Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2014-08-31.

MISSING IMAGE

Material Information

Title:
Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2014-08-31.
Physical Description:
Book
Language:
english
Creator:
Dzierzewski, Joseph M
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Psychology, Clinical and Health Psychology
Committee Chair:
Mccrae, Christina S
Committee Co-Chair:
Marsiske, Michael
Committee Members:
Bowers, Dawn
Roberts, Beverly L

Subjects

Subjects / Keywords:
Clinical and Health Psychology -- Dissertations, Academic -- UF
Genre:
Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility:
by Joseph M Dzierzewski.
Thesis:
Thesis (Ph.D.)--University of Florida, 2012.
Local:
Adviser: Mccrae, Christina S.
Local:
Co-adviser: Marsiske, Michael.
Electronic Access:
INACCESSIBLE UNTIL 2014-08-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0044480:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 COGNITIVE CHANGE IN OLDER ADULTS: PRACTICE EFFECTS, SHORT TERM VARIABILITY, AND THEIR ASSOCIATION WITH SLEEP By JOSEPH MICHAEL DZIERZEWSKI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIA L FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 201 2

PAGE 2

2 201 2 Joseph Michael Dzierzewski

PAGE 3

3 This dissertation is dedicated t o my mother and father, for everything and more ; t o my entire famil y and closest friends for motivation, encouragement, support, ; and l astly, to my inspiration in research and clinical care older adults.

PAGE 4

4 ACKNOWLEDGMENTS T here are many, many people that have contr ibuted to this project in one way or another. First, and foremost, I must thank my mentors, Drs. Christina McCrae and Michael Marsiske. They have given me much throughout my graduate school tenure. I also must express my gratitude to the complete team of r esearchers whom all have been intricately involved in every phase of the Active Adult Mentoring Program (AAMP; the parent study to this dissertation): Christina McCrae, Michael Marsiske, Beverly Roberts, Peter Giacobbi, Adrienne Aiken Morgan, and Mathew Bu man. The experience of working within this exceptional group of multidisciplinary researchers has been informative and rewarding on many different levels. I have also been fortunate to have enlisted the aid of a fantastic group of undergraduate research as sistants, whom at many times went above and beyond the call of duty. There are many sources of funding that need acknowledgement. First, this dissertation project was supported primarily by a National Institute on Aging (NIA) Aging Research Dissertation A ward to Increase Diversity (1R36AG029664 01 PI: Aiken Morgan). In addition, this study was supported by other sources that include: (1) a Research Opportunity Fund in the College of Health and Human Performance at the University of Florida (PI: Giacobbi), (2) an Age Network Multidisciplinary Research Enhancement grant at the University of Florida (PI: McCrae), (3) a Mentorship Opportunity Grant from the Graduate Student Council at the University of Florida (PI: Buman), (4) a Graduate Student Research Grant, University of Florida, College of Public Health and Health Professionals (PI: Dzierzewski), (5) a Division 20, American Psychological Association and

PAGE 5

5 Retirement Research Foundation Thesis Proposal Award (PI: Dzierzewski), and (5) a Division 20, American P sychological Association and Retirement Research Foundation Dissertation Research Award (PI: Dzierzewski). Lastly, I have been fortunate enough to have been funded by both an Institutional Training Grant (T32 AG 020499, National Institute on Aging) and an Individual Training Grant (F31 AG 032802 01A1, National Institute on Aging) throughout the course of this project. At this point I have to turn my appreciation to my parents, Steve and Karen Dzierzewski, for never holding back. They were always there, alw ays caring, and always believing in me. They have given me everything and asked for nothing everything. Lastly, I must thank Zvinka Zlatar for being a dependable resource thro ughout the many ups and downs that came throughout the process of completing this dissertation project.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF T ABLES ................................ ................................ ................................ ........... 10 LIST OF FIGURES ................................ ................................ ................................ ........ 12 ABSTRACT ................................ ................................ ................................ .................... 14 CHAPTER 1 GENERAL INTRO DUCTION ................................ ................................ ..................... 16 Overview ................................ ................................ ................................ ................. 16 Late life Cognitive Change ................................ ................................ ................ 18 Public Heal th Importance of Learning and Short term Fluctuation in Late life ................................ ................................ ................................ .......... 21 Paper 1: Intensive Cognitive Practice in Older Adults: Gains, Structure, Predictors, and Transfer ................................ ................................ ....................... 23 Paper 2: Sleep Variables as Predictors of Initial Level and Gain in Late Life Cognition ................................ ................................ ................................ ....... 26 Paper 3: Do Weekly Deviations in Sleep Impact Subsequent Co gnitive Functioning in Older Adults? ................................ ................................ ................ 29 Summary ................................ ................................ ................................ ................. 32 2 INTENSIVE COGNITVE PRACTICE IN OLDER ADULTS: GAINS, STRUCTURE, PREDICTORS, AND TRANSFER. ................................ .................. 33 Introduction ................................ ................................ ................................ .............. 33 Practice Related Learning Gains ................................ ................................ ...... 35 Practice Related Learning Structure ................................ ................................ 38 Practice Related Learning Predictors ................................ ................................ 41 Practice Related Learning Transfer ................................ ................................ .. 42 The Current Investigation ................................ ................................ .................. 44 Methods ................................ ................................ ................................ ................... 45 General Study Design ................................ ................................ ....................... 45 Procedure ................................ ................................ ................................ ......... 46 Participants ................................ ................................ ................................ ....... 47 Inclusion/ e xclusion c riteria ................................ ................................ .......... 47 Demographic and d escriptive m easures ................................ ..................... 48 Descriptive s tatistics ................................ ................................ ................... 48 Measure s ................................ ................................ ................................ .......... 49 Practiced c ognitive m easures ................................ ................................ ..... 49 Processing speed ................................ ................................ ....................... 50 Executi ve processing ................................ ................................ .................. 50

PAGE 7

7 Near t ransfer c ognitive m easures ................................ ............................... 51 Processing s peed ................................ ................................ ....................... 51 Executive p rocessing ................................ ................................ .................. 52 Far t ransfer m easures ................................ ................................ ................ 52 Analyses ................................ ................................ ................................ ........... 53 Preliminary a nalyses ................................ ................................ .................. 53 Main a nalyses ................................ ................................ ............................. 54 Results ................................ ................................ ................................ .................... 57 Prelim inary Analyses ................................ ................................ ......................... 57 Normality ................................ ................................ ................................ .... 57 Missing d ata ................................ ................................ ............................... 57 Attrition ................................ ................................ ................................ ....... 58 Data s tructure ................................ ................................ ............................. 58 Main Analyses ................................ ................................ ................................ ... 60 Aim 1: Individual g rowth c urves ................................ ................................ .. 60 Aim 2: Curves of f actors ................................ ................................ ............. 64 Aim 3: Curves of f actors: I ndividual d ifferences ................................ .......... 65 Aim 4: Curves of f actors: T ransfer ................................ .............................. 66 Discussion ................................ ................................ ................................ ............... 68 Practice Related Learning ................................ ................................ ................. 68 Practice Related Learning Structure ................................ ................................ 7 0 Practice Related Learning Predictors ................................ ................................ 72 Practice Relat ed Learning Transfer ................................ ................................ .. 74 Limitations ................................ ................................ ................................ ......... 75 Future Directions ................................ ................................ ............................... 77 Summ ary ................................ ................................ ................................ ........... 78 3 SLEEP VARIABLES AS PREDICTORS OF INITIAL LEVEL AND GAIN IN LATE LIFE COGNITION ................................ ................................ ....................... 112 Introduction ................................ ................................ ................................ ............ 112 Late Life Practice Related Learning ................................ ................................ 113 Sleep and Cognitive Functioning ................................ ................................ .... 115 Natura l s leep ................................ ................................ ............................ 116 Retrospective ................................ ................................ ............................ 117 Prospective ................................ ................................ ............................... 119 Objective: PSG ................................ ................................ ......................... 122 Objective: Actigraphy ................................ ................................ ................ 125 Experimentally m anipulated s leep ................................ ............................ 125 Sleep d eprivation ................................ ................................ ...................... 126 Sleep r estriction ................................ ................................ ........................ 134 Sleep c ognition t heories ................................ ................................ ........ 138 The c urrent i nvestigation ................................ ................................ ................. 140 Methods ................................ ................................ ................................ ................. 142 General Study Design ................................ ................................ ..................... 142 Procedure ................................ ................................ ................................ ....... 142 Participants ................................ ................................ ................................ ..... 143

PAGE 8

8 Inclusion/exclusion c riteria ................................ ................................ ........ 143 Demographic and d escripti ve m easures ................................ ................... 144 Descriptive s tatistics ................................ ................................ ................. 145 Measures ................................ ................................ ................................ ........ 145 Cognitive m easures ................................ ................................ .................. 145 Sleep m easures ................................ ................................ ........................ 147 Analyses ................................ ................................ ................................ ......... 148 Prelimina ry a nalyses ................................ ................................ ................ 148 Main a nalyses ................................ ................................ ........................... 150 Results ................................ ................................ ................................ .................. 152 Preliminary Analyses ................................ ................................ ....................... 152 Normality ................................ ................................ ................................ .. 152 M issing d ata ................................ ................................ ............................. 153 Attrition ................................ ................................ ................................ ..... 154 Systematic s leep c hanges ................................ ................................ ........ 155 Main Analyses ................................ ................................ ................................ 155 Aim 1: Unconditional growth c urves ................................ ......................... 155 Aim 2: Conditional growth c urves ................................ ............................. 159 Discussion ................................ ................................ ................................ ............. 164 Practice Related Learning ................................ ................................ ............... 165 Predictors of Starting Level in Cognitive Functioning ................................ ...... 166 Predictors of Practice Learning in Cognitive Functioning ................................ 169 Limitations ................................ ................................ ................................ ....... 172 Future Directions ................................ ................................ ............................. 175 Summary ................................ ................................ ................................ ......... 177 4 DO WEEKLY DEVIATIONS IN SLEEP IMPACT SUBSEQUENT COGNITIVE FUNCTIONING IN OLDER ADULTS? ................................ .............. 207 Introduction ................................ ................................ ................................ ............ 207 Sleep and Cognitive Functioning ................................ ................................ .... 209 Short term Fluctuation in Late Life Sleep and Cognitive Functioning ............. 210 The Current Investigation ................................ ................................ ................ 214 Methods ................................ ................................ ................................ ................. 216 General Study Design ................................ ................................ ..................... 216 Procedure ................................ ................................ ................................ ....... 217 Participants ................................ ................................ ................................ ..... 217 Inclusion/ e xclusion c riteria ................................ ................................ ........ 218 Demographics ................................ ................................ .......................... 219 Descriptive s tatistics ................................ ................................ ................. 219 Measures ................................ ................................ ................................ ........ 219 Cognitive m easures ................................ ................................ .................. 219 Sleep m easures ................................ ................................ ........................ 221 Analyses ................................ ................................ ................................ ......... 222 Preliminary a nalyses ................................ ................................ ................ 222 Main a nalyses ................................ ................................ ........................... 224 Results ................................ ................................ ................................ .................. 226

PAGE 9

9 Preliminary Analyses ................................ ................................ ....................... 226 Normality ................................ ................................ ................................ .. 226 Missing d ata ................................ ................................ ............................. 226 Attrition ................................ ................................ ................................ ..... 227 Main Analyses ................................ ................................ ................................ 227 Aim 1: Amount of s hort term f luctuation ................................ .................... 227 Aim 2: Temporal s leep c ognition a ssociations ................................ .......... 228 Discussion ................................ ................................ ................................ ............. 232 Amount of Fluctuati on ................................ ................................ ..................... 232 Temporal Sleep Cognition Associations ................................ ......................... 234 Limitations and Future Directions ................................ ................................ .... 236 Summary ................................ ................................ ................................ ......... 239 5 GENERAL DISCUSSION ................................ ................................ ........................ 252 LIST OF REFERENCES ................................ ................................ .............................. 258 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 276

PAGE 10

10 LIST OF TABLES Table page 2 1 Descriptive/ d emographic s tatistics ................................ ................................ ...... 80 2 2 Measures ................................ ................................ ................................ ............ 81 2 3 Descriptive, n ormality, and r aw c ognition v alues p rior to and p ost d ata c leaning. ................................ ................................ ................................ .............. 83 2 4 Compar ison of means and standard deviations of baseline study variables among completed and attrited subjects. ................................ .............. 92 2 5 Baseline correlations between practiced and unpracticed (i.e., transfer) cognitive measures. ................................ ................................ .............. 93 2 6 Rotated f actor p attern of c ognitive m easures at b aseline. ................................ .. 94 2 7 Cognitive b lock d ata ................................ ................................ ............................ 96 2 8 Parameter e stimates for l atent g rowth c urve m odels. ................................ ......... 97 2 9 Parameter e stimates for c urve of factors m odel. ................................ ............... 100 2 10 Standardized c oefficients p redicting c ommon l evel and c hange in g eneralized l earning. ................................ ................................ ......................... 104 2 11 Standardized c oefficients for t ransfer in g eneralized l earning to n on practiced m easures (N = 68). ................................ ................................ ............ 110 3 1 Sleep c ognition r elationship. ................................ ................................ .......... 178 3 2 Descriptive/ d emographic s tatistics. ................................ ................................ ... 180 3 3 Between p erson c orrelations a mong 5 b lock s leep v ariables and d escriptive s leep d ata. ................................ ................................ ...................... 181 3 4 Descriptive and n ormality v alues for c ognitive d ependent v ariabl es and s leep i ndependent v ariables p rior to and p ost d ata c leaning. ..................... 184 3 5 Comparison of m eans and s tandard d eviations of b aseline s tudy v ariables among c ompleted and a ttrited s ubjects. ................................ ............ 191 3 6 Unconditional m ultilevel g rowth m odels e xamining c hange in s leep Over Time. ................................ ................................ ................................ ........ 192 3 7 Parameter e stimates for u nconditional l atent g rowt h c urve m odels. ................. 193

PAGE 11

11 3 8 Parameter e stimates for f inal c onditional l atent g rowth c urve m odels. .............. 196 3 9 Model f it for c onditional l atent g rowth c urve m odels. ................................ ......... 203 4 1 Demographic s tatistics. ................................ ................................ ..................... 241 4 2 Interindividual ( b elow d iagonal) and i ntraindividual ( a bove d iagonal) c o rrelations a mong s leep v ariables and d escriptive s leep d ata. ....................... 242 4 3 Two l evel m ultilevel m odel p redicting n umber c opy p erformance. .................... 247 4 4 Two l evel m ultilevel m odel p redicting s ymbol d igit p erformance. ...................... 248 4 5 Two l evel m ultilevel m odel p redicting l etter s eries p erformance. ...................... 249 4 6 Two l evel m ultilevel m odel p redicting s imple RT p erformance. ......................... 250 4 7 Two l evel m ultilevel m odel p redicting c hoice RT p erformance. ......................... 251

PAGE 12

12 LIST OF F IGURES Figure page 2 1 Path diagram depicting higher order latent growth model ( a im 2, center circles/large dashed circle), predictor variables ( a im 3, bottom rectangles/solid rectangle ), and transfer learning ( a im 4, top rectangles/double lined rectangle). ................................ ................................ ..... 82 2 2 Graphical r epresentation of m issing d ata across the s tudy p eriod. ..................... 91 2 3 Posttest c onfirmatory f actory a nalysis of c ognitive v ariables. .............................. 95 2 4 Model i mplied c hange in c ognitive f unctioning a cross s tudy p eriod. ................... 99 2 5 Model i mplied c hange in g eneral c ognitive f unctioning a cross s tudy p eriod. ................................ ................................ ................................ ............... 102 2 6 Curve of f actors p ath d iagram with s tandardized c oefficients. .......................... 103 2 7 Complete p ath d iagram of i ndividual d ifference p redictors of o verall c hange in g eneralized l earning. ................................ ................................ ........ 105 2 8 Standardized l oadings and s ignificant l evels of i ndividual d ifferences in g eneralized l earning. N ote: ** p < .01, p < .05. ................................ ............... 106 2 9 Graphical r epresentation of g eneralized l earning for both a h igh (i.e., 90th %ile) and l ow (i.e., 10 th %ile) e ducated (left) and s tate a nxious (right) s ubject. ................................ ................................ ................................ ... 107 2 10 Complete p ath d iagram of t ransfer of g eneralized l earning in c ognitive f unctioning. ................................ ................................ ................................ ........ 108 2 11 Standardized l oadings and s ignificant l evels of t ransfer in g eneralized l earning. Note: *** p < 0.01, p < 0.05, ~ p < 0.10. N = 68. ................................ 109 2 12 Graphical r epresentation of g eneralized l earning t ransfer for both a h igh (i.e., 80 th %ile) and l ow (i.e., 20 th %ile) c hange 1 b ack RT s ubject (left) and 2 b ack RT s ubject (right). ................................ ................................ ... 111 3 1 Graphical r epresentation of u ni variate u nconditional g rowth c urve. .................. 182 3 2 Graphical r epresentation of f inal u nivariate c onditional g rowth c urve. .............. 183 3 3 Model i mpli ed c hange in c ognitive f unctioning a cross s tudy p eriod. ................. 195 3 4 Graphical r epresentation of a ssociation between TWT and b lock 1 s ymbol d igit. ................................ ................................ ................................ ...... 204

PAGE 13

13 3 5 Graphical r epresentation of s ssociation between TWT and b lock 1 l etter s eries. ................................ ................................ ................................ ....... 205 3 6 Graphical r epresentation of p ractice l earning in s imple RT for both a h igh (i.e., 90th %ile) and l ow (i.e., 10th %ile) TWT s ubject. .............................. 206 4 1 Gantt c hart i llustrating the t iming of s leep and c ognitive d ata c ollection. ................................ ................................ ................................ .......... 240 4 2 Gr aphical r epresentation of w eekly and TST and TWT: r aw d ata p lots of s w eekly s leep v alues. ................................ ................................ ...... 243 4 3 Graphical r epresentation of m issing d ata across the s tudy period. ................... 244 4 4 Graphical r epresentation of the a mount of i ntra and i nterindividual f luctuation and the p roportion of i ntra to i nterindividual v ariability in c ognitive f unctioning. ................................ ................................ ......................... 245 4 5 Graphical r epresentation of f luctuation in c ognitive f unctioning: r aw d ata p lots of s m ean w eekly p erformance. ................................ ............ 246

PAGE 14

14 Abstract of Dissertation Presented to the Graduate School of the Univers ity of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy COGNITIVE CHANGE IN OLDER ADULTS: PRACTI CE EFFECTS, SHORT TERM VARIABILITY, AN D THEIR ASSOCIATION WITH SLEEP By Joseph Michael Dzierzewski August 2012 Chair: Christina S. McCrae Cochair: Michael Marsiske Major: Psychology T his multipaper dissertation examine d t wo types of late life cognitive change (1) late life learning and (2) short term fluctuation in late life cognitive functioning. The papers examine d the structure, predictors and potential consequences of practice related learning in late ability to acquire skills through practice, and how an aberrant week of sleep affect s subsequent cognitive functi oning. Future implications of the work relate to both a deeper understanding of how effectively practice functions as a cognitive intervention strategy (in terms of producing broad cognitive enhancement effects), and how sleep predicts the ability to profi t from practice interventions. The dissertation successfully replicated previously reported findings in cognitive aging literature (i.e., older adults are capable of demonstrating significant practice related learning, and there is substantial short term fluctuation present in the cognitive performance of older adults), while simultaneously addressing unanswered questions regarding late life cognitive plasticity (i.e., practice related learning and short term fluctuations in late life cognitive

PAGE 15

15 functioning ). Both processing speed measures and an executive processing measure demonstrated significant gains associated with repeated practice, individuals who gained more in one ability were more likely to gain in another, both education and state anxiety were as from repeated practice, and generalized practice related learning demonstrated limited near transfer for processing speed. We also discovered that older adults who spent more time awake during the night on average had lower average cognitive performance. Interestingly, the same older adults that spent increased time awake during the night were found to also benefit the most from cognitive practice on a very simple task reaction time task. No relationships were observed between weekly fluctuations in sleep and cognitive functioning in late life. As a whole, these finding suggest several conclusions and directions for future investigation. Older adults are very capable of engendering improvements in their cognitive func tioning through self administered practice, and sleep is vitality. Future research examini ng sleep and late life cognitive functioning should be conducted to further explicate the complicated association between these important quality of life indicators.

PAGE 16

16 CHAPTER 1 GENERAL INTRODUCTION Overview In this General Introduction it is argued that l ate life cognitive changes can occur at three broad levels of functioning: (1) development, (2) learning, and (3) short term fluctuation. This document examined cognitive changes in older adults at the latter two of these three levels learning and short term fluctuation. Investigation into learning and short term fluctuation is an important area of investigation given their potential to impact daily functioning and quality of life in older adults. The first two individual studies conducted in this docum ent examined late life learning, which may be important for maintenance of independence [i.e., cognitive interventions may slow the decline in activities of daily living in late life (Willis et al., 2006)] and has traditionally been investigated through in terventional work (Ball et al., 2002). However, older adults appear to be able to benefit considerably from practice alone (Yang, Krampe, & Baltes, 2006). Thus, the first two individual studies sought to focus on practice related learning because: (1) self guided practice is likely a simpler, more cost effective, and a more widely scalable approach to cognitive interventions than traditional tutor based approaches, and (2) practice appears to be the active ingredient in many cognitive training studies (and should be investigated in its purest form). While studies of late life practice related learning have existed for several decades these studies were innovative in its focus on the more general nature of practice effects, and the identification of predicto rs of individual differences in practice

PAGE 17

17 related gain. More specifically, through its focus on practice related learning, the set of studies attempted to characterize the structure (i.e., is practice related learning domain general or domain specific?), pr edictors (i.e., what predict ed practice related learning?), and potential consequences (i.e., does practice related learning transfer to non practiced tests?) of practice related learning in late life. The longer term practical implications of this set of studies are (a) characterization of the effectiveness of practice as an intervention approach for a wide set of cognitive outcomes; (b) deeper understanding of the extent to which practice effects are measure specific or more reflective of broad, domain g eneral learning processes that may transfer to other, non practiced cognitive outcomes; and (c) identification of attributes and characteristics of older adults that may make them differentially able to profit from practice. The first paper in this dissert ation describes the magnitude, generality, and demographic/personality level predictors of practice related learning on a wide set of outcomes. Following this examination of practice related learning, the second study in this document then turned specifica lly to the investigation of related learning. The evidence linking sleep to cognitive functioning in late life is mixed (Altena, Van Der Werf, Strijers, & Van Someren, 2008; Tworoger, Lee, Sche rnhammer, & Grodstein, 2006); however, no known research has examined sleep as a yield important information regarding optimal characteristics of sleep necessary

PAGE 18

18 for olde r adults to achieve the most advantageous results from engaging in cognitive exercise (i.e., practicing cognitive tasks). Lastly, the third study in this set of studies investigated late life cognitive changes at the level of short term fluctuation in lat e life cognitive functioning. As fluctuation in cognitive performance has been demonstrated to be an overall negative indicator of health (Hultsch, Strauss, Hunter, & MacDonald, 2008) and relates to adaptive fitness at any given time (such that when an ind ividual is performing closer to their intrapersonal mean levels of performance their adaptive fitness may be higher than when they are performing at lower levels of performance), it is important to understand potential predictors of fluctuations in cogniti ve functioning. This study could result in potential treatment recommendations regarding sleep that are aimed at regulating the cognitive performance of elders and potentially increasing independence. Late life Cognitive Change The psychology of aging is a sub discipline of psychology that is devoted to the empirical investigation of psychological change associated with the aging process The vast majority of both basic and applied investigation s in the psychology of aging ha ve been focused on some aspect of late life cognitive functioning. Cognitive change, like all behavioral change, can be characterized in three broad ways : (1) development, (2) learning, and (3) short term fluctuation As Li and colleagues (2004) have described, the varying types of cog nitive (Li, Huxhold, & Schmiedek, 2004) Investigation of cognitive change, like

PAGE 19

19 investigation of any type of change, is an exploration into plasticity or the malleability of cog nitive functioning. The first type of cognitive change, d evelopmental maturation is expected to be the result of cumulative, long term processes that are universal or nearly universal and are resistant to efforts to reverse the change (Li et al., 2004; Nesselroade, 1991) This type of change has been investigated through both cross sectional and longitudinal investigation of age differences and age related changes in cognitive functioning [ i.e., (Park et al., 1996; Schaie, Willis, & Caskie, 2004)]. Resu lts of such investigations have found t here is a general cognitive decline experienced with increasing age (Craik & Byrd, 1982; Hasher & Zacks, 1988; Lindenberger & Baltes, 1994; Salthouse, 1996). This decline has been demonstrated to be pervasive, affecti ng many sub domains of cogniti on including reaction time, sensory processing, attention, memory, re asoning, and executive control. Park and colleagues (1996) collected cross sectional data from adults ranging in age from 20 years old to 90 years old in the domains of processing speed, working memory, long term memory, and vocabulary. They found generally linear declines in all fluid intelligence (processing speed, working memory, long term memory) constructs across the lifespan. However, their lone indicato r of crystallized intelligence (i.e., vocabulary) displayed no such decline (Park, Smith, Lautenschlager, et al., 1996). Several large scale longitudinal studies of cognitive functioning in late life have provided evidence to corroborate the cross sectiona l findings of significant decreases in cognitive performance across the lifespan (e.g., Christensen, Mackinnon, Jorm, et al., 2004; Lovden,

PAGE 20

20 Ghisletta, & Lindenberger, 2004; Schaie, Willis, & Caskie, 2004). While much is known regarding developmental change s in cognitive functioning, relatively little is known regarding developmental changes in learning potential, with the exception that advanced age may be related to reduced training related gains in cognitive functioning (McArdle & Prindle, 2008; Yang et a l., 2006). The second type of cognitive change, learning, likely occurs over a short to moderate time frame (i.e., within a day, across several days and across months) and is semi reversible (that is, learned behavior can be forgotten and/or unlearned) ( Li et al., 2004; Nesselroade, 1991) In late life, learning is commonly the result of targeted strategies aimed at engendering noticeable improvements in functioning (though it can also be the result of practice alone) This type of change has been investi gated through direct manipulation of variables thought to affect late life cognitive functioning [such as cognitive process training (i.e., Ball et al., 2002), exercise training (i.e., McAuley, Kramer & Colcombe, 2004), and metacognition training (i.e., Je nnings & Jacoby, 2003)] and has been referred to as intervention related learning. Cognitive interventions all of which contain significant practice components (i.e., practice with sample problems is usually a key constituent of most cognitive training p rograms), have been described as an life (Hertzog, Kramer, Wilson, & Lindenberger, 2008) Results of such investigations have revealed: (a) that late li fe learning can be substantial, (b) that transfer of learning in late life is limited, (c) that late life learning can be

PAGE 21

21 maintained long term, and (d) that late life learning is reduced in older adults with dementias (Hertzog et al., 2008) The third type of cognitive change is short term fluctuation (i.e., inconsistency or intraindividual variability) As the name implies, this type of change is typically characterized by fluctuation in functioning that occurs over a relatively brief time frame and is tho ught to be highly reversible in nature (Li et al., 2004; Nesselroade, 1991) This type of change has been examined through repeated assessments spaced in close proximity to one another [such as multiple cognitive assessments repeated daily [ i.e., ( Allaire & Marsiske, 2005) ] and/or examination of inconsistency from trial to trial within a given assessment (i.e., (Hultsch, MacDonald, & Dixon, 2002) ]. Results of investigations into this type of late life cognitive change have revealed: (1) that older adults di splay large amounts of short term fluctuation in cognitive performance [ i.e., ( Hultsch, MacDonald, & Dixon, 2002) ] (2) that short term cognitive fluctuation is related to disease status [ i.e., (Hultsch, MacDonald, Hunter, Levy Bencheton, & Strauss, 2000) ] and (3) that relative peaks and valleys of cognitive functioning can be predicted [ i.e., (Neupert, Almeida, Mroczek, & Spiro, 2006) ] Public Health Importance of Learning and Short term Fluctuation in Late life This dissertation aimed to examine two of t he three types of late life cognitive change discussed above : (1) late life learning (specifically, practice related late life learning), as well as (2) short term fluctuation (i.e., inconsistency) in late life cognitive performance. Examination of the se t wo types of late life cognitive changes (i.e., learning and short term fluctuation) is important for

PAGE 22

22 several reasons. Understanding practice related learning (i.e., change independent of intervention) can shed needed light on the improve ment in cognition prior to the onset of a specific tutor based intervention (Hertzog et al., 2008); in other words, the degree of improvement in cognitive functioning that older adults can achieve on their own. Understanding practice related learning also allows for an examination of predictors of individual difference s in plasticity, while investigation into short term found to be a prominent characteristic of cognitive functio ning in late life [ e.g., (Gamaldo, Weatherbee, & Allaire, 2008; Weatherbee, Gamaldo, & Allaire, 2009)] and which (i.e., may reflect the level of cognitive effort that a person might bring to bea r to manage daily living challenges on a given day) Further, there is considerable, and growing, national interest in behavioral and cognitive plasticity in late life. This set of studies examined two different aspects of late life cognitive functioning ( i.e., learning and short term fluctuation) that may relate to adaptability and maintenance of independence in old age [i.e., receiving a cognitive intervention may slow the rate of decline in activities of daily living and older adults who demonstrate more short term fluctuation in cognitive functioning perform worse on laboratory measures of everyday functioning (Burton, Strauss, Hultsch, & Hunter, 2009; Willis et al., 2006)]. This is important both socially and economically, as the cost of providing both formal (i.e., paid) and informal (i.e., unpaid, typically provided by a blood relative) care for older adults who lose their

PAGE 23

23 ability to maintain independence has been estimated at over $200 billion (LaPlante, Harrington, & Kang, 2002). Paper 1: Intensive Cognitive Practice in Older Adults: Gains, Structure, Predictors, and Transfer Emerging trends in the late life cognitive learning literature have examined the following topics: (a) long term maintenance [ e.g., (Unverzagt et al., 2009)] and transfer [ e.g., (Willis et al., 2006)] of learning, (b) what older adults can self achieve (via computerized home based intervention tools) [ e.g., (Mahncke et al., 2006; Smith et al., 2009)] and (c) individual difference characteristics as predictors of training respons ivity [ e.g., (Unverzagt et al., 2007)] However, these late life learning: practice. With several notable recent [ i.e., (Yang & Krampe, 2009; Yang et al., 2006; Yang, Reed, Russo, & Wilkinson, 2009) and historical [i.e., (Baltes, Sowarka, & Kliegl, 1989; Hofland, Willis, & Baltes, 1981)] exceptions the vast majority of literature examining late life learning has viewed practice related improvements in cognitive functioning a s a nuisance that needs to be controlled methodologically and statistically [i.e., (Ball et al., 2002; Jaeggi, Buschkuehl, Jonides, & Perrig, 2008; Jennings & Jacoby, 2003)] However, practice alone has been empirically demonstrated to engender gains equiv alent to focused cognitive interventions (Baltes et al., 1989). Practice performance improvement through retest practice) (Yang et al., 2006) differs from traditional cognitive training in sev eral substantive ways, including the absence of: (1) adaptive feedback, (2) strategy training, and (3) goals. Thus,

PAGE 24

24 questions have been posed as to whether practice alone is sufficient to be ed the mechanisms responsible for practice related learning (Yang et al., 2009) and whether practice related learning is merely the result of memorization of specific test items (Salthouse, Schroeder, & Ferrer, 2004) For practice related learning to be co nsidered a cognitive intervention it must be found to meet some of the following criteria : (a) demonstrate substantial gains in cognitive functioning, (b) demonstrate durability over time and (c) demonstrate at least as much transfer to other unpracticed measures of cognition as training related learning does Correspondingly, the first paper in this document examined practice related learning in older adults, and specifically addressed: (a) the magnitude of gains, (b) whether gains in various practiced co gnitive abilities improve together, (c) individual differences in gains, and (d) whether level of improvement in practiced cognitive measures predicted amount of improvement in unpracticed measures. The first overall goal (i.e., Paper 1/Chapter 2) of this document wa s to examine practice related learning in community dwelling older adults. Specific aims of this study include d : (1) To replicate the presence of significant practice related learning in both processing speed and reasoning in older adults, (2) T o examine whether a higher order practice related learning construct was evident (i.e., do practiced cognitive measures improve together?) (3) To examine individual difference predictors of practice related learning (i.e., who was most likely to improve f rom cognitive practice?), and (4) To examine whether individual

PAGE 25

25 differences in practice related learning predict ed changes in unpracticed measures Cognitive functioning in late life has demonstrated an ability to respond to cognitive interventions [i.e. (Ball et al., 2002)] However, some of the gains illustrated through these interventions may be the result of practice related learning (i.e., learning independent of instruction) given that even untrained controls show pre posttest improvements An em erging goal of cognitive training research has been to demonstrate perform tasks of daily living (Willis et al., 2006). The hope was that improvements in basic functions (like memory, reasoning, speed) might yield benefits in areas like Instrumental Activities of Daily Living or, more generally, health, institutionalization rates, mortality, etc. However, for at least the past 100 years (Thorndike, 1903), it has also been known that generalization of training effects from one cognitive skill to another is very difficult to achieve; indeed, this has been the normative finding in cognitive intervention research across the lifespan (Baltes & Willis, 1982; Owen et al., 2010). More re cent research has suggested that cognitive training in speed and reasoning may generalize to everyday functioning, in that persons who received such training showed slower rates of decline in self reported Instrumental Activities of Daily Living. (Willis e t al., 2006). The current investigation could not examine these broader transfer questions, in part due to the short time frame (18 weeks) of the investigation. However, as a first look at the question of transfer, we examined whether

PAGE 26

26 extensive practice ( the 18 week duration of practice in this study exceeded the dosage of most cognitive training studies) on selective cognitive tasks might generalize to other non practiced cognitive measures. Demonstration of even local (i.e., cognitive) generalization of transfer might suggest that practice is a route to more broadly improving cognitive functioning. Previous research has shown substantial gains in cognitive functioning following practice that can be maintained for eight months (Yang & Krampe, 2009), whil e tutor guided cognitive training gains have been shown to persist for up to five years (Willis et al., 2006). However, examination of transfer of gains (i.e., how learning on one task is related to learning on a separate task) as well as the magnitude and predictors individual differences in gains has yielded mixed results (Baltes et al., 1989; Yang et al., 2006). Paper 2: Sleep Variables as Predictors of Initial Level and Gain in Late Life Cognition In addition to characterizing practice related learnin g (see above), it is also from practice ; since practice is an important component of training related learning, findings in this area are likely to be relevant to predic ting more general ability to profit from interventions In fact, a key question in the emerging gerontological cognitive intervention literature concerns individual differences in learning potential and the identification of predictors of such differences (Hertzog et al., 2008) Understanding individual differences in response to repeated exposure to cognitive stimuli (i.e., practice related learning) is essential for two main reasons. First, a solid understanding of who responds best to practice

PAGE 27

27 would enab le the empirically guided selection of individuals best suited to partake in cognitive interventions such that individuals with cha racteristics known to be related to optimal performance gains through practice might be identified and encouraged to partake in such activities Secondly identification of factors and individuals unlikely to respond to intervention would allow for potential remediation through modifiable factors. For example, if it is discovered that individuals with poor mood or lengthy wake time during the night respond poorly to cognitive practice, one could envision that efforts to improve mood and sleep While researchers are interested in individual differences in response to cognitive training (and presumably practice related learning), sleep has not been widely explored in the gerontological literature on this topic. Sleep represents an intriguing individual difference variable as it has been shown to relate to various as pects of cognitive functioning [ e.g., (Pilcher & Huffcutt, 1996)] and has demonstrated an ability to be malleable well into the later years of life [ e.g., ( Past research has revealed that sleep is related to lev el of cognitive functioning in older adults (Blackwell et al., 2006; Tworoger et al., 2006) long term decline in late life cognitive functioning (Cricco, Simonsick, & Foley, 2001) and learning in children (O'Brien, 2009) and adults (Maquet, 2001) These studies have generally indicated that more/better sleep is associated with positive outcomes. However, the sleep cognitive functioning relationship in late life is not without controversy. While retrospective recall of sleep typically yields the expected results [i.e., worse/less sleep

PAGE 28

28 associated with poorer cognitive functioning (Tworoger et al., 2006)], older adults with poor sleep do not routinely exhibit poorer performance on neuropsychological measures [e.g., verbal fluency, vigilance, etc.) than non complaining older adults (Altena, Van Der Werf et al., 2008; Altena, Werf et al., 2008)]. Thus, this study was unique in its methodology (i.e., intense repeated measurement) and outcome (i.e., practice related learning) in its examination of the sleep c ognitive functioning relationship in old age. The second overall goal (i.e., Paper 2/Chapter 3) of this document wa s to examine sleep as an individual difference predictor of practice related learning in late life. Specific aims of this paper include d exa sleep relates to : (1) their initial level of cognitive performance, and ( 2 ) their ability to benefit from repeated exposure to cognitive stimuli (i.e., ability to demonstrate practice related learning) after controlling for salient individual difference variables (i.e., age, education, estimated IQ, etc.). There is currently much ambivalence in the literature regarding the precise nature of the association between sleep and cognition, particularly in later life. While certa in aspects of sleep have been experimentally associated with learning [ i.e., (Maquet, 2001)] practice. This research could result in potential necessity for pre cognitive interventions to address sleep to enable older adults to exhibit maximal gains from cognitive practice.

PAGE 29

29 Paper 3: Do Weekly Deviations in Sleep Impact Subsequent Cognitive Fun ctioning in Older Adults? In addition to the examination of late life learning (i.e., practice related learning) and sleep as an individual difference predictor of late life learning the third paper in this dissertation switches to a very different level analysis. Specifically, in the cognitive aging literature, there has been growing attention to the topic of short term fluctuation in cognitive performance in late life and its possible sources. Short term fluctuation an important indicator of overall health and well being, including cognitive health (Hultsch et al., 2008) Some investigators have argued that short term fluctuation in cognitive performance may be indicative of unde rlying dysfunction ; correspondingly, increased short term fluctuation in cognitive performance has been shown to be related to poorer overall performance on many cognitive tasks (Hultsch et al., 2002; Hultsch et al., 2000; Nesselroade & Salthouse, 2004; Sa lthouse, Nesselroade, & Berish, 2006) Further, it has been found that individuals with known neurological disorders display more short term fluctuation in cognitive performance than non demented older adults (Burton, Hultsch, Strauss, & Hunter, 2002; Sliwinski, Hofer, & Hall, 2003) One retrospective study recently reported that individuals who were closer to death also displayed more sho rt term fluctuation in cognitive performance in addition to poorer levels of overall functioning (MacDonald, Hultsch, & Dixon, 2008)

PAGE 30

30 The parent study from which the data for these papers was drawn was optimally designed to investigate short term fluctua tion in cognitive functioning (i.e., cognitive inconsistency, cognitive intraindividual variability) Multiple repeated assessments of cognitive functioning permitted us to examine week to To underst and week to week variations in cognition, one must also examine concurrent weekly fluctuations in predictors. Extant investigations into understanding and predicting cognitive inconsistency in late life have largely focused on stress (Sliwinski, Smyth, Ho fer, & Stawski, 2006) and blood pressure (Gamaldo et al., 2008) These investigations found that o n days when a stressful event occurred, older adults had worse attention/concentration performance. On rapersonal mean, they also performed worse cognitively. Given that sleep quality is associated with improvements mood (McCrae et al., 2008) and blood pressure (Endeshaw, White, Kutner, Ouslander, & Bliwise, 2009; Silva, Moreira, Bicho, Paiva, & Clara, 2000) in older adults it would seem reasonable to extrapolate that better recent sleep may also be related to subsequent cognitive functioning in older adults. As with the under studied role of sleep in predicting learning in late r life, so too sleep has been under investigated as a potential source of cognitive inconsistency. With the exception of work done by Christina McCrae (NIH AG0244591) and Gamaldo and colleagues (2010) virtually no aging studies have done careful ex aminations of the occasion to occasion variations in sleep and how they are coupled with

PAGE 31

31 occasion to occasion variations in cognitive functioning (Dzierzewski, 2007; Gamaldo, Allaire, & Whitfield, 2010) The work by McCrae and colleagues, however, has most ly focused on older adults with disordered sleep while Thus, the current set of studies offers the possibility of extending our knowledge of the sleep cognition coupling to a normal comm unity based sample of the aging population The third overall goal (i.e., Paper 3/Chapter 4) of this dissertation was therefore to examine various dynamic association s (i.e., mean level and week to week) between sleep and short term changes in cognition i n late life. Specific aims of this paper include: (1) To explore the amount of short term fluctuation in cognitive functioning across an 18 week period, and (2) To explore various different temporal associations (i.e., mean level and weekly deviations in s leep) that may exist between self reported sleep and cognitive functioning in older adults. In Paper 3, we further detailed the aspects of sleep that may be particularly germane to understanding inconsistency in cognition. Short term fluctuation in cogniti ve functioning has been related to a multitude of poor outcomes in late life (Hultsch et al., 2008). If sleep predicted fluctuations in cognition, sleep treatment (which should produce more consistent sleep) might constitute one available approach for prac titioners if regulating cognitive functioning becomes a future treatment goal. Thus, the results of the third paper may result in treatment recommendations involving sleep that in

PAGE 32

32 addition to improving sleep as a primary outcome may have meaningful secon dary benefits in reducing the in Summary The study from which th is dissertation was derived repeatedly assessed cognitive functioning across an 18 week lifestyle intervention in a sample of older adults (Aiken Morgan, 2008; Buman, 2008; Buman et al., 2011) Thus, this study can provide insight into both the magnitude of practice related change (and its predictors), as well as short term fluctuation in cognitive performance. Both type of changes (practice related learning and short term fluctuation) may have tasks (Burton et al., 2009; Hultsch et al., 2008; Willis et al., 2006). In general, this dissertation attempt ed to examine two o ut of the three types of late life cognitive change (1) late life learning and (2) short term fluctuation in late life cognitive functioning. The papers that follow examine d the structure, predictors and potential consequences of practice related learning in late life, aberrant week of sleep may affect a subsequent cognitive functioning. The results of each paper independently are likely to offer considerable new evidence r egarding the magnitude, nature, and predictors of learning and cognitive inconsistency in older adults. Future implications of the work will relate to both a deeper understanding of how effectively practice functions as a cognitive intervention strategy (i n terms of producing broad cognitive enhancement effects), and how sleep might predict the ability to profit from practice interventions.

PAGE 33

33 CHAPTER 2 INTENSIVE COGNITVE P RACTICE IN OLDER ADU LTS: GAINS, STRUCTUR E, PREDICTORS, AND TRAN SFER. Introduction As a ge increases, many aspects of cognitive functioning (i.e., processing speed, reasoning, attention, executive functioning, and memory) tend to decrease. In cross sectional studies, these changes appear to begin in the mid twenties; however, cognitive declin e (where cognitive losses become discernible to oneself or others) is typically considered a problem of late life (Salthouse, 2004) While decline may be considered a prominent feature of aging, cognitive functioning remains plastic well into late life (He rtzog et al., 2008) In adults as old as the eighth and ninth decade of life, achieving cognitive improvements through training appears to remain normatively possible (Ball et al., 2002) Common findings from the literature on intervention related learning in late life [summarized by (Hertzog et al., 2008) ] suggest that late life learning : (a) can be substantial, (b) demonstrates limited transfer or generalizability to untrained skills/abilities [ e.g., (Willis et al., 2006)] (c) can be maintained long term [ e.g., up to 5 7 years (Unverzagt et al., 2009)] and (d) is reduced in older adults with dementias [ e.g., (Unverzagt et al., 2007)]. When discussing late life learning, it is crucial to distinguish between learning as a result of focused cognitive interv entions (i.e., intervention related learning) and learning due to repeated practice (i.e., practice related learning). This distinction is complicated by the fact that all cognitive training interventions contain substantial practice components. With seve ral notable exceptions [ i.e., (Baltes et al., 1989; Hofland et al., 1981; Yang & Krampe, 2009; Yang et al., 2006; Yang et al., 2009)] the vast majority of the late life learning literature has considered practice related improvements in cognitive

PAGE 34

34 function ing as a nuisance in need of methodological and statistical control Understanding practice related learning (i.e., learning independent of focused intervention) can shed adults can achieve on their own, without formal training. Consequently, practice related related learning, mea ning that questions remain regarding (1) the magnitude of gains associated with repeated practice, (2) the effects of practice on underlying cognitive constructs (do improvements on a test generalize to other measures of related abilities?), (3) predictors of practice related effects (do untrained abilities also improve). The overarching goal of this first paper was an in depth investigation into practice related learning in late life. Specifically, this paper sought to replicate previous studies (Baltes et al., 1989; Yang et al., 2006) that have found that practice can result in significant gains in the cognitive performance of older adults. Subsequently, this paper investigated w hether the effects of practice differ by cognitive domain (i.e., does practicing speed tasks yield more or less gain than practicing reasoning tasks?) and by individual characteristics of participants. Lastly, this paper turned to an investigation of poten tial transfer of practice related gains in cognitive functioning. As used in this study, transfer refers to the effects of learning on one task on another task (McArdle & Prindle, 2008). As practice is a key component of focused cognitive interventions [se e (Ball et al., 2002; Jaeggi et al., 2008; Jennings & Jacoby, 2003; Mahncke et al., 2006) for examples of focused cognitive interventions that contain substantial practice

PAGE 35

35 components], it is often difficult to disentangle the effects of strategy, feedback, adaptive difficulty, etc. versus the effects of repeated exposure to demanding cognitive exercises. Practice Related Learning Gains In one of the earliest examinations of practice related learning, it was found that two measures of reasoning ability, fig ural relations and inductive reasoning both display ed significant improvements (i.e., 0 .75 standard score increase in performance) when practiced for eight sessions by older adults age 60 84 years old (Hofland et al., 1981) With each subsequent practic e session, older adults attempted and correctly answered more items, with no increase in commission errors even though item difficulty increased. Thus, in that study, repeated practice appear ed to result in increased speed of responding and ability to answ er increasingly difficult items (Hofland et al., 1981) Yang, Krampe, and Baltes (2006) investigated practice related learning in measures of processing speed, reasoning, and visual attention across 6 retest session s in a sample of older adults 70 91 ye ars of age and reported that practice related learning was substantial. In general (across all three cognitive domains), the average improvement from initial retest to the last retest was a 1standard score gain in performance (Yang et al., 2006) Yang and Krampe (2009) subsequently retested available subjects from this study of practice learning [i.e., (Yang et al., 2006)] to examine maintenance of practice learning in older adults (Yang & Krampe, 2009) They reported that older adults were able to maintain over 50% of their original practice gains following an eight month delay. Interestingly, the authors reported that the maintenance in reasoning ( 0 .45 standard score) was comparable to the immediate training gain ( 0 .48 standard score) in tutor guided train ing in ACTIVE 10 sessions (Ball et al., 2002) or the estimated age related decline over a 14 year period (Yang & Krampe, 2009)

PAGE 36

36 Given that practice related learning may be the result of simple memorization of specific test items (Salthouse et al., 2004) Yang and colleagues (2009) examined practice related learning in the absence of item specific effects by creating parallel forms of all practice measures (Yang et al., 2009) Administration of the alternate forms (in reasoning, processing speed, and atte ntion) occurred over 8 practice sessions. All three domains demonstrated significant learning. The authors examined the effects of item specific learning by comparing their results to those achieved in previous work (Yang et al., 2006) in which the same me asures were admin i stered repeatedly instead of alternate forms. They found that only reasoning learning was significantly greater when same items were repeatedly administered, and thus concluded that older adults are able to demon strate significant gains ( i.e., 0 .33 1.35 standard score increases) in cognitive functioning that do not appear the result of mere memorization (Yang et al., 2009) An important caveat to note, however, is that even when the specific content of items changes, the algorithm for so lution may not. For example, in a series reasoning m b a n b d r q e r item content has changed, the specific algorithms and the general understanding of the need for algorithms may not have changed. Said differently, there may be algorithm specific learning tha t looks like more generalizable learning because it appears to be free of identical items; however, this would not be learning that will generalize to tasks with other solution algorithms. As such, distinguishing learning effects from memorization effects appears difficult.

PAGE 37

37 Baltes and colleagues (1986) noticed that a control group in a tutor guided cognitive intervention (i.e., ADEPT Protocol) which received only pre testing post testing and two follow up assessments of figural relations and inductive rea soning exhibited significant, and comparable, gains in cognitive functioning as compared to the intervention group (i.e., 0.85 standard score increase). However, the practice control group was outperformed by the tutor guided intervention group on the most difficult items (Baltes, Dittmann Kohli, & Kliegl, 1986) Baltes, Dittmann Kohli, and Kliegl (1988) performed a follow up study to compare three groups. Eighty seven older adults aged 60 86 years were assigned to receive either (1) 10 hours of tutor gui ded training (2) 10 hours of retest only practice on figural relations and inductive reasoning or (3) a no contact control condition. Both training conditions demonstrated significant gains in both inductive reasoning and figural relations Interestingly the practice group also demonstrated significant gains in two additional transfer measures of perceptual speed (that the tutor guided group did not). When item difficulty was examined, the practice group outperformed the training group on medium difficul ty items; the two groups did not differ on easy or hard items. However, the training group had better accuracy (i.e., lower error rates) on half of the outcome measures than did the practice group. The authors concluded, based on the preponderance of the e vidence (i.e., practice group had better transfer, better performance on medium difficulty items, and that tutor training contains practice), practice (Baltes, Kliegl, & Dittmann Kohli, 1988). Additional work has examined the effects of cognitive training and extended practice on cognitive functioning (Dittmann Kohli, Lachman, Kliegl, & Baltes, 1991) The

PAGE 38

38 authors compared four groups: (1) a no contact control group; (2) a pre test only group; (3) a group that received 10 sessions of tutor guided training, and (4) a group that received 10 session s of no feedback practice Participants were 116 health y adults aged 63 89 years old They reported that both the training and practice groups signif i cantly cognitively outperformed the other two groups on all criterion tasks (Dittmann Kohli et al., 1991) I n a nother head to head comparison of five session intervention related learning versus five session practice related learning in 72 older adults ag ed 63 90 years, investigators reported that older adults receiving practice alone gained equally as much as older adults who received tutor guided training in figural relations performance (i.e., approximately 0.5 standard score increase) (Baltes et al., 1989) Further, no differences emerged between the practice and tutor groups on accuracy, number of errors, and ability to answer questions of varying difficulty levels (i.e., easy, medium, hard). Again, o be capable of producing gains by themselves that were comparable to those following tutor (Baltes et al., 1989) Thus, practice related learning appears to be substantial in size (and may be comparable to intervention related learning co gnitive improvements). In fact, many of the effect sizes reported from practice studies are comparable to those found in focused cognitive interventions [i.e., ACTIVE intervention (Ball et al., 2002)]. Practice Related Learning Structure The underlying g oal of late life cognitive intervention (whether through active training or practice) is to engender noticeable changes in both the specific tasks being trained and the constructs underlying these specific tasks (McArdle & Prindle, 2008). As such, the goal of cognitive training is not only to improve performance on specific

PAGE 39

39 cognitive tasks (e.g., processing speed tasks), but rather the goal is to also improve general cogniti ve functioning. In this sense, the structure of practice related learning is of importance. When considering tutor guided cognitive interventions, results typically demonstrate very narrow ability gains [i.e., training in reasoning produces gains in reas oning and not in other domains (Ball et al., 2002)]. This is likely due to the domain focused nature of the interventions employed. However, because practice does not require specialized instruction, participants who enroll in practice studies typically ar e able to practice in multiple cognitive domains. Thus, the potential for broader gains in multiple domains may actually be greater. To date, there are no known studies of practice related learning that examined whether generalized learning or underlying constructs improved with practice. However, there is indirect evidence rega r ding the structure of practice related learning. Hofland and colleagues (1981) reported that gains in both figural relations and inductive reasoning displayed constant linear impro vements with no sign of asymptote being reached across eight testing sessions (Hofland et al., 1981) Interestingly, others have reported significant linear and quadratic ( signifying potential diminishing gains) trends in practice related learning of reaso ning, processing speed, and attention across six sessions (Yang et al., 2006) In a follow up examination Yang, Reed, Russo, and Wilkinson (2009) reported that reasoning, processing speed, and attention all demonstrated significant linear improvements acro ss 8 testing sessions; however, only processing speed demonstrated a significant quadratic trend (Yang et al., 2009) Thus,

PAGE 40

40 there is some evidence suggesting that practice related learning typically follows a pattern of linear improvements across retest se ssions (sometimes demonstrating a generalized learning. I n addition to simple descriptions of the overall shape of practice related learning, it is useful to also examin e the rate of practice related learning. Yang and colleagues (2009) examined the rate of practice related learning in reasoning, processing speed, and attention They reported that while attention and processing speed did not differ from each other in lear ning rate (although it was marginal, p = .09), they both displayed a steeper rate of learning than reasoning (over twice as steep) (Yang et al., 2009) Such results suggest that practice may affect cognitive domains distinctly. Several practice related le arning investigations have repeatedly administered multiple measures within a single cognitive domain. Yang and colleagues administered multiple measures of reasoning and examined practice related learning in a composite measure (simple average) of reasoni ng (Yang et al., 2006) They reported steady linear and quadratic time trends in this composite measure. In a subsequent investigation multiple measures of both reasoning and speed were administered (Yang et al., 2009) Simple composites (formed through a veraging) demonstrated significant linear (reasoning) and linear and quadratic (speed) time trends. Additionally, it was reported that practice related learning across the domains of reasoning, processing speed, and attention are highly correlated (i.e., c orrelations among practice gains of 0.66 0.79, all p <0.01) (Yang et al., 2006) These shared trajectories seem to open the possibility that some broader trans ability (general cognitive) learning may have occurred. The present

PAGE 41

41 study sought to extend th ese findings by (a) examining the extent to which a single common trajectory (latent growth model) can be fitted to multiple practice trajectories, and (b) whether there is evidence that individual differences in practice related improvement on a specific set of measures transferred to other, unpracticed measures (i.e., correlated gain). Practice Related Learning Predictors repeated cognitive practice could shed much needed lig ht on the potential mechanisms underlying practice related learning. Further, it could also aid in the empirically guided selection of individuals into cognitive practice interventions. It has been reported that practice related learning in processing spee d, reasoning, and visual attention is substantial in the young old (70 79 years) and oldest old (80 91 years) (Yang et al., 2006). However, younger age and higher levels of initial functioning have both been associated with increased rates of practice rela ted learning. For reasoning both lower age and higher functioning is associated with increased practice related learning. For processing speed only younger age was associated with improved practice related learning. For attention, neither age nor level of functioning was found to be associated with practice related learning (Yang et al., 2006). Additionally, older adults with higher levels of cognitive functioning do not benefit more from practice than those with lower levels of cognitive functioning (Yang et al., 2006). Similarly, it was reported that maintenance of practice related gains following an 8 month delay differed according to age classification (young old vs. oldest old) (Yang & Krampe, 2009). The young old maintained slightly, but significantly more than did the oldest old. Additionally, individuals who gained more during initial practice also

PAGE 42

42 maintained the most during follow up ( correlations between initial gain and maintenance of .52 0.63) (Yang & Krampe, 2009). Both state anxiety (Hofland et al., 1981) and trait anxiety (Yang et al., 2009) have been examined as a potential predictors of response to practice; however, neither state nor trait anxiety significantly contributed to or explained the patterns of practice related learning in late life. Thus, while evidence from a handful of studies by Yang and colleagues have identified several important individual differences variables in practice related learning, additional research is need to further explicate these relationships. When focuse d cognitive interventions are examined for predictors of training gains, mixed results are reported. Results from the largest clinical trial of cognitive interventions in late life [i.e., ACTIVE (Ball et al., 2002)] suggests that most demographic variables (i.e., age, sex, education, visual functioning, and mental status) are not related to training related gains (Ball et al., 2002). However, subsequent analysis did reveal that specific impairment in the trained domain does reduce training gains in that dom ain [i.e., memory impairment impact memory training gains (Unverzagt et al., 2007)]. However, other work has demonstrated largest training related gains in individual with the lowest initial ability levels (Jaeggi et al., 2008). Thus, even in the realm of focused cognitive interventions in late life, much remains to be explored regarding individual difference predictors of training response. Practice Related Learning Transfer Transfer of learning can be defined as the influence of learning one skill on the performance of another skill (Salomon & Perkins, 1989; Woodworth & Thorndike, 1901). Transfer can be further examined in terms of distance along a continuum from near to far transfer. Simplistically stated, near transfer refers to the transfer of learning to tasks

PAGE 43

43 that are similar in structure, content, and the processes required to complete them, and far transfer refers to transfer of learning to tasks that are different in structure, content, and the processes required to complete them [i.e., (Blieszner, Willis, & Baltes, 1981; Salomon & Perkins, 1989)]. As such, transfer is more likely to occur for near measures as opposed to far measures. While the concept of transfer has received much attention in the cognitive training literature [e.g., (Hertzog et al ., 2008)], results are not very promising. With several notable exceptions [i.e., (Jaeggi et al., 2008; Smith et al., 2009; Willis et al., 2006)], transfer of training in cognitive interventions appears to be the exception rather than the norm. While th e most recent examinations [i.e., (Yang & Krampe, 2009; Yang et al., 2006; Yang et al., 2009)] of practice related learning have not examined potential transfer of learning, the foundational works examining practice related learning [i.e., (Baltes et al., 1986; Baltes et al., 1988; Baltes et al., 1989; Dittmann Kohli et al., 1991; Hofland et al., 1981) have all included assessment of transfer. In examining transfer of learning through a tutor guided cognitive intervention Baltes and colleagues (1986) repor ted that the control group (which only received practice) demonstrated comparable transfer of learning to near ability measures of figural relations and reasoning ( B altes et al., 1986). In a subsequent study it was found that equivalent time spent in pract ice as spent in tutor guided training produced identical patterns of transfer to near ability measures (again in figural relations and reasoning). Additionally, the practice group displayed additional transfer to processing speed measures that were not dem onstrated by the tutor guided group (Baltes et al., 1988). Dittmann Kohli and colleagues (1991) reported that ten sessions of practice produced

PAGE 44

44 more near transfer as compared to ten session of tutor guided training. However, tutor guided training produced transfer to self efficacy beliefs while practice did not (Dittmann Kohli et al., 1991). Five sessions of practice and tutor guided training have been shown to produce equivalent gains in near transfer measures of figural relations and inductive reasoning ( Baltes et al., 1989). While it appears that transfer of practice related learning to near ability tasks is possible, definitive conclusions regarding transfer of practice related learning to novel tasks/skills cannot be drawn from the extant literature due to limited investigations and mixed results. The Current Investigation Previous research has revealed that older adults can achieve substantial gains from practice related learning (Baltes et al., 1989; Hofland et al., 1981; Yang & Krampe, 2009; Yang et al., 2006; Yang et al., 2009) and can maintain those gains for relatively long time frames (Yang & Krampe, 2009). To our knowledge, no known research has examined whether practice related learning has an effect on generalized learning in older adults; thu s, conclusions regarding the nature of practice related learning cannot be drawn. While both young old and oldest old, as well as high and low functioning older adults, respond to practice with substantial gains (Yang et al., 2006), it is unknown how other practice. Similarly, the paucity of research attempting to demonstrate transfer of practice related learning (Baltes et al., 1989) precludes any definitive conclusions. Kno wn practice related learning investigations have utilized relatively brief numbers of practice sessions [eight (Hofland et al., 1981; Yang et al., 2009), five (Baltes et al., 1989), and six sessions (Yang & Krampe, 2009; Yang et al., 2006)]. Thus, question s remain regarding older adults response to more intensive practice (and the gains,

PAGE 45

45 structure, predictors, and transfer within a more intensive practice paradigm), and, optimal practice dosage needed to achieve cognitive gains. The current investigation seeks to address four specific aims: (1) To replicate the presence of significant practice related learning in both speed and reasoning abilities in older adults, ( 2 ) To examine whether a higher order practice related learning construct is evident (i.e., g eneralized learning) and whether such a construct can account for practice related learning in individual measures (i.e., What was the effect of practice related learning on overall or generalized learning?) (3) To examine common individual difference pr edictors (i.e., age, education, and mood) of practice related learning (i.e., What predicted practice related learning?), and (4) To examine whether individual differences in practice related learning predict changes in unpracticed measures (i.e. Does prac tice related learning transfer to unpracticed tests?) We hypothesized that: (1) Practice related learning would be substantial in both processing speed and reasoning [based on prior research (Baltes et al., 1986; Baltes et al., 1988; Baltes et al., 1989; Yang et al., 2006; Yang et al., 2009)]; (2) Repeated practice would engender changes generalized learning; (3) Age, education, and mood would predict response to practice; and (4) Practice related learning would transfer to near ability measures but not fa r ability measures. Methods General Study Design This study represents a secondary analysis of the Active Adult Mentoring Program (Project AAMP). The primary objective of Project AAMP was to test the efficacy of a social cognitive lifestyle intervention t o increase moderate intensity exercise in older adults. Participants were randomly assigned to either an Active Lifestyle intervention

PAGE 46

46 arm (receiving weekly, group based behavioral counseling) or a Health Education arm (receiving appropriately matched heal th education). The current study utilizes data from the initial 18 weeks of the study, including a baseline week of observation prior to group assignment, 16 weeks of intervention, and a subsequent week of observation following the intervention period. The study protocol was approved by the appropriate university institutional review boards. Procedure Individuals who expressed interest in study participation were initially screened by telephone (see below). Following telephone screening, qualified particip ants were consented and completed a baseline assessment. This baseline assessment included both computerized cognitive assessment and traditional paper pencil cognitive assessment of both practice cognitive measures (i.e., measures that were subsequently a ssessed weekly) and transfer cognitive measures (i.e., cognitive measures that were only reassessed at post treatment). Next, participants were randomized to either the Active Lifestyle or Health Education arm of the intervention. Each intervention arm con sisted of sixteen weekly group meetings [ see (Buman, 2008 ; Buman et al., 2011) for more information]. Either before or after each group meeting all participants completed a computerized cognitive battery that included all practice measures (see below). Las tly, each participant completed a post treatment assessment that included both computerized cognitive assessment and traditional paper pencil cognitive assessment of both practice and transfer cognitive measures. Thus, this study included 18 consecutive we eks of cognitive practice, paired with pre/post cognitive assessment of potential transfer measures.

PAGE 47

47 Participants Study participants include 90 adults aged 50 years and greater who participated in Project AAMP. Potential enrollees responded to community b ased health promotion recruitment delivered through local media outlets. To ensure their suitability for the study, subjects went through a thorough screening process that included many inclusion and exclusion criteria. Inclusion/Exclusion Criteria All po tential participants were screened by telephone to exclude individuals based on the following criteria: severe dementing illness, history of significant head injury (loss of consciousness for more than 5 minutes), neurological disorders disease), inpatient psychiatric treatment, extensive drug or alcohol abuse, use of an anticholinesterase inhibitor (such as Aricept) severe uncorrected vision or hearing impairments, terminal illness with life expectancy less than 12 months, major medical illnesses cardiovascular disease, pulmonary disease requiring oxygen or steroid treatment, and ambulation with assistive devices. Telephone screening include d the 11 item Telephone Interview for Cognitive Status (TICS ), with a cut off score of 30 points being employed to differentiate mild dement ia from cognitive ly intact (Brandt, Spencer, & Folstein, 1988) All study participants were required to self report sedentary lifestyle 0 minutes/week of moderate or vigorous physical activity during the previous 6 months ( Physical Activity Guidelines Advisory Committee Report, Part A: Executive Summary 2009) ]. A note from primary care physician s acknowledging ability to participate in th e study prior to formal enrollment was required

PAGE 48

48 Demographic and Descriptive Measures Each study participant provided demographic data through means of a telephone screening instrument. Information regarding participant age (measured in years since birth) gender (male or female), and education level (years of education ) were collected. During the first in person session, all individuals completed the North American Adult Reading Test (NAART) (Blair & Spreen, 1989) yielding a pre morbid IQ estimate (that has been found to correlate between 0.40 and 0.80 with other measures of intelligence). The Beck Depression Inventory, Second Edition [BDI II; (Beck, Steer, & Brown, 1996) ] was administered to assess depressive symptomatology. The BDI II consists of 21 gro ups of statements related to cognitive and somatic depression symptoms. The BDI II is a commonly used self report measure of depression in both younger (Beck et al., 1996) and older adults (Segal, Coolidge, Cahill, & O'Riley, 2008) T he State Trait Anxiety Inventory [ STAI ; (Spielberger, 1983) ] was administered to assess current (state) and typical (trait) anxiety symptoms. The STAI consists of 40 statements to which participants respond based on the degree to which they feel like each statement either Descriptive Statistics The final sample included 90 adults aged 50 years and older. Mean age for the entire sample was 63.56 years, range = 50 86 years. The sample was highly educated, average years of education of 16.12, predominate ly female, 82%, of above average intelligence, and evinced few depressive and anxiety symptoms (i.e., few depressive and anxiety symptoms). Please refer to Table 2 1 for a complete list of demographic/descriptive statistics.

PAGE 49

49 Measures Cognitive measures pr esented below are organized into two main subheadings: (1) Practiced Cognitive Measures, and (2) Transfer Cognitive Measures. Please refer to Table 2 2 for a schematic of practiced measures, and near and far transfer measures. Practice cognitive measures w ere administered once per week for 18 consecutive weeks. Transfer cognitive measures were administered twice, once at pretest and once at posttest (and are further subdivided into near and far transfer measures). Near transfer measures represent tasks that rely on similar underlying abilities as practiced measures. Far transfer measures are tasks that tap into unpracticed abilities. Practiced Cognitive Measures All weekly cognitive measures (except Simple Reaction Time and Complex Reaction Time) were sli ghtly modified to allow for computer administration. Computerization of the cognitive measures was done using DirectRT experimental generation program (Jarvis, 2008a) All computerized cognitive measures were then compiled and administered via MediaLab exp erimental implementation program (Jarvis, 2008b). In an attempt to minimize practice effects due to memorization commonly found in repeated cognitive assessments (Salthouse et al., 2004) fourteen alternate forms of each test were used and rotated such tha t the same version of any given test was not given within 6 weeks of each other. The alternate forms were constructed to be comparable in difficulty and cognitive resources needed to complete them and have been shown to have high test retest reliabilities (i.e., Letter Series = 0.95 0.98; Symbol Digit = 0.40 0.91) (Allaire & Marsiske, 2005; McCoy, 2004) Processing speed and reasoning domains were selected as practiced cognitive domains due to their inclusion

PAGE 50

50 in previous practice related learning investigat ions (Yang & Krampe, 2009; Yang et al., 2006; Yang et al., 2009). Processing speed Processing speed was practiced with the Symbol Digit and Digit Copy tests (Smith, 1982). These tests consist of matching symbols that are paired with nu mber s (Symbol Digit) or numbers paired with same numbers (Digit Copy) as quickly as possible There was a 120 second time limit for each task and the performance score was the total number of correct pairings made by the participant Processing Speed was also practiced with the Simple and Choice Reaction Time task (Hultsch et al., 2000) The Simple Reaction Time (SRT) task present ed of the screen. Participants were instructed to press a key with their preferred hand as quickly as possible when the signal stimulus (+) appears. The Choice Reaction Time (CRT) task present ed to the left and right of the center of the screen. Af d into a circle and the location of the circle was randomly equalized across trials. Respondents were instructed to press a key corresponding to the location of the circle as quickly as possible. Ten practice trials we re fo llowed by 50 test trials. The outcome measure s for SRT and CRT were the mean latenc ies for correct test trials. Executive processing Executive processing was practiced with the Letter Series task (Thurstone, 1962) which is primarily a measure of reasonin g In this task, participants ha d to identify the pattern for a series of letters. Participants we re asked to choose the letter that would continue an established pattern (A B D A B D A B ___?) in a series of letters from five

PAGE 51

51 answer choices. Participants were given four minutes to complete as many items as possible. The performance score was the number of correct responses. Near Transfer Cognitive Measures Transfer is conceptualized as the influence of learning one skill on the performance of another skill (Salomon & Perkins, 1989; Woodworth & Thorndike, 1901). Transfer measures described below are those cognitive measures that were not practiced; that is, they were only assessed at pre and posttest. Transfer can be further examined in terms of distance alo ng a continuum from near to far. Near transfer refers to the transfer of learning to measures that are similar in structure, content, and the processes required to complete them. Far transfer refers to transfer of learning to measures that are different in structure, content, and the processes required to complete them (Blieszner et al., 1981; Salomon & Perkins, 1989). Processing Speed To assess near transfer for processing speed (i.e., tasks that assess processing speed) the Trail Making Test A and B was administered (Reitan, 1992) This task required individuals to connect circles containing numbers, and numbers and letters as quickly as possible. The outcome measure s were the time to complete the tasks. Near transfer for processing speed was also assess ed with the N Back Task (1 back and 2 back conditions) The N Back was computerized via DirectRT (Jarvis, 2008a) and administered via MediaLab (Jarvis, 2008b). In the 1 back condition, participants had to judge whether the current letter matches the immedi ately preceding letter. In the 2 back condition, participants had to judge whether the current letter matches the letter presented 2 letters previously, as quickly as possible. As such, it was a measure of speed of processing (Cohen et al., 1994). The outc ome measure s were the average

PAGE 52

52 reaction times for correct responses in the 1 Back and 2 Back conditions Both Trail Making Test A and B and N Back (1 back and 2 Back) represent near transfer measures of processing speed due to their heavy reliance on speed of responding. Executive Processing To assess near transfer for executive processing (i.e., tasks that measure executive processes) the Letter Number Sequencing subtest of the WMS III (Wechsler, 1997) which is a working memory tasks requiring mental mani pulation and serial rearrangement of numbers and letters presented aurally, was administered The outcome measure was the total number of correct sequences recalled. The Control Oral Word Association task (COWA) was used to assess phonemic fluency (Benton & Hamsher, 1989) which is considered an executive function. Participants were read a proper nouns beginning with that letter as possible within sixty seconds. The outcome measure was the total n umber of correct words produced for each letter The above described measures are all near transfer measures for executive processing due to the demanding cognitive skill set required for their completion and their common reliance on frontal lobe functioni ng. Far Transfer Measures One far transfer tasks (i.e., a task that measures neither processing speed nor executive processing) was administered. The Logical Memory subtest of the WMS III (Wechsler, 1997) measures verbal memory. Two short stories were rea d to participants. Free recall of the stories occurred immediately after story presentation and again 30 minutes later. One of the two stories was presented twice to measure learning. Delayed recall and recognition trials for each story wer e completed foll owing a 25 35 minute

PAGE 53

53 delay interval. The outcome measure s are the total number of propositions produced at immediate recall and delayed recall. Analyses Preliminary Analyses Prior to statistical analyses to address each main aim, preliminary analyses were conducted. Preliminary analyses: (1) examined normality of the data, (2) examined rates of missingness among the data, (3) examined attrition (differences between attrited and non attrited subjects), and ( 4 ) examined whether the proposed structure of the cognitive variables (i.e., processing speed, executive processing, and near and far transfer) held true across time. Preliminary analysis examined the data for normality. All data were screened for outliers at the intraindividual level (i.e., within perso n from trial to trial). Further, all cognitive variables were also screened for potential outliers at the interindividual level (i.e., between persons). Interindividual outliers were replaced with their respective 3 standard deviation values, while intrain dividual outliers were simply removed from the dataset prior to calculation of occasion specific values. Skewness and kurtosis values were also examined using generally agreed upon criteria (i.e., skewness and kurtosis values less than 1.0) (Field, 2005). Rates of missingness were calculated as a descriptor of the data, because all available data was utilized. All structural equation modeling (SEM) was implemented through AMOS (Arbuckle & Wothke, 1999) via full information maximum likelihood (FIML) estimati on methodology. Analysis of attrition was conducted to examine whether there were differences between study completers (those whom were present at posttest) and study attriters (those whom dropped out before reaching posttest) on all cognitive and descrip tive/demographic variables.

PAGE 54

54 The proposed structure of the cognitive variables was examined via a n exploratory factory analysis (EFA) for all cognitive variables obtained during baseline testing ( i.e., practiced and unpracticed measures ). In order to deter mine if the structure of the cognitive variables was maintained throughout the course of the study, a subsequent confirmatory factor analysis (CFA) was conducted with all cognitive variables obtained during posttest testing ( i.e., practiced and unpracticed measures ) with the baseline structure imposed. As such, the EFA suggested cognitive structure was estimated at posttest, and the fit was evaluated. The parent study (Project AAMP) to the current paper was a test of the efficacy of a social cognitive li festyle intervention to increase moderate intensity exercise in older adults. Participants were assigned to either an Active Lifestyle intervention or a Health Education. As such, concerns regarding differing cognitive trajectories between these groups may be present. However, previous reports have reported few baseline to posttest fitness effects by group status (Buman, 2008) and no baseline to posttest cognitive effects by group status (Aiken Morgan, 2008). Thus, group status was not controlled for in sub sequent analyses. Main Analyses To determine if practice related learning was evident in each individual repeated cognitive measure (A im 1), separate univariate growth curves were estimated. Both linear and quadratic estimates of change were examined. Sig nificant practice related learning was indicated by tests of the slope being non zero. Further, random slopes and intercepts were estimated and were examined to determine if the slope related or intercept related variance was non zero. There must be signif icant individual differences (i.e., random variance) in the slopes and/or intercepts prior to estimation of models that

PAGE 55

55 predict either an intercept or slope term. Model fit was evaluated by examining the following fit statistics: 2 (chi squared), ratio of 2 value to degrees of freedom (CMIN/DF), comparative fit index (CFI), Tucker Lewis index (TLI), incremental fit index (IFI), and root mean square error of approximation (RMSEA). Good model fit was indicated by models with non si gnificant chi square statistic, CMIN/DF < 2, CFI, TLI, and IFI > 0.90, and RMSEA < .08. To examine whether practice related learning in individual cognitive measures results in gains in generalized learning (Aim 2) a h igher order latent growth model (curve of factors ) was estimated. Initially, a factor of curves model was proposed; however, this model was not identifiable and it has been reported that factor of curves and curve of factors models typically produce very similar results while addressing compar able questions (Duncan, Duncan, & Strycker, 2006) This model explored if at each assessment point, for each cognitive measure that was practiced, the measures could be made to load on a single common occasion specific factor, and then whether these common occasion specific factors could be made to load on a common intercept and linear and quadratic slope. If the unique univariate measures load on common time factors that load on an overall intercept and linear and quadratic slope it would be suggestive tha t practice related learning in the individuals tasks results in a common (or generalized) gain in cognitive functioning. Significant higher order practice related learning was indicated by tests of the common factor slopes being non zero. Model fit was eva luated by examining common fit statistics (i.e., 2 CMIN/DF, CFI, TLI, IFI, and RMSEA). Please refer to the dashed center circle in Figure 2 1 for a graphical representation of this analysis.

PAGE 56

56 Frequently used predictors of individual difference in cognition were next added to the model built in Aim 2 to examine their predictive utility on practice related learning in late life (Aim 3). That is, Aim 3/Analysis 3 explored predictors of practice related learning. The variables of age, education, depression, and state anxiety were examined. It was initially h ypothesized that the relevant predictors of growth to study would include gender, predicted IQ, and trait anxiety. In the models that follow, these terms are not included because such models did not converge. As these variables were found to have little pr edictive value or overall effect on model fit, they were removed from further analyses. A single model in which all four predictors were estimated concurrently was estimated, as there was no hypothesized hierarchical ordering among the predictors. Each va riable was evaluated based on the significance of their paths to (a) common level practice intercept, (b) common level practice linear slope, and (c) common level practice quadratic slope. Significant paths from these exogenous variables to common level in tercept and slopes indicate significant prediction of practice related learning. Variables were also be evaluated based on their overall consequences to model fit statistics (i.e., 2 CMIN/DF, CFI, TLI, IFI, and RMSEA). Please refer to the solid lower rec tangle in Figure 2 1 for a graphical representation of this analysis. To examine any potential transfer of practice related learning to non practiced cognitive measures (Aim 4) the above described transfer measures were added to the model built in Aim 3. T ransfer measures were allowed to have correlated residuals if improvement in model fit resulted from such correlated residuals. Transfer measures were calculated as simple pre/post change scores. Each transfer measure (both near

PAGE 57

57 and far transfer measures) became a dependent variable that was predicted by (a) the common level linear slope and (b) the common level quadratic slope. Significant paths from either slope indicated significant transfer of learning. Paths from the common level slopes were evaluated based on their significance levels and overall effects on model fit (i.e., 2 CMIN/DF, CFI, TLI, IFI, and RMSEA). Please refer to the double lined upper rectangle in Figure 2 1 for a graphical representation of this analysis. Results Preliminary Analyses Normality Descriptive statistics revealed the presence of numerous sample outliers and suggested data were both skewed and kurtic. An iterative approach to outlier trimming resulted in data that were normally distributed. Please refer to Table 2 3 for descriptive and normality statistics pre and post outlier trimming procedures and for raw cognitive scores for each measure at each occasion. In total, 118 variables were cleaned for 90 subjects. This produced a total of 10,620 potential data points that could have been trimmed. Only 87 actual data points were trimmed; resulting i n 0.82% of the data being altered. Following data cleaning procedures, all RT based data (i.e., N Back, Simple and Choice RT, and Trails A and B) were transformed such that higher scores were indicative of better performance. This was done to place all cog nitive measures on a similar metric and to improve interpretability of the main study Aims. Missing Data Ninety subjects were enrolled in the current investigation. Of these 90 subjects, twenty two (or approximately 25% of the baseline sample) did not com plete post testing. As is typical for intraindividual variability studies, throughout the repeated

PAGE 58

58 assessments, participation waxed and waned considerably. In general, participation was at its highest during the earlier weeks of assessment. Week 1 had the lowest percent of missing data with just over 12% missing. In contrast, participation was at its lowest at the end of the repeated assessments. Week 15 had the highest rate of missing data with just over 49% missing. Overall, the average level of missing d ata was approximately 31%, while the median amount of missing data was also approximately 31% missing. Please refer to Figure 2 2 for a graphical representation of missing data across the study period. Attrition As is shown in Table 2 4, comparing demogra phic/descriptive characteristics of individuals who completed posttest (n = 68) to those that failed to complete posttest (i.e., attriters; n = 22), independent samples t reported significantly more depressive sym ptoms than did study completers, t (88) = 2.41, p = 0.02. Study completers and attriters did not significantly differ in their performance on any baseline cognitive measure, all > 0.05. Full information maximum likelihood (FIML) estimation methodology was employed in all subsequent analyses, as such methodology has been shown to produce reliable estimates even when data are not missing at random and is preferable to listwise deletion and purposive estimation (Graham, 2009). Data Structure As a prelimina ry exploration of the relatedness of the baseline cognitive variables a set of bivariate correlations were ran among practiced and non practiced (i.e., transfer) cognitive measures. These correlations generally supported a high degree of associations among the baseline cognitive measures. Please see Table 2 5 for a

PAGE 59

59 complete list of correlation coefficients among baseline cognitive variables. An exploratory factor analysis (EFA) with principal axis factoring and Promax rotation with Kaiser normalization was conducted to examine the structure of all cognitive variables at baseline assessment. Four factors had eigenvalues greater than the recommended level of 1 (Tabachnick & Fidell, 2001); however, interpretability, loadings of at least .30, and percent varianc e explained in the indicators all suggested a three factor explanation to be parsimonious. All three factors had eigenvalues greater than 1 (4.75, 1.82, and 1.27, respectively) and explained 60.32% of the variance between measures (36.55%, 14.01%, and 9.76 %, respectively). While the variance explained by the three factor solution fell short of recommended guidelines (Gorsuch, 1983), interpretability of structure loadings suggested the three factor model to be the most desirable solution. Please refer to Tab le 2 6 for the pattern matrix of factor loadings from the three factor solution. In general, the proposed structure of the cognitive data, with processing speed and executive processing (both practiced and unpracticed) measures along with far transfer meas ures (i.e., memory measures), was confirmed. To explore the stability of the cognitive structure across the repeated assessments a confirmatory factor analysis (CFA) was conducted on the posttest data. Model parameters were estimated using maximum likelih ood. Evaluation of model fit was done through commonly used fit statistics including chi square statistic, ratio of chi square value to degrees of freedom (CMIN/DF), comparative fit index (CFI), Tucker Lewis index (TLI), incremental fit index (IFI), and ro ot mean square error of approximation (RMSEA). The three 2 (57) = 51.15, p < .05,

PAGE 60

60 the 2 Back, and as depicted in Figure 2 3, fa ctor loadings (ranging from .34 to .95) and explained variance (ranging from .12 to .90) were good. As such, it appears that the factor structure was maintained from baseline to posttest. Main Analyses Initially, parameterization of the latent growth curv es (LGC), one for each repeatedly assessed cognitive variable, was planned with each individual assessment occasion (i.e., 18 total occasions) being modeled. However, preliminary analyses revealed that these models, without exception, failed to converge an d thus did not provide robust estimates. As such, composite cognitive scores composed of blocks of adjacent occasions were computed and used in all subsequent analyses of Aims 1 4. Blocks were constructed in the following manner: Block 1 = baseline, week 1, and week 2; Block 2 = week 3, week 4, week 5, and week 6; Block 3 = week 7, week 8, week 9, and week 10; Block 4 = week 11, week 12, week 13, and week 14; and Block 5 = week 15, week 16, and posttest. The blocks were constructed in this manner so that each block would contain roughly equal numbers of occasions and to allow the earliest occasions (i.e., occasions 1 3) and the latest occasions (i.e., occasions 15 17) to have slightly more influence on their respective block scores. Five blocks were co nstructed as this number of occasions is in accordance with other higher order latent growth models published in the psychology and aging literature (Christensen et al., 2004; Hofer et al., 2002). Please see Table 2 7 for mean performance on each cognitiv e measure at each block. Aim 1: Individual Growth Curves All latent growth curve analysis results are presented in Table 2 8. Please refer to this table for all intercept and slope estimates, and correlation coefficients between

PAGE 61

61 intercept and slope estima tes. Additionally, please see Figure 2 4 for a graphical representation of standardized model implied change in each cognitive measure across the study period. Number Copy. Fit statistics indicated good model fit for the Number Copy LGC 2 (10) = 8.79, p = 0.00. The mean intercept was 42.56, p < .01, suggesting that the average Block 1 starting value was s ignificantly greater than zero. The mean estimate of the linear slope was 0.13, p > .05, suggesting non significant linear growth across blocks in Number Copy performance. However, the mean estimate of the quadratic slope was 1.85, p < .01, suggesting sig nificant quadratic change across blocks in Number Copy performance. Specifically, significant growth was observed during the first two blocks of assessment, followed by decelerated growth throughout the end of the study period. The intercept related varian ce was 16.23, p < .01, suggesting significant individual differences in Block 1 Number Copy performance. The linear slope related variance was 0.57, p > .05, suggesting non significant individual differences in linear change in Number Copy performance. How ever, the quadratic slope related variance was 6.23, p < .05, suggesting significant individual differences in quadratic change in Number Copy performance. Symbol Digit. Fit statistics indicate d good model fit for the Symbol Digit LGC 2 (10) = 13.99, p RMSEA = 0.067. The mean intercept was 23.44, p < .01, suggesting that the average Block 1 starting value was significantly greater than zero. The mean estimate of the linear slope was 0.03, p > .05, suggesting non significant linear growth across blocks in

PAGE 62

62 Symbol Digit performance. However, the mean estimate of the quadratic slope was 2.24, p < .01, suggesting significant quadratic c hange across blocks in Symbol Digit performance. Specifically, significant growth was observed during the first two blocks of assessment, followed by decelerated growth throughout the end of the study period. The intercept related variance was 19.07, p < 01, suggesting significant individual differences in Block 1 Symbol Digit performance. The linear slope related variance was 1.82, p < .05, suggesting significant individual differences in linear change in Symbol Digit performance. Lastly, the quadratic sl ope related variance was 9.38, p < .05, suggesting significant individual differences in quadratic change in Symbol Digit performance. Letter Series. The LGC model for Letter Series would not properly converge with correlations to the quadratic slope bein g estimated. This was likely due to negative quadratic slope variance. As a result, the correlations between the intercept and linear slope and the quadratic slope were not estimated and the quadratic slope variance was set to zero. Fit statistics indicate d adequate model fit for the Letter Series LGC model. 2 (13) = 43.81, p RMSEA = 0.16. The mean intercept was 7.85, p < .01, suggesting that the average Block 1 starting value was sign ificantly greater than zero. The mean estimate of the linear slope was 0.14, p > .05, suggesting non significant linear growth across blocks in Letter Series performance. However, the mean estimate of the quadratic slope was 2.14, p < .01, suggesting signi ficant quadratic change across blocks in Letter Series performance. Specifically, significant growth was observed during the first two blocks of assessment, followed by decelerated growth throughout the end of the study period.

PAGE 63

63 The intercept related varian ce was 10.82, p < .01, suggesting significant individual differences in Block 1 Letter Series performance. The linear slope related variance was 0.47, p < .05, suggesting significant individual differences in linear change in Letter Series performance. Las tly, as stated above, the quadratic slope related variance was set to zero. Simple RT. The LGC model for Simple RT would not properly converge with correlations between the intercept, linear slope and quadratic slope being estimated. As a result, these co rrelations were not estimated. Fit statistics indicated good model fit for 2 (13) = 13.73, p > .05, CMIN/DF = 1.06, CFI, TLI, and IFI p < .01, suggesting that the averag e Block 1 starting value was significantly greater than zero. The mean estimate of the linear slope was 6.79, p < .05, suggesting significant linear change across blocks in Simple RT performance. The mean estimate of the quadratic slope was 26.54, p < .01 suggesting significant quadratic change across blocks in Simple RT performance. Specifically, significant growth was observed during the first two blocks of assessment, followed by a leveling off throughout the end of the study period. The intercept rela ted variance was 2466.41, p < .01, suggesting significant individual differences in Block 1 Simple RT performance. The linear slope related variance was 85.22, p < .05, suggesting significant individual differences in linear change in Simple RT performance Lastly, the quadratic slope related variance was 181.90, p > .05, suggesting non significant individual differences in quadratic change in Simple RT performance.

PAGE 64

64 Choice RT. Fit statistics indicated good model fit for the Choice RT LGC model. 2 (10) = 5.08, p 0.00. The mean intercept was 353.81, p < .01, suggesting that the average Block 1 starting value was si gnificantly greater than zero. The mean estimate of the linear slope was 17.15, p < .01, suggesting significant linear change across blocks in Choice RT performance. The mean estimate of the quadratic slope was 54.53, p < .01, suggesting significant quadr atic change across blocks in Choice RT performance. Specifically, significant growth was observed during the first two blocks of assessment, followed by a slight decline throughout the end of the study period. The intercept related variance was 4557.28, p < .01, suggesting significant individual differences in Block 1 Choice RT performance. The linear slope related variance was 349.68, p < .05, suggesting significant individual differences in linear change in Choice RT performance. Lastly, the quadratic slo pe related variance was 2604.88, p < .05, suggesting significant individual differences in quadratic change in Choice RT performance. Aim 2: Curves of Factors The Curve of Factors model was fit with both a linear and quadratic slope. The Curve of Factors model would not converge due to negative deviance related variances at Block 1 and Block 5. As a result, these variances were set to zero. Fit statistics indicated adequate model fit for the model 2 (240) = 433.90, p < .001; however, CMIN/DF = 1.81, CFI a nd IFI = 0.92, TLI = 0.89. Lastly, RMSEA = 0.095. Each block of each measure significantly loaded on its respective overall block, all < .005. The mean intercept was 369.03 p < .01, suggesting that the average Block 1 starting value was significantly greater than zero. The mean estimate of the linear slope was 8.49 p < .01, suggesting significant linear growth across blocks in overall generalized learning

PAGE 65

65 The mean estimate of the quadratic slope was 53.76 p < .01, suggesting significant quadratic ch ange across blocks in generalized learning Specifically, significant growth was observed during the first two blocks of assessment, followed by a leveling off throughout the end of the study period. The intercept related variance was 2317.34 p < .01, sug gesting significant individual differences in Block 1 general cognitive performance. The linear slope related variance was 37.96 p < .05, suggesting significant individual differences in linear change in generalized learning Lastly, the quadratic slope r elated variance was 554.71, p > .05, suggesting non significant, though marginal, individual differences in quadratic change in generalized learning Please refer to Table 2 9 for all intercept and slope estimates, correlation coefficients between intercep t and slope estimates and standardized loadings for each measure to each general block Please see Figure 2 5 for a graphical representation of model implied change in overall generalized learning across the study period. See Figure 2 6 for the standardiz ed coefficients in model path diagram. Aim 3: Curves of Factors: Individual Differences The goal of Aim 3 was to examine common individual difference predictors of the general cognitive functioning level (i.e., common intercept) and change in generalized learning (i.e., common slopes) modeled in the previous analysis (i.e., Aim 2). In order to examine the predictive utility of widely used individual difference variables on the level and rate of change in generalized learning age, education, depressive sym ptomology, and state anxiety were added to the previously parameterized curve of factors model. In addition to the negative deviance related variances at Block 1 and Block 5 that were observed in Aim 2, the current model had convergence problems due to neg ative

PAGE 66

66 deviance related variance at Block 2. As a result, Blocks 1, 2, and 5 variances were set to zero. 2 (337) = 576.27, p < .001; however, CMIN/DF = 1.71, CFI and IFI = 0.90, TLI = 0.87. Lastly, RMSEA = 0.089. The level of the common factor was associated with age ( 0.50, p < .001), education (0.19, p < .05), an d depressive symptoms ( 0.30, p < .05). This indicated that the intercept of the common factor was lower in older adults, less educated adults, and older adults with more depressive symptoms. The linear slope of the generalized learning factor was not asso ciated with any of the individual difference variables. The quadratic slope of the generalized learning factor was associated with education ( 0.40, p < .05) and state anxiety symptoms (0.41, p < .05). This indicated that older adults with more years of ed ucation experienced less pronounced deceleration in generalized learning, while older adults with more state anxiety related symptoms experience increased deceleration in generalized learning. Please refer to Table 2 10 for a listing of standardized coeffi cients predicting the common level and change in generalized learning Please see Figure 2 7 for the complete path diagram modeling individual differences in common level and change in generalized learning See Figure 2 8 for a path diagram of standardized loadings and significance levels of the individual difference predictors. Refer to Figure 2 9 for a graphical representation of generalized learning for both a high (i.e., 90 th %ile) and low (i.e., 10 th %ile) educated and state anxious subject. Aim 4: Cur ves of Factors: Transfer The goal of Aim 4 was to examine potential transfer of the practice related gains in generalized learning to changes in unpracticed cognitive measures (i.e., do common

PAGE 67

67 slopes predict pre/post changes in unpracticed measures?). In o rder to examine the potential transfer of generalized learning, simple pre post change scores in the unpracticed cognitive measures were added to the previously parameterized curve of factors model in Aim 3. Paths were subsequently drawn from the common in tercept, linear and quadratic slopes to the pre/post change scores. Like in previous models, negative deviance related variances at Block 1, 2, and 5 were present. As a result, Blocks 1, 2, and 5 variances were set to zero. Fit statistics indicated adequat 2 (572) = 944.23, p < .001; however, CMIN/DF = 1.65, CFI = 0.85, IFI = 0.86, and TLI = 0.82. Lastly, RMSEA = 0.085. There were no statistically significant associations between either the common intercept or common slope and any pre/post chang e scores. However, there were significant associations between the common level quadratic slope and pre post change in 1 Back RT (0.91, p < .01) and 2 Back RT (0.36, p < .05). This indicated that subjects whom experienced more positive growth in generalize d learning also experience greater levels of pre post improvement in 1 Back and 2 Back RT. This transfer of practice related learning represents near transfer for processing speed. Please refer to Table 2 11 for a listing of standardized coefficients for t ransfer in generalized learning to non practiced measures Please see Figure 2 10 for the complete path diagram modeling transfer as a result of change in generalized learning See Figure 2 11 for a path diagram of standardized loadings and significance le vels of transfer. Refer to Figure 2 12 for a graphical representation of generalized learning transfer for both a high (i.e., 80 th %ile) and low (i.e., 20 th %ile) change 1 Back RT and 2 Back RT subject.

PAGE 68

68 Discussion The current study investigated improvemen ts in cognitive functioning in older adults that were the result of repeated practice with cognitive stimuli (i.e., practice related learning). Specifically, this study sought to: (a) replicate the findings of significant practice related learning in late life, (b) investigate if practice in specific cognitive measures resulted in improvements in overall cognitive functioning/generalized learning, (c) determine if individual difference variables predict improvements in any generalized learning observed, and (d) examine whether improvements in generalized learning led to subsequent transfer of learning to non practiced cognitive measures. It was found that both processing speed measures (i.e., Number Copy, Symbol Digit, Simple RT, and Choice RT) and an execu tive processing measure (i.e., Letter Series) demonstrated significant gains associated with repeated practice. Further, the significant practice related gains observed in these individual measures were found to result in generalized learning. Specificall y, it was possible to extract a common growth model that suggested the existence of correlated growth across all the measures; individuals who gained more in one ability were more likely to gain in another. Regarding predictors of generalized learning, it was found that both education and state very limited near transfer for processing speed was observed. Practice Related Learning The first overall goal of the current inv estigation was to replicate previous research demonstrating the presence of significant practice related learning in late life. Previous investigation has demonstrated significant practice related learning in figural relations

PAGE 69

69 and inductive reasoning (i.e. .75 standard score increase in performance) across eight practice trails (Hofland et al., 1981) processing speed, reasoning, and visual attention gains (i.e., 1standard score gain in performance ) across 6 retest session s (Yang et al., 2006), reasoning, processing speed, and attention (i.e., .33 1.35 standard score increases) over 8 practice sessions (Yang et al., 2009), figural relations and inductive reasoning improvements ( i.e., 0.85 standard score increase) across 4 practice sessions, and figural re lations improvements (i.e., approximately 0.5 standard score increase) over 5 testing sessions (Baltes et al., 1989) The current investigation successfully replicated these previous findings. Significant practice related learning (i.e., 0.48 1.40 standa rd score increases) was observed in processing speed and executive processing measures over 18 practice sessions. As Baltes and colleagues (1986, 1988) have reported, practice related learning has been shown to compare favorably to tutor guided cognitive intervention in head to head trials (Baltes et al., 1986; Baltes et al., 1988). In line with these previous investigations, the gains demonstrated in the current study (i.e., 0.48 1.40 standard score increases) are comparable to the average immediate tra ining gain s (.48 standard score) in tutor guided training in ACTIVE 10 sessions (Ball et al., 2002) and are larger than the estimated age related decline over a 14 year period (Yang & Krampe, 2009) Of importance, due to speculation that practice related learning may be the result of mere item memorization by older adults (Salthouse et al., 2004), in the current investigation no specific test item was repeated within 6 weeks of itself. As such, it appears highly unlikely that the practice related gains ob served are the result of item memorization. Our results corroborate those of Yang and colleagues (2009) whom also

PAGE 70

70 found significant practice related learning in the absence of item specific effects (Yang et al., 2009) While it cannot be guaranteed that th ere were no item specific learning effects in the current study, the number of items presented and the lag between same items greatly reduced the likelihood of such learning. Practice Related Learning Structure There is little previous knowledge regarding the effects of cognitive interventions, through either practice or tutor guided instruction, on generalized learning. Further, the knowledge that does exist in this domain is mixed. In general, tutor guided cognitive interventions typically result in very narrow ability gains (Ball et al., 2002). This has been speculated to be a result of the domain focused nature of the interventions employed. As practice does not require specialized instruction and is therefore not necessarily domain specific (i.e., prac tice can occur across many domains), ability gains may be more dispersed. As such, the potential for gains in general cognitive functioning remain. come from practice related learning investi gations that have reported that growth in various different cognitive constructs (i.e., figural and inductive reasoning, processing speed, attention, etc.) all appear to occur in similar fashion (Hofland et al., 1981; Yang et al., 2006). That is, it has be en reported that separate cognitive constructs all improve with linear and quadratic trends across repeated testing sessions. While this is only a cursory way to comment on the potential effects of practice on generalized learning, it does suggest that pra ctice potentially results in common changes to general functioning. In the current investigation, we progressed beyond noting similar trajectories to the finding of correlated trajectories; that is, as individuals improved on some cognitive measures, so to o did they improve on all other cognitive measures. Similarly to these previous

PAGE 71

71 investigations, the current study also reported that across 5 distinct processing speed and executive processing measures practice resulted in linear and quadratic improvements in functioning. Further evidence of the effects of practice on generalized learning come from Yang and colleagues (2006, 2009), who similarly showed that cognitive composites demonstrate significant practice related learning (Yang et al., 2009), and that gains across composite measures are highly correlated with one another (Yang et al., 2006) The current investigation directly examined the question of whether or not practice related learning engenders changes in overall (or generalized) cognitive functi oning through modeling change at the latent factor level. In this sense, the combined effect of repeated cognitive practice of multiple cognitive tests on generalized learning was directly modeled. Significant linear and quadratic growth was noted in gener alized learning as a result of repeated practice. The generalized cognitive growth was substantial in size (i.e., 0.84 standard score increase), with the vast majority of gains occurring within the first 4 weeks of practice (i.e., 0.77 standard score incre ase in performance). As such, it appears that generalized learning can be strengthened with as little as 4 weeks of practice. Further, lengthier practice seems to offer little additional bonus in terms of improved cognitive functioning. However, such a cla im is deserving of future investigations. The goal of late life cognitive interventions is to produce changes in both the specific tasks being training and overall cognitive functioning (McArdle & Prindle, 2008). The current study found evidence of genera lized learning resulting from cognitive practice in older adults. As such, future investigations/interventions aimed at improving

PAGE 72

72 late life cognitive functioning may do well to consider utilization of a cognitive practice paradigm. Not only is practice eas y to implement (due to the absence of tutors and specific techniques), it also appears to affect general cognitive functioning in late life. This effect on general cognitive functioning may have important functional and quality of life implications, though such questions are in need of further investigation. Practice Related Learning Predictors Previous research has examined predictors of response to test specific practice related learning in older adults. In this realm it has been reported that both the y oung old and old old demonstrate practice related learning in processing speed, reasoning, and visual attention (Yang et al., 2006). However, age and level of initial functioning have been found to be differentially related across these various cognitive d omains (Yang et al., 2006). Recently, it was reported that reasoning, spatial visualization, episodic memory, perceptual speed and vocabulary show significant practice related gains, but more pronounced gains in younger individuals (Salthouse, 2010). A ddit ionally, neither state nor trait anxiety significantly predict ed the patterns of practice related learning in late life (Hofland et al., 1981; Yang et al., 2006). As no previous research has directly modeled changes in gains that transcend multiple measu res, so too no previous investigation has attempted to examine individual difference predictors of changes in generalized learning as a result of practice. However, if task specific practice related improvements are related across measures (i.e., generaliz ed learning), then it stands to reason that individual difference predictors found to be related to task specific improvements may also be related to generalized learning improvements. The current investigation found that individuals with higher levels of education displayed more constant improvements in generalized learning than

PAGE 73

73 those with lower levels of educations. Further, and contrary to previous examinations in task specific practice related learning, individuals with increased levels of state anxiety were found to show a decreased rate of generalized learning. Selection of subjects optimally suited to participate in practice related cognitive interventions in late life may be guided by the current results: individuals with higher levels of education and low levels of anxiety appear best suited to gain maximal benefits from practice. While educational attainment (which is most typically attained in early adulthood) is not something that can be altered/intervened in during late life, anxiety appears ver y modifiable in older adults (Barrowclough et al., 2001; Stanley, Beck, & Glassco, 1996; Wetherell, 1998). As such, it appears warranted to recommend that prior to enrollment in practice related cognitive interventions in late life older adults should be s creened appropriately for anxiety symptoms and, if present, should receive appropriate evidence based treatments to ameliorate the anxiety prior to engaging in cognitive practice. The findings that educational attainment and anxiety are related to practic e related improvements in generalized learning may shed light on the potential mechanisms underlying practice related learning. Educational level is thought to be a proxy for reserve, which is the ability to optimize or maximize performance (Stern, 2002). Reserve is hypothesized to serve as a protective agent to late life cognitive decline (Stern, 2006). As such, educational level may not only be a protective agent against cognitive decline it may also be an active ingredient in cognitive improvement. As ed ucation is a proxy for reserve, it is suggestive that practice related improvements in generalized learning may be due to either brain reserve or optimization of neural

PAGE 74

74 networks and/or recruitment of alternative networks (Stern, 2002, 2006). General anxiet y and state anxiety have been shown to be negatively related to cognitive functioning in younger (Elliman, Green, Rogers, & Finch, 1997) and older adults (Beaudreau & O'Hara, 2009; Wetherell, Reynolds, Gatz, & Pedersen, 2002). Theorists speculate that anxi and intrusive thoughts that compete for cognitive resources (i.e., inhibition and attention deficits) (Eysenck & Calvo, 1992; Eysenck, Derakshan, Santos, & Calvo, 2007). As anxiety was r elated to lower gains in practice related learning, it may be that changes in generalized learning rely on attentional and inhibitory processes. Additional research is needed to further explicate the mechanisms underlying practice related changes in genera lized learning Practice Related Learning Transfer Transfer of learning in a major goal of nearly all cognitive interventions. However, it is a goal that is seldom achieved (Hertzog et al., 2008). With very few exceptions [i.e., (Jaeggi et al., 2008; Smit h et al., 2009; Willis et al., 2006)] tutor guided interventions have not demonstrated transfer of learning. Conversely, several recent investigations of practice related learning investigations [i.e., (Yang & Krampe, 2009; Yang et al., 2006; Yang et al., 2009)] have not attempted to examine potential transfer. However, the earliest investigations into practice related learning [i.e., (Baltes et al., 1986; Baltes et al., 1988; Baltes et al., 1989; Dittmann Kohli et al., 1991; Hofland et al., 1981)] all inve stigated transfer and demonstrated narrow transfer to near abilities. This demonstrated transfer represents transfer from practiced specific tasks to non practiced specific tasks of the same underlying ability.

PAGE 75

75 The current investigation examined transfer from practice related gains in generalized learning to non practiced measures of both near and far abilities. It was found that individuals whom showed more constant improvements in generalized learning also demonstrated the highest levels of pre/post impr ovements in 1 Back and 2 Back RT. As such, near transfer from generalized learning to a processing speed generalized learning through simple repeated practice may be a viable mea ns to improving other areas of cognitive functioning. [Note: This set of analyses was hampered by a reduced N. Please see Limitations Section below for further discussion of transfer in the current investigation.] Limitations There are several limitations to the current study that need to be recognized. The current study was embedded within a larger trial of a lifestyle intervention (Aiken Morgan, 2008; Buman, 2008; Buman et al., 2011). There was no effect of intervention group on cognitive performance acr oss the entirety of the study. However, all subjects were given a free gym membership for the length of the investigation. As physical activity and physical fitness has been shown to be related to cognitive functioning in late life (McAuley, Kramer, & Colc ombe, 2004), it is possible that some of the variance associated with growth in generalized learning is related to physical activity/fitness levels, and/or changes in physical activity/fitness. While previous studies have demonstrated an absence of relati onships between changes in physical activity/fitness and cognition in the current sample (Aiken Morgan, 2008), the lack of such relationships does not suggest the lack of other relationships between physical activity/fitness and cognition (i.e., correlated changes across time, etc.).

PAGE 76

76 Another limitation is related to the practiced and unpracticed cognitive measures employed in the study. As demonstrated by the EFA/CFA, only two cognitive domains were assessed repeatedly (i.e., processing speed and executive processing). It would have been advantageous to examine multiple other cognitive domains in the context of practice related learning. Further, far transfer was assessed by a single measure (i.e., Logical Memory) which resulted in two indices of verbal mem ory. Again, it would have been advantageous to assess more than one potential far transfer measure. Lastly, related to the cognitive assessments, the number of specific measures assessing each cognitive domain was extremely unbalanced. It would have been b eneficial to have multiple executive processing measures practiced so that practice related learning could have been examined at the individual, construct, and overall levels. Another major limitation of the current investigation was its relatively small sample size for its complex analytical approach. This limitation bares direct consequences to some of the reported results and interpretations. For example, only one instance of significant transfer of learning was detected. The standardized path coefficie nts in structural equation modeling are analogous to Pearson's r (Tabachnick & Fidell, 2001). Examination of the coefficients observed in the analysis of transfer suggests that 7 associations between change in generalized learning and pre/post changes in t ransfer measures were of medium effect sizes or larger (Cohen, 1992). As such, it appears as though the current investigation (particularly Aim 4 which had an even smaller N due to pre/post attrition) was hindered by a small sample size which may have affe cted power and the ability to detect significant results. In fact, studies employing similar analytic techniques (i.e., Factor of Curves) with much larger samples have resulted in significant

PAGE 77

77 associations of much weaker strength than those reported in the current study (Christensen et al., 2004). However, we believe the density/richness of the current dataset out weighs the limitations imposed by reduced power. Future Directions Future investigations should examine individual difference predictors of task specific practice related learning (Please see Chapter 3/Paper #2) in addition to predictors of generalized learning. Such examination would allow for the direct examination of potential differences in predictors of task specific versus generalized cogniti ve change. Any differences observed may allow for targeted cognitive interventions for special populations of older adults (i.e., those with comorbid psychological disorders, those with low educational attainment, etc.). Furthermore, future work would be well suited to investigate the effects of improved generalized learning resulting from practice on both functional (i.e., driving, self care activities, falls, etc.) and quality of life (i.e., mood, self efficacy, interpersonal relations, etc.) outcomes. R esearch is need that examines the real world consequences, and hopefully benefits, of laboratory based interventions with older adults. Additionally, as the current investigation reported significant gains resulting from short term practice (i.e., 4 weeks) and little additional gains from continued practice, the optimal length of practice should receive further investigation. Older adults should be randomized to varying lengths of cognitive practice to examine not only gains in cognitive functioning, but al so individual difference predictors and potential transfer resulting from differing lengths of cognitive practice. The addition of a tutor guided comparison group would allow for compelling comparisons to be made. Direct comparison between older adults re

PAGE 78

78 cognitive gains, transfer, and impact on functioning. Without such a direct comparison, such statements are hypothetical and subject to severe limit ations. Lastly, future investigation should recruit a large, diverse sample of older adults. Obtainment of a larger sample would allow for robust tests of potential transfer and predictors of gain. Further, without the limitations on modeling resulting fr om a restricted sample, estimation of the relationships between intercept and slopes for the higher order cognitive construct (i.e., generalized learning) may be possible. Such estimations would allow for examination of whether individuals who initially ha d better general cognitive functioning also demonstrated the most gains in generalized learning. Additionally, future investigations would likely benefit from the collection of multiple indicators of both near and far transfer in addition to multiple indic ators of each cognitive domain assessed. Summary The current investigation examined the gains, structure, predictors, and transfer of cognitive practice in older adults. We confirmed the presence of significant practice related learning in late life and e xtended this work by demonstrating that practice in individual cognitive tasks can result in positive generalized learning. Furthermore, we ability to demonstrate generali zed learning as a result of practice. Lastly, very limited transfer was demonstrated to near ability measures of processing speed. Recommendations regarding screening and treatment of anxiety prior to enrollment in cognitive interventions were made. Potent ial mechanisms of reserve and attentional and inhibitory processing were identified and suggested for future investigation.

PAGE 79

79 In summation, it appears that older adults remain very plastic well into late life. Given the vast age range included in the presen t investigation (i.e., 50 years to 87 years) and the lack of association of age with change in generalized learning, it would appear that middle aged and older adults are capable of demonstrating the same benefits of cognitive practice. While this result i s in contrast to other recent studies demonstrating diminished practice effects with increasing age (Salthouse, 2010), our results were obtained on the correlated cognitive functioning level. As such, older adults may be less plastic than younger adults in specific cognitive domains; however, it appears they remain as capable of positive changes in cognitive functioning and as such, research into methods of improving cognition in late life appear to warranted.

PAGE 80

80 Table 2 1. Descriptive/ d emographic s tatistics Mean (SD) Age 63.56 ( 8.45 ) Education 16.12 ( 2.25 ) Gender 1.82 ( 0.38 ) NAART 113.17 ( 6.45 ) BDI II 6.52 ( 5.39 ) State Anxiety 30.3 ( 7.94 ) Trait Anxiety 31.54 ( 8.77 ) Notes: Age measured in years since birth; Education measured in years; Gender: 1 = male, 2 = female; NAART = premorbid IQ estimate; BDI II = Beck Depression Inventory, 2 nd Edition.

PAGE 81

81 Table 2 2. Measures Domain Practiced Measures Near Transfer Measures Far Transfer Measures Processing Speed Simple RT Choice RT Number C opy Symbol Digit Trails A and B N Back Logical Memory Executive Processing Letter Series Letter Number Sequencing COWA Notes: Domain is the hypothesized underlying construct common to all practiced and near transfer tests. Practiced measures were asses sed weekly for 18 consecutive weeks. Transfer measures were assessed pre/post only. Near transfer measures are those tests hypothesized to tap similar cognitive processes as the practiced cognitive measures. Far transfer measures are those tests hypothesiz ed to require unpracticed cognitive skills needed for completion.

PAGE 82

82 Figure 2 1 Path diagram depicting higher order latent growth model ( a im 2, center circles/large dashed circle), predictor variables ( a im 3, bottom rectangles/solid rectangle), and transfer learning ( a im 4, top rectangles/double lined rectangle). LS = l etter s eries; NC = n umber c opy; DS = d igit s ymbol; SRT = s imple r eaction t ime; CRT = c hoice r eaction t ime; BDI = Beck Depression Inventory ; STAI = State Trait Anxiety Inventory; Lette rNum = l etter n umber s equencing; COWA = Controlled Oral Word Association; LogicMem = l ogical m emory.

PAGE 83

83 Table 2 3. Descriptive, n ormality, and r aw c ognition v alues p rior to and p ost d ata c leaning. Measure Data c leaned Minimum Maximum Mean (SD) Skewness ( SE) Kurtosis (SE) Logical Memory Immediate pre No --------Yes 17.00 59.00 41.01 (8.73) 0.05 (0.25) 0.35 (0.50) Logical M emory Delay pre No --------Yes 6.00 44.00 25.91 (7.54) 0.07 (0.25) 0.39 (0.50) FAS p re No 16.00 70.00 39.49 (10.42) 0.22 (0.25) 0.10 (0.50) Yes 16.00 70.00 39.49 (10.42) 0.22 (0.25) 0.10 (0.50) Trails A pre No 17.00 73.62 31.58 (10.67) 1.35 (0.25) 2.67 (0.50) Yes 17.00 61.69 31.40 (10.07) 1.00 (0.25) 0.97 (0.50) Trails B pre No 33.3 2 237.85 78.09 (38.06) 2.18 (0.25) 6.06 (0.50) Yes 33.32 163.62 76.02 (30.85) 1.19 (0.25) 1.04 (0.50) Letter Number Sequencing pre No 5.00 20.00 10.73 (2.76) 0.93 (0.25) 1.15 (0.50) Yes 5.00 19.01 10.72 (2.72) 0.85 (0.25) 0.80 (0.50) 1 Back pre No 44 9.67 1638.32 867.60 (204.30) 0.97 (0.26) 1.57 (0.51) Yes 449.67 1460.19 865.59 (197.42) 0.72 (0.26) 0.45 (0.51) 2 Back pre No 298.05 4867.96 1715.46 (714.17) 1.90 (0.25) 6.16 (0.50) Yes 298.05 3593.69 1688.95 (613.57) 0.89 (0.25) 1.00 (0.50) Number C opy pre No 1.00 53.00 41.43 (6.72) 3.15 (0.25) 16.08 (0.50) Yes 27.08 53.00 41.86 (4.83) 0.73 (0.25) 0.95 (0.50) Number Copy week 01 No 24.00 88.00 43.91 (8.36) 3.10 (0.27) 16.54 (0.53) Yes 29.30 57.06 43.26 (4.90) 0.30 (0.27) 1.69 (0.53) Number C opy week 02 No 26.00 52.00 43.61 (4.57) 0.94 (0.28) 2.15 (0.55) Yes 29.90 52.00 43.66 (4.38) 0.62 (0.28) 0.68 (0.55) Number Copy week 03 No 29.00 89.00 45.00 (7.14) 3.74 (0.31) 24.26 (0.60) Yes 31.55 57.61 44.53 (4.45) 0.20 (0.31) 1.65 (0.60) Numb er Copy week 04 No 30.00 53.00 44.66 (3.88) 1.14 (0.30) 2.84 (0.60) Yes 34.68 53.00 44.74 (3.63) 0.69 (0.30) 0.95 (0.60) Number Copy week 05 No 27.00 52.00 43.90 (4.40) 1.03 (0.29) 2.37 (0.57) Yes 30.70 52.00 43.95 (4.20) 0.69 (0.29) 0.82 (0.57)

PAGE 84

84 Table 2 3. Continued. Measure Data c leaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Number Copy week 06 No 30.00 93.00 45.20 (7.13) 4.62 (0.30) 32.07 (0.59) Yes 33.15 55.22 44.67 (3.88) 0.32 (0.30) 0.81 (0.59) Number Copy week 0 7 No 33.00 54.00 45.23 (4.14) 0.64 (0.30) 1.36 (0.60) Yes 33.00 54.00 45.23 (4.14) 0.64 (0.30) 1.36 (0.60) Number Copy week 08 No 38.00 53.00 45.54 (3.69) 0.15 (0.31) 0.26 (0.61) Yes 38.00 53.00 45.54 (3.69) 0.15 (0.31) 0.26 (0.61) Number Copy week 09 No 36.00 56.00 45.16 (4.11) 0.10 (0.32) 0.23 (0.62) Yes 36.00 56.00 45.16 (4.11) 0.10 (0.32) 0.23 (0.62) Number Copy week 10 No 37.00 56.00 45.58 (3.84) 0.16 (0.31) 0.33 (0.61) Yes 37.00 56.00 45.58 (3.84) 0.16 (0.31) 0.33 (0.61) Number Copy week 11 No 3.00 57.00 45.09 (6.91) 4.34 (0.32) 26.27 (0.64) Yes 34.01 57.00 45.67 (4.04) 0.42 (0.32) 1.52 (0.64) Number Copy week 12 No 3.00 54.00 44.82 (6.50) 4.63 (0.31) 29.39 (0.61) Yes 34.50 54.00 45.34 (3.75) 0.57 (0.31) 0.92 (0.61) Nu mber Copy week 13 No 37.00 52.00 45.14 (3.31) 0.39 (0.34) 0.17 (0.66) Yes 37.00 52.00 45.14 (3.31) 0.39 (0.34) 0.17 (0.66) Number Copy week 14 No 33.00 54.00 45.82 (3.56) 0.62 (0.33) 2.59 (0.66) Yes 36.72 54.00 45.90 (3.32) 0.06 (0.33) 0.70 (0.6 6) Number Copy week 15 No 38.00 56.00 46.87 (3.66) 0.07 (0.35) 0.43 (0.69) Yes 38.00 56.00 46.87 (3.66) 0.07 (0.35) 0.43 (0.69) Number Copy week 16 No 36.00 55.00 45.78 (4.26) 0.43 (0.34) 0.02 (0.67) Yes 36.00 55.00 45.78 (4.26) 0.43 (0.34) 0.02 (0.67) Number Copy post No 34.00 56.00 45.85 (4.29) 0.50 (0.29) 0.33 (0.57) Yes 34.00 56.00 45.85 (4.29) 0.50 (0.29) 0.33 (0.57) Symbol Digit pre No 0.00 32.00 21.99 (6.30) 1.66 (0.25) 3.33 (0.50) Yes 6.38 32.00 22.22 (5.59) 1.14 (0.25) 1.14 (0.5 0) Symbol Digit week 01 No 14.00 48.00 24.49 (5.12) 2.07 (0.27) 9.15 (0.53) Yes 14.00 35.49 24.19 (3.95) 0.11 (0.27) 1.12 (0.53)

PAGE 85

85 Table 2 3. Continued. Measure Data c leaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Symbol Digit we ek 02 No 16.00 33.00 25.37 (3.55) 0.53 (0.28) 0.30 (0.56) Yes 16.00 33.00 25.37 (3.55) 0.53 (0.28) 0.30 (0.56) Symbol Digit week 03 No 9.00 45.00 26.26 (4.72) 0.04 (0.30) 5.75 (0.60) Yes 14.85 37.61 26.23 (3.99) 0.31 (0.30) 1.57 (0.60) Symbol Digi t week 04 No 15.00 32.00 26.19 (3.23) 0.82 (0.30) 1.28 (0.60) Yes 16.51 32.00 26.22 (3.15) 0.64 (0.30) 0.50 (0.60) Symbol Digit week 05 No 18.00 32.00 25.96 (3.37) 0.38 (0.29) 0.36 (0.57) Yes 18.00 32.00 25.96 (3.37) 0.38 (0.29) 0.36 (0.57) Sym bol Digit week 06 No 17.00 55.00 26.80 (4.95) 2.94 (0.30) 15.75 (0.59) Yes 17.00 36.83 26.52 (3.69) 0.48 (0.30) 0.76 (0.59) Symbol Digit week 07 No 22.00 34.00 26.52 (3.20) 0.36 (0.30) 0.93 (0.60) Yes 22.00 34.00 26.52 (3.20) 0.36 (0.30) 0.93 (0.60) Symbol Digit week 08 No 19.00 34.00 26.64 (3.44) 0.05 (0.31) 0.37 (0.61) Yes 19.00 34.00 26.64 (3.44) 0.05 (0.31) 0.37 (0.61) Symbol Digit week 09 No 10.00 33.00 26.61 (4.01) 1.27 (0.32) 3.86 (0.62) Yes 14.58 33.00 26.69 (3.71) 0.65 (0.32) 0.7 2 (0.62) Symbol Digit week 10 No 16.00 34.00 26.54 (3.54) 0.59 (0.31) 0.73 (0.61) Yes 16.00 34.00 26.54 (3.54) 0.59 (0.31) 0.73 (0.61) Symbol Digit week 11 No 0.00 35.00 26.63 (4.94) 2.80 (0.32) 15.26 (0.64) Yes 15.17 35.00 26.91 (3.67) 0.20 (0.3 2) 0.92 (0.64) Symbol Digit week 12 No 16.00 35.00 27.08 (3.40) 0.45 (0.30) 1.06 (0.60) Yes 16.87 35.00 27.09 (3.36) 0.34 (0.30) 0.70 (0.60) Symbol Digit week 13 No 21.00 34.00 27.04 (3.15) 0.06 (0.34) 0.46 (0.66) Yes 21.00 34.00 27.04 (3.15) 0.06 (0.34) 0.46 (0.66) Symbol Digit week 14 No 19.00 35.00 27.12 (3.24) 0.13 (0.33) 0.54 (0.66) Yes 19.00 35.00 27.12 (3.24) 0.13 (0.33) 0.54 (0.66) Symbol Digit week 15 No 21.00 37.00 28.37 (3.59) 0.08 (0.35) 0.38 (0.69) Yes 21.00 37.00 28.37 (3.59) 0.08 (0.35) 0.38 (0.69)

PAGE 86

86 Table 2 3. Continued. Measure Data c leaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Symbol Digit week 16 No 14.00 36.00 27.02 (4.07) 0.87 (0.34) 1.85 (0.67) Yes 15.03 36.00 27.04 (4.01) 0.76 (0.34) 1.47 (0.67) Symbol Digit post No 16.00 36.00 28.64 (3.93) 0.82 (0.29) 0.85 (0.57) Yes 16.84 36.00 28.65 (3.89) 0.76 (0.29) 0.59 (0.57) Letter Series pre No 0.00 13.00 6.02 (3.09) 0.09 (0.26) 0.60 (0.51) Yes 0.00 13.00 6.02 (3.09) 0.09 (0.26) 0.60 (0. 51) Letter Series week 01 No 1.00 19.00 8.85 (3.62) 0.15 (0.27) 0.15 (0.53) Yes 1.00 19.00 8.85 (3.62) 0.15 (0.27) 0.15 (0.53) Letter Series week 02 No 0.00 22.00 9.85 (4.64) 0.02 (0.28) 0.34 (0.55) Yes 0.00 22.00 9.85 (4.64) 0.02 (0.28) 0.34 ( 0.55) Letter Series week 03 No 2.00 22.00 10.89 (4.98) 0.10 (0.30) 0.54 (0.60) Yes 2.00 22.00 10.89 (4.98) 0.10 (0.30) 0.54 (0.60) Letter Series week 04 No 0.00 23.00 10.37 (5.02) 0.47 (0.30) 0.09 (0.60) Yes 0.00 23.00 10.37 (5.02) 0.47 (0.30) 0.09 (0.60) Letter Series week 05 No 0.00 21.00 8.84 (5.10) 0.17 (0.29) 0.34 (0.57) Yes 0.00 21.00 8.84 (5.10) 0.17 (0.29) 0.34 (0.57) Letter Series week 06 No 2.00 19.00 9.89 (4.00) 0.28 (0.30) 0.91 (0.59) Yes 2.00 19.00 9.89 (4.00) 0.28 (0.30) 0. 91 (0.59) Letter Series week 07 No 1.00 21.00 11.42 (3.90) 0.45 (0.30) 0.12 (0.60) Yes 1.00 21.00 11.42 (3.90) 0.45 (0.30) 0.12 (0.60) Letter Series week 08 No 1.00 21.00 11.54 (4.48) 0.24 (0.31) 0.55 (0.61) Yes 1.00 21.00 11.54 (4.48) 0.24 (0.3 1) 0.55 (0.61) Letter Series week 09 No 1.00 24.00 12.34 (4.36) 0.05 (0.31) 0.52 (0.62) Yes 1.00 24.00 12.34 (4.36) 0.05 (0.31) 0.52 (0.62) Letter Series week 10 No 1.00 8.00 4.05 (2.05) 0.07 (0.31) 0.91 (0.61) Yes 1.00 8.00 4.05 (2.05) 0.07 (0.3 1) 0.91 (0.61) Letter Series week 11 No 3.00 23.00 12.36 (5.38) 0.02 (0.32) 1.05 (0.63) Yes 3.00 23.00 12.36 (5.38) 0.02 (0.32) 1.05 (0.63)

PAGE 87

87 Table 2 3. Continued. Measure Data c leaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Letter Series week 12 No 3.00 22.00 12.55 (4.77) 0.11 (0.31) 0.55 (0.61) Yes 3.00 22.00 12.55 (4.77) 0.11 (0.31) 0.55 (0.61) Letter Series week 13 No 1.00 19.00 12.10 (4.24) 0.69 (0.34) 0.21 (0.66) Yes 1.00 19.00 12.10 (4.24) 0.69 (0.34) 0.21 ( 0.66) Letter Series week 14 No 4.00 23.00 13.31 (5.20) 0.02 (0.33) 1.07 (0.66) Yes 4.00 23.00 13.31 (5.20) 0.02 (0.33) 1.07 (0.66) Letter Series week 15 No 5.00 23.00 13.91 (4.86) 0.01 (0.35) 0.82 (0.69) Yes 5.00 23.00 13.91 (4.86) 0.01 (0.35) 0. 82 (0.69) Letter Series week 16 No 2.00 21.00 14.08 (5.30) 0.46 (0.34) 0.89 (0.67) Yes 2.00 21.00 14.08 (5.30) 0.46 (0.34) 0.89 (0.67) Letter Series post No 1.00 25.00 10.87 (5.51) 0.50 (0.29) 0.26 (0.57) Yes 1.00 25.00 10.87 (5.51) 0.50 (0.29) 0.26 (0.57) Simple RT pre No 220.04 518.12 322.97 (66.77) 0.96 (0.25) 0.51 (0.50) Yes 220.04 518.12 322.97 (66.77) 0.96 (0.25) 0.51 (0.50) Choice RT pre No 341.97 921.78 473.24 (106.89) 2.20 (0.25) 6.53 (0.50) Yes 341.97 705.62 466.85 (84.45) 1.03 ( 0.25) 0.78 (0.50) Simple RT week 01 No 228.70 596.82 308.09 (58.76) 1.96 (0.27) 7.00 (0.53) Yes 228.70 464.32 306.11 (50.90) 0.91 (0.27) 0.95 (0.53) Choice RT week 01 No 321.30 869.43 442.67 (95.05) 2.63 (0.27) 8.93 (0.53) Yes 321.30 637.41 435.42 (6 8.49) 1.09 (0.27) 1.48 (0.53) Simple RT week 02 No 212.42 697.93 304.05 (70.47) 2.73 (0.28) 13.02 (0.56) Yes 212.42 469.78 300.73 (55.70) 0.88 (0.28) 0.97 (0.56) Choice RT week 02 No 305.62 774.88 424.00 (77.00) 2.40 (0.28) 8.80 (0.56) Yes 305.62 594 .10 419.39 (59.44) 0.86 (0.28) 1.04 (0.56) Simple RT week 03 No 186.55 754.91 313.75 (106.53) 3.05 (0.30) 10.47 (0.60) Yes 186.55 477.95 301.37 (61.66) 1.04 (0.30) 1.89 (0.60)

PAGE 88

88 Table 2 3. Continued. Measure Data c leaned Minim um Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Choice RT week 03 No 304.57 546.49 412.20 (53.39) 0.57 (0.30) 0.06 (0.60) Yes 304.57 546.49 412.20 (53.39) 0.57 (0.30) 0.06 (0.60) Simple RT week 04 No 191.87 448.58 289.78 (56.58) 0.68 (0.30) 0.42 (0. 60) Yes 191.87 448.58 289.78 (56.58) 0.68 (0.30) 0.42 (0.60) Choice RT week 04 No 307.41 506.80 399.39 (48.78) 0.33 (0.30) 0.65 (0.60) Yes 307.41 506.80 399.39 (48.78) 0.33 (0.30) 0.65 (0.60) Simple RT week 05 No 183.61 635.68 294.17 (70.67) 1.94 ( 0.29) 7.05 (0.57) Yes 183.61 480.13 291.88 (61.36) 0.85 (0.29) 0.64 (0.57) Choice RT week 05 No 309.17 943.55 407.31 (80.98) 4.53 (0.29) 28.67 (0.57) Yes 309.17 545.78 401.46 (50.17) 0.95 (0.29) 1.02 (0.57) Simple RT week 06 No 179.62 816.25 302.11 ( 92.33) 3.07 (0.30) 14.84 (0.59) Yes 179.62 502.18 296.96 (69.52) 1.09 (0.30) 1.68 (0.59) Choice RT week 06 No 311.31 627.01 403.67 (55.68) 1.15 (0.30) 2.95 (0.59) Yes 311.31 570.69 402.79 (52.44) 0.70 (0.30) 0.70 (0.59) Simple RT week 07 No 147.12 42 8.40 279.67 (63.16) 0.29 (0.30) 0.06 (0.60) Yes 147.12 428.40 279.67 (63.16) 0.29 (0.30) 0.06 (0.60) Choice RT week 07 No 313.59 738.21 401.47 (68.63) 2.40 (0.30) 9.45 (0.60) Yes 313.59 546.59 397.36 (53.04) 0.78 (0.30) 0.74 (0.60) Simple RT week 0 8 No 127.03 512.68 279.58 (70.63) 0.59 (0.31) 0.97 (0.61) Yes 127.03 491.47 279.22 (69.47) 0.47 (0.31) 0.50 (0.61) Choice RT week 08 No 321.07 778.58 395.57 (67.06) 3.32 (0.31) 17.68 (0.61) Yes 321.07 547.09 391.65 (48.45) 0.79 (0.31) 0.68 (0.61) Sim ple RT week 09 No 146.48 844.86 286.44 (100.46) 3.09 (0.32) 16.39 (0.62) Yes 146.48 516.73 280.69 (73.74) 0.59 (0.32) 0.90 (0.62) Choice RT week 09 No 312.99 552.93 398.92 (52.23) 0.68 (0.32) 0.27 (0.62) Yes 312.99 552.93 398.92 (52.23) 0.68 (0.32) 0. 27 (0.62) Simple RT week 10 No 142.13 586.61 282.55 (76.02) 1.16 (0.31) 3.41 (0.62)

PAGE 89

89 Table 2 3. Continued. Measure Data c leaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Yes 142.13 510.62 281.24 (71.19) 0.65 (0.31) 1.00 (0 .62) Choice RT week 10 No 315.03 709.75 405.43 (67.60) 1.97 (0.31) 6.29 (0.62) Yes 315.03 570.95 403.03 (58.48) 1.02 (0.31) 0.84 (0.62) Simple RT week 11 No 111.36 458.75 285.48 (71.00) 0.08 (0.32) 0.26 (0.64) Yes 111.36 458.75 285.48 (71.00) 0.08 (0.32) 0.26 (0.64) Choice RT week 11 No 321.68 566.76 406.70 (57.74) 0.75 (0.32) 0.08 (0.64) Yes 321.68 566.76 406.70 (57.74) 0.75 (0.32) 0.08 (0.64) Simple RT week 12 No 124.19 527.99 288.07 (72.64) 0.67 (0.31) 1.07 (0.61) Yes 124.19 506.00 287.70 (71.46) 0.55 (0.31) 0.60 (0.61) Choice RT week 12 No 305.72 656.34 399.03 (60.26) 1.45 (0.31) 4.32 (0.61) Yes 305.72 579.79 397.75 (55.32) 0.80 (0.31) 0.80 (0.61) Simple RT week 13 No 167.63 489.66 292.26 (73.60) 0.47 (0.34) 0.00 (0.66) Yes 167.63 4 89.66 292.26 (73.60) 0.47 (0.34) 0.00 (0.66) Choice RT week 13 No 296.42 549.24 404.22 (53.54) 0.60 (0.34) 0.38 (0.66) Yes 296.42 549.24 404.22 (53.54) 0.60 (0.34) 0.38 (0.66) Simple RT week 14 No 127.43 731.01 283.15 (95.02) 2.03 (0.33) 8.73 (0.66) Yes 127.43 524.29 279.10 (78.51) 0.50 (0.33) 0.60 (0.66) Choice RT week 14 No 266.00 556.15 394.88 (56.08) 0.55 (0.33) 0.65 (0.66) Yes 266.00 556.15 394.88 (56.08) 0.55 (0.33) 0.65 (0.66) Simple RT week 15 No 105.77 1147.66 298.90 (158.96) 3.93 (0.35) 19.08 (0.69) Yes 105.77 529.11 280.87 (83.37) 0.96 (0.35) 2.16 (0.69) Choice RT week 15 No 318.84 498.83 394.48 (46.33) 0.62 (0.35) 0.39 (0.69) Yes 318.84 498.83 394.48 (46.33) 0.62 (0.35) 0.39 (0.69) Simple RT week 16 No 118.86 645.68 282.06 (87.5 4) 1.31 (0.34) 5.02 (0.67) Yes 118.86 518.88 279.48 (77.91) 0.36 (0.34) 0.75 (0.67) Choice RT week 16 No 297.04 690.02 400.33 (68.22) 2.04 (0.34) 6.16 (0.67) Yes 297.04 574.05 397.96 (59.41) 1.24 (0.34) 1.66 (0.67)

PAGE 90

90 Table 2 3. Continued. Measure Data c leaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Simple RT week 17 No 129.90 594.69 293.66 (85.13) 0.89 (0.29) 1.94 (0.57) Yes 129.90 539.62 292.74 (82.09) 0.64 (0.29) 0.96 (0.57) Choice RT week 17 No 303.34 1133.68 419. 14 (111.11) 4.43 (0.29) 26.09 (0.57) Yes 303.34 595.09 409.07 (63.21) 0.98 (0.29) 0.79 (0.57) Logical Memory Immediate post No --------Yes 23.00 66.00 45.76 (9.10) 0.27 (0.29) 0.00 (0.57) Logical memory Delay post No -------Yes 11.00 45.00 26.67 (7.32) 0.30 (0.29) 0.17 (0.58) FAS post No 23.00 80.00 42.07 (11.54) 0.84 (0.35) 1.41 (0.69) Yes 23.00 76.69 41.99 (11.31) 0.70 (0.35) 0.84 (0.69) Trails A post No 16.00 49.72 30.29 (7.94) 0.34 (0.35) 0.73 (0.69) Ye s 16.00 49.72 30.29 (7.94) 0.34 (0.35) 0.73 (0.69) Trails B post No 18.25 202.00 77.22 (38.57) 1.31 (0.35) 1.87 (0.69) Yes 18.25 187.37 76.90 (37.57) 1.18 (0.35) 1.26 (0.69) Letter Number Sequencing post No 7.00 16.00 10.80 (2.24) 0.59 (0.35) 0.34 (0 .69) Yes 7.00 16.00 10.80 (2.24) 0.59 (0.35) 0.34 (0.69) 1 Back post No 467.98 2179.14 790.13 (227.56) 3.66 (0.29) 20.70 (0.58) Yes 467.98 1239.92 775.03 (155.92) 0.81 (0.29) 0.99 (0.58) 2 Back post No 633.49 3662.74 1586.14 (663.94) 1.09 (0.29) 1.1 6 (0.58) Yes 633.49 3557.87 1584.58 (659.08) 1.05 (0.29) 0.99 (0.58) Note: All data reported in raw metric (i.e., RT data not recoded to positive = better format). Data presented prior to and po st cleaning procedures to allow interested readers to ins pect changes in descriptive, normality, and raw data values due to cleaning procedures.

PAGE 91

91 Figure 2 2. Graphical r epresentation of m issing d ata across the s tudy p eriod. 0 10 20 30 40 50 60 % Missing Data Session

PAGE 92

9 2 Table 2 4. Comparison of means and standard deviations of baseline study variable s among completed and attrited subjects. Total Sample Completed Sample Attrited Sample Df t 2 p value Characteristic n = 90 n = 68 n = 22 Demographic/Descriptive Age (SD), years 63.56 (8.45) 63.88 (8.75) 62.55 (7.56) 88 0.64 0.52 Education (SD), years 16.12 (2.25) 16.22 (2.22) 15.82 (2.34) 88 0.73 0.47 Gende r, n (% female) 72 (82) 57 (84) 17 (77) 1 0.69 0.49 Estimated IQ (SD) 113.17 (6.45) 113.10 (6.50) 113.41 (6.43) 88 0.19 0.85 BDI II (SD) 6.52 (5.38) 5.76 (5.17) 8.86 (5.47) 88 2.41 0.02 State Anxiety (SD) 30.30 (7.94) 29.85 (7.47) 31.68 (9.3 0) 88 0.94 0.35 Baseline Cognition Logical Memory Immediate (SD) 41.01 (8.73) 41.05 (8.50) 40.86 (9.63) 88 0.09 0.93 Logical Memory Delay (SD) 25.91 (7.54) 25.75 (7.57) 26.41 (7.61) 88 0.35 0.72 FAS Total (SD) 39.49 (10.42) 39.56 ( 10.89) 39.27 (9.04) 88 0.11 0.91 Trails A (SD) 168.60 (10.07) 167.82 (10.37) 170.99 (8.88) 88 1.29 0.20 Trails B (SD) 123.98 (30.85) 121.48 (31.10) 131.72 (29.41) 88 1.36 0.18 LN Sequencing (SD) 10.72 (2.72) 10.71 (2.81) 10.77 (2.49) 88 0.10 0.92 1 Back (SD) 3134.41 (197.42) 3153.27 (192.86) 3076.95 (204.50) 87 1.59 0.12 2 Back (SD) 2311.05 (613.57) 2270.16 (623.21) 2437.45 (578.12) 88 1.11 0.27 Number Copy (SD) 41.86 (4.83) 41.93 (4.74) 41.64 (5.23) 88 0.24 0.81 Symbol Digit ( SD) 22.22 (5.59) 22.44 (5.39) 21.52 (6.24) 88 0.67 0.50 Letter Series (SD) 6.02 (3.09) 5.94 (3.05) 6.27 (3.25) 87 0.44 0.66 Simple RT (SD) 477.03 (66.77) 482.94 (63.08) 458.75 (75.74) 88 1.49 0.14 Choice RT (SD) 387.80 (53.39) 340.34 (84.93 ) 310.93 (80.79) 88 1.43 0.16 Notes : Mean (Standard Deviation); BDI II = Beck Depression Inventory, 2 nd Edition; LN = Letter Number; RT = Reaction Time.

PAGE 93

93 Table 2 5 Baseline correlations between practiced and unpracticed (i.e., transfer) cognitive measures. Number Copy Symbol Digit Letter Series Simple RT Choice RT Logical Memory Immediate 0.45 ** 0.35 ** 0.39 ** 0.07 0.07 Logical Memory Delay 0.39 ** 0.33 ** 0.39 ** 0.08 0.10 FAS Total 0.26 0.04 0.32 ** 0.13 0.08 Trails A 0.27 ** 0.32 ** 0.23 0.39 ** 0.44 ** Trails B 0.39 ** 0.47 ** 0.48 ** 0.35 ** 0.39 ** LN Sequencing 0.33 ** 0.32 ** 0.47 ** 0.17 0.24 1 Back RT 0.42 ** 0.54 ** 0.29 ** 0.45 ** 0.47 ** 2 Back RT 0.16 0.11 0.14 0.22 0.17 Notes: p < .05; ** p < .01.

PAGE 94

94 Table 2 6 Rotated f actor p atter n of c ognitive m easures at b aseline. Measures Factor 1: Processing Speed Factor 2: Executive Processing Factor 3: Far Transfer Choice RT 0.98 0.16 0.11 Simple RT 0.80 0.13 0.13 1 Back 0.57 0.08 0.05 Symbol Digit 0.50 0.06 0.29 Number Copy 0.4 0 0.08 0.33 2 Back 0.23 0.02 0.07 Logical Memory Immediate 0.06 0.14 1.0 0 Logical Memory Delay 0.06 0.03 0.85 Letter Number Sequencing 0.05 0.91 0.12 FAS 0.13 0.66 0.03 Letter Series 0.02 0.45 0.28 Trails A 0.38 0.38 0.11 Trails B 0.35 0.36 0.13 Note: Extraction Method: Principal Axis Factoring. Rotation Method: Promax with Kaiser Normalization. Coeffiecients with values greater than .32 are bolded.

PAGE 95

95 Figure 2 3. Posttest c onfirmatory f actory a nalysis of c ognitive v ariables. Note: All factor loadings are statistically significant (all < 0.05), except Speed 2Back, p = 0.57. Factor correlations are statistically significant (all < 0.05), except Speed Memory, p = 0.11.

PAGE 96

96 Table 2 7 Cognitive b lock d ata Mean (SD) Number Copy Block 1 42.57 ( 4.32 ) Number Copy Block 2 44.5 ( 3.40 ) Number Copy Block 3 45.05 ( 3.54 ) Number Copy Block 4 45.59 ( 3.33 ) Number Copy Block 5 45.89 ( 3.89 ) Symbol Digit Block 1 23.41 ( 4.53 ) Symbol Digit Block 2 26.17 ( 3.15 ) Symbol Digit Block 3 26.46 ( 3 .04 ) Symbol Digit Block 4 26.99 ( 3.20 ) Symbol Digit Block 5 27.92 ( 3.61 ) Letter Series Block 1 7.92 ( 3.47 ) Letter Series Block 2 9.85 ( 4.40 ) Letter Series Block 3 11.4 ( 4.26 ) Letter Series Block 4 12.44 ( 4.71 ) Letter Series Block 5 12.24 ( 5.40 ) Sim ple RT Block 1 486.85 ( 53.60 ) Simple RT Block 2 504.67 ( 56.14 ) Simple RT Block 3 515.64 ( 63.50 ) Simple RT Block 4 515.37 ( 70.83 ) Simple RT Block 5 512.21 ( 74.82 ) Choice RT Block 1 353.75 ( 69.80 ) Choice RT Block 2 394.63 ( 48.10 ) Choice RT Block 3 399 .37 ( 51.32 ) Choice RT Block 4 398.96 ( 52.65 ) Choice RT Block 5 397.67 ( 55.24 ) Note: Higher values = better performance on all cognitive variables.

PAGE 97

97 Table 2 8. Parameter Estimates for Latent Growth Curve Models. Estimate SE z Value Number Copy Mean Intercept 42.56 0.46 93.32 Linear Slope 0.13 0.21 0.61 Quadratic Slope 1.85 0.47 3.96 Variance Intercept 16.23 2.79 5.82 Linear Slope 0.57 0.65 0.87 Quadratic Slope 6.23 3.16 1.98 Correlations Inter cept Linear Slope 0.35 -1.13 Intercept Quadratic Slope 0.72 -2.69 Linear Slope Quadratic Slope 0.61 -1.00 Symbol Digit Mean Intercept 23.44 0.48 49.07 Linear Slope 0.08 0.21 0.37 Quadratic Slope 2.24 0.44 5.1 Var iance Intercept 19.07 3.05 6.26 Linear Slope 1.82 0.58 3.12 Quadratic Slope 9.38 2.61 3.6 Correlations Intercept Linear Slope 0.42 -2.45 Intercept Quadratic Slope 0.62 -3.66 Linear Slope Quadratic Slope 0.88 -3 .07 Letter Series Mean Intercept 7.85 0.37 20.33 Linear Slope 0.14 0.22 0.67 Quadratic Slope 2.14 0.4 5.41 Variance Intercept 10.82 1.87 5.79 Linear Slope 0.47 0.12 3.8 Quadratic Slope 0 --Correlations In tercept Linear Slope 0.64 -3.96 Intercept Quadratic Slope ---Linear Slope Quadratic Slope ---

PAGE 98

98 Table 2 8. Continued Estimate SE z Value Simple RT Mean Intercept 486.38 5.72 85.07 Linear Slope 6.79 2.94 2.3 1 Quadratic Slope 26.54 5.55 4.78 Variance Intercept 2466.41 411.52 5.99 Linear Slope 85.22 40.73 2.09 Quadratic Slope 181.9 146.71 1.24 Correlations Intercept Linear Slope ---Intercept Quadratic Slope ---Linear Slope Quadratic Slope ---Choice RT Mean Intercept 353.81 7.36 48.01 Linear Slope 17.15 2.92 5.87 Quadratic Slope 54.53 6.85 7.96 Variance Intercept 4557.28 722.59 6.31 Linear Slope 349.68 115.92 3.02 Q uadratic Slope 2604.88 636.45 4.09 Correlations Intercept Linear Slope 0.65 -3.62 Intercept Quadratic Slope 0.75 -4.52 Linear Slope Quadratic Slope 0.93 -3.38 Note: z scores greater than 1.96 are statistically significant ( p < .05).

PAGE 99

99 Figure 2 4. Model i mplied c hange in c ognitive f unctioning a cross s tudy p eriod. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1 2 3 4 5 Z score Blocks (trials in parentheses) Eighteen week Cognitive Change (Standardized Metric) (0 2) (3 6) (7 10) (11 14) (15 17) Letter Series Symbol Digit Number Copy Choice RT Simple RT

PAGE 100

100 Table 2 9 Parameter e stimates for c urve of factors m odel. Estimate SE z Value Mean Intercept 369.026 5.88 62.8 Linear Slope 8.49 1. 31 6.46 Quadratic Slope 53.76 6.46 8.33 Variance Intercept 2317.34 434.41 5.33 Linear Slope 37.96 15.29 2.48 Quadratic Slope 554.71 308.45 1.8 Correlations Intercept Linear Slope 0.11 -0.53 Intercept Quadratic Slope 0.55 -2.25 Linear Slope Quadratic Slope 0.76 -2.34 Standardized Estimate z Value Block 1 Choice RT 0.81 -Simple RT 0.76 9.21 Letter Series 0.27 2.95 Number Copy 0.51 6.98 Symbol Digit 0.45 6.07 Bl ock 2 Choice RT 0.92 -Simple RT 0.68 9.21 Letter Series 0.19 2.95 Number Copy 0.55 6.98 Symbol Digit 0.48 6.07 Block 3 Choice RT 0.94 -Simple RT 0.68 9.21 Letter Series 0.19 2.95 Number Copy 0.57 6.98 Symbol Digit 0.53 6.07 Block 4 Choice RT 0.93 -Simple RT 0.67 9.21 Letter Series 0.19 2.95 Number Copy 0.58 6.98 Symbol Digit 0.52 6.07 Block 5

PAGE 101

101 Table 2 9 Continued Estimate SE z Value Choice RT 0.9 -Simple RT 0.65 9.21 Letter Series 0.17 2.95 Number Copy 0.56 6.98 Symbol Digit 0.51 6.07 Note: z scores greater than 1.96 are statistically significant ( p < .05).

PAGE 102

102 Figure 2 5. Model i mplied c hange in g ene ral c ognitive f unctioning a cross s tudy p eriod. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 2 3 4 5 Z Score Blocks Z score Change in Overall Cognition

PAGE 103

103 Figure 2 6. Curve of f actors p ath d iagram with s tandardized c oefficients.

PAGE 104

104 Table 2 10. Standardized c oefficients p redicting c ommon l evel and c hange in g eneralized l earning. Common Factor Loadings Level Linear Quadratic Age 0.50*** 0.03 0.03 Education 0.19* 0.03 0.40* Depression 0.26* 0.02 0.19 Anxiety 0.08 0.05 0.41* Notes: *** p < .001; p < .05.

PAGE 105

105 Figure 2 7. Complete p ath d iagram of i ndividual d ifference p redictor s of o verall c hange in g eneralized l earning.

PAGE 106

106 Figure 2 8. Standardized l oadings and s ignificant levels of i ndividual d ifferences in g eneralized l earning Note: ** p < .01, p < .05.

PAGE 107

107 Figure 2 9. Graphical r epresentation of g eneralized l earning f or both a h igh (i.e., 90th %ile) and l ow (i.e., 10th %ile) e ducated (left) and s tate a nxious (right) s ubject.

PAGE 108

108 Figure 2 10. Complete p ath d iagram of t ransfer of g eneralized l earning in c ognitive f unctioning.

PAGE 109

109 Figure 2 11. Standardized l oadings and s ignificant l evels of t ransfer in g eneralized l earning. Note: *** p < 0.01, p < 0.05, ~ p < 0.10. N = 68.

PAGE 110

110 Table 2 11 Standardized c oefficients for t ransfer in g eneralized l earning to n on practiced m easures (N = 68). Pre Post Change Scor es Common Factor Path Loadings Intercept Linear Quadratic Trails A 0.17 0.37 ~ 0.04 Trails B 0.22 ~ 0.04 0.21 1 Back 0.04 0.51 0.91 ** 2 Back 0.02 0.01 0.36 Letter Number Sequencing 0.11 0.37 ~ 0.29 ~ FAS Total 0.08 0.29 0.13 Logical Memory Immediate 0.05 0.04 0.15 Logical Memory Delay 0.06 0.15 0.05 Notes: p < .05; ** p < .01; ~ p < .10.

PAGE 111

111 Figure 2 12. Graphical r epresentation of g eneralized l earning t ransfer for both a h igh (i.e., 80 th %ile) an d l ow (i.e., 20 th %ile) c hange 1 b ack RT s ubject (left) and 2 b ack RT s ubject (right).

PAGE 112

112 CHAPTER 3 SLEEP VARIABLES AS P REDICTORS OF INITIAL LEVEL AND GAIN IN LA TE LIFE COGNITION Introduction Late life cognitive improvement can be the result of either targe ted intervention strategies [i.e., intervention related learning (Ball et al., 2002)] or repeated practice with cognitive stimuli [i.e., practice related learning (Hofland et al., 1981)]. While a relatively large amount of research has been conducted exami ning intervention related cognitive improvements in late life, a comparatively small amount of research has examined practice related learning in old age. However, practice with cognitive stimuli is a major part of any cognitive intervention and practice r elated learning has been demonstrated to be comparable to intervention related learning in magnitude and ability to be maintained (Baltes et al., 1988; Baltes et al., 1989; Yang & Krampe, 2009). As such, it is responses to repeated practice, but to also understand predictors of that response. A key question in the emerging gerontological cognitive intervention literature concerns individual differences in learning potential and the identification of predictors of such differences (Hertzog et al., 2008) Consequently, ability to benefit from practice. Sleep represents an intriguing individual difference variable as it has been shown to rel ate to various aspects of cognitive functioning [ e.g., (Pilcher & Huffcutt, 1996)] and has demonstrated an ability to be improved well into the later years of life [ e.g., (Dzierzewski et al., 2010)] As such, the overarching goal of the current investigati on was to examine sleep as an individual difference predictor of both initial level of late life

PAGE 113

113 cognitive functioning and gain in late life cognitive functioning as a result of repeated practice. Late Life Practice Related Learning Repeated practice wit h cognitive stimulus has been shown to produce significant gains in cognitive functioning. The domains of reasoning, processing speed, and attention have all demonstrated substantial (i.e., .45 to 1.35 standard score improvements) practice related learning (Baltes et al., 1989; Hofland et al., 1981; Yang et al., 2006; Yang et al., 2009). Given that the magnitude of practice related learning gains is comparable in size to gains demonstrated following focused cognitive interventions (Baltes et al., 1986; Balt es et al., 1988; Baltes et al., 1989), it is just as important to understand individual difference predictors of response to practice related learning as it is to understand individual difference predictors of response to cognitive interventions (Hertzog e t al., 2008). A small number of investigations have examined individual difference variables in relation to practice related learning in late life. Understanding the characteristics of an om repeated cognitive practice may well illuminate the mechanisms basic to practice related learning and/or the facilitating conditions for optimal practice related learning to occur. For example, given the known involvement of scientifically manipulated s leep (i.e., sleep deprivation) on learning and plasticity (Drummond et al., 2000; Maquet, 2001; Walker, 2008), differences in subjective sleep as a predictor of practice related learning would suggest that individual differences in naturally occurring slee p are important to learning as well. Further, an appreciation of individual differences in practice related learning could also aid in the empirically guided selection of individuals into cognitive practice interventions,

PAGE 114

114 and/or may suggest avenues through which response to cognitive practice may be bolstered. For example, if it is discovered that individuals with less than average time spent asleep respond poorly to cognitive practice in a particular domain, one could envision that efforts to improve sleep through practice. While the extant literature regarding individual difference predictors of practice related learning is limited, several factors have been investigated. It has been previously report that bo gains in processing speed, reasoning, and visual attention (Yang et al., 2006). While young old (70 79 years), oldest old (80 91 years), and older adults with higher and lower leve ls of cognitive functioning all benefit from repeated practice, younger age and higher levels of initial functioning both have been associated with increased rates of practice related learning in reasoning and to a lesser extent processing speed (Yang et a l., 2006). This finding is consistent with other studies in which similar individual differences in cognitive training related gains were observed (McArdle & Prindle, 2008) Both stable anxiety (i.e., trait anxiety) and momentary anxiety (i.e., state anxi ety) have neither type of anxiety significantly predicted patterns of practice related learning in late life (Yang et al., 2009) When considering predictors of focused cognitive interventions, results have been mixed. In particular, it has been reported that common demographic/descriptive indicators (i.e., age, sex, education, visual functioning, and mental status) do not predict response to cognitive training (Ball et a l., 2002). It has also been reported that

PAGE 115

115 individuals with localized impairments (e.g., memory impairments) do not respond to cognitive training in the impaired domain (Unverzagt et al., 2007), while others have reported largest training related gains from individuals with the lowest initial ability levels (Jaeggi et al., 2008). Thus, it appears that much is yet to be known regarding predictors of practice and training related learning in late life. While sleep has not been investigated as a predictor of l ate life practice related learning (or cognitive training related learning) its relation to late life cognition (in general) has received some empirical attention. As sleep is associated with frontal lobe functioning (Jones & Harrison, 2001), and frontal lobe functioning is critical for higher order cognitive functioning and learning (Cabeza & Nyberg, 2000), it stands to reason that sleep should be related to cognitive functioning and learning in late life. Similarly, sleep is critical for hippocampal proc essing of memory formations (Hobson & Pace Schott, 2002), that are important for learning. The assertion that sleep is related to cognitive functioning appears correct [i.e., (Banks & Dinges, 2007; Goel, Rao, Durmer, & Dinges, 2009; Maquet, 2001; Walker, 2 008)]; however, the direction and strength of this relationship is still relatively not well understood. Sleep and Cognitive Functioning The vast majority of investigations examining the link between nocturnal sleep and daytime cognitive functioning hav e been conducted through either correlation/quasi experimental designs, where differences in sleep serve as independent variables to predict cognitive functioning (i.e., sleep is naturally occurring and simply measured) or experimental designs in which sle ep is manipulated experimentally to test effects on cognitive performance (i.e., sleep deprivation paradigms). Correlation and quasi experimental designs have allowed for the measurement of naturally sleeping

PAGE 116

116 individuals to determine how aspects of sleep a re related to aspects of cognitive functioning. Experimental deprivation paradigms have allowed for direct manipulation of sleep and subsequent examination of how experimentally manipulated changes in sleep relate to cognitive functioning. Thus, the litera ture regarding the sleep cognition relationship will be summarized as those in which sleep was naturally occurring and those in which sleep was experimentally manipulated (with emphasis drawn to studies of older adults when available). Natural Sleep Natura lly occurring sleep can be measured in several different manners. Researchers can ask research subjects to retrospectively self report their habitual sleep characteristics via simple questions [e.g., Over the past year, did you have difficulty falling asle ep or staying asleep? (Tworoger et al., 2006)] or through well validated retrospective sleep questionnaires [e.g., Pittsburg Sleep Quality Index; PSQI (Nebes, Buysse, Halligan, Houck, & Monk, 2009)]. Subjects can also prospectively self report their sleep characteristics nightly through use of widely used instruments called sleep diaries (Lichstein, Riedel, & Means, 1999). Sleep diaries require subjects to record various aspects of their sleep (length of onset, quality, etc.) each morning upon awakening for a period of time defined by the research team (typically 1 2 weeks). Additionally, it is possible to objectively assess the naturally occurring sleep of research subjects through both actigraphy and polysomnography (PSG). Actigraphy attempts to quantify sleep through precise measurement of limb movements subsequently applied to mathematical computations. PSG quantifies sleep through multiple physiological indices, including muscle tone, eye movements, EEG recordings, respiration, etc.

PAGE 117

117 Retrospective In a study of 157 community dwelling older adults, it was found that poor sleepers (as identified by the PSQI global score) performed worse than good sleepers on several tests of executive functioning (i.e., RBANS, TONI, Trails B, and N Back); however, there w ere no differences between the groups on indicators of processing speed, inhibition, or verbal memory. Interestingly, habitual sleep fragmentation indicators (i.e., SOL and SE) were related to lower scores on RBANS, TONI, and N Back while habitual sleep du ration was not related to any cognitive measure. Such findings led the authors to conclude that the unwanted intruding wakefulness experienced, rather than the total number of minutes of sleep the (Nebes et al., 2009). However, such a statement is in direct contention with several large epidemiological studies that have reported associations between habitual sleep duration and cognitive functioning. Several large scale epidemio logical studies have included questions which have garnered information regarding habitual sleep duration and difficulty. In a study of over 3000 older Spanish adults, sleep duration (as reported in response to the question many hours do you usually ) was associated with overall cognitive functioning (assessed via MMSE) (Faubel et al., 2009). Cognition was found to decrease linearly from sleepers of 7 h ou rs per night through sleepers of 11 h ou rs per night, such that sleeping 11 h ou r s pe r night was equivalent to 10 years of ag ing No association with short sleep duration and cognitive functioning was observed. However, the use of a single, global indicator of cognitive functioning may have truncated any potential relationships. In a simil ar study of over 5000 adults in Finland, sleep duration sleep in 24h ) was

PAGE 118

118 associated with verbal fluency and list memory, such that both long and short sleep duration were associated with po orer performance (Kronholm et al., 2009). The authors of these studies suggest that sleep duration may be related to cognitive functioning through changes in sleep architecture, fragmentation, quality, and neurological conditions (Faubel et al., 2009; Kron holm et al., 2009). Tworoger and colleagues (2006) reported that in a large sample of community dwelling older women, sleeping less than 5 hours per night (as reported in response to sleep in a 24h period ) was associ ated with poorer score composite of many cognitive tests) when compared to women sleeping 7 hours or more per night. It was also reported that these short sleeping older women performed significant worse across many of the individua l indicators of cognitive functioning (i.e., TICS, East Boston Memory Test, Category Fluency, and Digit Span Backwards ). Additionally, women who responded affirmatively also scored lower on related cognitive difficulties were reported to be equivalent to 4 gnitive functioning (Tworoger et al., 2006). Two additional studies examined habitual sleep difficulties reported through retrospective recall and general cognitive functioning (measured with global assessment tools, the Short Portable Mental Status Quest ionnaire and the MMSE) in community dwelling older adults (Cricco et al., 2001; Jelicic et al., 2002). Both investigations reported that the presence of a sleep complaint (as indicated through response to

PAGE 119

119 simple questions regarding difficulty initiating or maintaining sleep) was associated with increased risk for cognitive decline at 3 year follow up. The authors suggest that may be the result o f (Cricco et al., 2001; Jelicic et al., 2002). I n summation, examination of the association between retrospective self reported recall of habitual sleep duration/difficulty and cognitive functioning has resulted in reports of contradictory findings. Good and poor sleepers have been shown to differ on se veral indicators of executive performance while performing similarly on a measure of verbal memory. Interestingly, no association between sleep duration and cognitive functioning was reported (Nebes et al., 2009b). However, other groups have reported relat ionships between short sleep lengths (Kronholm et al., 2009; Tworoger et al., 2006) and cognitive performance, while others have failed to observe the relationship between short sleep duration, but did note relationships between long sleep duration and cog nitive functioning (Faubel et al., 2009). In general, there does appear to be a relationship between self reported retrospective recall of habitual sleep duration and cognitive functioning; however, the exact nature of that relationship cannot be ascertain ed from the extant literature. Prospective The use of prospective sleep diaries to investigate the relationship between sleep and cognitive functioning is a relatively rare occurrence in the literature. In addition, most of the work using prospective self report of sleep characteristics tend to focus on either poor sleepers or differences between good and poor sleepers. This is most likely due to the subjective nature of insomnia, the utility of sleep diaries in quantifying sleep

PAGE 120

120 characteristics over time, and the requirement of a sleep complaint (Martin & Ancoli Israel, 2002). In a study of 60 older adults (grouped into good sleepers and those with insomnia) it was found that better sleep diary measured depth (i.e., self reported ratings of depth of sleep) SQR, and SE was associated with better attention and concentration (digit forwards and backwards) and worse memory (verbal paired associates, visual reproductions) in the good sleepers. Additionally, better SOL and TWT was related to better speed (digit symbol, pegboard, Trails A and B, RT) in good sleepers However, the same relationships were not observed for the poor sleepers. It was discovered that i ncreased d epth and lower TWT were related to increased memory for poor sleepers (Bastien et al., 2003). The authors concluded that However, they speculate that the c ounter intuitive results (i.e., poor sleep being related to better cognitive functioning in some cases) he fact that these individuals invest more effort in maintaining good performance when the In a study of 49 younger subjects (grouped into poor sleepers and good sleepers), self reported sleep diary measures differed between the groups (i.e., poor sleepers consistently reported worse sleep); however, the groups did not differ on any of the numerous cognitive measures administers [ Attention (Stroop, Digit Span, Brief Test of Attention, LN Sequencing, S ustained Attention), Speed (Digit Symbol, Trails), Verbal Fluency (COWA), Verbal Memory (HVLT) ] (Orff, Drummond, Nowakowski, & Perlis, if hyperarousal extends to daytime hours, it aids in maintaining performance Additio nally, they hypothesize that a lack of group

PAGE 121

121 differences may indicate (Orff et al., 2007). In a set of excellently designed experiments, Altena and colleagues (2008, 2008) examined the brain functioning of older adult poor and good sleepers (Altena, Van Der Werf et al., 2008; Altena, Werf et al., 2008). Insomnia diagnosis was based on prospective sleep diary and accompanying complaint. Poor sleep ing elders were demonstrated to have superior verbal fluency performance while simultaneously exhibiting hypoactivation in prefrontal regions. The authors suggest that p refrontal hypoactivation in the absence of behavioral differences may be related to in dividually (Altena, Werf et al., 2008). Interestingly, these same poor sleeping elders performed faster on a simple RT task than good sleepers but performed slower on a comple x RT than good sleepers However, following a multicomponent behavioral treatment to improve sleep (which was successful), treated subjects became slower on the simple RT task and faster on the complex RT task ( i.e., matched the good sleepers) The authors suggest a may be responsible for the interesting results (Altena, Van Der Werf et al., 2008). Similar to the retrospective recall of sleep summarized above, examination of the association between prospective se lf reported sleep characteristics and cognitive functioning has resulted in reports of contradictory findings. Indicators of poor sleep have been associated with both better and worse performance on various cognitive tasks (Bastien et al., 2003). Good and poor sleepers have been shown to not differ across many measures of cognitive functioning (Altena, Werf et al., 2008; Orff et al.,

PAGE 122

122 2007). And, sometimes poor sleepers have been found to outperform their good sleeping counterparts, while other times they pe rform at lower levels than good sleepers (Altena, Van Der Werf et al., 2008). In general, the relationship between self reported prospective sleep and cognition is currently unclear. While there appears to be a relationship between these variables, the exa ct nature of that relationship cannot be ascertained from the extant literature. Objective: PSG In the above described study of 49 younger subjects (grouped into poor sleepers and good sleepers), sleep was also measured objectively via PSG (in concert wit h the sleep diary assessments already discussed) (Orff et al., 2007). While good and poor sleepers did not differ in any measure of cognitive functioning, canonical correlation revealed that o bjectively measured sleep indicators (SE, SOL, WASO, REM latency NWAK, TST, % in various sleep stages) w ere associated with speed indicators (Digit Symbol, Trails) (Orff et al., 2007). However, the nature of the inferential statistics prevents a more detailed description of this association. Several well designed exp eriments have allowed subjects to sleep naturally (measuring sleep with PSG) while manipulating the time of cognitive testing/training to either allow sleep to occur following testing or not in order to examine any effects of post testing/training sleep on subsequent testing. In a study of 66 younger, healthy subjects individual who slept following training on a number reduction task were twice as likely as individuals who spent comparable time awake to spontaneously discover gner, Gais, Haider, Verleger, & Born, 2004). While no PSG concluded that

PAGE 123

123 extraction of explicit knowledge and insightf In a similarly designed experiment, 62 younger, healthy subjects underwent training in various finger tapping sequences either early in the morning or at night, and were retested 12 hours later (Walker, Brakefield, Morgan, Hobson, & Stickgold 2002). It was reported that a night of sleep provide d a 20% increase in motor skills, while equivalent time awake offer ed no benefits. Further, a p ositive correlation between time spent in Stage 2 NREM sleep and motor skill learning was observed The aut hors suggest that Stage 2 sleep and sleep spindle activity may be important for learning simple tasks, while SWS and REM sleep may be more important for complex task learning (Walker et al., 2002) In an attempt to extend the findings of optimized performan ce following sleep, Tucker and colleagues (2011) trained 16 healthy older adults in the finger tapping sequencing employing the above described experimental paradigm (Tucker, McKinley, & Stickgold, 2011). Older adults performed significantly better followi ng sleep than 12 hours of not sleeping, suggesting sleep dependent motor skill performance in the elderly. Interestingly, older adults showed similar rate s of improvement as was found in younger samples; however, s leep characteristic s did not correlate wit performance ( in contrast, Stage 2 sleep and sleep spindle activity does correlate in younger adults ) The authors concluded that sleep in the elderly does optimize motor skill learning; however, it may do so differently than in younger adults (Tucker et al., 2011). In the above described study of 60 older adults (grouped into good sleepers and those with insomnia), sleep was also measured objectively via PSG (in concert with the sleep diary assessments already discussed) (Bastien et al., 2003). It was reported

PAGE 124

124 poor er SOL was related to poor er verbal memory ( for both the good and poor sleepers ), executive functioning (i.e., Wisconsin Card Sorting Test; for the good sleepers) and psychomotor speed and attention ( for the poor sleepers ). Higher amo unts of TWT were related to lower psychomotor speed ( for both the good and poor sleepers ) and memory ( g ood sleepers only ) (Bastien et al., 2003). In a study of 64 older adults (grouped into good sleepers and those with insomnia), SWS measured with PSG was unrelated to performance on a Simple RT task, Continuous Performance task, and Switching Attention test in the good sleeping elderly. However, older poor sleepers with the slowest RTs were found to have comparative deficits in the amount of time spent in S WS (Crenshaw & Edinger, 1999). The authors stipulate that wave sleep deficits may contribute to lowered cognitive performances ; however, why this relationship was only found in poor sleeping elders was not discussed (Crenshaw & Edinger, 1999). Inves tigation into the relationship between naturally occurring sleep, assessed via PSG, and waking cognitive functioning has provided mixed results. Sleep does appear to facilitate learning/insight (Tucker et al., 2011; Wagner et al., 2004; Walker et al., 2002 ); however, the aspects of sleep that contribute to this optimization vary across studies (Wagner et al., 2004; Walker et al., 2002) and may change with age (Tucker et al., 2011). Further, some investigators have demonstrated a general association between PSG measured sleep and cognitive functioning (Orff et al., 2007), while other have failed to demonstrate such associations, or only demonstrated them in a unique subset of subjects (Crenshaw & Edinger, 1999). As such, it appears additional research

PAGE 125

125 is need to further explicate the relationship between PSG measured naturally occurring sleep and cognitive functioning. Objective: Actigraphy In a study of nearly 3000 older community dwelling women actigraphy measured sleep (SE, SOL, WASO, and napping) were ass ociated with an increased risk of poorer MMSE and Trails B performance such that worse sleep was related to worse cognitive functioning (Blackwell et al., 2006) However, TST was not related to cognitive functioning which led the authors to conclude that (Blackwell et al., 2006). In a different, but complementary vein, 7 nights of actigraphy were used to compute sleep/wake patterns in 144 community dwelling older adults. It was repor ted that older adults who displayed many shifts from rest to activity performed worse on composites of executive functioning (Digit backwards, Stroop interference, Trails B), memory (AVLT, Digit Span, Pattern Recognition), and speed (Stroop, T rail M aking T est ) than elderly with more consistent rest activity patterns (Oosterman, Van Someren, Vogels, Van Harten, & Scherder, 2009) It does appear that there is a relationship between objectively measured naturally occurring sleep measured with actigraphy and cog nitive performance. However, the scarcity of research studies examining these associations and the differing methodologies of the two studies that did examine these associations preclude any definitive conclusions from being drawn Experimentally Manipulat ed Sleep For reasons of experimental control, most studies of the sleep cognition relationship employ either a sleep deprivation or sleep restriction paradigm. In these

PAGE 126

126 paradigms, subjects are typically cognitively assessed initially, kept awake in a labor atory for an extended period of time ( 1 3 days) or allowed to sleep for a restricted amount of time for several nights (e.g., 5 nights of 4 hours sleep per night) and then cognitively assessed again. However, for several reasons, the majority of experimen tally manipulated sleep studies utilize convenience samples that exclude elders. Potential reasons for the lack of older adult subjects in such studies include: difficulty recruiting older adults due to the high demand of the studies, concerns about older of results obtained from these types of studies among older participants. On this last point, many deprivation/restriction studies focus specifically on young samples because they are more often employed in jobs in which sleep deprivation and/or restriction is required (e.g., the military, truck driving, etc.) and there is funding for research in these populations. As a note, while many of the below discussed experimen ts utilize PSG measurement of sleep, few examine the association between PSG indicators and cognitive functioning. Sleep deprivation/restriction paradigms are achieved through direct manipulation of time spent in bed. PSG devices are typically employed in these experiments as screening devices to detect the presence of occult sleep disorders. Sleep Deprivation following a single night of sleep deprivation to that of 12 youn g, healthy subjects allowed to sleep as usual (Horne, 1988). He reported that 1 night of sleep deprivation impaired performance on all subtest of the Torrance Test of Creative Thinking ( flexibility, strategy change, originality, generation of unusual ideas etc.) and also impaired word fluency.

PAGE 127

127 He concluded that a night of sleep loss can negatively impact divergent thinking ability (Horne, 1988). In a similar experiment, 30 younger, healthy subjects underwent 40 hours of total sleep deprivation (i.e., 1 ni ght) and were repeatedly assessed with a computerized version of the Stroop Color Word test (Cain, Silva, Chang, Ronda, & Duffy, 2011). The authors reported that reaction times were slowest in the incongruent trials and fastest in the congruent trials. Int erestingly, while performance deteriorated as a function of time awake, e xtended wakefulness did not significantly change the additional time needed to respond when the color and word did not match (Stroop interference), nor did it change the amount of fac ilitation when color and word matched. Consequently, the authors concluded that one night of sleep deprivation influences performance on the Stroop task by an overall increase in response time, but does not appear to impact the underlying processes of inte rference or facilitation. Interpretation may be that the degree to which executive function studied and the degree to which it is subserved by the prefrontal cortex (Cain et al., 2011). However, Harrison and Horne (1997, 1998, 1999) have run several sleep deprivation experiments in which they demonstrate acute sleep loss has a negative t he first experiment, 9 healthy, younger subjects underwent 36 hours of total sleep deprivation and sleep as usual (in a counterbalanced design). Word fluency and reading measures were repeatedly administered. Following a single night of total sleep depriva tion, fluency was significantly impaired compared to following normal sleep. The

PAGE 128

128 authors suggested that reductions in cerebral metabolic rate in the prefrontal cortex may be responsible for the effects of sleep loss on language skills (Harrison & Horne, 19 97). In a similar experiment, 50 younger, healthy subjects underwent 1 night total sleep deprivation and were tested using a sentence completion and word generation task. Results indicated that following 1 night of sleep loss, performance was hindered and Lastly, these authors tested younger, healthy subjects following either 36 hours of total ynamic, realistic marketing de cision game and a reasoning test Their results suggested n o effects of sleep deprivation on reasoning; however, following a single night of sleep deprivation, subjects performed worse on the marketing tasks especially in maneuvers that required flexibili ty, innovation, and updating information They again suggested sleep loss has a negative impact on the prefrontal cortex (Harrison & Horne, 1999). These effects of sleep deprivation on cognitive functioning have led to the conclusion that in young, healthy adults, 1 night of sleep deprivation produces similar effects on the prefrontal cortex as 60 years of aging (Harrison, Horne, & Rothwell, 2000). A study of healthy, younger adults has demonstrated that following 43 hours of total sleep deprivation sugges accurate from inaccurate information that was previous presented demonstrates dramatic declines (Blagrove, 1996). May and Kline (1987) designed an experiment to examine the effects of sleep loss on cognitive functioning, as compared to normal sleep and fatigue without sleep loss (May & Kline, 1987). They had 135 younger, healthy subjects complete 15 cognitive tests (measuring the domains of f lexibility, speed, verbal

PAGE 129

129 functioning, fluency, reasonin g, working memory, perceptual speed, and visuospatial ability ) following 2 nights total sleep deprivation, 1 night of normal sleep, and 1 day of fatigue producing activity with no sleep loss. The authors reported that f ollowing 2 nights of sleep deprivatio n, 8 cognitive tests showed decrements (visuospatial tasks, scanning, speed, originality, attention, working memory) 6 showed improvements ( logic, reasoning, directions) and 1 showed no change compared to no differences between normal sleep and fatigue w ithout sleep loss. The authors suggested that t he more complex and interesting a task the more incentive subjects ha d to complete it and the more effort they put forth thus compensating for any effects of sleep loss (May & Kline, 1987) In a study compa ring younger, healthy subjects undergoing 1 night total sleep deprivation to younger, healthy subjects allowed adequate sleep in the domains of f igural relations, working memory, trial making, auditory discrimination, and recognition it was found that fig ural relations, trail making, and letter recognition task performance were impacted negatively following sleep loss The authors suggest that sleep serves a function of cognitive restitution, particularly in the maintenance of attentional mechanisms (Wimme r, Hoffmann, Bonato, & Moffitt, 1992). While the above studies have shown some select impairments following total sleep deprivation (while also demonstrating a lack of effects on some tasks), other studies have failed entirely to find any cognitive conseq uences of total sleep loss. In a study of 61 younger, healthy adults (half of which underwent 1 night total sleep deprivation) no differences were found between groups on measures of intelligence (WAIS III), sustained attention (PASAT), or executive perfor mance (COWA, WCST, Booklet

PAGE 130

130 Category). Interestingly, subjects rated their effort equivalently across groups; however, the sleep deprived subjects rated their cognitive performance as worse than the normally sleeping subjects (Binks, Waters, & Hurry, 1999). The authors speculate that s leep loss only affects attention which can easily be compensated for through additional effort ; however, they fail to reconcile their findings of equal effort ratings across groups (Binks et al., 1999). As sleep deprivation ha s been suggested to differentially impact executive, or frontally mediated, cognitive tasks, Pace Schott and colleagues (2009) extensively examined executive functioning and working memory in young, healthy sleep deprived and non sleep deprived subjects (P ace Schott et al., 2009). They reported that on v erb al fluency, logical reasoning, T ower of London, temporal memory, digit span, N Back, Iowa Gaming, object alterations, sentence completion and Stroop tasks, there were no significant differences between i ndividuals who underwent 35 39 hours of sleep deprivation and those that slept normally. The authors suggest an unknown compensatory mechanism may be responsible for the lack of group differences (Pace Schott et al., 2009). Despite these null results, a 20 10 meta analysis confirmed that short term total (Lim & Dinges, 2010). The authors grouped cognitive functioning across 6 domains ( simple attention, complex attention, w orking memory, processing speed, short term memory, and reasoning ) from 70 published empirical investigations. They reported significant small to large effect sizes across domains and noted that

PAGE 131

131 complexity [were] following sleep deprivation (Lim & Dinges, 2010). However, all the aforementioned sleep deprivation studies focused on young, healthy samples. Webb and Levy (1982) and Webb (1989) conducted experiments to examine potential age differences in cognitive res ponse to sleep deprivation (Webb, performance on a variety of cognitive tasks (i.e., addition, visual search, word memory, word detection, reasoning, object uses, remote assoc iative s auditory vigilance, judgment, digit symbol, anagrams, long term memory ) were compared following 2 nights of sleep deprivation. It was reported that the o lder adult group showed greater deterioration following sleep deprivation than did the younger adults in vigilance, visual search, reaction times, word detection, addition, anagrams, and objects uses. This led larger performance decrements associated with deprivation in (Webb, 1985; Webb & Levy, 1982). While t he above summarized behavioral accounts of the cognitive consequences of sleep loss have provided evidence allowing researchers to speculate about the potential neural underpinnings, there have been a growing number of sleep deprivation studies have paired neuropsychological assessment and imaging protocols in the past decade. In a set of well designed experiments, Sean Drummond and colleagues (1999, 2000, 2001) examined both the behavioral and functional correlates of sleep deprivation in healthy, younger adults ( Drummond & Brown, 2001; Drummond et al., 2000; Drummond et al., 1999; Drummond, Gillin, & Brown, 2001). Using fMRI, the researchers tested young, healthy good sleeping adults following a night of good sleep and a night

PAGE 132

132 of sleep deprivation (35 hrs) with a verbal learning, divided attention, and arithmetic task. For verbal learning and divided attention, total sleep deprivation was associated with poorer performances and increased activation in the bilateral prefrontal cortex and parietal lobes. Furt her, lower level s of verbal learning impairments (i.e., better performance following sleep deprivation) were associated with increased additional activation. However, the arithmetic task following total sleep deprivation resulted in impaired performance an d decreased activation in the bilateral prefrontal cortex and parietal lobes. Thus cognitive performance following TSD [total sleep deprivation] with the specific pattern of adaptation depending o The authors hypothesized that of a cognitive task following TSD [total sleep deprivation] represents an adaptive cerebral compensatory res (Drummond & Brown, 2001; Drummond et al., 2000; Drummond et al., 1999; Drummond et al., 2001). In order to delineate the potential neural response to sleep deprivation, a follow up conducted (Drummond, Meloy, Yanagi, Orff, & Brown, 2005). Cerebral response to easy words on both sleep and sleep deprived nights were identical. However, following sleep deprivation bilateral i nferior frontal gyrus, bilateral dorsolateral prefrontal cor tex, and bilateral inferior parietal lobe showed increased activation to hard words. Complicating the matter, the authors concluded that (Drummond & Brown, 200 1).

PAGE 133

133 In another experiment employing fMRI technology, 8 healthy, young subjects were tested on a sustained attention tasks following 1 night total sleep deprivation, normal sleep, and normal sleep plus caffeine (Portas et al., 1998). Interestingly, f ollowi ng sleep deprivation the ventrolateral thalamus showed the highest levels of activation compared to both normal sleep and caffe i ne (which did not differ from each other) No differences were observed between the groups on attentional task performance. The authors maintain performance (Portas et al., 1998). In a complimentary line of research, the functional neural response to sleep deprivation has been examined using PET metho dology (Patiau, 1998; Thomas et al., 2000). In these studies younger, healthy subjects underwent one night sleep deprivation and performed either a serial addition/subtraction task or word generation task while being scanned. In both instances, performance was significantly negatively impacted by sleep deprivation. Additionally, PET scan revealed decreased blood flow to prefrontal areas and d ecreased whole brain metabolism and regional metabolisms (thalamus and prefrontal cortex and posterior parietal lobes ) during cognitive tasks. The authors neurobehavioral function of sleep in humans is (Patiau, 1998; Thomas et al., 2000). Like the methodology pr eviously summarized, investigation into the relationship between experimentally manipulated sleep, via sleep deprivation, and waking cognitive functioning has provided mixed results. Authors have reported differences between sleep deprived subjects and nor mally sleeping subjects on all [e.g., (Drummond et al.,

PAGE 134

134 2000; Drummond et al., 1999; Drummond et al., 2001; Harrison & Horne, 1998; Harrison & Horne, 1997, 1999)], some [e.g., (Cain et al., 2011; May & Kline, 1987; Wimmer et al., 1992)], and none [e.g., (B inks et al., 1999; Pace Schott et al., 2009)] of the cognitive measures administered. Further, some studies have indicated hypoactivation (Drummond et al., 2000; Drummond et al., 1999; Harrison & Horne, 1997; Patiau, 1998; Thomas et al., 2000), while other s suggest hyperactivation (Drummond et al., 2000; Drummond et al., 1999; Drummond et al., 2001; Drummond et al., 2005; Portas et al., 1998), following sleep deprivation. Jones and Harrison (2001) summarized the extant sleep deprivation work nicely by stati ng present many inconsistencies, task classification is often ambiguous and, in the absence of any unifying explanation at the level of cognitive mechanisms, the overall picture is one of a disparate range of impairment following sle ep loss Harrison, 2001). Sleep Restriction In response to criticism that total sleep deprivation studies impose unrealistic conditions on subjects and, as a result, cognitive consequences observed are ecologically unimportant, researchers have begun employing sleep restriction paradigms more frequently. Sleep restriction (sometimes referred to as partial sleep deprivation) is an experimental methodology which curtails the amount of time an individual is allowed to sleep across multiple days. As such, sleep restriction is believed to be a better proxy for naturally occurring sleep. However, these studies are still less abundant than total sleep deprivation studies, likely due to their time and resource intensive nature.

PAGE 135

135 In one sleep restriction e xperiment, the sleep of 25 healthy, young adults was restricted to 6 hours per night for 7 consecutive nights (Vgontzas et al., 2004). Daytime cognitive functioning was assessed via the Psychomotor Vigilance Test. The researchers also monitored proinflamma tory cytokine levels with daily blood draws. Results indicated reductions in Psychomotor Vigilance Test performance and increases in proinflammatory cytokine levels (i.e., IL 6 and TNF a) across the 7 days of sleep restriction (Vgontzas et al., 2004). Such results indicate that very modest restriction to sleep can have negative consequences for both cognitive functioning and neurobiology. Dinges and colleagues (1997) restricted the sleep of 16 young, healthy subjects approximately 33% of their normal sleep (i.e., to about 5 hours per night) for 7 consecutive nights (Dinges, 1997). Again, daytime cognitive functioning was assessed with the Psychomotor Vigilance Test. Psychomotor Vigilance performance (RTs and lapses) declined across the 7 days. There was mil d evidence of asymptote being reached after 3 days (i.e., leveling off of deterioration ); however, t he authors note d that the predominate time trend was linear, suggesti ve of a cumulative effect of restricted sleep on vigilance. The authors concluded that consecutive nights had a clearly measurable effect on neurobehavioral markers of (Dinges, 1997). In an elaborate design, the effects of rate of 8 hours of sleep loss [i.e., no sleep loss, slow sleep loss ( 6 h ou rs TIB/night for 4 nights ) intermediate ( 4 h our s TI B for 2 nights ) and rapid sleep loss ( 0 h ou rs TIB for 1 night )] on a p robed recall memory task, psychomotor vigilance task, and divided attention task was investigated (Drake et al., 2001). It was found tha t more rapid sleep loss produced the most substantial deficits in

PAGE 136

136 alertness, memory, and attention. The authors suggest that because the cognitive compensatory adaptive mechanism operating with the While these mechanisms are unknown, potential candidates include: changes in sleep architecture and additional effort (Drake et al., 2001). While the above discussed study focus on young, healthy adults, Bliese and colleagues ( 2006) examined age related changes in response to modest sleep restriction. Again, cognitive functioning was measured with the Psychomotor Vigilance Test. Interestingly, older adults displayed less pronounced effects of sleep restriction on their RT s than younger adults H owever, 62 was the oldest adult in cluded in the study so aging effects must be interpreted cautiously Never the less, the authors suggest that o differences in perf (Bliese, Wesensten, & Balkin, 2006). In summation, the experimental evidence regarding the sleep cognition relationship gained through studies employing sleep restriction methodology have consistently yielded results indicating an impact on vigilance (Bliese et al., 2006; Dinges, 1997; Drake et al., 2001; Vgontzas et al., 2004), which may be blunted in older adults (Bliese et al., 2006). However, the evidence is equivocal regarding whether (Drake et al., 2001) or not (Dinges, 1997) adapti on to sleep restriction is normative. Regarding potential mechanisms underlying the relationship between sleep restriction and cognitive functioning, Banks and Dinges (2007) summarized the evidence by definitive evidence of what is accumulating and destabilizing cognitive functions over time when sleep is regularly restricted

PAGE 137

137 Several studies have directly compared the cognitive consequences of sleep restriction to total sleep deprivation. Forty eight younge r, healthy adults were randomized to either 3 days of total sleep restriction or 14 days of sleep restriction to either 4, 6, or 8 hours per night (Van Dongen, 2003) Cognitive functioning was measured with the Psychomotor Vigilance Test (sustained attenti on), Digit S ymbol (working memory), and serial addition/subtraction (cognitive throughput) It was found that c hronic sleep restriction resulted in cognitive deficits in all domains. These deficits were linear and did not reach any sign of leveling off. Re ducing sleep to 4 hours per night for 14 nights resulted in working memory and attention deficits equivalent to 2 days total sleep deprivation while cognitive throughpu t was equivalent to 1 day total sleep deprivation Attention and working memory deficit s of 6 h ou rs restriction were found to be equivalent to 1 night total sleep deprivation. Based on this evidence, it can be concluded that chronically restricting sleep has serious neurobehavioral consequences (Van Dongen, 2003). However, in another study of chronic sleep restriction which compared the cognitive functioning (i.e., l ogical reasoning, vigilance, visual search ) of 1 night total sleep deprivation to that of long term sleep reduction (5.2 h ours per night for 21 night s 4.3h ours per night for 4 n igh ts and 5.3h ours per night for 18 night s ) it was reported that sleep restriction did not impact reasoning or vigilance, while total sleep deprivation had an impact on all cognitive measures (Blagrove, Alexander, & Horne, 1995). Like earlier investigati ons with null results, the authors speculate that added effort and compensation during sleep restriction was responsible for a lack of effect of sleep restriction.

PAGE 138

138 Sleep Cognition Theories The above referenced studies examining the relationship between sleep (both subjective and objective naturally occurring and experimentally manipulated sleep) and daytime cognitive functioning (across many domains) has led to the formation of several informative hypotheses concerning sleep and cognitive functioning. Th e controlled attention hypothesis (Pilcher, Band, Odle Dusseau, & Muth, 2007) attempts to synthesize the extant literature and reconcile discrepant findings by suggesting that p complex and difficult tasks (which often times fail to show effects of sleep loss) which An alternativ e synthesis of the literature could be labeled the neuropsychological hypothesis (Harrison & Horne, 2000; Jones & Harrison, 2001; Lim & Dinges, 2010). This model suggests that sleep loss results in focal impairment in functions subserved by the prefrontal cortex, beyond any impairment in attention or vigilance. In essence, this stance posits that executive tasks are vulnerable to sleep loss beyond any detriments that may be reasonably expected from lowered levels of arousal. Evidence supporting the neurops ychological hypothesis comes from studies demonstrating select impairments following sleep loss on tasks believed to be frontally dependent (Harrison & Horne, 2000; Harrison et al., 2000) and imaging studies demonstrating marked changes in functional activ ity in the frontal lobes following sleep loss (Drummond et al., 2000; Drummond et al., 1999; Drummond et al., 2001). Yet another proposed explanation for the observed sleep cognition relationships is the vigilance/arousal hypothesis. Attention, which is needed for the performance of

PAGE 139

139 many other cognitive tasks, is mediated by arousal -a common correlated feature of disturbed sleep (Bonnet & Arand, 2010; Richardson 2007 ) In general, individuals with a disturbed sleep display physiological hyperarousal. This hyperarousal that disrupts nighttime sleep may also promote daytime cognitive functioning (Bonnet & Arand, 2010; Richardson 2007 ) While arousal may benefit some aspects of cognition, other cognitive abilities may continue to be negatively affected by sleep disturbance (Bonnet & Arand, 2010). Consistent findings of altered vigilance and arousal levels following sleep loss (i.e., Psychomotor Vigilance Test performance) lend support for this hypothesis (Dinges, 1997; Durmer & Dinges, 2005). Proponents of this theory suggest that most higher order cognitive abilities are dependent, to some extent, on proper levels of arousal and vigilance. Lastly, the extant evidence examining the relationship between sleep and cognitive functioning has been synthesized into the wake state instability hypothesis (Dinges, 1997; Durmer & Dinges, 2005; Goel et al., 2009). This theory suggests that the cognitive deficits observed as a result of sleep loss occur due to the interaction of the drive to maintain alertness and th e homeostatic drive to initiate sleep. As the sleep drive increases in strength, it becomes more and more of a hindrance to the maintenance of maintain alertness is These various theories about the role of sleep processes in regulating and maintaining cognitive functioning may not be mutually exclusive (Lim & Dinges, 2010). The controlled attention and vigilance/arousal hypotheses are essentially parallel descriptions of the same phenomena. As previously stated, many higher order cognitive

PAGE 140

140 functions may rely on the appropriate levels of attention and arousal. Similarly, it has been suggested that imp airment of the prefrontal cortex may cause decrements in attention and vigilance (Boonstra, Stins, Daffertshofer, & Beek, 2007). As such, it would appear that all three of these theories (attention control, vigilance/arousal, and neuropsychological hypothe ses) acknowledge the presence of detriments in lower level and executive functions following sleep loss, and only differ in their stance on which impairment is preeminent. Lastly, the wake state instability hypothesis is more of a description of the proces s through which impairments may occur, than an account of specific mechanisms responsible for the impairments. In sum, sleep appears to impede arousal/vigilance/attention and prefrontal functioning, potentially through instability of the neurobiological sy stems responsible for attentional and sleep drives. Please refer to Table 3 1 for a summary of the potential mechanisms linking sleep and cognitive functioning. The Current Investigation It has been reported that age and level of cognitive functioning ar e related to practice related learning in older adults (Yang et al., 2006), while trait and state anxiety are not (Yang et al., 2009). Additionally, sleep has been related to level (Blackwell et al., 2006; Tworoger et al., 2006) and long term decline in l ate life cognitive functioning (Cricco et al., 2001). N o known ability to learn through exposure (i.e., demonstrate practice related learning). In fact, scholars have suggested participants are well trained in the (Harrison & Horne, 2000). However, as practice related learning has and sleep depth and continuity are associated with frontal

PAGE 141

141 lobe functioning (Jones & Harrison, 2001), neuronal plasticity (Wagner et al., 2004), hippocampal memory processing (Hobson & Pace Schott, 2002), and attention/arousal (Bonnet & Arand, 2010), the association betw een sleep and practice related learning appears plausible and worth investigation. The current investigation seeks to examine subjective sleep as an individual difference predictor of practice related learning in late life. Specific aims of this study inc lude: (1) To replicate previous findings (Yang & Krampe, 2009; Yang et al., 2006; Yang et al., 2009) of significant practice related learning in processing speed and reasoning in late life, ( 2 s to their: (a) initial level of cognitive performance, and (b) their ability to benefit from repeated exposure to cognitive stimuli (i.e., ability to demonstrate practice related learning) after controlling for salient individual difference variables (i.e ., age, education, estimated IQ, etc.). Given the previously described equivocal association between sleep and cognitive life practice related learning is not possible. It may be that indi viduals whose sleep is better (i.e., more time asleep, less awake time) will display higher initial levels of cognitive functioning and will also benefit the most from repeated cognitive practice, because not only is sleep loss associated with cognitive de crements [i.e., (Banks & Dinges, 2007; Goel et al., 2009)] but sleep gain is associated with positive cognitive changes [i.e., (Ellenbogen, 2005; Wagner et al., 2004; Walker et al., 2002)] Conversely, individuals who spend more time awake during the night may also display higher levels of cognitive functioning and may display more pronounced practice related learning due to compensatory mechanisms

PAGE 142

142 (Drake et al., 2001; Drummond et al., 2000) or heightened arousal responsible for both difficulty sleeping and increased daytime vigilance (Altena, Van Der Werf et al., 2008; Bonnet & Arand, 2010). While the current investigation will be able to examine the relationship between subjective sleep variables and late life cognitive functioning, it does not directly me asure any of the suspected underlying mechanisms that may be responsible for any observed associations. As such, discussion of potential mechanisms will occur only at the theoretical level. Methods General Study Design This study represents a secondary an alysis of the Active Adult Mentoring Program (Project AAMP). The primary objective of Project AAMP was to test the efficacy of a social cognitive lifestyle intervention to increase moderate intensity exercise in older adults. Participants were randomly ass igned to either an Active Lifestyle intervention arm (receiving weekly, group based behavioral counseling) or a Health Education arm (receiving appropriately matched health education). The current study utilizes data from the initial 18 weeks of the study, including a baseline week of observation prior to group assignment, 16 weeks of intervention, and a subsequent week of observation following the intervention period. The study protocol was approved by the appropriate university institutional review boards Procedure Individuals that expressed interest in study participation were initially screened by telephone (see below). Following telephone screening, qualified participants were consented and completed a baseline assessment. This baseline assessment inc luded seven consecutive nights of self reported sleep monitoring, one in person computerized

PAGE 143

143 cognitive assessment, and administration of demographic and descriptive self report questionnaires. Next, participants were randomized to either the Active Lifesty le or Health Education arm of the intervention. Each intervention arm consisted of sixteen weekly group meetings [see (Aiken Morgan, 2008; Buman, 2008 ; Buman et al., 2011) for more information]. During the intervention period, each participant continued to monitor their sleep nightly. Prior to each group meeting, all sleep logs were collected and checked for compliance. Similarly, either before or after each group meeting all participants completed the computerized cognitive battery. Lastly, each participan t completed a post treatment week of assessment that included seven nights of sleep monitoring and one computerized cognitive assessment. Thus, this study included 18 consecutive weeks of sleep monitoring (i.e., up to 126 days), paired with 18 weekly cogni tive assessments. Participants Study participants include 87 adults aged 50 years and greater who participated in Project AAMP. Potential enrollees responded to community based health promotion recruitment delivered through local media outlets. While part icipants were not recruited or screened based on their sleep characteristics, older adults present with great heterogeneity and have increased prevalence of disturbed sleep (Morgan, 2000). Thus, the sample is likely to include a wide range of elder sleeper s, including both good and poor sleepers. To ensure their suitability for the study, subjects went through a thorough screening process that included many inclusion and exclusion criteria. Inclusion/Exclusion Criteria All potential participants were screen ed by telephone to exclude individuals based on the following criteria: severe dementing illness, history of significant head injury (loss

PAGE 144

144 of consciousness for more than 5 minutes), neurological disorders disease), inpatient psychiatric treatment, extensive drug or alcohol abuse, use of an anticholinesterase inhibitor (such as Aricept) severe uncorrected vision or hearing impairments, terminal illness with life expectancy less than 12 months, major medical illnesses cardiovascular disea se, pulmonary disease requiring oxygen or steroid treatment, and ambulation with assistive devices. Telephone screening include d the 11 item Telephone Interview for Cognitive Status [ TICS; (Brandt et al., 1988) ], utilizing a cut off score of 30 points to d ifferentiate mild dement ia from cognitive ly intact (Brandt et al., 1988) All study participants were required to se lf report sedentary lifestyle [defined of moderate or vigorous physical activity during the previous 6 months ( Physical Activity Guidelines Advisory Committee Report, Part A : Executive Summary 2009) ]. All i ndividuals were required to come to the laboratory with a note from their primary care physician acknowledging their ability to participate in the study prior to formal enrollment Demographic and Descriptive Measures Each study participant provided demographic data through means of a telephone screening instrument. Information regarding participant age (measured in years since birth), gender (male or female), and education level (years of education ) were collected. During the first in person session, all individuals completed the North American Adult Reading Test [ NAART ; (Blair & Spreen, 1989) ], yielding a pre morbid IQ estimate (that has been found to correlate between 0.40 and 0.80 with other measures of intelligence). Th e Beck Depression Inventory, Second Edition [BDI II; (Beck et al., 1996) ] was administered to assess depressive symptomatology. The BDI II consists of 21 groups

PAGE 145

145 of statements related to cognitive and somatic depression symptoms. The BDI II is a commonly u sed self report measure of depression in both younger (Beck et al., 1996) and older adults (Segal et al., 2008) T he State Trait Anxiety Inventory [ STAI ; (Spielberger, 1983) was administered to assess current (state) and typical (trait) anxiety symptoms. D escriptive Statistics The final sample included 87 adults aged 50 years and older. Mean age for the entire sample was 63.33 years, range = 50 87 years. The sample was highly educated, average years of education of 16.14, predominately female, 82%, of above average intelligence, and evinced few depressive and anxiety symptoms (i.e., few depressive and anxiety symptoms). Please refer to Table 3 2 for a complete list of demographic/descriptive statistics. Measures Cognitive Measures The Letter Series task (Th urstone, 1962) is primarily a measure of reasoning In this task, participants ha d to identify the pattern for a series of letters. Participants we re asked to choose the letter that would continue an established pattern (A B D A B D A B ___?) in a series o f letters from five answer choices. Participants were given four minutes to complete as many items as possible. The performance score wa s the number of correct responses. Processing speed was assessed with the Symbol Digit and Digit Copy tests (Smith, 1982 ). These tests consist of matching symbols that are paired with numbers ( Symbol Digit) or numbers paired with same numbers (Digit Copy) as quickly as

PAGE 146

146 possible There was a 120 second time limit for each task and the performance score was the total number o f correct pairings made by the participant The Simple and Choice Reaction Time task (Hultsch et al., 2000) was used to assess reaction time/processing speed The Simple Reaction Time (SRT) task presents y a signal stimulus (+) in the middle of the screen. Participants were instructed to press a key with their preferred hand as quickly as possible when the signal stimulus (+) appear ed The Choice Reaction Time (CRT) task present ed respondents with a warnin d into a circle and the location of the circle was randomly equalized across trials. Respondents were instructed to press a key c orresponding to the location of the circle as quickly as possible. 10 practice trials were followed by 50 test trials. The outcome measure s for SRT and CRT was the mean latency of the total correct test trials. All cognitive measures were computer adminis tered. Administration occurred once per week for 18 consecutive weeks. Computerization of the cognitive measures was d one using DirectRT experimental generation program (Jarvis, 2008a) All computerized cognitive measures were then compiled and administere d via MediaLab experimental implementation program (Jarvis, 2008b) In an attempt to minimize practice effects d ue to memorization commonly found in repeated cognitive assessments (Salthouse et al., 2004) fourteen alternate forms of each test were used an d rotated such that the same version of any given test was not given within 6 weeks of each other. The alternate forms were constructed to be comparable in difficulty and cognitive resources needed to complete them and have been shown to have high test ret est reliabilities (Allaire &

PAGE 147

147 Marsiske, 2005; McCoy, 2004) Both processing speed and reasoning measures were selected to be practiced due to their inclusion in previous practice related learning investigations (Yang & Krampe, 2009; Yang et al., 2006; Yang et al., 2009). Sleep Measures Participants completed sleep diaries (Lichstein et al., 1999) each morning for the entirety of Project AAMP (i.e., 18 weeks/126 days). The sleep diaries provide subjective estimates of the following sleep parameters: (1) tota l sleep time (TST): the total amount of time asleep during the night (TST = total time in bed sleep onset latency wake time after sleep onset and terminal wakefulness ); (2) total wake time (TWT): the total awake time in bed (TWT = sleep onset latency + wake time after sleep onset + terminal wakefulness ); (3) sleep onset latency (SOL): the time it took to fall asleep after laying down with the intention of going to sleep; (4) wake time after sleep onset (WASO): the total amount of time spent awake durin g the night from the time sleep was first initiated until the final wake up (5) sleep quality rating (SQR): overall rating of the quality of sleep (from 1= very poor to 5 = excellent) (6) terminal wakefulness (TWAK): the amount of time spent awake laying in bed in the morning following final wake up, (7) number of awakenings (NWAK): total number of nocturnal awakenings, and time in bed (TIB): total amount of time spent in bed For each participant, average level sleep variables (averaged up from 5 blocks of sleep data; see below) were calculated. Due to conceptual [i.e., potential consequences of sleep loss and sleep gain (Ellenbogen, 2005)] and pragmatic (i.e., high multicollinearity) issues, only TST and TWT were employed as individual difference predict ors of practice related learning. Please see Table 3 3 for a complete listing of inter correlations among sleep variables and raw mean level sleep values over 5 blocks of assessment.

PAGE 148

148 Analyses Preliminary Analyses Prior to statistical analyses to address e ach main aim, preliminary analyses were conducted. Preliminary analyses: (1) examined normality of the data, (2) examined rates of missingness among the data, (3) examined attrition (differences between attrited and non attrited subjects), and ( 4 ) examined any systematic changes in sleep across the study period. Preliminary analysis examined the cognitive data (i.e., dependent variables) for normality. All cognitive data were screened for outliers at the intraindividual level (i.e., within person from tria l to trial). Further, all cognitive variables were also screened for potential outliers at the interindividual level (i.e., between persons). Interindividual outliers were replaced with their respective 3 standard deviation values, while intraindividual ou tliers were simply removed from the dataset prior to calculation of occasion specific values. Skewness and kurtosis values were also examined using generally agreed upon criteria (i.e., skewness and kurtosis values less than 1.0) (Field, 2005). Normality i n independent variables (i.e., covariates and sleep predictors) is less of a concern in structural equation modeling (SEM) analyses (Tabachnick & Fidell, 2001), and as such these variables were only screened to detect impermissible values that likely repre sent data entry errors and to render the data more appropriate for parametric analyses. Extreme outliers (cases representing beyond the top and bottom 0.25% of the distribution) were rescaled to the cut points for the 0.25 and 99.75 percentiles of the dist ribution. Such a practice was performed to preserve participant rank orders, improve skewness and kurtosis, and reduce the potential leverage and influence of extreme outlier cases on relationships to be investigated (Hastings,

PAGE 149

149 Mosteller, Tukey, & Winsor, 1947). For the purposes of this report, data were cleaned at the daily and weekly level. Rates of missingness were calculated as a descriptor of the data, because all available data was utilized. All structural equation modeling (SEM) was implemented throu gh AMOS (Arbuckle & Wothke, 1999) via full information maximum likelihood (FIML) estimation methodology. Analysis of attrition was conducted to examine whether there were differences between study completers (those whom were present at posttest) and study attriters (those whom dropped out before reaching posttest) on all cognitive, descriptive/demographic variables, and sleep variables from baseline to posttest. To examine for significant systematic changes in sleep over the course of the 5 block observat ion period, a multilevel growth model was estimated (Bryk & Raudenbush, 1992; Singer & Willett, 2003) for each sleep independent variable. We first examined whether there were significant temporal trends in sleep over the 5 block (i.e., 126 day) period. I f no systematic time related changes were found, then models of occasion to occasion variability in sleep would not need to control for temporal trends; if temporal trends were found, then corrections for change (e.g., mean level sleep with temporal trends residualized out) would be preferable. Thus, a final step prior to the main analyses below was to estimate an unconditional growth model for sleep. Non significant temporal trends indicated that mean level sleep was an unbiased estimate. The presence of s ignificant temporal trends in sleep suggested alternative indices of sleep (i.e., mean level of sleep with temporal trends residualized out) may be preferable individual difference predictors. For consistency with the cognitive growth models, the

PAGE 150

150 sleep var iables were combined to form blocks of data. These blocks were formed in the same manner as the cognitive blocks (i.e., Block 1 = baseline, week 1, and week 2; Block 2 = week 3, week 4, week 5, and week 6; Block 3 = week 7, week 8, week 9, and week 10; Blo ck 4 = week 11, week 12, week 13, and week 14; and Block 5 = week 15, week 16, and posttest). The parent study (Project AAMP) for the current paper was designed to test the efficacy of a social cognitive lifestyle intervention to increase moderate intensit y exercise in older adults. Participants were assigned to either an Active Lifestyle intervention or a Health Education. As such, concerns regarding differing cognitive trajectories between these groups may be present. However, previous work s have shown fe w baseline to posttest fitness effects by group status (Buman, 2008) and no baseline to posttest cognitive effects by group status (Aiken Morgan, 2008). Thus, group status was not controlled for in subsequent analyses. Main Analyses To attempt to replica te the findings of significant practice related learning in older adults (Yang et al., 2006), separate univariate growth curves ( one for each cognitive variable) w ere parameterized to model the level and rate of change in cognitive functioning across the 1 8 week study period (Aim 1) This was done through a SEM approach (Tabachnick & Fidell, 2001). Both linear and quadratic estimates of change were examined. As these models contained no predictor variables they are termed unconditional growth models. Signif icant practice related learning was indicated by tests of the slope being non zero. Further, random slopes were estimated and were examined to determine if the slope related variance is non zero. There must be significant individual differences (i.e., rand om variance) in the slopes and/or intercepts

PAGE 151

151 prior to estimation of models that predict either an intercept or slope term. Model fit was evaluated by examining the following fit statistics: 2 (chi squared), ratio of 2 value to degrees of freedom (CMIN/DF ), comparative fit index (CFI), Tucker Lewis index (TLI), incremental fit index (IFI), and root mean square error of approximation (RMSEA). Good model fit was indicated by models with non significant chi square statistic, CMIN/DF < 2, CFI, TLI, and IFI > 0 .90, and RMSEA < .08. Please see Figure 3 1 for a graphical representation of the univariate unconditional growth curves. Next, to examine sleep as an individual difference predictor of practice related learning in late life (Aim 2), predictor variables w ere added to the unconditional growth models estimated in Aim 1 to form conditional growth models. The individual difference covariates of age, gender, education, estimated IQ, anxiety, and depression were added to control for common factors associated wit h late life cognitive functioning (Yang et al., 2006; Yang et al., 2009) Additionally, mean level TST and TWT were added to aim a) of cognitive functioning and rate of change (Aim 2, sub aim b) in cognitive functioning. Variables were entered hierarchically to the models. Initially, all covariates (i.e., age, gender, education, estimated IQ, anxiety, and depression) were added to the models. Subsequently, TST and TWT were both added to the models. Such hierarchical model building allowed for examination of potential improvements in model fit. Each variable was evaluated based on the significance of their paths to (a) practice intercept, (b) practice linear slope, and (c) practice quad ratic slope. Significant paths from these exogenous variables to intercept and slopes indicate significant prediction of practice related learning. Variables were also be evaluated based on their overall consequences

PAGE 152

152 to model fit statistics (i.e., 2 CMIN /DF, CFI, TLI, IFI, and RMSEA). Please see Figure 3 2 for a graphical representation of the final univariate conditional growth curves. Based on previous research (See Chapter 2) all growth curve analyses (i.e., Aim 1 and Aim 2) utilized cognitive data for med into composite blocks of surrounding occasions. Blocks were constructed in the following manner: Block 1 = baseline, week 1, and week 2; Block 2 = week 3, week 4, week 5, and week 6; Block 3 = week 7, week 8, week 9, and week 10; Block 4 = week 11, wee k 12, week 13, and week 14; and Block 5 = week 15, week 16, and posttest. The blocks were constructed in this manner so that each block would contain roughly equal numbers of occasions and to allow the earliest occasions (i.e., occasions 1 3) and the lat est occasions (i.e., occasions 15 17) to have slightly more influence on their respective block scores. Five blocks were constructed as this number of occasions is in accordance with other higher order latent growth models published in the psychology and aging literature (Christensen et al., 2004; Hofer et al., 2002). Results Preliminary Analyses Normality Descriptive statistics revealed the presence of numerous sample outliers and suggested the cognitive data were both skewed and kurtic. An iterative approach to outlier trimming resulted in cognitive data that were normally distributed. Please refer to Table 3 4 for cognitive descriptive and normality statistics pre and post outlier trimming procedures, and for raw cognitive scores for each measure at each occasion. In total, 90 cognitive variables were cleaned for 87 subjects. This produced a total of 7,830 potential data points that could have been trimmed. Only 69 actual data points were

PAGE 153

153 trimmed; resulting in 0.88% of the data being altered. Followin g data cleaning procedures, all RT based data (i.e., N Back, Simple and Choice RT, and Trails A and B) were transformed such that higher scores were indicative of better performance. This was done to place all cognitive measures on a similar metric and to improve interpretability of the main study Aims. Inspection of the TWT data revealed 18 data points in need of trimming. Inspection of the TST data revealed 41 data points in need of trimming. In total, 2 sleep variables (i.e., TST and TWT) were trimmed f or 126 occasions for 87 subjects. This produced a total of 21,924 potential data points that could have been replaced. Only 59 actual data points were replaced d ue to non permissible values or extreme values, result ing in 0. 27 % of the sleep data being Wins orized (Hastings et al., 1947) Please refer to Table 3 3 for TST and TWT descriptive and normality statistics pre and post Winsorizing procedures. Of note, while normality statistics improved post data replacement, TWT was still largely skewed and kurtic. However, as the normality of independent variables is not as critical in SEM analyses (Tabachnick & Fidell, 2001), and as the normality concerns arise out of TWT data being confined to normal bounds, raw sleep data was used in all subsequent analyses (whe n significant temporal trends were not present). Missing Data Eighty seven subjects were enrolled in the current investigation. Throughout the repeated cognitive assessments, participation waxed and waned considerably. While missing data rates were relativ ely high for each cognitive measure across the 18 consecutive weeks of testing (see Chapter 2, Figure 2 3), the current paper utilized composite scores of adjacent occasions of cognitive data formed into 5 blocks. If any data point was present in any occas ion for a given block, that block would not be

PAGE 154

154 missing. As a result, the blocked cognitive dependent variables contain very little missing data. Block 1 contained 100% data, Block 2 contained 94% data, Block 3 contained 85% data, Block 4 contained 78% data and Block 5 contained 80% data. As the sleep variables employed in the current paper represent person level mean sleep, and as every subject contributed at least one week of sleep data, there are no missing sleep data for the current analyses. Similarly, no demographic covariates contained missing data points. Attrition As is shown in Table 3 5, comparing demographic/descriptive characteristics (age, education, gender, estimated IQ, depression, and anxiety) of individuals who completed posttest (n = 68) t o those that failed to complete posttest (i.e., attriters; n = 19), independent samples t more depressive symptoms than did study completers, t (85) = 2.17, p = 0.03. It is important to note that if correction for multiple comparisons had been conducted (e.g., Bonferroni), even this difference would not have reached significance. Study completers and attriters did not significantly differ in their performance on any Block 1 cognitive measure (Number Copy, Symbol Digit, Letter Series, Simple and Choice RT) or in their mean TST or mean TWT, all > 0.05. Full information maximum likelihood (FIML) estimation methodology was employed in all subsequent analyses, as such methodology has been shown to produce reliable estimates even when data are not missing at random and is preferable to listwise deletion and purposive estimation (Graham, 2009).

PAGE 155

155 Systematic Sleep Changes To examine the daily TST and TWT data for any systematic changes across the 5 b lock study period separate multilevel growth models (Bryk & Raudenbush, 1992; Singer & Willett, 2003) were estimated. These models predicted change in sleep with time (linear block and quadratic block) and error indicators alone through implementation of t he following equation: Sleep it = 00 + 10 (Time) + 20 (Time 2 ) + r 1i (Time) + r 2i (Time 2 ) + r oi + e it The multilevel growth model for TST revealed no systematic changes across the 5 block (i.e., 126 day) study period, Linear Time, = 1.85, SE = 1.72, t (245.5 7) = 1.07, p > .05 Quadratic Time, = 0.31, SE = 1.54, t (182.06) = 0.20, p > .05 As such, raw mean level 5 block TST was employed in future analyses. The multilevel growth model for TWT also revealed no systematic changes across the 5 block (i.e., 126 day) study period, Linear Time, = 2.24, SE = 1.21, t (285.17) = 1.85, p > .05 Quadratic Time, = 0.75, SE = 1.04, t (222.72) = 0.73, p > .05 As such, raw mean level 5 block TWT was employed in future analyses. Please refer to Table 3 6 for a complete listing of estimates obtained from the multilevel growth models. Main Analyses Aim 1: Unconditional Growth Curves All unconditional latent growth curve analysis results are presented in Table 3 7. Please refer to this table for all intercept and slope es timates, and correlation coefficients between intercept and slope estimates. Additionally, please see Figure 3 3 for a graphical representation of standardized model implied change in each cognitive measure across the study period.

PAGE 156

156 Number Copy. Fit statist ics indicated good model fit for the Number Copy 2 (10) = 8.86, p > .05, CMIN/DF = 0.87, CFI, TLI, and IFI all p < .01, suggesting that the average Block 1 starting value was significantly greater than zero. The mean estimate of the linear slope was 0.10, p > .05, suggesting non significant linear growth across blocks in Number Copy performance. However, the mean estimate of the quadratic slope was 1.75, p < .01, suggesting s ignificant quadratic change across blocks in Number Copy performance. Specifically, significant growth was observed during the first two blocks of assessment, followed by decelerated growth throughout the end of the study period. The intercept related vari ance was 13.99, p < .01, suggesting significant individual differences in Block 1 Number Copy performance. The linear slope related variance was 0.54, p > .05, suggesting non significant individual differences in linear change in Number Copy performance. T he quadratic slope related variance was 5.66, p = .066, suggesting marginal individual differences in quadratic change in Number Copy performance. Symbol Digit. Fit statistics indicated good mode l fit for the Symbol Digit 2 (10) = 13.92, p > .05, CMIN/DF = 1.40, CFI, TLI, and IFI all p < .01, suggesting that the average Block 1 starting value was significantly great er than zero. The mean estimate of the linear slope was 0.01, p > .05, suggesting non significant linear growth across blocks in Symbol Digit performance. However, the mean estimate of the quadratic slope was 2.00, p < .01, suggesting significant quadrati c change across blocks in Symbol Digit performance. Specifically, significant growth was observed

PAGE 157

157 during the first two blocks of assessment, followed by decelerated growth throughout the end of the study period. The intercept related variance was 12.62, p < .01, suggesting significant individual differences in Block 1 Symbol Digit performance. The linear slope related variance was 1.68, p < .01, suggesting significant individual differences in linear change in Symbol Digit performance. Lastly, the quadratic slope related variance was 7.82, p < .01, suggesting significant individual differences in quadratic change in Symbol Digit performance. Letter Series. The LGC model for Letter Series would not converge with correlations to the quadratic slope being esti mated. This was likely due to negative quadratic slope variance. As a result, the correlations between the intercept and linear slope and the quadratic slope were not estimated and the quadratic slope variance was set to zero. Fit statistics indicated adeq uate model fit for the Letter Series LGC model. 2 (13) = 44.61, p RMSEA = 0.17. The mean intercept was 8.03, p < .01, suggesting that the average Block 1 starting value was significan tly greater than zero. The mean estimate of the linear slope was 0.18, p > .05, suggesting non significant linear growth across blocks in Letter Series performance. However, the mean estimate of the quadratic slope was 2.10, p < .01, suggesting significant quadratic change across blocks in Letter Series performance. Specifically, significant growth was observed during the first two blocks of assessment, followed by decelerated growth throughout the end of the study period. The intercept related variance was 10.20, p < .01, suggesting significant individual differences in Block 1 Letter Series performance. The linear slope related variance was 0.47, p < .01, suggesting significant individual differences in linear change in Letter

PAGE 158

158 Series performance. Lastly, a s stated above, the quadratic slope related variance was set to zero. Simple RT. The LGC model for Simple RT would not properly converge with correlations between the intercept, linear slope and quadratic slope being estimated. As a result, these correlat ions were not estimated. Fit statistics indicated good model fit for 2 (13) = 13.71, p > .05, CMIN/DF = 1.05, CFI, TLI, and IFI p < .01, suggesting that the average Blo ck 1 starting value was significantly greater than zero. The mean estimate of the linear slope was 6.77, p < .05, suggesting significant linear change across blocks in Simple RT performance. The mean estimate of the quadratic slope was 26.43, p < .01, sug gesting significant quadratic change across blocks in Simple RT performance. Specifically, significant growth was observed during the first two blocks of assessment, followed by a leveling off throughout the end of the study period. The intercept related v ariance was 2509.00, p < .01, suggesting significant individual differences in Block 1 Simple RT performance. The linear slope related variance was 85.43, p < .05, suggesting significant individual differences in linear change in Simple RT performance. Las tly, the quadratic slope related variance was 181.52, p > .05, suggesting non significant individual differences in quadratic change in Simple RT performance. Choice RT. Fit statistics indicated good model fit for the Choice RT LGC model. 2 (10) = 5.07, p 0.00. The mean intercept was 357.01, p < .01, suggesting that the average Block 1 starting value was sig nificantly greater than zero. The mean estimate of the linear slope

PAGE 159

159 was 16.52, p < .01, suggesting significant linear change across blocks in Choice RT performance. The mean estimate of the quadratic slope was 52.57, p < .01, suggesting significant quadra tic change across blocks in Choice RT performance. Specifically, significant growth was observed during the first two blocks of assessment, followed by a slight decline throughout the end of the study period. The intercept related variance was 4398.31, p < .01, suggesting significant individual differences in Block 1 Choice RT performance. The linear slope related variance was 343.62, p < .05, suggesting significant individual differences in linear change in Choice RT performance. Lastly, the quadratic slop e related variance was 2545.51, p < .01, suggesting significant individual differences in quadratic change in Choice RT performance. Aim 2: Conditional Growth Curves All conditional latent growth curve analysis results are presented in Tables 3 8 and 3 9. Please refer to these tables for all intercept and slope estimates, correlation coefficients between exogenous predictor variables, regression weights, standardized regression weights, and changes in model fit. Models were all constructed hierarchically, such that initially only covariates (age, gender, education, estimated IQ, depression, and anxiety) were included. Person mean TST and TWT were together subsequently added to the models. Below, final model estimates (i.e., models including sleep variables) are presented for each cognitive DV separately. For no cognitive DV did the model fit change significantly from the model containing covariates only to the model also including sleep variables (see Table 3 9). All associations between sleep and cognition (level and change) are displayed in graphical formats for ease of interpretation (please see Figures 3 4, 3 5, and 3 6).

PAGE 160

160 Number Copy. Fit statistics indicated barely acceptable model fit for the final 2 ( 54 ) = 108.49 p < .0 0 1, CMIN/DF = 2.01 CFI = 0. 84 TLI = 0. 74 and IFI 0. 86 and RMSEA = 0. 11 The mean intercept was 40.56, p < .01, suggesting that the average Block 1 starting value was significantly greater than zero. The mean estimate of the linear slope was 9.32, p < .05, suggesting significant linear change across blocks in Number Copy performance. The mean estimate of the quadratic slope was 18.59, p > .05, suggesting non significant quadratic change across b locks in Number Copy performance. The intercept related variance was 8.09, p < .01, suggesting significant individual differences in Block 1 Number Copy performance. The linear slope related variance was 0.07, p > .05, suggesting non significant individual differences in linear change in Number Copy performance. The quadratic slope related variance was 1.63, p < .05, suggesting individual differences in quadratic change in Number Copy performance. Only age was a significant individual difference predictor o f Number Copy intercept, B = 0.17, p < .01; suggesting older adults had lower Number Copy starting performances than younger adults. There were no significant predictors of change in Number Copy performance across time. Symbol Digit. Fit statistics indic ated barely acceptable model fit for the final 2 ( 54 ) = 115.96 p < .01, CMIN/DF = 2.15 CFI = 0. 86 TLI = 0. 77 and IFI 0. 87 and RMSEA = 0. 12 The mean intercept was 320.03, p < .01, suggesting that the average Block 1 starting value was significantly greater than zero. The mean estimate of the linear slope was 1.22, p > .05, suggesting non significant linear growth across blocks in Symbol Digit performance. Similarly, the mean estimate of the quadratic slope was 2.03 p > .05, suggesting non significant quadratic

PAGE 161

161 change across blocks in Symbol Digit performance. The intercept related variance was 5.503, p < .01, suggesting significant individual differences in Block 1 Symbol Digit performance. The linear slope related variance was 0.20, p > .05, suggesting non significant individual differences in linear change in Symbol Digit performance. The quadratic slope related variance was 0.48, p > .05, suggesting non significant individual differences in quadratic change in Sy mbol Digit performance. Age, B = 0.19, p < .01, education level, B = 0.33, p < .05, and TWT, B = 0.19, p < .05, were significant individual difference predictors of Symbol Digit intercept; suggesting older, less educated adults and those that spend more time awake during the night had lower Symbol Digit starting performances than younger, more highly educated adults and those that spent less time awake during the night. Education level was also a significant predictor of linear change in Symbol Digit per formance, B = 0.17, p < .05; suggesting more highly educated older adults exhibited steeper linear changes in Symbol Digit with repeated practice than did lower educated older adults. Similarly, education level was also a significant predictor of quadratic change in Symbol Digit performance, B = 0.43, p < .05; suggesting more highly educated older adults experienced less pronounced deceleration in Symbol Digit learning than less educated older adults. Refer to Figure 3 4 for a graphical representation of S ymbol Digit level for both a high (i.e., 90 th %ile) and low (i.e., 10 th %ile) TWT subject. Letter Series. Fit statistics indicated barely acceptable model fit for the final Letter 2 ( 54 ) = 120.82 p < .0 0 1, CMIN/DF = 2.24 CFI = 0. 88 TLI = 0. 79 and IFI 0. 89 and RMSEA = 0. 12 The mean intercept was 3.23, p > .05,

PAGE 162

162 suggesting that the average Block 1 starting value was not significantly greater than zero. The mean estimate of the linear slope was 0.60, p > .05, suggesting non significant linear growth across blocks in Letter Series performance. Similarly, the mean estimate of the quadratic slope was 2.04, p > .05, suggesting non significant quadr atic change across blocks in Letter Series performance. The intercept related variance was 6.99, p < .01, suggesting significant individual differences in Block 1 Letter Series performance. The linear slope related variance was 0.32, p > .05, suggesting no n significant individual differences in linear change in Letter Series performance. The quadratic slope related variance was 0.33, p > .05, suggesting non significant individual differences in quadratic change in Letter Series performance. Age, B = 0.16, p < .01, predicted IQ, B = 0.24, p < .05, and TWT, B = 0.24, p < .05, were significant individual difference predictors of Letter Series intercept; suggesting older adults with lower intelligence and those that spent more time awake during the night had l ower Letter Series starting performances than younger adults with higher intelligence and those that spent less time awake during the night. There were no significant predictors of change in Letter Series performance across time. Refer to Figure 3 5 for a graphical representation of Letter Series level for both a high (i.e., 90 th %ile) and low (i.e., 10 th %ile) TWT subject. Simple RT. Fit statistics indicated adequate model fit for the final Simple RT 2 ( 54 ) = 84.01 p < .0 1 CMIN/DF = 1. 56 CFI = 0.9 4 TLI = 0. 8 9, and IFI 0.9 4 and RMSEA = 0.0 8 The mean intercept was 25.55, p < .01, suggesting that the average Block 1 starting value was significantly greater than zer o. The mean estimate of the linear slope was 20.72, p > .05, suggesting non significant

PAGE 163

163 linear growth across blocks in Simple RT performance. Similarly, the mean estimate of the quadratic slope was 10.93, p > .05, suggesting non significant quadratic chan ge across blocks in Simple RT performance. The intercept related variance was 2262.30, p < .01, suggesting significant individual differences in Block 1 Simple RT performance. The linear slope related variance was 82.81, p < .05, suggesting significant ind ividual differences in linear change in Simple RT performance. The quadratic slope related variance was 158.79, p > .05, suggesting non significant individual differences in quadratic change Simple RT performance. There were no significant predictors of le vel in Simple RT performance. TWT, B = 0.54, p < .05, was a significant predictor of quadratic change in Simple RT performance; suggesting that older adults who spent more time awake during the night benefited more from repeated practice than older adults who spent less time awake during the night. Refer to Figure 3 6 for a graphical representation of Simple RT learning for both a high (i.e., 90 th %ile) and low (i.e., 10 th %ile) TWT subject. Choice RT. Fit statistics indicated barely adequate model fit for the final Choice 2 ( 55 ) = 131.00 p < .0 0 1, CMIN/DF = 2. 38 CFI = 0.8 5 TLI = 0. 74 and IFI 0. 86 and RMSEA = 0.1 3 The mean intercept was 42.47, p < .01, suggesting that the average Block 1 starting value was significantly greater than z ero. The mean estimate of the linear slope was 138.31, p < .05, suggesting significant linear growth across blocks in Choice RT performance. Similarly, the mean estimate of the quadratic slope was 333.84, p < .05, suggesting significant quadratic change a cross blocks in Choice RT performance. The intercept related variance was 2786.46, p < .01, suggesting significant individual differences in Block 1 Choice RT performance. The

PAGE 164

164 linear slope related variance was not estimated due to model non convergence. Th e quadratic slope related variance was 302.46, p < .05, suggesting significant individual differences in quadratic change Choice RT performance. Age, B = 1.78, p < .05, and estimated intelligence, B = 2.48, p < .05, were significant individual difference predictors of Choice RT intercept; suggesting older adults with higher estimated intelligence had lower Choice RT starting performances than younger adults with lower estimated intelligence. Estimated intelligence, B = 1.15, p < .05, and anxiety symptom s, B = 0.74, p < .05, were significant individual difference predictors of Choice RT linear slope; suggesting older adults with lower estimated intelligence and lower anxiety symptoms benefited more from repeated practice than older adults with higher est imated intelligence and higher anxiety symptoms. Gender, B = 29.30, p < .05, estimated intelligence, B = 2.43, p < .05, depressive symptoms, B = 2.63, p < .05, and anxiety symptoms, B = 1.47, p < .05, were significant predictors of quadratic change in Cho ice RT performance; suggesting that females, older adults with lower estimated intelligence, higher depressive symptoms, and lower anxiety symptoms benefited more from repeated practice than males, older adults with high estimated intelligence, lower depre ssive symptoms, and higher anxiety symptoms. Discussion The current study investigated improvements in cognitive functioning in older adults that were the result of repeated practice with cognitive stimuli (i.e., practice related learning). Specifically, this study sought to: (1) replicate previous findings (Yang & Krampe, 2009; Yang et al., 2006; Yang et al., 2009) of significant practice related learning in processing speed and reasoning in late life, and (2) determine how an

PAGE 165

165 f time spent a sleep (i.e., TST) and time spent awake (i.e., TWT) relates to their: (a) initial level of cognitive performance, and (b) their ability to benefit from repeated exposure to cognitive stimuli after controlling for salient individual difference variables (i.e., age, education, gender, estimated IQ, and depression and anxiety related symptoms ). A prominent question in the late life cognitive intervention literature revolves around the identification individual differences in learning potential (He rtzog et al., 2008) It was found that both processing speed measures (i.e., Number Copy, Symbol Digit, Simple RT, and Choice RT) and an executive processing measure (i.e., Letter Series) demonstrated significant gains associated with repeated practice. T he various individual difference variables (i.e., age, education, gender, estimated IQ, depressive symptoms, and anxiety symptoms) were found to be differentially related to level and rate of change across the cognitive dependent variables. Similarly, TWT (but not TST) was related to both level and rate of change in cognitive functioning for certain cognitive dependent variables. Practice Related Learning The first overall goal of the current investigation was to replicate previous research demonstrating th e presence of significant practice related learning in late life. Paper 1 of this multipaper dissertation also included this same goal. Further, while the samples utilized in Paper 1 and Paper 2 differ by 3 subjects (due to missing sleep data), the results of corresponding Aim 1s are identical. Readers are referred to the discussion of Paper 1, Aim 1 for a discussion of practice related learning in late life.

PAGE 166

166 Predictors of Starting Level in Cognitive Functioning The second aim of the current paper was to e xamine individual difference predictors of initial level of cognitive functioning in late life. The current investigation found that the common individual difference variables of age, education, and estimated IQ were related to initial level of cognitive f unctioning across the domains of processing speed and reasoning in late life. Such findings are congruent with decades of cross sectional and longitudinal examinations into factors related to late life cognitive functioning [e.g., (Christensen et al., 2004 ; Salthouse, 1991, 2004; Schaie, 1994, 1996)]. However, the current investigation also included examination into sleep variables as potential predictors of initial level of cognitive functioning. Previous research into the relation between self reported s leep and cognitive functioning in late life has generally reported that worse sleep (i.e., indicators of sleep disruption) is associated with poor cognitive functioning (Cricco et al., 2001; Nebes et al., 2009; Tworoger et al., 2006). These associations ha ve been reported for executive functioning measures (i.e., RBANS, TONI, Trails B, and N Back) and composite cognitive functioning indices, but not for perceptual speed tasks (Cricco et al., 2001; Nebes et al., 2009; Tworoger et al., 2006). Similarly, we fo und that older individuals who spent more time awake on average during the night also performed significantly worse on a reasoning and processing speed measure. Interestingly, we did not find any associations between our indicator of sleep achieved (i.e., total sleep time) and cognitive functioning. Previous research has also failed to capture this relationship (Cricco et al., 2001; Nebes et al., 2009; Tworoger et al., 2006). There are several plausible explanations for the observed relationships between TW T and reasoning and processing speed. Disrupted sleep has been hypothesized to

PAGE 167

167 impact cognitive functioning through inhibiting the brain restoration that occurs during healthy sleep (Nebes, Buysse, Halligan, Houck, & Monk, 2009; Tworoger et al., 2006). As such, individuals who chronically spend more time awake during the night may experience the cumulative effects of a lack of nighttime brain restoration. Another potential explanation was suggested by Cricco and colleagues (2001) who hypothesized that sleep difficulty may be related to cognitive functioning through potential reductions in cognitively engaging activity resulting from a lack of restorative sleep (Cricco et al., 2001). Chronic lack of intellectual engagement could result in poor sleeping older adults performing worse than their good sleeping counterparts. The lack of association between TST and initial level of cognitive functioning may indicate that even in the presence of significant nightly awake time older adults are still achieving suffici ent nocturnal sleep. In fact, subjects in the current study achieved nearly 7 hours and 15 minutes of sleep, on average, per night. This level of sleep is well within commonly accepted recommendations for sleep obtainment (Ferrara & De Gennaro, 2001). This level of sleep is suspected to be sufficient to achieve the cognitive benefits of nighttime sleep. As such, it is unlikely that there would be any untoward effects of the average levels of TST reported in the current study on daytime cognitive functioning Interestingly, TWT was related to Letter Series and Symbol Digit level of functioning. These two tasks represent the most complex of the five cognitive tasks examined. These might have been the only tasks sufficiently difficult enough to not allow for s uspected compensatory mechanisms (Drake et al., 2001) to alleviate any performance deficits as a result of sleep loss. Unwanted nocturnal wake time is believed

PAGE 168

168 to be resultant from hyperarousal (Bonnet & Arand, 2010), and this hyperarousal is suspected to not only be present at night, but also be present during daytime hours (Bonnet & Arand, 2010); as such, it may aid in the completion of relatively simple cognitive tasks (Altena, Werf et al., 2008). However, the difficulty level of the Letter Series and Sy mbol Digit tasks may have been too demanding for heightened levels of arousal to adequately compensate. Concretely, elevated arousal levels may impede night time sleep. These same elevated arousal levels may provide compensation for nightly sleep disrupti on on relatively simple daytime tasks. However, with increasing task difficulty it becomes less likely that increased arousal can fully compensate for the lack of potential restoration provided by nocturnal sleep. Future investigations that manipulate task difficulty are necessary to answer such questions. Similarly, compensation for increased wake time during the night may come in the form of added effort during cognitive tasks (Drake et al., 2001). Disturbed night time sleep may be easily compensated for through increased effort on relatively simple cognitive tasks; however, more complex or cognitively demanding tasks may require more cognitive resources than a simple increase in effort can provide. Again, future empirical endeavors that assess effort are required to address this concept. Such a finding of wake time being negatively related to more complex cognitive functioning is consistent with the neuropsychological hypothesis of the effects of sleep loss on cognitive functioning (Harrison & Horne, 2000 ; Harrison et al., 2000). This theory suggests that sleep loss differentially impacts higher order cognitive functioning. Perhaps any detriments in arousal/attention due to increased time spent awake during

PAGE 169

169 the night can be compensated for with additional effort, while the focal impairments in more executive tasks are impervious to such increased effort. Predictors of Practice Learning in Cognitive Functioning The second aim of the current paper also included an examination of individual difference predict ors of domain specific practice related learning in late life. The current investigation found that the common individual difference variables of education, estimated IQ, gender, anxiety symptoms, and depression symptoms were related to practice related le arning in various processing speed measures. The previous research examining predictors of response to practice related learning is limited. Younger age has been reported to be related to practice related learning in processing speed, reasoning, visual att ention, episodic memory, and vocabulary (Salthouse, 2010; Yang et al., 2006). Both state and trait anxiety were previously found to fail to significantly predict the patterns of practice related learning in late life (Hofland et al., 1981; Yang et al., 20 06). Interestingly, we did not find age to be a significant predictor of practice related learning, and we did find that state anxiety was a significant predictor of practice related learning. These results are promising, as it appears that age is related not their ability to benefit from practice. Such a finding suggests that cognitive plasticity remains intact well into later life. Further, late life anxiety appears very modifiable (Barrowclo ugh et al., 2001; Stanley et al., 1996; Wetherell, 1998). Prior to enrollment in practice related cognitive interventions older adults should be screened for anxiety symptoms and, if present, should receive treatments to improve the anxiety prior to engagi ng in cognitive practice.

PAGE 170

170 Individual with higher levels of educational attainment were found to benefit more from practice than those individuals with lower levels of educations attainment. As discussed in Chapter 2 (Paper 1) such a finding may suggest a p ossible role of cognitive/brain reserve (Stern, 2002, 2006) in late life practice related learning. However, why individuals with lower estimated IQs and women would benefit more from cognitive practice is unknown. Future studies would do well to investiga te these individual difference variables further. Such unanticipated findings must be viewed in relation to what is known regarding individual differences in response to cognitive training (which has been more widely investigated than individual difference s in response to cognitive practice). It has been reported that individuals with specific impairments do not respond to cognitive training in the impaired domain (Unverzagt et al., 2007); however, it has also been reported that the largest training related gains can be expected from individuals with the lowest initial ability levels (Jaeggi et al., 2008). As such, much is yet to be known regarding predictors of practice and training related learning in late life. No known study has examined sleep as a pred benefit from cognitive practice. We discovered that individuals who spent more time awake during the night on average (i.e., had higher average TWT) benefited more from repeated practice on the Simple RT task than indivi duals who did not spend as much time awake during the night. While previous investigations have reported that poor sleepers (i.e., those with the most time spent awake during the night) sometimes outperform good sleepers (Altena, Van Der Werf et al., 2008; Altena, Werf et al., 2008), this is the first investigation to extend this finding to learning. Altena and colleagues

PAGE 171

171 (2008) reported that poor sleepers outperformed good sleepers on multiple cognitive tasks (e.g., attention and speed tasks); however, the y noted that the poor sleepers had their largest performance advantage on the simplest tasks. So too, Simple RT was the most simple cognitive task practiced in the current study. Several plausible explanations might account for the positive effect of time spent awake during the night on practice related learning in reaction time. Compensation through effort, altered brain activation, and arousal levels represent the most likely explanatory mechanisms. Neural pruning, hippocampal processing, and neurotransm itter and cytokine functioning are alternative explanatory factors (Krueger, Rector, Churchill, Phelps, & Korneva, 2008; McAfoose & Baune, 2009). However, these mechanisms do not explain why nightly wake time may benefit learning for a simple task and not for complex tasks. As Simple RT is a relatively simple task, individuals who spend more time awake during the night may be able to put forth additional resources or increased effort to compensate for any detrimental effects of potential sleep loss (Drake et al., 2001). However, because it is such a simple task, they may actually outperform and learn more than their better sleeping counterparts. Another potential explanation involves possible altered brain functioning as a response to the added nightly awak e time. Previous reports (Altena, Werf et al., 2008; Drummond et al., 2000) have described altered brain activation associated with disrupted sleep. This is suspected to be a compensatory mechanism resulting from sleep loss. Again, due to the simplistic na ture of the Simple RT task, such additional neural resources may allow for increased learning. Lastly, as hyperarousal appears to be at the core of nocturnal wake time (Bonnet & Arand, 2010)

PAGE 172

172 it may be possible that this heightened level of arousal may bene fit performance on very simple tasks. However, more complex tasks may rely on higher order cognitive complexity/difficulty it is increasingly less likely that arousal will recom pense for the lack of nocturnal restoration due to wake time. Future research is needed before definitive conclusion can be drawn. Limitations There are several limitations to the current study that need to be recognized. Only two cognitive domains were a ssessed repeatedly (i.e., processing speed and executive processing). It would have been advantageous to examine multiple other cognitive domains in the context of practice related learning. So too, the number of specific measures assessing each cognitive domain was extremely unbalanced (i.e., 4 measures of processing speed and 1 measure of executive processing). It would have been beneficial to have multiple executive processing measures practiced so that practice related learning could have been examined at the individual and construct levels. Lastly, assessment of the varying cognitive domains with measures differing in known degree of difficulty would be beneficial. It is unknown how difficult subjects perceived each task. Having this information would h ave allowed for examination into difficulty levels both within and across domains. A limitation of the current investigation was its relatively small sample size for it s SEM analytical approach. This limitation bares direct consequences to some of the reported results and interpreted. The standardized path coefficients in structural equation modeling are analogous to Pearson's r (Tabachnick & Fidell, 2001).

PAGE 173

173 Examination o f the coefficients observed in the analysis of predictors of initial level and change in cognitive functioning reveal that 27 associations between individual difference variables and level and rate of change were of medium effect sizes (i.e., 0.24) or larg er (J. Cohen, 1992). As such, it appears as though the current investigation was hindered by a small sample size which may have affected power and the ability to detect significant results. Another limitation of the current investigation that needs to be acknowledged is the reported sleep. Sleep diaries provide a non invasive measure of sleep that allowed for the repeated assessments employed in the present study. However, as suggested by the reviewed literature, much of wh at is known regarding the sleep cognition relationship has been garnered through investigations which utilize objective measures of nocturnal sleep actigraphy and PSG. As such, how findings and theories based on studies that employ alternative measures of sleep translate to subjective measures of sleep is questionable. Yet, there is evidence to suggest a relatively high concordance between the various measures of sleep (Ancoli Isreal et al., 2003). Correlations between actigraphy measured TST and PSG me asured TST have been estimated as high as 0.97 (Jean Louis et al., 1996). For older adults, correlations between actigraphy and PSG measured TST have been estimated at 0.81 0.91 (Ancoli Israel, Clopton, Klauber, Fell, & Mason, 1997). Actigraphy appears t o be less sensitive to detection of waketime than PSG; however, estimates of the correlation between these measures reaches as high as 0.94 (Cole, Kripke, Gruen, Mullaney, & Gillin, 1992). Recently, Sanchez Ortuno and colleagues (2010) demonstrated that ac tigraphy and PSG measured sleep (SOL, WASO, TST, and TST) displayed highly

PAGE 174

174 significant correlations both between individuals and within individual across time (Sanchez Ortuno, Edinger, Means, & Almirall, 2010). Importantly, these associations were observed for both good and poor sleepers. In a recent study of depressed individuals with insomnia, actigraphy, PSG, and sleep diary were only modestly related (correlations range 0.22 0.59), though they were significantly associated with one another (McCall & M cCall, 2010). Further sleep diary rated sleep estimates have been demonstrated to have high agreement rates with PSG measured sleep in healthy subjects (Rogers, Caruso, & Aldrich, 1993). While it is undeniable that sleep diary, actigraphy, and PSG measure sleep in a non uniform manner, it does appear that there is a strong enough association between the three measures to allow for one to draw inferences across measures. Additionally, knowledge of the relationship between subjectively reported sleep and co gnitive functioning is of high importance due to its ecological validity. Objective measures of sleep are expensive to implement and require a great deal of specialized knowledge and training. However, self reported sleep is the most common way to assess s leep and represents the standard by which common diagnoses, like insomnia, are assessed. As such, self reported sleep is an important topic of study due to its ecological validity and real world applications. A lack of assessment of potential sleep disord ers could be considered a limitation of the current investigation. Simply asking subjects whether or not they had a sleep complaint would have allowed for additional group comparisons that may have led to interesting results. As individuals with sleep comp laints are known to exhibit both nocturnal and daytime hyperarousal (Bonnet & Arand, 2010), such an assessment

PAGE 175

175 would have potentially contributed insight into the mechanisms underlying the wake time cognition associations. Thus, lack of subjective assess ment of presence or absence of a sleep complaint is a limitation. Further, an assessment of potential sleep disorders would have yielding information regarding the probability of sleep apnea. Sleep apnea has known associations with cognitive functioning (B eebe & Gozal, 2002; Findley et al., 1986; Mathieu et al., 2008), which include executive dysfunction and female (82%), and that the prevalence of sleep apnea is much greater in males (Jordan & Doug McEvoy, 2003), the risk of our observed sleep cognition associations being driven by the presence of sleep apnea is reduced. Yet, the prevalence of sleep apnea appears to increase with age (Bixler et al., 2001; Bixler, Vgontzas, Ten Ha ve, Tyson, & Kales, 1998), though there still appears to be a gender difference. Future research is needed that includes additional objective assessment for potential occult sleep disorders. Future Directions iation to late life learning would be well suited to assess for sleep complaint. As stated above, such a complaint may be related to hyperarousal (Bonnet & Arand, 2010), and therefore may yield useful information. Further, objective measurement of sleep, i n conjunction with self report sleep, could yield potentially useful information regarding the mechanisms underlying the sleep cognition association. Previous examinations have found associations between the various sleep stages and cognitive functioning ( Ellenbogen, 2005; Hobson & Pace Schott, 2002; Peigneux, Laureys, Delbeuck, & Maquet, 2001; Tononi & Cirelli, 2006; Wagner et al., 2004; Walker et al., 2002), perhaps similar relationships exist

PAGE 176

176 between sleep stages and practice related learning. Improved s leep assessment may very well allow for a better understanding of sleep cognition relationship in older adults. This question deserves empirical attention. The present study confirmed the presence of practice related learning in late life. Future investig ations may want to follow older adults longitudinally to examine whether transition to mild cognitive impairment or dementia is impacted by practice related learning. Such information would have substantial public health implications. Additionally, investi gation into predictors of such practice related learning would then allow for the optimal selection of individuals to participate in practice interventions with the goal of deterring or preventing late life cognitive decline. On a similar line, future inve stigations would be well suited to assess any real world impact of improved laboratory based cognitive functioning resulting from practice. Sleep and cognitive functioning were repeatedly assessed in the current study. Further, both sleep (Buysse et al., 2010; Dzierzewski et al., 2008) and cognitive functioning (Hultsch et al., 2000; Hultsch et al., 2008) have been show to display high levels of intraindividual fluctuations in late life. As such, future investigation should of sleep and cognitive functioning over time. In fact, some sleep researchers have called for future researchers to in future investigations of the sleep cognition relationship (Bliese et al., 2006). Lastly, future investigation should recruit a large, diverse sample of older adults. Obtainment of a larger sample would allow for robust tests of predictors of gain and allow for the inclusion of additional predictors within a single model. Not only should

PAGE 177

177 futur e researchers collect a larger sample, this sample should also be of increased diversity. The present sample may have limited generalizability due to its over representation with highly educated white females. Examination of practice related learning, and predictors of practice related learning, in less educated, non white males is necessary. Summary The current investigation examined the gains and predictors of cognitive practice in older adults. We confirmed the presence of significant practice related l earning in late life, reported associations between learning and education, estimated IQ, gender, anxiety symptoms, and depression symptoms, and extended this work by examining sleep variables as potential predictors of late life cognition (both initial le vel and learning). Older adults were found to exhibit substantial gains from cognitive practice alone. Amount of time spent awake during the night was found to be related to both initial level of cognitive functioning and practice related learning. Hyperar ousal, which has been associated with time spent awake during the night, may play a central role in Arand, 2010), though future investigations are needed to clarify these ambivalent relationships. Continued investigation into practice impact on practice related learning appear warranted.

PAGE 178

178 Table 3 1. Sleep c ognition r elationship Sleep Aspect Mechanism/Process Cognitive Result Direc tion Reference 1. Poor sleep, sleep complaint, inability to initiate or maintain sleep Physiological hyperarousal (i.e., increased temperature, heart rate, EEG, brain metabolism, etc.) Attention and speed tasks may be better in worse sleepers. Poor slee pers outperform good sleepers, especially on simple tasks. (+) (Altena, Van Der Werf et al., 2008; Altena, Werf et al., 2008; Bonnet & Arand, 2010; Richardson 2007 ) 2 Actigraphy measured SOL, WASO, SE, Napping, and TST NO MECHANISM SUGGESTED* Cognitiv e impairment (i.e., MMSE) and speed (Trails B) lower in worse sleepers. No association with TST. ( ) (Blackwell et al., 2006) 3 Sleep quality (i.e., PSQI), SOL, SE, TST. Sleep duration and sleep difficulty. NO MECHANISM SUGGESTED (attentional resou rce allocation hinted at). Brain region responsible for sleep perception potentially involved in cognitive functioning. Decreased brain restoration during sleep. Decreased cognitive engagement.* Worse executive performance (i.e., RBANS, TONI, Trails B, N Back) for poor sleepers. No difference in processing speed, inhibition, or verbal memory. SOL and SE related to cognition. TST not related to cognition. Short sleep duration and sleep difficulty associated with impaired overall cognitive functioning. Slee p difficulty related to cognitive decline (i.e., SPMSQ) ( ) (Cricco et al., 2001; Nebes et al., 2009a; Tworoger et al., 2006) 4 Total sleep loss and sleep reduction (experimentally manipulated loss of sleep), sleep fragmentation Frontal lobe/prefrontal functioning (i.e., increased or decreased blood flow, activation, deactivation, EEG, etc.)* Reduced executive functioning (verbal fluency, creativity, planning), working memory, attention, processing speed though compensatory mechanisms and stable incr eased performance has been noted. (~) (Altena, Werf et al., 2008; Sean P. A. Drummond et al., 2000; Durmer & Dinges, 2005; Horne, 1988; Jones & Harrison, 2001; June J. Pilcher & Huffcutt, 1996; Van Dongen, 2003) 5 Non rapid eye movement sleep (NR EM): sleep spindles Long term potentiation; hippocampal processing; plasticity Learning and memory formation enhanced ( ) (Hobson & Pace Schott, 2002; Walker et al., 2002)

PAGE 179

179 Table 3 1. Continued. Sleep Aspect Mechanism/Process Cognitive Result Direction Reference 6 Rapid eye movement sleep (REM) Hippocampal processing of memory traces; neuronal replay ; Cortical plasticity Increased learning and insight; procedural memory ( ) (Ellenbogen, 2005; Hobson & Pace Schott, 2002; Peigneux et al. 2001; Stickgold, Hobson, Fosse, & Fosse, 2001; Wagner et al., 2004) 7. Slow wave activity (SWA) Synaptic homeostasis, pruning.* Cognitive functioning and learning is facilitated. ( ) (Tononi & Cirelli, 2006) 8. Wakefulness during sleep Neurotr ansmitter and cytokine regulation. Inability of sleep to restore cognitive resources chemistry and biology not restored to optimum. Decreased synaptic plasticity.* Decreased cognitive performance, processing speed and higher order functions. ( ) (Baune et al., 2008; Haensel et al., 2009; G. P. Krueger, 1989; J. M. Krueger, Obl, Fang, Kubota, & Taishi, 2001; McAfoose & Baune, 2009; Vgontzas et al., 2004; Wilson, Finch, & Cohen, 2002) Notes: Indicates that hyperarousal may also be implicated a s a reasonable mechanism for stated process. (+) Indicates that worse sleep is generally associated with better cognitive functioning. ( ) Indicates that worse sleep is generally associated with worse cognitive functioning. (~) Indicates that worse sleep h as been associated with both better and worse cognitive functioning.

PAGE 180

1 80 Table 3 2 Descriptive/ d emographic s tatistics Mean (SD) Age 63.33 (8.49) Education 16.14 (2.25) Gender 1.82 (0.39) NAART 113.23 (6.49) BDI II 6.41 (5.40) State A nxiety 30.36 (7.99) Trait Anxiety 31.44 (8.76) Notes: Age measured in years since birth; Education measured in years; Gender: 1 = male, 2 = female; NAART = premorbid IQ estimate; BDI II = Beck Depression Inventory, 2 nd Edition.

PAGE 181

181 Table 3 3 Between p erson c orrelations a mong 5 b lock s leep v ariables and d escriptive s leep d ata. SOL WASO TWAK NWAK SQR TIB TST TWT SE Mean (SD) SOL 1.00 18.65 (15. 02 ) WASO 0. 38 ** 1.00 17.80 ( 13.08 ) TWAK 0 .45 ** 0.30 ** 1.00 18.3 4 ( 14.18 ) NWAK 0.10 0.45 ** 0.17 1.00 1.33 (0.79 ) SQR 0.25* 0.39 ** 0.26 0.24 1.00 3.67 (0.5 4 ) TIB 0.39** 0.20 0.24 0.02 0.15 1.00 490.7 5 ( 50.36 ) TST 0.13 0.27 0.26 ** 0.09 0.41 ** 0.79** 1.00 436.02 ( 49.16 ) TWT 0.82 ** 0.70 ** 0.76 ** 0.15 0.39 ** 0. 37** 0.27* 1.00 54.96 ( 32.65 ) SE 0.78 ** 0.72** 0.75 ** 0.19 0.47** 0.20 0.44** 0.98** 1.00 8 9.00 (6. 15 ) Notes: ** p < .01; p < .05; SOL = sleep onset latency; WASO = wake time after sleep onset; TWAK = terminal wakefulness; NWAK = number of awak enings; SQR = sleep quality rating; TIB = time in bed; TST = total sleep time; TWT = total wake time; SE = sleep efficiency. Sleep measured in minutes.

PAGE 182

182 Figure 3 1. Graphical r epresentation of u nivariate u nconditional g rowth c urve.

PAGE 183

183 Figur e 3 2 Graphical r epresentation of f inal u nivariate c onditional g rowth c urve.

PAGE 184

184 Table 3 4 Descriptive and Normality Values for Cognitive Dependent Variables and Sleep Independent Variables Prior to and Post Data Cleaning. Measure Data Cleaned Minimum Ma ximum Mean (SD) Skewness (SE) Kurtosis (SE) Number Copy pre No 1.00 53.00 41.43 (6.72) 3.15 (0.25) 16.08 (0.50) Yes 27.08 53.00 41.86 (4.83) 0.73 (0.25) 0.95 (0.50) Number Copy week 01 No 24.00 88.00 43.91 (8.36) 3.10 (0.27) 16.54 (0.53) Yes 29.30 57.06 43.26 (4.90) 0.30 (0.27) 1.69 (0.53) Number Copy week 02 No 26.00 52.00 43.61 (4.57) 0.94 (0.28) 2.15 (0.55) Yes 29.90 52.00 43.66 (4.38) 0.62 (0.28) 0.68 (0.55) Number Copy week 03 No 29.00 89.00 45.00 (7.14) 3.74 (0.31) 24.26 (0.60) Yes 31 .55 57.61 44.53 (4.45) 0.20 (0.31) 1.65 (0.60) Number Copy week 04 No 30.00 53.00 44.66 (3.88) 1.14 (0.30) 2.84 (0.60) Yes 34.68 53.00 44.74 (3.63) 0.69 (0.30) 0.95 (0.60) Number Copy week 05 No 27.00 52.00 43.90 (4.40) 1.03 (0.29) 2.37 (0.57) Ye s 30.70 52.00 43.95 (4.20) 0.69 (0.29) 0.82 (0.57) Number Copy week 06 No 30.00 93.00 45.20 (7.13) 4.62 (0.30) 32.07 (0.59) Yes 33.15 55.22 44.67 (3.88) 0.32 (0.30) 0.81 (0.59) Number Copy week 07 No 33.00 54.00 45.23 (4.14) 0.64 (0.30) 1.36 (0.60) Yes 33.00 54.00 45.23 (4.14) 0.64 (0.30) 1.36 (0.60) Number Copy week 08 No 38.00 53.00 45.54 (3.69) 0.15 (0.31) 0.26 (0.61) Yes 38.00 53.00 45.54 (3.69) 0.15 (0.31) 0.26 (0.61) Number Copy week 09 No 36.00 56.00 45.16 (4.11) 0.10 (0.32) 0.23 (0.62) Yes 36.00 56.00 45.16 (4.11) 0.10 (0.32) 0.23 (0.62) Number Copy week 10 No 37.00 56.00 45.58 (3.84) 0.16 (0.31) 0.33 (0.61) Yes 37.00 56.00 45.58 (3.84) 0.16 (0.31) 0.33 (0.61) Number Copy week 11 No 3.00 57.00 45.09 (6.91) 4.34 (0.32) 2 6.27 (0.64) Yes 34.01 57.00 45.67 (4.04) 0.42 (0.32) 1.52 (0.64) Number Copy week 12 No 3.00 54.00 44.82 (6.50) 4.63 (0.31) 29.39 (0.61) Yes 34.50 54.00 45.34 (3.75) 0.57 (0.31) 0.92 (0.61) Number Copy week 13 No 37.00 52.00 45.14 (3.31) 0.39 (0. 34) 0.17 (0.66)

PAGE 185

185 Table 3 4 Continued. Measure Data Cleaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Yes 37.00 52.00 45.14 (3.31) 0.39 (0.34) 0.17 (0.66) Number Copy week 14 No 33.00 54.00 45.82 (3.56) 0.62 (0.33) 2.59 (0.66) Yes 36.72 54.00 45.90 (3.32) 0.06 (0.33) 0.70 (0.66) Number Copy week 15 No 38.00 56.00 46.87 (3.66) 0.07 (0.35) 0.43 (0.69) Yes 38.00 56.00 46.87 (3.66) 0.07 (0.35) 0.43 (0.69) Number Copy week 16 No 36.00 55.00 45.78 (4.26) 0.43 (0. 34) 0.02 (0.67) Yes 36.00 55.00 45.78 (4.26) 0.43 (0.34) 0.02 (0.67) Number Copy post No 34.00 56.00 45.85 (4.29) 0.50 (0.29) 0.33 (0.57) Yes 34.00 56.00 45.85 (4.29) 0.50 (0.29) 0.33 (0.57) Symbol Digit pre No 0.00 32.00 21.99 (6.30) 1.66 (0.25) 3.33 (0.50) Yes 6.38 32.00 22.22 (5.59) 1.14 (0.25) 1.14 (0.50) Symbol Digit week 01 No 14.00 48.00 24.49 (5.12) 2.07 (0.27) 9.15 (0.53) Yes 14.00 35.49 24.19 (3.95) 0.11 (0.27) 1.12 (0.53) Symbol Digit week 02 No 16.00 33.00 25.37 (3.55) 0.53 (0. 28) 0.30 (0.56) Yes 16.00 33.00 25.37 (3.55) 0.53 (0.28) 0.30 (0.56) Symbol Digit week 03 No 9.00 45.00 26.26 (4.72) 0.04 (0.30) 5.75 (0.60) Yes 14.85 37.61 26.23 (3.99) 0.31 (0.30) 1.57 (0.60) Symbol Digit week 04 No 15.00 32.00 26.19 (3.23) 0.82 (0.30) 1.28 (0.60) Yes 16.51 32.00 26.22 (3.15) 0.64 (0.30) 0.50 (0.60) Symbol Digit week 05 No 18.00 32.00 25.96 (3.37) 0.38 (0.29) 0.36 (0.57) Yes 18.00 32.00 25.96 (3.37) 0.38 (0.29) 0.36 (0.57) Symbol Digit week 06 No 17.00 55.00 26.80 (4.9 5) 2.94 (0.30) 15.75 (0.59) Yes 17.00 36.83 26.52 (3.69) 0.48 (0.30) 0.76 (0.59) Symbol Digit week 07 No 22.00 34.00 26.52 (3.20) 0.36 (0.30) 0.93 (0.60) Yes 22.00 34.00 26.52 (3.20) 0.36 (0.30) 0.93 (0.60) Symbol Digit week 08 No 19.00 34.00 26.64 (3.44) 0.05 (0.31) 0.37 (0.61) Yes 19.00 34.00 26.64 (3.44) 0.05 (0.31) 0.37 (0.61)

PAGE 186

186 Table 3 4 Continued. Measure Data Cleaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Symbol Digit week 09 No 10.00 33.00 26.61 (4.01) 1.27 (0.32) 3.86 (0.62) Yes 14.58 33.00 26.69 (3.71) 0.65 (0.32) 0.72 (0.62) Symbol Digit week 10 No 16.00 34.00 26.54 (3.54) 0.59 (0.31) 0.73 (0.61) Yes 16.00 34.00 26.54 (3.54) 0.59 (0.31) 0.73 (0.61) Symbol Digit week 11 No 0.00 35.00 26.63 (4.94) 2.80 (0.32) 15.26 (0.64) Yes 15.17 35.00 26.91 (3.67) 0.20 (0.32) 0.92 (0.64) Symbol Digit week 12 No 16.00 35.00 27.08 (3.40) 0.45 (0.30) 1.06 (0.60) Yes 16.87 35.00 27.09 (3.36) 0.34 (0.30) 0.70 (0.60) Symbol Digit week 13 No 21.00 34.00 27.04 ( 3.15) 0.06 (0.34) 0.46 (0.66) Yes 21.00 34.00 27.04 (3.15) 0.06 (0.34) 0.46 (0.66) Symbol Digit week 14 No 19.00 35.00 27.12 (3.24) 0.13 (0.33) 0.54 (0.66) Yes 19.00 35.00 27.12 (3.24) 0.13 (0.33) 0.54 (0.66) Symbol Digit week 15 No 21.00 37.00 28. 37 (3.59) 0.08 (0.35) 0.38 (0.69) Yes 21.00 37.00 28.37 (3.59) 0.08 (0.35) 0.38 (0.69) Symbol Digit week 16 No 14.00 36.00 27.02 (4.07) 0.87 (0.34) 1.85 (0.67) Yes 15.03 36.00 27.04 (4.01) 0.76 (0.34) 1.47 (0.67) Symbol Digit post No 16.00 36.00 28.64 (3.93) 0.82 (0.29) 0.85 (0.57) Yes 16.84 36.00 28.65 (3.89) 0.76 (0.29) 0.59 (0.57) Letter Series pre No 0.00 13.00 6.02 (3.09) 0.09 (0.26) 0.60 (0.51) Yes 0.00 13.00 6.02 (3.09) 0.09 (0.26) 0.60 (0.51) Letter Series week 01 No 1.00 19.00 8 .85 (3.62) 0.15 (0.27) 0.15 (0.53) Yes 1.00 19.00 8.85 (3.62) 0.15 (0.27) 0.15 (0.53) Letter Series week 02 No 0.00 22.00 9.85 (4.64) 0.02 (0.28) 0.34 (0.55) Yes 0.00 22.00 9.85 (4.64) 0.02 (0.28) 0.34 (0.55) Letter Series week 03 No 2.00 22.00 10.89 (4.98) 0.10 (0.30) 0.54 (0.60) Yes 2.00 22.00 10.89 (4.98) 0.10 (0.30) 0.54 (0.60) Letter Series week 04 No 0.00 23.00 10.37 (5.02) 0.47 (0.30) 0.09 (0.60) Yes 0.00 23.00 10.37 (5.02) 0.47 (0.30) 0.09 (0.60)

PAGE 187

187 Table 3 4 Continued. Measure Data Cleaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Letter Series week 05 No 0.00 21.00 8.84 (5.10) 0.17 (0.29) 0.34 (0.57) Yes 0.00 21.00 8.84 (5.10) 0.17 (0.29) 0.34 (0.57) Letter Series week 06 No 2.00 19.00 9.89 (4.00) 0.2 8 (0.30) 0.91 (0.59) Yes 2.00 19.00 9.89 (4.00) 0.28 (0.30) 0.91 (0.59) Letter Series week 07 No 1.00 21.00 11.42 (3.90) 0.45 (0.30) 0.12 (0.60) Yes 1.00 21.00 11.42 (3.90) 0.45 (0.30) 0.12 (0.60) Letter Series week 08 No 1.00 21.00 11.54 (4.48) 0.24 (0.31) 0.55 (0.61) Yes 1.00 21.00 11.54 (4.48) 0.24 (0.31) 0.55 (0.61) Letter Series week 09 No 1.00 24.00 12.34 (4.36) 0.05 (0.31) 0.52 (0.62) Yes 1.00 24.00 12.34 (4.36) 0.05 (0.31) 0.52 (0.62) Letter Series week 10 No 1.00 8.00 4.05 (2 .05) 0.07 (0.31) 0.91 (0.61) Yes 1.00 8.00 4.05 (2.05) 0.07 (0.31) 0.91 (0.61) Letter Series week 11 No 3.00 23.00 12.36 (5.38) 0.02 (0.32) 1.05 (0.63) Yes 3.00 23.00 12.36 (5.38) 0.02 (0.32) 1.05 (0.63) Letter Series week 12 No 3.00 22.00 12.5 5 (4.77) 0.11 (0.31) 0.55 (0.61) Yes 3.00 22.00 12.55 (4.77) 0.11 (0.31) 0.55 (0.61) Letter Series week 13 No 1.00 19.00 12.10 (4.24) 0.69 (0.34) 0.21 (0.66) Yes 1.00 19.00 12.10 (4.24) 0.69 (0.34) 0.21 (0.66) Letter Series week 14 No 4.00 23.0 0 13.31 (5.20) 0.02 (0.33) 1.07 (0.66) Yes 4.00 23.00 13.31 (5.20) 0.02 (0.33) 1.07 (0.66) Letter Series week 15 No 5.00 23.00 13.91 (4.86) 0.01 (0.35) 0.82 (0.69) Yes 5.00 23.00 13.91 (4.86) 0.01 (0.35) 0.82 (0.69) Letter Series week 16 No 2.00 21.00 14.08 (5.30) 0.46 (0.34) 0.89 (0.67) Yes 2.00 21.00 14.08 (5.30) 0.46 (0.34) 0.89 (0.67) Letter Series post No 1.00 25.00 10.87 (5.51) 0.50 (0.29) 0.26 (0.57) Yes 1.00 25.00 10.87 (5.51) 0.50 (0.29) 0.26 (0.57) Simple RT pre No 220.04 518 .12 322.97 (66.77) 0.96 (0.25) 0.51 (0.50) Yes 220.04 518.12 322.97 (66.77) 0.96 (0.25) 0.51 (0.50)

PAGE 188

188 Table 3 4 Continued. Measure Data Cleaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Choice RT pre No 341.97 921.78 473.24 (106.8 9) 2.20 (0.25) 6.53 (0.50) Yes 341.97 705.62 466.85 (84.45) 1.03 (0.25) 0.78 (0.50) Simple RT week 01 No 228.70 596.82 308.09 (58.76) 1.96 (0.27) 7.00 (0.53) Yes 228.70 464.32 306.11 (50.90) 0.91 (0.27) 0.95 (0.53) Choice RT week 01 No 321.30 869.43 442.67 (95.05) 2.63 (0.27) 8.93 (0.53) Yes 321.30 637.41 435.42 (68.49) 1.09 (0.27) 1.48 (0.53) Simple RT week 02 No 212.42 697.93 304.05 (70.47) 2.73 (0.28) 13.02 (0.56) Yes 212.42 469.78 300.73 (55.70) 0.88 (0.28) 0.97 (0.56) Choice RT week 02 No 3 05.62 774.88 424.00 (77.00) 2.40 (0.28) 8.80 (0.56) Yes 305.62 594.10 419.39 (59.44) 0.86 (0.28) 1.04 (0.56) Simple RT week 03 No 186.55 754.91 313.75 (106.53) 3.05 (0.30) 10.47 (0.60) Yes 186.55 477.95 301.37 (61.66) 1.04 (0.30) 1.89 (0.60) Choice R T week 03 No 304.57 546.49 412.20 (53.39) 0.57 (0.30) 0.06 (0.60) Yes 304.57 546.49 412.20 (53.39) 0.57 (0.30) 0.06 (0.60) Simple RT week 04 No 191.87 448.58 289.78 (56.58) 0.68 (0.30) 0.42 (0.60) Yes 191.87 448.58 289.78 (56.58) 0.68 (0.30) 0.42 (0 .60) Choice RT week 04 No 307.41 506.80 399.39 (48.78) 0.33 (0.30) 0.65 (0.60) Yes 307.41 506.80 399.39 (48.78) 0.33 (0.30) 0.65 (0.60) Simple RT week 05 No 183.61 635.68 294.17 (70.67) 1.94 (0.29) 7.05 (0.57) Yes 183.61 480.13 291.88 (61.36) 0.85 (0.29) 0.64 (0.57) Choice RT week 05 No 309.17 943.55 407.31 (80.98) 4.53 (0.29) 28.67 (0.57) Yes 309.17 545.78 401.46 (50.17) 0.95 (0.29) 1.02 (0.57) Simple RT week 06 No 179.62 816.25 302.11 (92.33) 3.07 (0.30) 14.84 (0.59) Yes 179.62 502.18 296.96 (69.52) 1.09 (0.30) 1.68 (0.59) Choice RT week 06 No 311.31 627.01 403.67 (55.68) 1.15 (0.30) 2.95 (0.59) Yes 311.31 570.69 402.79 (52.44) 0.70 (0.30) 0.70 (0.59) Simple RT week 07 No 147.12 428.40 279.67 (63.16) 0.29 (0.30) 0.06 (0.60) Yes 147.12 428.40 279.67 (63.16) 0.29 (0.30) 0.06 (0.60)

PAGE 189

189 Table 3 4 Continued. Measure Data Cleaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Choice RT week 07 No 313.59 738.21 401.47 (68.63) 2.40 (0.30) 9.45 (0.60) Yes 313.59 546.59 397.3 6 (53.04) 0.78 (0.30) 0.74 (0.60) Simple RT week 08 No 127.03 512.68 279.58 (70.63) 0.59 (0.31) 0.97 (0.61) Yes 127.03 491.47 279.22 (69.47) 0.47 (0.31) 0.50 (0.61) Choice RT week 08 No 321.07 778.58 395.57 (67.06) 3.32 (0.31) 17.68 (0.61) Yes 321.07 547.09 391.65 (48.45) 0.79 (0.31) 0.68 (0.61) Simple RT week 09 No 146.48 844.86 286.44 (100.46) 3.09 (0.32) 16.39 (0.62) Yes 146.48 516.73 280.69 (73.74) 0.59 (0.32) 0.90 (0.62) Choice RT week 09 No 312.99 552.93 398.92 (52.23) 0.68 (0.32) 0.27 (0.62 ) Yes 312.99 552.93 398.92 (52.23) 0.68 (0.32) 0.27 (0.62) Simple RT week 10 No 142.13 586.61 282.55 (76.02) 1.16 (0.31) 3.41 (0.62) Yes 142.13 510.62 281.24 (71.19) 0.65 (0.31) 1.00 (0.62) Choice RT week 10 No 315.03 709.75 405.43 (67.60) 1.97 (0.31 ) 6.29 (0.62) Yes 315.03 570.95 403.03 (58.48) 1.02 (0.31) 0.84 (0.62) Simple RT week 11 No 111.36 458.75 285.48 (71.00) 0.08 (0.32) 0.26 (0.64) Yes 111.36 458.75 285.48 (71.00) 0.08 (0.32) 0.26 (0.64) Choice RT week 11 No 321.68 566.76 406.70 (57. 74) 0.75 (0.32) 0.08 (0.64) Yes 321.68 566.76 406.70 (57.74) 0.75 (0.32) 0.08 (0.64) Simple RT week 12 No 124.19 527.99 288.07 (72.64) 0.67 (0.31) 1.07 (0.61) Yes 124.19 506.00 287.70 (71.46) 0.55 (0.31) 0.60 (0.61) Choice RT week 12 No 305.72 656. 34 399.03 (60.26) 1.45 (0.31) 4.32 (0.61) Yes 305.72 579.79 397.75 (55.32) 0.80 (0.31) 0.80 (0.61) Simple RT week 13 No 167.63 489.66 292.26 (73.60) 0.47 (0.34) 0.00 (0.66) Yes 167.63 489.66 292.26 (73.60) 0.47 (0.34) 0.00 (0.66) Choice RT week 13 No 296.42 549.24 404.22 (53.54) 0.60 (0.34) 0.38 (0.66) Yes 296.42 549.24 404.22 (53.54) 0.60 (0.34) 0.38 (0.66) Simple RT week 14 No 127.43 731.01 283.15 (95.02) 2.03 (0.33) 8.73 (0.66) Yes 127.43 524.29 279.10 (78.51) 0.50 (0.33) 0.60 (0.66)

PAGE 190

190 Table 3 4 Continued. Measure Data Cleaned Minimum Maximum Mean (SD) Skewness (SE) Kurtosis (SE) Choice RT week 14 No 266.00 556.15 394.88 (56.08) 0.55 (0.33) 0.65 (0.66) Yes 266.00 556.15 394.88 (56.08) 0.55 (0.33) 0.65 (0.66) Simple RT week 15 No 105.77 1147.66 298.90 (158.96) 3.93 (0.35) 19.08 (0.69) Yes 105.77 529.11 280.87 (83.37) 0.96 (0.35) 2.16 (0.69) Choice RT week 15 No 318.84 498.83 394.48 (46.33) 0.62 (0.35) 0.39 (0.69) Yes 318.84 498.83 394.48 (46.33) 0.62 (0.35) 0.39 (0.69) Simp le RT week 16 No 118.86 645.68 282.06 (87.54) 1.31 (0.34) 5.02 (0.67) Yes 118.86 518.88 279.48 (77.91) 0.36 (0.34) 0.75 (0.67) Choice RT week 16 No 297.04 690.02 400.33 (68.22) 2.04 (0.34) 6.16 (0.67) Yes 297.04 574.05 397.96 (59.41) 1.24 (0.34) 1.66 (0.67) Simple RT week 17 No 129.90 594.69 293.66 (85.13) 0.89 (0.29) 1.94 (0.57) Yes 129.90 539.62 292.74 (82.09) 0.64 (0.29) 0.96 (0.57) Choice RT week 17 No 303.34 1133.68 419.14 (111.11) 4.43 (0.29) 26.09 (0.57) Yes 303.34 595.09 409.07 (63.21) 0. 98 (0.29) 0.79 (0.57) Total Sleep Time Mean No 292.14 533.85 435.59 (49.41) 0.21 (0.26) 0.07 (0.51) Yes 297.93 533.85 43 6.02 (49.1 6 ) 0.20 (0.26) 0.15 (0.51) Total Wake Time Mean No 5.99 206.81 55.43 (33.40) 1.77 (0.26) 4.96 (0.51) Yes 5.99 201.54 54.96 (32.65 ) 1.69 (0.26) 4.49 (0.51) Note: All data reported in raw metric (i.e., RT data not recoded to positive = better format). Data presented prior to and po st cleaning procedures to allow interested readers to inspect changes in descriptive, norm ality, and raw data values due to cleaning procedures.

PAGE 191

191 Table 3 5 Comparison of m eans and s tandard d eviations of b aseline s tudy v ariables among c ompleted and a ttrited s ubjects. Total Sample Completed Sample Attrited Sample Df t 2 p value Ch aracteristic n = 87 n = 68 n = 19 Demographic/Descriptive Age (SD), years 63.33 ( 8.49 ) 63.88 ( 8.75 ) 61.37 ( 7.33 ) 85 1.14 0.26 Education (SD), years 16.14 ( 2.25 ) 16.22 ( 2.22 ) 15.84 ( 2.39 ) 85 0.65 0.52 Gender, n (% female) 71 (82) 57 ( 84 ) 1 4 ( 7 4) 1 1.00 0.32 Estimated IQ (SD) 113.23 ( 6.49 ) 113.10 ( 6.50 ) 113.68 ( 6.60 ) 85 0.34 0.73 BDI II (SD) 6.41 ( 5.40 ) 5.76 ( 5.17 ) 8.74 ( 5.70 ) 85 2.17 0.03 State Anxiety (SD) 30.36 ( 7.99 ) 29.85 ( 7.47 ) 32.16 ( 9.62 ) 85 1.11 0.2 7 Trait Anxiety (SD) 31.44 ( 8.76 ) 30.66 ( 8.08 ) 34.21 ( 10.63 ) 85 1.57 0.12 Block 1 Cognition Number Copy (SD) 42.78 ( 4.05 ) 42.77 ( 3.88 ) 42.84 ( 4.71 ) 85 0.07 0.94 Symbol Digit (SD) 23.89 ( 3.74 ) 23.72 ( 3.71 ) 24.49 ( 3.91 ) 85 0.79 0.43 Letter Series (SD) 8.10 ( 3.35 ) 7.82 ( 3.23 ) 9.08 ( 3.68 ) 85 1.45 0.15 Simple RT (SD) 487.33 ( 54.04 ) 492.28 ( 52.71 ) 470.06 ( 54.06 ) 85 1.71 0.09 Choice RT (SD) 356.96 ( 68.66 ) 360.22 ( 71.33 ) 333.76 ( 62.16 ) 85 1.56 0.12 Sleep Variables TST 436.02 ( 49.1 6) 433.49 ( 49.06 ) 442.59 ( 50.03 ) 85 0.71 0.48 TWT 5 4.96 ( 32. 65) 55.14 ( 35.01 ) 55.74 ( 24.53 ) 85 0.07 0.94 Notes : Mean (Standard Deviation); BDI II = Beck Depression Inventory, 2 nd Edition; LN = Letter Number; RT = Reaction Time; TST = Total Sleep Time; TWT = Total Wake Time.

PAGE 192

192 Table 3 6 Unconditional m ultilevel g rowth m odels e xamining c hange in s leep o ver t ime. Parameter Symbol Estimate SE Estimate SE Total Sleep Time Total Wake Time Fixed Effects I ntercept 00 433. 04 3.34 *** 61.44 2.73*** Linear Time 10 1.85 1.72 2.24 1.21 Quadratic Time 20 0. 3 1 1.54 0. 75 1.04 Random Effects Intercept r oi 4490.65 72.23 *** 1 578.01 25.45 *** Residual e it 2261.00 364.61 ** 1337.46 212.93 *** Linear Time r 1i 0.04 0.02 ** 0.04 0.01 *** Quadratic Time r 2i 0.0001 0.00002 *** 0.00001 0.00001 ** Note: p < .05, ** p < .01 *** p < .0 0 1

PAGE 193

193 Table 3 7 Parameter e stimates for u nconditional l atent g rowth c urve m odels. Estimate SE z Value Number Copy Mean Intercept 42. 78 0.4 4 98.34 Linear Slope 0.1 0 0.21 0. 49 Quadratic Slope 1. 7 5 0.4 6 3.77 Variance Intercept 13.99 2. 50 5.60 Linear Slope 0.5 4 0.65 0.8 3 Quadratic Slop e 5.66 3. 08 1. 84 Correlations Intercept Linear Slope 0.3 0 -0.92 Intercept Quadratic Slope 0. 56 -2.37 Linear Slope Quadratic Slope 0. 70 -0 92 Symbol Digit Mean Intercept 23. 92 0.4 0 59.59 Linear Slope 0.0 1 0.21 0.02 Quadratic Slope 2. 00 0.4 2 4.77 Variance Intercept 12.62 2.12 5.96 Linear Slope 1. 68 0.5 6 2.99 Quadratic Slope 7.82 2. 38 3.29 Correlations Intercept Linear Slope 0. 32 -1.86 Intercept Quadratic Slope 0. 51 -2 .92 Linear Slope Quadratic Slope 0.8 7 -2.85 Letter Series Mean Intercept 8.03 0.3 8 20.94 Linear Slope 0.1 8 0.22 0. 83 Quadratic Slope 2.1 0 0.4 0 5.29 Variance Intercept 10.2 0 1.8 0 5.66 Linear Slope 0.47 0.12 3.80 Quadratic Slope 0 .00 --Correlations Intercept Linear Slope 0.64 -3.94 Intercept Quadratic Slope ---Linear Slope Quadratic Slope ---Simple RT

PAGE 194

194 Table 3 7 Continued Estimate SE z Value Mean Intercept 486. 84 5. 86 83.10 Linear Slope 6.7 6 2.9 5 2.29 Quadratic Slope 26. 43 5.5 6 4.76 Variance Intercept 2 509 00 4 24.25 5.91 Linear Slope 85. 43 40.7 5 2.10 Quadratic Slope 18 0 52 146.7 7 1.2 3 Correlations Intercept Linear Slo pe ---Intercept Quadratic Slope ---Linear Slope Quadratic Slope ---Choice RT Mean Intercept 357.01 7.36 48.52 Linear Slope 1 6. 5 2 2.92 5.65 Quadratic Slope 5 2 .5 7 6.8 6 7.67 Variance Intercept 4398 31 7 10 83 6.19 Linear Slope 34 4 .6 2 115. 16 2.98 Quadratic Slope 2 5 4 5 51 6 29 67 4.04 Correlations Intercept Linear Slope 0.6 4 -3.54 Intercept Quadratic Slope 0.7 4 -4.42 Linear Slope Quadratic Slope 0.9 2 -3.33 Note: z s cores greater than 1.96 are statistically significant ( p < .05).

PAGE 195

195 Figure 3 3. Model i mplied c hange in c ognitive f unctioning a cross s tudy p eriod. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1 2 3 4 5 Z score Blocks (trials in parentheses) Eighteen week Cognitive Change (Standardized Metric) (0 2) (3 6) (7 10) (11 14) (15 17) Letter Series Symbol Digit Number Copy Choice RT Simple RT

PAGE 196

196 Table 3 8. Parameter e stimates for f inal c onditional l atent g rowth c urve m odels. Parameter Standar d Estimate Estimate SE z Value Number Copy Mean Intercept -40.56 7.88 5.15 Linear Slope -9.32 4.67 2.00 Quadratic Slope -18.59 9.62 1.93 Variance Intercept -8.09 1.48 5.46 Linear Slope -0.07 0.18 0.37 Quadratic Slope -1.63 0.74 2.19 Regression Weights Age ->Intercept 0.42 0.17 0.04 4.05 Age ->Linear Slope 0.14 0.01 0.03 0.45 Age ->Quadratic Slope 0.07 0.01 0.05 0.28 Gender ->Intercept 0.01 0.06 0.91 0.07 Gender ->Li near Slope 0.32 0.55 0.54 1.02 Gender ->Quadratic Slope 0.13 0.59 1.11 0.53 Education ->Intercept 0.13 0.20 0.17 1.17 Education ->Linear Slope 0.27 0.08 0.10 0.80 Education ->Quadratic Slope 0.29 0.23 0.21 1.13 NAART ->Intercept 0 .17 0.09 0.06 1.56 NAART ->Linear Slope 0.59 0.06 0.04 1.74 NAART ->Quadratic Slope 0.35 0.10 0.07 1.36 Depression ->Intercept 0.08 0.05 0.07 0.73 Depression ->Linear Slope 0.03 0.004 0.04 0.09 Depression ->Quadratic Slope 0.06 0 .02 0.09 0.22 Anxiety ->Intercept 0.18 0.08 0.05 1.59 Anxiety ->Linear Slope 0.26 0.02 0.03 0.77

PAGE 197

197 Table 3 8. Continued Parameter Standard Estimate Estimate SE z Value Anxiety ->Quadratic Slope 0.27 0.06 0.06 1.03 TST ->Intercep t 0.06 0.004 0.01 0.58 TST ->Linear Slope 0.18 0.002 0.004 0.54 TST ->Quadratic Slope 0.17 0.01 0.01 0.70 TWT ->Intercept 0.01 0.001 0.01 0.06 TWT ->Linear Slope 0.37 0.01 0.01 1.15 TWT ->Quadratic Slope 0.34 0.02 0.01 1.38 Co rrelations Predicted IQ Education -0.37 -3.20 State Anxiety Depression -0.40 -3.44 TST TWT -0.27 -2.43 Symbol Digit Mean Intercept -32.03 6.47 4.95 Linear Slope -1.22 3.91 0.31 Quadratic Slo pe -2.03 7.54 0.27 Variance Intercept -5.50 0.99 5.57 Linear Slope -0.20 0.12 1.58 Quadratic Slope -0.48 0.46 1.05 Regression Weights Age ->Intercept 0.50 0.19 0.03 5.44 Age ->Linear Slope 0.07 0.01 0.02 0.27 Age ->Quadratic Slope 0.12 0.02 0.04 0.53 Gender ->Intercept 0.12 1.00 0.75 1.34 Gender ->Linear Slope 0.19 0.35 0.45 0.78 Gender ->Quadratic Slope 0.34 1.25 0.87 1.43 Education ->Intercept 0.23 0.33 0.14 2.35

PAGE 198

198 Table 3 8. Continued Parameter Standard Estimate Estimate SE z Value Education ->Linear Slope 0.52 0.17 0.08 1.97 Education ->Quadratic Slope 0.67 0.43 0.16 2.67 NAART ->Intercept 0.07 0.03 0.05 0.69 NAART ->Linear Slope 0.06 0.01 0.03 0.24 NAART ->Quadratic Slope 0.10 0.02 0.06 0.39 Depression ->Intercept 0.18 0.11 0.06 1.80 Depression ->Linear Slope 0.15 0.02 0.04 0.56 Depression ->Quadratic Slope 0.22 0.06 0.07 0.85 Anxiety ->Intercept 0.16 0.07 0.04 1.65 Anx iety ->Linear Slope 0.18 0.02 0.02 0.68 Anxiety ->Quadratic Slope 0.32 0.06 0.05 1.24 TST ->Intercept 0.004 0.000 0.01 0.04 TST ->Linear Slope 0.44 0.01 0.004 1.75 TST ->Quadratic Slope 0.20 0.01 0.01 0.84 TWT ->Intercept 0.19 0.02 0.01 1.98 TWT ->Linear Slope 0.39 0.01 0.01 1.55 TWT ->Quadratic Slope 0.36 0.02 0.01 1.49 Correlations Predicted IQ Education -0.37 -3.20 State Anxiety Depression -0.40 -3.44 TST TWT -0.27 -2.43 L etter Series Mean Intercept -3.23 7.36 0.44 Linear Slope -0.60 4.62 0.13 Quadratic Slope -2.04 8.72 0.23 Variance Intercept -6.99 1.26 5.55

PAGE 199

199 Table 3 8. Continued Parameter Standard Estimate Estimate SE z Value Linear Slope -0.32 0.17 1.90 Quadratic Slope -0.33 0.59 0.56 Regression Weights Age ->Intercept 0.40 0.16 0.04 4.02 Age ->Linear Slope 0.12 0.01 0.02 0.44 Age ->Quadratic Slope 0.28 0.04 0.05 0.76 Gender -> Intercept 0.03 0.26 0.85 0.31 Gender ->Linear Slope 0.12 0.24 0.53 0.45 Gender ->Quadratic Slope 0.02 0.06 1.01 0.06 Education ->Intercept 0.19 0.28 0.16 1.75 Education ->Linear Slope 0.54 0.18 0.10 1.82 Education ->Quadratic Slop e 0.50 0.24 0.19 1.26 NAART ->Intercept 0.24 0.12 0.06 2.25 NAART ->Linear Slope 0.08 0.01 0.03 0.26 NAART ->Quadratic Slope 0.34 0.06 0.07 0.85 Depression ->Intercept 0.02 0.01 0.07 0.18 Depression ->Linear Slope 0.14 0.02 0 .04 0.46 Depression ->Quadratic Slope 0.52 0.10 0.08 1.29 Anxiety ->Intercept 0.07 0.03 0.05 0.61 Anxiety ->Linear Slope 0.40 0.04 0.03 1.35 Anxiety ->Quadratic Slope 0.54 0.07 0.05 1.35 TST ->Intercept 0.03 0.002 0.01 0.31 TST ->Linear Slope 0.02 0.000 0.004 0.06 TST ->Quadratic Slope 0.20 0.004 0.01 0.51 TWT ->Intercept 0.24 0.03 0.01 2.34 TWT ->Linear Slope 0.10 0.002 0.01 0.36 TWT ->Quadratic Slope 0.07 0.002 0.01 0.19 Correlations

PAGE 200

200 Table 3 8. Continued Parameter Standard Estimate Estimate SE z Value Predicted IQ Education -0.37 -3.20 State Anxiety Depression -0.40 -3.44 TST TWT -0.27 -2.43 Simple RT Mean Intercept -625.55 125.16 5.00 Linear Slope -20.72 64.88 0.32 Quadratic Slope -10.93 122.05 0.09 Variance Intercept -2262.30 384.98 5.88 Linear Slope -82.81 39.01 2.12 Quadratic Slope -158.79 139.32 1.14 Regression Weights Ag e ->Intercept 0.19 1.10 0.66 1.65 Age ->Linear Slope 0.27 0.37 0.34 1.07 Age ->Quadratic Slope 0.20 0.48 0.65 0.75 Gender ->Intercept 0.02 2.95 14.44 0.20 Gender ->Linear Slope 0.10 3.07 7.49 0.41 Gender ->Quadratic Slope 0.14 7.27 14.08 0.52 Education ->Intercept 0.03 0.73 2.69 0.27 Education ->Linear Slope 0.28 1.46 1.39 1.05 Education ->Quadratic Slope 0.24 2.24 2.62 0.85 NAART ->Intercept 0.01 0.09 0.93 0.09 NAART ->Linear Slope 0.18 0.31 0.48 0.65 NAART ->Quadratic Slope 0.17 0.53 0.91 0.58 Depression ->Intercept 0.20 1.83 1.14 1.61 Depression ->Linear Slope 0.24 0.51 0.59 0.86 Depression ->Quadratic Slope 0.42 1.60 1.11 1.44

PAGE 201

201 Table 3 8. Continued Parameter Standard Est imate Estimate SE z Value Anxiety ->Intercept 0.06 0.39 0.77 0.50 Anxiety ->Linear Slope 0.10 0.15 0.40 0.37 Anxiety ->Quadratic Slope 0.06 0.17 0.75 0.22 TST ->Intercept 0.16 0.17 0.12 1.39 TST ->Linear Slope 0.01 0.003 0.06 0.04 TST ->Quadratic Slope 0.07 0.03 0.12 0.26 TWT ->Intercept 0.05 0.08 0.18 0.46 TWT ->Linear Slope 0.42 0.15 0.09 1.67 TWT ->Quadratic Slope 0.54 0.34 0.17 2.01 Correlations Predicted IQ Education -0.37 -3.20 Sta te Anxiety Depression -0.40 -3.44 TST TWT -0.27 -2.43 Choice RT Mean Intercept -742.47 135.09 5.50 Linear Slope -138.31 53.86 2.57 Quadratic Slope -333.84 114.37 2.92 Variance Intercept -2786.46 459.83 6.06 Linear Slope -0.00 0.00 0.00 Quadratic Slope -302.46 73.85 4.10 Regression Weights Age ->Intercept 0.26 1.78 0.72 2.49 Age ->Linear Slope 0.06 0.07 0.29 0.25 Age ->Quadratic Slope 0.13 0.45 0.61 0. 75 Gender ->Intercept 0.13 18.77 15.59 1.20 Gender ->Linear Slope 0.29 7.45 6.22 1.20

PAGE 202

202 Table 3 8. Continued Parameter Standard Estimate Estimate SE z Value Gender ->Quadratic Slope 0.38 29.30 13.20 2.22 Education ->Intercept 0.14 3.63 2.90 1.25 Education ->Linear Slope 0.09 0.38 1.16 0.33 Education ->Quadratic Slope 0.04 0.53 2.46 0.22 NAART ->Intercept 0.27 2.48 1.01 2.46 NAART ->Linear Slope 0.75 1.15 0.40 2.88 NAART ->Quadratic Slope 0.52 2.43 0.8 5 2.85 Depression ->Intercept 0.07 0.78 1.23 0.64 Depression ->Linear Slope 0.51 0.94 0.49 1.92 Depression ->Quadratic Slope 0.47 2.63 1.04 2.53 Anxiety ->Intercept 0.03 0.23 0.83 0.27 Anxiety ->Linear Slope 0.60 0.74 0.33 2.2 4 Anxiety ->Quadratic Slope 0.39 1.47 0.70 2.10 TST ->Intercept 0.00 0.01 0.13 0.04 TST ->Linear Slope 0.07 0.01 0.05 0.28 TST ->Quadratic Slope 0.01 0.01 0.11 0.08 TWT ->Intercept 0.17 0.30 0.19 1.56 TWT ->Linear Slope 0.09 0.0 3 0.08 0.34 TWT ->Quadratic Slope 0.11 0.11 0.16 0.64 Correlations Predicted IQ Education -0.37 -3.20 State Anxiety Depression -0.40 -3.44 TST TWT -0.27 -2.43 Note: z scores greater than 1.96 are statistically signi ficant ( p < .05).

PAGE 203

203 Tables 3 9. Model f it for c onditional l atent g rowth c urve m odels. 2 (df) p 2 (df) p C MIN /DF CFI TLI IFI RMSEA AIC Number Copy Covariates Only 110.22 ( 60 ) 0.001 -( -) -1.84 0.86 0.78 0.87 0.10 198.22 Sleep Added 108.49 ( 54 ) 0.001 1.73 ( 6 ) > 0.05 2.01 0.84 0.74 0.86 0.11 208.49 Symbol Digit Covariates Only 123.29 ( 60 ) 0.001 -( -) -2.06 0.86 0.79 0.87 0.11 211.29 Sleep Added 115.96 ( 54 ) 0.001 7.34 ( 6 ) > 0.05 2.15 0.86 0.77 0.87 0.12 215.96 Letter Series Covariates Only 126.31 ( 60 ) 0.001 -( -) -2.11 0.88 0.81 0.88 0.11 214.31 Sleep Added 120.82 ( 54 ) 0.001 5.49 ( 6 ) > 0.05 2.24 0.88 0.79 0.89 0.12 220.82 Simple RT Covariates Only 89.10 ( 60 ) 0.009 -( -) -1.49 0.94 0.91 0.94 0.08 177.10 Sleep Added 84.01 ( 54 ) 0.006 5.09 ( 6 ) > 0.05 1.56 0.94 0.89 0.94 0.08 184.01 Choice RT Covariates Only 136.24 ( 61 ) 0.001 -( -) -2.23 0.85 0.77 0.86 0.12 222.24 Sleep Added 131.00 ( 55 ) 0.001 5.24 ( 6 ) > 0.05 2.38 0.85 0.74 0.86 0.13 229.00 Notes: No models displayed significant change in fit from cova riates only model to sleep added model.

PAGE 204

204 Figure 3 4. Graphical r epresentation of a ssociation between TWT and b lock 1 s ymbol d igit

PAGE 205

205 Figure 3 5. Graphical r epresentation of a ssociation between TWT and b lock 1 l etter s eries

PAGE 206

206 Figure 3 6. Graphica l r epresentation of p ractice l earning in s imple RT for both a h igh (i.e., 90th %ile) and l ow (i.e., 10th %ile) TWT s ubject. 435 455 475 495 515 535 555 575 595 615 1 2 3 4 5 Raw Score Block (trials) Simple RT and TWT Low TWT High TWT (0 2) (3 6) (7 10) (11 14) (15 17)

PAGE 207

207 CHAPTER 4 DO WEEKLY DEVIATIONS IN SLEEP IMPACT SUBS EQUENT COGNITIVE FUNCTIONING IN OLDER ADULTS? Introduction Possible links betwe en aspects of sleep and cognitive functioning have long been theorized about and studied. There exists, however, a lack of consensus regarding the role of sleep on cognitive functioning and the effects of sleep loss on cognitive functions, especially in la te life (Bruce & Aloia, 2006). This lack of consensus may be the result of the numerous ways in which both sleep (i.e., objective and subjective measurement, and naturally occurring and experimentally manipulated sleep) and cognitive functioning (objective and subjective measurement across many different sub domains) can be conceptualized and measured. Further, both sleep and cognitive functioning can be measured across many different time scales, which may result in inconsistencies across research studies. Generally speaking, both sleep architecture (Morgan, 2000) and cognitive functioning (Salthouse, 2004) demonstrate considerable age related changes. As such, it is a common idea that age related changes in sleep contribute to cognitive decline In fact, sleep and cognitive functioning do appear to be related in late life; however, this relationship is not without many inconsistencies and seemingly contradictory findings [i.e., reports of poor sleep being related to both better and worse cognitive performa nce exist (Goel et al., 2009; Jones & Harrison, 2001)]. Such contradictory findings may be the result of diverging approaches to the conceptualization and measurement of both sleep and cognition. While the sleep cognition relationship has garnered empiric al attention, a lack of attention to the inherent short term fluctuations found in both sleep (Buysse et al., 2010;

PAGE 208

208 Dzierzewski et al., 2008) and cognition (Hultsch et al., 2008) in favor of studies of between person associations may have resulted in a res tricted conceptualization of the impact of sleep on cognitive performance in late life. Said differently, much of the extant literature has looked at between person associations we know, however, that there is significant amounts of short term fluctuatio n in aspects of sleep and subtypes of cognition thus, there is relatively little that we may be able to generalize from understanding how between person differences in sleep and cognition are related to how short term fluctuation in aspects of sleep and cognition might be related. Disturbed sleep often follows a pattern of several nights of poor sleep followed by a night of higher quality sleep (Buysse et al., 2010). This pattern of naturally occurring short term fluctuations in sleep are analogous to a f orm of naturally occurring experimentally manipulates chronic partial sleep deprivation (see Sleep and Cognitive Functioning section below). The above referenced pattern of short term fluctuation in sleep (i.e., up and down nights of sleep) likely obscures the relationship between cognition and sleep To maximize the power in establishing the sleep cognition relationship, micro longitudinal studies that investigate the time varying nature of sleep and cognitive performance on several occasions are needed. As such, the overall goal of this paper was to examine weekly temporal associations that may exist between sleep and cognitive functioning in late life [i.e., how prior reported sleep was related to dwelling older adults]. Rationale and evidence for the potential relationship between weekly sleep and subsequent cognitive func tioning is provided below. Briefly, the

PAGE 209

209 weekly level of analysis was chosen based on pragmatic rational (i.e., timing involved in the parent study) and theoretical consideration (i.e., extrapolating from chronic partial sleep restriction studies). Sleep a nd Cognitive Functioning For a detailed review of the relationship between nocturnal sleep and daytime cognitive functioning please refer to Chapter 3 (Paper 2). In general, no matter what the manner of sleep assessment or the domains of cognitive function ing assessed, results of studies examining the link between aspects of sleep and various cognitive domains have been mixed. Naturalistic designs, or designs that do not pose an experimental manipulation on sleep, typically employ rather crude measurement of sleep and/or disregard the dynamic nature of sleep [e.g., (Blackwell et al., 2006; Cricco et al., 2001; Nebes et al., 2009; Orff et al., 2007)]. Conversely, experimentally manipulated sleep studies have resulted in detailed information regarding brain f unction (i.e. prefrontal activity and cytokine density), cognitive functioning (i.e., attention, vigilance, executive functions, and reaction time), and very specific sleep characteristics [e.g., (Banks & Dinges, 2007; Goel et al., 2009; Walker, 2008)]. Ho wever, the majority of experimentally manipulated sleep studies did not manipulate sleep in a manner that mirrors actual is, researchers have called for examination into the relationship between naturally occurring characteristics of sleep (or proxies thereof) and various components of cognitive functioning. In an attempt to examine the cognitive consequences of more ecolog ically valid sleep loss, researchers have employed chronic partial sleep deprivation paradigms. Chronic partial sleep deprivation (i.e., reducing the amount of time allowed to sleep at

PAGE 210

210 night for several consecutive nights), which most realistically resembl reportedly mixed effects on cognitive performance. In general, chronic partial sleep deprivation has been found to negatively affect psychomotor vigilance, attention, processing speed, and working memory (Banks & Dinges, 2007). In fact Pilcher and Huffcutt (1996) performed a meta analysis on the effects of sleep deprivation on cognitive performance. The results of their analysis found that chronic partial sleep deprivation, getting less than 5 hours of sleep a night for several nights, resulted in the largest decrease in cognitive performance (Pilcher & Huffcutt, 1996). Such evidence suggests that there may be a naturally occurring relationship between fluctuations in several nights of sleep and subsequent cognitive functioning. As such examination into the relationship between self reported weekly deviations in sleep and subsequent cognitive functioning represents a natural extension of the extant chronic partial sleep deprivation literature. Such research not only represents an extens ion of previous research and addresses direct calls to move towards more ecological validity in research, but due to the inherent short term fluctuation in both sleep and cognitive functioning in late life research examining the association between these c onstructs over time is both timely and theoretically substantive. Short term Fluctuation in Late Life Sleep and Cognitive Functioning Sleep and cognitive functioning are both very dynamic processes. Experts have commented on the relevance of short term f luctuation in sleep patterns for broadening understanding of sleep (Espie, 1991; Pallesen, Nordhus, & Kvale, 1998). However, short term fluctuation in sleep has yet to be largely investigated. T he majority of research has focused on other aspects of sleep [such as measuring sleep in laboratory versus in the home (Edinger et al., 1997) or characterizing good and poor sleepers

PAGE 211

211 (McCrea et al., 2003)] and then commented on the considerable variability observed in the data. In studying the effects of measuring s leep via polysomnography at home or in laboratory several researchers have found that poor sleepers tend to display highly variable sleep while at home compared to normal sleepers and compared to their own sleep in lab (Edinger et al., 1997; Edinger, Mars h, Mccall, Erwin, & Lininger, 1991; Frankel, Coursey, Buchbinder, & Snyder, 1976; Hauri & Wisbey, 1992) Further, it has been shown that sleep in older adults exhibits great short term fluctuation in both self reported quantity and quality (Buysse et al., 2010; D zierzewski et al., 2008). In this sense, from night to night, the sleep of an older adult may vary considerably. As such, single occasion measurements or aggregate indices of sleep may not fully capture the dynamics present in the real world and ass ociations of such measurements with cognitive functioning may not be fully accurate. Interestingly, r esearch has demonstrated that it may be possible to classify poor sleepers by the predictability of their nightly sleep based on levels of variability (Val lieres, Ivers, Bastien, Beaulieu Bonneau, & Morin, 2005), as poor sleepers tend to exhibit more variable sleep (Edinger et al., 1997; Edinger et al., 1991; Frankel et al., 1976; Hauri & Wisbey, 1992) than good sleepers (Edinger et al., 1997; McCrae et al., 2005; McCrea et al., 2003) No known research has examined short term fluctuation in sleep from week to week. However, weekly deviations in self reported sleep appear plausible as previous research has shown that several nights of sleep may vary in predic table pattern in older adults (Buysse et al., 2010). In contrast to the limited empirical research on short term fluctuation in sleep, short term fluctuation in cognitive functioning in late life is relatively well studied. It has

PAGE 212

212 been found that older ad ults exhibit a great amount of short term fluctuation (i.e., over 50% of the amount of between person fluctuation) in performance in vocabulary, speed, reasoning, and memory (assessed with 13 cognitive tests) (Salthouse et al., 2006). Short term fluctuatio n across many cognitive domains (i.e., nonverbal and verbal reaction time, perceptual speed, working memory, episodic memory, visual motor dexterity, processing speed, and mental computation and flexibility) has been found to relate to increased age (Hults ch et al., 2002; Nesselroade & Salthouse, 2004), level of cognitive functioning in those domains and in general fluid and crystallized composite scores (Hultsch et al., 2002; Nesselroade & Salthouse, 2004; Salthouse et al., 2006), diagnosis status (i.e., A D, PD, MCI, etc.) (Burton et al., 2002; Hultsch et al., 2000; Sliwinski et al., 2003), and impending death (MacDonald, Nyberg, Sandblom, Fischer, & Backman, 2008). Short term fluctuation in visual motor dexterity, perceptual speed, and executive functionin g has also been found to be a stable characteristic of individuals that is, people who are more inconsistent at one point in time may be more inconsistent at a future time (Nesselroade & Salthouse, 2004). As such, short term fluctuation in cognitive func tioning (i.e., cognitive performance across the above listed domains) is considered an important indicator of overall health (indexed by time to death, cognitive vitality, etc.) (Hultsch et al., 2008). Due to the large amounts of short term fluctuation pre sent within late life cognitive functioning, sleep deprivation and cross sectional studies of the sleep cognition relationship in late life may not accurately capture the actual relationship between sleep and cognitive functioning. Concretely, observed rel ationships may represent aberrant circumstances due to high levels of short term

PAGE 213

213 fluctuations or the lack of any observed relationships may be the result of inaccurately modeling the variables of interest. Researchers have attempted to capture the covar within individuals. That is, researchers have attempted to examine whether peaks and valleys in one behavioral function are concurrent with peaks and valleys in another. Such investigations are of value in a sense that u nderstanding from a coupled variability perspective what outside variables might predict peaks and valleys in cognition may in turn lead to the identification of modifiable correlates that can be manipulated in future studies. If a modifiable factor, like sleep, is related to fluctuations in cognitive functioning, perhaps treatments may be developed to stabilize late life cognitive performance. Thus, the search for correlates that explain within person variability in cognition has yielded several initial f indings. For example, it has been reported that o n days when a stressful event occurred, older adults had worse cognitive performance intrapersonal mean, they also perform ed worse cognitively (Gamaldo et al., 2008) Additionally, sleep quality has been found to relate to improvements in next (McCrae et al., 2008) and blood pressure (Endeshaw et al., 2009; Silva et al., 2000). Regarding the dynamic association between sleep and cognitive functioning, very little work has been conducted. It has been reported that following a night of above average self reported SOL and terminal wakefulness older adults with poor sleep perform better on measures of re asoning and processing speed (Dzierzewski, 2007). In older African American elders it has been reported that the greater an individual

PAGE 214

214 deviated from their usual self re composite) would decline (Gamaldo et al., 2010). While these two studies represent the only known attempts to examine the time varying association between sleep and cognition (especially in late life), others have remarked on the necessity for addition al research into these associations. Bliese and colleagues have called for their colleagues to varying in future investigations of the sleep cognition relationship (Bliese et al., 2006). The current investigation is a n attempt to answer this call for added attention regarding the potential dynamic association between sleep and cognitive functioning in late life. The Current Investigation Results from studies employing a sleep deprivation and sleep restriction paradigm have limited ecological validity due to the nature of the manipulation (i.e., artificially imposed sleep loss may not mimic real world sleep loss). Further, the lack of sleep deprivation/restriction studies including older participants, limits generalizat ion to this segment of the population. Results from cross sectional investigations have largely ignored the inherent short term fluctuation (i.e., night to night inconsistency) in sleep of older adults (Buysse et al., 2010; Dzierzewski et al., 2008). Furth ermore, both sleep deprivation/restriction and cross sectional investigations regarding the relationship between sleep and cognitive functioning, have not accounted for the short term fluctuation (i.e. day to day inconsistency) in cognitive functioning of older adults (Hultsch et al., 2008). Thus, how deviations in sleep may relate to subsequent cognitive functioning in community dwelling older adults remains largely unknown. This question is of vital importance given the highly variable nature of sleep in late life (Buysse et al.,

PAGE 215

215 2010; Dzierzewski et al., 2008), and the limited ecological validity of both experimental and correlational/quasi experimental investigations into the sleep cognition relationship. Both sleep (Buysse et al., 2010; Dzierzewski et al., 2008) and cognitive functioning (Hultsch et al., 2008) display significant amounts of short term fluctuations in old age. Current investigations into understanding cognitive inconsistency in late life have reported that both fluctuations in stress (Sliwinski et al., 2006) and blood pressure (Gamaldo et al., 2008) are related to fluctuations in cognitive functioning Given that (McCrae et al., 2008) and blood press ure (Endeshaw et al., 2009; Silva et al., 2000) it seems reasonable to suggest that deviations in sleep may also relate to next day cognitive functioning in late life. Previous research has demonstrated such a relationship, though within specialized segm ents of older adults (i.e., poor sleepers and African Americans) (Dzierzewski, 2007; Gamaldo et al., 2010). Sleep restriction studies have demonstrated that several days of experimentally reduced sleep has a negative impact on vigilance (Bliese et al., 20 06; Dinges, 1997; Drake et al., 2001; Vgontzas et al., 2004); though this impact may be reduced in elders (Bliese et al., 2006). Whether adaption to several days of experimentally reduced sleep is possible is currently debatable (Dinges, 1997; Drake et al. 2001). This study represents a test of the association between naturally occurring sleep restriction and cognitive functioning in late life. Stated more specifically, the current investigation sought to extend the experimental chronic partial sleep depri vation research into the domains of naturally occurring, self reported sleep. Through examination of

PAGE 216

216 reported sleep and subsequent cognitive functioning, the work on experimental sleep restriction ma y be examined in elders under naturally occurring situations (due to the natural fluctuations in the sleep of older adults). The overall goal of this paper was to examine the week to week association between time spent asleep (sleep gained) and time spen t awake (i.e., sleep loss) and cognition functioning, specifically processing speed and reasoning, in a sample of older adults Specific aims of this study include d : (1) To explore the amount of short term fluctuation in cognitive functioning (i.e., proces sing speed and reasoning) across a 1 7 week period, and (2) To explore how weekly deviations in self reported total sleep time and total wake time may be related to subsequent cognitive functioning (i.e., processing speed and reasoning) in older adults. It was hypothesized that: (1) older adults would display substantial amounts of short term variability in cognitive functioning across the 17 week study period, and (2) relative changes in level of self reported sleep would be associated with subsequent cogn itive functioning. However, given the lack of previous studies employing a similar methodology and the equivocal results of previous studies into the sleep with cognitive functioning cannot be made. Methods General Study Design This study represents a secondary analysis of the Active Adult Mentoring Program (Project AAMP). The primary objective of Project AAMP was to test the efficacy of a social cognitive lifestyle intervention to increase moderate intensity exercise in older adults. Participants were randomly assigned to either an Active Lifestyle intervention

PAGE 217

217 arm (receiving weekly, group based behavioral counseling) or a Health Education arm (receiving appropriately matched healt h education). The current study utilizes data from the initial 17 weeks of the study, including a baseline week of observation prior to group assignment and 16 weeks of intervention. The study protocol was approved by the appropriate university institution al review boards. Procedure Individuals whom expressed interest in study participation were initially screened by telephone (see below). Following telephone screening, qualified participants were consented and completed a baseline assessment. This baselin e assessment included seven consecutive nights of self reported sleep monitoring followed by one in person computerized cognitive assessment. Next, participants were randomized to either the Active Lifestyle or Health Education arm of the intervention. Eac h intervention arm consisted of sixteen weekly group meetings [ see (Aiken Morgan, 2008; Buman, 2008; Buman et al., 2011) for more information]. During the intervention period, each participant continued to monitor their sleep nightly. Prior to each group m eeting, all sleep logs were collected and checked for compliance. Similarly, either before or after each group meeting all participants completed the computerized cognitive battery. Thus, this study included 17 consecutive weeks of sleep monitoring, paired with 18 weekly cognitive assessments, such that 7 nightly sleep logs were completed prior to each cognitive assessment session. Please refer to Figure 4 1 for a Gantt chart depicting the flow of sleep and cognitive data collection. Participants Study p articipants include 87 adults aged 50 years and greater who participated in Project AAMP. Potential enrollees responded to community based health promotion

PAGE 218

218 recruitment delivered through local media outlets. While participants were not recruited or screened based on their sleep characteristics, older adults present with great heterogeneity and have increased prevalence of disturbed sleep (Morgan, 2000). Thus, the sample is likely to include a wide range of elder sleepers, including both good and poor sleeper s. To ensure their suitability for the study, subjects went through a thorough screening process that included many inclusion and exclusion criteria. Inclusion/Exclusion Criteria All potential participants were screened by telephone to exclude individuals based on the following criteria: severe dementing illness, history of significant head injury (loss of consciousness for more than 5 minutes), neurological disorders disease), inpatient psychiatric treatment, extensive drug or alcohol ab use, use of an anticholinesterase inhibitor (such as Aricept) severe uncorrected vision or hearing impairments, terminal illness with life expectancy less than 12 months, major medical illnesses cardiovascular disease, pulmonary disease requiring oxygen or steroid treatment, and ambulation with assistive devices. Telephone screening include d the 11 item Telephone Interview for Cognitive Status [ TICS; (Brandt et al., 1988) ], utilizing a cut off score of 30 points to differentiate mild dement ia from cogniti ve ly intact (Brandt et al., 1988) All study participants were required to self report sedentary lifestyle [defined of moderate or vigorous physical activity during the previous 6 months ( Physical Activity G uidelines Advisory Committee Report, Part A: Executive Summary 2009) ]. All i ndividuals were required to come to the laboratory with a note from their primary care physician acknowledging their ability to participate in the study prior to formal enrollment

PAGE 219

219 Demographics Each study participant provided demographic data through means of a telephone screening instrument. Information regarding participant age (measured in years since birth), gender (male or female), and education level (years of education ) were collected. Descriptive Statistics The final sample included 87 adults aged 50 years and older. Mean age for the entire sample was 63.33 years, range = 50 87 years. The sample was highly educated, average years of education of 16.14, and predominately fem ale, 82%. Please refer to Table 4 1 for a complete list of demographic/descriptive statistics. Measures Cognitive Measures All cognitive measures were computer administered. Administration occurred once per week for 18 consecutive weeks. Computerization o f the cognitive measures was done using DirectRT experimental generation program (Jarvis, 2008a) All computerized cognitive measures were then compiled and administered via MediaLab experimental implementation program (Jarvis, 2008b). In an attempt to min imize practice effects due to memorization commonly found in repeated cognitive assessments (Salthouse et al., 2004) fourteen alternate forms of each test were used and rotated such that the same version of any given test was not given within 6 weeks of e ach other. The alternate forms were constructed to be comparable in difficulty and cognitive resources needed to complete them and have been shown to have high test retest reliabilities (Allaire & Marsiske, 2005; McCoy, 2004) All cognitive measures were c hosen for inclusion in the present study due to their reliance on frontal lobe functioning [which is impacted by sleep (Goel et al., 2009; Jones & Harrison, 2001)] and their utility in measuring cognitive

PAGE 220

220 domains (i.e., attention, processing speed, and exe cutive functions) with previous reported associations with acute changes in sleep (Banks & Dinges, 2007; Durmer & cognitive functioning, cognitive testing included tasks that likely vary their intrinsic motivation due to difficulty (Pilcher et al., 2007), are differentially supported by the prefrontal regions of the brain (Harrison & Horne, 2000; Jones & Harrison, 2001; Lim & Dinges, 2010), and rely largely on vigilance/arousa l (Bonnet & Arand, 2010; Durmer & Dinges, 2005). The Letter Series task (Thurstone, 1962) is primarily a measure of reasoning In this task, participants ha d to identify the pattern for a series of letters. Participants we re asked to choose the letter that would continue an established pattern (A B D A B D A B ___?) in a series of letters from five answer choices. Participants were given four minutes to complete as many items as possible. The performance score was the number of correct responses. Processing speed and attention were assessed with the Symbol Digit and Digit Copy tests (Smith, 1982). These tests consist of matching symbols that are paired with numbers (Symbol Digit) or numbers paired with same numbers (Digit Copy) as quickly as possible There was a 120 second time limit for each task and the performance score was the total number of correct pairings made by the participant The Simple and Choice Reaction Time task (Hultsch et al., 2000) was used to assess reaction time The Simple Reaction Time (SRT) task present ed respondents screen. Participants were instructed to press a key with their preferred hand as quickly

PAGE 221

221 as possible when the signal stimulus (+) appears The Choice Reaction Time (CRT) d into a circle and the location of the circle was random ly equalized across trials. Respondents were instructed to press a key corresponding to the location of the circle as quickly as possible. 10 practice trials are followed by 50 test trials. The outcome measure s for SRT and CRT was the mean latency of the t otal 6 0 test trials. Sleep Measures Participants completed sleep diaries (Lichstein et al., 1999) each morning for the entirety of Project AAMP (i.e., 18 weeks/126 days). The sleep diaries provide subjective estimates of the following sleep parameters: (1 ) total sleep time (TST): the total amount of time asleep during the night (TST = total time in bed sleep onset latency wake time after sleep onset and terminal wakefulness ); (2) total wake time (TWT): the total awake time in bed (TWT = sleep onset lat ency + wake time after sleep onset + terminal wakefulness ); (3) sleep onset latency (SOL): the time it took to fall asleep after laying down with the intention of going to sleep; (4) wake time after sleep onset (WASO): the total amount of time spent awake during the night from the time sleep was first initiated until the final wake up (5) sleep quality rating (SQR): overall rating of the quality of sleep (from 1= very poor to 5 = excellent) (6) terminal wakefulness (TWAK): the amount of time spent awake laying in bed in the morning following final wake up, (7) number of awakenings (NWAK): total number of nocturnal awakenings, and (8) time in bed (TIB): total amount of time spent in bed Due to conceptual [i.e., potential consequences of sleep loss and sl eep gain (Ellenbogen, 2005)] and pragmatic (i.e., high multicollinearity) issues, only TST and

PAGE 222

222 TWT were employed as individual difference predictors of cognitive functioning For each participant, two unique parameterizations of TST and TWT were calculated and utilized in subsequent analyses. For the purposes of this report TST and TWT were parameterized in the following ways: (1) Person overall mean level of sleep across the entire study period), and (2) Week level c entered grand mean). As such, sleep was parameterized to include: (1) Overall mean assessmen average weekly TST and TWT values from their overall mean level of TST and TWT). Please see Table 4 2 for a complete listing of inter correlations among sleep variables (interindividual and intraindividual correlations) and descriptive sleep values. See Figure 4 2 for a graphical representation of weekly deviation in TST and TWT across the study period. Analyses Preliminary Analyses Prior to statistical analyses to address each main aim, preliminary analyses were conducted. Preliminary analyses: (1) examined normality of the dependent variables, (2) examined rates of missingness among the data, and (3) examined attrition (differences between attrited and non attrited subjects). Prelimina ry analysis examined the cognitive data (i.e., dependent variables) for normality. All cognitive data were screened for outliers at the intraindividual level (i.e., within person from trial to trial). Further, all cognitive variables were also screened for potential outliers at the interindividual level (i.e., between persons). Interindividual

PAGE 223

223 outliers were replaced with their respective 3 standard deviation values, while intraindividual outliers were simply removed from the dataset prior to calculation of occasion specific values. While outlier trimming in a study of within person fluctuation may appear counterintuitive, practices like trimming at the +/ 3 SD level have previously been employed in studies of intraindividual variability in late life cognit ive functioning [i.e., (Bunce, MacDonald, & Hultsch, 2004; Hultsch et al., 2002)]. Skewness and kurtosis values were also examined using generally agreed upon criteria (i.e., skewness and kurtosis values less than 1.0) (Field, 2005). While multilevel mode ling (MLM) uses all available data, and is valid for making inferences to the population of origin when data are missing at random, the impact of non random missingness will increase with higher levels of missigness. Thus it was important to begin by char acterizing the extent of missingness within the data (Bryk & Raudenbush, 1992). Analysis of attrition was conducted to examine whether there were differences between study completers (those whom provided data on any occasion after baseline) and study attr iters (those whom dropped out and did not provide data following baseline) on all cognitive, descriptive/demographic variables, and sleep variables at baseline. The parent study (Project AAMP) for the current paper was designed to test the efficacy of a s ocial cognitive lifestyle intervention to increase moderate intensity exercise in older adults. Participants were assigned to either an Active Lifestyle intervention or a Health Education. As such, concerns regarding differing cognitive trajectories betwee n these groups may be present. However, previous work have shown few baseline to posttest fitness effects by group status (Buman, 2008) and no

PAGE 224

224 baseline to posttest cognitive effects by group status (Aiken Morgan, 2008). Thus, group status was not controlle d for in subsequent analyses. Main Analyses To examine the amount of short term fluctuation (Aim 1) in cognitive functioning in older adults (above and beyond learning trends), unconditional growth curves (with predictors of linear and quadratic time onl y) were estimated. These models produced intercept and residual variance estimates. Such variance estimates were subsequently used to compute intraclass correlation coefficients (ICC) which serves as an estimate of the proportion of the total variance at the interindividual (i.e., between subjects) level in a given variable (Bryk & Raudenbush, 1992) subtracted from 1, providing an estimate of the concomitant amount of intraindividual variability within the same variable. As s uch, 1 ICC is a representation of the relative proportion of within person variation (in the absence of an overall trend toward increase/decrease, this can be interpreted as the amount short term fluctuation) in cognitive functioning. To explore various different temporal associations (i.e., mean level sleep and weekly deviations in sleep) that may exist between self reported sleep and cognitive functioning in older adults (Aim 2), separate ( one for each cognitive variable) conditional multilevel models (MLM) were parameterized (Singer & Willett, 2003). MLM, also referred to as mixed effects modeling or hierarchical linear modeling [ HLM; (Bryk & Raudenbush, 1992)] is an extension of the general linear model, and does not require observations to be indep endent. Model building was conducted in a hierarchical manner, such that each dependent variable (cognitive functioning) was predicted by four increasingly compl ex models. Model 1 included no predictor estimates and w as

PAGE 225

225 parameterized to allow for estimatio n of subsequent fit statistics (i.e., null model) Model 2 permitted estimates of time trends (e.g., gain or decline) to control for any systematic growth in the data growth model) Mod el 3 include d the covariates of age, gender, and education (i.e., conditional model with covariates only), as previous research has suggested that these variables are associated with late life cognitive functioning. Model 4 include d the previously entered variables, with the addition of mean level and weekly deviation TST and TWT variables (i.e., conditional model with predictors of interest). The final model (Model 4) predict ed weekly cognition with: average level of 00 ) linear and quadratic time function s to control for growth trajectories in 10 20 ) demographic variables [age ( 01 ) gender ( 02 ) education ( 03 ) ], person level TST ( 04 ), weekly centered TST ( 30 ) person level TWT ( 0 5 ), weekly centered TWT ( 4 0 ), random coefficients of the time functions ( r 1i r 2i ) and within person TST (weekly) and TWT (weekly) functions ( r 3i r 4 i ), random error term ( e it ), and random residual component ( r oi ). The specifics of MLM are beyond t he scope of this paper. Interested readers are referred to other sources (Bryk & Raudenbush, 1992) All variables were evaluated based on their significance levels, effects on overall model fit (i.e., 2LL), and their effect on intercept and residual relat ed variance estimates (Bryk & Raudenbush, 1992; Singer & Willett, 2003). Final model equation was: Cognition it 00 10 20 (Time 2 01 (Age i 02 (Gender i 03 (Education i ) + 04 ( ) 0 5 ( ) 30 ( 4 0 ( ) + r 1i (Time) + r 2i (Time 2 ) + r 3i ( )+ r 4 i ( ) + r oi + e it

PAGE 226

226 Results Preliminary Analyses Normality Normality of the cognitive data was thoroughly examined in previous studies. Please refer to Paper 2 (Chapter 3) for full details of this process. Interested readers are encouraged to inspect Table 3 4 for cognitive descriptive and normality statistics pre and post outlier trimming procedures, and for raw cognitive scores for each measure at each occasion. As stated previously, outlier trimming has historical grounding in variability studies [i.e., (B unce et al., 2004; Hultsch et al., 2002)]. In fact, such a practice person fluctuation (Hultsch et al., 2002). In total, the data cleaning procedure resulted in much more normally distribut ed data and required very little of the data being manipulated (i.e., 0.88%), which falls well below the level of data manipulation resulting from outlier trimming reported in previous within person fluctuation studies (Bunce et al., 2004; Hultsch et al., 2002). Following data cleaning procedures, all RT based data (i.e., N Back, Simple and Choice RT, and Trails A and B) were transformed such that higher scores were indicative of better performance. This was done to place all cognitive measures on a similar metric and to improve interpretability of the main study Aims. Missing Data Eighty seven subjects were enrolled in the current investigation. Throughout the repeated cognitive assessments, participation waxed and waned considerably. In general, participa tion was at its highest during the earlier weeks of assessment. Week 1 had the lowest percent of cognitive missing data with just over 10% missing. In contrast,

PAGE 227

227 participation was at its lowest at the end of the repeated assessments. Week 15 had the highest rate of cognitive missing data with just over 47% missing. Overall, the average level of cognitive missing data was approximately 29%, while the median amount of cognitive missing data was also approximately 28% missing. A similar pattern of increasing mi ssing data as the study progressed was observed for the sleep variables. Missing weekly sleep was lowest at the beginning of the trial (approximately 3%). The average level of sleep missing data was approximately 18%. Please refer to Figure 4 2 for a graph ical representation of missing cognitive and sleep data across the study period. Attrition Only 1 subject failed to provide any data following baseline assessment. All subsequent analyses are based on all available participants minus this one non responde nt. Main Analyses Aim 1: Amount of Short term Fluctuation The ICC for Number Copy was 0.6455, indicating that 64.55% of the variance in Number Copy is between persons while 35.45% of the variance is within persons (across occasions). The ICC for Symbol Di git was 0.6955, indicating that 69.55% of the variance in Symbol Digit is between persons while 30.45% of the variance is within persons (across occasions). The ICC for Letter Series was 0.6619, indicating that 66.19% of the variance in Letter Series is be tween persons while 33.81% of the variance is within persons (across occasions). The ICC for Simple RT was 0.6944, indicating that 69.44% of the variance in Simple RT is between persons while 30.56% of the variance is within persons (across occasions). The ICC for Choice RT was 0.7746,

PAGE 228

228 indicating that 77.46% of the variance in Choice RT is between persons while 22.54% of the variance is within persons (across occasions). Please see Figures 4 3 and 4 4 for graphical representations of the amount of short ter m fluctuation in cognitive functioning across the study period. Aim 2: Temporal Sleep Cognition Associations Predictor estimates, significance levels, and model parameters are presented in Table 4 3 for Number Copy, Table 4 4 for Symbol Digit, and Table 4 5 for Letter Series, Table 4 6 for Simple RT, and Table 4 7 for Choice RT. All models were constructed with quadratic time residualized from linear time, to control for multicollinearity among the time trends. TWT and TST (both mean level and weekly devia tion) were included in all models in their raw metric. Correlations amongst the various sleep predictors, at the various levels, were found to be low (see Table 4 2). Further, to confirm that collinearity between the sleep variables was not artificially bi asing the results, all models were conducted once more splitting the TWT and TST predictors into separate models. These separated models yielded identical patterns of results (in terms of significance levels and direction of associations). Thus, only the c ombined model is shown below. In the final MLM predicting Number Copy, no individual sleep predictors was found to be significant; however, the addition of TST and TWT to the model did significantly improve model fit, 2 = 197.83(4), p < .001. Age, = 0 .16, SE = 0.04, t (76.63) = 3.90, p < .001 was a significant between person predictor suggesting that older individuals had lower than average Number Copy performance At the within person level both Linear Time, = 0.19, SE = 0.03, t (72.32) = 7.04, p < .001 and Quadratic Time, = 0.01, SE = 0.005, t (77.05) = 2.35, p < .05 were significant predictors of Number Copy

PAGE 229

229 performance, suggesting that across the 18 week study period individuals experienced improvements in Number Copy performance that decele rated as the study progressed The linear temporal and quadratic trends in Number Copy performance were qualified by significant random effects, suggesting that there were individual differences in the 17 week Number Copy trends. The model explained 29% of the within person variance, 38% of the between person variance, and 34% of the total variance in Number Copy performance. In the final MLM predicting Symbol Digit, Age, = 0.15, SE = 0.03, t (78.31) = 4.47, p < .001 and person mean TWT, = 0.03, SE = 0.01, t (78.54) = 3.38, p < .001 were significant between person predictors suggesting that older individuals and individuals who spent more time awake during the nigh t on average had lower than average Symbol Digit performance No other sleep variable (either fixed or random effect) was found to be a significant predictor of cognitive functioning. However, the addition of TST and TWT to the model did significantly impr ove model fit, 2 = 198.81(4), p < .001. At the within person level both Linear Time, = 0.20, SE = 0.02, t (79.34) = 8.16, p < .001 and Quadratic Time, = 0.02, SE = 0.005, t (83.85) = 4.90, p < .001 were significant predictors of Symbol Digit performance, suggesti ng that across the 18 week study period individuals experienced improvements in Symbol Digit performance that decelerated as the study progressed The linear temporal and quadratic trends in Symbol Digit performance were qualified by significant random eff ects, suggesting that there were individual differences in the 17 week Symbol Digit trends. The model explained 38% of the within person variance, 49% of the between person variance, and 44% of the total variance in Symbol Digit performance.

PAGE 230

230 In the final MLM predicting Letter Series, Age, = 0.15, SE = 0.04, t (81.34) = 3.61, p < .001 Education, = 0.39, SE = 0.16, t (81.03) = 2.49, p < .05 and person mean TWT, = 0.03, SE = 0.01, t (81.32) = 2.66, p < .01 were significant between person predictors suggesting that older individuals, individuals with lower educational levels, and individuals who spent more time awake during the night on average had lower than average Letter Series performance No other sleep variable (either fixed or random effect) was found to be a significant predictor of cognitive functioning. However, the addition of TST and TWT to the model did significantly improve model fit, 2 = 181.47(4), p < .001. At the within person level both Linear Time, = 0.38, SE = 0.03, t (78.79) = 13.14, p < .001 and Quadratic Time, = 0.01, SE = 0.004, t (909.20) = 2.80, p < .01 were significant predictors of Letter Series performance, suggest ing that across the 18 week study period individuals experienced improvements in Letter Series performance that decelerated as the study progressed The linear temporal trend in Letter Series performance was qualified by significant random effect, suggesti ng that there were individual differences in the 17 week Letter Series trend. The model explained 43% of the within person variance, 47% of the between person variance, and 46% of the total variance in Letter Series performance In the final MLM predicting Simple RT, n o sleep variables (either fixed or random effect) was found to be a significant predictor of cognitive functioning. While no individual sleep predictors was found to be significant, the addition of TST and TWT to the model did significantly im prove model fit, 2 = 402.95(4), p < .001. Additionally, there were no significant between person predictors At the within person level both Linear Time, = 1.21, SE = 0.50, t (76.89) = 2.49, p < .05 and Quadratic Time, = 0.25, SE =

PAGE 231

231 0.06, t (69.90) = 4.34, p < .001 we re significant predictors of Simple RT performance, suggesting that across the 17 week study period individuals experienced improvements in Simple RT performance that decelerated as the study progressed The linear temporal trend in Simple RT performance w as qualified by significant random effect, suggesting that there were individual differences in the 18 week Simple RT trend. The model explained 29% of the within person variance, 41% of the between person variance, and 37% of the total variance in Simple RT performance In the final MLM predicting Choice RT, the addition of TST and TWT to the model did significantly improve model fit, 2 = 365.60(4), p < .001. N o sleep variables (either fixed or random effect) were found to be a significant predictor of cog nitive functioning. Age, = 1.58, SE = 0.67, t (81. 40) = 2.36, p < .05 was a significant between person predictor suggesting that older individuals had lower than average Choice RT performance At the within person level Linear Time, = 1.70, SE = 0. 35, t (68.86) = 4.86, p < .001 Quadratic Time, = 0.50, SE = 0.08, t (76.38) = 6.64, p < .001 were significant predictors of Choice RT performance. The time trends suggest that across the 17 week study period individuals experienced improvements in Choi ce RT performance that decelerated as the study progressed .. The linear temporal trends in Choice RT performance were qualified by significant random effects, suggesting that there was individual differences in the 17 week Choice RT trends. The model expla ined 45% of the within person variance, 19% of the between person variance, and 28% of the total variance in Choice RT performance

PAGE 232

232 Discussion The current study sought to investigate aspects of short term fluctuations in late life cognitive functioning. Spe cifically, this study aimed to (1) replicate previous findings (Hultsch et al., 2002; Hultsch et al., 2008; Nesselroade & Salthouse, 2004; Salthouse et al., 2006) of large amounts of short term fluctuations (i.e., within person or intraindividual variabili ty) in late life cognitive functioning, and (2) investigate various different temporal associations (i.e., mean level and weekly) that may exist between self reported TST and TWT and cognitive functioning in older adults. As short term fluctuations in cogn itive functioning have been previously demonstrated to be important predictors of cognitive and overall health in late life (Hultsch et al., 2002; Hultsch et al., 2008; MacDonald et al., 2008; MacDonald et al., 2008; Nesselroade & Salthouse, 2004; Sliwinsk i et al., 2003), identifying predictors (especially modifiable predictors) of this short term fluctuation becomes paramount. It was found that both processing speed measures (i.e., Number Copy, Symbol Digit, Simple RT, and Choice RT) and an executive proc essing measure (i.e., Letter Series) demonstrated large amounts of short term (i.e., week to week) fluctuation. Persons whom spent more time awake on average during the night were found to perform worse, on average, on some cognitive variables. However, no relationships were observed between weekly fluctuations in TST and TWT and cognitive functioning. Amount of Fluctuation We observed large amounts of short term fluctuation in each cognitive variable. Number Copy, Symbol Digit, Letter Series, Simple RT, and Choice RT displayed 35%, 30%, 34%, 31%, and 23% of the total variance within persons (across occasions). These estimates of short term fluctuation are slightly lower than previously reported

PAGE 233

233 estimates in the cognitive aging literature, which typicall y approach or exceed 50% of the size of between person variability (Li et al., 2004; MacDonald et al., 2008; Nesselroade & Salthouse, 2004; Salthouse et al., 2006). While smaller, the estimates from the current investigation are of a similar magnitude. Due to differences in the method used to calculate the amount of short term fluctuation, estimates are presented as the percent of the total amount of variance (as opposed a proportion of the amount of between person fluctuation). If placed in this between pe rson metric [through simple conversion using the following formula: short term fluctuation/(100 short term fluctuation)], Number Copy, Symbol Digit, Letter Series, Simple RT, and Choice RT displayed 55%, 44%, 53%, 44%, and 29.% of the amount of between persons variability within persons (across occasions). Such estimates are congruent with previous investigations into the amount of short term fluctuation in cognitive functioning in late life (Li et al., 2004; MacDonald et al., 2008; Nesselroade & Salthou se, 2004; Salthouse et al., 2006). Such findings add confirmation to the growing evidence suggestive of the presence of short term fluctuation in late life cognitive functioning (Hultsch et al., 2008). This is important due to the fact that previous resea rch has demonstrated that short term fluctuation in cognitive functioning is typically associated with negative outcomes (Hultsch et al., 2000; MacDonald et al., 2008; MacDonald et al., 2008; Sliwinski et al., 2003). As such, inspection of factors related to short term changes in late life cognition appears needed and warranted. It has been found that blood pressure (Gamaldo et al., 2008), mood (Sliwinski et al., 2006), and physical functioning (Li et al., 2004) are associated with the up and down, short te rm fluctuation in late life cognition. Two

PAGE 234

234 previous investigations have directly examined the potential associations between deviations in sleep and cognitive functioning in late life (Dzierzewski, 2007; Gamaldo et al., 2010), though both studies were cond ucted on specialized segments of the older adult population. This study represents a frontline attempt at examination into the dynamic association between self reported sleep and cognitive functioning in community dwelling elders. Temporal Sleep Cognition Associations We observed that individuals who spent longer amounts of time awake during the night, on average, performed worse on a measure of processing speed (i.e., Symbol Digit) and reasoning (i.e., Letter Series). Such relationships between time spent awake during the night and poorer cognitive functioning make intuitive sense, and are supported by previous reports of poor sleep being associated with lower levels of cognitive functioning (Blackwell et al., 2006; Cricco et al., 2001; Nebes et al., 2009; Tworoger et al., 2006). These associations are suggestive of nocturnal wake time impeding normally occurring processes during sleep that may either maintain or promote healthy cognitive functions. Decreased brain restoration due to nocturnal wake time has been proposed as a potential mechanism responsible for the association between poor sleep and decreased levels of cognitive functioning (Cricco et al., 2001; Nebes et al., 2009; Tworoger et al., 2006). The negative association between TWT and processing s peed and reasoning was observed at the person mean level of analysis (i.e., TWT was the average of 119 days of self reported sleep). Such an association is indicative of a chronic relationship between nightly wake time and waking cognitive functioning. As in chronic sleep restriction (Durmer & Dinges, 2005), the effects of

PAGE 235

235 additional time spent awake during the night may have a cumulative effect on Interestingly, we observed no relationships between either TST or TWT and cognitive functioning at the weekly level of analysis. Such null findings indicate that following a week of above or below average TST or TWT older adults in the current sample did not perform better or worse on cognitive measures of processing speed or reasoning. While the absence of significant results does not suggest the null hypothesis is correct, several interesting possibilities for the lack of an association at the weekly level abound. Previous research has demonstrated a rath er consistent negative effect of experimentally induced sleep restriction on vigilance (Bliese et al., 2006; Dinges, 1997; Drake et al., 2001; Vgontzas et al., 2004). There might exist a fundamental difference in the way naturally occurring deviations in s leep occur and are perceived as opposed to restrictions in sleep that are arbitrarily imposed by an outside source. Conversely, naturally occurring sleep changes may need to be more chronic in nature to accumulate a noticeable effect on cognitive functioni ng. Previous research has suggested that sleep restriction may have a reduced impact in older adults (Bliese et al., 2006). As the deviations in sleep were naturally occurring in the present study, perhaps a reduced effect of sleep restriction would be to o small to be reliably detected. While there is debate regarding whether individuals are able to adapt to several days of experimentally reduced sleep (Dinges, 1997; Drake et al., 2001), the potential for adaptation to naturally occurring deviations in sle ep is very possible. In the present study, the modal weekly deviation in TST was approximately 35

PAGE 236

236 minutes, while the modal deviation in TWT was only approximately 3 minutes. Such estimates are considerably smaller than experimentally reduced levels of slee p. Lastly, perhaps sleep may be operating on a more acute basis, as is the case in the plethora of sleep deprivation experiments. Weekly associations between sleep and cognition may be both too short to observe any chronic effects and too long to detect a suggested by sleep researchers to lprit for the lack of associations between sleep and cognitive functioning in previous investigations (Crenshaw & Edinger, 1999). The current investigation was not designed to examine such fine grained associations, and as such we were unable to confirm th e time sequence of data collection at a level to allow for such an analysis. Limitations and Future Directions There are several limitations to the current study that need to be recognized. Only two cognitive domains were assessed repeatedly (i.e., proces sing speed and executive processing). It would have been advantageous to examine multiple other cognitive domains in the context of potential dynamic associations between sleep and cognition. Given some of the known associations between sleep and hippocamp al processing and memory formations (Hobson & Pace Schott, 2002; Stickgold et al., 2001; Wagner et al., 2004; Walker et al., 2002), it would have been intriguing to examine the potential dynamic associations between sleep and memory functioning in older ad ults. Future investigations would do well to include repeated assessments of memory functioning. Another limitation of the present investigation is its lack of assessment of potential sleep disorders. Asking subjects whether or not they had a sleep compla int would have

PAGE 237

237 allowed for additional group comparisons that may have led to interesting results. As individuals with sleep complaints are known to exhibit both nocturnal and daytime hyperarousal (Bonnet & Arand, 2010), such an assessment would have potent ially contributed insight into the mechanisms underlying the observed sleep cognition associations. Further, individuals with and without sleep complaints may exhibit distinct dynamic associations between sleep and daytime cognitive functioning. Future r esearch that clearly assesses for the presence of a sleep complaint is needed to examine such hypotheses. As such, lack of subjective assessment of the presence or absence of a sleep complaint is a limitation. Assessment of potential sleep disorders would have yielded information regarding probable sleep apnea. Sleep apnea has known associations with cognitive functioning (Beebe & Gozal, 2002; Findley et al., 1986; Mathieu et al., 2008), which include executive dysfunction and cognitive slowing. However, g iven the preponderance of our greater in males (Jordan & Doug McEvoy, 2003), the risk of our observed sleep cognition association being driven by the presence of sleep apnea i s reduced. Yet, the prevalence of sleep apnea appears to increase with age (Bixler et al., 2001; Bixler et al., 1998), though there still appears to be a gender difference. Future research is needed that includes additional objective assessment for potenti al occult sleep disorders. Additionally, repeated objective assessments of sleep would allow for examination of the short term fluctuations in sleep apnea symptoms and their potential associations yield potential insightful information regarding the role of nocturnal oxygenation for daytime cognitive

PAGE 238

238 functioning. Future research may be well suited to include repeated polysomnographic assessments of sleep. Additionally, repeated objective measurem ent of sleep would yield potentially useful information regarding the mechanisms underlying the dynamic sleep cognition association. Previous examinations have found static (i.e., occurring at a single time point) associations between the various sleep sta ges and cognitive functioning (Ellenbogen, 2005; Hobson & Pace Schott, 2002; Peigneux et al., 2001; Tononi & Cirelli, 2006; Wagner et al., 2004; Walker et al., 2002), perhaps similar relationships exist between sleep stages and cognitive functioning across time. Improved sleep assessment may very well allow for a better understanding of sleep cognition relationship in older adults. This question deserves empirical attention. Another limitation of the current investigation that needs to be acknowledged is the reliance on self reported sleep. Subjective (i.e., self reported measures) and objective (PSG and actigraphy measures) indicators of sleep vary in several important ways. As such, information gathered from the various different measurement methodologie s may yield information may be consistent or inconsistent across populations and assessment modalities. Interested reviews are referred to the Limitation and Future Directions section of Paper 2/Chapter 3 for a more detailed discussion of the relationship between objective and subjective measurement of sleep. The present study was unable to confirm the presence of week to week associations between sleep and cognitive functioning in late life. This lack of an observed association may be the result of seve ral different factors, as discussed above. However, future investigations would be well suited to carefully collect both sleep and

PAGE 239

239 cognitive functioning data across time in order to allow for the precision necessary to examine how the preceding night of sl eep may impact next day cognitive functioning. A wealth of knowledge extracted from sleep deprivation studies suggests such a relationship may exist. Lastly, future studies should recruit a larger, more diverse sample of older adults. Obtainment of a larg er sample would allow for the inclusion of additional predictors of late life cognitive functioning. Not only should future researchers collect a larger sample, this sample should also be of increased diversity in terms of both ethnic/cultural background a nd levels of cognition and sleep assessed. The present sample may have limited generalizability due to its over representation with highly educated white females. Examination of the dynamic association between sleep and late life cognitive functioning in l ess educated, non white males is necessary. Summary The current investigation examined the amount of short term fluctuation displayed in the cognitive functioning of older adults. Our results confirm the presence of substantial amounts (i.e., approximately 50% of the amount of between person variability) of within person variability in late life cognitive functioning. We then investigated how TST and TWT were related to late life cognitive functioning. The relationship between sleep and cognitive functionin g was examined at multiple different temporal levels. Average amount of time spent awake during the night was found to be negatively related to processing speed and reasoning performance. No relationships were observed between weekly fluctuations in sleep and weekly cognitive performance.

PAGE 240

240 Figure 4 1. Gantt c hart i llustrating the t iming of s leep and c ognitive d ata c ollection.

PAGE 241

241 Table 4 1 Demographic s tatistics Mean (SD) Age 63.33 (8.49) Education 16.14 (2.25) Gender (% fem ale) 82 -Notes: Age measured in years since birth; Education measured in years; Gender: 1 = male, 2 = female; NAART = premorbid IQ estimate; BDI II = Beck Depression Inventory, 2 nd Edition.

PAGE 242

242 Table 4 2 Interindividual ( b elow diagonal) and i nt raindividual ( a bove d iagonal) c orrelations a mong s leep v ariables and d escriptive s leep d ata. SOL WASO TWAK NWAK SQR TIB TST TWT SE Mean (SD) SOL 1 0.43** + 0.55** + 0.12 0.19 0.43** + 0.11 0.84** + 0.79** + 18.66 ( 15.37 ) WASO 0.09 1 0.32** 0.45** + 0.32 ** 0.20 0.28* 0.72** + 0.73** + 18.02 ( 13.75 ) TWAK 0.06 0.01 1 0.15 0.28* 0.25* 0.28* 0.80** + 0.79** + 18.84 ( 14.35 ) NWAK 0.07 0.37** + 0.02 1 0.22* 0.04 0.07 0.28* 0.19 1.31 ( 0. 75 ) SQR 0.18 0.32** 0.10 0.23* 1 0.12 0.34* 0.33** 0.40** + 3.6 8 ( 0. 53 ) TIB 0.23* 0.22* 0.19 0.11 0.05 1 0.78** + 0.38** + 0.21 49 1.57 ( 5 0.45 ) TST 0.13 0.21 0.12 0.06 0.16 0.79** + 1 0.28* 0.44** + 436.3 3 ( 48.95 ) TWT 0.54** + 0.64** + 0.53** + 0.44** + 0.34** 0.35** 0.28* 1 0.98** + 5 5.35 (33.40) SE 0.49** + 0.6 2* + 0.51** + 0.27* 0.36** + 0.18 0.44** + 0.95** + 1 8 8 96 ( 6.2 5 ) Notes: ** p < .01; p < .05; + indicates correlation would remain significant following Bonferroni correction for multiple comparisons. SOL = sleep onset latency; WASO = wake time after s leep onset; TWAK = terminal wakefulness; NWAK = number of awakenings; SQR = sleep quality rating; TIB = time in bed; TST = total sleep time; TWT = total wake time; SE = sleep efficiency. Sleep measured in minutes.

PAGE 243

243 Figure 4 2. Graphical r e presentation of w eekly and TST and TWT: r aw d ata p lots of s w eekly s leep v alues.

PAGE 244

244 Figure 4 3. Graphical r epresentation of m issing d ata across the s tudy period.

PAGE 245

245 Figure 4 4. Graphical r epresentation of the a mount of i ntra and i nterindividual f luctuation and the p roportion of i ntra to i nterindividual v ariability in c ognitive f unctioning. 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 Symbol Digit Number Copy Letter Series Simple RT Choice RT % Variance Cognitive Measure Different Depictions of Amount of Variability Within Between Proportion

PAGE 246

246 Figure 4 5. Graphical r epresentation of f luctuation in c ognitive f unctioning: r aw d ata p lots of s m ean w eekly p erformance.

PAGE 247

247 Table 4 3 T wo l evel m ultilevel m odel p redicting n umber c opy p erformance. Parameter Symbol Estimate SE Estimate SE Estimate SE Estimate SE Model 1 Model 2 Model 3 Model 4 Fixed Effects Intercept 00 44.38 0.37*** 43.13 0.40*** 51.08 3.23*** 51.78 4.20*** L inear Time 10 0.19 0.03*** 0.18 0.03*** 0.19 0.03*** Quadratic Time 20 0.02 0.005** 0.01 0.005** 0.01 0.005* Age a 01 0.16 0.04*** 0.16 0.04*** Gender b 02 0.70 0.89 0.70 0.87 Education c 03 0.15 0.15 0.09 0.15 T ST person d 04 0.002 0.007 TST week e 30 0.001 0.003 TWT person d 05 0.02 0.01 TWT week e 40 0.004 0.005 Random Effects Residual e it 8.93 0.41*** 6.37 0.32*** 6.43 0.32*** 6.39 0.33*** Intercept r oi 11.14 1. 97*** 11.60 2.08*** 7.72 1.42*** 7.04 1.32*** Linear Time r 1i 0.03 0.01*** 0.03 0.01*** 0.02 0.01** Quadratic Time r 2i 0.001 .0003** 0.001 .0003* 0.001 .0004** TST week e r 3i --g TWT week e r 5i 0.0001 0.0002 Model Fit 2LL ( df ) f 5471.85 (3) 5270.42 (7) 5154.08 (10) 4956.25 (16) Notes: a Measured in years since birth. b 0 = male, 1 = female. c Measured in total years. d variables is person level mean indicator. e indicates variables is weekly, grand mean centered indicator. TST = total sleep time; TWT = total wake time. f 2LL = 2 Log Likelihood (an indicator of model fit). g Variance too small to be estimated. The final Hessian matrix is not positive definite a lthough all convergence criteria are satisfied. The test statistic and confidence interval cannot be computed.*** p < .001; ** p < .01; p < .05, + p < .10.

PAGE 248

248 Table 4 4 T wo l evel m ultilevel m odel p redicting s ymbol d igit p erformance. Parameter Symbol E stimate SE Estimate SE Estimate SE Estimate SE Model 1 Model 2 Model 3 Model 4 Fixed Effects Intercept 00 25.84 0.35*** 24.54 0.38*** 31.42 2.93*** 35.04 3.61*** Linear Time 10 0.20 0.02*** 0.20 0.02*** 0.20 0.02*** Quadratic Ti me 20 0.03 0.005*** 0.02 0.005*** 0.02 0.004*** Age a 01 0.17 0.04*** 0.15 0.03*** Gender b 02 1.40 0.80+ 1.36 0.74+ Education c 03 0.26 0.14+ 0.14 0.13 TST person d 04 0.003 0.006 TST week e 30 0. 002 0.003 TWT person d 05 0.03 0.01*** TWT week e 40 0.01 0.01 Random Effects Residual e it 7.50 0.35*** 4.68 0.23*** 4.72 0.24*** 4.69 0.24*** Intercept r oi 10.04 1.77*** 10.69 1.88*** 6.42 1.16*** 5.17 0.96*** Linear Time r 1i 0.03 0.01*** 0.02 0.01*** 0.02 0.01** Quadratic Time r 2i 0.0008 0.0003** 0.0008 0.0003** 0.0001 0.0003** TST week e r 3i --g TWT week e r 4i 0.0002 0.0002 Model Fit 2LL ( df ) f 5304.93 (3) 5304.93 (3) 4867.79 (10 ) 4668.98 (16) Notes: a Measured in years since birth. b 0 = male, 1 = female. c Measured in total years. d indicates variables is person level mean indicator. e centered indi cator. TST = total sleep time; TWT = total wake time. f 2LL = 2 Log Likelihood (an indicator of model fit). g Variance too small to be estimated. The final Hessian matrix is not positive definite although all convergence criteria are satisfied. The tes t statistic and confidence interval cannot be computed.*** p < .001; ** p < .01; p < .05, + p < .10.

PAGE 249

249 Table 4 5 T wo l evel m ultilevel m odel p redicting l etter s eries p erformance. Parameter Symbol Estimate SE Estimate SE Estimate SE Estimate SE Model 1 Model 2 Model 3 Model 4 Fixed Effects Intercept 00 10.56 0.43*** 8.10 0.39*** 11.15 3.37** 13.89 4.41** Linear Time 10 0.37 0.03*** 0.37 0.03*** 0.38 0.03*** Quadratic Time 20 0.01 0.004** 0.01 0.004** 0.01 0.004** Age a 01 0.17 0.04*** 0.15 0.04** Gender b 02 0.35 0.92 0.40 0.91 Education c 03 0.48 0.16 0.39 0.16* TST person d 04 0.002 0.01 TST week e 30 0.001 0.003 TWT person d 05 0.03 0.01** TWT week e 40 0.005 0.005 Random Effects Residual e it 10.40 0.45*** 5.92 0.28*** 5.82 0.28*** 5.90 0.28*** Intercept r oi 15.07 2.57*** 11.59 1.97*** 8.62 1.54*** 7.92 1.44*** Linear Time r 1i 0.04 0.01*** 0.03 0.01*** 0.03 0.01** Quadratic Time r 2i --g --g --g TST week e r 3i --g TWT week e r 4i --g Model Fit 2LL ( df ) f 5689.70 (3) 5203.40 (7) 5070.48 (10) 4889.01 (16) Notes: a Measured in years since birth. b 0 = male, 1 = female. c Measured in total years. d indicates variables is person level mean indicator. e centered indicator. TST = total sleep time; TWT = total wake time. f 2LL = 2 Log Likelihood (an indica tor of model fit). g Variance too small to be estimated. The final Hessian matrix is not positive definite although all convergence criteria are satisfied. The test statistic and confidence interval cannot be computed.*** p < .001; ** p < .01; p < .05, + p < .10.

PAGE 250

250 Table 4 6 T wo l evel m ultilevel m odel p redicting s imple RT p erformance. Parameter Symbol Estimate SE Estimate SE Estimate SE Estimate SE Model 1 Model 2 Model 3 Model 4 Fixed Effects Intercept 00 502.92 6.26*** 494.41 5.64** 587.36 51.76*** 663.24 68.25*** Linear Time 10 1.38 0.44** 1.32 0.45** 1.21 0.50* Quadratic Time 20 0.25 0.06*** 0.24 0.06*** 0.25 0.06*** Age a 01 1.45 0.63* 0.98 0.65 Gender b 02 12.17 14.16 8.02 13.97 Education c 03 0.05 2.42 1.06 2.40 TST person d 04 0.16 0.12 TST week e 30 0.002 0.04 TWT person d 05 0.31 0.17+ TWT week e 40 0.06 0.07 Random Effects Residual e it 1428.66 65.48*** 1063.15 52.59*** 1047.77 5 2.10*** 1018.91 52.55*** Intercept r oi 3250.35 519.00*** 2416.70 409.03*** 2066.63 361.44*** 1929.56 342.37*** Linear Time r 1i 10.36 2.31*** 10.72 2.37*** 11.92 2.70*** Quadratic Time r 2i 0.05 0.04 0.05 0.04 0.05 0.04 TST week e r 3i --g TWT week e r 4i 0.02 0.04 Model Fit 2LL ( df ) f 10777.70 (3) 10596.03 (7) 10396.50 (10) 9993.55 (16) Notes: a Measured in years since birth. b 0 = male, 1 = female. c Measured in total years. d var iables is person level mean indicator. e centered indicator. TST = total sleep time; TWT = total wake time. f 2LL = 2 Log Likelihood (an indicator of model fit). g Variance too small to be esti mated. The final Hessian matrix is not positive definite although all convergence criteria are satisfied. The test statistic and confidence interval cannot be computed.*** p < .001; ** p < .01; p < .05, + p < .10.

PAGE 251

251 Table 4 7 T wo l evel m ultilevel m ode l p redicting c hoice RT p erformance. Parameter Symbol Estimate SE Estimate SE Estimate SE Estimate SE Model 1 Model 2 Model 3 Model 4 Fixed Effects Intercept 00 385.25 5.66*** 370.84 5.91*** 478.19 53.15*** 528.46 70.31*** Linear Time 10 2.03 0.31*** 1.85 0.34*** 1.70 0.35*** Quadratic Time 20 0.49 0.07*** 0.47 0.07*** 0.50 0.08*** Age a 01 1.86 0.65** 1.58 0.67* Gender b 02 22.78 14.54 20.23 14.43 Education c 03 0.84 2.49 0.28 2.48 TST perso n d 04 0.07 0.12 TST week e 30 0.008 0.04 TWT person d 05 0.33 0.17+ TWT week e 40 0.04 0.06 Random Effects Residual e it 1360.95 62.52*** 793.88 39.68*** 768.95 38.56*** 747.99 37.62*** Intercept r oi 2634. 69 430.52*** 2727.85 449.95*** 2261.80 380.92*** 2136.26 362.01*** Linear Time r 1i 4.72 1.27*** 4.99 1.30*** 4.86 1.38*** Quadratic Time r 2i 0.29 0.07*** 0.25 0.06*** 0.28 0.07*** TST week e r 3i 0.02 0.02 TWT week e r 4i 0 .01 0.02 Model Fit -2LL ( df ) f 10704.17 (3) 10361.05 (7) 10139.31 (10) 9773.71 (16) Notes: a Measured in years since birth. b 0 = male, 1 = female. c Measured in total years. d variables is person level mean indica tor. e centered indicator. TST = total sleep time; TWT = total wake time. f 2LL = 2 Log Likelihood (an indicator of model fit). g Variance too small to be estimated. The final Hessian matrix is not positive definite although all convergence criteria are satisfied. The test statistic and confidence interval cannot be computed.*** p < .001; ** p < .01; p < .05, + p < .10.

PAGE 252

252 CHAPTER 5 GENERAL DISCUSSION The previous Papers/Chapters have successf ully replicated previously reported findings in the cognitive aging literature (i.e., older adults are very capable of demonstrating significant practice related learning, and there is substantial short term fluctuation present in the cognitive performance of older adults), while simultaneously attempting to address unanswered questions regarding late life cognitive plasticity (i.e., practice related learning and short term fluctuations in late life cognitive functioning). We found that both processing spee d measures and an executive processing measure demonstrated significant gains associated with repeated practice, that individuals who gained more in one ability were more likely to gain in another, that both education and state anxiety were associated with that generalized practice related learning demonstrated limited near transfer for processing speed. We also discovered that older adults who spent more time awake during the night on average had lower a verage cognitive performance. Interestingly, the same older adults that spent increased time awake during the night were found to also benefit the most from cognitive practice on a very simple task (i.e., Simple RT). However, no relationships were observed between weekly fluctuations in sleep and cognitive functioning. As a whole, these finding suggest several conclusions and directions for future investigation. First, these findings suggest that older adults are capable of much positive change. A numbe r of scholars have suggested that late life is a time of perpetual loss, absent of gains (Olshansky, Carnes, & Grahn, 1998). Additionally, the utility of continued efforts to extend life into very late 85 years old

PAGE 253

253 and o nward), has come under scrutiny (Baltes & Smith, 2003). However, the present set of studies has shown that older adults are very capable of positive change, even in the complete absence of directed intervention. While very few participants in the current s et suggesting the potential of significant plasticity well into old age. Fu ture investigations should specifically examine practice related learning as a potential avenue for preserving and/or enhancing the cognitive functioning and lives of the oldest old. Practice related learning has been tive late life cognition prior to the onset of a specific tutor based interventio n appears to be demonstrated in tutor based interventions (Ball et al., 2002). Practice is a major component of tutor based interventions. Future investigations shoul d examine whether tutor provided interventions result in improvements above and beyond those found through practice related learning. S elf guided practice is a simpler, more cost effective, and a more widely scalable approach to cognitive interventions tha n traditional tutor based approaches. If tutor based interventions do not significantly add to the gains found through practice alone, monies currently devoted to funding tutor based interventions could be better partitioned to wide scale dissemination of practice learning programs. Such intervention may have to potential to reach numerous elders and influence their quality of life and long term independence (Willis et al., 2006).

PAGE 254

254 A key question in the emerging gerontological cognitive intervention literatu re concerns individual differences in learning potential and the identification of predictors of such differences (Hertzog et al., 2008) Increased time spent awake during the night was associated with increased benefit from cognitive practice. We do not r ecommend focused efforts to disrupt the sleep of elders in an attempt to heighten learning potential. Instead future work should investigate the potential factors common to nocturnal wake time and increased plasticity. Global hyperarousal is a good candida te for further inspection (Altena, Van Der Werf et al., 2008; Bonnet & Arand, 2010). The idea of increasing arousal without impacting night time sleep is daunting; however, this may not be entirely necessary. The nocturnal hyperarousal that has been observ ed in poor sleepers is often accompanied by daytime levels of increased arousal (Bonnet & Arand, 2010). Increased levels of daytime arousal are likely responsible for improved cognitive functioning in individuals with poorer sleep (however, this is an empi rical question worthy of investigation). In fact, previous research has demonstrated that stimulants, like caffeine, may provide cognitive enhancement effects for older adults (Jarvis, 1993; Johnson Kozlow, Kritz Silverstein, Barrett Connor, & Morton, 2002 ; Ryan, Hatfield, & Hofstetter, 2002). Future endeavors to modify arousal levels such that daytime arousal may be heightened and nighttime arousal may be lessened appear timely and to self promote their own cognitive vitality through practice. One intriguing potential avenue to increase daytime arousal and decrease nighttime arousal may be exercise. Exercise has been demonstrated to have beneficial effects on late life cognitive fun ctioning (Colcombe, Kramer, McAuley, Erickson, & Scalf, 2004), has been shown to promote healthy sleep

PAGE 255

255 in late life (King, Oman, Brassington, Bliwise, & Haskell, 1997; King et al., 2008), and has demonstrated effects on physiological daytime arousal (Kamij o et al., 2004). Perhaps exercise may be impacting late life cognitive functioning through alterations in the arousal levels of elders. Such a hypothesis is worthy of direct investigations. Future investigations should examine this possibility through stru ctured assessment of arousal, physical activity, cognitive performance, and sleep characteristics. term fluctuation may be an important indicator of overall health and well being, including cognitive health (Hultsch et al., 2008) Short term fluctuation may serve as a proxy for at any given time point (i.e., may reflect the level of cognitive effort that a person might bring to bear to manage daily living challenges during a given sit uation/on a given day) Following a week of above or below average sleep, elders cognitive functioning was not found to respond systematically. Several potential explanations for the lack of an observed association were hypothesized in Paper 3/Chapter 4 of this document. As sleep remains very malleable in late life (Dzierzewski et al., 2010) and as short term fluctuation in cognitive functioning is associated with worse performance on measures of everyday functioning (Burton et al., 2009; Willis et al., 200 6), continued investigation into the potential dynamic association between sleep and late life cognitive functioning is warranted. Late life sleep treatments likely stabilize sleep. As such, there is potential that sleep treatments may stabilize late life cognitive functioning. To date, there have been no known interventions aimed at promoting consistent cognitive performance. If such research is embarked upon, sleep appears a likely avenue for potential intervention, though future research is needed to con clude such a stance.

PAGE 256

256 Amount of time spent awake has consistently demonstrated reliable improvements following sleep interventions (Dzierzewski et al., 2010). We have shown that older adults who spend less time awake during the night perform better acro ss several cognitive tasks. Whether reduction in time spent awake during the night could improve cognitive functioning remains unknown. A common difficulty in interpreting the observed sleep cognition relationships reported in the current set of investig ations is the fact that the present sample was comprised of older adults of unknown sleep complaint status. However, as sleep tends to display age correlated negative changes (Morgan, 2000), the present set of studies likely included a heterogeneous sample of older adults with varying degrees of sleep complaint (i.e., from no complaint to moderate complain to severe complaint). Further, given the above noted age correlated negative changes in sleep, there would likely be room for improvements in the sleep o f even non unwanted awake time initiating sleep and unwanted awake time during the night (Morgan, 2000). Additionally, complaint status may be unrelated to actual quantitative sleep characteristics in older adults (McCrea et al., 2003). While theoretically possible, no know research has attempted to intervene in the sleep of non complaining elders. Such an attempt would prove highly original, and would be consist with the positi ve psychology movement aimed at optimization of behavior and functioning rather than mere remediation of problem behavior. Practice related learning of elders completing a healthy sleep program could be compared to elders not completing such a program. D ifferences in the magnitude or slope of practice related learning would suggest a role of modifiable sleep.

PAGE 257

257 In general, this dissertation examined the plastic nature of late life cognitive functioning. It was found that older adults are very capable of en gendering improvements in their cognitive functioning through self administered practice. Further, would appear primed for further investigation as a potential target towards the life cognitive functioning should be conducted to further explicate the complicated association between these important quality of life indicators.

PAGE 258

258 LIST OF REFERENCES Aiken Morgan, A. (2008). Effects of improved physical fitness on cognitive/psychosocial functioning in community dwelling, sedentary middle aged and older adults. Unpublished doctoral Dissertation, University of Florida, Gainesville, Florida. Al laire, J. C., & Marsiske, M. (2005). Intraindividual Variability May Not Always Indicate Vulnerability in Elders' Cognitive Performance. Psychology and Aging, 20 (3), 390 401. Altena, E., Van Der Werf, Y. D., Strijers, R. L. M., & Van Someren, E. J. W. (200 8). Sleep loss affects vigilance: effects of chronic insomnia and sleep therapy. Journal of Sleep Research, 17 (3), 335 343. Altena, E., Werf, Van Der Werf, Y. D., Sanz Arigita, E. J., Voorn, T. A., Rombouts, S. A. R. B., Kuijer, J. P. A., et al. (2008). Pr efrontal Hypoactivation and Recovery in Insomnia. Sleep, 31 (9), 1271 1276. Ancoli Israel, S., Clopton, P., Klauber, M. R., Fell, R., & Mason, W. (1997). Use of wrist activity for monitoring sleep/wake in demented nursing home patients. Sleep, 20 (1), 24 27. Ancoli Isreal, A., Cole, R., Alessi, C., Chambers, M., Moorcroft, W., & Pollak, C. P. (2003). The role of actigraphy in the study of sleep and circadian rhythms. SLEEP, 26 (3), 342 392. Arbuckle, J. L., & Wothke, W. (1999). Amos user's guide Chicago, IL: Smallwaters Corporation. Ball, K., Berch, D. B., Helmers, K. F., Jobe, J. B., Leveck, M. D., Marsiske, M., et al. (2002). Effects of cognitive training interventions with older adults: A randomized controlled trial. JAMA: Journal of the American Medical As sociation, 288 (18), 2271 2281. Baltes, P. B., Dittmann Kohli, F., & Kliegl, R. (1986). Reserve capacity of the elderly in aging sensitive tests of fluid intelligence: Replication and extension. Psychology and Aging, 1 (2), 172 177. Baltes, P. B., Kliegl, R. & Dittmann Kohli, F. (1988). On the Locus of Training Gains in Research on the Plasticity of Fluid Intelligence in Old Age. Journal of Educational Psychology, 80 (3), 392 400. Baltes, P. B., & Smith, J. (2003). New Frontiers in the Future of Aging: From S uccessful Aging of the Young Old to the Dilemmas of the Fourth Age. Gerontology, 49 (2), 123 135.

PAGE 259

259 Baltes, P. B., Sowarka, D., & Kliegl, R. (1989). Cognitive Training Research on Fluid Intelligence in Old Age: What Can Older Adults Achieve by Themselves? Psy chology and Aging, 4 (2), 217 221. Baltes, P. B., & Willis, S. L. (1982). Enhancement (plasticity) of intellectual functioning in old age: Penn State's Adult Development and Enrichment Project (ADEPT). In F. M. L. Craik & S. E. Trehub (Eds.), Aging and cong itive processes (pp. 353 389). New York: Plenum Press. Banks, S., & Dinges, D. F. (2007). Behavioral and physiological consequences of sleep restriction. Journal of Clinical Sleep Medicine, 3 (5), 519 528. Barrowclough, C., King, P., Colville, J., Russell, E., Burns, A., & Tarrier, N. (2001). A Randomized Trial of the Effectiveness of Cognitive -Behavioral Therapy and Supportive Counseling for Anxiety Symptoms in Older Adults. Journal of Consulting and Clinical Psychology, 69 (5), 756 762. Bastien, C. H., For tier Brochu, E., Rioux, I., LeBlanc, M., Daley, M., & Morin, C. M. (2003). Cognitive performance and sleep quality in the elderly suffering from chronic insomnia: Relationship between objective and subjective measures. Journal of psychosomatic research, 54 (1), 39 49. Baune, B. T., Ponath, G., Golledge, J., Varga, G., Arolt, V., Rothermundt, M., et al. (2008). Association between IL 8 cytokine and cognitive performance Study. Neurobiology of Aging, 29 (6), 937 944. Beaudreau, S A., & O'Hara, R. (2009). The association of anxiety and depressive symptoms with cognitive performance in community dwelling older adults. Psychology and Aging, 24 (2), 507 512. Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Beck Depression Inventory Second Edition San Antonio, TX: Psychological Corporation. Beebe, D. W., & Gozal, D. (2002). Obstructive sleep apnea and the prefrontal cortex: towards a comprehensive model linking nocturnal upper airway obstruction to daytime cognitive and behavioral d eficits. Journal of Sleep Research, 11 (1), 1 16. Benton, A., & Hamsher, K. (1989). Multilingual Aphasia Examination Iowa City, IA: AJA Associates. Binks, P. G., Waters, W. F., & Hurry, M. (1999). Short term total sleep deprivation does not selectively imp air higher cortical funcitoning. SLEEP, 22 (3), 328 334.

PAGE 260

260 Bixler, E. O., Vgontzas, A. N., Lin, H., Ten Have, T., Rein, J., Vela Bueno, A., et al. (2001). Prevalence of Sleep disordered Breathing in Women Effects of Gender. Am. J. Respir. Crit. Care Med., 1 63 (3), 608 613. Bixler, E. O., Vgontzas, A. N., Ten Have, T., Tyson, K., & Kales, A. (1998). Effects of Age on Sleep Apnea in Men I. Prevalence and Severity. Am. J. Respir. Crit. Care Med., 157 (1), 144 148. Blackwell, T., Yaffe, K., Ancoli Israel, S., Sc hneider, J. L., Cauley, J. A., Hillier, T. A., et al. (2006). Poor Sleep Is Associated With Impaired Cognitive Function in Older Women: The Study of Osteoporotic Fractures. Journals of Gerontology: Series A: Biological Sciences and Medical Sciences, 61 (4), 405 410. Blagrove, M. (1996). Effects of Length of Sleep Deprivation on Interrogative Suggestibility. Journal of Experimental Psychology: Applied, 2 (1), 48 59. Blagrove, M., Alexander, C., & Horne, J. A. (1995). The effects of chronic sleep reduction on t he performance of cognitive tasks sensitive to sleep deprivation. Applied Cognitive Psychology, 9 (1), 21 40. Blair, J. R., & Spreen, O. (1989). Predicting pre morbid IQ: A revision of the National Adult Reading Test. The Clinical Neuropsychologist, 3 129 136. Bliese, P. D., Wesensten, N. J., & Balkin, T. J. (2006). Age and individual variability in performance during sleep restriction. Journal of Sleep Research, 15 (4), 376 385. Blieszner, R., Willis, S. L., & Baltes, P. B. (1981). Training research in agin g on the fluid ability of inductive reasoning. Journal of Applied Developmental Psychology, 2 (3), 247 265. Bonnet, M. H., & Arand, D. L. (2010). Hyperarousal and insomnia: State of the science. Sleep Medicine Reviews, 14 (1), 9 15. Boonstra, T., Stins, J., Daffertshofer, A., & Beek, P. (2007). Effects of sleep deprivation on neural functioning: an integrative review. Cellular and Molecular Life Sciences, 64 (7), 934 946. Brandt, J., Spencer, M., & Folstein, M. (1988). The Telephone Interview for Cognitive Sta tus. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, 1 111 117. Bruce, A. S., & Aloia, M. S. (2006). Sleep and cognition in older adults. Sleep Medicine Clinics, 1 207 220. Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods Thousand Oaks: Sage.

PAGE 261

261 Buman, M. P. (2008). Evaluation of a peer assisted social cognitive physical activity intervention for older adults. Unpublished doctoral Dissertation, University of Florida, Gainesville, Flori da. Buman, M. P., Giacobbi, P. R., Dzierzewski, J. M., Aiken Morgan, A., McCrae, C. S., Roberts, B. L., et al. (2011). Peer volunteers improve long term maintenance of physical activity with older adults: A randomized controlled trial. Journal of Physical Activity and Health, 8 (Suppl 2), S257 S266. Bunce, D., MacDonald, S. W. S., & Hultsch, D. F. (2004). Inconsistency in serial choice decision and motor reaction times dissociate in younger and older adults. Brain and Cognition, 56 (3), 320 327. Burton, C. L. Hultsch, D. F., Strauss, E., & Hunter, M. A. (2002). Intraindividual variability in physical and emotional functioning: Comparison of adults with traumatic brain injuries and healthy adults. The Clinical Neuropsychologist, 16 (3), 264 279. Burton, C. L., Strauss, E., Hultsch, D. F., & Hunter, M. A. (2009). The relationship between everyday problem solving and inconsistency in reaction time in older adults. Aging, Neuropsychology, and Cognition, 16 (5), 607 632. Buysse, D. J., Cheng, Y., Germain, A., Moul, D E., Franzen, P. L., Fletcher, M., et al. (2010). Night to night sleep variability in older adults with and without chronic insomnia. Sleep Medicine, 11 (1), 56 64. Cabeza, R., & Nyberg, L. (2000). Neural bases of learning and memory: functional neuroimagi ng evidence. Current Opinion in Neurology, 13 (4), 415 421. Cain, S. W., Silva, E. J., Chang, A., Ronda, J. M., & Duffy, J. F. (2011). One night of sleep deprivation affects reaction time, but not interference or facilitation in a Stroop task. Brain and Cog nition, 76 (1), 37 42. Christensen, H., Mackinnon, A., Jorm, A. F., Korten, A., Jacomb, P., Hofer, S. M., et al. (2004). The Canberra longitudinal study: Design, aims, methodology, outcomes and recent empirical investigations. Aging, Neuropsychology, and Co gnition, 11 (2 3), 169 195. Cohen, J. (1992). A Power Primer. Psychological Bulletin, 112 155 159. Cohen, J. D., Forman, S. D., Braver, T. S., Casey, B. J., Servan Schreiber, D., & Noll, D. C. (1994). Activation of prefrontal cortex in a non spatial workin g memory task with functional MRI. Human Brain Map, 1 293 304. Colcombe, S., Kramer, A., McAuley, E., Erickson, K., & Scalf, P. (2004). Neurocognitive aging and cardiovascular fitness. Journal of Molecular Neuroscience, 24 (1), 9 14.

PAGE 262

262 Cole, R. J., Kripke, D F., Gruen, W., Mullaney, D. J., & Gillin, J. C. (1992). Automatic sleep/wake identification from wrist actigraphy. SLEEP, 15 (5), 461 469. Crenshaw, M. C., & Edinger, J. D. (1999). Slow Wave Sleep and Waking Cognitive Performance Among Older Adults With a nd Without Insomnia Complaints. Physiology & Behavior, 66 (3), 485 492. Cricco, M., Simonsick, E. M., & Foley, D. J. (2001). The impact of insomnia on cognitive functioning in older adults. Journal of the American Geriatrics Society, 49 (9), 1185 1189. Dinge s, D. F., Pack, F., Williams, K., Gillen, N. A., Powell, J W., Ott, G E., Aptowicz, C., & Pack, A. I. (1997). Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4 5 hours per night Sleep, 20 267 277. Dittmann Kohli, F., Lachman, M. E., Kliegl, R., & Baltes, P. B. (1991). Effects of Cognitive Training and Testing on Intellectual Efficacy Beliefs in Elderly Adults. Journal of Gerontology, 46 (4), P162 P164. Drake, C. L., Roehrs, T. A ., Burduvali, E., Bonahoom, A., Rosekind, M., & Roth, T. (2001). Effects of rapid versus slow accumulation of eight hours of sleep loss. Psychophysiology, 38 979 987. Drummond, S. P. A., & Brown, G. G. (2001). The effects of total sleep deprivation on cer ebral responses to cognitive performance. Neuropshopharmacology, 25 68 73. Drummond, S. P. A., Brown, G. G., Gillin, J. C., Stricker, J. L., Wong, E. C., & Buxton, R. B. (2000). Altered brain response to verbal learning following sleep deprivation. Nature 403 (6770), 655 657. Drummond, S. P. A., Brown, G. G., Stricker, J. L., Buxton, R. B., Wong, E. C., & Gillin, J. C. (1999). Sleep deprivation induced reduction in cortical functional response to serial subtraction. NeuroReport, 10 (18), 3745 3748. Drummond S. P. A., Gillin, J. C., & Brown, G. G. (2001). Increased cerebral response during a divided attention task following sleep deprivation. Journal of Sleep Research, 10 (2), 85 92. Drummond, S. P. A., Meloy, M. J., Yanagi, M. A., Orff, H. J., & Brown, G. G. (2005). Compensatory recruitment after sleep deprivation and the relationship with performance. Psychiatry Research, 140 (3), 211 223. Duncan, T. E., Duncan, S. C., & Strycker, L. A. (2006). An Introduction to Latent Varaible Growth Curve Modeling: Concept s, Issues, and Applications (Second ed.). Mahwah, NJ: Lawrence Erlbaum Associates.

PAGE 263

263 Durmer, J. S., & Dinges, D. F. (2005). Neurocognitive Consequences of Sleep Deprivation. Semin Neurol, 25 (01), 117 129. Dzierzewski, J. M. (2007). Intraindividual variabilit y in objective and subjective sleep and cognitive perfromance in older adults with insomnia. Unpublished Masters thesis University of Florida, Gainesville, FL. Dzierzewski, J. M., McCrae, C. S., Rowe, M., Marsiske, M., McCoy, K., McNamara, J., et al. (200 8). Nightly sleep patterns in older aults: Night to night fluctuations in the sleep if older good noncomplaining, good complaining, poor noncomplaining, poor complaining/insomnia, and caregivers. SLEEP, 31 A104. & McCrae, C. S. (2010). Tackling sleeplessness: Psychological treatment options for insomnia in older adults. Nature and Science of Sleep, 2 47 61. Edinger, J. D., Fins, A. I., Sullivan, R. J., Jr., Marsh, G. R., Dailey, D. S., Hope, T. V., et al. (1997 ). Sleep in the laboratory and sleep at home: Comparisons of older insomniacs and normal sleepers. Sleep, 20 1119 1126. Edinger, J. D., Marsh, G. R., Mccall, W. V., Erwin, C. W., & Lininger, A. W. (1991). Sleep variability across consecutive nights of hom e monitoring in older mixed DIMS patients. Sleep, 14 (13 17). Ellenbogen, J. M. (2005). Cognitive benefits of sleep and their loss due to sleep deprivation. Neurology, 64 (7), E25 27. Elliman, N. A., Green, M. W., Rogers, P. J., & Finch, G. M. (1997). Proces sing efficiency theory and the working memory system: Impairments associated with sub clinical anxiety. Personality and Individual Differences, 23 (1), 31 35. Endeshaw, Y. W., White, W. B., Kutner, M., Ouslander, J. G., & Bliwise, D. L. (2009). Sleep Disord ered Breathing and 24 Hour Blood Pressure Pattern Among Older Adults. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 64A (2), 280 285. Espie, C. A. (1991). The psychological treatment of insomnia Chichesster, UK: J. Wiley a nd Sons. Eysenck, M. W., & Calvo, M. G. (1992). Anxiety and performance: The processing efficiency theory. Cognition and Emotion, 6 (6), 409 434. Eysenck, M. W., Derakshan, N., Santos, R., & Calvo, M. G. (2007). Anxiety and cognitive performance: Attentiona l control theory. Emotion, 7 (2), 336 353.

PAGE 264

264 Faubel, R., Lopez Garcia, E., Guallar Castillon, P., Graciani, A., Banegas, J. R., & Rodriguez Artalejo, F. (2009). Usual sleep duration and cognitive function in older adults in Spain. Journal of Sleep Research, 1 8 (4), 427 435. Ferrara, M., & De Gennaro, L. (2001). How much sleep do we need? Sleep medicine reviews, 5 (2), 155 179. Field, A. (2005). Discovering statistics using SPSS (2nd ed.) Thousand Oaks, CA US: Sage Publications, Inc. Findley, L. J., Barth, J. T. Powers, D. C., Wilhoit, S. C., Boyd, D. G., & Suratt, P. M. (1986). Cognitive impairment in patients with obstructive sleep apnea and associated hypoxemia. Chest, 90 (5), 686 690. Frankel, B. L., Coursey, R. D., Buchbinder, R., & Snyder, F. (1976). Record ed and reported sleep in chronic primary insomnia. Archives of General Psychiatry, 33 615 623. Gamaldo, A. A., Allaire, J. C., & Whitfield, K. E. (2010). Exploring the Within Person Coupling of Sleep and Cognition in Older African Americans. Psychology an d Aging, 25 (4), 851 857. Gamaldo, A. A., Weatherbee, S. R., & Allaire, J. C. (2008). Exploring the within person coupling of blood pressure and cognition in elders. 63 P386 389. Goel, N., Rao, H., Durmer, J. S., & Dinges, D. F. (2009). Neurocognitive cons equences of sleep deprivation. Semin Neurol, 29 (4), 320 339. Gorsuch, R. L. (1983). Factor analysis (Second ed.). Hillsdale, NJ: Lawrence Earlbaum Associates. Graham, J. W. (2009). Missing data analysis: Making it work in the real world. Annual Review of P sychology, 60 549 576. Haensel, A., Bardwell, W., Mills, P., Loredo, J., Ancoli Israel, S., Morgan, E., et al. (2009). Relationship between inflammation and cognitive function in obstructive sleep apnea. Sleep and Breathing, 13 (1), 35 41. Harrison, Y., & Horne, J. (1998). Sleep loss impairs short and novel language tasks having a prefrontal focus. Journal of Sleep Research, 7 (2), 95 100. Harrison, Y., & Horne, J. A. (1997). Sleep deprivation affects speech. SLEEP 20 (10), 871 877. Harrison, Y., & Horne, J. A. (1999). One Night of Sleep Loss Impairs Innovative Thinking and Flexible Decision Making. Organizational Behavior and Human Decision Processes, 78 (2), 128 145.

PAGE 265

26 5 Harrison, Y., & Horne, J. A. (2000). The impact of sleep deprivation on decision making: A r eview. Journal of Experimental Psychology: Applied, 6 (3), 236 249. Harrison, Y., Horne, J. A., & Rothwell, A. (2000). Prefrontal neuropsychological effects of sleep deprivation in young adults -a model for healthy aging? Sleep, 23 (8), 1067 1073. Hastings, C., Mosteller, F., Tukey, J. W., & Winsor, C. P. (1947). Low moments for small samples: A comparative study of order statistics. Annals of Mathematical Statistics 18 413 426. Hauri, P. J., & Wisbey, J. (1992). Actigraphy and insomnia: A closer look. Sleep 15 293 301. Hertzog, C., Kramer, A. F., Wilson, R. S., & Lindenberger, U. (2008). Enrichment effects on adult cognitive development: Can the functional capacity of older adults be preserved and enhanced? Psychological Science in the Public Interest, 9 (1 ), 1 65. Hobson, J. A., & Pace Schott, E. F. (2002). The cognitive neuroscience of sleep: neuronal systems, consciousness and learning. Nat Rev Neurosci, 3 (9), 679 693. Hofer, S. M., Christensen, H., Mackinnon, A. J., Korten, A. E., Jorm, A. F., Henderson, A. S., et al. (2002). Change in Cognitive Functioning Associated With ApoE Genotype in a Community Sample of Older Adults. Psychology and Aging, 17 (2), 194 208. Hofland, B. F., Willis, S. L., & Baltes, P. B. (1981). Fluid intelligence performance in the e lderly: Intraindividual variability and conditions of assessment. Journal of Educational Psychology, 73 (4), 573 586. Horne, J. A. (1988). Sleep loss and 'divergent' thinking ability. Sleep, 11 528 536. Hultsch, D. F., MacDonald, S. W. S., & Dixon, R. A. ( 2002). Variability in reaction time performance of younger and older adults. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 57 (2), P101 PP115. Hultsch, D. F., MacDonald, S. W. S., Hunter, M. A., Levy Bencheton, J., & Str auss, E. (2000). Intraindividual variability in cognitive performance in older adults: Comparison of adults with mild dementia, adults with arthritis, and healthy adults. Neuropsychology, 14 (4), 588 598.

PAGE 266

266 Hultsch, D. F., Strauss, E., Hunter, M. A., & MacDon ald, S. W. S. (2008). Intraindividual variability, cognition, and aging. In F. I. M. Craik & T. A. Salthouse (Eds.), The handbook of aging and cognition (3rd ed.). (pp. 491 556). New York, NY, US: Psychology Press. Jaeggi, S. M., Buschkuehl, M., Jonides, J ., & Perrig, W. J. (2008). Improving fluid intelligence with training on working memory. Proceedings of the National Academy of Sciences, 105 (19), 6829 6833. Jarvis, B. G. (2008). DirectRT (Version 2008). New York, NY: Empirisoft. Jarvis, B. G. (2008). Med iaLab (Version 2008). New York, NY: Empirisoft Corporation. Jarvis, M. (1993). Does caffeine intake enhance absolute levels of cognitive performance? Psychopharmacology, 110 (1), 45 52. Jean Louis, G., von Gizycki, H., Zizi, F., Fookson, J., Spielman, A., N unes, J., et al. (1996). Determination of sleep and wakefulness with the actigraph data analysis software (ADAS). SLEEP, 19 (9), 739 743. Jelicic, M., Bosma, H., Ponds, R., Van Boxtel, M. P. J., Houx, P. J., & Jolles, J. (2002). Subjective sleep problems in later life as predictors of cognitive decline. Report from the Maastricht Ageing Study (MAAS). International Journal of Geriatric Psychiatry, 17 (1), 73 77. Jennings, J. M., & Jacoby, L. L. (2003). Improving memory in older adults: Training recollection. N europsychological Rehabilitation: An International Journal, 13 (4), 417 440. Johnson Kozlow, M., Kritz Silverstein, D., Barrett Connor, E., & Morton, D. (2002). Coffee Consumption and Cognitive Function among Older Adults. American Journal of Epidemiology 156 (9), 842 850. Jones, K., & Harrison, Y. (2001). Frontal lobe function, sleep loss and fragmented sleep. Sleep medicine reviews, 5 (6), 463 475. Jordan, A., & Doug McEvoy, R. (2003). Gender differences in sleep apnea: epidemiology, clinical presentation and pathogenic mechanisms. Sleep medicine reviews, 7 (5), 377 389. Kamijo, K., Nishihira, Y., Hatta, A., Kaneda, T., Kida, T., Higashiura, T., et al. (2004). Changes in arousal level by differential exercise intensity. Clinical Neurophysiology, 115 (12), 26 93 2698.

PAGE 267

267 King, A. C., Oman, R. F., Brassington, G. S., Bliwise, D. L., & Haskell, W. L. (1997). Moderate Intensity Exercise and Self rated Quality of Sleep in Older Adults. JAMA: The Journal of the American Medical Association, 277 (1), 32 37. King, A. C., Pruitt, L. A., Woo, S., Castro, C. M., Ahn, D. K., Vitiello, M. V., et al. (2008). Effects of Moderate Intensity Exercise on Polysomnographic and Subjective Sleep Quality in Older Adults With Mild to Moderate Sleep Complaints. The Journals of Gerontology S eries A: Biological Sciences and Medical Sciences, 63 (9), 997 1004. Kronholm, E., Mikael, S., Suutama, T., Sulkava, R., Pertti, E., & Partonen, T. (2009). Self reported sleep duration and cognitive functioning in the general population. Journal of Sleep Re search, 18 436 446. Krueger, G. P. (1989). Sustained work, fatigue, sleep loss and performance: A review of the issues. Work & Stress, 3 (2), 129 141. Krueger, J. M., Obl, F., Fang, J., Kubota, T., & Taishi, P. (2001). The Role of Cytokines in Physiolog ical Sleep Regulation. Annals of the New York Academy of Sciences, 933 (1), 211 221. Krueger, J. M., Rector, D. M., Churchill, L., Phelps, & Korneva. (2008). Cytokines and Sleep. In NeuroImmune Biology (Vol. Volume 6, pp. 213 240): Elsevier. LaPlante, M. P., Harrington, C., & Kang, T. (2002). Estimating Paid and Unpaid Hours of Personal Assistance Services in Activities of Daily Living Provided to Adults Living at Home. Health Services Research, 37 (2), 397 415. Li, S. C., Huxhold, O., & Schmie dek, F. (2004). Aging and Attenuated Processing Robustness. Gerontology, 50 (1), 28 34. Lichstein, K. L., Riedel, B. W., & Means, M. K. (1999). Psychological treatment of late life insomnia. In R. Schulz, Maddox, G., & Lawton, M. P. (Ed.), Annual review of gerontology and geriatrics: Vol. 18. Focus on interventions research with older adults New York: Springer. Lim, J., & Dinges, D. F. (2010). A Meta Analysis of the Impact of Short Term Sleep Deprivation on Cognitive Variables. Psychological Bulletin, 136 ( 3), 375 389. MacDonald, S. W. S., Hultsch, D. F., & Dixon, R. A. (2008). Predicting impending death: Inconsistency in speed is a selective and early marker. Psychology and Aging, 23 (3), 595 607.

PAGE 268

268 MacDonald, S. W. S., Nyberg, L., Sandblom, J., Fischer, H., & Backman, L. (2008). Increased response time variability is associated with reduced inferior parietal activation during episodic recognition in aging. Journal of Cognitive Neuroscience, 20 (5), 779 786. Mahncke, H. W., Connor, B. B., Appelman, J., Ahsanuddi n, O. N., Hardy, J. L., Wood, R. A., et al. (2006). Memory enhancement in healthy older adults using a brain plasticity based training program: A randomized, controlled study. Proceedings of the National Academy of Sciences, 103 (33), 12523 12528. Maquet, P (2001). The Role of Sleep in Learning and Memory. Science, 294 (5544), 1048 1052. Martin, J. L., & Ancoli Israel, S. (2002). Assessment and diagnosis of insomnia in non pharmacological intervention studies. Sleep medicine reviews, 6 (5), 379 406. Mathieu, A., Mazza, S., Decary, A., Massicotte Marquez, J., Petit, D., Gosselin, N., et al. (2008). Effects of obstructive sleep apnea on cognitive function: A comparison between younger and older OSAS patients. Sleep medicine, 9 (2), 112 120. May, J., & Kline, P. ( 1987). Measuring the effects upon cognitive abilities of sleep loss during continuous operations. British Journal of Psychology, 78 (4), 443. McAfoose, J., & Baune, B. T. (2009). Evidence for a cytokine model of cognitive function. Neuroscience and biobehav ioral reviews, 33 (3), 355 366. McArdle, J. J., & Prindle, J. J. (2008). A latent change score analysis of a randomized clinical trial in reasoning training. Psychology and Aging, 23 (4), 702 719. McAuley, E., Kramer, A. F., & Colcombe, S. J. (2004). Cardiov ascular fitness and neurocognitive function in older Adults: A brief review. Brain, Behavior, and Immunity, 18 (3), 214 220. McCall, C., & McCall, W. V. (2010). Comparison of actigraphy with polysomnography and sleep logs in depressed insomniacs. Journal of Sleep Research no no. McCoy, K. J. M. (2004). Understanding the transition from normal cognitive aging to mild cognitive impairment: Comparing the intraindividual variability in cognitive function. Unpublished doctoral Dissertation, University of Florida Gainesville, Florida.

PAGE 269

269 McCrae, C. S., McNamara, J. P. H., Rowe, M. A., Dzierzewski, J. M., Dirk, J., Marsiske, M., et al. (2008). Sleep and affect in older adults: Using multilevel modeling to examine daily associations. Journal of Sleep Research, 17 (1), 42 53. McCrae, C. S., Rowe, M. A., Tierney, C. G., Dautovich, N. D., DeFinis, A. L., & McNamara, J. P. H. (2005). Sleep complaints, subjective and objective sleep patterns, health, psychological adjustment, and daytime functioning in community dwelling old er adults. Journal of Gerontology (60). McCrea, C. S., Wilson, N. M., Lichstein, K. L., Durrence, H. H., Taylor, D. J., without insomnia complaints. Journal of Psychosomatic Resear ch, 54 11 19. Morgan, K. (2000). Sleep and Aging. In K. Lichstein & C. Morin (Eds.), Treatment of late life insomnia. Thousand Oaks, CA: Sage Publications. Nebes, R. D., Buysse, D. J., Halligan, E. M., Houck, P. R., & Monk, T. H. (2009). Self Reported Sle ep Quality Predicts Poor Cognitive Performance in Healthy Older Adults. Journals of Gerontology Series B Psychological Sciences and Social Sciences, 64 (2), 180 187. Nesselroade, J. R. (1991). The warp and woof of the developmental fabric. In R. Downs, L. L iben & D. Palermo (Eds.), Visions of development, the environment, and aesthetics: The legacy of Joachim F. Wohlwill (pp. 213 240). Hillsdale, NJ: Erlbaum. Nesselroade, J. R., & Salthouse, T. A. (2004). Methodological and theoretical implications of intrai ndividual variability in perceptual motor performance. The Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 59 (2), P49 PP55. Neupert, S. D., Almeida, D. M., Mroczek, D. K., & Spiro, A., III. (2006). Daily stressors and memory failures in a naturalistic setting: Findings from the va normative aging study. Psychology and Aging, 21 (2), 424 429. O'Brien, L. M. (2009). The Neurocognitive Effects of Sleep Disruption in Children and Adolescents. Child and Adolescent Psychiatric Clinic s of North America, 18 (4), 813 823. Olshansky, S. J., Carnes, B. A., & Grahn, D. (1998). Confronting the boundaries of human longevity. American Scientist, 86 52 61. Oosterman, J. M., Van Someren, E. J. W., Vogels, R. L. C., Van Harten, B., & Scherder, E. J. A. (2009). Fragmentation of the rest activity rhythm correlates with age related cognitive deficits. Journal of Sleep Research, 18 (1), 129 135.

PAGE 270

270 Orff, H. J., Drummond, S. P., Nowakowski, S., & Perlis, M. L. (2007). Discrepancy between subjective symptom atology and objective neuropsychological performance in insomnia. SLEEP, 30 (9), 1205 1211. Owen, A. M., Hampshire, A., Grahn, J. A., Stenton, R., Dajani, S., Burns, A. S., et al. (2010). Putting brain training to the test. Nature, 465 (7299), 775 778. Pace Schott, E. F., Hutcherson, C. A., Bemporad, B., Morgan, A., Kumar, A., Hobson, J. A., et al. (2009). Failure to Find Executive Function Deficits Following One Night's Total Sleep Deprivation in University Students Under Naturalistic Conditions. Behavioral Sleep Medicine, 7 (3), 136 163. Pallesen, S., Nordhus, I. H., & Kvale, G. (1998). Nonpharmacological interventions for insomnia in older adults: A meta analysis of treatment efficacy. Psychotherapy: Theory/Research/Practice/Training, 35 (4), 472 482. Park, D. C., Smith, A. D., Lautenschlager, G., Earles, J. L., Frieske, D., Zwahr, M., et al. (1996). Mediators of long term memory performance across the life span. Psychology and Aging, 11 (4), 621 637. Patiau, C. (1998). Modifications of frontal temporal conne ctivity during a verb generation task after 30 hr total sleep deprivation. A PET study. Journal of Sleep Research, 7 (Supplemental 2), 208. Peigneux, P., Laureys, S., Delbeuck, X., & Maquet, P. (2001). Sleeping brain, learning brain. The role of sleep for m emory systems. NeuroReport, 12 (18), A111 A124. Physical Activity Guidelines Advisory Committee Report, Part A: Executive Summary 114 120 (2009). Pilcher, J. J., Band, D., Odle Dusseau, H. N., & Muth, E. R. (2007). Human Performance Under Sustained Operat ions and Acute Sleep Deprivation Conditions: Toward a Model of Controlled Attention. Aviation, Space, and Environmental Medicine, 78 B15 B24. Pilcher, J. J., & Huffcutt, A. J. (1996). Effects of sleep deprivation on performance: A meta analysis. Sleep: Jo urnal of Sleep Research & Sleep Medicine, 19 (4), 318 326. Portas, C. M., Rees, G., Howseman, A. M., Josephs, O., Turner, R., & Frith, C. D. (1998). A specific role for the thalamus on mediating the interaction of attention and arousal in humans. Neuroscien ce, 18 (21), 8979 8989. Reitan, R. M. (1992). Trail Making Test: Manual for Administration and Scoring Tucson, AZ: Reitan Neuropsychology Lab.

PAGE 271

271 Richardson, G. S. (2007). Human physiological models of insomnia. SLEEP, 8 (4), 9 14. Rogers, A. E., Caruso, C. C. & Aldrich, M. S. (1993). Reliability of sleep diaries for assessment of sleep/wake parameters. Nursing Research, 42 (6), 368 372. Ryan, L., Hatfield, C., & Hofstetter, M. (2002). Caffeine Reduces Time of Day Effects on Memory Performance in Older Adults. Psychological Science, 13 (1), 68 71. Salomon, G., & Perkins, D. N. (1989). Rocky Roads to Transfer: Rethinking Mechanism of a Neglected Phenomenon. Educational Psychologist, 24 (2), 113 142. Salthouse, T. A. (1991). Theoretical perspectives on cognitive a ging. Hillsdale, NJ: Erlbaum. Salthouse, T. A. (2004). What and When of Cognitive Aging. Current Directions in Psychological Science, 13 (4), 140 144. Salthouse, T. A. (2010). Influence of Age on Practice Effects in Longitudinal Neurocognitive Change. Neuro psychology, 24 (5), 563 572. Salthouse, T. A., Nesselroade, J. R., & Berish, D. E. (2006). Short term variability in cognitive performance and the calibration of longitudinal change. The Journals of Gerontology: Series B: Psychological Sciences and Social S ciences, 61 (3), P144 PP151. Salthouse, T. A., Schroeder, D. H., & Ferrer, E. (2004). Estimating Retest Effects in Longitudinal Assessments of Cognitive Functioning in Adults Between 18 and 60 Years of Age. Developmental Psychology, 40 (5), 813 822. Sanchez Ortuno, M. M., Edinger, J. D., Means, M. K., & Almirall, D. (2010). Home is where sleep is: an ecological approach to test the validity of actigraphy for the assessment of insomnia. Journal of clinical sleep medicine : JCSM : official publication of the Am erican Academy of Sleep Medicine, 6 (1), 21 29. Schaie, K. W. (1994). The Course of Adult Intellectual Development. American Psychologist, 49 (4), 304 313. Schaie, K. W. (1996). Intellectual development in adulthood. In J. E. Birren & K. W. Schaie (Eds.), Handbook of psychology and aging. (4th ed., pp. 266 286). San Diego: Academic Press.

PAGE 272

272 Schaie, K. W., Willis, S. L., & Caskie, G. I. L. (2004). The Seattle longitudinal study: Relationship between personality and cognition. Aging, Neuropsychology, and Cognit ion, 11 (2), 304 324. Segal, D. L., Coolidge, F. L., Cahill, B. S., & O'Riley, A. A. (2008). Psychometric properties of the Beck Depression Inventory II (BDI II) among community dwelling older adults. Behavior Modification, 32 (1), 3 20. Silva, A. P., Moreir a, C., Bicho, M., Paiva, T., & Clara, J. G. (2000). Nocturnal sleep quality and circadian blood pressure variation. Portuguese Journal of Cardiology, 19 (10), 991 1005. Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: modeling cha nge and event occurrence Oxford: New York: Oxford University Press. Sliwinski, M. J., Hofer, S. M., & Hall, C. (2003). Correlated and Coupled Cognitive Change in Older Adults With and Without Preclinical Dementia. Psychology and Aging, 18 (4), 672 683. Sli winski, M. J., Smyth, J. M., Hofer, S. M., & Stawski, R. S. (2006). Intraindividual coupling of daily stress and cognition. Psychology and Aging, 21 (3), 545 557. Smith, A. (1982). Symbol Digit Modalities Test Revised: Manual. Los Angeles: Western Psych ological Services Smith, G. E., Housen, P., Yaffe, K., Ruff, R., Kennison, R. F., Mahncke, H. W., et al. (2009). A Cognitive Training Program Based on Principles of Brain Plasticity: Results from the Improvement in Memory with Plasticity based Adaptive Co gnitive Training (IMPACT) Study. Journal of the American Geriatrics Society, 57 (4), 594 603. Spielberger, C. D. (1983). Manual for the State Trait Anxiety Inventory (STAI) Palo Alto, CA: Consulting Psychologists Press. Stanley, M. A., Beck, J. G., & Glass co, J. D. (1996). Treatment of generalized anxiety in older adults: A preliminary comparison of cognitive behavioral and supportive approaches. Behavior Therapy, 27 (4), 565 581. Stern, Y. (2002). What is cognitive reserve? Theory and research application o f the reserve concept. Journal of the International Neuropsychological Society, 8 (03), 448 460. Stern, Y. (2006). Cognitive Reserve and Alzheimer Disease. Alzheimer Disease & Associated Disorders, 20 S69 S74. Stickgold, R., Hobson, J. A., Fosse, R., & Fos se, M. (2001). Sleep, Learning, and Dreams: Off line Memory Reprocessing. Science, 294 (5544), 1052 1057.

PAGE 273

273 Tabachnick, B. G., & Fidell, L. S. (2001). Using Multivariate Statistics (4th ed.). Boston, MA: Allyn & Bacon. Thomas, M., Sing, H., Belenky, G., Holco mb, H., Mayberg, H., Dannals, R., et al. (2000). Neural basis of alertness and cognitive performance impairments during sleepiness. I. Effects of 24 h of sleep deprivation on waking human regional brain activity. Journal of Sleep Research, 9 (4), 335 352. T horndike, E. L. (1903). Educational psychology New York, NY: Lemcke & Buechner. Thurstone, T. G. (1962). Primary mental ability for Grades 9 12 (Revised ed.). Chicago, IL: Science Research Associates. Tononi, G., & Cirelli, C. (2006). Sleep function and s ynaptic homeostasis. Sleep medicine reviews, 10 (1), 49 62. Tucker, M., McKinley, S., & Stickgold, R. (2011). Sleep Optimizes Motor Skill in Older Adults. Journal of the American Geriatrics Society, 59 (4), 603 609. Tworoger, S. S., Lee, S., Schernhammer, E. S., & Grodstein, F. (2006). The association of self reported sleep duration, difficulty sleeping, and snoring with cognitive function in older women. Alzheimer Dis Assoc Disord, 20 (1), 41 48. Unverzagt, F. W., Kasten, L., Johnson, K. E., Rebok, G. W., Mar siske, M., Koepke Mann, K., et al. (2007). Effect of memory impairment on training outcomes in ACTIVE. Journal of the International Neuropsychological Society, 13 (06), 953 960. Unverzagt, F. W., Smith, D. M., Rebok, G. W., Marsiske, M., Morris, J. N., Jone s, R., et al. (2009). The Indiana Alzheimer Disease Center's Symposium on Mild Cognitive Impairment. Cognitive Training in Older Adults: Lessons from the ACTIVE Study. Current Alzheimer Research, 6 375 383. Vallieres, A., Ivers, H., Bastien, C. H., Beaul ieu Bonneau, S., & Morin, C. M. (2005). Variability and predictability in sleep patterns of chronic insomniacs. Journal of Sleep Research, 14 447 453. Van Dongen, H. P. A., Maislin, G., Mullington, J. M., & Dinges, D. F. (2003). The cumulative cost of add itional wakefulness: does response effects on neurobehavioral functions and sleep physiology from chronic sleep restriction and total sleep deprivation. Sleep, 26 117 126. Vgontzas, A. N., Zoumakis, E., Bixler, E. O., Lin, H. M., Follett, H., Kales, A., e t al. (2004). Adverse Effects of Modest Sleep Restriction on Sleepiness, Performance, and Inflammatory Cytokines. J Clin Endocrinol Metab, 89 (5), 2119 2126.

PAGE 274

274 Wagner, U., Gais, S., Haider, H., Verleger, R., & Born, J. (2004). Sleep inspires insight. Nature, 427 (6972), 352 355. Walker, M. P. (2008). Cognitive consequences of sleep and sleep loss. Sleep Medicine, 9 (Supplement 1), S29 S34. Walker, M. P., Brakefield, T., Morgan, A., Hobson, J. A., & Stickgold, R. (2002). Practice with Sleep Makes Perfect: Sleep D ependent Motor Skill Learning. Neuron, 35 (1), 205 211. Weatherbee, S. R., Gamaldo, A. A., & Allaire, J. C. (2009). Exploring the Within Person Coupling of Reading Vision and Cognition in the Elderly. Aging, Neuropsychology, and Cognition, 16 (6), 671 682. Webb, W. B. (1985). A Further Analysis of Age and Sleep Deprivation Effects. Psychophysiology, 22 (2), 156 161. Webb, W. B., & Levy, C. M. (1982). Age, Sleep Deprivation, and Performance. Psychophysiology, 19 (3), 272 276. Wechsler, D. (1997). Wechsler Memo ry Scale Third Edition: Manual San Antonio, TX: The Psychological Corporation. Wetherell, J. L. (1998). Treatment of Anxiety in Older Adults. Psychotherapy: Theory/Research/Practice/Training, 35 (4), 444 458. Wetherell, J. L., Reynolds, C. A., Gatz, M., & Pedersen, N. L. (2002). Anxiety, Cognitive Performance, and Cognitive Decline in Normal Aging. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 57 (3), P246 P255. Willis, S. L., Tennstedt, S. L., Marsiske, M., Ball, K., El ias, J., Koepke, K. M., et al. (2006). Long term Effects of Cognitive Training on Everyday Functional Outcomes in Older Adults. JAMA, 296 (23), 2805 2814. Wilson, C. J., Finch, C. E., & Cohen, H. J. (2002). Cytokines and Cognition The Case for A Head to Toe Inflammatory Paradigm. Journal of the American Geriatrics Society, 50 (12), 2041 2056. Wimmer, F., Hoffmann, R. F., Bonato, R. A., & Moffitt, A. R. (1992). The effects of sleep deprivation on divergent thinking and attention processes. Journal of Sleep Res earch, 1 (4), 223 230. Woodworth, R. S., & Thorndike, E. L. (1901). The influence of improvement in one mental function upon the efficiency of other functions. (I). Psychological Review, 8 (3), 247 261.

PAGE 275

275 Yang, L., & Krampe, R. T. (2009). Long term Maintenance of Retest Learning in Young Old and Oldest Old Adults. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 64B (5), 608 611. Yang, L., Krampe, R. T., & Baltes, P. B. (2006). Basic forms of cognitive plasticity extended into th e oldest old: Retest learning, age, and cognitive functioning. Psychology and Aging, 21 (2), 372 378. Yang, L., Reed, M., Russo, F. A., & Wilkinson, A. (2009). A New Look at Retest Learning in Older Adults: Learning in the Absence of Item Specific Effects. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 64B (4), 470 473.

PAGE 276

276 BIOGRAPHICAL SKETCH scholarly interests in the aging process began when he was an undergraduate at the University of Nevada, Las Vegas. T hrough the combination of witnessing the aging of his grandmother and concurrently being enrolled in a seminar on cognitive aging a spark was lit in him that has never gone out. Over the course of his junior and senior year he was awarded grants from the N ational Institute on Health (NIH) and the National Science Foundation (NSF) to conduct independent research. After graduating magna cum laude and earning a Bachelor of Arts degree in psychology he quickly applied to graduate school. Dr. Dzierzewski was acc epted into the Department of Clinical and Health Psychology doctoral program in Clinical Psychology at the University of Florida (UF). In the s pring of 2007 he was award a Masters in Science from the University of Florida. While at UF he was continuously funded through national fellowships [ three years of support from a National Institute on Aging Training Grant ( T32 AG 020499 ) and two years of support from an Individual Training Grant from the National Institute on Aging ( F31 AG 032802 01A1) ] Throughou t his years at UF Dr. Dzierzewski received several prestigious awards, including: Thesis, American Psycholog Aging) Research and Retirement Award for Most Outstanding Completed Dissertation, Excellence in Research by a Stud ent Member Award, Most Outstanding Poster award by the college of Public Health and Health Professionals, Graduate Student Research College of Public Health and Health Professionals, and the Department

PAGE 277

277 erall Best Student Researcher Award. His research interests are many, but always share the common theme of aging. He is cognitive performance, and short term variability. This work represents the end of a long and winding road. Yet, it is just the beginning Following completion of a predoctoral internship at the Bruce W. Carter Veterans Affairs Medical Center in Miami, FL, he was awarded his Ph.D. from the University of F Advanced Fellow ship in Geriatrics at the Veterans Affairs Greater Los Angeles Healthcare System where he will continue in scholarly endeavors and clinical care with older adults