|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 ITEM AND ASSOCIATIVE MEMORY IN NORMAL AGING AND IN DIVIDUALS WITH MILD COGNITIVE IMPAIRMENT By YU-LING CHANG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008
2 2008 Yu-Ling Chang
3 To my parents, for their love and support
4 ACKNOWLEDGMENTS My deepest appreciation goes to m y chair and mentor, Dr. Russe ll M. Bauer. He has been an inspiring role model and I thank him for his constant encouragement and fostering of independence, and for the research and clinical skills that he helped me to develop throughout my graduate training. Many thanks to Dr. David Loring for his astu te guidance and assistance in recruiting participants throughout the development of this work. I would like to thank Dr. Michael Marsiske for offering his constructive comments and guidance in statistics. I also thank my other doctoral committee members: Drs. Kenn eth Heilman, Tom Kerkhoff, and Lise Abrams. They have watched me progress through my edu cation and research, ch allenging, guiding, and encouraging me to think more critically. W ith their help, my dissertation project became stronger. I feel very fortunate to have tr ained in such an enriching environment. A very special thanks goes to Dr. Bruce Crosson who has been my mentor in the neuroimaging world. He took me under his wing and provided me with countless opportunities that allowed me to experience different aspects of our profession in an academic environment that was unusually cooperative and intellectually motivating. I have been truly blessed to be able to work with him and the gang in the BIRC laboratory. I also thank Dr. Mau-Sun Hua, who inspired and encouraged me to pur sue a career in neuropsychology. I would like to express my appreciation to Deborah Townsend and Emily King for their assistance with participant recruitment and data collection. Mostly im portantly, I thank my parents for defining love for me as a verb and gi ving me the freedom to pursue my dreams. I am indebted to you both for your many years of unrel enting encouragement, support and belief, even in the face of challenge and obstacles. I also want to thank my husband Yu-Feng Tu for his love and rock solid support. It is because of him th at I can sail so smoothly, accomplish any goal, and tackle lifes toughest challenges with my sanity still intact.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES.........................................................................................................................9 CHAP TER 1 INTRODUCTION..................................................................................................................13 Item versus Associative Memory........................................................................................... 13 Association Memory and Its Neural Correlates ..............................................................15 Frontal lobe and associative memory....................................................................... 15 Hippocampus and associative memory.................................................................... 16 Can Item Memory and Associative Memo ry be Disproportionately Im paired?............. 20 Aging and Memory............................................................................................................... ..24 Individual Difference among Older Adults..................................................................... 26 Mild Cognitive Impairment............................................................................................. 27 Associative Memory in Normal and Pathological Aging................................................ 28 2 PURPOSES AND AIMS OF THE PRESENT STUDY ........................................................32 3 METHODOLOGY................................................................................................................. 35 Participants.............................................................................................................................35 Mild Cognitive Impairment Participants......................................................................... 35 Healthy Older Control Participants................................................................................. 38 Differential Contributions of the Fr ontal and Medial Tem poral lobes................................... 39 General Procedure.............................................................................................................. ....42 Standardized Clinical Ne uropsychological Measures ............................................................42 Neuropsychological and Questionnaire Data.........................................................................45 4 THE EFFECT OF INCIDENTALVERSUS INTENTIONAL LEARNING ON THE RETRIEVAL OF ASSOCIATION I NFORMATION........................................................... 48 Introduction................................................................................................................... ..........48 Methods..................................................................................................................................49 Participants......................................................................................................................49 Design..............................................................................................................................49 Materials and Procedure.................................................................................................. 50 Hypotheses and Predictions.................................................................................................... 52 Results.....................................................................................................................................53 Analyses on the HY, HO, and MCI Groups....................................................................53 Analyses on Groups Divided by MTL-FL Status............................................................ 58
6 Discussion...............................................................................................................................61 5 MEMORY OF ASSOCIATIONS BETWEE N ITEMS OF T HE SAME KIND USING SEMANTIC RELATED OR UN RELATED WORD PAIRS................................................ 64 Introduction................................................................................................................... ..........64 Methods..................................................................................................................................65 Participants......................................................................................................................65 Design..............................................................................................................................65 Materials and Procedure.................................................................................................. 66 Hypotheses and Predictions.................................................................................................... 67 Results.....................................................................................................................................68 Discussion...............................................................................................................................72 6 ASSOCIATIONS BETWEEN DIFFERENT KI NDS OF INFORMATION CROSSREGION ASSOCIATIONS................................................................................................... 73 Associations between Objects and Locations.........................................................................73 Introduction................................................................................................................... ..73 Methods...........................................................................................................................74 Participants...............................................................................................................74 Design.......................................................................................................................74 Materials and Procedure........................................................................................... 75 Hypotheses and Predictions............................................................................................. 75 Results........................................................................................................................ .....76 Analyses on HY, HO, and MCI groups...................................................................76 Analyses on groups divided by MTL-FL status....................................................... 79 Associations between Faces and Houses--Relational Inform ation......................................... 82 Introduction................................................................................................................... ..82 Method.............................................................................................................................82 Participants...............................................................................................................82 Design.......................................................................................................................83 Materials and procedure...........................................................................................83 Hypotheses and Predictions............................................................................................. 84 Results........................................................................................................................ .....85 Analyses on the HY, HO, and MCI groups.............................................................. 85 Analyses on groups divided by MTL-FL status....................................................... 86 Association between Temporal Inform ation and Novel Faces............................................... 89 Introduction................................................................................................................... ..89 Methods...........................................................................................................................90 Participants...............................................................................................................90 Design.......................................................................................................................90 Materials and procedure...........................................................................................90 Hypotheses and Predictions............................................................................................. 92 Results........................................................................................................................ .....92 Analyses on the HY, HO, and MCI groups.............................................................. 92 Analyses on groups divided by MTL-FL status....................................................... 93
7 Association between Temporal Info rm ation and Historical Events.......................................96 Methods...........................................................................................................................96 Participants...............................................................................................................96 Design.......................................................................................................................96 Materials and procedure...........................................................................................96 Hypotheses and Predictions............................................................................................. 97 Results........................................................................................................................ .....97 Analyses on the HY, HO, and MCI groups.............................................................. 98 Analyses on groups divided by MTL-FL status....................................................... 98 Correlations between tempor al order test and neuropsychological m easures.......... 99 Discussion..................................................................................................................... .100 7 CONCLUSIONS AND DISSCUSIONS .............................................................................. 106 Disproportionate Impairment on the Associat ive Mem ory Compared to Item Memory..... 107 Semantic Memory versus Episodic Memory........................................................................ 111 Frontal Lobe Involvement in Associative Mem ory and the Heterogeneity Among Older Adults................................................................................................................................113 Study Limitations.............................................................................................................. ....114 Future Direction....................................................................................................................116 Conclusions...........................................................................................................................117 LIST OF REFERENCES.............................................................................................................118 BIOGRAPHICAL SKETCH.......................................................................................................129
8 LIST OF TABLES Table page 3-1 Mean characteristics of groups based on cognitive status .................................................38 3-2 Mean characteristics of groups based on frontal lobe (FL) function and m edial temporal lobe (MTL) function........................................................................................... 41 3-3 Numbers of people classified as normal old or MCI in each of the four groups ............... 42 3-4 Neuropsychological measures included in the present study ............................................ 43 3-5 Means and standard deviations for neuropsychological non-mem ory measures............... 46 3-6 Means and standard deviations fo r neuropsychological m emory measures...................... 46 3-7 Means and standard deviations for subjective mood and memory questionnaires............ 47 4-1 Data for the incidental and intentio nal learning conditions in three groups. ..................... 54 5-1 Means and standard deviations for the va riables o btained on the semantic related or novel word pair list in both immediate and delayed conditions on the young, healthy old, and MCI groups.......................................................................................................... 70 6-1 Means and standard deviations of vari ables obtained on the Object Location Test on the three groups ..................................................................................................................77 7-1 Effect sizes for all associative measures..........................................................................109
9 LIST OF FIGURES Figure page 3-1 Frequency distribution of m edial temporal lobe compos ite scores for total elderly sample (n=35).................................................................................................................. ..40 3-2 Frequency distribution of frontal lobe com posite scores for total elderly sample (n=35).................................................................................................................................40 4-1 Data of the immediate item recognition hits for the three groups..................................... 55 4-2 Data for the immediate and delayed recogni tion of the pairs (hits ) in the incidental and intentional learning condi tions for the three groups. .................................................. 57 4-3 Data for the immediate and delayed recogniti on false alarm s (false positive) errors of the pairs in the incidental and intentiona l learning conditions for the three groups.......... 58 4-4 Data for the immediate and delayed recogni tion of the pairs (hits ) in the incidental and intentional learning c onditions for the four groups divided by their MTL-FL function levels. ...................................................................................................................60 4-5 Data for the immediate and delayed rec ognition false positive score of the pairs in the incidental and intentional learning c onditions for the four groups divided by their MTL-FL function levels. ................................................................................................... 60 5-1 Data for free recall, cued recall, and recognition test on the sem antic related or novel word pair list in both immediate and delayed conditions for the young, healthy old, and MCI groups.................................................................................................................71 6-1 The display of the namable objects and th eir locations for the object-location test. ......... 76 6-2 Data for object recognition hit and ob ject-location recogni tion tests in both imm ediate and delayed conditions fo r the young, normal old, and MCI groups............... 78 6-3 Data for object, location, and object-locat ion free recall tests in both imm ediate and delayed recall for the young, normal old, and MCI groups............................................... 79 6-4 Data for object, location, and object-locat ion free recall tests in both imm ediate and delayed recall for the four groups divide d based on their function on MTL and FL factors.................................................................................................................................81 6-5 Data for object recognition hit and ob ject-location recogni tion tests in both imm ediate and delayed conditions for the f our groups divided based on their function on MTL and FL factors......................................................................................................81 6-6 An example of the stimuli and the task for the face house association test ..................... 84
10 6-7 Data for immediate and delayed recall on th e face-house association test across three groups.................................................................................................................................86 6-8 Data for face-house association test in both imm ediate and delayed conditions for the four groups divided based on their function on MTL and FL factors............................... 87 6-9 Data for face-house association test in both imm ediate and delayed conditions for the two groups divided based on their function on medial te mporal lobe factor..................... 88 6-10 Data for face-house association test in both imm ediate and delayed conditions for the two groups divided based on their f unction on frontal lobe factor.................................... 88 6-11 Data of the between-lis t and within-list trials at two tim e points for the young, healthy old, and the MCI groups........................................................................................ 94 6-12 Data of the recency task for the four groups divid ed by their frontal and medial temporal lobe function levels............................................................................................. 95 6-13 Data of the difference scores on the hist orical event tem poral order test for the young, healthy old, and the MCI groups............................................................................98 6-14 Data of the difference scor es on the historical event tem poral order test for the four groups divided by their m edial temporal l obe and frontal lobe function levels................ 99
11 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ITEM AND ASSOCIATIVE MEMORY IN NORMAL AGING AND IN DIVIDUALS WITH MILD COGNITIVE IMPAIRMENT By Yu-Ling Chang August 2008 Chair: Russell M. Bauer Major: Psychology Efficient and accurate prediction of dementia in the preclinical stage is important for social and medical reasons. The foregoing literature review suggests that associative memory may serve as a tool for early detection of individuals who may develop dementia later in life. Individuals with mild cognitive impairment (MCI) may be in a transitional stage between normal aging and dementia. The primary goal of the current study was to systematically investigate associative memory in healthy younger adults, healthy older adults, and individuals with Amnestic MCI. We further examined heterogeneity in frontal functioning documented in literature and its relationship with the performan ce on associative memory among ol der individuals. Thirty-five older adults (11 with Amnestic MCI) and 20 younger adults comple ted a series of associative memory experiments which assessed associations between information of the same kind (wordword pairs), associations between different kinds of informati on (e.g., object-location pairs), and relational associations (an association establis hed between two items through a third item). Additionally, each participant completed a neurop sychological screening battery which served the purpose of classification into different groups (Amnestic MCI or healthy control). Results revealed that older adults with Amnestic MCI de monstrated the greatest and a disproportionate impairment of associative memory with relativ ely spared item memory. Second, the impaired
12 associative memory was found in all kinds of associative memory tasks. Third, the results suggest that normal healthy old a dults, but not MCI individuals, benefited from explicit encoding instruction when performing associative memo ry. Furthermore, free recall compared to recognition was a more sensitive paradigm when examining the associative memory. Finally, a parallel analysis on groups esta blished by the older individuals f unctional level (high vs. low) in the frontal /executive functioning and medial tempor al lobe functioning rev ealed that the medial temporal lobe function is critical for associ ative memory. The level of frontal/ executive function became important only when the inte grity of the medial temporal lobe was compromised. Overall, our results suggest that associative memory functioning may serve as a reliable and sensitive indicator for early detection of people who may develop dementia later in life.
13 CHAPTER 1 INTRODUCTION Item versus Associative Memory Com plex events consist of multiple kinds of information that are related or bound together as a unitary whole. An event can incl ude the semantic content, information about the time in which the event took place, the place in which it occurred, the acting agents and their characteristics, and so on. All of these aspects integrated with the internal cognitive state of the person are encoded as an episode. Remembering such an episode requires that at least some of the bound components be retained, as well as their relationships with each other. As people age, their memory for recent events tends to become less precise. Although they might know that a particular event occurred or have the knowledge of a particular f act, they may be less likely to recall other information associated with the even t (e.g., where or when the event took place). The distinction between item and associative information has a long history in psychology and has been supported by several experiments that yielded different patter ns of results for the two types of information (Hockley, 1991, 1992; Ho ckley & Cristi, 1996). The items in question are typically well-integrated stimuli such as wo rds, letters, numbers, or pictures, whereas associative information refers variously to th e information of new items by binding features together, to the links among severa l items, or to the integration of an event with its context of occurrence. It could be argued th at vast areas of memory resear ch, including contextual aspects of memory and memory for source, basically i nvolve the study of associative memory. For example, the relationships of two messages can be conceptualized within a classic pairedassociation paradigm. The relationships of a messa ge and the characteristics of the acting agents (e.g., the voice or the face) can be probed as memory for context. Memory for a message in a specific time and space is investigated as memory for source. All of these examples constitute
14 the establishment of connections (associations) among single representationa l units at one level or another. Although the distinction between item and associative information is well supported in the literature, methodological issues have raised questions concerni ng the interpretation of many of the existing studies of item and associative memo ry. First, because retrieving associations is commonly more difficult than retrieving single item s, individuals with poor memories, such as amnesic patients and older adults, may show associative memory impairments because of task difficulty, rather than because of differential impairment in associative memory capacity per se. Second, memory for associative information has us ually been assessed only for those items that are remembered. Thus, item and associative memo ry tests have not been independent. Third, the forms of the tests for item and associative info rmation are not often equated. For example, in some studies, item memory has been tested in recognition, whereas associative memory has been tested in recall. In general, studies have s hown impaired associative memory when item memory is either normal or equated across groups, but have not shown impaired item memory with normal associative memory. Typically, recogniti on of single items is matched between patient and control groups by giving th e patients extended study time or by testing them following shorter delays than control subject s. The ability to remember the associative information is then investigated under the equivalent item memory conditions and, if it is impaired, it can be concluded that memory for this kind of inform ation is disproportionately impaired, relative to item recognition. The validity of this conc lusion, however, depends on insuring that the matching procedure does not affect one kind of me mory more than the other in normal subjects (Mayes, Meudell, & Pickering, 1985).
15 Association Memory and Its Neural Correlates The actual m echanisms of integration or bi nding are still poorly understood, although we know that content and context must be integrated at encoding, and information about their cooccurrence must be stored in memory. Inte grative encoding processe s that often involve conscious attentional processes may depend on the frontal lobe (Stuss & Benson, 1986). Alternatively, the binding of multip le components of a stimulus event into an integrated trace, which is often considered as an automatic pr ocess may be a key function of the hippocampalmedial temporal lobe memory system (C ohen & Eichenbaum, 1993; Cohen, Poldrack & Eichenbaum, 1997). Frontal lobe and associative memory Previous res earch has clearly established that certain types of frontal lesions can produce deficits in associative memory, particularly in temporal or source context memory tasks, when compared to lesions of the temporal lobe (But ters, Kaszniak, Glisky, Eslinger, & Schacter, 1994; Milner, 1982; Milner, Corsi & Leon ard, 1991). This pattern is also seen in certain spatial context memory tasks (Owen, Downes, Sahakian, Polk ey, & Robbins, 1990; Owen, Sahakian, Semple, Polkey, & Robbins, 1995). Cabeza et al. (1997) used positron emission tomography (PET) to directly contrast the neural co rrelates of items and temporal order memory. In their study, subjects studied a list of words and were then scanned while retrieving information about what words were in the list or when they occurred within the list (recency j udgment). They found item retrieval was related to increased neural act ivity in medial temporal and basal forebrain regions, whereas temporal-order re trieval was associated with ac tivations in dorsal prefrontal, cuneus/precuneus, and right posterior parietal regions. The di ssociation between temporal and frontal lobe regions confirms and extends previo us lesion data. Although historically important, these results leave some gaps. First, the double dissociation between tem poral lobe and frontal
16 lobe patients has been difficult to replicate. Mi lner, Corsi, & Leonard (1991) showed that fontal patients were more impaired in recency than on recognition judgments, but the opposite pattern in temporal lobe patients was not clearly observed. Patients w ith temporal lobe lesions were surprisingly unimpaired in recognition memory for words or represen tational drawings, and although temporal lobe patients we re significantly impaired on r ecognition of abstract pictures compared to control subjects, they also showed a decrement in recent memory for these items. Hippocampus and associative memory The discovery that the m edial temporal br ain regions, particularly the hippocampal formations, are essential for human memory provi ded the basis for neuros cientific theories and experimental practice during the past several decades. In th e years following this discovery, research with amnesic patients led to the finding that memory is not a unitary system but is divided into subsystems, each supported by di fferent but partially overlapping neuronal networks. The hippocampus is thought to be primarily involved in declarative memory, though its specific function has been variously descri bed as involving conso lidation (Squire & ZolaMorgan, 1991) episodic memory (Vargha-Khadem et al., 1997), novelty detection (Tulving et al., 1994, 1996), and spatial learning (Maguire et al., 1998). At the same time, learning experiments with rats (Cohen & Eichenbaum, 1993; Eichenbaum et al., 1994; Eichenbaum, 1997) as well as with amnesic patients (Kroll et al., 1996; Mayes et al., 2001; Vargha-Khadem et al., 1997) indicated that the hippocampal forma tion is important for the establishment of associations between component s of episodes in memory. Ne uroimaging evidence in younger adults also suggests that medial-temporal re gions are critically i nvolved in associativerecognition tasks for words, both at encoding (Jackson & Schacter, 2004) and at retrieval (Giovanello, Schnyer & Verfaellie, 2004; Heckers et al., 2004; Pres ton et al., 2004; Yonelinas et al., 2001).
17 The experience of an episode typically invol ves the simultaneous processing of diverse sensory inputs, bodily sensations, thoughts, and emotions in distri buted cortical re gions, creating patterns of coactivations in the cortex. The composition and temporal contiguity of these coactivations needs to be stored in memory for the later recovery of some or all aspects of that episode. The anatomy and physiology of the hip pocampal system lends itself to store such patterns of neuronal co activations temporarily (Cohen & Eichenbaum, 1993; Eichenbaum, 1997; Squire & Zola-Morgan, 1991). Inputs from th e various sensory association cortices are channeled through the perirhinal and parahippocampal cortices to the entorhinal cortex (or directly to the entorhinal cortex) and, from there, to the hippocampal formation. The hippocampus is reciprocally conn ected with these sensory areas and therefore can reactivate them. This suggests that the hippocampus perf orms the most complex mnemonic computations such as associating the different inputs with one another and reactivati ng the involved cortical sites for consolidation and retr ieval of information. Moscovitch and Winocur (1995) also suggested that medial temporal ar eas are at the core of an associative system that automatically binds together what is consciously apprehended They indicated that at any point in time a number of distinct neural netw orks/ pathways are activated via both internally generated and/or externally available stimuli. The medial temp oral system has been viewed as binding these distinct patterns (act ivation patterns) in a relatively automatic/modular manner to produce a record of the conscious experience. Although the hippocampal system seems to be important for the process of binding information from different inputs, it is unclear whether hippocampal damage impairs all kinds of associative memory to the same extent. Thr ee types of associative memories have been described commonly across studies, although the terms in different st udies are sligh tly different:
18 1) intra-item association. An example of such type of associations is the associations among parts or components within a face. 2) associatio ns between the same kinds of items or withinregion associations, i.e., associa tions between information stored in the same cortical region. Examples of this kind of association include word -word or face-face associations. 3) associations between different kinds of inform ation or cross-region associati ons, i.e., associations between information represented in distinct cortical regions. Examples fo r this kind of associations are associations between objects and their location, words and their temporal information, and faces and names etc. Describing this three-level ta xonomy of associative memory is not meant to imply that associative memories cannot be ca tegorized in other meaningful ways (e.g., intramodal vs. cross-modal associations, etc.). Kroll et al. (1996) examined recognition memo ry for associations between components of schematic faces or syllables of known words (e.g ., study valley and bar ter and test either with these target words, new words, or the sy llabic recombination barle y). They found their patient with bilateral hippocam pal damage could discriminate between studied words or faces and completely novel foils, but was significantly impaired at discrimi nating between studied items and recombined foils for both the word and face items. Their results suggest that hippocampal lesions disrupt the ability to form intra-item associative memory. In contrast, Vargha-Khadem et al. (1997) found that early hippocampal lesions in childhood impairs the recognition of associations betw een different kinds of informa tion (e.g., face-voice and objectplace associations) but spares the recognition of associations between th e same kinds of items (e.g., word-word or face-face associations). Si milarly, Mayes et al. (2001, 2004) reported the case of an adult amnesic patient with bilateral hi ppocampal atrophy who showed normal intraitem associations and associative recognition of word or face pair s but failed in the recognition
19 of associations between differe nt kinds of information after controlling the difficulty level between different tasks. Based on these fi ndings, the suggestion has been made that hippocampal damage disrupts memory for cross-region associations but spares memory for single items and memory for intra-item and within-region associations. The Kroll et al. (1996) study did not control for task difficulty; in th eir study, normal subjects found the recombination task harder than discriminating targets from novel foils. Furthermore, their patient was impaired at discrimination of studied patt erns comprising several components from recombination foils. It is unclear whether this test required memory for intra-item associations or inter-item association because the patterns may not have b een perceived as items. Intere stingly, Turriziani et al. (2004) found that patients with hippocampal lesions perfo rmed more poorly than the normal controls on associations between the same kinds of items and associations between different kinds of information, and there was no indication that the impairment was relatively more severe for the latter than for the former kind of association. Deficits in source memory (association be tween content and context) have been occasionally reported in patients with diencepha lic or temporal lobe lesions, though it is controversial as to whether such deficits should be attributed to conc omitant frontal pathology. On the basis of correlational data and also comparisons with patients with frontal lobe lesions, various authors have taken the view that associ ation memory for temporal order and source are dependent upon frontal functi on (Janowsky, Shimamura, & S quire, 1989; Schacter, 1987; Schacter, Harbluk, & McLachlan, 1984; Shimamura & Squire, 1987; Shimamura, Janowsky, & Squire, 1990). However, Kopelman (1989) Parkin, Leng, & Hunkin (1990), Kopelman, Stanhope, & Kingsley (1997), and Ma yes et al. (2001) found that te mporal association memory was more strongly correlated with target memo ry performance in amne sic patients than with
20 performance on tests of frontal lobe function. Pickering, Mayes and Fair bairn (1989) reported a similar finding for modality context, and Shoqeir at & Mayes (1991) and Ho ldstock et al. (2002) had similar findings for spatial context. Like wise, Chalfonte, Verfae llie, Johnson and Reiss (1996) failed to find a correlation between frontal function and spatial location memory. These latter findings indicate that associative memory impairment s eems to be more closely associated with a core memory deficit in amnesic patient s than with concomitant executive or frontal dysfunction per se. In summary, although the precis e roles of hippocampus and fr ontal lobe in associative memory are still unclear, available evidence suggests that both brain regions are needed for successful coding and storage of an entire event. The frontal lobe is needed to initiate and carry out cognitive control pro cesses to ensure that all aspects of an experience are encoded; the medial temporal lobe-hippocampus is critical fo r the rapid formation of associative memories and may serve to physiologically integrate informa tion recorded in many different areas of the brain. Although there is insuffi cient evidence to indicate that hippocampus damage impairs one type of associative memory ove r the other, available data su ggests that the hippocampus may play a more important role in item-item associat ions (involving both same and different items) than it does in intra-item associa tion (i.e., linking parts of items together). This latter function may be more likely handled by modality-specific association cortex and may be computed prior to hippocampal processing. Can Item Memory and Associative Memor y be Disproportionately Impaired? As suggested by the findings reviewed a bove, hippocam pus may have as its most important function the rapid formation of associ ative memories. However, the literature on the effects of relatively selective hippocampal le sions on item memory in humans is also well documented (Reed & Squire, 1997), although the effects on item recognition vs. recall are
21 conflicting (Mayes et al., 2002). One issue that remains unresolved is whether any impairment in associative memory following amnesic patie nts with selective hippocampal lesions is disproportionate to their defic it in memory for individual items (Cave & Squire, 1991; Mayes et al., 2004; Squire, 1982; Vargha -Khadem et al., 1997). Although studies have consisten tly shown that amnesic patien ts are impaired at list discrimination and temporal sequencing tests, performance on these tests has not always been found to be disproportionately impaired relative to item recognition in patients with medial temporal lobe (MTL) damage. Early wo rk by Squire et al. (1981, 1982) reported a disproportionate deficit in memory for temporal context in patie nts with Korsakoffs syndrome, whereas the performance of MTL patients was no worse than woul d have been predicted by their poor recognition memory. Based on such findings, these authors suggested that hippocampal, more extensive MTL, and midline diencephalic lesions should each disrupt temporal order memory and item recognition to equivalent extents. This unitary view predicts that temporal order memory will be more impaired than item recognition when any of these lesions is accompanied by frontal cortex damage that fu rther impairs temporal order memory in a relatively selective way. However, some caution in accepting the conclusions of Squire et al. (1981, 1982) is warranted because their MTL group was recruited from a sample of depressed patients who had recently undergone ECT, in whom it was claimed would be likely to suffer a functional disturbance of MTL structures. Als o, a matching procedure was not used in their study. Despite these limitations, several further studie s have also supported the contention that disproportionate deficits in lis t discrimination memory seen in patients with Korsakoffs syndrome need not be related to frontal lobe dysfunction (Hunkin & Parkin, 1993; Hunkin,
22 Parkin, & Longmore, 1994). For example, Hunkin, Parkin, & Longmore (1994) found that although amnesics with Korsakoffs syndrome a nd those with MTL damage were similarly impaired in their recognition memory for target items, the Korsakoff amnesics were more impaired than the MTL patients on a test of lis t discrimination. The disproportional impairment found in Korsakoff amnesic did not correlate with performance on tests associated with frontal lobe functioning. Based on the result s, Parkin et al. (1992) argued that retrieval of contextual information, defined as spatiotemporal and other in formation that allows memories for events to be differentiated, plays a role in the recall/recognition of target (o r attended) information and that encoding of contextual information into memo ry is disrupted by the kind of damage to diencephalic structures that is f ound in Korsakoff patients. In c ontrast, the role of the MTL is postulated to be that of conso lidating target information, cont ext information, and presumably the associations between them, into memory. According to this hypothesis, patients with MTL damage would therefore be equally impaired at consolidating target (item) and temp oral as well as other kinds of contextual and associative information into memory. MTL lesi ons of any kind should, therefore, disrupt these different forms of memory to the same degree. Their hypothesis differs from the unitary view in predicting that midline diencephalic lesions will disrupt context information more than item recognition even when there is no additional impa irment of context memory caused by frontal cortex damage. The results found in Hunki n, Parkin, & Longmore (1994) support their argument, however, as was the case with the Squire et al. (1981, 1982) study, Hunkin et al. (1994) did not use matching procedure and theref ore the results may be confounded by difficulty issues.
23 Several recent studies based on matching pr ocedures have found that patients with selective hippocampal lesions disp layed disproportional associative memory impairment relative to their item memory. For example, Vargha-K hadem et al. (1997) found that patients with selective hippocampal lesions showed fairly nor mal item recognition, but that their recognitions of object-location and face-voice asso ciations were impaired. Kro ll et al. (1996) also examined recognition memory for associations between sy llables of known words (e.g., study valley and barter and test either with these target wo rds, new words, or the syllabic recombination barley) or components of schematic faces. They found their patient who reportedly had bilateral hippocampal damage fo llowing an anoxic episode could discriminate between studied words or faces and completely novel foils, but was significantly impaired at discriminating between studied items and recombined foils for both the word and face items. Similarly, Mayes et al. (2004) found that a patie nt who suffered a sele ctive bilateral lesion to the hippocampus showed relatively preserved verbal and visual ite m recognition memory. However, this patient demonstrated disproportio nal impairment at recognition of associations between different kinds of information (e.g., association between objects and their location, words and their temporal order, and faces and voices, etc.). The views of Squire and his colleagues (1981, 1982), and Parkin et al. (1992) cannot explain th e above findings. Alternatively, those resu lts seem to fit with the hypothesi s, forwarded by Aggleton and Brown (1999), that the role of the extended hippocam pal system (hippocampus, fornix, mammillary bodies, anterior thalamus, and possibly parts of th e cingulate cortex) is to link target information with contextual information such as spatial, te mporal, and source information in memory so that particular episodes are uniquely characterized. As such, the extended hippocampal system is postulated to mediate memory for all kinds of associations that unde rlie the process of
24 recollection (item-context retrieva l) on which free recall and a ssociative recognition are based. On the other hand, perirhinal cortex of the MTL and its projection to the dorsomedial nucleus of the thalamus (DM) are postulated to mediate th e process of familiarity on which recognition of single-item information is based. Their view pr edicts that both hippocampal and more extensive MTL lesions would disrupt temporal order memory to a greater extent than item recognition even when there is insignificant damage to the frontal lobes. Furtherm ore, hippocampal lesions will not disrupt item recognition at all if this mainly depends on item familiarity rather than recollection. It also predicts that midline di encephalic lesions will have the same effect as hippocampal lesions if they only affect structur es in the extended hippocampal system, and the same effect as more extensive MTL lesions if they additionally damage structures in the perihinal cortex-DM system. In summary, the review of the literature reveal s conflicting results with regard to the issue of whether associative memory is more impaired than item memory following selective hippocampal lesions, although several recent studies found the associative memory is disproportionately impaired when compared to item memory based on the careful use of procedures that match for initial level of learning. Aging and Memory Mem ory loss is one of the main complaints of normal aging. Older adults often experience substantial difficulty compared with young adults in remembering many types of event details. For example, there is evidence of age-related me mory deficits for spatial information (Kirasic, 2001), voices (Kausler & Puckett, 1981), color (Park & Puglisi, 1985), temporal information (Parkin, Hunkin, & Walter, 1995; Schacter et al., 1991), and persons (Ferguson, Hashtroudi, & Johnson, 1992). Several hypotheses have been attemp ted to explain the relatively poor memory performance of old adults, includ ing a deficit in semantic proce ssing, a failure of metamemory, a
25 failure of deliberate recollec tion, a reduction in processing re sources (Light, 1991), impaired processing speed (Salthous, 1996), and a failure of inhibitory processes (Hasher & Zacks, 1988). Although all of the above relate to older adults episodic memory deficits, none of them alone can provide an explanation for the full range of phenomena associated with age-related declines in memory. Lesion (Pruall, Gabrieli, & Bunge, 1999) a nd imaging (Cabeza & Nyberg, 2000) studies indicate that episodic memory draws on a widesp read network of brain structures, including the hippocampus and related regions, diencephalon, basal forebrain, anterior cingulate gyrus, precuneus and other parietal st ructures, cerebellum, and frontal lobe. The susceptibility of episodic memory to aging may thus reflect that performance could be disrupted because of changes at multiple sites in a large distributed netw ork. Brain structures that may be particularly relevant to episodic memory are known to be affected by aging. Specifically, moderate agerelated atrophy of hippocampus (Golomb et al., 1 994) and a more sizable loss of frontal volume in old age (Raz et al., 1997) ha ve been reported. In genera l, attempts to link volumetric measurements of these structures to age-relate d episodic memory deficits have been successful, with most studies reporting m oderate to strong relationships. The pattern of outcome from studies in which the brain activity of young and older adults is compared during encoding and retrieval is less consistent, likely reflecting differences across studies in tasks, methods, and subject select ion procedures. Many st udies additionally suffer from small sample sizes. Nevertheless, there are several findings that may contribute to our understanding of age-related episodic memory defic its. Grady et al. (19 95) found an age-related decrease in hippocampal activity during encoding of faces. Such a decrease was not observed during recognition. This suggest s that age-related functional changes in the hippocampus are
26 more likely to occur at encoding than at retrie val. There is also evidence of age-related decreases of activity in left inferior frontal co rtex during episodic encoding (Cabeza et al., 1997). This region has been implicated in semantic el aboration during encoding in imaging research with young adults (Kapur et al., 1994). Also of interest is the finding that young adults show decreased activity in left frontal cortex under conditions of divide d attention (Idaka et al., 2000). In addition, in young adults prefront al activation is lateralized for episodic memory, such that left prefrontal activity dominates during encoding and ri ght prefrontal activity dominates at retrieval (Nyberg, Cabeza & Tulving, 1996). Several studie s demonstrate that this encoding-retrieval asymmetry is markedly reduced in old age (Back man et al., 1997; Cabeza et al., 1997), possibly reflecting age-related decreases in specificity of neural processing dur ing episodic remembering (Cabeza, 2001). Individual Difference among Older Adults Although older adults, in general, perform more poorly than young adults in episodic memory tasks, there is great variability within the older population with re gard to the size of the impairment. Many researchers have focused on demographic, lifesty le, and health-related factors to explain these indivi dual differences (Backman et al., 1999). However, knowledge of these factors is not very useful in determining the sources of age-related performance deficits, although it is informative in understanding late life variability in memory performance. As suggested by literature, there are data s uggesting dysfunction of both medial temporal and frontal lobe regions in olde r adults. Moreover, the link betw een volumetric measurements of these structures to age-related episodic memory deficits has been successful. Recently, Glisky, Polster & Routhieaux (1995) divi ded their older adults into groups based on two dimensions of functioningfrontal lobe functioning and me dial temporal functioning to address the individual difference i ssues in the topic of source memory. Their data suggested that
27 source memory is deficient only in a subset of older adults, who ha d lower frontal lobe functioning compared to younger adu lts. Although their measures of FL and MTL function were based on neuropsychological, not neuroanatomical evidence, and so th ey could provide only indirect indication of the involve ment of specific brain regions, their study actually demonstrated an effective way to explain the variable results among literatures and the individual differences among older population. Mild Cognitive Impairment In the pas t several years, there has been c onsiderable debate in the aging literature surrounding what appears to be a continuum of memory performance between normal aging and pathological aging. Mild cognitive impairmen t (MCI) is a controversial concept which is generally referring to persons who do not fulf ill criteria for Alzheimers disease (AD) or dementia, but who exhibit some form of cognitive impairment. Evidence from both neuropsychological and neuroimagi ng studies have suggested that MCI may represent a clinical prodrome to degenerative dementias such as AD. For example, both cross-sectional and longitudinal neuropsychological research has sugg ested that people with MCI have significant memory impairment beyond what is expected in normal aging, but less than that seen in dementia syndromes, and that such cognitive defi cits may increase the likelihood that dementia will develop later (Bischkopf, Bu sse & Angermeyer, 2002). In addition, significant hippocampal and entorhinal cortex volume reductions, whic h are well-established risk factors for the development of AD, are consistently found in s ubjects with MCI as compared with cognitively unimpaired controls (Wolf et al., 2003). Although criteria have been specified for the classification of MCI, these criteria differ across studies, and the potential of miscla ssification is high (Busse & Angermeyer, 2002, Petersen et al., 2001). It has been suggested that MCI may re present a highly heterogeneous
28 group, the members of which cannot be classified perfectly by current criteria. With the subclassifications of MCI, which includes MCI-amnestic, MCI-multiple domains slightly impaired, and MCI-single nonmemory domain, proposed at the Current Concepts in MCI Conference in 1999 (Petersen et al., 2001) a numbe r of shortcomings of the classical MCI concept was addressed. Cognitive impairment in MCI is now clearly related to age and education-specific norms and is not limited ex clusively to isolated memory impairment. Although the subclassifications of MCI by Petersen et al. (2001) wa s a further step for scientists to understand the inherent nature of MCI, Bu sse et al. (2003) recently found that none of the three subclassifications alone pr oved to have significant rela tive predictive power in the prediction of dementia. Thus, further populatio n-based studies addressing the heterogeneous nature of MCI are needed, as is the development of cognitive test and biological markers. If persons with MCI are in a prodrom al or at risk stage for devel oping dementia, we will need to develop criteria with a high sensitivity and specificity for identifying these individuals so that they can be offered effective pharmacological a nd behavioral interventions for delaying the onset and progression of the disease. Associative Memory in Norm al and Pathological Aging Chalfonte and Johnson (1996) have suggested that part of ol der adults deficient m emory performance stems from their difficulty in bind ing new information into complex memories. Similarly, Naveh-Benjamin (2000) has proposed an a ssociative deficit hypothe sis that states that older adults have poorer memory because of a de ficiency in creating and retrieving links between single units of information. The basic units can be two items, an item and its context, two contextual elements, or more generally, th e representation of two mental codes. Age differences in associative memory appear quite consistent. For example, in comparison with their young counterparts, norm al older adults experience difficulties in
29 remembering the source of information (Spencer & Raz, 1995) and the temporal order of information (Fabiani & Friedman, 1997; Newman et al., 2001). Older adults also tend to forget specific contextual details such as the case format of stimulus text (Kausler & Puckett, 1980), the color of stimulus material (Park & Puglisi, 1985), or the spatial location of items (Park, Puglisi & Lutz, 1982). Meta-analytic evidence suggests that although frontal dysfunction has been observed in older adults, intentional encoding in structions do not necessarily put healthy older adults at a disadvantage on memory tasks as we see in patients with significant frontal lobe lesion (Verhaeghen et al., 1993) Both content memory and context memory were actually enhanced by intentional encodi ng strategies (Kausler et al., 1985; Kausler & Phillips, 1989; Schmitter-Edgecombe & Simpson, 2001), and age di fferences seem to remain stable across manipulations that affect the ove rall level of performance. Ma nipulations of study time and of the delay between study and test also show the same pattern for young and old adults (Perlmutter et al., 1981). Overall, th e existing literature indicates that ag e difference represent a quantitative rather than a qualitative decrease in associative memory. However, the important question is not onl y whether older adults are less adept in associative memory than young adults, but also wh ether this impairment is disproportionately greater than age-associated impairment in memo ry for items (or content). The literature yields conflicting answers to this question. Some res earch has suggested that normal older adults demonstrate a disproportionately greater deficit in the memory for associative relationships among items compared to their memory for the ite ms themselves (Craik et al., 1990; Chalfonte and Johnson, 1996; Naveh-Benjamin, 2000). In ot her studies, however, differential age effects on associative memory are relatively small (Denne y et al., 1991). A potential moderator of age effects is the type of associative information to be remembered. With the wide variety of
30 contextual details (e.g., perceptual spatial and temporal), it is possible that age differences are not uniform across materials and encoding moda lities. Unfortunately, there is practically no research on this question. Research on age-related changes in the neural substrates of associative memory could bring new insight into the mechan isms of disproportionately diffe rent age-related declines in associative memory. Based on the frontal defic it hypothesis, many studies have concluded that age differences in the probability of associating items with re levant contextual information mainly result from the frontal lobe dysfunction in older people (Spencer & Raz, 1995). This conclusion is based on the idea th at the frontal lobe is in ch arge of strategic information processing, including the implementation of elab orative and organizational schemes to help encode the associative informati on. However, as indicated earlie r, not only functions mediated by frontal lobe system but also by medial tempor al/hippocampus are affected by aging. Most of the literature does not disc uss the relationship between medial temporal/hippocampus dysfunction and the performance in associative memory in hea lthy aging population. According to our literature review, we know that medial temporal/hippocampus has its important role in binding events together by simple temporal contiguity into memory traces and this system works in a rather automatic fashion under incidental conditions. A recent study by Naveh-Benjamin, Hussain, Gu ez, & Bar-On (2003) demonstrated that dividing attention in younger adul ts fails to produce an impair ment of associative memory equivalent to that seen in older adults perfor ming the same task with full attention. Since dividing attention is thought to impair deliberate, elabora tive encoding (e.g., the frontal contribution to learning), this study suggests that frontal impairment itsel f cannot account for the full degree of associative memory impairment in ol der adults and that anot her factor is involved.
31 One possibility is that although frontal lobe functions may be taxed during divided attention procedure, the binding capabilities of the hi ppocampus may not be taxed by this manipulation and may thus provide a means whereby younger a dults can successfully process associative information in the memory task. If the literature reviewed above is correct, older adults may not succeed on the task because of incipient impairment in this binding mechanism. MCI, characterized by substantial medial temporal/ hippocampus atrophy relative to healthy aging, is considered a tr ansitional state between dementia and normal aging. Given the role of the hippocampus system in forming asso ciations or integrati ng stimuli from multiple sensory sources into distinct episodes and events it will be reasonable to expect to see the relatively poor performance on associative memo ry compared to healthy elders. Even if associative information is available to people with MCI, frontal dysfunction, as suggested in some aging literature, might impair their ability to strategically encode information and to oppose familiarity-based false recognition. To date, there is no research that addresses such hypotheses in people with MCI.
32 CHAPTER 2 PURPOSES AND AIMS OF THE PRESENT STUDY The literature review suggests th at associative m emory may se rve as an important gateway for researchers to understand age-related differen ces in episodic memory. Thus far, cognitive mechanisms of age-related differences as well as pathological aging in associative memory are poorly understood. To date, very few studies ha ve dealt with these discrepancies by accounting for, or measuring, possible heterogeneity in th e anatomical substrates (i.e., hippocampus and surrounding area) thought to underl ie associative memory in aging. It is unknown whether pathological aging is characterized by disproportional impairment on associative memory (context memory) compared to item memory (content memory), since only item memory has been widely investigated across both cross-sectional a nd longitudinal studies. Moreover, it is unclear whether age effects on context recall are larger than those on r ecognition of contextual information, as is commonly found in memory for cont ent. There is also a lack of data regarding whether age differences in memory for context ar e affected by intentiona l encoding instructions when context becomes another target and compet es with content for attention resources. Correlating behavioral performance on experimental tasks of associative memory with clinical neuropsychological measures as well as anat omic measures of hippocampus and surrounding area will provide important data to the associat ive memory literature, and to the memory and aging literature. Moreover, it can potentially yield new behavioral mark ers that will become useful for early detection of the pathological aging population. This current study had several sp ecific aims. First, we aimed to replicate the age-related declines in associative memory in other studies as well as to investigate whether patients with MCI showed impairment in associative memory co mpared to healthy older individuals. We also intended to determine whether such impairment was disproportionately impaired compared to
33 item memory. The performance of normal young i ndividuals, normal aging individuals, and individuals with MCI on tasks of different types of associa tive memory was investigated. Specifically, performances on tasks investigatin g association between items of the same kind, association between different kinds of information, and relational a ssociation were investigated. In the tasks related to associa tion between different kinds of in formation, we intended to include both verbal and visual (e.g., faces, objects, a nd spatial information), recent and remote information, as well as temporal information. We hypothesized that increasing associative memory impairment, particularly the associations between different kind s of information, would be found in the mild cognitive impairment popula tion. In addition, we intended to investigate the effect of delayed recall on the associative memory, which was assumed to involve hippocampus function. Second, we aimed to clarify the degree to wh ich associative memory for novel and familiar or semantically related information depends on medial temporal lobe stru ctures in a continuum of memory impairment represented by normal h ealthy aging and individuals with MCI. We hypothesized that performance on no vel rather than semantically related association produced greater difference between MIC individuals and normal healthy aging individuals based on the widely-accepted view that the hippocampus was not needed for act ivation and retrieval of preexisting (long-term) knowledge. Third, we wanted to know whether age eff ects on associative information recall were larger than those on recognition. We expected larg er age differences in fr ee recall, a smaller one in cued recall, and stil l a smaller one in recognition. Fourt h, we wanted to investigate whether performance on associative memory was aff ected by intentional encoding instructions, particularly when considering both frontal lobe functioning. Frontal lobe functioning is
34 associated with strategic information processing, while medial temporal functioning is thought to establish associations in an automatic way. Finally, we aimed to determine the degree to which associative memory depended on medial temporal lobe functioni ng and frontal lobe functioning in a sense of considering the possible heterogeneity among individuals. We ther efore would evaluate our data based on the Glisky et al. (1995) frontal-temporal classification scheme in both individual and group analyses.
35 CHAPTER 3 METHODOLOGY Participants Participants were 20 younger adults and 35 elderl y individuals recruited from north-central Florida or from standing populations that ex ist through the Department of Neurology Memory Clinic at the University of Florida. Inclusi on criteria were an age range of 18 to 25 for the younger individuals and age 60 for the elderly individuals. Exclusion criteria include: 1) past history of central nervous system disease (e.g., cerebrovascular accident, Parkinsons disease or Alzheimers disease), 2) history of head injury with loss of c onsciousness, 3) history of selfreported substance use with adverse social, medica l, or occupational consequences, 4) history of significant psychiatric illness re quiring hospitalization, 5) histor y of recent heart attack (i.e., within 6 months of participation), 6) history of learning disability or developmental disability. The mean age of the younger adults was 20.30 years with a standard deviation of 1.56 years. Mean education level was 13.70 years with a standard deviation of 1.26 years. Nine participants were male and eleven were female. Seventeen of the participants were Caucasian (85%) and three were African-American (15%). The mean age of the elderly individuals was 73.91 years with a standard deviation of 6.90 years. Mean education level was 15.23 years with a standard deviation of 2.40 years. Fourt een participants were female and 21 were male. All of the participants were Caucasian. Mild Cognitive Impairment Participants Diagnosis of MCI was made based on a for mal consensus panel comprised of 11 professional raters including two board-certified neuropsychologists, one expert in cognitive aging, two post-doctoral fellows, and six graduate students in the Neuropsychology, Neurorehabilitation, and Clinical Neuroscien ce area of concentration in the APA-accredited
36 clinical psychology doctoral program at the Univer sity of Florida. The consensus conference was held after all data were collected, and thus diagnosis of MCI was made retrospectively. During the consensus conference, each rater was b linded to participant identity but was presented with demographic (age, gender, educational leve l) and neuropsychological test data for each elderly participant. Test data was presented in th e form of both raw and st andard scores (age and education corrected where available) on all ps ychological and clinic al neuropsychological measures administered during the participants la boratory sessions. Additi onally, all raters were aware of, and were presented with the published criteria for Amnestic MCI and other subclassifications of MCI (Pet ersen et al., 1999, 2001). The criteria for Amnestic MCI-single or multiple domains included 1) Evidence of a subjective memory impairment which was demonstrated through use of the MAC-Q (Crook, Feher, & Larrabee, 1992) and through the Clinical Dementia Rating Scale (CDR) (Morri s, 1993), 2) evidence of an objective memory impairment demonstrated through performance on the California Verbal Learning Test-Second Edition (CVLT-II, Delis et al., 2000) and the Wech sler Memory Scale (Logical Memory I & II, Visual Reproduction I & II, Verbal Pair Associat es I & II, Visual Pair Associates, Wechsler, 1997), 3) Evidence of intact (for Amnesic MCI-single domain subtype) or impaired (for Amnesic MCI -multiple domains subtype) non-memory cognitive function demonstrated through performance on non-memory measures including th e Wechsler Abbreviated Scale of Intelligence (WASI), language tests (e.g., Boston Naming test Verbal Fluency, Controlled Oral Word Association), attention tests (e.g., Trail Making Test and Digit Span), 4) No evidence of impairment in Activities of Daily Living (ADL), demonstrated through the CDR (Morris, 1993), and 5) No evidence of dementia which was demonstrated through a score > 24 on the Mini Mental State Examination (MMSE, Folstein, 1975), and a CDR rating < 0.5. Criteria for the
37 diagnosis of probable Alzheimers disease based on NINCDS-ADRDA criteria were also available for consensus conference participants. These included: 1) Dementia, established by clinical examination and documented by MMSE and Clinical Dementia Rating Scale, and confirmed by neuropsychological tests, 2) Defi cits in two or more areas of cognition, 3) Progressive worsening of memory or other cognitive function, 4) No disturbance of consciousness, 5) onset between 40 and 90 years of age. The NINCDS-ADRDA criteria were provided to give participants specific information about th e upper boundary between MCI and frank dementia. After viewing the data, each rate r voted whether the participant qualified as: 1) healthy control, 2) MCI, or 3) probable AD. Consensus was established when the majority of the raters agreed. If there were more than two di ssenting opinions, the dissenters explained the reasoning behind their vote, discussion ensued, and the voting was repeated until consensus was achieved. The conference resulted in a diagnosis of four Amne sic MCI-single domain, seven Amnesic MCI-multiple domain, and one Non-Amnesi c MCI of the 35 elderly participants. The age range of the MCI participants, which included all subtypes of MCI, was 66 to 88 years. The mean age was 77.50 years (SD=6.43), and mean education level was 15.42 years (SD= 1.83). Four of the participants were female and eight were male. All 12 participants were Caucasian. Because the main focus of the current study wa s to compare Amnestic MCI with normal aging, the one individual who was classified as NonAmnesic MCI during the conference was excluded from the subsequent three-group (young, normal old control, and MCI) data analyses. The remaining 11 MCI individuals with diagnosis of either Amnestic MCI-single domain or Amnestic MCI-multiple domains were combined into a MCI group for subsequent data analyses.
38 The final MCI group had a mean age of 77.73 year s (SD = 4.90) and mean education level of 15.73 years (SD =1.56). Three of the participants were female and eight were male. Healthy Older Control Participants As a result o f the consensus conference, 23 el derly individuals were classified as healthy elderly controls without memory difficulty. Compared to MCI individuals, the healthy elderly group was significantly younger [t (32) = -2.24, p<.05]. Therefore, 20 among the 23 healthy older individuals who were older than the age of 65 years were then selected to serve as a comparison group for the 11 MCI in dividuals to match the age, education, sex, scores on the Mini-Mental Status Examination (MMSE), and WA SI Full Scale IQ scores of the two groups. The age range of the final healthy older control group was 65 to 82 years, with a mean age of 73.95 (SD =6.25). The mean education level was 15.50 years (SD =2.67). Eight were females and twelve were males. A ll twenty were Caucasian. Table 3-1 shows the demographic information for the full older sample and each group. Compared to normal older and younger control, th e MCI group demonstrated significantly lower scores on the MMSE [F (2,48) =11.80, P<.001] but had equivale nt Full-Scaled IQ scores (p>.05). The younger group demonstrated a lowe r educational level as a group compared to both normal old control and MCI groups (p>.05). Table 3-1. Mean characteristics of groups based on cognitive status Total older sample (n=35) Total NonMCI individuels (n=23) Healthy elderly control (n=20) MCI (n=11) Younger control (n=20) Variable M (SD) M (SD) M (SD) M (SD) M (SD) Age 73.91 (6.90) 72.39 (7.13) 73.95 (6.25) 77.73 (4.90) 20.30 (1.56) Education 15.23 (2.40) 15.13 (2.69) 15.50 (2.67) 15.72 (1.56) 13.70 (1.26) MMSE 28.94 (1.42) 29.26 (0.81) 29.40 (0.68) 28.27 (1.42) 29.80 (0.52) FSIQ 1 15.28 (13.74) 118.26 (12.42) 120.30 (11.52) 110.38 (13.11) 118.40 (7.81) Gender (F/M) 14/21 10/13 8/12 3/8 11/9
39 Differential Contributions of the F rontal and Medial Temporal lobes Each study participant was assi gned two scores based on the wo rk of Glisky, et al.(1995), one representing relative performa nce on a group of tests associat ed with frontal lobe (FL) executive function and the other representing re lative performance on a gr oup of tests thought to represent memory-relevant aspects of medial temporal lobe (MTL) function. The tests contributing on the FL factor included: 1) numbe r of categories achieved on the Wisconsin Card Sorting Test, 2) the total number of words generated in a word fluency test, using initial letters F, A, and S (Spreen & Benton, 1997), 3) Mental Arithm etic from the Wechsler Adult Intelligence Scale-Revised (Wechsler, 1987), 4) Mental Control from WMS-R, and 5) Backward Digit Span from the WMS-R. The tests contributing to the MTL factor included: 1) Logical Memory I from the WMS-R, 2) Long-Delay Cued Recall measure fr om the CVLT (Delis et al., 1987), 3) Visual Paired Associates II from the WMS-R, 4) Verbal Paired Associates I from the WMS-R. In this present study, we used the latest version of thes e tests. For example, we used the Wechsler Memory Scale-third version (WMS-III) instead of the Wechsler Memory Scale-Revised version (WMS-R). All elderly participants perf ormance, including the one Non-Amnestic MCI individuals, on each of these tests was converted to a z score based on available norms, such as Heaton norms, WMS-III norms, or the norms establis hed by Glisky et al. (1995), wherever was available. The z scores were then averaged ac ross the tests contributing to each factor. The resulting composite z scores thus represented an individual participants relative performance on each factor. Based on these procedures describe d above, each individual was then assigned into one of the four different subgroups based on their scores on the tw o factors were above or below the group composite means. The resulting distribu tion of composite z scores on the frontal lobe factor was approximately normal with a mean z score of .06 (SD= .65); 16 of the 35 participants
40 had scores above the mean. The distribution of scores on the medial temp oral lobe factor was somewhat negatively skewed with a mean z score of .24 (SD= .71) and 19 participants scored above the mean. Figure 3-1 displa ys the frequency distribution of composite frontal lobe scores and Figure 3-2 shows the frequency distribution of composite tempor al lobe scores for the total elderly sample. Figure 3-1. Frequency dist ribution of medial tem poral lobe composite scores for total elderly sample (n=35) Figure 3-2. Frequency distributi on of frontal lobe composite scores for total elderly sample (n=35) Medial Temporal Lobe Composite Score1.50 1.00 0.50 0.00 -0.50 -1.00 -1.50 Frequency12 10 8 6 4 2 0 Frontal Lobe Composite Score2.00 1.50 1.00 0.50 0.00 -0.50 -1.00 -1.50 -2.00 Frequency6 4 2 0
41 Classifying participants according to their sc ores on the two composite measures yielded a distribution of twelve who were above the m ean on both factors, ten who were below on the mean on both factors, four who were above the m ean on the FL factor and below the mean on the MTL factor, and nine who were above the mean on the MTL factor and below the mean on the FL factor. Characteristics of the four groups ar e presented in Table 3-2. Table 3-3 indicates the actual number of people in each of four groups who were previous identified as normal healthy old individuals, Amnestic MCI, or Non-Amnestic MCI. Separate one-way between-subject analyses of variance (ANOVAs) indi cated that there were no diffe rences in age [F (3,31) = .98, P >.05)], education[F(3,31) = 1.87, P >.05)], or scores on the MMSE [F(3,31) = 2.46, P >.05)] among the four groups. However, the Full Scale IQ scores were significantly higher in the HighMTL-High FL group when compared to the Low MTL-Low FL group [t (17) = -3.35, p <.005]; no other group differences involving FSIQ were found. The High and Low FL groups differed significantly on their com posite FL scores [t (33)= -6.74, p < .001] but not on their MTL scores (p >.05), whereas the High and Low MTL groups differed on MTL scores [t (33)= -8.58, p < .001] but not on FL scores ( p>.05). Table 3-2. Mean characteristic s of groups based on frontal l obe (FL) function and medial temporal lobe (MTL) function High FL function Low FL function High MTL function (n=12) Low MTL function (n=4) High MTL function (n=9) Low MTL function (n=10) Variable M SD M SD M SD M SD Age 72.83 5.67 78.00 3.74 71.89 8.42 75.40 7.57 Education 15.33 2.87 17.50 1.91 14.22 2.11 15.10 1.79 MMSE 29.33 0.65 29.00 0.82 29.22 0.83 28.20 1.55 FSIQ 122.67 10.33 122.50 11.70 110.00 14.14 105.29 11.91 FL score 0.62 0.50 0.75 0.63 -0.24 0.31 -0.60 0.48 MTL score 0.76 0.31 -0.03 0.20 0.57 0.26 -0.58 0.45
42 Table 3-3. Numbers of people cl assified as normal old or MCI in each of the four groups Low FL High FL Low MTL 3 Normal old 2 Amnestic MCI-single domain 5 Amnestic MCI-multiple domains 1 Normal old 2 Amnestic MCI-single domain 1 Amnestic MCI-multiple domains High MTL 7 Normal old 1 Amnestic MCI-single domain 1 Non-Amnestic MCI 12 Normal old Data analyses described below were base d on one of the two described methods of grouping subjects. In some analyses, three c linical groups (younger a dults, healthy elderly, MCI) were used. In others, the four groups re sulting from the Glisky, et al. (1995) approach (i.e., High MTL-High FL, Low MTL-High FL, High MTL-Low FL, and Low MTL-Low FL) were used. General Procedure All par ticipants were tested in a quiet clinic or laboratory room at the Department of Clinical Psychology at the University of Florid a. Each person completed the standardized clinical neuropsychological measures first. A re turn visit was scheduled for completion of the experimental tasks. Each participant provided informed consent, and was given the opportunity to decline participation prior to, during, or after administrati on of the neuropsychological and experimental measures. Participants were reim bursed 10 dollars per hour for their participation. Standardized Clinical Ne uropsychological Measures Participants underwent a battery of neuropsyc hological tests that 1) broadly sam pled a variety of cognitive domains and brain regions and 2) emphasized those cognitive domains (e.g., memory and executive functioning) that are se nsitive to the earliest change in MCI. Measurement of non-memory domains provided th e basis for ruling out (or in) alternative
43 hypotheses about the basis of early memory dysf unction in preclinical dementia. Whenever possible, normative standards that corrected for demographic characteri stics (e.g., age, gender, education level, and race) were used. Neurops ychological tests administrated are outlined in Table 3-4, which provides a reference for each and a description of the cognitive construct/domain that each te st was designed to measure. Table 3-4. Neuropsychological measur es included in the present study Overall construct Test name Description of test measure Dependent variables recorded General mental status Mini Mental State Exam (Folstein et al., 1975) A brief screening of mental status Total possible score=30 Observer/ interview ratings Clinical Dementia Rating Scale (Morris, 1993) Disease severity rating: includes informant & participants subjective/objective elements CDR rating (0-3); collected during semistructured interview Intellectual functioning WASI: Vocabulary, Similarities, Block Design, and Matrix Reasoning (Wechsler, 1999) Together, these were used to predict general intelligence, verbal and performance abilities. Predicted FSIQ, VIQ, and PIQ; Voc. & Info. Wechsler Memory ScaleIII (Logical memory I & II; Visual Reproduction I & II; Verbal paired associates I & II) (Wechsler, 1997) Measure of verbal and visual/figure memory Scores for immediate and delayed recall California Verbal Learning Test2nd Edition (Delis et al., 2000) Verbal memory test that assesses learning strategy, immediate and delayed recall, recognition, interference, and errors Slop of learning trails, savings score, discriminability Index Memory functioning Wechsler Memory Scale-R (Visual Paired Associates I & II, Wechsler, 1987) Measure of paired verbal and visual memory Total correct score Boston Naming Test-2nd Edition (Goodglass & Kaplan, 2001) Confrontation naming using large ink drawings Total correct score Controlled Oral Word Association (COWA, Spreen & Benton, 1997) Verbal fluency to alphabet letter Total correct examples Language functioning Category Fluency-Animals (Tombaugh et al., 1999) Verbal fluency to a semantic category Total correct examples
44 Table 3-4. Continued Overall construct Test name Description of test measure Dependent variables recorded WASI-Block Design (Wechsler, 1999) Visuoconstruction measure requiring reproduction of two dimensional designs with blocks Total score based on time limits Visuoperceptual/ visuoconstruction Test of Facial Recognition (Benton & Van Allen, 1986) Visual perception using facial matching paradigm Total correct score Frontal/executive monitoring functioning WAIS-III Digit Span, (Wechsler, 1997) Auditory attention span Total correct score and longest digits Attention/ concentration WMS-III Spatial Span Visual attention span Total correct score and longest digits WAIS-III Letter-Number Sequencing Ability to hold auditory information for a period of time and mental manipulation or numbers and letters Total correct score Working memory WMS-III Mental Control Ability to manipulate familiar information Total correct score Psychomotor speed Trail Making A & B Tests (Reitan, 1958) Visuomotor speed, mental shifting between numbers and letters Total time to completion; number of errors WASI-Similarities (Wechsler, 1999) Abstract thinking; requiring participant to tell how two words are alike Total correct score WASIMatrix Reasoning (Wechsler, 1999) Measure of visual reasoning ability Total correct score Abstract thinking/ problem solving Wisconsin Card Sorting Test Participants sort cards based on their own principle; measure of mental flexibility and problem solving. Number of categories achieved; errors; number to finish first category Memory Assessment Clinic Questionnaire (MAC-Q; Crook et a., 1992) Subjective memory complaints questionnaire Total possible score=30, score >=25 as a cutoff. Lawton IADL scale (Lawton & Brody, 1969) IADL; informant and participant self-reports on IADL abilities. Sum of endorsed items from both participant and informant. Geriatric Depression Scale (Sheikh & Yesavage, 1986) 30item self-evaluation questionnaire assessing elements of depression Total number of items endorsed. Total possible=24, conventional cutoff =10 Beck Depression Scale-II (Beck, Steer, & Brown, 1996 ) 21item self-evaluation questionnaire assessing elements of depression Total number of items endorsed. Self-reported measures State Trait Anxiety Inventory (STAI; Speilberger, 1968) 40-item self-evaluation questionnaire assessing state and trait anxiety Total number of items endorsed
45 Neuropsychological and Questionnaire Data Tables 3-5 and 3-6 show m ean performan ce on neuropsychological measures for all participants. The data for th e total group of older individu als were also included. Raw neuropsychological scores were converted to standard scores using normative data appropriate for the participants whenever possible. Standard sc ores or raw scores if standard scores were not available for non-memory neuropsychological measures are present in Table 3-5, standard scores or raw scores if standard scores were not avai lable for memory measures are presented in Table 3-6, and standard or raw scores for mood and memory questionnaires are presented in Table 3-7. ANOVAs were conducted between young, normal old control, and MCI using standard scores whenever possible followed by post-hoc t-tests. Compared to normal old control group, the MCI group demonstrated significantly lower scores on the WMS-III spatial span test, Trails A & B, number of category completed on the Wisconsin Card Sorting Test (WCST), Boston Naming Test, and animal fluency. The normal older control group completed fewer categories on the WCST test when compared to younger contro ls. Compared to nor mal old individuals, individuals with MCI de monstrated significantly lower scores on most of the memory measures, including Logical Memory I & II, Verbal Paired Associates II, Visual Reproduction I & II, CVLT-2 total recall, CVLT-2 short delayed free recall, and CVLT-2 long delayed free recall, which was not surprising given that these measures were used to classify individuals as MCI. Healthy old controls demonstrated significantly lower scores on Visu al Paired Associates I when compared to younger controls. MCI subjects had lower CDR total sc ores than their healthy older counterparts.
46 Table 3-5. Means and standard deviations for neuropsychological non-memory measures Total older adult sample (n=35) MCI (n=11) Normal old control (n=20) Younger control (n=20) Measure M SD M SD M SD M SD Digit Span total (AcSS) 11.14 2.88 10 .27 1.56 11.85 3.41 11.00 2.99 Spatial Span total (AcSS) 12.06 2.96 10.67* 1.87 13.45 2.67 12.65 1.98 Trails A (T-score) 45.51 12.03 36.73** 15.78 49.70 7.24 48.15 10.99 Trails B (T-score) 48.83 14.25 36.45** 15.23 55.50 10.15 51.70 7.65 Mental Control (AcSS) 12.11 2.74 11.73 3.61 12.90 2.00 11.20 1.85 Letter-Number Sequencing (AcSS) 10.71 2.27 9.82 2.14 11.50 2.06 12.25 3.02 Arithmetic (AcSS) 12.69 2.37 11.81 2.64 13.50 1.88 12.45 2.50 WCST-category Completed (raw) 3.86 1.91 2.82** 1.78 4.55# 1.73 5.90 0.45 BFT (short version, raw) 21.39 2.54 19.33 3.30 22.05 1.96 23.70 1.81 BNT (T-score) 59.89 11.68 51.27* 8.39 64.75 9.80 50.80 9.20 COWA total (T-score) 50.49 10.76 48.45 9.81 52.55 11.68 51.35 10.61 Animal Fluency (Tscore) 49.14 11.49 43.27* 11.69 53.65 9.80 51.35 10.74 Note: BFT=Benton Facial Recognition Test; BNT=Bo ston Naming Test; COWA= Controlled Oral Word Association; AcSS= Age-Corrected Scaled Score. ANOVAs co nducted between groups using standard scores followed by t-tests post-hoc comparisons. p<.05 for comparisons between MCI and old control group. ** p < 05 for comparisons between MCI and the other two control groups. # p<.05 for comparisons between old and young control groups. Table 3-6. Means and standard deviations for neuropsychological memory measures Total older adult sample (n=35) MCI (n=11) Normal old control (n=20) Younger control (n=20) Measure M SD M SD M SD M SD LM I (AcSS) 11.97 2.86 9.18** 2.40 13.05 2.16 12.05 2.21 LM II (AcSS) 12.26 2.91 9.55** 2.62 13.35 2.27 12.45 2.28 Verbal PA I (AcSS) 10.51 2.70 9.27^ 2.76 11.25 2.69 13.05 2.28 Verbal PA II (AcSS) 11.49 2.81 9.64** 2.80 12.25 2.55 11.95 0.22 VR I (AcSS) 10.86 3.23 8.36** 2.91 12.30 2.62 11.65 2.16 VR II (AcSS) 12.66 3.39 9.82** 2.75 14.55 2.58 12.35 3.08 Visual PA I (raw) 13.38 4.71 9.09^ 4.13 12.05# 4.74 17.35 1.04 Visual PA II (raw) 5.16 1.43 4.00^ 1.67 5.10 1.48 6.00 0.00 CVLT2-total recall (Tscore) 53.23 10.06 43.91** 8.90 58.15 7.46 60.60 7.48 CVLT2-SDFR (Z-score) 0.03 1.28 -1.18** 0.72 0.65 1.08 0.68 0.82 CVLT2-LDFR (Z-score) -0.20 1.18 -1.41** 0.83 0.43 0.88 0.53 0.85 CVLT2-recognition d (Z-score) -0.37 0.83 -0.86** 0.32 -0.50 0.94 0.33 0.54 Note: LM= Logical Memory; Verbal PA= Verbal Paired Associates; VR= Visual Reproduction; Visual PA= Visual Paired Associates; SDFR= short delayed free recall; LDFR =long delayed free recall; AcSS= Age-Corrected Scaled Score. All ANOVAs conducted between groups using standard scores, except for visual paired associated I & II tests in which raw scores were used, followed by t-tests po st-hoc comparisons. ** p < 05 for comparisons between MCI and the other two control groups. # p<.05 for comparisons between old and young control groups. ^ p<.05 for comparisons between MCI and young control groups.
47 Table 3-7. Means and standard deviations for subjective mood and memory questionnaires Total older adult sample (n=35) MCI (n=11) Normal old control (n=20) Younger control (n=20) Measure M SD M SD M SD M SD BDI-2 (raw score) ------------5.15 4.34 GDS (raw score) 3.09 2.92 3.60 3.34 2.85 2.70 ----STAI-state (z score) -0.49 0.72 -0.19 1.15 -0.48 0.60 -0.65 0.58 STAI-trait (z score) -0.51 0.78 -0.51 0.84 -0.78 0.56 -0.26 0.90 MAC-Q (raw score) 22.94 5.09 25.36^ 4.65 23.30 5.58 19.92 3.80 CDR (total score) 0.21 0.25 0.50* 0.00 0.08 0.18 ----Note: BFT-Benton Facial Recognition Test; BNT=Bo ston Naming Test; COWA= Controlled Oral Word Association. ANOVAs conducted between groups using standard scores followed by t-tests post-hoc comparisons. p<.05 for comparisons between MCI and old control group. ^ p<.05 for comparisons between MCI and young control groups.
48 CHAPTER 4 THE EFFECT OF INCIDENTALVERSUS INTENTIONAL LEARNING ON THE RETRIEVAL OF AS SOCIATION INFORMATION Introduction The purpose of this experim ent was to examine the effect of incidental and intentional learning on the associative memory performance in healthy young (HY), healthy older (HO) individuals, and individu als with MCI. Based on the literatu re, we hypothesized that the medial temporal lobe (or more specifically, hippocam pus) is important for the encoding of novel information (both item and associative informa tion) into memory and that such a process proceeds in an automatic, rather than in a deli berate or controlled way. In contrast, neural systems of the frontal lobe have been rec ognized as important for controlled, strategic information processing, including the implementati on of elaborative and organizational schemes to help encode to-be-remembered information. A generation of research in cognitive psychology has established that intentional encoding often produces better memory performance compared to incidental encoding in both young and healt hy old individuals (Che rry & Park, 1993). However, results in normal aging studies are mo re mixed than in young people, in which giving intentional instruction to older people either has no effect or improves older adults memory performance (Kausler et al., 1985; Kausler & Phillips, 1989; Schmitter-Edgecombe & Simpson, 2001). One interesting question was whether disrup tion of hippocampal function would decrease the benefit derived from receiving intentional in struction in terms of encoding of associative information. Most studies of this issue have been based on studying item-based information rather than associative information. For example, we know that H.M.s memory performance for items did not differ under incide ntal and intentional learning conditions (Smith, 1988), despite
49 extensive bilateral hippocampal damage. Whether similar results could be obtained in learning associative information has yet to be established. In this experiment, we manipulated the en coding condition (encoding items or pairs) and the recognition tasks (item vs. pairs) to examine the effect of incident al versus intentional learning on the retrieval of associative informati on. We also examined the individual differences in neuropsychological correlates of temporal and frontal lobe functioning and how the differences could affect the perf ormance on the specific tasks. Methods Participants There were 51 participants, of which 20 were younger, 20 were healthy older and 11 were MCI individuals. The m ean age and number of y ears of education appear in Table 3-1. For the second types of analyses based on the individuals frontal lobe and me dial temporal lobe function level, 35 older adults were included. Th e details of demographic information and factor scores for each of the four groups can be found in Chapter 3. Design The design of this experim ent is a 3 (Gr oup; healthy young [HY], healthy old [HO], mild cognitive impairment [MCI]) 2 (Encoding Conditio n: items vs. pairs) 2 (Test Type: Items, Pairs) Repeated Measures Analysis of Vari ance. Group and Study Instructions are betweensubject variables, while Test Type was a w ithin-subject variable. Recognition memory was tested immediately following the l earning phase as well as after a 30 -minute delay for all tests. The design results in two conditions in which encoding and recognition are matched (Encode ItemTest Item, Encode Pair-Test Pair) and tw o conditions in which they are mismatched (Encode Item-Test Pair, Encode Pair-Test Item).
50 Materials and Procedure Stim uli comprised from the Nelson Norms of Free Association (Nel son et al., 1998) and all of the words were high-frequency nouns. A tota l of 96 words were selected and 48 of them were used to create two lists of 12 novel word pa irs. Pairing was done so that pairs reflected no semantic relationship. The association strength of the novel word pairs were very low or zero and therefore thought to be novel and unique, not accompanied by a pre-existing memory representation. Words belonging to different pairs were also unr elated to each other in any apparent way. Additionally, these word pairs were orthographically and phonemically unlike (avoiding perceptually and phonemically similar pairs, e.g., bookgood). Within each list, the first and second word of each pair was balanced for length, frequency, and concreteness. Word pairs were also matched across lists for lengt h, frequency, and concreteness. The remaining 48 words were used as foil trials for the immediat e and delayed item recognition tests. In both learning conditions, recombinati on of pairs for the pair reco gnition tests was done randomly, with the constraint that no resulting pair ha d any obvious orthographic, phonemic, or semantic relationship. All participants were given th e incidental learning condition first, and then at least 45 minutes later, the intentional learning conditi on was given. The list us ed for intentional or incidental pair learning was counterbalanced between subjects w ith half of the participants using one list as the incidental list and half of the partic ipants using the same list as the intentional list. All the stimuli were presented visually on a computer screen. The procedures for the incidental and intenti onal learning conditions we re identical except for the instructions during the study phase. In the incidental condition, partic ipants were told to study the words in each pair indi vidually (items) in preparatio n for an upcoming item recognition
51 test. In the intentional condition, participants were instructed to study the pairs in preparation for an upcoming paired associ ated recognition test. After participants questions were answered, the experiment al phase started, in which participants were presented w ith the 12 experimental pairs vi sually on a computer, two words (one pair) at a time with a 400 ms pause between pairs. On each trial, the word pair was presented simultaneously in a horizontal left-right format. Presentation rate for each list was selfpaced with a limit of seven seconds per pair Each participant was given two learning trials in order to prevent floor effect and participants were asked to free recall words after each trial. After the end of the study pha se, the item and pair recognition tests listed below were administered to all of the participants with the order of item recognition test first and then followed by the pair recognition test. 1) Item recognition test: In this test, 12 ta rgets and 12 distracters were presented to participants one at a time. Part icipants were asked to give Yes/No responses to indicate if the word was one that appeared on the study list. Half of the targets were the words that appeared on the right side of the computer screen during the study phase, and ha lf were words that appeared on the left side. No two words from the same pair were used as targets in the test. 2) Pair recognition test: In this test, 12 intact (targets) pairs from the study phase (i.e., the two words that appeared together on the comput er screen in the study phase) and 12 recombined (distracters) pairs (i.e., they were composed of words taken fr om different study pairs) were presented. Participants were told that all of th e items appeared in the study phase and that their task was to successfully identify those same pairs that appeared in the study phase. Participants had as much time as they needed to complete the two memory tests.
52 After a 30-minute delay, all participants were asked to perform the free recall, item recognition, and pair recognition te sts in order to test their lo ng-term memory. Similar to immediate recall condition, participants had as mu ch time as they needed to complete these memory tests. Hypotheses and Predictions We predicted that the three groups (HY, HO, and MCI) would have com parable performance on the item memory (i.e., single word recognition) in both inte ntional and incidental encoding of association information conditions because item recognition could be achieved through familiarity recognition without heavy reliance on the hippocampus. On the associative memory tests, we pr edicted that the HO group would demonstrate slightly poorer performance on the incidental learning condition and th at giving intentional instructions would either ha ve no effect or slightly improve the HO groups memory for associative information. The pa ttern of improvement in the pe rformance, if there was any, should be similar between the HY and the HO groups. The memory decline over time (from immediate to delayed testing) should be minimal in the HO group. Similarly, comparing the MCI group w ith the HO group, the MCI group should demonstrate poorer performance in learning associ ative information in both manipulated learning conditions due to compromise in hippocampus function, and the intentional instruction would not improve their performance to a significan t degree. The MCI group would demonstrate significantly reduced recall for both item and a ssociative information in the delayed recall condition when compared to the HO group. A second set of analyses was conducted usi ng the groupings derived from Glisky et al. (1995) frontal and temporal lobe factors. These analyses used a 4 (Group) 2 (Encoding Condition: items or pairs) 2 (T est Type: items, pairs) mixed factorial ANOVA. We predicted
53 that High MTL-High FL group would demonstrate the best performance on associative information compared to other groups for both le arning conditions, and performance on the two conditions would be equivalent Similarly, the High MTL-Low FL group should demonstrate good associative memory in both learning conditi ons, although a slight increase in performance in the intentional condition was expected. In contrast, the Low MTL-Low FL group should show poor associative memory on both lear ning conditions and performance on the two conditions should be equivalent. The Low MT L-High FL group would show poor associative memory on both conditions; however, they would show slightly better or equal performance in the intentional condition than in the incidental condition. Results For all analyses described in th e current study, the leve l of statistical sign ificance was set at p= .05. Tests for sphericity were carried out in each analysis Huynh-Feldt or Greenhouse and Geisser co rrected significance levels were reported for any effect for which the sphericity test was significant. Instead of runni ng omnibus post-hoc tests, two pl anned contrast analyses were carried out on significant main effects of th e between-subject variab le (groups) and their interactions with time of test (immediate v. dela y). These two planned contrast analyses served to accomplish two goals: 1) to examine differe nces between young adult group and normal older adults group, and 2) to compare differences between normal older adults and individuals with MCI. Similar analyses were also conducted by dividing individuals into four groups based on their FL-MTL function levels. Analyses on the HY, HO, and MCI Groups Table 4-1 describes the data for the two m a nipulated learning conditions in the three groups. A Group Learning Condition (immediate item recognition on the incidental and intentional conditions) repeated ANOVA was conducted to examine the item memory. A Group
54 Learning Condition (incidental an d intentional conditions) Measur es (pair hits and pair false positive) Time (immediate and delayed re call) repeated ANOVA was conducted to examine the effect of encoding instruction on the associative memory. Table 4-1. Data for the incide ntal and intentional learning conditions in three groups. MCI (n=11) Normal old control (n=20) Younger control (n=20) Measure M SD M SD M SD Incidental learning Immediate recall Learning trial 1 3.09 1.58 6.80 3.38 11.95 3.49 Learning trial 2 5.90 2.88 10.9 4.47 18.15 3.48 Item recognition hits 11.10 0.88 10.65 1.27 11.45 0.94 Item recognition-false positives 3.30 4.67 0.45 0.60 0.05 0.22 Item recognitiontotal 19.80 4.26 22.20 1.64 23.40 0.99 Pair association recognition-hits 9.60 1.96 9.85 1.57 11.45 0.10 Pair association recognition-false positives (recombined pairs) 7.30 2.54 4.75 2.65 0.55 0.83 Pair association recognition-total 14.30 1.16 17.10 3.51 22.90 1.62 Delayed recall Item free recall 2.70 2.36 8.80 5.21 16.90 5.17 Item recognition hits 9.22 1.30 10.45 1.31 11.80 0.41 Item recognition-false positives 2.78 1.30 0.85 1.18 0.30 0.57 Item recognitiontotal 18.44 1.23 21.60 1.90 23.50 0.61 Pair association recognition-hits 8.67 1.73 9.75 2.38 11.70 0.57 Pair association recognition-false positives (recombined pairs) 4.78 1.64 4.20 2.88 0.60 1.05 Pair association recognition-total 15.89 1.36 17.55 3.99 23.10 1.48 Intentional learning Immediate recall Learning trial 1 3.60 2.63 5.50 3.53 10.35 3.79 Learning trial 2 6.33 2.24 13.20 4.94 19.00 3.21 Item recognition hits 10.00 1.32 11.15 1.04 11.85 0.49 Item recognition-false positives 1.11 1.17 0.35 0.75 0.00 0.00 Item recognitiontotal 20.89 1.76 23.00 1.08 23.85 0.49 Pair association recognition-hits 9.56 1.42 10.95 1.50 11.70 0.57 Pair association recognition-false positives (recombined pairs) 3.44 3.17 1.35 1.95 0.30 0.66 Pair association recognition-total 18.11 3.86 21.60 3.05 23.50 0.76 Delayed recall Item free recall 2.80 2.90 8.00 4.84 15.05 5.38 Item recognition hits 9.00 1.12 10.75 1.41 11.50 0.89 Item recognition-false positives 2.33 1.32 1.45 1.28 0.60 0.82 Item recognitiontotal 18.67 1.50 21.30 2.36 22.90 1.52 Pair association recognition-hits 9.22 1.56 10.55 2.11 11.70 0.57 Pair association recognition-false positives (recombined pairs) 4.89 2.57 1.90 2.15 0.15 0.37 Pair association recognition-total 16.33 3.28 20.65 4.13 23.55 0.76
55 We first conducted a Group x Learning C ondition mixed factorial ANOVA to examine subjects immediate item memory in both learning c onditions. The analysis yielded a significant main effect for Group [F (2, 46)=9.29, p <.001, eta2=.29] and a significant Group Learning Condition interaction effect [F (2,46)=4.17, p <.05, eta2=.15]. Post-hoc planned contrast comparisons (HY vs. HO; HO vs. MCI) were co nducted and revealed that the HY and the HO groups significantly differed in single word reco gnition in intentional [t (47) = 2.351, p< .05] and incidental learning [t (46)=2.40, p< .05] conditio ns, although the difference of the mean numbers of items recognized in the two groups was less than one item. When comparing the HO and the MCI group, the results demonstrated that the two groups did not differ significantly in item memory in the incidental condition [t (47)=-1.08, p>.05]. However, in the intentional learning condition, the MCI group demonstrated a signifi cantly lower score on the item recognition [t (46)=3.11, p<.005]. Figure 4-1 demons trates the data for the item memory for the three groups. Figure 4-1. Data of the immediate item recognition hits for the three groups. Due to the differences in item memory betw een groups, the subsequent analyses in the associative memory were conducted by entering the immediate incidental item hits as a covariate. A Group Measures (pair hits and false positives) Learning Condition (intentional Item memroy: immediate item hits 0 2 4 6 8 10 12 14 MCIHOHYgroup item hit incidental learning intentional learning
56 vs. incidental) Time (immediate and de layed) repeated ANOVA was conducted and the analysis yielded a significant main e ffect for Group [F (2,45)=6.13, p < .01, eta2=.21], and significant Group Learning Condi tion [F (2,45)=4.03, p <.05, eta2=.15], Group Measure [F (2,45)=18.74, p <.001, eta2=.45], Group Time Learning Condition [F (2,45)=3.32, p <.05, eta2=.13], and Group Learning Condition Measure [F (2,45)=9.26, p <.001, eta2=.29] interaction effects. Further planned contrast comparisons were conducted on the two th ree-way interaction effects. The analyses between the HY and the HO groups showed that the HO group had significantly lower scores on the hits measure in the incidental learning condition compared to the HY group (p<.05); however, the difference disa ppeared when giving the intentional learning instruction, indicating that the HO group showed improvement on the pair recognition hits when giving intentional instruction wh ile the HY group remained stable and showed high performance with or without the intentional instruction. In the delay recognition condition, the HO group demonstrated significantly lower performance compared to the HY group regardless whether under incidental (p<.005) or intent ional (p<.05) learning condition. In the analyses for the false positive meas ure, the HO group demonstrated higher numbers of false positives in the incidental condition during both immediate (p< .001) and delayed (p <.001) testing when compared to the HY group. However, in the intent ional learning condition, the HO group significantly decreased the numbers of false positive errors during both immediate (p< .005) and delayed (p <.001) test ing situations, and as a result, the differences in terms of numbers of false positive errors previously existed between the HO and HY groups disappeared under the intentional instruction. When comp aring the HO and the MCI groups, the two groups did not differ significantly in the hits scor e under the incidental condition during either
57 immediate or delayed testing situations. Howeve r, when giving specific instructions to encode the pairs (intentional condition), the HO group, but not the MCI group, demonstrated significant improvement (p <.05) in the pair hits score, but only in the imme diate testing situation. The analyses conducted on the false positive errors between the HO a nd the MCI groups showed that the MCI group had significantly higher number of false positive errors in both learning conditions (both p <.05) during the immediat e testing situation, although both groups demonstrated the trend of decreas ing numbers of false positive erro rs in the intentional condition compared to the incidental condition. During th e delayed testing situati on, the two groups did not differ significantly in terms of making false positive errors in the incidental condition; however, the intentional instruction helped th e HO group to decrease numbers of false positive errors significantly (p<.05) while the MCI group made a similar number of false positive errors as in the incidental condition. Th e data for the pair r ecognition hits and fa lse positive score in the two learning conditions at the two time points for the three groups is presented in Figures 4-2 and 4-3. Figure 4-2. Data for the immediate and delayed recognition of the pairs (hits) in the incidental and intentional learning condi tions for the three groups. 7 8 9 10 11 12 MCIHOHY groupcorrect score incidental intentional Immediate Pair Recognition Hits Delayed Pair Recognition Hits7 8 9 10 11 12 MCIHOHY groupcorrect score incidental intentional
58 Figure 4-3. Data for the immediate and delayed recognition false alar ms (false positive) errors of the pairs in the incidental and intentiona l learning conditions for the three groups. Analyses on Groups Divided by MTL-FL Status A second set of analyses was conducted by divi ding individuals into four groups based on their FL-MT L function levels. It was hypothe sized that the High M TL-High FL group would demonstrate the best performance on associative information compared to other groups for both learning conditions, and performance on the two c onditions would be equivalent. Similarly, it was predicted that the High MTL-Low FL group w ould demonstrate good associative memory in both learning conditions, although a slightly increased performan ce in the intentional condition was expected. In contrast, the Low MTL-Lo w FL group should show poor performance of associative information on both learning condi tions and performance on the two conditions should be equivalent. The Low MTL-High FL group would show poor associative memory on both conditions; however, they would show slig htly better or equal performance in the intentional condition than in the incidental condition. Immediate Pair Recognition False Alarms0 1 2 3 4 5 6 7 8 MCI HO HY groupnumber of false positive incidental intentional Delayed Pair Recognition False Alarms0 1 2 3 4 5 6 7 8 MCI HO HY groupnumber of false positive incidental intentional
59 The analyses on item memory demonstrated that the four groups did not differ significantly on the single word recognition (hit s) in either the intentional or the incidental learning condition [F (3,29)=2.35, p >.05, eta2=.20], indicating equivalent item memory. In order to examine associative memory, a Group x Measures (pair hits and false positive) x Learning Condition (intentional vs. incidental) x Time (immed iate and delayed) repeated ANOVA was conducted and the analysis yield significant main effects for Measure [F (1,28)=207.09, p < .001, eta2=.88] and Learning Condition [F (1,28)=25.10, p < .001, eta2=.47], and significant Group x Measure[F (3,28)=8.62, p < .001, eta2=.48], Learning Condition x Measure[F (1,28)=28.86, p < .001, eta2=.51], Time x Learning Condition x Measure [F (1,28)=10.80, p < .005, eta2=.28] interaction effects. These interaction effects were examined further by conducting post-hoc comparisons. The results showed that the four groups did not differ significantly in the pair hits measure in either learning condition during the immediate or delayed testing condition. However, in the false positive measure, all four groups demonstrated a similar degree benefit from the intentional instructions and decreased the number of fals e positive errors during the immediate testing situation. During the delayed tes ting situation, the Low MTL-Low FL group did not benefit from the intentional learning instruction and made a si milar number of false positive errors, while the High MTL-Low FL group showed a significantly lower number of errors (p<.01) and the Low MTL-High FL showed a marginal effect of decreasing number of errors (p=.052) in the intentional condition. The data of the pair r ecognition hits and false pos itive score in the two learning condition at the two time points for the four groups divided by th eir medial temporal lobe and frontal lobe function are pr esented in Figures 4-4 and 4-5.
60 Figure 4-4. Data for the immediate and delayed recognition of the pairs (hits) in the incidental and intentional learning c onditions for the four groups divided by their MTL-FL function levels. Figure 4-5. Data for the immediate and delayed recognition false positive score of the pairs in the incidental and intentional learning c onditions for the four groups divided by their MTL-FL function levels. Immediate Pair Hits6 7 8 9 10 11 12 Low-LowLow-HighHigh-LowHigh-High MTL-FL groupcorrect score incidental intentional Delayed Pair Hits6 7 8 9 10 11 12 Low-LowLow-HighHigh-LowHigh-High MTL-FL groupcorrect score incidental intentional Immediate False Alarms0 1 2 3 4 5 6 7 8 Low-LowLow-HighHigh-LowHigh-High MTL-FL groupnumber of false positive error incidental intentional Delayed False Alarms0 1 2 3 4 5 6 7 8 Low-LowLow-HighHigh-LowHigh-High MTL-FL groupnumber of false positive error incidental intentional
61 Discussion In this experim ent, we manipul ated the learning instructions (whether explicit instructions about learning the word pairs were given) and attempted to exam ine the relationship of encoding instructions and the performance on the associat ive memory using verbal materials. After controlling the item memory between groups, the overall findings indicated that healthy older adults have a decreased ability to encode associative information automatically when compared to younger adults. However, when thei r attention was directed to the associative information (intentional encoding), they demons trated significant improvement in performance on the associative memory. The younger individuals maintained similar levels of performance regardless of the instruction types. The me mory improvement evident in the intentional encoding condition in healthy older adults indicates that healthy ol der adults could use strategic behavior (controlled process) to help them enc ode associative information which was in accord with previous results on age-rela ted difference in content and context memory (Kausler et al., 1985; Kausler & Phillips, 1989; Schmitter-Edgecombe & Simpson, 2001). Additionally, the results also corresponded to the notion that front al lobe/ executive functioning is differentially affected by the normal aging process and the comp romised frontal lobe system could impede the use of self-initiated strategic behavior duri ng the incidental encodi ng condition (Stuss, 1986). Despite the promising effect observed during the immediate recall c ondition for the healthy older adults, their performance on the delayed memory did not show a significant difference between the two encoding instructions. In cont rast to the healthy ol der individuals, the MCI individuals did not benefit from the explicit encoding instruction and remained at a lower level of performance across instruction conditions during the immediate r ecall phase. Although a trend toward better memory unde r the intentional in struction condition wa s observed in both normal older group and the MCI group during the de layed recall phase, the healthy older adults
62 did not demonstrate a significantly larger perf ormance gain under the in tentional instruction condition compared to the MCI group. The second major finding of the present study was the group differenc e in the occurrence of false positive errors. The findings indicated that the explicit encoding instructions not only improved the healthy old individuals performance on the pair hits measure, but also effectively decreased their false positive erro rs. In other words, the instru ction improved the healthy older adults discrimination ability on this task, and su ch improvement could actually carry over to the delayed recall phase. In contrast the MCI individuals did not be nefit from the explicit encoding instruction as the healthy old individuals did and still displayed high numbers of false positive errors in both immediate and delayed recall c onditions. Similar analyses were conducted on groups divided by their FL-MTL function levels, and the results suggested th at while the explicit encoding instructions helped all four groups to decreas e the occurrence of fa lse positive errors on immediate recall, the benefit did not carry ov er to delayed recall for the Low MTL-Low FL group. Such benefits were carried over the delayed recall phase for the High MTL-Low FL group and Low MTL-High FL group, although the la ter group only demonstrated a marginal effect (p= .05). The High MTL-High FL group exhibited low error rates across conditions. Overall, the collective results were consistent with the predictions. The findings obtained from the incidental encoding condition suggest ed that the healthy old adults showed a disproportionate impairment relative to younger adu lts in performance on the associative tests compared with performance on item tests wh en hippocampus was assumed to encode novel information in an automatic fashion. This replic ated prior findings that ol der adults clearly have deficits in binding discrete units of information together. However, rather than a generalized decrement in associative memory across manipulat ions, the healthy older adults could overcome
63 the disadvantage and perform at the same leve l as the younger people did if provided specific encoding instruction, although the benefit did not last for long On the other hand, the MCI individuals, as predicted, did not benefit from such instructions and showed lower performance and higher error rates when compar ed to the healthy old adults. This finding is consistent with the notion that the hippocampus is a key structure for binding information and forming the association. Even providing the explicit instruction during the encoding stage, individuals with putatively compromised hippocampal function were st ill unable to improve their memory for the associative information.
64 CHAPTER 5 MEMORY OF ASSOCIATIONS BETWEE N ITEMS OF T HE SAME KIND USING SEMANTIC RELATED OR UNRELATED WORD PAIRS Introduction In this current experim ent, we wanted to test whether the less a task requires the creation of episodic associations between components, the le ss pronounced the associative deficits would be in healthy old and individuals with MCI. To te st this prediction, we presented younger and older adults with pairs of words, eith er unrelated or related semantica lly, and tested them later with item and associative recognition tests. The two t ypes of word pair materials have often been used to study verbal memory: one is material be lieved to pre-experimentally exist in memory stores (preexisting, or familiar information sem antic related pairs), and the other type of material is novel information (novel pairs). Bower and Schacter (1993) define novel information as material introduced to memory for the first time during a study phase of an experiment. This can include stimuli such as a nonword or a new association between previously unrelated words (island and broom). A large body of literatures indicates that context-rich episodic memories, but not context-free semantic memories, require processing provided by the hippocampal circuit (Squire & Zola-Morgan, 1991; Vargha-Khadem et al., 1997). Accordingly, we predicted that older adults a nd MCI individuals would show a re latively smaller deficit in the associative recognition test when semantic related pairs were used compared to the novel pairs, since this task allowed them to rely more on preexisting associations and less on establishing new associations. Second, we wanted to investigate the effect of different test formats on recall of associative information. There is a large body of evidence sugge sting that age-related difficulty in retrieval is a major reason for poorer memory performance in older adults. One might expect recognition performance, where the information is re-represe nt during the test phase and hence could help
65 induce the appropriate mental operations necessary for retrieval. This c ould not be affected by age as much as recall performance, where participants must generate their own cues because no retrieval cues are present (Cra ik & McDowd, 1987; Naveh-Benjam in, 2000). Some studies have demonstrated that contextual or associative memo ry plays a larger role in free recall than it does in recognition memory, which can rely on the retrieval of item representations involving minimal contextual elaboration (Jacoby, 1991). In this current experime nt, the HY, HO, and MCI groups were given three formats of associative memory tasks: free recall, cued recall, and recognition by using word pairs as stimuli. On the free recall test, individuals had to generate their own cues because no retrieval cues were present. On the cued recall test, the first word of each word pair was given, and then the participant had to give the second word of the pair. On the recognition test, the word pairs were presented and participants had to answer yes/no for each pair to indicate whether the pair they saw was the same pair they were asked to remember during the study phase. According to previous literature, age-re lated decline in associative memory should be more closely related to age differences in the free recall than in the r ecognition tasks (Craik & McDowd, 1987; Naveh-Benjamin, 2000). Methods Participants Participants were 20 young, 20 healthy old, and 11 individuals with MCI, who were taken from the same pools as those run in experime nt 1. Their mean ages and other relevant demographic data appear in Chapter 3. Design The design was a 3 (Group) 2 (Pair Type; re lated, unrelated) (T est Type; free recall, cued recall, recognition) (St udy-Test Interval; immediate, dela y) Repeated Measures Analysis
66 of Variance. Group was a between-subject variable while Pair Type, Test Type, and Study-Test Interval were within-subjects variables. Materials and Procedure The study phase for each of the tasks included the presentation of a list of 14 word-pairs shown on the com puter screen as the experimental stimuli. For the word pair recognition test, 14 additional word pairs (seven semantically relate d and seven semantically unrelated pairs) were chosen. All of the words were high-frequency no uns. Half of the word pairs (e.g., seven pairs) in each of the lists were two words that were sema ntically unrelated. The other half of the word pairs were two semantically rela ted words (e.g., cage-bird), which we re considered as preexisting (or familiar) semantic associations because prev ious knowledge could support the creation of this kind of association. Previous work has provided the following gui delines for association: .40 (SD=.11) (strong), .17 (SD=.03) (moderate) and .07 (SD=.03) (weak) (Nelson, McEvoy, & Schreiber, 1998). Association strength for critical semantic pair s in the current study ranged from .17-.22, with a mean of .19 (SD=.02). Therefore, these semantica lly related word pairs were thought to reflect some preexis ting semantic knowledge. Within the list, the semantically unrelated pairs and the semantically related pairs were matched for length, frequency, and concreteness. Additionally, among the list, the semantically rela ted pairs and the semantically unrelated pairs were intermixed. The list of 14 pairs was presen ted during the study phase at a rate of one pair every 7 seconds for all the participants. On each trial, the word pair was presented simultaneously in a horizontal left-right fo rmat. The task was run under inte ntional learning instructions. Participants were told that each trial consists of a cue word (on the left) and a target word (on the right) presented together on the computer screen. In all of the lists, participants were told to try to learn the cue-target pairs, but to pay special attention to the target word in each pair. They
67 were also told to pay attention to the cue word because it could help them memorize and retrieve the target word. Participants were told befo re the beginning of the st udy phase about the three upcoming memory tasks:1) Free recall task : In this task, participants were asked to recall as many words as possible, particularly the target words (the second word in each pair). 2) Cuedrecall task : In this task, participants were presented, in random order, each of the cue words from each pair and were instructed to recall the target word that was a ssociated with each cue. 3) Recognition task: In this task, participants were shown 42 pairs of words, one pair at a time. Fourteen pairs were the intact pairs (those th ey studied during the study phase) with seven semantically related pairs and seven semantically unrelated pairs, 14 were recombined pairs (all words appeared during the study phase, but they are recombined, 7 were drawn from the semantically-related pairs, and 7 were from the unrelated pairs), and 14 were novel foil pairs. Participants were asked to answer yes if the pa ir was previously studied and no if it was not. For all of the tests, participants had as much time as they needed to provide a response for each test item. The same measures we re given again after 30minute delay. Hypotheses and Predictions Semantically unrelated pairs (novel pairs) : Due to previous findings that the establishm ent of novel associations depends upon hippocampal f unction, it was expected that HY participants would demonstrate the best performance on the n ovel pair tasks, and MCI participants should show the worst performance. We expected that the HO group would perform in an intermediate position. Also, we expected a large age effect in fr ee recall, a smaller effect in cued recall, and a still smaller one in recognition. A similar patte rn would carry over a 30-minute delayed recall condition. Semantically related pairs: In the preexisting semantic related association condition, previous knowledge could suppor t the creation of associations which was assumed to not
68 depend on the hippocampus. We expected much sm aller differences, if any, between HY, HO, and MCI groups across the three types of tasks (free recall, cued reca ll, and recognition) on learning and retention of semantically related pa irs than was expected on semantically unrelated pairs. A similar difference would carry over a 30-minute de layed recall condition. Results In this experim ental task, we predicte d that group differences would appear for semantically unrelated pairs in which the HY group should demonstrate the best memory followed by the HO and MCI groups, respectively. Als o, we expected a larger age effect in free recall, a smaller age effect in cued recall and in the recognition condition. In contrast to the semantically unrelated pairs (novel pairs), the three groups memory of the semantically related pairs should be more equivalent. If there we re any between-group differences, the differences should be much smaller than memory for the novel pairs, particularly in the recognition condition. A within-group difference between the se mantically related and novel pairs should be stronger in the HO and the MCI groups than in the HY group. A similar difference would carry over a 30-minute delayed recall condition. To te st this hypothesis, a mixed factorial Group Material (semantic related or novel) Task (f ree recall, cued recall and recognition) Time (immediate and delayed recall) ANOVA was conduc ted. Group was a between-subject variable, and Material, Task, and Time were within-subject variables. Significant effects were further examined using planned c ontrast comparisons. Table 5-1 presents the means and standard devi ations for the variables obtained on the task in both immediate and delayed conditions for the three groups. The analysis yielded significant main effects for Group [F (2,47)=36.17, p <.001, eta2=.61], Task [F (2,46)=421.59, p <.001, eta2=.95], and Material [F (1,47)=142.36, p <.001, eta2=.75] as well as significant Group Task [F (4,94)=9.11, p <.001, eta2=.28], Group Material [F (2,47)=22.32, p <.001, eta2=.49], Time
69 Task [F (2,46)=8.09, p <.005, eta2=.26], Task Material[F (2,46)=30.03, p <.001, eta2=.57], and Group Task Material [F (4,94)=6.50, p <.001, eta2=.22] interactions. The planned contrast comparisons on the in teraction effects revealed that during the immediate recall condition, the HY group had better memory regardless the type of materials (semantically related or novel) than did the HO group in both free (p<.005) and cued (p<.005) recall conditions. However, in the recognition test, the HY group demonstrated a better memory for the novel materials compared to the HO group (P <.01), but for the semantically related materials, the two groups did not differ significa ntly. During the delayed recall condition, the age effect was significant for bot h types of materials in the fr ee recall test (p < .005 in both materials). However, when giving a cued reca ll or recognition test, the HO group demonstrated similar performance as the HY group (Figure 5-1). During the immediate recall condition, when comparing the HO and the MCI groups, the MCI demonstrated poorer memory on the semantic material (p <.05) but comparable memory on the novel materials on the free recall test. On the cued recall procedure, both the MCI and the HO groups demonstrated benefits from this proce dure on the semantically related materials and the group difference found on the free recall proc edure disappeared. In contrast to the performance gain on the semanti cally related materials, the MCI did not benefit significantly from the cued procedure while the HO group di d when asked to recall the novel materials, resulting in a significant gr oup effect (p< .001). The MCI and the HO groups did not differ significantly on the recognition tests for either ty pe of materials. Du ring the delayed recall condition, the HO group demonstrated better perf ormance on both free recall and cued recall across both types of materials. On the recognition test, the MCI showed comparable
70 performance on the semantic related material s but poorer performance on the novel materials (p<.001). Table 5-1. Means and standard de viations for the variables obtain ed on the semantic related or novel word pair list in both immediate and delayed conditions on the young, healthy old, and MCI groups. MCI (n=11) Normal old control (n=20) Younger control (n=20) Measure M SD M SD M SD Immediate recall: Free recall (target word): Semantic word 0.80 1.03 1.90 1.12 3.60 1.14 Novel word 0.40 0.52 1.40 1.31 3.15 1.66 Cued recall : Semantic word 4.60 1.50 5.50 1.70 6.60 0.50 Novel word 0.80 0.79 3.30 1.95 5.90 1.12 Recognition: Semantic pairs 6.50 0.71 6.65 0.59 6.75 0.44 Novel pairs 4.40 1.65 5.20 1.47 6.70 0.66 Delayed recall: Free recall (target word): Semantic word 0.60 1.07 2.00 1.38 3.90 1.77 Novel word 0.20 0.63 1.75 1.29 3.40 1.39 Cued recall : Semantic word 4.90 1.91 5.95 1.28 6.65 0.59 Novel word 1.20 1.23 3.65 2.18 6.40 0.94 Recognition: Semantic pairs 6.30 1.06 6.75 0.44 6.80 0.41 Novel pairs 4.00 1.83 5.35 1.39 6.65 0.81
71 Figure 5-1. Data for free recall, cued recall, and recognition test on the semantic related or novel word pair list in both immediate and delayed conditions for the young, healthy old, and MCI groups. Word Pair Immediate Recall0 1 2 3 4 5 6 7 8measurecorrect scoreFree RecallSemanticNovelHY HO MCI HY HO MCI HY HO MCICured Recall Recognition SemanticNovel SemanticNovel Word Pair Delayed Recall0 1 2 3 4 5 6 7 8measurecorrect scoreFree RecallSemanticNovelHY HO MCI HY HO MCI HY HO MCICured Recall Recognition SemanticNovel SemanticNovel
72 Discussion The m ajor finding in this experiment demonstr ated distinct group differences on different types of test recall (i.e., free recal l, cued recall, or yes/ no rec ognition). In general, the three groups all followed the pattern with better performance on the recognition followed by the cued recall and then free recall on the same test, which was consistent with previous literatures (Craik and McDowd, 1987; Jacoby, 1991; Naveh-Benjamin, 2000). These evidences suggested that associative memory plays a larger role in free recall than it does in recognition memory. The pattern was particularly cl ear in the healthy old and the MCI groups: the MCI groups demonstrated the lowest level of performance compared to the healt hy old adults and the impairment was particularly clear using the free recall and cued recall pa radigms. Similarly, the healthy older individuals perfor med more poorly than younger par ticipants on free recall of word pairs. A similar pattern was also shown in the delayed recall. Rather than a generalized decrement in memory across all manipulations, olde r adults showed specific deficits that were attenuated in certain conditions. Second, this memory deficit was mediated by the semantic relatedness of the pairs, in which inherent semantic structure in related word pairs lent cohesion to associative memory, and therefore, the group di fference became smaller or even disappeared. In fact, even with substantial difficulty in learning the nove l word pairs that did not have any semantic relatedness, the MCI group was able to achieve similar level of performance as the healthy older adults did on lear ning and recalling the semantically related pairs during the immediate recall condition. This finding was co nsistent with longstanding theories, which postulate that semantic memory or memory for semantic related materials is more corticalized, and thus less dependent on the medial temporal lobes which support the encoding and acquisition of memory for novel material s. (Squire & Zola-Morgan, 1991; Vargha-Khadem et al., 1997).
73 CHAPTER 6 ASSOCIATIONS BETWEEN DIFFERENT KI NDS OF INFORMATI ON CROSS-REGION ASSOCIATIONS There is evidence that hippocampus damage does not impair all kinds of associations in an equal way. Vargha-Khadem et al. (1997) found that early hippocampal damage in childhood impairs the recognition of associations between items with different formats (e.g., face-voice and object-place associations) more than it does the recognition of a ssociations between the same kinds of items (e.g., word-word or face-face associat ions). Associations between different kinds of information are different from associations of the same kinds in the sense that the components in the former associations were probably represen ted in distinct neocortical regions rather than represented within one neocor tical region. Available eviden ce suggests that hippocampal damage preferentially disrupts associative memo ry that depends on the binding of information across cortical regions as compar ed to associations that depend on within-region binding. In this experiment, we investigated several types of associative information, including object-location, face-house, and item (pictures of faces and historical events) and temporal information association. Associations between Objects and Locations Introduction The aim of the present task was to inves tigate object-location memory in young, healthy old, and MCI individuals. Object-location memory concerns knowledge of the exact position of objects and their relative relati onship with each other (Kessels, Dehaan, Kappelle, & Postma, 2001). Previous literature has in dicated that two major cortico-co rtical processing pathways: the socalled dorsal and ventral processing str eams (Ungerleider and Mi shkin, 1982) are involved when performing such a test. The dors al pathway is directed into the posterior parietal cortex and is important for spatial perception/ localization The ventral pathway is directed into the inferior
74 temporal cortex and is important for visual object recognition (Ungerleider and Mishkin, 1982). Inputs from the two streams are integrated in hippocampal formation wh ich bind these distinct components together and form new memory (Mos covitch & Winocur, 1995). Overall, we were interested to examine whether healthy ol d and individuals with MCI demonstrated disproportionate impairment on this task. Methods Participants There were 51 participants, of which 20 were younger, 20 were healthy older and 11 were MCI individuals. The m ean age and number of y ears of education appear in Table 3-1. For the second types of analyses based on the individuals frontal lobe and me dial temporal lobe function level, 35 older adults were included. Th e details of demographic information and factor scores for each of the four groups can be found in Chapter 3. Design One set of 12 pictures of nam eable natural and manmade objects presented on a 6 x 6 grid were used for this experiment. Memory for th e studied objects was tested using free recall (object recall, subjects had to name them) and yes/no object recognition paradigms by presenting pictures of target and foil objects. Memory for st udied locations on the grid was tested using free recall. Memory for the object and location association was tested by fr ee recall (object-location recall) and forced-choice recogn ition tests (forced-choice object-l ocation recognition). Memory was tested immediately following th e study phase and again after a 30-minute delay for all tests. The design was a 3 (Group) 3 (Item Type, obj ect, location, object-lo cation) Study-Test Interval (immediate, delay) Repeated Measures Analysis of Variance. Group was a betweensubject variable, and Item Type and Study-Test Interval were within -subject variables.
75 Materials and Procedure One set of 12 pictures of nam eable natural and manmade objects was placed in a predetermined random position on a 6 grid (12 gr id positions were occupied by pictures, while 24 positions were blank). The participants had two minutes to study the entire display and they were instructed to remember both objects and their locations. After the two-minute study phase, subjects were given several t ypes of tests in order: 1) Object Free Recall and Recognition Tests : subjects had to name as many as objects as they could remember from the study phase and then perform a yes/no object recognition test. If the participant did not reac h a minimal criterion of 80% accuracy on the object recognition test, the study phase and recognition tests were repeated (before giving other tests) until 80% accuracy was achieved. 2) Location Free Recall Test : subjects were asked to place wood dots on those locations on the grid that were occupied by objects during the study phase. 3) Object-Location Association Test (OLAT) : In the OLAT recall test, subjects were provided with the pictures of 12 studied obj ects and they had to place the objects in their studied locations. Then, the su bject were given the OLAT recognition test in which each picture were shown in its studied loca tion and in three other locations (foils) which had been occupied by other studied pictures. Pa rticipants were asked to select the correct location for that individual pi cture. Memory was tested imme diately following the study phase and also after a 30-minute delay for all tests. Figu re 6-1 presents the stim uli used in this task. Hypotheses and Predictions Since es tablishment of associations between di fferent kinds of information was believed to depend on the hippocampus, we exp ected that HO participants would demonstrate slightly reduced performance compared to HY, while participants with MCI would demonstrate worst performance compared to the other groups. Furthe rmore, by dividing indivi duals into four groups based on their FL-MTL function, we predicted th at individuals with Low MTL would show
76 worst performance on the associative test. Howeve r, slight variation might be seen in those individuals with Low MTL depending on thei r FL function. Those Low MTL-High FL individuals would demonstrate slight better performance than those Low MTLLow FL individuals. Figure 6-1. The display of the namable objects and their locati ons for the object-location test. Results Analyses on HY, HO, and MCI groups In this experim ent, we hypothesized th at three groups (HY, HO, and MCI) should demonstrate equivalent memory for items indicated by their performance on the object recognition test, but that the MCI group should show disproportiona te impairment in associating objects and locations in free recall and recognition test s of this ability. To test this hypothesis, three mixed factorial ANOVAs were conducted separa tely for object free recall and recognition, location free recall, and object-l ocation free recall and recogniti on tests. Group was a betweensubject variable, and Tasks and Time (immedia te and delayed recall) were within-subject variables. Significant effects were further exam ined using planned contrast comparisons. Table 6-1 presents the means and standard deviations of the variables obtained on the Object Location Test in both immediate and delaye d conditions on th e three groups.
77 Table 6-1. Means and standard deviations of variables obtained on the Object Location Test on the three groups MCI (n=11) Normal old control (n=20) Younger control (n=20) Measure M SD M SD M SD Immediate recall: Object free recall 6.00 1.90 8.70 1.92 10.55 1.50 Object recognition hit 11.18 0.98 11.50 0.89 11.75 0.79 Object recognition-false positive 0.45 0.69 0.50 1.40 0.00 0.00 Object recognition-total 28.73 1.35 29.00 1.52 29.75 0.18 Location free recall 5.36 2.11 8.20 2.57 10.55 1.85 Object-location free recall 4.36 2.54 7.80 3.37 10.25 2.86 Object-location recognition 8.45 2.34 10.30 1.87 11.40 1.50 Delayed recall: Object free recall 6.60 2.99 9.35 2.28 11.70 0.66 Object recognition hit 11.70 0.67 11.90 0.31 11.95 0.22 Object recognition-false positive 0.30 0.67 0.25 1.12 0.00 0.00 Object recognition-total 29.40 1.07 29.65 1.14 29.95 0.22 Location free recall 5.90 1.79 8.35 2.62 11.10 1.59 Object-location free recall 4.70 2.41 7.70 2.89 10.85 2.18 Object-location recognition 9.20 1.81 10.70 1.38 11.40 1.50 All participants used only one learning trial to reach 80% accuracy cr iteria. Analyses on the Object Free Recall and Recognition Tests (hits) revealed a significant main effect of Group [F (2,47)=26.17, p <.001, eta2=.53], Test [F(2,47)=0.30, p <.05, eta2=.59; all groups performed better on recognition than free recall] and Time [F (1,47)=9.84, p <.005, eta2=.17; all groups did better on delayed than on immediate recall] A significant Test X Group interaction [F(2,47)=26.23, p <.001, eta2=.53] was also found and the posthoc comparisons revealed that all groups had equivalent scores on the recognition tests (both immediate and delayed recall) which suggested an equivalent item memory. However, on the free recall te sts, the younger control group had the best scores followed by normal ol d control and then MCI [F(2,47)=28.25, p <.001, eta2=.55]. On the Location Free Recall Test, a significant Group effect was found in which the younger control group has the best scores followed by normal old control and then MCI [F (2,47)=27.70, p <.001, eta2=.54]. No significant interaction effect was found [F(2,47)=.17, p >.05, eta2=.01]. Analyses on the Object-Location Free Recall and Recognition Test revealed a significant main effects of Group [F(2,47)=13.89, p <.001, eta2=.37] and Measure (free recall vs.
78 recognition) [F(1,47)=153.94, p <.001, eta2=.76] in which all three groups demonstrated better performance on the recognition test compared to the free recall test. A Group x Measure interaction was also signif icant [F(2,47)=21.69, p <.001, eta2=.48 ]. The planned contrast analyses carried out on the Group and Group x Measure interaction showed that the HY group had significantly better performance on free r ecall tests (immediate: t (48)=2.57, p <.05 and delayed: t(47)=3.94, p <.001) but similar level of performance on the recognition tests (p>.05) when compared to the HO group. The HO group dem onstrated significantly better performance on free recall (immediate: t (48) = 3.04, p <.005 and delayed: t (47)= 3.06, p <.005) as well as recognition tests (immediate: t( 48)=2.66, p <.05 and delayed: t( 47)=2.55, p <.05) compared to the MCI group. Figure 6-2 shows scores of obj ect recognition tests (i tem hits) and objectlocation recognition tests across gro up and two time points. Figure 6-3 shows that scores of each group on object, location, and objectlocation free recall tests. Figure 6-2. Data for object recognition hit and object-location rec ognition tests in both immediate and delayed conditions fo r the young, normal old, and MCI groups. Object-Location Test: Item and Associative Memory0 2 4 6 8 10 12measurecorrect score Object Recognition hit Object-Location RecognitionImmediate delayed MCI HO HY MCI HO HY Object Recognition hit Object-Location Recognition
79 Figure 6-3. Data for object, location, and objectlocation free recall test s in both immediate and delayed recall for the young, normal old, and MCI groups. Analyses on groups divided by MTL-FL status A second analysis was conducted by dividing individuals into four groups based on their FL-MTL function levels. It wa s hypothesized that individuals with Low MTL should show worst perform ance on the associative test. Howeve r, slight variation might be seen in those individuals with Low MTL depending on thei r FL function. Those Low MTL-High FL Immediate Recall 0 2 4 6 8 10 12 14 MCI HO HYgroupcorrect score object free recall location free recall object-location free recall Delayed Recall 0 2 4 6 8 10 12 14 MCI HO HYgroupcorrect score object free recall location free recall object-location free recall
80 individuals would demonstrat e slightly better performance than those Low MTL-Low FL individuals. Analyses on the Object Free Recall and Recogni tion Tests (hit) revealed a significant main effect of Group [F (3,31)=4.78, p <.01, eta2=.32] and Test [F(1,31)=84.66, p <.001, eta2=.73] in which all groups showed a better performance on the recognition tests compared to the free recall tests. A significant Group x Test interaction [F (3,31)=4.75, p <.01, eta2=.32] was also found and the post-hoc comparisons revealed that all groups ha d equivalent scores on the recognition tests (both immediate and delayed recall), which suggested an equivalent item memory. However, on the free recall tests, the Low MTL-Low FL group, but not the Low MTLHigh FL group, showed significantly poorer perf ormance than the two groups with High MTL function. On the Location Free Recall Test, a significant Group effect was found in which the Low MTL-Low FL group had significantly lower sc ores compared to th e two groups with High MTL function. No significant in teraction effect was found. Analyses on the Object-Location Free Recall and Recognition Test revealed significant main effects of Group [F(3,31)=9.02, p <.001, eta2=.47] and Measures (free recall vs. recognition) [F(1,31)=112.37, p <.001, eta2=.78], in which all gr oups demonstrated better performance on the recognition test than on the free recall test. A Group Measure interaction was also significant [F(3,31)=3.73, p <.05, eta2=.27 ]. The post-hoc contrast comparisons showed that the Low MTL-Low FL group had si gnificantly lower performances on both free recall and recognition tests compar ed to the two groups with High MTL function. Compared to the Low MTL-High FL group, the Low MTL-Low FL group demonstrated a similar level of performance on the free recall tests, but on the recognition test, the former group had significantly better performance than the later group (Figures 6-4 and 6-5).
81 Figure 6-4. Data for object, location, and objectlocation free recall test s in both immediate and delayed recall for the four groups divide d based on their function on MTL and FL factors. Figure 6-5. Data for object recognition hit and object-location rec ognition tests in both immediate and delayed conditions for the f our groups divided based on their function on MTL and FL factors. Immediate Recall0 2 4 6 8 10 12 Low-Low Low-HighHigh-LowHigh-High MTL-FL groupcorrect score location free recall object free recall object-location free recall Delayed Recall0 2 4 6 8 10 12 Low-Low Low-HighHigh-LowHigh-High MTL-FL groupcorrect score location free recall object free recall object-location free recall Immediate Recall0 2 4 6 8 10 12 14 Low-Low Low-HighHigh-LowHigh-High MTL-FL groupcorrect score object item hit object-location recognition Delayed Recall0 2 4 6 8 10 12 14 Low-Low Low-HighHigh-LowHigh-High MTL-FL groupcorrect score object item hit object-location recognition
82 Associations between Faces and Houses--Relational Information Introduction Both anim al and human functional imaging studies indicate a unique role for the hippocampus in the flexible expression of d eclarative memories because no other medial temporal region demonstrated a similar effect (Bunsey & Eichenbaum, 1996; Heckers et al., 2004; Preston et al., 2004). The memory for rela tional association is an ability to infer the relationships between indirectly related items th at have not been pres ented together, based on previous learning of overlapping pa irs (e.g., A and B are associated, if A is associated with C, then B is also associated with C) (Heckers et al., 2004). Alth ough several studies have evaluated the relational association or transitive asso ciation in hippocampal lesion cases or normal individuals using functional imaging, to our knowledge, no study has examined the relational association in aging population as well as in individuals with mild cognitive impairment. Accordingly, the goal of the pr esent experiment was to exam ine whether older adults and individuals with MCI de monstrated disproportionate impairme nt in relational association with relative sparing ability of the item memory. Method Partic ipants There were 51 participants, of which 20 were younger, 20 were healthy older and 11 were MCI individuals. The mean age and number of y ears of education appear in Table 3-1. For the second types of analyses based on the individuals frontal lobe and me dial temporal lobe function level, 35 older adults were included. Th e details of demographic information and factor scores for each of the four groups can be found in Chapter 3.
83 Design The design was a 3 (Group) 2 (Study-Test Interval; imm ediate, delay) Repeated Measures Analysis of Variance. Group was a between-subject vari able, while Study-Test Interval was a within-subject variable. Materials and procedure The m aterials and design used by Preston, Shrager, Dudukovic, & Gabrieli (2004) was adapted in the current experiment. The stimu li consisted of 124 black and white photographs of faces (12 male, 12 female) obtained from yearbooks and 15 colorful pictures of real houses. Twenty-four photographs of faces and 12 photographs of houses were used to construct two sets of paired associates: two sets of face-house pairs and one set of face-face pair s. The first set of face-house pairs consisted of 12 f aces (B, 6 men and 6 women) and 12 houses (A). The second set of face-house pairs consisted of a set of 12 different faces (C, 6 men and 6 women) and the same 12 houses (A) used in the first set. Therefore, the face-house pairings resulted in two faces that shared an association with the same house (A). Four of the remaining photographs of faces and three photographs of the houses were used to construct the practi ce trial and the other 96 faces served as foils for the item recognition tests. Figure 6-6 demonstrates an example of stimuli and the task for this experiment. During the study phase, the set of face-house pairs was shown to subjects at a rate of one pair every six seconds. Participants were instruct ed to remember the single face, the relationship between the face and the house, and were particularly told to try and make the connection of the two faces that shared the same houses. The set of stimuli was presented to the subject three times on a computer to avoid a floor effect. Following the study phase, participants received a yes/ no face recognition test to assess their learning of individual f aces. In the recognition test, if the participant did not reach a minimal criterion of 80% accuracy on this test, the study phase
84 and recognition tests were repeated until 80% accuracy was achieved. If after repeating three more trials from the initial learning trials, the subject still could not re ach 80% accuracy, s/he was excluded from the experiment. Following the face recognition test, subjects were asked to do an association recognition test in which there we re five previously learned faces in each item (one target and four choices). They were asked to identify one face from the four choices that shared the same house as the target face. The picture of the shared house was not presented. A practice trial was given prior to the formal task to ensure that the pa rticipants understand the nature of the task. Both the face and associati on recognition tests were ad ministrated again after a 30-minute delay. Figure 6-6. An example of the stimuli and the task for the face house association test Hypotheses and Predictions Available evidence (Eichenbaum & Cohen, 2001; Heckers et al., 2004; Preston et al., 2004) suggests that the hippocampus contributes uniquely to the ab ility to infer relationships between elements of learned associations while other brain regions (e.g., prefrontal cortex and pre-supplementary motor area) also contribute to performance on this type of task. As a result, A The 4 faces were all previously learned in the study phase B A C
85 we would predict individuals with MCI would demonstrate poorest performance compared to other subjects, following by HO, and then HY in th e recognition tests. Furthermore, by dividing individuals into four groups ba sed on their FL-MTL function, we would predict that individuals with low MTL (and possibly Low FL) function would show poorer performance on relational task compared to individuals with High MTL or/and FL function. Results Analyses on the HY, HO, and MCI groups In this experim ental task, with the manipula tion of repeated learning trials, we predicted that three groups (HY, HO, and MCI) should demons trate equivalent item memory as indicated by their performance on the face recognition test However, the three groups should differ significantly on the face-house association test, in which HO participants would demonstrate slightly reduced performance compared to HY, while participants with MCI would demonstrate lower performance than HO. To test this hypoth esis, we first examined the equivalence of the item memory and then conducted a mixed factor ial ANOVA for the face-house association tests with group as the between-subject variable and time (immediate vs. delayed recall) as the withinsubject variable. On the face recognition test (hit), a significant Group effect was found [F (2,47)= 3.31, p<.05] even after providing extra expos ure to stimuli to people needed. Post-hoc comparisons were conducted and revealed that the MCI group had significantly lower scores on the face recognition test compared to the young control group, but no group differences were found between young control and old control gro ups, or old control group and the MCI group. Due to the difference in face recognition perf ormance between MCI and HY group, the omnibus test was skipped and two planned Helmert cont rast comparisons (HY vs. HO; HO vs. MCI) were conducted. For the HO vs. MCI comparisons, results revealed significant main effects for Time
86 [F(1,28)= 6.39, P<.05, eta2=.19], Group [F(1,28)= 7.99, P<.01, eta2=.22], and a significant Time x Group interaction [F(1,28)= 4.57, P<.05, eta2=.14]. The post-hoc comparisons showed that the MCI group and the NO group did not differ signifi cantly on the immediat e association test, although the p values was marginal [t (48)=2.04, p =.052]. After a 30 minutes delay, the MCI group showed significant decline on the same test compared to the NO group [t(47)=4.37, p <.001]. No significant main eff ects or interactions were seen in the HY vs. HO comparisons. Figure 6-7 shows the data of immediate and dela yed association tests across the three groups. Figure 6-7. Data for immediate and delayed reca ll on the face-house association test across three groups. Analyses on groups divided by MTL-FL status A second analysis was conducted by dividing individuals into four groups based on their FL-MTL function levels. It was hypothesized that individuals with Low MTL or Low FL should have poorer perform ance on the associative test compared to individuals with High MTL or/and FL function. Face-House Association Test0 2 4 6 8 10 12 HY HO MCI groupcorrect score delayed recall immediate recall
87 On the face recognition test (hit), the f our groups were found to have comparable performance [F (3,30)= 1.36, p >.05], indicating a si milar level of item memory for the faces. The analyses on the Face-House association tests revealed that only the main effect for Group was significant [F (3,31)= 3.09, p <.05, eta2=.23]. However, when conducting post-hoc comparisons, the group differences were not reve aled. The data of the four groups on the association tests are presented on Figure 6-8. Figure 6-8. Data for face-house as sociation test in both immediate and delayed conditions for the four groups divided based on thei r function on MTL and FL factors. In order to further examine the relative invol vement of frontal lobe and medial temporal lobe functioning, two separate analyses were also conducted by comparing Low MTL individuals with High MTL individuals regardless of their functi onal status on the frontal lobe factor, and Low FL individuals with High FL individuals regardless of their functional status on the MTL factor. The analyses revealed a sign ificant Group main effect [F (1,33)= 9.13, p <.01, eta2=.22] when comparing the Low MTL group with the High MTL group, suggesting that the Low MTL group had lower scores compared to the High MTL group in both immediate and delayed recall of the face -house association test. No significant interaction effect was observed. Face-House Association Test0 1 2 3 4 5 6 Low-LowLow-HighHigh-LowHigh-High MTL-FL groupcorrect score delayed recall Immediate recall
88 Figure 6-9 presents the data for Low MTL and Hi gh MTL groups on the association tests. When comparing Low FL and High FL individuals, no significant main effects or interaction effects were found. Figure 6-10 presents the data fo r Low FL and High FL groups on the association tests. Figure 6-9. Data for face-house as sociation test in both immediate and delayed conditions for the two groups divided based on their f unction on medial temp oral lobe factor. Figure 6-10. Data for face-house association test in both immediate and delayed conditions for the two groups divided based on thei r function on frontal lobe factor. 0 1 2 3 4 5 6 immediate delayed test conditioncorrect score High MTL Low MTL MTL involvement on the Association Test 0 1 2 3 4 5 6 immediate delayed test conditioncorrect score High FL Low FL FL involvement on the Association Test
89 Association between Temporal Information and Novel Faces Introduction The purpose of the present task as well as the hi storical event tem poral order test used in the next session was to assess healthy older adults and individuals with MCIs ability to process temporal information. Temporal memory is ofte n associated with fontal lesions in humans (Shimamura et al., 1990). Similar to patients with frontal-lobe lesions, normally aging subjects show performance decrements (which are presumed to derive from diminished frontal lobe function) on tasks requiring the re trieval and organization of contextual information, such as the spatial and temporal source or c ontext in which an item was initially learned. On the other hand, hippocampus has been heavily implicated in the modulation of asso ciative memory and, consequently, is hypothesized to provide a memo ry mechanism for temporal events as well, particularly if these events invol ve novel materials (Eichenbaum et al., 1994). In fact, Floresco et al. (1997) found that the inte raction between the hippocampus and medial prefrontal cortex (mPFC) was necessary for rats to perform inte gration of sequential re sponses with a 30-minute delay. Once a task requires prospective coding of a sequence of several sp atial items during an intermediate-term delay (i.e., 30 min), based on memory of previously coded items, the interactive communication between the hippocampus and medial prefrontal cortex (mPFC) might become intensified (Floresco et al., 1997). Accordingly, temporal or sequence memory may be a function that depends on both mPFC and medial te mporal lobe and the length of delay may be one of the critical factors in dissociating para llel coding between the hippocampus and PFC. In the field of cognitive aging, several studie s of temporal memory have been conducted, in which an age-related effect was reported on the temporal memory and the effect was often associated with frontal dysfunction, although its relationship with some frontal type tasks/executive tasks is still unsure (Cabeza et al., 2000; Fabiani & Friedman, 1997; Glisky et al.,
90 1995; Schmitter-Edgecombe & Simpson, 2001). Ho wever, similar studies conducted in MCI population are still lacking. Given the involve ments of both hippocampus and frontal lobe systems in aging process, it was our goal to exam ine the temporal memory in healthy aging and individuals with mild cognitive impairment. Methods Participants There were 51 participants, of which 20 were younger, 20 were healthy older and 11 were MCI individuals. The m ean age and number of y ears of education appear in Table 3-1. For the second types of analyses based on the individuals frontal lobe and me dial temporal lobe function level, 35 older adults were included. Th e details of demographic information and factor scores for each of the four groups can be found in Chapter 3. Design This task in volves the presentation of tw o unfamiliar face lists for study followed by a forced choice recognition test comprised of three types of test trials: recognition, list discrimination, and within-list recency trials. Th e lags (the distances between the two items) between the list discrimination a nd within-list recency trials we re matched. The design is a 3 (Group: HY, HO, MCI) 4 (Trials Type: recogni tion, list discriminati on, within-list recency judgment, and recall) Repeated Me asures Analysis of Variance. Materials and procedure The m aterials consisted of two 16face lists. All faces were unfamiliar to subjects, having been derived from high school yearbooks from dist ant towns. Subjects firs t studied a list of 16 faces presented at a rate of 1 item every five seconds on a computer screen. After three repetitions of the first list, a s econd list was presented and repeated three times. The lag between the last item of the first list a nd the first item of the second list was fixed at 45 seconds. For all
91 participants, following presentation of one item, there was a blank inte rval (400 ms) before presentation of the next item. All of the stimuli were run under intentional learning instructions, in which participants were told that they w ould be shown two lists of faces, following which their memory for the temporal order relati onship among these faces would be tested. A recognition test with 124 trials was give to all participants. During the recognition test, two items appeared on the computer screen together and the subjects task was to indicate which of the two stimuli was presented most recentl y. In the case of recognition, only one item was previously seen and therefore was, by definition, th e most recent. Subjects were instructed that in some test trials there was only one previously seen stimulus. There were three types of test trials: 1) Recognition trials (32 trials): subjects were presente d with a previously seen stimulus and a foil (i.e. a new stimulu s) at the same time. 2) Betweenlist discrimination trials (44 trials): subjects were shown two faces that they had previously studied. One face was from the fi rst list and one was from the second list. 3) Within-list discrimination trials (48 trials): subjects were shown with two stimuli from the same list (i.e., either from the first list or the second list). The most recent stimulus (or the old stimul us for recognition trials) appeared equally frequently on the left and the right side of the screen, and relative positions were also counterbalanced across face-face lag. The three types of trials described above were intermixed randomly throughout the sequence. The advantage of this paradigm was that subjects did not need to be aware of whether the trial was te sting recency or recognition memory, since the procedure was identical in the two cases and the trials were intermixed. Thus, any performance
92 difference observed between recency and recognition could not be attributed to difference in task requirements. In the recency judgment trials, the lags (distances between two items) used in the withinlist recency trials and list discrimination trials were matched. Thus, the performance difference observed between the two recency judgment tasks was not confounded by the lags between stimuli. Hypotheses and Predictions Previous findings suggested that hippocam pal lesions can result in disproportional impairment (compared to item recognition) on temporal order memory, which included list discrimination task and withinlist recency judgment. Based on such findings, we expected that HY would demonstrate best performance on the tw o temporal order tasks followed by HO. The MCI group would demonstrate poorest performance among these groups. It was also expected that a between-list discrimination would be easie r than a within-list discrimination, leading to better performance for all three groups. In addition, by dividing individuals into four groups based on their FL-MTL function, we predicted th at individuals with either Low MTL or/and Low FL would show poor performance based on past literature, in which both types of lesions have been suggested to impair memory for temp oral information of novel information, compared to individuals in the High MTL-High FL group. Results Analyses on the HY, HO, and MCI groups In this experim ent, participants were aske d to do a recency judgment test with novel pictures of faces as stimuli, it was predicted th at the three groups would have similar levels of item memory performance (as seen in the item r ecognition score). We also predicted that the HY should perform better on the judgment of th e temporal information compared to the HO
93 group, who would in turn show a better perf ormance than the MCI group. The group effect should be larger after a 30-minute delay. In addition, it was expected that the between-list discrimination items would be easier than a within -list discrimination items for all three groups. To examine these predictions, item memory across groups was tested first and the result showed a significant Group effect [F (2,47)= 12.59, p <.001]. Two planned contrast comparisons indicated that the HY group and the HO group show ed comparable item memory, but the MCI group had significantly lower score on th e item memory than did the HO group. Due to the group differences in item memo ry, a Group x Trial Type (between-list discrimination and within-list discrimination ite ms) x Time (immediate and delayed recall) mixed factorial ANOVA was conducted w ith item memory (immediate item hit) as a covariate. The results revealed a significant main effect for Group [F (2,46)= 5.35, p <.01, eta2=.19] and a significant Group Trial Type inte raction [F (2,46)= 5.36, p <.01, eta2=.19]. The planned contrast comparisons showed that, both immedi ately and after a delay, the HY outperformed the HO on the between-list items[imme diate recall: t (47)= 2.61, p < .05; delayed recall: t(47)=3.52, p<.005], but not on the within-list items. When comparing the MCI and the HO groups, a similar pattern was revealed in which the HO group demonstrated a better performance on the between-list items in the immediate [t (47)= 3.34, p <.005] and delayed recall [t (47)= 3.52, p <.005] conditions. The two groups did not differ significantly on the within-list items. Figure 6-11 presents the data of the between-list and within-list trials for the young, healthy old, and the MCI groups. Analyses on groups divided by MTL-FL status A second set of analyses was conducted using groups divided by their frontal and m edial temporal lobe function levels. It was hypothesize d that individuals with either Low MTL or/and
94 Low FL would show poorer performance based on past literature compared to individuals in the High MTL-High FL group. Figure 6-11. Data of the betw een-list and with in-list trials at two time points for the young, healthy old, and the MCI groups. Similar to previous three-group analyses, the item memory was first examined among the four groups and the results indicated that the four groups did not differ sign ificantly on the item memory score [F (3,30)= 2.35, p >.05]. A Group x Trial Type (between-list discrimination and within-list discrimination items) x Time (immed iate and delayed recall) mixed factorial ANOVA was conducted. The result demonstrated a signi ficant main effect for Group [F (3,30)=4.30, p <.05, eta2 = .30A Measure x Time interaction [F (1,30)=5.28, p <.05, eta2 = .15] indicated that memory for the between-list items declined more rapidly over time than did memory for the within-list items. Two separate analyses on the two frontal lobe groups regardless of their MTL function level (i.e. high frontal lobe group vs. low frontal lobe group) and the two MTL groups regardless of their FL functi on level (i.e., High MTL group vs Low MTL group) were also conducted. The results showed that the Low a nd High FL groups differed significantly only on 0 10 20 30 40 50 60 70 80 MCIHOHY groupcorrect [%] Between List Within List Immediate Recall on the Recency Test 0 10 20 30 40 50 60 70 80 MCIHOHY groupcorrect [%] Between List Within List Delayed Recall on the Recency Test
95 memory for the within-list item s during the immediate recall cond ition [t (32) = -2.07, p <.05], in which the High FL group showed a better memory on this measure. Comparing the Low and High MTL groups, the Low MTL group demonstrated significantly lower sc ores on the betweenlist during the delayed re call condition compared to the High MTL group [t (30)=-2.58, p< .05]. The effect on the memory for within-list items in the immediate recall condition was marginal (p=.05), in which the High MTL group demonstrat ed a better performance than the Low MTL group. Figure 6-12 displays the data on the recency task for the four groups divided by their frontal and medial temporal lobe function levels. Figure 6-12. Data of the recency task for the four groups divided by th eir frontal and medial temporal lobe function levels. 0 10 20 30 40 50 60 70 Low-LowLow-HighHigh-LowHigh-High MTL-FL groupcorrect score Within List Between List Delayed Recall on the Recency Test 0 10 20 30 40 50 60 70 Low-LowLow-HighHigh-LowHigh-High MTL-FL groupcorrect score Within List Between List Immediate Recall on the Recenc y Tes t
96 Association between Temporal In format ion and Historical Events Methods Participants There were 51 participants, of which 20 were younger, 20 were healthy older and 11 were MCI individuals. The m ean age and number of y ears of education appear in Table 3-1. For the second types of analyses based on the individuals frontal lobe and me dial temporal lobe function level, 35 older adults were included. Th e details of demographic information and factor scores for each of the four groups can be found in Chapter 3. Design This task in volves the presen tation of 15 historical events The design is a one-way analysis by Group (HY, HO, and MCI). Materials and procedure This test was m odified from Bauer (1984) & Shimamura, Janowski, & Squire (1990) and the events used for the experiment were selected based on pilot data. A total of 15 items were used, consisting of two to three events from each decade from the 1940s to the 2000s. Each item was presented on a separate card. The subject was presented with a set of 15 cards, arranged in a random order. The subject was then asked to order the events ac cording to the time of occurrence, starting with the most remote event a nd completing the set with the most recent one. A global arrangement score, which was a vector score based on the distance between correct response and the subjects response, was obtained for each subject. For each item, the absolute difference was taken between the position give n to the item by the subjects and its correct position. For example, a response (A1, A3, A2, A4, A5, A6, A7assuming all correct responses after A7) would receiv e a score of 0+1+1+0+0+0+0+0+ =2. The scores could range
97 between 0 (correct response) and 112 (total inversion of the correct or der). The subject was allowed to use as much time as they needed to finish this task. Hypotheses and Predictions Due to the fact th at knowledge of historical events was consid ered as one type of semantic memory stored in the semantic network and wa s believed not to rely on hippocampus or medial temporal lobe structures, the th ree groups (HY, HO, and MCI) w ould not be expected to show significant group difference on this measure. Fu rthermore, performance on this task was expected to correlate with the score on Glisky et al.s frontal factor (since poor performance on this task would reflect source memory impairme nt or a failure in st rategic processing). By dividing individuals into four groups based on their FL-MTL function, we expected that Low FL subjects would have poorer performance on this task compared to High FL subjects. Results In this experim ent, we asked participants to arrange 15 historical events based on the chronological order. It wa s hypothesized that three groups (HY, HO and MCI) would not demonstrate significant group differences on th is measure because semantic memory was believed not rely on hippocampus or medial temporal lobe memory system. However, performance on this task was expected to correl ate with the frontal func tion. To test this prediction, a correlational analysis between the task and clinical frontal measures as well as the frontal lobe composite scores was conducted. Additionally, by dividing individuals into four groups based on their medial tempor al lobe and frontal lobe functi on levels, it was expected that subjects in the lower frontal function groups would have poorer performance on this task compared to subjects who were in the higher frontal lobe function groups
98 Analyses on the HY, HO, and MCI groups An absolute difference s core was calculated between the subjects answer and the actual order for the events. The higher scores indicated the poorer performance. A Group Task ANOVA was then conducted and yi eld a significant Group main e ffect [F (2,48)=7.03, p< .005]. The post-hoc comparison revealed that the MC I group showed significantly poorer performance on this measure than did the HY and HO groups, whic h did not differ from each other. Figure 613 presents the results for the three groups on th e historical event temporal order test. Figure 6-13. Data of the difference scores on th e historical event tempor al order test for the young, healthy old, and the MCI groups. Analyses on groups divided by MTL-FL status A separate analysis was conducte d on four groups divided by th eir frontal lobe and medial tem poral lobe function levels. An omnibus test was skipped and instead two planned contrast comparisons (i.e., high frontal l obe group vs. low frontal lobe group; high temporal lobe group vs. low temporal lobe group) were also conducted in order to ex amine the relative contributions of frontal lobe and medial temporal lobe functio n. The analyses revealed that the high and low frontal lobe groups did not differ significantly on this measure [t (31) =1.13, p >.05]; however, Historical Event Temporal Order Test0 10 20 30 40 50 60 MCIHOHYgroupdifference socre
99 the High MTL group demonstrated a significantly better performance on this task compared to the Low MTL group[t (31) =2.07, p <.05]. Figure 614 shows the data for the four groups on the measure. Figure 6-14. Data of the differen ce scores on the historical event te mporal order test for the four groups divided by their medial temporal lobe and frontal lobe function levels. Correlations between temporal order te st and neuropsychological measures The correlational relationship (P earson correlations) between the event temporal order measure and frontal/ executive function m easures was conducted through a one-tailed correlational analysis and the results revealed that the performance i ndicating by the absolute different scores (higher score = worse performa nce) on the event temporal order measure was negatively correlated with Similarities subtest (WASI) raw scores (r = -.26, p <.05), Matrix Reasoning subtest (WASI) raw scor es (r = -.25, p <.05), Arithmetic subtest (WAIS) (r = -.35, p <.01), Letter-Number Sequencing subtest (WMS-III) (r = -.23, p <.05), COWA (r = -.27, p <.05), WCST total errors raw (r = .23, p <.05), and Fron tal lobe composite scores (r = -.30, p <.05). The event temporal order measure was also pos itively correlated with Trails A time (r = .24, p <.05) and Trails B time (r = .31, p <.05). Historical Event Temporal Order Test0 10 20 30 40 50 60 Low-LowLow-HighHigh-LowHigh-HighMTL-FL groupdifference socre
100 Discussion Several experim ental tasks related to associa tions between different kinds of information were conducted. On the object-loca tion association task, the results showed that the three groups demonstrated comparable item memory demonstrated through a recognition paradigm, although a clear pattern of group difference was shown in free recall. Memory for the object-location association followed the predic ted pattern with the HY group having the best memory followed by the HO group and then the MCI group. Unexp ectedly, the MCI group did not demonstrate significant forgetting for items or item-location associations over a 30-minute delay. On the parallel analyses conducting on the four groups divided by their FL-MTL function status, the results indicated that the four groups showed comparable item memory. However, the Low MTL-Low FL group, but not the Low MTL-High FL group, demonstrated significantly lower scores on the associative memory. The overall findings indicated that MCI individuals showed disproportionate difficulty in the associative memory, even though they demonstrated similar item memory compared to the healthy older adults Furthermore, impairment on tasks associated with medial temporal lobe appeared to have th e major contribution to this specific task. The frontal lobe appeared to play a secondary ro le on the associative memory in this specific measure. High vs. Low FL status came into pl ay only on those comparisons involving the Low MTL groups. The face-house relational association task in the present experiment was a very difficult task. Although the list was pres ent at least three times for all participants, the healthy older participants scores on the face-house relational as sociative items were ha lf these of the younger people. In fact, all of younger participants only n eeded 3 trials (minimal exposures) to reach an 80% accuracy criteria on the f ace recognition test, while the hea lthy old individuals needed an average number of 3.85 (SD =0.93, range: 3 to 7 trials) trials and the MCI individuals needed 4.9
101 trials (SD = 1.14, range: 4 to 7 trials) to reach si milar performance on the item recognition test. Despite the difficulty of this memory task, th e HO and the HY groups were able to demonstrate comparable item memory. In contrast to item memory, the healthy older group was able to get only half of the associative items correct as the younger people did during the immediate recall phase, although their memory for the associativ e items was as well-maintained over a 30-minute delay, as was that of the younger group. The HO and MCI groups demonstrated comparable item memory. The MCI group demonstrated a trend (p=.05) toward lower scores on the immediate associative memory compared to the HO group. After a 30-minute delay, the MCI group showed substantial forgetting on the associative memory, which actually fell below the chance level. In summary, with comparab le item memory be tween groups, the younger individuals demonstrated the best memory for the relational association followed by the healthy old group and then the MCI individuals. The MCI group was further char acterized by a faster forgetting rate on the relational associative tas k, while the other two groups remained stable memory for the same material. On the parallel analyses conducted on the four groups divided by their FL-MTL function status, the four groups demonstrated comparab le item memory. When further examining the relative involvements of medial temporal lobe and frontal lo be on the relational association memory, the Low MTL group, which combined the groups with high and low FL factor scores, demonstrated significantly poorer performance on the associative memory compared to the High MTL group, although the Low MTL group did not de monstrate further forgetting after a 30minute delay. When comparing the two groups w ith either High or Low FL factor scores, regardless of their MTL function level, th ere was no significant difference found.
102 Overall, these findings indicated that medial temporal lobe rather than the frontal lobe function has more substantial involvement in re lational associative memory. MCI individuals who were characterized by compromised M TL function demonstrated relatively lower performance compared to the healthy older cont rol, and such memory dysfunction was further accentuated in a delayed recall condition. Indeed, pr evious studies in animals indicated a role for the hippocampus in the flexible expression of declarative memory (Bunsey and Eichenbaum, 1996). Studies conducted on patients or healthy a dults also have suggested that hippocampus contributed to the process that conjoin novel relations between elem ents of learned associations (OReilly and Rudy, 2001; Preston et al., 2004). Two experimental tasks were conducted, one wi th novel face materials and the other with historical events, in order to examine the tem poral order of information in the episodic and semantic memory systems. The face recency task in the present experiment was also a very difficult task. Although each of the two lists was presented three times, the younger participants could identify only approximately 75 % of items correctly and the healthy older group could identify only approximately 60 % of items corre ctly. The performance of the MCI group was actually close to the chance level. Despite the difficulty of this task, the results still demonstrated a pattern as predicted, with th e younger group showing th e best performance, followed by the HO group and then the MCI group on the novel face recency task. This indicates that the MCI group was impaired in de termining which of two stimuli had been seen more recently on a test using novel faces as stimu li and such impairment was disproportionately impaired compared to the item memory. In fact the group effect was mainly derived from the difference on the between-list item discriminations rather than on the within-list item
103 discrimination, even after controlling the lags between two stimuli on the two discrimination conditions. On the semantic temporal order test, an une xpected result was found in which the HY and HO groups showed a comparable performance that was better than the performance seen in MCI group. Overall, the results obtained from the two tests related to temporal order judgment indicated that the MCI individua ls ability in judging temporal information was impaired for both novel materials and materials that depende d on the semantic system. Further, the disproportionately impaired temporal memory suggested that whatever process mediates judgment about when a stimulus occurred (i.e., recency judgment) was not identical to that which mediates judgments that a stimulus occurred (e.g., recognition judgment). A parallel analysis was also conducted on gr oups classified by thei r FL and MTL function levels on the two temporal ordering tests. The findings indicated that on the novel face recency test, the Low FL group had a lower score on th e within-list items in the immediate recall condition compared to the High FL group. The Low MTL group had a significantly lower score on the between-list discrimination during the delayed recall, and a tendency of a lower score on the within-list items during th e immediate recall compared to the High MTL group. Another piece of evidence that supported both FL and M TL involvements came from the correlational analyses. Several neuropsychological measures that were assumed to tap frontal /executive functioning (e.g., Letter-Number Sequencing test, Animal Fluency, and Trails A & B tests) and medial temporal lobe memory functioning (e.g., Logical Memory, Verbal and Visual Paired Associates, and Visual Reproduction, and CVLT-II) were highly correlated with performance on the temporal order task using novel materials. Furthermore, the results found during the immediate recall, but not delayed recall, be tween High FL versus Low FL groups, and both
104 immediate (marginal, p =.05) and delayed recal l effects between the High MTL versus Low MTL were interesting. Such findings were simila r to the findings found in the Floresco et al. (1997) study, only their study was conducted with anim als, and indicated the length of delay may be one of the critical factor to tease apart the role of hippocampus on the temporal memory (Floresco et al., 1997). On the historical event temporal orde r test, the group with higher MTL function demonstrated a better performance than did the group with lower MTL function. The high and Low FL groups did not demonstrate sign ificant differences on this measure, although significant correlations were f ound between several frontal/executive function measures (e.g., Letter-Number Sequencing test, Trails B, and Arit hmetic test) and performance on the temporal order test using public events. Overall, the evidence obtained from the group analysis and the correlational analysis indicated that both medial temporal and frontal lobe played a role in performance of a temporal order test that used semantic familiar materials as stimuli. Collectively, these results indicated that MCI involves a loss of temporal order information beyond what is predictable from a general loss of item memory. Such information is fragile in MCI individuals, but it is fragile as well in normal old adul ts when compared to younger individuals. Additionally, thes e findings also support the view that association memory for temporal order and source depend additionally on frontal function (Milner et al., 1991; Schacter, 1987; Shimamura & Squire, 1987), as well as memory performance mediated by medial temporal lobe (Kopelman et al., 1997). In sum, the findings obtained from the four associative memory tests indicated that the MCI group demonstrated dispropor tionate difficulties in differ ent aspects of information associative memory as compared to the healthy older adults, who then demonstrated a poorer associative memory compared to healthy younger a dults. Additionally, the evidence that medial
105 temporal lobe plays a primary role in the associ ative memory for different kinds of information emerged consistently from different experiment al tasks mentioned above. However, the evidence regarding the role of fr ontal lobe system on associative memory for different kinds of information was less cohesive.
106 CHAPTER 7 CONCLUSIONS AND DISSCUSIONS The present study aim ed to overcome shortcomings of previous studies by systematically investigating multiple measures of associa tive memory in normal aging and MCI. The underlying question in this study wa s whether associative memory is disproportionately impaired in MCI, and whether impairment in associative memory could serv e as a meaningful predictor of risk to develop dementia later in life. While the la tter question awaits lo ngitudinal confirmation, this study attempted to determine whether associative memory (memory for associated items) was disproportionately impaired compared to me mory for single items in individuals with MCI on several associative memory te sts. Previous work has shown that associative memory, particularly cross-modal associative learni ng, is a sensitive indicator of hippocampus involvement. In addition, structural imaging st udies have demonstrated that MCI individuals have significant hippocampal volume reductions compared to normal healthy elderly and young adults. Therefore, we predicted that MCI individuals would dem onstrate greatest and disproportional difficulty in associ ative memory tasks, particularly for the associations between different kinds of information (cross-modal) co mpared to healthy elde rly. In addition, we expected that MCI individuals would show a greatest decline on the delayed recall condition compared to the immediate recal l condition in both withinand betweensubject analyses, given the well-established role of hippocampus in the long-term memory consolidation process (Frankland & Bontempi, 2005). Additionally, the current study also attempted to determine the degree to which associative memory depends on medial temporal lobe f unctioning and frontal l obe functioning, given possible heterogeneity among older individuals. First, we examin ed heterogeneity in frontal
107 functioning using methods descri bed by Glisky and her colleagues (1995) among healthy elderly and MCI individuals. We then examined its relati onship with associative me mory performance. We predicted that older indivi duals with higher frontal-execu tive skills would demonstrate better performance on associative memory tasks than would elders with lower but still normal frontal-executive skills. Past studies have shown that giving explicit instruction during the encoding phase can improve memory pe rformance only for people who have good frontal/executive functioning. In this present study, we also ma nipulated the instruction types during the encoding phase to explore the phenomen a in normal aging and MCI individuals using an associative memory paradigm. We further e xpected to see an inter action effect of group by instruction, in which the High FL older adults would demonstrate a greater benefit from intentional than incidental instructions (a within-subject comparison) compared to Low FL elders (a between-subject comparis on) on associative memory tasks in both conditions. Disproportionate Impairment on the Associ ative Memory Compared to Item Memory Across experim ental tasks, our results clear ly show that the MCI group demonstrated a disproportionate impairment in associative me mory. Furthermore, th e pattern of greater impairment of associative than item recognition shown on the MCI individuals is consistent with the pattern of performance displayed by pa tients with hippocampal pathology reported by Holdstock et al. (2005), Mayes et al. (2004), Turrizian et al. ( 2004) and Vargha-Khadem et al. (1997). It is a well-established finding that the hippocampus is critically important in associational processing (Eichenbaum, 1997; Moscovitch and Winocur, 1995). Numerous investigations have demonstrated that one of the primary functions of the hippocampus in memory is to bind elements together to form a memory that can later be retrieved. In addition, the MCIs performance on several of the dela yed recall tests (e.g., face-house association) dropped by a significantly larger amount than that of the healthy cont rols from the initial test to
108 the retest. It has been argued that the retest (delayed recall) relies mo re on recollection of the association of the item to its study context than does the initia l test (Aggleton et al., 2000). The pattern of disproportionate impairment on the associative memory shown in the MCI group was also seen in the healthy older c ontrol group when compared to the younger group, which was consistent with previous literature (Kirasic, 2001). One possible reason for the group difference found between young and healthy old groups was that moderate age-related atrophy of hippocampus occurs in a continuous fashion. Second, younger control subjects may use more elaborative encoding strategies than older subjects in memory tasks (Naveh-Benjamin, 2000). Elaborative encoding has been reported to benef it recollection more than familiarity; consistent with this, aging has been found to have a greater detrimenta l effect on recollection than familiarity (Yonelinas, 2001). In the present study, we conducted several t ypes of association tests which included associations between information stored in the same cortical region (e.g., word-word association), association between information re presented in distinct cortical regions (e.g., object-location association), and relational association which focu sed on the flexible expression of relational information (e.g., face-face asso ciation through a face-house connection). One question raised in the literature review was whether hippocampal damage impairs all kinds of associative to the same extent. Collectively, our findings showed that the MCI group, compared with healthy old adults, demonstrated impaired performance on the word-word association, the associations that involved information represented in distinct cortical regions (including the object-location association and temp oral order judgment tests), a nd the relational association.
109 Table 7-1. Effect sizes fo r all associative measures Associative memory measures Cohens d description of relative size Intentional word pair recognition 1 1.09 Large effect Intentional word pair recognition 2 1.15 Very large effect Incidental word pair recognition 1 0.98 Large effect Incidental word pair recognition 2 0.50 Medium effect Semantic-novel pair free recall 1 1.34 Very large effect Semantic-novel pair cued recall 1 1.14 Very large effect Semantic-novel pair recognition 1 0.56 Medium effect Semantic-novel pair free recall 2 1.43 Very large effect Semantic-novel pair cued recall 2 1.07 Large effect Semantic-novel pair recognition 2 1.14 Very large effect Object-location recogniti on 1 0.94 Large effect Object-location recogniti on 2 1.01 Large effect Face house association1 0.62 Medium effect Face house association2 1.40 Very large effect Face recency judgment 1 1.28 Very large effect Face recency judgment 2 1.14 Very large effect Historical event temporal order 1.42 Very large effect Note: The calculation of effect sizes were based on the comparison between healthy old and MCI groups. 1 indicates immediate recall; 2 indicates delayed recall Table 7-1 displays the effect size calculated based on the gr oup effect between healthy old and individuals with MCI for each associative memory measure. The effect sizes for the wordword association (intentional word pair) range from large effect (i mmediate recall) to very large effect (delayed recall). Similarly, the effect si zes for the tasks measure as sociations that involved cross-region information as well as the relational association memory were generally in the large to very large range, particularly for the delaye d recall condition. From th e effect size standpoint, the current data indicate that different associ ative memory measures used in the current study were sensitive to hippocampus dysfunction to a co mparable level. The overall finding in the current study was consistent with findings repo rted by Turriziani et al. (2004) who found that patients with hippocampal lesions performed poorly on associations of same kind information as well as associations of diffe rent kinds of information.
110 Our findings based on experiment 2 and findi ngs on the object-location association test also suggested that free recall was a more se nsitive measure compared to recognition when investigating associative memory. This was consistent with the view that the hippocampus is critical for recollection whereas neocortical re gion are sufficient to support familiarity-based recognition memory decisions (Holdstock et al ., 2002, Mayes et al., 2002). Interestingly, older participants continued to exhi bit poorer associative memory on several associative memory tasks, even when tested usi ng a recognition test which was s upposed to reduce the retrieval demands involved in the free recall tests. Such finding indicated that providing greater amounts of contextual support (i.e., giving recognition tests) does not help the older adults, particularly individuals with MCI, to overcome their memory difficulties for associative information. This might suggest an encoding proficiency impairment among those older participants when performing the associative memory tests. We cannot rule out the possibility that retrieval difficulties are also accounting for the decline in associative memory observed in both normal aging and individuals with mild cognitive impairment. Moreover, healthy older adults benefited from the explicit encoding instruction and the eff ect has the tendency to last for a period of time while the MCI individuals only rece ived a short-term benefit. To summarize, the results of the study ar e in line with findings based on hippocampus lesion patients and lends additional support to th e role of hippocampus in multiple aspects of associative memory. Specifica lly, the findings indicate a disp roportionate impairment of associative memory relative to it em memory in the MCI individua ls (when compared to healthy aging) and normal aging (when compared to youn ger subjects), indicati ng that the associative memory measure may be a sensitive measure to differentiate individuals w ho may be at risk to
111 develop dementia later in life. Verifying this suggestion will, of c ourse, await longitudinal confirmation. Semantic Memory versus Episodic Memory Num erous studies have suggested that the m ild cognitive impairment is characterized by impairment in episodic memory, particularly de layed recall of newly learned materials, with relatively sparing of semantic memory. As a re sult, the findings based on the historical event temporal order test were interesting and une xpected. The healthy old adults demonstrate comparable performance as the younger adul ts; however, the MCI group demonstrated significantly impaired performance on this measur e compared to the healthy older adults. The puzzling question was whether the impaired perfo rmance observed in the MCI group reflected an impaired ability in discriminating the temporal relationship between items, an early sign of breakdown of semantic system, or both. From the temporal relationship standpoint, so me studies have indicated that the aging process affects ones ability to discriminate the temporal sequence (Parkin et al., 1995; Schacter et al., 1991) and frontal lobe dysfunction has been proposed to account for the impairment observed in aging studies. Indeed, the result s based on the recency test with novel faces as stimuli showed that the ability to discrimi nate the temporal sequence between items was decreased in both normal aging and the MCI gro up even after controlling for item memory. Furthermore, the time-tagging mechanism may be compromised in both h ealthy old adults and the MCI individuals based on above description; however, the healthy olde r adults, who have a relatively intact hippocampus comp ared to the MCI individuals, can still rely on trace strength information to make recency judgment while the MCI individuals appeared unable to do so. An alternative possibility is that the impairment in MCI somehow affects trace strength, thereby rendering it unreliable as a s ource of recency judgments.
112 One additional theoretical explanation for the poor performance of MCI patients on the public events ordering test is th at the semantic system, which is affected dramatically in Alzheimers disease, has begun to break down once the patient reaches the MCI stage. Indeed, several recent studies suggest that semantic knowledge brea kdown might start ap pearing as early as in the stage of MCI. For example, a r ecent study conducted by Kraut et al. (2007) reported that 10 out of 35 Amnestic MCI individuals in their sample demonstrated impaired semantic memory measured by a semantic object retrieval test. They further indicated that a major distinction between the MCI indi viduals with semantic impair ment, MCI individuals without semantic impairment, and healthy aging control is the relatively large number of false-positive memory errors. A relatively large false positive ra te was seen in our MCI subjects in Experiment 1 of the present study. Similarly, Estvez-Gonzlez et al. (2004) reporte d that MCI individuals who were diagnosed two years later with Alzhei mers Disease performed significantly worse in a task involving semantic knowledge of famous people than those MCI w ho did not convert to dementia later. The evidence obtained from the studies mentioned above seems to point out that there is a subgroup of Amnestic MCI who demo nstrated semantic knowledge impairment and can be distinguished from the other Amnes tic MCI who do not have semantic knowledge impairment and may have a better prognosis in th e future. Although, with a small sample size, it is likely that some, if not a ll, individuals with MCI in th e current study showed semantic knowledge impairment. In deed, in our MCI sample, we included both MCI amnestic single domain as well as MCI amnestic multiple domains in the analyses. It will be interesting to follow up those people with MCI in the present study and observe th eir prognosis in the future. Overall, the evidence seems to suggest th at impaired time tagging mechanism and semantic knowledge could both account for the findi ngs obtained in the historical event temporal
113 order task. As a result, it is somewhat difficult to tease apart th e two different processes without including other semantic knowledge tests; doing so might be worthy of further study. Frontal Lobe Involvement in Associative Memory and the Heterog eneity Among Older Adults The primary goal of the current study was to examine associative memory in normal aging and individuals with mild cognitive impairment. The role of medial temporal lobe /hippocampus on associative memory is well established and can account for the impaired performance found in the MCI group. In addition to the role of medial temporal lobe in associative memory, we were also interested in investigating the a dditional contributions ma de by frontal-executive impairment that is prevalent in the aging populati on. In the present invest igation, our interest in the heterogeneity among older adults and adults with MCI was mainly focused on the variation of the frontal /executive function and its effect on associative me mory in these populations. In order to examine the relative invol vements of frontal l obe and medial temporal lobe, we adapted Gliskys method and divided subjects based on th eir levels on the frontal lobe and media temporal lobe composite scores. Overall, th e most significant differences were found between the High vs. Low MTL function groups inst ead of the High vs. Low FL function group, suggesting that the medial temporal lobe functio ning plays a more critical role in associative memory function. However, we also found the frontal lobe may contribute to associative memory and that its importance usually does not become salient until the functional integrity of medial temporal lobe was compromised. For exam ple, in the object-locati on association test, the two High MTL groups with either High or Low FL function did not differ si gnificantly in their ability to recall the object-location association. However, in individuals with Low MTL functioning, an intact frontal lobe system may he lp mitigate impairment to some degree and may even allow performances comparable to that seen in patients with High MTL functioning. Such
114 findings are consistent with the view that the frontal lobe is important in initiation and execution of cognitive control processes or strategy usage during a memory process; however, the medial temporal lobe, particularly the hippocampus, is cr itical for the rapid formation of associative memory. An additional implication of these findings is that the Lo w MTL-Low FL group may be a specifically at-risk group that needs to be followed up from both diagnosis and intervention perspectives. Those individuals ma y have higher risk to convert to cognitive disorders, such as dementia, in the future. Moreover, some nove l preventative memory intervention techniques may be found most helpful for people who are in the Low MTL-High FL or High MTL-Low FL group. Of note, the current findings were base d on small sample sizes and the frontal lobe composite scores were not normally distributed in our sample. This might indicate an unrepresentative sample, suggesting caution in inte rpreting the current resu lts. A larger sample size study is warranted in order to confirm the findings reported in the present study. Study Limitations One key limitation to this study is the sam ple size. Challeng es in recruiting individuals with MCI, due to population prevalence in the lo cal community and to the difficulty identifying them prior to the neuropsychological assessmen t and interview phase, resulted in a relatively small sample size for this group as well as a lack of ethnically diverse part icipants. In addition, the individuals who participated in the current study were healthy, highly educated older adults. In fact, the average education le vel of the older adults sample was higher than that of the younger control group. The high education level may se rve as a protective factor, producing cognitive reserve that would mask the underl ying pathological aging process in some individuals. Overall, the small sample size of the MCI group may resu lt in reduced power, since in some analyses trends were observed that did not reach significance. The sample size issue also appeared in the four-group analyses using G liskys factor approach.
115 Additionally, the MCI group had a significantly older mean age compared to the healthy old control group, which was consistent with pr evious literatures that suggested that the incidence of MCI increased signifi cantly with age. To deal with the unmatched age issue, we choose to conduct analyses on healthy contro l that were older than 65 years. Several studies have pointed out the issue of unreliable or unstable diagnosis of MCI, particularly for a cross-sec tional study. We intended to minimize the possibility of misclassifying normal individuals as MCI or vise versa by including a la rge set of clinical neuropsychological battery data which incl uded not only multiple cognitive domains (e.g., language, executive function, and vi suospatial function) and mood measures, but also contained multiple memory measures in conjunction with stri ct exclusion criteria. Moreover, we adapted a consensus conference approach, consisting of a larger number of professionals, to classify individuals as normal or MCI. By doing this, we hoped to achieve a more stable and reliable group classification. Another limitation of the study is the length of the study prot ocol. The study required at least two sessions to complete the Neuropsycholog ical assessment and the experimental tasks. The time required to complete the two sessions ranged between six to nine hours. Some participants, particularly those w ith poor memory, tended to need l onger time to finish the tests, which might have further introduced a confounding factor of cognitive fatigue. Additionally, those individuals with poorer memory often felt frustrated during the test ing sessions. Some of the participants were not able to finish all re quired tasks due to the re asons mentioned above. Although we do not have any data to indicate whether the frustr ation they experience may have directly affected test results, the HO and the MC I group did not differ sign ificantly on measures of depression or state anxi ety collected during testing.
116 While attempting to provide tasks that were neither too difficult for the impaired participants, nor too easy for the cognitivel y intact individuals, including the younger participants, through several pilots during the task development st age, some of the tests still presented substantial difficulty fo r the participants, particularly the older ones. For example, even with the significant group effect, the face rece ncy test seemed to be challenging for the two older groups, in which the MCI demonstrated almost a chance level of performance. The difficulty of the tasks also contributed to the fact that the HO and the MCI groups demonstrated different levels of item memory even w ith multiple exposures during the study phase Future Direction The findings of the current study have provide d additional insight into the nature of associative memory in normal aging and individu als with mild cognitive impairment. In the short-term, a replication of the current study with a large sample size and different subtypes of mild cognitive impairment or groups based on th eir frontal lobe and medial temporal lobe function levels will help to examine the stability of the findings reported in the current study. A longitudinal follow-up of the HO and MCI partic ipants might be able to provide a unique opportunity to investigate the diagnostic and predictive utility of the eval uation of associative memory on eventual conversion to dementia. A post-hoc analysis of MCI individuals who convert to dementia as well as to individuals who show no degenerative decline in cognitive functioning in the future would be valuable a nd may also result in in creased opportunities for intervention. Data from both st ructural and functional imaging in additional to the behavioral data may provide insights regarding the involvem ent of frontal lobe and medial temporal lobe structures in the associative memory.
117 Conclusions In summ ary, in the current study, older adults with Amnestic MCI demonstrated a pattern of disproportionate impairment on the associativ e relative to item memory. Similar but smaller effects were found in our normal elderly sample. The impaired associative memory found in the MCI group applied to association between items of the same kind, association between different kinds of information, inferred relationships, as well as to novel and pre-existing (semantic) associations. These findings suggest that associative memory impairment in MCI transcends the boundary between episodic and semantic memory. Furthermore, our data also indicate the frontal lobe involvement in the associative memory process, although its role is more likely to be secondary to the medial temporal lobe.
118 LIST OF REFERENCES Aggleton, J. P., & Brown, M. (1999). Episodic m e mory, amnesia, and the hippocampal-anterior thalamic axis. The Behavioral and Brain Sciences, 22 425-489. Aggleton, J. P., Vann, S. D., Oswald, C. J., Good, M. (2000). Identifying cortical inputs to the rat hippocampus that subserve allocentric spatial processes: a simple problem with a complex answer. Hippocampus 10, 466-474. Backman, L., Almkvist, O., Andersson, J. L. R., Nordberg, A., Winblad, B., Reineck, R., & Langstrom, B. (1997). Brain activation in young and older adults during implicit and explicit retrieval. Journal of Cognitive Neuroscience, 9 378-391. Backman, L., Small, B. J., Wahlin, A., & Larsson, M. (1999). Cognitive functioning in very old age. In F. I. M. Craik & T. A. Salthouse (Eds.), Handbook of cognitive aging (Vol. 2, pp. 499-558). Mahwah, NJ: Erlbaum. Baltes, P. B., & Lindenberger, U. (1997). Emer gence of a powerful connection between sensory and cognitive functions across the adult life span: A new window to the study of cognitive aging? Psychology and Aging, 12, 12-21. Beck, A. T.; Steer, R. A.; & Brown, G. K. ( 1996). Beck Depression Inve ntory-II; San Antonio, The Psychological Corporation. Benton, A.L. & Van Allen, M.W. (1968). Impairment in facial recognition in patients with cerebral disease. Cortex, 4, 344-358. Bischkopf, J., Busse, A., & Angermeyer, M. C. ( 2002). Mild cognitive impairment-a review of prevalence, incidence and outcome according to current approaches. Acta Psychiatrica Scandinavica, 106 403-414. Bower, J.S. & Schacter, D.L. (1993). Priming of nove l information in amnesic patients. In G. P. Masson & M. E. Masspn (Eds.), Implicit memory: New directions in cognition, development and neuropsychology (pp. 303-326)). New Jersey, Lawrence Erlbaum Associates. Breier, J. I., Leonard, C. M., Ba uer, R. M., Roper, S., Lucas, T. H., & Gilmore, R. L. (1996). Quantified volumes of temporal lobe structures in patients with epilepsy. Journal of neuroimaging, 6, 108-114. Bunsey, M. & Eichenbaum, H. (1996). Conserva tion of hippocampal memo ry function in rats and humans. Nature, 379, 255-257. Busse, A., Bischkopf, J., Riedel-Heller, S. G., & Angermeyer, M. C. (2003). Subclassifications for mild cognitive impairment: prevalence and predictive validity. Psychological Medicine, 33, 1029-1038.
119 Butters, M. A., Kaszniak, A. W., Glisky, E. L., Eslinger, P. J. and Schacter, D. L. (1994). Recency discrimination deficits in frontal lobe patients. Neuropsychology, 8 343-353. Cabeza, R. (2001). Functional neuroimaging of cognitive aging. In R. Cabeza & A. Kingstone (Eds.), Handbook of functional ne uroimaging of cognition (pp. 331-337). Cambridge, MA: MIT Press. Cabeza, R., Grady, C. L., Nyberg, L., McIntos h, A. R., Tulving, E., Kapur, S., Jennings, J., Houle, S., & Craik, F. I. M. (1997). Age-re lated differences in neural activity during memory encoding and retrieval: A positron emission tomography study. Journal of Neuroscience, 17 391-400. Cabeza, R., & Nyberg, L. (2000). Imaging cogn ition II: An empirical review of 275 PET and fMRI studies. Journal of Cognitive Neuroscience, 12 1-47. Castel, A. D. & Craik, F. I. M. (2003). The eff ects of aging and divided attention on memory for item and associative information. Psychology and Aging 18, 873-885. Cave, C. B., & Squire, L. R. (1992). Intact ve rbal and nonverbal short term memory following damage to the human hippocampus. Hippocampus, 2, 151. Chalfonte, B. L., & Johnson, M. K. (1996). Feature memory and binding in young and older adults. Memory & Cognition, 24 403-416. Chalfonte, B. L., Verfaellie, M., Johnson, M. K., & Reiss, L. (1996). Spatial location memory in amnesia: Binding item and location information under incidental and intentional encoding conditions. Memory, 4 591-614. Charlson, M. E., Pompei, P., Ales, K. L. & MacKenzie, C. R. (1987). A new method of classifying prognostic comorbidity in longitu dinal studies: development and validation. Journal of Chronic Disease, 40 373-383. Cherry, K. E., & Park, D. C. (1993). Individual difference and contextual variables influence spatial memory in younger and older adults. Psychology and Aging, 8 517-526. Cohen, J. E. (1988). Estimating the effects of successful malarial control programmes on mortality. Population Bulletin of the United Nations, 25 6-26. Cohen, N. J. & Eichenbaum, H. (1993). Memory, amnesia and the hippocampal system Cambridge, MA: MIT Press. Cohen, N. J., Poldrack, R. A., & Eichenbaum, H. (1997). Memory for items and memory for relations in the procedural/d eclarative memory framework. Memory, 5, 131-178. Craik, F. I. M., & Morris, L. W., Morris, R. G., & Loewen, E. R. (1990). Relations between source amnesia and frontal lobe functioning in older adults. Psychology and Aging, 5 148151.
120 Craik, F. I. M., & McDowd, J.M. (1987). Age differences in recall and recognition. Journal of Experimental Psychology: Learning, Memory, and Cognition 13, 474-479. Crook, T. H., Feher, E. P., & Larrabee, G. J. (1992). Assessment of memory complaint in ageassociated memory impairment: The MAC-Q. International Psychogeriatrics, 4 165-76. Delis, D. C., Kramer, J. H., Kaplan, E., & Ober, B.A. (1987). The California Verbal Learning Test San Antonio, TX: The Psychological Corporation. Delis, D.C., Kramer, J.H., Kaplan, E., & Ober, B.A. (2000). California Verbal Learning Test II. San Antonio, TX: Psychological Corporation. Denney, N. W., Miller, B. V., Dew, J. R., & Le vav, A. L.(1991). An adult development study of contextual memory. Journal of Gerontology, 46 44-50. Eichenbaum, H. (1997). Declarative memo ry: Insight from cognitive neurobiology. Annual Review of Psychology, 48 547-572. Eichenbaum, H., & Cohen, N. J. (2001). From conditioning to conscious recollection. Memory systems of the brain Oxford: Oxford University Press. Eichenbaum, H., Otto, T., & Cohen, N.J. (1994). Two functional components of the hippocampal memory system. The Behavioral and Brain Sciences, 17 449-517. Estvez-Gonzlez, A., Garca-Snchez, C., Boltes, A., Otermn, O., Pascual-Sedano, B., Gironell, A., & Kulisevsky J. (2004). Seman tic Knowledge of Famous People in Mild Cognitive Impairment and Progression to Alzheimer's Disease. Dementia and Geriatric Cognitive Disorders, 17, 188-195. Fabiani, M., & Friedman, D. (1997). Dissocia tions between memory for temporal order and recognition memory in aging. Neuropsychologia, 35 129-141. Ferguson, S. A., Hashtroudi, S., & Johnson, M. K. (1992). Age difference in using sourcerelevant cues. Psychology and Aging, 7 443-452. Floresco S. B., Seamans, J. K., Phillips, A. G. (1997). Selective roles for hippocampal prefrontal cortical, and ventral striatal circuits in ra dial-arm maze tasks with or without a delay. Journal of Neuroscience, 17 ,1880-1890. Folstein, M. F., Folstein, S. E., & Mc Hugh, P. R. (1975). "Mini Mental State." Journal of Psychiatric Research, 12 189-198. Fortin, N., Agster, K., & Eichenba um, H. (2002). Critical role of the hippocampus in memory for sequences of events. Nature Neuroscience, 5 458-462. Frankland, P. W. & Bontempi, B. (2005). The organization of recent and remote memories. Nature Reviews Neuroscience 6, 119-130.
121 Gilmore, R. L., Childress, M. D., Leonard, C ., Quisling, R., Roper, S., Eisenschenk, S., & Mahoney, M. (1995). Hippocampal vol umetrics differentiate patients with temporal lobe epilepsy and extratemporal lobe epilepsy. Archives of Neurology, 52 819-824. Giovanello, K. S., Schnyer, D. M ., & Verfaellie, M. (2004). A crit ical role for the anterior hippocampus in relational memory: evidence from an fMRI study comparing associative and item recognition. Hippocampus, 14 5-8. Glisky, E. L., Polster, M. R., & Routhieaux, B. C. (1995). Double dissoci ation between item and source memory. Neuropsychology, 9, 229-235. Golomb, J., Kluger, A., de Leon, M.J., Ferris, S. H., Convit, A., Mittelman, M. S., Cohen, J., Rusnick, H., De Santi, S., & George, A. E.(1994). Hippocampal formation size in normal human aging: A correlate of delaye d secondary memory performance. Learning & Memory, 1, 45-54. Grady, C. L., McIntosh, A. R., Horwitz, B., Mais og, J. M., Ungerleider, L. G., Mentis, M. J., Pietrini, P., Schapiro, M. B., & Haxby, J. V. (1995). Age-related reductions in human recognition memory due to impaired encoding. Science, 269 218-221. Hasher, I., & Zacks, R. T., (1988). Working memo ry, comprehension, and ag ing: A review and a new view. In G. H. Bower (Ed.), The psychology of learning and motivation: advances in research and theory (vol. 22, pp.193-225). San Diego: Academic Press. Heckers, S. Zalesak, M. Weiss, A. P., Ditma n, T., & Titone, D. (2004). Hippocampal activation during transitive inference in humans. Hippocampus, 14, 153-162. Helveston, W., Gilmore, R., Roper, S., Mastin, S., Quisling, R., Drane, W., Eikman, E., Leonard, C., Browd, S., & Childress, D. (1996). Intracta ble temporal lobe epilepsy: comparison of positron emission tomography with qu alitative and quantitative MR. American Journal of Neuroradiology, 17 1515-1521. Hockley, W. E. (1991). Recognition memory for ite m and associative information: a comparison of forgetting rates. In W. E. Hockley & S. Lewandowsky (Eds.), Relating theory and data: Essays on human memory in honor of Bennet B. Murdock (pp. 227-248). Hilladale, NJ: Erlbaum. Hockley, W. E. (1992). Item versus associative information: Further comparisons of forgetting rates. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18 13211330. Hockley, W. E., & Cristi, C. (1996). Tests of encoding tradeoffs between item and associative information. Memory & Cognition, 24 202-216. Holdstock, J. S., Mayes, A. R., Roberts, N., C ezayirli, E., Isaac, C. L., O'Reilly, R., & Norman, K. (2002). Under what conditions is recognition spared relative to recall after selective hippocampal damage in humans? Hippocampus, 12 341-351.
122 Hughes, C., Berg, L., Danziger, W. L., Cohen, L.A., & Martin,R. L. (1982). A new clinical scale for the staging of dementia. British Journal of Psychiatry 140 566-572. Hunkin, N. M., & Parkin, A. J. (1993). Recency judgments in WernickeKorsakoff and post encephalitic amnesia: influences of proact ive interference and retention interval. Cortex, 29, 485-499. Hunkin, N. M., Parkin, A. J., Longmore, B. E. (1994). Aetiological variation in the amnesic syndrome: comparisons using the list discrimination task. Neuropsychologia, 32 819-825. Idaka, T., Anderson, N. D., Kapur, S., Cabeza, R., & Craik, F. I. M. (2000). The effect of divided attention on encoding and retr ieval in episodic memory re vealed by positron emission tomography. Journal of Cognitive Neuroscience, 12 267-280. Insausti, R., Juottonen, K., Soininen, H., Insausti, A. M., Partanen, K., Vainio, P., Laakso, M. P., & Pitkanen, A. (1998). MR volum etric analysis of the human entorhinal, perirhinal, and temporopolar cortices. American Journal of Neuroradiology, 19, 659-671. Jackson, O. III., & Schacter, D. L. (2004). Encodi ng activity in anterior medial temporal lobe supports subsequent associative recognition. Neuroimage, 21, 456-462. Jacoby, L. L. (1991). A process dissociation fram ework: Separating automatic from intentional use of memory. Journal of Memory and Language, 30 513-541. Jacoby, L. L. & Dallas, M. (1981). On the rela tionship between autobiographical memory and perceptual learning. Journal of Experimental Psychology: General, 110 306. Jacoby, L. L., Woloshyn, V., & Kelley, C. (1989) Becoming famous without being recognized unconscious influences of memory produced by dividing attention. Journal of Experimental Psychology: General, 118 115-125. Janowsky, J. S., Shimamura, A. P., & Squire, L. R. (1989). Source memory impairment in patients with frontal lobe lesions. Neuropsychologia, 27 1043-1056. Kaplan, E. R., Goodglass, H., Weintraub, S. (1983). The Boston Naming Test (2 ed.) Philadelphia: Lea & Febiger. Kapur, S., Craik, F. I. M., Tulving, E., Wils on, A. A., Houle, S., & Brown, G. M. (1994). Neuroanatomical correlates of encoding in ep isodic memory: levels of processing effect. Proceedings of the National Academy of Science U.S.A., 91 2008. Kausler, D. H., Lichty, W., & Davis, R. T. (1985). Temporal memory for performed activities: Intentionality and a dult age differences. Developmental Psychology, 21 1132-1138. Kausler, D. H., & Phillips, P. L. (1989). Inst ructional variation and adult age differences in activity memory. Experimental Aging Research, 14 195-199.
123 Kausler, D. H., & Puckett, J. M. (1980). A dult age difference in recognition memory for a nonsemantic attribute. Experimental Aging Research, 6 349-355. Kausler, D. H., & Puckett, J. M. (1981). Adu lts age difference in memory for sex of voice. Journal of Gerontology, 36 44-50. Kessels, R. P. C., Dehann, E. H.F., Kappelle, L. J., & Postma, A. (2001). Varieties of human spatial memory: A meta-analysis on the effects of hippocampal lesions. Brain Research Reviews 35, 295-303 Killiany, R. J., Hyman, B. T., Gomez-Isla. T., Moss, M. B., Kikinis, R., Jolesz, F., Tanzi, R., Jones, K., & Albert, M. S. (2002). MRI measur es of entorhinal cortex vs. hippocampus in preclinical AD. Neurology, 58 1188-1196. Kirasic, K.C. (2001). Aging and spatial behavior in the elderly adult. In: Kitchin, R. and Freundschun, S. (Eds.). Cognitive mapping: past, present, and future. Routledge frontiers of cognitive science (pp. 166-178). Taylor and Francis, London. Kopelman, M. D. (1989). Remote and autobiogra phical memory, temporal context memory, and frontal atrophy in Korsakoff and Alzheimer patients. Neuropsychologia, 27 437-460. Kopelman, M. D., Stanhope, N., & Kingsley, D. ( 1997). Temporal and spatial context memory in patients with focal frontal, temporal lobe, and diencephalic lesions. Neuropsychologia, 35 1533-1545. Kraut, M. A., Cherry, B., Pitcock, J. A., Anand, R., Li, J., Vestal, L., Henderson, V. W., & Hart, J. (2007). The semantic object retrieval test (SORT) in amnestic mild cognitive impairment. Cognitive and Behavioral Neurology, 20 62-67. Kroll, N. E. A., Knight, R. T., Metcalf. J., Wolf, E. S., & Tulving, E. (1996). Cohesion failure as a source of memory illusions. Journal of Memory and Language, 35 176-196. Lawton, M. P. (1988). Scales to measur e competence in everyday activities. Psychopharmacology Bulletin, 24 609-614. Lawton MP, Brody EM. (1969). Assessment of older people: self-maintaining and instrumental activities of daily living. Gerontologist, 9 179186. Light, L. L. (1991). Memory and aging: Four hypotheses in search of data. Annual Review of Psychology, 43, 333-376. Maguire, E. A., Burgess, N., Donnett, J. G., Fr ackowiak, R. S., Frith, C. D., & OKeefe, J. (1998). Knowing where and getting there: A human navigation network. Science, 280 921-924. Mandler, G. (1980). Recognizing: The judgment of previous occurrence. Psychological Review, 87, 252.
124 Mangels, J. A. (1997). Strategic processing and memory for tem poral order in patients with frontal lobe lesions. Neuropsychology, 11, 207-211. Mayes, A. R., Holdstock, J. S., Isaac, C. L., Hunkin, N. M., & Roberts, N. (2002). Relative sparing of item recognition memory in a pati ent with adult-onset damage limited to the hippocampus. Hippocampus, 12, 325-340. Mayes, A. R., Holdstock, J. S., Isaac, C. L., Montaldi, D., Grigor, J., Gummer, A., Cariga, P., Downes, J. J., Tsivilis, D., Gaffan, D., & Norman, K. A. (2004). Associative recognition in a patient with selective hippocampal lesions and relatively normal item recognition. Hippocampus, 14, 763-784. Mayes, A. R., Isaac, C. L., Downes, J. J., Holdstock, J. S., Hunkin, N. M., Montaldi, D., MacDonald, C., Cezayirli, E., & Roberts, J. N. (2001). Memory for single items, words pairs and temporal order in a patient with selective hippocampal lesions. Cognitive Neuropsychology, 18 97-123. Mayes, A. R., Meudell, P. R., & Pickering, A. (1985). Is organic amnesia caused by a selective deficit in remembering contextual information? Cortex, 21, 167-202 Milner, B. (1982). Some cognitive eff ects of frontal-lobe lesions in man. Philosophical Transactions of the Royal Society of L ondon. Series B, Biological Sciences, 298 211. Milner, B., Corsi, P., & Leonard, G. (1991). Fron tal-lobe contribution to recency judgments. Neuropsychologia, 29, 601-618. Morris, J. C. (1993). The Clinical Dementia Ra ting (CDR): current vers ion and scoring rules. Neurology, 43 2412 Moscovitch, M. (2000). Theories of memory and c onsciousness. In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory (pp. 609-625). New York: Oxford University Press. Moscovitch, M., & Winocur, G. (1992). The neurops ychology of memory and aging. In F. I. M. Craik & T.A. Salthouse (Eds.), The handbook of aging and cognition (pp. 315-372). Hillsdale, NJ: Erlbaum. Moscovitch, M., & Winocur, G. (1995). Frontal lobes, memory, and aging. Annals of the New York Academy of Sciences. 769, 119-150. Moser, D. J., Bauer, R. M., Gilmore, R. L., Dede D. E., Fennell, E. B., Algina, J. J., Jakus, R., Roper, S. N., Zawacki, T. M., & Cohen, R. A. (2000). Electroencephalographic, volumetric, and neuropsychologica l indicators of seizure focu s lateralization in temporal lobe epilepsy. Archives of Neurology, 57 707-712. Murdock, B. B., Jr. (1982). A theory for the storage and retrieval of item and associative information. Psychological Review, 89 609-626.
125 Naveh-Benjamin, M. (2000). Adult age differen ces in memory performance: Tests of an associative deficit hypothesis. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26 1170-1187. Naveh-Benjamin, M., Hussain, Z., Guez, J., & Bar-On, M. (2003). Adult age differences in episodic memory: Further support for an associative-deficit hypothesis. Journal of Experimental Psychology: Learning, Memory, and Cognition, 29, 826-837. Nelson, D. L., McEvoy, C. L., & Schreiber, T. A. (1998). The University of South Florida word association, rhyme, and word fragment norms. Retrieved July, 20, 2004, from the University of South Florida Web site: http://www.usf.edu/FreeAssociation/ Newman, M. C., Allen, J. J. B., & Kaszniak, A. W. (2001). Tasks for assessing memory for temporal order versus memory for items in aging. Aging, Neuropsychology, and Cognition, 8 72-78. Nyberg, L., Cabeza, R., & Tulving, E. (1996). P ET studies of encoding and retrieval: The HERA model. Psychonomic Bulletin and Review, 3 135-148. OReilly, R.C. & Rudy, J.W. (2001). Conjunctive representations in learning and memory: principles of cortical and hippocampal function. Psychology Review, 108 311-345. Owen, A. M., Downes, J. D., Sahakian, B. J., Polkey, C. E., & Robbins, T. W (1990). Planning and spatial working memory followi ng frontal lobe lesions in man. Neuropsychologia, 28 1021-1034. Owen, A. M., Sahakian, B. J., Semple, J., Polkey C. E., & Robbins, T. W. (1995). Visuospatial short term recognition memory and learning af ter temporal lobe excisions, frontal lobe excision or amygdalo-hippocampectomy in man. Neuropsychologia, 33, 1-24. Park, D. C., & Puglisi, J. T. (1985). Older adults memory for th e color of pictures. Journal of Gerontology, 40, 198-204. Park, D. C., Puglisi, J. T., Lutz, R. (1982). Spatial memory in older adults: effects of intentionality. Journal of Gerontology, 37 330-335. Parkin, A. J., Hunkin, N. M., & Walter, B. M. (1995). Relationships between normal aging, frontal lobe function, and memory for temporal and spatial information. Neuropsychology, 9 304-312. Parkin, A. J., & Leng, N. R. C. (1992). Neuropsychology of amnesic syndromes Basingstoke, UK: Taylor & Francis. Parkin, A. J., Leng, N. R. C., & Hunkin, N.M. (1990). Differential sensitivity to context in diencephalic and temporal lobe amnesia. Cortex, 26, 373-380. Perlmutter, M., Metzger, R., Nezworski, T., & M iller, K. (1981). Spatial and temporal memory in 20 and 60 year olds. Journal of Gerontology, 36 59-65.
126 Petersen, R. C., Doody, R., Kurz, A., Mohs, R. C ., Morris, J. C., Rabins, P. V., Ritchie, K., Rossor, M., Thal, L., & Winblad, B. (2001) Current concepts in mild cognitive impairment. Archives of Neurology, 58 1985-1992. Petersen, R. C., Smith, G. E., Waring, S. C., Ivni k, R. J., Tangalos, E. G., & Kokmen, E. (1999). Mild cognitive impairment: clinical characterization and outcome. Archives of Neurology, 56, 303-308. Pickering, A., Mayes, A. R., & Fairbairn, A. F. (1989). Amnesia and memory for modality information. Neuropsychologia, 27, 249. Preston, A. R., Shrager, Y., Dudukovic, N. M., & Gabrieli, J. D. E. (2004). Hippocampal contribution to the novel use of relationa l information in declarative memory. Hippocampus, 14, 148-152. Prull, M. W., Gabrieli, J. D. E., & Bunge, S. A. (1999). Age-related changes in memory: A cognitive neuroscience perspective. In F.I.M. Craik & T.A. Salthouse (Eds.), Handbook of cognitive aging (Vol. 2, 99. 91-153). Mahwah, NJ: Erlbaum. Rabinowitz, J. C., Ackerman, B. P., Craik, F. I., & Hinchley, J. L. (1982). Aging and metamemory: the roles of relatedness and imagery. Journal of Gerontology, 37 688-695. Raz, N., Gunning, F. M., Head, D., Dupuis, J. H., McQuain, J., Briggs, S. D., Loken, W. J., Thornton, A. E., & Acker, J. D. (1997). Se lective aging of the human cerebral cortex observed in vivo: Differential vulnerabil ity of the prefrontal gray matter. Cerebral Cortex, 7, 268-282. Reed, J. M., & Squire, L. R. (1997). Impaired recognition memory in patients with lesions limited to the hippocampal formation. Behavioral Neuroscience, 111 667. Reinitz, M. T., & Alexander, R. (1996). Mechanisms of facilitation in primed perceptual identification. Memory and Cognition, 24 129-135. Reitan, R. M. (1958). Validity of the Trail Making te st as an indicator of organic brain damage. Perceptual and Motor Skills, 8 271-276. Salthouse, T. A., Hancock, H. E., Meinz, E. J., & Hambrick, D. Z. (1996). Interrelations of age, visual acuity, and cognitive functioning. Journal of Gerontology: Psychological Sciences, 51, 317-330. Schacter, D. L. (1987). Memory, amne sia, and frontal lobe dysfunction. Psychobiology, 15, 2137. Schacter, D. L., Harbluk, J. L., & McLachlan, D. R. (1984). Retrieval wi thout recollection: An experimental analysis of source amnesia. Journal of Verbal Lear ning and Verbal Behavior, 23, 593-611.
127 Schacter, D. L., Kaszniak, A. W ., Kihlstrom, J. F., & Valdiserri M. (1991). The relation between source memory and aging. Psychology and Aging, 6 559-568. Schmitter-Edgecombe, M., & Simpson, A. L. ( 2001). Effects of age and intentionality on content memory and temporal memory for performed activities. Aging, Neuropsychology, and Cognition, 8 81-97. Sheikh, J.I. & Yesavage, J.A. (1986). Geriatri c Depression Scale: Recent evidence and development of a shorter version. Clinical Gerontologist, 5, 165-173. Shimamura, A. P., Janowski, J. S., & Squire, L. R. (1990). Memory for the temporal order of events in patients with frontal lobe lesions and amnesic patients. Neuropsychologia, 28 803-813. Shimamura, A. P., & Squire, L. R. (1987). A neuropsychological study of fact memory and source amnesia. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 464-473. Shoqeirat, M. A., & Mayes, A. R. (1991). Dispr oportionate incidental sp atial-memory and recall deficits in amnesia. Neuropsychologia, 29, 749-769. Smith, M. L. (1988). Recall of spatial location by the amnesic patient H.M. Brain Cognition, 7 178-183. Spielberger C, Gorsuch R, Lushene R. (1970). STAI manual for State-Trait Anxiety Inventory Palo Alto, CA: Consulting Psychologists Press. Spencer, W. D., & Raz, N. (1995). Differentia l effects of aging on memory for content and context: A meta-analysis. Psychology and Aging, 10 527-539. Spreen, O. & Benton, A. L. (1997). Neurosensory Center Comprehensive Examination for Aphasia (NCCEA)-revised University of Victoria Neuropsychological Laboratory, Victoria. Squire, L. R. (1982). Comparisons between form s of amnesia: Some deficits are unique to Korsakoffs syndrome. Journal of Experimental Psychology: Learning, Memory and Cognition, 8 560-571. Squire, L. R., Nadel, L., & Slater, P.C. (1981) Anterograde amnesia and memory for temporal order. Neuropsychologia, 19 141-145. Squire, L. R., & Zola-Morgan, S. (1991). Th e medial temporal lobe memory system. Science, 253, 1380-1386. Stuss, D.T., & Benson, D.F., (1986). The frontal lobes New York: Raven Press.
128 Tombaugh, T.N., Kozak, J., & Rees L. (1999). Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming. Archives of Clinical Neuropsychology, 14, 167-177. Tulving, E., Kapur, S., Markowitsch, H. J., Cr aik, F. I., Habib, R., & Houle, S. (1994). Neuroanatomical correlates of retrieval in ep isodic memory, audito ry sentence recognition. Proceedings of the National Academy of Science U.S.A., 91, 2012-2015. Tulving, E., Markowitsch, H. J., Craik, F. E., Habib, R., & Houle, S. (1996). Novelty and familiarity activations in PET studies of memory encoding and retrieval. Cerebral Cortex, 6, 71-79. Turriziani, P., Fadda, L., Caltagirone, C., Carlesimo, G. A. (2004). Recognition memory for single items and for associat ions in amnesic patients. Neuropsychologia, 42 426-433. Ungerleider, L. G. & Mishkin, M.(1982). Two cortical visual systems. In: D. J. Ingle, M.A. Goodale, & R. J.W. Mansfield (Eds.). Analysis of visual behavior (pp 549-586). Cambridge (MA): MIT Press. Vargha-Khadem, F., Gadian, D. G., Watkins, K. E., Connelly, A., Wan Paesschen, W., & Mishkin, M. (1997). Differentia l effects of early hippocam pal pathology on episodic and semantic memory. Science, 277 376-380. Verfaellie, M., & Treadwell, J. R.(1993). St atus of recognition memory in amnesia. Neuropsychology, 7 5. Wechsler, D. (1987). Wechsler memory scale revised New York: Psychological Corporation. Wechsler, D. (1997). Wechsler adult intelligence scale III San Antonio, TX: Psychological Corporation, Harcourt Brace. Wechsler, D. (1999). Wechsler abbreviated scale of intelligence San Antonio, TX: Psychological Corporation, Harcourt Brace. Wolf, H., Jelic, V., Gertz, H. J., Nordberg, A., Julin, P., and Wahlund, L. O. (2003). A critical discussion of the role of neuroimaging in mild cognitive impairment. Acta Neurologica Scandinavica. Supplementum 179, 52-76. Yonelinas, A. P., Hopfinger, J. B., Buonocore, M. H., Kroll, N. E., & Baynes, K. (2001). Hippocampal, parahippocampal and occipitaltemporal contributions to associative and item recognition memory: An fMRI study. NeuroReport, 12, 359-363.
129 BIOGRAPHICAL SKETCH Yu-Ling Chang was born in Ping-Tong, Taiwan. She received her Bachelor of Science degree in psychology in 1997 and her Master of Science degree in psychology with a concentration in neuropsychology in 2001 from the National Taiwan University. She completed a pre-doctoral internship at the University of California, San Diego/VAMC La Jolla and received her Ph.D. in clinical psychology w ith a concentration in neuropsychology from the University of Florida in the summer of 2008. Her research in terests include cognitive functioning in older adults and patients with neurolog ical conditions, and the applicati on of structural and functional neuroimaging techniques to study cognitive function and brain plasticity.