presentation of unrecalled

words.









4.4.1 Prestimulus Ongoing Activity, Theta Oscillations and Memory Encoding

Studies on the neuronal mechanisms of memory encoding have placed emphasis on

stimulus-evoked responses and their ability to predict memory performance (Klimesch et al.,

1996; Fernandez et al., 1999; Friedman and Johnson, 2000; Tendolkar et al., 2000; Weiss and

Rappelsberger, 2000; Jensen and Tesche, 2002; Sederberg et al., 2003; Takashima et al., 2006;

Sederberg et al., 2007). In many areas of cognitive neuroscience recent work has begun to

elucidate the importance of ongoing brain activity immediately before stimulus onset in defining

a state of the brain and the effect of such a state on subsequent stimulus processing (Engel et al.,

2001; Liang et al., 2002; Dehaene and Changeux, 2005; Ohara et al., 2006; Ohara et al., 2008;

Zhang et al., 2008b).

Ongoing brain activity is replete with oscillatory phenomena. In particular, the role of

hippocampal and limbic theta-band (4-9 Hz) EEG activity in learning and memory has been

extensively studied (Buzsaki, 1989; Huerta and Lisman, 1993; Staubli and Xu, 1995; Buzsaki,

1996, 2002; Vertes et al., 2004; Hasselmo, 2005; Vertes, 2005). Pharmacological blocking of

acetylcholine and medial septum lesions decrease theta activity and lead to memory impairments

(Winson, 1978; Givens and Olton, 1990; Berry and Seager, 2001; Siok et al., 2006) while

pharmacological blocking of serotonin enhances theta rhythms and leads to improved

performance in spatial memory tasks (Staubli and Xu, 1995; Berry and Seager, 2001; Seager et

al., 2002). At the cellular level, episodes of theta oscillations represent a strong depolarizing

influence on NMDA receptor-containing cells. Glutamate release triggered by the stimulus input,

when temporally coupled with the theta-induced depolarization, activates the theta-facilitated

NMDA channels and leads to stronger Ca2+ influx, which in turn facilitates LTP and memory

trace formation (Grunwald et al., 1999; Seager et al., 2002; Vertes et al., 2004; Honey et al.,

2005; Vertes, 2005; Gallinat et al., 2006; Lisman et al., 2007; Meador, 2007). This suggests that









the level of theta activity prior to the presentation of stimulus is an index of tissue excitability

and should be positively correlated with the efficacy of subsequent memory encoding and recall

(Otten et al., 2006; Guderian et al., 2009). Our first finding is consistent with this hypothesis. In

addition, our data demonstrates that the theta-facilitation principle can be extended beyond

animal preparations and the temporal lobe to manifest strongly in human neocortical structures

important for memory encoding such as the prefrontal cortex.

4.4.2 Inter-Areal Synchrony and Network Excitability

PFC, MTL and LTL are part of a network of brain areas involved in memory encoding,

consolidation and retrieval (Scoville and Milner, 1957; Janowsky et al., 1989; Tulving et al.,

1994; Kapur et al., 1996; Wagner et al., 1998; Kirchhoff et al., 2000; Otten et al., 2002; Strange

et al., 2002; Simons and Spiers, 2003). While stimulus-triggered inter-areal synchronization and

its influence on memory performance has been examined in the past (Sarnthein et al., 1998;

Weiss and Rappelsberger, 2000; Summerfield and Mangels, 2005), little is known about how

inter-areal synchrony prior to stimulus onset serves as a gating mechanism to facilitate

subsequent stimulus-driven network operations. Our second finding can be seen as offering the

first evidence in humans demonstrating that increased inter-areal synchrony in the

fronto-temporal network prior to word presentation is associated with higher likelihood of the

word being recalled in the subsequent test (see Ohara et al., 2008 for report of a similar analysis

in pain perception). The changes in inter-areal synchrony, taken as an indicator of changes in

network excitability, may be a result of spontaneous fluctuations in the level of anticipatory

attention. Recent studies have shown that anticipatory attention can modulate phase synchrony in

a diverse set of cortical networks (Engel et al., 2001; Fries et al., 2001; Ohara et al., 2006;

Rajagovindan and Ding, 2008b) and differences in prestimulus inter-areal synchrony are

correlated with conscious perception of stimuli in different sensory modalities (Hanslmayr et al.,









2007a; Ohara et al., 2008; Zhang and Ding, in press). It is worth noting that, unlike the

theta-facilitating hypothesis where activity in a single frequency band is implicated in the

indexing of tissue excitability, the prestimulus facilitation at the network level is across a broad

range of frequencies, including theta, alpha, beta and low gamma.


Table 4-1. ECoG electrode information. Number of recording channels within each grid
overlying the prefrontal cortex (PFC), lateral temporal lobe (LTL) and medial
temporal lobe (MTL) in each of the five subjects.
Subject # PFC LTL MTL
1 14 15 4
2 20 9 6
3 43 -
4 -25 7
5 -9


Table 4-2. ECoG bipolar electrode derivation summary. Total number of bipolar derivations in
each area and total number of inter-grid pairwise combinations of bipolar signals.
# bipolar derivations # inter-grid combinations
Subject # PFC LTL MTL PFC-LTL PFC-MTL LTL-MTL
1 26 29 6 754 156 174
2 55 20 7 1100 385 140











Subject


Figure 4-1. Approximate placement of electrode grids for each of the five subjects.


D FR
30 s 50 s





Word 1
10Y


Word 2
.I.


D FR
30 s 50 s





SWord 20
.I 1


D Recognition


60 s


40 Targets + 40 Foils
Y/N response


D Distraction period

FR Free Recall period

M Prestimulus period


Figure 4-2. Schematic of the experimental paradigm. The sequence of events within a single
block of the subsequent memory paradigm is given and the encoding phase (shaded in
gray) is expanded to highlight the prestimulus interval (-500 to 0 ms) which is the
focus of this work.


" '


""


""
""


""
""














4..
CO0






AI


S20



S10

5

0


0 5 10 15 20
Frequency (Hz)


25 30


5 10 15 20 25
Frequency (Hz)


Figure 4-3. Prestimulus differences in spectral power between the recalled and unrecalled
ensembles of words. (A) The test statistic for the difference in prestimulus power
between the recalled and unrecalled ensembles across frequencies from 0-30 Hz
obtained from a representative frontal electrode. Notice that for this channel the
power in the theta band was significantly higher preceding the subsequently free-
recalled words relative to the unrecalled words (shaded in gray). The horizontal gray
line marks the p=0.1 level of significance. (B) The number of electrodes exhibiting
significant differences in prestimulus power between the two ensembles of words
across frequencies from 3-28 Hz.












PFC


0.5


0.4
z


Recalled Unrecalled


Figure 4-4. Comparison of theta band power. Group-averaged prestimulus theta band power for
the recalled and the unrecalled ensembles of words over the three brain regions. Here
PFC = prefrontal cortex, LTL = lateral temporal lobe, and MTL = medial temporal
lobe.


PFC LTL


* a


Recalled Unrecalled


PFC MTL


. 0
I. 0.1



0


LTL- MTL


.......................... .....


PFC- LTL


Recalled Unrecalled


PFC MTL







Recall Unrec


Recalled Unrecalled


Figure 4-5. Comparison of inter-areal synchrony. Inter-grid phase locking value (PLV) results
for Subject 1, (A) and Subject 2, (B). (A) and (B) Theta band PLV values for every
pairwise combination of bipolar signals for each inter-grid pair is plotted as black
dots. From left to right are PFC-MTL, PFC-LTL, and LTL-MTL. Height of gray bars
and white bars is the mean PLV for the recalled and the unrecalled conditions
respectively. The gray horizontal lines indicate the p=0.001 level of statistical
significance for different grid pairs.


0.5
0
IL

"0.4
0
z

0.3


LTL- MTL


'
'
'
'.'.,. .-C.













A PFC LTL


Unrecalled


Unrecalled


PFC LTL


Recalled Unrecalled


PFC LTL


B

30











C

80
0






0
0L-


100

2 80
s.



01
- 20
0


PFC MTL


Recalled Unrecalled


PFC MTL


Recalled Unrecalled


Recalled


LTL- MTL


Recalled Unrecalled


LTL- MTL


100

E 80

S60

| 40

0
-20


Recalled Unrecalled


Unrecalled


Figure 4-6. Comparison of percentage of bipolar signal pairs with significant PLVs. Average
percent of inter-grid bipolar signal pairs whose theta (A), alpha/low beta (B) and
beta/low gamma (C) PLV exceeds the significance threshold for the recalled (gray
bars) and the unrecalled (white bars) conditions. (D) Schematic of the relation
between levels of prestimulus inter-areal synchrony, represented by the thickness of
the curves connecting different brain regions, and subsequent memory performance.


Unrecalled


Recalled Unrecalled


PFC MTL


LTL- MTL









CHAPTER 5
FROM PRESTIMULUS ALPHA OSCILLATION TO VISUAL EVOKED RESPONSE: AN
INVERTED-U FUNCTION AND ITS ATTENTIONAL MODULATION

5.1 Introduction

Research into the neuronal mechanisms of cognition has mainly relied on stimulus evoked

responses and their modulation by higher mental functions such as attention and memory. Recent

work has begun to explore how ongoing neural activity prior to stimulus onset influences

stimulus processing and behavior. In the human visual cortex the relation between prestimulus

alpha activity (8-12 Hz) and the subsequent event-related potential (ERP) has been the subject of

extensive investigation. The phase-reset model supposes that the ERP is generated as a result of

stimulus-triggered alpha phase realignment and predicts a positive correlation between the

prestimulus alpha power and the magnitude of early visual evoked response (Makeig et al., 2002;

Hanslmayr et al., 2007b) also see (Mazaheri and Jensen, 2006) Two lines of empirical evidence

appears to be at variance with this prediction: (1) a state of low alpha is found to improve

detection and discriminability of threshold-level stimuli (Ergenoglu et al., 2004; Hanslmayr et

al., 2007a; Romei et al., 2008a; Van Dijk et al., 2008) and (2) covert attention reduces

prestimulus alpha amplitude and at the same time increase stimulus-evoked response (Mangun

and Hillyard, 1991; Foxe et al., 1998; Hillyard et al., 1998; Mangun et al., 1998; Marrufo et al.,

2001; Sauseng et al., 2005; Thut et al., 2006). The hypothesis that prestimulus alpha power

should be inversely proportional to stimulus evoked response proves to be also contradicted by

studies where (1) alpha episodes over the parieto-occipital areas immediately preceding the

stimulus onset enhanced the ability to detect a threshold-level visual stimuli (Babiloni et al.,

2006) and (2) higher alpha power led to stronger evoked response (Brandt and Jansen, 1991;

Brandt et al., 1991; Paul et al., 2007). These divergent reports, in conjunction with recent

findings in the somatosensory domain where the mu rhythm (7-13 Hz) is found to exhibit an









inverted-U relationship with the event-related potential and behavior (Ho and Destexhe, 2000;

Linkenkaer-Hansen et al., 2004; Zhang and Ding, in press), suggest that a new assessment of the

relation between alpha and visual information processing is needed. An added benefit of such a

reexamination is that it may yield insights into the fundamental question of how cognitive

operations such as attention modulate baseline ongoing neural activity to achieve enhanced

stimulus information processing.

Noninvasively recorded EEG and invasively recorded local field potentials are a measure

of dendritic activity in populations of neurons. Computational as well as in vitro and in vivo

studies in animal preparations have shown that synaptic background activity can exert a

significant influence on the responsiveness of sensory neurons to stimulus input (Ho and

Destexhe, 2000; Chance et al., 2002; McCormick et al., 2003; Wolfart et al., 2005; Haider et al.,

2007). Our approach to investigating the relation between prestimulus activity and stimulus

evoked response starts by considering a theoretical model in which synaptic input to a sensory

neuronal ensemble comes in two forms: external stimulus (exogenous) or other brain structures

endogenouss). On a moment by moment basis, the endogenous brain activity fluctuates

randomly, and the effective processing of a stimulus depends on the state of the brain defined by

these fluctuations at its onset. Supposing that, in the absence of sensory input, the firing rate of

the neuronal ensemble is related to the magnitude of endogenous synaptic fluctuations through a

sigmoidal function, and sensory input has an additive synaptic effect, it follows that the stimulus

evoked response corresponds to the derivative of the sigmoidal function, referred to as the gain,

which is an inverted-U function of the prestimulus level of background activity. Reformulated in

terms of EEG variables where alpha oscillations and the P1 component in the event-related

potential are treated as indices of background synaptic activity and stimulus-evoked response,









the model makes a number of predictions, which were then tested by recording EEG data from

human volunteers performing a trial-by-trial cued covert spatial visual attention task. Finally, a

mechanism on how changes in the global state of the brain, such as selective attention,

influenced the aforementioned sigmoidal curve was proposed. Through experimental data and

computational modeling, the reduced magnitude and trial-to-trial variability of alpha activity is

linked to an increase in cortical excitability brought about by attention, which enables an

increase in the overall slope of the sigmoidal curve and hence higher gain to sensory input.

5.2 Materials and Methods

5.2.1 Subjects

The experimental protocol was approved by the University of Florida Institutional Review

Board. Nineteen subjects, free from movement and neurological disorders and with normal or

corrected-to-normal vision, gave written informed consent and participated in the study. Data

from twelve subjects (20-31 years of age, 25.16 +3.5 years, 5 females and 7 males, all right

handed) were included in the analyses reported here. The remaining 7 participants were excluded

for a combination of three reasons: (1) poor behavioral performance (target detection rates less

than 75%; 3 subjects), (2) excessive eye blinking, eye movement, and muscle activity or

electrolyte bridging (3 subjects), and (3) inability to complete the experiment (1 subject).

5.2.2 Paradigm

Seated in a dimly lit, acoustically and electrically shielded room, the subject was instructed

to fixate a central cross on a VGA monitor placed approximately 85 cm from the eye. Two

square boxes, each delineated by white dots placed at the four corners in the upper left visual

field and in the upper right visual field, were displayed permanently on the monitor (Figure 5-1).

Each box, centered 4.70 above the horizontal meridian and 7.20 lateral to the central fixation,

subtended a visual angle of 3.35. A trial began with the onset of a cue, which consisted of either









a left or right pointing arrow of 200 ms in duration at the location of the central fixation cross,

instructing the subject to deploy covert attention to the square box on the side indicated by the

arrow. After a random time delay between 1800 ms and 2200 ms, a standard or a target stimulus

of 100 ms in duration appeared either inside the attended square box (valid trial) or inside the

square box on the opposite side (invalid trial). The standard stimulus was a circular checkerboard

subtending a 3.30 visual angle, and the target stimulus was also a circular checkerboard, but of a

slightly smaller diameter. The subject was required to press a button with their right index finger

in response to the valid target stimulus and withhold response to any other stimuli. The standards

appeared 80% of the time with 50% validity and the targets appeared 20% of the time with 66%

validity. The interval between the cue offset of the previous trial and the cue onset of the

successive trial was randomly varied between 4700 to 5700 ms. The entire experiment was

comprised of 15 to 16 blocks of approximately 6 min each (60 trials). Breaks were given

between blocks. Subjects received practice sessions of 150 trials to familiarize themselves with

the task and to minimize the effect of learning.

Four of the twelve subjects also took part in an additional behavioral experiment in which

they responded to targets for both valid and invalid trials. The purpose was to establish the

validity effect on the reaction time. In this task, the standards appeared 80% of the time and the

targets appeared 20% of the time, both with 50% validity. This additional experiment consisted

of 10 blocks of 64 trials each.

5.2.3 Data Acquisition

The electroencephalogram (EEG) data was recorded with a 128-channel BioSemi

ActiveTwo System at a sampling rate of 1024 Hz. Four additional electrooculogram (EOG)

electrodes were placed around the eyes to monitor eye blinks and horizontal and vertical eye

movements. EEG and EOG data was amplified and lowpass-filtered with a cutoff set at 205 Hz.









The stimulus and the response was delivered and registered by a BeriSoft Experimental Run

Time System (ERTS) and by a Berisoft EXKEY microprocessor logic pad respectively. In

addition to the attention experiment, five minutes of EEG activity during relaxed fixation was

also recorded for baseline purposes.

5.2.4 Data Preprocessing

Analysis was performed using a combination of BESA 5.1 (MEGIS software GmbH,

Munich, Germany) and custom scripts written in MATLAB 7.4 (MathWorks Inc.,

Massachusetts). The EEG data was filtered off-line between 1 and 83 Hz and downsampled to

250 Hz. The channels were re-referenced off-line against the average reference (Nunez et al.,

1997). Trials contaminated by any of the following factors were excluded from further analysis:

(1) incorrect behavioral response, (2) excessive muscle or movement artifact, (3) eye blinks

during a 150 ms period around the onset of the cue and the standard stimulus, and (4) horizontal

eye movement following the cue onset and during the period -1000 ms to 300 ms with 0 ms

denoting the presentation of the standard stimuli. The average trial rejection rate was 24%. The

remaining eye blinks were removed using the adaptive artifact correction procedure in BESA

5.1. The high rejection rate is necessary because the single-trial based analysis protocol here

places stringent requirements on the quality of the EEG signal (see below). The data from the

fixation only condition was segmented into 300ms long non-overlapping artifact free epochs. For

the spatial attention task, the continuous EEG data was epoched from -1000 ms to 800 ms. For

event-related potential (ERP) analysis, the data epochs were baseline corrected by subtracting the

mean voltage over the interval -150 ms to 0 ms.

There are four conditions for comparison: a) Attend Left cuedd left and standard stimulus

appearing on the left; valid cue), b) Attend Right cuedd right and standard stimulus appearing on

the right; valid cue), c) Ignore Left cuedd right and standard stimulus appearing on the left;









invalid cue), and d) Ignore Right cuedd left and standard stimulus appearing on the right).

Stimulus evoked neural activity measured by ERP as well as ongoing neural activity from the

prestimulus time period during which preparatory attention is engaged were studied. Between

two to four neighboring channels overlying the occipital and parieto-occipital areas with

comparable magnitudes of Pl evoked component (peak, 90-160 ms) and prestimulus alpha band

power (8-12 Hz) exhibiting attentional modulation were chosen for analysis.

5.2.5 Behavior and Event-related Potential Analysis

Behavior was quantified by three measures: (1) target detection rate defined as the number

of correctly identified valid targets divided by the total number of valid targets presented, (2)

false alarm rates for the conditions where the participants had to withhold their responses was

defined as the number of responses to (a) invalid targets, (b) valid standards and (c) invalid

standards, each divided by total number of trials in each of these conditions, and (3) mean

reaction time for correctly detected valid targets. The visual event-related potential (ERP) was

computed by averaging the EEG response elicited by the left and right field standard stimuli for

the attended and ignored conditions. The amplitude of the ERP components were estimated using

the baseline-to-peak method where the baseline period was defined to be (-150 to 0 ms) (Luck,

2005). Wilcoxon's signed-rank test was used to evaluate the statistical significance for the

attentional enhancement of the ERP components over the occipital and parieto-occipital areas.

5.2.6 Correlation Between Prestimulus Alpha Power and P1 Amplitude

A key objective of this study is to investigate how the state of the brain immediately

preceding the onset of stimulus affects stimulus processing. The former is characterized by the

power of alpha oscillation (8-12 Hz) during the time period from -300 ms to 0 ms and the latter

by the amplitude of the visually evoked P1 component (90-160 ms). The method for

accomplishing the objective is correlation analysis and it has the following steps. (1) Prestimulus









power for each of the pre-selected channels overlying the parieto-occipital area contra-lateral to

the visual field of stimulation was computed on a trial by trial basis using the multitaper spectral

estimation technique (Thomson, 1982; Mitra and Pesaran, 1999). This power was then averaged

over the alpha frequency band and across the set of pre-selected channels. (2) The trials from

each subject were then sorted, according to the magnitude of the prestimulus alpha power, from

the smallest to the largest, into 5 groups of equal size (Liang et al., 2002; Zhang et al., 2008b)

with 50% overlap. Each group, containing 30% of the total trials, will be henceforth referred to

as a power group. (3) For every power group, the group-mean prestimulus alpha power and the

baseline-to-peak group visually evoked P1 component were estimated for each subject. (4) To

minimize the effect of inter-subj ect variability on population averaging, the normalized P1

amplitude was obtained by dividing the evoked P1 for each power group by the largest evoked

P1 component across all power groups under the attend condition for every individual subjects.

(5) The normalized evoked P1 component was averaged across subjects and plotted against the

power group index. The relation between the two variables, reflecting prestimulus and

poststimulus activities respectively, was assessed from the scatter plot.

5.2.7 Characterization of Ongoing Alpha Activity

Ongoing brain activity in the visual cortex is dominated by oscillations in the alpha band.

On a trial by trial basis, the amplitude of alpha activity varies markedly; collectively, these

amplitudes form an alpha power distribution. Past work has focused on the mean of this

distribution (mean alpha power) and its modulation by attention. Here, in addition to the mean,

the variance of the distribution is also given physiological importance and compared across

conditions. Three conditions are considered: attend, ignore, and relaxed fixation. For the first two

conditions, the 300 ms interval prior to the onset of the standard stimulus constitutes a trial. For

the relaxed fixation, continuous recordings were divided into 300 ms nonoverlapping epochs,









and each epoch is considered a trial. The same method as that described above was used to

estimate the single-trial alpha power for each hemisphere of every subject. To test the difference

in alpha variance between conditions, single-trial alpha power was log transformed to ensure an

approximate normal distribution. Population differences across subjects in the magnitude of the

estimates between a pair of conditions were tested for statistical significance using Wilcoxon's

signed-rank test.

5.2.8 Computational Modeling

A computational model was employed to further test the validity of the proposed

framework at a single neuron level. To assess the influence of change in excitability on the slope

of synaptic input-output curve, the point-conductance model of background activity introduced

by Destexhe et.al., (Destexhe et al., 2001) was employed here. The model comprises of a single

compartment Hodgkin-Huxley type neuron with excitatory and inhibitory conductance

representing the collective effect of a large number of synaptic inputs and the membrane

potential is described by the following equation:

dV 1
c-I -I --I (5-1)
C = _ILeak Na IKd- IM Iyn (5-1)
dt a
where C = 1 pF/cm2 is the specific membrane capacitance, V is the membrane potential and

a=34636 mm2 is the total membrane area of a representative cortical pyramidal cell (Destexhe et

al., 2001). INa is the voltage-dependent Na+ current, IKd is the delayed rectifier K current, IM the

noninactivating K+ current, and ILeak the leak current. The parameters of the passive and voltage-

dependent currents were taken from (Destexhe et al., 2001). The total synaptic current in Eq. 5-1,

Isyn, is decomposed into a sum of two independent currents:

I,, = ge(t)(V- Ee) + g, (t)(V- EL) (5-2)









where ge(t) and gi(t) in Eq. 5-2 are time-dependent global excitatory and inhibitory conductance,

respectively, with reversal potentials of Ee = 0 mV and E = -75 mV. The synaptic conductance

are described by Ornstein-Uhlenbeck stochastic processes (Destexhe et al., 2001):

dg, 12
de (g, go)+ xe(t)
dt Tr rt,
dt ze (5-3)
dg, 1 2
g 1(g, _go)+ \ _(t)
dt V ,7
where, geo and gio, oe and ci, Te and Ti, are the average conductance, standard deviations, and time

constants, for the excitatory and inhibitory inputs, respectively. xe(t) and xi(t) are zero mean

Gaussian white noise processes of unit standard deviation. The parameter values of Eq. 5-3 were

taken as Te = 2.73 ms and Ti = 10.49 ms, gio = 0.06 iS, geo was varied between 0.01lS and 0.0364

in steps of 0.0033 resulting in 8 levels of increasing excitability state. 25 levels of background

synaptic activity were simulated by changing the standard deviation of inhibitory conductance oi

between 0 and 0.0375 S in steps of 0.0015 S. The parameter Ge, the standard deviation for

excitatory conductance, was kept constant at 0.006.S. Qualitatively, similar results were

obtained when Ge was varied with ci kept constant or when both ci and ,e were varied in tandem.

Note that varying ci or cG also shares a correspondence with varying input correlation between

individual synapses (Destexhe et al., 2001). The firing rates were collected for 40 simulation

trials of 200 ms duration each, for 25 values of oi for 8 excitability levels.

5.3 Results

The relation between prestimulus alpha oscillation and stimulus evoked response and its

attentional modulation were investigated by combining a theoretical model with human visual

attention experiment. Below, basic experimental findings are reported first, which is followed by

model analysis and test of model predictions.









5.3.1 Behavior

The average target detection rate across subjects was 87.27% ( 3.2% SEM) for attended

targets appearing in the left visual field and 87.44% ( 3.68% SEM) for attended targets

appearing in the right visual field. These results suggest that the level of task difficulty was

equated between the attend-left and the attend-right conditions (p=0.8, two-tailed Wilcoxon's

signed-rank test). The average false alarm rate stemming from responses to the target appearing

in the ignored location, to the attended standard stimulus, and to the ignored standard stimulus

was 1.58%, 4% and 0.15%, respectively. The mean reaction time (RT) was 540.8 ms (+36ms

SEM) to attended targets appearing in the left visual field and 501.5 ms (+27.8ms SEM) to

attended targets appearing in the right visual field. The difference between the two mean RTs is

not statistically significant. For the 4 subjects who performed the additional task of responding to

targets appearing in either the left or right visual field after the cueing period, the mean RT to

validly cued targets was significantly faster than to invalidly cued targets: 358.825 (56.67 ms

SEM) versus 550.422 (+104.21 ms SEM), (p=0.0078, Wilcoxon's signed-rank test),

demonstrating the behavioral benefit of cued preparatory attention.

5.3.2 ERP Analysis

Event-related potential (ERP) profiles elicited by the standard stimuli under the attend and

ignore conditions were compared. Waveforms from a single channel of a representative subject

are shown in Figure 5-2A. Consistent with previous reports, the P1 component (90-160 ms),

reflecting early response of the extrastriate cortex to stimulus input, is significantly higher for

attend versus ignore condition over the posterior electrodes (p = 0.0005, right hemisphere for

attend left versus ignore left and p = 0.00024, left hemisphere for attend right versus ignore

right); see Figure 5-2. The scalp distribution of the P1 attention effect has an amplitude

maximum over the lateral occipital sites contralateral to the visual field of stimulation for the









attend-left condition (Figure 5-3A). For the attend-right condition the pattern is more bilateral

(Figure 5-3B). Attentional enhancement of the N1 component (160-210 ms) over the parieto-

occipital areas was also observed but will not be pursued here. Further our emphasis below is on

the attentional modulation of prestimulus baseline activity and its relation to the early

stimulus-evoked P1 component.

5.3.3 Prestimulus Alpha Oscillations

Grand-average power spectra from the left and right occipital areas are shown in

Figures 5-4A and 5-4B, respectively, for three conditions: attend, ignore and fixation only. Data

used for the attend and ignore conditions were taken from the prestimulus time period, -300 to

0 ms, with 0 ms denoting the onset of the standard stimulus. For the fixation only condition,

continuous recording was divided into non-overlapping epochs of 300 ms in duration. The mean

spectral peak frequency averaged across subjects is 9.5 + 0.39 Hz and no frequency difference is

seen between conditions. The power of alpha activity for both attend and ignore conditions is

significantly lower than that for the fixation condition. Alpha power is further reduced with

attention when the attend condition is compared to the ignore condition (p<0.002). These effects

were established shortly after the cue onset and persisted until the onset of the stimulus. The

scalp distribution of alpha power difference between the conditions is shown in Figure 5-4. A

hemisphere-specific modulation is seen in Figure 5-4 where alpha activity was maximally

suppressed over the parieto-occipital areas contralateral to the attended visual field. Specifically,

for the right occipital areas, the alpha power was significantly lower when attention was directed

to the left visual field compared to when attention was directed to the right visual field, ignoring

the left visual field (p=0.0034 Wilcoxon's signed-rank test). For the left occipital areas, the alpha

band power was significantly lower when attention was directed to the right visual field

compared to when attention was directed to the left visual field, ignoring the right visual field









(p=0.0024 Wilcoxon's signed-rank test). Figures 5-4D and 5-4E show the comparison with the

fixation condition where hemispheric-specific lateralization patterns are also seen.

5.3.4 From Ongoing Activity to Evoked Response: A Model

Attention enhanced the stimulus-evoked P1 component and reduced alpha power prior to

stimulus onset. Presumably, the prestimulus alpha power reduction contributed to the

subsequently improved stimulus processing. The physiological mechanism that mediates such

improvement remains to be understood. The first step in our attempt to address this issue

consisted of putting forth a plausible model linking ongoing prestimulus activity with stimulus

evoked response and validating the model by testing its prediction. The model was then used to

delineate different mechanisms of prestimulus attention modulation.

5.3.4.1 The sigmoidal input-output function

Consider a neuronal ensemble in the sensory cortex. It is well-accepted that the level of

synaptic input S it receives is related to its output O measured in firing rate through a sigmoidal

function 0(S) in Figure 5-5A (Freeman, 1979; Destexhe et al., 2001). The synaptic input S can

come in two forms: external stimulus (exogenous) or other brain structures endogenouss). Let SN

denote the amount of endogenous prestimulus background activity at the time of stimulus onset

and Sx the amount of synaptic activity induced by the sensory input. From the schematic in

Figure 5-5A, the stimulus-evoked response is the difference: O(SN+Sx)-O(SN). For a constant

sensory stimulus, as is the case here, Sx is approximately the same from trial to trial over the

physiological range of S. The stimulus-evoked response is then proportional to the first

derivative of the sigmoidal input-output function 0(S), [O(SN+Sx)-O(SN)]/Sx, referred to as the

gain function. As shown in Figure 5-5B, the gain function has an inverted-U shape, suggesting

that SN has an effective operational range, denoted by W, beyond which stimulus input evokes a

negligible response. Also based on this model, it is worth noting that for a very strong input









(high Sx) the response to the stimulus may saturate for all levels of endogenous prestimulus

background activity hence giving rise to a purely decreasing gain with increasing SN. An

estimate of Wis provided by the distribution of SN on a trial by trial basis. Knowledge of this

distribution, combined with the magnitude of the stimulus-evoked response which reflects the

slope of the sigmoidal input-output function, forms the basis for evaluating how attention

modulates the sensory cortex to achieve enhanced input processing.

5.3.4.2 Model reformulation in terms of electroencephalogram variables

The abscissa and ordinate of the model in Figure 5-5 are synaptic input and firing rate.

While these variables can be precisely controlled and measured in computational models and in

slice preparations, they are not readily available in noninvasive human experiments. Thus, the

model in Figure 5-5 requires reformulation to be applicable to our EEG data. First, consider the

abscissa. Physiologically, EEG reflects the fluctuating extracellular field due to spatially

summated dendritic currents which are in turn caused by synaptic input (Niedermeyer and Lopes

Da Silva, 2005). There is evidence suggesting that the amplitude of EEG fluctuation is

proportional to the level of correlated synaptic input (Ho and Destexhe, 2000; Niedermeyer and

Lopes Da Silva, 2005). In this sense, the abscissa of the model in Figure 5-5A may be

equivalently expressed by the amplitude of EEG fluctuation, which is further replaced here by

the power of the alpha oscillations. Next, consider the ordinate. Firing rates are not accessible.

This means that the sigmoidal input-output curve during the ongoing brain state cannot be

directly established with EEG. But the consequence of this curve on stimulus processing can be

assessed via the early ERP component P1 (90-160 ms). Dipole modeling and imaging studies

have identified the ventral-lateral extrastriate cortex as the putative source for P1 (Heinze et al.,

1994; Clark and Hillyard, 1996; Mangun et al., 1997). Invasive recordings in animals revealed

that, upon receiving afferent input from the thalamus or lower visual areas, layer 4 stellate cells









fire action potentials which excite the supragranular pyramidal cells at their basal dendrites

spanning layer 3 or portion of layer 4 (Gilbert and Wiesel, 1983; Mitzdorf, 1985; Callaway,

1998; Thomson and Lamy, 2007). This activation causes an extracellular sink in layer 3/4 and an

extracellular source near the apical dendrite in layer 1/2, with the latter manifesting as a positive

going potential at nearby surface electrodes. See Figure 5-6A for an illustration. This dendritic

response in the superficial layers is the summated EPSPs to the impinging action potentials

arriving from layer 4 and the magnitude of dendritic response is proportional to the intensity of

the action potential input (Jagadeesh et al., 1993; Cash and Yuste, 1999; Araya et al., 2006). The

reformulated model in Figure 5-6B establishes the relation between prestimulus EEG power and

stimulus-evoked P1 amplitude, where an effective operational range Wfor baseline activity can

again be defined.

5.3.4.3 Experimental test of model predictions

In our experiment, although the stimulus is kept constant, the background synaptic activity

prior to stimulus onset, assessed by the amplitude of alpha oscillations, varies from trial to trial,

leading to significant trial-to-trial variability in stimulus-evoked responses. Given the model in

Figure 5-5, and its reformulation in terms of EEG variables in Figure 5-6, it is expected that the

amplitude of the P1 component follows an inverted-U function of the prestimulus alpha power.

This prediction is tested in Figure 5-7 for the attend condition. For each trial the amplitude of the

prestimulus (-300 to 0 ms) EEG activity in the frequency range of 8 to 12 Hz was estimated,

rank-ordered from the smallest to the largest, and sorted into equal-sized groups. The average

normalized P1 amplitude in each prestimulus power group was plotted against group index.

Consistent with the model prediction, an inverted-U relation is clearly seen. This relationship

was robust across all 12 subjects, with the exception of subject 2 and subject 10, where a

non-systematic and a negatively correlated relationship, respectively, were observed over the









right parieto-occipital regions. Quantitatively, for the left occipital areas, when the subject was

validly cued to the right visual field, the P1 amplitude for power group 3 (intermediate alpha

power) is 63% and 104% higher than that for power group 1 (lowest alpha power) and power

group 5 (highest alpha power) (p=0.0005 and p =0.00025 respectively, one-tail Wilcoxon's

signed-rank test). For the right occipital areas, when the subject is validly cued to the left visual

field, the normalized P1 amplitude for power group 3 is 57% and 165% higher than that for

power group 1 and power group 5 (p=0.02 and p = 0.0025 respectively, one-tail Wilcoxon's

signed-rank test). Quadratic fits appear as solid smooth curves in Figure 5-7 to guide the eye.

5.3.5 Attentional Modulation of Baseline Ongoing Activity: Possible Mechanisms

Based on the model in Figures 5-6 and results in Figure 5-7, it is plausible that deployment

of effortful covert attention through top-down compensatory mechanisms modulates the

prestimulus ongoing neural activity, hence optimizing the cortical excitability (Bressler et al.,

2008; Fischer et al., 2008; Lakatos et al., 2008) to increase the overall slope of the sigmoidal

function O(S) in Figures 5-5 and 5-6, as illustrated in Figure 5-8A. This global change in the

sigmoidal function, corresponding to an increase in global gain, can account for both reduced

alpha power and enhanced P1 by attention (Figure 5-8A). Two predictions can be made

according to Figure 5-8A. First, the attention-induced increase in local gain is not uniformly

distributed over the level of the prestimulus EEG power. Specifically, the ratio between the two

gain functions (attend versus ignore) has again an inverted-U shape, as shown in Figure 5-8B.

Second, the overall increased steepness in the slope (gain) is accompanied by a reduced effective

operational range W, from W1 to W2, implying a reduction in the trial-to-trial variability of

prestimulus alpha activity by attention to maximize stimulus-evoked response.

Test of prediction 1: The data in Figures 5-7A and 5-7B, together with the data obtained

in the same way from the ignore condition, are plotted in Figures 5-9A and 5-9B. It is evident









that the gain reflecting the slope of the sigmoidal function is enhanced for the attend condition

relative to the ignore condition. The ratios of Pl amplitude between the two conditions for

different prestimulus alpha power groups are shown in Figures 5-9C and 5-9D. As expected, the

gain ratio is maximal for the intermediate power group, in agreement with the model prediction

in Figure 5-8B. Quantitatively, for the left occipital areas, the P1 amplitude ratio for power group

3 (intermediate alpha power) is 52% and 68% higher than that for power group 1 (lowest alpha

power) and power group 5 (highest alpha power). For the right occipital areas, the P1 amplitude

ratio for power group 3 is 16% and 60% higher than that for power group 1 and power group 5.

It is worth noting that the P1 component is relatively small for the ignore condition. Its

subensemble estimation is further complicated by the fact that each power group in a given

subject has only about 30 trials. These factors may explain the weak and sometimes even the

lack of inverted-U appearance in Figures 5-9A and 5-9B, respectively, for the ignore condition.

Test of prediction 2: Alpha band power in the interval -300 to 0 ms was estimated on a

single-trial basis for each of the three conditions over the parieto-occipital sites and log-

transformed. Figure 5-10A shows the Gaussian fits to the empirical distributions from a

representative subject. The left-ward shift of the distributions under task conditions corresponds

to the reduced mean alpha power reported in Figures 5-4 and 5-5. In addition, both the attend and

ignore conditions have narrower alpha power distributions relative to that of the fixation

condition, with the distribution for the attend condition further narrowed from that for the ignore

condition. Figures 5-10B, 5-10C and 5-10D summarizes the variance of the empirical alpha

distribution across all subjects for the attend and ignore conditions. The variance reduction when

attention is deployed to the contralateral visual field (attend) relative to the ipsilateral visual field

(ignore) was significant at p = 0.003 over both hemispheres, and the reduction relative to the









fixation condition was significant at p = 0.0005. The variance over the left hemisphere for the

attend-right condition was significantly lower than that for the attend-left condition and for the

fixation condition at p = 0.00075 and 0.0005 respectively. The variance over the right

hemisphere for the attend-left condition was lower than that for the attend-right condition and for

the fixation condition at p = 0.088 and 0.021 respectively. In addition, the variance for the ignore

condition collapsed over both hemispheres is lower than that for the fixation condition at

p= 0.01.

5.3.6 Role of Excitability: Validation in a Single Neuron Model

A computational model was constructed as described in section 5.2.8. The firing rates were

collected and averaged over 40 simulation trials of 200 ms duration each, for 25 increasing

values of oi (variance of inhibitory synaptic input) in the range of 0-0.0375pS, for 8 levels of

increasing excitability (mean membrane depolarization. Firing rate over the 200 ms interval was

averaged across trials and plotted as a function of the variance of background synaptic activity in

Figure 5-11A for three levels of excitability: low excitability (less depolarized, red curve);

intermediate excitability (more depolarized, black curve) and high excitability (excessively

depolarized, blue curve). The derivatives of all three sigmoidal functions are given in

Figure 5-1 1B. Notice from Figure 5-11A and 5-11B that a moderate increase in depolarization

increases gain, but too much excitability causes a decrease in gain. This nonlinear relation is

further demonstrated by plotting the firing rate as a function of all 8 levels of excitability and the

25 levels of variance of background synaptic activity in Figure 5-11C. The average gain, defined

as the weighted mean of the inverted-U function in Figure 5-11B, is plotted for each level of

excitability in Figure 5-1 ID, which suggests the existence of an optimal range of excitability for

enhanced stimulus processing. The results of this modeling study confirm the influence of

excitability (mean membrane depolarization) on the gain curve, where within optimal ranges, an









increase in excitability effects an increase in gain. The results obtain through this modeling study

is in complete agreement with the observations of attentional modulation at the scalp EEG level

hence providing further validation of the proposed model.

5.4 Discussion

We considered the relation between visually evoked response and prestimulus alpha

activity. Three results were found: (1) the P1 amplitude is shown to be an inverted-U function of

the prestimulus alpha power (Figure 5-7), providing key evidence for the validity of the

sigmoidal function model in Figures 5-5 and 5-6; (2) attentional enhancement of the P1

amplitude is interpreted as reflecting an increase in the overall slope of the sigmoidal function;

and (3) in addition to amplitude reduction, the trial-to-trial variability of alpha is also reduced by

attention, corresponding to a reduced working range as a consequence of the attentional

modulation of the sigmoidal function (Figure 5-8). In light of these results some of the seemingly

conflicting observations in relation to alpha oscillations and stimulus processing may be

reconsidered. First, in a given behavioral state (e.g. attend, ignore, or fixation), the inverted-U

relation in Figures 5-6 and 5-7 indicates that intermediate levels of prestimulus alpha activity

lead to strongest stimulus-evoked response. This may explain why both high and low levels of

prestimulus alpha activity are not conducive to the detection of a threshold level stimulus

(Ergenoglu et al., 2004; Babiloni et al., 2006; Romei et al., 2008b; Romei et al., 2008a). This

explanation, together with reports in the somatosensory domain where the rate of detecting a

weak stimulus is highest for intermediate levels of mu activity (Linkenkaer-Hansen et al., 2004;

Zhang and Ding, in press), appears to suggest a mechanism of sensory processing that is

modality independent. Second, attentional reduction of alpha power, possibly mediated by top-

down compensatory mechanisms, is taken to indicate the increased overall slope of the sigmoidal









function and hence the higher global gain to input, and should not be extrapolated to imply that

prestimulus alpha should be inversely proportional to stimulus evoked response.

5.4.1 Theoretical Model

The state of the brain fluctuates in a seemingly random manner. The fate of a sensory

stimulus depends on the waxing and waning of the physiological variables characterizing these

fluctuations (Lakatos et al., 2005; Lakatos et al., 2008). While behavior such as attention

modulates the global dynamics of these variables (Foxe et al., 1998; Sauseng et al., 2005; Thut et

al., 2006; Fischer et al., 2008), on a given trial, the processing of a stimulus is affected by the

magnitude of these variables immediately preceding its onset. In this work, the global state of a

neuronal ensemble is assumed to be defined by a sigmoidal function, where the horizontal axis is

the level of background synaptic activity over a suitably defined time interval and the vertical

axis is the output firing rate over the same interval. On a trial by trial basis, the stimulus-evoked

response is proportional to the derivative of the sigmoidal curve referred to as local gain, which

is an inverted-U function of background synaptic activity level preceding the stimulus

(Figure 5-5). Attention is hypothesized to enhance the global gain to sensory input by increasing

the overall slope of the sigmoidal function. This model, while inspired by recent computational

studies on the impact of mean and variance of membrane potential fluctuations on a neuron's

responsiveness to input (Ho and Destexhe, 2000; Chance et al., 2002; McCormick et al., 2003;

Wolfart et al., 2005; Haider et al., 2007), differs from these models in two important ways. First,

the horizontal axis in Figure 5-5, rather than being the external input of varying intensity, is the

level of moment by moment synaptic background activity assessable by dendritic field

potentials. Second, an explicit mechanism is given, which stipulates that the stimulus-evoked

response is an inverted-U function of the level of background synaptic activity immediately

preceding stimulus onset.









5.4.2 Model Testing and Validation

In Figure 5-6, the model in Figure 5-5 is reformulated in which the amplitude of the P1

component is equated with the intensity of stimulus-evoked firing and the prestimulus alpha

power with the level of background synaptic activity. A single-trial based analysis was carried

out, and as predicted, an inverted-U relationship between the P1 amplitude and the prestimulus

alpha power was found for both hemispheres (Figure 5-7). Considered a validation of the model,

this finding suggests that the observed increase in P1 amplitude to the attended stimulus reflects

increased global gain, which is defined by an increase in the overall slope of the sigmoidal

function (Figures 5-8 and 5-9). The change in the sigmoidal function further predicted a possible

leftward shift of the sigmoidal function and a reduced effective operational range, both of which

were confirmed by computing and comparing across experimental conditions the mean

prestimulus alpha power (Figure 5-4) and the variance of the alpha power distribution estimated

on a trial by trial basis (Figure 5-10). It is worth noting that qualitatively similar results were

observed when alpha power was replaced by total signal variance to index background activity.

The alpha activity was chosen as a proxy for the strength of background synaptic activity

because (1) all subjects exhibited prominent spectral power in this band accounting for most of

the total signal variance and (2) robust attentional modulation was observed in this band. In our

data attentional modulation in the other frequency bands including theta, beta and gamma was

either completely absent or were not statistically significant across subjects. In addition, power in

higher frequency bands such as beta and gamma are too small to serve as a representative of total

synaptic background activity.

5.4.3 P1 Genesis and Attentional Modulation

The visually evoked P1 component reflects the early response of the extrastriate cortex and

is enhanced by spatial attention. See Figures 5-2 and 5-3. Concerning the generating mechanisms









of Pl, Mitzdorf (Mitzdorf, 1985) showed in cats that a surface positive potential may arise due to

excitatory synaptic activation of the supragranular pyramidal cells at their basal dendrites

spanning layer 3 or portion of layer 4, which causes an extracellular sink at layer 4 and a

superficial source sensed by the surface electrode. In behaving macaques, Schroeder et al.

(Schroeder et al., 1991) localized the P60 component, the monkey equivalent of the P1

component in humans, to pattern evoked and pattern reversal VEP to layer 2/3. Associated with

the P60 was an increase in MUA activity signifying that it is predominantly a depolarization of

the supragranular pyramidal cells in the superficial layers. The excitatory input to the superficial

layers is considered to be from the layer 4 stellate cells which receive afferent input from either

thalamocortical projections or inputs from earlier sensory areas (Figure 5-6A). According to

these results, the increased P1 amplitude by attention reflects the increased firing of neuronal

populations in layer 4, which is the basis for the choice of Pl for our analysis. The origins of the

N1 and other later evoked components (Clark and Hillyard, 1996; Di Russo et al., 2002;

Di Russo et al., 2003) are more complex. Their generation, involving reentrant processing and

the interaction among multiple brain areas (Garrido et al., 2007), depends to a limited degree on

visual alpha. These components are not pursued further.

5.4.4 Alpha Oscillation: Neuronal Mechanisms and Attentional Modulation

The genesis of alpha oscillations in visual cortex is debated. While early investigators

focused on the pacemaking role of the thalamus, more recent work stresses the cortical origin of

the alpha oscillations. Bollimunta et al.(2008), analyzing laminar field potentials and multiunit

activity from extrastriate cortical areas V2 and V4 of behaving macaques, found local generators

of alpha in superficial layers, layer 4 and deep layers, and showed that the activities of these

generators are highly synchronized. According to this view, the widely observed decrease of

EEG alpha power in areas contralateral to the attended visual field (Figures 5-4), likely reflects









similar decreases in the entire cortical column, allowing us to establish a link between scalp

measured alpha power and layer 4 neuronal firing, the latter being proportional to the magnitude

of the P1 component.

Physiologically, the suppression of alpha is taken to imply increased cortical excitability

(Foxe et al., 1998; Bastiaansen and Brunia, 2001; Marrufo et al., 2001; Sauseng et al., 2005;

Thut et al., 2006; Klimesch et al., 2007) (Jones et al., 2000; Worden et al., 2000; Goldman et al.,

2002; Kelly et al., 2006; Rihs et al., 2007; Jones et al., 2008; Romei et al., 2008a). Cortical

excitability is often understood in terms of mean depolarization levels of cellular membrane

potentials and it is known to have a complex relation with stimulus-evoked response. For

example, the absence or very low levels of spontaneous oscillatory activity may fail to bring

local neuron populations closer to firing threshold and excessive depolarization due to high

levels of spontaneous activity may cause short-term synaptic depression, both of which can lead

to diminished response to sensory input (Abbott et al., 1997; Chung et al., 2002; Petersen et al.,

2003; Dehaene and Changeux, 2005). While enhanced stimulus processing by attention in our

experiment is thought to be mediated by increase in the slope of the sigmoidal function or gain,

as defined in Figure 5-8, excitability and gain in the current framework are not mutually

exclusive concepts, and their relationship was explored in a computational study (Figure 5-11).

The results show that (1) there is an optimal range in the mean membrane depolarization which

gives rise to highest overall slope of the sigmoidal function, and (2) both insufficient and

excessive depolarization reduce the slope of the sigmoidal function, thereby hindering the gain to

sensory input. This computational study, together with the experimental data, suggests that (1)

stimulus processing benefits directly from increased global gain and (2) attention optimally

modulates excitability to enable such gain increase.























t 200 1800-2200 t 100 2800+0/200/400 t 200 (ms)
Cue Preparatory attention period Imperative ISI Cue

Figure 5-1. Experimental task. At the beginning of each trial, the cue (left/right pointing arrow)
presented at the fixation instructs the subjects to attend to one of two demarcated
locations. After 1800 to 2200 ms a unilateral imperative stimulus (target/standard) is
presented on one of the locations. The subjects had to indicate by a speeded response
to the occurrence of the target stimulus at the attended location.











A B
Attend
I- gnore 3.5

3

2.5

2-

1.5
I I II

-1 0V
100 ms 0.5

S ATR IGR ATL IGL
Left Occipital areas Right Occipital areas
C D

1 1
0.9 0.9
0.8 0.8
0.7 0.7
1 0.6 1 0.6
0.5 0.5
B B
I 0.4 N 0.4
S0.3 0.3
o o
0.2 0.2
0.1 0.1

1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12
Subjects Subjects


Figure 5-2. ERP attention effect. (A) Typical waveform of visual ERP over the parieto-occipital
region contralateral to the visual field of stimulation. Note that attending to the
stimulus location produces an enhanced P1 and N1 component. (B) Grand average
peak P1 amplitudes averaged over channels overlying the occipital / parieto-occipital
areas contralateral to visual field of stimulation, indicating attentional enhancement
when the stimulus location was attended relative to ignored. ATL -attend left, IGL -
ignore left; ATR -attend right, IGR -ignore right. Panels (C) and (D) summarizes the
normalized visual evoked P1 component over 12 subjects over the left parieto-
occipital areas (C) and right parieto-occipital areas (D) when attention was paid to the
contralateral visual field (filled black-Attend), contralateral visual field was ignored
(filled white-Ignore). Normalization was performed only for display purposes and
was obtained by dividing the respective conditions by the sum of estimates of both
conditions.











A B
1.5
1
0.8
0.6
l ** 0.4
0.5 0.2

0
0 *-0.2
-0.4
-0.5 -0.6
-0.8
-1 -1


Figure 5-3. Scalp distribution of attentional enhancement. Difference ERPs of grand average
over 12 subjects at 120 ms after presentation of the standard stimulus. Note the
posterior P1 (shaded red) and fronto-central N1 (shaded blue) attention effect. (A)
Attend minus ignore difference ERP for left visual field input and (B) Attend minus
ignore difference ERP for right visual field input.










A B
6 -Attend I Attend
5 -Ignore 5 Ignore
M -Fixation Fixation


22
2 I
C1 A

10 20 30 40 10 20 30 40
Frequency (Hz) Frequency (Hz)

C D E











-0.5 0 0.5 -2.5 -2 -1.5 -1 -5 0

Figure 5-4. Attentional modulation of alpha band activity. Grand average power spectrum over
12 subjects over left parieto-occipital (A) and right parieto-occipital areas (B) when
attention was paid to the contralateral visual field (black), contralateral visual field
was ignored (red) and for relaxed fixation condition (blue). (C)-(E) highlights the
scalp topography of difference in alpha band power (-300 to 0 ms prior to standard
stimulus) averaged across subjects. (C) Attend left minus attend right, blue shading
over the right parieto-occipital areas indicate lower alpha band power to attend left
(contralateral visual field) relative to attend right (ipsilateral visual field) and red
shading over the left parieto-occipital areas indicate lower alpha band power to attend
right relative (contralateral visual field) to attend left (ipsilateral visual field). (D)
Attend left (ATL) minus relaxed fixation condition (FIX), blue shading over the right
parieto-occipital areas indicate a reduction in the alpha band with attention as against
relaxed fixation. (E) Attend right (ATR) minus relaxed fixation condition, both left
and right parietal-occipital regions show reduction in alpha band power (blue
shading) when attention is directed to the right visual field as compared to relaxed
fixation condition.










A B

Stimulus evoked
response Input-output
curve



response
O(SN+S.) :3
Stimul -------
evoked
response II
O(SN)



SN Sx Synaptic Input (S) Synaptic Input (S)


Figure 5-5. Schematic illustration of theoretical model. (A) For a neuronal ensemble the
spontaneous output firing rate O measured over a suitably defined time interval is a
sigmoidal function of endogenous background synaptic activity level SN measured
over the same interval. Letting SX denote stimulus-induced synaptic activity, the
stimulus-evoked firing rate, O(SN+SX) O(SN), is proportional to the derivative
(local gain) of the sigmoidal curve, [O(SN+SX) O(SN)] / SX, shown as the red
curve in (B). In this model, when the stimulus is kept constant, the stimulus-evoked
response is an inverted-U function of the prestimulus level of background synaptic
activity. The variable W in (B) defines an effective range of prestimulus level of
synaptic activity beyond which stimulus evokes a negligible response.













0 0 0 00
A oB




P1 amplitude / *
0 .I \ Input-output
Curve
0o II / A
-oo

III \"
--v /


Cortical
feedforward V
input W

VI |Prestimulus EEG Power


Figure 5-6. Model reformulation in terms of EEG variables. (A) Neural genesis of visually
evoked P1. Feedforward activation of layer 4 stellate cells excites the basal dendrites
of the superficial layer pyramidal cells, leading to an extracellular source comprised
mainly of passive return current near the cortical surface, which is sensed by the scalp
electrode as a positive-going ERP component. (B) Reformulated model in which
prestimulus EEG power in alpha band is equated with level of background synaptic
activity and magnitude of Pl with intensity of stimulus-evoked response. This model
predicts that the P amplitude is an inverted-U function of prestimulus alpha power
with the variable W defining a range of effective operation.












0.9 F


1 2 3 4 5
Alpha Power Group


1 2 3 4
Alpha Power Group


Figure 5-7. Relationship between prestimulus alpha power and stimulus processing. Trials with
attended standard stimuli were considered (valid trials). Single trials were rank
ordered by the magnitude of prestimulus alpha power (-300 ms to 0 ms) and sorted
into 5 groups of equal size. The P1 component was determined for each group,
normalized and averaged across subjects, and plotted as function of group index in
(A) for left parieto-occipital areas and (B) for right parieto-occipital areas. An
inverted-U relationship is clearly seen where the solid lines represent quadratic fits.
Error bars are the standard error of the mean.


Sy W E
Synaptic Input (Prestimulus EEG Power)


Synaptic Input
(Prestimulus EEG Power)


--- Ignore Ignore (Gain)
--- Attend Attend (Gain)


Figure 5-8. Schematic illustration of attentional modulation mechanisms of ongoing activity. (A)
Attention shifts the sigmoidal curve in Figures 4 and 5 left-ward and increases its
overall slope (global gain). (B) The ratio of the gain curves in (A) for attend and
ignore conditions is again an inverted-U function.


A








O
-o
5.6
E

a-
0


N


0
z


0.9

0.8

0.7

0.6















- Attend
- Ignore


0.9 r


0
S0.8

E
U 0.7
9-
'a
N 0.6


o 0.5
z


1 2 3 4
Alpha Power Group


1 2 3 4 5
Alpha Power Group


1.8

1.7

S1.6


*0
Cm1.5

C 1.4
0
1.3


. 1.1

0 1

0.9


- Attend
- Ignore


! 3 4
Alpha Power Group


1 2 3 4 5
Alpha Power Group


Figure 5-9. Gain ratio as a function of prestimulus alpha. P1 amplitude as a function of
prestimulus alpha power is plotted for both attend and ignore conditions in (A) and
(B). The ratios between the two curves are shown in (C) and (D). Maximum gain is
achieved for intermediate levels of prestimulus alpha activity.


0.8


0.7


0.6


C
1.7

1.6
0
C 1.5

*I 1.4





1.1


0.9
0.9


[










A B
0.1 Attend
Ignore
Fixation 0.5
0.08



2 ^6 1, I I
2 0.4
a 0.04
z
0.02


-20 -10 0 10 20 0.3
Alpha Power (dB) ATR IGR ATL IGL
Left Occipital areas Right Occipital areas


Figure 5-10. Distribution of alpha power. (A) Distributions of log-transformed single trial alpha
power for attend, ignore and fixation conditions over left parieto-occipital areas from
a representative subject. Variance of the single trial alpha power distribution is
compared over subjects between attend and ignore conditions in (B) where a
reduction in variance is evidence of decreased trial-to-trial alpha power variability.



























Background synaptic ac
- Low Excitability


S 0.6

0.4


0.24

20 25
tivity
- Intermediate Excitabilitl


Background synaptic activity
y High Excitability


25
100 15 ac9 i
5 ouod 011 toP~'


1 2 3 4 5 6 7
Membrane depolarization


Figure 5-11. Computational study of relation between excitability and gain. (A) Firing rate over
a 200 ms interval as function of background synaptic level for three levels of mean
membrane depolarization (blue: high excitability, black: intermediate excitability, and
red: lower excitability). (B) Derivative of the sigmoidal curves in (A) showing
nonlinear relationship between gain and excitability. (C) Firing rate as a function of
mean membrane depolarization levels and variance of background synaptic activity.
(D) Average gain for increasing levels of membrane depolarization.









CHAPTER 6
TOP-DOWN CONTROL OF HUMAN VISUAL CORTEX IN ANTICIPATORY SPATIAL
VISUAL ATTENTION

6.1 Introduction

Early studies into the neuronal mechanisms of attention focus on attention's role in the

selective processing of stimuli. How attention modulates endogenous anticipatory/preparatory

processes that precede the appearance of stimuli is less known. Emerging reports indicate that

preceding changes in stimulus-evoked response properties, deployment of attention modulates

the excitability of the sensory areas that facilitate subsequent stimulus processing (Liang et al.,

2002; Dehaene and Changeux, 2005; Padilla et al., 2006; Weissman et al., 2006; Bestmann et al.,

2007; Eichele et al., 2008; Zhang et al., 2008b; Gregoriou et al., 2009; Zhang and Ding, in

press). However, such goal-directed prestimulus modulation of sensory cortical activity is not an

inherent property of the early sensory areas. Instead, it is believed that the frontal-parietal

network facilitates an enhancement of the excitability of the task-relevant and feature-specific

visual areas and perhaps a suppression of activity in task-irrelevant visual areas in advance of

visual stimulation by modulating its baseline activity (Chelazzi et al., 1993; Luck et al., 1997;

Hillyard et al., 1998; Kastner et al., 1999; Hopfinger et al., 2000; Kastner and Ungerleider, 2001;

Corbetta and Shulman, 2002; Gazzaley et al., 2005; Giesbrecht et al., 2006; Grent-'T-Jong and

Woldorff, 2007; Klimesch et al., 2007). In line with this notion, it has been identified that

deployment of covert spatial visual attention is associated with increased activation in a network

involving dorsal frontal and parietal cortices with concomitant increases in activation over

retinotopic early visual areas (Desimone, 1996; Gitelman et al., 1999; Kastner et al., 1999;

Hopfinger et al., 2000; Corbetta and Shulman, 2002; Giesbrecht et al., 2003).

While prior neuroimaging and electrophysiological studies of the top-down control of

spatial visual attention have offered insights into the areas in the brain that exert control over









stimulus processing, the neural mechanisms by which this is implemented are still unclear. Apart

from prestimulus modulation of activity over visual areas that facilitate stimulus processing,

putatively induced by top-down signals from the frontal-parietal network, empirical evidence of

such inter-areal interaction is lacking with the exception of (Fuster et al., 1985; Siegel et al.,

2008; Gregoriou et al., 2009). However, recent microstimulation and rTMS studies in monkeys

and humans respectively on frontal eye fields (FEF) and intraparietal sulcus (IPS) have provided

casual evidence of top-down influences of these areas on activity in the visual cortices

(Armstrong et al., 2006; Taylor et al., 2007; Capotosto et al., 2009) during anticipatory attention.

Additional evidence of frontal dependent tonic top-down modulatory influence on visual

processing in humans comes from lesion studies (Knight, 1997; Knight et al., 1999; Barcelo et

al., 2000; Yago et al., 2004) which is also supported by anatomical evidences of neural substrates

that enable such long-distant communication coming from axonal tract-tracing and diffusion

tensor imaging in animals and humans (Ungerleider et al., 1989; Felleman and Van Essen, 1991;

Petrides and Pandya, 1999; Barbas, 2000; Rempel-Clower and Barbas, 2000; Petrides and

Pandya, 2002).

To identify the neural signals that mediate the interaction between the fronto-parietal and

early visual areas during anticipatory attention, we studied the activity during the delay period

between an attention directing cue and the subsequent imperative stimulus from human

volunteers performing a trial-by-trial cued spatial visual attention task. Physiologically, EEG

field potential oscillations reflect the rhythmic shifting in excitability due to fluctuating

extracellular dendritic currents which are in turn caused by synaptic input (Niedermeyer and

Lopes Da Silva, 2005). Recent studies on animals and humans have shown that the alpha rhythm

(8-12 Hz) over the visual areas is attenuated in a topographically-specific fashion dependent on









the hemifield of attentional cueing and this state of reduced alpha activity is considered to reflect

increased cortical excitability (Foxe et al., 1998; Worden et al., 2000; Bastiaansen and Brunia,

2001; Marrufo et al., 2001; Sauseng et al., 2005; Kelly et al., 2006; Thut et al., 2006; Jokisch and

Jensen, 2007; Klimesch et al., 2007; Rihs et al., 2007; Romei et al., 2008b; Romei et al., 2008a;

Van Dijk et al., 2008). While synchronization of neuronal oscillations within and across cortical

areas has long been suggested as a potent mechanism of attentional selection and sensory gating

(Salinas and Sejnowski, 2000; von Stein et al., 2000; Engel et al., 2001; Fries et al., 2001;

Salinas and Sejnowski, 2001; Fries, 2005; Buschman and Miller, 2007; Lakatos et al., 2007;

Melloni et al., 2007; Lakatos et al., 2008; Rajagovindan and Ding, 2008b; Rajagovindan and

Ding, 2008a; Siegel et al., 2008; Gregoriou et al., 2009), evidence as to whether and how

selective attention modulates the rhythmic synchronization between frontal, parietal, and early

visual regions in the human brain remains sparse (Doesburg et al., 2008; Siegel et al., 2008).

Further the exact neural mechanism of how this increased excitability over the visual sensory

areas is brought about has not been clearly established. The goal of this study is to examine the

neural mechanism by which the visual cortical excitability is controlled by the frontal-parietal

network. In light of prior anatomical and lesion evidence (Gonchar and Burkhalter, 2003; Knight

et al., 1999), two competing hypotheses may be put forth: increased visual excitability and

reduced alpha rhythm is mediated through (1) increased excitatory top-down input to local

pyramidal neurons or (2) decreased excitatory top-down input to local inhibitory interneurons. In

particular, three questions are considered: (1) what is the signal that implements the frontal-

parietal control in visual cortex, (2) how functional connectivity between executive and sensory

areas is modulated by attention, and (3) whether top-down control of visual activity is transient

or sustained. These questions were addressed by recording scalp EEG from healthy volunteers









performing a trial-by-trial cued spatial visual attention task and by performing multivariate

spectral analysis to assess the regional and the inter-areal oscillatory synchronization and causal

interaction between the fronto-parietal and the visual areas during a period of sustained covert

attention.

6.2 Materials and Methods

6.2.1 Subjects

The experimental protocol was approved by the University of Florida Institutional Review

Board. Nineteen subjects, free from movement and neurological disorders and with normal or

corrected-to-normal vision, gave written informed consent and participated in the study. Data

from twelve subjects (20-31 years of age, 25.16 +3.5 years, 5 females and 7 males, all right

handed) were included in the analyses. The remaining 7 participants were excluded due to poor

performance or excessive artifacts.

6.2.2 Paradigm

Seated in a dimly lit, acoustically and electrically shielded room, the subject was instructed

to fixate on a central cross on a VGA monitor placed approximately 85 cm from the eye. Two

square boxes, each delineated by white dots placed at the four corners in the upper left visual

field and in the upper right visual field, were displayed permanently on the monitor (Figure 6-1).

Each box, centered 4.70 above the horizontal meridian and 7.20 lateral to the central fixation,

subtended a visual angle of 3.35. A trial began with the onset of a cue, which consisted of either

a left or right pointing arrow of 200 ms in duration at the location of the central fixation cross,

instructing the subject to deploy covert attention to the square box on the side indicated by the

arrow. After a random time delay between 1800 ms and 2200 ms, a standard or a target stimulus

of 100 ms in duration appeared either inside the attended square box (valid trial) or inside the

square box on the opposite side (invalid trial). The standard stimulus was a circular checkerboard









subtending a 3.30 visual angle, and the target stimulus was also a circular checkerboard, but of a

slightly smaller diameter. The subject was required to press a button with their right index finger

in response to the valid target stimulus and withhold response to any other stimuli. The standards

appeared 80% of the time with 50% validity and the targets appeared 20% of the time with 66%

validity. The interval between the cue offset of the previous trial and the cue onset of the

successive trial was randomly varied between 4700 to 5700 ms. The entire experiment was

comprised of 15 to 16 blocks of approximately 6 min each (60 trials). Breaks were given

between blocks. Subjects received practice sessions of 150 trials to familiarize themselves with

the task and to minimize the effect of learning.

6.2.3 Data Acquisition

The electroencephalogram (EEG) data was recorded with a fully DC-coupled 128-channel

BioSemi ActiveTwo System at a sampling rate of 1024 Hz. Four additional electrooculogram

(EOG) electrodes were placed around the eyes to monitor eye blinks and horizontal and vertical

eye movements. EEG and EOG data was amplified and lowpass-filtered with a cutoff set at 205

Hz. The stimulus and the response was delivered and registered by a BeriSoft Experimental Run

Time System (ERTS) and by a Berisoft EXKEY microprocessor logic pad respectively.

6.2.4 Data Preprocessing

Analysis was performed using BESA 5.2 and custom scripts written in MATLAB 7.4. The

EEG data was low-pass filtered off-line with a cutoff set at 83 Hz and downsampled to 250 Hz.

The channels were re-referenced off-line against the average reference (Nunez et al., 1997;

Ferree, 2006). Trials contaminated by muscle artifacts, movement artifacts, or excessive eye

blinks were excluded from further analysis. Trials with correct and incorrect behavioral

responses to validly cued target stimulus were analyzed separately. The voltage time series was

transformed into continuous current source density (CSD) estimates (Perrin et al., 1989; Nunez









et al., 1997; Scherg et al., 2002) to mitigate the adverse impact of volume conduction and

common reference on functional connectivity measures. The continuous CSD data was epoched

to yield two datasets (a) from -300 ms to 1500 ms around the cue onset with 0 ms indicating the

presentation of the left-pointing or right-pointing arrow and (b) from -300 ms to 0 ms with 0 ms

indicating the onset of the imperative stimuli (standards and targets).

6.2.5 Data Analysis

Spectral measures: Three analyses were performed on the EEG data. The multivariate

autoregressive (MVAR) modeling method was used in all three analysis (Ding et al., 2006). For

an ensemble of trials at each recording site the ensemble mean was removed from each trial. This

allowed the residuals to be treated as coming from a zero mean stochastic process required by

MVAR data modeling. Model order between 23 and 25 were found to be optimal across subjects,

determined both by the Akaike information criterion and comparing the spectral estimates

obtained by Fourier based methods (Thomson, 1982; Mitra and Pesaran, 1999). Hence a fixed

model order of 24 was chosen for all subjects. Three spectral quantities were derived from the

model: power, coherence, and Granger causality. The regional oscillatory activity quantified by

the power was averaged across the preselected channels overlying the regions of interest and

further averaged within the frequency bands of interest. The inter-areal coherence was obtained

by averaging the coherence across all pairs of sites representing the different regions of interest

and further averaged within a given frequency band. Granger causality was calculated in which

the set of channels representing a given area (frontal, parietal, or visual) was treated as a block

(Wang et al., 2007b). From the model, Granger causality spectra for Frontal -Occipital,

Parietal- Occipital, Occipital -Frontal, Occipital -Parietal, Frontal -Parietal and

Parietal -Frontal were derived and averaged over the frequency band of interest. As with any

other physiological variables, power, coherence and Granger causality values varied markedly









from subject to subject. To remove the possible adverse effect of this variability on population

averaging, the spectral measures for each condition were first normalized by dividing by the

corresponding measures from the ignore condition, and then averaged across subjects. The

Wilcoxon's rank-sum test was then applied to assess the statistical significance of the differences

between conditions.

Analysis 1: Attentional modulation of power, coherence and Granger causality in the time

period -300 ms to 0 ms prior to the onset of the imperative stimulus was analyzed for trials with

correct behavioral responses. Five regions of interest were identified: anterior medial frontal,

medial frontal, parietal and lateral occipital areas where each area is represented by six

neighboring channels (Figure 6-2D).

Analysis 2: The behavioral consequences of anticipatory oscillatory activity were

examined. Trials with validly cued targets were rank-ordered according to the reaction time (RT)

and divided into two non-overlapping groups of equal size: fast RT group and slow RT group.

Error trials were given a reaction time of infinity. Spectral measures over a period of -300 to

0 ms prior to stimulus onset were estimated and the Wilcoxon's rank-sum test was then applied

to assess the statistical significance of the differences in the ongoing oscillatory activity

preceding the two classes of trials.

Analysis 3: Temporal functions of power, coherence and Granger causality were estimated

using a sliding window approach. Each window was 140 ms (35 sample points) in length and

stepped forward in 4 ms increment. A total of 416 windows resulted for the time period -300 ms

to 1500 ms around the onset of the attention directing cue. The time designation of each window

corresponded to the middle of the covered interval.









Significance testing: For interdependence measures the following random permutation

procedure was adopted for statistical significance assessment (Nichols and Holmes, 2002;

Brovelli et al., 2004). (1) Consider the data from two recording sites (A and B). Randomly

pairing data for site A with data for site B from a different trial leads to the formation of a

synthetic ensemble of trials for which there is no interdependence. Performing such random

pairing with many different permutations (500 permutations used for this study) will result in the

null-hypothesis distribution for the interdependence quantity of interest from which a threshold

can be derived that corresponds to a given significance level (a = 0.01 chosen for this study). The

value from the actual data is compared with the threshold for the assessment of statistical

significance. This procedure was implemented for the coherence measures over each subject

during the prestimulus time period for Analysis 3. Eleven out of the twelve subjects exhibited

significant alpha band coherence between the frontal, parietal and occipital areas, revealing a

coherently oscillating network. Granger causality was then estimated between the regions of

interest over the eleven subjects who exhibited coherent oscillations in the alpha band between

areas implicated in the visual attention network. (2) Between the attend and ignore conditions

there are two temporal functions of averaged alpha band spectral estimates for Analysis 3. The

statistical significance of their difference was assessed by employing the Wilcoxon's sign-rank

test over each sliding window. The differences were deemed statistically significant only if the

differences were significant over at least 34 consecutive sliding windows to minimize potential

false positives due to multiple comparisons. Time periods of significant difference were

indicated on the plots for comparison.

6.3 Results

Twelve subjects performed the task according to instructions. The average target detection

rate was 87% and the mean reaction time (RT) was 521 ms. The average false alarm rate









stemming from (1) response to the target appearing in the ignored location, (2) response to the

attended standard stimulus, and (3) response to the ignored standard stimulus was 1.6%, 4.0%

and 0.2%, respectively.

Analysis 1: For the time period -300 to 0 ms prior to the onset of the standard stimulus all

subjects exhibited prominent power spectral peaks in the alpha band (8-12 Hz) accounting for

most of the total signal variance. No consistent power spectral peaks were seen in theta (4-8 Hz),

beta (13-25 Hz), and gamma (25 to 40 Hz) frequency bands. In agreement with previous work,

alpha activity for the attend condition is significantly lower than for the ignore condition

(p=1.2e-7), and maximum alpha suppression was seen over the parieto-occipital areas

contralateral to the attended visual field. While theta and beta power were reduced by attention

(p=0.04 and p=1.6e-3 respectively), given the lack of consistent spectral peaks in theta and beta

frequencies across participants, it is highly probable that their attentional modulation stems from

power leakage associated with the spectral estimation procedure from the alpha band. Gamma

band power was not different between the attend and the ignore conditions (p=0.59). Frontal-

Occipital and Parietal-Occipital coherence in each frequency band was evaluated using the

channels overlying the anterior medial frontal, medial frontal, parietal, and lateral occipital areas

in Figure 6-2D and the difference between the two conditions were not statistically significant

(p>0.13). However, Granger causality spectra in Figures 6-2A and 6-2B between the same set of

channels for a typical subject shows a reduction in the alpha band when the attend is compared

with the ignore condition. This reduction of Frontal-*Occipital and Parietal-*Occipital causal

influences in the alpha band was consistent across subjects as shown in Figure 6-2C.

Statistically, collapsed across left and right occipital areas, alpha range Frontal-*Occipital and

Parietal-*Occipital for attend condition is significantly lower than for ignore condition (p=0.035









and p=0.06 respectively). The difference in causal influences in the opposite direction,

Occipital--Frontal and Occipital-*Parietal, between attend and ignore condition for the same

time period failed to reach statistical significance (p>0.79). In addition, as summarized in

Table 1, Granger causality values in theta, beta and gamma bands are not modulated by attention,

suggesting that the alpha oscillatory activity is likely the signal that mediates top-down

attentional influences.

Analysis 2: To identify if the alpha oscillatory activity during the anticipatory attention

period preceding the presentation of the targets is correlated with behavioral performance,

power, coherence and Granger causality in alpha band were derived for the period -300 to 0 ms

before validly cued target stimulus for the fast and slow reaction time groups (Figure 6-3). In line

with the result of Analysis 1, we found that for the trials in the fast reaction time group where the

subjects correctly identified the targets and made a quicker response, the alpha band power over

the occipital areas contralateral to the attended location, and the alpha band coherence between

the fronto-parietal areas and the occipital areas contralateral to the attended location, were

significantly reduced relative to the slow reaction time group (p=2.4e-5 and 2.2e-4 respectively).

Similarly, the causal influence from the anterior medial frontal region into occipital areas

contralateral to the attended location was significantly reduced for the fast reaction time group

compared to the slow reaction time group (p=0.035). However, no statistically significant

difference in causal influence from the medial frontal areas sampled around the electrode Fz and

from the medial parietal areas into the occipital areas were observed (p>0.12). Statistically

significant difference in causal influence from occipital regions into the frontal and parietal

regions, possibly reflecting bottom-up influence, were observed (p=0.035) with the faster

reaction time group trials exhibiting reduced









Analysis 3: Figure 6-4A compares the time course of alpha band power over the occipital

areas when attention was allocated to the contralateral visual field (attend, solid line) with when

the contralateral visual field was ignored (ignore, dotted line). Periods of significant differences

between the two conditions (p < 0.05) are indicated in Figure 6-4. Post-cue alpha power

decreased markedly shortly after the presentation of the cue over the occipital regions with the

attend condition having greater alpha power reduction compared to the ignore condition and this

difference was sustained throughout the anticipatory attention period.

The time courses of coherence and causality between the combined fronto-parietal areas

(anterior medial frontal, medial frontal and parietal) and the occipital visual areas are shown in

Figures 6-4B-6-4D. These functional connectivity measures exhibited marked decrease post-cue

for both attend and ignore conditions, suggesting a substantial weakening of the oscillatory

network in the alpha band. Although a similar post-cue decrease in coherence and causal

influences between the frontal and parietal regions were observed, there were no significant

differences between the attend and ignore conditions, consistent with the notion that both attend

and ignore conditions involve effortful attention deployment and the frontal-parietal areas are

equally engaged in both conditions (Hopfinger et al., 2000; Corbetta and Shulman, 2002;

Bressler et al., 2008).

6.4 Discussion

Anticipatory attention is a cognitive process critical for top-down control. The maintenance

of anticipatory attention following an attention directing cue has been shown to entail sustained

delay period activity in the fronto-parietal network hypothesized to enable preparatory biasing of

the task-relevant sensory areas (Luck et al., 1997; Kastner et al., 1999; Kastner and Ungerleider,

2001; Corbetta and Shulman, 2002). Further support for the tonic frontal top-down regulation

comes from selective cooling in animals (Fuster, 1990) and human lesion studies where the loss









of frontal control impaired both inhibitory and excitatory influences on early sensory and sensory

association areas (Knight et al., 1995; Knight et al., 1999; Barcelo et al., 2000; Yago et al.,

2004). With accumulating evidence for the hypothesis, we predicted that with anticipatory visual

attention, the frontal areas facilitated the excitability of the posterior visual areas contralateral to

the attended location in preparation for the to-be-detected visual target stimulus. In this study, we

evaluated whether visual cortex receives top-down modulation from frontal-parietal areas during

anticipatory attention using scalp EEG recordings from healthy volunteers performing a trial-by-

trial cued spatial visual attention task.

6.4.1 Alpha Band Oscillation: A Neural Correlate of Top-Down Attentional Control

It has been amply demonstrated that attention modulates ongoing brain activity prior to

stimulus onset. In the human visual system, depending on the hemifield of attention deployment,

alpha band power is shown to be reduced in a topographically selective manner over the

extrastriate areas. In this study we consider whether visual alpha modulation is subjected to the

control of higher-order executive structures. Four results were found. First, alpha band inter-areal

synchrony and Granger causality between frontal-parietal areas and occipital visual areas were

decreased by attention following an attention-directing cue. Second, the decrease was sustained

throughout the anticipatory attention period. Third, the degree of fronto-parietal-occipital alpha

band interaction is correlated with behavioral performance. Fourth, regional visual cortical

activity and inter-areal interaction in other frequency bands (theta (4-8 Hz), beta (13-25 Hz), and

gamma (25 to 40 Hz)) did not show consistent differences between attend and ignore conditions

and were not systematically correlated with behavior across participants. Taken together, it is

suggestive that the alpha oscillatory activity is likely a signal that mediates top-down attentional

influences over the visual areas. Recent evidence of the causal role of the frontal-parietal

network in the attentional modulation of the visual alpha oscillation (Capotosto et al., 2009) and









visual ERP (Taylor et al., 2007) using rTMS in humans further supports this notion, where the

application of rTMS on FEF and IPS during the anticipatory attention period aimed at disrupting

their top-down influence over early visual areas was characterized by the lack of the customary

alpha power reduction and was associated with deteriorated attentional performance and target

detection (Capotosto et al., 2009).

6.4.2 Top-Down Control of Visual Alpha: Excitation or Disinhibition?

Attention-induced alpha power reduction occurred approximately 200 ms after the cue and

this state of suppressed alpha activity was sustained throughout the anticipatory period between

cue and imperative stimulus. Reduced alpha power led to enhanced visual evoked response (not

shown) and better behavioral performance (Figure 6-3). This is consistent with the notion that

reduced alpha activity indexes increased excitability of the task-relevant visual sensory areas

(Foxe et al., 1998; Bastiaansen and Brunia, 2001; Marrufo et al., 2001; Goldman et al., 2002;

Laufs et al., 2003; Sauseng et al., 2005; Thut et al., 2006; Klimesch et al., 2007; Romei et al.,

2008b; Romei et al., 2008a) and that a state of increased alpha activity potentially reflects

functional inhibition (Worden et al., 2000; Kelly et al., 2006; Klimesch et al., 2007; Rihs et al.,

2007).

In addition to regional modulation in oscillatory activity, interareal synchronization

between the regions involved in attentional orienting, maintenance and processing is modulated

in a topographic-specific manner. Specifically, we observed a strong reduction in interareal

synchronization in the alpha frequency band between the higher-order areas represented by

frontal and parietal areas and early visual extrastriate regions contralateral to the attended visual

field relative to both the pre-cue baseline level and when the contralateral visual field was

ignored for the same network. This reduction suggests a counterintuitive dissolution of the

frontal-parietal-occipital network with deployment of attention. However given the evidence of









alpha band oscillation reflecting inhibitory control (Klimesch et al., 2007) and evidence from

neuroimaging and lesion studies (Knight et al., 1995; Knight et al., 1999; Barcelo et al., 2000;

Pliszka et al., 2000; Sieroff et al., 2004; Yago et al., 2004; Gazzaley et al., 2005; Murias et al.,

2007) of tonic frontal inhibitory influence over early sensory areas either through cortico-cortical

projections (Jones et al., 2000; Gonchar and Burkhalter, 2003) or through cortico-thalamo-

cortical projections (Yingling and Skinner, 1975; Skinner and Yingling, 1976; Yingling and

Skinner, 1976; LaBerge, 1997, 2005), it is likely that the observed synchronized network in the

alpha band prior to attention deployment reflects tonic frontal-parietal inhibition of extrastriate

areas. The subsequent attention related reduction may reflect a release from tonic inhibition, to

facilitate an optimal excitability state over the sensory cortex, favoring a disinhibition

mechanism.

Coherence being a symmetric measure does not offer distinct information on the

contributions of the different directions of neural interaction between the regions. To dissociate

top-down from bottom-up interaction and to characterize the attentional modulation and its time

course, the data was subjected to Granger causality analysis. Granger causality in the alpha band

between the frontal-parietal areas and occipital areas in both directions were found to be reduced

as early as 200 ms following the presentation of the attention directing cue. Comparison of

magnitude of Granger causality in the alpha band for frontal and parietal driving extrastriate

regions revealed a further reduction when the attend condition is compared to the ignore

condition reflecting differences in attentional deployment and attentional load. Given the

evidence of long distance projections, e.g. from higher-order areas to occipital lobe, tend to arise

from pyramidal neurons and terminate on interneurons (Jones et al., 2000; Gonchar and

Burkhalter, 2003), this reduced Granger causality with deployment of attention likely reflects









reduced excitatory drive on local intemeurons, suggesting a disinhibitory role of attention in

which its goal-oriented deployment leads to the release of the visual cortex from tonic top-down

inhibitory influence (Knight et al., 1999). EEG predominantly measures the activity due to

pyramidal cells. Prior to the attention directing cue, frontal/parietal mediated tonic excitatory

drive to the visual cortical intemeurons would result in inhibitory postsynaptic potential (IPSP)

in the pyramidal cells over the visual areas sensed by scalp EEG. The temporal precedence of

activity between the fronto-parietal areas and the visual areas would manifest as strong pre-cue

frontal driving posterior Granger causality characterizing the tonic inhibitory influence. The

putative reduction in excitatory drive to the visual interneurons with deployment of goal directed

attention would result in the release of tonic inhibition over the visual pyramidal cells and thus

increase in excitability. This reduction in the tonic drive upon deployment of attention would

account for the decreased fronto-parietal driving posterior Granger causality post-cue.

Further insights on the role of alpha band synchronization comes from low frequency

rTMS (-1 Hz) studies known to induce inhibitory effects (Hilgetag et al., 2001; Brignani et al.,

2008) where it has been shown to induce long-lasting increases in alpha band power and cortico-

cortical alpha band coherence with associated decreases in evoked potential (Strens et al., 2002;

Brignani et al., 2008). Gathering from evidences across modalities it emerges that a state of

reduced excitability or functional inhibition is accompanied by increased alpha band activity and

enhanced interareal coherence in the alpha frequency band. The reduction in the alpha band

power, frontal-parietal-occipital coherence and reduced frontal-parietal driving extrastriate with

deployment of attention strongly suggests an attentional control through a disinhibition

mechanism, which likely subserves prestimulus biasing and maintenance of attentional state. It is

however worth noting that long-range alpha band desynchronization has also been explained in









terms of an enhanced direct or indirect excitatory facilitation of local neural population, which

due to reciprocal interaction between local pyramidal cells and interneurons and complex

interplay of h and T ionic currents resulted in varying levels of jitter in the firing time of cells

leading to asynchrony and hence a reduction in local and interareal synchronization in the alpha

band (Jones et al., 2000). Unlike recent animal (Gregoriou et al., 2009) and human attention

studies (Siegel et al., 2008), the differences in higher frequencies between conditions were not

statistically significant across subjects. The inability of scalp EEG to clearly resolve gamma band

activity (25-90 Hz) may be a contributing factor. It should also be pointed out that, while

supporting a role of frontal-parietal areas in exerting top-down modulatory influence on the

excitability of visual cortex, our data do not have the requisite spatial resolution to resolve

whether this influence is anatomically mediated by the cortico-cortical or cortico-thalamo-

cortical forms of signal transmission (Watson et al., 1981; LaBerge, 1997; Heilman et al., 2000;

LaBerge, 2005; Zikopoulos and Barbas, 2007).

Additionally, bottom-up influence of the early visual areas on parietal and frontal areas

showed a drastic reduction following attention directing cue and further showed a significant

reduction with attention to contralateral visual field relative to when it was ignored specifically

in the alpha band (Figures 6-3 and 6-4D). Given that feedforward projections from lower-order

areas to higher-order areas are predominantly excitatory and that they terminate on excitatory

cells, it is surmised that this dissolution of causal influence from the early visual areas reflect a

weakening of the bottom-up excitatory drive supporting a reduction in the information transfer in

the bottom-up direction during anticipatory attention (Bressler et al., 2008) serving to reduce

interference due to distracters. Alternatively if the alpha band synchrony emerges from feed-

forward inhibition, this reduction in alpha band coupling similar to the top-down drive, may









reflect a opening up of a more efficient bottom-up communication channel enabling a more

effective sensory-perceptual and higher-order processing of the subsequent stimuli in the

attended domain.

6.4.3 Top-Down Control of Visual Alpha: Transient or Sustained?

Several areas including prefrontal cortex (PFC), frontal eye field (FEF), intraparietal sulcus

(IPS) and superior parietal sulcus (SPL) have been implicated in the top-down control of early

visual areas in covert spatial visual attention (Gitelman et al., 1999; Hopfinger et al., 2000;

Corbetta and Shulman, 2002). However, the question of whether the top-down biasing

originating from these higher-order areas to extrastriate regions during sustained attention is

transient or sustained remains unclear. It has been hypothesized that areas such as FEF and SPL

which are involved in shifting of attentional focus, may invoke a transient top-down control

(Luks et al., 2008) but see (Gregoriou et al., 2009) where as regions such the anterior PFC,

dorsolateral PFC and IPS implicated in maintaining anticipatory attention to spatial locations

may be continuously engaged throughout the anticipatory attention period (Kastner et al., 1999;

Hopfinger et al., 2000; Corbetta and Shulman, 2002; Luks et al., 2008). Although, our data do

not have the requisite spatial resolution to resolve the anatomical structure mediating the top-

down influence, it is likely that the frontal scalp electrodes sample activity from the anterior

PFC, FEF and the anterior cingulate regions and the parietal scalp electrodes sample activity

from the SPL and IPS among other neighboring structures. Our data suggests that upon initiation

of attention orienting activity, both the frontal, parietal and the posterior visual regions appear to

maintain sustained levels of activity throughout the cue-imperative interval indexing a continued

engagement of the frontal-parietal-occipital regions in the maintenance of the attentional state

and a tonic preparatory biasing of the relevant visual sensory areas. Further, any momentary

lapse within this frontal-parietal-occipital anticipatory attention network characterized by









increased magnitude of visual alpha power and interareal interaction resulted in deteriorated

behavioral performance evidenced by response errors and slowed response. This weakening of

top-down biasing may either be actively controlled in a goal-directed manner as is the case

during the ignored condition or may be due to a momentary lapse in attention resulting in

deficient visual processing of the subsequent stimulus and hence in errors.



Table 6-1. Attention modulation of oscillatory activity. p values of Wilcoxon rank-sum test
result (two-tailed) for difference in magnitude of Granger causality between attend
and ignore conditions over theta, alpha, beta and gamma frequency bands during the
time period 300 to 0 ms immediately preceding the onset of the standard stimulus
(Analysis 1).
Granger Causality Region of interest Theta Alpha Beta Gamma
(4-8 Hz) (8-12 Hz) (13-25 Hz) (25-40 Hz)
Top-down Granger Frontal -Occipital 0.71 0.036* 0.71 0.71
Causality Parietal-Occipital 0.29 0.06 0.12 0.29
Bottom-up Granger Occipital -Frontal 0.13 0.79 0.45 0.45
Causality Occipital -Parietal 0.49 1.00 0.30 0.49
















I II
t 200 1800-2200 t 100 2800+0/200/400 t 200 (ms)
Cue Preparatory attention period Imperative ISI Cue
Figure 6-1. Experimental task. At the beginning of each trial, a cue (left/right pointing arrow)
was presented at the fixation point instructing the subjects to covertly attend one of
two predefined locations in the upper left and upper right visual fields. After 1800 to
2200 ms of delay, a unilateral imperative stimulus of either target or standard type
appeared at one of the two locations, and the subjects indicated the occurrence of the
target stimulus at the attended location by a speeded keypress response.










Parietal -> Occipital areas


0 10 20 30 40 50
Frequency (Hz)


60 70


0 10 20 30 40 50 60 70
Frequency (Hz)


Frontal -, Occipital Parietal -, Occipital


AT IG


I rF


* Frontal (Fpz)
* Frontal (Fz)


AT IG


* Parietal (Pz)
* Occipital


Figure 6-2. Top-down attentional control of visual areas. Granger causality spectra for (A)
(Frontal -Occipital) and (B) (Parietal--Occipital) from a representative subject are
shown for the attend (black) and the ignore (gray) conditions. (C) Normalized
Granger causality in the alpha frequency band for (Frontal -Occipital) and
(Parietal--Occipital) are averaged across subjects and compared between attend
(black) and ignore (white) conditions. The error bars are the standard error of the
mean. (D) Regions of interest marked by shaded colors correspond to frontal (Fpz,
Fz), parietal (Pz) and occipital cortices. denotes statistically significant difference
(p<0.05).


C


I

ii 0.9

O 0.8
0

C,
S0.7


0.6L


Frontal --) Occipital areas
















1j1 1I I 1I
0.8



0.6
Power Coherence Top-down Bottom-up
Causality Causality
Fast RT Z Slow RT
Figure 6-3. Pre-target alpha band activity predicts performance. The figure summarizes the
regional alpha band oscillatory synchrony over the Occipital electrodes, averaged
coherence between (Frontal-Occipital) and (Parietal-Occipital) areas, Granger
causality measurements of (Frontal --Occipital) indexing top-down influence and
average of (Occipital -Frontal) and (Occipital -Parietal) indexing the bottom-up
interaction during the 300 ms period immediately prior to the presentation of the
validly cued targets for two classes of trials: correct and quicker responded trials
(black) and incorrect and slower responded trials (white). The figure highlights the
notion that a state of increased regional and network-wide alpha band oscillatory
synchrony and interaction reflects a weakening or disruption of top-down biasing and
hence impaired performance. denotes statistically significant difference (p<0.05).












-Attend -Attend
/. -- Ignore 0.7 Ignore
o.8 0.

o D
0.7 0. Ignore
( ,,o \/


S0.0.5
Z o
0.5 z
0.4

0.4
0 500 1000 0 500 1000
Time (ms) Time (ms)

C D

--Attendl -Attend
---- Ignore 0.8 ( Ignore

a b- A ttend A tte n d
S0.8 0_7
o oo.7


S090.6
Zo
z z
0.7

0 500 1000 0 500 1000
Time (ms) Time (ms)


Figure 6-4. Temporal evolution of regional and network-level alpha band activity. (A)
Normalized spectral power in the alpha frequency band over lateral Occipital areas
are averaged across subjects and compared between attend (solid black) and ignore
(dotted lines) conditions. Normalized coherence (B), and Granger causality indexing
top-down (C) and bottom-up (D) influences in the alpha frequency band collapsed
across (Frontal-Occipital) and (Parietal-Occipital) are averaged across subjects and
compared between attend (solid black) and ignore (dotted lines) conditions. The
shaded time periods denote periods that exhibit statistically significant differences
between conditions.









CHAPTER 7
CONCLUSION

The dissertation investigated the functional role of ongoing oscillations in stimulus

processing and higher-order cognitive processes such as anticipatory attention and encoding of

memory along four cogent studies. The evidences garnered from these studies debunk the long

held notion that ongoing neural activity preceding stimulus onset is noise and irrelevant to

subsequent stimulus processing. It is also amply shown that the prestimulus activity is not merely

a random fluctuation in excitability but one that is actively controlled in a goal-directed manner

through neural synchronization across widespread functionally relevant cortical networks. In

addition, a biophysically inspired framework to assess and characterize the dependence of

sensory processing on the excitability state of the brain indexed by the prestimulus ongoing brain

activity was proposed and tested.

In the first study, we sought to identify the factors that lead to near zero-lags in

synchronous networks and resolve the ambiguity of such observations in unidirectional

networks. We decomposed neural synchrony into components quantifying directed interaction

and instantaneous correlation between the regions and further, quantified their influence on

phase-lag between the areas. From our findings, we confirmed our hypothesis that two factors:

(a) instantaneous causality, which may be due to common input, such as cortical or subcortical

sources which may or may not be sampled in the recordings, and / or (b) reciprocal interaction

pattern between the regions, account for the coherent activity with zero-lag across long distances.

In cases where the interaction is clearly unidirectional determined by Granger causality analysis,

the phenomenon of zero-lag was completely explained by instantaneous causality. For cases

wherein the interaction is bidirectional, the reciprocal nature of the network in addition to

instantaneous causality may cause near zero time lags. Hence we show that zero-lag synchrony is









an emergent phenomenon dependent on the topology of the interacting network of neural

ensembles and is not representative of the conduction delay between the interacting areas as

often misconstrued.

In a following study examining the role of prestimulus ongoing oscillatory brain activity in

encoding of memory we provided the first evidence in humans demonstrating that increased

inter-areal synchrony in the fronto-temporal network prior to item presentation is associated with

higher likelihood of the item being recalled in the subsequent test. The changes in inter-areal

synchrony, taken as an indicator of changes in network excitability, may be a result of

spontaneous fluctuations in the level of anticipatory attention and hence waxing and waning of

top-down influence. Possibly as a consequence of top-down influence the level of regional theta

activity indexing depolarizing influence on NMDA receptor-containing cells and hence tissue

excitability over the fronto-temporal structures prior to the presentation of the item was

positively correlated with the efficacy of subsequent memory encoding and later recall. Based

on rodent studies on theta oscillations and memory, we propose that the temporal coupling of the

sensory input triggered glutamate release and theta-induced depolarization activates the theta

facilitated NMDA channels leading to stronger calcium influx, which in turn facilitates LTP and

thus leads to more effective encoding-related processes.

As part of the third study, we provided a thorough characterization of the relationship

between the prestimulus ongoing oscillatory activity specifically the alpha band activity

(8-12 Hz) and the subsequent early visual-evoked response. A mechanism linking the dynamics

of prestimulus ongoing oscillatory activity and the generation of early evoked response was

proposed by invoking the sigmoidal relationship between the endogenous synaptic fluctuations

and the firing rate of the neural ensemble, and that external sensory input has an additive









synaptic effect. It follows that the stimulus-evoked response corresponds to the derivative of the

sigmoidal function, which is an inverted-U function of the prestimulus level of background

activity. Reformulated in terms of EEG variables where field oscillations and the early

component of event-related potential are treated as indices of background synaptic activity and

stimulus-evoked response, an inverted-U relationship between early evoked response and

prestimulus field oscillations was predicted and confirmed empirically through noninvasive EEG

studies of human visual processing. Attention has been thought to selectively enhance the

perceptual salience of attended stimuli through competing mechanisms of sensory gain control.

Based on the mechanism linking pre and poststimulus activity and using computational modeling

and empirical evidence from EEG studies of visual attention we show that attention through

active control of (a) mean membrane depolarization (excitability) and (b) the variance of

background synaptic level (membrane potential fluctuation) over the early sensory areas enable

preferential facilitation of processing in the attended domain over the unattended domain.

Although the previous aim addressed the attention modulation of excitability of the early

visual areas and hence the subsequent visual processing, such goal-directed prestimulus

modulation is not an inherent property of the early sensory areas. Further, the neural mechanisms

by which this is implemented are still unclear. As part of the fourth and final study, we assessed

how attention modulates endogenous anticipatory processes employing time-frequency spectral

analysis. Applying coherence analysis to the anticipatory attention period we identified a

synchronous frontal parietal-occipital network in the low frequency range (alpha band), where

Granger causality analysis further defined the direction of causal interaction. The degree of

inter-areal synchrony and the magnitude of Granger causality were found to be decreased by

attention. In light of prior lesion and EEG/MEG studies and anatomical evidences this reduced









fronto-parietal driving occipital Granger causality is seen to lend support to the disinhibitionn

mechanism', where attention serves to increase excitability through reduced excitatory drive on

local intemeurons, suggesting a disinhibition function of attention in which its goal-oriented

deployment leads to the release of the visual cortex from tonic top-down inhibitory influence.

It is evident from our studies and from other previous studies that all cognitive function

depends on the coordinated interaction among neural assemblies distributed across the brain.

Although we have exclusively focused on neural oscillations and synchrony to address the

mechanisms of executive processes and other cognitive processes in healthy human volunteers, it

is equally relevant to address pathological brain states in diseases including Schizophrenia,

Alzheimer's disease, Parkinson's disease and other developmental diseases causing

neurocognitive impairments such as autism and ADHD among others. A common thread among

all these diseases is the deficits in coordinated processing across distributed cortical networks

leading to specific cognitive deficits especially in executive functioning, attention and memory.

Measures of inter-areal connectivity such as coherence and Granger causality in addition to

anatomical metrics quantifying white matter and thus neuroanatomical substrates of inter-areal

connectivity may enable objective measures of functional integrity of pathways between

different areas of the brain and thus may serve as biomarkers offering diagnostic potential.

Further the findings from our studies regarding the critical role of spontaneous brain state on

subsequent stimulus processing and behavioral performance have implications in

neurodiagnostic and neurofeedback based intervention capability to detect and treat cognitive

decline and dementia caused by aging and possibly ameliorate psychiatric conditions afflicting

the emotion circuits by employing brain-state contingent training which may effect a facilitatory

pattern of brain activity and hence behavioral improvement. Thus, synchronous oscillations









quantifying regional and large-scale integration of neural activity as assessed from noninvasive

and invasive electrophysiological recordings may be of critical value not only in elucidating

mechanism related to normal cognitive functioning but also to address cognitive impairments

brought about by multitude of debilitating neurological and psychiatric disorders.









REFERENCES


Abbott LF, Varela JA, Sen K, Nelson SB (1997) Synaptic depression and cortical gain control.
Science 275:220-224.

Anderson K, Rajagovindan R, Ghacibeh G, Meador KJ, Ding M (in press) Theta oscillations
mediate interaction between prefrontal cortex and medial temporal lobe in human memory.
Cerebral Cortex.

Araya R, Eisenthal KB, Yuste R (2006) Dendritic spines linearize the summation of excitatory
potentials. Proceedings of the National Academy of Sciences 103:18799-18804.

Arieli A, Sterkin A, Grinvald A, Aertsen A (1996) Dynamics of ongoing activity: Explanation of
the large variability in evoked cortical responses. Science 273:1868-1871.

Armstrong KM, Fitzgerald JK, Moore T (2006) Changes in visual receptive fields with
microstimulation of frontal cortex. Neuron 50:791-798.

Averbeck BB, Lee D (2004) Coding and transmission of information by neural ensembles.
Trends in Neurosciences 27:225-230.

Babiloni C, Vecchio F, Bultrini A, Romani GL, Rossini PM (2006) Pre- and poststimulus alpha
rhythms are related to conscious visual perception: A high-resolution EEG study. Cerebral
Cortex 16:1690-1700.

Baccala LA, Sameshima K (2001) Partial directed coherence: A new concept in neural structure
determination. Biological Cybernetics 84:463-474.

Barbas H (2000) Connections underlying the synthesis of cognition, memory, and emotion in
primate prefrontal cortices. Brain Research Bulletin 52:319-330.

Barcelo F, Suwazono S, Knight RT (2000) Prefrontal modulation of visual processing in
humans. Nature Neuroscience 3:399-403.

Bastiaansen MCM, Brunia CHM (2001) Anticipatory attention: an event-related
desynchronization approach. International Journal of Psychophysiology 43:91-107.

Bechara A, Damasio H, Tranel D, Anderson SW (1998) Dissociation of working memory from
decision making within the human prefrontal cortex. Journal of Neuroscience 18:428-437.

Berry SD, Seager MA (2001) Hippocampal theta oscillations and classical conditioning.
Neurobiology of Learning and Memory 76:298-313.

Bestmann S, Ruff CC, Blakemore C, Driver J, Thilo KV (2007) Spatial attention changes
excitability of human visual cortex to direct stimulation. Current Biology 17:134-139.

Bokil H, Purpura K, Schoffelen JM, Thomson D, Mitra P (2007) Comparing spectra and
coherences for groups of unequal size. Journal of Neuroscience Methods 159:337-345.









Bollimunta A, Chen Y, Schroeder CE, Ding M (2008) Neuronal mechanisms of cortical alpha
oscillations in awake-behaving macaques. Journal of Neuroscience 28:9976-9988.

Brandt ME, Jansen BH (1991) The relationship between prestimulus alpha-amplitude and visual
evoked-potential amplitude. International Journal of Neuroscience 61:261-268.

Brandt ME, Jansen BH, Carbonari JP (1991) Pre-stimulus spectral EEG patterns and the visual
evoked-response. Electroencephalography and Clinical Neurophysiology 80:16-20.

Bressler SL, Coppola R, Nakamura R (1993) Episodic multiregional cortical coherence at
multiple frequencies during visual task-performance. Nature 366:153-156.

Bressler SL, Tang W, Sylvester CM, Shulman GL, Corbetta M (2008) Top-down control of
human visual cortex by frontal and parietal cortex in anticipatory visual spatial attention.
Journal of Neuroscience 28:10056-10061.

Brignani D, Manganotti P, Rossini PM, Miniussi C (2008) Modulation of cortical oscillatory
activity during transcranial magnetic stimulation. Human Brain Mapping 29:603-612.

Brovelli A, Ding MZ, Ledberg A, Chen YH, Nakamura R, Bressler SL (2004) Beta oscillations
in a large-scale sensorimotor cortical network: Directional influences revealed by Granger
causality. Proceedings of the National Academy of Sciences of the United States of
America 101:9849-9854.

Burgess AP, Gruzelier JH (1997) Short duration synchronization of human theta rhythm during
recognition memory. Neuroreport 8:1039-1042.

Buschman TJ, Miller EK (2007) Top-down versus bottom-up control of attention in the
prefrontal and posterior parietal cortices. Science 315:1860-1862.

Buzsaki G (1989) 2-Stage model of memory trace formation a role for noisy brain states.
Neuroscience 31:551-570.

Buzsaki G (1996) The hippocampo-neocortical dialogue. Cerebral Cortex 6:81-92.

Buzsaki G (2002) Theta oscillations in the hippocampus. Neuron 33:325-340.

Buzsaki G (2006) Rhythms of the Brain. New York: Oxford University Press.

Cadotte AJ, DeMarse TB, He P, Ding M (2008) Causal measures of structure and plasticity in
simulated and living neural networks. PLoS ONE 3:1-14.

Cadotte AJ, Mareci TH, DeMarse TB, Parekh MB, Rajagovindan R, Ditto WL, Talathi SS,
Hwang D-U, Carney PR (2009) Temporal lobe epilepsy: Anatomical and effective
connectivity. IEEE Transactions in Neural System and Rehabilitation Engineering 17:214-
223.









Callaway EM (1998) Local circuits in primary visual cortex of the macaque monkey. Annual
Review of Neuroscience 21:47-74.

Capotosto P, Babiloni C, Romani GL, Corbetta M (2009) Frontoparietal cortex controls spatial
attention through modulation of anticipatory alpha rhythms. Journal of Neuroscience
29:5863-5872.

Cash S, Yuste R (1999) Linear summation of excitatory inputs by CA1 pyramidal neurons.
Neuron 22:383-394.

Cassidy M, Brown P (2003) Spectral phase estimates in the setting of multidirectional coupling.
Journal of Neuroscience Methods 127:95-103.

Chance FS, Abbott LF, Reyes AD (2002) Gain modulation from background synaptic input.
Neuron 35:773-782.

Chawla D, Friston KJ, Lumer ED (2001) Zero-lag synchronous dynamics in triplets of
interconnected cortical areas. Neural Networks 14:727-735.

Chelazzi L, Miller EK, Duncan J, Desimone R (1993) A neural basis for visual-search in inferior
temporal cortex. Nature 363:345-347.

Chen YH, Bressler SL, Ding MZ (2006) Frequency decomposition of conditional Granger
causality and application to multivariate neural field potential data. Journal of
Neuroscience Methods 150:228-237.

Chung S, Li XR, Nelson SB (2002) Short-term depression at thalamocortical synapses
contributes to rapid adaptation of cortical sensory responses in vivo. Neuron 34:437-446.

Clark VP, Hillyard SA (1996) Spatial selective attention affects early extrastriate but not striate
components of the visual evoked potential. Journal of Cognitive Neuroscience 8:387-402.

Cohen NJ, Ryan E, Hunt C, Romine L, Wszalek T, Nash C (1999) Hippocampal system and
declarative (relational) memory: Summarizing the data from functional neuroimaging
studies. Hippocampus 9:83-98.

Corbetta M, Shulman GL (2002) Control of goal-directed and stimulus-driven attention in the
brain. Nature Reviews Neuroscience 3:201-215.

Dasilva FL (1991) Neural mechanisms underlying brain waves from neural membranes to
networks. Electroencephalography and Clinical Neurophysiology 79:81-93.

Dehaene S, Changeux J-P (2005) Ongoing spontaneous activity controls access to consciousness:
A neuronal model for inattentional blindness. PLoS Biology 3:910-927.

Desimone R (1996) Neural mechanisms for visual memory and their role in attention.
Proceedings of the National Academy of Sciences: 13494-13499.









Destexhe A, Rudolph M, Fellous JM, Sejnowski TJ (2001) Fluctuating synaptic conductances
recreate in vivo-like activity in neocortical neurons. Neuroscience 107:13-24.

Di Russo F, Martinez A, Hillyard SA (2003) Source analysis of event-related cortical activity
during visuo-spatial attention. Cerebral Cortex 13:486-499.

Di Russo F, Martinez A, Sereno MI, Pitzalis S, Hillyard SA (2002) Cortical sources of the early
components of the visual evoked potential. Human Brain Mapping 15:95-111.

Ding M, Chen Y, Bressler SL (2006) Granger causality: Basic theory and application to
neuroscience. In: Handbook of Time Series Analysis (Winterhalder M, Schelter B and
Timmer J, eds), pp 451-474. Berlin: Wiley-VCH Verlag.

Ding MZ, Bressler SL, Yang WM, Liang HL (2000) Short-window spectral analysis of cortical
event-related potentials by adaptive multivariate autoregressive modeling: Data
preprocessing, model validation, and variability assessment. Biological Cybernetics 83:35-
45.

Doesburg SM, Roggeveen AB, Kitajo K, Ward LM (2008) Large-scale gamma-band phase
synchronization and selective attention. Cerebral Cortex 18:386-396.

Dringenberg HC (2000) Alzheimer's disease: More than a 'cholinergic disorder' evidence that
cholinergic-monoaminergic interactions contribute to EEG slowing and dementia.
Behavioural Brain Research 115:235-249.

Eckhorn R, Bauer R, Jordan W, Brosch M, Kruse W, Munk M, Reitboeck HJ (1988) Coherent
oscillations: A mechanism of feature linking in the visual cortex? Multiple electrode and
correlation analyses in the cat. Biological Cybernetics 60:121-130.

Efron B (1985) Bootstrap confidence-intervals for a class of parametric problems. Biometrika
72:45-58.

Efron B (2000) The bootstrap and modern statistics. Journal of American Statistical Association
95:1293-1296.

Efron B, Gong G (1983) A leisurely look at the bootstrap, the jacknife, and cross-validation. The
American Statistician 37:36-48.

Eichele T, Debener S, Calhoun VD, Specht K, Engel AK, Hugdahl K, von Cramon DY,
Ullsperger M (2008) Prediction of human errors by maladaptive changes in event-related
brain networks. Proceedings of the National Academy of Sciences of the United States of
America 105:6173-6178.

Engel AK, Fries P, Singer W (2001) Dynamic predictions: Oscillations and synchrony in top-
down processing. Nature Reviews Neuroscience 2:704-716.









Ergenoglu T, Demiralp T, Bayraktaroglu Z, Ergen M, Beydagi H, Uresin Y (2004) Alpha rhythm
of the EEG modulates visual detection performance in humans. Cognitive Brain Research
20:376-383.

Felleman DJ, Van Essen DC (1991) Distributed hierarchical processing in the primate cerebral
cortex. Cerebral Cortex 1:1-47.

Fernandez G, Effern A, Grunwald T, Pezer N, Lehnertz K, Dumpelmann M, Van Roost D, Elger
CE (1999) Real-time tracking of memory formation in the human rhinal cortex and
hippocampus. Science 285:1582-1585.

Ferree T (2006) Spherical splines and average referencing in scalp electroencephalography.
Brain Topography 19:43-52.

Fischer T, Langner R, Birbaumer N, Brocke B (2008) Arousal and attention: Self-chosen
stimulation optimizes cortical excitability and minimizes compensatory effort. Journal of
Cognitive Neuroscience 20:1443-1453.

Foxe JJ, Simpson GV, Ahlfors SP (1998) Parieto-occipital similar to 10 Hz activity reflects
anticipatory state of visual attention mechanisms. Neuroreport 9:3929-3933.

Freeman WJ (1979) Non-linear gain mediating cortical stimulus-response relations. Biological
Cybernetics 33:237-247.

Friedman D, Johnson R (2000) Event-related potential (ERP) studies of memory encoding and
retrieval: A selective review. Microscopy Research and Technique 51:6-28.

Fries P (2005) A mechanism for cognitive dynamics: Neuronal communication through neuronal
coherence. Trends in Cognitive Sciences 9:474-480.

Fries P, Reynolds JH, Rorie AE, Desimone R (2001) Modulation of oscillatory neuronal
synchronization by selective visual attention. Science 291:1560-1563.

Furey ML, Pietrini P, Haxby JV (2000) Cholinergic enhancement and increased selectivity of
perceptual processing during working memory. Science 290:2315-2319.

Fuster JM (1990) Behavioral electrophysiology of the prefrontal cortex of the primate. Progress
in Brain Research 85:313-324.

Fuster JM, Bauer RH, Jervey JP (1985) Functional interactions between inferotemporal and
prefrontal cortex in a cognitive task. Brain Research 330:299-307.

Gallinat J, Kunz D, Senkowski D, Kienast T, Seifert F, Schubert F, Heinz A (2006)
Hippocampal glutamate concentration predicts cerebral theta oscillations during cognitive
processing. Psychopharmacology 187:103-111.

Garrido MI, Kilner JM, Kiebel SJ, Friston KJ (2007) Evoked brain responses are generated by
feedback loops. Proceedings of the National Academy of Sciences 104:20961-20966.









Gazzaley A, Rissman J, D'Esposito M (2004) Functional connectivity during working memory
maintenance. Cognitive Affective & Behavioral Neuroscience 4:580-599.

Gazzaley A, Cooney JW, McEvoy K, Knight RT, D'Esposito M (2005) Top-down enhancement
and suppression of the magnitude and speed of neural activity. Journal of Cognitive
Neuroscience 17:507-517.

Gazzaley A, Rissman J, Cooney J, Rutman A, Seibert T, Clapp W, D'Esposito M (2007)
Functional interactions between prefrontal and visual association cortex contribute to top-
down modulation of visual processing. Cerebral Cortex:I125-I135.

Gelfand IM, Yaglom AM (1959) Calculation of the amount of information about a random
function contained in another such function. American Mathematical Society Translations
Series 2:199-246.

Gershberg FB, Shimamura AP (1995) Impaired use of organizational strategies in free-recall
following frontal lope damage. Neuropsychologia 33:1305-1333.

Gevins A, Smith ME, McEvoy L, Yu D (1997) High-resolution EEG mapping of cortical
activation related to working memory: Effects of task difficulty, type of processing, and
practice. Cerebral Cortex 7:374-385.

Geweke J (1982) Measurement of linear dependence and feedback between multiple time series.
Journal of American Statistical Association 77:304-314.

Giesbrecht B, Woldorff MG, Song AW, Mangun GR (2003) Neural mechanisms of top-down
control during spatial and feature attention. Neuroimage 19:496-512.

Giesbrecht B, Weissman DH, Woldorff MG, Manqun GR (2006) Pre-target activity in visual
cortex predicts behavioral performance on spatial and feature attention tasks. Brain
Research 1080:63-72.

Gilbert CD, Wiesel TN (1983) Functional-organization of the visual-cortex. Progress in Brain
Research 58:209-218.

Gitelman DR, Nobre AC, Parrish TB, LaBar KS, Kim YH, Meyer JR, Mesulam MM (1999) A
large-scale distributed network for covert spatial attention further anatomical delineation
based on stringent behavioral and cognitive controls. Brain 122:1093-1106.

Givens BS, Olton DS (1990) Cholinergic and gabaergic modulation of medial septal area effect
on working memory. Behavioral Neuroscience 104:849-855.

Goldman RI, Stern JM, Engel J, Cohen MS (2002) Simultaneous EEG and fMRI of the alpha
rhythm. Neuroreport 13:2487-2492.

Gonchar Y, Burkhalter A (2003) Distinct GABAergic targets of feedforward and feedback
connections between lower and higher areas of rat visual cortex. Journal of Neuroscience
23:10904-10912.









Grady CL, McIntosh AR, Craik FIM (2003) Age-related differences in the functional
connectivity of the hippocampus during memory encoding. Hippocampus 13:572-586.

Granger CWJ (1969) Investigating causal relations by econometric models and cross-spectral
methods. Econometrica 37:414-438.

Gray CM, Konig P, Engel AK, Singer W (1989) Oscillatory responses in cat visual-cortex
exhibit inter-columnar synchronization which reflects global stimulus properties. Nature
338:334-337.

Green JJ, McDonald JJ (2008) Electrical neuroimaging reveals timing of attentional control
activity in human brain. PLoS Biology 6:730-738.

Gregoriou GG, Gotts SJ, Zhou H, Desimone R (2009) High-frequency, long-range coupling
between prefrontal and visual cortex during attention. Science 324:1207-1210.

Grent-'T-Jong T, Woldorff MG (2007) Timing and sequence of brain activity in top-down
control of visual-spatial attention. PLoS Biology 5:114-126.

Grunwald T, Beck H, Lehnertz K, Blumcke I, Pezer N, Kurthen M, Fernandez G, Van Roost D,
Heinze HJ, Kutas M, Elger CE (1999) Evidence relating human verbal memory to
hippocampal N-methyl-D-aspartate receptors. Proceedings of the National Academy of
Sciences of the United States of America 96:12085-12089.

Guderian S, Schott BH, Richardson-Klavehn A, Duzel E (2009) Medial temporal theta state
before an event predicts episodic encoding success in humans. Proceedings of the National
Academy of Sciences of the United States of America 106:5365-5370.

Guo S, Wu J, Ding M, Feng J (2008) Uncovering interactions in the frequency domain. PLoS
Computational Biology 4:1-10.

Haider B, Duque A, Hasenstaub AR, Yu YG, McCormick DA (2007) Enhancement of visual
responsiveness by spontaneous local network activity in vivo. Journal of Neurophysiology
97:4186-4202.

Hanslmayr S, Aslan A, Staudigl T, Klimesch W, Herrmann CS, Bauml KH (2007a) Prestimulus
oscillations predict between and within subjects. Neuroimage 37:1465-1473.

Hanslmayr S, Klimesch W, Sauseng P, Gruber W, Doppelmayr M, Freunberger R, Pecherstorfer
T, Birbaumer N (2007b) Alpha phase reset contributes to the generation of ERPs. Cerebral
Cortex 17:1-8.

Hasselmo ME (2005) What is the function of hippocampal theta Rhythm? Linking behavioral
data to phasic properties of field potential and unit recording data. Hippocampus 15:936-
949.









Hasselmo ME, McGaughy J (2004) High acetylcholine levels set circuit dynamics for attention
and encoding and low acetylcholine levels set dynamics for consolidation. Progress in
Brain Research 145:207-231.

Heilman KM, Valenstein E, Watson RT (2000) Neglect and related disorders. Seminars in
Neurology 20:463-470.

Heinze HJ, Mangun GR, Burchert W, Hinrichs H, Scholz M, Munte TF, Gos A, Scherg M,
Johannes S, Hundeshagen H, Gazzaniga MS, Hillyard SA (1994) Combined spatial and
temporal imaging of brain activity during visual selective attention in humans. Nature
372:543-546.

Hermer-Vazquez L (2008) Tracing 'driver' versus 'modulator' information flow throughout large-
scale, task-related neural circuitry. Journal of Combinatorial Optimization 15:242-256.

Herrmann CS, Munk MHJ, Engel AK (2004) Cognitive functions of gamma-band activity:
Memory match and utilization. Trends in Cognitive Sciences 8:347-355.

Hilgetag CC, Theoret H, Pascual-Leone A (2001) Enhanced visual spatial attention ipsilateral to
rTMS-induced 'virtual lesions' of human parietal cortex. Nature Neuroscience 4:953-957.

Hillyard SA, Vogel EK, Luck SJ (1998) Sensory gain control (amplification) as a mechanism of
selective attention: electrophysiological and neuroimaging evidence. Philosophical
Transactions of the Royal Society of London Series B-Biological Sciences 353:1257-1270.

Ho N, Destexhe A (2000) Synaptic background activity enhances the responsiveness of
neocortical pyramidal neurons. Journal of Neurophysiology 84:1488-1496.

Honey GD, Honey RAE, O'Loughlin C, Sharar SR, Kumaran D, Suckling J, Menon DK, Sleator
C, Bullmore ET, Fletcher PC (2005) Ketamine disrupts frontal and hippocampal
contribution to encoding and retrieval of episodic memory: An fMRI study. Cerebral
Cortex 15:749-759.

Hopfinger JB, Buonocore MH, Mangun GR (2000) The neural mechanisms of top-down
attentional control. Nature Neuroscience 3:284-291.

Huerta PT, Lisman JE (1993) Heightened synaptic plasticity of hippocampal CA1 neurons
during a cholinergically induced rhythmic state. Nature 364:723-725.

Jagadeesh B, Wheat HS, Ferster D (1993) Linearity of summation of synaptic potentials
underlying direction selectivity in simple cells of the cat visual cortex. Science 262:1901-
1904.

Janowsky JS, Shimamura AP, Squire LR (1989) Source memory impairment in patients with
frontal-lobe lesions. Neuropsychologia 27:1043-1056.

Jean-Philippe L, Eugenio R, Jacques M, Francisco JV (1999) Measuring phase synchrony in
brain signals. Human Brain Mapping 8:194-208.









Jensen O, Tesche CD (2002) Frontal theta activity in humans increases with memory load in a
working memory task. European Journal of Neuroscience 15:1395-1399.

Jokisch D, Jensen O (2007) Modulation of gamma and alpha activity during a working memory
task engaging the dorsal or ventral stream. Journal of Neuroscience 27:3244-3251.

Jones MW, Wilson MA (2005) Theta rhythms coordinate hippocampal-prefrontal interactions in
a spatial memory task. PLoS Biology 3:2187-2199.

Jones SR, Pinto DJ, Kaper TJ, Kopell N (2000) Alpha-frequency rhythms desynchronize over
long cortical distances: A modeling study. Journal of Computational Neuroscience 9:271-
291.

Jones SR, Pritchett DL, Stufflebeam SM, Hamalainen M, Moore CI (2008) Influence of Pre-
Stimulus Somatosensory Mu-Rhythm on Evoked Response Gain: A Combined MEG And
Computational Modeling Study. Society for Neuroscience Meeting Abstract 34:858.14.

Kahana MJ, Sekuler R, Caplan JB, Kirschen M, Madsen JR (1999) Human theta oscillations
exhibit task dependence during virtual maze navigation. Nature 399:781-784.

Kaminski M, Ding MZ, Truccolo WA, Bressler SL (2001) Evaluating causal relations in neural
systems: Granger causality, directed transfer function and statistical assessment of
significance. Biological Cybernetics 85:145-157.

Kapur S, Tulving E, Cabeza R, McIntosh AR, Houle S, Craik FIM (1996) The neural correlates
of intentional learning of verbal materials: A PET study in humans. Cognitive Brain
Research 4:243-249.

Kastner S, Ungerleider LG (2001) The neural basis of biased competition in human visual
cortex. Neuropsychologia 39:1263-1276.

Kastner S, Pinsk MA, De Weerd P, Desimone R, Ungerleider LG (1999) Increased activity in
human visual cortex during directed attention in the absence of visual stimulation. Neuron
22:751-761.

Kelly SP, Lalor EC, Reilly RB, Foxe JJ (2006) Increases in alpha oscillatory power reflect an
active retinotopic mechanism for distracter suppression during sustained visuospatial
attention. Journal of Neurophysiology 95:3844-3851.

Kimura F (2000) Cholinergic modulation of cortical function: A hypothetical role in shifting the
dynamics in cortical network. Neuroscience Research 38:19-26.

Kirchhoff BA, Wagner AD, Maril A, Stem CE (2000) Prefrontal-temporal circuitry for episodic
encoding and subsequent memory. Journal of Neuroscience 20:6173-6180.

Klimesch W (1999) EEG alpha and theta oscillations reflect cognitive and memory performance:
A review and analysis. Brain Research Reviews 29:169-195.









Klimesch W, Sauseng P, Hanslmayr S (2007) EEG alpha oscillations: The inhibition-timing
hypothesis. Brain Research Reviews 53:63-88.

Klimesch W, Doppelmayr M, Russegger H, Pachinger T (1996) Theta band power in the human
scalp EEG and the encoding of new information. Neuroreport 7:1235-1240.

Knight RT (1997) Distributed cortical network for visual attention. Journal of Cognitive
Neuroscience 9:75-91.

Knight RT, Grabowecky MF, Scabini D (1995) Role of human prefrontal cortex in attention
control. In: Advances in Neurology: Epilepsy and the functional anatomy of the frontal
lobe (Jasper HH, Riggio S, Goldman-Rakic PS, eds), pp 21-36. New York: Raven Press.

Knight RT, Staines WR, Swick D, Chao LL (1999) Prefrontal cortex regulates inhibition and
excitation in distributed neural networks. Acta Psychologica 101:159-178.

Kocsis B, Kaminski M (2006) Dynamic changes in the direction of the theta rhythmic drive
between supramammillary nucleus and the septohippocampal system. Hippocampus
16:531-540.

Kocsis B, Bragin A, Buzsaki G (1999) Interdependence of multiple theta generators in the
hippocampus: a partial coherence analysis. Journal of Neuroscience 19:6200-6212.

LaBerge D (1997) Attention, awareness, and the triangular circuit. Consciousness and Cognition
6:149-181.

LaBerge D (2005) Sustained attention and apical dendrite activity in recurrent circuits. Brain
Research Reviews 50:86-99.

Lakatos P, Chen CM, O'Connell MN, Mills A, Schroeder CE (2007) Neuronal oscillations and
multisensory interaction in primary auditory cortex. Neuron 53:279-292.

Lakatos P, Karmos G, Mehta AD, Ulbert I, Schroeder CE (2008) Entrainment of neuronal
oscillations as a mechanism of attentional selection. Science 320:110-113.

Lakatos P, Shah AS, Knuth KH, Ulbert I, Karmos G, Schroeder CE (2005) An oscillatory
hierarchy controlling neuronal excitability and stimulus processing in the auditory cortex.
Journal of Neurophysiology 94:1904-1911.

Laufs H, Kleinschmidt A, Beyerle A, Eger E, Salek-Haddadi A, Preibisch C, Krakow K (2003)
EEG-correlated fMRI of human alpha activity. Neuroimage 19:1463-1476.

Ledberg A, Bressler SL, Ding MZ, Coppola R, Nakamura R (2007) Large-scale visuomotor
integration in the cerebral cortex. Cerebral Cortex 17:44-62.

Lfeachor EC, Jarvis BW Digital signal processing: A practical approach. Essex: Prentice Hall.









Liang HL, Bressler SL, Ding MZ, Truccolo WA, Nakamura R (2002) Synchronized activity in
prefrontal cortex during anticipation of visuomotor processing. Neuroreport 13:2011-2015.

Linkenkaer-Hansen K, Nikulin VV, Palva S, Ilmoniemi RJ, Palva JM (2004) Prestimulus
oscillations enhance psychophysical performance in humans. Journal of Neuroscience
24:10186-10190.

Lisman JE, Raghavachari S, Tsien RW (2007) The sequence of events that underlie quantal
transmission at central glutamatergic synapses. Nature Reviews Neuroscience 8:597-609.

Loh M, Pasupathy A, Miller EK, Deco G (2008) Neurodynamics of the prefrontal cortex during
conditional visuomotor associations. Journal of Cognitive Neuroscience 20:421-431.

Luck SJ (2005) An introduction to the event-related potential technique. Cambridge: MIT Press.

Luck SJ, Chelazzi L, Hillyard SA, Desimone R (1997) Neural mechanisms of spatial selective
attention in areas VI, V2, and V4 of macaque visual cortex. Journal of Neurophysiology
77:24-42.

Luks TL, Sun FT, Dale CL, Miller WL, Simpson GV (2008) Transient and sustained brain
activity during anticipatory visuospatial attention. Neuroreport 19:155-159.

Lungarella M, Sporns O (2006) Mapping information flow in sensorimotor networks. PLoS
Computational Biology 2:1301-1312.

Macklin JT (1982) An investigation of the properties of double radio sources using Spearman
partial rank correlation coefficient. Royal Astronomical Society, Monthly Notices
199:1119-1136.

Makeig S, Westerfield M, Jung TP, Enghoff S, Townsend J, Courchesne E, Sejnowski TJ (2002)
Dynamic brain sources of visual evoked responses. Science 295:690-694.

Malsburg CVD (1995) Binding in models of perception and brain function. Current Opinion in
Neurobiology 5:520-526.

Mangun GR, Hillyard SA (1991) Modulations of sensory-evoked brain potentials indicate
changes in perceptual processing during visual spatial priming. Journal of Experimental
Psychology-Human Perception and Performance 17:1057-1074.

Mangun GR, Buonocore MH, Girelli M, Jha AP (1998) ERP and fMRI measures of visual
spatial selective attention. Human Brain Mapping 6:383-389.

Mangun GR, Hopfinger JB, Kussmaul CL, Fletcher EM, Heinze HJ (1997) Covariations in ERP
and PET measures of spatial selective attention in human extrastriate visual cortex. Human
Brain Mapping 5:273-279.

Marrufo MV, Vaquero E, Cardoso MJ, Gomez CM (2001) Temporal evolution of alpha and beta
bands during visual spatial attention. Cognitive Brain Research 12:315-320.









Mazaheri A, Jensen O (2006) Posterior alpha activity is not phase-reset by visual stimuli.
Proceedings of the National Academy of Sciences of the United States of America
103:2948-2952.

McCormick DA, Shu YS, Hasenstaub A, Sanchez-Vives M, Badoual M, Bal T (2003) Persistent
cortical activity: Mechanisms of generation and effects on neuronal excitability. Cerebral
Cortex 13:1219-1231.

Meador KJ (2007) The basic science of memory as it applies to epilepsy. Epilepsia 48:23-25.

Melloni L, Molina C, Pena M, Torres D, Singer W, Rodriguez E (2007) Synchronization of
neural activity across cortical areas correlates with conscious perception. Journal of
Neuroscience 27:2858-2865.

Miltner WH, Braun C, Arnold M (1999) Coherence of gamma-band EEG activity as a basis for
associative learning. Nature 397:917-924.

Mitra P, Bokil H (2007) Observed brain dynamics. New York: Oxford University Press.

Mitra PP, Pesaran B (1999) Analysis of dynamic brain imaging data. Biophysical Journal
76:691-708.

Mitzdorf U (1985) Current source-density method and application in cat cerebral-cortex -
Investigation of evoked-potentials and EEG Phenomena. Physiological Reviews 65:37-
100.

Munk MHJ, Roelfsema PR, Konig P, Engel AK, Singer W (1996) Role of reticular activation in
the modulation of intracortical synchronization. Science 272:271-274.

Murias M, Swanson JM, Srinivasan R (2007) Functional connectivity of frontal cortex in healthy
and ADHD children reflected in EEG coherence. Cerebral Cortex 17:1788-1799.

Nichols TE, Holmes AP (2002) Nonparametric permutation tests for functional neuroimaging: A
primer with examples. Human Brain Mapping 15:1-25.

Niedermeyer E, Lopes Da Silva F (2005) Electroencephalography: Basic principles, clinical
applications, and related fields. Philadelphia: Lippincott Williams & Wilkins.

Nunez PL, Srinivasan R, Westdorp AF, Wijesinghe RS, Tucker DM, Silberstein RB, Cadusch PJ
(1997) EEG coherency. 1: Statistics, reference electrode, volume conduction, Laplacians,
cortical imaging, and interpretation at multiple scales. Electroencephalography and
Clinical Neurophysiology 103:499-515.

Ohara S, Crone NE, Weiss N, Lenz FA (2006) Analysis of synchrony demonstrates 'pain
networks' defined by rapidly switching, task-specific, functional connectivity between
pain-related cortical structures. Pain 123:244-253.









Ohara S, Crone NE, Weiss N, Kim JH, Lenz FA (2008) Analysis of synchrony demonstrates that
the presence of "pain networks" prior to a noxious stimulus can enable the perception of
pain in response to that stimulus. Experimental Brain Research 185:353-358.

Okatan M, Wilson MA, Brown EN (2005) Analyzing functional connectivity using a network
likelihood model of ensemble neural spiking activity. Neural Computation 17:1927-1961.

Otten LJ, Henson RNA, Rugg MD (2002) State-related and item-related neural correlates of
successful memory encoding. Nature Neuroscience 5:1339-1344.

Otten LJ, Quayle AH, Akram S, Ditewig TA, Rugg MD (2006) Brain activity before an event
predicts later recollection. Nature Neuroscience 9:489-491.

Padilla ML, Wood RA, Hale LA, Knight RT (2006) Lapses in a prefrontal-extrastriate
preparatory attention network predict mistakes. Journal of Cognitive Neuroscience
18:1477-1487.

Palva JM, Palva S, Kaila K (2005) Phase synchrony among neuronal oscillations in the human
cortex. Journal of Neuroscience 25:3962-3972.

Paul MD, Simon PK, John JF, Richard BR, Ian HR (2007) Optimal sustained attention is linked
to the spectral content of background EEG activity: Greater ongoing tonic alpha
(approximately 10 Hz) power supports successful phasic goal activation. European Journal
of Neuroscience 25:900-907.

Perrin F, Pemier J, Bertrand O, Echallier JF (1989) Spherical splines for scalp potential and
current-density mapping. Electroencephalography and Clinical Neurophysiology 72:184-
187.

Petersen CCH, Hahn TTG, Mehta M, Grinvald A, Sakmann B (2003) Interaction of sensory
responses with spontaneous depolarization in layer 2/3 barrel cortex. Proceedings of the
National Academy of Sciences of the United States of America 100:13638-13643.

Petrides M, Milner B (1982) Deficits on subject-ordered tasks after frontal-lobe and temporal-
lobe lesions in man. Neuropsychologia 20:249-262.

Petrides M, Pandya DN (1999) Dorsolateral prefrontal cortex: Comparative cytoarchitectonic
analysis in the human and the macaque brain and corticocortical connection patterns.
European Journal of Neuroscience 11:1011-1036.

Petrides M, Pandya DN (2002) Comparative cytoarchitectonic analysis of the human and the
macaque ventrolateral prefrontal cortex and corticocortical connection patterns in the
monkey. European Journal of Neuroscience 16:291-310.

Pfurtscheller G, da Silva FHL (1999) Event-related EEG/MEG synchronization and
desynchronization: Basic principles. Clinical Neurophysiology 110:1842-1857.









Pliszka SR, Liotti M, Woldorff MG (2000) Inhibitory control in children with attention-
deficit/hyperactivity disorder: Event-related potentials identify the processing component
and timing of an impaired right-frontal response-inhibition mechanism. Biological
Psychiatry 48:238-246.

Poulet JFA, Petersen CCH (2008) Internal brain state regulates membrane potential synchrony in
barrel cortex of behaving mice. Nature 454:881-885.

Raghavachari S, Kahana MJ, Rizzuto DS, Caplan JB, Kirschen MP, Bourgeois B, Madsen JR,
Lisman JE (2001) Gating of human theta oscillations by a working memory task. Journal
of Neuroscience 21:3175-3183.

Rajagovindan R, Ding M (2008a) Selective visual attention increases neuronal excitability and
gain in humans: A high-density EEG study. Society for Neuroscience Abstract 34:436.12.

Rajagovindan R, Ding M (2008b) Decomposing neural synchrony: Toward an explanation for
near-zero phase-lag in cortical oscillatory networks. PLoS ONE 3:1-8.

Rajagovindan R, Anderson K, Ghacibeh G, Meador KJ, Ding M (2007) Prestimulus theta
oscillations in human cortex during encoding predict subsequent memory recall. Society
for Neuroscience Abstract 33:635.14.

Rempel-Clower NL, Barbas H (2000) The laminar pattern of connections between prefrontal and
anterior temporal cortices in the rhesus monkey is related to cortical structure and function.
Cerebral Cortex 10:851-865.

Rihs TA, Michel CM, Thut G (2007) Mechanisms of selective inhibition in visual spatial
attention are indexed by alpha-band EEG synchronization. European Journal of
Neuroscience 25:603-610.

Rodriguez E, George N, Lachaux JP, Martinerie J, Renault B, Varela FJ (1999) Perception's
shadow: Long-distance synchronization of human brain activity. Nature 397:430-433.

Roelfsema PR, Engel AK, Konig P, Singer W (1997) Visuomotor integration is associated with
zero time-lag synchronization among cortical areas. Nature 385:157-161.

Romei V, Rihs T, Brodbeck V, Thut G (2008a) Resting electroencephalogram alpha-power over
posterior sites indexes baseline visual cortex excitability. Neuroreport 19:203-208.

Romei V, Brodbeck V, Michel C, Amedi A, Pascual-Leone A, Thut G (2008b) Spontaneous
fluctuations in posterior alpha-band EEG activity reflect variability in excitability of
human visual areas. Cerebral Cortex 18:2010-2018.

Roskies AL (1999) The binding problem. Neuron 24:7-9.

Saalmann YB, Pigarev IN, Vidyasagar TR (2007) Neural mechanisms of visual attention: How
top-down feedback highlights relevant locations. Science 316:1612-1615.









Salinas E, Sejnowski TJ (2000) Impact of correlated synaptic input on output firing rate and
variability in simple neuronal models. Journal of Neuroscience 20:6193-6209.

Salinas E, Sejnowski TJ (2001) Correlated neuronal activity and the flow of neural information.
Nature Reviews Neuroscience 2:539-550.

Sarnthein J, Petsche H, Rappelsberger P, Shaw GL, von Stein A (1998) Synchronization between
prefrontal and posterior association cortex during human working memory. Proceedings of
the National Academy of Sciences of the United States of America 95:7092-7096.

Sauseng P, Klimesch W, Stadler W, Schabus M, Doppelmayr M, Hanslmayr S, Gruber WR,
Birbaumer N (2005) A shift of visual spatial attention is selectively associated with human
EEG alpha activity. European Journal of Neuroscience 22:2917-2926.

Schack B, Rappelsberger P, Anders C, Weiss S, Moller E (2000) Quantification of
synchronization processes by coherence and phase and its application in analysis of
electrophysiological signals. International Journal of Bifurcation and Chaos, 10:2565-
2586.

Scherg M, Ille N, Bornfleth H, Berg P (2002) Advanced tools for digital EEG review: Virtual
source montages, whole-head mapping, correlation, and phase analysis. Journal of Clinical
Neurophysiology 19:91-112.

Schroeder CE, Mehta AD, Foxe JJ (2001) Determinants and mechanisms of attentional
modulation of neural processing. Frontiers in Bioscience 6:D672-D684.

Schroeder CE, Tenke CE, Givre SJ, Arezzo JC, Vaughan HG (1991) Striate cortical contribution
to the surface-recorded pattern-reversal VEP in the alert monkey. Vision Research
31:1143-1157.

Scoville WB, Milner B (1957) Loss of recent memory after bilateral hippocampal lesions.
Journal of Neurology Neurosurgery and Psychiatry 20:11-21.

Seager MA, Johnson LD, Chabot ES, Asaka Y, Berry SD (2002) Oscillatory brain states and
learning: Impact of hippocampal theta-contingent training. Proceedings of the National
Academy of Sciences of the United States of America 99:1616-1620.

Sederberg PB, Kahana MJ, Howard MW, Donner EJ, Madsen JR (2003) Theta and gamma
oscillations during encoding predict subsequent recall. Journal of Neuroscience 23:10809-
10814.

Sederberg PB, Schulze-Bonhage A, Madsen JR, Bromfield EB, McCarthy DC, Brandt A, Tully
MS, Kahana MJ (2007) Hippocampal and neocortical gamma oscillations predict memory
formation in humans. Cerebral Cortex 17:1190-1196.

Seth AK (2005) Causal connectivity of evolved neural networks during behavior. Network-
Computation in Neural Systems 16:35-54.









Shimamura AP (1995) Memory and the prefrontal cortex. Structure and Functions of the Human
Prefrontal Cortex 769:151-159.

Shimamura AP, Jurica PJ, Mangels JA, Gershberg FB, Knight RT (1995) Susceptibility to
memory interference effects following frontal-lobe damage findings from tests of paired-
associate learning. Journal of Cognitive Neuroscience 7:144-152.

Siapas AG, Lubenov EV, Wilson MA (2005) Prefrontal phase locking to hippocampal theta
oscillations. Neuron 46:141-151.

Siegel M, Donner TH, Oostenveld R, Fries P, Engel AK (2008) Neuronal synchronization along
the dorsal visual pathway reflects the focus of spatial attention. Neuron 60:709-719.

Sieroff E, Piquard A, Auclair L, Lacomblez L, Derouesne C, Laberge D (2004) Deficit of
preparatory attention in frontotemporal dementia. Brain Cognition 55:444-451.

Simons JS, Spiers HJ (2003) Prefrontal and medial temporal lobe interactions in long-term
memory. Nature Reviews Neuroscience 4:637-648.

Singer W, Gray CM (1995) Visual feature integration and the temporal correlation hypothesis.
Annual Review of Neuroscience 18:555-586.

Siok CJ, Rogers JA, Kocsis B, Hajos M (2006) Activation of alpha 7 acetylcholine receptors
augments stimulation-induced hippocampal theta oscillation. European Journal of
Neuroscience 23:570-574.

Skinner JE, Yingling CD (1976) Regulation of slow potential shifts in nucleus reticularis thalami
by mesencephalic reticular-formation and frontal granular cortex. Electroencephalography
and Clinical Neurophysiology 40:288-296.

Slotnick SD, Moo LR, Kraut MA, Lesser RP, Hart J (2002) Interactions between thalamic and
cortical rhythms during semantic memory recall in human. Proceedings of the National
Academy of Sciences of the United States of America 99:6440-6443.

Staubli U, Xu FB (1995) Effects of 5-HT3 receptor antagonism on hippocampal theta-rhythm,
memory, and LTP induction in the freely moving rat. Journal of Neuroscience 15:2445-
2452.

Strange BA, Otten LJ, Josephs O, Rugg MD, Dolan RJ (2002) Dissociable human perirhinal,
hippocampal, and parahippocampal roles during verbal encoding. Journal of Neuroscience
22:523-528.

Strens LHA, Oliviero A, Bloem BR, Gerschlager W, Rothwell JC, Brown P (2002) The effects
of subthreshold 1 Hz repetitive TMS on cortico-cortical and interhemispheric coherence.
Clinical Neurophysiology 113:1279-1285.

Summerfield C, Mangels JA (2005) Coherent theta-band EEG activity predicts item-context
binding during encoding. Neuroimage 24:692-703.









Takashima A, Jensen O, Oostenveld R, Maris E, van de Coevering M, Fernandez G (2006)
Successful declarative memory formation is associated with ongoing activity during
encoding in a distributed neocortical network related to working memory: A
magnetoencephalography study. Neuroscience 139:291-297.

Tallon-Baudry C, Mandon S, Freiwald WA, Kreiter AK (2004) Oscillatory synchrony in the
monkey temporal lobe correlates with performance in a visual short-term memory task.
Cerebral Cortex 14:713-720.

Taylor PCJ, Nobre AC, Rushworth MFS (2007) FEF TMS affects visual cortical activity.
Cerebral Cortex 17:391-399.

Tendolkar I, Rugg M, Fell J, Vogt H, Scholz M, Hinrichs H, Heinze HJ (2000) A
magnetoencephalographic study of brain activity related to recognition memory in healthy
young human subjects. Neuroscience Letters 280:69-72.

Thomson AM, Lamy C (2007) Functional maps of neocortical local circuitry. Frontiers in
Neuroscience 1:19-42.

Thomson DJ (1982) Spectrum estimation and harmonic analysis. Proceedings of the IEEE
70:1055-1096.

Thomson DJ (2007) Jackknifing multitaper spectrum estimates. Signal Processing Magazine,
IEEE 24:20-30.

Thut G, Nietzel A, Brandt SA, Pascual-Leone A (2006) Alpha-band electroencephalographic
activity over occipital cortex indexes visuospatial attention bias and predicts visual target
detection. Journal of Neuroscience 26:9494-9502.

Truccolo W, Eden UT, Fellows MR, Donoghue JP, Brown EN (2005) A point process
framework for relating neural spiking activity to spiking history, neural ensemble, and
extrinsic covariate effects. Journal of Neurophysiology 93:1074-1089.

Tulving E, Kapur S, Craik FIM, Moscovitch M, Houle S (1994) Hemispheric encoding/retrieval
asymmetry in episodic memory positron emission tomography findings. Proceedings of
the National Academy of Sciences of the United States of America 91:2016-2020.

Ungerleider LG, Gaffan D, Pelak VS (1989) Projections from inferior temporal cortex to
prefrontal cortex via the uncinate fascicle in rhesus-monkeys. Experimental Brain
Research 76:473-484.

Van Dijk H, Schoffelen JM, Oostenveld R, Jensen O (2008) Prestimulus oscillatory activity in
the alpha band predicts visual discrimination ability. Journal of Neuroscience 28:1816-
1823.

Varela F, Lachaux JP, Rodriguez E, Martinerie J (2001) The brainweb: Phase synchronization
and large-scale integration. Nature Reviews Neuroscience 2:229-239.









Vertes RP (2005) Hippocampal theta rhythm: A tag for short-term memory. Hippocampus
15:923-935.

Vertes RP, Hoover WB, Di Prisco GV (2004) Theta rhythm of the hippocampus: Subcortical
control and functional significance. Behavioral and Cognitive Neuroscience Reviews
3:173-200.

Volgushev M, Chistiakova M, Singer W (1998) Modification of discharge patterns of neocortical
neurons by induced oscillations of the membrane potential. Neuroscience 83:15-25.

von Stein A, Samthein J (2000) Different frequencies for different scales of cortical integration:
from local gamma to long range alpha/theta synchronization. International Journal of
Psychophysiology 38:301-313.

von Stein A, Chiang C, Konig P (2000) Top-down processing mediated by interareal
synchronization. Proceedings of the National Academy of Sciences of the United States of
America 97:14748-14753.

Wagner AD, Poldrack RA, Eldridge LL, Desmond JE, Glover GH, Gabrieli JDE (1998)
Material-specific lateralization of prefrontal activation during episodic encoding and
retrieval. Neuroreport 9:3711-3717.

Wang X, Chen YH, Ding MZ (2007a) Testing for statistical significance in bispectra: A
surrogate data approach and application to neuroscience. IEEE Transactions on Biomedical
Engineering 54:1974-1982.

Wang X, Chen YH, Bressler SL, Ding MZ (2007b) Granger causality between multiple
interdependent neurobiological time series: Blockwise versus pairwise methods.
International Journal of Neural Systems 17:71-78.

Watson RT, Valenstein E, Heilman KM (1981) Thalamic neglect possible role of the medial
thalamus and nucleus reticularis in behavior. Archives of Neurology 38:501-506.

Weiss S, Rappelsberger P (2000) Long-range EEG synchronization during word encoding
correlates with successful memory performance. Cognitive Brain Research 9:299-312.

Weissman DH, Roberts KC, Visscher KM, Woldorff MG (2006) The neural bases of momentary
lapses in attention. Nature Neuroscience 9:971-978.

Wiener N (1956) The theory of prediction. New York: McGraw-Hill.

Winson J (1978) Loss of hippocampal theta rhythm results in spatial memory deficit in rat.
Science 201:160-163.

Witham CL, Wang M, Baker SN (2007) Cells in somatosensory areas show synchrony with beta
oscillations in monkey motor cortex. European Journal of Neuroscience 26:2677-2686.









Wolfart J, Debay D, Le Masson G, Destexhe A, Bal T (2005) Synaptic background activity
controls spike transfer from thalamus to cortex. Nature Neuroscience 8:1760-1767.

Womelsdorf T, Fries P, Mitra PP, Desimone R (2006) Gamma-band synchronization in visual
cortex predicts speed of change detection. Nature 439:733-736.

Worden MS, Foxe JJ, Wang N, Simpson GV (2000) Anticipatory biasing of visuospatial
attention indexed by retinotopically specific alpha-band electroencephalography increases
over occipital cortex. Journal of Neuroscience 26:9494-9502.

Wu JH, Liu XG, Feng JF (2008) Detecting causality between different frequencies. Journal of
Neuroscience Methods 167:367-375.

Yago E, Duarte A, Wong T, Barcelo F, Knight RT (2004) Temporal kinetics of prefrontal
modulation of the extrastriate cortex during visual attention. Cognitive Affective &
Behavioral Neuroscience 4:609-617.

Yingling CD, Skinner JE (1975) Regulation of unit-activity in nucleus reticularis thalami by
mesencephalic reticular-formation and frontal granular cortex. Electroencephalography and
Clinical Neurophysiology 39:635-642.

Yingling CD, Skinner JE (1976) Selective regulation of thalamic sensory relay nuclei by nucleus
reticularis thalami. Electroencephalography and Clinical Neurophysiology 41:476-482.

Zhang Y, Ding M (in press) Detection of a weak somatosensory stimulus: Role of the
prestimulus mu rhythm and its top-down modulation. Journal of Cognitive Neuroscience.

Zhang Y, Chen Y, Bressler SL, Ding M (2008a) Response preparation and inhibition: The role of
the cortical sensorimotor beta rhythm. Neuroscience 156:238-246.

Zhang Y, Wang X, Bressler SL, Chen YH, Ding MZ (2008b) Prestimulus cortical activity is
correlated with speed of visuomotor processing. Journal of Cognitive Neuroscience
20:1915-1925.

Zikopoulos B, Barbas H (2007) Circuits for multisensory integration and attentional modulation
through the prefrontal cortex and the thalamic reticular nucleus in primates. Reviews in the
Neurosciences 18:417-438.









BIOGRAPHICAL SKETCH

Rajasimhan Rajagovindan was born in 1983, in Chennai, India. He grew up in Chennai,

graduating from Padma Seshadri Bala Bhavan Senior Secondary School in 2000. He earned the

Bachelor of Engineering degree in electronics and communication engineering from the

University of Madras, Chennai in 2004 following which Raj enrolled in the graduate program at

the University of Florida where he earned the Master of Science degree in biomedical

engineering in 2008. He continued at the University of Florida to earn his Doctor of Philosophy

in the fall of 2009 in biomedical engineering specializing in cognitive neuroscience under the

mentorship of Dr.Mingzhou Ding. Raj intends to pursue his research interests in functional

biomarker discovery to develop diagnostic tools to identify and assess the progression of

neurological and psychiatric illnesses and the effectiveness of therapeutic interventions to enable

therapy augmentation.





PAGE 1

FUNCTIONAL ROLE OF NEURAL OSCILLATIONS IN ATTENTION AND MEMORY IN HUMANS By RAJASIMHAN RAJAGOVINDAN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009 1

PAGE 2

2009 Rajasimhan Rajagovindan 2

PAGE 3

To my Mom and Dad, for their love and support 3

PAGE 4

ACKNOWLEDGMENTS This project could be accomp lished only with the outstan ding guidance and patience and generous support of my mentor Dr.Mingzhou Ding to whom I am deeply indebted for imbibing in me an exceptional perspective on highest qu ality research, and also for his constant encouragement and constructive role throughout my stay at UF. I cannot thank him enough for the numerous opportunities he has presented me with through all these years that have helped me gain invaluable experience and confidence. I sincerely thank Prof.Venka teswaran Nagarajan of WARFT, for nurturing my early interest to study the brain and w ith whose motivation I began to pur sue studies in neuroscience. I also owe a deep debt of grat itude to Dr.Kimford Meador, Dr.B enzi Kluger, Dr.Linda Hermer, Dr. Thomas DeMarse, Dr.Georges Ghacibeh, Dr.Andreas Keil, Dr.Hans van Oostrom, Dr.Paul Carney and Dr. Govindan Rangarajan for their ch eerful enthusiasm, constant support and with whom I have the had the pleasure of collaborati ng and learn on numerous occasions that has contributed immensely to my scientific development. I am also grateful to Dr.Clifford Saron and Dr.George Mangun at UC Davis for their kind reception, the many insightful discussions that we had during my short visit to Davis and for sharin g a ton of their excellent experimental data. A special thanks to Kristopher Anderson for an extremely fruitful collaboration over the past years and for making the project with the ep ilepsy patients happen. I am also thankful to Dr.Kenneth Heilman, Dr.Susan Marino and Dr.Dav id Loring for all the insightful discussions during our work with the patien t population. I would like to th ank all the epilepsy monitoring unit staff, Cathy, Bonnie, Becky, Andrea, and Augus ta at the Shands Hospital at the University of Florida, for their kind assistance throughout that study. I would also like to thank Tasmeer Hassan and Joshua Proemsey for their support with data acquisition for th e PD-fatigue project. 4

PAGE 5

I am deeply grateful to all the volunteers who graciously endured our studies, without whose participation none of the studies woul d have been possible. I am also extremely thankful to all my colleagues and friends including Xue Wang, Yonghong Chen, Mukesh Dhamala, Alex Cadotte, Xianzhi Shao, Yan Zhang, Anil Bollimunta, Sahng Min Han, Hariharan Nalatore, Mo Jue, Yuelu Lui, Xiaotong Wen, Chao Wang, Chien JuiHong, Il Park, Sanjeev Malkani and Krishna Bhar ath for the good times and for the many fruitful discussions that we have had not just limited to science that has been an integral part of my experience at UF. I owe a tremendous amount of gratitude to the past and present BME family including Dr.Hans van Oostrom (once again!), Katherine Whitesides, Tifiny Dyer, Anide Pierre-Louis, April Derfinyak, Laura Studstill, Michelle Gr iffin, Mary Wall, Danielle Wise and Art Bautista-Hardman who handled all the financial, administrative, computing related resources and so many other nightmares that they made sure I did not even have to worry about and their immense patience and delightful support while I re peatedly goofed up on my registration, all my paper work and not to mention the computer mishap s! Above all, my heartf elt thanks to my mom and dad and all my teachers through all these y ears who enriched me and my experience, for their constant encouragement and support, their infinite patience, their affection and their unwavering confidence in me. 5

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4 LIST OF TABLES ...........................................................................................................................9 LIST OF FIGURES .......................................................................................................................10 ABSTRACT ...................................................................................................................... .............12 CHAPTER 1 INTRODUCTION ................................................................................................................ ..14 2 GENERAL METHODS .........................................................................................................21 2.1 Parametric Multivariate Spectral Estimation ................................................................22 2.2 Granger Causality Analysis...........................................................................................25 2.2.1 Time Domain Formulation ................................................................................25 2.2.2 Frequency Domain Formulation .......................................................................27 2.2.3 Mathematical Formulation of Block Granger Causality ...................................32 2.2.4 Interpretation .....................................................................................................34 2.2.5 Assessment of Statistical Significance ..............................................................35 2.3 Multitaper Spectral Estimation .....................................................................................35 3 DECOMPOSING NEURAL SYNCHRONY: TOWARD AN EXPLANATION FOR NEAR-ZERO PHASE-LAG IN CORTIC AL OSCILLATORY NETWORKS ....................39 3.1 Introduction ...................................................................................................................39 3.2 Materials and Methods ..................................................................................................41 3.2.1 Simulation .........................................................................................................41 3.2.1.1 Setup ...................................................................................................41 3.2.1.2 Positively correlated common input ...................................................41 3.2.1.3 Bidirectional interaction .....................................................................42 3.2.2 Experiment ........................................................................................................42 3.2.2.1 Behavioral paradigm ...........................................................................42 3.2.2.2 Data acquisition ..................................................................................43 3.2.2.3 Data set selection ................................................................................43 3.2.2.4 Time series decomposition .................................................................43 3.2.2.5 Data analysis protocol .........................................................................45 3.2.3 Logic of Analysis Protocol ...............................................................................46 3.3 Results ...........................................................................................................................47 3.3.1 Simulation .........................................................................................................47 3.3.2 Experiment ........................................................................................................47 3.3.2.1 Network identification ........................................................................47 3.3.2.2 Phase-lag distribution .........................................................................48 6

PAGE 7

3.3.2.3 Synchrony decomposition and near-zero phase-lag ...........................48 3.3.2.4 Instantaneous causality and phase-lag ................................................49 3.3.2.5 Bidirectional intera ction and phase-lag ..............................................49 3.4 Discussion .....................................................................................................................50 4 PRESTIMULUS THETA OS CILLATIONS IN FRONTO-TEMPORAL NETWORK PREDICT SUBSEQUENT MEMO RY RECALL IN HUMANS ..........................................58 4.1 Introduction ...................................................................................................................58 4.2 Materials and Methods ..................................................................................................60 4.2.1 Subjects and Electrode Placement ....................................................................60 4.2.2 Paradigm ...........................................................................................................60 4.2.3 Data Preprocessing ............................................................................................61 4.2.4 Spectral Analysis ...............................................................................................62 4.2.4.1 Power estimation ................................................................................62 4.2.4.2 Significance test for differen ce in power betw een conditions ............62 4.2.4.3 Group power analysis .........................................................................63 4.2.4.4 Inter-areal synchrony analysis ............................................................63 4.2.4.5 Random permutation test fo r statistical si gnificance ..........................64 4.3 Results ...........................................................................................................................65 4.3.1 Power ................................................................................................................65 4.3.2 Inter-Areal Phase Synchrony ............................................................................66 4.4 Discussion .....................................................................................................................67 4.4.1 Prestimulus Ongoing Activity, Theta Os cillations and Memory Encoding ......68 4.4.2 Inter-Areal Synchrony a nd Network Excitability .............................................69 5 FROM PRESTIMULUS ALPHA OSCILLATION TO VISUAL EVOKED RESPONSE: AN INVERTED-U F UNCTION AND ITS ATTENTIONAL MODULATION .................................................................................................................... .75 5.1 Introduction ...................................................................................................................75 5.2 Materials and Methods ..................................................................................................77 5.2.1 Subjects .............................................................................................................77 5.2.2 Paradigm ...........................................................................................................77 5.2.3 Data Acquisition ...............................................................................................78 5.2.4 Data Preprocessing ............................................................................................79 5.2.5 Behavior and Event-re lated Potential Analysis.................................................80 5.2.6 Correlation Between Prestimulus Alpha Power and P1 Amplitude..................80 5.2.7 Characterization of Ongoing Alpha Activity ....................................................81 5.2.8 Computational Modeling ..................................................................................82 5.3 Results ...........................................................................................................................83 5.3.1 Behavior ............................................................................................................84 5.3.2 ERP Analysis ....................................................................................................84 5.3.3 Prestimulus Alpha Oscillations .........................................................................85 5.3.4 From Ongoing Activity to Evoked Response: A Model ...................................86 5.3.4.1 The sigmoidal input-output function ..................................................86 5.3.4.2 Model reformulation in terms of electroencephalogram variables .....87 7

PAGE 8

5.3.4.3 Experimental test of model predictions ..............................................88 5.3.5 Attentional Modulation of Ba seline Ongoing Activity: Possible Mechanisms ......................................................................................................89 5.3.6 Role of Excitability: Validation in a Single Neuron Model ..............................91 5.4 Discussion .....................................................................................................................92 5.4.1 Theoretical Model .............................................................................................93 5.4.2 Model Testing and Validation ...........................................................................94 5.4.3 P1 Genesis and Attentional Modulation ...........................................................94 5.4.4 Alpha Oscillation: Neuronal Mech anisms and Attentional Modulation ...........95 6 TOP-DOWN CONTROL OF HUMAN VISUAL CORTEX IN ANTICIPATORY SPATIAL VISUAL ATTENTION .......................................................................................107 6.1 Introduction .................................................................................................................107 6.2 Materials and Methods ................................................................................................110 6.2.1 Subjects ...........................................................................................................110 6.2.2 Paradigm .........................................................................................................110 6.2.3 Data Acquisition .............................................................................................111 6.2.4 Data Preprocessing ..........................................................................................111 6.2.5 Data Analysis ..................................................................................................112 6.3 Results .........................................................................................................................114 6.4 Discussion ...................................................................................................................117 6.4.1 Alpha Band Oscillation: A Neural Correlate of Top-Down Attentional Control ............................................................................................................118 6.4.2 Top-Down Control of Visual Al pha: Excitation or Disinhibition? .................119 6.4.3 Top-Down Control of Visual Alpha: Transient or Sustained?........................123 7 CONCLUSION .................................................................................................................. ...129 REFERENCES .................................................................................................................... ........134 BIOGRAPHICAL SKETCH .......................................................................................................153 8

PAGE 9

LIST OF TABLES Table page 3-1 Influence of increased instan taneous causality on phase-lag .............................................53 3-2 Influence of increased bidirec tional interaction on phase-lag ...........................................53 3-3 Mean beta band phase-lag (in ra dians) for monkeys GE and LU ......................................54 3-4 Contribution of intrinsic and exogenous components to total interdependence ................54 3-5 Correlation between instantaneous causality and phase-lag ..............................................54 3-6 Partial correlation between instan taneous causality and phase-lag ...................................54 4-1 ECoG electrode information.. ............................................................................................70 4-2 ECoG bipolar electr ode derivation summary.. ..................................................................70 6-1 Attention modulation of oscillatory activity. ...................................................................124 9

PAGE 10

LIST OF FIGURES Figure page 3-1 Granger causality spectra ................................................................................................. ..553-2 Schematic Granger causality graph ...................................................................................553-3 Phase distribution ........................................................................................................ .......563-4 Influence of common input on phase-lag ...........................................................................563-5 Coherence and phase spectra for site pair (S1,7b) in monkey GE .....................................573-6 Phase locking modes ....................................................................................................... ...574-1 Approximate placement of electrode gr ids for each of the five subjects. ..........................714-2 Schematic of the experimental paradigm ...........................................................................714-3 Prestimulus differences in spectral power between the recalled and unrecalled ensembles of words ............................................................................................................ 724-4 Comparison of theta band power .......................................................................................734-5 Comparison of inter-areal synchrony.................................................................................734-6 Comparison of percentage of bipolar signal pairs with significant PLVs .........................745-1 Experimental task.......................................................................................................... .....975-2 ERP attention effect ...........................................................................................................985-3 Scalp distribution of attentional enhancement ...................................................................995-4 Attentional modulation of alpha band activity .................................................................1005-5 Schematic illustration of theoretical model .....................................................................1015-6 Model reformulation in terms of EEG variables ..............................................................1025-7 Relationship between prestimulus al pha power and stimulus processing .......................1035-8 Schematic illustration of attentional modulation mechanisms of ongoing activity .........1035-9 Gain ratio as a function of prestimulus alpha ..................................................................1045-10 Distribution of alpha power .............................................................................................1 05 10

PAGE 11

5-11 Computational study of relation between excitability and gain .......................................1066-1 Experimental task.......................................................................................................... ...1256-2 Top-down attentional co ntrol of visual areas ...................................................................1266-3 Pre-target alpha band activ ity predicts performance .......................................................1276-4 Temporal evolution of regional a nd network-level alpha band activity ..........................128 11

PAGE 12

Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy FUNCTIONAL ROLE OF NEURAL OSCILLATIONS IN ATTENTION AND MEMORY IN HUMANS By Rajasimhan Rajagovindan December 2009 Chair: Mingzhou Ding Major: Biomedical Engineering Synchronized oscillatory neural activity involving widespread neural ensemble is common in the central nervous system, and is hypothesized to enable the integration of various attributes of the sensory input, resulting in a unified per cept. While activity evoked to external sensory input has been instantly recognized to refl ect stimulus processing, ongoing neural activity preceding stimulus onset is often considered as background noise and irrelevant to subsequent stimulus processing. Our study seeks to advance the idea that ongoing neur al oscillations are critical indices of the brain st ate and, as such, play an important role in stimulus-evoked cognitive processing. Applying advanced multivaria te spectral analysis to invasive and noninvasive electrophysiological reco rdings from animals and humans, we address the mechanisms and the functional roles of synchr onized prestimulus oscillatory ac tivity in processes such as anticipatory attention, visual processing and encoding of memory. First, applying computational techniques to in vasive local field pot ential recordings from primates we expound the mechanisms leading to zero-lag synchrony across widely separated cortical areas, a phenomenon critic ally implicated in effective neuronal communication and thus higher-order perceptual processe s. Second, we assessed the functi onal role of prestimulus theta band oscillation in memory related stimulus en coding in the human cortex using invasive 12

PAGE 13

13 electrocorticogram. Third, a physiological model linking prestimulus oscillations and stimulusevoked response is proposed and is tested on non invasive studies of human visual processing. Finally, the neural control of anticipatory attention a nd stimulus selection using electroencephalogram is investigated. The objectiv e here is to expound the signals and the neural mechanisms that mediate these top-down processes that enable selective sensory integration during attentive behavior in humans. In conclusion, our findings suggest that effective sensory-pe rceptual processing and thus cognitive and behavioral performance to an impe nding stimulus, depends on the excitability state of the brain defined by these neur al oscillations at its onset. Further, it is shown that the prestimulus activity is not merely a random fluctu ation in excitability bu t one that is actively controlled in a goal-directed manner through ne ural synchronization across widespread functionally relevant cortical networks. Thus, this study bol sters the functional role of prestimulus ongoing activity in subsequent sens ory-perceptual and higher-order cognitive processing.

PAGE 14

CHAPTER 1 INTRODUCTION The ubiquitous nature of oscillatory neural act ivity in the human and animal brain is well established. In 1989, Gray, Singer and colleagues (1989), showed that neurons with distinct receptive fields in the primary visual cortex when jointly stimulated with one perceptual object, produced synchronized oscillatory responses. On the other hand, when stimulated by two distinct perceptual objects in their respective receptive fi elds, the firing rates remained unaltered, and the synchrony was abolished. Ever since this early study, it has been shown consistently that all cognitive processes involve transi ent integration of several widely distributed cortical areas (Singer and Gray, 1995; Malsburg, 1995; Rosk ies, 1999; von Stein et al., 2000). These synchronous oscillations, resul ting from dynamic interaction betw een distant yet functionally relevant cortical regi ons, have been suggested as a put ative mechanism for the large-scale integration of sensory inputs, known as the bin ding problem (Eckhorn et al., 1988; Sarnthein et al., 1998; Weiss and Rappelsberge r, 2000; Engel et al., 2001; Varela et al., 2001; Fries, 2005; Jones and Wilson, 2005). Numerous studies have recently reported synchrony with zero time lag across widely separated cortical regions, and have interpreted the observation as precise timing of neural activation within sub-millisecond range across functionally relevant cortical regions facilitating binding and functional integration (R oelfsema et al., 1997; Chawla et al., 2001). However the mechanisms leading to such zero-lag synchrony across spatially distant areas are yet unclear. The field potential of neuronal ensembles as measured by invasive macro-electrodes or by non-invasive EEG / MEG, exhibits a broad range of oscillatory activity, usually described in terms of frequency bands, delta (1-3 Hz), theta (4-9 Hz), alpha (8-12 Hz), beta (13-25 Hz) and gamma (25-90 Hz). The oscillator y neural activity both in huma ns and animals exhibit task 14

PAGE 15

related modulation in their strength and synchrony. Theta and alpha range oscillations have been long shown to reflect memory processes (K ahana et al., 1999; Klimesch, 1999; Weiss and Rappelsberger, 2000; Raghavachari et al., 2001; Jensen and Tesche, 2002; Sederberg et al., 2003; Summerfield and Mangels, 2005; Vertes, 2005) Alpha oscillations have been associated with alertness, arousal and attentional demands (Ray and Cole, 1985; LaBerge, 1997; Foxe et al., 1998; Bastiaansen and Brunia, 2001; Babiloni et al., 2006; Keil et al., 2006; Kelly et al., 2006; Thut et al., 2006; Fan et al., 2007; Klimesch et al., 2007). The beta rhythm has been shown to reduce in strength preceding movement and s ubsequently increase in strength after the movement is complete (Pfurtscheller and da Silva, 1999). The gamma range oscillations are often found in early sensory area s and have been suggested to be involved in the binding of sensory information (Miltner et al., 1999; Mull er et al., 2000; von Stein and Sarnthein, 2000; Fell et al., 2003; Herrmann et al., 2004; Palva et al ., 2005; Womelsdorf et al., 2006; Hermer-Vazquez et al., 2007; Landau et al., 2007; Doesburg et al., 2008). Increasing number of studies s uggest that neural oscillati ons play significant roles in cognitive functions such as at tention (Worden et al., 2000; Frie s et al., 2001; Womelsdorf and Fries, 2007; Doesburg et al., 2008), anticipati on (Foxe et al., 1998; Wo rden et al., 2000; Bastiaansen and Brunia, 2001; Lia ng et al., 2002), asso ciative learning (Mi ltner et al., 1999), memory (Burgess and Gruzelier, 1997; Gevins et al., 1997; Klimesch, 1999; Jensen and Tesche, 2002; Herrmann et al., 2004; Jones and Wilson, 2005) motor functions (Roelfsema et al., 1997; Farmer, 1998; Brovelli et al., 2004; Hermer-Vazqu ez et al., 2007) and perception (Eckhorn et al., 1988; Rodriguez et al., 1999; Babiloni et al., 200 6; Melloni et al., 2007). In particular, the neuronal oscillatory activ ity is considered the substrate for communication between functionally relevant cortical areas (Roelfsema et al., 1997; Miltner et al., 1999; Fries, 2005). 15

PAGE 16

Although neural oscillations have been st udied extensively, emphasis has traditionally been on stimulus-evoked responses and stimulus-rel ated changes in oscillatory activity while the ubiquitous spontaneous oscillati ons were regarded as noise and irrelevant for subsequent stimulus related processing. A landmark study by Ar ieli et al., (1996) usi ng optical recordings showed for the first time that the state of the brain at the onset of an external stimulus predicts the effectiveness of the ensuing stimulus proces sing, providing the early impetus for a systematic investigation of the role of spontaneous brain activity in stimulus processing. Recent work has begun to explore this issue. Com putational modeling as well as in -vitro and in-vivo single unit recordings in animal preparations have e xplored mechanisms by which background synaptic activity can influence the responsiveness of cortical neurons to afferent input (Ho and Destexhe, 2000; Chance et al., 2002; McCormick et al., 20 03; Wolfart et al., 2005; Haider et al., 2007). How these mechanisms manifest in hu mans is not yet clearly established. This dissertation explores the role of pres timulus ongoing oscillations as well as their relationship to subsequent stim ulus processing and higher-order cognitive processes along four specific aims. Aim 1: To identify factors contributing to near ze ro time lags in synchronous oscillatory cortical networks. Given that the conduction de lay between two brain areas is only a small fraction of the oscillation cycle, a near-zero phase-l ag relation could stem e ither from reciprocal communication between the two area s (bidirectional interaction) or from the two areas being readied to communicate by a third set of areas (common input). Th e testing of these possibilities has not been carried out empiri cally. The main reason is that the commonly used methods such as cross correlation and coherence lack the abilit y to decompose neural in teractions into their constituent components. We expound the mechan isms leading to zero-lag synchrony across 16

PAGE 17

widely separated cortical areas by introducing Ge wekes time series decomposition theorem into the analysis of multivariate neural data. Employing Gewekes decomposition theorem, we decompose the neural synchrony between two ar eas (A and B) into its constituent terms (A B), (B A) describing the direction of neural interaction and a term representing the instantaneous correlation (instantaneous common input). We then characterized the infl uence of each of the constituent interaction terms on the phase-lag to a ddress the factors contribu ting to near-zero lags across widely separated cortical areas using simulation examples and local field potential data recorded from behaving monkeys performing a visuomotor pattern discrimination task. Aim 2: To investigate the functional role of prestimulus theta band oscillation in memoryrelated sensory encoding. It is well known that when instructed to memorize a set of items, we typically are capable of recollecting only a fracti on of the items previously presented. Toward addressing the question of why so me words are remembered better while others are not, it was shown that the level of physiological responses evoked by the visual word presentation in the prefrontal cortex (PFC) and th e medial temporal lobe (MTL) is predictive of successful subsequent recall and can thus can be considered an indicator of encoding efficacy. However, the question that still remains to be addressed is that, why are some words encoded better while others are not? We hypothesize that the efficacy of memory encoding in humans depends on the level of excitability over the frontal-medial temporal memory network and that the strength of frontal-medial temporal interaction immediatel y prior to stimulus presentation, possibly reflecting the strength of top-down attentional control originating in the frontal cortex (Furey et al., 2000; Kimura, 2000; Hasselmo and McGaughy, 2004), is positively correlated with the efficacy of stimulus encoding and the likelihood of its subsequent recal l. Considering that noninvasive electroencephalogram (EEG) lacks the spatial resoluti on and the ability to sample 17

PAGE 18

areas proximal to the medial temporal lobe, the hypotheses were instead tested by recording electrocorticogram (ECoG) from multiple implante d girds in five patients undergoing presurgical evaluation for intractable epilep sy who performed a subsequent memory task. In this study we characterize how ongoing oscillatory brain activities in the fronto-temporal network immediately preceding stimulus presentation affect me mory encoding and thus later recall. Aim 3: To investigate the func tional role of ongoing oscilla tions during anticipatory attention and their relati on to evoked responses in a spatial vi sual attention task. While numerous EEG/MEG studies have considered the role of pres timulus oscillations in the genesis of sensory evoked potentials, no consensus has emerged, and disparate reports remain unreconciled (Brandt and Jansen, 1991; Foxe et al., 1998; Marrufo et al., 2001; Makeig et al., 2002; Ergenoglu et al., 2004; Linkenkaer-Hansen et al., 2004; Sauseng et al., 2005; Babiloni et al., 2006; Mazaheri and Jensen, 2006; Thut et al., 2006; Hanslmayr et al., 2007a; Hanslmayr et al., 2007b; Paul et al., 2007; Romei et al., 2008b; Romei et al., 2008a; Van Dijk et al., 2008; Zhang and Ding, in press). Further, sensory information processing depends to a great extent on to what and where an individual deploys attention. It is well known from early studies that responses to stimuli at attended locations and attended sensory modalities are enhanced relative to unattended stimuli. It has also been shown in recent years that attent ion modulates baseline ongoing brain activity in advance of sensory input (Kastner et al., 1999; Dehaene and Changeux, 2005; Bestmann et al., 2007). However, the mechanisms linking the pre a nd poststimulus attentional effects remain not well understood. Our approach to investigating the relation between prestimulus activity and stimulus evoked response starts by considering a theoretica l model in which synaptic input to a sensory neuronal ensemble comes in two forms: external stimulus (exogenous) or other brain structures 18

PAGE 19

(endogenous). Supposing that, in the absence of se nsory input, the firing rate of the neuronal ensemble is related to the magnitude of endoge nous synaptic fluctuations through a sigmoidal function, and sensory input has an additive synap tic effect, it follows that the stimulus-evoked response corresponds to the derivativ e of the sigmoidal function, referr ed to as the gain, which is an inverted-U function of the prestimulus level of background activity. Reformulated in terms of EEG variables where field oscillations and the early component of event-related potential are treated as indices of backgr ound synaptic activity and stimulus-evoked response, the model makes a number of predictions, which were th en tested by recording EEG data from human volunteers performing a trial-by-trial cued covert spatial visual attention task. The systematic model based characterization of the pattern of relationship between the early visual-evoked responses and the prestimulus oscilla tory activity also offers an a dded benefit in that it may yield insights into the fundamental question of how cognitive operations such as attention modulate baseline ongoing neural activity to achieve enhanced stimulus information processing. Based on the proposed model linking prestimu lus oscillatory activity and subsequent stimulus-evoked response, a mechanism on how changes in the global state of the brain, such as selective attention, influenced the aforem entioned sigmoidal curve was proposed. Through experimental data and comput ational modeling we present a mechanism of how attention modulates baseline ongoing activity over ea rly sensory areas to enable enhanced sensory-perceptual processing. Aim 4: To investigate the functional role of ongoing oscillati ons in top-down control of human visual cortex during anticipatory spatia l visual attention. The theoretical framework adopted thus far emphasizes the role of ongoing EEG oscillations over early sensory areas, in this case the visual cortex, and addresses the mechanisms linking the attention induced 19

PAGE 20

20 prestimulus modulations to stimulus-evoked attentional enhancements. However, such goal-directed prestimulus modulation of sensory co rtical activity is not an inherent property of the early sensory areas. Instead, it is believed that the hi gher-order areas such as the frontal-parietal network facilitate s an enhancement of the excitabi lity of the task-relevant and feature-specific visual areas a nd perhaps a suppression of activity in task-irrelevant visual areas in advance of visual stimulation by modulating its baseline activity (Chela zzi et al., 1993; Luck et al., 1997; Hillyard et al., 1998; Kastner et al ., 1999; Hopfinger et al., 2000; Kastner and Ungerleider, 2001; Corbetta and Shulman, 2002; Gazzaley et al., 2005; Gi esbrecht et al., 2006; Grent-'T-Jong and Woldorff, 2007; Klimesch et al., 2007). In line with this notion, it has been identified that deployment of covert spatial vi sual attention is associated with increased activation in a network i nvolving dorsal frontal and parietal cortices with concomitant increases in activation over retinotopic earl y visual areas (Desimone, 1996; Gitelman et al., 1999; Kastner et al., 1999; Hopfinger et al., 2000; Corbetta and Shulman, 2002; Giesbrecht et al., 2003). While prior neuroimaging and le sion studies of the top-down c ontrol of visu al attention have offered insights into the areas in the brai n that exert control over stimulus processing, the neural mechanisms of how this increased excita bility over the visual sensory areas is brought about has not been clearly estab lished. What is the neural sign al that mediates the top-down biasing influence? Is the incr eased excitability achieved by increasing the excitatory drive on local pyramidal neurons (excitat ion) or by decreasing excitato ry drive on local inhibitory interneurons (disinhibition)? Is the top-down influence transient or sustained? These questions are addressed by applying time-frequency analysis of power, coherence a nd Granger causality to scalp EEG recordings from healthy volunteers performing a spatial visual attention task.

PAGE 21

CHAPTER 2 GENERAL METHODS The primary goals of this dissertation invol ve the investigation of prestimulus ongoing activity and its relation to stimulus evoked resp onses and hence cognitive performance. Neural ensemble activity exhibit a wide range of oscill ations and the role of these oscillations in cognitive information processing have been increas ingly studied in recent years. Physiologically, cortical field oscillations reco rded by scalp EEG/MEG reflect the rhythmic fluctuations in the correlated background synaptic activity and provide an index of the excitability of the underlying neural population (Dasilva, 1991). EEG/MEG si gnals consist of several simultaneous oscillations, which have traditionally been s ubdivided into frequency bands such as delta (1-3 Hz), theta (4-9 Hz), alpha (8-12 Hz), beta (about 13-25 Hz ) and gamma (25-90 Hz) each of which are distinctly related to cognitive information processing and hence behavior. The oscillatory nature of the signal necessitates the need for a spectr al domain approach to study this problem. Further, it is well ag reed upon that complex cognitive processes involve complex interaction between disparate regions of the brain. Synchr ony (functional connectivity) and Granger causality (effective conne ctivity) analysis have emerged as principled approach of identifying and quantifying inter-a real interaction and infer caus al influence among distributed networks (Eckhorn et al., 1988; Schack et al., 2000; von Stein and Sarnthein, 2000; Baccala and Sameshima, 2001; Kaminski et al., 2001; Varela et al., 2001; Liang et al., 2002; Brovelli et al., 2004; Tallon-Baudry et al., 2004; Pa lva et al., 2005; Seth, 2005; Ding et al., 2006; Wang et al., 2007b; Bollimunta et al., 2008; Cadotte et al., 2008; Hermer-Vazquez, 2008; Rajagovindan and Ding, 2008b; Zhang et al., 2008a; Zhang et al., 2008b; Cadotte et al., 2009). Granger causality is a statistical approach that d ecomposes neural synchrony into causal directional interactions, which provides a non-invasive altern ative to identify the direction of neural interaction. Below, 21

PAGE 22

some of the spectral estimation techniques employe d in this study to quan tify neural activity and also the mathematical framework of Granger cau sality with special emphasis on its spectral decomposition is reviewed. Practical issues concerning how to estimate such measures from time series data are then discussed. Th e description in the following sect ions follow that of (Granger, 1969; Geweke, 1982; Ding et al., 2000; Chen et al., 2006; Ding et al., 2006; Wang et al., 2007b; Rajagovindan and Ding, 2008b) 2.1 Parametric Multivariate Spectral Estimation Multivariate autoregressive (MVAR) modeling is a parametric spectral analysis method in which time series models are estimated from data sets having either one long realization or a number of realizations. In the la tter case, the fundamental assumption of this algorithm is that the short-window time series can be treated as realiz ations of an underlying stationary stochastic process. As cognitive information processing in volves transient change s in neural activity, MVAR can be used to investigate time series in short window size (<100 ms). The following is the procedure of MVAR (Ding et al., 2000). Let ()[(1,),(2,),,(,)]T X tXtXtXmt be a m -dimensional jointly stationary random process. T denotes matrix transposition. In the case of neural recordings, m refers to the total number of recording channels to be analyzed. Assuming () X tis a zero mean stationary process, () X t can be modeled by the following pth order autoregressive equations: ()(1)(1)(2)(2)()()() X tAXtAXtApXtpEt (2-1) where p is the model order, and are the unknown () Aimm coefficient matrices and is the uncorrelated noise term with covariance matrix () Et Multiplying ()T X tk to Eq. 2-1 and taking expectation on both sides we arri ve at the Yule-Walker equations. ()(1)(1)()()0 RkARkApRkp (2-2) 22

PAGE 23

where () R j is covariance matrices of ()()T X tXtj with lag j. Also note that since ())0 Etk(TXt () E tis an uncorrelated noise process. The unbiased estimator of the covariance matrix in Eq. 2-2 fo r a single realization of the X process is given by 11 ()()()Nn T i R nxix Nn in In the case of multiple realizations of the process as is usually the case, the covariance matrix is computed for each individual reali zation and then averaged across all the realizations to obtain the most robust estimate. In the case of short window of data, i.e. short sample length where N is small, it is apparent that the estimation from a single realization is poor. However, with increasing number of realizations this problem can be offset for short sample length time series. In the limiting case with number of realizations approaching infi nity, the sample length of each realization may be as s hort as the model order plus one ( N = p+1). Coefficient matrices and covariance matrix () Ai of noise term is obtained by solving Eq. 2-2 through the Levinson, Wiggins and Robinson (LWR) algorithm. The noise covariance matrix is obtained as part of the LW R algorithm. One may also obtain as () Et 1(0)()()p i R AiRi It is instructive to note that the Eq. 2-2 contain a total of pm2 unknown model coefficients to be estimated from the same number of simultaneous linear equations. Although the primary objective is to fit the best model to the da ta, which is to minimize the residual noise variance, this constraint alone ma y result in over-parameterization. The reason for this is because the variance of the residual te rm decreases monotonically with incorporating more and more past values of the process. Ov er-parameterization leads to inaccuracies in the estimation of the model coefficient besides the obvious increase in the computational 23

PAGE 24

complexity. Thus a penalization scheme to avoid over-parameterization is employed. Criterions that incorporate both minimization of the varian ce of the residual term and penalize excessive coefficients are the Akiake Information Crite rion (AIC) and Bayesian Information Criterion (BIC) among several other similar cr iterions. The AIC is defined as 22 ()2logdet()totalmp AICp N where Ntotal is the total number of data points from all the trials. The first term on the right hand side accounts for the variance minimization and the second term correspond to the cost associated with increasing para meters. Plotted as a function of p, the optimum model order corresponds to the minimum of this func tion. For typical neurobiological data Ntotal is very large and hence for practical ranges of p, the AIC function does not achieve a minimum due to inadequate penalization as the second term vanishes with large values of Ntotal. An alternative criterion is the Bayesian Information Criterion (BIC), which is defined as 22log ()2logdet()total totalmpN BICp N This criterion can compensate for the large num ber of data points and may perform better in neuroscientific applications. A final step and cr itical step necessary for determining whether the autoregressive time series model is suited for a given data set, is to check whether the residual term is white. Here the residual term is obtained by computing the difference between the models predicted values and the actually measured values. Once an autoregressive model is adequately estimated, it becomes the basis for both time domain and spectral domain causality analysis Spectral features ar e derived from MVAR models after acquiring () A i and estimates. Taking the Fourier transform, Eq. 2-1 can be rewritten in the spectral domain as 24

PAGE 25

()()() AfXfEf, Defining then 1 2 0()()m ijf iHfAie ()()() X fHfEf where is the transfer function. The spectral matrix can be readily derived from () Hf **1 ()lim[()()]()()NSfEXfXfHfHf N where means both transpose and complex conjugate. Spectral power is contained in the diagonal terms of the spectral matrix and the offdiagonal terms represent the cross spectra. Coherence spectra between two random process (,) X it and (,) X jt is defined as: 2() () ()()ij ij iijjSf Cf SfSf If the coherence value is equal to 1 or 0, the two processes ar e maximally interdependent or independent, respectively. Once the transfer func tion, the noise covariance and the spectral matrix are estimated, Granger causality may be derived according to the procedures outlined in the following sections. 2.2 Granger Causality Analysis 2.2.1 Time Domain Formulation Consider two jointly stationary stochastic processes t X and each having an autoregressive representation given by tY111 1 111 1, var() var()tjtjtt j tjtjtt jXaX YdY x y (2-3) 25

PAGE 26

Jointly, they can be represented by the fo llowing bivariate au toregressive model 22 11 22 11 tjtjjtj jj tjtjjt j jjXaXbY YcXdY2 2 t t (2-4) where,2t and 2t are uncorrelated over time, and their contemporaneous covariance matrix is x xxy y xyy If is not uniformly zero for jb1,2,..., j then is said to have a causal influence on tYt X Likewise, t X is said to have a causal influence on if is not uniformly zero for The total interdependence between tYjc1,2, ..., j t X and is defined as (Gelfand and Yaglom, 1959) tY ,ln x y XYF where denotes the determinant of the enclosed matr ix. It can be easily verified that according to this definition when the two time series are independent, and when they are not. Considering Eq. 2-3, 2-4, ,0XYF ,0XYF x represents the prediction error of t X based on its past values and x xrepresents the prediction error, wh en predicting the present value of t X based on past values of t X and. If tY x xis less than x then is said to have causal influence on tYt X quantified by (Granger, 1969; Geweke, 1982) ln x YX x xF 26

PAGE 27

It may be noticed that when there is no causal influence from Y to X and otherwise. Similarly the causal drive from X to Y may be defined as (Granger, 1969; Geweke, 1982) 0YXF0YXF lny XY yyF If indicating that the noise terms 0xyyx2t and 2t are correlated at time t then the interdependence between t X and has another contributor that is not explained by the interaction between tYt X and. This component possibly repres enting common input from a third system, will be referred to as the instan taneous causality, defined by (Geweke, 1982) tY ln x xyy XYF Notice that when the instantaneous causality is zero and greater than zero otherwise. 0xyyxIt follows from definition that the total interd ependence is the sum of mutual interaction between X and Y and a component exogenous to the system given by (Geweke, 1982) XYXYYXXYFFFF 2.2.2 Frequency Domain Formulation To derive Gewekes formulation of Granger causality in the frequency domain it is convenient to repose the autoregressive equations in terms of lag operator. The lag operator L is defined as Rewriting Eq. 2-1in terms of lag operator 1ttLXX2 22 2 22()() ()()t ttX aLbL Y cLdLt (2-5) where, Fourier transforming both sides of Eq. 2-5 2222(0)1,(0)0,(0)0,(0)1 abcd 27

PAGE 28

22 22() ()() ()() () () ()() ()()x yE ab XX A E cd YY where, 22222222 111()1,()1,()1,()1ij ij ij ij jjj jjjaaebbeccedd1j je Further simplifying ()()() () ()()() ()xx xy x yx yy yHHE X HHE Y where, the transfer function 1()() HA and 22 2211 ()(),()() 11 ()(),()()xx xy yx yyHdHb AA HcH AA a Performing proper ensemble averaging, th e spectral matrix can be derived as xx xy yx yySS SS SH H, where denotes complex conjugate and matrix transpose. The diagonal terms of the spectra l matrix denote the auto spec tra (power spectrum) and the off-diagonal terms denote the cross spectra. If X and Y are independent, the off-diagonal elements are zero. The spectral domain represen tation of the total interdependence between t X and at frequency (=2 f) is given by (Geweke, 1982) tY ,()() ()ln ln(1()) ()xxyy XYSS fC S 28

PAGE 29

where 2() () ()()xy xxyyS C SS is the coherence function. The phase-lag between t X and at a given frequency is given bytY 1Im() tan Re()yx yxS S Further, ()()()()()xxyy xyyxSSSS S and *()()xy yxSS Notice that when X and Y are independent, () S reduces to the product of the auto spectra, hence it can be verified that the total inter dependence is zero. E xpanding the entries of the spectral matrix, **()()()2Re(()())()xx xxxxxx xyxxxy xyyyxySHHHHHH* (2-6) consider the case when that is when there is no instantaneous causality. Then, 0xy ()xxS reduces to the sum of the first and the third term in the right hand side of Eq. 2-6. Intuitively, this can be understood as decomposing the auto spectr um into an intrinsic part and a component representing the causal drive from the other time series. However if 0xy this form of interpretation may not be obvious. Consider a transformation introduced by Geweke (Geweke, 1982) referred to as normalization, that removes the effects of the cross terms and enables the identification of a similar intrinsic power te rms and a causal power term possible. Left multiplying on both sides of Eq. 2-5 by the transformation matrix P given by (Geweke, 1982) 10 1xy xxP we obtain 22 33() ()() ()() () ()() ()() ()x yE ab XX A cd YY E 29

PAGE 30

322322()()(),()()(),()()()xy xy xy y yx xx xx xxccaddbEEE The new transfer function ()H is the inverse of the ()A 32 32()() ()() 1 () ()() ()()xx xy yx yydb HH ca A HH H Also, A= A thus we have ()()(),()() ()()(),()()xy xx xy xx xy xy xx xy yx xy yx xx xy xxHHHHH HHHHH Note that x E and yE are uncorrelated. The variance of the residual term for the normalized equation for is tY 2 x y yy yy x x Thus after normalization the auto spectrum of t X can be written as *()()()()xx xx yy xx xx xyxySHHHH As earlier, the auto spectra can now be interpre ted as the sum of intrinsic power and causal power component of t X due to Based on this formulation we now define the frequency domain causal influence from to tYtYt X at frequency as (Geweke, 1982; Ding et al., 2006) *() ()ln ()()xx XY xx xx xxS f HH (2-7) Notice that this definition of causal influence is expressed in terms of the intrinsic power rather than the causal power. It is expressed in this way so that the causal influence is zero when the causal power is zero (i.e., the intrinsic power equals the total power), and 30

PAGE 31

increases as the causal power increases (i.e., the intrinsic power decreases). Now by taking the transformation matrix as 1 01xy yy and performing the same operations as earlier, we get the influence of t X toY given by t *() ()ln ()()yy YX yy yy yyS f HH (2-8) where, ()()()xy yy yy yx yyHHH Having identified the total interdependence and both the directional influences, the spectral decomposition of the instantaneous causal ity can now be readily determined as **(()())(() ()ln ()xx xx yyyy xx yy XYHHHH f S (2-9) The above set of variables are relate d through (Geweke, 1982; Ding et al., 2006) ,()()()()XYXYYXXYffff Under general conditions, the spectral measures derived above are relate d to the time domain measures as ,,1 2XY XYFf d 1 () 2XY XYFf d 1 () 2YX YXFf d 31

PAGE 32

1 () 2XY XYFf d The invertible nature of Gewekes formula tion between the time domain and frequency domain representation is a desirabl e characteristic that is not po ssessed by other current causality measures. For practical purposes, th e same quantities listed above (Eq. 2-7, 2-8, 2-9) with simple algebraic modification may be expr essed as (Brovelli et al., 2004) 2 2() ()ln(1 ) ()yx xx yx yy XY xxH f S 2 2() ()ln(1 ) ()xy yy xy xx YX yyH f S 22 **()()()()()()()() ()ln( ) ()xy yx xx xyyy xy yy yxxx yx xx yy XYSHHSHH f S 2.2.3 Mathematical Formulation of Block Granger Causality Application of Granger causality to the inves tigation of causal influences between different brain structures have so far mainly utilized pairwise analysis (Brovelli et al., 2004) in which two time series are analyzed at a time. However, pairwise analysis may be considered inadequate given large sets of multivariate neural time series data. A more pragmatic approach for many purposes is to combine the time series recorded from different brain regions of interest into blocks, and then to analyze th e relations between the blocks. Although it is possible to achieve blockwise analysis by combining th e results of repeated pairwise analyses, a more effective, and theoretically more elegant, approach is to addr ess the relation between tw o blocks of time series 32

PAGE 33

directly (Wang et al., 2007b). This blockwis e approach is contai ned in the general methodological framework developed by Geweke (1982). Let the block of signals from tw o recording sites be denoted byt X and with dimensions k and l. In the case of pairwise Granger causality, tY t X andtY t are two one-dimension time series (k=1, l=1); while in the case of block Granger causality, X andtY are two sets of time series of dimensions k and l typically not equal to 1 and ha ve no overlap. Jointly, they can be represented by the following multivariate autoregressive model -11 -11 pp tt jt j t jj jj pp tt jt j t jj jjXaXbY YcXdY (2-10) where t and t are uncorrelated over time, and their contemporaneous covariance matrix is x xxy y xyy If is not uniformly zero for then jb 1,2,..., j ptY is said to have a causal influence on t X Likewise, t X is said to have a causal influence on tY if is not uniformly zero for jc 1,2, ..., jpThe model order p is determined using either the Ak aike Information Criterion (AIC) or the Bayesian Information Criterion (BIC) which reflects a tradeoff between sufficient spectral resolution and over-parameterization. Fourie r transforming Eq. 2-10 and performing proper ensemble average, we obtain the spectral matrix xx xy yx yySS SH SSH 33

PAGE 34

where denotes complex conjugate and matrix transpose, and H is the transfer function matrix. Similar to the procedure described in the prev ious section, the Gra nger causality spectrum of the two blocks of time series, for t X driving tY is given by, 2 2-( () -ln(1) ()yx xx yx yy XY xxH f S ) which can be interpreted as the proportion of t X s causal contribution to the power of the tY series at frequency The logarithm is taken to preserve certain favorable statistical properties. Similarly, the causality spectrum for tY driving t X is given by 2 2-( () -ln(1) ()xy yy xy xx YX yyH f S )time series, one repeats the process to address the question of causal in fluence in the opposite To obtain the magnitude of causali ty for a particular frequency ba nd, the above spectrum may be averaged over the desired frequency band. 2.2.4 Interpretation Statistically, for two simultaneously measured ti me series, one series can be called causal to the other if we can better pr edict the second series by incorporating past knowledge of the first one (Wiener, 1956). This concept was formalized by (Granger, 1969) in the context of linear regression models of stochastic pr ocesses. Specifically, if the variance of the prediction error for the second time series at the present time is reduced by including past measurements from the first time series in the linear regression model, then the first time series can be said to have a causal (directional or driving) in fluence on the second time series. Reversing the roles of the two 34

PAGE 35

direction. From this definition it is clear that th e flow of time plays an essential role in allowing inferences to be made about directional causal in fluences from time series data. Alternatively, improvement in prediction can also be viewed from the perspective of comparing relative estimates of conditional probability. Recent work has modeled the relation between multip point processes along this view (O katan et al., 2005; Truccolo et al., 2005). In our analysis, th EEG data constitute the time series, and Granger cau sal influence is equated with the direction o synaptic transmission between neuronal ensembles. 2.2.5 Assessment of Statistical Significance le e f ctra we have adopted a random permutation appro ted 2 from a mporal is per Spectral Estimation Multitaper spectral estimation (Thomson, 1982) re fers to a set of non parametric methods for estimating power spectra, coherences and re lated spectral quantitie s using an orthogonal set For coherence and Granger causality spe ach (Brovelli et al., 2004) to build a baseline null-hypot hesis distribution from which statistical significance can be derived. Consider two channels of recordings with many repea trials We can reasonably assume that the data from different trials are approximately independent of one another. Randomly pairing data for channel 1 with data for channel different trial leads to the creation of a synthetic ensemble of trials for which there is no interdependence between the two channels based on construction, while preserving the te structure within a channel. Performing such ra ndom pairing with many different permutations will result in a distribution of coherence or cau sality spectra corresponding to the null hypothes of no statistical interdependence. The calculated va lue for a given statistic from the actual data is compared with this baseline null hypothesis dist ribution for the assessment of significance levels. Bootstrap resampling techniques (Efron and G ong, 1983; Efron, 1985, 2000) were used to estimate confidence intervals. 2.3 Multita 35

PAGE 36

of data tapers, in specific, the n sequences and their appro lable es timation techniques. The problem of optimum spectr as ther ss th, ach s ing t to be discrete prol ate spheroidal a.k.a Slepia ximate minimum bias sine tapers. Before proceeding to the mathematical formulation of the Multitaper spectra estimation techniques it is worth noting the several favorab le properties that motivated the use of this approach from among several other avai al estimation is confronted with several key issues including that of minimizing the (a) bi of the estimators (b) variance of the estimators and (c ) the spectral leakag e among several o considerations. For further details on each of these considerations the readers may refer to (Lfeachor and Jarvis; Thomson, 1982; Mitra and Pesaran, 1999; Mitra and Bokil, 2007). For example consider one of the conventional nonparametric spectral analysis techniques, the Welchs method. In order to reduce the variance in the estimation, the approach involves splitting the data into overlapping segments follo wed by the estimation of the power or the cro spectrum for each segment and then averaging over all the segments. For data of short leng such approaches suffer from severe bias in th e estimation. In comparison, multitaper appro does not suffer from these limitations. Both reduc tion in bias and variance of the estimation is achieved since averaging the spectral estimates over different tapers enable reduction in the variance and since each taper is applied to the entire data instead of short segments of the data a in Welchs method the resulting bias in the es timation is smaller in comparison. Further the choice of the orthonormal Slepia n tapers enables maximum side lobe suppression hence offer the least spectral leakage in comparison to all other spectral estimation techniques. Thus multitaper approach offers the optimum spectral estimation in the face of short sample length data or short sample length and limited realizations of data as often encountered in neuroscientific applications as is the case in this study. It is appare nt from the aims set ou 36

PAGE 37

addressed that it is crucial to be able to achieve single-trial estimation of the spectra this crucial requirement, this study utilizes the full advantage of this technique to enable single-trial estimation of spectral features duri ng short prestimulus time periods which otherwise would not have been possible to estimate with sufficient accuracy using either the parame framework based on autoregressive models presented earlier or through other common non parametric approaches. A brief exposition of the mathematical formul ation of the multitaper spectral estimatio technique for univariate c In light of tric n ase is as follows. The ex tension to multivariate case is straightforward. Consi will be unstable and may n value problem der time series x(t) (t = 1,2N) a zero mean second order stationary random process and let w()kt denote the kth Slepian taper (window se quence) and k = 1,2K. Step 1: Let us define the time-bandwidth product given by C = NW where N is the sample length and W the desired spectral bandwidth. If C is too small, the estimate ot have sufficient dynamic range and if C is too large, the estimate may not have adequate frequency resolution. For a given choice of C there are K = 2C data taper sequences. Since the energy concentration of higher orde r tapers is poorer than the low order tapers, in practice it is common to choose K = 2C-1 or K = 2C-2 Step 2: The Slepian tapers having the maximum energy concentration in the bandwidth W is determined by solving the following eigen 1 0sin[2()] () () ()N jWtj wt wj tj for t = 1, 2N. kk kStep 3: The eigen coefficients of this problem re duces to the Fourier transform of the data sequence multiplied by the tapers (similar to periodogram). The windowed Fourier transform of the data x(t) is given by 37

PAGE 38

38 -1 0 () ()()N j t t kk k X xtwte Step 4: The spectrum then estimated as the average of the spectra obtained through each taper is given by 1 0 kK 1 ()()()K kkSXX alizations, the above esti mate is further averaged across all realizations. For further treatment of this topic refer (Thomson, 1982; Thomson, 2007). In the case of multiple re

PAGE 39

CHAPTER 3 DECOMPOSING NEURAL SYNCHRONY: TOWARD AN EXPLANATION FOR NEAR-ZERO PHASE-LAG IN CORTICAL OSCILLATORY NETWORKS 3.1 Introduction Cortical information processing involves the coordinated activity among many distinct regions of the brain. Statistic ally, this coordinated activity manifests as correlated or synchronized co-variations in the recorded multivariate data. Early studies in animals have shown that stimulus-evoked short-range synchro ny between neurons in th e primary visual area subserves perceptual binding of sensory info rmation (Singer and Gray, 1995; Tallon-Baudry et al., 2004). Simultaneous action potentials fired by lower order neurons (Malsburg, 1995; Roskies, 1999) provide an effective drive on higher order neur ons, the activations of which enable pattern discrimination and conscious awareness. In huma ns, similar observations have been made (Slotnick et al., 2002) where highly synchronized EEG activity occurs in response to stimulus input. In all these cases, a near-zero pha se-lag relation between di fferent data series is observed, reflecting the millisecond or even sub-millisecond precision required for feature integration (Singer and Gray, 1995; Munk et al., 1996; Roelfsema et al., 1997; Witham et al., 2007). Increasingly, long-range synchr onization of oscillatory field activity with near-zero phaselag, often in the absence of s timulus input, has been reported (B ressler et al., 1993; Roelfsema et al., 1997; von Stein et al., 2000; von Stein and Sarnthein, 2000; Brovelli et al., 2004). Depending on the task, the strength of the synchronization can influence the efficiency of both sensory and motor processing (von Stein et al., 2000; Engel et al., 2001; Liang et al., 2002; Womelsdorf et al., 2006), suggesting that it has a functional role. To date, however, an explanation of the observed near-zero phase-lag rela tion in these large-scale networ ks has not been forthcoming. Recent work have considered this issue from the perspective of neuronal communication. 39

PAGE 40

Physiologically, field oscillations provide an in dex of the excitability level of a neuronal ensemble (Lakatos et al., 2007; Lakatos et al., 2008). During the excitable phase of the oscillation cycle, presynaptic neurons are more li kely to fire action potentials, whereas for the postsynaptic neuron, action potentials received during the excitable phase are more effective at integrating the synaptic input and triggering a response (Volgushev et al., 1998; Fries, 2005; Lakatos et al., 2007). This suggests that l ong-range synchrony coul d serve as a gating mechanism of information flow in cortical circ uits (Salinas and Sejnowski, 2001; Averbeck and Lee, 2004; Fries, 2005; Siapas et al., 2005). Give n that the conduction de lay between two brain areas is only a small fraction of the oscillation cycle, a near-zero phase-lag relation could stem either from reciprocal communication between the two areas (bidirectional interaction) or from the two areas being readied to communicate by a third set of areas (common input). Mathematically, it is intuitively clear that a positively correlated common input, if strong enough, can drive the two areas into near-zero phase-lag synchrony. Alternatively, a recent computational model based on the anatomical conn ectivity pattern in the visual system proposes that the emergence of near-zero phase-lag synchr ony can also result from two or more cortical areas experiencing bidirectiona l interaction (Chawla et al., 2001). The testing of these possibilities has not been carried out. The main reason is that the commonly used methods such as cross correlation and coherence lack the ability to decompose neural interactions into their constituent components. In this paper we attempt to address this problem by introducing Gewekes time series decomposition theorem into the analysis of multivariate neural data. Let the two brain areas be denoted by A and B. The interaction between thes e two areas may be medi ated by A influencing B (AB), B influencing A (BA), and/or A and B both receiv ing a common input. Gewekes 40

PAGE 41

theorem states that the total in terdependence (synchrony) between two stochastic processes from A and B can be expressed as the sum of the three components: (AB)+(BA)+(common input). Here the arrow is underst ood in the sense of Granger causa lity and the common input is represented by the instantaneous causality (Geweke, 1982; Chen et al., 2006; Ding et al., 2006). In this framework it is hypothesized that bi directional interaction or common input or a combination of both contributes to the estab lishment of near-zero phase-lag synchrony. In particular, when the interaction is clearly unidirectional (e.g. AB equals zero but BA does not), the phenomenon of near-zero phase-lag can only occur as a result of strong positively correlated common input, arising exogenously to A and B. We tested these ideas by analyzing local field potential data recorded from be having monkeys performing a visuomotor pattern discrimination task. 3.2 Materials and Methods 3.2.1 Simulation 3.2.1.1 Setup Auto-regressive model of the form in Eq. 3-1 was used to generate all the simulated time series. Two representative types of interaction pattern, namely (a) unidirectional interaction with positively correlated instantaneous common input and (b) bidirectional interaction, were considered. The phase-lag as a function of a ppropriate model parameters for both cases was studied. 3.2.1.2 Positively correlated common input A bivariate AR(3) process [p = 3 in Eq. 3-1] in which X drives Y was used. The coefficients of the model were a1 = 0.4428, a2 = -0.5134, a3 = 0, d1 = 0.506, d2 = -0.6703, d3 = 0, b1 = b2= b3 = 0, c1 = c2 = 0, c3=0.1, and xx=yy = 1. The cross terms in the noise covariance matrix, yx=xy, reflecting the strength of positi vely correlated common input, was 41

PAGE 42

systematically varied. The parameter choice abov e enabled the model to oscillate at 40 Hz for which the phase-lag was computed. The dataset consisted of 100 epochs of 200 sample points each. The sampling rate was assumed to be 200 Hz. 3.2.1.3 Bidirectional interaction A bivariate AR(4) process [p = 4 in Eq. 3-1] was used. The coefficients of the model were a1 = 0.9, a2 = -0.5, a3 = 0, a4 = 0, d1 = 0.8, d2 = -0.5, d3 = d4=0, xx= yy = 1, and xy = yx = 0. The coefficients of the interaction terms b1,2,3,4 and c1,2,3,4 were varied in tandem to achieve simultaneous increase in the strength of both feed -forward and feed-back interaction. The model exhibited narrow frequency band oscillations with a frequency peak at 32 Hz for which the phase-lag was computed. The dataset consisted of 100 epochs of 200 sample points each. The sampling rate was assumed to be 200 Hz. 3.2.2 Experiment 3.2.2.1 Behavioral paradigm Two monkeys (GE and LU) were trained to perform a GO/NO-GO visual pattern discrimination task in the Laboratory of Neur opsychology at the National Institute of Mental Health (Bressler et al., 1993; Ledberg et al ., 2007). Animal care was in accordance with institutional guidelines at the time. The monkey initiated each trial by depressing a hand lever and maintained its depression while anticipating the onset of a visual stimulus. Four squares arranged in either a line (lef t-slanting and right-slanting) or a diamond (left-slanting and rightslanting) shaped formation appear ed on a visual display after a random time interval triggered by the lever depression. The monkey made either a GO (lever release) or a NO-GO (maintaining lever depression) response upon discriminating th e input pattern. For th e GO trials, the time between stimulus onset and the lever release is defined as the response time (RT). The experiment was conducted in sessio ns of approximately 1000 trials. 42

PAGE 43

3.2.2.2 Data acquisition Local field potentials (LFPs) were record ed with bipolar tefl on-coated platinum microelectrodes (51-m diameter and 2.5-mm tip separation) fr om up to 15 distributed sites located in the hemisphere cont ralateral to the dominant hand (right hemisphere in monkey GE and left hemisphere in monkey LU). The data coll ection period started 90 ms before the stimulus onset and ended approximately 500 ms after (B ressler et al., 1993). L FPs were amplified by Grass P511J amplifiers (-6 dB at 1 and 100 Hz, 6 dB/octave falloff) and digitized at 200 Hz. As this study is concerned with the phase between tw o signals, the bipolar derivation carries certain arbitrariness. Reversing the order of the two s ubtracting recording lead s can change the phase from 0 to or vice versa. This can affect the form ulation of the hypothesis to be tested. See Results and Discussion sections for more details. 3.2.2.3 Data set selection Previous analysis of the same experiment (B rovelli et al., 2004) has identified a coherent beta (14 to 30 Hz) oscillatory network in the sensorimotor cortex invo lving both preand postcentral sites during the prestimulus time period. Fo r this work the three recording sites that are common to both monkeys were selected for furt her analysis: primary somatosensory area (S1), primary motor area (M1) area, a nd posterior parietal area 7b. Tria ls contaminated with artifacts or associated with incorrect be havioral responses were rejected. To achieve a sufficient number of trials, different sessions having similar RT dist ributions were combined to yield a data set of approximately 2400 and 1400 trials for monkeys GE and LU, respectively. The prestimulus time interval from -90 ms to 20 ms was considered. 3.2.2.4 Time series decomposition Let the LFP data from two recording sites be denoted byt X and Jointly, they can be represented by the following biva riate autoregressive model tY 43

PAGE 44

11 11 pp tjtjjtj jj pp tjtjjt-j jjXaXbY YcXdYt t (3-1) where t and t are uncorrelated over time, and their contemporaneous covariance matrix is x xxy yxyy If j b is not uniformly zero for 1,2,..., j p then is said to have a causal influence on t tY X Likewise, t X is said to have a causal influence on Y if t j c is not uniformly zero for 1,2, ..., j p 0xyyx indicating that the noise terms t. If and t are correlated at tim then the interdependence betweente t X and tY has another contributor that is not explained interaction between by the t X and This contributor, possibly representing influences exogenous to the (X,Y) system such as a common input from a third system, will be referred to as the instantaneous causality (Geweke, 1982; Ding et al., 2006). It can be shown that the above set of variables are related th rough (see section 2.2.2) tY,()()()()xyxyyxxyffff. (3-2) Intuitively, the decomposition in Eq. 3-2 m eans that the total am ount of statistical synchrony between two LFP signals is the sum of their causal drives on one another and a common input component. The expression in Eq. 3-2 can be integrated over the entire frequency domain to yield the time-domain counterpart. While the frequency-domain formulation is convenient for estimation, the above decompositi on is more readily interpretable in the timedomain. 44

PAGE 45

3.2.2.5 Data analysis protocol For each monkey there are 3 distinct pairs of re cording sites: (M1, S1), (M1, 7b) and (S1, 7b). For each pair, previous work (Brovelli et al., 2004) has identified a prominent coherence peak in the beta frequency range (14 to 30 Hz). Except for (M1, S1) in monkey LU the coherence in other five channel pairs is concentrated in the be ta frequency range. These five channel pairs are further analyzed as the time-domain quantities are e quivalent to that in the beta range. The relative phase, referred to as the phase -lag in this study, is well-defined for the peak frequency. The dependence of this phase-lag on the factors in Eq. 32 was investigated by carrying out the following analysis. (1) The phase-lag in the beta frequency range was estimated for each single trial by Fast F ourier Transform (FFT) and sorted according to its value. The sorted trials were grouped into subensembles w ith 30% overlap, resulting in 33 and 19 groups in monkeys GE and LU, respectively. Each sube nsemble contained 100 trials. (2) For each subensemble the ensemble mean was estimated and removed from the individual trials within the subensemble. This is to ensure that the da ta may be treated as coming from a zero-mean stochastic process. An AR model of order p =9 was fit to the mean-re moved data. (3) For each subensemble, power, coherence, causality spectr a as well as phase-lag at the peak frequency were derived from the AR model (Chen et al., 2006; Ding et al., 2006). (4) Spearman rank correlation (SRCC) and Spearman rank partia l correlation (Macklin, 1982) (SRPCC) were computed between the instantaneous causality m easure and the estimated phase-lag to assess the prediction that the two variables are negatively co rrelated. To remove the effect of directional influences a partial correlation analysis was performed. Significance was determined through one tail t-test with p<.05. (5) For channel pairs determined to have bidirectional interactions, Spearman rank correlation (SRCC) and Spearman rank partial correla tion (Macklin, 1982) 45

PAGE 46

(SRPCC) were computed between the magnitude of the sum of the two di rectional influences and the estimated phase-lag to assess wh ether they exhibit negative correlation. Pairwise Granger analysis has been adopted since phase-lag is a bivariate phenomenon. However, evaluating causal inter action patterns in a bivariate framework may ignore potential influences from other areas that may or may not be sampled. In order to assess more reliable causal interaction patterns, a multivariate approa ch applied to all the relevant simultaneously sampled data may be resorted to and one such approach is the condi tional Granger causality (Chen et al., 2006; Ding et al., 2006). In the br oader context of conditional Granger causality analysis applied to multivariate data, it is only reliable to submit channel pairs that exhibit significant nonzero conditional Granger causality to the data analysis protocol described above. 3.2.3 Logic of Analysis Protocol The goal of this work is to test that (a) positively correlated common input and (b) bidirectional interaction contributes to near-zero phase-lag. In the simulation examples, this is accomplished by changing the strength of input correlation and bidirect ional interaction and observing the corresponding change in the phase-lag. For actual data, while Gewekes theorem allows the extraction of various causal influences through the decomposition of synchrony, the strength of input correlation a nd bidirectional interaction is no t easily manipulated. The sorting of trials according to their phase-lag is the strate gy to deal with this problem. Each subensemble of trials gives rise to a different phase-lag value. Performing Gewekes decomposition for each subensemble provides the avenue to observe the correlation between phase-lag and common input/bidirectional interaction. It is important to note that, for a given pair of recording sites, both instantaneous causality and the two directional influences may change as functions of the sorted phase-lag. A simple pairwise correlation anal ysis may thus become confounded. Partial 46

PAGE 47

correlation is used here to make possible the examination of one fact ors contribution to near-zero phase-lag while the contribution of other factors is statistically removed. 3.3 Results 3.3.1 Simulation The impact of (a) positively correlated comm on input and (b) direct ional interaction on phase-lag is examined using simulated time series and summarized in Table 3-1 and Table 3-2. In the case of unidirectional in teraction, increased common input to the network, measured by instantaneous causality, reduces the magnitude of phase-lag from 2.19 radians to a near-zero value of 0.16 radians (Table 3-1). Similar effect of instantaneous common input was observed in networks with bidirectional interaction as well (not shown). In the case of the system with bidirectional interaction and no instantaneous common input, the parametric increase in the model coefficients b1,2,3,4 and c1,2,3,4 (Eq.3-1) resulted in corres ponding increase in the estimated feed-forward and feed-back interaction patterns a nd a decrease in the magnitude of phase-lag to near-zero values (Table 3-2). Ho wever, not all reciprocally inte racting systems exhibit near-zero phase-lag synchrony. The actual phase-lag in a network depends on many factors including delays involved in the feed-forward and feed-bac k pathways and the strength/type of coupling. The approach outlined in this study enables one to empirically test the influences of each of the factors on phase-lag in any given data. 3.3.2 Experiment 3.3.2.1 Network identification Granger causality spectra are shown in Fi gure 3-1 for a pair of sites experiencing unidirectional interaction (A) a nd another pair of sites undergo ing bidirectional interaction. Figure 3-2 shows the interaction pa tterns for all five sites pair s in both monkeys in the beta frequency range where the same significance thres hold criterion described in (Brovelli et al., 47

PAGE 48

2004) were used. Except for S1 and 7b in monkey GE and M1 and S1 in monkey LU, the remaining site pairs in both monkeys exhibite d unidirectional interaction. An earlier study examining the causal patterns on the same dataset in a multivariate context using conditional Granger causality has shown that the network analyzed here is robust and free of any significant mediated influences from other simultaneously recorded areas that were not included in the pairwise analysis pursued in this study. Unlike the other five site pairs where the interaction is concentrated in the beta range (see Figure 3-2) the (M1,S1) pair in m onkey LU also exhibited significant interaction in the gamm a frequency range, in addition to that in the beta range. For this pair, the causal influences in the time-domain where instanta neous causality is most readily interpreted, are confounded a nd is thus excluded from furt her analysis. Functionally, the observation that S1 and 7b play a pivotal role in the organization of the network has led to the hypothesis that the beta network supports the maintenance of lever depression by facilitating sensorimotor integration (Brovel li et al., 2004; Ding et al., 2006). 3.3.2.2 Phase-lag distribution For a given site pair, the phase-lag at the peak beta frequency was estimated for each trial. Figure 3-3 shows the phase-lag di stribution for two different pair s of sites in monkey LU and GE. Both distributions are unimodal. In particul ar, despite a unidirectional interaction pattern between M1 and 7b (Figure 3-2B), the phase-lag is approximately cente red around zero with a mean of 0.04 radians. This suggests that for su ch pairs the instantaneous causality may contribute significantly to the overa ll degree of synchrony. Table 3-3 summa rizes the mean phase lag in the beta band for all five pair s of recording sites. 3.3.2.3 Synchrony decomposition and near-zero phase-lag For recording sites A and B, according to Eq. 3-2, the total synchrony, which is derived from the coherence function can be written as the sum of two dire ctional influences (AB) and 48

PAGE 49

(BA) and instantaneous causality (A.B) (Table 34). Intuitively, positiv ely correlated common input measured by the instantaneous causality has the effect of br inging phase-lag close to zero. This is particularly so for pairs experiencing un idirectional causal influe nce. In monkey LU, the phase-lag between M1 and 7b is near zero (Figure 3-2B and Figure 3-3B), and not surprisingly, the instantaneous causality in this case makes up 72% percent of total interdependence, a substantial percentage (Table 3-4). Below we tested the idea by carrying out an analysis for the site pairs characterized by unidirectional interacti on with the analysis prot ocol outlined in section 3.2.2.5. 3.3.2.4 Instantaneous causality and phase-lag For each site pair the phase lag was estimated for each trial and the estimated value was used to sort all trials into subensembles. The phase-lag and the instantaneous causality measure for each subensemble constituted a point on a scatter plot. Figure 3-4 shows the result for (S1,7b) in monkey GE and LU. Clearly, th e two quantities are negatively co rrelated, indicating that as the instantaneous causality increases, the phase-l ag decreases and, in fact, approaches zero. Spearmans rank correlation and Spearmans rank partial correlation coefficients were computed for all the site pairs and listed in Table 3-5 and 3-6. All correla tion coefficients were negative. Except for (M1,7b) in monkey GE, they were statis tically significant at p=0.05 level (one tail ttest) (Table 3-5). By partialing out the effects of the directiona l influences, the correlation for (M1,7b) also became significant (Table 3-6). 3.3.2.5 Bidirectional interaction and phase-lag Inspection of the causality spectrum between the site pair (S1,7b) in monkey GE revealed the presence of bidirectional in teraction in the beta band (Fig ure. 3-1B and Figure 3-2B and Table 3-4). The coherence function and the associated phase spect rum are shown in Figure 3-5. This channel pair was further analyzed to iden tify the effect of recipr ocal interaction on the 49

PAGE 50

phase-lag. The strength of directional interaction is expressed as the sum of feedforward and feedback influences. After par tialing out the influence of the common input the phase-lag was negatively correlated with the mutual interacti on (r = -0.455 p = 0.0045). This result supports our early assertion that in addition to instantane ous causality, bidirectiona l interaction may also lead to near-zero phase-lag. 3.4 Discussion The relative phase between two ne ural signals A and B, referred to as phase-lag here, can be calculated from the cross-spectrum. For decades, the sign and magnitude of phase-lag have been used to infer direction of information transmission and delay (Cassidy and Brown, 2003). Increasingly, phase-lag is found to be near zero in synchronous cortical networks, sometimes involving distant sites. Such phenomenon renders th e use of phase-lag as a measure to identify directional influences ambiguous. The phenomenon of zero phase-lag has been examined from the point of view of neuronal communication an d is considered a manifestation of the brain integrating information from diverse sources (Singer and Gray, 1995; Roelfsema et al., 1997; Engel et al., 2001; Brovelli et al., 2004). Two f actors can be identified that contribute to near-zero phase-lag: (a) positively correlated co mmon input and (b) bidirectional interaction. Recent reports of near-zero phase -lag arising in networks with a predominantly unidirectional interaction pattern further highlights the importance of the first factor. The influence of the two factors on phase-lag was tested on simulated datasets generated by bivariate autoregressive models. First, it was observed that, for a linear system with unidirectional interaction, a near-zero phase-lag was unlik ely in the absence of an instantaneously positively corre lated input. As such an input is introduced and increased, phase-lag is seen to decrease and approach ze ro. A similar influence on phase-lag was also observed in bidirectionally inte racting networks. Second, for the case of a bidirectionally 50

PAGE 51

interacting system with no comm on input, increase in the strength of bot h feed-forward and feed-back interaction leads to the redu ction in the magnitude of phase-lag. The empirical testing of the above ideas f aces considerable challenge as a standard correlation or coherence analysis do not offer sufficient inform ation on the relation between the two signals A and B. Recently, advanced connec tivity tools have been proposed (Baccala and Sameshima, 2001; Kaminski et al., 2001; Seth, 2005; Chen et al., 2006; Lungarella and Sporns, 2006; Wang et al., 2007b; Guo et al., 2008; Wu et al., 2008) which aim at parsing the synchrony into directional interaction. Mathematically a theorem proven by Geweke promises deeper insights. It states that the total interdependence between A and B can be written as the sum of three contributing factors: (AB), (BA) and (A.B). The arrow is understood in the sense of Granger causality and (A.B) signifies instantaneous causality interpreted as reflecting a common input. In the present study Gewe kes theorem was applied to study the contribution of the two factors identified above to near-zero phase-lag. Local field potentials from primary soma tosensory (S1), primary motor (M1), and posterior parietal (7b) area s were recorded in two monke ys performing a sensorimotor integration task. A beta oscillat ory network involving all three sites was identified by coherence. The total interdependence was then decomposed into its directional components. Out of five distinct pairs of recording sites studied, four exhibited predom inantly unidirectional interaction in the beta band. The phase-lag was nearly zero fo r one of the five pairs and relatively small for another three. Unlike simulated mo dels, neither the strength of i nput correlation nor the strength of feed-forward / feed-back inte raction can be manipulated to in fer their influences on phase-lag. The sorting of trials according to their phase-lag is a strategy to deal with this problem. By sorting the trials according to single trial estimated phase-lag, a negative correlation was found 51

PAGE 52

between the phase-lag and instantaneous causality for all pairs, implying that the stronger is the common input the closer to zero is the phase-lag. Despite this tendency, the actual value of the phase-lag for a given pair depends on the relati ve contribution of the each of the factors in Eq. 3-2, and may vary broadly [from 0.04 (near-zer o) to -1.34 (far-fromzero), see Table 3-6]. If the common input is negatively correlated, then the stronger is this input the close to is the phase-lag (Figure 3-6). A careful inspection of the five pairs of re cording sites revealed that the noise terms in their respective autoregressive models (see Eq. 3-1) were all positively correlated with the exception of the channel pa ir (S1,7b) in GE where the noise terms was negatively correlated. Since the order of the reco rding leads used for the bipolar derivation was arbitrary, the signal from S1 was reversed in po larity, which is equivalent to a depth-to-surface recording. This correction enable s the data from all five pairs to be considered under the same hypothesis. Channel pair (S1, 7b) in monkey GE also has anothe r differing characteristic: the interaction is bidirectional in the beta range. In light of the earlier di scussion, the reciprocal interaction in addition to common input coul d also contribute to the observed near-zero phase-lag. This prediction was confirmed by a pa rtial correlation analysis between the phase-lag and the strength of feedforward and feedback interaction after removi ng the influence of the instantaneous common input. In sum, based on our simulated as well as e xperimental data, for two cortical regions engaged in unidirectional interaction, instanta neous causality/common input is likely a main contributor for the near-zero phase-lag between th e sites. On the other hand, for two cortical areas engaged in bidirectional interaction, near-zero phase-lag s ynchrony can emerge as a result of reciprocal interacti on or instantaneously correlated in fluence or a combination of both. Gewekes decomposition theorem, combined with the analysis protocol outlined in the Methods 52

PAGE 53

section can help to ascertain the exact network mechanism for a given problem. Each measure obtained through this decomposition technique has the desirable feature that they all have clear physiological correspondence. For ex ample, bidirectional interacti on is highly interpretable in terms of the anatomical connectivity principle of reciprocity in the co rtex (Felleman and Van Essen, 1991). Instantaneous causality/common i nput may be taken to collectively reflect activation of a distinct cortical or subcortical region that project to the sampled sites. Volume conduction, while a possible contribu tor to instantaneous causality, is an unlikely factor in the present study as bipolar derivati on localizes neural activity to its generator. However, for scalp EEG, the influence of volume conduction is like ly to be significant and must be carefully considered. Table 3-1. Influence of increased instantaneous causality on phase-lag Instantaneous causality Phase-lag (radians) 0.00 2.19 0.04 1.16 0.29 0.44 1.02 0.16 Simulated data were generated by a bivariate model with unidirectional interaction. Here instantaneous causality characterizes the strength of common input. Magnitude of phase-lag is seen to decrease with increase in instantaneous causality. Table 3-2. Influence of increased bidirectional intera ction on phase-lag FX Y / FY X FX Y + FY X Phase-lag (radians) 0.038 / 0.081 0.119 1.31 0.085 / 0.117 0.202 0.94 0.267 / 0.131 0.398 0.36 Simulated data were generated by a bivariate model with bidirectional interaction. FXY and FYX quantify the feedforward and feed-back components estimated by Granger ca usality analysis. Magnitude of phase-lag is seen to decrease with increase in strength of bidirectional inte raction measured by feed-for ward and feed-back Granger causality. 53

PAGE 54

Table 3-3. Mean beta band phase-lag (in radians) for monkeys GE and LU Site pairs GE LU M1,S1 -1.34 M1,7b 0.9 0.04 S1,7b -0.42 -0.4 Table 3-4. Contribution of intrinsic and exogenous compone nts to total in terdependence indicates instantaneous causality, the exogenous component Causality GE (%) LU (%) M1S1 13 29 S1 M1 75 27 M1 S1 12 44 M17b 1 72 7b M1 86 25 M1 7b 13 3 S1 7b 17 48 7b S1 17 9 S1 7b 66 43 indicates directed interaction, the intrinsic components. Table 3-5. Correlation between inst antaneous causality and phase-lag Strength of Spearmans rank correlation coefficient betw een instantaneous causality and magnitude of phase-lag in the beta band between site pairs specified in column 1 for monkeys GE and LU. The results that did not satisfy significance at p<0.05, are included with their corresponding p-value. denotes the channel pair that exhibits bidirectional interaction in the beta frequency band. Site pairs GE LU M1,S1 -0.79 M1,7b -0.33 (.058) -0.70 S1,7b -0.68 -0.73 Table 3-6. Partial correlation between instantaneous causality and phase-lag Site pairs GE LU M1,S1 -0.79 M1,7b -0.31 -0.44 S1,7b -0.41 -0.81 Strength of Spearmans partial rank correlation coefficient between instantaneous causality and magnitude of phaselag in the beta band between site pairs specified in column 1. The partialing is against the directional influences for monkeys GE and LU. The results that did not satisfy significance at p<0.05, are included with their corresponding pvalue. denotes the channel pair that exhibits bi directional interaction in the beta frequency band. 54

PAGE 55

Figure 3-1. Granger causality spectra. (A) a pair of sites expe riencing unidirectional interaction and (B) a site pair experiencing bidirectional interaction in the beta frequency band. The threshold level for significance at p =,0.005 is overlaid as a flat line. Figure 3-2. Schematic Granger causality gr aph. (A) monkey GE and (B) monkey LU. Solid arrows indicate directions of causal influe nce in the beta frequency band that were significant at p = 0.005. 55

PAGE 56

Figure 3-3. Phase distribution. Hi stogram of single trial phase-lag values at peak beta frequency for site pairs (A) (S1, 7b) in monke y GE and (B) (M1,7b) in monkey LU. Figure 3-4. Influence of common input on pha se-lag. Scatter plot showing strong negative correlation between instantaneous causality (IC) and magnitude of phase-lag between site pairs (S1,7b) in monkeys (A) GE and (B) LU. 56

PAGE 57

57 Figure 3-5. Coherence and phase spectra for s ite pair (S1,7b) in m onkey GE. (A) Coherence spectra for site pair (S1,7b) in monkey GE indicating strong beta band synchrony. (B) The relative phase spectra for the same site pair. The near-zero phase-lag in the beta band is the result of both positively correlated common input and reciprocal interaction. Figure 3-6. Phase locking modes. Schematic illustration of different modes of phase locking for a network with unidirectional interacti on pattern and instantaneously correlated inputs. AB denotes the phase-lag between sites A and B.

PAGE 58

CHAPTER 4 PRESTIMULUS THETA OSCILL ATIONS IN FRONTO-TEMPORAL NETWORK PREDICT SUBSEQUENT MEMORY RECALL IN HUMANS 4.1 Introduction Items presented under identical environmenta l conditions are sometimes remembered and sometimes not. Research into the origin of this variability has linked the efficacy of memory encoding in the prefrontal cortex (PFC) and th e medial temporal lobe (MTL) with memory performance. In particular, it has been shown th at the level of physiological responses in PFC and MTL to presented items reliably predicts the likelihood of these items subsequently being remembered (Fernandez et al., 1999; Friedman and Johnson, 2000; Tendolkar et al., 2000; Weiss and Rappelsberger, 2000; Sederber g et al., 2003; Takashima et al ., 2006; Sederberg et al., 2007). While most studies emphasized the stimulus-evo ked activity and its variability as a neural correlate of memory performance, research has gradually expanded to include the characterization of the state of the brain prior to stimulus input and its impact on memory encoding (Seager et al., 2002; Guderian et al., 2009). Ongoing br ain activity is replete with oscillations of various freque ncies (Niedermeyer et al., 1999; Buzsaki, 2006). Summarizing the extensive literature on the theta rhythm (4-9 Hz) in the rodent hippocampus, Vertes (2005) proposed that theta oscillations represent a depolarizing drive on NMDA containing cells, and a stimulus presented during str ong theta activity leads to enha nced response and facilitated long-term potentiation (LTP) (Buzsaki, 1989; Hu erta and Lisman, 1993; Staubli and Xu, 1995; Buzsaki, 1996; Kocsis et al., 1999; Berry and S eager, 2001; Buzsaki, 2002; Seager et al., 2002; Vertes et al., 2004; Hasselmo, 2005; Kocsis a nd Kaminski, 2006). How th is hypothesis manifests in human memory is not well understood. The first goa l of the present investig ation is to test the hypothesis that the efficacy of memory encoding in humans depends on the level of excitability 58

PAGE 59

indexed by the strength of theta activity in PFC and MTL immediately preceding the stimulus onset, with stronger theta predic ting better subsequent recall. Memory is a network phenomenon. Formation of memory traces depends on the coordinated activity in many brai n areas. In particular, the impor tance of PFC-MTL interaction in memory encoding has been supported by evid ence from a variety of studies using lesion (Petrides and Milner, 1982; Gershberg and Shimamura, 1995; Shimamura, 1995; Shimamura et al., 1995; Bechara et al., 1998), neuroimaging (Kapur et al., 1996; Wagner et al., 1998; Cohen et al., 1999; Grady et al., 2003; Gazzaley et al ., 2004), and electrophysiological techniques (Sarnthein et al., 1998; Weiss a nd Rappelsberger, 2000; Anderson et al., in press). Similar to the earlier hypothesis where theta osci llations serve to index the level of tissue ex citability in single brain areas, it is expected that the level of neuronal interacti on and synchrony prior to stimulus onset acts as a gating mechanism and provides an index of increased processing capacity at a network level (Ohara et al., 2006; Ohara et al., 2008). The second goal of this investigation is to test the hypothesis that the st rength of PFC-MTL interaction immediately prior to stimulus presentation, possibly reflecting th e strength of top-down attenti onal control originating in the PFC (Furey et al., 2000; Kimura, 2000; Hasselm o and McGaughy, 2004), is positively correlated with the efficacy of stimulus encoding a nd the likelihood of its subsequent recall. The above goals are accomplished by recording electrocorticogram (ECoG) also known as intracranial electroencephalogram (iEEG), from multiple implanted electrode grids in five patients undergoing presurgical eval uation for intractable epilepsy who performed a subsequent memory task. The subjects were given a series of words to remember. Then, following a distraction period, the subjects were asked to recall the words from memory. Data from the period of initial presen tation of words (encoding period) was analyzed where the level of 59

PAGE 60

regional prestimulus oscillatory activity and interregional inter action immediately prior to the presentation of the words was compared between those that were subsequently recalled and those that were not. 4.2 Materials and Methods 4.2.1 Subjects and Electrode Placement Five patients (4 female and 1 male, mean age 38 years) with medically intractable epilepsy, undergoing invasive video-EEG mon itoring for localization of the epileptic focus and subsequent epilepsy surgery, participated in the study. A ll participants gave informed consent. The experimental and recording pr otocol was approved by the inst itutional review board of the University of Florida. Figure 4-1 illustrates th e approximate positions of the implanted electrode grids in each of the five subjects. Subjects 1, 2 and 3 had subdural grids covering the lateral prefrontal cortex (PFC), Subject s 1, 2 and 4 had subdural grids c overing the lateral temporal lobe (LTL), Subjects 1, 2 and 4 had subdural subtemporal strips on the ventral surface of the temporal lobe, and Subject 5 had bilateral depth electrodes inserted thr ough the occipital lobes into the hippocampi. Data from the electrodes of th e subtemporal strips and hippocampal depth electrodes in Subjects 1, 2, 4 a nd 5 were treated as proxies of medial temporal lobe (MTL) activity. For Subjects 1 and 2 the frontal and the temporal regions were simultaneously sampled. 4.2.2 Paradigm The subjects were asked to perform a subs equent memory task. An LCD monitor placed approximately 90 cms in front of the subjects wa s used to present the stimuli. A fixation cross remained in the center of the screen throughout the experiment. The entire experiment was divided into blocks. Each block, as illustrated in Figure 4-2, consisted of a presentation (encoding) period, followed by a distraction peri od, then a free-recall period, and finally a recognition period. The encoding period consisted of the sequen tial presentation of 20 words 60

PAGE 61

chosen from the Kucera and Francis word pool (Kucera and Francis 1967). Each word was displayed on the screen for 2 seconds with a random delay between words of either 1 or 2 seconds (mean 1.5 seconds). Subjects were inst ructed to memorize as many of the words as possible. The distraction period lasted 30 sec onds and consisted of counting aloud backwards by threes, starting at a random number. This distra ction period served to at tenuate both the recency effect and further encoding by rehearsing. During the free-recall period, subjects were given 50 seconds to recall aloud as many of the previously studied words as they could. This was followed by a second presentation period of a different set of 20 words, followed by another distraction period, and a free recall period. Finally, a longer dist raction period which lasted 60 seconds was given followed by the recognition period that co nsisted of the presen tation of 80 words sequentially on the screen, including the 40 prev iously studied words along with 40 foil words not seen previously. The subjects were asked to respond by button presses to indicate whether or not they recognized the word as being a previously presented word or a new word. The word stayed on the screen until the subject responded. Each response was followed by a random delay of either 1 or 2 seconds (mean 1.5 seconds) be fore the appearance of the next word. Subjects performed either two or three bloc ks depending on their tolerance. The focus of this investigation is how brain activity preceding each word to be memorized (s haded interval in the expanded encoding period in Figure 4-2) influences subsequent encoding and later recall. 4.2.3 Data Preprocessing Data were sampled at 400 Hz by a Nicolet amp lifier system, band pass filtered from 0.1 to 100 Hz, which was subsequently subject to an o ffline lowpass filter with a cutoff at 30 Hz and then downsampled to 250 Hz. Only channels (ele ctrodes) free of pathological abnormalities and epileptic spikes were included for analysis. Ta ble 4-1 summarizes the number of the studied electrodes in each cortical area for each of the five subjects. After rejecting data segments 61

PAGE 62

contaminated by movement and othe r artifacts, the remaining artifact-free data recorded during the encoding period was epoched fr om -500 ms to 0 ms (length of the shaded interval in the expanded encoding period in Figur e 4-2), with 0 ms denoting the onset of word presentation. Epochs where the words were subsequently free-re called were combined into an ensemble of trials, referred to as recalled ensemble, and epochs where the words were not recalled were combined into another ensemble of trials, referred to unrecalled ensemble. Power and synchrony estimates (see below) were obtained for each ensemble for the prestimulus time period -500 ms to 0 ms (see Figure 4-2). The statistical differenc es in these variables be tween the recalled and unrecalled conditions were assessed. 4.2.4 Spectral Analysis 4.2.4.1 Power estimation A multitaper approach was used to estimate pow er spectra for each of the two ensembles of trials (recalled versus unrecalled) from frontal, lateral temporal and medial temporal channels in each of the five subjects (Thomson, 1982; Mitra and Pesaran, 1999; Mitra and Bokil, 2007). Three tapers were found to give the optimal balance between spectral smoothness and frequency resolution. 4.2.4.2 Significance test for diff erence in power between conditions For a given electrode, the differe nce between spectral power for the recalled ensemble and that for the unrecalled ensemble was tested with a nonparametric Jackknife procedure (Bokil et al., 2007; Thomson, 2007). Figure 4-3A illustrates th e behavior of the test statistic for the difference in power across all frequencies betw een the recalled and unrecalled ensembles (dark line) for a single electrode from th e frontal grid of Subject 3. The horizontal gray line denotes the 90% confidence interval based on Gaussian assumption of the log spectra. If the test statistic lies within the 90% confidence band, then the null hypothesis that no difference in power between 62

PAGE 63

the two conditions is not rejected. On the other hand, if the test statistic lies outside the confidence band, then the null hypothesis of equal power is rejected at the significance level of =0.1. To reduce false positives as a result of multiple comparisons, the rejection of the null hypothesis for any particular frequency (f) was carried out only if the test statistic lay outside the confidence band for a contiguous frequency range be tween (f-2) and (f+2) Hz (Figure 4-3A). For the channel illustrated in Figure 4-3A significant differences in pow er were observed in the theta band (4-9 Hz). The aforementioned procedure was carried out for the freq uency range between 3 and 28 Hz. Figure 4-3B summarizes the total number of electrodes over all five subjects that exhibited significant difference in power betwee n the recalled and unrecalled ensembles for each frequency in the 3-28 Hz range. 4.2.4.3 Group power analysis The channels having significant differences in power between the recalled and unrecalled conditions were subjected to group analyses. Band averaged power for each condition was normalized by dividing by the sum of band averag ed power from both conditions and averaged across electrodes within each of the three brain regions (PFC, LTL, MTL) which was further averaged across subjects. The group average was compared between the two conditions and all statistical comparisons were perf ormed using Students t test. 4.2.4.4 Inter-areal synchrony analysis The raw data from each implanted electrode were recorded against a common reference which is typically not free of neural ac tivity. Volume conduction presented a further complicating factor for connectivity analysis. To overcome these problems, bipolar derivations or signals, which were computed as the differences between horizont ally, vertically, and diagonally neighboring electrodes, were used for synchrony analysis. Table 4-2 summarizes the number of bipolar derivations for the two subjects (Subject 1 and 2) who had simultaneous PFC 63

PAGE 64

and temporal lobe recordings. For each pair of inter-grid bipolar signa ls, a frequency domain interdependence measure called the phase locking value (PLV) (Jean-Philippe et al., 1999) was used to evaluate the degree of phase synchr ony between the two brain regions in a given frequency band, with larger PLV values indicat ing higher levels of pha se synchrony, which is taken to imply greater functional connectivity. To calculate PLVs, the data were bandpass-fi ltered between 4 and 9 Hz for theta band, 10 and 14 Hz for the alpha/low beta band and 21 and 25 Hz for the high beta/low gamma band. The instantaneous phase of the signal was extracted using the Hilbert transform on a trial-by-trial basis and the difference in phase between the bipolar pairs of interest was computed. Following an ensemble averaging of the estimate, the instantaneous phase locking measure was derived according to 12 11 ()exp(((,)(,)))N nPLVt jtntn N where (,)itn is the phase of the i-th signal at time t a nd trial n (i=1 or 2). This PLV measure was calculated for the recalled ensemble and unrecalled ensemble separately for each subject and averaged over the 500 ms period im mediately prior to word presenta tion (i.e. -500 ms to 0 ms). The values of PLV range from 0 to 1, where 1 indicates perfect sync hrony across trials, and 0 indicates that the phase rela tionship is completely random from trial to trial. 4.2.4.5 Random permutation test for statistical significance To test the significance of inter-grid PLV, the following procedure was followed (JeanPhilippe et al., 1999; Brov elli et al., 2004), the aim of which is to create a null hypothesis distribution for the PLV measure in the frequenc y band of interest. (1) For each subject, the 500 ms prestimulus epochs were numbered from 1 to N where N is the total number of epochs across both recalled and unrecalled ensembles. (2) For a specific inter-grid comparison, the 64

PAGE 65

epoch index of one of the grids was permuted rando mly to create a synthetic data set where it is reasonable to assume that no interdependence ex ists between any pair of inter-grid bipolar derivations. (3) All pairwise PLV was calculated fo r the synthetic data set and the single largest PLV for the frequency range of interest from a ll bipolar signal pairs was selected. (4) Steps (2) and (3) were repeated 400 times. (5) The null hypothesis distribution was estimated by fitting an extreme-value distribution to the histogram of the largest PLV measures (Wang et al., 2007a). Estimated PLV from the experimental data were considered significant if they exceeded the 99.9th percentile value of the null hypothesis distri bution (p<0.001). Note that testing the many combinations of bipolar signal pairs between grids presents a multiple comparison problem. Choosing the maximum value in Step (3) is a way to account for this problem (Nichols and Holmes, 2002). 4.3 Results Preliminary analysis of this experiment has appeared previously in an abstract form (Rajagovindan et al., 2007). The subjects were attentive throughout the experiment and completed the task according to instructions. On average, 31% of the words were free-recalled, and 88% of the words were correctly identified during the recognition period. These behavioral results are in line with published reports (Sederberg et al., 2003). 4.3.1 Power Spectral power over the interval (-500 ms to 0 ms) preceding word presentation in the encoding period (shaded interval in the expande d encoding period in Figure 4-2) was estimated for the recalled ensemble of trials and for the unrecalled ensemble of trials. The number of recording channels exhibiting si gnificant differences between the two conditions across all three brain areas and subjects is summa rized in Figure 4-3B. Strongest m odulation is seen in the theta band (4-9 Hz) with smaller effects also observed in the alpha/low beta band (10-14 Hz) and the 65

PAGE 66

high beta/low gamma band (21-25 Hz). Figure 44 shows the group-averaged theta band power from the significantly modulated channels in Figure 4-3B for each of the three brain regions. Over the PFC region, theta band power was signif icantly higher prior to the words that were subsequently free recalled relative to words th at were not (p=0.026). Results in the MTL and LTL regions are mixed. While the theta band power is generally higher for the recalled ensemble compared to the unrecalled ensemble, the differe nce failed to reach the statistical significance level of p=0.05. However, by pooling LTL and MTL channels together, the overall temporal lobe theta band power was found to be significantly higher for the recalled words relative to unrecalled words (p=0.046). Similar analysis for the alpha/ low be ta band power (10-14 Hz) and high beta/low gamma band power (2 1-25 Hz) revealed no statistically significant difference in any of the three brain regions (p>0.1). 4.3.2 Inter-Areal Phase Synchrony For the two subjects (Subjects 1 and 2 in Figure 4-1) with si multaneous coverage of PFC, MTL and LTL, the pattern of in teraction between th ese areas over the prestimulus interval (-500 ms to 0 ms) was evaluated by computing th e PLV measure in theta, alpha/ low beta, and high beta/low gamma bands between all possible pairwise combin ations of (a) PFC-LTL bipolar derivations, (b) PFC-MTL bipolar derivations, and (c) MTL-LTL bi polar derivations. Each black dot in Figure 4-5A (Subject 1) and Figure 4-5B (Subject 2), represents the PLV value for a distinct pair of bipolar signals, a nd the height of the rectangular ba rs is the mean of all black dots for a given condition. Significantly higher prestim ulus theta band synchrony is present for the recalled ensemble relative to the unrecalled ensemble in all three pairwise combinations of brain areas (PFC-MTL, PFC-LTL, MTL-LTL) (p<7e-6). Another way of quantif ying the task-related modulation of inter-areal theta synchrony is the number of pair wise combinations whose theta band PLV values exceed the threshold correspondi ng to the significance of p=0.001 (horizontal 66

PAGE 67

lines in Figure 4-5A and 4-5B; see Methods). Since the number of bipolar signals in each grid varied between subjects, the percentages of pair wise bipolar combinations above threshold were calculated and averaged across subjects. As seen in Figure 4-6A, the percen tage of pairs showing significant theta synchrony is significantly higher for the re called ensemble than for the unrecalled ensemble (p=0.0028) in all pairwise co mbinations of the thr ee brain areas: PFC-MTL, PFC-LTL and LTL-MTL. The same synchrony analysis in the alpha/low beta band (Figure 4-6B) and high beta/low gamma band (Figure 4-6C) reveal ed a similar pattern of higher prestimulus synchrony between frontal and temp oral lobe structures for wo rds that were subsequently recalled than words that were not (p=5e-5 and p=0.0002, respectively). A schematic illustrating the relation between the level of prestimulus interregional synchrony and subsequent memory performance is given in Figure 4-6D. 4.4 Discussion In this study, we investigated how neuronal activity preceding the events to be memorized affects their subsequent encoding and recall. ECoG signals from the lateral prefrontal cortex (PFC), medial temporal lobe (MTL), and late ral temporal lobe (LTL) were recorded from implanted subdural electrodes in five epilepsy patients who performed a subsequent memory task. Applying nonparametric spectral methods to the data we obtained two results. First, free-recalled words were associated with higher theta oscillation prior to their presentation over the frontal and temporal lobe structures comp ared to unrecalled word s. Second, inter-areal synchrony in theta as well in alpha/low beta and high beta/low gamma bands preceding the presentation of free-recalled words was higher than that preceding the presentation of unrecalled words. 67

PAGE 68

4.4.1 Prestimulus Ongoing Activity, Theta Oscillations and Memory Encoding Studies on the neuronal mechanisms of memory encoding have placed emphasis on stimulus-evoked responses and their ability to predict memory performance (Klimesch et al., 1996; Fernandez et al., 1999; Friedman and J ohnson, 2000; Tendolkar et al., 2000; Weiss and Rappelsberger, 2000; Jensen and Tesche, 2002; Se derberg et al., 2003; Takashima et al., 2006; Sederberg et al., 2007). In many areas of cognitive neuroscience recent work has begun to elucidate the importance of ongoing brain activity immediately before stimulus onset in defining a state of the brain and the effect of such a state on subsequent stimulus processing (Engel et al., 2001; Liang et al., 2002; Dehaene and Changeux, 2005; Ohara et al., 2006; Ohara et al., 2008; Zhang et al., 2008b). Ongoing brain activity is replete with oscillat ory phenomena. In particular, the role of hippocampal and limbic theta-band (4-9 Hz) EEG activity in learning and memory has been extensively studied (Buzsaki, 1989; Huerta and Lisman, 1993; Staubli and Xu, 1995; Buzsaki, 1996, 2002; Vertes et al., 2004; Hasselmo, 2005; Vertes, 2005). Pharmacological blocking of acetylcholine and medial septum lesions decrease theta activity and lead to memory impairments (Winson, 1978; Givens and Olton, 1990; Berry and Seager, 2001; Si ok et al., 2006) while pharmacological blocking of serotonin enhances theta rhythms and leads to improved performance in spatial memory tasks (Staubli and Xu, 1995; Berry and Seager, 2001; Seager et al., 2002). At the cellular level, episodes of theta oscillations represent a strong depolarizing influence on NMDA receptor-containing cells. Glut amate release triggered by the stimulus input, when temporally coupled with th e theta-induced depolarization, activates the th eta-facilitated NMDA channels and leads to stronger Ca 2+ influx, which in turn facilitates LTP and memory trace formation (Grunwald et al., 1999; Seager et al., 2002; Vertes et al., 2004; Honey et al., 2005; Vertes, 2005; Gallinat et al ., 2006; Lisman et al., 2007; Mea dor, 2007). This suggests that 68

PAGE 69

the level of theta activity prior to the presentation of stimulus is an index of tissue excitability and should be positively correlated with the effi cacy of subsequent memory encoding and recall (Otten et al., 2006; Guderian et al., 2009). Our first finding is cons istent with this hypothesis. In addition, our data demonstrates that the theta-facilitation principle can be extended beyond animal preparations and the temporal lobe to ma nifest strongly in human neocortical structures important for memory encoding such as the prefrontal cortex. 4.4.2 Inter-Areal Synchrony and Network Excitability PFC, MTL and LTL are part of a network of brain areas involved in memory encoding, consolidation and retrieval (Sco ville and Milner, 1957; Janowsky et al., 1989; Tulving et al., 1994; Kapur et al., 1996; Wagner et al., 1998; Kirchhoff et al., 2000; Otten et al., 2002; Strange et al., 2002; Simons and Spiers, 2003). While stim ulus-triggered inter-area l synchronization and its influence on memory performance has been ex amined in the past (Sarnthein et al., 1998; Weiss and Rappelsberger, 2000; Summerfield an d Mangels, 2005), little is known about how inter-areal synchrony prior to stimulus onset serves as a gating mechanism to facilitate subsequent stimulus-driven network operations. Our second finding can be seen as offering the first evidence in humans demonstrating that increased inter-areal synchrony in the fronto-temporal network prior to word presentation is associated with higher likelihood of the word being recalled in the subsequent test (see Ohara et al., 2008 for report of a similar analysis in pain perception). The changes in inter-areal synchrony, taken as an indicator of changes in network excitability, may be a re sult of spontaneous fluctuations in the level of anticipatory attention. Recent studies have shown that anticipatory attention can modulate phase synchrony in a diverse set of cortical networks (Engel et al., 2001; Fries et al., 2001; Ohara et al., 2006; Rajagovindan and Ding, 2008b) and differences in prestimulu s inter-areal synchrony are correlated with conscious perception of stimuli in different sensory modalities (Hanslmayr et al., 69

PAGE 70

2007a; Ohara et al., 2008; Zhang and Ding, in press). It is worth noting that, unlike the theta-facilitating hypothesis where activity in a single frequenc y band is implicated in the indexing of tissue excitability, the prestimulus f acilitation at the network level is across a broad range of frequencies, including th eta, alpha, beta and low gamma. Table 4-1. ECoG electrode information. Numb er of recording channels within each grid overlying the prefrontal cort ex (PFC), lateral temporal lobe (LTL) and medial temporal lobe (MTL) in each of the five subjects. Subject # PFC LTL MTL 1 14 15 4 2 20 9 6 3 43 4 25 7 5 9 Table 4-2. ECoG bipolar electr ode derivation summary. Total number of bipolar derivations in each area and total number of inter-grid pair wise combinations of bipolar signals. # bipolar derivations # inter-grid combinations Subject # PFC LTL MTL PFC-LTL PFC-MTL LTL-MTL 1 26 29 6 754 156 174 2 55 20 7 1100 385 140 70

PAGE 71

Figure 4-1. Approximate placement of electro de grids for each of the five subjects. Figure 4-2. Schematic of the experimental pa radigm. The sequence of events within a single block of the subsequent memory paradigm is given and the encoding phase (shaded in gray) is expanded to highlight the prestimulus interval (-500 to 0 ms) which is the focus of this work. 71

PAGE 72

Figure 4-3. Prestimulus diffe rences in spectral power between the recalled and unrecalled ensembles of words. (A) The test statistic for the difference in prestimulus power between the recalled and unrecalled ensembles across frequencies from 0-30 Hz obtained from a representative frontal electrode. Notice that for this channel the power in the theta band was significantly higher preceding the subsequently freerecalled words relative to the unrecalled word s (shaded in gray). The horizontal gray line marks the p=0.1 level of significance. (B) The number of electrodes exhibiting significant differences in prestimulus pow er between the two ensembles of words across frequencies from 3-28 Hz. 72

PAGE 73

Figure 4-4. Comparison of thet a band power. Group-averaged prestimulus theta band power for the recalled and the unrecalled ensembles of words over th e three brain regions. Here PFC = prefrontal cortex, LTL = lateral tem poral lobe, and MTL = medial temporal lobe. Figure 4-5. Comparison of interareal synchrony. Inter-grid pha se locking value (PLV) results for Subject 1, (A) and Subject 2, (B). (A ) and (B) Theta band PLV values for every pairwise combination of bipolar signals fo r each inter-grid pair is plotted as black dots. From left to right are PFC-MTL, PF C-LTL, and LTL-MTL. Height of gray bars and white bars is the mean PLV for the recalled and the unrecalled conditions respectively. The gray horizontal lines indicate the p=0.001 level of statistical significance for differe nt grid pairs. 73

PAGE 74

74 Figure 4-6. Comparison of per centage of bipolar signal pairs with significant PLVs. Average percent of inter-grid bipolar signal pairs whose theta (A), alpha/low beta (B) and beta/low gamma (C) PLV exceeds the significance threshold for the recalled (gray bars) and the unrecalled (wh ite bars) conditions. (D) Schematic of the relation between levels of prestimulus inter-area l synchrony, represented by the thickness of the curves connecting different brain regi ons, and subsequent memory performance.

PAGE 75

CHAPTER 5 FROM PRESTIMULUS ALPHA OSCILLATION TO VISUAL EVOKED RESPONSE: AN INVERTED-U FUNCTION AND ITS ATTENTIONAL MODULATION 5.1 Introduction Research into the neuronal mechanisms of cognition has mainly relied on stimulus evoked responses and their modulation by higher mental functions such as attention and memory. Recent work has begun to explore how ongoing neural activity prior to stimul us onset influences stimulus processing and behavior In the human visual cortex the relation between prestimulus alpha activity (8-12 Hz) and the s ubsequent event-relate d potential (ERP) has been the subject of extensive investigation. The phase -reset model supposes that the ERP is generated as a result of stimulus-triggered alpha phase realignment and predicts a positive correlation between the prestimulus alpha power and the magnitude of ea rly visual evoked respon se (Makeig et al., 2002; Hanslmayr et al., 2007b) also see (Mazaheri and Jensen, 2006) Tw o lines of empirical evidence appears to be at variance with this prediction: (1) a state of low alpha is found to improve detection and discriminability of threshold-level s timuli (Ergenoglu et al., 2004; Hanslmayr et al., 2007a; Romei et al., 2008a; Van Dijk et al., 2008) and (2) covert attention reduces prestimulus alpha amplitude and at the same time increase stimulus-evoked response (Mangun and Hillyard, 1991; Foxe et al., 1998; Hillyard et al., 1998; Mangun et al., 1998; Marrufo et al., 2001; Sauseng et al., 2005; Thut et al., 2006). The hypothesis th at prestimulus alpha power should be inversely proportional to stimulus evoked response proves to be also contradicted by studies where (1) alpha episodes over the pariet o-occipital areas immediately preceding the stimulus onset enhanced the ability to detect a threshold-level visual s timuli (Babiloni et al., 2006) and (2) higher alpha power led to stronger evoked respon se (Brandt and Jansen, 1991; Brandt et al., 1991; Pa ul et al., 2007). These divergent re ports, in conjunction with recent findings in the somatosensory domain where the mu rhythm (7-13 Hz) is found to exhibit an 75

PAGE 76

inverted-U relationship with th e event-related potential and behavior (Ho and Destexhe, 2000; Linkenkaer-Hansen et al., 2004; Zhang and Ding, in press), suggest that a new assessment of the relation between alpha and visual information processing is needed An added benefit of such a reexamination is that it may yield insights into the fundamental question of how cognitive operations such as attention modulate baseline ongoing neural activity to achieve enhanced stimulus information processing. Noninvasively recorded EEG and invasively recorded local fi eld potentials are a measure of dendritic activity in p opulations of neurons. Computational as well as in vitro and in vivo studies in animal preparati ons have shown that synaptic background activity can exert a significant influence on the res ponsiveness of sensory neurons to stimulus input (Ho and Destexhe, 2000; Chance et al., 2002; McCormick et al., 2003; Wolfart et al ., 2005; Haider et al., 2007). Our approach to investigating the relati on between prestimulus activity and stimulus evoked response starts by considering a theoretical model in which synaptic input to a sensory neuronal ensemble comes in two forms: external stimulus (exogenous) or other brain structures (endogenous). On a moment by moment basis, the endogenous brain ac tivity fluctuates randomly, and the effective proce ssing of a stimulus depends on th e state of the brain defined by these fluctuations at its onset. Supposing that, in the absence of sensory input, the firing rate of the neuronal ensemble is relate d to the magnitude of endogenous synaptic fluctuations through a sigmoidal function, and sensory inpu t has an additive synaptic effect, it follows that the stimulus evoked response corresponds to the derivative of the sigmoidal function, referred to as the gain, which is an inverted-U function of the prestimulus level of bac kground activity. Reformulated in terms of EEG variables where alpha oscillations and the P1 co mponent in the event-related potential are treated as indices of background synaptic activity and stimulus-evoked response, 76

PAGE 77

the model makes a number of predictions, which we re then tested by reco rding EEG data from human volunteers performing a trial-by-trial cued covert spatial visual attention task. Finally, a mechanism on how changes in the global state of the brain, such as selective attention, influenced the aforementioned sigmoidal curv e was proposed. Through e xperimental data and computational modeling, the reduced magnitude and trial-to-trial variability of alpha activity is linked to an increase in corti cal excitability brought about by attention, which enables an increase in the overall slope of the sigmoidal curve and hence hi gher gain to sensory input. 5.2 Materials and Methods 5.2.1 Subjects The experimental protocol was approved by th e University of Florida Institutional Review Board. Nineteen subjects, free fr om movement and neurological disorders and with normal or corrected-to-normal vision, gave wr itten informed consent and participated in the study. Data from twelve subjects (20-31 years of age, 25.16 .5 years, 5 females and 7 males, all right handed) were included in the analyses reported he re. The remaining 7 participants were excluded for a combination of three reas ons: (1) poor behavioral performa nce (target detection rates less than 75%; 3 subjects), (2) excessive eye blin king, eye movement, and muscle activity or electrolyte bridging (3 su bjects), and (3) inability to comple te the experiment (1 subject). 5.2.2 Paradigm Seated in a dimly lit, acoustically and electrica lly shielded room, the subject was instructed to fixate a central cross on a VGA monitor pl aced approximately 85 cm from the eye. Two square boxes, each delineated by white dots placed at the four corners in the upper left visual field and in the upper right visual field, were displayed permanen tly on the monitor (Figure 5-1). Each box, centered 4.7 above the horizontal meri dian and 7.2 lateral to the central fixation, subtended a visual angle of 3.35. A trial began with the onset of a cue, which consisted of either 77

PAGE 78

a left or right pointing arrow of 200 ms in duration at the location of the central fixation cross, instructing the subject to deploy covert attention to the square box on the side indicated by the arrow. After a random time delay between 1800 ms a nd 2200 ms, a standard or a target stimulus of 100 ms in duration appeared either inside the attended squa re box (valid trial) or inside the square box on the opposite side (invalid trial). Th e standard stimulus was a circular checkerboard subtending a 3.3 visual a ngle, and the target stimulus was also a circular checkerboard, but of a slightly smaller diameter. The subject was require d to press a button with their right index finger in response to the valid target stimulus and with hold response to any other stimuli. The standards appeared 80% of the time with 50% validity and the targets appeared 20 % of the time with 66% validity. The interval between the cue offset of the previous trial and the cue onset of the successive trial was randomly varied between 4700 to 5700 ms. The entire experiment was comprised of 15 to 16 blocks of approximately 6 min each (60 trials). Breaks were given between blocks. Subjects received practice sessions of 150 trials to familiarize themselves with the task and to minimize th e effect of learning. Four of the twelve subjects also took part in an additional behavioral experiment in which they responded to targets for both valid and invalid trials. The purpose was to establish the validity effect on the reaction time. In this task the standards appeared 80% of the time and the targets appeared 20% of the time, both with 50% validity. This additional experiment consisted of 10 blocks of 64 trials each. 5.2.3 Data Acquisition The electroencephalogram (EEG) data wa s recorded with a 128-channel BioSemi ActiveTwo System at a sampling rate of 1024 Hz. Four additional electrooculogram (EOG) electrodes were placed around the eyes to monitor eye blinks and horizontal and vertical eye movements. EEG and EOG data was amplified a nd lowpass-filtered with a cutoff set at 205 Hz. 78

PAGE 79

The stimulus and the response was delivered and registered by a BeriSo ft Experimental Run Time System (ERTS) and by a Berisoft EXKEY microprocessor logic pad respectively. In addition to the attention experiment, five minutes of EEG activity during relaxed fixation was also recorded for baseline purposes. 5.2.4 Data Preprocessing Analysis was performed using a combination of BESA 5.1 (MEGIS software GmbH, Munich, Germany) and custom scripts wr itten in MATLAB 7.4 (MathWorks Inc., Massachusetts). The EEG data was filtered offline between 1 and 83 Hz and downsampled to 250 Hz. The channels were re-referenced off-lin e against the average reference (Nunez et al., 1997). Trials contaminated by any of the following factors were excluded from further analysis: (1) incorrect behavioral respons e, (2) excessive muscle or m ovement artifact, (3) eye blinks during a 150 ms period around the on set of the cue and the standard stimulus, and (4) horizontal eye movement following the cue onset and dur ing the period -1000 ms to 300 ms with 0 ms denoting the presentation of the standard stimu li. The average trial rejection rate was 24%. The remaining eye blinks were removed using the ad aptive artifact correctio n procedure in BESA 5.1. The high rejection rate is necessary because the single-trial based analysis protocol here places stringent requirements on the quality of the EEG signal (see below). The data from the fixation only condition was segmente d into 300ms long non-overlapping artifact free epochs. For the spatial attention task, the continuous EEG data was epoche d from -1000 ms to 800 ms. For event-related potential (ERP) analysis, the data e pochs were baseline corrected by subtracting the mean voltage over the inte rval -150 ms to 0 ms. There are four conditions for comparison: a) A ttend Left (cued left and standard stimulus appearing on the left; valid cue), b) Attend Righ t (cued right and standard stimulus appearing on the right; valid cue), c) Ignore Left (cued right and standard stimulus appearing on the left; 79

PAGE 80

invalid cue), and d) Ignore Ri ght (cued left and standard s timulus appearing on the right). Stimulus evoked neural activity measured by ERP as well as ongoing neural activity from the prestimulus time period during which preparatory attention is engaged we re studied. Between two to four neighboring channels overlying the occi pital and parieto-oc cipital areas with comparable magnitudes of P1 evoked component (peak, 90-160 ms) and prestimulus alpha band power (8-12 Hz) exhibiting attentional modulation were chosen for analysis. 5.2.5 Behavior and Event-related Potential Analysis Behavior was quantified by three measures: (1) target detection rate defined as the number of correctly identified valid targets divided by the total number of valid targets presented, (2) false alarm rates for the conditi ons where the participants had to withhold their responses was defined as the number of responses to (a) invali d targets, (b) valid standards and (c) invalid standards, each divided by tota l number of trials in each of these conditions, and (3) mean reaction time for correctly detect ed valid targets. The visual ev ent-related potential (ERP) was computed by averaging the EEG re sponse elicited by the left and ri ght field standard stimuli for the attended and ignored conditions. The amplitude of the ERP components were estimated using the baseline-to-peak method where the baseline pe riod was defined to be (-150 to 0 ms) (Luck, 2005). Wilcoxons signed-rank test was used to ev aluate the statistica l significance for the attentional enhancement of the ERP components over the occipital and parieto-occipital areas. 5.2.6 Correlation Between Prestimulus Alpha Power and P1 Amplitude A key objective of this study is to investig ate how the state of the brain immediately preceding the onset of stimulus affects stimulus processing. The former is characterized by the power of alpha oscillation (8-12 Hz) during the time period from -300 ms to 0 ms and the latter by the amplitude of the visually evoked P1 component (90-160 ms). The method for accomplishing the objective is correlation analysis and it has the following steps. (1) Prestimulus 80

PAGE 81

power for each of the pre-selected channels overlying the parieto-oc cipital area contra-lateral to the visual field of stimulation was computed on a trial by trial basis using the multitaper spectral estimation technique (Thomson, 1982; Mitra and Pesaran, 1999). This power was then averaged over the alpha frequency band and across the set of pre-selected channels. (2) The trials from each subject were then sorted, according to the magnitude of the prestimulus alpha power, from the smallest to the largest, in to 5 groups of equal size (Liang et al., 2002; Zhang et al., 2008b) with 50% overlap. Each group, c ontaining 30% of the total trials, will be henceforth referred to as a power group. (3) For every power group, th e group-mean prestimulus alpha power and the baseline-to-peak group visually evoked P1 component were estimated for each subject. (4) To minimize the effect of inter-subject variabil ity on population averaging, the normalized P1 amplitude was obtained by divi ding the evoked P1 for each power group by the largest evoked P1 component across all power groups under the attend condition fo r every individual subjects. (5) The normalized evoked P1 component was aver aged across subjects a nd plotted against the power group index. The relation between the two variables, reflec ting prestimulus and poststimulus activities respectively, wa s assessed from the scatter plot. 5.2.7 Characterization of Ongoing Alpha Activity Ongoing brain activity in the visu al cortex is dominated by osci llations in the alpha band. On a trial by trial basis, the amplitude of alpha activity varies markedly; collectively, these amplitudes form an alpha power distribution. Past work has focused on the mean of this distribution (mean alpha power) a nd its modulation by attention. He re, in addition to the mean, the variance of the distributi on is also given physiological importance and compared across conditions. Three conditions are cons idered: attend, ignore, and rela xed fixation. For the first two conditions, the 300 ms interval prior to the onset of the standard stimulus constitutes a trial. For the relaxed fixation, continuous recordings were divided into 300 ms nonoverlapping epochs, 81

PAGE 82

and each epoch is considered a trial. The same method as that described above was used to estimate the single-trial alpha power for each hemis phere of every subject. To test the difference in alpha variance between conditions, single-trial alpha power was log transformed to ensure an approximate normal distribution. Population differe nces across subjects in the magnitude of the estimates between a pair of conditions were tested for statistical significance using Wilcoxons signed-rank test. 5.2.8 Computational Modeling A computational model was employed to fu rther test the validity of the proposed framework at a single neuron level. To assess the influence of cha nge in excitability on the slope of synaptic inputoutput curve, the point-c onductance model of background activity introduced by Destexhe et.al., (Destexhe et al., 2001) was employed here. Th e model comprises of a single compartment Hodgkin-Huxley type neuron with excitatory and inhibitory conductance representing the collective e ffect of a large number of syna ptic inputs and the membrane potential is described by the following equation: 1 -----LeakNaKdMsyndV CIIII dt a I (5-1) where C = 1 F/cm2 is the specific membrane capacitan ce, V is the membrane potential and a=34636 mm2 is the total membrane area of a representa tive cortical pyramidal cell (Destexhe et al., 2001). INa is the voltage-dependent Na+ current, IKd is the delayed rectifier K+ current, IM the noninactivating K+ current, and ILeak the leak current. The paramete rs of the passive and voltagedependent currents were taken from (Destexhe et al., 2001). The total synaptic current in Eq. 5-1, Isyn, is decomposed into a sum of two independent currents: ()(-)()(-) s yne ei i I gtVEgtVE (5-2) 82

PAGE 83

where ge(t) and gi(t) in Eq. 5-2 are time-dependent global excitatory and inhibitory conductance, respectively, with reversal potentials of Ee = 0 mV and Ei = -75 mV. The synaptic conductance are described by OrnsteinUhlenbeck stoc hastic processes (D estexhe et al., 2001): 2 0 2 02 1 -(-)( 2 1 -(-)()ee ee e e ii ii i iidg ggxt dt dg ggxt dt )e (5-3) where, ge0 and gi0, e and i, e and i, are the average conductance, standard deviations, and time constants, for the excitatory and inhibitory inputs, respectively. xe(t) and xi(t) are zero mean Gaussian white noise processes of unit standard de viation. The parameter values of Eq. 5-3 were taken as e = 2.73 ms and i = 10.49 ms, gi0 = 0.06 S, ge0 was varied between 0.01S and 0.0364 in steps of 0.0033 resulting in 8 levels of incr easing excitability state. 25 levels of background synaptic activity were simulated by changing the standard deviation of inhibitory conductance i between 0 and 0.0375S in steps of 0.0015S. The parameter e, the standard deviation for excitatory conductance, was kept constant at 0.006S. Qualitatively, similar results were obtained when e was varied with i kept constant or when both i and e were varied in tandem. Note that varying i or e also shares a correspondence with varying input correlation between individual synapses (Destexhe et al., 2001). The firing rates were collected for 40 simulation trials of 200 ms duration each, for 25 values of i for 8 excitability levels. 5.3 Results The relation between prestimulus alpha osci llation and stimulus evoked response and its attentional modulation were inve stigated by combining a theoreti cal model with human visual attention experiment. Below, basic experimental findings are reported first, which is followed by model analysis and test of model predictions. 83

PAGE 84

5.3.1 Behavior The average target detection rate across subjects was 87.27% ( 3.2% SEM) for attended targets appearing in the left visual field and 87.44% ( 3. 68% SEM) for attended targets appearing in the right visual fi eld. These results sugge st that the level of task difficulty was equated between the attend-left and the attendright conditions (p=0.8, two-tailed Wilcoxons signed-rank test). The average fals e alarm rate stemming from res ponses to the target appearing in the ignored location, to the attended standard stimulus, and to the ignored standard stimulus was 1.58%, 4% and 0.15%, respectively. The mean reaction time (RT) was 540.8 ms (36ms SEM) to attended targets appear ing in the left visual field and 501.5 ms (.8ms SEM) to attended targets appearing in the right visual fi eld. The difference between the two mean RTs is not statistically significant. For the 4 subjects who performed th e additional task of responding to targets appearing in either the left or right visu al field after the cueing period, the mean RT to validly cued targets was significan tly faster than to invalidly cued targets: 358.825 (.67 ms SEM) versus 550.422 (.21 ms SEM), (p=0.0078, Wilcoxons signed-rank test), demonstrating the behavioral benef it of cued preparatory attention. 5.3.2 ERP Analysis Event-related potential (ERP) profiles elicited by the standa rd stimuli under the attend and ignore conditions were compared Waveforms from a single channe l of a representative subject are shown in Figure 5-2A. Consis tent with previous reports, the P1 component (90-160 ms), reflecting early response of the extrastriate cortex to stimulus input, is significantly higher for attend versus ignore condition over the posterior electrodes (p = 0.0005, right hemisphere for attend left versus ignore left and p = 0.00024, left hemisphere for attend right versus ignore right); see Figure 5-2. The scal p distribution of the P1 attention effect has an amplitude maximum over the lateral occipital sites contralate ral to the visual field of stimulation for the 84

PAGE 85

attend-left condition (Fi gure 5-3A). For the attend-right cond ition the pattern is more bilateral (Figure 5-3B). Attentional enhancement of th e N1 component (160-21 0 ms) over the parietooccipital areas was also observed but will not be pursued here. Further our emphasis below is on the attentional modulation of prestimulus ba seline activity an d its relation to the early stimulus-evoked P1 component. 5.3.3 Prestimulus Alpha Oscillations Grand-average power spectra from the left and right occipital areas are shown in Figures 5-4A and 5-4B, respectively, for three conditions: attend, ignore and fixation only. Data used for the attend and ignore conditions were ta ken from the prestimulus time period, -300 to 0 ms, with 0 ms denoting the onset of the sta ndard stimulus. For the fixation only condition, continuous recording was divided into non-overlapping epochs of 300 ms in duration. The mean spectral peak frequency averaged across subjects is 9.5 0.39 Hz and no frequency difference is seen between conditions. The power of alpha ac tivity for both attend and ignore conditions is significantly lower than that for the fixation c ondition. Alpha power is further reduced with attention when the attend conditi on is compared to the ignore condition (p<0.002). These effects were established shortly after th e cue onset and persisted until the onset of the stimulus. The scalp distribution of alpha power difference betw een the conditions is shown in Figure 5-4. A hemisphere-specific modulation is seen in Figure 5-4 where alpha activity was maximally suppressed over the parieto-occipita l areas contralateral to the atte nded visual field. Specifically, for the right occipital areas, th e alpha power was significantly lowe r when attention was directed to the left visual field compared to when attention was directed to the right visual field, ignoring the left visual field (p=0.0034 W ilcoxons signed-rank test). For the le ft occipital areas, the alpha band power was significantly lowe r when attention was directed to the right visual field compared to when attention was directed to the le ft visual field, ignori ng the right visual field 85

PAGE 86

(p=0.0024 Wilcoxons signed-rank test). Figures 54D and 5-4E show the comparison with the fixation condition where hemispheric-specific lateralization patterns are also seen. 5.3.4 From Ongoing Activity to Evoked Response: A Model Attention enhanced th e stimulus-evoked P1 component a nd reduced alpha power prior to stimulus onset. Presumably, the prestimulus alpha power reduction contributed to the subsequently improved stimulus processing. Th e physiological mechanism that mediates such improvement remains to be understood. The first step in our attempt to address this issue consisted of putting forth a plausible model linking ongoing prestimulus activity with stimulus evoked response and validating the model by testing its prediction. The model was then used to delineate different mechanisms of prestimulus attention modulation. 5.3.4.1 The sigmoidal input-output function Consider a neuronal ensemble in the sensory co rtex. It is well-accepte d that the level of synaptic input S it receives is related to its output O measured in firing rate through a sigmoidal function O(S) in Figure 5-5A (Freeman, 1979; Destexhe et al., 2001). The synaptic input S can come in two forms: external stimulus (exoge nous) or other brain stru ctures (endogenous). Let SN denote the amount of endogenous prestimulus backgr ound activity at the time of stimulus onset and SX the amount of synaptic activity induced by the sensory input. From the schematic in Figure 5-5A, the stimulus-evoked response is the difference: O(SN+SX)-O(SN). For a constant sensory stimulus, as is the case here, SX is approximately the same from trial to trial over the physiological range of S. The stimulus-evoked response is then proportional to the first derivative of the sigmoidal input-output function O(S), [O(SN+SX)-O(SN)]/SX, referred to as the gain function. As shown in Figure 5-5B, the gain function has an inverted-U shape, suggesting that SN has an effective oper ational range, denoted by W, beyond which stimulus input evokes a negligible response. Also based on this model, it is worth notin g that for a very strong input 86

PAGE 87

(high SX ) the response to the stimulus may satura te for all levels of endogenous prestimulus background activity hence giving rise to a purely decreasing gain with increasing SN An estimate of W is provided by the distribution of SN on a trial by trial basis. Knowledge of this distribution, combined with th e magnitude of the stimulus-ev oked response which reflects the slope of the sigmoidal inputoutput function, forms the basis for evaluating how attention modulates the sensory cortex to achieve enhanced input processing. 5.3.4.2 Model reformulation in terms of electroencephalogram variables The abscissa and ordinate of the model in Fi gure 5-5 are synaptic input and firing rate. While these variables can be prec isely controlled and measured in computational models and in slice preparations, they are not readily availa ble in noninvasive human experiments. Thus, the model in Figure 5-5 requires reformulation to be applicable to our EEG data. First, consider the abscissa. Physiologically, EEG reflects the fluctuating extracellular field due to spatially summated dendritic currents which are in turn ca used by synaptic input (Niedermeyer and Lopes Da Silva, 2005). There is evidence suggesting that the amplitude of EEG fluctuation is proportional to the level of corre lated synaptic input (Ho and De stexhe, 2000; Niedermeyer and Lopes Da Silva, 2005). In this sense, the ab scissa of the model in Figure 5-5A may be equivalently expressed by the amplitude of EEG fluctuation, which is further replaced here by the power of the alpha oscillations. Next, consider the ordinate. Firing rates are not accessible. This means that the sigmoidal input-output curve during the ongoing brain state cannot be directly established with EEG. But the consequen ce of this curve on stimulus processing can be assessed via the early ERP component P1 (90160 ms). Dipole modeli ng and imaging studies have identified the ventral-lateral extrastriate co rtex as the putative source for P1 (Heinze et al., 1994; Clark and Hillyard, 1996; Mangun et al., 1997). Invasive recordings in animals revealed that, upon receiving afferent input from the thalamus or lower visual areas, layer 4 stellate cells 87

PAGE 88

fire action potentials which excite the supragranular pyramidal cells at their basal dendrites spanning layer 3 or portion of layer 4 (Gilber t and Wiesel, 1983; Mi tzdorf, 1985; Callaway, 1998; Thomson and Lamy, 2007). This activation cause s an extracellular sink in layer 3/4 and an extracellular source near th e apical dendrite in layer 1/2, with the latter manifesting as a positive going potential at nearby surface electrodes. See Figure 5-6A for an illustration. This dendritic response in the superficial layers is the summ ated EPSPs to the impinging action potentials arriving from layer 4 and the magnitude of dendriti c response is proportional to the intensity of the action potential input (Jagad eesh et al., 1993; Cash and Yuste, 1999; Araya et al., 2006). The reformulated model in Figure 5-6B establishes the relation between pres timulus EEG power and stimulus-evoked P1 amplitude, where an effective operational range W for baseline activity can again be defined. 5.3.4.3 Experimental test of model predictions In our experiment, although the stimulus is ke pt constant, the backgr ound synaptic activity prior to stimulus onset, assessed by the amplitude of alpha oscillations varies from trial to trial, leading to significant trial-to-tri al variability in stimulus-evoke d responses. Given the model in Figure 5-5, and its reformulation in terms of EEG variables in Figur e 5-6, it is expected that the amplitude of the P1 component follows an inve rted-U function of the prestimulus alpha power. This prediction is tested in Figure 5-7 for the at tend condition. For each trial the amplitude of the prestimulus (-300 to 0 ms) EEG activity in the fr equency range of 8 to 12 Hz was estimated, rank-ordered from the smallest to the largest, and sorted into equalsized groups. The average normalized P1 amplitude in each prestimulus power group was plotted against group index. Consistent with the model prediction, an invert ed-U relation is clearly seen. This relationship was robust across all 12 subjects, with the ex ception of subject 2 and subject 10, where a non-systematic and a negatively co rrelated relationship, respectiv ely, were observed over the 88

PAGE 89

right parieto-occipital regions. Quantitatively, for the left occipital areas, when the subject was validly cued to the right visual field, the P1 amplitude for power group 3 (intermediate alpha power) is 63% and 104% higher th an that for power group 1 (l owest alpha power) and power group 5 (highest alpha power) (p=0.0005 and p =0.00025 respectively, one-tail Wilcoxons signed-rank test). For the ri ght occipital areas, when the subject is validly cued to the left visual field, the normalized P1 amplitude for power group 3 is 57% and 165% higher than that for power group 1 and power group 5 (p=0.02 a nd p = 0.0025 respectively, one-tail Wilcoxons signed-rank test). Quadratic fits appear as solid smooth curves in Figure 5-7 to guide the eye. 5.3.5 Attentional Modulation of Baseline Ongoing Activity: Po ssible Mechanisms Based on the model in Figures 5-6 and results in Figure 5-7, it is plau sible that deployment of effortful covert attention through top-dow n compensatory mechanisms modulates the prestimulus ongoing neural activity, hence optimizing the cortical excitabili ty (Bressler et al., 2008; Fischer et al., 2008; Lakato s et al., 2008) to increase the overall slope of the sigmoidal function O(S) in Figures 5-5 and 5-6, as illustra ted in Figure 5-8A. This global change in the sigmoidal function, corresponding to an increas e in global gain, can account for both reduced alpha power and enhanced P1 by attention (F igure 5-8A). Two predictions can be made according to Figure 5-8A. First, the attention-in duced increase in local gain is not uniformly distributed over the level of the prestimulus EEG power. Specifi cally, the ratio between the two gain functions (attend versus i gnore) has again an inverted-U sh ape, as shown in Figure 5-8B. Second, the overall increased steepness in the sl ope (gain) is accompanied by a reduced effective operational range W, from W1 to W2, implying a reduction in the trial-to-trial variability of prestimulus alpha activity by attention to maximize stimulus-evoked response. Test of prediction 1: The data in Figures 5-7A and 5-7B, togeth er with the data obtained in the same way from the ignore condition, are pl otted in Figures 5-9A and 5-9B. It is evident 89

PAGE 90

that the gain reflecting the slope of the sigmoidal function is enhanced for the attend condition relative to the ignore condition. The ratios of P1 amplitude between the two conditions for different prestimulus alpha power groups are show n in Figures 5-9C and 5-9D. As expected, the gain ratio is maximal for the intermediate power group, in agreement w ith the model prediction in Figure 5-8B. Quantitatively, for the left occip ital areas, the P1 amplitude ratio for power group 3 (intermediate alpha power) is 52% and 68% higher than that for power group 1 (lowest alpha power) and power group 5 (highest alpha power). For the right occipital areas, the P1 amplitude ratio for power group 3 is 16% and 60% higher than that for pow er group 1 and power group 5. It is worth noting that the P1 component is relatively small for the ignore condition. Its subensemble estimation is further complicated by the fact that each power group in a given subject has only about 30 trials. These factors may explain the weak and sometimes even the lack of inverted-U appearance in Figures 5-9A and 5-9B, respectively, fo r the ignore condition. Test of prediction 2: Alpha band power in the interval -300 to 0 ms was estimated on a single-trial basis for each of the three conditions over the pa rieto-occipital sites and logtransformed. Figure 5-10A shows the Gaussian fits to the empirical distributions from a representative subject. The left -ward shift of the distributions under task conditions corresponds to the reduced mean alpha power reported in Figures 5-4 and 5-5. In addition, both the attend and ignore conditions have narrower alpha power di stributions relative to that of the fixation condition, with the distribution for the attend condition further narrowed from that for the ignore condition. Figures 5-10B, 5-10C and 5-10D summ arizes the variance of the empirical alpha distribution across all subjects for the attend and ignore conditions The variance reduction when attention is deployed to the contra lateral visual field (attend) relative to the ipsilateral visual field (ignore) was significant at p = 0.003 over both hemispheres, and the reduction relative to the 90

PAGE 91

fixation condition was significant at p = 0.0005. Th e variance over the left hemisphere for the attend-right condition wa s significantly lower than that for the attend-left condition and for the fixation condition at p = 0.00075 and 0.0005 re spectively. The vari ance over the right hemisphere for the attend-left condition was lowe r than that for the atte nd-right condition and for the fixation condition at p = 0.088 and 0.021 respectively. In additi on, the variance for the ignore condition collapsed over both hemispheres is lower than that for the fixation condition at p = 0.01. 5.3.6 Role of Excitability: Validation in a Single Neuron Model A computational model was constructed as described in section 5.2.8. The firing rates were collected and averaged over 40 simulation tria ls of 200 ms duration each, for 25 increasing values of i (variance of inhibitory syna ptic input) in the range of 0~0.0375S, for 8 levels of increasing excitability (mean memb rane depolarization. Firing rate over the 200 ms interval was averaged across trials and plotte d as a function of the variance of background synaptic activity in Figure 5-11A for three levels of excitability: lo w excitability (less depolarized, red curve); intermediate excitability (more depolarized, bl ack curve) and high excitability (excessively depolarized, blue curve). The derivatives of all three sigmoidal functions are given in Figure 5-11B. Notice from Figure 5-11A and 5-11B that a moderate increase in depolarization increases gain, but too much excitability causes a decrease in gain. This nonlinear relation is further demonstrated by plotting th e firing rate as a function of all 8 levels of excitability and the 25 levels of variance of background synaptic activ ity in Figure 5-11C. The average gain, defined as the weighted mean of the i nverted-U function in Figure 5-11B, is plotted for each level of excitability in Figure 5-11D, which suggests the ex istence of an optimal range of excitability for enhanced stimulus processing. The results of this modeling study conf irm the influence of excitability (mean membrane depo larization) on the gain curve, where within optimal ranges, an 91

PAGE 92

increase in excitability effects an increase in gain. The results obtain through this modeling study is in complete agreement with the observations of attentional modulation at the scalp EEG level hence providing further validation of the proposed model. 5.4 Discussion We considered the relation between visual ly evoked response and prestimulus alpha activity. Three results were found: (1) the P1 amplitu de is shown to be an inverted-U function of the prestimulus alpha power (F igure 5-7), providing key evid ence for the validity of the sigmoidal function model in Figures 5-5 and 56; (2) attentional enhancement of the P1 amplitude is interpreted as reflecting an increase in the overall slope of the sigmoidal function; and (3) in addition to amplitude reduction, the tria l-to-trial variability of alpha is also reduced by attention, corresponding to a reduced working range as a consequence of the attentional modulation of the sigmoidal function (Figure 5-8). In light of thes e results some of the seemingly conflicting observations in rela tion to alpha oscillations a nd stimulus processing may be reconsidered. First, in a given behavioral state (e.g. attend, ignore, or fixation), the inverted-U relation in Figures 5-6 and 5-7 i ndicates that intermediate levels of prestimulus alpha activity lead to strongest stimulus-evoked response. This may explain why both high and low levels of prestimulus alpha activity are not conducive to the detection of a threshold level stimulus (Ergenoglu et al., 2004; Babiloni et al., 2006; Romei et al., 2008b; Romei et al., 2008a). This explanation, together with reports in the somatosensory domain where the rate of detecting a weak stimulus is highest for intermediate leve ls of mu activity (Linke nkaer-Hansen et al., 2004; Zhang and Ding, in press), appears to suggest a mechanism of sensory processing that is modality independent. Second, attentional reduct ion of alpha power, possibly mediated by topdown compensatory mechanisms, is taken to indicat e the increased overall slope of the sigmoidal 92

PAGE 93

function and hence the higher global gain to input, and should not be extrapolated to imply that prestimulus alpha should be inversely pr oportional to stimulus evoked response. 5.4.1 Theoretical Model The state of the brain fluctuates in a seem ingly random manner. The fate of a sensory stimulus depends on the waxing and waning of th e physiological variables characterizing these fluctuations (Lakatos et al., 2005; Lakatos et al., 2008). While behavior such as attention modulates the global dynamics of these variables (Foxe et al., 1998; Sauseng et al., 2005; Thut et al., 2006; Fischer et al., 2008), on a given trial, the processing of a stimulus is affected by the magnitude of these variables immediately preceding its onset. In this work, the global state of a neuronal ensemble is assumed to be defined by a sigmoidal function, where the horizontal axis is the level of background synaptic activity over a su itably defined time interval and the vertical axis is the output firing rate over the same inte rval. On a trial by trial basis, the stimulus-evoked response is proportional to the deri vative of the sigmoidal curve refe rred to as local gain, which is an inverted-U function of background syna ptic activity level pr eceding the stimulus (Figure 5-5). Attention is hypothesized to enhance the global gain to sensory input by increasing the overall slope of the sigmoidal function. This model, while inspired by recent computational studies on the impact of mean and variance of membrane potential fluctuations on a neurons responsiveness to inpu t (Ho and Destexhe, 2000; Chance et al., 2002; McCormick et al., 2003; Wolfart et al., 2005; Haider et al ., 2007), differs from these models in two important ways. First, the horizontal axis in Figure 5-5, rather than bei ng the external input of varying intensity, is the level of moment by moment synaptic backgr ound activity assessable by dendritic field potentials. Second, an explicit mechanism is given, which stipulates that the stimulus-evoked response is an inverted-U function of the le vel of background synaptic activity immediately preceding stimulus onset. 93

PAGE 94

5.4.2 Model Testing and Validation In Figure 5-6, the model in Figure 5-5 is refo rmulated in which the amplitude of the P1 component is equated with the intensity of stimulus-evoked firing and the prestimulus alpha power with the level of background synaptic activ ity. A single-trial base d analysis was carried out, and as predicted, an inverted-U relationship between the P1 amplitude and the prestimulus alpha power was found for both hemi spheres (Figure 5-7). Considered a validation of the model, this finding suggests that the observed increase in P1 amplitude to the attended stimulus reflects increased global gain, which is defined by an in crease in the overall slope of the sigmoidal function (Figures 5-8 and 5-9). The change in the sigmoidal function further predicted a possible leftward shift of the sigmoidal function and a reduced effective operationa l range, both of which were confirmed by computing and comparing across experimental conditions the mean prestimulus alpha power (Figure 5-4) and the va riance of the alpha power distribution estimated on a trial by trial basis (Figure 510). It is worth noting that qu alitatively similar results were observed when alpha power was replaced by total signal variance to index background activity. The alpha activity was chosen as a proxy for the strength of bac kground synaptic activity because (1) all subjects exhibite d prominent spectral power in th is band accounting for most of the total signal variance and (2) robust attentional modulation was observed in this band. In our data attentional modulation in the other freque ncy bands including thet a, beta and gamma was either completely absent or were not statistically significant across subjects. In addition, power in higher frequency bands such as beta and gamma are too small to serv e as a representative of total synaptic background activity. 5.4.3 P1 Genesis and Attentional Modulation The visually evoked P1 component reflects the ea rly response of the extrastriate cortex and is enhanced by spatial attention. See Figures 5-2 and 5-3. Concerning the generating mechanisms 94

PAGE 95

of P1, Mitzdorf (Mitzdorf, 1985) showed in cats th at a surface positive potential may arise due to excitatory synaptic activation of the supragra nular pyramidal cells at their basal dendrites spanning layer 3 or portion of layer 4, which causes an extracellular sink at layer 4 and a superficial source sensed by the surface electrode. In behaving macaques, Schroeder et al. (Schroeder et al., 1991) localized the P60 co mponent, the monkey equivalent of the P1 component in humans, to pattern evoked and patter n reversal VEP to layer 2/3. Associated with the P60 was an increase in MUA activity signifyi ng that it is predominantly a depolarization of the supragranular pyramidal cells in the superficial layers. The exci tatory input to the superficial layers is considered to be from the layer 4 stella te cells which receive afferent input from either thalamocortical projections or inputs from earlier sensory areas (Figure 5-6A). According to these results, the increased P1 amplitude by attention reflects the increased firing of neuronal populations in layer 4, which is the basis for the choice of P1 for our analysis. The origins of the N1 and other later evoked components (Clark and Hillyard, 1996; Di Russo et al., 2002; Di Russo et al., 2003) are more complex. Thei r generation, involving re entrant processing and the interaction among multiple brai n areas (Garrido et al., 2007), depends to a limited degree on visual alpha. These components are not pursued further. 5.4.4 Alpha Oscillation: Neuronal Mechanisms and Attentional Modulation The genesis of alpha oscillations in visual cortex is debate d. While early investigators focused on the pacemaking role of the thalamus, more recent work stresses the cortical origin of the alpha oscillations. Bollimunta et al.(2008), an alyzing laminar field potentials and multiunit activity from extrastriate cortical areas V2 a nd V4 of behaving macaques, found local generators of alpha in superficial layers, layer 4 and deep layers, and showed that the activities of these generators are highly synchronize d. According to this view, th e widely observed decrease of EEG alpha power in areas contralateral to the atte nded visual field (Figures 5-4), likely reflects 95

PAGE 96

similar decreases in the entire cortical column, allowing us to establish a link between scalp measured alpha power and layer 4 neuronal firing, the latter being proportional to the magnitude of the P1 component. Physiologically, the suppression of alpha is take n to imply increased cortical excitability (Foxe et al., 1998; Bastiaansen and Brunia, 2001; Marrufo et al ., 2001; Sauseng et al., 2005; Thut et al., 2006; Klimesch et al., 2007) (Jones et al., 2000; Worden et al., 2000; Goldman et al., 2002; Kelly et al., 2006; Rihs et al., 2007; Jones et al., 2008; Ro mei et al., 2008a). Cortical excitability is often understood in terms of mean depolarizati on levels of cellular membrane potentials and it is known to have a complex relation with stimulus-evoked response. For example, the absence or very low levels of s pontaneous oscillatory ac tivity may fail to bring local neuron populations closer to firing threshold and excessi ve depolarization due to high levels of spontaneous activity may cause short-te rm synaptic depression, both of which can lead to diminished response to sensory input (Abbott et al., 1997; Chung et al., 2002; Petersen et al., 2003; Dehaene and Changeux, 2005). While enhanced stimulus processing by attention in our experiment is thought to be medi ated by increase in the slope of the sigmoidal function or gain, as defined in Figure 5-8, exci tability and gain in the current framework are not mutually exclusive concepts, and their relationship was ex plored in a computati onal study (Figure 5-11). The results show that (1) there is an optimal range in the mean membrane depolarization which gives rise to highest overall slope of the sigmoidal function, and (2) both insufficient and excessive depolarization reduce the slope of the sigmoidal function, thereby hindering the gain to sensory input. This computational study, together with the experimental data, suggests that (1) stimulus processing benefits directly from increased global ga in and (2) attention optimally modulates excitability to enable such gain increase. 96

PAGE 97

Figure 5-1. Experimental task. At the beginning of each trial, th e cue (left/right pointing arrow) presented at the fixation inst ructs the subjects to attend to one of two demarcated locations. After 1800 to 2200 ms a unilateral im perative stimulus (target/standard) is presented on one of the locations. The subjects had to indicate by a speeded response to the occurrence of the target s timulus at the attended location. 97

PAGE 98

Figure 5-2. ERP attention effect. (A) Typical waveform of visual ERP ove r the parieto-occipital region contralateral to the visual field of stimulation. Note that attending to the stimulus location produces an enhanced P1 and N1 component. (B) Grand average peak P1 amplitudes averaged over channels overlying the occipital / parieto-occipital areas contralateral to visual field of stim ulation, indicating attentional enhancement when the stimulus location was attended re lative to ignored. ATL attend left, IGL ignore left; ATR attend right, IGR ignore right. Panels (C) and (D) summarizes the normalized visual evoked P1 component ove r 12 subjects over the left parietooccipital areas (C) and right parieto-occipita l areas (D) when attention was paid to the contralateral visual field (filled black-Attend), contralateral visual field was ignored (filled white-Ignore). Normalization was pe rformed only for display purposes and was obtained by dividing the respective c onditions by the sum of estimates of both conditions. 98

PAGE 99

Figure 5-3. Scalp distribution of attentional enhancement. Di fference ERPs of grand average over 12 subjects at 120 ms after presentati on of the standard stimulus. Note the posterior P1 (shaded red) and fronto-central N1 (shaded blue) attention effect. (A) Attend minus ignore difference ERP for left visual field input and (B) Attend minus ignore difference ERP for ri ght visual field input. 99

PAGE 100

Figure 5-4. Attentional modula tion of alpha band activity. Gra nd average power spectrum over 12 subjects over left parietooccipital (A) and right pari eto-occipital areas (B) when attention was paid to the contralateral visual field (black), contralateral visual field was ignored (red) and for relaxed fixati on condition (blue). (C)-(E) highlights the scalp topography of difference in alpha band power (-300 to 0 ms prior to standard stimulus) averaged across subjects. (C) A ttend left minus attend right, blue shading over the right parieto-occipita l areas indicate lower alpha band power to attend left (contralateral visual field) relative to atte nd right (ipsilateral vi sual field) and red shading over the left parieto-occipital areas indicate lower alpha band power to attend right relative (contralateral visual field) to attend left (ipsilateral visual field). (D) Attend left (ATL) minus relaxed fixation cond ition (FIX), blue shading over the right parieto-occipital areas indicat e a reduction in the alpha band with attentio n as against relaxed fixation. (E) Attend right (ATR) mi nus relaxed fixation condition, both left and right parietal-occipital regions sh ow reduction in alpha band power (blue shading) when attention is directed to the right visual field as compared to relaxed fixation condition. 100

PAGE 101

Figure 5-5. Schematic illustration of theoretic al model. (A) For a neuronal ensemble the spontaneous output firing rate O measured over a suitably define d time interval is a sigmoidal function of endogenous background synaptic activity level SN measured over the same interval. Letting SX denote stimulus-induced synaptic activity, the stimulus-evoked firing rate, O(SN+SX) O(SN), is proportional to the derivative (local gain) of the sigmoidal curve, [O (SN+SX) O(SN)] / SX, shown as the red curve in (B). In this model, when the s timulus is kept constant, the stimulus-evoked response is an inverted-U function of th e prestimulus level of background synaptic activity. The variable W in (B) defines an effective range of pr estimulus level of synaptic activity beyond which stimulus evokes a negligible response. 101

PAGE 102

Figure 5-6. Model reformulation in terms of EEG variables. (A ) Neural genesis of visually evoked P1. Feedforward activation of layer 4 stellate cells excite s the basal dendrites of the superficial layer pyramidal cells, le ading to an extracellular source comprised mainly of passive return current near the co rtical surface, which is sensed by the scalp electrode as a positive-going ERP compone nt. (B) Reformulated model in which prestimulus EEG power in alpha band is e quated with level of background synaptic activity and magnitude of P1 with intensity of stimulus-evoked response. This model predicts that the P1 amplitude is an inve rted-U function of prestimulus alpha power with the variable W defining a range of effective operation. 102

PAGE 103

Figure 5-7. Relationship between prestimulus alpha power and stimulus processing. Trials with attended standard stimuli were considered (valid trials). Single trials were rank ordered by the magnitude of prestimulus al pha power (-300 ms to 0 ms) and sorted into 5 groups of equal size. The P1 component was determined for each group, normalized and averaged across subjects, and plotted as function of group index in (A) for left parieto-occipita l areas and (B) for right pa rieto-occipital areas. An inverted-U relationship is clearly seen where the solid li nes represent quadratic fits. Error bars are the standard error of the mean. Figure 5-8. Schematic illustration of attenti onal modulation mechanisms of ongoing activity. (A) Attention shifts the sigmoidal curve in Fi gures 4 and 5 left-ward and increases its overall slope (global gain). (B) The ratio of the gain curves in (A) for attend and ignore conditions is again an inverted-U function. 103

PAGE 104

Figure 5-9. Gain ratio as a function of pres timulus alpha. P1 amplitude as a function of prestimulus alpha power is plotted for bot h attend and ignore conditions in (A) and (B). The ratios between the two curves ar e shown in (C) and (D). Maximum gain is achieved for intermediate levels of prestimulus alpha activity. 104

PAGE 105

Figure 5-10. Distribution of alpha power. (A) Distributions of l og-transformed single trial alpha power for attend, ignore and fixation conditions over left parieto-occipital areas from a representative subject. Variance of the single trial alpha power distribution is compared over subjects between attend and ignore conditions in (B) where a reduction in variance is evidence of decrease d trial-to-trial alpha power variability. 105

PAGE 106

106 Figure 5-11. Computational study of relation between excitability a nd gain. (A) Firing rate over a 200 ms interval as function of background s ynaptic level for three levels of mean membrane depolarization (blue: high excitability, black: intermed iate excitability, and red: lower excitability). (B) Derivative of the sigmoidal curv es in (A) showing nonlinear relationship between gain and excita bility. (C) Firing rate as a function of mean membrane depolarization levels and variance of background synaptic activity. (D) Average gain for increasing leve ls of membrane depolarization.

PAGE 107

CHAPTER 6 TOP-DOWN CONTROL OF HUMAN VISUAL CORTEX IN ANTICIPATORY SPATIAL VISUAL ATTENTION 6.1 Introduction Early studies into the neuronal mechanisms of attention focus on attentions role in the selective processing of stimuli. How attention modulates endogenous anticipatory/preparatory processes that precede th e appearance of stimuli is less know n. Emerging reports indicate that preceding changes in stimulus-evoked response pr operties, deployment of attention modulates the excitability of the sensory areas that facili tate subsequent stimulus processing (Liang et al., 2002; Dehaene and Changeux, 2005; Padilla et al., 2006; Weissman et al., 2006; Bestmann et al., 2007; Eichele et al., 2008; Zha ng et al., 2008b; Gregor iou et al., 2009; Zhang and Ding, in press). However, such goal-directed prestimulus m odulation of sensory cortical activity is not an inherent property of the early sensory areas. In stead, it is believed th at the frontalparietal network facilitates an enhancemen t of the excitability of the ta sk-relevant and feature-specific visual areas and perhaps a suppres sion of activity in ta sk-irrelevant visual areas in advance of visual stimulation by modulating it s baseline activity (Chelazzi et al., 1993; Luck et al., 1997; Hillyard et al., 1998; Kastner et al., 1999; Hopfi nger et al., 2000; Kastner and Ungerleider, 2001; Corbetta and Shulman, 2002; Gazzaley et al., 20 05; Giesbrecht et al., 2006; Grent-'T-Jong and Woldorff, 2007; Klimesch et al., 2007). In line with this notion, it has been identified that deployment of covert spatial visual attention is associated with increased activation in a network involving dorsal frontal and parietal cortices with concomitant increases in activation over retinotopic early visual areas (Desimone, 1996; Gitelman et al., 1999; Kastner et al., 1999; Hopfinger et al., 2000; Corbetta and Shul man, 2002; Giesbrecht et al., 2003). While prior neuroimaging and electrophysiologi cal studies of the top-down control of spatial visual attention have offered insights into the areas in the brain that exert control over 107

PAGE 108

stimulus processing, the neural mechanisms by which this is implemented are still unclear. Apart from prestimulus modulation of activity over visual areas that facilitate stimulus processing, putatively induced by topdown signals from the frontalparietal network, empirical evidence of such inter-areal interaction is lacking with the exception of (Fuster et al., 1985; Siegel et al., 2008; Gregoriou et al., 2009). Ho wever, recent microstimulation and rTMS studies in monkeys and humans respectively on frontal eye fields (FEF ) and intraparietal sulcus (IPS) have provided casual evidence of top-down influences of th ese areas on activity in the visual cortices (Armstrong et al., 2006; Taylor et al., 2007; Capotosto et al., 2009) during anticipatory attention. Additional evidence of frontal dependent toni c top-down modulatory influence on visual processing in humans comes from lesion studies (Knight, 1997; Knight et al., 1999; Barcelo et al., 2000; Yago et al., 2004) which is also supported by anatomical evidences of neural substrates that enable such long-distant communication coming from axonal tract-tracing and diffusion tensor imaging in animals and humans (Ungerlei der et al., 1989; Felleman and Van Essen, 1991; Petrides and Pandya, 1999; Barbas, 2000; Rempel -Clower and Barbas, 2000; Petrides and Pandya, 2002). To identify the neural signals that mediate the interaction between the fronto-parietal and early visual areas during anticipatory attention, we studied the activity during the delay period between an attention directing cue and the subsequent imperative stimulus from human volunteers performing a trial-by-trial cued spat ial visual attention ta sk. Physiologically, EEG field potential oscillations reflect the rhythmic shifting in excitability due to fluctuating extracellular dendritic currents which are in turn caused by synaptic input (Niedermeyer and Lopes Da Silva, 2005). Recent studies on animals and humans have shown that the alpha rhythm (8-12 Hz) over the visual areas is attenuated in a topographica lly-specific fashion dependent on 108

PAGE 109

the hemifield of attentional cueing and this state of reduced alpha activity is considered to reflect increased cortical exci tability (Foxe et al., 1998; Worden et al., 2000; Bastiaansen and Brunia, 2001; Marrufo et al., 2001; Sause ng et al., 2005; Kelly et al., 2006; Thut et al., 2006; Jokisch and Jensen, 2007; Klimesch et al., 2007; Rihs et al., 2007; Romei et al., 2008b; Romei et al., 2008a; Van Dijk et al., 2008). While sync hronization of neuronal oscillati ons within and across cortical areas has long been suggested as a potent mechanism of attentional selection and sensory gating (Salinas and Sejnowski, 2000; von Stein et al., 2000; Engel et al., 2001; Fries et al., 2001; Salinas and Sejnowski, 2001; Fries, 2005; Buschman and Miller, 2007; Lakatos et al., 2007; Melloni et al., 2007; Lakatos et al., 2008; Rajagovindan a nd Ding, 2008b; Rajagovindan and Ding, 2008a; Siegel et al., 2008; Gregoriou et al., 2009), evid ence as to whether and how selective attention modulates the rhythmic sync hronization between frontal, parietal, and early visual regions in the human brain remains spar se (Doesburg et al., 2008 ; Siegel et al., 2008). Further the exact neural mechanis m of how this increased excitability over the visual sensory areas is brought about has not b een clearly established. The goal of this study is to examine the neural mechanism by which the visual cortical ex citability is controlled by the frontal-parietal network. In light of prior anat omical and lesion evidence (Goncha r and Burkhalter, 2003; Knight et al., 1999), two competing hypothe ses may be put forth: increas ed visual excitability and reduced alpha rhythm is mediated through (1) in creased excitatory top-down input to local pyramidal neurons or (2) decreased excitatory top-down input to lo cal inhibitory interneurons. In particular, three questions are considered: (1) wh at is the signal that implements the frontalparietal control in visual cort ex, (2) how functional connectivity between executive and sensory areas is modulated by attention, and (3) whether top-down control of visual activity is transient or sustained. These questions were addressed by recording scalp EEG from healthy volunteers 109

PAGE 110

performing a trial-by-trial cued spatial visual attention task and by performing multivariate spectral analysis to assess the regional and the inter-areal oscillatory synchronization and causal interaction between the fronto-pari etal and the visual areas during a period of sustained covert attention. 6.2 Materials and Methods 6.2.1 Subjects The experimental protocol was approved by th e University of Florida Institutional Review Board. Nineteen subjects, free fr om movement and neurological disorders and with normal or corrected-to-normal vision, gave wr itten informed consent and participated in the study. Data from twelve subjects (20-31 years of age, 25.16 .5 years, 5 females and 7 males, all right handed) were included in the analyses. The remaining 7 participants were excluded due to poor performance or excessive artifacts. 6.2.2 Paradigm Seated in a dimly lit, acoustically and electrica lly shielded room, the subject was instructed to fixate on a central cross on a VGA monitor pl aced approximately 85 cm from the eye. Two square boxes, each delineated by white dots placed at the four corners in the upper left visual field and in the upper right visual field, were displayed permanen tly on the monitor (Figure 6-1). Each box, centered 4.7 above the horizontal meri dian and 7.2 lateral to the central fixation, subtended a visual angle of 3.35. A trial began with the onset of a cue, which consisted of either a left or right pointing arrow of 200 ms in duration at the locati on of the central fixation cross, instructing the subject to deploy covert attention to the square box on the side indicated by the arrow. After a random time delay between 1800 ms and 2200 ms, a standard or a target stimulus of 100 ms in duration app eared either inside the attended squa re box (valid trial) or inside the square box on the opposite side (invalid trial). Th e standard stimulus was a circular checkerboard 110

PAGE 111

subtending a 3.3 visual angle, a nd the target stimulus was also a circular checkerboard, but of a slightly smaller diameter. The subject was require d to press a button with their right index finger in response to the valid target stimulus and with hold response to any other stimuli. The standards appeared 80% of the time with 50% validity and the targets appeared 20 % of the time with 66% validity. The interval between the cue offset of the previous trial and the cue onset of the successive trial was randomly varied between 4700 to 5700 ms. The entire experiment was comprised of 15 to 16 blocks of approximately 6 min each (60 trials). Breaks were given between blocks. Subjects received practice sessions of 150 trials to familiarize themselves with the task and to minimize th e effect of learning. 6.2.3 Data Acquisition The electroencephalogram (EEG) data was r ecorded with a fully DC-coupled 128-channel BioSemi ActiveTwo System at a sampling rate of 1024 Hz. Four additional electrooculogram (EOG) electrodes were placed around the eyes to monitor eye blinks and horizontal and vertical eye movements. EEG and EOG data was amplified and lowpass-filtered with a cutoff set at 205 Hz. The stimulus and the response was delivered and registered by a BeriSoft Experimental Run Time System (ERTS) and by a Berisoft EXKE Y microprocessor logic pad respectively. 6.2.4 Data Preprocessing Analysis was performed using BESA 5.2 and custom scripts written in MATLAB 7.4. The EEG data was low-pass filtered off-line with a cutoff set at 83 Hz and downsampled to 250 Hz. The channels were re-referenced off-line against the average reference (Nunez et al., 1997; Ferree, 2006). Trials contaminated by muscle ar tifacts, movement artifacts, or excessive eye blinks were excluded from furt her analysis. Trials with correct and incorrect behavioral responses to validly cued target stimulus were analyzed separately. The voltage time series was transformed into continuous current source dens ity (CSD) estimates (Pe rrin et al., 1989; Nunez 111

PAGE 112

et al., 1997; Scherg et al., 2002) to mitigat e the adverse impact of volume conduction and common reference on functional connectivity meas ures. The continuous CSD data was epoched to yield two datasets (a) from -300 ms to 1500 ms around the cue onset w ith 0 ms indicating the presentation of the left-pointing or right-pointing arrow and (b) from -300 ms to 0 ms with 0 ms indicating the onset of the imperative stimuli (standards and targets). 6.2.5 Data Analysis Spectral measures: Three analyses were performed on the EEG data. The multivariate autoregressive (MVAR) modeling method was used in all three analysis (Ding et al., 2006). For an ensemble of trials at each recording site th e ensemble mean was removed from each trial. This allowed the residuals to be treated as coming from a zero mean stochastic process required by MVAR data modeling. Model order between 23 a nd 25 were found to be optimal across subjects, determined both by the Akaike information cr iterion and comparing the spectral estimates obtained by Fourier based methods (Thomson, 1 982; Mitra and Pesaran, 1999). Hence a fixed model order of 24 was chosen for all subjects. Th ree spectral quantities were derived from the model: power, coherence, and Gran ger causality. The regional os cillatory activity quantified by the power was averaged across th e preselected channels overlying the regions of interest and further averaged within the frequency bands of interest. The inter-areal coherence was obtained by averaging the coherence across all pairs of site s representing the different regions of interest and further averaged within a given frequency band. Granger causality was calculated in which the set of channels representing a given area (frontal, parietal, or visual) was treated as a block (Wang et al., 2007b). From the model, Granger causality spectra for FrontalOccipital, ParietalOccipital, OccipitalFrontal, OccipitalParietal, FrontalParietal and ParietalFrontal were derived and averaged over the frequency band of interest. As with any other physiological variables, pow er, coherence and Granger causal ity values varied markedly 112

PAGE 113

from subject to subject. To remove the possible adverse effect of this variability on population averaging, the spectral measures for each cond ition were first normalized by dividing by the corresponding measures from th e ignore condition, and then av eraged across subjects. The Wilcoxons rank-sum test was then applied to asse ss the statistical signifi cance of the differences between conditions. Analysis 1: Attentional modulation of power, coherence and Granger causality in the time period -300 ms to 0 ms prior to the onset of the imperative stimulus was analyzed for trials with correct behavioral responses. Five regions of in terest were identified: anterior medial frontal, medial frontal, parietal and lateral occipita l areas where each area is represented by six neighboring channels (Figure 6-2D). Analysis 2: The behavioral consequences of anticipatory oscillatory activity were examined. Trials with validly cu ed targets were rank-ordered acco rding to the reaction time (RT) and divided into two non-overlapping groups of equal size: fast RT group and slow RT group. Error trials were given a reacti on time of infinity. Spectral me asures over a period of -300 to 0 ms prior to stimulus onset were estimated a nd the Wilcoxons rank-sum test was then applied to assess the statistical significance of the differences in the ongoi ng oscillatory activity preceding the two classes of trials. Analysis 3: Temporal functions of power, coheren ce and Granger causality were estimated using a sliding window approach. Each window was 140 ms (35 sample points) in length and stepped forward in 4 ms increment. A total of 416 windows resulted for the time period -300 ms to 1500 ms around the onset of the attention directing cue. The time designation of each window corresponded to the middle of the covered interval. 113

PAGE 114

Significance testing: For interdependence measures the following random permutation procedure was adopted for st atistical significance assessmen t (Nichols and Holmes, 2002; Brovelli et al., 2004). (1) Consider the data fr om two recording site s (A and B). Randomly pairing data for site A with data for site B from a different trial leads to the formation of a synthetic ensemble of trials for which there is no interdependence. Performing such random pairing with many different permut ations (500 permutations used fo r this study) will result in the null-hypothesis distribution for the interdependen ce quantity of interest from which a threshold can be derived that corresponds to a given significance level ( = 0.01 chosen for this study). The value from the actual data is compared with the threshold for the assessment of statistical significance. This procedure was implemented for the coherence measures over each subject during the prestimulus time period for Analysis 3. Eleven out of the twelve subjects exhibited significant alpha band coherence between the fronta l, parietal and occipital areas, revealing a coherently oscillating network. Granger causality was then estimated between the regions of interest over the eleven subject s who exhibited cohere nt oscillations in the alpha band between areas implicated in the visual attention netw ork. (2) Between the attend and ignore conditions there are two temporal functions of averaged alpha band spectral estimates for Analysis 3. The statistical significance of their difference was assessed by employing the Wilcoxons sign-rank test over each sliding window. The differences we re deemed statistically significant only if the differences were significant over at least 34 consecutive slidi ng windows to minimize potential false positives due to multiple comparisons. Time periods of significant difference were indicated on the plots for comparison. 6.3 Results Twelve subjects performed the task according to instructions. The average target detection rate was 87% and the mean reaction time (RT) was 521 ms. The average false alarm rate 114

PAGE 115

stemming from (1) response to the target appeari ng in the ignored location, (2) response to the attended standard stimulus, and (3) response to the ignored standard stimulus was 1.6%, 4.0% and 0.2%, respectively. Analysis 1: For the time period -300 to 0 ms prior to the onset of the standard stimulus all subjects exhibited prominent power spectral p eaks in the alpha band (8-12 Hz) accounting for most of the total signal variance. No consistent po wer spectral peaks were s een in theta (4-8 Hz), beta (13-25 Hz), and gamma (25 to 40 Hz) freque ncy bands. In agreement with previous work, alpha activity for the attend condi tion is significan tly lower than for the ignore condition (p=1.2e-7), and maximum alpha suppression wa s seen over the parieto-occipital areas contralateral to the attended visual field. While theta and beta power were reduced by attention (p=0.04 and p=1.6e-3 respectively), given the lack of consistent spectral peaks in theta and beta frequencies across participants, it is highly probable that their a ttentional modulation stems from power leakage associated with the spectral es timation procedure from the alpha band. Gamma band power was not different between the attend and the ignore conditions (p=0.59). FrontalOccipital and Parietal-Occipita l coherence in each frequency band was evaluated using the channels overlying the anterior medial frontal, medial frontal, pari etal, and lateral occipital areas in Figure 6-2D and the difference between the tw o conditions were not statistically significant (p>0.13). However, Granger causality spectra in Figures 6-2A and 6-2B between the same set of channels for a typical subject shows a reduction in the alpha band when the attend is compared with the ignore condition. This reduction of Frontal Occipital and Parietal Occipital causal influences in the alpha band was consistent across subjects as shown in Figure 6-2C. Statistically, collapsed across left and ri ght occipital areas, alpha range Frontal Occipital and Parietal Occipital for attend condition is significantly lower than for ignore condition (p=0.035 115

PAGE 116

and p=0.06 respectively). The difference in cau sal influences in the opposite direction, Occipital Frontal and Occipital Parietal, between attend and ignore condition for the same time period failed to reach statistical significan ce (p>0.79). In addition, as summarized in Table 1, Granger causality values in theta, beta and gamma bands are not modulated by attention, suggesting that the alpha oscill atory activity is likely the sign al that mediates top-down attentional influences. Analysis 2: To identify if the alpha oscillatory activity during the anticipatory attention period preceding the presentation of the targets is correlated w ith behavioral performance, power, coherence and Granger causality in alpha ba nd were derived for the period -300 to 0 ms before validly cued target stimulus for the fast and slow reaction time gr oups (Figure 6-3). In line with the result of Analysis 1, we found that for the trials in the fast reaction time group where the subjects correctly identified the targets and ma de a quicker response, the alpha band power over the occipital areas contralateral to the attended location, and the alpha band coherence between the fronto-parietal area s and the occipital areas contralate ral to the attended location, were significantly reduced relative to the slow reaction time group (p=2.4e-5 and 2.2e-4 respectively). Similarly, the causal influence from the anteri or medial frontal regi on into occipital areas contralateral to the attended location was signifi cantly reduced for the fast reaction time group compared to the slow reaction time group (p=0.0 35). However, no statistically significant difference in causal influence from the medial frontal areas sampled around the electrode Fz and from the medial parietal areas into the occip ital areas were observed (p>0.12). Statistically significant difference in causal influence from oc cipital regions into the frontal and parietal regions, possibly reflecting bottom-up influence, were observed (p=0.035) with the faster reaction time group tria ls exhibiting reduced 116

PAGE 117

Analysis 3: Figure 6-4A compares the time course of alpha band power over the occipital areas when attention was allocated to the contralateral visual field (attend, solid line) with when the contralateral visual field was ignored (ignore dotted line). Periods of significant differences between the two conditions (p < 0.05) are indicated in Figur e 6-4. Post-cue alpha power decreased markedly shortly after the presentation of the cue over the occipital regions with the attend condition having greater alph a power reduction compared to the ignore condition and this difference was sustained throughout the anticipatory attention period. The time courses of coherence and causality between the combined fronto-parietal areas (anterior medial frontal, medial frontal and parietal) and the occi pital visual areas are shown in Figures 6-4B-6-4D. These functional connectivity measures exhibited marked decrease post-cue for both attend and ignore conditions, suggesting a substantial weakening of the oscillatory network in the alpha band. Although a similar post-cue decrease in coherence and causal influences between the frontal and parietal re gions were observed, th ere were no significant differences between the attend a nd ignore conditions, consistent w ith the notion that both attend and ignore conditions involve effortful attenti on deployment and the front al-parietal areas are equally engaged in both conditi ons (Hopfinger et al., 2000; Co rbetta and Shulman, 2002; Bressler et al., 2008). 6.4 Discussion Anticipatory attention is a cogni tive process critical for topdown control. The maintenance of anticipatory attention following an attention di recting cue has been shown to entail sustained delay period activity in the front o-parietal network hypothesized to enable preparatory biasing of the task-relevant sensory areas (Luck et al., 1997 ; Kastner et al., 1999; Ka stner and Ungerleider, 2001; Corbetta and Shulman, 2002). Further supp ort for the tonic frontal top-down regulation comes from selective cooling in animals (Fuster, 1990) and human lesion studies where the loss 117

PAGE 118

of frontal control impaired both inhibitory and excitatory influences on early sensory and sensory association areas (Knight et al ., 1995; Knight et al., 1999; Ba rcelo et al., 2000; Yago et al., 2004). With accumulating evidence for the hypothesis, we predicted that with anticipatory visual attention, the frontal areas facilitated the excitability of the posterior visual areas contralateral to the attended location in preparation for the to-be-detected visual targ et stimulus. In this study, we evaluated whether visual cortex receives top-do wn modulation from fronta l-parietal areas during anticipatory attention using scal p EEG recordings from healthy volunteers performing a trial-bytrial cued spatial visual attention task. 6.4.1 Alpha Band Oscillation: A Neural Co rrelate of Top-Down Attentional Control It has been amply demonstrated that atten tion modulates ongoing brain activity prior to stimulus onset. In the human visual system, de pending on the hemifield of attention deployment, alpha band power is shown to be reduced in a topographically sele ctive manner over the extrastriate areas. In th is study we consider whether visual alpha modulation is subjected to the control of higher-order executive structures. Four results were found. First, alpha band inter-areal synchrony and Granger causality be tween frontal-parietal areas and occipital visual areas were decreased by attention following an attention-dir ecting cue. Second, the decrease was sustained throughout the anticipatory attention period. Third, the degree of fronto-parietal-occipital alpha band interaction is correlated w ith behavioral performance. Fourth, regional visual cortical activity and inter-areal interaction in other frequenc y bands (theta (4-8 Hz), beta (13-25 Hz), and gamma (25 to 40 Hz)) did not show consistent differences between attend and ignore conditions and were not systematically correlated with beha vior across participants. Taken together, it is suggestive that the alpha oscillat ory activity is likely a signal that mediates top-down attentional influences over the visual areas. Recent evid ence of the causal role of the frontal-parietal network in the attentional modulation of the visu al alpha oscillation (Capotosto et al., 2009) and 118

PAGE 119

visual ERP (Taylor et al., 2007) using rTMS in humans further supports this notion, where the application of rTMS on FEF and IPS during the an ticipatory attention peri od aimed at disrupting their top-down influence over earl y visual areas was characterize d by the lack of the customary alpha power reduction and was asso ciated with deteriorated atte ntional performance and target detection (Capotosto et al., 2009). 6.4.2 Top-Down Control of Visual Alpha: Excitation or Disinhibition? Attention-induced alpha power reduction occurred approximate ly 200 ms after the cue and this state of suppressed alpha activity was sust ained throughout the antic ipatory period between cue and imperative stimulus. Reduced alpha powe r led to enhanced visual evoked response (not shown) and better behavioral performance (Figure 6-3). This is consistent with the notion that reduced alpha activity indexes increased excitabili ty of the task-relevant visual sensory areas (Foxe et al., 1998; Bastiaansen and Brunia, 2001; Marrufo et al., 2001; Goldman et al., 2002; Laufs et al., 2003; Sauseng et al ., 2005; Thut et al., 2006; Klimes ch et al., 2007; Romei et al., 2008b; Romei et al., 2008a) and that a state of increased alpha activity potentially reflects functional inhibition (Worden et al., 2000; Kelly et al., 2006; Klimes ch et al., 2007; Rihs et al., 2007). In addition to regional modulation in osci llatory activity, interareal synchronization between the regions involved in attentional orienting, maintena nce and processing is modulated in a topographic-specific manner. Specifically, we observed a strong redu ction in interareal synchronization in the alpha frequency band be tween the higher-order areas represented by frontal and parietal areas and early visual extrastriate regions contralateral to the attended visual field relative to both the precue baseline level and when the contralateral visual field was ignored for the same network. This reduction suggests a counterintui tive dissolution of the frontal-parietal-occipital networ k with deployment of attention. However given the evidence of 119

PAGE 120

alpha band oscillation reflecting inhibitory control (Klimesch et al., 2007) and evidence from neuroimaging and lesion studies (K night et al., 1995; Knight et al., 1999; Barcelo et al., 2000; Pliszka et al., 2000; Sieroff et al., 2004; Yago et al., 2004; Gazzaley et al ., 2005; Murias et al., 2007) of tonic frontal inhibitory influence over early sensory areas either through cortico-cortical projections (Jones et al., 2000; Gonchar and Burkhalter, 2003) or through cortico-thalamocortical projections (Yingling and Skinner, 1975; Skinner and Yingli ng, 1976; Yingling and Skinner, 1976; LaBerge, 1997, 2005), it is likely th at the observed synchronized network in the alpha band prior to attention deployment reflects tonic frontal-parietal inhi bition of extrastriate areas. The subsequent attention re lated reduction may reflect a rel ease from tonic inhibition, to facilitate an optimal excitability state ove r the sensory cortex, favoring a disinhibition mechanism. Coherence being a symmetric measure does not offer distinct information on the contributions of the different directions of neur al interaction between th e regions. To dissociate top-down from bottom-up interaction and to char acterize the attentional modulation and its time course, the data was subjected to Granger causal ity analysis. Granger causality in the alpha band between the frontal-parietal areas and occipital areas in both directions were found to be reduced as early as 200 ms following the presentation of the attention directi ng cue. Comparison of magnitude of Granger causality in the alpha band for frontal and parietal driving extrastriate regions revealed a further reduc tion when the attend condition is compared to the ignore condition reflecting differences in attentional deployment and attentional load. Given the evidence of long distance projections, e.g. from hi gher-order areas to occipital lobe, tend to arise from pyramidal neurons and terminate on interneurons (Jones et al., 2000; Gonchar and Burkhalter, 2003), this reduced Gr anger causality with deployment of attention likely reflects 120

PAGE 121

reduced excitatory drive on local interneurons, suggesting a disinhi bitory role of attention in which its goal-oriented deployment leads to the re lease of the visual cortex from tonic top-down inhibitory influence (Knight et al., 1999). EEG predominantly measures the activity due to pyramidal cells. Prior to the attention directing cue, frontal/parietal mediated tonic excitatory drive to the visual cortical interneurons would result in inhibi tory postsynaptic potential (IPSP) in the pyramidal cells over the visual areas sensed by scalp EEG. The temporal precedence of activity between the fronto-parietal areas and the visual areas would manifest as strong pre-cue frontal driving posterior Grange r causality characterizing the t onic inhibitory influence. The putative reduction in excitatory drive to the visual interneurons with depl oyment of goal directed attention would result in the release of tonic inhibition over the visual pyramidal cells and thus increase in excitability. This reduction in th e tonic drive upon deploym ent of attention would account for the decreased fronto-parietal driv ing posterior Granger causality post-cue. Further insights on the role of alpha band synchronization comes from low frequency rTMS (~1 Hz) studies known to i nduce inhibitory effect s (Hilgetag et al., 2001; Brignani et al., 2008) where it has been shown to induce long-lasting increases in alpha band power and corticocortical alpha band coherence with associated decreases in evoked poten tial (Strens et al., 2002; Brignani et al., 2008). Gathering from evidences across modalitie s it emerges that a state of reduced excitability or functional inhibition is accompanied by increased alpha band activity and enhanced interareal coherence in the alpha frequency band. The reduction in the alpha band power, frontal-parietal-occipital c oherence and reduced frontal-parie tal driving extrastriate with deployment of attention strongly suggests an attentional control through a disinhibition mechanism, which likely subserves prestimulus biasi ng and maintenance of atte ntional state. It is however worth noting that long-ra nge alpha band desynchronizati on has also been explained in 121

PAGE 122

terms of an enhanced direct or indirect excitatory f acilitation of local neural population, which due to reciprocal interaction between local pyramidal cells and interneurons and complex interplay of h and T ionic currents resulted in varying levels of jitter in the firing time of cells leading to asynchrony and hence a reduction in local and interareal synchronization in the alpha band (Jones et al., 2000). Unlike recent animal (Gregoriou et al., 2009) and human attention studies (Siegel et al., 2008), th e differences in higher frequenc ies between conditions were not statistically significant across subjects. The inabi lity of scalp EEG to clearly resolve gamma band activity (25-90 Hz) may be a contributing factor It should also be pointed out that, while supporting a role of frontal-par ietal areas in exerting top-dow n modulatory influence on the excitability of visual cortex, our data do not have the requisite spatial resolution to resolve whether this influence is anatomically mediat ed by the cortico-cortic al or cortico-thalamocortical forms of signal transmission (Watson et al., 1981; LaBerge, 1997; Heilman et al., 2000; LaBerge, 2005; Zikopoulos and Barbas, 2007). Additionally, bottom-up influence of the early visual areas on parietal and frontal areas showed a drastic reduction following attention di recting cue and further showed a significant reduction with attention to contralateral visual field relative to when it was ignored specifically in the alpha band (Figures 6-3 and 6-4D). Given that feedforward projec tions from lower-order areas to higher-order areas are predominantly ex citatory and that they terminate on excitatory cells, it is surmised that this dissolution of causa l influence from the early visual areas reflect a weakening of the bottom-up excitatory drive suppor ting a reduction in the information transfer in the bottom-up direction during an ticipatory attention (Bressler et al., 2008 ) serving to reduce interference due to distracters. Alternatively if the alpha ba nd synchrony emerges from feedforward inhibition, this reducti on in alpha band coupling simila r to the top-down drive, may 122

PAGE 123

reflect a opening up of a more efficient bottom-up communication channel enabling a more effective sensory-perceptual and higher-order processing of the subsequent stimuli in the attended domain. 6.4.3 Top-Down Control of Visual Alpha: Transient or Sustained? Several areas including prefrontal cortex (PFC), frontal eye fiel d (FEF), intraparietal sulcus (IPS) and superior parietal sulcus (SPL) have be en implicated in the top-down control of early visual areas in covert spatial visual attention (Gitelman et al., 1999; Hopfinger et al., 2000; Corbetta and Shulman, 2002). However, the question of whether the top-down biasing originating from these higher-order areas to extr astriate regions during sustained attention is transient or sustained remains unc lear. It has been hypothesized th at areas such as FEF and SPL which are involved in shifting of attentional focus, may invoke a transient top-down control (Luks et al., 2008) but see (Gregor iou et al., 2009) where as re gions such the anterior PFC, dorsolateral PFC and IPS implicated in maintain ing anticipatory attentio n to spatial locations may be continuously engaged thro ughout the anticipatory attention period (Kastner et al., 1999; Hopfinger et al., 2000; Corbetta and Shulman, 2002; Luks et al., 2008) Although, our data do not have the requisite spatial resolution to reso lve the anatomical structure mediating the topdown influence, it is likely that the frontal s calp electrodes sample activity from the anterior PFC, FEF and the anterior cingulate regions a nd the parietal scalp electrodes sample activity from the SPL and IPS among other neighboring structures. Our data suggests that upon initiation of attention orienting activity, bo th the frontal, parietal and the pos terior visual regions appear to maintain sustained levels of activity throughout the cue-imperative interval indexing a continued engagement of the frontal-parieta l-occipital regions in the mainte nance of the attentional state and a tonic preparatory biasing of the relevant visual sensor y areas. Further, any momentary lapse within this frontal-parietal-occipital an ticipatory attention network characterized by 123

PAGE 124

increased magnitude of visual al pha power and interareal intera ction resulted in deteriorated behavioral performance evidenced by response e rrors and slowed respons e. This weakening of top-down biasing may either be actively controlled in a goal-directed manner as is the case during the ignored condition or may be due to a momentary lapse in attention resulting in deficient visual processing of the subse quent stimulus and hence in errors. Table 6-1. Attention modulati on of oscillatory activity. p va lues of Wilcoxon rank-sum test result (two-tailed) for difference in magn itude of Granger causality between attend and ignore conditions over theta, alpha, beta and gamma frequency bands during the time period 300 to 0 ms immediately precedi ng the onset of the standard stimulus (Analysis 1). Granger Causality Region of interest Theta (4-8 Hz) Alpha (8-12 Hz) Beta (13-25 Hz) Gamma (25-40 Hz) Top-down Granger Causality FrontalOccipital 0.71 0.036* 0.71 0.71 ParietalOccipital 0.29 0.06 0.12 0.29 Bottom-up Granger Causality OccipitalFrontal 0.13 0.79 0.45 0.45 OccipitalParietal 0.49 1.00 0.30 0.49 124

PAGE 125

Figure 6-1. Experimental task. At the beginning of each trial, a cue (left/right pointing arrow) was presented at the fixation point instructi ng the subjects to c overtly attend one of two predefined locations in the upper left and upper right visual fields. After 1800 to 2200 ms of delay, a unil ateral imperative stimulus of e ither target or standard type appeared at one of the two locations, and the subjects indicated the occurrence of the target stimulus at the attended loca tion by a speeded keypress response. 125

PAGE 126

Figure 6-2. Top-down attentiona l control of visual areas. Granger causality spectra for (A) (FrontalOccipital) and (B) (ParietalOccipital) from a representative subject are shown for the attend (black) and the i gnore (gray) conditions. (C) Normalized Granger causality in the alpha frequency band for (FrontalOccipital) and (ParietalOccipital) are averaged across subj ects and compared between attend (black) and ignore (white) conditions. The e rror bars are the sta ndard error of the mean. (D) Regions of interest marked by shaded colors correspond to frontal (Fpz, Fz), parietal (Pz) and occipi tal cortices. denotes stat istically significant difference (p<0.05). 126

PAGE 127

Figure 6-3. Pre-target alpha band activity predicts performance. The figure summarizes the regional alpha band oscillatory synchrony ove r the Occipital electrodes, averaged coherence between (Frontal-Occipital) a nd (Parietal-Occipital) areas, Granger causality measurements of (FrontalOccipital) indexing top-down influence and average of (OccipitalFrontal) and (OccipitalParietal) indexing the bottom-up interaction during the 300 ms period immediately prior to the presentation of the validly cued targets for two classes of tr ials: correct and quic ker responded trials (black) and incorrect and slower responded trials (white). The figure highlights the notion that a state of increased regional and network-wide alpha band oscillatory synchrony and interaction refl ects a weakening or disrupti on of top-down biasing and hence impaired performance. denotes st atistically significant difference (p<0.05). 127

PAGE 128

128 Figure 6-4. Temporal evolu tion of regional and networklevel alpha band activity. (A) Normalized spectral power in the alpha fr equency band over late ral Occipital areas are averaged across subjects and compared between attend (so lid black) and ignore (dotted lines) conditions. Normalized coherence (B), and Granger causality indexing top-down (C) and bottom-up (D) influences in the alpha frequency band collapsed across (Frontal-Occipital) a nd (Parietal-Occipital) are averaged across subjects and compared between attend (solid black) and ignore (dotted lines) conditions. The shaded time periods denote periods that exhibit statistically significant differences between conditions.

PAGE 129

CHAPTER 7 CONCLUSION The dissertation investigated the functional role of ongoing oscillations in stimulus processing and higher-order cognitiv e processes such as anticipato ry attention and encoding of memory along four cogent studies. The evidences garnered from these studies debunk the long held notion that ongoing neural act ivity preceding stimulus onset is noise and irrelevant to subsequent stimulus processing. It is also amply shown that the pr estimulus activity is not merely a random fluctuation in excitability but one that is actively controlled in a goal-directed manner through neural synchronization ac ross widespread functionally re levant cortical networks. In addition, a biophysically inspired framework to assess and characte rize the dependence of sensory processing on the excitability state of the brain indexed by the prestimulus ongoing brain activity was proposed and tested. In the first study, we sought to identify th e factors that lead to near zero-lags in synchronous networks and resolve the ambiguity of such observations in unidirectional networks. We decomposed neural synchrony into components quantifying directed interaction and instantaneous correlation be tween the regions and further, quantified their influence on phase-lag between the areas. From our findings, we confirmed our hypothesis that two factors: (a) instantaneous causality, which may be due to common input, such as cortical or subcortical sources which may or may not be sampled in the recordings, and / or (b) reciprocal interaction pattern between the regions, account for the coherent activity with zero-lag across long distances. In cases where the interaction is clearly unidirectional determined by Granger causality analysis, the phenomenon of zero-lag was completely expl ained by instantaneous causality. For cases wherein the interaction is bidirectional, the re ciprocal nature of the network in addition to instantaneous causality may cause near zero time lags. Hence we show that zero-lag synchrony is 129

PAGE 130

an emergent phenomenon dependent on the topo logy of the interacting network of neural ensembles and is not representative of the c onduction delay between the interacting areas as often misconstrued. In a following study examining the role of pr estimulus ongoing oscillatory brain activity in encoding of memory we provided the first evid ence in humans demonstrating that increased inter-areal synchrony in the frontotemporal network prior to item presentation is associated with higher likelihood of the item being recalled in th e subsequent test. The changes in inter-areal synchrony, taken as an indicator of changes in network excitability may be a result of spontaneous fluctuations in th e level of anticipatory attenti on and hence waxing and waning of top-down influence. Possibly as a consequence of top-down influence the level of regional theta activity indexing depolarizing influence on NM DA receptor-containing cells and hence tissue excitability over the fronto-temporal structures prior to the presentation of the item was positively correlated with the efficacy of subsequent memory encoding and later recall. Based on rodent studies on theta oscillations and memory, we propose that the te mporal coupling of the sensory input triggered glutamat e release and theta-induced depol arization activates the theta facilitated NMDA channels leading to stronger calc ium influx, which in turn facilitates LTP and thus leads to more effectiv e encoding-related processes. As part of the third study, we provided a thorough characterization of the relationship between the prestimulus ongoing oscillatory act ivity specifically th e alpha band activity (8-12 Hz) and the subsequent early visual-evok ed response. A mechanism linking the dynamics of prestimulus ongoing oscillatory activity and the generation of early evoked response was proposed by invoking the sigmoidal relationship between the endogenous synaptic fluctuations and the firing rate of the neural ensemble, a nd that external sensory input has an additive 130

PAGE 131

synaptic effect. It follows that the stimulus-evoked response corresponds to the derivative of the sigmoidal function, which is an inverted-U function of the prestimulus level of background activity. Reformulated in terms of EEG variab les where field oscill ations and the early component of event-related poten tial are treated as indices of background synaptic activity and stimulus-evoked response, an inverted-U relationship betwee n early evoked response and prestimulus field oscillations was predicted a nd confirmed empirically through noninvasive EEG studies of human visual processing. Attention has been thought to se lectively enhance the perceptual salience of attended stimuli through competing mechanis ms of sensory gain control. Based on the mechanism linking pre and poststimu lus activity and using computational modeling and empirical evidence from EEG studies of visu al attention we show that attention through active control of (a) mean membrane depolariz ation (excitability) and (b) the variance of background synaptic level (membrane potential fluc tuation) over the early sensory areas enable preferential facilitation of processing in the attended domain over the unattended domain. Although the previous aim addressed the attent ion modulation of exc itability of the early visual areas and hence the subs equent visual processing, su ch goal-directed prestimulus modulation is not an inherent property of the ear ly sensory areas. Further, the neural mechanisms by which this is implemented are still unclear. As part of the fourth and final study, we assessed how attention modulates endogenous anticipator y processes employing time-frequency spectral analysis. Applying coherence analysis to the anticipatory attention period we identified a synchronous frontal parietal-occip ital network in the low freque ncy range (alpha band), where Granger causality analysis further defined the direction of causal inte raction. The degree of inter-areal synchrony and the magnitude of Granger causality were found to be decreased by attention. In light of prior lesion and EEG/MEG studies and anat omical evidences this reduced 131

PAGE 132

fronto-parietal driving occipital Granger causality is seen to lend support to the disinhibition mechanism, where attention serves to increase excitability through reduc ed excitatory drive on local interneurons, suggesting a disinhibition f unction of attention in which its goal-oriented deployment leads to the release of the visual cort ex from tonic top-down inhibitory influence. It is evident from our studies and from othe r previous studies that all cognitive function depends on the coordinated interaction among neur al assemblies distributed across the brain. Although we have exclusively focused on neural oscillations and synchrony to address the mechanisms of executive processes and other cogn itive processes in healthy human volunteers, it is equally relevant to address pathological brain states in di seases including Schizophrenia, Alzheimers disease, Parkinsons disease and other developmental diseases causing neurocognitive impairments such as autism and ADHD among others. A common thread among all these diseases is the deficits in coordinate d processing across distribu ted cortical networks leading to specific cognitive deficits especially in executive functioning, attention and memory. Measures of inter-areal connectivity such as coherence and Granger causality in addition to anatomical metrics quantifying white matter and t hus neuroanatomical substrates of inter-areal connectivity may enable objective measures of functional integrity of pathways between different areas of the brain a nd thus may serve as biomarkers offering diagnostic potential. Further the findings from our studi es regarding the critical role of spontaneous brain state on subsequent stimulus processing and behavi oral performance have implications in neurodiagnostic and neurofeedback based interv ention capability to detect and treat cognitive decline and dementia caused by aging and possibl y ameliorate psychiatric conditions afflicting the emotion circuits by employing br ain-state contingent training wh ich may effect a facilitatory pattern of brain activity and hence behavioral improvement. Thus, synchronous oscillations 132

PAGE 133

133 quantifying regional and large-scale integration of neural activity as assessed from noninvasive and invasive electrophysiological re cordings may be of critical value not only in elucidating mechanism related to normal cognitive functi oning but also to address cognitive impairments brought about by multitude of debilitating ne urological and psychiatric disorders.

PAGE 134

REFERENCES Abbott LF, Varela JA, Sen K, Nelson SB (1997) S ynaptic depression and co rtical gain control. Science 275:220-224. Anderson K, Rajagovindan R, Ghacibeh G, Mea dor KJ, Ding M (in press) Theta oscillations mediate interaction between prefrontal cortex and medial temporal l obe in human memory. Cerebral Cortex. Araya R, Eisenthal KB, Yuste R (2006) Dendritic spines linearize the summation of excitatory potentials. Proceedings of the National Academy of Sciences 103:18799-18804. Arieli A, Sterkin A, Grinvald A, Aertsen A (1996) Dynamics of ongoing activity: Explanation of the large variability in evoked cort ical responses. Science 273:1868-1871. Armstrong KM, Fitzgerald JK, Moore T (2006) Ch anges in visual receptive fields with microstimulation of frontal cortex. Neuron 50:791-798. Averbeck BB, Lee D (2004) Coding and transmission of information by neural ensembles. Trends in Neurosciences 27:225-230. Babiloni C, Vecchio F, Bultrini A, Romani GL, Rossini PM (2006) Preand poststimulus alpha rhythms are related to consci ous visual perception: A high-resolution EEG study. Cerebral Cortex 16:1690-1700. Baccala LA, Sameshima K (2001) Partial directed c oherence: A new concept in neural structure determination. Biologica l Cybernetics 84:463-474. Barbas H (2000) Connections underlying the synthesis of cognition, memory, and emotion in primate prefrontal cortices. Brai n Research Bulletin 52:319-330. Barcelo F, Suwazono S, Knight RT (2000) Pref rontal modulation of visual processing in humans. Nature Neuroscience 3:399-403. Bastiaansen MCM, Brunia CHM (2001) Antic ipatory attention: an event-related desynchronization approach. International Journal of Psychophysiology 43:91-107. Bechara A, Damasio H, Tranel D, Anderson SW (1998) Dissociation of working memory from decision making within the human prefrontal cortex. Journal of Neuroscience 18:428-437. Berry SD, Seager MA (2001) Hippocampal thet a oscillations and cl assical conditioning. Neurobiology of Learning and Memory 76:298-313. Bestmann S, Ruff CC, Blakemore C, Driver J, Thilo KV (2007) Spatial attention changes excitability of human visual cortex to direct stimulat ion. Current Biology 17:134-139. Bokil H, Purpura K, Schoffelen JM, Thomson D, Mitra P (2007) Comparing spectra and coherences for groups of unequal size. J ournal of Neuroscience Methods 159:337-345. 134

PAGE 135

Bollimunta A, Chen Y, Schroeder CE, Ding M (2008) Neuronal mechanisms of cortical alpha oscillations in awake-behaving macaque s. Journal of Neuroscience 28:9976-9988. Brandt ME, Jansen BH (1991) Th e relationship between prestimulus alpha-amplitude and visual evoked-potential amplitude. Internationa l Journal of Neuroscience 61:261-268. Brandt ME, Jansen BH, Carbonari JP (1991) Pre-stimulus spectra l EEG patterns and the visual evoked-response. Electroencephalography and Clinical Neurophysiology 80:16-20. Bressler SL, Coppola R, Nakamura R (1993) Ep isodic multiregional cortical coherence at multiple frequencies during visual task-performance. Nature 366:153-156. Bressler SL, Tang W, Sylveste r CM, Shulman GL, Corbetta M (2008) Top-down control of human visual cortex by frontal and parietal cortex in anticipato ry visual spatial attention. Journal of Neuroscience 28:10056-10061. Brignani D, Manganotti P, Rossini PM, Miniussi C (2008) Modulat ion of cortical oscillatory activity during transcranial magnetic stimulation. Hu man Brain Mapping 29:603-612. Brovelli A, Ding MZ, Ledberg A, Chen YH, Nakamu ra R, Bressler SL (2004) Beta oscillations in a large-scale sensorimotor cortical network: Directional influences revealed by Granger causality. Proceedings of the National Academ y of Sciences of the United States of America 101:9849-9854. Burgess AP, Gruzelier JH (1997) Short duration synchronization of huma n theta rhythm during recognition memory. Neuroreport 8:1039-1042. Buschman TJ, Miller EK (2007) Top-down vers us bottom-up control of attention in the prefrontal and posteri or parietal cortices Science 315:1860-1862. Buzsaki G (1989) 2-Stage model of memory trac e formation a role for noisy brain states. Neuroscience 31:551-570. Buzsaki G (1996) The hippocampo-neocortica l dialogue. Cerebral Cortex 6:81-92. Buzsaki G (2002) Theta oscillations in the hippocampus. Neuron 33:325-340. Buzsaki G (2006) Rhythms of the Brain. New York: Oxford University Press. Cadotte AJ, DeMarse TB, He P, Ding M (2008) Caus al measures of structure and plasticity in simulated and living neural networks. PLoS ONE 3:1-14. Cadotte AJ, Mareci TH, DeMarse TB, Parekh MB, Rajagovindan R, Ditto WL, Talathi SS, Hwang D-U, Carney PR (2009) Temporal l obe epilepsy: Anatomical and effective connectivity. IEEE Transactions in Neural Sy stem and Rehabilitation Engineering 17:214223. 135

PAGE 136

Callaway EM (1998) Local circuits in primary visual cortex of the macaque monkey. Annual Review of Neuroscience 21:47-74. Capotosto P, Babiloni C, Romani GL, Corbetta M (2009) Frontopari etal cortex controls spatial attention through modulation of anticipatory alpha rhythms. Journal of Neuroscience 29:5863-5872. Cash S, Yuste R (1999) Linear summation of excitatory inputs by CA1 pyramidal neurons. Neuron 22:383-394. Cassidy M, Brown P (2003) Spectral phase estimates in the setting of multidirectional coupling. Journal of Neuroscien ce Methods 127:95-103. Chance FS, Abbott LF, Reyes AD (2002) Gain m odulation from background synaptic input. Neuron 35:773-782. Chawla D, Friston KJ, Lumer ED (2001) Zero -lag synchronous dynamics in triplets of interconnected cortical areas Neural Networks 14:727-735. Chelazzi L, Miller EK, Duncan J, Desimone R (1993) A neural basis for visual-search in inferior temporal cortex. Nature 363:345-347. Chen YH, Bressler SL, Ding MZ (2006) Frequency decomposition of conditional Granger causality and application to multivariate neural field potential data. Journal of Neuroscience Methods 150:228-237. Chung S, Li XR, Nelson SB (2002) Short-term depression at thalam ocortical synapses contributes to rapid adaptati on of cortical sensory respons es in vivo. Neuron 34:437-446. Clark VP, Hillyard SA (1996) Spatia l selective attention affects early extrastriate but not striate components of the visual evoked potential. Journal of Cognitive Neuroscience 8:387-402. Cohen NJ, Ryan E, Hunt C, Romine L, Wszalek T, Nash C (1999) Hippocampal system and declarative (relational) memory: Summarizi ng the data from functional neuroimaging studies. Hippocampus 9:83-98. Corbetta M, Shulman GL (2002) Control of goal-directed and stimulus-driven attention in the brain. Nature Review s Neuroscience 3:201-215. Dasilva FL (1991) Neural mechanisms underlying brain waves from neural membranes to networks. Electroencephalography and Clinical Neurophysiology 79:81-93. Dehaene S, Changeux J-P (2005) Ongoing spontane ous activity controls access to consciousness: A neuronal model for inattentiona l blindness. PLoS Biology 3:910-927. Desimone R (1996) Neural mechanisms for visu al memory and their role in attention. Proceedings of the National Academy of Sciences:13494-13499. 136

PAGE 137

Destexhe A, Rudolph M, Fellous JM, Sejnowski TJ (2001) Fluctuating sy naptic conductances recreate in vivo-like activity in neoc ortical neurons. Neuroscience 107:13-24. Di Russo F, Martinez A, Hillyard SA (2003) Sour ce analysis of event-related cortical activity during visuo-spatial attention. Cerebral Cortex 13:486-499. Di Russo F, Martinez A, Sereno MI, Pitzalis S, Hillyard SA (2002) Cortical sources of the early components of the visual evoked potential. Human Brain Mapping 15:95-111. Ding M, Chen Y, Bressler SL (2006) Granger causality: Basic theory and application to neuroscience. In: Handbook of Time Series An alysis (Winterhalder M, Schelter B and Timmer J, eds), pp 451-474. Berlin: Wiley-VCH Verlag. Ding MZ, Bressler SL, Yang WM, Liang HL (2000) Short-window sp ectral analysis of cortical event-related potentials by adaptive multivariate autoregressive modeling: Data preprocessing, model validation, and variability assessment. Biological Cybernetics 83:3545. Doesburg SM, Roggeveen AB, Kitajo K, Ward LM (2008) Large-scale gamma-band phase synchronization and selective atten tion. Cerebral Cortex 18:386-396. Dringenberg HC (2000) Alzheimer' s disease: More than a 'cholin ergic disorder' evidence that cholinergic-monoaminergic interactions c ontribute to EEG slowing and dementia. Behavioural Brain Research 115:235-249. Eckhorn R, Bauer R, Jordan W, Brosch M, Kr use W, Munk M, Reitboeck HJ (1988) Coherent oscillations: A mechanism of feature linking in the visual cortex? Multiple electrode and correlation analyses in the cat. Bi ological Cybern etics 60:121-130. Efron B (1985) Bootstrap confidence-intervals for a class of parametric problems. Biometrika 72:45-58. Efron B (2000) The bootstrap and m odern statistics. Journal of Am erican Statistical Association 95:1293-1296. Efron B, Gong G (1983) A leisurely look at the bootstrap, the jac knife, and cross-validation. The American Statistician 37:36-48. Eichele T, Debener S, Calhoun VD, Specht K, Engel AK, Hugdahl K, von Cramon DY, Ullsperger M (2008) Prediction of human errors by maladaptive cha nges in event-related brain networks. Proceedings of the National A cademy of Sciences of the United States of America 105:6173-6178. Engel AK, Fries P, Singer W (2001) Dynamic pred ictions: Oscillations and synchrony in topdown processing. Nature Re views Neuroscience 2:704-716. 137

PAGE 138

Ergenoglu T, Demiralp T, Bayraktaroglu Z, Erge n M, Beydagi H, Uresin Y (2004) Alpha rhythm of the EEG modulates visual detection perf ormance in humans. Cognitive Brain Research 20:376-383. Felleman DJ, Van Essen DC (1991) Distributed hierarchical processing in the primate cerebral cortex. Cerebral Cortex 1:1-47. Fernandez G, Effern A, Grunwald T, Pezer N, Lehnertz K, Dumpelmann M, Van Roost D, Elger CE (1999) Real-time tracking of memory fo rmation in the human rhinal cortex and hippocampus. Science 285:1582-1585. Ferree T (2006) Spherical splines and average referencing in scalp electroencephalography. Brain Topography 19:43-52. Fischer T, Langner R, Birbaumer N, Brocke B (2008) Arousal and attention: Self-chosen stimulation optimizes cortical excitability a nd minimizes compensatory effort. Journal of Cognitive Neuroscience 20:1443-1453. Foxe JJ, Simpson GV, Ahlfors SP (1998) Parietooccipital similar to 10 Hz activity reflects anticipatory state of visu al attention mechanisms. Neuroreport 9:3929-3933. Freeman WJ (1979) Non-linear gain mediating cortical stimulus-response relations. Biological Cybernetics 33:237-247. Friedman D, Johnson R (2000) Event-related pote ntial (ERP) studies of memory encoding and retrieval: A selective review. Micros copy Research and Technique 51:6-28. Fries P (2005) A mechanism for cognitive dynamics: Neuronal communication through neuronal coherence. Trends in Cognitive Sciences 9:474-480. Fries P, Reynolds JH, Rorie AE, Desimone R (2001) Modulation of oscillatory neuronal synchronization by selective visual attention. Science 291:1560-1563. Furey ML, Pietrini P, Haxby JV (2000) Cholinergic enhancement and increased selectivity of perceptual processing during work ing memory. Science 290:2315-2319. Fuster JM (1990) Behavioral elec trophysiology of the prefrontal co rtex of the primate. Progress in Brain Research 85:313-324. Fuster JM, Bauer RH, Jervey JP (1985) Functio nal interactions between inferotemporal and prefrontal cortex in a cognitive task. Brain Research 330:299-307. Gallinat J, Kunz D, Senkowski D, Kienast T, Seifert F, Schubert F, Heinz A (2006) Hippocampal glutamate concentration predicts ce rebral theta oscillat ions during cognitive processing. Psychopha rmacology 187:103-111. Garrido MI, Kilner JM, Kiebel SJ, Friston KJ (2007) Evoked br ain responses are generated by feedback loops. Proceedings of the Na tional Academy of Sciences 104:20961-20966. 138

PAGE 139

Gazzaley A, Rissman J, D'Esposito M (2004) Fu nctional connectivity during working memory maintenance. Cognitive Affective & Behavioral Neuroscience 4:580-599. Gazzaley A, Cooney JW, McEvoy K, Knight RT D'Esposito M (2005) Top-down enhancement and suppression of the magnitude and speed of neural activity. Journal of Cognitive Neuroscience 17:507-517. Gazzaley A, Rissman J, Cooney J, Rutman A, Seibert T, Clapp W, D'Esposito M (2007) Functional interactions between prefrontal and visual association cort ex contribute to topdown modulation of visual proce ssing. Cerebral Cortex:I125-I135. Gelfand IM, Yaglom AM (1959) Calculation of the amount of information about a random function contained in another such function. American Mathematical Society Translations Series 2:199-246. Gershberg FB, Shimamura AP (1995) Impaired us e of organizational strategies in free-recall following frontal lope damage Neuropsychologia 33:1305-1333. Gevins A, Smith ME, McEvoy L, Yu D (1997) High-resolution EEG mapping of cortical activation related to working memory: Effects of task difficulty, type of processing, and practice. Cerebral Cortex 7:374-385. Geweke J (1982) Measurement of linear dependence and feedback between multiple time series. Journal of American Statistical Association 77:304-314. Giesbrecht B, Woldorff MG, Song AW, Mangun GR (2003) Neural mechanisms of top-down control during spatial and feature attention. Neuroimage 19:496-512. Giesbrecht B, Weissman DH, Wo ldorff MG, Manqun GR (2006) Pretarget activity in visual cortex predicts behavioral performance on spatial and feature attention tasks. Brain Research 1080:63-72. Gilbert CD, Wiesel TN (1983) Functional-organiza tion of the visual-cortex. Progress in Brain Research 58:209-218. Gitelman DR, Nobre AC, Parrish TB, LaBar KS, Kim YH, Meyer JR, Mesulam MM (1999) A large-scale distributed network for covert spatial attention further anatomical delineation based on stringent behavioural and cognitive controls. Brain 122:1093-1106. Givens BS, Olton DS (1990) Cholin ergic and gabaergic modulation of medial septal area effect on working memory. Behavior al Neuroscience 104:849-855. Goldman RI, Stern JM, Engel J, Cohen MS (20 02) Simultaneous EEG and fMRI of the alpha rhythm. Neuroreport 13:2487-2492. Gonchar Y, Burkhalter A (2003) Distinct GABAergic targets of feedforward and feedback connections between lower and higher areas of rat visual cortex. Journal of Neuroscience 23:10904-10912. 139

PAGE 140

Grady CL, McIntosh AR, Craik FIM (2003) Ag e-related differences in the functional connectivity of the hippocampus during memory encoding. Hippocampus 13:572-586. Granger CWJ (1969) Investigating causal relati ons by econometric models and cross-spectral methods. Econometrica 37:414-438. Gray CM, Konig P, Engel AK, Singer W (1989) Oscillatory responses in cat visual-cortex exhibit inter-columnar synchronization which reflects global stimulus properties. Nature 338:334-337. Green JJ, McDonald JJ (2008) Electrical neuroi maging reveals timing of attentional control activity in human brain. PLoS Biology 6:730-738. Gregoriou GG, Gotts SJ, Zhou H, Desimone R (2009) High-frequency, long-range coupling between prefrontal and visual cortex during attention. Science 324:1207-1210. Grent-'T-Jong T, Woldorff MG ( 2007) Timing and sequence of brain activity in top-down control of visual-spatial attention. PLoS Biology 5:114-126. Grunwald T, Beck H, Lehnertz K, Blumcke I, Pezer N, Kurthen M, Fernandez G, Van Roost D, Heinze HJ, Kutas M, Elger CE (1999) Evidence relating human verbal memory to hippocampal N-methyl-D-aspartate receptors. Proceedings of the National Academy of Sciences of the United Stat es of America 96:12085-12089. Guderian S, Schott BH, Richardson-Klavehn A, Duzel E (2009) Medial temporal theta state before an event predicts episodic encoding su ccess in humans. Proceedings of the National Academy of Sciences of the Unite d States of America 106:5365-5370. Guo S, Wu J, Ding M, Feng J (2008) Uncovering interactions in the frequency domain. PLoS Computational Biology 4:1-10. Haider B, Duque A, Hasenstaub AR, Yu YG, McCormick DA (2007) Enhancement of visual responsiveness by spontaneous local network ac tivity in vivo. Jour nal of Neurophysiology 97:4186-4202. Hanslmayr S, Aslan A, Staudigl T, Klimesch W, Herrmann CS, Bauml KH (2007a) Prestimulus oscillations predict between and with in subjects. Neuroimage 37:1465-1473. Hanslmayr S, Klimesch W, Sauseng P, Gruber W, Doppelmayr M, Freunberger R, Pecherstorfer T, Birbaumer N (2007b) Alpha phase reset contri butes to the generati on of ERPs. Cerebral Cortex 17:1-8. Hasselmo ME (2005) What is the function of hippocampal theta Rhythm? Linking behavioral data to phasic properties of field potential and unit recording data. Hippocampus 15:936949. 140

PAGE 141

Hasselmo ME, McGaughy J (2004) High acetylcholine le vels set circuit dynamics for attention and encoding and low acetylcholine levels se t dynamics for conso lidation. Progress in Brain Research 145:207-231. Heilman KM, Valenstein E, Watson RT (2000) Neglect and related disorders. Seminars in Neurology 20:463-470. Heinze HJ, Mangun GR, Burchert W, Hinrichs H, Scholz M, Munte TF, Gos A, Scherg M, Johannes S, Hundeshagen H, Gazzaniga MS, Hillyard SA (1994) Combined spatial and temporal imaging of brain ac tivity during visual selective attention in humans. Nature 372:543-546. Hermer-Vazquez L (2008) Tracing 'driver' vers us 'modulator' information flow throughout largescale, task-related neural circuitry. Journal of Combin atorial Optimization 15:242-256. Herrmann CS, Munk MHJ, Engel AK (2004) Cognitive functions of gamma-band activity: Memory match and utilization. Tren ds in Cognitive Sciences 8:347-355. Hilgetag CC, Theoret H, Pascual-Leone A (2001) E nhanced visual spatial a ttention ips ilateral to rTMS-induced 'virtual lesions' of human pari etal cortex. Nature Neuroscience 4:953-957. Hillyard SA, Vogel EK, Luck SJ (1998) Sensory ga in control (amplificati on) as a mechanism of selective attention: electrophysiological and neuroimaging evidence. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 353:1257-1270. Ho N, Destexhe A (2000) Synaptic backgr ound activity enhances th e responsiveness of neocortical pyramidal neurons. J ournal of Neurophysiology 84:1488-1496. Honey GD, Honey RAE, O'Loughlin C, Sharar SR, Kumaran D, Suckling J, Menon DK, Sleator C, Bullmore ET, Fletcher PC (2005) Ke tamine disrupts frontal and hippocampal contribution to encoding and retrieval of episodic memory: An fMRI study. Cerebral Cortex 15:749-759. Hopfinger JB, Buonocore MH, Mangun GR (2000) The neural mechanisms of top-down attentional control. Natu re Neuroscience 3:284-291. Huerta PT, Lisman JE (1993) Heightened synaptic plasticity of hippocampal CA1 neurons during a cholinergically induced r hythmic state. Nature 364:723-725. Jagadeesh B, Wheat HS, Ferster D (1993) Linear ity of summation of synaptic potentials underlying direction selectivity in simple cells of the cat visual cortex. Science 262:19011904. Janowsky JS, Shimamura AP, Squire LR (1989) Source memory impairment in patients with frontal-lobe lesions. Neur opsychologia 27:1043-1056. Jean-Philippe L, Eugenio R, Jacques M, Fran cisco JV (1999) Measuring phase synchrony in brain signals. Human Br ain Mapping 8:194-208. 141

PAGE 142

Jensen O, Tesche CD (2002) Frontal theta activity in humans increases with memory load in a working memory task. European Journal of Neuroscience 15:1395-1399. Jokisch D, Jensen O (2007) Modulation of ga mma and alpha activity du ring a working memory task engaging the dorsal or ventral stream. Journal of Neuroscience 27:3244-3251. Jones MW, Wilson MA (2005) Theta rhythms coordi nate hippocampal-prefront al interactions in a spatial memory task. PLoS Biology 3:2187-2199. Jones SR, Pinto DJ, Kaper TJ, Kopell N (2000) Alpha-frequency rhythms desynchronize over long cortical distances: A modeling study. Journal of Computational Neuroscience 9:271291. Jones SR, Pritchett DL, Stufflebeam SM, Hamalain en M, Moore CI (2008) Influence of PreStimulus Somatosensory Mu-Rhythm on Evoked Response Gain: A Combined MEG And Computational Modeling Study. Society for Neuroscience Meeting Abstract 34:858.14. Kahana MJ, Sekuler R, Caplan JB, Kirschen M, Madsen JR (1999) Human theta oscillations exhibit task dependence during virtua l maze navigation. Nature 399:781-784. Kaminski M, Ding MZ, Truccolo WA, Bressler SL (2001) Evaluating causal relations in neural systems: Granger causality, directed transf er function and statis tical assessment of significance. Biological Cybernetics 85:145-157. Kapur S, Tulving E, Cabeza R, McIntosh AR, Ho ule S, Craik FIM (1996) The neural correlates of intentional learning of verbal materi als: A PET study in humans. Cognitive Brain Research 4:243-249. Kastner S, Ungerleider LG (2001) The neural basis of biased competition in human visual cortex. Neuropsychologia 39:1263-1276. Kastner S, Pinsk MA, De Weerd P, Desimone R, Ungerleider LG (1999) Increased activity in human visual cortex during dire cted attention in the absence of visual stimulation. Neuron 22:751-761. Kelly SP, Lalor EC, Reilly RB, Foxe JJ (2006) In creases in alpha oscill atory power reflect an active retinotopic mechanism for distracter suppression during sust ained visuospatial attention. Journal of Neurophysiology 95:3844-3851. Kimura F (2000) Cholinergic modulation of cortical function: A hypothetical role in shifting the dynamics in cortical network. Neuroscience Research 38:19-26. Kirchhoff BA, Wagner AD, Maril A, Stern CE (2000) Prefrontal-temporal circuitry for episodic encoding and subsequent memory. J ournal of Neuroscience 20:6173-6180. Klimesch W (1999) EEG alpha and theta oscillati ons reflect cognitive and memory performance: A review and analysis. Brai n Research Reviews 29:169-195. 142

PAGE 143

Klimesch W, Sauseng P, Hanslmayr S (2007) EEG alpha oscillations: The inhibition-timing hypothesis. Brain Resear ch Reviews 53:63-88. Klimesch W, Doppelmayr M, Russegger H, Pachi nger T (1996) Theta band power in the human scalp EEG and the encoding of new information. Neuroreport 7:1235-1240. Knight RT (1997) Distributed cortical network for visual attention. Journal of Cognitive Neuroscience 9:75-91. Knight RT, Grabowecky MF, Scabini D (1995) Role of human prefrontal cortex in attention control. In: Advances in Neurology: Epilepsy and the functio nal anatomy of the frontal lobe (Jasper HH, Riggio S, Goldman-Rakic PS, eds), pp 21-36. New York: Raven Press. Knight RT, Staines WR, Swick D, Chao LL (1999) Prefrontal cortex re gulates inhibition and excitation in distributed neural networks. Acta Psychologica 101:159-178. Kocsis B, Kaminski M (2006) Dynamic changes in the direction of the theta rhythmic drive between supramammillary nucleus and the septohippocampal system. Hippocampus 16:531-540. Kocsis B, Bragin A, Buzsaki G (1999) Interdependence of multiple theta generators in the hippocampus: a partial coherence analysis Journal of Neuroscience 19:6200-6212. LaBerge D (1997) Attention, awareness, and the triangular circuit. Consciousness and Cognition 6:149-181. LaBerge D (2005) Sustained attention and apical dendrite activity in r ecurrent circuits. Brain Research Reviews 50:86-99. Lakatos P, Chen CM, O'Connell MN, Mills A, Sc hroeder CE (2007) Neuronal oscillations and multisensory interaction in primary auditory cortex. Neuron 53:279-292. Lakatos P, Karmos G, Mehta AD, Ulbert I, Schroeder CE (2008) Entrainment of neuronal oscillations as a mechanism of atte ntional selection. Science 320:110-113. Lakatos P, Shah AS, Knuth KH, Ulbert I, Ka rmos G, Schroeder CE (2005) An oscillatory hierarchy controlling neuronal ex citability and stimulus proce ssing in the auditory cortex. Journal of Neurophysiology 94:1904-1911. Laufs H, Kleinschmidt A, Beyerle A, Eger E, Salek-Haddadi A, Preibi sch C, Krakow K (2003) EEG-correlated fMRI of human alpha activity. Neuroimage 19:1463-1476. Ledberg A, Bressler SL, Ding MZ, Coppola R, Nakamura R (2007) Large-scale visuomotor integration in the cerebral co rtex. Cerebral Cortex 17:44-62. Lfeachor EC, Jarvis BW Digital signal processing: A practical approach. Essex: Prentice Hall. 143

PAGE 144

Liang HL, Bressler SL, Ding MZ, Truccolo WA, Nakamura R (2002) Synchronized activity in prefrontal cortex during anticipation of vi suomotor processing. Neuroreport 13:2011-2015. Linkenkaer-Hansen K, Nikulin VV, Palva S, Ilmoniemi RJ, Palva JM (2004) Prestimulus oscillations enhance psychophys ical performance in humans. Journal of Neuroscience 24:10186-10190. Lisman JE, Raghavachari S, Tsien RW (2007) Th e sequence of events that underlie quantal transmission at central glutamatergic syna pses. Nature Reviews Neuroscience 8:597-609. Loh M, Pasupathy A, Miller EK, Deco G (2008) Neurodynamics of the pref rontal cortex during conditional visuomotor associations. J ournal of Cognitive Neuroscience 20:421-431. Luck SJ (2005) An introduction to the event-related potential t echnique. Cambridge: MIT Press. Luck SJ, Chelazzi L, Hillyard SA, Desimone R (1997) Neural mechanisms of spatial selective attention in areas V1, V2, and V4 of maca que visual cortex. Journal of Neurophysiology 77:24-42. Luks TL, Sun FT, Dale CL, Miller WL, Simp son GV (2008) Transient and sustained brain activity during anticipatory visuospatial attention. Neuroreport 19:155-159. Lungarella M, Sporns O (2006) Mapping information flow in se nsorimotor networks. PLoS Computational Biology 2:1301-1312. Macklin JT (1982) An investigation of the pr operties of double radio sources using Spearman partial rank correlation coefficient. Royal Astronomical Society, Monthly Notices 199:1119-1136. Makeig S, Westerfield M, Jung TP, Enghoff S, Townsend J, Courchesne E, Sejnowski TJ (2002) Dynamic brain sources of visual evoked responses. Science 295:690-694. Malsburg CVD (1995) Binding in m odels of perception and brain function. Current Opinion in Neurobiology 5:520-526. Mangun GR, Hillyard SA (1991) Modulations of sensory-evoked brain potentials indicate changes in perceptual processing during visual spatial priming. Journal of Experimental Psychology-Human Perception and Performance 17:1057-1074. Mangun GR, Buonocore MH, Girelli M, Jha AP (1998) ERP and fMRI measures of visual spatial selective a ttention. Human Brain Mapping 6:383-389. Mangun GR, Hopfinger JB, Kussmaul CL, Fletcher EM, Heinze HJ (1997) Covariations in ERP and PET measures of spatial selective attention in human extrastriate visual cortex. Human Brain Mapping 5:273-279. Marrufo MV, Vaquero E, Cardoso MJ, Gomez CM (2001) Temporal evolution of alpha and beta bands during visual spatial attention. Cognitive Brain Research 12:315-320. 144

PAGE 145

Mazaheri A, Jensen O (2006) Posterior alpha activity is not phase-reset by visual stimuli. Proceedings of the National Academy of Scie nces of the United States of America 103:2948-2952. McCormick DA, Shu YS, Hasenstaub A, SanchezVives M, Badoual M, Bal T (2003) Persistent cortical activity: Mechanisms of generation a nd effects on neuronal ex citability. Cerebral Cortex 13:1219-1231. Meador KJ (2007) The basic science of memory as it applies to epileps y. Epilepsia 48:23-25. Melloni L, Molina C, Pena M, Torres D, Si nger W, Rodriguez E ( 2007) Synchronization of neural activity across cortical areas correla tes with conscious pe rception. Journal of Neuroscience 27:2858-2865. Miltner WH, Braun C, Arnold M (1999) Coherence of gamma-band EEG activity as a basis for associative learning. Nature 397:917-924. Mitra P, Bokil H (2007) Observed brain dynamics. New York: Oxford University Press. Mitra PP, Pesaran B (1999) Analysis of dyna mic brain imaging data. Biophysical Journal 76:691-708. Mitzdorf U (1985) Current source-density method and application in cat cerebral-cortex Investigation of evoked-poten tials and EEG Phenomena. P hysiological Reviews 65:37100. Munk MHJ, Roelfsema PR, Konig P, Engel AK, Si nger W (1996) Role of reticular activation in the modulation of intracortical sy nchronization. Science 272:271-274. Murias M, Swanson JM, Srinivasan R (2007) Functi onal connectivity of front al cortex in healthy and ADHD children reflected in EEG cohere nce. Cerebral Cortex 17:1788-1799. Nichols TE, Holmes AP (2002) N onparametric permutation tests for functional neuroimaging: A primer with examples. Human Brain Mapping 15:1-25. Niedermeyer E, Lopes Da Silva F (2005) Electr oencephalography: Basic principles, clinical applications, and related fields. Philadelphia: Lippincott Williams & Wilkins. Nunez PL, Srinivasan R, Westdor p AF, Wijesinghe RS, Tucker DM Silberstein RB, Cadusch PJ (1997) EEG coherency.1: Statis tics, reference electrode, vo lume conduction, Laplacians, cortical imaging, and interpretation at multiple scales. Electroencephalography and Clinical Neurophysiology 103:499-515. Ohara S, Crone NE, Weiss N, Lenz FA (2006) Analysis of synchrony demonstrates 'pain networks' defined by rapidly switching, task -specific, functional connectivity between pain-related cortical st ructures. Pain 123:244-253. 145

PAGE 146

Ohara S, Crone NE, Weiss N, Kim JH, Lenz FA (2008) Analysis of sync hrony demonstrates that the presence of "pain networks" prior to a noxious stimulus can enab le the perception of pain in response to that stimulus. E xperimental Brain Re search 185:353-358. Okatan M, Wilson MA, Brown EN (2005) Analyz ing functional connectivity using a network likelihood model of ensemble neural spiking activity. Ne ural Computation 17:1927-1961. Otten LJ, Henson RNA, Rugg MD (2002) State-rela ted and item-related neural correlates of successful memory encoding. Nature Neuroscience 5:1339-1344. Otten LJ, Quayle AH, Akram S, Ditewig TA, Ru gg MD (2006) Brain activity before an event predicts later reco llection. Nature Ne uroscience 9:489-491. Padilla ML, Wood RA, Hale LA, Knight RT (2 006) Lapses in a prefrontal-extrastriate preparatory attention network predict mist akes. Journal of Cognitive Neuroscience 18:1477-1487. Palva JM, Palva S, Kaila K (2005) Phase sync hrony among neuronal oscillations in the human cortex. Journal of Neuroscience 25:3962-3972. Paul MD, Simon PK, John JF, Richard BR, Ian HR (2007) Optimal sustained attention is linked to the spectral content of background EEG activity: Greater ongoing tonic alpha (approximately 10 Hz) power supports successful phasic goal activation. European Journal of Neuroscience 25:900-907. Perrin F, Pernier J, Bertrand O, Echallier JF ( 1989) Spherical splines for scalp potential and current-density mapping. Electroencephalogr aphy and Clinical Neurophysiology 72:184187. Petersen CCH, Hahn TTG, Mehta M, Grinvald A, Sakmann B (2003) Interaction of sensory responses with spontaneous depolarization in layer 2/3 barrel cortex. Proceedings of the National Academy of Sciences of the United States of America 100:13638-13643. Petrides M, Milner B (1982) De ficits on subject-ordered tasks after frontal-lobe and temporallobe lesions in man. Neuropsychologia 20:249-262. Petrides M, Pandya DN (1999) Dors olateral prefrontal cortex: Comparative cytoarchitectonic analysis in the human and the macaque brai n and corticocortical connection patterns. European Journal of Neuroscience 11:1011-1036. Petrides M, Pandya DN (2002) Comparative cytoarchitectonic analysis of the human and the macaque ventrolateral prefrontal cortex a nd corticocortical conn ection patterns in the monkey. European Journal of Neuroscience 16:291-310. Pfurtscheller G, da Silva FHL (1999) Ev ent-related EEG/MEG synchronization and desynchronization: Basic principles Clinical Neurophysiology 110:1842-1857. 146

PAGE 147

Pliszka SR, Liotti M, Woldorff MG (2000) Inhibitory control in children with attentiondeficit/hyperactivity disorder: Event-related potentials iden tify the processing component and timing of an impaired right-frontal response-inhibition mechanism. Biological Psychiatry 48:238-246. Poulet JFA, Petersen CCH (2008) Internal brain state regulates membrane potential synchrony in barrel cortex of behaving mice. Nature 454:881-885. Raghavachari S, Kahana MJ, Rizzuto DS, Caplan JB, Kirschen MP, Bourgeois B, Madsen JR, Lisman JE (2001) Gating of human theta osci llations by a working memory task. Journal of Neuroscience 21:3175-3183. Rajagovindan R, Ding M (2008a) Selective visual a ttention increases neur onal excitability and gain in humans: A high-density EEG study. So ciety for Neuroscience Abstract 34:436.12. Rajagovindan R, Ding M (2008b) Decomposing neural synchrony: Toward an explanation for near-zero phase-lag in cortical oscillatory networks. PLoS ONE 3:1-8. Rajagovindan R, Anderson K, Ghacibeh G, Meador KJ, Ding M (2007) Prestimulus theta oscillations in human cortex during encoding predict subsequent memory recall. Society for Neuroscience Abstract 33:635.14. Rempel-Clower NL, Barbas H (2000) The laminar pattern of connections be tween prefrontal and anterior temporal cortices in the rhesus monkey is related to cortical structure and function. Cerebral Cortex 10:851-865. Rihs TA, Michel CM, Thut G (2007) Mechanisms of selective inhibiti on in visual spatial attention are indexed by al pha-band EEG synchronization. European Journal of Neuroscience 25:603-610. Rodriguez E, George N, Lachaux JP, Martinerie J, Renault B, Varela FJ (1999) Perception's shadow: Long-distance synchronization of human brain activity. Nature 397:430-433. Roelfsema PR, Engel AK, Konig P, Singer W (1997) Visuomotor integration is associated with zero time-lag synchronization among cortical areas. Nature 385:157-161. Romei V, Rihs T, Brodbeck V, Thut G (2008a) Resting electroencephalogram alpha-power over posterior sites indexes baseline visual cortex excitab ility. Neuroreport 19:203-208. Romei V, Brodbeck V, Michel C, Amedi A, Pascual-Leone A, Thut G (2008b) Spontaneous fluctuations in posterior alpha -band EEG activity reflect vari ability in excitability of human visual areas. Cere bral Cortex 18:2010-2018. Roskies AL (1999) The binding problem. Neuron 24:7-9. Saalmann YB, Pigarev IN, Vidyasagar TR (2007) Ne ural mechanisms of vi sual attention: How top-down feedback highlights releva nt locations. Science 316:1612-1615. 147

PAGE 148

Salinas E, Sejnowski TJ (2000) Impact of corre lated synaptic input on output firing rate and variability in simple neuronal models. Journal of Neuroscience 20:6193-6209. Salinas E, Sejnowski TJ (2001) Correlated neuronal activity and the flow of neural information. Nature Reviews Neuroscience 2:539-550. Sarnthein J, Petsche H, Rappelsberger P, Sh aw GL, von Stein A (1998) Synchronization between prefrontal and posterior associ ation cortex during human wo rking memory. Proceedings of the National Academy of Sciences of the United States of America 95:7092-7096. Sauseng P, Klimesch W, Stadler W, Schabus M, Doppelmayr M, Hanslmayr S, Gruber WR, Birbaumer N (2005) A shift of visual spatial a ttention is selectively associated with human EEG alpha activity. European Jour nal of Neuroscience 22:2917-2926. Schack B, Rappelsberger P, Anders C, Weiss S, Moller E (2000) Quantification of synchronization processes by coherence and pha se and its applicati on in analysis of electrophysiological signals. International Journal of Bi furcation and Chaos, 10:25652586. Scherg M, Ille N, Bornfleth H, Berg P (2002) Advanced tools for digital EEG review: Virtual source montages, whole-head mapping, correlation, and phase analysis. Journal of Clinical Neurophysiology 19:91-112. Schroeder CE, Mehta AD, Foxe JJ (2001) Determinants and mechanisms of attentional modulation of neural processing. Fr ontiers in Bioscience 6:D672-D684. Schroeder CE, Tenke CE, Givre SJ, Arezzo JC, Vaughan HG (1991) Striate cortical contribution to the surface-recorded pattern-reversal VEP in the alert monkey. Vision Research 31:1143-1157. Scoville WB, Milner B (1957) Loss of recent me mory after bilateral hippocampal lesions. Journal of Neurology Neurosurge ry and Psychiatry 20:11-21. Seager MA, Johnson LD, Chabot ES, Asaka Y, Be rry SD (2002) Oscillat ory brain states and learning: Impact of hippocampal theta-conti ngent training. Proceedings of the National Academy of Sciences of the Unite d States of America 99:1616-1620. Sederberg PB, Kahana MJ, Howard MW, Donn er EJ, Madsen JR (2003) Theta and gamma oscillations during encoding predict subseque nt recall. Journal of Neuroscience 23:1080910814. Sederberg PB, Schulze-Bonhage A, Madsen JR, Br omfield EB, McCarthy DC, Brandt A, Tully MS, Kahana MJ (2007) Hippocampal and neocor tical gamma oscillations predict memory formation in humans. Cere bral Cortex 17:1190-1196. Seth AK (2005) Causal connectivity of evolve d neural networks during behavior. NetworkComputation in Neural Systems 16:35-54. 148

PAGE 149

Shimamura AP (1995) Memory and the prefrontal cortex. Structure and F unctions of the Human Prefrontal Cortex 769:151-159. Shimamura AP, Jurica PJ, Mangels JA, Gershberg FB, Knight RT (1995) Susceptibility to memory interference effects following frontal-l obe damage findings from tests of pairedassociate learning. Journal of Cognitive Neuroscience 7:144-152. Siapas AG, Lubenov EV, Wilson MA (2005) Prefrontal phase locking to hippocampal theta oscillations. Neuron 46:141-151. Siegel M, Donner TH, Oostenveld R, Fries P, Engel AK (2008) Neuronal synchronization along the dorsal visual pathway re flects the focus of spatial attention. Neuron 60:709-719. Sieroff E, Piquard A, Auclair L, Lacomblez L, Derouesne C, Laberge D (2004) Deficit of preparatory attention in frontotemporal dementia. Brain Cognition 55:444-451. Simons JS, Spiers HJ (2003) Prefrontal and medi al temporal lobe interactions in long-term memory. Nature Reviews Neuroscience 4:637-648. Singer W, Gray CM (1995) Visual feature integration and the temporal correlation hypothesis. Annual Review of Neuroscience 18:555-586. Siok CJ, Rogers JA, Kocsis B, Hajos M (2006) Activation of alpha 7 acetylcholine receptors augments stimulation-induced hippocampal th eta oscillation. European Journal of Neuroscience 23:570-574. Skinner JE, Yingling CD (1976) Regul ation of slow potential shifts in nucleus reticularis thalami by mesencephalic reticular-forma tion and frontal granular co rtex. Electroencephalography and Clinical Neurophysiology 40:288-296. Slotnick SD, Moo LR, Kraut MA, Lesser RP, Hart J (2002) Interactions between thalamic and cortical rhythms during semantic memory r ecall in human. Proceedings of the National Academy of Sciences of the Unite d States of America 99:6440-6443. Staubli U, Xu FB (1995) Effects of 5-HT3 receptor antagonism on hippocampal theta-rhythm, memory, and LTP induction in the freely m oving rat. Journal of Neuroscience 15:24452452. Strange BA, Otten LJ, Josephs O, Rugg MD, Do lan RJ (2002) Dissociable human perirhinal, hippocampal, and parahippocampal roles during verbal encoding. Journal of Neuroscience 22:523-528. Strens LHA, Oliviero A, Bloem BR, Gerschla ger W, Rothwell JC, Brown P (2002) The effects of subthreshold 1 Hz repetitive TMS on cortic o-cortical and interhemispheric coherence. Clinical Neurophysiology 113:1279-1285. Summerfield C, Mangels JA ( 2005) Coherent theta-band EEG ac tivity predicts item-context binding during encoding. Neuroimage 24:692-703. 149

PAGE 150

Takashima A, Jensen O, Oostenveld R, Mari s E, van de Coevering M, Fernandez G (2006) Successful declarative memory formation is associated with ongoing activity during encoding in a distributed neocortical network related to working memory: A magnetoencephalography study. Neuroscience 139:291-297. Tallon-Baudry C, Mandon S, Fr eiwald WA, Kreiter AK (2004) Oscillatory synchrony in the monkey temporal lobe correlate s with performance in a visu al short-term memory task. Cerebral Cortex 14:713-720. Taylor PCJ, Nobre AC, Rushworth MFS (2007) FEF TMS affects visual cortical activity. Cerebral Cortex 17:391-399. Tendolkar I, Rugg M, Fell J, Vogt H, Sc holz M, Hinrichs H, Heinze HJ (2000) A magnetoencephalographic study of brain activity related to recognition memory in healthy young human subjects. Neuroscience Letters 280:69-72. Thomson AM, Lamy C (2007) Func tional maps of neocortical lo cal circuitry. Frontiers in Neuroscience 1:19-42. Thomson DJ (1982) Spectrum estimation and harmonic analysis. Proceedings of the IEEE 70:1055-1096. Thomson DJ (2007) Jackknifing multitaper spec trum estimates. Signal Processing Magazine, IEEE 24:20-30. Thut G, Nietzel A, Brandt SA, Pascual-Leone A (2006) Alpha-band electroencephalographic activity over occipital cortex inde xes visuospatial atte ntion bias and predic ts visual target detection. Journal of Neuroscience 26:9494-9502. Truccolo W, Eden UT, Fellows MR, Donoghue JP, Brown EN (2005) A point process framework for relating neural spiking activity to spiking history, neural ensemble, and extrinsic covariate effects. J ournal of Neurophysiology 93:1074-1089. Tulving E, Kapur S, Craik FIM, Moscovitch M, Houle S (1994) Hemispheric encoding/retrieval asymmetry in episodic memory positron emission tomography findings. Proceedings of the National Academy of Sciences of the United States of America 91:2016-2020. Ungerleider LG, Gaffan D, Pela k VS (1989) Projections from in ferior temporal cortex to prefrontal cortex via the uncinate fascic le in rhesus-monkeys. Experimental Brain Research 76:473-484. Van Dijk H, Schoffelen JM, Oostenveld R, Jens en O (2008) Prestimulus oscillatory activity in the alpha band predicts visual discrimina tion ability. Journal of Neuroscience 28:18161823. Varela F, Lachaux JP, Rodriguez E, Martiner ie J (2001) The brainweb: Phase synchronization and large-scale integr ation. Nature Review s Neuroscience 2:229-239. 150

PAGE 151

Vertes RP (2005) Hippocampal theta rhythm: A tag for short-term memory. Hippocampus 15:923-935. Vertes RP, Hoover WB, Di Prisco GV (2004) Theta rhythm of the hippocampus: Subcortical control and functional significance. Behavioral and Cognitive Neuroscience Reviews 3:173-200. Volgushev M, Chistiakova M, Singer W (1998) Modi fication of discharge pa tterns of neocortical neurons by induced oscillations of the membrane potential. Neuroscience 83:15-25. von Stein A, Sarnthein J (2000) Diffe rent frequencies for different sc ales of cortical integration: from local gamma to long range alpha/theta synchronization. Intern ational Journal of Psychophysiology 38:301-313. von Stein A, Chiang C, Konig P (2000) Topdown processing medi ated by interareal synchronization. Proceedings of the National Ac ademy of Sciences of the United States of America 97:14748-14753. Wagner AD, Poldrack RA, Eldridge LL, De smond JE, Glover GH, Gabrieli JDE (1998) Material-specific lateralization of prefr ontal activation during episodic encoding and retrieval. Neuroreport 9:3711-3717. Wang X, Chen YH, Ding MZ (2007a) Testing fo r statistical significan ce in bispectra: A surrogate data approach and a pplication to neuroscience. IEEE Transactions on Biomedical Engineering 54:1974-1982. Wang X, Chen YH, Bressler SL, Ding MZ (2007b) Granger causality between multiple interdependent neurobiologi cal time series: Blockwise versus pairwise methods. International Journal of Neural Systems 17:71-78. Watson RT, Valenstein E, Heilman KM (1981) Thal amic neglect possible role of the medial thalamus and nucleus reticularis in be havior. Archives of Neurology 38:501-506. Weiss S, Rappelsberger P (2000) Long-range EEG synchronization during word encoding correlates with successful memory performance. Cognitive Brain Research 9:299-312. Weissman DH, Roberts KC, Visscher KM, Woldor ff MG (2006) The neural bases of momentary lapses in attention. Natu re Neuroscience 9:971-978. Wiener N (1956) The theory of prediction. New York: McGraw-Hill. Winson J (1978) Loss of hippocampal theta rhythm results in spatial memory deficit in rat. Science 201:160-163. Witham CL, Wang M, Baker SN (2007) Cells in somatosensory areas show synchrony with beta oscillations in monkey motor cortex. Eur opean Journal of Ne uroscience 26:2677-2686. 151

PAGE 152

152 Wolfart J, Debay D, Le Masson G, Destexhe A, Bal T (2005) Synaptic background activity controls spike transfer from thalamus to cortex. Nature Neuroscience 8:1760-1767. Womelsdorf T, Fries P, Mitra PP, Desimone R (2006) Gamma-band sync hronization in visual cortex predicts speed of change detection. Nature 439:733-736. Worden MS, Foxe JJ, Wang N, Simpson GV ( 2000) Anticipatory biasing of visuospatial attention indexed by retinotopically specific alpha-band elect roencephalography increases over occipital cortex. Journa l of Neuroscience 26:9494-9502. Wu JH, Liu XG, Feng JF (2008) Detecting causality between different frequencies. Journal of Neuroscience Methods 167:367-375. Yago E, Duarte A, Wong T, Barcelo F, Knight RT (2004) Temporal kinetics of prefrontal modulation of the extrastriate cortex during visual attention. Cognitive Affective & Behavioral Neuros cience 4:609-617. Yingling CD, Skinner JE (1975) Regulation of un it-activity in nucleus reticularis thalami by mesencephalic reticular-formation and frontal granular cortex. Electroencephalography and Clinical Neurophysiology 39:635-642. Yingling CD, Skinner JE (1976) Selective regulati on of thalamic sensory relay nuclei by nucleus reticularis thalami. Electroencephalogra phy and Clinical Neurophysiology 41:476-482. Zhang Y, Ding M (in press) Detection of a weak somatosensory stimulus: Role of the prestimulus mu rhythm and its top-down m odulation. Journal of Cognitive Neuroscience. Zhang Y, Chen Y, Bressler SL, Ding M (2008a) Re sponse preparation and i nhibition: The role of the cortical sensorimotor beta rhythm. Neuroscience 156:238-246. Zhang Y, Wang X, Bressler SL, Chen YH, Ding MZ (2008b) Prestimulus cortical activity is correlated with speed of visuomotor processing. Journal of Cognitive Neuroscience 20:1915-1925. Zikopoulos B, Barbas H (2007) Circuits for multisensory integration and attentional modulation through the prefrontal cortex a nd the thalamic reticul ar nucleus in primates. Reviews in the Neurosciences 18:417-438.

PAGE 153

BIOGRAPHICAL SKETCH Rajasimhan Rajagovindan was born in 1983, in Chennai, India. He grew up in Chennai, graduating from Padma Seshadri Bala Bhavan Senior Secondary School in 2000. He earned the Bachelor of Engineering degr ee in electronics and communication engineering from the University of Madras, Chennai in 2004 following which Raj enrolled in th e graduate program at the University of Florida where he earned th e Master of Science degree in biomedical engineering in 2008. He continued at the University of Florida to earn his Doctor of Philosophy in the fall of 2009 in biomedical engineering specializing in cognitive neurosciences under the mentorship of Dr.Mingzhou Ding. Raj intends to pursue his research interests in functional biomarker discovery to devel op diagnostic tools to identify and assess the progression of neurological and psychiatric illnesses and the eff ectiveness of therapeutic interventions to enable therapy augmentation. 153