Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Explicit/implicit memory performance and memory strategies in Alzheimer patients Gallie, Karen A. 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1993_fall_phd_gallie_karen.pdf [ 10.97MB ]
JSON: 831-1.0098809.json
JSON-LD: 831-1.0098809-ld.json
RDF/XML (Pretty): 831-1.0098809-rdf.xml
RDF/JSON: 831-1.0098809-rdf.json
Turtle: 831-1.0098809-turtle.txt
N-Triples: 831-1.0098809-rdf-ntriples.txt
Original Record: 831-1.0098809-source.json
Full Text

Full Text

EXPLICIT/IMPLICIT MEMORY PERFORMANCE AND MEMORY STRATEGIES INALZHEIMER PATIENTSbyKAREN ANN GALLIEB.Sc., University of Victoria, 1980M.A., University of British Columbia, 1985A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESNEUROSCIENCE PROGRAMWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIASeptember 1993© Karen A. Gallie, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature) Department of ^Net/ROSCie/ICe MyrathThe University of British ColumbiaVancouver, CanadaDateDE-6 (2/88)iiABSTRACTResearch has shown that implicit memory can be spared insome amnesic patients, even when explicit memory is severelyimpaired. In contrast, results from studies with Alzheimerpatients are mixed. This investigation re-explored these latterfindings in a series of experiments, many that differed fromthose used in previous work, to find out how the performance ofAD patients with different levels of functional impairmentcompares to that of non-institutionalized controls.The first part of this dissertation explored Alzheimerpatients (AD) and age-matched controls' performance on fourparallel forms of explicit and implicit memory tests employingwritten and spoken words, pictures, and common objects (e.g.,toothbrush). For each test, subjects encoded critical targets byfirst identifying, and then generating a personal meaning foreach item. The second part of this dissertation consists ofthree experiments that explored whether encoding and retrieval(i.e., test) strategies could be used to elevate the explicitmemory performance of AD patients.The average age of the 20 AD patients (9 possible and 11probable using NINCDS-ADRDA criteria; 13 female, 7 male) was 70.8years (range = 50-84 years, SR = 8.2) with a mean of 12.5 yearsof education (range = 6-18 years, SD = 3.3). The FunctionalRating Scale (Tuokko & Crockett, 1989) indicated that 6 of theseiiifunctionally impaired. The average age of the 40 controlsubjects (27 female, 13 male) was not statistically differentfrom that of the AD patients (67.8 years, range = 51-89 years, 2p= 9.9) although controls reported a significantly greater numberof years'of education (i.e., H = 14.83 years, range = 8-27 years,= 4.2).The overall hypothesis of part one was that differences inthe memory performance of AD patients versus controls would besmaller on implicit than explicit tests. Results showed thatthis hypothesis was supported for tests of written and spokenwords (i.e., using category completion and category cued recalltests) and for common objects in a perceptual learning condition(i.e., the same objects re-presented over two trials using atactile identification and a tactile recognition test). Incontrast, this hypothesis was not supported for tests usingpicture materials (i.e., using a picture fragment and a picturerecognition test) or for common objects in a skill-based learningcondition (i.e., different objects presented over two trials in atactile identification and a tactile recognition test).These findings support previous reports that AD patientsexhibit impaired performance on picture fragment completiontests. In contrast, these results do not support previous workwhich has indicated that performance on semantic priming tests isalso impaired in these patients. This study extends current workby finding that implicit memory for spoken words and commonobjects need not be impaired in these patients. In addition,ivthat AD patients with mild, moderate, and severe levels offunctional impairment can show a similar magnitude of priming onsome tests. Results from this study are discussed in terms ofthe influence that the encoding task, stimuli, and test form mayhave had on memory performance.The overall hypothesis for part two was that the explicitmemory performance of AD patients would be elevated usingspecific encoding and retrieval conditions. This hypothesis wassupported in the levels-of-processing and partially supported forthe Subject Performed-Experimenter Performed (i.e., SPT-EPT) andthe multisensory (i.e., See, Say, and Do) experiments. Resultsfrom this study extends previous work by showing that ADpatients' explicit memory abilities can be significantly elevatedwhen performance and meaning-generated encoding strategies areused in combination with a cued recall or recognition testretrieval condition. In addition, the ability of these encodingand retrieval strategies to raise the explicit memory performanceof AD patients was found to decline as these patients' level offunctional impairment increased. These findings are discussed interms of the possible mechanisms by which these encoding andretrieval strategies influenced the memory performance of ADpatients.TABLE OF CONTENTSABSTRACT^ page iiTABLE OF CONTENTS^ page vLIST OF TABLES page viiLIST OF FIGURES^ page ixLIST OF APPENDICES page xACKNOWLEDGEMENTS^ page xiCHAPTER ONE: INTRODUCTION AND OVERVIEWStudy OneStudy TwoSummary of ResultsRemaining Chapterspage 1page 3page 6page 7page 9CHAPTER TWO: INTRODUCTION TO ALZHEIMER'S DISEASEChapter Overview^ page 11Behavioral and Anatomical Characteristics page 11Neuropathological Development andChanges in Memory Associated withDamage to the: Hippocampus^page 14Hippocampus andSurrounding Areas^page 22Outside theHippocampus^page 28CHAPTER THREE: EXPLICIT AND IMPLICIT MEMORY page 34Chapter Overview^ page 34Introduction page 34Multiple Systems Model^ page 36Process Model page 37Neuropsychological Investigations^page 44Cognitive Investigations^page 59Study One^ page 67CHAPTER FOUR: MEMORY STRATEGIES page 70Chapter Overview^ page 70Introduction page 70Retrieval Strategies^ page 70Encoding Strategies page 71Subject Performed Tasks (SPTs)^page 73-Levers—af—Procesoing— ^page^77Theoretical Explanations page 82Study Two^ page 84CHAPTER FIVE: METHOD page 86Chapter Overview page 86Subjects page 86Recruitment and Selection Criteria page 86Comparison of Controls and AD Patients page 90Medications page 90Demographics page 91Explicit and Implicit Memory Tests page 91Materials page 91Procedure page 95Memory Strategy Experiments page 105Materials page 105Procedure page 107CHAPTER SIX: RESULTSChapter Overview page 120Overall Design and Analyses page 120Memory Tests page 122Tests of Written Word Materials page 123Tests of Spoken Word Materials page 129Tests of Picture Materials page 133Tests of Object Materials page 138Memory Strategy Experiments page 145Levels of Processing page 146SPT/EPT page 151Multisensory page 157CHAPTER SEVEN: DISCUSSION page 164Chapter Overview page 164Part One page 165Part Two page 177Limitations of Work page 187Future Work page 190Contributions page 191REFERENCES^ page 193APPENDICES page 218viviiList of TablesPAGETable 1^Demographic Characteristics of Controls and^113Alzheimer Patients.Table 2^Medications taken by Controls and Alzheimer^114Patients.Table 3^Statistical Comparison of Controls versus ADPatients.^ 115Table 4^Explicit and Implicit Memory Tests Used in StudyOne. 116Table 5^Counterbalancing Methods Used for Explicit andImplicit Memory Tests.^ 117Table 6 Summary of Memory Strategy Experiments Used inStudy Two.^ 118Table 7 Counterbalancing Methods Used for Memory Strategy 119Experiments.TableG-8a Controls' and AD Patients' Performance on^306Category Cued Recall and Category CompletionTests for Written Word Materials.TableG-8b Two Factor Repeated Measures MANCOVA of Controls' 307and AD Patients' Performance on Category CuedRecall and Category Completion Tests for WrittenWord Materials.TableG-9a Controls' and AD Patients' Performance on^308Category Cued Recall and Category CompletionTests for Spoken Word Materials.TableG-9b Two Factor Repeated Measures MANCOVA of Controls' 309and AD Patients' Performance on Category CuedRecall and Category Completion Tests for SpokenWord Materials.TableG-10a Controls' and AD Patients' Performance on^310Picture Recognition and Picture FragmentCompletion Tests.TableG-10b Two Factor Repeated Measures MANCOVA of^312Controls' and AD Patients' Performance on PictureRecognition and Picture Fragment Completion Tests.viiiList of TablesTableG-ila Controls' and AD Patients' Performance on^313Tactile Recognition and Tactile IdentificationTests.TableG-lib Two Factor Repeated Measures MANCOVA of Control 315and AD Patients' Performance on the TactileRecognition and Tactile Identification Tests (OldMaterials Condition).TableG-lic Two Factor Repeated Measures MANCOVA of Control 316and AD Patients' Performance on the TactileRecognition and Tactile Identification Tests (NewMaterials Condition).TableG -12a Controls' and AD Patients' Performance in the^317Levels of Processing Memory Strategy Experiment.TableG-12b Three Factor Repeated Measures MANCOVA of^318Controls' and AD Patients' Performance in theLevels of Processing Experiment.TableG-13a Controls' and AD Patients' Performance in the^319SPT-EPT Memory Strategy Experiment.TableG-13b Three Factor Repeated Measures MANCOVA of^320Controls' and AD Patients' Performance in theSPT-EPT Memory Strategy Experiment.TableG -14a Controls' and AD Patients' Performance in the^322Multisensory Memory Strategy Experiment.TableG-14b Three Factor Repeated Measures MANCOVA of^323Controls' and AD Patients' Performance in theMultisensory Memory Strategy Experiment.ixList of FiguresPAGEFigure 1 Diagram of the Brain and Hippocampal Formation.^21Figure 2 Controls' and Alzheimer Patients' Performance on 125Category Cued Recall and Completion Tests forWritten Word Materials.Figure 3 Controls' and Alzheimer Patients' Performance on 130Category Cued Recall and Completion Tests forSpoken Word Materials.Figure 4 Controls' and Alzheimer Patients' Performance on 135Picture Recognition and Fragment CompletionTests.Figure 5 Controls' and Alzheimer Patients' Performance on 141Tactile Recognition and Identification Tests.Figure 6 Controls' and Alzheimer Patients' Performance on 148Free Recall and Recognition of targets in theLevels of Processing Memory Strategy Experiment.Figure 7 Controls' and Alzheimer Patients' Free and Cued 153Recall of targets in the Subject Performed MemoryStrategy Experiment (SPT-EPT).Figure 8 Alzheimer Patients and Controls' Free and Cued^159Recall of targets in the Multisensory MemoryStrategy Experiment.LIST OF APPENDICESAppendix A^Letters of contact and consent for^page 218Alzheimer patients, controls andAlzheimer caregivers.Appendix B NINCDS-ADRDA guidelines for making^page 224a diagnosis of Alzheimer's Disease andFunctional Rating Scale criteriafor measuring stage of FunctionalImpairment.Appendix C^Procedure and questions used to^page 228select volunteers to act asnon-demented control subjects.Appendix D^Materials used in Explicit andImplicit Memory Tests.Appendix E^Test booklet to record subjectresponses.Appendix F^Materials used inMemory Strategy 231page 290page 301Appendix G^Tables reporting descriptive andinferential statistics.^ page 305xiACKNOWLEDGEMENTSThis thesis would not have been possible without thepersonal and financial commitment of several individuals andorganizations. First, without the support of Dr. B. LynnBeattie, Director of the Clinic for Alzheimer Disease and RelatedDisorders-University Hospital-UBC site, this study would neverhave happened. Dr. Beattie provided generous access to theclinic and integral help with patient recruitment. I would alsolike to acknowledge the work of Dr. Peter Graf, my programsupervisor; Despite our very different academic backgroundsPeter persisted in his attempts to navigate me through whatturned out to be a difficult process for both of us. Finally, Iam indebted to Dr. Holly Tuokko for her guidance during thearduous task of data collection and to Dr. Jonathan Berkowitz forstatistical advice.Financial support for this study came from several sources.Two research grants from Sigma Xi and funding from the I.O.D.E.supported project and equipment expenses. The Alzheimer Societyof B.C. provided generous assistance that supported me for twoyears during the collection of data for this study. I am alsoindebted to UCLA and the University of Southern California fortravel support that allowed me to interact with researchersdealing with issues directly related to this study.Last, but of equal significance is the personal support Ireceived from friends that kept me writing during the difficultfinal stages of this work. Special thanks to Drs. Paula Brookand Sandra Clark, Ken Gallie and Eunice Williams.1CHAPTER 1: INTRODUCTION AND OVERVIEWPatients with Alzheimer's disease (AD) provide a uniquemodel for investigating the relationship between human memory andthe brain. This is because the neuropathology caused by thisdisease follows a consistent pattern that begins in thehippocampus before spreading to specific cortical associationsites (Ball, 1977; Damasio, Van Hoesen, & Hyman, 1990; Lewis,Campbell, Terry, & Morrison, 1987). This allows the researcherto investigate an intentional or explicit form of memory that isassociated with hippocampal functioning (Squire, 1992a; Zola-Morgan & Squire, 1992, p. 333) and to chart how this type ofmemory changes as damage accumulates in the hippocampus andsurrounding cortical areas (cf. Squire, 1992a). It also allowsthe investigator to examine a non-intentional or implicit form ofmemory that research suggests may depend on modality-specificcortical association areas (Squire, 1992a; Tulving & Schacter,1990).To date, our interpretation of the way memory is changed byAD has been influenced by our existing knowledge of the functionsthat these anatomical regions are presumed to have on memoryperformance. For example, AD patients' ability to performexplicit memory tests is always impaired in comparison to that ofneurally-intact controls and^ •2to damage to hippocampal regions (Damasio, Van Hoesen, & Hyman,1990; Van Hoesen & Damasio, 1987). In addition, AD patients'ability to perform implicit memory tests (with the exception ofmotor-based tasks) also appears to be impaired (Butters, Heindel,& Salmon, 1990; Salmon & Heindel, 1992). The retained ability toperform implicit motor-based tests has been related to the factthat the cortical areas associated with somatosensory and motorprocessing, along with related subcortical areas, are spared byAD (Damasio et al., 1990; Eslinger & Damasio, 1986). Thesefindings and their interpretations have been integral to thedevelopment of a deficit memory model of AD -- that is, memory .abilities are irrevocably lost as a direct cause of neuronaldeath.A new phase in our thinking about memory processes may beemerging since researchers are now advocating that greaterconsideration be given to the role that influences like attentionand stimulus characteristics have on memory performance (Light &La Voie, in press; Nebes, 1992; Zola-Morgan, 1993). These ideasseem particularly well-suited to study with AD patients who, itis now being recognized, also experience attentional, as well asother types of general cognitive impairments (Filoteo, Delis,Massman, Demadura, Butters, & Salmon, 1992; Graf, Tuokko, &Gallie, 1990; Salmon & Heindel, 1992).3Study One The first study of this investigation explored the memoryperformance of AD patients using an encoding method that ensuredthat these attentionally-impaired patients had attended to, andprocessed critical target stimulus (refer to chapters 3 & 5).Research suggests that AD patients will not spontaneously encodenew information and most previous investigations have not usedstudy methods that compensate for these patients' attentionaldeficits (Rohling, Ellis, & Scogin, 1991; Strauss, Weingartner, &Thompson, 1985).Memory performance was examined on eight different explicitand implicit tests. Based on previous research, the premise wasthat explicit tests engaged primarily hippocampal-relatedactivity and implicit tests required predominantly non-hippocampal or cortical association activity (Squire, 1992b).Because priming or implicit test performance may be dependent onmodality-specific perceptual processes (Squire, 1992a) thematerials used in these tests were chosen for their ability totap different sensory modalities. For example, written word andpicture materials were chosen since they primarily tap a visualtype of processing, whereas spoken word and tactilely presentedobjects required that subjects engage primarily auditory andsomatomotor processing, respectively.4Research from cognitive psychology shows that the same testforms should be used when comparing explicit and implicit testperformance since variations in the way that targets are studiedand retrieved can have very different effects on memoryperformance (Howard, Fry, & Brune, 1991; Roediger, 1990a, 1990b).For this reason the explicit and implicit tests for each type ofstimulus material (i.e., written and spoken word, picture, andobjects) were designed so that the same test forms were used.Specifically, the same encoding conditions and materials wereused for each set of explicit and implicit tests but theydiffered in the retrieval conditions that occurred. For example,the explicit forms of each test employed retrieval instructionsthat encouraged the subject to intentionally recollect previouslypresented stimuli and the implicit forms of each test did not(refer to Chapter 3).The overall hypothesis of this study was that differences inthe memory performance of AD patients and controls would besmaller on implicit than explicit tests. This hypothesis wasbased on the idea that the hippocampus, which is integral toperforming explicit tests, is destroyed early in AD, whereasareas outside the hippocampus, important to performing implicittests are damaged later (Ball, 1987; Price, Davis, Morris, &White, 1991; refer to Chapter 2). This hypothesis was examinedexplicit and implicit tests-for-muditarally,^5visually, and somatomotor processed materials. Follow-upanalyses were used to determine whether the explicit and implicittest performance of AD patients was significantly different fromcontrols.The results from this study and the second study in thisinvestigation were based on the performance of 20 patientsdiagnosed with possible or probable AD using NINCDS-ADRDAcriteria (McKhann et al., 1984) and 40 non-institutionalized,age-matched controls. The 20 AD patients representedapproximately equal numbers of mildly (n = 6), moderately (n =7), and severely (n = 7) functionally impaired (F.I.) patients asindexed using the Functional Rating Scale developed by Tuokko andCrockett (FRS: 1989; see Appendix B for a copy of the FRS).Employing approximately twice the number of AD patients thathave been included in most extant studies on implicit andexplicit memory, as well as obtaining equal representation fromdifferent F.I. levels, decreased the chances of this being abiased patient sample. Secondly, it allowed the investigation ofwhether the implicit test performance of AD patients remainsstable or is variant across F.I. groups (F.I. was presumed toprovide a rough index of the extent of neuropathology as perPrice, Davis, Morris, & White, 1991 and Morris et al., 1991). Inprevious studies we have found that the explicit test performanceof patients declines from mild to severe F.I. groups, as one6would expect in a condition where there is an increasing amountof neuropathology to the hippocampus and surrounding corticalareas (Gallie, Tuokko, & Graf, 1991; Tuokko, Gallie, & Crockett,1990; cf. Squire, 1992a).Using the same patients and controls in study one and instudy two ensured that performance variation due to individualdifferences or different levels of neuropathology was heldconstant across the memory tests and experiments reported here.Study TwoThe second study in this investigation included threedifferent experiments that explored the ability of various typesof encoding and retrieval strategies to elevate the explicitmemory performance of AD patients. These experiments alsoenabled a closer examination of some of the assumptions andresults that were obtained from the first study. The firstexperiment examined whether requiring AD patients to studycritical targets for their meaning significantly elevated theirmemory test performance. This was examined in a levels ofprocessing design modelled after the work of Craik and Lockhart(1972; Lockhart & Craik, 1990). The second and third experimentsexamined whether AD patients recollected more critical targetsthey had encoded using a performance versus a non-performancebased encoding task. The second experiment was based on thesubject-performed, experimenter-performed work of Cohen (1983;7Cohen & Bryant, 1991) and Backman and Nilsson (Backman & Nilsson,1989; Nilsson & Backman, 1991). The third experiment was amultisensory experiment and was not based on previous work.All three experiments allowed me to investigate anadditional question and this was whether the memory problems ofAD patients were irrevocably lost or impaired due to a combinedretrieval and encoding deficit (Gallie, Graf, & Tuokko, 1990).The idea for this work was based on existing knowledge of thebrain areas that are damaged in AD and their presumed functionalroles, as well as theories found in the cognitive literature.Specifically, the hippocampus plays a role in the retrieval ofinformation and other brain areas such as the locus coereleus,basal forebrain, hippocampus, and parietal lobe play a role inthe attentional focusing or ability to encode target stimuli(Squire, 1992a; Mountcastle, 1978; Van Hoesen & Damasio, 1987).Thus, the premise in study two was that when AD patients wereengaged in tasks that assisted them to encode and retrieveinformation this would elevate their memory performance.Summary of Results The results obtained from study one provided evidence that:(1) contrary to previous findings, implicit memory forwritten and spoken word materials is not alwaysimpaired in AD patients.(2) support was found for Larry Squire's (1992a) ideathat priming or implicit memory performance may be AAtflummmera- by slight changes in the form in whichthe stimulus is presented at study and test. Thiswas shown by the fact that AD patients' performance8on the implicit test of object materials was notsignificantly different from that of controls whenpriming was based on a re-presentation of the samematerials over two trials. However, significantdifferences were found when priming was based onthe study-test performance of two different setsof object stimuli.(3) as found in previous studies, AD patients' implicittest performance for picture materials wassignificantly impaired when compared to that ofcontrols. Results suggest that the heavyvisuospatial and abstract reasoning demands of thistask may have contributed to this finding.(4) unlike explicit test performance, the implicittest performance of mildly, moderately, andseverely F.I. patients was similar on severaltests used in this study (note, this performancewas not at floor).The results obtained from study two provided evidence that:(1) the encoding method employed in study onethat required subjects to identify andthen generate the meaning of critical .targets significantly elevated the memoryperformance of AD patients.(2) somatomotor encoding strategies are more effectivethan non-performance strategies in elevating thememory performance of AD patients in certainconditions.(3) contrary to pre-existing ideas, the explicit memoryperformance of AD patients can be significantlyelevated when a combination of encoding andretrieval strategies are used. However, theextent to which these strategies elevateperformance appears to depend on the patients'F.I. level.9Overview of Remaining Chapters This thesis includes six additional chapters.Chapter 2 examines some of the advancements in our knowledgeof how AD changes the brain. Emphasis is placed on the brain-behavior relationship of AD and the specific changes inintentional or explicit memory performance that occur in thiscondition.Chapter 3 reviews the neuropsychological investigations thathave provided evidence of the brain regions that might beresponsible for performing specific types of implicit memorytests. Work that has focused on the cognitive processes that areinvolved in performing these memory tests is considered and theoverall hypothesis of study one is introduced.Chapter 4 introduces the concept of encoding and retrieval(i.e., test types) strategies. The limited studies that haveused memory strategies with AD patients are reviewed, togetherwith the theories addressing how these strategies might work.Study two and its overall hypothesis is then introduced.Chapter 5 describes the method used in this dissertation.The first section discusses the criteria used to select ADpatients and controls and the demographic characteristics ofthese subject groups. The next sections describe how materialswere chosen and the procedures that were employed in the memorytest and strategy experiments.10Chapter 6 reports the results of this investigation. Thefirst section describes the statistical design and generalapproach taken in analyzing results. The second and thirdsections report the results of the memory test and strategyexperiments, respectively.Chapter 7 is a general discussion of the results that wereobtained. In this chapter the main findings from study one andfrom study two are discussed. The limitations as well as the waythis investigation guides my future work is also considered.This chapter ends with a summary of the global and specificcontributions of this work.11CHAPTER 2: INTRODUCTION TO ALZHEIMER'S DISEASEThis chapter examines some of the behavioral andneuroanatomical characteristics of Alzheimer's Disease. Emphasisis placed on brain-behavior relationships and the intentional orexplicit memory changes that occur in this condition.Behavioral and Neuroanatomical CharacteristicsAlois Alzheimer first described the condition we now callAlzheimer's Disease at a German Psychiatric conference in 1906(Alzheimer, 1907; Hippius, 1990). In his presentation Alzheimerreported the clinical symptoms of a 51 year old patient,Frau A.D., who over a four year period showed changes in both thetype and severity of her behavioral symptoms. Frau A.D.'ssymptoms began with problems in remembering recent events, likewhat she had bought at the store that day. With time thesememory problems changed and she developed an inability torecollect information learned many years previous, such as herhouse address. During the time Alzheimer observed this patienthe noticed that she also developed other symptoms. One symptomwas visuospatial confusion when attempting to find the rooms inthe house she had lived in for many years. The other symptom wasthe development of a progressive language impairment shown byFrau A.D.'s increasing inability to find the correct words toexpress herself.12Alzheimer's presentation of Frau A.D. became significantwhen he suggested that the behavioral symptoms observed in thispatient might be related to the results of her autopsy (Hippius,1990). Using a recently developed silver stain, Alzheimer haddetected neuritic plaques and neurofibrillary tangles in thecerebral cortex and hippocampus of this patient (Hafner, 1990).Contemporary research has provided additional details about theseplaques and tangles, and their presumed role in causing thebehavioral symptoms of AD patients.Neuritic Plaques and Neurofibrillary Tangles Neuritic plaques (NPs) are extracellular brain lesions thatare currently regarded as the primary sites of damage caused byAD (Mann, 1989). In contrast, neurofibrillary tangles (NFT) areintracellular lesions that occur within neurons thought tosynapse with cells involved in NP formation (Jellinger, 1990;Mann, 1989). Together the presence of NPs and NFTs are requiredbefore a definite diagnosis of AD can be made (Iqbal, 1991;Jellinger, 1990). Although there is no consensus on what causesNPs and NFTs to form, this has not stopped researchers fromacquiring a basic understanding of their structure (Iqbal, 1991;Jellinger, 1990).Structure. Neuritic or senile plaques (NPs) areapproximately 50 - 200 micrometer in size and most have a central_corexd'_anyloid  protein fLandomr& Kidd, 19239; Mann, 1989). All13plaques are surrounded by neurites that contain paired helicalfilaments similar to those found in NFTs (Mann, 1989). Presentlythere is no consensus regarding whether NPs and NFTs are relatedstructures (Arnold, Hyman, Flory, Damasio, & Van Hoesen, 1991;Mann, 1989).The method by which NPs develop is not known, but one of themore prominent theories suggests that it is the presence ofamyloid protein that triggers NP formation (see Blass, Li-wen Ko,& Wisniewski, 1991; Landon & Kidd, 1989). The mechanism by whichamyloid enters the brain is not known, but its presence isthought to sever axonal and dendritic connections and eventuallycause neuronal death (Mann, 1989). As these neurons die theymove towards the amyloid to form the plaque (Wisniewski, 1988).Neurofibrillary tangles (NFTs) are the other histologicalsign of AD (Flament-Durand & Brion, 1990; Lewis, Campbell, Terry,& Morrison, 1987). These structures are composed of paired 10 nmfilaments that wind about each other with a periodicity of 80 nm(Mann, 1989). As NFTs develop within the neuron they disruptaxonal transport and metabolic functioning (Flament-Durand &Brion, 1990). At some point these tangles impede the neuron'sability to maintain its metabolic functions, and this results incell death (Flament-Durand & Brion, 1991; Mann, 1989).Thus, in combination, the development of NPs and NFTs causebrain neurons to ^TTbIeddho-GA-6-ca, 1988; Wisniewski, 1988).14Since neurons are postmitotic, dying cells are not replaced andthis causes serious disruptions in interneuronal communication(Damasio, Van Hoesen, & Hyman, 1990; Kandel & Schwartz, 1985).Although surviving neurons may compensate for cell death viaaxonal sprouting, the ability to reinstate interneuronalcommunication is limited (Gertz & Cervos-Navarro, 1990).Evidence that axonal sprouting provides limited functionalcompensation is provided in the following discussions of theareas that NPs and NFTs develop and the paralleldisruption/distortion of behavior (Arnold et al., 1991; VanHoesen & Damasio, 1987).NPs and NFTs and the Behavioral Symptoms of AD Patients NP and NFT Development Begins in Hippocampal Regions.Autopsy results show that NPs and NFTs always occur in thehippocampus of demented, but not in that of non-demented elderly(Hyman, Van Hoesen, Kramer, & Damasio, 1986; Mann & Esiri, 1988;Morris et al., 1991; Price et al., 1991). These consistentfindings have led to the accepted notion that NPs and NFTs signalthe presence of AD (Arnold, Hyman, Flory, Damasio, & Van Hoesen,1991; Lewis, Campbell, Terry, & Morrison, 1987).Evidence that AD originates in hippocampal and/or closelylocated brain structures (i.e., the olfactory bulbs and amygdala)has been slower to develop since this information has accruedrice e^1991). Initially, autopsies• brom15performed on mildly cognitively impaired patients (whichclinicians assume are in the early stages of neuropathology) hadshown that NP and NFT development was restricted to hippocampalregions (Ball, 1977; Ball et al., 1985; Price et al., 1991).Autopsy results based on moderately and severely impairedpatients revealed a pattern of pathology that suggested that NPsand NFTs had originated in the hippocampus before spreading toother brain regions (Price et al., 1991).Recent advances in magnetic resonance imaging methods haveallowed investigators to obtain more accurate images of thehippocampal region (deLeon, George, Stylopoulous, Miller, &Smith, 1990a). These advances have provided researchers withdirect evidence that the hippocampus is the first area in whichbrain changes (i.e., atrophy) occur in AD patients (deLeon etal., 1990b).The Hippocampus and Memory. Milner's extensive testing ofpatient H.M. (Milner, 1958), Squire's work with patient R.B.(Squire, 1992a; Zola-Morgan, Squire, & Amarall, 1986) togetherwith research employing animal models (i.e., Mahut, Zola-Morgan,& Moss, 1982; Mishkin, 1978) have consistently shown that thehippocampus is integral to intentional or explicit memoryperformance (see Squire, 1992a for an extensive review on thistopic). This would suggest that if the pathology associated withAD originates in hippocampal regions, then memory problems should16be one of the first symptoms to be exhibited by these patients.This correspondence between brain pathology and behavioralsymptoms is supported by clinical research (Van Hoesen & Damasio,1987; Morris & Rubin, 1991).Specific analyses of the hippocampal regions in which NPsand NFTs form, and the pattern in which they develop appears tobe closely correlated to the intentional memory changes thatoccur in AD patients: for example, the initial fluctuations inthe ability to recollect recently acquired information thateventually develop into a consistent problem when recollectinginformation acquired in the distant past (McKhann et al., 1984;Reisberg, 1983).Impairment in Intentional and Short Term Memory Abilities.Neuropathological investigations show that one of the first areasthat NPs and NFTs develop is in the CA1 subfield of thehippocampus (Morris et al., 1991; Price et al., 1991). Since theinitial brain changes caused by AD are restricted to the CA1region (Morris et al., 1991; Price et al., 1991) this permits acomparison to be made with patient R.B.The most extensively documented case of bilateral damagerestricted to the entire rostral-caudal extent of CA1 was patientR.B. (Zola-Morgan, Squire, & Amaral, 1986). Comprehensiveneuropsychological testing of this patient showed that his memoryproblems  were gemerally-confIned-to-an- inability t^o intentionally17recollect recently acquired information (Zola-Morgan, Squire, &Amaral, 1986). For example, R.B. suffered from a short-termmemory impairment for all types of stimuli whether it wasauditory, visual, pictorial, or somatosensory in nature. Thiswas shown by this patient's impaired performance on tests ofstory and word recall, immediate reproduction of the Rey-Osterreith figure, as well as an inability to remember which ofhis fingers had recently been touched by the examiner (i.e.,behind a partition; Zola-Morgan, Squire, & Amaral, 1986).In contrast, R.B. did not appear to have a profound long-term memory impairment. For example, his performance was similaror better than that of controls on the famous faces and thetelevision test that required that he correctly identifycelebrities and television programs that had been prominentbefore his ischemic attack (Cohen & Squire, 1981; Zola-Morgan,Squire, & Amaral, 1986). R.B. also exhibited a retained abilityto engage in the free recall and the recognition of both publicand personal events that had occurred before his accident(Squire, 1987). However, there was some evidence that thispatient might have had a slight retrograde amnesia for eventsthat had occurred one to two years before his accident (Zola-Morgan, Squire, & Amaral, 1986, p. 2954). This period ofretrograde amnesia corresponds to that found in patientsundergoimselem:ta=lonstuas4mv-therapy and to - idem5 regarding18consolidation and the ongoing role of the hippocampus for a timeafter the initial learning event (Squire, 1987; Squire & Cohen1979; Squire, Slater, & Chace, 1975).Additional testing of R.B. did not detect any significantcognitive impairment other than for memory. For example, thispatients' performance on most scales of the Weschler AdultIntelligence Scale was above normal for both the verbal and theperformance components of this test (Zola-Morgan, Squire, &Amaral, p. 2954). Use of a non-traditional memory test (i.e.,word stem completion) further indicated that R.B. showed anability to non-intentionally recollect recently studied items(Zola-Morgan, Squire, & Amaral, 1986, p. 2964). Thus, theneuropsychological assessment of R.B. indicated that hiscognitive impairment was generally restricted to an inability tointentionally recollect recently experienced information.Memory deficits similar to R.B.'s have now been documentedin patients with damage to CA1 and surrounding hippocampal areas(see Squire, 1992a; Victor & Agamanolis, 1990). The consistencyof these findings support the idea that the memory impairmentsdocumented in R.B. were related to damage to the hippocampus andthat this area performs "some computation upon newly processedinformation" (Squire, 1987, p. 194). Furthermore, since thefirst memory problems that AD patients encounter is an inabilityto intentionally recdII -dbt newly encountered information, it19seems reasonable to assume, based on our present knowledge, thatthese changes are related to NP and NFT development in the CA1-hippocampal area (Damasio et al., 1990).Insidious Onset. Differences between R.B.'s relativelysudden, and AD patients' gradual inability to intentionallyrecollect newly acquired information can be explained in terms ofthe processes that damage CA1 regions. The ischemic episodesexperienced by R.B. appear to have occurred over a relativelyshort time period (i.e., months) in comparison to AD which isestimated to take anywhere between 3 to 15 years to completelydevelop (Huff, Growdon, Corkin, & Rosen, 1987; Van Dijk, Dippel,& Habbema, 1991). Additionally, the ischemic episodesexperienced by R.B. appear to have totally destroyed his CA1regions (Zola-Morgan, Squire, & Amaral, 1986). In contrast, anincremental development of NFTs and NPs occur in the CA1 regionsof AD patients (Price et al., 1991). Since the functionalability of CA1 regions presumably depends on the remaining numberof unaffected neurons (see Damasio, Van Hoesen, & Hyman, 1990;Van Hoesen & Damasio, 1987) an incremental accumulation ofneuropathology should result in an insidious, rather than anabrupt onset of memory problems. This is the pattern of memoryimpairment that is observed in AD patients (McKhann et al.,1984).20Increasing Severity. The severity of R.B.'s intentionalmemory deficits did not appear to drastically change over time,unlike that of AD patients (McKhann et al., 1984; Reisberg, 1983;Van Hoesen & Damasio, 1987). Recent work with monkeys incombination with neuropathological investigations of AD patientsreveals some of the neural substrates that might be responsiblefor the increasing memory difficulties encountered by thesepatients (Squire, 1992a; Price et al., 1991). A brief review ofthe neuroanatomy of the hippocampus and surrounding structures isrequired to understand this work.Anatomical Relationships. The hippocampus is a member ofthe functional unit referred to as the hippocampal formation(Barr & Kiernan, 1988). This formation, depicted in Figure 1,consists of several structures including the parahippocampalgyrus that curves into the subiculum, which in turn merges toform the hippocampal and CA1 regions (Barr & Kiernan, 1988). Thehippocampal formation is connected to the amygdala and itsoverlying perirhinal cortex via the entorhinal cortex (Barr &Kiernan, 1988; Carpenter, 1985). As the entorhinal cortex curvesto form the parahippocampal gyrus, it brings with it major fibrebundles that form the perforant pathway that runs within the1 R.B. may have experienced a more severe memory impairmentduring the first year of his ischemic attack than in subsequent• • • •:•P• owever,this pattern is opposite to that observed in AD patients where thememory impairment increases with time (McKhann et al., 1984).Inferior Born of theLateral VentricleIlk-Caudate Nucleus/ &Rippocampusif \/:,^Perforant Pathway/ ''rhi.. _..41 cortexParahippocampal CyrusFimbria/FornixSubiculum• AmygdalaPerirhinal Cortexapproximate areaLocus ceruleus of cerebellumBERSCBEL'S GYROS(auditory)BASAL FOREBRAIN(nucleus basali■of Meynert)approximate areaof basal •anglia PRIMARY MOTOR AREASOMATOSENSORYPARIETAL LOBEOCCIPITAL CORTEXOptic TractFigure 1. Schematic drawing in coronal perspective of theRippmcampal Formation with medial surface at left.• Location of Amgdala not exact.TEMPORAL LOBEapproximate areaof spinal cordLateral View of the Brain(stiple indicates distribution and location of NeurofibrillaryPlaques and Tangles as per Brun. 1983)2122hippocampus (Carpenter, 1985). Based on investigations withmonkeys, it appears that the perforant pathway is comprised offibres that originate from isocortical association areas andcarry most of the auditory, visual, somatosensory and motor-related information that enters the hippocampus (Arnold et al.,1991; Carpenter, 1985; Hyman, Van Hoesen, Kromer, & Damasio,1986; Insausti, Amaral, & Cowan, 1987; Squire, 1992a). Theperforant pathway, in a series of intrinsic intrahippocampalconnections, ends in the subiculum and CA1 regions (Arnold etal., 1991). From the subiculum and CAI regions, a reciprocalpathway exits the hippocampus carrying information back to thecortex, thalamus, and hypothalamus (Arnold et al., 1991; Braak &Braak, 1990). Information from other brain areas (primarilysubcortical) enter and exit the opposite end of the hippocampalformation through the fimbria/fornix (Barr & Kiernan, 1988).Animal model investigations proffer more direct informationon some of the neuroanatomical structures responsible forvariations in the severity of memory impairments (Squire, 1992a;Squire & Zola-Morgan, 1991). In these investigations monkeys andrats have been subjected to bilateral, circumscribed lesionslimited to particular structures, or to a combination ofneuroanatomical structures located in the medial temporal lobe(Squire & Zola-Morgan, 1991; Zola-Morgan & Squire, 1992). Theeffects^ of-these anatomical- removals on memory performance are23then analyzed by having these lesioned animals perform memorytasks in which amnesic patients exhibit similar memoryperformance (Squire & Zola-Morgan, 1991; Zola-Morgan, 1993).In many of these animal studies the delayed nonmatching-tosample (DNMS) task has been employed (Squire & Zola-Morgan,1991). Both AD patients and lesioned monkeys show similarperformance on DNMS tasks (Albert & Moss, 1984; Albert, Moss, &Milberg, 1989). In combination, the results obtained from theseanimal investigations have shown that it is the extent of damageto cortical areas surrounding the hippocampus and amygdala (i.e.,entorhinal and perirhinal cortex, respectively) that iscorrelated to the level of performance on the DNMS test (Squire &Zola-Morgan, 1991; Zola-Morgan, Squire, Amaral, & Suzuki, 1989).For example, monkeys with bilateral damage to the hippocampus,amygdala, and the cortical areas surrounding these structures(designated as a H+A+ lesion) exhibit a more severe memoryimpairment than animals with bilateral lesions restricted to thehippocampus and its surrounding cortex (i.e., an H+ lesion)(Mishkin, 1978; Zola-Morgan & Squire, 1985, 1986). The maindifference in these two models is damage to the amygdala andsurrounding perirhinal cortex. Since animals with bilaterallesions restricted to the amygdala (i.e., an A lesion) shownormal performance on the DNMS task (Zola-Morgan, Squire, &Amaral, 1989) damage to the perirhinal cortex appears responsible24for the increased memory deficit (Squire, 1992a; Squire & Zola-Morgan, 1991; Zola-Morgan, Squire, Amaral, & Suzuki, 1989).Additional support for the theory that damage to corticalareas surrounding the amygdala and hippocampus influence memoryseverity is provided by other studies. For example, the DNMSperformance of animals with H+A+ lesions has been found to besimilar to H++ animals when damage is restricted to thehippocampus and to cortical areas surrounding the hippocampus andamygdala (Glower, Zola-Morgan, & Squire, 1990). Similarly, theDNMS performance of animals with lesions restricted to theperirhinal-parahippocampal gyrus (i.e., the PRPH lesion) issimilar to that of animals with H+A+ lesions (Zola-Morgan,Squire, Amaral, & Suzuki, 1989).In combination, these findings have led animal investigatorsto conclude that damage to the perirhinal and surroundingcortical areas -- not the amygdala -- are responsible for thegreater memory impairment observed in H+A+ animals (Squire,1992a; Squire & Zola-Morgan, 1991, Zola-Morgan & Squire, 1992).Subsequent re-analysis of the lesioned areas produced in theseinvestigations support this conclusion. Specifically, that "theperirhinal cortex [is] the only area where the H++ and PRPHmonkeys sustained more damage than the H+ and the H+A monkeys"(Squire & Zola-Morgan, 1991, p. 1383). As Squire (1992a) hasastutely-Gernmented- -"herein-lies-25more amnesic than other amnesic study patients, including R.B.H.M. sustained an H+A+ lesion, but R.B. sustained a lesioninvolving only a portion of the hippocampus" (p. 201) that didnot include adjacent perirhinal and entorhinal cortices.Studies based on monkeys indicate that approximately twothirds of all cortical input to the hippocampus travels throughthe entorhinal cortex (Insausti, Amaral, & Cowan, 1987; Squire1992a; Zola-Morgan, 1993). Thus, depending on the extent ofdamage to the entorhinal cortex and areas that project to it(i.e., the perirhinal cortex) this would result in a relateddecline in the amount of information entering the hippocampus.This would be exhibited as a memory impairment that waspresumably proportional to the extent of damage that had occurred(Hyman, Van Hoesen, Kromer, & Damasio, 1986; Squire, 1992a).Price et al. (1991) have recently uncovered a pattern ofneuropathology in AD patients that corresponds to the anatomicaldamage induced to the perirhinal and entorhinal areas in animalinvestigations. In a rare investigation that compared thedistribution of NFTs and NPs in AD patients at different levelsof memory severity, Price et al. (1991) found evidence that NFTsand NPs spread from the CA1 region to the perirhinal andentorhinal areas surrounding the hippocampus.The pattern of neuropathology found by Price et al.,suggests that the number of NPs and NFTs in perirhinal and26entorhinal areas corresponds to the degree of memory impairmentexhibited by AD patients. To illustrate, Price et al. found anaverage of 10.5 NPs and NFTs per mm2 in the perirhinal andentorhinal cortex of mildly memory impaired patients (n = 6; fage = 86 years). In comparison, an average of 16.55 NPs and NFTsper mm2 were found in the perirhinal and entorhinal cortex ofmoderately and severely memory impaired patients (n = 6; } age =76.17 years). In contrast, very few NFTs and no NPs were foundin the same anatomical areas of young (n = 5; f age = 59.2 years)and older (n = 8; f age = 79.13 years) non-demented elders (Priceet al., 1991). Together Price's findings support the idea thatthere is a correlation between the number of NPs and NFTs thatdevelop in the perirhinal and entorhinal cortices and theseverity of the memory impairment experienced by AD patients.Price et al. also found around .15 NFTs per mm2 in the youngnon-demented subjects compared to 4.2 NFTs per mm2 in the oldernon-demented subjects. These results are similar to those foundby other investigators who have shown that whileneuropathological changes are not found in the hippocampus andparahippocampal regions of thirty year olds, they can be found inbetween 5 - 15% of the brains sampled from the fifth, and inaround 50% of the brains sampled from the seventh decade of life(Squire, 1987; Tomlinson, Blessed, & Roth, 1968; Tomlinson,1972), Together- tbese- findings suggest -that the subtle, but27increasing memory impairment experienced by some "healthy" peoplewith advancing age might be attributed, at least in part, to anincrease in the number of NFTs that accumulate in these brainareas (Davis & Bernstein, 1992; Price et al., 1991; Squire,1987).Long-term Memory Impairment. Other AD researchers havechosen to focus their attention on the neuropathologicaldevelopment that occurs within the neural pathways of thehippocampal formation (see Damasio et al., 1990). What this workhas shown is that NFTs selectively accumulate in the cells oforigin of the perforant pathway (i.e., the entorhinal cortex)while NPs accumulate in this pathway's area of termination (i.e.,the hippocampal formation; Hyman, Van Hoesen, Kromer, & Damasio,1986). These findings have led Damasio and his associates (1990;Van Hoesen et al., 1986) to suggest that while damage isoccurring to the cortical areas surrounding the hippocampus(i.e., the entorhinal and perirhinal cortices) there is aconcurrent process of neuropathological destruction that isoccurring within the hippocampus. The destruction of the majorneural pathways within the hippocampus serves to completelydisconnect it from other brain regions. That is, information isessentially prevented from entering and leaving the hippocampusthrough the entorhinal and fornix routes.28As Damasio et al. (1990) have noted, the memory deficitsthat are observed in AD patients are more extensive than thatseen in most amnesic conditions (p. 93). Specifically, thedevelopment of a long-term memory impairment that is (at somepoint in the development of AD) transposed onto a pre-existingproblem when intentionally recollecting recently acquiredinformation. The exact reasons for the gradual development ofthis long-term memory impairment is not known, but it seemslikely that this is due to the far reaching areas that aredirectly, and indirectly influenced by the disconnection of thehippocampus from cortical areas (Damasio et al., 1990).Neuropathological Developments Outside the Hippocampus.Proponents of an anatomical pathogenesis of AD suggest that it isduring the process of hippocampal disconnection that NPs and NFTsspread from the hippocampus to other brain regions via cortico-cortical pathways2 (Damasio, Van Hoesen, & Hyman, 1990; Esiri,1989; Jellinger, 1990; Lewis, Campbell, Terry, & Morrison, 1987;Rogers & Morrison, 1985). Cortico-cortical pathways originatefrom the cell bodies of pyramidal neurons and serve to indirectlyconnect the hippocampus to many cortical and subcortical areas(Van Hoesen & Damasio, 1987). The cortico-cortical progression2 This is one of several theories regarding the development ofAD (see Hardy & Davies, 1988; Toledano-Gasca, 1988). To date theanatomical pathogenesis model provides the most consistentexplanation for the pattern of neuropathology observed in ADpatients (Mann, 1991).29theory is supported by the fact that the brain areas that incurthe largest amount of NFT development are those that have theheaviest connections to these pathways (Lewis et al., 1987; VanHoesen & Damasio, 1987). For example, in later stages of AD theareas with the largest numbers of NPs and NFTs are the temporaland parietal lobes, basal forebrain and hippocampal regions(Brun, 1983; Lewis, et al., 1987; see Figure 1).There is also a discrete pattern of neuropathology foundwithin cortical areas (Lewis et al., 1987; Van Hoesen & Damasio,1987). That is, within a cortical area, it is the region thatreceives the most input from cortico-cortical pathways that havethe largest numbers of NFTs. To illustrate, Lewis et al. (1987)compared the amount of neuropathology that occurred in theprimary visual cortex (i.e., Brodmann's area 17), to the adjacentsecondary visual association area (i.e., Brodmann's area 18) andto the tertiary or higher-order visual association areas (i.e.,Brodmann's area 20; see Figure 1). These areas receive anincremental amount of cortico-cortical input as one goes fromBrodmann's area 17 to 20, which is presumably related to theincreasing complexity in the information these areas process(Kandel, 1985; Lewis et al., 1987). Lewis et al. found that whencomparing the brains of 8 AD patients (stage of disease notspecified) there was a corresponding incremental relationship inthe number of NFTs that were found. For example, per 25030micrometer area there were, on average, around .9 NFTs found inthe primary, 19.7 NFTs in the secondary, and 35.5 NFTs in thetertiary visual areas. Similar findings have been reported forother areas including the auditory cortex (Esiri, Pearson, &Powell, 1986; Pearson, Esiri, Hiorns, Wilcock, & Powell, 1985).Reaffirming this relationship between cortico-corticalconnections and AD is the fact that the brain areas that are notas heavily connected to these pathways incur less neuropathology(Van Hoesen & Damasio, 1987). For example, the relative sparingof Brodmann's area 3, 1, and 2 for the somatic modality andBrodmann's area 4 and 6 of the motor cortices illustrates this(Brun & Englund, 1981; Brun, 1983; Lewis et al., 1987; Van Hoesen& Damasio, 1987).To synopsize, the neuropathology associated with AD appearsto follow a consistent pattern that begins in the hippocampusbefore spreading along cortico-cortical pathways to other brainregions. This pattern of pathology serves to eventuallydisconnect both the hippocampus and the tertiary corticalassociation areas responsible for visual and auditory processing,from other brain areas (Damasio et al., 1990; Esiri et al., 1986;Lewis et al., 1987). In addition, other brain centers are alsoinvolved. For example, cortical areas including the temporal,parietal, and the basal aspects of the frontal lobe (i.e.,nucleus basalis_of_14eynert4-as-well as^areas31including the locus coeruleus, basal ganglia, and the thalamusare also invaded by NFTs and NPs (Brun & Englund, 1981; Davies &Maloney, 1976; Damasio et al., 1990; Forno, 1978; Rudelli,Ambler, & Wisniewski, 1984).The brain areas outside the hippocampus that are affected byAD are thought to be responsible for the fact that the memorydeficits of these patients are more extensive than those observedin most amnesic patients (Damasio et al., 1990). Some of theseregions are believed to work together in the coordinated activityof general purpose cognitive functions that have not beenconsidered to directly influence the memory impairment of ADpatients. For example, the locus coeruleus, nucleus basalis ofMeynert, hippocampus, frontal, and parietal lobes have beenascribed various roles in attention (Filoteo et al., 1992;Mountcastle, 1978). As subsequent sections will examine,attention and memory are inseparately linked in that, amoungother things, if the organism has not attended to (or focusedattention on) critical targets, later recollection of the targetwill not occur (Graf, Tuokko, & Gallie, 1990). Thus, there is apossibility that some of the memory impairments observed in ADpatients may be influenced by these patients' attentionaldeficits.In contrast, other areas of the brain appear to berelatively- -spared-by- PM-and-this-is presumably because they32receive less input from cortico-cortical pathways (Van Hoesen &Damasio, 1987). These areas include those responsible forsomatomotor type processing and this has been correlated to ADpatients' ability to learn some motor-based tasks (Damasio etal., 1990; Eslinger & Damasio, 1986).What implications does this pattern of neuropathology havefor the memory researcher? While many researchers have used thefunctional or cognitive impairment level of AD patients as arough index of the stage of neuropathological progression of thisdisease, in reality this provides only a gross measure. There iscurrently no practical method for determining (short of autopsy)the precise locations and amount of neuropathology present in ADpatients while they are alive. Thus, unlike memory researchersusing animal models, those investigating AD patients are leftwith the dilemma of what brain areas are correlated to the memoryimpairment. Comparisons of the memory performance of AD patientswith patients with localized lesions has provided a wealth ofinformation with respect to the intentional memory impairmentsthat occur as a result of damage to hippocampal regions.However, AD patients are unique in that they provide theresearcher with a model for investigating the implications thatother brain regions also have on memory performance. For thepattern of neuropathology that occurs in AD however, one wouldneed a -"-tout"- that could separate hippocampal from non-hippocampal functions as well as the influences that attentionand modality specific processing might have on memoryperformance. The next chapter introduces such a tool which iscontained under the framework known as explicit and implicitmemory.3334CHAPTER 3: EXPLICIT AND IMPLICIT MEMORYThe focus of this chapter is on the brain regions and thecognitive processes that might be required to perform implicit(non-intentional) memory tests. Neuropsychologicalinvestigations that use a "multiple systems" model and cognitivestudies using a "process" interpretation of the dissociationsthat occur in explicit and implicit test performance arereviewed. In the final section I show how ideas generated byresearch guided by these two models were used in the first studyof this investigation.IntroductionMany ways have been used to categorize and investigatememory (Lovelace, 1990; Richardson-Klavelin & Bjork, 1988). Forexample, memory has been described in terms of the duration oftime that has elapsed between the "learned" event and itssubsequent recollection (i.e., short term vs. long term memory).In recent years an additional way of classifying memory hasbecome prominent because of its usefulness in investigating boththe neural and the cognitive processes of memory; this is thedistinction made between explicit and implicit memory (Masson &Graf, in press).The terms explicit and implicit were first introduced byWilliam McDougall (1924) and describe a memory dissociation thathad been known to clinicians for many years (i.e., Freud, 1890;35Korsakoff, 1889). For instance, the neurologist Claparede (1911,1951) observed that an amnesic patient would not shake his handafter he had pricked her with a pin, even though she did not knowwhy she should refuse. These patients appeared to retain animplicit recollection of an event they could not explicitlyrecount.The terms explicit and implicit were later re-introduced byGraf and Schacter (1985) as "descriptive concepts that focus onthe person's psychological experience at the time of retrieval"(Schacter, 1987, p. 501). Graf & Schacter (1985) definedimplicit memory as being revealed when previous experiencesfacilitated performance on tasks that did not require a consciousor intentional recollection of those experiences. In contrast,explicit memory was revealed when performance on a task requiredthe intentional recollection of a previous event. By definitionimplicit memory was indexed by implicit memory tests and explicitmemory by explicit tests (Graf & Schacter, 1985). Priming, orthe facilitation in performance as a result of recentlyencountered information (Shimamura, 1986, in press) was used asan index of implicit memory .The terms explicit and implicit (as introduced by Graf &Schacter) were intended to refer only to the cognitive processesengaged by the subject during the retrieval of information.However,iTivestigators interested in the biological substrates of36memory have found the explicit and implicit distinction to beuseful in their work 3 (Squire, 1992a). As a result, two mainways for interpreting dissociations between explicit and implicittest performance have evolved (Schacter, 1987; Squire, 1987).One of these methods has been referred to as the "multiplesystems" model (Mishkin, Malamut, & Bachevalier, 1984; Squire,1987; Shimamura, in press) and the other as the "process" view(Roediger, 1990a, 1990b).Multiple Systems Researchers interested in the biological substrates ofmemory have mainly adopted the "multiple systems" model (e.g.,Cohen & Squire, 1980; Schacter, 1989; Squire, 1992a; Tulving,1986). The main assumption of this model is that different brainstructures are required to perform explicit and implicit tests(Shimamura, in press). This idea receives support from amnesicpatients like R.B. who can perform implicit but not explicitmemory tests (Zola-Morgan, Squire, & Amaral, 1986). In this caseexplicit tests are viewed as tapping the integrity of areas suchas the hippocampus and diencephalic regions that are typicallydestroyed in amnesia (Squire, 1992a). In contrast, implicittests are viewed as tapping non-hippocampal regions such as the3 As Squire, Knowlton and Musen (1993) have discussed, Grafand Schacter's (1985) definition of explicit and implicit dependon the - arse- of language to -influence the subject's -fiFtent atrecollection. This is not useful in animal investigations wherethe focus must be on the tasks that are used.37cortical association areas (Squire, 1987; Squire, Knowlton, &Musen, 1993).Performance dissociations also occur on implicit tests (cf.Butters, Heindel, & Salmon 1990) and Squire (1992a) has suggestedthat this is because priming occurs in many different brainareas. Assumptions are made that the ability to finddissociations in implicit test performance is related to thegeographical closeness (i.e., functional separateness) of theneural systems required to perform these tests (Shimamura, inpress; Squire, 1987).Process Model In contrast, the process model is used primarily bycognitive psychologists who employ neurally-intact subjects fortheir investigations (Craik, 1983; Graf & Mandler, 1984;Roediger, Weldon, & Challis, 1989). The main assumption of theprocess model is that different cognitive processes, not neuralregions, are required to perform explicit and implicit tests.Dissociations in test performance are therefore viewed as tappingthe different cognitive processes required to perform these tests(Roediger, 1990b; Roediger, Weldon, & Challis, 1989). Unlikesystems researchers whose focus is on isolating anatomicalregions, the goal of the cognitive psychologist is to identifythe cognitive activities that influence test performancetHoediger_, _I-990br-Shimamura- in prebb). They do tnis by38manipulating the encoding, retrieval, and test materials (i.e.,written words, pictures) that are used and measure the effectsthis has on performance (Roediger, 1990b).Transfer Appropriate Processing Framework. Many cognitivepsychologists use the theoretical framework called TransferAppropriate Processing to guide their experimentalinterpretations (TAP; Graf & Gallie, 1992; Roediger, 1990b). Theconcept of TAP was first introduced by Morris, Bransford, andFranks (1977) who postulated that performance on various taskswas maximized when the cognitive operations engaged at studyrecapitulated those that were engaged at test. To illustrate,the most effective way to learn how to paraglide is to paraglide,not to read a book on the subject. Placed in other terms,performance on an implicit memory task will be maximized when thecognitive processes engaged at study are those that are engagedat test.As Graf and Ryan (1990) have argued, in this general formTAP is unable to explain differences between explicit andimplicit test performance because it does not state whichcognitive processes are involved. In recent years severalproposals regarding the nature of these processes have been madeincluding the widely held view of Roediger and his colleagues(Roediger, 1990b; Roediger & Blaxton, 1987). Roediger views-cognitive- processing requirements as varying along a continuum39where at one end, to perform a memory test requires that thesubject focus on the physical features of the test stimulus(i.e., engage in data driven processes) and at the other end, thefocus is on the conceptual features of the stimulus (i.e., engagein conceptually driven processes). However, as Graf and hiscolleagues have shown, when the only difference between implicitand explicit tests are the test instructions provided to subjects(i.e., the test forms do not differ) performance dissociationsare found that are not predicted by Roediger's classification(Graf & Mandler, 1984; Graf & Schacter, 1987). Thus, it appearsthat Roediger's data driven - conceptually driven distinction maynot capture all the critical features that distinguish thecognitive processes required for performing implicit and explicittests.Alternatively, Graf and his colleagues (Graf & Gallie, 1992;Graf & Mandler, 1984; Mandler, 1980) have focused theirinterpretations in terms of the requirement for integrative andelaborative-type processing. The advantages of this proposal isthat it is able to address cognitive, neural, and test-relatedconsiderations.Integrative-type Processina. By definition, integrationresults from processing that bonds the features of a target intoa pre-existing whole or unitized representation (Graf & Gallie,19921. For example_, main v • -.^ are40required to complete the following; PSYCH OGY.Elaborative-type Processing. In contrast, elaborative-typeprocessing results in the formation of a new relation between atarget and its pre-existing mental contents (Graf & Gallie,1992). This can occur when a target is encoded in relation torelevant prior knowledge. For example, when you learn that a newdopaminergic (DA) receptor has just been discovered (i.e., DA100) and you place this information into your pre-existingknowledge of DA receptors.Explicit and Implicit Tests. The assumption that is made isthat memory tests vary in the degree to which they engageelaborative or integrative processes (Graf & Gallie, 1992).Explicit tests such as free and cued recall are viewed asrequiring mainly elaborative-type processing since they requirethe retrieval of targets that have been associated, or tagged, tospecific prior knowledge (Roediger, Weldon, & Challis, 1989). Incontrast, implicit tests such as word stem completion are viewedas requiring more integrative-type processing (Graf & Gallie,1992). However, within each test type (i.e., explicit, implicit)there will be variations in the amount of integrative orelaborative processing that is required (Graf & Gallie, 1992).For example, free recall tests are viewed as requiring moreelaborative type processing than cued recall tests (Schacter,1987).41Which Model? There has been an ongoing debate regarding the "best" modelto use in the interpretation of explicit and implicit memory testperformance (see Roediger, 1990a, 1990b; Schacter, 1990b). Themajor issue with the process model is seen as its reliance on theTAP framework (Schacter, 1990a). TAP does not enable theinvestigator to test experimental hypotheses since all findingscan be explained by this theoretical framework (Graf & Gallie,1992; Schacter, 1992). In contrast, cognitive psychologists donot see the point of postulating different neural systems toaccount for findings that are accountable by one processingsystem (Roediger, 1990b; Roediger, Weldon, & Challis, 1989).In reality, each model is equally able to explain the mainfindings in the literature (Tulving & Schacter, 1990; Schacter,1990b). For example, the fact that amnesic patients' performanceon explicit tests is much lower than that of controls can beexplained with a systems or process interpretation. A systemsbased explanation would be that the brain regions required toperform explicit tests (i.e., hippocampus and/or diencephalon)are intact in the control but not the amnesic patient. Incontrast, a process account would postulate that the control, butnot the amnesic patient, was able to engage in the elaborative-type processing required to perform this test. In other terms,the explicit test was too difficult for the amnesic  patient to^42perform.Similarly, both a systems and a process interpretation canbe used to guide interpretations regarding implicit testdissociations. A systems interpretation would be that differentneural systems are required to perform different implicit testsand dissociations simply identify the integrity of these systems.For example, Huntington Disease (HD) patients can performimplicit tests of visually-presented words but not motor-basedtasks (Butters, Heindel, & Salmon, 1990). A systems explanationfor this finding would be that the neural areas required toprocess visually-presented words are intact in HD patients. Incontrast, those areas required to perform the motor-based taskare somehow impaired. This latter interpretation would appear tocorrelate closely with the loss of neurons that occur in thebasal ganglia of HD patients (although some cortical atrophy mayalso occur; Kandel & Schwartz, 1985). In contrast, a processinterpretation would explain these performance differences interms of the encoding, retrieval, and/or stimulus characteristicsof these tests. Impaired test performance with visuallypresented word stimuli would be viewed in terms of how this testdiffered from the motor-based task. For example, is theresomething inherently different in the way a visually versusmotor-based task is encoded? retrieved? etc.43As Tulving and Schacter (Schacter, 1992; Tulving & Schacter,1990) have maintained, there is no incompatibility betweensystems or process approaches, they just serve to ask differentquestions about memory. Squire has made similar comments byadvocating that the focus should not be a philosophical debate ofhow to best classify memory, but rather, on finding the best wayto approach empirical questions regarding brain function (Squire,1987, p. 160).As the next sections show, research using systems andprocess interpretations can provide information on how implicitmemory is affected in AD patients. Neuropsychologicalinvestigations reveal the implicit tests (and therefore themethods) in which AD patients perform poorly in comparison tocontrols, and some of the neuroanatomical areas that might beresponsible. Cognitive investigations (with older adults)suggest that impaired performance on some implicit tests may bedue to the cognitive processes that are involved. The focus ofthese sections is on test results since the theory thataccompanies these findings is not well established. This stateof affairs is reflected in a recent quote by Roediger on implicitmemory (1990b, p. 1054)The new mass of knowledge is still formless, incomplete,lacking the essential threads of connection, displayingmisleading signals at every turn, riddled with blind alleys.There are fascinating ideas all over the place, irresistibleexperiments beyond numbering, all sorts of new ways into theme44The following sections attempt to bring some cohesion to twodifferent bodies of research that (1) are both in the earlystages of development, and (2) use memory tests to ask differentquestions about memory.Neuropsychological Investigations Dissociations in Explicit and Implicit Test Performance.Warrington and Weiskrantz (1968, 1970) were one of the firstgroups to discover that different brain regions might be requiredto perform explicit and implicit memory tests. In one of aseries of experiments, four amnesic patients (three patients withKorsakoff's syndrome causing damage to diencephalic regions andone patient with a temporal lobectomy) were provided with listsof words to read. At test these patients were provided withwords in which pieces of the printed letters were missing. Whenasked to use these fragments as clues to remembering previouslystudied targets (i.e., engage in intentional recollection)amnesic patients were found to remember significantly fewer itemsin comparison to control subjects (Warrington & Weiskrantz, 1968,1970). In contrast, when the same patients were allowed to treatthe task as a guessing game (i.e., engage in non-intentionalrecollection) they showed performance that was similar tocontrols (Warrington & Weiskrantz, 1968, 1970).Subsequent research has shown that patients with amnesiacaused by etiologies other  than Korsakoff's and temporal45lobectomies (i.e., chronic electroconvulsive therapy,encephalitis, ischemic attacks) show a consistent ability toperform implicit, but not explicit tests (Graf, Squire, &Mandler, 1984; Squire, Shimamura, & Graf, 1985; Warrington &Weiskrantz, 1978). Research has also shown that it doesn'tmatter what modality information is presented in (i.e., auditory,visual, somatomotor) amnesic patients can show normal performanceon the implicit, but not the explicit version of the same test(Graf, Shimamura, & Squire, 1985; Moscovitch, 1982; Johnson, Kim,& Risse, 1985). It is from these consistent findings thatresearchers adopting the systems model have concluded that themedial temporal lobe and/or diencephalic areas -- areas typicallydamaged in amnesic patients -- are required to perform explicit,but not implicit memory tests (Keane, Gabrieli, Fennema, Growdon,& Corkin, 1991; Salmon & Heindel, 1992; Shimamura, in press).In contrast, a systems interpretation would suggest thatareas outside the temporal and thalamic regions are involved inimplicit test performance' (Butters, Heindel, & Salmon, 1990;Schacter, 1992). This idea has been supported by the results ofinvestigations with patients that have brain damage to areasoutside, or in addition to hippocampal and diencephalic regions(Butters, Heindel, & Salmon, 1990). Much of this information hasThese_ideas have recently been supported by Positron PanissinnTomography (e.g., Squire, Ojeman, Miezin, Petersen, Videen, &Raichle, 1992b).46been developed from the work of a collaborative team ofresearchers who have compared the test performance of patients inthe early stages of Huntington's disease (HD) and AD (seeButters, Heindel, & Salmon, 1990). In these studies HD patientsare used as brain models where damage occurs primarily to thebasal ganglia 5 (i.e., subcortical damage) and AD patients wheredamage is to the cortical association areas (i.e., corticaldamage). Dissociations in the implicit test performance of thesetwo patient groups are used to deduce the neural substrates thatmight be required to perform these tests.Dissoc i ations in Implicit Test Performance. One of thefirst groups to compare the implicit test performance of AD andHD patients was Shimamura, Salmon, Squire, & Butters (1987). Inthis study Shimamura et al. gave eight patients in mild tomoderate AD, eight HD, and seven Korsakoff (KS) patients andtheir age-matched controls a word stem completion test. The samestimuli and procedures used in a previous study where amnesicpatients' performance was found to be similar to that of controlswas used (Graf, Squire, & Handler, 1984).At study Shimamura et al. asked subjects to read 10 words(e.g., motel, abstain) and rate how much they liked each word5 In reality, the neuronal changes that occur in the brainsof Huntington and Alzheimer patients are not restricted to theseareas. Cortical atrophy can occur  in Huntington  patients and .subcortical damage occurs in Alzheimer patients (see Brun, 1983;Kandel & Schwartz, 1985; Price et al., 1991).47using a five point scale (i.e., 1 = dislike extremely to 5 = likeextremely). Subjects were then shown 20, three-letter word stems(e.g., MOT^, ABS^) and requested that they complete eachstem with the first word that came to mind. For each stem therewere at least 10 endings that would complete the word (i.e.,mother, motive, motel, motor, etc.). Ten of the stems could becompleted with words presented at study. The remaining ten stemswere new and were used to provide an index of what subjectperformance might be like without previous exposure to words(i.e., baseline performance). Subjects were also given the ReyAuditory Verbal Learning Test to obtain an index of explicitmemory performance (Rey, 1964).Shimamura et al. found that the AD, HD, and KS patients allshowed an impaired level of performance on the Rey AuditoryVerbal Learning Test. On the word stem completion test the HDand KS patients completed between 30 to 40% more of the stemsfrom the studied (i.e., target) versus the non-studied condition(i.e., baseline) and this performance was similar to that of age-matched controls. In contrast, AD patients completedapproximately 10% more of the studied than the non-studied itemsand this performance was significantly lower than this group'sage-matched controls. However, baseline guessing rates werefound to be similar for the AD (i.e., baseline = 6%) and the HDand KS patient groups (i.e.,  baseline = 5-11%1.48Shimamura et al. noted that the AD patients had been able tocomplete 98 percent of the stems with words, just not with thewords that had been previously studied. This finding suggestedthat the AD patients' poor test performance was not due to aninability to perform the basic task of completing word stems.Other investigators have reported similar findings usingdifferent AD patients and the word stem completion test (Bondi &Kaszniak, 1991; Heindel, Salmon, Shults, Walicke, & Butters,1989; Randolf, 1991; Salmon, Shimamura, Butters, & Smith, 1988).When considering the brain regions that might be requiredfor performing the word stem completion task, Shimamura et al.reasoned that since the AD, but not the KS patients had showedimpaired levels of priming, this likely reflected damage toregions other than the medial temporal lobe or diencephalicstructures (p. 350). In turn, since HD patients had showedpriming levels that were similar to their controls it appearedthat the integrity of the basal ganglia was not critical fornormal performance on the word stem completion test. This meantthat the impaired performance of AD patients on the word stemtest was likely a result of damage to cortical areas (p. 350).When considering what AD patients' impaired word stemcompletion performance meant in terms of memory stores, Shimamuraet al. employed ideas found in the spreading activation theory ofsemantic processing (cf. Collins_kLoftus, 1975; Nebes, 1992). ^49As conceptualized by Tulving (1986) semantic memory is the bodyof knowledge that people possess about words, concepts, theirmeanings, associations, and the rules that govern themanipulation of these words and concepts. Information insemantic memory is viewed as being arranged as an organizednetwork of concept nodes that are interconnected to each other onthe basis of their semantic relationships. The ability to locateitems in this network has been viewed as being impeded by either(1) a breakdown in the structure of semantic knowledge (i.e., aloss rather than an inability to access information; cf. Salmon &Heindel, 1992), (2) an inability to access items as a result ofmemory-related processes (cf. Nebes, 1992), or more recently, (3)an inability to access items due to non-memory processes (i.e.,attentional and/or visuospatial impairments; cf. Nebes, 1992).A main assumption of investigators is that the presentationof either a word or of a picture target automatically activatesits mental representation in the semantic network (Martin, 1992).Another assumption is that this activation automatically spreadsto related concept nodes, and by doing so serves to increasetarget retrievability (Martin, 1992). Presumably this activationestablishes a new and highly specific mental representation ofthe target. This new representation could account for thespecificity and temporal persistence of priming events (Schacter,1990b). The spreading activation to related concept nodes would50presumably correspond to priming that can occur for items thatare semantically related to the original target (i.e., target =orange and the subject responds with lemon).Shimamura reasoned that the AD patients poor performance onthe word stem completion test was more likely due to a memoryrather than to a non-memory impairment (i.e., a global intellector cognitive deficit). This was because AD patients had beenable to perform the basic task of completing word stems, hadshowed baseline performances that had been similar to the othersubject groups, and that their level of dementia had not appearedto be related to their priming performance. Shimamura et al.felt that the AD patients' memory impairment was due to aninability to activate nodal representations and that this mightbe due to a breakdown in the structure of lexical knowledgecaused by damage to cortical brain areas.Salmon, Shimamura, Butters, and Smith (1988) continued theirstudy of implicit memory in AD patients by giving many of thesame patients from the Shimamura study a word pair task. Thistest was thought to provide an index of memory for semanticrelationships since it required that subjects focus on therelationship between word pairs.Nine AD and 10 HD patients, along with similar numbers ofage-matched controls, were asked to rate a set of 12 word pairs(i.e., bird-robin, needle-thread) twice on the basis of how51related each word was to each other (i.e., 1 = not related to 5 =very closely related). Testing was started immediately uponcompletion of the study phase. At test subjects were told thatsingle words would be presented visually, and that they were torespond verbally with the first word that came to mind inresponse to each stimulus word (i.e., bird , needle^ ).As in the Shimamura et al. study, subjects were shown wordsbelonging to word pairs they had studied (the target condition)along with words they had not studied (the baseline condition).Results showed that only the AD patients showed impairedpriming on this free association task when compared to theircontrol group. Specifically, Salmon et al. found that ADpatients completed only about 2% more of the word associationsfrom the target than from the baseline condition (baseline =18%). In contrast, AD patients' age-matched controls completed30% more of the target than the baseline word associations(baseline = 18%). These findings were much different than forthose found for the HD patients. HD patients correctly completedaround 25% more of the target than the baseline associates(baseline = 17%). HD patients' performance was similar to theirage-matched controls who showed a 20% level of priming abovebaseline (i.e., baseline = 21%). Since the baseline performanceof HD, AD, and their age-matched controls was not significantlydifferent, the magnitude of these priming scores was based on52similar performance levels. This provided stronger evidence thatpriming was impaired in AD patients.Salmon et al. concluded that the AD patients' impairment onthe word association test must reflect damage to brain regionsthat were not disrupted in HD patients who were able to performthis test (p. 492). Furthermore, in view of the languagedisruptions and the neuropathological changes that occur in thetemporo-parietal cortices of AD patients this was the area thatwas felt to be responsible for their impaired test performance(p. 492).Salmon et al. also concluded that the cue "bird" had beenunable to evoke the paired associate "robin" because theassociation between the two words has been greatly "weakened" (p.490). Salmon et al. based this idea on the fact that AD patientshad been able to provide appropriate responses for non-studiedassociates and this suggested that the semantic memory networkhad not been totally destroyed. Salmon et al. went no further intheir interpretations of why this system was weakened but theunderlying assumption of neuropsychologists would be thatneuronal loss caused by AD would somehow be responsible (i.e., adisruption in the memory network rather than an inability toaccess it; see also Salmon & Heindel, 1992).Heindel, Salmon, and Butters (1990) continued this series ofinvestigations by administering a picture fragment completion53test to 12 AD and 12 HD patients and their age-matched controls.Their hypothesis was that if lexical, semantic, and pictorialpriming tasks share a common neurological substrate, then ADpatients, but not HD patients, would show impaired performance onthe picture test (p. 284).At study subjects were asked to name items depicted in a setof 15 simple line drawings of common objects. At test theexperimenter placed a binder in front of each subject with therequest that they state the first thing they thought of when theysaw the fragmented stimuli.Heindel et al. results showed that HD patients had similarlevels of priming when compared to their controls. In contrast,AD patients had significantly lower levels of priming which is afinding that was later corroborated by Bondi and Kaszniak (1991).Since AD patients had correctly identified 99.3% of the studypictures Heindel et al. reached a similar conclusion to thatoffered by Shimamura et al., and Salmon et al. Specifically,that AD patients' performance on the picture fragment completiontask was because they were impaired in their ability to activatepre-existing representations of pictures. Additional evidencefor dissociations in implicit test performance was provided byHeindel, Salmon, Shults, Walicke, and Butters (1989). In thisstudy they gave a rotary pursuit and a word stem completion testto 16 AD, 13 HD, 8 demented and 9 non-demented Parkinson patients54(PD) and a group of 10 middle-aged and 12 elderly controls (age-matched to patient groups). The rotary pursuit task consisted ofhaving subjects maintain a stylus on a rotating target for 6blocks of 4 trials lasting 20 seconds. To standardize thedifferent performance abilities of each subject group therotation speed of the target was varied. This speed wasdetermined from practice trials and set at a level where thatgroup, on average, had maintained the stylus on target for atleast 25% of the trial duration. The word stem completion testwas the same as that used in the Shimamura et al. investigation.Heindel et al. found that whereas AD patients showedimpaired performance on the word stem completion task, this wasnot the case for the pursuit rotor task. The average time thatAD patients maintained the stylus on the rotating target steadilyincreased from the first to the sixth block of trials. Theseresults implied that AD patients had "learned" or retained somebenefits from previous exposure to the task.In contrast, Heindel et al. found opposite results for theHD group. Whereas HD patients showed similar performance totheir age-matched controls on the word stem test, they showedlittle improvement in their ability to maintain the stylus on therotating target over the six blocks of trials. In combinationwith the findings from the AD patients, these results were usedto suggest that motor skill learning might be mediated by the55corticostriatal system (which is damaged in HD patients) whereaslexical priming may depend on the integrity of neocorticalassociation areas (which are destroyed in AD patients). Theadditional finding that the demented PD patients showed impairedperformance on both the word stem and rotary pursuit tests seemedto confirm this conclusion since the brain pathology of thesepatients included areas that are affected in both AD and HDpatients (Agid, Ruberg, Dubois, & Pillon, 1987; Hakin & Matheson,1979; Gaspar & Gray, 1984). Similar findings to that of Heindelet al. (1989) have now been reported by investigators who haveeither employed AD patients (Eslinger & Damasio, 1986), AD andnon-demented PD patients (Bondi & Kaszniak, 1991) or AD and HDpatients (Heindel, Butters, & Salmon, 1988).To summarize, studies adopting a systems model approach haveprovided information on some of the memory tests that AD patientsperform poorly in comparison to controls. Because these studiesfocus on mapping test performance to functionally impaired brainareas (i.e., lost neuronal function) they have not been designedto ask questions other than those that would address astructurally damaged network model (i.e., a memory deficitmodel). That is, neuropsychological studies have not generallyfocussed on whether access to the memory network is impaired bynon-memory factors or whether access can be achieved throughother routes. As such, these studies have been unable to provide56direct explanations for the reason AD patients can perform thetask but not the memory demands of tests. As well, these studiesdo not address why the memory performance of some AD patientsfluctuate during the day. For example, some patients may showgreater memory impairments in a morning than during an afternoontest session (personal observations). Additionally, with someinvestigators now deviating from traditional neuropsychologicalmethods and using different tests and changing the methods thatare used different findings are beginning to appear.One study that has shown that the word stem completionperformance of AD patients may not be impaired, relative tocontrols, was conducted by Partridge, Knight, and Feehan (1990).In this study 15 AD patients and 15 age matched controlsperformed a word stem task similar to that used in the Shimamuraet al. (1987), Salmon et al. (1988), and Heindel et al. (1989)investigations. The basic difference in the Partridge et al.,study was that, instead of encoding targets by providing apleasantness rating, subjects responded with the targets'meaning.Partridge et al. found no statistical difference betweentheir AD patients' and their controls' performance on the wordstem completion task. Since both groups' baseline performancewas not statistically different this confirmed that the priminglevels of each group were based on similar performance levels.57Unlike previous studies, Partridge found that their ADpatients had slightly higher priming levels than that of theircontrols (i.e., 31.4% vs. 26%). Since this patient group mayhave been more demented than those in the Shimamura et al., theSalmon et al., and the Heindel et al., investigations (p. 117)the reason for this finding remains unknown. One possibility isthat since the controls used in the Partridge study were obtainedfrom nursing homes, they may have been a low functioning controlgroup in comparison to those employed in the other studies. Theabsence of information on the functional abilities of Partridge'scontrols as well as precise information on the impairment levelsof the AD patients that are used in these type of studiesprecludes a decision about whether Partridge's findings might bedue to the subjects or to the encoding method that she used.A second study to show that priming for word stimuli may not beimpaired in AD patients employed a word identification task (seeKeane, Gabrieli, Fennema, Growdon, & Corkin, 1991). In thisinvestigation Keane et al. had 12 AD patients and 12 age-matchedcontrols view a series of 32 words, presented one at a time. Atstudy each word appeared on the computer and subjects were askedto read it to the experimenter. At test subjects were told thatthey would now perform a different task in which words would beflashed on the computer screen and subjects were to simplyidentify the word.58Keane et al. found that AD patients and controls identifiedstudied words faster than non-studied words and that themagnitude of the priming effects did not differ between subjectgroups. Thus, the Keane et al. results showed that AD patientscan show a similar magnitude of priming to controls.Adopting a systems-based explanation Keane et al. decidedthat their results provided evidence for the existence of animplicit memory system mediated by the occipital lobe (p. 340).In contrast, it is less clear how the postulation of anadditional neural system could account for the Partridge et al.results given that they simply changed the way in which ADpatients studied target information. The next section shows howrecent information from the field of cognitive psychology providealternative, but complementary explanations for these findings.Cognitive Investigations DissociAtions in Explicit and Implicit Test Performance. Asecond body of research that focuses on explicit and implicittest dissociations is found in the gerontological literature.The typical study compares the test performance of neurologicallyintact adults around the age of 70, to younger adults in theirearly 20's. The age-correlated incidence of AD (Treves, 1991)and the similar pattern of brain changes that occur in theelderly as compared to AD patients (albeit not to the samedegree; Corkin, 1982; Squire, 1987) makes this work pertinent to59this investigation.Irrespective of the type of explicit test that is used(i.e., free and cued recall, recognition) the memory performanceof older adults is found to consistently decline with advancingage (see Craik, 1978; Light, 1991; Poon, 1985, for reviews).This explicit memory pattern is similar to that discussed for ADpatients and has also been related to the brain changes thatoccur around hippocampal and surrounding cortical regions (i.e.,Davis & Bernstein, 1992; Price et al., 1991). However, unlikethat found for AD patients the age-related performanceimpairments on explicit tests disappear when implicit versions ofthe same test are used. To date this finding has beenestablished in a large number of studies using different types oftests and test materials. For example, older and younger adults'performance has been found to be similar in studies that haveused the word identification (Light & Singh, 1987), word stemcompletion (Java & Gardiner, 1991; Light & Singh, 1987), categorycompletion (Light & Albertson, 1989) and the picture fragmentcompletion test (Mitchell, in press). Similarly, the performanceof old and young adults has been found to be similar when wordand picture stimuli have been used (Java & Gardiner, 1991;Mitchell, in press).There does however appear to be one implicit test where thepriming levels of older adults are not always similar to younger- -60groups, and this is the paired associates test (Howard, 1988;Light & Burke, 1988; Light & LaVoie, in press). The pairedassociates test taps memory for newly experienced, unrelated wordassociations (i.e., queen-stairs, author-project; Howard, 1988).A cognitive interpretation for the reason performance on thistest differs from the category completion or word identificationtest is that, while all are implicit tests, they differ in theamount of elaborative-type processing they require. That is, thepaired associates test requires more elaborative (i.e.,attentionally demanding) processing than the category completiontest and older adults can meet the processing demands of thelatter but not former test. This does however leave the questionof why older adults are able to perform some, but not all formsof the paired associates test.According to Light and LaVoie (in press) the few studiesthat have used the paired associates test have used differentencoding methods, materials (i.e., words vs. non-words) and non-identical test forms (i.e., implicit and explicit). Since thecognitive processes required to perform this test has differed onseveral conditions that are important to the cognitivepsychologist (i.e., encoding, retrieval, material type) thismakes it impossible to pinpoint the source of these performancedifferences. A recent investigation by Howard, Fry, and Brune(1991) using similar tests and materials, but varying the61encoding conditions, indicates that one reason for theseperformance variations may be the extent to which targets areencoded for their meaning.Dissociations in Implicit Test Performance. In the Howardet al. study 20 young (approximately 20 years old) and 20 olderparticipants (approximately 69 years of age) studied pairs ofunrelated nouns (i.e., dog-apple). The same test form was usedto index explicit and implicit memory so that the main differencein cognitive processing requirements was whether the subjectintentionally or non-intentionally recollected targets. Forexample, subjects were provided with one word and a word stem(i.e., queen-sta^) and then asked to either complete the stemwith the first word that came to mind (i.e., the implicit or wordstem completion test) or to fill in the blank with a word theyhad previously studied (i.e., the explicit or wordstem cuedrecall test).In experiment one subjects were asked to listen to simplesentences containing unrelated word pairs and then make a quickaddition to this sentence (i.e., add a short phrase). Forexample, for the sentence containing the unrelated nouns house and zky, an addition might be ... are common Qbjects. Experimenttwo was different in that word pairs were presented to thesubject who used them to compose their own sentences. Subjectssubsequently rated how difficult that sentence had been for them62to develop (i.e., 1 = easy to 5 = difficult). For example, house : sky; My friend painted her house in a colour that reminds meof the gky after a thunderstorm. Howard et al. postulated thatthis latter task forced their subjects to attend more to themeaning of the critical targets than the encoding task used inexperiment one.Results showed that in experiment one the older subjectsrecollected significantly fewer word stem completion and cuedrecall stems than did the younger subjects. In contrast, inexperiment two the older subjects' performance on the word stemcompletion test was not statistically different from that of theyounger subjects. Nevertheless, performance on the word stemcued recall test remained significantly lower than that of theyoung subjects. Together these findings suggested that theencoding task used in experiment two may have been responsiblefor the elevation in the implicit test performance of the oldersubjects.The general gerontological literature supports theassumption that the encoding tasks may have differentlyinfluenced the implicit test performance of older adults. Thereis a large body of literature that shows that aging isaccompanied by attentional deficits which may be related to thefact that older adults do not spontaneously engage inelaborative-type encoding processes (Craik & Rabinowitz, 1984;63Kausler & Litchty, 1988; Plude & Hoyer, 1985; Rankin & Collins,1986).Since memory and attention are inseparately linked theattentional problems that accompany aging could be influencingmemory test performance at either encoding and/or retrieval(Graf, Tuokko, & Gallie, 1990). To explain further, to remembera particular event requires that subjects be able to focus andsustain attention during encoding processes. Specifically, thesubject must attend to those features of the target thatdistinguish it from non-essential characteristics whileconcurrently monitoring (i.e., attending to) other processingactivities. Similar attentional demands characterize memory testperformance (i.e., retrieval). That is, the subject must focusand maintain attention on the goal of the task, select anappropriate retrieval strategy and monitor its effectiveness,while ignoring irrelevant information in the test environment(Graf, Tuokko, & Gallie, 1990, p. 528).What connection does the relatio ship between attention and wemory have on ttle _results obtained by Howard et al.?  Aprocessing model interpretation is that the encoding method thatHoward et al. used in their second experiment was able to guidethe older subject in attending to targets at encoding (i.e., itassisted them in completing the encoding processingrequirements). However, it wasn't until this encoding method was64combined with a retrieval condition (i.e., non-intentionalrecollection) that was also less attentionally demanding (i.e.,required more integrative processes) that the overall processingdemands of the memory task reached a level that the oldersubjects could perform at levels that were similar to youngeradults.How are these findings relevant to AD patients? Firstconsider the question of attention. To date there are three mainlines of evidence to support the claim that AD patientsexperience attentional deficits. The first comes from clinicalobservations that some AD patients will not spontaneously encodetarget information. However, when the experimenter guides themthrough this processing by asking them questions about the targetthey often can complete the task. In addition, some AD patientsexhibit behavioral fluctuations that have been said to reflecttransient fluctuations in attentional focusing (cf. Graf, Tuokko,& Gallie, 1990). Second, as discussed in Chapter 2, theneuroanatomical damage associated with AD infiltrates areas thathave been ascribed various roles in attention (cf. Filoteo etal., 1992; Mountcastle, 1978). Third, there have been a fewinvestigators that have directly looked at the attentionalabilities of AD patients using different types of tasks. Forexample, in one study I used a computer to assess AD patients'ability to perform a vigilance, selective, and divided attention65task (Gallie, Graf, & Tuokko, 1991). The findings from thisstudy indicated that as the attentional demands of the task wereincreased, AD patients experienced greater difficulty incompleting these tasks in comparison to controls. In a secondinvestigation I followed the memory and attention performance ofmore than 100 AD patients during an 18 month period (Gallie,Tuokko, & Graf, 1991). Results from this study indicated that ADpatients experienced attentional deficits that were more severethan that of controls and that there was a relationship betweenexplicit memory performance, attention, and the patient's levelof F.I. That is, as AD patients changed from mild to severelevels of F.I., their performance on memory and attention tasksprogressively declined. Other investigators have chosen to lookat specific aspects of attention and a recent investigation byFiloteo et al. (1992) has found that AD patients encounterproblems when shifting their attention across target stimuli.How might these attentional deficits relate to previous findings of impaired implicit memory performance in AD patients? Most investigators have simply required that AD patients encodetargets by rating how pleasant that target is to them. Forexample, patients are asked to respond by saying five if theyagree that the target is pleasant or one if they think the targetis extremely unpleasant. While studies have found that thepleasantness encoding task results in normal priming levels for66KS and HD patients, work by Strauss, Weingartner, and Thompson(1985) and later, similar findings by Rohling, Ellis, and Scogin(1991) show that only KS and HD patients retain the ability tospontaneously and effortfully encode targets. Thus, thepossibility exists that one reason why studies have found that ADpatients show impaired priming on some tests is that they havenot used an appropriate encoding method with these patients.To synopsize, the information contained in this chapterprovides early evidence that the explicit/implicit distinctionmay be the tool, or Rosetta stone, that enables neuroscientists'to unravel further relationships between memory and the brain.To date this framework has shown that the priming abilities of ADpatients (unlike some other neurologically impaired groups) areimpaired on implicit tests of word and picture, but not motor-based materials. Damage to temporal and parietal cortical areaswith relative sparing of motor and somatosensory areas may be thereason for these findings. In turn, we reviewed evidence thatthe cognitive processes engaged by older subjects at encoding,retrieval, and in performing different implicit tests can changetheir memory performance. These findings show that memoryperformance changes when different cognitive processes areactivated on the same biological substrate. This idea is thepremise on which study one was based.Study One 67The goal of study one was to re-examine the implicit testperformance of AD patients. This study extends existing researchin that it combines both a process and a systems approach. Thiswas carried out by administering eight explicit and implicitmemory tests that were developed on the following principles.First, given that AD patients suffer from attentionaldeficits an encoding method was used with all memory tests thatensured that patients had attended to, and processed, criticaltargets. This method simply required that subjects identify andthen tell the experimenter what the target meant to them.Existing research shows that AD patients can effectively generatethe meaning of targets until the later stages of their disease(albeit with diminishing comprehension; Nelson & McKenna, 1978).Second, in order to tap hippocampal and non-hippocampalfunctioning explicit and implicit memory tests were used. Toensure that the primary differences in the processes these teststapped were those that were engaged at retrieval, parallel formsof each test were developed.Third, because existing research suggests that priming maybe dependent on modality-specific processing the test materialsused in this study were chosen for their ability to tap differentsensory modalities. For example, written word and pictures werechosen for their ability to tap a primarily visual type ofprocessing. Spoken words and tactilely presented objects were68used to engage primarily auditory and somatomotor processing. Toensure that maximum priming levels were achieved by subjects, allmaterials were presented in the same modality at study and test.Thus, unlike previous studies that have mixed the modalities oftheir study and test stimuli (i.e., study visually presentedstimuli and test auditorally) this study did not.The overall hypothesis of study one was that differences inthe memory performance of AD patients and controls would besmaller on implicit than on explicit tests using written andspoken word, picture, and object materials. This hypothesis wasbased on two main findings regarding the neuroanatomy of AD andone regarding the effects of encoding processes on implicit testperformance. First, previous research suggests that theneuropathology of AD begins in the hippocampus before spreadingto other brain regions that may be integral for performingimplicit memory tests. Thus, the explicit test performance ofpatients should always be lower than that of controls. second,the hippocampus processes multimodal information and so explicittest performance will be lower in AD patients compared tocontrols for all the test materials used (i.e., even for objectmaterials although the neuropathology of motor and somatomotorareas is generally preserved in AD patients). Third, theimplicit test performance of AD patients should be similar tothat of controls on all tests for the following reasons -- first,an encoding method was used that ensured that targets had beenencoded, and second, damage to those cortical areas presumedresponsible for priming appears to occur mainly in advancedstages of this disease (cf. Price et al., 1991).69CHAPTER 4: MEMORY STRATEGIESThis chapter introduces the concept of encoding andretrieval strategies (i.e., test types). The limited studiesthat have used memory strategies with AD patients are thenreviewed, together with the theories addressing how thesestrategies might work. Study two and its overall hypothesis isthen introduced.IntroductionMemory strategies can be defined as any activity that isused to increase the amount of information that can be explicitlyremembered (Bellezza, 1988). To date, systematic studies on howthese strategies should be classified have not been conducted(West & Tomer, 1989) and for the purposes of this chapter theywill be divided into those that are used during the acquisition(i.e., encoding) and those used during the retrieval ofinformation.Retrieval Strategies Retrieval strategies basically act to provide informationabout the to-be-remembered target and they have both a practicaland a formal use (Bellezza, 1988). The practical use ofretrieval strategies includes leaving notes (to ourselves) on thefridge or on calenders. In this application the effectiveness ofthese retrieval strategies depends on the extent to which we have7071provided enough clues to reaccess the original memory event. Theformal use of retrieval strategies occurs with the selection ofdifferent test types and this application is more familiar to theeducator or clinician. For example, depending on the patient (orthe class) the wise clinician (and educator) might choose arecognition (i.e., multiple choice test) rather than a freerecall test (i.e., or essay) to ensure that the patient's (orclass's) performance is not at floor. As in the previousexample, the principle at work is that more retrieval clues areprovided by a recognition (or cued recall) than by a free recalltest and this may assist some patients (and students) torecollect information that may not have been accessible withoutthis assistance. The underlying assumption is that a memory foran event has occurred, but there is a difficulty in reaccessingit. To compare this situation with the semantic network model ofthe previous chapter, there is either a problem in the memory ornon-memory (i.e., attention) related processes that are requiredto relocate the mental representation of the "learned" event.Encoding Strategies In contrast, encoding strategies influence the cognitiveprocesses that subjects engage to study targets (Bellezza, 1988).Many types of encoding strategies exist, including those wherethe subject simply repeats information (i.e., repetition) towhere more effort is employed (Bellezza, 1988). In this latter72case generating the meaning of a target or composing a mentalimage of the event (Bellezza, 1988).Research on memory strategies has been primarily focused onthe use of different encoding strategies with young children(Matlin, 1983). However, there is now an increasing interest inthe use of memory strategies with healthy older adults (Hill,Storant, & Simeone, 1990; Norris & West, 1991). Together thiswork has shown that strategies are most effective when they arematched to the cognitive abilities of the subject group they areto be used with (Duke, Weathers, Caldwell, & Novack, 1990). Toillustrate, rehearsal strategies are more effective in elevatingthe memory performance of young children, but not that of olderadults (Matlin, 1983). Since older adults have acquired a moreexpansive knowledge of the world, a strategy requiring them tothink more about the meaning of targets in relationship to whatthey already know is the more effective strategy for them(Matlin, 1983; Poon, 1985).In contrast, there have been very few investigations thathave used memory strategies with AD patients (Camp & McKitrick,1992). Several reasons can be cited for this, including theevidence that AD patients' memory abilities are progressively andirrevocably lost (cf. McKhann et al., 1984). While it has beenrecognized that AD patients' performance on recognition and cuedrecall tests is often higher than that found using free recall73tests, there have been few systematic studies on this effect (cf.Bellchambers, 1990). The next section reviews the equallylimited work on the use of encoding strategies with AD patients.This section is selective in that it reviews those strategiesthat engage the abilities that are best retained in AD patients;that is, their ability to perform tasks and generate the meaningof targets (cf. Eslinger & Damasio, 1986; Nelson & McKenna,1978). Due to the limited work with AD patients relevant studiesusing healthy older adults are selectively reviewed.Subject Performed Tasks (SPTs) One encoding strategy that emphasizes AD patients' retainedmotor abilities is referred to as subject performed tasks(Karlsson et al., 1989; Nyberg, Nilsson, & Backman, 1991).Subject performed tasks (or SPTs) are comprised of simple action-focused sentences like "Put the pen in the case" that the subjectperforms, and then is later asked to recollect (see Cohen, 1981).Investigators can judge the effectiveness of SPTs by comparingsubjects' performance on similar tasks that do not include motorenactment. For example, one condition is to simply ask subjectsto listen to, or to read the command sentences without performingthe task (i.e., VTs or verbal tasks). Another requires thatsubjects simply watch the experimenter perform the minitasks(referred to as EPTs or experimenter-performed tasks).74Investigators have found that both young and old subjectsremember more SPTs than VTs (Engelkamp & Cohen, 1991). Thus, itappears that performing a task provides superior recollection ofthe event compared to simply reading or listening to it,irrespective of the subject's age (Backman & Nilsson, 1984).SPTs also appear to be a particularly effective strategy forolder adults since most studies have found that old and youngadults remember equivalent numbers of these tasks (Backman &Nilsson, 1985; Cohen & Stewart, 1982). SPTs also benefitcognitively impaired young adults who often recollect similarnumbers when compared to non-impaired controls (Cohen & Bean,1983). In combination, this research suggests that SPTs may beeffective in elevating the memory performance of cognitively-impaired older adults, such as AD patients.To date only two groups of investigators have used SPTs withAD patients and they report very different results. The firstgroup was Karlsson et al. (1989) who found that AD patientsremembered significantly more SPTs than VTs. However, ADpatients did not remember as many SPTs as did their age-matchedcontrols.When Karlsson et al. (1989) gave these AD patientsadditional clues (i.e., retrieval clues) regarding the SPTs theyhad performed, this more than doubled the number of items theywere able to remember. However, the combined use of an encoding75and retrieval strategy still did not raise AD patients'performance to levels that were comparable to those of their age-matched controls.In this study Karlsson et al. (1989) also investigated if ADpatients in different levels of impairment might be experiencingdifferent memory benefits from the SPTs. Their results showedthat this was the case. Mildly impaired AD patients (n = 5)remembered approximately 36 percent of the SPTs they hadperformed with cueing. In contrast, the moderately impaired ADpatients (n = 8) remembered approximately 28 percent of the SPTswith cueing, and severely impaired patients (n = 7) rememberedapproximately 20 percent.The only other study to investigate SPT versus VTperformance in AD patients was Dick, Kean, and Sands (1989). Intheir investigation AD patients (n = 18) performed a total of 36minitasks (in comparison to Karisson et al. who used 25). Dicket al. results showed no significant difference in the number ofSPTs and VTs that were freely recalled by their AD patients. Thediscrepancy in the results reported by Dick et al. (1989) andKarisson et al. (1989) may be attributed to the different methodsthat they used.Recent investigations using healthy older adults have foundthat memory for SPTs declines when the experimental conditionsare made more demanding (Engelkamp & Cohen, 1991; Kausler,76Lichty, Hakami, & Freund, 1986; Norris & West, 1991). Forexample, when the number of SPTs are increased older adults'memory performance for SPTs decline (Engelkamp & Cohen, 1991).Since the method used by Dick et al. (1989) employed 36 SPTs incomparison to Karlsson et al. (1989) who used 25 this may be onereason for the different findings they report. Another reasonmay be the retrieval conditions that were used.Dick et al. required subjects to freely recall SPTs and VTs,whereas Karlsson et al. (1989) cued patients for target items.Research shows that cued recall conditions always elevate theamount of information that AD patients remember in comparison towhen they are required to freely recall an event (Bird & Luszcz,1991; Grober, Gitlin, Bang, & Buschke, 1992). Thus, when oneconsiders that the Karlsson et al. study used a lower number ofSPTs in addition to a cued recall test condition, this may haveproduced a memory task that was easier for AD patients than didthe Dick et al. (1989) study.Dick et al. (1989) also compared the performance of ADpatients in an SPT versus EPT task condition. In a free recalltest condition Dick et al. again found no difference in thenumber of SPTs and EPTs that were recollected by AD patients (p.80). Once again, the large number of tasks and the retrievalcondition Dick et al. used is a likely factor in producing thisresult.77To conclude, the two studies that have looked at the waythat SPTs change AD patients' memory performance have producedconflicting results. There is some evidence that requiringpatients to perform a reasonable number of tasks and providingthem with retrieval clues can elevate their explicit memoryperformance. In contrast, when a larger number of items areperformed in a free recall test condition, the memory performanceof AD patients is not elevated in comparison to VT or to EPTencoded tasks.Levels of ProcessingAD patients retain the ability to generate the meaning ofwords until the later stages of their disease (Bayles & Kaszniak,1987; Nelson & McKenna, 1978). This finding, in combination withresearch guided by the levels of processing framework, suggeststhat an encoding strategy that requires these patients togenerate the meaning of targets might be useful in elevatingtheir memory performance.Briefly, the levels of processing framework (Craik &Lockhart, 1972; Lockhart & Craik, 1990) views memory as theproduct of cognitive processes. One can engage in a type ofsensory-level processing that does not result in a long-lastingmemory trace; for example, when studying words for the number ofvowels or letters it contains. In contrast, a person can engage78in a semantic-level of processing, and this occurs when targetsare studied for their meaning. Semantic processing is viewed asresulting in a longer-lasting memory trace than sensory-levelprocessing.Support for Craik and Lockhart's (1972; Lockhart & Craik,1990) idea that semantically encoded information increases thedurability of a memory trace (i.e., increases the amount ofinformation that is remembered) comes from several experiments.The first was a series of experiments conducted by Craik andTulving (1975). In their first experiment Craik and Tulving(1975) found that more words were remembered when they werestudied for their meaning than for their physicalcharacteristics. College students were asked to view targetsthat were exposed by a tachistoscope, and as each target waspresented, they were asked to answer one of three questions. Toencourage a non-semantic type of processing subjects were asked"Is the word in capital letters?". To encourage subjects toengage in processing that Craik and Tulving (1975) considered asintermediate (i.e., between non-semantic and semantic processing)they used the question, "Does the word rhyme with^?". Thequestion "Is the word a type of^ ?" was used to encouragesubjects to process the target word at a more semantic level.At test, subjects were asked to remember all the targetwords they had viewed. Irrespective of whether free recall, cued79recall, or recognition test conditions were used, young adultsshowed that they remembered significantly more targets from thesemantic than from the non-semantic encoding conditions.Older adults also remember more words that they encode fortheir semantic versus their non-semantic characteristics (Craik,1978; Erber, Herman, & Botwinick, 1980). However, research showsthat several conditions must be present before a semanticencoding strategy will raise the memory performance of olderadults to levels that are equivalent to younger adults (Craik,1978). One condition is that older adults must be tested in arecognition or cued recall format and the second is that they areguided at encoding (Craik, 1978; Craik & Byrd, 1982). Thisguidance can be provided by simply having the subject state whatthe target means to them (Rankin & Collins, 1986).To date, there have been a limited number of studies thathave used semantic encoding strategies with AD patients (Bird &Luszcz, 1991). Similar to the case of SPTs, some studies havefound that AD patients recollect more meaning than non-meaning(or phonetically encoded targets) and others have not. Thissituation is illustrated by the following two investigations.Martin, Brouwers, Cox, and Fedio (1985) had AD patients (n =14) and age-matched control subjects (n = 11) respond to 9 targetwords by either telling the experimenter where the object couldbe found (semantic condition), or what the word rhymed with80(phonetic condition). As expected, control subjects rememberedsignificantly more words than AD patients in both the semanticand the phonemic encoding conditions. Of importance to thisdiscussion is that AD patients recollected more items they hadencoded in the semantic than in the non-semantic condition.In contrast, Corkin (1982) reported the absence of a levelseffect in AD patients (i.e., more meaning than non-meaningtargets being recollected). This study differed from the Martinet al. study in several ways including (1) the number of targetsthat were employed (n = 30), (2) the way that targets wereencoded, and (3) the retrieval condition that was used.Corkin had mildly (n = 11), moderately (n = 8), and severelyimpaired AD patients (n = 4) make yes/no judgements as to whether(1) a man or woman would say the target word (i.e., the non-meaning condition), (2) whether the target word rhymed with "X"(i.e., the phonetic condition), or whether (3) the target was atype of "X" (the semantic condition). At test, AD patients wereonly required to recognize target from distractor words.Similar to the Martin et al. investigation, Corkin (1982)found that AD patients did not remember as many targets as didcontrol subjects. Of importance to this discussion is that,unlike the Martin et al. investigation, Corkin found that ADpatients did not show a levels effect (p. 157).81Since Corkin (1982) used mildly, moderately, and severelyimpaired AD patients, this made it possible to examine whetherher severely impaired patients were influencing her findings.Results from these comparisons showed that the meaning (versusnon-meaning) performance of mildly impaired AD patients wasboosted by approximately ten percent, moderately impairedpatients showed a five percent increase, and severely impairedpatients showed no increase. In combination these findingsindicate that all patients were performing at levels that were ator near floor. This makes it unlikely that the inclusion of theseverely impaired patients accounted for this study's results.There are additional possibilities for why Corkin may nothave found a levels effect and one is that her patients may have,on average, been more impaired than those used in the Martin etal. study. Since Martin et al. did not ascertain the impairmentlevels of his patients it is not possible to comment further onthis possibility. What does appear obvious is that similarencoding conditions were not used in these two investigations.By requesting subjects to make a yes/no response Corkin wasunable to ensure that her subjects had encoded targets at theencoding level directed by her questions. This was unlike Martinet al. who required that subjects make completed responses to theexperimenter. An additional factor is that Martin et al.employed a smaller number of targets than did Corkin which,82similar to that observed in the SPT task, may have influencedtest results. This idea is supported by the work of Cermak andReale (1978) who found that amnesic patients exhibited a levelseffect only when target items were kept to a minimum (i.e., n =12).Theoretical Explanations SPT encode strategies. Several theories have been used toexplain why memory for SPTs are higher than VTs (Engelkamp &Cohen, 1991). Although there is little consistency in thesetheories they do share one common theme -- that SPTs engage fewerattentionally-demanding processes than VTs (Backman & Nilsson,1985; Cohen, 1985; Engelkamp & Zimmer, 1985).To date Backman and Nilsson (1984, 1985) have offered themost comprehensive explanation for why SPTs might require fewerattentionally demanding processes than VTs. Backman and Nilssonpropose that performing a task (i.e., SPTs) engages severalsensory modalities. For example, visual processing as you reachfor the target, and somatomotor processes as you grasp andperform the task. In some cases auditory, olfactory, and tastemodalities may also be activated. This multimodal processingpresumably acts to (1) guide the encoding of targets and (2)produce a stronger representation of the target event (Backman &Nilsson, 1989). A stronger representation of the event (i.e.,SPT) is presumed to occur because there are many variables on83which the target is processed; for example, on the basis of itscolour, weight, shape, texture, smell, etc. This type of multi-informational representation presumably makes retrieval of an SPTevent easier since a memory of the target can be reactivated fromseveral sources (Backman & Nilsson, 1989).In contrast, VTs require more attentionally-demandingprocesses than SPTs. This is because VTs (and presumably EPTs)are typically processed unimodally via visual or auditorymodalities (Backman & Nilsson, 1989). Unimodal processing ispresumed to provide less guidance at encoding as well as toproduce a more limited representation of the target event(Backman & Nilsson, 1989). A restricted representation of the VTevent occurs because the informational features on which they areencoded are limited to semantic, phonemic and/or non-semanticevent. Assumptions that VTs engage more attentionally-demandingprocessing than SPTs is supported by the results that have beenobtained from vigilance and divided attention tasks (Backman &Nilsson, 1989).Meaning encode strategies. Cohen, Sandler, and Schroeder -(1987) have offered their explanation why the semantic encodingof targets elevates memory performance. When subjects generatethe meaning of a target this guides their processing while at thesame time producing a "richer" representation of the event. Thericher representation makes the target easier to retrieve and it84lowers the attentional demands required to recollect the event.In contrast, encoding targets for their phonemic and physicalattributes does not provide a similar level of guidance atencoding. This results in a mental representation that is not aseasily retrieved, and thus, increases the attentional demands forrecollecting the event.Test or retrieval strategies. According to Perlmutter(1978) cued recall and recognition tests elevate memoryperformance because they provide information that helps to locatethe target, and this reduces the attentional demands of the task.Decreasing the number of items to-be-remembered also lowers theattentional demands of the task since fewer targets presumablyneed to be located (Backman & Nilsson, 1989).Study Two The goal of study two was to examine the effects thatdifferent encoding and retrieval strategies had in elevating theexplicit memory performance of AD patients. This study extendsprevious research in several ways. First, existing research hasprovided conflicting results that make it impossible to determinewith any certainty whether the SPT or the meaning encodedstrategies significantly elevate the explicit memory performanceof AD patients. Two different memory strategy experiments weredesigned to address this situation. One experiment examined thecontroversy between meaning and non-meaning encoded tasks, and85the second addressed SPT versus EPT tasks. These two experimentsextended previous research by (1) requiring that patients inknown levels of functional impairment (2) encode an ideal numberof targets (i.e., n = 12) in (3) conditions that enabled theseparate study of the effects that the encoding strategy and theretrieval (i.e., test) condition, as well as the combined effectsthat the encoding and the retrieval strategy had on memoryperformance.Second, this study extends existing research byinvestigating the effects of a new encoding strategy in elevatingAD patients' memory performance. This strategy required subjectsto engage increasing amounts of multimodal processing inconditions that required subjects to either silently read (theSee condition), read and say (the Say condition), or see, say,and perform (the Do condition) tasks. Similar methods (i.e.,number of stimuli, retrieval conditions) to that employed in theprevious two experiments were used in this third experiment.The overall hypothesis of study two was that AD patientswould recollect more targets from the meaning, SPT, and Do encodeconditions and that more targets would be recollected in eitherthe recognition and the cued recall, than in the free recall testcondition.86CHAPTER 5: METHODThis chapter has three main sections. The first describeshow volunteers were recruited and the subject sample that wasused. The next section discusses the materials used for theexplicit and implicit memory tests and the method used toadminister these tests. The final section describes thematerials used for the memory strategy experiments and the methodused to administer these experiments. All tables referred to inthese sections are located at the end of the chapter.Subjects Two groups of subjects participated in this investigation.One group consisted of Alzheimer patients from the Clinic forAlzheimer and Related Disorders, University Hospital-UBC site andthe second group served as non-demented, non-institutionalized,age-matched controls. Both groups were volunteers who met thefollowing general criteria: (1) they resided in the lowermainland or Vancouver Island region, (2) they had hearing andvision levels appropriate for psychological testing, and (3) theywere manifestly right-handed (i.e., wrote with their right hand).Alzheimer PatientsPatient Recruitment. Information from the Clinic forAlzheimer Disease and Related Disorders was used to identifypotential AD patients to participate in this study. After87qualified patients were identified (see criteria below), lettersinviting them to be a member of this study were sent to theirdesignated caregiver (see Appendix A for copies of theseletters). Of the 250 patients contacted by letter, 12 agreed toparticipate. Eight additional patients were recruited by Dr. B.Lynn Beattie, Director of the Clinic for Alzheimer Disease andRelated Disorders, who asked qualifying patients and theirfamilies to participate.Patient Selection Criteria. To be asked to participate inthis study patients had to meet the following criteria: (1) theyhad visited the Clinic for Alzheimer Disease and RelatedDisorders in the last 18 months, (2) they had a diagnosis ofpossible or probable Dementia of the Alzheimer's Type usingNINCDS-ADRDA criteria (McKhann, Drachman, Foistein, Katzman,Price, & Stadlan, 1984; see Appendix B for NINCDS-ADRDAcriteria), (3) there was no evidence of multiple strokes, cardiacinsufficiency, or previous head injury recorded in their medicalrecords, (4) they could perform basic psychological tests asindicated by the results from the neuropsychological part oftheir clinical assessment, and (5) both the patient and theirfamily agreed that the patient could participate.As part of their clinic assessment all patients had theirlevel of functional impairment evaluated using the FunctionalRating Scale (FRS: Tuokko & Crockett, 1989). The FRS has sound88psychometric properties (i.e., an inter-rater reliability of 0.94and validity of 0.84; Tuokko & Crockett, 1991) and scores fromthis scale were used to categorize patients into mild, moderate,and severe levels of functional impairment (F.I.). Categorizingpatients on the basis of their FRS score enabled the furtherinvestigation of possible differences in memory performance andstrategy effects due to F.I. Of the 20 AD patients whoparticipated in this study 6 were mildly, 7 were moderately, and7 were severely F.I. using criteria that we have established inother investigations (see Gallie, Tuokko, & Graf, 1991; Tuokko,Gallie, & Crockett, 1990). See Appendix B for a copy of the FRSand description of how scores were assigned to patients and usedto categorize them into F.I. groups.Table 1 summarizes the characteristics of the Alzheimerpatients who participated in this investigation. As can be seenfrom this table, Alzheimer patients had an average age of 70.8years (range = 50-84) and 12.5 years of education (range = 6-18).There were almost twice as many women as men and for fourpatients English was not their first language. Additionally, thenumber of patients with possible and probable diagnoses of ADwere not statistically different (see Table 3). Table 1 alsoshows that patients with mild, moderate, and severe levels ofF.I. generally had similar characteristics. The main exceptionwas the mildly impaired group which was slightly younger, had89more women than men, and took less medication.Control Subjects Control Subject Recruitment. Volunteers were recruited fromtalks given to local seniors' groups, radio interviews, and adsplaced in community newsletters and "UBC Reports". Afterindicating their interest in participating all volunteers werelater telephoned and asked several questions to determine theirsuitability for the study. (See Appendix C for the procedure andquestions asked during the telephone interview).Control Subject Selection Criteria. Only individuals whowere: (1) functioning independently in their home and notreceiving institutional care or services, (2) provided a self-report of good health including eyesight and hearing, (3) had nohistory of psychiatric or neurological disturbances includingdepression, memory problems, or major head injury, (4) were 50years of age or older, and (5) were manifestly right handed(i.e., wrote with their right hand) were selected to act ascontrols. From a total of 80 initial volunteers, the first 40who met selection criteria were chosen to act as non-dementedcontrols for this investigation'. In total, approximately 75% ofIt was not possible to corroborate control subjects'responses since most lived alone. Thus, to ensure that controlswere not demented they were asked a series of questions that aregood indicators of normal functioning (i.e., absence of self-reported memory problems, current employment, active social life,non-institutional living arrangements, etc.; Reisberg, 1983). Inaddition, controls' test performances were examined for low scoresthat might indicate dementia. If volunteers had failed to pass90these people had responded to ads placed in UBC Reports. Twelvecontrol subjects accepted a ten dollar honorarium at the end ofthe study (28 subjects declined the honorarium).Control Subject Characteristics. Table 1 summarizes thedemographic characteristics of control subjects who participatedin this investigation. As can be seen from this table, controlsreported an average age of 67.8 years (range = 51-89) and 14.83years of education (range = 8-27). As in the AD patient groupthere were twice as many women as men in this control group.Comparison of Control and AD Patient Characteristics Medications. All participants were asked about themedications they were taking to ensure the absence of a drug-induced memory impairment. The CPA's Compendium ofPharmaceuticals and Specialties (CPA, 1992) and Goodman andGilman's Pharmacological Basis of Therapeutics (Gilman, Goodman,Rall, & Murad, 1985) indicated that none of the medicationsreported by participants were likely to exert anti-mnemoniceffects.Medications fell into the six general categories listed inTable 2 (i.e., analgesic/anti-inflammatory, anti-arrythmic, anti-depressant, anti-epileptic, anti-hypertensive, and hormonal). Asthis table shows, although the control group reported taking alarger number of medications than did AD patients (i.e., 25 vs.these measures they would have been excluded from the study.917, respectively) there was a similar percentage of individuals ineach group who were not taking drugs. Table 2 also shows thatcontrols and AD patients took similar percentages of all drugs(within 5 percent) with the exception of analgesics/anti-inflammatories and anti-hypertensives (controls took 15% and 10%more of these drugs, respectively). In the case ofanalgesic/anti-inflammatories this difference was due to a largenumber of controls using aspirin for arthritic purposes.Statistical analysis indicated that there were no significantdifferences in the total number of medications taken by thecontrol and the AD patient group (see Table 3).Demographic Characteristics. Statistical comparisons ofcontrol and AD patients' demographic characteristics are reportedin Table 3. As the results of these analyses indicate, nostatistical differences were found in the age, proportion offemale and male participants, or occupation of control and ADpatients. The only characteristic on which these groups werestatistically different was that controls reported an average of2.33 more years of education than AD patients (i.e., t(58) =2.16, g < .04).Materials and Procedures for Explicit and Implicit Memory TestMaterials The same materials were used to develop explicit andimplicit versions of each memory test (i.e., for written and92spoken words, pictures, and objects). This was possible sincethe main difference between explicit and implicit tests was theinstructions provided to subjects at recollection.Written and Spoken Words. To investigate explicit andimplicit memory for written and spoken words, category cuedrecall and category completion tests were developed, respectively(see Appendix D). These tests were chosen since they enabled thesame materials and study procedures to be used to index explicitand implicit memory performance. The 32 categories required todevelop these tests were selected from Battig and Montague's(1969) well-established norms.To be selected, categories had to meet the followingcriteria: (1) they were common in everyday language (e.g., typeof weather), (2) they were unlikely to produce ambiguousresponses (e.g., type of fruit was chosen but pot , type ofscience), and (3) categories were unlikely to produce cohort-biased responses (e.g., type of tree was chosen but pot type ofdance).'Once 32 categories were selected they were randomlyorganized to form eight sets of four categories. Next, three' Results from Howard's (1980) study where adults fromdifferent age groups made responses to 21 of Battig and Montague'scategories were used to select 13 non age-biased categories. Theremaining categories that were required were chosen based on thedecisions of five independent raters that they were unlikely toproduce cohort-biased responses.93words that had been provided as responses to each category by atleast 30 percent of Battig and Montague's (1969) subjects werechosen (see Appendix D for the percentage of subjects respondingto each item). In total, 96 words were selected from Battig andMontagues's (1969) word category norms (i.e., 3 words/category x4 categories/set x 8 sets = 96 words in total). These words werethen printed in 8 mm block lettering on blank index cards (7.5 cmby 13 cm; see Appendix D). These stimulus cards were used forwritten and spoken word target and baseline conditions.Pictures. To investigate explicit and implicit memory forpictures, picture recognition and picture fragment completiontests were developed, respectively (see Appendix D). These testswere chosen since they enabled the same materials and studyprocedures to be used to index explicit and implicit memoryperformance. To develop these tests 48 pictures from Snodgrassand Corwin's (1988) well-known fragmented picture materials wereused (see Appendix D).'8 To develop these fragmented pictures Snodgrass andVanderwaart (1980) had 219 subjects rate 260 black and whitepictures on four variables known to be of central relevance tomemory and cognitive functions (i.e., name and image agreement,familiarity, and visual complexity). Snodgrass and Corwin (1988)then changed these pictures using a computer that randomly deletedsuccessively greater areas of critically important visualinformation. Snodgrass and Corwin (1988) then used this processto develop pictures that had eight levels of increasingly morecomplete visual information ranging from 8% at level one to 100%at level eight (see Appendix D for stimuli and informationregarding fragmentation levels).94To be selected pictures had to meet the following criteria:(1) they were correctly named by more than 70% of Snodgrass andCorwin's subjects, and (2) they had received ratings of 2 orgreater for both item familiarity and image consistency (scoresranged from 1 = very unfamiliar to 5 = very familiar). Overall,selected pictures were consistently named by a large percentageof subjects (i.e., 1 = 93.9%), were very familiar (11 rating =3.67), and closely matched subjects' mental image of how the itemshould look (M rating = 3.72; refer to Appendix D for ratings ofindividual pictures). These pictures were then semi-randomlydivided to make four sets of 12 pictures that had similar namingconsistency, item familiarity, and image consistencycharacteristics (see Appendix D).Assessing explicit and implicit memory required thatpictures be in two different forms, a 100% intact form (for thepicture recognition test) and eight levels of picture fragments(for the picture fragment completion test). Thus, each of the 48pictures was developed into a 100% intact and a fragmentedversion set (with eight levels of fragments per picture). Eachof the 48 intact pictures (7 cm by 7.5 cm in dimension) wasmounted on 12.5 cm by 13 cm black cardboard. In contrast, theeight fragmented versions of each picture (7 cm by 7.5 cm indimension) were arranged on 21 by 28 cm paper with the leastcompleted image in the lower right hand quadrant and successive95fragment levels continuing to the upper left corner (for examplessee Appendix D).Objects. To investigate explicit and implicit memory forobjects, a tactile recognition and a tactile identification testwas developed, respectively (see Appendix D). These tests werechosen since they allowed the same materials and study proceduresto be used to index explicit and implicit memory performance.The 48 stimuli required to develop these tests were chosen usingthe following criteria: (1) they were common everyday objects(e.g., spoon), (2) items weren't sharp to the touch, and (3) theywere small enough to fit behind a 90 cm by 46 cm by 20.5 cmcurtained partition. Chosen objects were then randomly arrangedto make four sets of 12 items (see Appendix D).Table 4, located at the end of this chapter, provides asummary of the explicit and implicit memory tests described inthis section.ProcedureGeneral. Since many of the volunteers selected toparticipate said they would not travel to U.B.C. all subjectswere given the choice of being examined at UBC or in their home.Approximately 90% of the AD patients and 40% of the controls weretested in their homes. All subjects were tested on an individualbasis.96Because fatigue factors affect the memory performance of ADpatients and older subjects, all participants were given thechoice of being tested in one or two sessions. Approximately 90%of the AD patients and 20% of the control subjects were tested intwo sessions. The average time it took control subjects tocomplete tests and experiments was approximately 4 hours comparedto 8 hours (with breaks) for AD patients.A pilot study where a counterbalanced order of tests andexperiments had been used indicated that most patients could notcomplete testing unless tests were staggered in an easy-difficultorder. Thus, a standard order of test presentation was adoptedthat is common to many AD investigations (cf. Heindel et al.,1989; Salmon et al., 1988). The following order of testpresentation was followed for both AD patients and controls:SPT-EPT experiment, memory tests for pictures, levels ofprocessing experiment, memory tests for written and spoken words,memory tests for objects, and the multisensory experiment (seeAppendix E for listing of test order).At the beginning of the first test session all participantswere verbally told of their rights as UBC subjects and wereprovided with a brief explanation of the study and the purpose oftheir involvement (i.e., to study memory in older adults). Theywere then asked to read a letter that contained the sameinformation and sign it to indicate their consent to participate.97AD patients and controls were provided with the same consentletter. Legal guardians of AD patients were provided with adifferent letter that requested that they ensure that the patientin their care understand the purpose of the study (see AppendixA). All participants were told they would receive a summary ofthe results upon completion of the investigation.Explicit and Ii,plicit Memory Tests. All memory tests beganwith two practice items that were used with all subjects (seeAppendix E). Practice items were used to ensure thatparticipants understood task requirements and had an opportunityto ask questions before studying critical target stimuli.Subjects who were unable to complete practice items were givenassistance and the experimenter returned to that item when theother practice item had been completed. If any subject had beenunable to complete the practice items with assistance, they wouldhave been excluded from the study. The same procedure was usedto administer target stimuli and all subject responses wererecorded by the experimenter in that subject's test booklet (seeAppendix E).Before beginning each test the experimenter shuffled cardscontaining the names or pictures of the stimuli that would beused. The order in which items appeared on these cardsdetermined the sequence of stimulus presentation. The exceptionwas object materials, which were randomly selected by the98experimenter from behind a partition that prevented the itemsfrom being viewed.Similar study procedures were followed for all memory tests.Subjects were first asked to identify targets before telling theexperimenter what that item meant to them. Requiring thatsubjects identify study items enabled the experimenter to be surethat subjects could provide the correct name for stimuli, whichis a problem for some AD patients and older subjects (see Bayles& Kaszniak, 1987). Thus, if subjects could name stimuli atstudy, but not a test, this suggested the presence of a memory,rather than a language impairment. Requiring that subjectsidentify stimuli also enabled the experimenter to check thatsubjects were attending to critical rather than non-criticaltargets. In addition, having subjects generate a personalmeaning for each critical target allowed the experimenter tocheck that subjects had focused on the target long enough toensure that items had been processed. A main premise of thisstudy was that the reason most previous investigations had notfound implicit test performance to be similar between AD patientsand controls was that they had not employed study methods whichensured that patients had focused on, and successfully encodedtarget stimuli.Implicit memory tests were performed before the explicittest for each material type (i.e., implicit test for pictures99before the explicit test for pictures). This order of testing isa standard method used to minimize the chance that subjects willdiscover that they can perform implicit tests by intentionallyrecollecting previously presented items (see Graf, Squire, &Mandler, 1984; Greene, 1988).Category Cued Recall and Category Completion Tests for Written and Spoken Word Materials. For these tests theexperimenter told subjects they would either hear or be shownwords written on cards. For each word that was presented theywere to first repeat the word, and then tell the experimenterwhat that word meant to them. For example, if the experimentersaid the word car, they might say, "Car, I bought my first carwhen I was 16, it was... etc."Cards designated as written practice items were visuallypresented to the subject; those selected as spoken practice itemswere read to the subject. Upon completion of the four practiceitems, the study phase for the category completion tests forwritten and spoken words began. All subjects were assigned eightsets of materials that had been counterbalanced across the targetand the baseline conditions of the category cued recall and thecategory completion tests for written and spoken word materials(see Table 5). During the study phase the 24 critical stimuliassigned to act as targets for the category completion tests forwritten and spoken word materials were studied using the same100procedure as practice items. Immediately upon completion of thestudy condition, the test phase for the category completion testsfor written and spoken word materials began.At test, subjects were told they would now start a differenttask. For this task they would either be visually or verballypresented with different category labels to which they were torespond with the first three exemplars that came to mind. Theexperimenter emphasized that subjects should provide only thosewords that first occurred to them since this task was like a wordassociation game. If subjects stated that they were attemptingto intentionally recollect previously presented targets they werediscouraged to do so by being told that this was a completely newtask.During the test phase subjects were presented with 16different category labels. Four of these labels corresponded tothe 12 critical targets assigned to the category completion testfor written word materials, and four categories matched the 12critical targets assigned to the category completion test forspoken word materials. The remaining eight categoriesrepresented 24 stimulus words that had not been presented atstudy and were assigned to act as baseline measures for thatsubject. In each case, the modality in which the category labelwas presented (i.e., visual or verbal) corresponded to theoriginal format in which the target had been presented.101Target scores were the number of critical target stimulithat were recollected from the written and the spoken wordconditions. Baseline scores corresponded to the number ofresponses that matched critical baseline stimuli selected fromBattig and Montague's (1969) norms (see Appendix D).The same general procedure used for category completiontests was repeated for the category cued recall tests with theexception of the test instructions that were provided tosubjects. For each category label subjects were asked to tellthe experimenter whether they had studied any words that wereexemplars of that category, and if so, what those three words hadbeen. Subjects were told that if they had previously been shownwords that belonged to that category label there would be threewords to remember. This information was provided to produce atest environment that was similar to that employed for thecategory completion test.Picture Recognition and Picture Fragment Completion Tests.The same general procedure used to administer the category cuedrecall and completion tests was followed for the picturerecognition and the picture fragment completion tests. A totalof four sets of 12 picture stimuli were counterbalanced acrossthe picture recognition and the picture fragment completiontarget and baseline conditions (see Table 5). After completingthe two standard practice items (see Appendix E) subjects were102then shown the 12 stimuli assigned to act as targets for thepicture recognition, and the 12 stimuli assigned to act astargets for the picture fragment completion tests. Immediatelyafter subjects named these critical target stimuli and told theexperimenter what these items meant to them, the test phase forthe picture fragment completion test began.At test, subjects were told that they would now start adifferent task. For this task they would be shown pictures thatvaried in their amount of completeness and the experimenter wouldask them what they thought each stimulus would make when it wascompleted. Similar to previous implicit tests, the experimenteremphasized that this was a game that was not related to anythingelse the subject had previously done.At test the experimenter exposed each fragmented stimuli bylifting one of the eight flaps of a cardboard sheet that coveredthe eight fragmented versions of each picture (see Appendix D).The experimenter began by exposing the most incomplete level ofeach picture (i.e., fragment level one) and continued to uncoversuccessive fragment levels (i.e., level two, etc.) until subjectshad correctly identified the stimulus. After the stimulus hadbeen correctly identified the experimenter then proceeded to thenext fragment level. If the subject again produced a correctresponse the first level in which that stimulus was correctlyidentified was recorded. The process of proceeding to the next103fragment level was employed to assure that subjects were certainof their responses. This procedure was repeated for the 12critical target and the 12 baseline stimuli assigned to thepicture fragment test condition for that subject.Target scores were the average fragment level at which alltargets had been correctly identified (i.e., the sum of thelevels in which fragmented pictures had been correctly identifieddivided by 12). Baseline scores were calculated in a similarmanner based on performance of the stimuli that had been assignedto that subject's baseline condition.Similar procedures to that used for the picture fragmentcompletion test were repeated for the picture recognition test.Two exceptions were that, (1) at test the set of intact (100%complete) pictures corresponding to the 12 critical target andthe 12 baseline stimuli assigned to that subject were used, and(2) subjects were requested to look at each picture and make ayes/no response regarding whether they remembered having beenshown that stimulus before. The number of target and distractorpictures correctly identified as well as the false positiveresponses made to distractor stimuli were recorded.Tactile Recognition and Tactile Identification Tests. Thesame general procedure described in previous sections was used toadminister the tactile recognition and the tactile identificationtests. A total of four sets of 12 object stimuli were104counterbalanced across the tactile recognition and the tactilecompletion target and baseline conditions (see Table 5). Aftercompleting the two standard practice items (see Appendix E)participants were then administered the 24 stimuli assigned toact as targets for the tactile recognition and the tactileidentification tests. Subjects were asked to put both of theirhands behind a curtain that prevented them from viewing theobjects. They were then told that the experimenter would placean object in their hands, one item at a time, and that theyshould identify the item as quickly as possible and then tell theexperimenter what that item meant to them. For example,"Baseball Hat", "You wear one of these to shade your eyes fromthe sun". Subjects were told to identify each object as quicklyas possible since the experimenter was measuring the time it tookfor them to identify each item. Using a stopwatch theexperimenter measured the time (in milliseconds) that elapsedbetween placing objects in the subject's hands until the item wascorrectly named.After completing the study phase the test phase of thetactile identification test was started. Subjects were informedthat they would now be asked to identify more objects. Theexperimenter placed 24 stimuli, one item at a time, in thesubject's hands with the request that they identify the object asquickly as possible. The 24 items consisted of the 12 target and105the 12 baseline items assigned to the tactile identificationcondition for that subject. The time until a correctidentification was made by the subject was again recorded.For the tactile recognition test a similar procedure to thatemployed for the tactile identification test was used. Theexception was that subjects were simply asked to tell theexperimenter if the item that was placed in their hands was anitem they had previously felt. Subjects were asked to makeyes/no responses to this request as the 12 target and the 12distractor items were randomly placed in their hands. The totalnumber of target and distractor objects correctly identified aswell as false positive responses made to distractor items wererecorded.Materials and Procedures Used in the Memory Strategy Experiments Materials This section contains information on the three memorystrategy experiments that comprised the second study of thisinvestigation. The materials that were used in these experimentsare described first. This is followed by a description of theprocedures used to conduct these experiments.yevels of Processing. For this experiment two sets of sixwords having similar letter length (mean = 5.5) and number ofvowels (mean = 2.5) were used. These two sets of materials werecounterbalanced across meaning and letter encode conditions.106Nine additional words with similar numbers of letters, vowels,and general meaning to the target stimuli served as distractorsthat were presented at test. All words were printed in 8 mmblock lettering on blank index cards (7.5 cm by 13.5 cm) (seeAppendix F).SPT-EPT Experiment. For this experiment 12 items fromCohen's (1981) list of minitasks were selected. The criteriaused to select these minitasks were that they: (1) were easy toperform and (2) were simple in meaning, for example, "Stretch theelastic band" and "Nod your head". Twelve minitasks wererandomly divided to make two sets of six tasks. These two setsof materials were counterbalanced across SPT and EPT conditions(see Appendix F).Multisensory Strategy. For this experiment four sets ofmaterials consisting of three sentences with similar verb stemswere created. For example, one set of sentences was: (a) touchyour nose, (b) touch your mouth, and (c) touch your ear. Thesecond set of sentences was (a) make a sign for goodbye, (b) makeA sign for yes, and (c) make a sign for no. These sentences wereprinted in 8 mm block lettering on white paper that was mountedon black cardboard (10 cm by 34 cm). One sentence was taken fromeach of the four sets of materials to make three final sets ofmaterials containing four different sentences (i.e., set one =touch your nose, make a sign for goodbye, put the pencil in the107case, make the clock say 3 o'clock). Each set of four sentenceswas assigned to either a (1) see, say and DO, (2) see and SAY or(3) a SEE only encode condition (see Appendix F).Procedure Similar to memory tests, all memory strategy experimentsbegan with two practice items to ensure that subjects understoodprocedures. Assistance was provided as required, stimuluspresentation was randomized, and the experimenter recorded allsubject responses in that subject's test booklet (see AppendixE).Jevels of Processing Experiment. This experimentinvestigated whether differences would occur in AD patients'abilities to remember words studied for their meaning versus thenumber of letters they contained. This is referred to as themeaning versus the letter encode strategy conditions. Theseencoding conditions were modelled after Craik and Lockhart's(1972; Lockhart & Craik, 1990) levels of processing framework.This experiment also investigated if AD patients would recollectmore targets under recognition versus free recall testconditions. Controls' performance in these encoding and testconditions was also examined.During the practice and study phases of this experimentsubjects were told they would be shown cards that had differentwords written on them. As in previous cases, the order of108stimulus presentation was random. Subjects were first asked toidentify the stimulus and then the experimenter would either askthem to count the number of letters in the word, or to tell theexperimenter what the word meant to them. Subjects were firstasked to identify the target word to ensure that the subject hadfocused on the stimulus. Each subject was assigned one set ofsix stimuli to act as targets for the meaning encode and one setfor the letter encode condition (see Appendix F). These two setsof materials were counterbalanced across each encoding condition(see Table 7).Immediately after the study phase finished the test phasebegan. At test the experimenter asked subjects to freely recallall the targets they had just studied. If the subject was unableto recollect all the targets, the experimenter then shuffled thestudy and distractor stimulus cards together and then presentedeach card to the subject, one card at a time. Subjects wereasked to respond by saying "yes" when the experimenter displayeda word they had seen at study, or "no" when it was a word theyhad not studied. This represented the recognition test of thisexperiment.The number of meaning encoded words that were correctlyidentified in the free recall and then the recognition conditionwere recorded. Similarly, the number of letter encoded wordsthat were remembered in the free recall and then the recognition109condition was recorded. Any target recollected in both freerecall and recognition conditions was included only in the freerecall score.SPT-EPT Experiment. This experiment investigated whetherdifferences would occur in AD patients' abilities to rememberminitasks they performed (Subject-Performed Minitask, SPT) versusthose they watched the experimenter perform (i.e., EPT). This isreferred to as the SPT versus the EPT encode strategy conditions.These encoding conditions were modelled after Cohen (1983) andBackman and Nilsson's (1984) work. This experiment alsoinvestigated if AD patients would recollect more targets undercued than free recall test conditions. Controls' performance inthese encoding and test conditions was also examined.At practice, and during the study phase of this experiment,subjects were told that they would be performing some simpletasks. Subjects were told that the experimenter had these taskswritten on cards, some of which the experimenter would do andsome that the subject would be asked to perform (see Appendix F).Before they could perform these tasks, however, subjects wouldneed to identify the physical objects they would use to performthese tasks. The experimenter then pointed to each object, oneat a time, and asked the subject to name each item. This ensuredthat failure to recollect minitasks was more likely due toproblems in recollecting the event rather than an inability to110produce the name for the objects involved in task performance.At study, all commands were read by the experimenter who hadrandomized their order of presentation by shuffling the stimuluscards. Each subject was assigned one set of materials thatserved as the targets for the SPT encode condition, and one setof materials for the EPT encode condition. These two sets ofmaterials were counterbalanced across each encoding condition(see Table 7).Upon completion of the study phase, the objects used toperform the minitasks were removed from the subject's view.Testing began immediately after the study phase and subjects wereasked to recollect all the minitasks that either they, or theexperimenter, had performed. Subjects were told they did notneed to remember who did what task, just what had been performed.This represented the free recall condition of this experiment.Tasks that were not remembered by the subject were cued by theexperimenter. This was done by placing the objects used toperform these tasks in front of the subject and the experimenterprovided verbal cues for items that had not been recollected.For example, the minitask "Cross your fingers" was verbally cuedby saying, "One of the commands said to cross something, do youremember what that was?" (see Appendix F for cues).The number of SPT-encoded minitasks freely recalled and thenthe remaining items recollected with cues were recorded.111Similarly, the number of EPT-encoded minitasks that wererecollected in the free recall and then the cued recallconditions were also recorded.Multisensory Experiment. This experiment investigatedwhether AD patients would remember more sentences if they (a)read the sentence to the experimenter and then performed what thesentence said (the DO condition), than if they (b) just read thesentence to the experimenter (the SAY condition), or if they (c)read the sentence silently to themselves (the SEE onlycondition). This is referred to as the Do, Say, and See encodeconditions, respectively. This experiment also investigatedwhether AD patients would recollect more sentences in cued thanfree recall conditipns.At practice and during the study phase of this experimentsubjects were told they would be shown cards with sentenceswritten on them, some of which they would be asked to readsilently to themselves, some of which they would be asked to readaloud to the experimenter, and some that they would be asked toread aloud and then do what the card said. Stimulus presentationwas randomized. Objects required to perform these sentences wereplaced in front of the subject. The experimenter pointed to eachof the objects, one item at a time, and asked subjects to namethe presented stimulus. As in previous experiments, thisprocedure was conducted to ensure that if stimuli were not112recollected it was more likely to be due to a memory, rather thanword finding difficulty. After the study phase of thisexperiment was completed objects were removed from the subject'sview.Each subject was assigned one set of materials to act astargets for the See condition, one set for the Say condition, andone for the Do condition (see Appendix F). These three sets ofmaterials were counterbalanced across each encode condition (seeTable 7).At test, subjects were asked to freely recall all thesentences they had been shown. Sentences that were notremembered by the subject were then cued by the experimenter(i.e., the cued recall condition). This was done by placing theobjects used to perform the sentence commands in front of thesubject and the experimenter then provided verbal cues for anytargets not recollected. For example, the sentence that said to"Touch your ear" was cued by the experimenter saying "Onesentence said to touch something, do you remember what that was?"The number of See, Say, and Do encoded sentences that werefreely recalled were recorded. Similarly, the number of See,Say, and Do encoded sentences that remained and were recollectedwith cueing were also recorded. See Appendix F for a listing ofthe materials used in memory strategy experiments and Tables 6and & 7 for a summary and listing of counterbalancing procedures.113Table 1Subject Characteristics.Controls(n=40IAD PatientsAll(n=20IMild*(n=6IModerate*(n=7I Severe*(n=7IWillM^67.8 70.8 68.2 70.0 73.9=^9.9 8.2 10.9 6.5 7.2Range^51-89 50-84 50-79 58-79 62-84EDUCATIONM^14.83 12.5 12.3 13.4 11.7ZD^4.2 3.3 2.6 4.1 3.2Range^8-27 6-18 10-17 6-18 8-16GENDERWomen^27 13 5 4 4Men 13 7 1 3 3AVERAGE NUMBEROF MEDICATIONSIS^0.63 0.35 0.16 0.42 0.425.12^0.95 0.59 0.41 0.55 0.55OCCUPATIONWhite Collar+ 24 11 3 4 4Blue Collar++ 10 4 2 1 1Housewife^6 5 1 2 2DIAGNOSISPossible^n/a 9 3 3Probable n/a 1 1 4 4Note. * Level of Functional Impairment+ White Collar (e.g., teacher, nurse)++Blue Collar (e.g., painter, bus driver)114Table 2Medications Taken by Controls and AD Patients.Medication Controls(n=40) AD Patients(n=20)Analgesic/N^I N IAnti-inflammatory 10 25 2 10Anti-arrythmic 3 7.5 1 5Anti-depressant 3 7.5 2 10Anti-epileptic 1 2.5 1 5Anti-hypertensive 4 10 0 0Hormonal* 4 10 1 5Total number ofMedications 25 7Number of IndividualsTaking Medications 15 37.5 6 30Note. This table summarizes medications taken by AD patients asrecorded in their medical files and updated by a spouse orsibling collaborator. For control subjects thisinformation was obtained on a self-report basis.* Medications in this category included estrogen and thyroidreplacement.EDUCATION^t= 2.162GENDER^X 0.00I MEDICATIONS j,= 1.182OCCUPATION^x 0.55POSSIBLE/^2PROBABLE Dx.^;X:= 0.95115Table 3Comparison of Control (n=40) versus AD Patients' (n=20)Characteristics.Variable^Statistic^ProbabilityAGE^t= -1.18 ++ 0.24 58++ 0.04* 58+ 1.00 1++ 0.24 580.76 20.62 2Note. * Significant at p <. 05+ Yates Correction For Continuity (Yates, 1934)++ Based on Pooled Variance Estimates to adjust forunequal sample sizes.116Table 4Explicit and Implicit Memory Tests Used in this Investigation.Material Type^Test Type^Test NameWritten words^Implicit^Category completionExplicit^Category cued recallSpoken words^Implicit^Category completionExplicit^Category cued recallPictures^Implicit^Picture fragment completionExplicit^Picture recognitionObjects^Implicit^Tactile identificationExplicit^Tactile recognition117Table 5Counterbalancing Method Used for Explicit and Implicit Tests.Four sets of written word, spoken word, picture, and objectmaterials were counterbalanced across Implicit and ExplicitMemory Test, target and baseline conditions. Note, for writtenand spoken word tests the same materials were used so that atotal of eight sets of items were counterbalanced across writtenand spoken word conditions.Counterbalancing was conducted across test conditions forthe AD patient and the control subject groups. The followingdiagram shows the counterbalancing procedure that was used. IMP designates the implicit and EXP the explicit test conditions.denotes target and B denotes baseline. Each line denotes thecounterbalancing schedule for one AD patient or one controlsubject. Thus, line one is for subject one, line two for subjecttwo, etc. Each number (i.e., 1, 2, 3, 4, etc.) denotes thematerial set number (see Appendix D for items contained in eachset of materials).Imp/Exp forWritten Words Imp/Exp forspoken Words Imp/Exp forPictures Imp/Exp forObjectsINE EKE IKE^EKE EXP ME^EKETB TB1 T$ T T TB TB1 2 3 4 5 6 7 8 1 2 3 4 1 2 3 42 1 4 3 6 5 8 7 2 3 4 1 2 3 4 13 4 5 6 7 8 1 2 3 4 1 2 3 4 1 24 3 6 5 8 7 2 1 4 1 2 3 4 1 2 35 6 7 8 1 2 3 4 etc. etc.6 5 8 7 2 1 4 37 8 1 2 3 4 5 68 7 2 1 4 3 6 5etc.118Table 6Summary of the Memory Strategy Experiments Used in thisInvestigation.Each of the three memory strategy experiments listed belowhad (1) two different encoding conditions, and (2) two differenttest retrieval conditions. The exception was the multisensoryexperiment which had three different encoding conditions.Name^Experimental ProcedureLevels of^In this experiment subjects studiedProcessing twelve words using two different encodingstrategies. (1) Six words were studiedby telling the experimenter what the wordmeant (= meaning encode condition), and sixwords by counting the number of letters theword contained (= letter encodecondition). (2) Retrieval conditions beganwith a free recall then recognitioncondition.SPT-EPT Strategy^In this experiment subjects studied twelveminitasks using two different encodingstrategies. (1) Six minitasks were studiedby having the subject perform the minitask(= SPT or active encode), and sixminitasks by watching this experimenterperform tasks (= EPT or passiveencode strategy). (2) Retrieval conditionsbegan with a free recall and then cued recallcondition.MultisensoryStrategyIn this experiment subjects studied twelvesimple sentences. (1) Four sentences werestudied having subjects read sentencessilently to themselves (= See encode),four sentences were studied having subjectsread sentences aloud to the experimenter(= Say encode) and four sentenceswere studied having subjects read thesentence to the experimenter and performwhat the sentence said (i.e., touch yournose = Do encode condition).(2) Retrieval conditions began with afree recall then cued recall condition.119Table 7Counterbalancing Method Used for the Memory Strategy Experiments.In the (1) Levels of Processing and (2) SPT-EPT MemoryStrategy Experiments two sets of materials were counterbalancedacross encoding conditions. Three sets of materials werecounterbalanced across the Multisensory Strategy Experiment.Similar to Implicit and Explicit Tests counterbalancing wasbetween subjects. That is, counterbalancing of AD patients wasmaintained separate from control subjects.Each line denotes the counterbalancing schedule for one ADpatient or one control subject. Thus, line one is for subjectone, line two for subject two, etc. Each number (i.e., 1, 2, 3)denotes the material set number. See Appendix F for itemscontained in each set of materials.Levels of Processing SPT-EPT^MultisensoryExperiment ^Bxperiment Experiment Mean / Letter^Subject/Experimenter^See/ Say / Do1^2 1^2^1^2^32 1 2 1 ^2^3^1etc. etc. 3^1^2etc.120CHAPTER 6: RESULTSThis chapter has three main sections. The first sectiondescribes the design and the analyses of results. The nextsection reports the results obtained from explicit and implicitmemory tests. In the final section I examine the data from thememory strategy experiments. Tables reporting descriptive andinferential statistics are located in Appendix G.Overall Design and Analyses Repeated measures Multivariate Analyses of Covariance(MANCOVA: Subjects' level of education as the covariate) with onebetween-subjects factor (Subjects: AD patients, Control Subjects)and one within-subjects factor (Test Type: Explicit, Implicit)were used, where described, to analyze the memory test results'.Similar statistical designs were used to analyze the resultsfrom the memory strategy experiments with the exceptions that thewithin-subjects factor for the levels of processing and the SPT-EPT experiments were Encoding Strategy: Retrieval Test, and eachof these factors had two levels. For the multisensory experimentthere were three levels to the Encoding Strategy factor (i.e.,the Do, Say, and See conditions).All multivariate analyses of covariance (MANCOVA) tests wererepeated using multivariate analyses of variance (MANOVA) todetermine covariate effects on results. In no case did MANCOVAand MANOVA analyses result in different conclusions being reached.Thus, the 2.33 years more education reported by control subjectsdid not appear to significantly influence the memory test resultsor the memory strategy experiment results.121Repeated Measures MANCOVA analyses were used since theyprovided the most powerful and relevant statistical tool foraddressing the hypotheses of memory tests and strategyexperiments (see Davidson, 1972; Kirk, 1968; O'Brien & Kaiser,1985). To guard against making Type 1 errors Bonferonnicorrections were applied. For study one the alpha level perMancova analyses was set at .01 (i.e., 5 analyses/.05) and forstudy two alpha was .0167 (i.e., 3 analyses/.05) following theprocedure recommended by Rosenthal and Rosnow (1991).Analyses Significant main effects were followed by post hoc tests andsignificant interactions were analyzed with main effect tests (asper Winer, Brown, & Michels, 1991). To correct for unequalsubject numbers the unweighted means approach using dummyvariables (as discussed in SPSS-X, 1990) was used in each MANCOVAanalysis (as per Howell, 1987; Rosenthal & Rosnow, 1991; Winer,Brown, & Michels, 1991). To test for homogeneity of varianceBox's test for multivariate F's and then the Bartlett-Box test"for univariate F's were performed (as per Glass & Hopkins, 1984;Howell, 1987). When heterogeneity of variance occurred, Box's(1954) procedure for adjusting the degrees of freedom for F" Several persons have made contributions to this test whichhas also been referred to as the Box test, Bartlett-Kendall, andScheffe test (Glass & Hopkins, 1984). This test was chosen sinceit has special applicability when group n's are not equal (Glass& Hopkins, 1984).122was employed as described by Howell" (1987, p. 297).When studying patients of mixed dementia severity it isimportant to determine if outlying scores have biased averagegroup results. Scrutiny of the results obtained in the memorytest and the strategy experiments showed that the standarddeviations of most variables were within acceptable ranges (i.e.,the majority are within 3 $Ds). Examination of raw scoresconfirmed the general absence of extreme scores except when notedin the discussion of the results. Additionally, checks onnormality were performed by establishing that AD patients' scoredistributions were similar in shape to those of control subjectsand devoid of extreme platykurtic and leptokurtic characteristicsfollowing the procedure described by Glass and Hopkins (1984).Study OneExplicit and Implicit Memory Tests The focus of this part of the investigation was to exploreAD patients and non-demented, age-matched control subjects'performance on four sets of explicit and implicit memory tests(see Table 4 of Method section for listing). The overallhypothesis was that differences in the memory performance of ADn Box (1953, 1954) has shown that when both variance and groupsize are unequal, a valid and conservative test of the significanceof group differences can be carried out if the degrees of freedomfor F mites are altered from df=(k-1, k(n-1)) to df=(1, n-1). Ifthis adjustment leads to a significant result then group means aresignificantly different regardless of the variances.123patients and controls would be smaller on implicit than explicittests, using written and spoken word, picture, and objectmaterials.A strict scoring criterion was applied to all memory tests.Responses were considered correct only if they matched theessential form of the original stimulus. For example, changes intense and plural forms were acceptable, but not substitute wordswith similar meaning.Category Cued Recall and Category Completion Tests for WrittenWord Materials Critical Measures. For the category cued recall and thecategory completion test the critical dependent measure was thenumber of written words identified in the target and the baselineconditions. Items were scored as correct if they matched one ofthe 12 words presented at study (the target condition) or one ofthe 12 words designated to act as baseline for that subject.Corrected scores were computed by subtracting target frombaseline performance. This is the standard method employed forobtaining an index of priming or implicit memory performance (seeGraf, Shimamura, & Squire, 1985; Salmon, Shimamura, Butters, &Smith, 1988). The corrected score provides several advantages tousing the target score in the analyses of results. One advantageis that corrected scores provide an index of the magnitude ofpriming that is not influenced by differences in the subjects'124baseline scores (Salmon et al., 1988). A second advantage isthat corrected scores are not as biased by guessing and/or chanceperformance on tests (see Graf et al., 1985; Salmon et al.,1988). Average percentage corrected scores were computed bydividing the corrected score by 12 and multiplying by 100.AD Patients and Control Subjects. Figure 2 shows theaverage percentage corrected scores for the AD patients and thecontrol subjects on the category cued recall test and on thecategory completion test for written word materials. Asexpected, the average percentage corrected category cued recallperformance of AD patients was much lower than that of the age-matched control subjects (i.e., N = 15.4%, 5D = 2.28 vs. M =72.3%, 5D = 1.83). In contrast, the average percentage correctedscores on the category completion test was more similar betweenAD patients and control subjects (M = 10.8%, 5D = 2.20 vs. M =13.5%, 5D = 2.39, respectively).Some additional points must be made regarding the results 'obtained from the category cued recall and the categorycompletion tests. The first is that patients' and controls'performance on both tests were not at floor. This was confirmedby inspecting the group where the smallest difference betweentarget and baseline performance had occurred to confirm that asignificant difference existed between scores [i.e., AD patientson the category completion test, t(19) = 2.36, p < .032].irAControl(n•40)All ADPatients(n-20)Mild^Mod.^Severe(n-6)^(n.7)^(n-7)Control^All AD^Mild^Mod.^SeverePatients(n-40)^(n-20)^(n-6)^(n-7)^(n-7)Implicit Test(Category Cued Recall) (Category Completion)Figure 2. Percentage corrected score of Alzheimer patients and of Control subjects on explicit and implicit tests for written word materials.Mildly, moderately and severely impaired patient groups' performance are shown separately. The error bars represent standarddeviation scores.Explicit Test0o -0co000'etOcsi0000CO0•0v.cooto0Ns'OOControl Subjects (n-40)All AD Patients (n-20)Mild (n-6)Moderate (n.7)Severe (n-7)'SV126Second, statistical inspection showed that there was asignificant difference between the baseline performance ofcontrols and AD patients on the category completion [E(1,58) =10.56, R < .005] but not on the category cued recall test[E(1,58) = .13, R > .5]. Since the corrected scores obtained forthe controls and the AD group on the category completion test wasnot based on similar baseline levels, some caution is required inthe interpretation of this test result. That is, although themagnitude of the corrected scores may be similar they are derivedfrom different performance levels that may reflect differentmemory processes. For instance, controls may have been engagingin the explicit recollection of items while performing implicittests although the method used in this study (i.e., presentingimplicit before explicit tests) would have minimized this effect.It should be noted that the magnitude of the corrected scoresobtained on the category completion test was similar to thoseobtained in a pilot investigation where a similar task was usedwith different subjects (see Graf, Tuokko, & Gallie, 1990).To examine if AD patients showed similar average percentagecorrected scores to controls on the category completion test,despite lower performance on the cued recall task, a repeatedmeasures Multivariate Analysis of Covariance (covariate factor =127education) was performed. The results of this MANCOVAl2 showedsignificant main effects for subject group" f(1,57) = 51.14, p <.001, MSe = 306.47 and test type f(1,58) = 110.33, p < .001, MSe = 385.07. A significant interaction between group and test typewas also detected, f(1,58) = 80.70, g < .001, MSe = 281.67.When the group by test interaction was analyzed with maineffect tests, significant differences were evident betweencontrol and AD patients' corrected performance on the categorycued recall test f(1,58) = 157.08, p < .001, MSe = 621.07, butnot on the category completion test, were found (see Table G-8b).Further examination confirmed these findings by showing thatthere was a larger effect of test type on subjects' performanceon the category cued recall (eta-squared = .73) than on thecategory completion test (eta-squared = .004).The next step was to decide whether the significantly lowercorrected scores of the AD patients on the category cued recalltest was influenced by the instructions provided at test, or toan inability to perform the basic task of providing categoryexemplars. Since AD patients and controls had provided astatistically similar number of responses to baseline items (aspreviously reported) this suggested that these patients could12 Results of the Box M and Bartlett Box tests confirmed thathomogeneity of variance assumptions were met in this analysis." One degree of freedom is lost from the within source ofvariation due to the covariate of education.128perform the non-memory, or task-related requirements of the test(see Salmon et al., 1988, p. 487; Shimamura et al., 1987, p.349).To summarize, these results supported the hypothesis thatdifferences in the memory performance of AD patients and controlswould be smaller on implicit than explicit tests of written wordmaterialS. These results indicate that when AD patients wereprovided with test instructions that did not require them tointentionally recollect written word materials, they showedmemory performance that was similar to non-demented, non-institutionalized, age-matched controls. In contrast, when thesame patients were instructed to intentionally recollect the samematerials, they showed a corrected score performance that wassignificantly lower than controls.Mildly, Moderately. and Severely Impaired AD Patients. Asecondary question was how AD patients with different levels offunctional impairment (F.I.) would perform on the categorycompletion test, despite expected variations in their performanceon the category cued recall test (see Gallie, Tuokko, & Graf,1991; Tuokko, Gallie, & Crockett, 1990). Figure 2 shows thatpatients that were in mild, moderate, and severe levels of F.I.displayed similar average percentage corrected scores for thecategory completion test. In contrast, there was a greatervariation in the average percentage corrected performances of129these patients on the category cued recall test.The small number of patients in each F.I. group precludedthe use of inferential analyses. However, similar findings werefound on subsequent implicit tests suggesting that the consistentlevels of performance in mildly, moderately, and severely F.I.patients was not artifactual in nature (note that thisperformance was not at floor). Refer to Table G-8a for acomplete listing of these results.Category Cued Recall and Cateaory Completion Tests for Spoken Word MaterialsCritical Measures. These were the same as described forwritten word materials with the exception that stimuli wereverbally presented..AD Patients and Control Subjects. Figure 3 shows theaverage percentage corrected scores for the AD patients and thecontrol subjects on the category cued recall and on the categorycompletion test for spoken word materials. Similar to theresults obtained using written word materials, the averagepercentage corrected performance on the category cued recall testfor AD patients was lower than that of age-matched controls(i.e., M = 20.4%, SD = 2.56 vs. M = 70%, SD = 2.67,respectively). In contrast, the performance of AD patients andcontrols was closer on the category completion test (M = 9.2%, SD= 1.55 versus M = 14.4%, SD = 2.48, respectively).Mild^Mod.^Severe(n.6)^(n-7)^(n-7)Explicit Test(Category Cued Recall)Control^All ADPatients(n-40)^(n-20)^(n-6)^(n-7)^(n-7)implicit Test(Category Completion)Mild^Mod.^SevereFigure 3. Percentage corrected score of Alzheimer patients and of Control subjects on explicit and implicit tests for spoken word materials.Mildly, moderately and severely impaired patient groups' performance are shown separately. The error bars represent standarddeviation scores.00 -OCOOOControl(n-40)All ADPatients(n-20)00 -OCOar8OtoO4.)rnal CorU00Control Subjects (n-40)All AD Patients (n-20)Mild (n-6)Moderate (n-7)Severe (n-7)7 ,,131Two additional points must be made regarding the resultsobtained from the category cued recall and the categorycompletion tests. The first is that patients' and controls'performance on both tests were not at floor. This was confirmedusing the same method that was previously described (i.e., ADpatients' category completion performance, t(19) = 2.78, p <.013). Second, statistical inspection showed that there was asignificant difference between the baseline performance ofcontrols and AD patients on the category completion [F(1,58) =11.95, p < .002] but not on the category cued recall test[E(1,58) = .053, p > .5]. Thus, the magnitude of the averagepercentage corrected scores obtained for AD patients and controlson the category completion test were not based on similarperformance levels.To examine whether AD patients showed similar averagepercentage corrected scores to controls on the categorycompletion test, despite lower performance on the category cuedrecall test, a Multivariate Analysis of Covariance (covariate =education) was performed. The results of this MANCOVA showed asignificant main effect for subject group E(1,57) = 36.44, p <.001, MSe = 232.71. Since the Bartlett-Box test indicated afailure to meet assumptions of homogeneity of variance on thefactor of implicit test type, Box's procedure for adjusting Fcritical was applied (Box, 1954). With Box's adjustment the main132effect of test type continued to be highly significant, E(1,57) =80.68, 12 < .001, MSe = 429.34 together with a significantinteraction between group and test type, E(1,57) = 35.52, R <.001, MSe = 189.04.When the group by test interaction was analyzed with maineffect tests and Box's adjustment, significant differencesbetween the controls' and the AD patients' average percentagecorrected performance on the category cued recall test Z(1,58) =68.01, p < .001, MSe = 472.03, but not on the category completiontest were found (see Table G-9b). Eta-squared analysis confirmedthat the effect of test type on subjects' performance was muchlarger for the category cued recall (.54) than on the categorycompletion test (.0fl. Statistical analysis of baselineperformance showed that AD patients and controls performed atsimilar levels on the category cued recall test (valuespreviously reported) suggesting that patients were able toperform the basic task of providing category exemplars but haddifficulty when intentionally recollecting target stimuli.To summarize, the results reported in this section supportedthe hypothesis that differences in the memory performance of ADpatients and controls would be smaller on implicit than explicittests of spoken word materials. Similar to conclusions reachedfrom tests using written word materials, it appeared that when ADpatients were directed to non-intentionally recollect targets133they showed memory performance that was similar in magnitude tonon-demented controls. In contrast, the same patients encountereddifficulty when directed to intentionally recollect the samestimulus materials.Mildly, Moderately, and Severely Impaired AD Patients. Asecondary question was whether mildly, moderately, and severelyF.I. patients would exhibit similar average percentage correctedscores on the category completion test despite expecteddifferences on the category cued recall test. As Figure 3 shows,a similar profile of results to that found with written wordmaterials was obtained. The average percentage corrected scoreon the category completion test was similar for mildly,moderately, and severely F.I. patients (note that thisperformance was not at floor). However, this was not the casefor the category cued recall test where variations in performanceacross F.I. groups can be observed. (Refer to Table G-9a for acomplete listing of these results).Picture Recognition and Picture Fragment Completion Tests Critical Measures. For the recognition test the criticaldependent measure was the number of pictures correctly identifiedin the target (n = 12) and the baseline conditions (n = 12). Inthe picture fragment completion test the critical dependentmeasure was the fragment level (of 8 possible levels) at whichthe target (n = 12) and baseline items (n = 12) were correctly134identified. For each of these tests, responses were scored ascorrect if they matched stimuli from the target and baselineconditions assigned to each subject.Average percentage corrected scores were computed for boththe picture recognition and the picture fragment completiontests. In the case of the picture recognition test, averagepercentage corrected scores were computed by subtracting target(hits) from false positive (or false alarm) responses, dividingby 12 and multiplying by 100. Subtracting target from falsepositive scores is the method recommended for correcting ADpatients' performance on recognition tests (see Braconnier, Cole,Spera, & De Vitt, 1982; Snodgrass & Corwin, 1988). Averagepercentage corrected scores for the picture fragment completiontest were computed by subtracting target from baseline scores,dividing by 12 and multiplying by 100.AD Patients and Control Subjects.  Figure 4 shows theaverage percentage corrected scores for the AD patients and thecontrol subjects on the picture recognition and the picturefragment completion tests. As expected, the average percentagecorrected picture recognition score of AD patients was much lowerthan that of controls (i.e., }1 = 49.2%, alp = 3.78 vs. 11 = 97.5%,5.p = .61, respectively). AD patients' average percentagecorrected picture fragment completion score was also lower thancontrols (i.e., X = 6.1%, SD = .64 vs. M = 13.1%, Sp = .65,respectively). These performance scores were similar in000co00003:12040le,o0,o,eti)0NT><0OCMO010 0Mod. Severe Control Mild^Mod.^Severe(n-40) (n.6)^(n.7)^(n-7)1011All ADPatients(n•20)Control^All AD^MildPatients(n•40)^(n.20)^(n-6)^(n.7)^(n.7)Control Subjects (n-40)All AD Patients (n-20)Mild (n•6)Moderate (n.7)Severe (n.7)Explicit Test(Picture Recognition)Implicit Test(Picture Fragment Completion)Figure 4. Alzheimer patients' and Control subjects' performance onscores are shown.Mildly, moderately and severely impaired patient groups'The error bars represent standard deviation scores.explicit and implicit tests for picture materials. Percentage correctedperformance are shown separately.136magnitude to those obtained by Bondi and Kaszniak (1991) usingthe picture fragment completion test with a pleasantness encodecondition (i.e., AD patients: n = 12, N = 4.2%; Controls: n = 16,M = 16%; M. Bondi, personal communication, June, 1992).Patients' and controls' performance was not at floor on thepicture recognition test (i.e., AD patients' target vs. baselineperformance, E(1,38) = 31.57, R < .001). However, AD patients'performance was at floor on the picture fragment completion test,t(19) = 1.87, R > .98. This finding corresponds to a similarproblem reported by Heindel et al. (1990) using a picturefragment completion test with 5 levels of picture fragments.Thus, using a test with 8 different fragment levels still did notprovide enough discriminating ability to detect differences in ADpatients' performance.The baseline performance of AD patients and controls wassignificantly different on both the picture recognition [E(1,58)= 16.81, p < .001], and the picture fragment completion tests[E(1,58) = 12.17, R <.002]. These findings are similar to thosereported by Bondi and Kaszniak (1991) and Heindel et al. (1990)and they suggest that AD patients were encountering greaterdifficulty than controls in performing the task components of thepicture recognition and picture fragment completion tests.To examine the hypothesis that differences in the memoryperformance between AD patients and controls would be smaller on137the picture fragment completion than the picture recognition testa repeated measures Multivariate Analyses of Covariance(covariate factor = education) was performed on subjects' averagepercentage corrected scores. The results of this MANCOVA showedsignificant main effects for subject group E(1,57) = 90.36, g <.001, MSe = 18233.71. Since the Bartlett-Box test indicated afailure to meet assumptions of homogeneity of variance on thefactor of explicit test type, Box's procedure for adjusting Fcritical was applied (Box, 1954). With Box's adjustment the maineffect of test type continued to be highly significant [i.e.,E(1,57) = 534.56, 12 < .001, MSe = 105139.41] together with asignificant interaction between group and test type, E(1,57) =60.57, 2 < .001, MSe = 11913.92.When the group by test interaction was analyzed with maineffect tests, significant differences between control and ADpatients' performance on both the picture recognition [E(1,57) =90.90, 12 < .001, MSe = 31148.15] and the picture fragmentcompletion test were found [E(1,58) = 9.36, p < .003, MSe =489.55; Refer to Table G-10b]. These findings did not supportthe hypothesis that there would be smaller differences betweenthe memory performance of AD patients and controls on theimplicit but not the explicit test of picture materials.Eta-squared analyses indicated that there had been a largercorrelation between subject performance on the explicit (eta-138squared = .61) than the implicit test (eta-squared = .14). Thesefindings in conjunction with those reported on differences inbaseline and floor effects suggest that the failure to findsupport for the hypothesis was due to the tests that were used.Specifically, that the picture fragment completion test was toodifficult for AD patients and the picture recognition test wastoo easy for controls. Together this produced a test environmentwhere there were large average corrected score performancedifferences between the AD patient and control group.Mildly, Moderately, and Severely Impaired AD Patients. AsFigure 4 shows, the average percentage corrected scoreperformance on the picture fragment completion test were similarfor mildly, moderately, and severely F.I. patients but thisperformance was at floor.In contrast, performance on the picture recognition test wasnot at floor and varied with F.I. group. Patients in the mildlyF.I. group showed, on average, much higher average percentagecorrected scores on this test than did the moderately andseverely impaired patients. (Refer to Table G-10a for a completelisting of these results).Tactile Recognition and Tactile Identification Tests Critical Measures. For the tactile recognition test thecritical dependent measure was the number of objects correctlyidentified in the target (n = 12) and the baseline conditions (n139= 12). For the tactile identification test the critical measureswere similar with the exception that the time required (inmilliseconds or ms) to identify stimuli were recorded.Average percentage corrected scores for the tactilerecognition test were computed using the same method employed forthe picture recognition test. For the tactile identificationtest timed responses were obtained for target items assigned toboth the tactile recognition and tactile identification testsenabling two different corrected scores to be computed. Acorrected score based on the re-identification of the sametargets (i.e., targets assigned to the study and test conditionsfor the tactile identification test; the old materials condition)and a corrected score based on targets assigned to the tactilerecognition test compared to targets assigned to the tactileidentification test (i.e., the new materials condition). Averagepercentage corrected scores were then computed for the old andnew materials condition based on a proportion of baselineperformance (i.e., the second presentation of targets assigned tothe tactile identification test acted as baseline for both thenew and old material conditions, refer to Table G-lla) which wasthen multiplied by 100.There were large variations in the timed responses ofsubjects on the tactile identification test. To decrease thisvariation a log transformation of the data was performed but this140resulted in the loss of subject data and mean square values ofzero in the MANCOVA analyses. Average median timed responseswere then computed as used in previous studies that have obtainedtimed response data from AD patients (see Knopman, 1991). Usingthe average median timed responses decreased some of thevariation in subject responses without the subsequent loss ofsubject data.AD Patients and Control Subjects. Figure 5 shows theaverage percentage corrected performance scores of AD patientsand control subjects on the tactile recognition and the tactileidentification test". As this figure shows, there were largedifferences in the performance of AD patients and controlsubjects on both memory tests. For the tactile recognition testAD patients recognized 49.17% (02 = 3.96) of the targets comparedto control subjects whose performance was at ceiling with 97.25%(ap = .94). This ceiling effect likely masked the true magnitudeof the differences in group performance.For the tactile identification test results depended onwhether priming was based on "old" or "new" stimulus materials(see Table G-lla). AD patients showed a similar averagepercentage corrected score to controls when performance was basedon the old material condition (i.e., if = 59.29%, 0 = 454.41 vs." Four AD patients were unable to complete the tactileidentification test and so their average performance scores wereremoved as per Howell, (1987).Control Subjects (n-40)All AD Patients (n-16)Mild (n-6)Moderate (n.5)Severe (n.5) 0cScy.000f■••••••••■■• 000300Control^All AD^Mild^Mod.^SeverePatients(n.40)^(n.16)^(n.6)^(n.5)^(n.5)Priming 1(Old Materials)Priming 2(New Materials)Explicit Test Implicit Test(Tactile Recognition) (Tactile Identification)Figure 5. Alzheimer patients' and Control subjects' performance on explicit and implicit tests tor objects. Percentage corrected scores areMildly. moderately and severely impaired patient groups' pertormance are shown separately.The error bars represent standard deviation scores.142N = 64.87%, SD = 103.04, respectively). In contrast, whenperformance was from the new materials condition control subjectsshowed much higher average percentage corrected scores than didAD patients (i.e., M = 73.96%, an = 84.80 vs. M = 13.19%, an =577.84).Additional statements must be made about these findings.The first is that despite computing average median timedresponses a large variation in subject responses remained.Similar findings have been reported by other investigators (e.g.,Nissen & Knoopman, 1987) and it suggests that these findings arehighly variable, especially in patients in severe levels of F.I.The large variances in these findings makes it impossible todetermine whether AD patients' performance was at floor orwhether significant differences existed between group baselineperformances.The first question that was addressed was whether theaverage percentage corrected performance of AD patients weresimilar to that of controls on the tactile identification testfor "old materials", despite lower performance on the tactilerecognition test. A repeated measures MANCOVA with Box'sadjustment detected only a significant group by test typeinteraction [f(1,53) = 9.56, R < .003, 4Se = 16976.41]. When thegroup by test interaction was analyzed with main effect testssignificant differences between the controls' and the AD143patients' average percentage corrected performance on the tactilerecognition [i.e., E(1,53) = 77.55, R < .001, MSe = 30880.21] butnot on the tactile identification test was found (see Table G-11b). Eta-squared analyses confirmed that the effect of testtype on subject performance was much larger for the tactilerecognition (.59) than on the tactile identification test (.007).To summarize, these results supported the hypothesis thatdifferences in the memory performance of AD patients and that ofcontrols would be smaller on the implicit test of "old materials"than on the explicit test. In contrast, much different resultswere found when tactile identification performance was based onthe "new materials" condition.A repeated measures MANCOVA of the average percentagecorrected performance scores of AD patients and controls on thetactile recognition and the tactile identification test of the"new materials" condition detected a significant main effect forsubject group [E(1,53) = 14.44, R < .001, MSe = 9878.21 and aBox-adjusted effect of test type [E(1,53) = 80.68, R < .001, MSe = 429.34. A significant interaction between group and test type[E(1,53) = 35.52, p < .001, MSe = 18851.02 was also found.When the group by test interaction was analyzed with maineffect tests significant differences between the controls' andthe AD patients' average percentage corrected performance on thetactile recognition [E(1,53) = 77.55, p < .001, MSe = 30880.21]144and the tactile identification test were found [i.e., E(1,54) =9.36, p < .003, MSe = 8470.10]. Eta-squared analyses indicatedthat the effect of test type on group performance had been muchlarger for the tactile recognition (.59) than on the tactileidentification test, but that the effect of the "new materials"condition had been much greater than for the "old materials"condition (i.e., .15 vs. .007, respectively).Mildly, Moderately, and Severely Impaired AD Patients.As Figure 5 shows, the performance of mildly, moderately, andseverely F.I. patients on the tactile identification test wasvery different to that found for the previous implicit tests.For example, the new material condition of the tactileidentification test was the only case where mildly F.I. patientsoutperformed moderately F.I. patients, who in turn outperformedthe severely F.I. group.Two main trends were found when comparing the performance ofthese patients on the new and old material conditions. The firstwas that the performance of mildly F.I. patients did not differacross the new and old material conditions. Thus, changingwhether the same or different objects were to be identified didnot affect mildly F.I. patients' performance. This was not thecase for moderately and severely F.I. patients whose averagepercentage corrected performance was much higher in the old thanin the new materials condition. In fact, severely F.I. patients145showed a negative performance in the new materials conditionsuggesting that this task was more difficult for them to performthan the old materials condition.Closer inspection reveals that the mildly F.I. patientsshowed a higher average percentage corrected performance than didcontrols in both the old and new material conditions. Thiscounter-intuitive result occurred because AD patients took longerto identify items than did controls and so there was a greatermargin for improvement in their performance. Similar resultshave been reported in other studies that have used timed taskswith AD patients and controls (cf. Eslinger & Damasio, 1986).Study TwoMemory Strategy Experiments The focus of this part of the investigation was to examinethe effects that different encoding strategies (or study tasks)and different retrieval cues (or test types) had on the explicitmemory performance of AD patients. This was addressed in threedifferent experiments where different study and test types wereused.Strict scoring criteria were applied to all memory strategyexperiments. For example, in the case of single word targets,changes in tense and plural forms were acceptable, but notsubstitute words with similar meaning. In the case of simplesentence commands, subjects' responses had to contain the correct146action verb and target nouns to be considered correct.Levels of Processing ExperimentThis experiment was designed to see if AD patients wouldremember more target words that they had studied on the basis ofmeaning than for the number of letters the word contained.Similarly, this experiment was designed to see if subjects wouldrecollect more targets in a recognition, rather than a freerecall test condition.Critical Measures. The critical scores for the encodecondition were the number of target words recollected from thesix words studied for their meaning and the six words studied forthe number of letters they contained. Previous work had shownthat using more than 12 stimuli would not result in a levelseffect (i.e., more meaning than letter encoded targets beingrecollected) in memory-impaired patients (Cermak & Reale, 1978).The critical scores for the retrieval condition were the numberof words freely recalled and recognized from the meaning andletter encode conditions.Subjects were engaged in the free recall of targets before arecognition phase for the remaining targets was begun. Thismeant that the total number of targets available to be recognizedwas dependent on the number that had not been freely recalled,and so a conditional or average percentage corrected score wascomputed. This score was obtained by subtracting the number of147targets that were recognized by the number of false alarms,dividing by the number of remaining targets, and multiplying by100. The average percentage corrected free recall score wasobtained by dividing the number of target words that were freelyrecalled by the number available for recall, and multiplying by100.AD Patients and Control Subjects. Figure 6 shows theaverage percentage corrected scores of the AD patients and thecontrol subjects on the free recall and recognition of items inthe levels of processing experiment. As this figure shows, ADpatients freely recalled, on average, 15.8% more targets from themeaning than from the letter study task condition (M = 27.5%versus X = 11.67%, respectively). Similar results can beobserved in the control subjects' performance. Controls freelyrecalled approximately 23.8% more targets from the meaning thanletter study task condition (i.e., M = 64.67% vs. M = 40.83%,respectively)".Similar trends were found for the recognition test. ADpatients recognized around 4.4% more targets from the meaningthan letter study task (i.e., X = 66.7% vs. fl = 62.26%).Similarly, control subjects recognized approximately 11.35% moretargets from the meaning than letter study task (i.e., X = 98.11%IS Standard deviations are not reported in this section sincethese are within group comparisons. Refer to relevant tables forthese values.Control(n-40)All^Mild^Mod.^SevereAD(n-20)^(n-6)^(n-7)^(n-7)r71Control^All^Mild^Mod.^SevereAD(n.40)^(n-20)^(n-6)^(n-7)^(n-7)Free Recall Recognition000coO00Encode Condition00:0)0co0co010000MeaningLetterrdpriiFigure 6. Percentage scores of Alzheimer patients (AD: all stages) and Control subjects on tree recall and recognition of itemsin the levels of processing expenment. The error bars represent standard deviation scores.149vs. M = 86.76%, respectively). It must be noted that controls'performance was at ceiling in the meaning encode condition and sothese results may not provide an accurate index of the magnitudeof the performance differences between the meaning and letter-encoded tasks.Figure 6 also shows that both AD patients and controlsrecognized more targets than they freely recalled. For example,AD patients had an average percentage corrected recognition scoreof 64.46% in comparison to 19.58% in the free recall testcondition (see Table G-12a). Control subjects showed a similareffect with an average percentage corrected recognition score of92.43% compared to 52.75% in the free recall condition.Finally, the results contained in Figure 6 indicated that,as expected, control subjects showed higher average percentagecorrected scores in all study task and test conditions than didAD patients.To examine the hypothesis that AD patients and controlswould recollect more target stimuli from the meaning than fromthe letter encode conditions, and from recognition than from thefree recall test conditions, a three factor repeated measuresMANCOVA (covariate factor = education) was performed. In thisanalysis the between-subjects factor was group (AD patients andcontrols) and the two within-subjects factors were study task(meaning, letter encode), and test type (recognition, free150recall). Results from this analysis indicated that there werefailures to meet multivariate assumptions of homogeneity ofvariance (i.e., Box M test = E(10,6902) = 7.98, R < .001). Whenthis result was further inspected with Bartlett-Box tests itappeared that these violations occurred on the recognition testfactor (i.e., meaning recognition f(1,7249) = 70.74, 112 < .001;letter recognition E(1,7249) = 8.05, 2 < .005). To correct forthese violations Box's procedure for adjusting the degrees offreedom for F critical was employed on factors involving test type(Box, 1954). Significant main effects of group E(1,57) = 39.68,< .001, MSe = 42093.20, study task, f(1,58) = 39.72, R < .001,MSe = 13546.87, and a Box adjusted test type factor, E(1,57) =165.54, R < .001, MSe = 103840.83 were found. No interactioneffects were significant (see Table G-12b).These results supported the hypothesis that AD patientswould have a higher average percentage corrected scoreperformance on the meaning than the letter study tasks and fromrecognition than free recall tests. A secondary and expectedfinding was that, although the meaning study task and recognitiontest condition resulted in a significant elevation of ADpatients' explicit memory performance, their performance wasnever raised to levels that were statistically similar tocontrols.151Mildly, Moderately. and Severely Impaired AD Patients.  Asecondary question was whether AD patients in different levels ofF.I. would also have a higher average percentage correctedperformance in the meaning than from the letter encode condition,and in the recognition than from the free recall test condition.Figure 6 shows this expected profile of results.These results can be summarized in three ways. Irrespectiveof the level of F.I., all AD patients had higher averagepercentage corrected scores: (1) in the meaning than in theletter encode condition and (2) in the recognition than in thefree recall test condition, and (3) there was a general trend formildly F.I. patients to outperform moderately and severely F.I.patients. With the exception of the recognition test moderatelyF.I. patients showed a higher average percentage correctedperformance than patients in severe levels of F.I.SPT-EPT ExperimentThis experiment was designed to see if AD patients wouldremember more minitasks that they had performed than theexperimenter had performed. These conditions are referred to asthe subject-performed task (i.e., SPT) and experimenter-performedtask conditions (i.e., EPT), respectively. Similarly, thisexperiment was designed to see if AD patients would recollectmore targets under cued than free recall conditions.152Critical Measures. The critical scores for the encodecondition were the number of minitasks recollected from the sixSPT and the six EPT items. This stimulus number was chosen basedon the results of a pilot study indicating the presence of floorand ceiling effects at smaller and larger numbers. The criticalscores for the retrieval condition were the number of SPT and EPTminitasks that were recollected with cueing in the free recalltest condition.Similar to the levels of processing experiment, subjectswere first engaged in the free recall of targets before the cuedrecall phase began. Thus, similar procedures to those used inthe previous experiment were employed to calculate averagepercentage corrected scores for the free and the cued recall testconditions. For example, the average percentage corrected freerecall score was the number of SPT and of EPT minitasks freelyrecalled, divided by 6, and multiplied by 100. The averagepercentage corrected cued recall score was the number of the SPTand of the EPT minitasks recollected with cues, divided by thenumber of remaining targets and multiplied by 100.X51 Patients and Control Subjects. Figure 7 shows theaverage percentage corrected scores of AD patients and controlsubjects on the free and cued recall of targets in the SPT-EPTmemory strategy experiment. As this figure indicates, ADpatients freely recalled the same number of average percentageEncode Condition8o0O0o -00300OSubject Performed (SPT)Experimenter Performed (EPT)VControl^AU^Mild^Mod.^SevereAD(n•40)^(n-20)^(n.6)^(n.7)^(n.7)^• Control^AllAD^( .40)^(n-20)Mild^Mod.^Severe(n.6)^(n•7)^(n.7)Free Recall Cued RecallFlours 7. Percentage scores of Alzheimer patients (AD: alt stages) and Control subjects on tree recall and cued recall of itemsM the subject performed memory strategy experiment (SPT:EPT). The error bars represent standard deviation scores.154corrected SPT and EPT encoded minitasks (i.e., 12.5% and 12.5%).In contrast, these patients recollected around 9.52% more SPTthan EPT corrected minitasks in the cued recall test condition(i.e., }1 = 56.19% vs. M = 46.67%, respectively).Two comments can be made about these results. The first isthat AD patients' performance was elevated by the provision ofadditional retrieval clues from the free to the cued recall testcondition for both the SPT and EPT encoded conditions. Thisconclusion was derived by comparing the average correctedperformance scores on the SPT and the EPT tasks in free and cuedrecall conditions (i.e., 12.5% and 12.5% versus 56.19% and46.67%, respectively). The second was that the cued recall testrevealed differences in the AD patients' abilities to recollectSPT versus EPT encoded tasks (i.e., M = 56.19% vs. If = 46.67%,respectively).Control subjects consistently recollected more correctedtargets from the SPT than the EPT study task condition,irrespective of the test condition. As Figure 7 shows, controlsrecollected around 9% more SPT than EPT encoded tasks in bothfree and cued recall test conditions (i.e., in free recall: 70.0%versus 61.67%; in cued recall: 79.44% versus 68.69%,respectively). Similar to AD patients, controls also benefitedfrom the provision of additional clues, albeit, not to the sameextent. Comparison of free and cued recall performance shows an155average elevation of around 8.0% in the average percentagecorrected scores of the cued versus the free recall testconditions (i.e., 74.06% versus 65.83%, respectively; see TableG-13a). In contrast, AD patients' explicit memory performancewas raised by approximately 39% in the cued versus the freerecall test condition (i.e., 51.43% versus 12.5%).To examine the hypothesis that AD patients and controlswould recollect more SPT than EPT encoded minitasks, and morecritical targets in the cued than in the free recall testcondition, a three factor repeated measures MANCOVA wasperformed. In this analysis the between-subjects factor wassubjects (AD patients, controls) and the two within-subjectsfactors were study task (SPT, EPT encode), and test type (i.e.,cued and free recall). Results from this analysis shows thatmultivariate and univariate assumptions of homogeneity ofvariance were met. A significant main effect for group, E(1,57)= 58.79, R < .001, MSe = 79227.64, study task, E(1,58) = 8.34, R< .005, MSe = 3325.02, and test type, E(1,58) = 70.80, p < .001,MSe = 38940.02 were found. Unlike the levels of processingexperiment, a significant group by test type interactionoccurred, E(1,58) = 15.92, R < .001, MSe = 8755.21.When this group by test type interaction was graphed itshowed that AD patients' free recall of the same percentage ofSPT and EPT encoded targets accounted this effect. This156conclusion was confirmed by the results of main effect tests thatshowed that AD patients and controls recollected significantlymore SPT than EPT encoded targets in the cued recall testcondition; f(1,19) = 63.15, p < .001, MSe = 31733.89 and E(1,39)= 14.09, p < .001, Hag = 8075.07, respectively. In contrast,main effect tests showed that only controls recollected more SPTthan EPT average percentage corrected targets in the free recalltest condition fE(1,39) = 9.01, p < .005, Nag = 4375.07).Main effect tests also revealed that AD patients andcontrols each recollected more average percentage correctedtargets in the cued than in the free recall condition. Forexample, AD patients and controls recollected more SPTs in cued,rather than free recall conditions; E(1,19) = 49.19, p < .001,14Se = 20400.28 and E(I,39) = 12.47, p < .001, Mae = 5335.56,respectively. AD patients also recollected significantly moreEPTs in cued versus free recall conditions, E(I,19) = 22.09, p <.001, MSe = 11902.50. This was not the case for controls and isdue to the fact that there was a smaller proportion of items thatwere left to be recollected in the cued recall condition (seeTable G-13b).To summarize, these results show that AD patientsrecollected significantly more targets in cued than free recalltest conditions. These results also show that although ADpatients recollected significantly more SPTs than EPTs in the157cued recall test condition, this was not the case for the freerecall test. Thus, the hypothesis that AD patients wouldremember more SPTs than EPTs and more items in the cued than inthe free recall test conditions was only partially supported.This hypothesis was also only partially supported for controls.Mildly. Moderately. and Severely Impaired AD Patients.A secondary question was whether AD patients in different levelsof F.I. would show similar trends to recollect more targets fromSPT than EPT, and in the cued versus the free recall testcondition. Figure 7 shows that patients in all stages of F.I.recollected more corrected targets in the cued than in the freerecall test condition. However, the SPT encode conditionproduced selective effects. Only mildly and moderately F.I.patients recollected more SPT than EPT encoded targets in thecued recall test condition. Since moderately and severely F.I.patients' performance in the free recall test was at floor littlesignificance can be placed in the fact that SPT performance wasslightly higher than EPT encoded targets in this test. (Refer toTable G-13a for a complete listing of these results).Multisensory ExperimentThis experiment was designed to see if AD patients andcontrol subjects would remember more sentence targets that theyread aloud to the experimenter and then performed (the See, Say,& Dca condition) than they read aloud to the experimenter (the See158& Say condition) or read silently to themselves (the See onlycondition). Similarly, this experiment was designed to see ifsubjects would recollect more targets in the cued than in thefree recall test condition.Critical Measures. The critical score was the number ofsentences recollected from (1) DQ, (2) Say, and (3) See conditions. There were four target sentences per encode taskcondition. As in the previous experiment, subjects were asked tofreely recall critical targets before the cued recall test began.Average percentage corrected scores were computed using the samemethod described in earlier sections with the exception thatthere were four targets per study task condition.AD Patients and Control Subjects. Figure 8 shows theaverage percentage corrected scores of AD patients and controlsubjects on the free and the cued recall tests of Do, Say, andSee encoded targets. As Figure 8 shows, AD patients recollectedmore Do encoded, than Say or See encoded targets, in both freeand cued recall conditions. For example, AD patients freelyrecalled 16.25% Do, 6.25% Say, and 5% See encoded targetscompared to 44.77% Do, 12% Say, and 11.84% See encoded targets inthe cued recall test condition. As these results show, there waslittle difference in AD patients' recollection of Say and Seeencoded targets, irrespective of the test condition. ComparisontiKFree Recall Cued RecallEncode ConditionFigure 8. Percentage corrected score of Alzheimer patients (AD: all stages) and Control subtects on free recall and cued recallof items n the multIsensory strategy experiment. The error bars represent standard deviation scores8Severe Control(ri."7)03300200Control^All^Mild^Mod.AD(n.40)^(n-20)^(n-6)^(n.7)All^Mild^Mod.^SevereAD(n-40)^(n-20)^(n.6)^(n.7)^(n-7)See. Say and DoSee and SaySee160of AD patients' cued versus free recall test performance showsthat cueing elevated the recollection of Do, but not of Say orSee encoded targets. AD patients recollected 44.77% of Doencoded targets in the cued recall and 16.25% in the free recalltest condition -- a difference of around 28.5%. In contrast, ADpatients' performance on Say and See encoded targets was at floorin the free recall test condition. That is, performance was6.25% and 5%, respectively and only slightly raised in the cuedrecall test condition (i.e., around 12% for both conditions).Similar trends to those described for AD patients were foundin the controls' performance in this experiment, with twoexceptions. Controls' performance was never at floor and therewas a similar performance in the Do, Say, and See encodedconditions in both the free and cued recall test conditions.Controls did not show the same elevation in performance as did ADpatients in the cued recall test condition because their highperformance in the free recall condition meant that there werefewer remaining items to be recollected in the cued recall testcondition. For example, controls recollected 72.5% versus 68% ofthe Do encoded targets in the free versus cued recall testconditions, 42% versus 36% in say encoded targets, and 34.5%versus 41% in see encoded conditions (see Table G-14a).To examine the hypothesis that subjects would recollect more(a) Do, than Say or See encoded targets, and (b) more targets in161cued rather than free recall test conditions, a three factorMANCOVA was performed. In this analysis the between-subjectsfactor was subjects (AD patients and controls) and the twowithin-subjects factors were study task with three levels (i.e.,Do, Say, and See), and test type with two levels (i.e., free andcued recall).This analysis indicated that there were failures to meetassumptions of homogeneity of variance and Bartlett-Box testsshowed that these violations occurred on the See free recall taskfactor (i.e., E(1,7249) = 11.67, R < .001). To correct for theseviolations Box's procedure for adjusting for degrees of freedomfor F criticia was employed on factors involving the study task andtest factor (Box, 1954; Howell, 1987). A significant main effectof group, E(1,57) = 47.26, p < .001, Hag = 91084.47, study task,E(1,115) = 28.46, p < .001, MSe = 24390.05, and test type E(1,57)= 9.41, p < .003, MSe = 8336.81. No interaction effects weresignificant. Since the performance of controls and AD patientswere similar for the See and Say encode conditions post hoc testsdid not need to be performed to determine where the significantdifferences in study task performance had occurred. Visualinspection of the average percentage corrected scores showed thatmore Do than Say or See encoded targets were recollected in bothfree and cued recall test conditions. Thus, it appeared thatrequesting subjects to see, say and perform simple sentences162produced a significant elevation in explicit test performancethan if subjects read the sentence to themselves or silently tothe experimenter.The hypothesis in this experiment was only partiallysupported in that subjects did not recollect significantly moreitems in the See than Say conditions.Mildly, Moderately, and Severely Impaired Patients.  Asecondary question was whether AD patients in different levels ofF.I. would recollect more corrected targets from Do, than Say orSee conditions and from cued than free recall tests. Figure 8shows that there was a consistent trend for all patients torecollect more Do than Say or See encoded targets. Whether moreSay than See encoded corrected targets were recollected wasdependent on the patients' level of F.I. For example, whereasmildly F.I. patients recollected more say than see encodedaverage percentage corrected targets in both cued and free recalltest conditions, this was not the case for moderately andseverely F.I. patients. Both moderately and severely F.I.patients showed Do, Say, and See performance that was at floor inthe free recall test. In the cued recall test condition themoderate F.I. group's performance in the Say condition was raisedfrom floor while the performance of the severe F.I. groupremained at floor. Thus, as in previous experiments there wasevidence of a slide-rule type influence in the effects thatencoding and test conditions had on the memory performance ofpatients in different levels of F.I. (Refer to Table G-14a).163164CHAPTER 7: DISCUSSIONThis chapter has six sections. The first reviews thegeneral research objectives and motivation for thisinvestigation. The second section reviews the hypothesis andmain findings from part one, how they relate to previous work,and the new things they may be telling us about implicit memoryin AD patients. The third section contains the same informationas it pertains to part two of the investigation. In the nextsection I consider some of the factors that limit thegeneralizability of the results from parts one and two. Thefifth section considers future research, and the last sectionsummarizes the general and specific contributions of this work.Research Objectives and MotivationThe main goal of this investigation was to re-explore theimplicit memory abilities and the effects of different memorystrategies on the explicit memory performance of AD patients. Toaddress the first part of this goal I examined AD patients' andcontrols' memory for different types of materials with implicitand explicit tests that differed in the instructions provided atretrieval. The second part of this goal was addressed with threememory strategy experiments that had different encoding andretrieval conditions.165The first part of this investigation was motivated by mixedreports regarding whether implicit memory is spared or impairedin AD patients. Variations in these findings may have beencaused by using tasks that differed in their ability to guidethese attentionally-impaired patients during the encoding ofcritical targets. The motivation behind the second part of thisinvestigation was similar to that of the first. A limited numberof studies have shown that the explicit memory performance of ADpatients can be elevated in some, but not in other conditions.The main premise of the second study was that strategies thatreduced the attentional demands of tasks at study and atretrieval would elevate the explicit memory performance of ADpatients.Part OneThe overall hypothesis of this study was that differences inthe memory performance of AD patients and controls would besmaller on implicit, than explicit tests, using written andspoken word, picture, and object materials. This hypothesis wasguided by two main ideas: First, that the neuropathology of ADbegins in hippocampal regions before advancing to brain areaswhich have been ascribed roles in implicit test performance.Thus, a greater magnitude of impairment on explicit than onimplicit tests should be observed in most AD patients. Second,that when an encoding method that compensated for AD patients'166attentional impairments was used, their implicit test performancewould be raised to levels that were similar to that of controls.Results from part one indicated that the explicit testperformance of AD patients was significantly lower than that ofcontrols on all tests, irrespective of the materials that wereused. This finding was expected since the hippocampus, whichplays a role in the explicit recollection of multimodalinformation appears to be damaged early in AD (Ball, 1977; Barr &Kiernan, 1988). In contrast, whether AD patients' implicit testperformance was found to be impaired depended on the materialsthat were used.Implicit Test ResultsAD patients' performance on category completion testsappeared not to be significantly different from that of controls.This test required that subjects provide exemplars to categories,and thus tapped a semantic type of knowledge similar to the wordassociation test used by Salmon et al. (1988). However, incontrast to the present study, Salmon et al. did not find thatpriming was spared in their AD patients.Several reasons can be offered for the different findings ofthe present experiment. One is that semantic priming was foundto be spared in AD patients because an encoding method thatrequired that patients identify and generate the meaning ofcritical targets (instead of a pleasantness rating as in the167Salmon study) was used. The premise here is that when a targetis not completely encoded its subsequent recollection is impeded(Graf, Tuokko, & Gallie, 1990). In the case of AD patients thepremise was that their inability to spontaneously encode targets(cf. Rohling, Ellis, & Scogin, 1991) would impede theirperformance, making them appear to be more impaired on implicittests than they really were. However, when an encoding methodthat guided AD patients during encoding was used, their memoryperformance would be elevated.There are at least two reasons why the meaning encodecondition could not be the sole factor responsible for findingthat priming for written and spoken words was spared in ADpatients. One reason is that this explanation would imply thatcortical association areas are not required since patients withdamage to this brain area (Brun, 1983) could perform theseimplicit tests. Since Positron Emission Tomographyinvestigations have shown that cortical association areas areactivated when controls perform wordstem completion tests (cf.Squire et al., 1992b) this makes it unlikely that requiring ADpatients to encode targets for their meaning is the sole reasonfor this study's findings. Second, AD patients' performance onthe implicit tests used in this study was not always found to besimilar to that of controls.168An alternative reason that accounts for encoding method,neuroanatomical substrates, and the fact that priming varied withthe test that was used is the following. Although implicit testsmay index similar types of memory (i.e., semantic as in the caseof category completion and word association), different cognitiveprocesses may be required to perform these tests. Whetherpriming is found to be spared or impaired in AD patients may beinfluenced by the particular combination of cognitive processesrequired to perform the specific test, as well as on the type ofmaterials that are used (i.e., pictures, words).Evidence that the word association test employed by Salmonet al. and the category completion tests used in thisinvestigation may have tapped different cognitive processes comesfrom a consideration of baseline performance. Salmon et al.(1988) found that the baseline performance of their AD patientsand controls was not statistically different. That is, both ADpatients and controls were able to provide a similar number ofwords to match previously unstudied word pairs. In contrast,this was not the case for the category completion tests. ADpatients provided significantly fewer exemplars to unstudiedcategories than did controls, for both written and spoken wordconditions.In combination, the above findings suggest that the taskrequirements of the word association test may have been easier169for the AD patients to achieve than those of the categorycompletion test. Thus, while both of these tests tap a semantictype of word knowledge, they may have done this via differentcognitive processes. Further support for the idea that ADpatients' performance on implicit tests may be influenced byfactors other than the stimulus materials that are used (i.e.,pictures or words) comes from a consideration of subjects'performance on the other tests in this investigation.Comparison of the performance of AD patients and that ofcontrols on the picture fragment completion test suggests thatpicture completeness, rather than picture materials per se,influenced this study's findings. Consideration of the baselineperformance of AD patients and controls suggest that whereas bothgroups were able to correctly identify a similar number ofcompleted picture stimuli, this was not the case when incompletedor fragmented pictures were used. AD patients requiredsignificantly more information (i.e., stimuli that were in a morecompleted form) than controls did to identify picture stimuli.Other investigators have also found that AD patientsexperience a difficulty in processing fragmented pictures. Forinstance, Kirshner, Webb, and Kelly (1984) found that AD patientswere able to identify completed, but not fragmented versions ofthe same items. Furthermore, AD patients may be unique in theirinability to identify fragmented pictures. As evidence, Bondi170and Kaszniak (1991) examined the average fragment level in whichAD, PD, and control subjects were able to identify a series of540 pictures. They found that AD patients required asignificantly more completed picture than PD patients andcontrols did to identify the same targets. Closer inspection ofthe stimuli contained in Appendix D reveals a likely reason forthese findings -- fragmented pictures are more abstract andvisuospatial in form than the completed version of the samepicture.Since problems in visuospatial and abstract reasoningcharacterize AD patients (Reisberg, 1983) the fragmentedcondition of these pictures rather than "an inability to activatepre-existing picture representations" (Heindel et al., 1990) maybe the reason that implicit memory for pictures was found to beimpaired in the AD patients in this, and in other investigations(cf. Bondi & Kaszniak, 1991; Heindel et al., 1990). A recentstudy supports this conclusion. When completed pictures wereemployed the implicit test performance of AD patients was notfound to be different from that of controls (Gabrieli,unpublished).Additional evidence that stimulus characteristics caninfluence implicit test performance comes from the tactileidentification test. When performance on this test was based onthe re-identification of the same objects (i.e., old materials171condition) AD patients showed a magnitude of priming that did notappear significantly different from controls. In contrast, whenthe identification of different objects was required (i.e., thenew materials condition) the same AD patients' performance wasnow found to be impaired. Since the same encoding method, test,and type of stimuli were present in both test conditions, thereason(s) for these performance differences is not clear. Onepossibility is that the re-identification of the same items madethe old materials condition an easier task than the new materialscondition; this idea is supported by the moderately and severelyF.I. patients' performance.Both the moderately and the severely F.I. patients showedbetter performance in the old versus the new materials condition.One interpretation of these findings is that the cognitiveprocesses required to perform the old materials condition wasmore attainable by these patients than those for the newmaterials condition. In contrast, there was no difference in themildly F.I. patients' performance in either the new or the oldmaterials condition, presumably because the cognitive processesrequired to perform both of these tasks did not exceed thesepatients' abilities"." The assumption made here is that the neuropathology of ADis less advanced in mildly, than in moderately or severely F.I.patients. As a result of this more limited pathology, the mildlyF.I. patients were able to meet the cognitive processing demandsof both versions of the tactile identification test.172Neuroanatomical Considerations What implications do the results of this study have forcurrent ideas regarding the neuroanatomical areas responsible forpriming? Any attempt to address this question is speculativegiven the absence of pathological data regarding the patients inthis investigation. However, if one is willing to accept that aconsistent pattern of neuropathology is caused by AD (cf. Lewiset al., 1987; Price et al., 1991) some reflection on thisquestion is possible.Several sources have provided evidence that corticalassociation areas may be integral to performing implicit tests oflexical, semantic, and picture materials (cf. Butters et al.,1990; Heindel et al., 1990; Squire et al., 1992b; Tulving &Schacter, 1990). In the case of AD, cortical association areasare eventually destroyed (Brun, 1983). Thus, to find that ADpatients' performance on category completion tests was of amagnitude that was not statistically different from that ofcontrols seems incongruous, until one considers the pattern ofdamage that occurs to cortical association areas. That is, theprimary and secondary cortical association areas of AD patientsappear to incur less pathology than the tertiary corticalassociation areas (cf. Esiri et al., 1986; Lewis et al., 1987;Van Hoesen & Damasio, 1987). Thus, the possibility exists thatwhen the appropriate level of cognitive demands is made (i.e., by173the combination of test type, encoding method, and stimuli thatare used) the primary and secondary cortical association areas ofsome AD patients may be able to support a magnitude of primingthat is similar to that achievable by controls (L.R. Squire,personal communication, April, 1993).It is also possible to be more specific regarding theinfluence that encoding method and stimuli characteristics mayhave had on neural substrates. With respect to encoding method,studies have shown that maximal activation of the nervous systemoccurs when subjects generate the meaning, rather than provide apleasantness or rhymed response to stimuli (Cohen & Waters,1985). Thus, when AD patients generated the meaning of criticaltargets, this presumably elevated the level of neural activity inbrain areas including those responsible for attentional focusing(i.e., locus coeruleus, nucleus basalis of Meynert, etc.). Under"normal"" circumstances these brain areas may not have beenactivated to a level capable of supporting the mental processesrequired to perform various implicit tests. Support for thisproposal comes from the Partridge et al. (1990) study which foundintact priming in AD patients when a meaning encode method wasused, in contrast to other investigators using the same test anda pleasantness encode task (cf. Shimamura et al., 1987; Salmon et" Where "normal" refers to the absence of specific encodinginstructions.174al., 1988; Heindel et al., 1990).Whereas a meaning encoding method may have served to changethe pattern of neural activity in AD patients, the same may betrue of the stimuli that was used. A completed picture may havemade less processing demands on the AD brain than the samepicture in its fragmented form. While the mechanisms by whichthis difference in cognitive processes may have been mediated hasyet to be determined, one possibility is the number of brainareas that were involved. That is, intact pictures may primarilytap the neural processing of areas such as the parietal,occipital, and temporal lobes whereas fragmented pictures might,in addition, make heavy demands on areas such as the frontallobes (Lezak, 1983). In the case of the AD brain whereneuropathology occurs in all these areas, the mental processesrequired to identify the fragmented (in contrast to thecompleted) picture stimuli may have surpassed the AD patients'neuronal processing abilities.Additionally, the progressive loss of neurons that occurs inAD patients' brains would presumably correspond with a decline inneural processing capabilities. This event presumably accountsfor the fact that patients in mild levels of F.I. (i.e., in earlystages of neuronal loss) outperformed those patients in laterstages of F.I.175Alternatively, principles of Transfer Appropriate Processingmust also be considered. Specifically, the stimulus form thatwas presented at study was not identical to that presented atretrieval in the picture fragment completion and in the newmaterials condition of the tactile identification test. Thissuggests that test performance might have been lowered on thesetests since, under the theoretical framework of TAP, an exactrecapitulation of the cognitive operations engaged at study andat test would not have occurred. This event may have beentranslated into more neural work for the AD brain in contrast toif the same stimulus form had been employed at study and at test.Positron Emission Tomography that have found that less neuralactivity is required to re-process identical (vs. non-identical)stimuli supports this idea (cf. Squire, 1992a; Squire, et al.,1992b).To review, the findings from this study have beeninterpreted from the view that implicit tests can vary in thecognitive processing (and thus the neural processing) demandsthey place on the AD patient. While this perspective is able toaccount for the discrepant findings regarding if implicit memoryis retained in AD patients (cf. Keane et al., 1991; Partridge etal., 1990) additional work on this issue is required.Specifically, a systematic investigation of the influence thatencoding and retrieval methods, as well as stimulus format has on176the test performance and brain activity of AD patients isrequired to determine the validity of my interpretations sinceother explanations are possible.Additional PossibilitiesThere are of course additional factors that could explainthe findings obtained in this study. One is that the AD patientsin this investigation were somehow different from those that havebeen used in other studies. For example, there may have been adifferent pattern and/or a less advanced degree of neuropathologypresent in this study's patients that could account for thepresent study's finding of spared priming in AD patients. Thisquestion of patient similarity pervades all AD investigations (L.R. Squire, personal communication, April, 1993) and is not aneasy issue to address. However, while information on theneuropathology of patients is not available to most researchersan indirect index via level of functional impairment isobtainable.The inclusion of mildly, moderately, and severely F.I.patients suggests that the average level of impairment in thepatients used in this study may have been greater than in otherinvestigations where mildly and moderately impaired patients havebeen used. As a result, had the findings from this study beensolely attributable to a unique patient sample, then primingshould have been found to be impaired rather than spared.177Additional explanations for this study's findings alsoexist. For example, the implicit tests used in this study mayhave not been sensitive to differences in control and patientperformance and this could account for finding preserved primingin AD patients. In the case of the picture fragment completiontest, AD patients' priming performance was at floor and thismakes it impossible to make conclusive statements regarding thistest result. Further, as is the case with any timed measure ofpriming, the large variability in subject performance on thetactile identification test may have masked the true magnitude ofthe differences between the controls and the AD patients. Thus,the findings from this study will need to be replicated before itwill be possible to determine whether the semantic network. of ADpatients is lost, or whether impaired priming reflects aninability to enter and access this network due to memory and non-memory related factors (i.e., attention).Part Two The overall hypothesis of this study was guided by the ideathat the attentional deficits encountered by AD patients caninfluence their explicit memory performance. Thus, it wasproposed that strategies that changed the way that targets wereencoded and retrieved (i.e., by presumably lowering theattentional processing demands that were required) should resultin an elevation of the explicit memory performance of AD178patients.The above hypothesis was tested in three differentexperiments. The first experiment used a levels of processingframework where target words were either studied for theirmeaning or for the number of letters they contained (i.e., thenon-semantic condition). Existing theories suggest that encodingtargets for their meaning (vs. non-meaning) might serve to lowerthe attentional processing demands required to encode targets(Cohen, Sandler, & Schroeder, 1987). The second experimentemployed a Subject Performed-Experimenter Performed (i.e., SPT-EPT) strategy where mini-tasks were either performed by thesubject or by the experimenter. Existing theories suggest thatperforming a task (in contrast to watching it being performed)engages multimodal processing that serves to lower theattentional demands of the task (Backman & Nilsson, 1985; Cohen,1985; Engelkamp & Zimmer, 1985). The third or muitisensoryexperiment was new. More multimodal processing was assumed tooccur when subjects read aloud and performed simple sentences(i.e., the DO condition) than when they either read aloud (i.e.,the Say condition) or silently read a target (i.e., the Seecondition). Thus, it was presumed that AD patients wouldrecollect more DO than Say or See encoded targets since the DOcondition should lower the attentional demands at encoding morethan the latter conditions did.179For the levels of processing experiment a recognition and afree recall test served as the retrieval conditions. For theSPT-EPT and multisensory experiments a cued recall and a freerecall test served as the two retrieval or test strategyconditions. Previous work suggested that recognition and cuedrecall conditions provided more specific clues to re-locatingtargets than did a free recall condition (Perlmutter, 1978). Inessence, the recognition and cued recall conditions could beviewed as lowering the attentional demands of the memory taskmore than did a free recall test. Thus, it was assumed that ADpatients would retrieve more targets in the recognition and cuedrecall conditions than when free recall test strategies wereused.Levels of processing experiment. Results from this part ofstudy two extended previous research by showing that AD patientscould recollect a larger number of targets that they had encodedon the basis of their meaning (vs. non-meaning) and in arecognition (vs. free recall) test condition. The overall memoryperformance of AD patients was also found to be highest when acombination of meaning encode and recognition conditions had beenemployed.Previous investigators have not been able to find evidencethat AD patients would recollect more targets that had beenstudied for their meaning (i.e., a levels effect; cf. Corkin,1801982) or to find a significant change in performance whendifferent tests were used (i.e., in a free versus cued recalltest condition; cf. Martin et al., 1985). Several factors couldaccount for the reason this study found that the explicit memoryperformance of AD patients was significantly elevated when ameaning encode and a recognition strategy was employed.Variations in the encoding method used in this study andthat used by Corkin (1982) might account for some of thedifferences found in the present experiment. That is, it islikely that Corkin was unable to find a levels effect (i.e., therecollection of more meaning than non-meaning encoded targets) inher AD patients because, in effect, there was no levels conditionin her study. To illustrate, the encoding method used by Corkinrequired that subjects provide a yes/no response to questionssuch as "Is the word a type of fruit ?" A similar situation tothat previously discussed for implicit tests is likely to havebeen present. That is, since AD patients are unable to engage inthe spontaneous elaboration of targets (cf. Rohling et al., 1991)requiring them to simply provide a yes/no response to a sentencessuch as "Is the word a type of fruit ?" would not have ensuredthat they had thought about the target's meaning. That is,patients might have responded yes or no without contemplating thetargets' meaning. In contrast, in the present study subjectswere required to generate the meaning of critical targets which181confirmed that AD patients had thought about the meaning oftargets.In comparison, Martin et al. (1985) who used a similarencoding method to that used in the present experiment and alsofound a levels effect, did not find a significant test effect.That is, the explicit memory performance of AD patients did notchange when a free and a cued recall test condition was used.This differs from what was found in this study where AD patientsrecollected significantly more targets in a recognition than in afree recall test condition. A comparison of the tests that wereused in the Martin et al. versus the present study provides onepossibility for the discrepant findings.Martin et al. required that AD patients engage in the freerecall of items before they were cued for the correct response.In this experiment AD patients were engaged in the free recall ofitems before they were asked to recognize the target presented atstudy. The recognition condition would have provided different"clues" about the stimulus than would a cued recall test. Toillustrate, presenting the stimulus that was shown at study wouldmake the process of re-locating a mental representation of theevent easier simply because a matching of the study to teststimulus would be taking place. In contrast, a cue such as "Whattype of fruit were you shown ?" would not be a clue thatpresumably matched the mental representation of the study target.182This latter condition would therefore theoretically require morecognitive effort or attentional processing, and thus be a moredifficult task for a cognitively impaired subject to perform. Inthe case of a patient experiencing serious retrievaldifficulties, the clues provided by a recognition versus a cuedrecall test might make the difference in whether the target wassuccessfully recollected.SPT EPT experiment. ResUlts from this experiment indicatedthat AD patients recollected more minitasks that they performed(i.e., SPTs) than the experimenter had performed (i.e., EPTs) ina cued recall test condition. However, AD patients freelyrecalled the same number of SPTs and EPTs. Closer inspection ofthe results presented two reasons for this latter finding.First, although both the moderately and the severely F.I.patients freely recalled more SPT than EPT encoded targets, theirperformance was at floor. These floor effects may have maskedthe magnitude of the SPT encoding effect. Second, inspection ofthe mildly F.I. patients' free recall performance shows thatthese patients recalled more EPTs than SPTs. While the reasonfor this latter finding could be attributable to many factors,including low subject numbers, it is possible that watching theexperimenter perform the tasks was a somehow unique or unusualevent. Since the mildly F.I. patients were presumably lesscognitively impaired than the moderately and severely F.I.183patients the "uniqueness" of the EPT tasks may have, in effect,produced a more memorable event that had significance only forthe mildly F.I. group.The present experiment confirmed the findings reported inthe only other investigation of this type. In this study Dick etal. (1989) also found that there was no difference in the numberof SPTs and EPTs that AD patients recollected in a free recallcondition. Thus, it appears that any memory benefits that ADpatients might obtain by performing a task may only be realizedif they are also provided with recognition or cued recall tests.That is, when compared to the presumed attentional processinglowering capabilities of a meaning encode condition, the SPTencode strategy may not be as effective. However, when the SPTand the cued recall conditions were combined, the explicit memoryperformance of some AD patients (i.e., those in mild and moderatestages of F.I.) was maximally raised. This finding appears tocorrespond with previous research in which we have found thatimpairments in retrieval are present in all AD patients, butproblems at encoding may only typify those in later stages ofF.I. (Tuokko, Gallie, & Crockett, 1990). Thus, in order toconsistently raise the explicit memory performance of any ADpatient a retrieval strategy may be required. However, thecombined use of an effective encoding and retrieval strategyshould produce the greatest elevation in the explicit memory184performance of this patient group.Multisensory experiment. The results obtained from thisexperiment indicated that the AD patients in this studyrecollected more targets that they had encoded in the DO than inthe Say or See condition. Also evident was the fact that the ADpatients' overall performance in the Say and See encodeconditions was not that different. Thus, seeing, saying, andperforming the target seems to have enhanced explicit memoryperformance more than if AD patients simply read the target. Asin the previous experiment, when retrieval in a cued rather thana free recall format was required, AD patients' explicit memoryperformance was maximally elevated.Implications The results from the three experiments contained in studytwo have provided some consistent findings. First, they haveshown that certain encoding and retrieval strategies cansignificantly elevate AD patients' explicit memory performance.That is, cued recall and recognition tests appear to be moreeffective in elevating explicit memory performance than freerecall conditions. Furthermore, the meaning and motorperformance encoding strategies may be more effective than non-meaning and non-motor performance events.While the exact processes by which the above encoding andretrieval strategies work is not known, one proposal has been185that they work by lowering the attentional or cognitiveprocessing requirements of the task (Backman & Nilsson, 1985;Cohen et al., 1987; Perlmutter, 1978). For the SPT and DO encodetasks,.the combination of multimodal processing and the fact thatthe brain areas associated with motor performance are relativelyspared in AD patients may be some of the factors responsible forthese memory-enhancing effects. In contrast, since the non-meaning, See, Say, and free recall conditions were not aseffective in elevating memory performance they presumably did notinfluence cognitive processing in the same way as the formertasks.A second finding that was consistent across the threeexperiments was that the combined use of specific encoding andretrieval strategies resulted in a maximal elevation of theexplicit memory performance of AD patients. However, bythemselves it was the cued and recognition retrieval strategiesthat were more effective than the encoding strategies in raisingthe explicit memory performance of AD patients. This findingappears to correspond to an issue that has been previouslydiscussed. That is, retrieval deficits may occur early, whereasencoding difficulties may arise in later stages of AD (Tuokko,Gallie, & Crockett, 1990). Thus it makes sense that, used bythemselves, the recognition and cued recall retrieval strategieswill always be more effective in elevating the explicit memory of186AD patients than the sole use of a meaning or performance-basedencoding strategy. While the exact reasons for this relationshipis unknown it might correspond to the neuropathologicaldevelopment of AD. That is, since AD begins in the hippocampus(which is integral for the intentional retrieval of information)the sole use of either a recognition or cued recall retrievalstrategy will always be more effective than either a meaning orperformance-based encoding strategy. In turn, since theneuropathology of AD eventually progresses to include brain areasassociated with various attention-related processes (Fedio etal., 1992; Mountcastle, 1978), the use of either a meaning orperformance-based encoding strategy should also be effective inraising the explicit memory performance of some AD patients(especially in combination with certain retrieval tests). Inaddition, the fact that neuronal loss increases as AD progressesmay be one reason that the benefits derived from these strategiesdeclined as the level of patient F.I. changed from mild to severestages.To summarize, together the results from studies one and twosuggest that the memory abilities of AD patients may be betterpreserved than previous research has indicated. Continuedinvestigation of the effects that different cognitive processeshave on the memory performance of AD patients will be required tocontinue the quest for a more accurate brain - memory model of187this disease.Limitations of this WorkThe environment one works in and the subjects one chooses tostudy provides every researcher with a unique spectrum ofchallenges. Working with AD patients is certainly no exceptionand many of the limitations present in this study are a directreflection of the characteristics of this patient group. I willselectively discuss the limitations of this work in terms ofthree main areas: (1) the method that was used, (2) subjectcharacteristics, and (3) statistical considerations.Limitations Attributable to Method Two aspects of the method limit my findings: i) the numberof stimuli that were used, and ii) the order of test andexperiment presentation. With respect to stimuli, 12 targetswere employed in each test and experiment and this limits thegeneralizability of the results. The main reason for employing12 items was that pilot work had indicated that floor effectsoccurred when more targets were included. Multiple blockstimulus presentations could have been employed to resolve thisproblem but wasn't for the following reasons. The focus of thisinvestigation was exploratory and so I wanted to use a largenumber of tests and experiments with the AD patients. The largenumber of tests in combination with the high fatigability of theAD patients meant that they would not have completed the188investigation if several blocks of stimulus presentations hadbeen used.Secondly, a standard order of test and experimentpresentation was followed and thus, there is the possibility thatorder effects are present in the data. A randomized test orderwas not employed for reasons similar to that already discussed.Pilot work had indicated that AD patients could not complete thetests and experiments unless they were staggered in analternating easy/difficult order. However, since mostneuropsychological investigations include a standard order oftest presentation the method used in this study is consistentwith current research practice.Subject Characteristics All researchers encounter a problem in recruiting largenumbers of AD patients for two main reasons. First, althoughmany patients visit Alzheimer clinics only a small proportionreceive a diagnosis of AD, and so there are few patients toselect from. Second, once diagnosed, AD patients are reluctantto participate in research that requires that they engage intasks that they will have difficulty performing. For these mainreasons research in this area have been based on small numbers ofAD patients. While there have been few studies that haveemployed more than 20 patients (the sample size employed in thisstudy) this still remains a small number of subjects in terms of189the statistical power it might provide (cf. Cohen, 1988). Thus,the possibility is that there was insufficient power toaccurately detect performance differences that might haveoccurred. However, since the size of the effect that was beingindexed (i.e., memory performance) was large, it is unlikely thatlow subject numbers unduly limited the ability to detectperformance differences when they existed.A second source of error is the possibility that some of thepatients did not have Alzheimer's Disease. Since a definitediagnosis of AD is made at autopsy (cf. McKhann et al., 1984) andthe patients in this study are still alive, it has not beenpossible to check the accuracy of their diagnosis. Thus, thepossibility exists that the memory deficits observed in patientscould be attributable to brain areas not associated with AD.Since all the patients included in this study were recruited froma clinic that has between an 85 to 90 percent diagnostic accuracyrate (B.L. Beattie, personal communication, April, 1993) it isunlikely that many of the patients in this study were not of theAlzheimer's type.Statistical Considerations There were floor and ceiling effects in some of the test andexperimental conditions in this investigation. While thissituation is common in AD research, it would have masked the truemagnitude of the performance differences between patients and190controls. Since these performance differences were generallylarge, it is unlikely that any masking of the true magnitude ofthis difference would have led to different conclusions fromthose that have been reported.In addition, Bonferroni family-wise error rates were used toreduce the likelihood of committing a Type I error. In turn, thelargest sample size that was reasonably possible, together withthe inferential test that was most relevant for the data (i.e.,repeated measures Multivariate Analysis of Covariance), wasemployed to reduce the possibility of making Type II errors.Future WorkThere are three main directions that I would like to pursuein future work. The first, and perhaps the most obvious, is todirectly investigate the assumption that I, and others have madethat the encoding method used by AD patients has a large impacton their implicit test performance (cf. Graf, Tuokko, & Gallie,1990; Partridge et al., 1990; Salmon & Heindel, 1992). Forexample, will AD patients show differences in performance whenthey are provided with no directions, requested to provide apleasantness rating, or to identify and generate their ownmeaning for the critical targets used on various implicit memorytests?The second and third studies represent life long researchprojects. One project would be to administer the tests developed191for study one to HD, KS, and PD and other memory-impaired patientgroups. Since many of the implicit tests that were employed inthis study have not been used elsewhere, a study of this typewould contribute to current efforts towards establishing uniquetest signatures for these patient types.The other project would be to continue to investigate, in amore direct manner, the relationship between cognitive or mentalprocesses and the neural substrates on which they areorchestrated. This would be done by systematically varyingspecific encoding and retrieval conditions as well as stimulusformats and correlating subject performance on these tasks withbrain activity as indexed by techniques such as Positron EmissionTomography. This general line of inquiry is already extendingour current understanding of brain : behavior relationships. Forexample, Larry Squire and his colleagues (1992b) have shown thatsubjects activate different brain regions when they intentionallyversus non-intentionally recollect events. In addition KenPaller (1987, 1990) has found that different patterns of neuralactivity are correlated with the processing of abstract andconcrete words.Main Contributions of this WorkThis investigation makes both a specific and a generalcontribution to the field of neuropsychology. With respect toits specific contribution, it has shown that the implicit and the192explicit memory abilities of AD patients may be better retainedthan previous research has indicated. That is, by simplychanging the way that targets were encoded, the clues that wereprovided at retrieval, as well as the stimulus materials thatwere used, the memory performance of AD patients weresignificantly changed. With respect to this study's generalcontribution, the findings from this investigation suggest thatfuture research will need to deviate from the standard tests andmethods that have been previously employed if we are to extendour current understanding of the brain - implicit memoryrelationships of patients with Alzheimer's disease.REFERENCESAdams, R.A., & Victor, M. (1985). Principles of Neurology (3rded.). New York: McGraw Hill. .Agid, Y., Ruberg, M., Dubois, B., & Pillon, B. (1987).Anatomoclinical and biochemical concepts of subcorticaldementia. In S.M. Stahl, S.D. Iversen, and E.C. Goodman(Eds.), Cognitive Neurochemistry (pp. 248-271). Oxford:Oxford University Press.Albert, M., & Moss, M. (1984). The assessment of memorydisorders in patients with Alzheimer's disease. In L.Squire & N. Butters (Eds.), The Neuropsycholoav of Memory(pp. 236-246). New York: Guilford Press.Albert, M., Moss, M., & Milberg, W. (1989). Memory testing toimprove the differential diagnosis of Alzheimer's Disease.In R. Iqbal, H.M. Wisniewski, & B. Winblad (Eds.),Alzheimer's disease and related disorders (pp. 55-69). NewYork: Alan R. Liss.Alzheimer, A. (1907). Veber eine eigenartige Erkrankung derhirnrinde. Allg Z Psychiatr, 645, 146-148.Arnold, S.E., Hyman, B.T., Flory, J., Damasio, A.R., & VanHoesen, G. (1991). The Topographical and NeuroanatomicalDistribution of Neurofibrillary Tangles and Neuritic Plaquesin the Cerebral Cortex of Patients with Alzheimer's Disease.Cerebral Cortex, 1, 103-116.Backman, L., & Nilsson, L.G. (1984). Aging effects in freerecall: An exception to the rule. Human Learning, 2, 53-69.Backman, L., & Nilsson, L.G. (1985). Prerequisites for lack ofage differences in memory performance. Experimental AgingResearch, 11, 67-73.Backman, L., & Nilsson, L.G. (1989). Encoding of verbal andnonverbal features of action events under conditions offocused and divided attention. In A.F. Bennett & K.M.McConkey (Eds.), Cognition in Individual and Social Contexts (pp.295-304). North Holland: Elsevier Publishers.Ball, M.J. (1977). Neuronal loss, neurofibrillary tangles andgranulovacuolar degeneration in the hippocampus with agingand dementia. Acta Neuropathology, 22, 111-118.193194Ball, M.J., Hachinski, V., Fox, A., Kirshen, A.J., Fishman, M.,Blume, W., Kral, V.A., & Fox, H. (1985). A new definitionof Alzheimer's disease: A hippocampal dementia. Lancet, 1,14-16.Barr, M.L., & Kiernan, J.A. (1988). The Human Nemvous System: An Anatomical Viewpoint (5th ed.). New York: J.B.Lippincott Company.Battig, W.F., & Montague, W.E. (1969). Category norms forverbal items in 56 categories: A replication and extensionof the Connecticut category norms. Journal of Experimental Psychology, la, 1-46.Bayles, K.A., & Kaszniak, A.W. (1987). Communication andCognition in Normal Aging and Dementia. Toronto: Little,Brown & Company.Bayles, K.A., Tomoeda, C.K., Kaszniak, A.W., & Trosset, M.W.(1991). Alzheimer's disease's effects on semantic memory:Loss of structure or impaired processing? Journal ofCognitive Neuroscience, 2, 166-182.Belichambers, M.G. (1990). An Investigation of Remote Memory and the Effectiveness of Cues on the Retrieval of Old Memories in Dementia of the Alzheimer Type. UnpublishedDoctoral Dissertation, Ohio State University.Bellezza, F.S. (1988). Mnemonic devices and memory schemas. InM.A. McDaniel & M. Pressley (Eds.), Duagery and RelatedMnemonic Processes: Theories, Individual Differences andApplications (pp. 34-55). New York: Springer-Verlag.Bird, M., & Luszcz, M. (1991). Encoding specificity, depth ofprocessing and cued recall in Alzheimer's disease. Journal of Clinical and Experimental NeuroTsychology, 12, 508-520.Blass, J.P., Li-wen Ko, B.V.M., & Wisniewski, H.M. (1991).Pathology of Alzheimer's Disease. In M. Davidson (Ed.), The Psychiatric Clinics of North America (pp. 397-420). London:W.B. Saunders Company.Box, G.E.P. (1953). Non-normality and tests on variances.Diometrika, A2, 318-335.195Box, G.E.P. (1954). Some theorems on quadratic forms applied inthe study of analysis of variance problems. Effect ofinequality of variance in the one-way classification.Annals of Mathematical Statistics, 25, 290-302.Braak, H., & Braak, E. (1990). Morphology of the cerebralcortex in relation to Alzheimer's disease. In K. Maurer, P.Riederer, & H. Beckmann (Eds.), Alzheimer's Disease, Epidemiology. Neuropathology, Neurochemistry and Clinics (pp. 85-92). New York: Springer-Verlag.Braconnier, R.J., Cole, J.0., Spera, K.F., & De Vitt, D.R.(1982). Recall and recognition as diagnostic indices ofmalignant memory loss in senile dementia: A Bayesiananalysis. Experimental Aging Research, 1, 189-193.Brun, A. (1983). An Overview of Light and Electron MicroscopicChanges. In B. Reisberg (Ed.), Alzheimer's Disease: The Standard Refe ence (pp. 37-47). New York: The Free Press.Brun, A., & Englund, E. (1981). Regional pattern ofdegeneration in Alzheimer's Disease: neuronal loss andhistopathological grading. Histopatholoay,^549-564.Butters, N., Heindel, W.C., & Salmon, D.P. (1990). Dissociationof implicit memory in dementia: Neurological implications.Bulletin of the Psychonomic Society, 21, 359-366.Camp, C.J., & McKitrick, L.A. (1992). Memory interventions inAlzheimer's type dementia populations: Methodological andtheoretical issues. In R.L. West & J.D. Sinnot (Eds.),Everyday Memory and Aging: Current Research and Methodology(pp. 155-172). New York: Springer-Verlag.Canadian Pharmaceutical Association. (1992). Compendium ofPharmaceuticals and Specialties (27th ed.). Ottawa:Canadian Pharmaceutical Association.Carpenter, M.B. (1985). Core Textbook of Neuroanatomy. London:Williams and Wilkins.Cermak, L.S. & Reale, L. (1978). Depth of Processing andRetention of Words by Alcoholic Korsakoff Patients. Journal "^ • •L •^r^- I^-^I^I •^- i •^• 1111165-174.196Claparede, E. (1951). Recognition and 'me-ness'. In D.Rapaport (Ed.), Organization and pathology of thought  (pp.58-75). New York: Columbia University Press. (Reprintedfrom Archives de Psychologie, 1911, 11, 79-90).Glower, R.P., Zola-Morgan, S., & Squire, L.R. (1990). Lesionsof the perirhinal cortex, but not lesions of the amygdala,exacerbate memory impairment in monkeys following lesions ofthe hippocampal formation. Society for NeuroscienceAbstracts,^616.Cohen, J. (1988). Statistical power analysis 'or the behavioral sciences. San Diego, CA: Academic Press.Cohen, N.J., & Squire, L.R. (1980). Preserved Learning andretention of pattern analyzing skill in amnesia:Dissociation of knowing how and knowing that. Science, 210,207-209.Cohen, N.J., & Squire, L.R. (1981). Retrograde amnesia andremote memory impairment. peuropsychologia, 119, 337-356.Cohen, A.A., & Waters, W.F. (1985). Psychophysiologicalcorrelates of levels and stages of cognitive processing.Neuropsychologia, 22, 243-256.Cohen, R.L. (1981). On the generality of some memory laws.Scandinavian Journal of Psychology, 22, 267-281.Cohen, R.L. (1983). The effect of encoding variables on thefree recall of words and action events. Memory andCognition, 11, 575-582.Cohen, R.L., & Bean, G. (1983). Memory in educable mentallyretarded adults: Deficit in subject or experimenter?Intelligence, 2, 287-298.Cohen, R.L., & Bryant, S. (1991). The role of duration inmemory and metamemory of enacted instructions.Psychological Research, al, 175-182.Cohen, R.L., Sandler, S.P., & Schroeder, K. (1987). Aging andMemory for Words and Action Events: Effects of ItemRepetition and List Length. psychology and Aging, 2, 280-285.Cohen, R.L., & Stewart, M. (1982). How to avoid developmentaleffects in free recall. Scandinavian Journal of Psychology,22, 9-16.197Collins, A.M., & Loftus, E.F. (1975). A spreading activationtheory of semantic processing. psychological Review, 12,407-428.Corkin, S. (1982). Some relationships between global amnesiasand the memory impairments in Alzheimer's Disease. In S.Corkin, K.L. Davis, J.H. Growdon, E. Usdin, & R.L. Wurtmen(Eds.), Aging. vol. 19. Alzheimer's Disease: A Report of Progress, (pp. 149-164). New York: Raven Press.Craik, F.I.M. (1978). Age differences in memory: The roles ofattention and depth of processing. In L.W. Poon, J.L.Fozard, L.S. Cermak, D. Arenberg, & L.W. Thompson (Eds.),New Directions in Memory and Aging: Proceedings of the George A. Talland Memorial Conference (pp. 95-112).Hillsdale, New Jersey: Erlbaum.Craik, F.I.M. (1983). On the transfer of information fromtemporary to permanent memory. philosophical Transactions of the Royal Society of London, p302, 341-359.Craik, F.I.M., & Byrd, M. (1982). Aging and cognitive deficits:The role of attentional resources. In F.I.M. Craik & S.Trehub (Eds.), Aging and Cognitive Processes (pp. 191-208).New York: Plenum Press.Craik, F.I.M., & Lockhart, R.S. (1972). Levels of processing:A framework for memory research. Journal of Verbal Learningand Verbal Behavior, 11, 671-684.Craik, F.I.M., & Rabinowitz, J.C. (1984). Age differences inthe acquisition and use of verbal information. In H. Houma& D.G. Bouwhuis (Eds.), Attention and performance (pp. 417-499). New York: Erlbaum.Craik, F.I.M., & Tulving, E. (1975). Depth of processing andthe retention of words in episodic memory. Journal ofexperimental Psycholoay: General, 104, 268-294.Damasio, A.R., Van Hoesen, G.W., & Hyman, B.T. (1990).Reflections on the selectivity of neuropathological changesin Alzheimer's disease. In M.P. Swartz (Ed.), ModularDeficits in Alzheimer Type Dementia (pp. 83-100).Cambridge, Massachusetts: MIT Press.Davies, P., & Maloney, A.J.F. (1976). Selective loss of centralcholinergic neurons in Alzheimer's Disease. Lancet,  2,1403.198Davis, H.P., & Bernstein, P.A. (1992). Age-Related Changes inExplicit and Implicit Memory. In L.R. Squire & N. Butters(Eds.), Neuropsychology of Memory, (pp. 249-261). New York:Guildford Press.deLeon, M.J., George, A.E., Stylopoulous, L.A., Miller, D.C., &Smith, G. (1990a). In vivo studies of hippocampal atrophyin Alzheimer's disease. In K. Maurer, P. Riederer, & H.Beckmann (Eds.), Alzheimer Disease, Epidemiology. Neuropathology. Neurochemistry. and Clinics (pp. 523-529).New York: Springer-Verlag.deLeon, M.J., George, A.E., Stylopoulous, L.A., Miller, D.C., &Smith, G. (1990b). Computed tomographic hippocampalatrophy: An early marker for Alzheimer's disease. In J.M.Orgogozo & C.G. Gottfries (Eds.), progress in Dementia Research (pp. 55-62). New York: The Parthenon PublishingGroup.Dick, M.B., Kean, M.L., & Sands, D. (1989). Memory for actionevents in Alzheimer type dementia: Further evidence of anencoding failure. Brain and Cognition, 2, 71-87.Duke, L.W., Weathers, S.L., Caldwell, S.G., & Novack, T.A.(1990). Cognitive rehabilitation after head injury trauma:Toward an integrated cognitive/behavioral perspective onintervention. In C.J. Long, L.K. Ross, & M.G. Mutchnick(Eds.), Head Trauma: Acute Care to Recovery (pp. 1-30). NewYork: Plenum Press.Engelkamp, J., & Cohen, R.L. (1991). Current issues in memoryof action events. Psychological Research, La, 175-182.Engelkamp, J., & Zimmer, H.D. (1985). Motor programs and theirrelation to semantic memory. ggrman Journal of Psychology,2, 239-254.Esiri, M.M. (1989). Patterns of cortical and subcorticalpathology in Alzheimer's disease. In D.C. Davies (Ed.),Alzheimer's Disease: Towards an Understanding of the Aetiology and Pathogenesis (pp. 33-41). London: JohnLibbey.Esiri, M.M., Pearson, R.C.A., & Powell, T.P.S. (1986). Thecortex of the primary auditory area in Alzheimer's Disease.Brain Research, 366, 385-387.199Eslinger, P.J., & Damasio, A.R. (1986). Preserved motorlearning in Alzheimer's disease: Implications for anatomyand behavior. Journal of Neuroscience, A, 3006-3009.Filoteo, J.V., Delis, D.C., Massman, P.J., Demadura, T., Butters,N., & Salmon, D.P. (1992). Directed and divided attentionin Alzheimer's Disease: Impairment in shifting of attentionto global and local stimuli. Journal of Clinical and Experimental Neuropsycholoay, A, 871-883.Flament-Durand, J., & Brion, J.P. (1990). Morphology ofneurofibrillary tangles and senile plaques. In K. Maurer,P. Riederer, and H. Beckmann (Eds.), Alzheimer's Disease. Epidemioloay, Neuropatholoay, Neurochemistry. and Clinics (pp. 105-122). New York: Springer-Verlag.Forno, L.S. (1978). The locus coeruleus in Alzheimer's Disease.Journal of Neuropathology, 22, 614.Gallie, K.A., Graf, P., & Tuokko, H. (1990, November). StudyStrategies and their ability to offset memory losses causedby Alzheimer's Disease. Paper presented at the CanadianAssociation of Gerontology/Acg, Victoria.Gallie, K.A., Graf, P., & Tuokko, H. (1991, June). Memory andattentional processes in healthy elderly versus those withAlzheimer's disease. Paper presented at the CanadianPsychological Association, Calgary.Gallie, K.A., Tuokko, H., & Graf, P. (1991, February).Longitudinal changes in memory and attention in Alzheimer'sdisease. Paper presented at the InternationalNeuropsychological Society, San Antonio, Texas.Gaspar, P., & Gray, F. (1984). Dementia in IdiopathicParkinson's Disease. Acta Neuropathologia, AA, 43-52.Gerzt, H.J., & Cervos-Navarro, J. (1990). Neuronal Plasticityof the septo-hippocampal pathway in patients suffering fromdementia of Alzheimer type. In K. Maurer, P. Riederer. andH.Beckman (Eds.), Alzheimer's Disease. Epidemiology, Neuropathology, Neurochemistry, and Clinics (pp. 99-104).New York: Springer-Verlag.Gilman, A.G., Goodman, L.S., Rall, T.W., & Murad, F. (1985).The Pharmacological Basis of Therapeutics. Toronto:MacMillan Publishing Company.200Glass, G.V., & Hopkins, K.D. (1984). Statistical Methods inEducation and Psychology. New Jersey: Prentice Hall.Grady, C.L., Grimes, A.M., Patronas, N., Sunderland, T., Foster,N.L., & Rapoport, S.I. (1989). Divided attention asmeasured by dichotic speech performance in Dementia of theAlzheimer Type. Archives of Neuroloay,  Ai, 317-320.Graf, P., & Gallie, K.A. (1992). A transfer appropriateprocessing account for memory and amnesia. In N. Butters &L.R. Squire, (Eds.), Neuropsycholoay of Memory (pp. 241-248). New York: Guilford Press.Graf, P., & Mandler, G. (1984). Activation makes words moreaccessible, but not necessarily more retrievable. Journal of Verbal Learning and Verbal Behavior, al, 553-568.Graf, P., Mandler, G., & Haden, P.E. (1982). Simulating amnesicsymptoms in normal subjects. Science, 218, 1243-1244.Graf, P., & Masson, M.E.J. (in press). Introduction: Lookingback and into the future. In P. Graf & M. Masson (Eds.),Implicit Memory: New Directions in Cognition. Developmentand Neuropsychology. Hillsdale, New Jersey: LawrenceErlbaum.Graf, P., & Ryan, L. (1990). Transfer-appropriate processingfor implicit and explicit memory. Journal of Experimental Psychology: Learning, Memory, and Cognition,^978-992.Graf, P., & Schacter, D.L. (1987). Selective effects ofinterference on implicit and explicit memory for newassociations. Journal of Experimental Psychology: Learning, Memory. and Cognition, 12, 45-53.Graf, P., & Schacter, D.L. (1985). Implicit and Explicit memoryfor New Associations in Normal and Amnesic Subjects.Journal of Experimental Psychology: Learning. Memory, and Cognition, 11, 501-518.Graf, P., Shimamura, A.P., & Squire, L.R. (1985). Primingacross modalities and priming across category levels:Extending the domain of preserved function in amnesia.Journal of Experimental Psychology: Learning. Memory, andCognition, 11, 386-396.201Graf, P., Squire, L.R., & Mandler, G. (1984). The informationthat amnesic patients do not forget. Journal ofExperimental Psychology: Learning, Memory and Cognition,  1Q,164-178.Graf, P., Tuokko, H., & Gallie, K.A. (1990). Attentionaldeficits in Alzheimer's disease and related dementias. InJ.T. Enns (Ed.), The Development of Attention: Research andTheory (pp. 527-544). Amsterdam, Elsevier SciencePublisher.Green, R.L. (1988). Word stems as cues in recall completiontasks. Ouarterly Journal of Experimental Psychology, 38A,663-673.Grober, E., Gitlin, H.L., Bang, S., & Buschke, H. (1992).Implicit and explicit memory in young, old, and dementedadults. Journal of Clinical and Experimental Neuropsycholocm, j4., 298-316.Grosse, D.A., Wilson, R.S., & Fox, J.H. (1990). Preserved word-stem completion priming of semantically encoded informationin Alzheimer's disease. psychology and Aging, 5, 304-306.Hafner, H. (1990). Epidemiology of Alzheimer's disease. In K.Maurer, P. Riederer, & H. Beckmann (Eds.), Alzheimer's Disease. Epidemiology, Neuropathology, Neurochemistry, and Clinics (pp. 24-39). New York: Springer-Verlag.Makin, A.M., & Matheson, G. (1979). Dementia in ParkinsonDisease: A Neuropathologic Study. Neurology, 22, 1209-1214.Hardy, J.A., & Davies, D.C. (1988). Alzheimer's disease.British Journal of Hospital Medicine, 22, 372-377.Hardy, J.A., Mann, D.M.A., Wester, P., & Winblad, B. (1986). Anintegrative hypothesis concerning the pathogenesis andprogression of Alzheimer's disease. Neurobiology of Aging,2, 489-502.Heindel, W.C., Butters, N., & Salmon, D.P. (1988). Impairedlearning of a motor skill in patients with Huntingtons'disease. behavioral Neuroscience, 122, 141-147.Heindel, W.C., Salmon, D.P., & Butters, N. (1990). Pictorialpriming and cued recall in Alzheimer's and Huntington'sdisease. Brain and Cognition, la, 282-295.202Heindel, W.C., Salmon, D.P., Shults, C.W., Walicke, P.A., &Butters, N. (1989). Neuropsychological evidence formultiple implicit memory systems: A comparison ofAlzheimer's, Huntington's, and Parkinson's disease patients.The Journal of Neuroscience,  „a, 582-587.Hill, R.D., Storandt, M., & Simeone, C. (1990). The effects ofmemory skills training and incentives on free recall inolder learners. Journal of Gerontology, 45, 227-232.Hippius, H. (1990). Aloys Alzheimer 14.06 1984-19.12 1915. InK. Maurer, P. Riederer, & H. Beckmann, (Eds.), Alzheimer's Disease: Epidemiology, Neuropathology. Neurochemistry andClinics (pp XIII-XIX). New York: Springer-Verlag.Howard, D.V. (1980). Category norms: A comparison of theBattig and Montague (1969) norms with the responses ofadults between the ages of 20 and 80. Journal of Gerontology, 15, 225-231.Howard, D.V. (1988). Implicit and explicit assessment ofcognitive aging. In M.L. Howe & E.J. Brainerd (Eds.),Cognitive Development in Adulthood: Progress in Cognitive Development Research (pp. 3-37). New York: Springer-Verlag.Howard, D.V., Fry, A.F., & Brune, C.M. (1991). Aging and Memoryfor New Associations: Direct Versus Indirect Measures.Journal of Experimental Psychology: Learning. Memory, andCognition, 12, 779-792.Howell, D.C. (1987). Statistical methods for Psychology.Boston: Duxbury Press.Huff, F.J., Growdon, J.H., Corkin, S., & Rosen, T.J. (1987).Age at Onset and Rate of Progression of Alzheimer's Disease.Journal of the American Geriatrics Association,  25, 27-30.Hultsch, D.F., Masson, M.E.J., & Small, B.J. (1991). Adult agedifferences in direct and indirect tests of memory.journals of Gerontology, 45, 22-30.Hyman, B.T., Van Hoesen, G.W., Kromer, L.J., & Damasio, A.R.(1986). Perforant pathway changes and the memory impairmentof Alzheimers' Disease. Annals of Neurology, 2Q, 472-481.Insausti, R., Amaral, D.G., & Cowen, W.M. (1987). TheEntorhinal Cortex of the Monkey: II. Cortical Afferents.Journal of Comparative Neurology, 264, 356-395.203Iqbal, K. (1991). Prevalence and neurobiology of Alzheimer'sdisease: Some highlights. In K. Iqbal, D.R.C. McLachlan, B.Winblad, & H.M. Wisniewski (Eds.), Alzheimer's Disease: Basic Mechanisms, Diagnosis and Therapeutic Strategies (pp.1-5). Toronto: John Wiley and Sons.Java, R.I., & Gardiner, J.M. (1991). Priming and aging:Further evidence of preserved memory function. American Journal of Psychology, 104, 89-100.Jellinger, K. (1990). Morphology of Alzheimer's disease andrelated disorders. In K. Maurer, P. Riederer, & H. Beckmann(Eds.), Alzheimer's Disease, Epidemiology. Neuropathology, Neurochemistry and Clinics (pp. 61-77). New York: Springer-Verlag.Johnson, M.K., Kim, J.K., & Risse, G. (1985). Do alcoholicKorsakoff's syndrome patients acquire affective reactions?Journal of Experimental Psychology: Learning. Memory, andCognition, 11, 27-36.Kandel, E.R., & Schwartz, J.H. (1985). Principles of Neural Science (2nd ed.). New York: Elsevier.Karisson, T., Backman, L., Herlitz, A., Nilsson, L.G., Winblad,B., & Osterlind, P.O. (1989). Memory improvement atdifferent stages of Alzheimer's disease. Neuropsychologia,2.2, 737-742.Katzman, R. (1986). Alzheimer's disease. The New EnglandJournal of Medicine, 314, 964-973.Kausler, D.H., & Litchty, W. (1988). Memory for activities:Rehearsal independence and aging. In M.L. Howe & C.L.Brainerd (Eds.), Cognitive Development in Adulthood: Progress in Cognitive Development Research (pp. 93-131).New York: Springer-Verlag.Kausler, D.H., Lichty, W., Hakami, M.K., & Freund, S.J. (1986).Activity duration and adult age differences for activityperformance. Psychology and Aging, 1, 80-81.Keane, M.M., Gabrieli, J.D.E., Fennema, A.C., Growdon, J.H., &Corkin, S. (1991). Evidence for a Dissociation betweenPerceptual and Conceptual Priming in Alzheimer's Disease.Behavioral Neuroscience, 2, 326-342.Kirk, R.E. (1968). Experimental Design for the Behavioral Sciences. Belmont, California: Brooks Cole Publishers.204Kirshner, H.S., Webb, W.G., & Kelly, M.P. (1984). The NamingDisorder of Dementia. Neuropsychologia, 22, 23-30.Kirsner, K., Milech, D., & Standen, P. (1983). Common andmodality-specific processes in the mental lexicon. Memoryand Cognition, 11, 621-630.Knopman, D. (1991). Long-term retention of implicitly acquiredlearning in patients with Alzheimer's disease. Journal of Clinical and Experimental Neuropsycholoay, 12, 880-894.Knopman, D.S., & Nissen, M.J. (1987). Implicit learning inpatients with probable Alzheimer's Disease. Neuroloay, 2z,784-788.Korsakoff, S.S. (1889). Medical-psychological study of a formof diseases of memory. Revue Philosophique, 2a, 501-530.Landon, M., & Kidd, M. (1989). The structure of proteinchemistry of senile plaques and cerebrovascular amyloid.Alzheimers' disease: Towards an understanding of the aetioloay and pathogenesis (pp. 65-75). London: JohnLibbey.Lassman, H., Fisher, P., Bancher, C., & Jellinger, K. (1990).Immunocytochemical and ultrastructural pathology of nervecells in Alzheimer's disease. Epidemiology, Neuropatholoay, Neurochemistry, and Clinics (pp. 171-179). New York:Springer-Wein.Lewis, D.A., Campbell, M.J., Terry, R.D., & Morrison, J.H.(1987). Laminar and Regional Distributions ofNeurofibrillary Tangles and Neuritic Plaques in Alzheimer'sDisease: A Quantitative Study of Visual and AuditoryCortices. The Journal of Neuroscience, Z, 1799-1808.Lezak, M.D. (1983). Neuropsychological Assessment. New York:Oxford University Press.Light, L.L. (1991). Memory and aging: Four hypothesis insearch of data. Annual Review of Psvcholoay, AZ, 333-376.Light, L.L., & Albertson, S.A. (1989). Direct and indirecttests of memory for category exemplars in young and olderadults. psychology and Aging, A, 487-492.205Light, L.L., & Burke, D.M. (1988). Patterns of Language andMemory in Old Age. In L.L. Light & D.M. Burke (Eds.),Language. Memory. and Aging (pp. 244-271). CambridgeUniversity Press.Light, L.L., & LaVoie, D. (in press). Direct and indirectmeasures of memory in old age. In P. Graf & M. Masson(Eds.), Implicit Memory: New Directions in Cognition.Development and Neuropsycholoay. Hillsdale, New Jersey:Erlbaum.Light, L.L., & Singh, A. (1987). Implicit and explicit memoryin young and older adults. Journal of Experimental Psychology: Learning, Memory and Cognition,  12, 531-541.Light, L.L., Singh, A., & Capps, J.L. (1986). The dissociationof memory and awareness in young and older adults. Journalof Clinical and Experimental Neuropsvcholoay,  1, 62-74.Lockhart, R.S., & Craik, F.I.M. (1990). Levels of processing :A retrospective commentary on a framework for memoryresearch. Canadian Journal of Psychology, AA, 87-112.Lovelace, E.A. (1990). Basic concepts in cognition and aging.In E.A. Lovelace (Ed.), Aging and Cognition: Mental Processes, Self Awareness and Interventions (pp. 1-28).North Holland: Elsevier Science Publishers.Mahut, H., Zola-Morgan, S., & Moss, M. (1982). Hippocampalresections impair associative learning and recognitionmemory in the monkey. Journal of Neuroscience, 2, 1214-1229.Mandler, G. (1980). Recognizing: The judgement of previousoccurrence. Psychological Review, gz, 252-271.Mandler, G. (1988). Memory: Conscious and unconscious. In P.R.Soloman, G.R. Goethals, C.M. Kelly, & R.B. Stephens (Eds.),Memory: Interdisciplinary Approaches (pp. 84-106). NewYork: Springer-Verlag.Mann, D.M.A. (1991). Is the pattern of nerve loss in ageing andAlzheimer's Disease a real or only an apparent selectivity?Neurobioloay of Aging, 12, 340-342.Mann, D.M.A. (1985). The Neuropathology of Alzheimer's Disease:A Review with Pathogenetic, Aetiological and TherapeuticConsiderations. Mechanisms of Ageing and Development,  fl,213-255.206Mann, D.M.A. (1989). The pathogenesis and progression of thepathological changes of Alzheimer's disease. In D.C. Davies(Ed.), Alzheimer's Disease: Towards an Understanding of the Aetiology and Pathogenesis (pp. 43-53). London: JohnLibbey.Mann, D.M.A., & Esiri, M.M. (1988). The site of the earliestlesions in Alzheimer's disease. New England Journal of Medicine, 318, 12.Martin, A. (1992). Degraded Knowledge Representations inPatients with Alzheimer's Disease: Implications for Modelsof Semantic and Repetition Priming. In L. R. Squire & N.Butters (Eds.), Neuropsycholoay of Memory (pp. 220-231).New York: Guilford Press.Martin, A., Brouwers, P., Cox, C., & Fedio, P. (1985). On thenature of the verbal memory deficit in Alzheimer's disease.Brain and Language, 2A, 323-341.Masson, M., & Graf, P. (in press). Introduction: Looking backand into the future. In P. Graf and M. Masson (Eds.),Implicit Memory: New Directions in Cognition, Developmentand Neuropsycholoay. Hillsdale, New Jersey: LawrenceErlbaum.Matlin, M. (1983). Cognition. New York: CBS CollegePublishers.McDougall, W. (1924). Outline of Psychology. New York:Charles Scribner & Sons.McKhann, G., Drachman, D., Foistein, M., Katzman, R., Price, D.,& Stadlan, E.M. (1984). Clinical diagnosis of Alzheimer'sdisease: Report of the NINCDS-ADRDA Work Group and HumanServices Task Force on Alzheimer's Disease. Neurology,  2A,939-944.Milner, B. (1958). Psychological deficits produced by temporallobe excision. Research publications. Association for Research in Nervous and Mental Disease,  2A, 244-257.Mishkin, M. (1978). Memory in Monkeys severely impaired bycombined but not by separate removal of amygdala andhippocampus. Nature, 273, 297-298.207Mishkin, M. (1989). Cerebral Organization Underlying Cognition.proceedings of the Seventeenth Annual International Neuropsychological Association (pp. 15). Vancouver, BritishColumbia, Canada: Swets & Zeitlinger B.V. Lisse.Mishkin, M., Malamut, B., & Bachevalier, J. (1984). Memoriesand Habits: Two Neural Systems. In J.L. McGaugh, N.M.Weinberger, & G. Lynch (Eds.), Neurobiology of Learning and Memory (pp. 65-77). New York: Guilford Press.Mitchell, D.B. (in press). Implicit and Explicit Memory forPictures: Multiple Views Across the Lifespan. In P. Graf &M.E.J. Masson (Eds.), Implicit Memory: New Directions in Cognition. Development. and Neuropsycholocry.•Hillsdale, NJ:Erlbaum.Morris, C.D., Bransford, J.D., & Franks, J.J. (1977). Levels ofprocessing versus transfer appropriate processing. Journal of Verbal Learning and Verbal Behavior, 1A, 519-533.Morris, R.G., & Kopelman, M.D. (1986). The memory deficits inAlzheimer type dementia: A review. The Ouarterly Journal ofExperimental Psychology, 384, 575-602.Morris, J.C., McKeel, D.W., Storandt, M., Rubin, E.H., Price,J.L., Grant, E.A., Ball, M.J., & Berg, L. (1991). VeryMild Alzheimer's disease: Informant-based clinical,psychometric, and pathologic distinction from normal aging.Neurology, Ai, 469-478.Morris, J.C., & Rubin, E.H. (1991). Clinical diagnosis andcourse of Alzheimer's disease. In M. Davidson (Ed.), The Psychiatric Clinics of North America (pp. 223-236). London:W.B. Saunders Company.Moscovitch, M. (1982). Multiple Dissociations of function inamnesia. In L.S. Cermak (Ed.), Human Memory and Amnesia.New York: Erlbaum.Moscovitch, M., Winocur, G., & McLachlan, D. (1986). Memory asassessed by recognition and reading time in normal andmemory impaired people with Alzheimer's Disease and otherneurological disorders. Journal of Experimental Psychology: General, 115, 331-347.Mountcastle, V.B. (1978). Brain Mechanisms for DirectedAttention. Journal of the Royal Society of Medicine, 21,14-28.208Nebes, R.D. (1992). Semantic Memory Dysfunction in Alzheimer'sDisease: Disruption of Semantic Knowledge or InformationProcessing Limitation? In L.R. Squire & N. Butters (Eds.),Neuropsycholoay of Memory (pp. 233-240). New York:Guilford Press.Nelson, H.E., & McKenna, P. (1978). The use of current readingability in the assessment of dementia. Dritish Journal of Social and Clinical Psychology, 14, 259-267.Nelson, D.L., Reed, V.S., & McEvoy, C.L. (1977). Learning toorder pictures and words: A model of sensory and semanticencoding. Journal of Expe imental Psychology: Human Learning and Memory, 2, 485-497.Nilsson, L.G., & Backman, L. (1991). Encoding dimensions ofsubject-performed tasks. Psychological Research, 52, 212-218.Norris, M.P., & West, R.L. (1990). Adult age differences inactivity memory: Cue and strategy utilization. In T.M.Hess (Ed.), Aging and Cognition: Knowledge Organization andUtilization (pp. 1-31). North Holland: Elsevier SciencePublishers.Norris, M.P., & West, R.L. (1991).* Age differences in therecall of actions and cognitive abilities: The effects ofpresentation rate and object cues. Psychological Research,A2, 188-194.Nyberg, L., Nilsson, L.G., & Backman, L. (1991). A componentanalysis of action events. psychological Research, 12, 219-225.Paller, K.A. (1990). . Recall and stem-completion priming havedifferent electrophysiological correlates and are modifieddifferently by directed forgetting. Journal of Experimental Psychology: Learning. Memory. and Cognition, IA, 1021-1032.Paller, K.A., Kutas, M., Shimamura, A.P., & Squire, L.R. (1987).Brain responses to concrete and abstract words reflectprocesses that correlate with later performance on a test ofstem-completion priming. In R. Johnson, Jr., J. W.Rohrbaugh, & R. Parasuraman (Eds.), Current trends in Event-related research (pp. 360-365). North Holland: ElsevierScience Publishers.209Parkin, A.J. (in press). Implicit memory across the lifespan.In P. Graf & M. Masson (Eds.), Implicit Memory: NewDirections in Cognition. Development and Neuropsycholoay.Hillsdale, New Jersey: Lawrence Erlbaum.Partridge, F.M., Knight, R.G., & Feehan, M. (1990). Direct andindirect memory performance in patients with seniledementia. Psychological Medicine, 22., 111-118.Pearson, R.C.A., Esiri, M.M., Hiorns, R.W., Wilcock, G.K.,. &Powell, T.P.S. (1985). Anatomical correlates of thedistribution of the pathological changes within theneocortex in Alzheimer's disease. proceedings of the National cademy of Sciences, 12, 4531-4534.Perlmutter, M. (1978). What is memory aging the aging of?Developmental Psychology, 1A, 330-345.Plude, D.J., & Hoyer, W.J. (1985). Attention and performance:Identifying and localizing age deficits. In N. Charness(Ed.), Aging and human performance (pp. 47-100). New York:Wiley.Poon, L.W. (1985). •Differences in human memory with aging:Nature, causes, and clinical implications. In J.E. Birren &K.W. Schaie (Eds.), Handbook of the psychology of aging(2nd ed., pp. 427-462). New York: Van Nostrand Reinhold.Price, J.L., Davis, P.B., Morris, J.C., & White, D.L. (1991).The Distribution of Tangles, Plaques and RelatedImmunohistochemical Markers in Healthy Aging and Alzheimer'sDisease. Neurobioloay of Aging, 12, 295-312.Randolph, C. (1991). Implicit, explicit and semantic memoryfunctions in Alzheimer's disease and Huntington's disease.Journal of Clinical and Experimental Neuropsycholoay, 11,479-494.Rankin, J.L., & Collins, M. (1986). Adult age differences inmemory elaboration. Journal of Gerontology, AQ, 451-458.Reisberg, B. (1983). Alzheimer's Disease: The StandardReference. New York: The Free Press.Rey, A. (1964). L'examen clinique en psychologie. [The clinicalpsychology examination]. Paris: Presses Universitaires deFrance.210Richardson-Klavehn, A., & Bjork, R.A. (1988). Measures ofmemory. Annual Review of Psvcholoay, 22, 475-543.Roediger, H.L. (1990a). Implicit memory: A commentary.bulletin of the Psychonomic Society, 21, 373-380.Roediger, H.L. (1990b). Implicit memory: Retention withoutremembering. American Psychologist, A5, 1043-1056.Roediger, H.L., Srinivas, K., & Weldon, M.S. (1989).Dissociations between implicit measures of retention. In S.Lewandowsky, J.C. Dunn, & K. Kirsner (Eds.), implicit Memory: Theoretical Issues (pp. 67-84). Hillsdale, NewJersey: Lawrence Erlbaum Associates.Roediger, H.L., Weldon, M.S., & Challis, B.H. (1989).Explaining dissociations between implicit and explicitmeasures of retention: A processing account. In H.L.Roediger & F.I.M. Craik (Eds.), Varieties of memory and,consciousness: Essays in honor of Endel Tulvinq (pp. 3-41).Hillsdale, New Jersey: Lawrence Erlbaum Associates.Rogers, J., & Morrison, J.H. (1985). Quantitative morphologyand regional and laminar distributions of senile plaques inAlzheimer's disease. Journal of Neuroscience,  5, 2801-2808.Rohling, M.L., Ellis, N.R., & Scogin, F. (1991). Automatic andEffortful Memory Processes in Elderly Persons with OrganicBrain Pathology. Journal of Gerontology: Psychological ,Sciences, 46, 137-143.Rosen, W.G., & Mohs, R.C. (1982). Evolution of cognitivedecline in dementia. In S. Corkin (Ed.), Alzheimer's Disease: A Report of Progress (pp. 183-188). New York:Raven Press.Rosenthal, R., & Rosnow, R.L. (1991). Essentials of Behavioral Research and Data Analysis. Montreal: McGraw-Hill.Rudelli, R.D., Ambler, M.W., & Wisniewski, H.M. (1984).Morphology and distribution of Alzheimer neurotic andamyloid plaques in striatum and diencephalon. ActaMeuropathologia, AA, 273-286.Salmon, D.P., & Heindel, W.C. (1992). Impaired Priming inAlzheimer's Disease: Neuropsychological Implications. InL.R. Squire & N. Butters (Eds.), Neuropsycholoay of Memory(pp. 179-187). London: Guilford Press.211Salmon, D.P., Shimamura, A.P., Butters, N., & Smith, S. (1988).Lexical and semantic priming deficits in patients withAlzheimer's Disease. Journal of Clinical and Experimental Neuropsychology, 12, 477-494.Schacter, D.L. (1987). Implicit Memory: History and CurrentStatus. Journal of Experimental Psychology: Learning, Memory, and Cognition, la, 501-518.Schacter, D.L. (1989). On the relation between memory andconsciousness: Dissociable interactions and consciousexperience. In H.L. Roediger & F.I.M. Craik (Eds.),Varieties of memory and consciousness: Essays in honor of Endel Tulvina (pp. 355-389). Hillsdale New Jersey:Lawrence Erlbaum Associates.Schacter, D.L. (1990a). Introduction to Implicit Memory:Multiple perspectives. bulletin of the Psychonomic Society,2A, 338-340.Schacter, D.L. (1990b). Perceptual representation systems andimplicit memory: Toward a resolution of the multiple memorysystems debate. ,Annals of the New York Academy of Sciences,608, 543-571.Schacter, D.L. (1992). Understanding Implicit Memory: ACognitive Neuroscience Approach. American Psychologist, AZ,559-569.Schacter, D.L., & Graf, P. (1989). Modality specificity ofimplicit memory for new associations. Journal of Experimental Psychology: Learning, Memory. and Cognition,2, 432-444.Shallice, T. (1990). From Neuropsycholoay to Mental Structure.Cambridge: Cambridge University Press.Shimamura, A.P. (1986). Priming effects in amnesia: Evidencefor a dissociable memory function. The Ouarterly Journal ofExperimental Psychology, 381, 619-644.Shimamura, A.P. (in press). Neuropsychological analysis ofimplicit memory: History, Methodology and TheoreticalInterpretations. In P. Graf & M. Masson (Eds.), ImplicitMemory: New Directions in Cognition, Development andNeuropsychology. Hillsdale, New Jersey: Lawrence Erlbaum.212Shimamura, A.P., Salmon, D.P., Squire, L.R., & Butters, N.(1987). Memory dysfunction and word priming in dementia andamnesia. Behavioral Neuroscience, 101, 347-351.Shimamura, A.P., & Squire, L.R. (1984). Paired-associatelearning and priming effects in amnesia: Aneuropsychological study. Journal of Experimental Psychology: General, 213, 556-570.Snodgrass, J.G., & Corwin, J. (1988). Perceptual identificationthresholds for 150 fragmented pictures from the Snodgrassand Vanderwart picture set. perceptual and Motor Skills,0., 3-36.Snodgrass, J.G., & Corwin, J. (1988). Pragmatics of measuringrecognition memory: Applications to dementia and amnesia.Journal of Experimental Psychology: General, 117, 34-50.Snodgrass, J.G., & Vanderwart, M. (1980). A standardized set of260 pictures: Norms for name agreement, familiarity, andvisual complexity. 'Journal of Experimental Psychology: Human Learning and Memory, 1, 174-215.SPSS-X Inc. (1990). Statistical Package for the Social Sciences: User's Guide. New York: McGraw-Hill.Squire, L.R. (1982). Comparisons between forms of amnesia:Some deficits are unique to Korsakoff's syndrome. Journal of Experim-ntal Psychology: Learning. Memory. and Cognition, 8, 560-571.Squire, L.R. (1987). Memory and Brain. New York: OxfordUniversity Press.Squire, L.R. (1992a). Memory and the hippocampus: A synthesisfrom findings with rats, monkeys, and humans. Psychological Review, 22, 195-231.Squire, L.R. (1992b). Declarative and Nondeclarative Memory:Multiple Brain Systems Supporting Learning and Memory.Journal of Cognitive Neuroscience, A, 232-243.Squire, L.R., & Butters, N. (1992). Neuropsycholoay of Memory(pp. 1-3). New York: Guilford Press.Squire, L.R., Knowlton, B., & Musen, G. (1993). The Structureand Organization of Memory. Annual Review of Psychology,AA, 453-495.213Squire, L.R., Ojeman, J.G., Miezin, F.M., Petersen, S.E., Videen,T.O., & Raichle, M.E. (1992b). Activation of thehippocampus in normal humans: A functional neuroanatomicalstudy of memory. Proceedings of the National Academy of ,Science, 12, 1837-1841.Squire, L.R., Shimamura, A.P., & Graf, P. (1985). Independenceof recognition memory and priming effects: Aneuropsychological analysis. Journal of Experimental psychology: Learning. Memory. and Cognition, U, 37-44.Squire, L.R., Slater, P.C., & Chace, P.M. (1975). Retrogradeamnesia: Temporal gradient in very long-term memoryfollowing electroconvulsive therapy. Science, 187, 77-79.Squire, L.R., & Zola-Morgan, S. (1991). The Medial TemporalLobe Memory System. Science, 253, 1380-1385.Squire, L.R., Wetzel, D., & Slater, P.C. (1978). Anterogradeamnesia following ECT: An analysis of the beneficial effectsof partial information. Neuropsychologia, lk, 339-348.Strauss, M.E., Weingartner, H., & Thompson, K. (1985).Remembering words and how often they occurred in memoryimpaired patients. Memory and Cognition, 11, 507-510.Stuss, D.T., Ely, P., Hugenholtz, H., Richard, M.T., LaRochelle,S., Poirier, C.A., & Bell, I. (1985). Subtleneuropsychological deficits in patients with good recoveryafter closed-head injury. Neurosurgery, 12, 41-47.Terry, R.D., & Katzman, R. (1983). Senile Dementia of theAlzheimer Type. Annals of Neurology, 1A, 497-506.Toledano-Gasca, A. (1988). Hypotheses concerning the aetiologyof Alzheimer's Disease. pharmacopsychiatry, 21, 17-25.Tomlinson, B.E. (1972). Morphological brain changes in non-demented old people. In H.M. VanPrag and A.F. Kalverbov(Eds.), Aging of the Central Nervous System. Biological anclPsychological Aspects (pp. 38-57). Bonn: Harlem Press.Tomlinson, B.E., Blessed, G., & Roth, M. (1968). Observationson the Brains of Non-demented old people. Journal ofNeurological Science, 2, 331-356.Treves, T.A. (1991). Epidemiology of Alzheimer's disease. InM. Davidson (Ed.), The Psychiatric Clinics of North America(pp. 251-265). London: W.B. Saunders Company.214Tulving, E. (1986). What kind of hypothesis is the distinctionbetween episodic and semantic memory? Journal of experimental Psycholoay: Learning, Memory, and Cognition. 12, 307-311.Tulving, E., & Schacter, D.L. (1990). Priming and Human MemorySystems. Science, p47, 301-306.Tuokko, H., & Crockett, D. (1989). Cued recall and memory indementia. Journal of Clinical and Experimental Neuropsychology, 11, 278-294.Tuokko, H., & Crockett, D. (1991). Assessment of everydayfunctioning in normal and malignant memory disorderedelderly. In D.E. Tupper & K.D. Cicerone (Eds.), The Neuropsychology of Everyday Life: Issues in Development andRehabilitation (pp. 135-182). Boston: Kluver AcademicPublishers.Tuokko, H., Gallie, K.A., & Crockett, D.J. (1990). Patterns ofmemory deterioration in normal and memory impaired elderly.Developmental Neuropsychologv, A, 291-300.van Dijk, P.T.M., Dippel, D.W.J., & Habbema, J.D. (1991).Survival of patients with dementia. Journal of the AmericanGeriatric Society, 22, 603-610.Van Hoesen, G.W., & Damasio, A.R. (1987). Neural Correlates ofcognitive impairment in Alzheimer's Disease. In V.B.Mountcastle, F. Plum, & S.R. Geiger (Eds.), Handbook of Physioloay: The Nervous System (pp. 871-898). Maryland:American Physiological Society.Victor, M., & Agamanolis, D. (1990). Amnesia due to Lesionsconfined to the Hippocampus: A Clinical-Pathologic Study.Journal of Cognitive Neuroscience, 2, 246-257.Warrington, E.K., & Weiskrantz, L. (1968). New method oftesting long-term retention with special reference toamnesic patients. Nature, 217, 972-974.Warrington, E.K., & Weiskranz, L. (1970). Amnesia:Consolidation or retrieval? Nature, 224, 628-630.Welch, B.L. (1947). The generalization of students' problemswhen several different population variances are involved.eiometrika, 21, 29-35.215West, R.L., & Tomer, A. (1989). Everyday memory problems ofhealthy older adults: Characteristics of a successfulintervention. In G.C. Gilmore, P.J. Whitehouse, & M.L.Wykle (Eds.), Memory, Aging and Dementia: Theory, Assessmeqtand Treatment (pp. 74-98). New York: Springer Publishing.Winer, B.J., Brown, D.R, & Michels, K.M. (1991). Statistical Principles in Experimental Design (3rd ed.). New York:McGraw-Hill Inc.Wisniewski, H.M., Currie, J.R., Wegiel, J., Silverman, W.P.,Merz, G.S., Jenkins, E.C., & Iqbal, K. (1990).Pathological, biochemical and genetic aspects of Alzheimer'sDisease. In J.M. Orgogozo & C.G. Gottfries (Eds.), Progress in Dementia Research (pp. 37-62). New Jersey: TheParthenon Publishing Group.Wisniewski, H.M., & Terry, R.D. (1973). Reexamination of thepathogenesis of the senile plaque. In H.M. Zimmerman (Ed.),Progress in Neuropathology, (pp. 1-26). New York: Gruneand Stratton.Woods, R.T., & Piercy, M. (1974). A similarity between amnesicmemory and normal forgetting. Neuropsychologia, la, 437-445.Yates, F. (1934). Contingency tables involving small numbersand the Chi-square test. Journal of the Royal Statistical Society Supplement, 1, 217-235.Zola-Morgan, S. (1993, March). UBC Neuroscience Lecture Series.Zola-Morgan, S., & Squire, L.R. (1985). Medial Temporal lesionsin monkeys impair memory in a variety of tasks sensitive tohuman amnesia. Behavioral Neuroscience, 22, 22-34.Zola-Morgan, S., & Squire, L.R. (1986). Memory impairment inmonkeys following lesions of the hippocampus. Behavioral Neuroscience, 100, 155-160.Zola-Morgan, S., Squire, L.R., & Amaral, D.G. (1986). HumanTemporal Region: Enduring Memory Impairments Following aBilateral Lesion limited to Field CAI of the Hippocampus.The Journal of Neuroscience, A, 2950-2967.Zola-Morgan, S., & Squire, L.R. (1992). The Components of theMedial Temporal Lobe Memory System. In N. Butters and L.R.Squire, (Eds.), Neuropsycholoay of Memory (pp. 325-335).New York: Guilford Press.216Zola-Morgan, S., Squire, L.R., & Amaral, D.G. (1986). Lesions ofthe amygdala that spare adjacent cortical regions do notimpair memory or exacerbate the impairment following lesionsof the hippocampal formation. Journal of Neuroscience, 2,1922-1936.Zola-Morgan, S., Squire, L.R., Amarall, D.G., & Suzuki, W.A.(1989). Lesions of Perirhinal and Parahippocampal Cortexthat Spare the Amygdala and Hippocampal Formation ProduceSevere Memory Impairment. The Journal of Neuroscience, 2,4355-4370.APPENDICES217Appendix ALetters of Contact and ConsentTable of Contents (1) Letter inviting AD patients and Control Subjects toparticipate in study (includes information about the studyand consent form).(2) Letter containing information about study for the patient'scaregiver (includes consent form).218219DateDear Patient/Control Subject:This letter invites you to participate in a project thatconcerns MEMORY FOR FACTS AND EPISODES IN OLDER ADULTS. Theproject is part of my doctoral dissertation, and is conductedunder the supervision of Dr. Holly Tuokko and Dr. Peter Graf.Both Dr. Tuokko and Graf are on the faculty of the University ofBritish Columbia. You can reach Dr. Tuokko at 822-7535 and Dr.Graf at 822-6635. I am Karen Gallie -- a senior graduatestudent; you can reach me at 822-2140.The project focuses on memory for common words and pictures;it is part of a larger study that examines changes in memory thatoccur in adulthood and in various clinical conditions. There aretwo parts to the project, each of which lasts about 45 minutes.The two parts will be conducted on different days. If you wouldlike to participate, I --Karen Gallie -- will contact you toarrange an appointment that is convenient to you.The project involves a series of short tasks that servedifferent purposes. The main purpose is to find out about twotypes of memory: the first has to do with such things asremembering items from a grocery list or the time of anappointment, whereas the second has to do with remembering thingsthat were learnt a long time ago, such as the rules of a game.To find out about these two types of memory, you will do a seriesof paper and pencil task that require saying and writing commonwords in response to questions such as "what things belong to thecategory sports?" For another task we show you cards with wordsor pictures of common objects such as a desk or a pencil and thenask you to remember these items. Your performance on these testsgives an indication of how different types of memory function inadulthood. 'Your results will be treated in strict confidence. For thispurpose the test forms will contain no information that wouldallow anyone else to find out about who you are. The resultswill be used to gain a better understanding of memory and toprepare scientific reports about memory functions in adulthood.Although you will participate as an individual, your individualresults will not be published in any form; they will be combinedwith the results of a large group of adults. If you areinterested, we will send you a copy of our findings at theconclusion of the study.220Participation in this project is completely voluntary anddoes not affect in any way your access to treatment or otherprograms. Participation requires about 2 times 45 minutes ofyour time with each 45 minute session arranged at yourconvenience. I (Karen Gallie) will visit you in your home toconduct each session. If at any time during the project youwould like to stop, you are free to do so. Your withdrawal willbe kept in strict confidence, will not be reported to anyone, andyou will not be penalized in any way.If you have any further questions about this project andabout its implications I will be glad to talk with you. You cancontact me at 822-2140; or you can call Dr. Graf at 822-6635 orDr. Tuokko at 822-7535.IF YOU WOULD LIKE TO PARTICIPATE PLEASE SIGN THIS FORM ANDCONTACT ME BY CALLING 822-2140, OR DR. GRAF AT 822-6635.Sincerely,Karen Gallie^ Peter GrafDoctoral Student Associate ProfessorIf you would like to participate in this project, pleaseprint and sign your name below and return the bottom part of thisform to me in the enclosed envelope. Your signature indicatesthat you have received and read a copy of this form.Name: ^Signature:^(please print)Date:^DateDear Caregiver:Our files at the Clinic for Alzheimer's Disease and RelatedDisorders list you as the primary caregiver of ^ . With theenclosed letter we are inviting ^ to participate in a newresearch project that concerns MEMORY FOR FACTS AND EPISODES INOLDER ADULTS. We are writing to you at this time to solicit yourhelp to ensure that ^ understands fully the content of theenclosed letter which describes the nature and purpose of theresearch project, where and when it will be conducted, and thatparticipation is entirely voluntary; in other words, we requestyour help in obtaining informed consent from the person in yourcare.The enclosed letter presents an outline of the project,describes the nature of the tasks that are involved, how muchtime is required of each participant, that participation isentirely voluntary, and it indicates what we plan to do with theresults. If ^ has problems reading and/or understanding thisletter, we would very much appreciate if you assist them in anyway possible.The project is conducted under the supervision of Dr. HollyTuokko and myself, Dr. Peter Graf. Both of us are on the facultyof the University of British Columbia. If you have any questionsyou can reach us by telephone: for Dr. Tuokko call 822-7535, andfor Dr. Graf call 822-6635.If ^ decides to participate in our project after readingthe enclosed letter will you please ask them to sign at thebottom of the page in the space provided. We would alsoappreciate your signature on the bottom of this letter toindicate to us that you have read the enclosed letter and thatyou have assisted the person in your care to make an informeddecision about participating in our project.If the person in your care would like to participate in ourproject, PLEASE CONTACT ME BY CALLING 822-2140 OR 822-6635.221We appreciate your assistance in this matter.Sincerely,Karen Gallie^Peter GrafPh.D. candidate Associate Professor222223I have read the enclosed project description and assisted ^to understand it and make an informed decision aboutparticipating in it.Name: ^Signature:^(please print)Date:^224Appendix BGuidelines Used to Establish Presence of AD and Level ofFunctional Impairment Table of Contents (1) NINCDS-ADRDA guidelines for making a Possible or ProbableDiagnosis of AD.(2) Copy of Functional Rating Scale and description of proceduresused to assign patients FRS scores and categorize them intomild, moderate, and severe levels of functional impairment.225Criteria for clinical diagnosis of Alzheimer's disease from the NINCDS-ADRDA1. The criteria for the clinical diagnosis of PROBABLE Alzheimer's disease include:— dementia established by clinical examination and documented by the MiniMental Test(Folstein et al., 1975), Blessed Dementia Scale (Blessed et al., 1968), or some similarexamination, and confirmed by neuropsychological tests— deficits in two or more areas of cognition— progressive worsening of memory and other cognitive functions— no disturbance of consciousness— owes between ages 40 and 90, most often after age 65— absence of systemic disorders or other brain diseases that in and of themselves couldact punt for the progressive deficits in memory and cognition11. The diagnosis of PROBABLE Alzheimer's disease is supported by:— progressive deterioration of specific cognitive functions such as language (aphasia), motorskills (apraxia), and perception (agnosia)— impaired activities ofdaily living and altered patterns of behavior— family history of similar disorders, particularly if confirmed neuropathologically— laboratory results of:— normal lumbar puncture as evaluated by standard techniques— normal pattern or nonspecific changes in EEG, such as increased slow-wave activity— evidence of cerebral atrophy on CT with progression documented by serial observationIll. Other clinical features consistent with the diagnosis of*PROBABLE Alzheimer's disease,after exclusion of causes of dementia other than Alzheimer's disease, include:— plateaus in the course ofprogression of the illness— associated symptoms of depression; insomnia; incontinence; delusions; illusions;hallucinations; catastrophic verbal: emotional, or physical outbursts; sexual disorders; andweight loss— other neurological abnormalities in some patients, especially with more advanced disease,and including motor signs such as increased muscle tone, myoclonus, or gait disorder— seizures in advanced disease— CT normal for age^'IV. Features that make the diagnosis of PROBABLE Alzheimer's disease uncertain or unlikelyinclude:— sudden, apoplectic onset— focal neurologic findings such as hemiparesis, sensory loss, visual field deficits, andincoordination early in the course of the illness— seizures or gait disturbances at the onset or very early in the course of the illnessV. Clinical diagnosis of POSSIBLE Alzhiemer's disease:— may be made on the basis of the dementia syndrome, in the absence of the neurologic,psychiatric, or systemic disorders sufficient to cause dementia, and in the presence ofvariations in the onset, in the presentation, or in the clinical course— may be made in the presence ofa second systemic or brain disorder sufficient to producedementia, which is not considered to be the cause ofthe dementia— should be used in research studies when a single, gradually progressive severe cognitivedeficit is identified in the absence of other identifiabk causeVI. Criteria for diagnosis of DEFINITE Alzheimer's disease arc:— the clinical criteria for probable Alzheimer's disease— histopathologic evidence obtained from a biopsy or autopsyVII. Classification of Alzheimer's disease for research purposes should specify features that maydifferentiate subtypes of the disorder, such as:— familial occurrence— onset before age of 65— presence of trisomy-21— coexistence of other relevant conditions such as Parkinson's diseaseNEWRY^CUESTIONAII.E(1) (7)NILO^MODERATE^NYE'(3) (4) (S)No deficit Orinceasisteet forget-fulness evident onlyee clinical interviewVariable ssipteisreported by patientor relative. seeminglyunrelated to level offunctioningMemory losses utitchinterfere with dailyliving. sere apparentfor recent events.Moderato mosery loss. onlyhighly leaned arterialretained. new materialrapidly lostUwe?* memory loss.needle to recallrelevant aspects ofcurreat life. wittyMnch, recall of pastlife.NEWAYNeither patient norrelatives suer. of anydeficitVariable levels offunctioning repottedby patient er relativesno objective evidenceof deficits in employ-meet or social .MegatonsPatient or relativeaware of decreasedperformaoce Indemendleg omplopmeetor social settings.spews menial tocasual infectionPatient or relative areaware of cooing deter-ioration. eses not pearmensal to objectiveobserver. unable to Performjob. little 1o/dependentfunctioning outside homeNarked Impairment ofsocial ladopoadoetfunctioning outsidehemsSOCIAL/COMMUNITYANDOCCUPATIONALNe thanes, noted bypatieet or relativeSlightly decreasedinvolvement lahousehold teaksand hobblesEngages le socialactivities to thishome but defialteimpaimmeat in somehousehold tests. samecomplicated hobbiesand laterests aband000dOnly dimple chores/hobbiespreserved. most complicatedhobbles/interestsabondowedNo isidepeedeet involve-meat to home sr hobbiesNOWANDHOMESPERSONALCAREfully capable ofself-careOccasional problemswith self-carereported by pottiest/relatives orobservedNeeds prompting tocomplete tasksadequately (1.e.dress)ng. feeding.hygiene)Requires supervisiee indressing. feeding. homilies.and keeping track ofpersonal effectsNeeds coaster* super-yule. and assistaace.with fee tag. dressialker hygiene. etc.LANGUAGESKILLSNo disturbance oftemples. reportedby patient orrelettlaiSubjective complaintof. er relativereports. levovseodeficits. usuallylimited to wardtimeless or Ramie,Fattest er relativeresorts verlobledisturbeaces or sockskills as articulationOf nosing. occasteoallihguage Westmontevident duringesamlostlem.Patient er relativereports consistentimposts disturtaace.language dIsturberceevident em examinatiemSevere lopeirmeat ofreceptive aer/ersapressive language.preductien ofwelatellitIbl* speechSolves everydayproblems adequatelyVariable lopelrmoatof problem-2000hdifferencesDifficulty in mealtimetemples problemsNarked Impstrnoot emtemples problma-solving tasks11061114, es 1101.0PfebION St 1W47 level,Via sad error behevieroften ebs4OVedPIONItEN SOLVINGANDREASONINGAFFECTMe change ineffect reported bypatient orrelativeApiropriatit concernwith reseset tosymptemateseyImfrogeent Changes inaffect (e.g. Irrita-bility) reported byflatfeet or relative.mould appear morsel toobjective observerFrequeet thong's Inaffect reported bypatient or relative.noticeeble to objectiveobserverSestolood elterotiemsof effect. imemirmdcontact with realityobserved er reportedORIENTATIONFully oriented Occasional diff-iculties with timerelatiershipsHerbed difficulty withtime relationshipsUsually disorleated totime *ad often to PlatoOriented arty toperson or net et 411(Tuokko & Crockett, 1989)227Assignment of FRS Scores to AD Patients Each patient was assigned an FRS score ranging from one(normal), to five (severe), for each of the eight dimensions ofthis scale. FRS scores were obtained from a consensual teamscore obtained from the clinic's geneticist, geriatrician,neurologist, neuropsychologist, psychiatrist, and social worker.These eight FRS scores were then averaged to produce one estimateof the patient's overall level of functional impairment (averageFRS score = AFRAS), or dementia severity.The following cut-off scores were used to assign patients tofunctional impairment groups: Mildly impaired: 2.13 < AFRAS <2.63; Moderately impaired: 2.63 < AFRAS < 3.13; and Severelyimpaired: 3.13 or greater.Appendix CTelephone InterviewTable of Contents (1) Procedure and questions used to screen volunteers whoacted as non-demented control subjects.228229Procedure and questions used to screen volunteers who would actas controls in this investigation.Procedure People who were interested in acting as volunteers were contactedby telephone. During the telephone conversation thisexperimenter began by identifying herself and her affiliationwith UBC and confirming that the person she was speaking to wasinterested in participating in a memory study. If the respondentsaid yes the nature of the study was explained to them includingthe fact that the investigation was designed to learn more aboutmemory in people with Alzheimer's disease including strategiesthat we were developing to help them remember more information.They were also informed of their rights as UBC subjects and thefact that if they participated they would receive a ten dollarhonorarium and a copy of the study's results upon completion.They were then told that to conduct this investigation werequired healthy older adults to act as controls. Controls wouldneed to be healthy people who were 50 years of age or older whodid not have any health or memory problems so that we couldestablish how normal people performed on our tasks in comparisonto people with Alzheimer's disease. To be considered as acontrol it would be necessary for volunteers to meet very strict230selection criteria and this meant that I would need to ask themsome very personal questions. They were told that they did nothave to answer any of these questions if they did not want to andthey could ask me any questions they wanted in return.The following are questions that were covered during thecourse of the informal telephone interview. In most casesvolunteers spontaneously provided information that made directquestioning unnecessary. In no case did a volunteer refuse toanswer any of the following questions.Questions1). How old are you? (Volunteers had to be 50 years of age orolder).2). How would you describe your memory abilities? (If theyreported concern but could not name a serious memory lossevent they were considered normal. If they reported arepetitive problem remembering events they were notconsidered).3). Do you, or have you in the past, suffered from a neurologicalor psychiatric problem such as depression, or stroke, or haveyou been in an accident where you hit your head? (If theyreported yes they were not included as acontrol).4). How would you describe your current health? (If theydescribed their health as being poor they were not acceptedto act as a control).5). Are you currently taking any medications? (If they reportedcontinuously taking more than three medications they werethanked for their time but told that, due to the nature ofthe study, they could not act as a control).Appendix DMaterials Used for Explicit and Implicit Memory TestsTable of Contents (1) Written and Spoken Word Materials(2) Picture Materials(3) Object Materials231232Implicit and Explicit Memory Test MaterialsSpoken and Written Word Materials Categories and three items named in response to it arecontained below. Numbers found in brackets indicate thefrequency in which that word was provided by 442 universitystudents. Please refer to the method section for the procedureused to select these categories from Battig and Montague's norms(1969).SET1 STATE PENNSYLVANIA (30%)TEXAS (33%)ILLINOIS (42%)SHIP CRUISER (32%)DESTROYER (36%)SAILBOAT (40%)FLOWER ORCHID (30%)DAISY (40%)CARNATION (41%)FOOTGEARSLIPPERS (36%)SANDALS (50%)SOCKS (58%) Set average = 39% SET 3 STONE PEARL (40%)SAPPHIRE (55%)EMERALD (74%)RELATIVE NIECE (31%)NEPHEW (32%)GRANDFATHER (68%)SET 2 VEHICLE BOAT (32%)MOTORCYCLE (39%)BICYCLE (44%)MONEY PENNY (55%)QUARTER (59%)DIME (59%)WEAPON RIFLE (37%)GUN (89%)KNIFE (92%)COUNTRY MEXICO (31%)SPAIN (36%)ITALY (36%) Set average = 50.7% aBT__1KITCHEN UTENSILSPATULA (35%)POT (46%)PAN (54%)METAL ZINC (30%)TIN (39%)SILVER (57%)READING MATERIALPAMPHLET (45%)NEWSPAPER (67%)BOOK (83%)CLOTH LINEN (32%)NYLON (48%)RAYON (50%) Set average = 52% SET 5 CITY BALTIMORE (37%)SAN FRANCISCO (39%)LOS ANGELES (43%)TREE BIRCH (30%)ELM (48%)PINE (48%)GIRL'S NAMEJANE (30%)ANN (36%)SUE (37%)VEGETABLE ASPARAGUS (31%)SPINACH (37%)LETTUCE (43%) Set average = 38% PART OF BUILDINGROOM (36%)CEILING (38%)FLOOR (54%)WEATHERSLEET (39%)HAIL (47%)SNOW (60%) Set average = 44.6% SET 6 BIRD CANARY (30%)BLUEBIRD (31%)CROW (34%)MUSICAL INSTRUMENTOBOE (33%)TROMBONE (39%)SAXOPHONE (40%)DWELLINGCAVE (35%)TENT (43%)APARTMENT (71%)FISHTUNA (31%)PERCH (32%)SALMON (32%) Set average = 37.6% 233SET 7^ SET 8TIME MILITARY TITLE WEEK (63%)^ ADMIRAL (33%)DECADE (72%) CORPORAL (38%)MONTH (73%) MAJOR (55%)DISTANCE ^ FRUIT MILLIMETER (41%)^ LEMON (30%)KILOMETER (48%) GRAPEFRUIT (35%)CENTIMETER (58%) PLUM (38%)4 FOOTED ANIMALPIG (32%)ELEPHANT (41%)TIGER (46%)COLOURBROWN (49%)PINK (50%)WHITE (62) Set average = 52.9% CLOTHINGTIE (31%)SWEATER (37%)HAT (45%)SPORT GOLF (35%)SOCCER (36%)SWIMMING (63%) Set average = 39.7% 234235Picture Materials Items selected from Snodgrass & Corwin's fragmented picturenorms (1988) that were consistently named by 70% or more of their219 subjects and were given a rating of 2 or more for itemfamiliarity and image agreement. A score of 1 indicates that thepicture was unfamiliar and there was no image match, and a scoreof 5 indicates that the picture was very familiar and was a goodmatch to the subject's mental image of that item.Refer to the method section for the procedure used to selectand administer these items.Percentage Item ImageNaming Consistency Familiarity ConsistencySET 1(Mean score) (Mean score)BALLOON 100 2.58 4.33BELT 98 4.12 4.05BOOK 100 4.75 4.33DESK 95 4.32 3.18DOG 100 4.60 3.05EYE 98 4.88 4.15MUSHROOM 98 2.88 3.78PENCIL 100 4.42 4.40RULER 98 3.58 3.98SUITCASE 79 3.65 2.98VEST 98 3.48 3.70VIOLIN 86 2.68 4.18AVERAGE 95.8% 3.83 3.84SET 2APPLE 98 3.98 4.05BED 100 4.72 3.65BOWL 95 4.18 3.76BREAD 83 4.40 4.02COMB 93 4.52 3.78FISH 100 3.28 3.58GORILLA 76 2.05 3.58GUITAR 98 3.58 4.20HOUSE 95 4.38 2.65KNIFE 90 4.45 3.25MOUNTAIN 90 2.70 3.52SHIRT 98 3.64 3.28AVERAGE 93j 3.82 3.61Percentage Item ImageNaming Consistency Familiarity ConsistencySET 3(Mean score) (Mean score)AXE 90 2.28 4.50BUS 100 4.50 4.08DUCK 95 2.75 3.85FRIDGE 93 4.68 3.85HAIR 90 4.59 2.71LEMON 100 3.25 4.35LIPS 93 4.50 4.10OWL 100 2.22 4.10PIPE 98 2.90 4.26SWEATER 83 4.48 2.78TRUCK 90 4.02 2.80VASE 95 2.78 2.72AVERAGE 93.9% 3.58 3.68SET 4BIKE 88 3.78 3.40BROOM 100 3.42 4.35CAR 81 4.70 3.10COAT 89 3.88 2.59FOOTBALL 100 3.55 4.18FROG 100 2.48 3.60HAT 98 3.18 3.65LOCK 88 3.18 3.51RABBIT 100 2.95 4.20ROLLING PIN 71 2.22 4.44TOASTER 100 4.08 3.92UMBRELLA 100 3.95 3.92AVERAGE 92.9% 3.45 3.73236Picture StimulusSet 1The following pictures are from Snodgrass and Vanderwarts'(1980) norms. These pictures have been scanned into an AppleMacIntosh computer and subjected to a fragmentation algorithmprogram. This program produced a fragment series for eachpicture by cumulatively deleting randomly selected 16 X.16 pixelblocks (see Snodgrass & Corwin, 1988). The percentage of deletedblocks followed an exponential function. Refer to the next pagefor the amount of picture displayed at each fragment level.237LEVEL 8 (100% intact)^ LEVEL 7 (70% intact)I41243mo tif ti •LC■ •244L/^N(O•a /••••I )j• r247 a-r=r -J^1= =r -^ r1248•;4t-; 1...^ •-; it■..:,*^/1 I■ t '. •iI^ 4t1■1^ iIt^ I......:-)1^ -ladi/' AP/Picture StimulusSet 2250I251\A1101•1IC. .11.:A.....Os.h .........■^... 741 . .. i' l 252..-.• ..._I - 7h-13aroarms25341•1111MI,e■•""`•254•■••••••■•...•■•••ro.\\V4^255a'"-4 ••ea.4 ... :;4: • it46.1■ ,/ ' • ' . '^.*..-.A ••4t'; • • • 4- A11%24 . 44L^1 'I1 •'1•VIE Ella^j 259J•••■ ...•f^::.-•,^••M•M a a ath: I I,ArIt= Awls .-_- Jiffl l EoriT `i= 4.1,7 j-is is =311.*•••=1•MUM^ • • \amea•• Nol. MEM•I■•■■■•1111111M• As•r" •260•••.00•.0"/ a. 4 •J, •••'../ • •• 1 •k■ I k■,^ ,Ik ■■^/ ■ It.kJ\ ' I /^^/J .or, "16^ A-.4.^ A.Picture StimulusSet 3263265aIV •l)Picture StimulusSet 4276288V1289Object Materials Items were selected because they were common articles thatweren't sharp on touch and could be placed behind a 46 cm by 20.5cm partition. Refer to the method section for the procedure usedto select and administer these items.SET 1^ SET 2 BOOT BANDAIDCLOTHESPIN BELTCUP^ BOOKELASTIC BAND^ BOWLFLAG CANDLEHAMMER DOLLHANGER^ FORKSCREW HATSOAP PENCILSOCK SCISSORSSPOON^ SCREWDRIVERTOOTHBRUSH THIMBLESET 3^ SET 4 BOW BALLBRUSH BASKETENVELOPE^ BOTTLEGLASS BOXGLOVE CARKEY GLASSESKNIFE^ LOCKNAILFILE PAINTBRUSHPIPE PAPERCLIPROLLERSKATE^ PENRULER^ PLIERSTOILET PAPER SHOEAppendix E290Test Booklet Used to Record ResponsesType^NameYears of education^Age^Occupation^AddressPhone NumberMedications291Results: CompletionPicture RecognitionLettersMeaningAuditory Category ProductionVisual Category ProductionTactile ProductionTactile RecognitionAuditory Category RecallVisual Category RecallSeeSee/SaySee/Say/DoExp/Sub.^2/. Picture Comp/Recog. 3/. Letters/MeaningExp. Implicit^T I lettersSub. meaningExplicit^TDAuditory/Visual Cat.^Product.ImplicitBTactile Ident/Recog.ImplicitDExplicitDAuditory/Visual Cat.^Recall.ExplicitB7/. Multi-Modal see ^see/saysee/say/do292EXPERIMENTER/SUBJECTProcedure:Instructions:STUDY:In this first task we're going to do some things together.I have some commands written on these cards, some of which I willdo, and some of which you will do. For some of the tasks we'llneed to use the items that are in front of us. When we'refinished I'm going to ask you to tell me all the things we didtogether. You don't have to remember who did what, just what wasdone. Any questions? Good, now for all of the tasks that you'llbe doing today we'll always start with some practice items so youunderstand what it is that I want you to do. Here are thepractice items for this task, I'll do the first one and I'll haveyou watch me do it. The first command is "raise your hands", nowwatch me as I do it. Good! Now you do the next one "put theglasses in the case" - Good - you've got the idea. Now we'll dothe following in the same way, but before we start I'd like you totell me the names of these items (point to each of the props andhave them tell you its name (i.e., elastic band, matches andmatchbox, paper, pen, thimble, toothpick).TEST:Now I'd like you to tell me all of the things that we just didtogether.1/ Free Recall 1)2)3)4)5)6)7)8)9)2/ Cue (physical items)^3/ Cue (words) PICTURE COMPLETION/RECOGNITION 293Procedure: A. Implicit/Explicit Targets^,^]B. Implicit Test (target b distracter) ]C. Explicit Test (target & distracter)]Instructions:STUDY:Now I'm going to show you some pictures that I'd like you tolook at. For each picture I want you to tell me what it is, andsomething about it. The more personal meaning you can tell meabout the picture the better. For example, this is a picture ofa ball and its the type of ball that I'd take to the beach. Thenext picture is an elephant, and the elephants are my favoriteanimal in the zoo. You do the next two (airplane, ashtray).That's good. Now I want you to do the same thing for these nextpictures. (show the 24 pictures constituting implicit andexplicit targets).IMPLICIT TEST:Now we're going to do a different task. I'm going to showyou some pictures that vary in their amount of completeness and Iwant you to tell me when you think you know what the picture is.(Now show them the elephant practice item and illustrate how thepicture starts out with not very much information and graduallyincreases in the amount of completeness. Tell them that some ofthe items will be more difficult than others and that it isimportant that they clear their minds and tell you whatever theythink the picture is. Tell them that this is a new task and thatit is important that they tell you whatever first pops to mindwithout thinking back to anything done previously). Record allresponses and level. When they've successfully identified itemproceed to the next level to ensure confidence of response.If they appear to be trying to remember the pictures at studydiscourage them by repeating instructions (i.e. it is importantthat you don't try to think of anything we've previously done.Just tell me what first comes to mind).EXPLICIT TEST:• Now I'm going to show you some more pictures, but this timeall I want you to do is say yes, if this was one of the picturesI've just shown you, or no if you don't remember seeing thispicture before.Implicit Test Explicit TestItem294Level^ Item^ItPL0121) 1)2) 2)3) 3)4) 4)5) 5)6) 6)7) 7)8) 8)9) 9)10) 10)11) 11)12) 12)13) 13)14) 14)15) 15)16) 16)17) 17)18) 18)19) 19)20) 20)21) 21)22) 22)23) 23)24) 24)295]]LETTERS/MEANINGProcedure:^fletters [meaning [Instructions:STUDY:For this task I'm going to show you some cards that havedifferent words written on them. For all of the words I'd likeyou to first tell me what the word is. Then, I'll either ask youto count the number of letters in the word, or I'll ask you totell me what that word means to you. When we're finished I'll askyou to tell me the words that were written on the cards. Nowlet's try some practice cards- (Bus- please tell me the # ofletters this word has. That's good. Now the next word is-showcar- I'd like you to tell me what meaning this word has for you).That's good, now I want you to do the same thing for these nextcards.TEST:1/ Words recalled^2/ Recognition. 1)2)3)4)5)6)7)8)9)10)11)12)296CATEGORY PRODUCTION/RECOGNITIONProcedure:^IMPLICIT CONDITIONA. Auditory/Visual Targets^,B. Auditory/Visual Test^( .Instructions:STUDY:For this task I'm either going to say some words, or showyou some that are written on cards. For each word that ispresented, I'd like you to first repeat the word, and then I'dlike you to tell what that word means to you. For example, if Isaid the word car, what would you say. That's good, or you couldhave also said something like, Car, I bought my first car when Iwas sixteen. If I showed you a card with the word cow written onit, what would you say ? Good! Now I'd like you to do the samething for these words.(present setf auditorally, and^visually, total=24 words).TEST:Now I'm going to show you cards that have differentcategories written on them, and I'd like you to tell me the first three words that are members of that category that come to mind when you see that category. Here's the first category, what three words come to mind? Category^Items (3) 1)2)3)4)5)6)7)8)13)14)15)16)297TACTILE IDENTIFICATION/RECALLProcedure:^A. Implicit/Explicit Targets [ , ]B. Implicit Test (target & distracter)C. Explicit Test (target E. distracter)Instructions:STUDY:Now I'd like you to put both of your hands behind thiscurtain - That's Good-. I'm going to place an object in yourhands and I'd like you to tell me what it is, and something aboutit. For example, let's practice with this item, tell me what itis, and something about it, [practice=loonie, packet of kleenex].(Present set # for implicit and # for explicit target,total=24items).IMPLICIT TEST:Now I'm going to place some more objects in your hands and I'dlike you to tell me what you think it is. I'll be timing you withthis stopwatch so I'd like you to tell me what the item is as soonas you know. After you've identified the object then haveyou tell me something about it or what you'd use it for. (Presentimplicit target and distracter #'es b , total=74).EXPLICIT TEST:Now I'm going to place some more objects in your hands andI'd like you to answer yes, if this was an item I gave you to feelbefore, or no, if you don't remember feeling this item before.(Present explicit target and distracter res , total=24).ITEMS:^ 298SET1: boot, clothespin, cup, elastic band, flag, hammer, hanger,screw, soap, sock, spoon, toothbrush.SET2: bandaid, belt, book, bowl, candle, doll, fork, hat, pencil,scissors, screwdriver,. thimble.SET3: bow, brush, envelope, glass, glove, key, knife, nailfile,pipe, rollerskate, ruler, toilet paper.SET4: ball, basket, bottle, box, car, glasses, lock, paintbrush,paperclip,^pen,^pliers,Implicit Testshoe.Explicit TestItem^Response 1^Response 2 Item^Yes/No1) 1)2) 2)3) 3)4) 4)5) 5)6) 6)7) 7)8) 8)9) 9)10) 10)11) 11)12) 12)13) 13)14) 14)15) 15)16) 16)17) 17)18) 18)19) 19)20) 20)21) 21)22) 22)23) 23)24) 24)Procedure: EXPLICIT CONDITIONA. Auditory/Visual TargetsB. Auditory/Visual Test [ ,299Instructions:STUDY: This task is similar to one that we did before. I'm eithergoing to say some words, or show you some that are written on cards . Like before I'd like you to first repeat the word andthen tell me something about that word. For example, if I saidthe word car you'd say, car, and then you might say something likeI bought my first car when I was sixteen. Any questions?TEST: Now I'm going to show you cards that have differentcategories written on them. First I'd like you to tell me ifyou've just studied any words that are members of that category,and then I'd like you to tell me what three words you just studiedthat belong to this category.Category^Items (3) 1)2)3)4)5)6)7)8)9)10)11)12)13)14)15)16)MULTISENSORY EXPERIMENT^300Procedure:^see [ ]see/say [ ]see/say/do [ ] .Instructions:STUDY:In this task I'm going to show you some cards - some of whichI will ask you to read silently to yourself, some of which I willask you to read out loud to me, and some that I'll have youread out loud to me and then do what the card says. For this taskyou may need to use the items you see in front of you (have themidentify items as in first task). Do you have any questions? -Good- now let's try a practice set. For this card I'd like you toread it silently to yourself (practice #1), I'd like you to readthis card out loud to me (practice #2), and for this card I'd likeyou to read it out loud to me and then do what it says (practice#3). Do you have any questions? Good, now let's begin. Whenwe're finished I'm going to ask you to remember as many of theitems as you can.1/ Free Recall 1)2)3)4)5)6)7)8)9)10)11)12)2/ Cue (physical items)^3/ Cue (words)Appendix FMaterials and Cues Used in Memory Strategy ExperimentsTable of Contents (1) Levels of Processing Experiment(2) SPT-EPT Experiment (i.e., Subject-Experimenter PerformedTasks)(3) Multisensory Experiment301302Memory Strategy MaterialsLevels of Processing Experiment Words contained in material sets 1 and 2 are target itemsthat were selected because they had similar numbers of lettersand vowels. Depending on the set, words were either studied forthe number of letters or personal meaning they had for thesubject. Sets 1 and 2 were counterbalanced across letter andmeaning conditions for control and patient groups.The distractor set is the nine words that were used in therecognition part of this experiment. Distractor items werechosen because they either had similar numbers of letters, vowelsor meaning to target items. Refer to the method section for theprocedures used to select and administer these items.SET 1^ SET 2 TIN PIGJANE PINERIFLE^ LEMONVIOLET BEETLESWEATER CEILINGKEROSINE VIRGINIADISTRACTOR SETJILLBILLDOGRAINBIKESWINGPURPLESPIDERCLOTHINGSPT-EPT ExperimentCommands contained in material sets 1 and 2 were eitherperformed by the subject or the experimenter. Command sets werecounterbalanced across subject and experimenter conditions sothat one subject would perform items in set 1 and watch theexperimenter perform set 2 and the next subject would perform set2 and watch set 1 performed etc. Refer to the method section forthe procedures used to select and administer these items.303SET^ SET2a) BLINK THREE TIMESb) POINT TO YOUR MOUTHc) CLAP YOUR HANDSd) BREAK THE TOOTHPICKe) PUT THE THIMBLE ON YOUR FINGERf) STRETCH THE ELASTIC BANDg) NOD YOUR HEADh) CROSS YOUR FINGERSi) SCRATCH YOUR NOSEj) PUT THE MATCH IN THE BOXk) DRAW A CIRCLE WITH THE PEN1) KNOCK ON THE TABLEOBJECTS USED TO PERFORM THESE MINITASKSELASTIC BANDMATCHES AND BOXPAPER AND PENFINGER THIMBLETOOTHPICKSCUES USED IN CUED RECALL CONDITION(e.g., cue a is for item a above)a) ONE OF THE COMMANDSWAS?b) ONE OF THE COMMANDSREMEMBER WHAT THATc) ONE OF THE COMMANDSWHAT THAT WAS?d) ONE OF THE COMMANDSWHAT THAT WAS?e) ONE OF THE COMMANDSREMEMBER WHAT THATf) ONE OF THE COMMANDSWHAT THAT WAS?g) ONE OF THE COMMANDSWHAT THAT WAS?h) ONE OF THE COMMANDSWHAT THAT WAS?i) ONE OF THE COMMANDSWHAT THAT WAS?j) ONE OF THE COMMANDSREMEMBER WHERE THATk) ONE OF THE COMMANDSWHAT THAT WAS?1) ONE OF THE COMMANDSREMEMBER WHERE THATSAID TO BLINK, DO YOU REMEMBER WHAT THATSAID TO NOD SOMETHING, DO YOU REMEMBERSAID TO CROSS SOMETHING, DO YOU REMEMBERSAID TO SCRATCH SOMETHING, DO YOU REMEMBERSAID TO PUT THE MATCH SOMEWHERE, DO YOUWAS?SAID TO DRAW SOMETHING, DO YOU REMEMBERSAID TO KNOCK ON SOMETHING, DO YOUWAS?SAID TO POINT TO SOMETHING, DO YOUWAS?SAID TO CLAP SOMETHING, DO YOU REMEMBERSAID TO BREAK SOMETHING, DO YOU REMEMBERSAID TO PUT THE THIMBLE SOMEWHERE, DO YOUWAS?SAID TO STRETCH SOMETHING, DO YOU REMEMBER304Multisensory Experiment Commands contained in material sets 1, 2, and 3 were eitherperformed by silently reading the item, reading the command tothe experimenter, or reading the item aloud and performing thecommand. Similar to previous tests these material sets werecounterbalanced across performance conditions. Refer to themethod section for the procedures used to select and administerthese items.SET1 a) TOUCH YOUR NOSEb) MAKE A SIGN FOR GOODBYEc) PUT THE PENCIL IN THE CASEd) MAKE THE CLOCK SAY 3 O'CLOCKSET2 a) TOUCH YOUR EARb) MAKE A SIGN FOR YESc) PUT THE PEN IN THE DRAWERd) MAKE THE CLOCK SAY 6 O'CLOCKSET3 a) TOUCH YOUR MOUTHb) MAKE A SIGN FOR NOc) PUT THE SCISSORS IN THE BOXd) MAKE THE CLOCK SAY 9 O'CLOCKOBJECTS USED TO PERFORM THESE SENTENCESBOXCLOCKCUPPENPENCILPENCIL CASESCISSORSCUES USED IN CUED RECALL CONDITION(e.g., cue a is for sentences a above)a) ONE SENTENCE SAID TO TOUCH SOMETHING, DO YOU REMEMBER WHATTHAT WAS?b) ONE SENTENCE SAID TO MAKE A SIGN FOR SOMETHING, DO YOUREMEMBER WHAT THAT WAS?C) ONE SENTENCE SAID TO PUT SOMETHING SOMEWHERE, DO YOU REMEMBERWHAT THAT WAS?d) ONE SENTENCE SAID TO MAKE THE CLOCK SAY SOMETHING, DO YOUREMEMBER WHAT THAT WAS?Appendix GTables 305306Table G -8aAD Patients* and Control Subjects' Performance on Category CuedRecall and Category Completion Tests for Written Words. *Mild,Moderate, and Severe indicates level of Functional Impairment.AD PatientsControls^All^Mild* Moderate* Severe*(n=40) 01=20) (n=6) (n=7) (n=7)EXPLICIT MEMORY/1.85(2.28)15.424.50(1.98)37.500.43(0.79)3.581.00(1.52)8.33Category CuedRecallCorrected Score8.68SD^(1.83)% Score^72.33TargetM^8.95 2.05 4.83 0.43 1.295D (1.75) (2.50) (2.22) (0.79) (1.89)% Score^74.58 17.08 40.25 3.58 10.75BaselineM^0.27 0.20 0.33 0.00 0.29D (0.85) (0.62) (0.82) (0.00) (0.76)% Score^2.25 1.67 2.75 0.00 2.42IMPLICIT MEMORY/Category CompletionPriming or Corrected ScoreM^1.62 1.30 1.33 1.28 1.2852 (2.39) (2.20) (3.39) (1.38) (1.98)% Score^13.50 10.83 11.08 10.67 10.67Target M^3.80 2.45 2.83 2.14 2.42ED (2.36) (2.16) (2.63) (1.46) (2.57)% Score^31.67 20.42 23.58 17.83 20.17BaselineM^2.18 1.15 1.50 0.86 1.14SD (1.11) (1.18) (1.38) (0.69) (1.46)% Score^18.17 9.58 12.50 7.17 9.50307Table G-8bTwo Factor Repeated Measures MANCOVA* with Subjects as theBetween Factor (AD Patients, Control Subjects) and Test Type(Explicit, Implicit) as the Within Factor for Written WordMaterials.Source of Variation df Mean SquareHomogeneity TestsBox M 3,34728 1.13 .334Bartlett-Box(Explicit) 1,^7249 1.24 .265(Implicit) 1,^7249 .16 .681Between Factor EffectsWithin cells 57 5.99Regression 1 .87 .15 .704Constant 1 82.58 13.78 .001Group 1 306.47 51.14 .001Within Factor EffectsWithin cells 58 3.49Test Type 1 385.07 110.33 .001Group by Test Type 1 281.67 80.70 .001Main Effects TestsExplicit Test by Group 1,58 621.07 157.08 .001Implicit Test by Group 1,58 1.41 0.26 .613*Note. Subjects' level of education is the covariate, CategoryCued Recall is the explicit, and Category Completion is theimplicit memory test.308Table G-9aAD Patients* and Control Subjects' Performance on Category CuedRecall and Category Completion Tests for Spoken Words. *Mild,Moderate, and Severe indicates level of Functional Impairment.AD PatientsControls^All^Mild* Moderate* Severe*EXPLICIT MEMORY/(n=40) (n=20) (n=6) (n=7) (n=7)Category CuedRecallCorrected Score8.40 2.45 4.17 1.14 2.28SD (2.67) (2.56) (2.71) (1.86) (2.49)% Score 70.Q0 20.42 34.75 9.50 19.00TargetM 8.53 2.55 4.17 1.14 2.57SD (2.39) (2.63) (2.72) (1.86) (2.69)% Score 71.08 21.25 34.75 9.50 21.42BaselineM 0.13 0.10 0.00 0.00 0.295D (0.52) (0.45) (0.00) (0.00) (0.76)% Score 1.08 0.83 0.00 0.00 2.42IMPLICIT MEMORY/Category CompletionPriming or Corrected ScoreM 1.73 1.10 1.17 1.20 1.29SD (2.48) (1.55) (1.72) (1.64) (1.79)% Score 14.41 9.17 9.75 10.00 10.75TargetM 3.63 2.10 2.67 1.86 1.86an (2.32) (1.51) (1.21) (1.35) (1.95)% Score 30.25 17.50 22.25 15.50 15.50BaselineM 1.90 1.00 1.50 0.66 0.57SD (1.01) (0.92) (0.84) (0.82) (0.97)% Score 15.83 8.33 12.50 5.50 4.75309Table G-9bTwo Factor Repeated Measures MANCOVA* with Subjects as theBetween Factor (AD Patients, Control Subjects) and Test Type(Explicit, Implicit) as the Within Factor for Spoken WordMaterials.Source of Variation df Mean Square F pHomogeneity TestsBox M 3,34728 1.86 .134Bartlett-Box(Explicit) 1,^7249 .03 .844(Implicit) 1,^7249 4.82 .028Between Factor EffectsWithin cells 57 6.39Regression 1 15.62 2.45 .123Constant 1 40.51 6.34 .015Group 1 232.71 36.44 .001Within Factor EffectsWithin cells 58 5.32Test Type 1** 429.34 80.68 .001Group by Test Type 1** 189.04 35.52 .001Main Effects TestsExplicit Test by Group 1,58 472.03 68.01 .001Implicit Test by Group 1,58** 5.21 1.06 .308Note. *Subjects' level of education is the covariate, CategoryCued Recall is the explicit, and Category Completion is theimplicit memory test. **Box-adjusted values reported in theresults section.310Table G -10aAD patients and Control Subjects' Performance on PictureRecognition and Picture Fragment Completion Tests. *Mild,Moderate, and Severe indicates level of Functional Impairment.AD PatientsControls^All^Mild*^Moderate* Severe*(n=40)^(n=20) (n=6)^(n=7)^(n=7)EXPLICIT MEMORY/Picture Recognition (don't know and false negative responsesare not reported)Corrected Score 11.70(0.61)97.505.90(3.78)49.208.50(3.45)70.804.15(3.76)34.504.40(3.26)36.67XZ2% ScoreTargetX 11.85 9.45 11.33 8.29 7.97W. (0.36) (2.96) (1.21) (4.11) (2.08)% Score 98.75 78.75 94.42 69.08 66.40DistractorX 11.70 7.80 8.50 7.57 7.4322 (1.43) (4.13) (3.39) (5.03) (4.31)% Score 97.50 65.00 70.83 63.08 61.92False PositiveM 0.15 3.55 2.83 4.14 3.57= (0.53) (3.69) (2.92) (4.91) (3.31)% Score 1.25 29.58 23.58 34.50 29.75^  continuedIMPLICIT MEMORY/Test (max. score = 8)Picture Fragment CompletionPriming or Corrected ScoreN^1.05 0.49 0.46 0.32 0.48aa (0.65) (0.64) (0.41) (0.53) (0.51)% Score 13.12 6.10 5.75 4.00 6.00TargetM 4.71 5.92 5.91 6.23 6.325_2baseline(0.66) (0.91) (0.97) (0.34) (0.67)M 5.76 6.41 6.37 6.55 6.805..12 (0.57) (0.73) (1.03) (0.57) (1.23)311312Table G -10bTwo Factor Repeated Measures MANCOVA* with Subjects as theBetween Factor (AD Patients, Control Subjects) and Test Type(Explicit, Implicit) as the Within Factor for Picture Materials.Source of Variation df Mean SquareHomogeneity TestsBox M 3,34728 26.59 .001Bartlett-Box(Explicit) 1,^7249 80.14 .001(Implicit) 1,^7249 .27 .602Between Factor EffectsWithin cells 57 201.78Regression 1 .10 0.01 .983Constant 1 15429.02 76.46 .001Group 1 18233.71 90.36 .001Within Factor EffectsWithin cells 58 196.68Test Type 1** 105139.41 534.56 .001Group by Test Type 1** 11913.92 60.57 .001Main Effects TestsExplicit Test by Group 1,58** 31148.15 90.90 .001Implicit Test by Group 1,58 489.55 9.36 .003Note. *Subjects' level of education is the covariate, PictureRecognition is the explicit, and Picture Fragment Completion isthe implicit memory test. ** Box-adjusted values reported in theresults section.313Table G-llaAD patients and Control Subjects' Mean Performance on the TactileRecognition Test.AD PatientsControls^All^Mild^Moderate Severe(n=40)^(n=16) (n=6) (n=5)*^(n=5)*EXPLICIT MEMORY/Tactile Recognition (don't know and false negative responsesare not reported).Corrected ScoreM 11.67 5.90 8.83 6.14 3.14SD (0.94) (3.96) (2.93) (4.29) (2.54)% Score 97.25 49.17 73.58 51.17 26.17TargetM 11.82 8.25 11.00 8.00 6.14SD (0.59) (3.71) (1.09) (4.08) (3.62)% Score 98.50 68.75 91.67 66.67 51.17DistractorM 11.82 9.15 9.17 9.86 8.43SD (0.59) (2.93) (3.90) (1.95) (3.05)% Score 98.50 76.25 76.42 82.17 70.25False PositiveM 0.15 2.35 2.17 1.86 3.00SD (0.43) (2.25) (2.78) (1.57) (2.52)% Score 1.25 19.58 18.08 15.50 25.00^continued314Table G -11a continuedAD patients and Control Subjects' Mean Performance on the TactileIdentification Test.AD PatientsControls^All^Mild^Moderate Severe(n=40)^(n=16) (n=6)^(n=5)*^(n=5)* IMPLICIT MEMORY/Tactile Identification test(median time savings in milliseconds% score as a proportion of baseline performance)014 Materials Priming or Corrected Score MSD119.65(103.04)64.87**288.31(454.41)59.29159.33(65.53)74.62**175.80(202.40)50.05304.40(823.82)32.00% $coreTarget304.07 774.56 372.83 527.00 1253.00SD (127.11) (812.87)(112.49) (202.20) (1366.93)Baseline184.42 486.25 213.50 351.20 948.60SD (68.47) (691.34) (68.82) (74.84) (1174.39)New MaterialsPriming or Corrected ScoreM^136.40 64.13 158.83 104.20 -89.60an (84.80) (577.84) (89.93) (133.84) (1085.87)% Score 73.96** 13.19 74.39** 29.67 -9.45Target320.83 550.38 372.33 455.40 859.005D (127.11) (412.63)(125.55) (151.68) (646.44)Baseline184.43 486.25 213.50 351.20 948.60$D (68.47) (691.34) (68.82) (74.84) (1174.39)Note. *Four patients could not perform the tactile identificationtest and so their performance was excluded. **Patients tooklonger to perform this test and so there was a greater marginavailable for improvement.315Table G-libTwo Factor Repeated Measures MANCOVA* with Subjects*** as theBetween Factor (AD Patients, Control Subjects) and Test Type(Explicit, Implicit) as the Within Factor for Old Materials.Source of Variation df Mean Square pHomogeneity TestsBox M 3,15033 17.44 .001Bartlett-Box(Explicit) 1,^4959 54.52 .001(Implicit) 1,^4959 .01 .990Between Factor EffectsWithin cells 53 1989.20Regression 1 1829.05 .92 .342Constant 1 64667.98 32.51 .001Group 1 7908.01 3.98 .051Within Factor EffectsWithin cells 54 1776.48Test Type 1** 44.24 .02 .875Group by Test Type 1** 16976.41 9.56 .003Main Effects TestsExplicit Test by Group 1,54** 30880.21 77.55 .001Implicit Test by Group 1,54 1221.67 .36 .549Note. *Subjects' level of education is the covariate, TactileRecognition is the explicit, and Tactile Identification is theimplicit memory test. **Box-adjusted values reported in theresults section. ***Four subjects dropped from analysis.316Table G -11cTwo Factor Repeated Measures MANCOVA* with Subjects as theBetween Factor (AD Patients, Control Subjects) and Test Type(Explicit, Implicit) as the Within Factor for New Materials.Source of Variation^df^Mean Square F pHomogeneity TestsBox M^ 3,15033 18.75 .001Bartlett-Box(Explicit) 1, 4959 60.14 .001(Implicit) 1, 4959 .03 .844Between Factor EffectsWithin cells^53^2454.61Regression 1 1669.13 .68 .542Constant 1^85935.89 35.01 .001Group^ 1 35444.57 14.44 .001Within Factor EffectsWithin cells^54^28971.84Test Type 1** 2337448.05 80.68 .001Group by Test Type^1** 1029079.76 35.52 .001Main Effects TestsExplicit Test by Group 1,54**^30880.21 77.55 .001Implicit Test by Group 1,54^8470.10 9.36 .003Note. *Subjects' level of education is the covariate, ObjectRecognition is the explicit, and Object Identification is theimplicit memory test. **Box-adjusted values reported in theresults section.317Table G-12aAD Patients and Controls' Performance in the Levels of ProcessingExpe i ent.  AD Patients Mild^Moderate SevereControls AllScore (maximum=6)Free Recall(n=401 (n=20) (n=6) (n=7) (n=7)Meaning EncodeM 3.88 1.65 2.33 1.86 0.86a2 (1.31) (1.69) (1.86) (1.95) (1.07)% Corrected 64.67 27.50 38.83 31.00 14.33Letter EncodeM 2.45 0.70 0.16 1.43 0.43ail (1.32) (1.26) (0.41) (1.90) (0.54)% Corrected 40.83 11.67 2.67 23.83 7.17Average Free Recall Score (%)52.75^19.58 20.75 27,42 10.75RecognitionMeaning EncodeM 2.08 2.90 3.33 2.29 3.57SD (1.27) (1.80) (1.63) (2.63) (1.81)% Score 34.67 48.33 55.50 38.17 59.50% Corrected 98.11 66.67 90.70 55.31 69.46Letter EncodeM 3.08 3.30 4.50 1.89 3.29SD (1.47) (2.36) (1.64) (1.68) (2.43)% Score 51.33 55.00 75.00 31.50 54.83% Corrected 86.76 62.26 77.05 41.36 59.06Average Cued Recall Score (%)92.43 64.46 83.88 48.34 64.26Note. Average Free Recall and Recognition scores are derived fromdifferent unit values and therefore do not sum to 100%.318Table G-12b^Levels of Processing ExperimentThree Factor Repeated Measures MANCOVA* with Subjects as theBetween Factor (AD Patients, Control Subjects) and Two WithinSubject Factors (Study Task, Test Type) each with Two Levels(Meaning & Letter Encode, Free Recall and Recognition).Source of Variation^df^Mean Square FHomogeneity TestsBox M^ 10,6902 7.98 .001Bartlett-Boxmeaning/free recall^1,7249 1.80 .180meaning/recognition 1,7249 70.74 .001letter/free recall^1,7249 .05 .821letter/recognition 1,7249 8.05 .005Between Factor EffectsWithin Cells^57^1060.73Regression 1 38.98 .04 .849Constant 1^62688.97 59.10 .001Group^ 1^42093.20 39.68 .001Within Factor Effects-Study TaskWithin Cells^58^341.06Study Task 1^13546.87 39.72 .001Group by Study Task^1 175.21 .51 .476Within Factor Effects-Test TypeWithin Cells^58^627.29Test Type 1** 103840.83 165.54 .001Group by Test Type^1**^925.93 1.48 .229Within Factor Effects-Study Task by Test TypeWithin Cells^58^364.72Study Task by Test Type^1**^792.25 2.17 .146Group by Study Task by^1**^245.58 .67 .415Test TypeNote. *Subjects' level of education is the covariate and onedegree of freedom is lost from the within cells source ofvariations as a result. **Box-adjusted values are reported inthe results section.319Table G-13aAD Patients and Controls' Performance in the SPT-EPT MemoryStrateay Experiment. AD PatientsControls^All^Mild^Moderate SevereLn=40)^(n=20) (n=6)^(n=7)^(n=7) Score (maximum=6)Free Recall SPT.Performed M^4.20^0.75^1.50^0.43^0.43aa (1.29)^(0.91) (1.05)^(0.54)^(0.79)% Corrected 70.00 12.50^25.00 7.15 7.16 EPT.Performed M^3.70^0.75^2.17^0.14^0.14aa (1.52)^(1.21)^(1.33)^(0.38) (0.38)% Corrected 61.67 12,50 36.17 2.38^2.33 Average Free Recall Score (%) 65.83^12.50^30.58^4.75^4.75Cued Recall SPT.PerformedM 1.43 2.95 3.33 2.86 2.71SD (1.08) (1.61) (1.03) (2.12) (1.60)% Score 23.83 49.17 55.50 47.67 45.17% Corrected 79.44 56.19 74.00 51.35 48.65EPT.PerformedM 1.58 2.45 1.83 2.43 3.00SD (1.24) (1.47) (1.17) (1.51) (1.63)% Score 26.33 40.83 30.50 40.50 50.00Corrected 68.69 46.67 47.78 41.47 51.19Average Free Recall Score (%) 74.06^51.43^60.89^46.41^49.92Note. SPT = subject performed, EPT = experimenter performedminitask. Average Free and Cued Recall scores are derived fromdifferent unit values and therefore do not sum to 100%.320Table G-13b^SPT-EPT ExperimentThree Factor Repeated Measures MANCOVA* with Subjects as theBetween Factor (AD Patients, Control Subjects) and Two WithinSubject Factors (Study Task, Test Type) each with Two Levels(Subject-Performed & Experimenter-Performed,Recall).Free and CuedSource of Variation^df Mean Square FHomogeneity TestsBox M^ 10,6902 1.19 .286Bartlett-BoxSPT/free recall^1,7249 2.71 .100SPT/cued recall 1,7249 .01 .922EPT/free recall^1,7249 1.26 .262EPT/cued recall 1,7249between Factor Effects.90 .342Within Cells^57 1347.75Regression 1 140.10 .10 .748Constant 1 44031.65 32.67 .001Group^ 1 79227.64 58.79 .001Within Factor Effects-Study TaskWithin Cells^58^398.64Study Task 1 3325.02 8.34 .005Group by Study Task^1 350.21 .88 .352Within Factor Effects-Test TypeWithin Cells^58 550.01Test Type 1 38940.02 70.80 .003.Group by Test Type^1 8755.21 15.92 .001Main Effects TestsAD PatientsSPT vs EPT Cued Recall^1,19 31733.89 63.15 .001SPT vs EPT Free Recall^1,19 568.89 1.26 .275SPT Free vs. Cued Recall 1,19 20400.28 49.19 .001EPT Free vs. Cued Recall 1,19 11902.50 22.09 .001^ continued321Table G-13b continued^SPT-EPT ExperimentThree Factor Repeated Measures MANCOVA* with Subjects as theBetween Factor (AD Patients, Control Subjects) and Two WithinSubject Factors (Study Task, Test Type) each with Two Levels(Subject-Performed & Experimenter-Performed, Free and CuedRecall).Source of Variation^df^Mean Square FMain Effects TestsControl Subjects SPT vs EPT Cued Recall 1,39 8075.07 14.09 .001SPT vs EPT Free Recall 1,39 4375.07 9.01 .005SPT Free vs. Cued Recall 1,39 5335.56 12.47 .001EPT Free vs. Cued Recall 1,39 2920.14 4.45 .041Within Factor Effects-Study Task by Test TypeWithin Cells^58^491.08Study Task by Test Type 1 741.69 1.51 .224Group by Study Task byTest Type 1 137.25 .28 .599322Table G-14aAD Patients and Controls' Performance in the MultisensoryExperiment (i.e., Do, Say, and See conditions).AD Patients Controls^All^Mild^Moderate SevereScore^(maximum=4)Free Recall(n=40) (n=20) (n=6) (n=7) (n=7)2.90(0.96)72.500.65(0.99)16.251.50(1.05)37.500.29(0.76)7.250.29(0.76)7.25DoMaa% CorrectedSayM 1.68 0.25 0.67 0.14 0.00an (1.21) (0.72) (1.21) (0.38) (0.00)% Corrected 42.00 6.25 16.75 3.50 0.00SeeM 1.38 0.20 0.33 0.14 0.14SD (1.13) (0.52) (0.82) (0.38) (0.38)% Corrected 34.50 5.00 8.25 3.50 3.50Average Free Recall Score (%) 20.83 4.75 3.58Cued Recall49.67 9.170.75(0.74)18.7568.181.50(1.28)37.5044.771.33(0.82)33.2553.201.86(1.57)46.5050.131.29(1.38)32.2534.77DoMSD% Score% CorrectedSay.,0.85 0.45 1.00 0.29 0.14MSD (0.92) (0.76) (1.09) (0.49) (0.38)% Score 21.25 11.25 25.00 7.25 3.50% Corrected 36.63 12.00 30.03 30.03 7.51SeeM 1.08 0.45 0.67 0.29 0.43SD (0.79) (0.89) (1.21) (0.76) (0.79)% Score 27.00 11.25 16.68 7.25 10.75% Corrected 41.22 11.84 18.26 7.51 11.14Average Cued Recall Score (%)44.37^22.02 31.58 21.35 4.24Note. Average Free and Cued Recall scores are derived fromdifferent unit values and therefore do not sum to 100%.323Table G-14b^Multisensory Experiment Three Factor Repeated Measures MANCOVA* with Subjects as theBetween Factor (AD Patients, Control Subjects) and Two WithinSubject Factors (Study Task, Test Type).^Study Task has ThreeLevels (Do, Say, and See) and Test Type has two Levels (Free andCued Recall).Source of Variation df^Mean Square FHomogeneity TestsBox M 21,^5551 3.05 .001Bartlett-Boxsee/free recall 1,7249 11.67 .001see/cued recall 1,7249 4.47 .035say/free recall 1,7249 5.87 .015say/cued recall 1,7249 2.08 .149do/free recall 1,7249 .03 .866do/cued recall 1,7249 .50 .478Between Factor EffectsWithin Cells 57^1927.50Regression 1 2314.78 1.20 .278Constant 1^17218.98 8.93 .004Group 1^91084.47 47.26 .001Within Factor Effects-Study TaskWithin Cells^116^856.86Study Task 2**^24390.05 28.46 .001Group by Study Task 2**^1720.29 2.01 .139Within Factor Effects-Test TypeWithin Cells^58^886.28Test Type 1**^8336.81 9.41 .003Group by Test Type* 1^1167.05 1.32 .256Within Factor _Effects-Study Task by Test TypeWithin Cells 116 560.40Study Task by Test Type 2** 542.82 .97 .383Group by Study Task by 2** 1266.59 2.26 .109Test Type*Note. * Subjects' level of education is the covariate and onedegree of freedom is lost from the within cells source ofvariation. **Box adjusted values reported in results section.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items