UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Semantic memory in Alzheimer's Disease Bakerink, Ronda Ann 1988

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

Item Metadata

Download

Media
831-UBC_1988_A6_7 B34.pdf [ 4.7MB ]
Metadata
JSON: 831-1.0097650.json
JSON-LD: 831-1.0097650-ld.json
RDF/XML (Pretty): 831-1.0097650-rdf.xml
RDF/JSON: 831-1.0097650-rdf.json
Turtle: 831-1.0097650-turtle.txt
N-Triples: 831-1.0097650-rdf-ntriples.txt
Original Record: 831-1.0097650-source.json
Full Text
831-1.0097650-fulltext.txt
Citation
831-1.0097650.ris

Full Text

SEMANTIC M E M O R Y IN ALZHEIMER'S  DISEASE  By RONDA ANN BAKERINK B.A., S i m o n Fraser University, 1986  A THESIS S U B M I T T E D IN P A R T I A L F U L F I L M E N T O F THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF SCIENCE  in T H E F A C U L T Y OF G R A D U A T E STUDIES S c h o o l of A u d i o l o g y and Speech Sciences  W e accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH C O L U M B I A September 1988 © R o n d a A n n Bakerink, 1988  In  presenting  degree  this  at the  freely available copying  of  department publication  of  in  partial  fulfilment  of  University of  British  Columbia,  I agree  for reference and study.  this or  thesis  thesis by  this  for scholarly  his thesis  or  her  the  purposes  may  representatives.  It  DE-6(3/81)  be is  an  advanced  that the Library shall make it  granted  for extensive  by the head  understood  that  of  my  copying  or  for financial gain shall not be allowed without my written  Audiology and Speech Sciences  The University of British Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3  Date  for  I further agree that permission  permission.  Department of  requirements  September 21, 1988  ii  ABSTRACT  A l z h e i m e r ' s Disease is characterized by a general decline i n cognitive  fanctioning.  Although  phonology are relatively unaffected, patients with A l z h e i m e r ' s Disease have been reported to have deficits of semantic memory. Thirteen patients w i t h dementia, five of w h o m had a confirmed diagnosis of dementia, participated i n the study. The purpose of this investigation was to replicate a study performed by M a r k B y r d (1984), using A l z h e i m e r ' s Disease patients. Subjects were presented w i t h category-word decision pairs, for w h i c h the task was to decide i f the w o r d was an exemplar of the category, and category-letter decision pairs for w h i c h the task was to generate an exemplar of the category beguining w i t h the letter. The dependent variable was reaction time. Results indicated that A l z h e i m e r ' s Disease patients and dementia patients had longer reaction times than a group of age-matched control subjects, and that the A l z h e i m e r ' s Disease and dementia patients showed a pattern of responses similar to that of the control subjects. A l l groups showed longer reaction times for the generation trials than the decision trials. The results are consistent w i t h the existence of a semantic memory deficit i n A l z h e i m e r ' s Disease, but other interpretations were discussed.  iii T A B L E OF CONTENTS Page ABSTRACT  ii  T A B L E OF CONTENTS  iii  LIST OF T A B L E S  v  LIST OF FIGURES  '  ACKNOWLEDGEMENT  CHAPTER 1 INTRODUCTION  vi vii  1  1.1 Semantic M e m o r y  1  1.2 A l z h e i m e r ' s Disease  3  CHAPTER 2 LITERATURE REVIEW  10  2.1 A l z h e i m e r ' s Disease and Language  10  2.2 A g i n g and Semantic M e m o r y  37  CHAPTER 3 OBJECTIVES  50  3.1 O r i g i n a l Objectives  50  3.2 R e v i s e d Objectives  53  CHAPTER 4 METHODS  55  4.1 Subjects  55  4.11 P i l o t Experiment  55  4.12 Experimental Subjects  55  4.13 Dementia Patients  56  4.2 S t i m u l i  57  4.3 Procedure  58  CHAPTER 5 RESULTS  60  5.1 Pilot Experiment  60  5.2 M a i n Experiment  61  5.3 Dementia Patients  64  CHAPTER 6 DISCUSSION  69  iv NOTES  71  BIBLIOGRAPHY  72  APPENDIX 1 SUBJECT INFORMATION  74  1 A P i l o t Subjects  74  I B A D Patients  74  1 C Dementia Patients  75  A P P E N D I X 2 STIMULUS PAIRS  76  A P P E N D I X 3 T R I A L LISTS  77  APPENDIX 4 INSTRUCTIONS  82  V  LIST OF T A B L E S Table  Page  I  A D Patients Data  62  LT  Dementia Patients Data  67  vi LIST OF FIGURES Figure  Page  1  Name Matching Task  16  2  Vocabulary Tests  26  3  Word Association Task  29  4  Experimental Tasks  52  5  Mean Reaction Times for A D Patients and Control Subjects - Generation Task  6  65  Mean Reaction Times for A D Patients and Control Subjects - Decision Task  66  vii ACKNOWLEDGEMENT I would like to express my thanks to all of those who contributed to this thesis. In particular, I would to thank: -John H.V. Gilbert, Ph.D. for his advice and encouragementthroughout my course of work on this project. -Dr. L. Beattie, Robyn Lawrence, and the staff of the U B C Health Sciences Centre Hospital Alzheimer's Clinic for providing and scheduling subjects. -Dr. S Holliday and the clinical research staff of Valleyview Hospital for providing subjects and allowing me to use their facilities. -Kathy Fuller for running the control subjects and providing me with data. -Hari Garududri, Ph.D. for developing the computer program. -Barbara Purves for reviewing the manuscript. -My friends and family for their unfailing support and understanding, particularly my Mother and my sister Amy.  1  CHAPTER 1 INTRODUCTION  1.1 Semantic Memory  In order to investigate the effects of A l z h e i m e r ' s Disease o n semantic memory, it is necessary to clearly state the nature o f semantic memory. M a s u r (1986) defined semantic memory ( S M ) as "...that aspect of memory concerned w i t h the representation and organization of w o r d meanings." (p. 1305-B) W h i l e this definition is not incorrect, other authors have described S M i n more detail. Ober, Dronkers, K o s s , Delis and Friedland (1986) defined S M as "...the associative network of permanent knowledge about the w o r l d w h i c h has been built up over one's lifetime.'' (p. 76) S M is more than memory f o r w o r d meanings. It is the store o f a person's  w o r l d knowledge, w h i c h  includes w o r d meanings. Perhaps the most beneficial w a y i n w h i c h to define S M is to contrast it w i t h episodic memory ( E M ) . Bayles (1987) conceived of episodic memory and semantic memory as t w o components of long-term memory. Nebes, M a r t i n and H o r n (1984) described episodic memory as a record of an i n d i v i d u a l ' s experiences. In E M , episodes are encoded i n relation to a temporal-spatial context. S M , on the other hand, is a context-free store o f knowledge about concepts, their associations and organization. Nebes et al. illustrated the distinction between S M and E M as follows: "...a person's recollection of seeing a canary i n a shop w i n d o w the preceding week, or o f hearing the word C A N A R Y among a list of 20 words given an hour earlier i n a memory study, involves episodic memory. B y contrast, the knowledge that a canary is a s m a l l y e l l o w b i r d often kept as a pet, or that the word C A N A R Y begins with the letter C and has three syllables, involves semantic memory...."1 Hannigan, Shelton, Franks and Bransford (1980) also contrasted E M and S M . They stated that EM  and S M  are t w o ways i n which past experience can influence performance.  Influences  attributable to a person's having experienced an event, those w h i c h are tied to a particular context,  2 are episodic. Irifluences w h i c h are not attributable to a particular event are semantic. Hannigan et al. claimed that certain types of acts require different types of memory. R e c a l l and recognition of previously presented material require episodic memory. Inferring and generalizing f r o m presented material require semantic memory. M o s t tasks w h i c h are used to study memory require one or the other type of memory, but not both. A recall task may be used to test episodic memory, while a verification task may be used to test semantic memory. The hypothesis of Hannigan et al. was that almost any cognitive act involves both semantic and episodic memory. B o t h semantic and episodic and semantic influences may affect the performance of  that act. Semantic  memory  affects recall performance  and  episodic  memory  affects the  performance of a verification task. Hannigan et al. found that both episodic and semantic memory influenced the identification of masked verbal stimuli. Thus, although S M and E M are different aspects of long-term memory and although they can be  investigated relatively separately, the performance  of  most  cognitive  acts  (certainly the  performance of complex cognitive acts such as creating and understanding discourse) involves the interaction of episodic and semantic memory. Bayles (1987) defined semantic memory i n greater detail. F o l l o w i n g T u l v i n g , she stated that S M is a subcomponent of procedural memory. Procedural memory involves the knowledge of how to do things, plans and skills. S M , o n the other hand, involves factual knowledge of the world. Bayles also stated that the contents of S M are different f r o m external sensory information about the world. Sensory information is transformed and coded i n a way that is readable by S M  processors.  "Semantic memory is that domain i n the nervous system in which concepts are represented and inferential processing takes place.... Input f r o m the perceptual modalities is accessed, actively crossreferenced, indexed, and sorted i n SM.... it is a cognitive store i n w h i c h knowledge is coded in an abstract way independently of the input modality. It is the functional system i n w h i c h ideation and inferential processing consciously occur. " 2 Thus, semantic memory involves m u c h more than the meanings of words. It is the storage system for abstract context-free mental representations. The highest-level, most abstract cognitive modality is stored i n S M ; conceptual knowledge.  3  1.2 Alzheimer's Disease  Several years ago memory loss and senility were thought to be natural consequences of aging. Scientists, however, began to notice that some elderly people became senile w h i l e others retained a l l their mental faculties. In 1906 a German neurologist, A l o i s A l z h e i m e r , observed neurofibrillary tangles i n the brain of a 55-year-old demented woman. Because of the lack o f any other brain pathology, he concluded that her dementia was due to these formations (Bayles, 1987;  Semple,  S m i t h and Swash, 1982). A t the time that A l z h e i m e r published his findings there was m u c h disagreement about his conclusions. However, it is n o w generally agreed that A l z h e i m e r ' s Disease ( A D ) is an identifiable disease syndrome, and that the physiological changes first noted by A l z h e i m e r are considered to be the hallmarks of the disease. T h e brains of A D patient atrophy and develop neurofibrillary tangles, neuritic plaques and areas of granulovacuolar degeneration. Bayles (1987) reviewed the research regarding the neuropathology of A D  and summarized the information as follows. The total volume of the brain, including both  grey and white matter, decreases. There is a loss of neurons i n the frontal and temporal regions, and the number of dendritic branches per neuron decreases. This results i n a decrease i n the efficiency of the transfer of neuronal messages. Nebes and M a d d e n (1988) studied the reaction times o f A D patients and normal elderly subjects on a variety of tasks. They found that A D acts to slow most cognitive functions by a constant proportion. That is, the cognitive s l o w i n g associated w i t h A D showed a pattern similar to the pattern associated with normal aging. However, there were some cognitive functions for w h i c h the performance of A D Nebes and M a d d e n suggested that: "The dissirnilarity of the effects that n o r m a l aging and A D have on the speed of cognitive performance suggests that A D is neither a component of normal human aging, nor a simple exaggeration of the processes that occur i n normal aging. "3  patients was disproportionately slow.  4 Neurofibrillary tangles are the most c o m m o n physiological change that occurs during A D . Bayles (1987) has described them i n the f o l l o w i n g manner. Neurofibrillary tangles are filaments i n the c e l l body w h i c h come together i n a helical manner, running throughout the cytoplasm. They are thought to be caused b y abnormal protein synthesis o r a reaction to damage elsewhere i n the nervous system. Neurofibrillary tangles are found i n the hippocampus and the amygdaloid nucleus, as w e l l as i n the hypothalamus and throughout the neocortex. Because they are found i n the brains of normal elderly subjects, there has been debate about the differentiation o f normal aging and A D . Bayles  (1987) also summarized the information about neuritic plaques and granulovacuolar  degeneration. Neuritic plaques are groups o f degenerating neurons surrounding an a m y l o i d core. They are found i n the brain areas where nettfofibrillary tangles are found. T h e centre o f the plaques consist o f proteins o r protein antibodies. O n e theory regarding the cause o f neuritic plaques states that plaques f o r m when white blood cells ingest antigen-antibody complexes. Plaques occur i n larger numbers i n the brains o f A D patients than normal elderly subjects. Granulovacuolar  degeneration refers  to the presence  o f fluid  filled  spaces  and granular  degeneration i n the brains o f A D patients. These are also present i n normal elderly brains, but not to the extent to w h i c h they are present i n A D patients. Bayles  also described certain neurochemical changes associated w i t h A D . T h e cholinergic  system is deficient, the noradrenergic system i s abnormal, and there i s a reduction i n the number of neuropeptides. There are several theories about the cause o f A D (Bayles, 1987). T h e infectious  agent theory  has received little support. T h e suggestion that A D may be caused b y an unconventional virus has suffered f r o m a lack o f evidence regarding the  tfadmissibility  of A D . The  aluminum toxicity  theory has received more support. There have been reports o f elevated levels o f aluminum i n the brains o f A D patiemts at autopsy. T h e genetic transmission theory has also received support. Some investigators have found that f a m i l y members  o f A D patients have  a greater risk o f  developing the disease than individuals without a f a m i l y history o f A D . Despite these findings, there is n o one theory w h i c h is generally accepted. T h e cause o f A D remains unclear and much more research is needed.  5 Researchers have recently begun to discover physiologic characteristics associated with A D . Researchers have noted that patients suspected of having A D show abnormalities on the f o l l o w i n g measures:  electroencephalogram,  evoked  potentials,  computerized  tomography,  and  positron  emission tomography. So far these associations have proven to be unreliable. Thus, only at autopsy can  AD  be  diagnosed  accurately. A t  autopsy,  neurofibrillary tangles, neuritic plaques  and  granulovacuolar degeneration can be detected. The c l i n i c a l diagnosis  of A D  is based on the presence of certain behavioural and cognitive  characteristics, and on the absence of evidence suggesting other causes of dementia (e.g. multiple cerebral infarcts, Parkinson's  Disease, severe depression). The physiologic measures  above are used mainly to rule out other possible causes  mentioned  of dementia. Hier, Hagenlocker  and  Shindler (1985) report that c l i n i c a l diagnosis of A D is 8 2 % accurate. The most c o m m o n l y  reported symptom of A D  is memory  disorder (Bayles,  1987).  Other  c o m m o n symptoms are disorientation, personality changes, and general cognitive decline including impaired concentration, difficulties with reasoning and judgement and an inability to abstract the central idea f r o m a problem or discussion (Semple et al., 1982). The onset of A D  is insidious, thus it is difficult to pinpoint the earliest symptoms.  Most  patients exhibit disturbed mental function one to three years before they are brought i n for diagnosis (Semple et al.). E v e n one to three years after the assumed onset of the disease, most patients are still i n the early stages of A D . 1. M i l d A D  In the early stages of the disease, the f o l l o w i n g symptoms are present: impairment  of recent and remote memory; personality changes such as irritability, hostility or apathy; impaired inferential ability; disorientation to place. The motor system is generally unimpaired, however (Bayles, 1987). 2. Moderate A D present:  global  judgement;  D u r i n g the second stage o f the disease, the f o l l o w i n g characteristics are  impairment  severely  of  memory  and  impaired problem solving  learning; indifference or  hostility; poor  ability; misperception; more  disorientation; motor system shows restless movements (Bayles, 1987).  social  prominent spatial  6 3. Severe A D  In the third stage of the disease, the patient's intellect is globally impaired,  personality is disorganized, the patient may be mute or echolalic, and the motor system exhibits rigidity or f l e x i o n (Bayles, 1987). Semple, S m i t h and Swash (1982) investigated the c l i n i c a l features of a group of A D patients. T h e y found that memory disorders were the most common feature, w h i l e many had language and communication deficits, disorientation, visual motor problems and personality deterioration. M i l l e r (1981) reviewed the literature regarding the nature of the cognitive deficit i n senile dementia. Before discussing his review, however, mention should be made about the distinction between A D  and dementia. The term dementia implies a cognitive deficit. Dementia has  causes, one of w h i c h is A D . Some other causes of dementia are Parkinson's Disease,  many  Huntingdon's  Disease, vascular dementia brought on by multiple cerebral infarcts, n o r m a l pressure hydrocephalus, hypothyroidism and intoxication. Although there are approximately fifty causes o f dementia, some of w h i c h are treatable and some of w h i c h are not, A D is the most c o m m o n (Bayles, 1987). Because there are so many causes of dementia and because the cause of A D is unknown, the c l i n i c a l diagnosis  of A D  is a diagnosis of exclusion. If the patient presents w i t h the  symptoms  discussed above and a l l other causes of dementia have been ruled out, then one can conclude that the patient has a dementing disease w h i c h is probably A D . The diagnosis can only be confirmed at autopsy. It is most difficult to diagnose  AD  i n the early stages of the disease. Early diagnosis  is  desirable, however. Cognitive deficits and memory disturbances i n particular are the earliest clinical symptoms of A D  (Martin, Brouwers, C o x and Fedio, 1985). Thus, it is important to be aware of  the cognitive deficits associated with dementia. M i l l e r (1981) summarized the general intellectual changes w h i c h occur i n dementia. Because AD  is a type of dementia, we can assume that these changes occur during the course of  Dementia patients show decreased mean I Q levels. It is generally found that V e r b a l I Q  AD.  exceeds  Performance IQ. There is debate as to whether the pattern of declining I Q is the same as that of n o r m a l aging subjects. M i l l e r also reviewed the evidence for perceptual abnormalities i n dementia. There have been few studies i n this area. One study revealed that about 4 %  of patients w i t h cerebral atrophy  7 showed some abnormalities when asked to fixate on a particular object. Dementia patients have also  shown  some m i n o r  gnostic  abnormalities, but no  consistent  pattern was  found.  Some  visuospatial deficits have been reported. Dementia patients have difficulty solving mazes and have an impaired understanding of reflected space. They have been found to be distracted by auditory, not visual stimuli, however. Thus, there is some evidence of perceptual impairment i n dementia, as w e l l as general cognitive decline. M i l l e r reviewed the literature regarding  the memory  abilities of dementia patients. M a n y  researchers have found dementia patients to have impaired short-term memory. Dememts  have  difficulty learning new material and are equally impaired regardless of whether the task involves recall or recognition. Based on evidence f r o m dichotic listening tasks, M i l l e r suggested that the difficulty lay i n acquisition, as opposed to retention. O n dichotic listening tasks, subjects are asked to recall two lists of items w h i c h are presented to both ears simultaneously. The system used to recall items f r o m the ear w h i c h is reported first must pass the information directly to the output system. The items f r o m the ear w h i c h is reported second must be held temporarily until the first ear's stimuli have been reported. T h i s  holding  system is a short-term store. W h e n compared with the performance of n o r m a l control subjects, dementia patients recalled items f r o m the ear w h i c h was reported first, as w e l l as the normal subjects. Dementia patients recalled items f r o m the ear w h i c h was reported second w i t h much reduced efficiency. This indicates a short-term memory deficit. O n a free recall task, dementia patients were impaired i n their recall of words f r o m both the beginning and the end of the list. R e c a l l f r o m the beginning of a list is assumed to reflect longterm memory w h i l e recall f r o m the end of a list is assumed to reflect short-term memory.  This  provides further evidence to support the theory that dementia patients have a short-term memory deficit. M i l l e r suggested that the long-term memory deficit indicated by these results was due to an interruption of the f l o w of words into long-term memory w h i c h i n turn was caused by the shortterm memory impairment. T o investigate this possibility, the rate of presentation of words  was reduced. F o r normal  subjects, reducing the rate of presentation allows short-term memory to cope with the f l o w of words  into long-term memory, and performance on tests of long-term memory  increases.  The  8 performance of dementia patients d i d not increase as the rate of presentation was decreased. These results suggested that dementia patients had an impairment of long-term memory as w e l l . There are several theories regarding the nature of short-term memory deficit i n dementia. It may be due to a disturbance at the input to the system, such as impaired attentional processes or disturbed i c o n i c memory. T h e efficiency w i t h w h i c h material entering short-term memory is coded may also account for the input disturbance. It was found that the acoustic similarity of stimuh had a less distracting effect on the recall of dementia patients as opposed to that of n o r m a l subjects. This indicates that dementia patients were less able to use acoustic coding i n short-term memory. The long-term memory abilities of dementia patients have received less attention. It has often been reported that memory for remote events is retained i n dementia. Preserved recall of remote events w o u l d i m p l y that the retrieval process  of long-term memory  must be intact. Objective  investigations, however, indicate that recall of remote events is probably disturbed. M i l l e r summarized the results of a study i n w h i c h dementia patients were presented a list of words three times and then administered a distracting task. Dementia patients performed poorly on the free recall and recognition tests. However, when cued w i t h the initial letters of the words, the recall of the dements was indistinguishable f r o m that of the controls. Because recall could be enhanced to n o r m a l levels, M i l l e r concluded that acquisition was not impaired and that the problem must lay i n retrieving information f r o m long-term memory. To  explain the retrieval deficit, M i l l e r suggested that new information entering  long-term  memory might be encoded inefficiently. Retrieval w o u l d then be enhanced i f cues at the time of recall were similar to those at the time of encoding. However, the finding that dementia patients made the same types of errors on recognition tasks as normal control subjects offered no support for the inefficient encoding hypothesis. Another  possible  explanation of the partial information effect is disinhibition.  Successful  performance on a memory task may depend, i n part, upon the ability to inhibit the recall of incorrect words.  Cues w h i c h supply  partial information w o u l d l i m i t the number  of  possible  incorrect words, thus partially negating disinhibition. O n a free recall task, dementia patients produced more intrusions than n o r m a l control subjects, as predicted by the disinhibition hypothesis. O n a recognition task, dementia patients became less  9 able to recognize the correct w o r d as the number of alternatives was increased. This was  also  predicted by the disinhibition hypothesis. The long-term memory deficit of dementia patients may be due to an inability to inhibit incorrect responses rather than an inability to recall the correct response. F r o m his review, M i l l e r concluded that dementia patients have deficits of both  short-term  memory and long-term memory, and that the difficulty probably lies i n acquisition, retention and retrieval. Thus, dementia patients, and therefor A D  patients, have a global deficit of episodic  memory. In conclusion, there are many cognitive deficits associated with A D , and these occur i n step w i t h the physical changes w h i c h occur i n the brains of A D  patients. A c c o r d i n g to Bayles (1987)  A D patients are characterized by the f o l l o w i n g symptoms: "...intellectual dysfunctions sufficient to interfere w i t h social behaviour, absence of the characteristics of delirium, memory impairment indication of brain damage, and at least one o f the f o l l o w i n g : personality change, impairment of abstract toiriking, poor judgement, presence of instrumental disorder (aphasia, apraxia, or agnosia)."4  10  CHAPTER 2 LITERATURE REVIEW  2.1 Alzheimer's Disease and Language  Recently researchers have begun to study the effect o f A l z h e i m e r ' s Disease on speech and language w i t h the goal of discovering more about the nature of A D , and o f language organization i n the brain. It is generally agreed that language is affected i n the early stages of A D w h i l e speech often remains unaffected u n t i l the disease has progressed significantly. Bayles (in Chapey, 1986) states that only i n the late stages of dementia does the patient make errors i n speech sounds and produce jargon words. Because the purpose of this study is to investigate the effects of A D on language, not speech, the literature reviewed i n this chapter w i l l be concerned w i t h the effects of A D on the f o l l o w i n g three levels o f language: the syntactic system, the lexicon, and the semantic system. M o s t researchers agree that syntax is relatively spared i n m i l d to moderate A D , w h i l e semantic memory is impaired even i n m i l d A D . Schwartz, M a r i n and Saffran (1979)report a case study i n w h i c h the patient demonstrated preserved syntax with profoundly impaired l e x i c a l function. The patient, W L P , a 62-year-old female, was seen over a 30 month period. O n initial assessment, she demonstrated a W A I S performance score of 95, and showed no trouble manipulating objects or pantomiming their use, drawing or writing. She exhibited deficits i n both verbal and nonverbal memory, had a limited vocabulary ( P P V T age score=6 years), produced frequent circumlocutions, and exhibited poor confrontation naming ability. Despite these deficits, her articulation was clear. She was able to read aloud and repeat sentences o f up to seven words i n length without error. She was diagnosed as having a primary presenile dementia w h i c h was not classified as to type. W h e n presented with 7 0 pictures of c o m m o n  objects W L P could name  only  one. She  overextended animal names. This brief description serves only to demonstrate W L P ' s severe lexical impairment.These findings w i l l be discussed i n more detail later i n this chapter. Despite her deficits, W L P demonstrated intact syntactic production and comprehension abilities and was able to transform sentences spoken b y the examiner into another sentence type; for  11 example, i f the examiner made a statement, she could transform it into a question (e.g. ' T h e boy is going.' became 'Is the b o y going?'). She was also able to use syntactic information to assign thematic roles. A semantically reversible sentence was read. The task was to point to the picture w h i c h depicted the correct meaning of the sentence. T h e choices showed both possibilities o f agent and object w i t h active and passive sentences being used. W L P ' s performance was compared to that of three B r o c a ' s aphasics. B r o c a ' s aphasics are k n o w n to have impaired syntactic comprehension. She performed significantly better than the aphasics and d i d equally w e l l on active and passive sentences. T h e aphasics conclude that W L P  h a d much more difficulty with the passive sentences. Schwartz et a l .  used syntactic information to identify the referents even when the referents  were those w h i c h she could not identify b y name. Further evidence o f W L P ' s preserved syntactic knowledge comes f r o m a disambiguation task. Thirty pairs o f homophones were embedded i n 3 contexts and read aloud. T h e task was to transcribe the words orthographically. The contexts were: 1) semantic triad i n w h i c h the target homophone  was presented w i t h two semantically related words (e.g. ocean lake /si/) 2) limited  semantic context i n w h i c h the homophone was presented with a paradigmatically related w o r d (e.g. a /noz/ vs. he /noz/) 3) f u l l sentence context i n w h i c h the homophone disambiguating  was embedded within a  sentence. T h e f u l l sentence context was four times as effective at eliciting the  appropriate w o r d as the semantic triad, but the limited syntactic context was virtually as effective as the f u l l sentences. Thus, W L P could use syntactic but not semantic information to disambiguate spoken  homophones.  Schwartz  et a l . conclude that syntactic and conceptual processing  dementia patient, syntax being spared and conceptual knowledge  are dissociated i n this  being impaired. T h e overall  pattern o f W L P ' s language was one o f preserved syntax and phonology  i n the face o f semantic  impairment. Schwartz et al. note a case study reported b y Whitaker (1976) i n w h i c h another profoundly demented w o m a n demonstrated a dissociation o f syntax and semantics. This patient was echolalic and displayed n o production or comprehension o f propositional language. She was, however, able to spontaneously correct syntactically or phonologically anomalous input sentences. She could not correct semantically anomalous input.  12 Schwartz et al. suggest that the dissociation of syntax and semantics characterizes the primary degenerative dementias, one of w h i c h is A l z h e i m e r ' s Disease. They suggest that the more automatic rule-governed aspects of language  (syntax and phonology)  can be isolated f r o m those  aspects  requiring cognition (semantics). Finally, they note that syntactic disturbance has been reported i n the late stages of dementia, but that this is not surprising given the heterogeneity  of dementia  patients and the variation i n the distribution of the pathology. A p p e l , Kertesz, and F i s m a n (1982) studied the language functioning o f A D patients using the Western  Aphasia  Battery. They  found  that patients  with m i l d A D  demonstrated  the  least  impairment on the fluency and articulation subtests and were most impaired on the naming and comprehension subtests. These findings also indicate preserved syntax i n the face of impaired semantics. Hier, Hagenlocker, and Shindler (1985) used several measures  of receptive and expressive  language to asses the language abilities of a group of A D patients ranging i n severity f r o m m i l d to severe. W h e n compared w i t h normal control subjects, the A D  patients demonstrated decreased  expressive and receptive vocabularies as w e l l as decreased confrontation naming and word fluency abilities. They produced fewer total words and fewer unique words i n a speech sample. T h e  AD  patients also reported fewer important facts i n a speech sample than d i d the n o r m a l controls. These results indicate an impairment of the semantic system i n A D . In apparent contrast to the conclusions of Schwartz et al., H i e r et al. found that their  AD  patients demonstrated poorer performance on the f o l l o w i n g measures of syntactic ability: l o g i c o grammatical comprehension,  MLU  i n a speech  sample, number  of subordinate  clauses  and  prepositional phrases i n a speech sample, mean clause length i n a speech sample, and number of sentence fragments i n a speech sample. The speech sample was a description of the cookie-theft picture f r o m the Boston Diagnostic Aphasia Examination. A l t h o u g h these results may indicate that syntax is also impaired i n A D , the difficulties i n producing syntax were m u c h less severe than those i n accessing the semantic system. H i e r et al. compared the performance of A D patients of differing severity and found that as A D severity increases, speech becomes more verbose yet contains less information while syntactic  13 complexity is w e l l maintained. Thus, the results indicate that A D  is characterized by relatively  preserved syntax i n the face of impaired semantic abilities. There is some evidence that the dissociation of syntax and semantics is not complete i n A D . In a review of the literature, Obler (1983) suggests that A D  does not effect specific linguistic  categories (i.e. syntax vs. semantics), but that the linguistic interface w i t h cognition -semantics- is disturbed. Since semantics can be expressed through the l e x i c o n or through syntactic constructions, disruptions may appear i n either area. A s evidence for this v i e w , O b l e r cites cases of dementia patients who, i n a repetition task, spontaneously correct sentences w i t h morphosyntactic errors (e.g. T h e boys is here.) but not sentences with logicosemantic errors (e.g. T h e boy comes yesterday.) or logicosyntactic errors (e.g. It is raining for I cannot go.). Thus, i n those cases where semantics and syntax interact, A D patients may also demonstrate syntactic deficits. The majority of researchers seem to agree that A D  is associated with preserved syntactic  abilities i n the face of impaired semantic functioning, but that when syntax is influenced by semantic  considerations  (as  in  Obler's  logicosemantic  and  logicosyntactic constructions)  then  syntactic impairment w i l l be associated with the semantic impairment. In this review of the hterature comparing the performance of A D  patients on syntactic and  semantic measures, the term 'semantic' has been applied loosely. It has been used to refer to both conceptual and l e x i c a l knowledge. The term is often used mdescriminantly i n the hterature to indicate concepts or the words 'meaning'  w h i c h represent them. G i v e n that a lay definition of 'semantics' is  or 'the meanings of w o r d s ' it is not surprising that the term is applied so loosely.  However, when a term is not w e l l defined, it creates confusion. F o r the purposes of this study it is necessary  to separate  measures  of lexical knowledge  and  conceptual knowledge  which  had  previously been regarded as measures of semantic knowledge. Bayles (1987) refers to conceptual knowledge as the semantic memory ( S M )  and distinguishes it f r o m the l e x i c a l store (or lexical  knowledge). S M is a conceptual store. Concepts are stored i n a system w h i c h is separate f r o m the storage system of the words w h i c h represent them (the l e x i c a l store). Evidence to support the c l a i m that concepts can be dissociated f r o m the words that represent them w i l l be presented later i n this chapter.  14 In the previous discussion, the c l a i m was made that syntax is preserved i n A D w h i l e semantics is impaired. G i v e n that the term 'semantics' was used to cover both l e x i c a l and semantic memory, we must then ask whether l e x i c a l or semantic memory (or both) are  impaired i n A D . If they are  both impaired, to what extent is each impaired? L e t us again begin by discussing W L P , the patient of Schwartz, M a r i n , and Saffran (1979). It was noted earlier that when asked to name pictures oif c o m m o n objects, W L P could name only one out of seventy pictures. She was able, however, to demonstrats her knowledge of the pictured objects by gesturing their use. These results indicate that although W L P  could not access the  l e x i c a l entries for these objects f r o m the l e x i c a l store, she had retained knowledge of the concepts. Thus, w h i l e her l e x i c a l memory appeared severely impaired, her semantic memory appeared much more intact. To  further investigate this  apparent dissociation of words  and concepts, Schwartz  et al.  presented W L P w i t h each picture and five words: the target, two unrelated names, a phonological distractor, and a semantic distractor (a w o r d f r o m the same semantic category as the target). Her task was to choose the name of the pictured object after reading the five words aloud. The task was untimed. The results indicate that W L P made errors on approximately 3 5 % of the words. The majority of her errors i n v o l v e d the choice of the semantic distractor. Thus, her error rate was significantly lower than that on the naming task. She was inconsistent i n her errors across two administrations of the name matching task, choosing the target one time and the semantic distracter the next time on a given trial. Schwartz et al. conclude that W L P had an impairment of semantic memory i n that terms f r o m the same semantic category no longer specified a particular referent, but were extended to a population of related referents. The task was administered on three subsequent occasions spanning the 30-month period during w h i c h W L P was followed. W h i l e the percentage of total errors gradually increased over time, the number of errors resulting f r o m the choice of the semantic distractor decreased. These results were c l a i m e d to be evidence for the hierarchical deterioration of semantic  features w h i c h define  referents, the more specific distmguishing features being lost before more general features.  15 Overall, Schwartz, M a r i n , and Saffran's results seem to indicate that this dementia patient had a semantic memory deficit i n addition to her l e x i c a l deficit. The naming task was a test of lexical memory and/or l e x i c a l access, not of semantic memory as defined by Bayles. W e can make this c l a i m because, although unable to generate the names o f items, W L P  clearly demonstrated her  understanding of the concepts w h i c h the names represented. W e can therefor conclude that W L P had a l e x i c a l deficit. D i d W L P have a deficit o f semantic memory? The name matching task does not give us a clear answer to this question. In order to perform this task, the patient may adopt one of two strategies (see Figure 1). U s i n g Strategy #1, the patient accesses the concept by means of the visual input (the picture), then maps the concept to the w o r d w h i c h represents it, then matches the word to one of the choices. If the patient adopts this strategy, it may be assumed that errors are due to a breakdown at any step i n the process. Errors may be due to a disrupted conceptual system or difficulty accessing the conceptual system, a semantic memory deficit. W L P ' s ability to m i m e the use of the objects makes this explanation unlikely. Errors may also be due to an inability to map the concept to the w o r d w h i c h represents it i n l e x i c a l memory. G i v e n W L P ' s severe naming deficit, this explanation seems more plausible.The choice of the semantic distractor may indicate that l e x i c a l memory is organized similarly to semantic memory, by semantic features, for ease of mapping.  Errors may also be due to an inability to map the correct l e x i c a l choice to the printed  words. However, given W L P ' s intact reading skills, this explanation is also unlikely. U s i n g Strategy #2, the patient accesses the name directly f r o m the v i s u a l input, completely bypassing  the conceptual system (see Figure 1). I f the patient adpots this strategy, it may be  assumed that errors are due to a perceptual deficit, a deficit of l e x i c a l memory, or difficulty mapping perceptual patterns to l e x i c a l memory. Since we have claimed, based on other evidence, that W L P  had a deficit o f l e x i c a l memory, and since her m i m i n g ability is evidence for an intact  perceptual system, a deficit of l e x i c a l memory seems to be the most plausible explanation of her performance on the name matching task. Thus, although it seems clear that W L P had a deficit i n l e x i c a l memory, the coexistence of a deficit i n semantic memory remains doubtful. It is difficult to design a task w h i c h tests semantic without also testing l e x i c a l memory.  16  Figure 1 Name Matching Task  Strategy #1 *  **  +  #  v i s u a l input~> S M - > L M - > name choice Strategy #2  * visual input key: * perceptual error * * S M deficit + mapping f r o m S M to L M disrupted # L M deficit  # > L M ~ > name choice  17  Schwartz et al. noticed that on several occasions W L P overextended the label D O G to include C A T S . D i d this mean that the concept C A T was lost or that the label D O G n o w covered both the concepts C A T and D O G ? T o investigate this question, a task was administered i n w h i c h W L P was shown pictures of various dogs cats and birds. She was presented with each picture one at a time and was shown two typed labels. The task was to move the picture to the appropriate label. A l l the instances of dogs were labelled correctly, while most instances of cats were also labelled D O G . A l l instances of birds were labelled correctly. In another condition, W L P was shown three photographs: one sample and two choices. The task was to indicate w h i c h was the same type of animal as the sample. W L P  was consistently  correct when the sample was a cat. W h e n the sample was a dog, she matched correctly only when the alternative was a b i r d and otherwise chose the cat. In this condition she overextended the category C A T to include dogs, w h i c h was the opposite of her performance i n the written label task. Schwartz et al. interpreted these results as mdicating that the underlying category structure was disrupted, but that the category structure d i d not predict the pattern of naming errors. T h e written label task, like the name matching task, could be accomplished by the two strategies mentioned above. It is difficult to decide whether W L P ' s errors on this task were due to a deficit of semantic or l e x i c a l memory, or both. In the picture-to-picture matching task, however, l e x i c a l memory is bypassed. The task can be accomplished by accessing the categories alone. Because W L P demonstrated deficits on this task, we can conclude that she had a semantic memory deficit as w e l l as a l e x i c a l memory deficit. It seems that the l e x i c a l memory deficit was more severe than the semantic memory deficit. G i v e n that the disrupted categories d i d not predict the pattern of naming errors, we can also conclude that W L P was unable to map the category to its label. A f i n a l aspect of Schwartz et al's study was the investigation of W L P ' s reading ability. She demonstrated preserved ability to read aloud. There are two methods to attain the pronunciation of a written word. The first is to sound out the word using letter-sound correlations. The second is to associate the written word directly with a lexical entry, referred to as sight reading. The finding that W L P c o u l d read nonsense words fluently i f they corresponded to normal spelling suggested  18 that she was able to use letter-sound correlations to attain the pronunciation. W h e n her task was to match a spoken word to three possible spellings, one o f w h i c h more closely approximated normal E n g l i s h spelling, W L P was able to match the majority o f words correctly. WLP  was also able to read irregularly spelled words fluently. O n e task required her to read  aloud pairs o f words w i t h similar spellings but different pronounciations (e.g.  home, come). I n this  task, consistent application o f spelling rules w o u l d result i n incorrect pronounciations. I f W L P correctly pronounced the words, then she must have been able to recognize them as words and access the pronounciation v i a the lexicon. That is, she must have been able to use the sight reading method. T h e result o f this investigation was that W L P pronounced a l l the words correctly without hesitation. Schwartz et al. concluded that W L P had intact phonological encoding abilities, suggesting that her naming difficulty must have arisen at a prephonological level. T h e phonological route to the l e x i c o n seemed to be open w h i l e the conceptual route was not. Contrary to the conclusions based on the naming task, this w o u l d suggest that l e x i c a l memory was not impaired but that perhaps mapping f r o m l e x i c a l memory to semantic memory was disrupted. Bayles (1987) commented on W L P ' s reading ability, suggesting that since W L P could read b y sight, she must have been retrieving words based on memory o f the visual configuration alone. W L P could read words w h i c h she could not understand. T h e ability to bypass the semantic system suggests the existence o f a brain circuit w h i c h  directly  connects  visual configurations and  phonological codes. I f dementia patients retain this circuit and i f sight-reading is analogous to naming an object by matching its visual configuration to a word, then i t is possible that dementia patients c a n name objects without activating semantic memory. Thus, Bayles concludes that naming tasks  cannot  be used  as evidence  regarding  the state  o f the semantic  memory  of A D  patients.Naming tasks test lexical, not semantic memory. There have been several studies w h i c h indicate that A D patients have poor confrontation naming abilities.In a review of the research, Obler (1983) states that naming disturbances are present i n virtually a l l cases o f dementia, and that naming is impaired i n the moderate stages of AD.  19 A p p e l l , Kertesz and Fisman (1982) found that a group of A D patients were more impaired on the naming subtest of the Western A p h a s i a Battery than on any other subtest. E v e n m i l d patients showed  a naming  impairment. The results indicated that naming  AD  showed the greatest  decline initially i n A D , but that other language functions declined more rapidly with time. Hier, Hagenlocker and Shindler (1985) found that dementia patients performed significantly less w e l l than normal control subjects on a confrontation naming task. M a r t i n and Fedio (1983) found that A D  patients had a mean score more than three standard deviations below the mean score of  n o r m a l subjects on the Boston N a m i n g  Test. A n error analysis revealed that the A D  patients  produced more synonyms, descriptions and semantic f i e l d errors than perceptual errors. M a r t i n and Fedio concluded that i n A D : " T h e ability to define a word using phrases or an appropriate synonym and knowledge of category membership were relatively more preserved than the ability to retrieve a specific referent." 1 The studies mentioned above investigated the overall pattern of language abilities and deficits in A D  and found confrontation naming to be impaired. There has been some debate about the  cause of the naming impairment. Is it due to a perceptual impairment or a semantic impairment? Bayles  and Tomoeda (1983) investigated this question. Dementia patients of varying type and  varying severity were asked to name twenty coloured pictures. Responses were classified according to the similarity to the target. The response could be unrelated, visually related, or linguistically related, w h i c h included phonemic and semantic similarities, to the target. The results showed that only the moderate A D control subjects  i n their naming  ability. F o r  patients differed significantly f r o m normal  a l l groups the errors  were more  l i k e l y to be  semantically associated than visually or phonemically associated. O f those responses that were visually similar, most were also semantically similar. The most c o m m o n type of semantic error was within-class substitution. Another word f r o m the same semantic category was chosen (e.g. for bus).  As  severity of dementia increased, extent of naming  truck  irnpairment and likelihood of  unrelated errors also increased. Bayles and Tomoeda concluded that their results supported the hypothesis that naming errors i n A D  patients are due to a linguistic-cognitive impairment, not a  20 perceptual impairment. They suggested that the linguistic-cognitive impairment was a deficit of semantic memory w h i c h resulted f r o m the erosion of the referential boundaries of words. Later, Bayles (1987) reconsidered her conclusions. Confrontation naming is a test of lexical memory, not semantic memory. It is possible that the w o r d f o r m can be accessed without accessing the concept w h i c h it represents. The predominance of semantic errors made by A D  patients may  indicate that the l e x i c o n is organized similarly to semantic memory. T o summarize, confrontation naming is impaired i n A D  and the impairment worsens as the disease progresses. This indicates  that l e x i c a l memory is impaired i n A D , but the use of confrontation naming as evidence of a semantic memory impairment is equivocal. Another task w h i c h has been used extensively i n the study of A D patients language abilities is verbal fluency. R o s e n (1980) defined verbal fluency as the ability to retrieve members belonging to a specified category w i t h i n a lhriited time period. A subject can be asked to retrieve members of a semantic category or words  beginning  with a specified letter. In their studies of the overall  language pattern i n A D , M a r t i n and Fedio (1983) and Hier, Hagenlocker and Shindler (1985) found that A D  patients scored significantly lower than n o r m a l control subjects on word fluency tasks.  H i e r et al. used animal naming w h i l e M a r t i n and Fedio used the category o f items found i n a supermarket (from the Mattis Dementia Rating Scale). M a r t i n and Fedio found that the normal control subjects not only produced more words than the A D patients, but also generated words f r o m more categories and more words per category. The A D patients switched categories more often and produced names instead of specific items. R o s e n (1980) investigated verbal fluency of A D patients i n more detail. A n animal naming task and an initial-letter ( C F L ) task were used. Subjects were given sixty seconds for each category: animals,words starting with C, words starting w i t h F, and words starting w i t h L. Results indicated that normal control subjects retrieved more words than the A D patients retrieved more words than moderate to severe A D  patients on both tasks. M i l d  AD  patients. O v e r a l l , more animal names  were given than C, F, or L words. N o r m a l controls and m i l d A D patients produced more animal names than C, F, or L words w h i l e moderate to severe A D patients showed no difference between tasks. This may have been due to the limited number of words produced by the moderate to severe A D patients.  21 Different patterns of retrieval were found for the normal controls, m i l d A D moderate to severe A D  patients and  patients. The normal elderly retrieved the most words during the first  fifteen-second interval than during the remaining three intervals. The performance of the normal elderly declined nonsignificantly over the next three intervals for the C F L words w h i l e performance declined significantly over the next three intervals for animal naming. M i l d A D patients retrieved more words during the first interval w i t h retrieval dechning nonsignificantly over the remaining time for both tasks. The  moderate to severe A D  patients showed  no  significant change  in  performance throughout the entire time for both tasks. R o s e n concluded that verbal fluency is impaired i n A D and performance declines as the disease progresses. The fact that both normal and m i l d A D patients produced more animal names than C F L words  suggested  to R o s e n that the different categories  are  structured differently. Semantic  categories have a core meaning composed of best exemplars. Therefore, animal names have many good exemplars w h i l e C F L words do not. R o s e n suggested two retrieval strategies: 1) retrieve the best exemplars, 2) go to subsets  of the category. F o r ardmal names, the best exemplars are  retrieved q u i c k l y and then the subsets are entered. F o r C F L words, immediate entrance into the category does not produce many words because there are few best exemplars. The subsets are then entered. R o s e n found that for C F L words there were two subsets; semantic (e.g.animals that begin with C ) and phonemic (e.g. words that begin w i t h CI). The retrieval of more animal names by the n o r m a l controls and the m i l d A D quicker recall of more numerous impaired i n m i l d A D  patients may indicate the use of retrieval strategies such as the best exemplars. R o s e n suggested  that subset entrance  was  w h i l e retrieval of best exemplars was not, and that both o f these strategies  were impaired i n moderate to severe A D .  He  interpreted these findings  as evidence for an  impairment of semantic memory i n A D . Ober, Dronkers, K o s s , Delis and Friedland (1986) also investigated verbal fluency i n  AD  patients. They used three different tasks: 1) a letter category task i n w h i c h subjects generate words beginning w i t h a certain letter, i n this case F, A  and S, 2) a semantic category task i n w h i c h  subjects generate words f r o m a given semantic category, i n this case animals and fruits, 3) a supermarket task i n w h i c h subjects generate the names of items found i n a supermarket. Ober et al.  22 expanded on the work o f R o s e n (1980) and M a r t i n and Fedio (1983) by administering a l l three types of task to the same group of A D patients and by analyzing correct responses and error types. The responses were scored i n one of the f o l l o w i n g ways: l)correct; 2)noncategory, when the response d i d not belong to the specified category; 3)morphological variant, when the response was morphologically similar to a w o r d given earlier e.g. F O O T and F E E T ; 4)categorical variant, when the response was similar i n category to a word given earlier e.g. B I R D and B L U E B I R D ;  5)  perseverations, when the response was a repetition of a previous response. Subjects were divided into three groups: normal controls, m i l d A D severe A D  patients and moderate to  patients. Results indicated that a l l groups produced more correct semantic category  items than letter category items during the first 15 seconds, but showed equal performance on the two tasks thereafter. A s expected, normal controls produced the most correct responses, m i l d A D patients produced fewer and moderate to severe A D patients produced the least correct responses. The effect of time interval, i n blocks o f 15 seconds, was significant i n a l l cases, subjects producing fewer correct responses over time. Ober et al. analyzed the responses o f the three groups according to category dominance and w o r d frequency  o f the exemplars. There were no  differences between the groups on  these  measures. Thus, differences i n the performance of the groups were not due to differences i n the accessibility  of l o w dominance  semantic category  members  or l o w frequency letter  category  members. F o r the semantic category task, normal controls demonstrated the highest proportion of correct responses and the lowest proportion of errors w h i l e m i l d A D and moderate to severe A D patients d i d not differ significantly f r o m each other i n proportion of correct responses to errors. F o r the letter category task, normal controls and m i l d A D  patients were not significantly  different f r o m each other, both having a higher proportion of correct responses than moderate to severe  AD  patients. Moderate  to severe  AD  patients demonstrated  a higher  proportion  of  noncategory and perseverative responses than m i l d A D patients and n o r m a l controls. The supermarket task was administered only to the A D patients since it is part of the Mattis Dementia Rating Scale. Moderate to severe A D patients performed significantly poorer than m i l d A D patients on number of correct responses, number of subcategories, items per subcategory, and  23 pairwise clusters (the number of responses f r o m the same category w h i c h are adjacent to one another i n the response protocol). T h e moderate to severe A D group produced significantly more perseverations than the m i l d A D  group.  Performance on the word fluency tasks was compared with results on a neuropsychological battery. W o r d fluency was significantly correlated w i t h results on the Tratt-Malring, D i g i t - S y m b o l and Vocabulary subtests of the W A I S ; the V i s u a l Attention subtest of the Mattis Dementia Rating Scale; N a m i n g  and Comprehension  subtests of the Boston Diagnostic A p h a s i a Examination; the  T o k e n Test; and B u s c h k e ' s verbal learning v i a selective reminding test. Ober et al. concluded that verbal fluency is impaired i n A D and that the impairment increases w i t h dementia severity, but that the impairment is not due to differences i n category dominance or w o r d frequency. The deficits i n performance are due to a decrease i n correct responses along with an increase i n errors. There is a qualitative difference between the performance of m i l d A D patients and moderate to severe A D patients produce  patients on the supermarket task i n that the moderate to severe  fewer items  per subcategory.  It  was  suggested that the results  AD  indicate a  breakdown i n the structure and/or processes of S M i n A D w h i c h increases w i t h the severity of the dementia. Ober et al. note, however, that the disruption may also be due i n part to problems with attention, learning, naming and language comprehension w h i c h have been shown to be impaired in AD. Thus, the consensus is that A D patients have difficulty with word fluency tasks and that this is due to a semantic memory deficit. W o r d fluency tasks w h i c h call for semantic categories require subjects to enter S M directly to L M ,  i n order to produce correct responses. Subjects cannot bypass S M ,  because  they must  enter the system through  SM  and use  SM  to  going  generate  subcategories, then map exemplars to forms i n L M . W o r d fluency tasks test S M , L M , and the mapping f r o m S M to L M . A n impairment of word fluency i n A D may be due to a S M deficit, a L M deficit or an inability to map concepts to the words w h i c h represent them. The letter category w o r d fluency task is somewhat different. Rosen (1980) found that the letter category task c o u l d be approached using two strategies, phonemic or semantic. If the phonemic strategy were used, subjects could bypass S M and perform a phonemic search of the lexicon (e.g. look for a l l words starting w i t h /cl/). A n impairment w o u l d then indicate a deficit of L M . If the  24 semantic strategy were used (e.g. look for a l l animals starting with /If), could not bypass S M  and  an impairment w o u l d indicate a deficit of S M , L M , or the ability to map between them. A l t h o u g h the evidence f r o m w o r d fluency tasks has been interpreted as indicating a S M deficit i n A D , the conclusions may not be warranted. Another c o m m o n finding is that A D  patients have diminished vocabularies, both expressively  and receptively, and that vocabulary declines during the course of the disease. Hier, Hagenlocker and Shindler (1985) found that, compared w i t h normal control subjects, dementia patients had significantly lower expressive vocabularies as measured by the W A I S vocabulary subtest. This test requires subjects to provide definitions for words. Dementia patients also demonstrated significantly lower receptive vocabularies than normal controls, as measured by the A m m o n s and A m m o n s Q u i c k Test. This test requires subjects to point to one of four pictures i n response to a spoken word. M a r t i n and F e d i o (1983) found that A D patients scored significantly lower than normal controls on both the vocabulary and similarities subtests of the W A I S , but that the scores were w i t h i n the average range according to the W A I S norms. In a review of the research regarding the vocabularies of A D patients, Bayles (1987) reported that moderate A D patients performed more poorly than normal controls on the W A I S  vocabulary  subtest. O n the Peabody Picture Vocabulary Test ( P P V T ) , a test of receptive vocabulary, m i l d A D patients performed significantly more poorly than normal controls. Bayles stated that to respond correctly to an expressive vocabulary test, a subject must activate concepts because the examiner is asking for the conceptual representation o f the word. A subject must access the l e x i c a l representation, then map it to the conceptual representation i n order to produce a definition (see F i g . 2). A n impairment of expressive vocabulary w o u l d indicate either a deficit of L M , S M , or the pathway between L M and S M , assuming that auditory perceptual skills are intact. T o respond correctly to a receptive vocabulary test, a subject may activate concepts v i a the l e x i c a l representation and then match the concept to the correct picture (path a) of F i g . 2). However, a subject may match the l e x i c a l representation directly to the v i s u a l configuration, i n the reverse of process  postulated for naming  (path b) of F i g . 2). Since most tests of receptive  25 vocabulary increase i n difficulty as the test progresses, it is unlikely that a direct mapping from v i s u a l memory to visual configuration w o u l d be possible for the more difficult abstract items. It is possible that A D strategy may  patients use strategy b), thus correctly responding to the simpler items. This  begin to f a i l as the items become more difficult and abstract. A D patients may then  attempt to switch to strategy a), but because of a S M deficit are unable to use this strategy. Thus, they w o u l d not be able to respond correctly to the more difficult items. In order to investigate this possibility, one w o u l d have to determine w h i c h items could be accessed directly f r o m l e x i c a l memory and w h i c h could not. If results indicated that A D  patients  responded correctly to those items w h i c h could be accessed directly f r o m l e x i c a l memory and not to those w h i c h could only be accessed through S M , one w o u l d have convincing proof of a  SM  deficit i n A D patients. A n impairment of receptive and/or expressive vocabulary is evidence that there is a deficit of S M , L M or the ability to map between L M and S M . Thus, the research indicates that A D patients may have a semantic memory deficit. Bayles (1987) noted that vocabulary measures such as the P P V T , w h i c h use line drawings, c o u l d be confounded by visual perceptual deficits. T o investigate this possibility, she asked patients to orally define the words to w h i c h they responded incorrectly on the P P V T .  AD  Results  showed that the patients could not define any of the words w h i c h they had missed. Thus, errors of receptive vocabulary were due to lack of knowledge of word meanings, not misperception. A s demonstrated by Schwartz, M a r i n and Saffran (1979), a subject's ability to pantomime can reveal information about his conceptual knowledge  without i n v o l v i n g l e x i c a l memory. If  AD  patients show an impairment i n their ability to pantomime, we can conclude that there is a  SM  deficit. B a y l e s ' (1987) review o f the literature revealed that while m i l d A D patients d i d not differ f r o m n o r m a l controls i n their ability to pantomime, moderate and severe A D patients showed a reduced ability to recognize or use pantomime. Because patients were able to imitate gestures, it was assumed that the impairment was not due to l i m b apraxia. A patient's ability to name an object was not a predictor of his ability to pantomime the use o f the object. W e can conclude with Bayles that patients with A D have a deficit of semantic as w e l l as l e x i c a l memory.  26  Figure 2 Vocabulary Tests  Expressive vocabulary auditory i n p u t — > L M — > S M — > v e r b a l output (word) (definition) Receptive vocabulary a) auditory i n p u t — > L M — > S M — > c h o i c e of picture (word) b) auditory i n p u t — > L M — > c h o i c e o f picture (word)  27 Another task w h i c h has been used with A D patients is word association. It is assumed that i f conceptual knowledge is intact the stimulus words w i l l elicit strongly associated words, as defined by the norms. G e r w i t h , Shindler and H i e r (1984) studied the responses o f 22 A D patients to 16 words f r o m the Palermo and Jenkins (1964) word association norms. T h e response latency was recorded and the responses were classified as popular or nonpopular. Popular was defined as one of the three most c o m m o n responses i n the normative sample. Responses were also classified as belonging to one o f the five f o l l o w i n g groups: l)paradigmatic-the response is semantically related to the stimulus i t e m and is o f the same grammatical class e.g.  hot, cold 2)syntagmatic- the  response is of a different grammatical class than the stimulus i t e m and w o u l d occur sequentially w i t h i n the same sentence e.g.  sit, down 3)identity- the response is echolalic or contains a minor  deviation f r o m the stimulus e.g. man, relationship to the stimulus item e.g.  men  4)idiosyncratic- the response  bears no discernable  hardly, yes 5)null- there is n o response w i t h i n 25 seconds of  the stimulus presentation. Results  indicated that as dementia progressed, the number  of popular responses  decreased  significantly. W i t h increasing severity of dementia, the number of paradigmatic responses decreased and the number o f idiosyncratic, identity and n u l l responses increased, but there was n o change i n the number of syntagmatic responses. T h e response latency increased significantly w i t h severity of dementia. G e r w i t h et al. suggested that paradigmatic responses are generated by conserving the syntactic markers o f a w o r d w h i l e making a minor change i n the semantic markers. A response w o u l d be of the same grammatical class as the stimulus while contrasting w i t h respect to a single semantic marker e.g.  hot, cold. T h e difficulty that A D patients experienced w i t h paradigmatic responses was  not due to a loss of syntactic markers. T h e normal control subjects produced the most paradigmatic responses when the stimulus i t e m was a noun, the most syntagmatic responses when the stimulus was a verb, and the most n u l l responses when the stimulus was an adverb. T h e A D patients produced the same pattern of responses, indicating that they were as sensitive to grammatical class as the normal subjects. T h e f a l l i n paradigmatic responses may have been due to a loss o f semantic markers. A l o n g with the stability of syntagmatic responses, these results l e d G e r w i t h et al. to conclude that syntactic knowledge was preserved i n A D while semantic knowledge was impaired.  28 T o respond paradigmatically to a word association task, a subject must access the lexical representation of the stimulus word, map the l e x i c a l representation to the conceptual representation in S M ,  complete a feature analysis  of the conceptual representation, f i n d another conceptual  representation w h i c h differs by only a few features, map the second conceptual representation to an item i n L M , and produce the response (see Fig.3). Failure to produce a popular response  may  indicate a m i l d deterioration of conceptual knowledge i n that common associations are not made, even though response  may  semantic  analyses  produce  indicate a breakdown  correct responses.  along  any  Failure to produce  a  paradigmatic  part of the path shown in Fig.3.  appropriate concepts may have been selected, the corresponding item i n L M  Although  may not have been  accessed. Perhaps deterioration of the conceptual store has resulted i n concepts no longer being related i n the usual manner. The mechanism  is not clear f r o m the results. W o r d  association  research is another area i n w h i c h results do not provide unequivocal evidence of a S M deficit i n AD. M a r t i n and F e d i o (1983) investigated the performance  of A D  patients on broad  category  judgement and a s y m b o l referent test. The broad category judgement task required subjects to rate the pleasantness o f 30 words on a 7-point rating scale. There were 10 pleasant, 10 neutral and 10 unpleasant words, as judged by normal individuals. Results showed that there was no significant difference i n the number of words rated as pleasant, neutral and  unpleasant by the A D  patients  and n o r m a l control subjects. B o t h groups rated the same words as pleasant, neutral and unpleasant. F o r the s y m b o l referent test, subjects were presented w i t h cards containing a w o r d and four stylized line drawings. The task was to select the drawing w h i c h best represented the meaning of the word. The correct response was determined by the drawing chosen most often by a group of college students. The stimulus items included actions, objects, emotions and modifiers. Results indicated that the A D  patients made significantly more errors than normal controls on actions,  objects and modifiers, but not on emotions. The authors concluded that A D  results i n a reduction i n the availability of attributes that  determine w o r d meanings ( A D patients performed more poorly on naming and word fluency than on broad category judgement). That is, word boundaries break d o w n while broader categorical information is preserved. This may indicate an underlying deficit i n the organization o f S M .  29  Figure 3 Word Association Task  feature analysis input  >LM >SM (word 1) (word 1)  >SM >LM >output (word 2) (word 2)  30 T o perform the broad category judgement task, a subject must access the l e x i c a l representation of the word, map the l e x i c a l representation to the conceptual representation, analyze the general aspects of the concept, and make a judgement regarding the pleasantness of the concept. Since the A D patients were able to do this, we can assume that L M and the general aspects o f S M are intact. T o perform the s y m b o l referent task, a subject must access the l e x i c a l representation, map the l e x i c a l representation to the conceptual representation, analyze more specific aspects of the concept, and choose a suitable drawing. Errors may indicate an impairment of L M , S M map  from L M  to S M .  However,  since  the A D  patients  demonstrated  or the ability to  intact L M  by  their  performance on the broad category judgement task and demonstrated that the more general aspects of S M  were intact, the errors on the s y m b o l referent task must be due to difficulty with specific  aspects of the concept or difficulty choosing a suitable drawing. The fact that A D  patients were  able to choose correctly for emotion words indicates that they were able to choose appropriate drawings and understood the task. Thus, the results of M a r t i n and F e d i o ' s (1983) study indicate that AD  patients have a deficit of semantic memory. In particular, A D  patients are unable to analyze  the attributes w h i c h determine word meaning, w h i l e knowledge of broader categorical information is preserved. M a s u r (1986) used a category exemplar task to examine the effects of A D  on  semantic  memory. Subjects were to decide whether sentences consisting of exemplar-category relationships were true or false, and their reaction times were measured. Based on a probabilistic semantic model, predictions were made about the reaction times for the sentences. W h i l e the normal control subjects produced reaction times consistent with those predicted by the model, the A D  patients  produced anomalous patterns. The results were interpreted as indicating that even m i l d A D patients had a marked disruption of semantic memory. To  perform a sentence verification task  representations  of both the category  representations i n S M ,  of this type, a subject must access  and the exemplar word, map  the  lexical  them to their conceptual  analyze the conceptual representations to decide i f the exemplar is  a  member of the category, and respond by indicating whether the sentence is true or false. The results f r o m this task do not provide unequivocal evidence for a S M deficit i n A D because errors may result f r o m a S M deficit, a L M deficit, or an inability to map between S M and L M .  31 Although  many  researchers  agree that A D  is associated w i t h a S M  deficit, some  studies  indicate that S M is preserved i n A D . Nebes, M a r t i n and H o r n (1984) concluded that S M is spared i n A D patients. They noted that most tasks used to investigate S M require attentional capacity and effortful processing. Subjects are required to make conscious decisions about the semantic features of stimuli. Nebes et al. suggested that tasks w h i c h are designed to assess S M  should require  automatic effortless processing, thereby placing fewer demands on attentional resources. Three tasks  designed  to elicit automatic processing  were administered to normal control  subjects and patients w i t h m i l d and moderate A D . F o r the semantic p r i m i n g task, subjects were required to name visually presented words. Pairs of words were either semantically related or unrelated. A f t e r a l l the pairs had been presented, subjects' recall and recognition of the words were tested. This task was meant to measure the intactness of semantic associations. It was predicted that the amount of semantic prirning w o u l d be equal for the normal controls and the A D patients. Results f r o m the prirning task indicated that both the normal controls and the A D  patients  showed a positive prirning effect. The response latency was significantly shorter when the target was preceded by a semantically related word. T h e normal subjects demonstrated shorter latencies than the A D  patients and remembered more words than the A D  patients on both the recall and  recognition tasks. The second task, approximation to English, required subjects to name letters presented one by one. Once the entire string of letters was presented, subjects were asked to recall the string i n the correct order.The strings varied i n their approximation to E n g l i s h spelling. The task was  designed  to measure a subject's ability to use his knowledge of E n g l i s h orthography to remember strings of letters. It was predicted that both normal controls and A D  patients w o u l d better remember the  strings w h i c h most closely approximated E n g l i s h spelling. Prior to performing the approximation to E n g l i s h task, subjects completed a standardized reading test. There was no significant difference between the mean scores of the A D patients and the normal control subjects. A l t h o u g h the control subjects and the A D patients were not matched for reading scores, they were matched for years of education. Nebes et al. d i d not report correlations between reading level and performance on the approximation to E n g l i s h and the approximation to text tasks.  32 Results f r o m this task indicated that although the normal subjects recalled more letters than the A D patients, the increasing approximation to English improved the percentage recall of both groups equally. T h e third task, approximation to text, was similar to the second. Subjects named a string of words presented one by one, and were then asked to recall them i n order. The strings varied i n their approximation to E n g l i s h w o r d order. This task was designed to measure a subject's ability to use his knowledge of E n g l i s h syntax to remember strings of words. It was expected that the more closely the words reflected E n g l i s h w o r d order the more easily they w o u l d be recalled by both n o r m a l subjects and A D patients. Results f r o m this task indicated that although normal control subjects recalled more words than the A D  patients, increasing the approximation to text improved the percentage recall for both  groups equally. Nebes et al. suggested that the A D patients had deficits of episodic memory, as evidenced by their poor performance on recall and recognition tasks. Evidence for an episodic memory deficit was presented i n Chapter 1.2. The A D  patients also performed poorly on a task i n v o l v i n g a  conscious search of semantic memory, namely verbal fluency. However, on the three experimental tasks, w h i c h required rninimal effortful processing, the A D  patients performed as w e l l as the  n o r m a l control subjects. The fact that the two groups demonstrated the same semantic priming effect indicated that the structure of semantic memory was grossly intact i n A D patients. The fact that the two groups demonstrated the same percentage improvement w i t h increased approximation to E n g l i s h orthography indicated that A D  patients were able to use their knowledge of English  spelling to encode strings of letters into larger units. Thus, Nebes et al. concluded that at least some aspects of S M remain intact i n A D . Kahneman (1973) proposed a theory of attentional resources. H e suggested that an individual has only a given amount of attentional resources. E a c h mental operation that is performed requires a certain amount of these resources. Some mental operations require few resources w h i l e others require large amounts  o f attentional resources. That is, they require effortful processing.  The  amount of attentional resource w h i c h is available at any one time is limited. This is evidenced by the fact that humans can perform only a limited number of operations at one time. W h e n a subject  33 attempts to perform more than this number of tasks, his performance w i l l suffer because the attentional resources needed to complete the task are not available. F o r example, recall of a list of words  w i l l be decreased i f subjects are asked to perform mathematical operations during the  interval between presentation and recall. The mathematical operations require a large amount of attentional resource and there is not enough resource available to rehearse the list of words. C r a i k and B y r d (1982) suggested that older normal adults have a reduced amount of attentional resources. T h e y presented evidence indicating that older subjects performed more poorly than y o u n g subjects o n tasks requiring effortful processing. If older n o r m a l subjects have a reduced amount of attentional resource, then it is possible that A D patients have an even further reduced amount of attentional resource. Perhaps the effortful processing required to perform most of the tasks used to test semantic memory depletes the available resources, causing semantic memory to break down. However, when the processing is not effortful, the resources needed to maintain S M are available and S M  remains intact. This w o u l d explain the discrepancy between the results of  Nebes et al. and the results discussed previously i n this chapter. M a r t i n , Brouwers, C o x and F e d i o (1985) used recall and recognition tasks to investigate the SM  of A D  patients. Subjects were to recall a list of 8 words. They were given five recall trials  then a recognition trial i n w h i c h there  were four choices: The target, a semantic distractor, a  phonemic distractor, and an unrelated word. The same procedure was repeated six times over a three week period. Results  showed  that A D  patients recalled fewer words  than n o r m a l control subjects, but  demonstrated the same pattern of recall. The mean number of words recalled increased with each trial. The A D patients also demonstrated a serial position curve w h i c h was no different than that of n o r m a l controls. B o t h primacy and recency effects were apparent. O n the recognition task, the A D patients recognized fewer words than the normal controls. The A D  patients chose the semantic  distractor significantly more often than the phonemic distractor or the unrelated word. If the target was not recognized, the A D  patients were most l i k e l y to choose a semantically related word.  M a r t i n et al. concluded that although the A D  patients had a reduced ability to learn word lists,  their pattern of performance d i d not differ f r o m that of normal control subjects. It was suggested  34 that the reduced recall of the A D  patients was due to an incomplete semantic analysis of verbal  material during learning. The second experiment of M a r t i n et al. (1985) was designed to investigate the effects of semantic encoding on recall. Subjects were required to recall four lists of nine words each. There were  four possible  encoding  conditions:  l)free encoding-  the subject was  given  no  special  instructions 2)rhyme- the subject was asked to say a w o r d w h i c h rhymes with the target 3)wherethe subject was asked to state where the object could be found 4)praxic- the subject was asked to pantomime the use o f the object. Subjects were asked for immediate recall of the lists. F o r those items w h i c h were not recalled, the subject was cued w i t h his o w n responses. Results  indicated that although A D  patients recalled fewer words than the normal control  subjects, they exhibited the same pattern of recall. Semantic encoding (the where and praxic conditions) was superior to phonemic encoding (the rhyme condition) i n prompting correct recall. M a r t i n et al. concluded that the memory abilities of A D patients are quantitatively, not qualitatively different f r o m those of normal subjects. If A D patients had S M deficits, then semantic encoding w o u l d not aid recall. Since semantic encoding d i d aid recall, we must conclude that A D patients do not have S M deficits. Thus, it seems that when A D patients are required to use S M to help them perform a task, they are able to do so; when asked to make a conscious search of S M , they are unable to do so. This adds further support to the speculation that effortful processing causes a breakdown of S M i n A D patients. In her 1987 review of the literature, Bayles suggests that priming may be the best paradigm for testing S M  because  SM  is  activated without effortful processing.  The  results  are equivocal,  however. W h i l e some researchers have found equivalent amounts of priming i n A D  patients and  n o r m a l controls, others have failed to find a semantic priming effect and some researchers have reported a negative priming effect for A D patients. The A D patients had longer response latencies when the target was preceded by a semantically related word. Bayles  suggested that the S M  of A D  patients is disrupted, but that the disruption  occurs  gradually. W h i l e i n m i l d A D the associative links between concepts i n S M may be weakened but  35 not broken, by severe A D the inability to use objects or even to imitate their use indicates a severe deficit of conceptual knowledge. Further evidence of a S M deficit i n A D (1986). They  investigated a phenomenon  comes f r o m a study by M i t c h e l l , Hunt and Schmitt called the generation  effect i n A D  patients. The  generation effect refers to the finding that internally generated information is better remembered than externally provided information. Another phenomenon, called reality monitoring, refers to the fact that subjects can be extremely accurate at identifying the source of the information, whether it was  externally provided or internally generated. Internally  generated and externally provided  information are treated differently psychologically. It has been proposed that the advantage  of  internally generated information depends on the activation of S M , because nonsense syllables w h i c h are internally generated do not elicit the generation effect. M i t c h e l l et al. hypothesized that i f the generation effect is produced by activating S M  and i f  A D patients have a disruption of S M , then A D patients w i l l f a i l to produce the generation effect. Several tasks were administered to the subjects. The generation task consisted of three sets of 20 sentences o f the f o r m S U B J E C T V E R B O B J E C T ( S V O ) . The S U B J E C T and V E R B were always provided w h i l e the O B J E C T position was either blank or supplied and underlined. Subjects read the sentence aloud and either read the O B J E C T  or generated an O B J E C T .  Subjects were then  administered the M i n i M e n t a l Status E x a m . Next, Subjects were asked to name as many  U.S.  presidents as they could i n two minutes. This served as a distractor and a test of semantic memory. Subjects were then given the S U B J E C T f r o m each S V O sentence and asked to recall the O B J E C T f r o m that sentence, a cued recall task. The reality monitoring task was the final task. Subjects were given the correct S O  pairs and were asked to judge whether they had read or generated the  OBJECT. Results f r o m the cued recall task indicated that w h i l e the generation effect was robust for the older normal subjects, the A D  patients showed no benefit f r o m generational processing. For the  reality monitoring task both younger and older normal subjects showed the reality monitoring effect, but the A D patients could not identify the information w h i c h they had generated at a level greater than chance. The fact that the A D  patients performed significantly more poorly than the  36 n o r m a l controls on the task w h i c h  required  them to name U S  presidents was  interpreted as  evidence of the A D patients' S M difficulties. M i t c h e l l et al. concluded that because the A D effect nor reality monitoring, S M  patients demonstrated neither the generation  must be disrupted i n A D . They note that the tasks w h i c h were  administered (excluding the presidents task) d i d not require effortful processing. M i t c h e l l et al. noted several sources of evidence to support the c l a i m that A D  patients demonstrated impaired  activation of and retrieval f r o m S M : 1) although the words generated by the A D patients d i d not differ significantly f r o m those generated by the normal subjects, the A D to 30 seconds to generate a word 2) the A D  patients often needed up  patients had m u c h difficulty recalling names of  presidents 3)although the A D patients generated suitable words i n the generation task, these words were not processed  i n the beneficial manner that normally produces  the generation effect i n  episodic recall. M i t c h e l l et al. proposed the f o l l o w i n g to explain their results. W h e n a subject  generates  information, he first activates the semantic representation then accesses the l e x i c a l representation. W h e n information is externally presented, the subject first activates the l e x i c a l representation then the semantic representation. Cued recall involves the activation of semantic memory, then lexical memory, and thus is more similar to the generation input. Therefore, recall for internally generated information is better than for externally presented information. M i t c h e D et al. suggested that the failure of A D  patients to produce the generation effect may be due to a difficulty of the lexical  decoding of a semantic representation. That is, the deficit may lay i n the ability to translate a semantic representation into a l e x i c a l representation. In this chapter findings have been discussed w h i c h indicate that patients w i t h A l z h e i m e r ' s Disease have deficits of l e x i c a l memory, semantic memory and the ability to map between lexical and semantic memory. The status of semantic memory i n A D patients remains an issue of debate. S o m e researchers have found it to be impaired while others have found it to be spared. W e have identified some variables w h i c h have important effects on the results and their interpretation: the severity of dementia, the type of task used and what it is testing, and the amount of effortful processing required to perform the task. The state of semantic memory patients remains unclear. The problem to be investigated is whether A D  i n A l z h e i m e r ' s Disease patients have a semantic  37 memory deficit. In order to investigate this problem, the variables of dementia severity, task type, and effortful processing must be controlled.  2.2 Aging and Semantic Memory  There is an extensive research literature concerning the effects o f aging  on  memory.The  majority of the research has examined episodic memory abilities of older adults. C r a i k and B y r d (1982) reviewed this literature, and concluded that older normal subjects have a deficit of episodic memory i n that they f a i l to carry out deeper inferential processing. W h e n asked to recall w o r d lists under various conditions, older subjects consistently performed more poorly than younger subjects. In tests of memory, results are scored according to the number of items recalled or recognized correctly. Another frequently used measure is reaction time. However, to interpret the results of reaction time studies  with older subjects  correctly, one  must  take into account the  general  behavioural s l o w i n g w h i c h occurs with age. If older subjects respond more s l o w l y than younger subjects, is it because of a memory deficit or behavioural slowing? Birren, Woods, and W i l l i a m s (1980) reviewed the research regarding behavioural slowing with age. The essence of their review was that there is a general decline i n speed of behaviour w i t h age, and that the slowness is evident not only i n simple motor responses and sensory phenomena, but also i n more complex behaviours such as memory and performance on tests of general intelligence. B i r r e n et al. reported that i n simple reaction tests, older subjects typically respond  13-20%  slower than younger subjects. This is perhaps the most replicated finding of behavioural change w i t h age. However, Birren et al. found little evidence to support the attribution of behavioural slowing to slower neural conduction velocity. The  fact that older subjects  show  slower reaction times during  complex cognitive  tasks  indicates that s l o w i n g is not l i m i t e d to simple movements. B i r r e n et al. discussed some of the early research on intellectual changes across adulthood. Although these studies showed that intellectual abilities declined w i t h age, it has proven impossible to eliminate the confounding effect o f cohort on the intellectual performance of older adults. One generally accepted finding i n this area of  38 research is that abilities w h i c h are the first to show decline with age are those w h i c h require speed: spatial visualization, inductive reasoning, and word fluency. In  order to avoid cohort effects, some researchers  have  used an information  processing  approach to investigate skills across adulthood. This approach divides cognitive processes into a sequence of steps, beginning w i t h stimulus acquisition and ending w i t h a response. T h e processing time required at each stage can be measured. A g e differences have been found for a l l of the five following  steps: peripheral feature analysis, central feature analysis, sensory store, short-term  memory, and long-term memory. B i r r e n et al. reported the results of a study i n w h i c h backward masking of visual stimuli was used to test peripheral processing. Older subjects showed significantly slower peripheral processing by their need for a longer interstimulus interval to escape masking. Central processing, occurs subsequent to peripheral processing, has also been investigated using a masking  which  paradigm.  Substantial age differences were found i n processing speed for central processing as w e l l . The next stage of information processing is the sensory store, also k n o w n as iconic memory. Iconic memory consists of an image of the stimulus w h i c h is holistic and not yet represented as a symbolic memory. B i r r e n et al. reported the results of several studies w h i c h indicated age-related differences i n iconic memory.  Certain aspects of iconic memory are selected to be transferred  to the next stage of information processing, i.e. short-term memory. B i r r e n et al. found no evidence to indicate a reduction i n the capacity of short-term memory with age, although there appears to be a reduction i n the speed at w h i c h items i n short-term memory are scanned. Scanning o f short-term memory is the sequential review of items i n short-term memory. W h e n young subjects are required to determine whether an i t e m has been presented among a list of items, their response latencies increase with the number of items i n the list, assuming that the number does not exceed the capacity of short-term memory. The latency is presumed to be a measure of short-term memory scanning speed. Several studies have indicated that for a given set size, the response latencies of older subjects are longer than those of younger subjects (Birren et al., 1980). These results indicate a slowing of the scanning rate i n short-term memory with age. B i r r e n et al. d i d not review Uterature regarding age differences i n long-term memory, but they noted that several studies  have  found  a correlation between memory  abilities and speed  of  39 behaviour. F o r example, the Weschler M e m o r y  Scale showed no correlation w i t h digit writing  speed for young subjects, while the two measures were significantly correlated for older subjects. A l t h o u g h behavioural slowing occurs with age, B i r r e n et al. noted that environmental factors also  have  an effect  on  speed  of behaviour.  Patients  w i t h hypertension,  heart disease,  and  cerebrovascular disease respond more slowly than patients without vascular problems. B i r r e n et al. suggested that the behavioural slowing  seen i n healthy  older adults  may  be due to  minor  cerebrovascular problems, since the relationship between age and atherosclerosis is well-known. E p i l e p t i c and brain-damaged subjects also showed slower response times than age-matched control subjects. One study found a significant interaction of the effects of age and brain damage  on  simple reaction time. B i r r e n et al. pointed that reaction time tests have been found to be very sensitive to both diffuse and focal brain damage. Based on these results, we can predict that dementia patients w o u l d show increased response latencies as compared to older normal subjects, and that reaction time measures may be sensitive to the presence o f A D . L a w s o n and Barker (1968) have shown reaction time to be a more sensitive measure of dementia than number of correct responses i n a naming task. L a w s o n and Barker assessed 100 dementia patients (etiology o f dementia was not reported) and 40 normal control subjects using a naming task. F o r half of the items, the use of the object was demonstrated and subjects were allowed to manipulate the object. The remaming  items were not demonstrated.  Reaction times were recorded using a stop watch. If an item had not been named 50 seconds after presentation, the trial was discontinued and scored as a failure. Results indicated that low-frequency words required longer reaction times for dementia patients and control subjects, and that dementia patients showed longer reaction times than control subjects for both l o w and high frequency words. The w o r d frequency effect was more pronounced for dementia patients than n o r m a l subjects. Demonstrating the use of the object decreased the reaction times for dementia patients but had no effect on the reaction times of the n o r m a l subjects. L a w s o n analyzed the results f r o m the reaction time measure and the pass/fail measure. The number of true and false positives (i.e. whether the task correctly identified the dementia patients) were calculated for both measures, and receiver-operator curves were created. Results f r o m this analysis indicated that use of the reaction time measure achieved a more accurate separation of  40 groups than the pass/fail measure. L a w s o n and Barker concluded that the use of reaction time measures was advantageous  i n distinguishing dementia patients f r o m normal age-matched adults.  They also concluded that the use of less common objects increases the efficiency of the test, while demonstrating the use of test items decreases the efficiency of the test. L a w s o n and Barker's results can be interpreted as evidence for increased behavioural slowing caused by diffuse brain damage associated with dementia. B i r r e n et al. reported that  Parkinson's  Disease, schizophrenia and depression were also associated w i t h a loss of speed. Birren et al. reported further evidence to support the theory that environmental factors interact w i t h general behavioural slowing. Several studies have indicated that the slowness of older persons may be attributable i n part to a lack of physical fitness. Some researchers have suggested that physical activity increases the arousal of the central nervous system, thus increasing  responsiveness.  B i r r e n et al. suggested several causes o f behaviuoral slowing w i t h age. 1) Older subjects may be less inclined to respond quickly because of the fear of making a mistake. 2) Older subjects' lower capacity to attend to stimuli may cause slowing of behaviour or slowness of behaviour may cause l o w e r attention. 3) There may be more than one cause of behaviuoral slowing: "Perhaps one should envisage a primary ontogenic slowing i n a subcortical structure that can be influenced by additional processes, such as learned cautiousness, or localized tissue damage i n a wide range of structures. "6 B i r r e n et al. also found evidence of neurobiological causes of behavioural slowing with age: 1) W i t h age, there is a loss of extrapyramidal cells i n humans. 2) There is a decrease i n the overall number of neurons  w i t h age, although no u n i f o r m pattern has been observed. 3) Reduced c e l l  counts i n the locus ceruleus have been reported. Interestingly, deficits i n this area of the brain have been implicated i n A D as w e l l (Bayles, 1987). 4) A reduction i n the number of synapses with age has been reported. This indicates that the number o f connections with other cells has been reduced. Reductions i n the number of intercellular connections i n the brains of A D patients have also been reported. 5) Neurotransmitters have been found i n reduced numbers i n the brains o f aging subjects, as w e l l as A D  patients. M o r p h o l o g i c a l changes, reduction of neuronal connectivity, and reduced  41 amounts of neurotransmitters can be associated w i t h a slowness of neural events and with memory changes. The similarity of these results with those f r o m A D patients deserves further investigation. B i r r e n et al. concluded that there is a general loss of speed of behaviour w i t h age.  This  slowing is reflected i n a number of behaviours, ranging f r o m simple motor responses to complex cognitive operations. Birren et al. suggested that the loss of speed is a reflection of a general mediating process i n the central nervous system. Behaviuoral slowing has been implicated i n the episodic memory deficits of older subjects. Because the purpose of this study is the investigation of semantic memory i n A D patients, it is necessary to review the research w h i c h has investigated the semantic memory  abilities of older  n o r m a l subjects. There is some evidence to indicate that older subjects exhibit slowing of semantic memory as w e l l as episodic memory. F o r tasks i n w h i c h the older subjects are required to make a motor response,  the investigators  must  control for slower motor response  time and  sensory  processing time i n order to accurately measure any age differences i n memory. B o w l e s and P o o n (1981) used several methods to control for motor response time i n a group of older subjects. The purpose of the study was to investigate the effect of aging on speed of l e x i c a l access, using response  a l e x i c a l decision task. Bowles  time during  and Poon noted that an age-related slowing  a l e x i c a l decision task could be due to sensorimotor  and/or  in  cognitive  processing. They used three methods to separate the sensorimotor and cognitive components: The subtraction method, the analysis of covariance, and the additive-factor method. In the subtraction method, latencies for an appropriate sensorimotor control task are subtracted f r o m the total latencies, supposedly revealing the latencies for the cognitive task. The success of this method depends on the control task duplicating the experimental task i n a l l but one processing stage. A two-choice reaction time task was used as a control for the l e x i c a l decision task. The reaction time task i n v o l v e d the information processing  stages of sensory processing, decision,  response selection, and motor response. The l e x i c a l decision task was assumed to include these four stages plus a l e x i c a l access stage. W h e n latencies for the control task are subtracted, the remaining latencies should reflect l e x i c a l access time. In the analysis of covariance method, the portion of the error variance due to the covariate is statistically removed. The covariate i n this study was sensorimotor processing, as measured by the  42 choice-reaction task. If a significant age effect is found by analyzing the covariance, it is attributed to processing other than the sensorimotor processing represented i n the covariate task. If no effect is found, the age difference is attributed to sensorimotor processing. In the additive-factor method, the reaction time is assumed to be the sum o f a l l the times for the component stages w h i c h make up total processing. If a variable affects the total reaction time, it must affect one or more of the component processes. If two independent variables affect the same stage of processing, they are assumed to interact. I f they affect different processing stages, their affects w o u l d be additive and there w o u l d be no interaction. B o w l e s and P o o n varied the word-frequency o f the stimuli. This variable is k n o w n to affect reaction time, w i t h high frequency words resulting i n faster reaction times than l o w frequency words. It was assumed that i f age and w o r d frequency interacted, then both variables affected the same processing stage, namely lexical access. If age and w o r d frequency d i d not interact, then it w o u l d be concluded that age did not affect l e x i c a l access time. Subjects were required to decide whether two strings of letters were both E n g l i s h words by lifting a finger o f f one of two buttons. H i g h and l o w frequency words were presented randomly i n pairs of two words per trial.  In the choice-reaction task, subjects were required to lift one finger  off a button i n response to the words U P P E R or L O W E R . Results indicated that age d i d not significantly affect accuracy, but there was a significant age difference i n reaction time. Analysis based on the subtraction method indicated that there was a s l o w i n g of l e x i c a l access associated with aging. However, B o w l e s and P o o n noted that the control task, choice-reaction time, was probably not an accurate replication of the sensorimotor, decision, and response selection stages involved i n lexical decision. They suggested that it is possible that the insertion of a new stage effects the processing time of the other stages. The finding of an age difference i n l e x i c a l access using this method, then, is not valid. The analysis of covariance indicated no effect of age on l e x i c a l access time. The variance associated w i t h the choice-reaction time task accounted for the variance i n the l e x i c a l decision task. There were problems w i t h this analysis as well. A n that the two variables are independent. B o w l e s  assumption of the analysis of covariance is  and Poon reported that some  of the variance  associated w i t h choice reaction time was shared w i t h the age variable. Because the two variables  43 were not independent, the failure to find an age deficit i n l e x i c a l access time using this method is not a v a l i d result. The additive-factor analysis indicated no effect of age on l e x i c a l access time. There was no age by word-frequency interaction, indicating that the two variables affected different processing stages. Because word-frequency has been shown to affect l e x i c a l access, age difference must affect a processing stage other than l e x i c a l access. B o w l e s and P o o n concluded that age had no effect on speed of l e x i c a l access, and that speed differences i n l e x i c a l decision were due to age-related slowing at other processing stages. These results are opposed to those of many researchers w h o have found that older subjects have w o r d finding difficulties and perform more poorly than young subjects on tests of naming. It was stated i n Chapter 2.1 that l e x i c a l decision tasks and naming tasks test l e x i c a l access  and/or lexical  memory. Obler and A l b e r t reviewed the Uterature regarding the l e x i c a l abilities of older subjects. They found that the number of correct responses on the Boston N a m i n g Test increased between the ages of 30 and 50, held steady among 60-year-olds, and decreased among 70-year-olds. Institutionalized elderly  subjects  made  significantly  more  errors  than  Interestingly, institutionalized elderly made more errors  those  who  were  not institutionalized.  of misperception and exhibited more  circumlocution, w h i l e noninstitutionalized elderly made more semantic association errors.  Two  types o f errors increased w i t h age: comments on the task and circumlocutions w h i c h described the item. There was no consistent pattern of response to semantic and phonemic cues across age. Thus, it seems that naming ability decreases w i t h age. A n action naming test developed by Obler and Albert, indicated that the pattern of results across ages was the same for verbs as it was for nouns. Obler and Albert also reviewed older subjects' definitional abilities. They  noted that the  vocabulary subtest of the W A I S , w h i c h requires subjects to provide definitions for stimulus words, is scaled for age. F o r example, a score of 43 for a 70-year-old is equivalent to a score of 50 for a 30-year-old. T h i s definitional  suggested  to Obler and  abilities exist. A f t e r some  Albert that small, but reliable age differences i n  investigation, however,  Obler  and  A l b e r t that  these  differences are due to scoring criteria. The criterion for a g o o d definition, according to the W A I S scoring procedures, is a one-word synonym. However, m u l t i w o r d responses increase with age. If  44 f u l l credit is given for correct m u l t i w o r d definitions, older subjects do not differ f r o m young subjects i n their ability to define words. A l t h o u g h naming ability appears to decline w i t h age, one aspect of semantic memory, definitional ability, appears to remain intact. Bowles  and P o o n (1985) investigated the effects of aging on the retrieval of words  from  semantic memory. They used a l e x i c a l decision task and a word retrieval task. Based on their earlier (1981) work, they predicted that older subjects w o u l d perform as w e l l as young subjects on the l e x i c a l decision task. The purpose of the study was to investigate the c l a i m made by many older subjects that they are unable to find the desired words when speaking, despite the fact that vocabulary tests have indicated that there is no age difference i n accessing the semantic network v i a the lexical network. B o w l e s and P o o n hypothesized that older people have a problem accessing the l e x i c a l network v i a the semantic network. That is, they may be unable to map f r o m S M to L M . T o investigate this hypothesis, B o w l e s and Poon used a word retrieval paradigm. Subjects were presented with the definition of a target w o r d and were required to produce the target w o r d that was defined. They predicted that there w o u l d be an age difference i n success at word retrieval. P r i m i n g was used i n both the word  retrieval and l e x i c a l decision tasks.  Results indicated that there was no significant difference between the young and older subjects i n accuracy and response latency for the lexical decision task. Neither was there a difference i n the size o f the priming effect due to semantically related primes. F o r the w o r d retrieval task, an agerelated deficit was found. The older subjects were significantly slower to respond than the young subjects i n every priming condition. B o w l e s and P o o n suggested that older subjects had difficulty mapping f r o m the conceptually organized semantic network to the orthographically and phonetically organized l e x i c a l network i n the absence of orthographic/phonetic information. The importance of orthographic information was evidenced by the fact that older subjects showed m u c h better performance on the word retrieval task when the prime was orthographically related to the target. B o w l e s and P o o n concluded that the older subjects' difficulty i n word retrieval indicated a breakdown i n the connection f r o m the semantic to the l e x i c a l networks, while both semantic and l e x i c a l memory remain intact.  45 B u r k e and Y e e (1984) also found the semantic memory of older subjects to be intact. They administered  a  lexical  decision  task  which  was  primed  by  semantically  related  sentences.Irnmediately after reading a sentence, subjects were required to decide whether a string of letters was a w o r d f English. The target words were either unrelated to the sentence, related to the episode described i n the sentence as a whole, instruments i m p l i e d by the action described i n the sentence, or related to only one w o r d i n the sentence. Results indicated that there were significant main effects for age and sentence-target condition. The  related response time was  faster than unrelated response time only  for the  instrument  condition. There were no age-related differences i n accuracy. B u r k e and Y e e also investigated the subjects' memory for sentence primes. O n a recognition test, older subjects made significantly more errors than younger subjects. Burke  and Y e e  concluded that their results indicated no  age-related changes i n  semantic  processing, but that there were clear age differences i n retention of sentences. Their results highlight the importance of separating comprehension and retention processes i n semantic memory testing. They suggested that the conflicting results regarding the status o f semantic memory  in  older subjects may be explained by the amount of effortful processing required by the tasks used. Tasks w h i c h require retention also require effortful processing, and on these tasks age-related deficits i n semantic processing have been found. These conclusions parallel those regarding the semantic memory deficit i n A D patients. W a l s h and B a l d w i n (1977) used a linguistic abstraction paradigm to investigate age differences i n semantic processing. Subjects were presented a set of sentences, each representing a partial meaning of a complete idea. A f t e r reading each sentence, subjects were asked an elliptical question about the information contained i n the sentence. A f t e r the set of sentences were presented, subjects were administered a recognition test i n w h i c h they judged whether a sentence had been presented previously. Studies have indicate that subjects abstract information f r o m separate, related sentences and retain an integrated representation of the idea. Results  indicated no  significant age-related differences i n the accuracy  of acquisition  of  information. B o t h young and older subjects understood the sentences equally well. There was no significant difference i n recognition of sentences. B o t h groups retained equivalent amounts of  46 information regarding the surface structure f o r m of the sentences. The results suggest that both groups integrated the linguistic ideas to the same degree. W a l s h  and B a l d w i n concluded that  ecologically v a l i d semantic memory functions may not decline w i t h age. Lorsbach and Simpson (1984) found evidence for age differences i n semantic memory. They used  a probe-recognition task  to investigate  the hypothesis  that presentation rate and  task  complexity have different effects on the memory skills of young and older adults. In the proberecognition task, subjects were presented w i t h a series of 10-word lists. The exposure time for each w o r d varied f r o m 350 to 1400 milliseconds. F o l l o w i n g each list, a cue and a probe word were presented. The cue indicated to the subject whether he was to judge i f the probe w o r d  was  identical to a w o r d presented i n the list, was a h o m o n y m or a synonym of a word i n the list. Lorsbach and Simpson predicted that i f complex semantic analyses require more processing time and i f older subjects have slower processing times, then older subjects should perform significantly poorer than on synonym recognition at shorter presentation rates. Accuracy synonym  results indicated that identical and h o m o n y m  probes  differed significantly f r o m  probes, but not f r o m each other. There was a small age effect, but an insignificant  presentation effect. A l t h o u g h accuracy was unaffected by age, response time was. Older subjects required significantly more time to respond to synonym probes, but presentation rate d i d not affect young and older subjects differentially. The results d i d not support the speed-loss hypothesis, but d i d support the hypothesis that older subjects are slower than young subjects i n gaming access to semantic information. These results do not indicate that the structure of semantic memory is deficient i n older adults, but that access to S M requires more processing time. The probe-recognition task requires effortful processing. Thus, it appears that older adults only show S M  deficits when effortful processing is  involved. T h i s lends support to C r a i k and B y r d ' s (1982) theory that aging produces a reduction i n attentional resources. C r a i k and B y r d (1982) suggested that: "...decreased availability of attentional resources i n older people reduces the amount of spontaneously initiated deep, elaborate, and inferential processing that is carried out and this reduction i n turn is associated with lower levels  47 of retention. W e do not wish to i m p l y that no semantic processing is carried out: rather, it seems that the semantic processing achieved is more general i n character. "7 T h e y have supported this theory by citing evidence w h i c h shows that on a recognition task, older subjects and subjects under divided attention conditions make the same types of errors. Specifically, they make more false alarm responses to synonyms of target words. O l d e r subjects, amnesic patients, and subjects under divided attention conditions a l l make more intrusions f r o m previous lists into their free recall of current lists. It is assumed that subjects under divided attention conditions have a reduction i n the amount of attentional resource available for the performance of the target task. The fact that older subjects respond similarly to subjects under conditions of d i v i d e d attention indicates that older subjects may also have a reduced amount of attentional resource available. Further evidence to support C r a i k and B y r d ' s theory comes f r o m the f i n d i n g that automatic aspects of information processing appear to be intact i n older subjects, w h i l e effortful, conscious processes are impaired. T h i s conclusion is indicated by the hterature reviewed i n this section and by the literature regarding A D and semantic memory (see Section 2.1 of this chapter). Based on this theory, B y r d (1984) predicted that older subjects w o u l d show poorer performance than young subjects on a test of semantic memory w h i c h required them to make a conscious effortful search of semantic memory. B y r d carried out two experiments to test his prediction. In the first experiment, subjects were presented a category name, f o l l o w e d after 3 seconds by either a letter or a category exemplar. W h e n provided with a category name plus an exemplar, the subject's task was to decide i f the exemplar was a member of the category (decision task). W h e n provided w i t h a category name plus a letter, the task was to generate an exemplar of the category that began w i t h the given letter (generation task). H a l f of the subjects were presented w i t h a random mixture of generation and decision trials (mixed trials condition), w h i l e the remaining subjects were presented with blocks  of generation and decision trials (blocked trials condition). B y r d  was  interested i n the effect of blocking because other investigators had found that knowledge of the question format for the generation trials allowed subjects to begin a search o f S M as soon as the category name was presented. This allowed a reduction i n the time necessary to f i n d a correct  48 answer. B y r d predicted that older subjects w o u l d benefit f r o m this condition to a greater extent than young subjects. Results indicated that older subjects showed longer response latencies than young subjects for both conditions. B o t h groups had longer response latencies i n the mixed trials condition than i n the b l o c k e d trials condition. B o t h groups o f subjects took longer to respond to l o w dominance than high dominance words. B o t h groups showed longer response latencies for the generation than the decision task. There was a significant interaction between age, condition and task, w i t h young subjects producing faster responses on the generation task i n the blocked trials condition. In contrast to B y r d ' s prediction, young subjects benefitted more f r o m the b l o c k e d trials condition of the generation task. T o explain this finding, B y r d noted that his subjects were informed of the nature of the b l o c k e d trials and were encouraged to use this information. Perhaps the young subjects were better able than the older subjects to use this information to initiate a search of S M . B y r d suggested that the differential performance on the two tasks as a function of age was due to an interaction between task demands and processing abilities.He suggested that the decision task i n v o l v e d more automatic, less effortful processing than the generation task because the given information i n the decision task directs the search of semantic memory. T h e decision task involves task-driven  processing,  w h i l e the generation task  involves  subject-driven processing.  In  the  generation task, the information is not enough to activate automatic processing, and the semantic memory structures must be consciously analyzed. If older subjects have less available attentional resource, then they w o u l d perform more poorly on the generation task than on the decision task. B y r d ' s second experiment was designed to: "...examine the use of information that is semantically related to the questions i n order to determine i f both groups are capable of facm^ating automatic information processing to the same degree."8 Subjects were presented w i t h category-word pairs as i n the decision task of Experiment 1, except that the members of the pairs were presented simultaneously. The pairs were presented w i t h zero, one, or two intervening items between two pairs of the same category (0,1, or 2 item lag). B y r d predicted that no  age  differences i n the size of priming w o u l d be observed because  older  49 individuals  should  be able to make use of automatic processing  strategies  as w e l l as  young  showed longer response latencies than  young  individuals. Results  indicated that while older subjects  subjects, the two groups responded equally to the priming. B o t h groups responded faster i n the 0 lag condition. B y r d concluded that although there was an age decrement i n the ability to activate S M , there was no decrement i n the ability to facilitate automatized information retrieval processes. H i s experiments demonstrated the usefulness of the effortful/automatic processing dichotomy interpreting results regarding age differences i n S M .  when  50  CHAPTER 3 OBJECTIVES  3.1 Original Objectives  The purpose o f the present study is to investigate the semantic memory o f a group o f patients w i t h the diagnosis o f probable A l z h e i m e r ' s Disease, b y replicating the first experiment o f B y r d ' s (1984) research. T h e a i m o f this study is to determine whether A D patients exhibit a deficit of semantic memory. In the discussion i n Chapter 2, it was concluded that A D patients appear to exhibit a deficit o f S M when performing tasks w h i c h require effortful processing. W h e n the task involves automatic processing, however, A D patients appear to have intact S M abilities. It has been suggested that n o r m a l older subjects have reduced attentional resources i n comparison to young subjects, and that perhaps A D patients have a further reduction o f attentional resources. G i v e n B y r d ' s suggestion that the generation task involves more effortful processing that the decision task, it is predicted that the A D patients w i l l perform more poorly on the generation task than on the decision task. F o l l o w i n g f r o m the suggestion that A D patients have a reduced capacity for effortful processing, it is predicted that the A D patients w i l l perform more poorly than a group of n o r m a l elderly subjects o n the generation task. The locus o f the verbal memory deficit was discussed i n Chapter 2. There is evidence that A D patients have a deficit o f l e x i c a l as w e l l as semantic memory. There is some evidence that A D patients have difficulty accessing L M v i a S M or S M v i a L M . O n e o f the important variables i n memory research is the type of memory w h i c h a task requires. D o the generation and decision tasks test S M , L M , or the ability to map between them? It is suggested that, to perform the  generation task a subject must access the l e x i c a l  representation o f the category name, map the l e x i c a l representation to the conceptual representation in S M , analyze the conceptual representation searching f o r members o f the category, then map the conceptual representation o f each category member to its l e x i c a l representation checking f o r a  51 member w h i c h begins with the specified letter, then choose the correct l e x i c a l f o r m and respond verbally (see Fig.4). It  is suggested that, to perform the decision task  a subject must  access  the lexical  representation o f the category name, map the l e x i c a l representation to its conceptual representation in  S M , then  access  the l e x i c a l representation  o f the exemplar, map it to its conceptual  representation i n S M , compare the two conceptual representations to determine i f the exemplar is a member o f the category, and respond verbally (see Fig.4). L e x i c a l access to the exemplar word is assumed to occur after the conceptual representation o f the category has been accessed because the category name is presented f o r 3 seconds prior to presentation o f the exemplar. However, given the extremely long response latencies o f some o f the A D patients, and given the theory that lexical access is impaired i n A D , w e cannot assume that 3 seconds is sufficient time to a l l o w a l l subjects access to S M . Therefore, f o r some subjects, access to the conceptual representations o f the category and exemplar may occur simultaneously. In contrast to a naming task, w h i c h can be performed by matching l e x i c a l representations directly to visual information thereby bypassing S M altogether, the generation and decision tasks i n v o l v e activation o f S M . I f A D patients have a deficit o f L M only, they w i l l perform better o n the decision task than the generation task because both l e x i c a l forms are provided i n the decision task. O n the generation task, however, only one f o r m is provided. T h e subject must generate the lexical f o r m o f the exemplar. I f A D patients have a deficit o f S M only, they w i l l perform equally poorly on the generation and decision tasks because S M must be accessed f o r both the category and the exemplar i n each task. If A D patients have a deficit i n the ability to access L M v i a S M , they w i l l perform more poorly o n the generation task than o n the decision task because the generation task provides less l e x i c a l information f o r the subjects to use as a cue i n accessing L M . If A D patients have reduced attentional resources, resulting i n a deficit o f semantic memory, then they w i l l demonstrate longer response latencies than normal older subjects o n the generation task. They w i l l demonstrate better performance o n the decision task than o n the generation task. The hypothesis f o r the purposes o f this study is: that patients with A D w i l l demonstrate a pattern of performance w h i c h is not different f r o m that o f a group o f normal older adults.  52  Figure 4 Exprimental Tasks  Generation Task  visual input  >LM >SM"—(category (category) name)  SM  >LM  A  > SM  >LM  ""~^SM  >LM  >vresponse \ choice  Decision Task (analysis of category membership) visual input  >LM >SM (category (category) name) visual input  >LM (exemplar name)  /  > decision T  >SM (exemplar)  >)  53  3.2 Revised Objectives  D u r i n g the course o f w o r k on this experiment, some unforseen problems arose. I n dealing with these problems, the original aims of the study were revised. 1) T h efirstproblem involved the difficulty o f finding subjects. T h e c l i n i c a l diagnosis o f A D is a long and complicated process. M a n y patients are referred to an A l z h e i m e r ' s  Disease clinic  because other sources, f a m i l y physicians and specialists, have been unable to discover the cause of the patient's memory problem. Since the process o f assessing a patient is so arduous, only a small number o f patients are seen over a short period o f time. O f the 13 subjects w h o were interviewed for this study, only 7 were diagnosed as having A l z h e i m e r ' s Disease. O f these 7 subjects, the data f r o m t w o subjects could not be used. T h e remaming 6 o f the 13 subjects exhibited dementia, but the etiology was unknown. Because o f the small number o f subjects, rigorous statistical analyses c o u l d not be applied. The subjects were divided into three groups: 1. subjects i n the pilot experiment, 2. the A D patients, and 3. the dementia patients. T h e f o l l o w i n g chapters  will  deal with these  groups  separately. 2) The  second problem became apparent when the first f e w subjects were interviewed. These  subjects responded inappropriately to the task b y making irrelevant comments, commenting about the task, or b y producing circumlocutions. They frequently asked h o w to respond to the task, or indicated b y their comments  that they d i d not understand the nature o f the task. A f t e r these  subjects were seen, the instructions were modified and more examples added. 3) T h e  third problem also became apparent w i t h the pilot subjects. T h e pilot subjects often  appeared unable to respond to the trials without help f r o m the experimenter. T h e experimenter provided prompts designed to elicit a response. Because it seemed likely that several subjects w o u l d require prompting and because the use o f a prompt to elicit a response affects reaction time, the method o f prompting was standardized. F o r the remaining subjects, a standard set of prompts were used, and a record was made o f each trial f o r w h i c h a prompt was used. Reaction time data f r o m prompted trials were excluded from the analyses.  54 4) Despite the size o f stimuli, some of the pilot subjects had difficulty reading the words. In order to carry out the experiment, the experimenter read the words aloud to subjects as they appeared on the computer screen. Because of the small number o f subjects, it was desirable not to exclude subjects on the basis of vision. Therefore, when it appeared that the subject was having difficulty reading the stimuli, the experimenter offered to read them aloud. 5) M a n y of the subjects presented with conditions w h i c h w o u l d complicate the diagnosis AD.  of  However, because of the s m a l l number of subjects available, subjects w i t h comphcating  conditions were not excluded. 6) Because of the s m a l l number of subjects available it was not possible to investigate the effect of severity of A D , although some general comments w i l l be made. The purpose of this study is to investigate the effects of A D because  of the s m a l l number  of subjects, rigorous  on semantic memory. However,  statistical analyses  of the many  factors  influencing semantic memory performance w i l l not be possible. The predictions made i n Section 3.1 cannot be tested with any accuacy, although some tentative conclusions may be drawn.  55  CHAPTER 4 METHODS  4.1 Subjects  4.11 Pilot Experiment  T w o male and one female subject participated i n the pilot experiment. They ranges i n age from 63 to 77 years (mean= 71.7 years). These subjects were drawn f r o m a population o f patients who were assessed at the University (HSCH)  Akheimer's  o f B r i t i s h C o l u m b i a ( U B C ) Health Sciences  C l i n i c (10). Subjects  Centre Hospital  i n both the pilot experiment and the main  study  participated during their initial assessment at the A l z h e i m e r ' s C l i n i c . Thus, diagnosis was unknown at the time that each subject participated. T w o subjects i n the pilot experiment received diagnoses of A D , although they presented w i t h conditions w h i c h complicated the diagnosis. T h e third subject received a diagnosis of Parkinson's Disease with associated dementia (see A p p e n d i x 1 A f o r more detailed information regarding the pilot subjects).  4.12 Experimental Subjects  F i v e patients with a diagnosis  o f A D participated i n the main study, one male and four  females. They ranged i n age f r o m 64 to 73 years (mean= 69.4 years). T h e subjects had a mean education l e v e l o f 11.9 years, w i t h a range of 10 to 13.5 years. One patient was diagnosed as moderately impaired w h i l e the other four patients exhibited m i l d A D . Three patients presented with conditions  w h i c h complicated the diagnosis  of A D  (see A p p e n d i x  IB  f o r information  about  individual subjects). These patients were also drawn f r o m the U B C H S C H A l z h e i m e r ' s C l i n i c and participated as part o f their initial assessment. A s w i t h the subjects w h o participated i n the pilot experiment, the diagnoses were made f o l l o w i n g the subjects' participation i n the study.  56 The group o f patients w h o were drawn from the U B C H S C H A l z h e i m e r ' s C l i m e included the pilot subjects, the A D patients and the dementia patients. Because subjects were seen during initial assessment and because it was not possible to determine the total number o f subjects prior to running  the experiment, subjects  were assigned to the blocked  and m i x e d  trials  conditions  alternately i n consecutive order. That is, subject #1 was assigned to the b l o c k e d trials condition, subject #2 was assigned to the m i x e d trials condition and so on. This procedure resulted i n the assignment  o f four A D patients to the m i x e d trials condition and one to the blocked trials  condition. A  group o f 2 0 n o r m a l older individuals, 10 male and 10 female, served as control subjects  (11). T h e control subjects ranged i n age f r o m 64 to 83 years (mean = 71.2 years) and had an average o f 13.2 years o f education (range= 6 to 2 0 years). H a l f were assigned to the b l o c k e d trials condition and h a l f to the m i x e d trials condition. A l l subjects were administered a short version o f the M i l l H i l l Vocabulary Test. The control subjects displayed slightly higher vocabulary scores (mean= 16.5) than the A D patients (mean= 14.0). this may have been attributable to the control subjects' higher mean l e v e l o f education.  4.13 Dementia Patients  O f the remaining five subjects, three received a diagnosis o f dementia f o r w h i c h the etiology was unknown, one received a diagnosis o f circumscribed memory deficit specific to the retrieval o f new information, and one had not received a diagnosis. She was referred to the U B C H S C H Alzheimer's  C l i n i c o n the basis of the f o l l o w i n g symptoms:  memory  disturbance, personality  disturbance, and increased dependence on her spouse. Three o f the dementia patients were drawn f r o m the U B C H S C H A k h e i m e r ' s C l i n i c and participated i n the experiment as part o f their initial assessment. Subjects #12 and #13 were inpatients at V a l l e y v i e w Hospital (12). A l l o f the dementia patients presented w i t h symptoms w h i c h w o u l d complicate a diagnosis o f etiology (see Appendix I C f o r more detailed information about the dementia patients). The dementia subjects ranged i n age f r o m 61 to 8 2 years (mean= 72.6 years)and had an average  o f 8.4 years  o f education (range= 6 to 11 years). This  group demonstrated  lower  57 vocabulary scores  (mean= 9.2) than the A D  patients and the control subjects. Four dementia  subjects were assigned to the blocked trials condition and one to the m i x e d trails condition.  4.2 Stimuli  The stimulus materials used i n this investigation are similar to those used i n B y r d ' s (1984) experiments. T w e l v e categories were chosen f r o m the Battig and Montague (1969) category norms (see A p p e n d i x 2). T w o high dominance and two l o w dominance exemplars were chosen f r o m each category. Those exemplars chosen by 5 0 % or greater o f the subjects i n the Battig and Montague norms  were considered to be high  dominance  exemplars. E q u a l numbers  of high  and  low  dominance exemplars were chosen i n order to control for the effect of category dominance on reaction time. F o r each category, the first letter of one high and one l o w dominance exemplar were used to create generation pairs. The remaining words were used i n f u l l to create decision pairs. Thus, 24 category-letter generation pairs and 24 category-word decision pairs were constructed. In addition, 24  incorrect category-word  decision  pairs  were  constructed f r o m  the  original  12  categories, to act as distracters i n the decision task (see A p p e n d i x 2 for a list of the stimulus items). S i x category-word decision pairs and three category-letter generation pairs were presented to the subjects as examples. The experimenter demonstrated the first few examples and then guided the subject through the remaining examples, explaining the response required. T e n category-word decision pairs and ten category-letter generation pairs served as practice trials. There were two conditions: blocked trials and m i x e d trials. In the blocked trials condition, blocks of 6 category-letter generation pairs or 12 category-word decision pairs were presented alternately. There were 4 blocks of generation trials and 4 blocks o f decision trials. In the mixed trials condition, there were 9 pairs per block, each containing a random mixture of generation and decision trials. The order of trials i n each b l o c k was randomly assigned for each subject (see A p p e n d i x 3).  58  4.3 Procedure  E a c h subject was seen for one 1-hour session. Subjects were informed that the purpose of the study was to investigate the ability to retrieve words f r o m memory. A standard set of instructions were read (see A p p e n d i x 4). Subjects were told to answer each question as q u i c k l y as possible, but to try to answer correctly. D u r i n g the administration of the example trials, subjects were instructed as to the proper type of response and were prompted often. Subjects i n the b l o c k e d condition were informed of the nature of the blocked trials and encouraged to use that information. A computer program written i n B A S I C (9) allowed stimuli to be presented on the screen of a Zenith 158 personal computer with N E C M u l t i s y n c colour monitor, w h i c h was controlled by the experimenter. W o r d s were approximately one half i n c h high, and the subject was seated two to three feet from the screen. A s noted i n the previous chapter, for those patients w h o could not read the stimuli, the experimenter read the words aloud. W h i l e this may have affected reaction time, none of the A D group required this modification. O n each trial, the category name appeared first. T h e category name (e.g. Fruit) remained on the screen for 3 seconds. T h e n either a letter or an exemplar appeared on the screen to the right of the category name. T h e appearance of the letter or exemplar was accompanied by a short beep. W h e n a category-letter pair appeared on the screen, the subject's task was to produce an exemplar of the category beginning  with that letter. W h e n a category-word pair appeared on the screen, the  subject's task was to decide i f the w o r d was an exemplar of the category, and respond verbally ' Y E S ' or ' N O ' . The pair remained on the screen until the subject responded. W h e n the subject responded, the examiner pressed a key on the computer w h i c h terminated the computer's timer and automatically recorded the reaction time. The examiner then coded the response as correct or incorrect, i n the case of decision trials, or recorded the exemplar produced i n generation trials. B y r d used a voice-activated timer to record response latency. This was felt to be inappropriate for A D  patients because they show a tendency to comment on the task or rehearse  responses aloud before producing a correct response. A l t h o u g h the experimenter's response time is also included i n the reaction time measure, it remained fairly constant across trials.  59  There was  a break of t w o minutes between each b l o c k of trials. D u r i n g  this time, the  experimenter conversed w i t h subjects i n an attempt to help them feel more comfortable. Prompts, i n the f o r m of questions, were administered when the subject responded to the stimulus items but not i n the manner required. F o r example, when presented with the stimulus VEGETABLE:  C H E R R Y , a subject might respond, " W e l l , there's vegetables and there's cherries."  Prompts were also administered when the subject indicated that he/she d i d not k n o w how to respond. F o r example, a subject might say, "I don't k n o w what y o u want me to do." Prompts were also administered when the subject was silent f o r several seconds. This criterion, however, could not be applied strictly because of the different response strategies of the subjects. W h i l e some subjects were silent only  when they needed help, other subjects remained silent until they  responded correctly, even i f they took over 30 seconds to respond. F o r the decision task, subjects were prompted w i t h the f o l l o w i n g question: Is there a C A T E G O R Y called E X E M P L A R ? e.g. Is there a fruit called apple? F o r the generation task, the f o l l o w i n g prompt was used: T e l l me a C A T E G O R Y starting w i t h L E T T E R . e.g. T e l l me a fruit starting with A . F o l l o w i n g the experimental task, subjects were administered a short vocabulary test. Twenty items of graded difficulty were taken f r o m the M i l l H i l l Vocabulary Test. Subjects saw one test item at a time. Testing was discontinued after three consecutive incorrect responses were made. The m a x i m u m score attainable was 20.  60 CHAPTER 5 RESULTS  5.1 Pilot Experiment  Results of the pilot study indicated a need for changes in the methodology of the experiment. These have been discussed in Chapter 3. The pilot subjects appeared to have much more difficulty completing the task than normal older adults. Byrd (1984) noted that errors, omissions and anticipatory responses accounted for 6.9% of the generation trials and 3.3% of the decision trials. Subjects #1 and #2 had no difficulty completing the decision task; they both responded correctly to every trial. On the generation task, however, subject #1 made errors on 25.0% of the trials and subject #2 erred on 21.0% of the trials. Both subjects required prompting to complete the tasks, a problem not encountered by Byrd with his older subjects. Reaction time data from subjects 1 and 2 could not be analyzed because no record of prompted trials was kept, making it impossible to distinguish valid and invalid reaction times. Subject #3 had greater difficulty completing the tasks. During a one-hour session, he was only able to complete two practice lists and three experimental lists. There are eight experimental lists for each condition. Subject #3 required one or more prompts for most trials. On only five of 32 decision trials did he respond appropriately without prompting. Of these, four of the five responses were correct, and all were 'YES' responses. Given his tendency to perseverate on 'YES', the five trials cannot be assumed to represent an understanding of the task. Subject #3 required prompting for each generation trial. With prompting, only three of 16 generation trials were responded to correctly. While some errors were purely phonological in nature (e.g. Tree:F produced the response 'FREE'), others seemed to represent an inability to focus attention on both the category and the letter simultaneously. For example, in response to the stimulus Bird:C, subject #3 responded 'CAT'. When the experimenter prompted him to tfiink of a bird begjTjjiing with C, he responded 'ROBIN'.  61 The results f r o m the pilot experiment indicated the importance of considering the number and type of errors made by subjects, as w e l l as the need to use standard prompts, excluding prompted trials f r o m analyses of reaction times.  5.2 Main Experiment  O v e r a l l , the A D  patients appeared to have less difficulty completing the tasks than the pilot  subjects did. The A D patients had a mean of 10.8% errors on the generation task and 4 . 2 % on the decision task. These error rates were slightly higher than those reported b y B y r d for the older n o r m a l subjects. The A D  patients also required prompting. Trials during w h i c h prompts were  supplied were excluded f r o m analyses of reaction times. Table I represents the reaction time data and the percent errors f o r the A D  patients i n both  conditions for each task. F o r each subject, the mean response latency for the generation task is longer than the mean latency for the decision task. The standard deviation values for each subject i n the generation task are larger than those i n the decision task, with the exception of subject #6. This indicates that there was a larger amount of variability i n the reaction times for the generation task than for the decision task. The range of reaction times for each subject i n the generation task is larger than the range i n the decision task, also indicating a greater degree of variability i n the reaction times for generation trials. A g a i n , subject #6 is an exception. The data f r o m subject #6 seem inconsistent w i t h the pattern of results seen for the other A D patients. This inconsistency can be explained by the fact that subject 6 had so f e w trials for which reaction time was a v a l i d measure. She required prompting on a l l but eight of 24 decision trails. One of the eight unprompted trials resulted i n an abnormally long reaction time. This long reaction time affected variability to a greater extent than w o u l d be expected i f there were a larger number of reaction times i n the sample. Subject #8 was diagnosed as having moderate A D , w h i l e the other A D patients were diagnosed as mild. H e r mean response latencies were greater than those of the m i l d A D patients for both the generation and decision tasks (disregarding the decision task data f r o m subject #6).  62  Table I AD Patients Data Generation Task Subject #  Condition  Errors  (%)  Reaction T i m e (msec) Mean SD Range  4  Mixed  4.2  6060  4112  1695-15,757  6  Mixed  8.3  5640  3256  2310-11,149  7  Blocked  8.3  5771  3256  1700-15,220  8  Mixed  16.7  10,901  8869  3020-32,410  10  Mixed  16.7  5975  5661  1650-19,169  Decision Task Subject #  Condition  Errors  (%)  Reaction T i m e (msec) Mean SD Range  4  Mixed  4.2  1796  665  992-3679  6  Mixed  8.3  3255  4139  1479-13,459  7  Blocked  0.0  1673  546  8  Mixed  4.2  3339  1958  10  Mixed  4.2  1713  220  1040-3080 1650-10,160 1259-2080  63 Subject #7 was assigned to the blocked trials condition w h i l e the other patients were assigned to the m i x e d trials condition. Inspection o f Table I reveals the pattern of results for subject #7 to be the same as those for subjects i n the mixed trials condition. F r o m inspection of Figures 5 and 6, i n w h i c h the reaction times of subjects 4,6,8, and 10 have been averaged, it appears that the b l o c k e d trials condition had the effect of decreasing response latency for both the generation and decision tasks. However, because the data for the blocked trials condition are based on the results f r o m only one subject, the difference may not be reliable. Figures 5 and 6 serve only to illustrate the similarity of the pattern of results for the A D patients and the control subjects. The control subjects showed longer response latencies for the generation task than the decision task, with b l o c k i n g reducing the response latencies on both tasks. B l o c k i n g showed a greater effect for the generation than the decision task. The mean response latencies for the control subjects were as follows: generation task, m i x e d trials  2174 msec  generation task, blocked trials  1766 msec  decision task, m i x e d trials decision task, blocked trials  1095 msec 988 msec.  Comparing these results with those i n Table I, it appears that each A D  patient demonstrated  longer response latencies than the mean response latencies of the control subjects on the generation and decision tasks regardless of condition. The majority of errors made by the A D patients were semantic i n nature. Semantic errors were those i n w h i c h the subject produced a word beginning with the correct letter, but not belonging to the specified category, e.g. a response to F l o w e n P was ' P L U M ' . Other semantic errors were more subtle, i n that the generated word was semantically related to the category, i f not a member of the category. F o r example, a c o m m o n response to M e t a l : N was ' N A I L S ' . Phonological errors were those i n w h i c h the subject produced a word w h i c h belonged to the correct category, but did not begin with the specified letter. Phonological errors were rare. Another c o m m o n error was the inabiLity to produce a response. Subjects were encouraged to try to produce a response i n order to avoid overuse of the n u l l response. There were a small number of idiosyncratic errors w h i c h seemed to result f r o m spelling difficulties. F o r example, one subject responded to  64 Four-footed  Animal: R with 'ORANGUTAN', which she pronounced 'rangutan'. Other than this  type of error, the initial letter was usually correct.  5.3 Dementia Patients  Reaction time and error rate data from the dementia patients are presented in Table n. the dementia patients appeared to have more difficulty completing the generation task than the A D patients. The dementia patients erred on an average of 25.0%  of the generation trials. Subject #13  had the highest percentage of errors (62.5%). On the decision task, however, the dementia patients performed as well as the A D patients. The  dementia patients showed the same pattern of results as the A D patients. The mean  response latency of each dementia patient was greater for the generation task than the decision task. The dementia patients also showed greater variability in response latencies for the generation task, as indicated by the larger ranges and standard deviations. Subject #12 was assigned to the mixed trials condition, while the other dementia patients were assigned to the blocked trials condition. Although his mean response latencies were greater than those of the subjects in the mixed trials condition for both tasks, the severity of his dementia may have confounded the results. Subject #12 was diagnosed as having moderate dementia, subject #15 had a diagnosis of mild dementia, and the remaining subjects had no diagnosis of severity. It is possible that the subjects i n the blocked condition were all mildly demented, and the observed difference between conditions may have been due to a difference in severity. Subject #9 was diagnosed as having a circumscribed memory deficit. Strictly speaking, he was not demented. He was included in this group of subjects for convenience. Given that subject 9 did not have dementia, he would be expected to perform better than a group of dementia subjects on a test of memory, but given that he showed a memory deficit, he would be expected to perform more poorly than a group of age-matched control subjects. As expected, his mean reaction times on both tasks were shorter than those of the dementia patients, but longer than those of the control subjects.  7500  Figure 5 Mean R e a c t i o n T i m e s f o r RD P a t i e n t s a n d C o n t r o l Subjects - Generation Task Reac t i on T i me <msec >  5000 -  2500 -  B l o c ked M i xed  0 Con t ro1 Sub jec t s  RD Group  Patients  7500  Figure 6 Mean R e a c t i o n T i m e s f o r RD P a t i e n t s a n d C o n t r o l S u b j e c t s - D e c i s i o n Task Reac t i on T i me < msec >  5000 ON '  E!500  0  fl  I Blocked M i xed  Con t ro1 Sub j e c t s  RD Group  Patients  67  Table II Dementia Patients Data Generation Task Subject #  Condition  Errors (%)  Reaction Time (msec') Mean SD Range  5  Blocked  4.2  6172  5218  1809-26,690  9  Blocked  8.3  3373  4098  1050-20,429  11  Blocked  29.2  5448  8145  1378-33,339  12  Mixed  20.8  6937  7358  1801-32,410  13  Blocked  62.5  4951  2733  2687-9558  Decision Task Subject #  Condition  Errors (%)  Reaction Time (msec') Mean SD Range  5  Blocked  0.0  2150  602  1490-4390  9  Blocked  8.3  1428  399  929-3019  11  Blocked  4.2  2086  862  1320-4339  12  Mixed  4.2  3385  1110  1757-4886  13  Blocked  4.2  2746  579  2261-4562  68 The pattern of errors made by the dementia subjects was also similar to that of the A D patients. The majority of errors were semantic errors or n u l l responses. In general, the initial letter was retained correctly. There were some examples  of phonological errors i n the responses of the  dementia patients; i n response to M e t a l : N one subject said ' S T E E L ' . Subject #13, who produced the greatest percentage of errors, produced mostly semantic errors i n w h i c h me initial letter was correct, but the word had no recognizable semantic relationship to the category. F o r example, i n response to C o l o u r : R she said ' R U T H ' and i n response to F i s h : H she said ' H A C K ' .  69  CHAPTER 6 DISCUSSION  In general, the predictions made i n Chapter 3 were upheld by the results of this investigation. The f i r s t prediction was that A D patients w o u l d perform more poorly on the generation task than on the decision task because of the reduced attentional resources supposedly available to them, and the greater degree of effortful processing required by the generation task. The A D perform more poorly on the generation task  patients d i d  as indicated by the results f r o m both the reaction time  and error rate measures. These results suggest that effortful processing is more impaired i n  AD  than automatic processing. The second prediction was that A D  patients w o u l d perform more poorly than normal older  subjects on the generation task. A D patients showed longer response latencies and more errors than the control subjects on the generation task. A l t h o u g h A D patients also performed more poorly than the control subjects on the decision task, the difference between groups was m u c h smaller than for the generation task. T h i s result provides further support for the notion that A D  results i n a  reduction o f attentional resources. O n the task w h i c h requires more effortful processing, the  AD  patients showed m u c h greater impairment. A D patients appear to have intact automatic processing abilities, as studies such Nebes, M a r t i n and H o r n ' s (1984) have also shown. The  third  prediction i n v o l v e d the locus of the verbal memory deficit i n A D . If A D patients had  a deficit of S M , as Bayles (1987) suggested, they were expected to perform equally w e l l on the generation and decision tasks. The A D  patients d i d not perform equally w e l l on both tasks,  therefore, they must not have a deficit of S M  only. The results are compatible w i t h a deficit of  L M , the mapping between L M and S M , or a deficit of S M i n conjunction w i t h these two. The pattern of results are consistent with an interpretation based on the interaction between S M and effortful processing. W h e n the task involves effortful processing, as i n the generation task, A D patients perform more poorly than normal age-matched subjects. W h e n the task involves automatic processing, as i n the decision task. A D Patients perform similarly to normal control subjects.  70 M o s t of the errors made by the A D patients were semantic i n nature. This may indicate a S M deficit. If S M  is disturbed, subjects w i l l generate exemplars f r o m the wrong categories, thus  producing errors. However, S M  may be intact. If subjects generate exemplars f r o m the correct  category, but unable to access L M v i a S M , semantic errors w i l l be produced. G i v e n the L M deficit demonstrated by A D  patients on naming tasks (Bayles and Tomoeda, 1983), A  deficit of  LM  The results o f this study d i d not provide evidence for a semantic memory deficit i n  AD,  w o u l d also produce semantically related errors.  although they are consistent w i t h its existence. The hypothesis of this study was rejected.  AD  patients (and dementia patients) performed differently from normal age-matched control subjects, indicating that the generation and decision tasks may be useful diagnostic tools for identifying dementia patients. It is difficult to tell i f the tasks used i n this investigation w o u l d differentiate A D patients from dementia patients of different etiologies, since most of the patients i n the dementia group were not diagnosed as to etiology. The generation task may be better suited to identifying dementia patients because it involves effortful processing, w h i c h these patients seem to have more difficulty with. It is suggested that future research involve larger numbers  of A D  patients, i n order to  investigate the effects o f the mixed and blocked trials conditions and the effects of severity. The s m a l l number o f subjects i n v o l v e d i n this study limits the conclusions w h i c h can be drawn. The results of this investigation must remain cursory. One conclusion w h i c h can be drawn with a fair amount of certainty is that the group of subjects suspected of suffering f r o m A l z h e i m e r ' s Disease is not homogeneous.  71 Notes  1. Nebes, R.D., M a r t i n , D.C., and H o r n , L.C. 1984. Sparimg A l z h e i m e r ' s Disease. Journal o f A b n o r m a l Psychology. 93 (3), pp. 321.  o f Semantic  Memory  in  2. Bayles, K.A. and Kaszniak, A . W . 1987. Communication A g i n g and Dementia. Boston: Little, B r o w n and Company. Pp. 51.  and Cognition i n N o r m a l  3. Nebes, R.D., M a d d e n , D.J. 1988. Different Patterns o f Cognitive S l o w i n g Produced by A l z h e i m e r ' s Disease and N o r m a l A g i n g . Psychology and A g i n g . 3 (1), pp. 104. 4. Bayles,K.A. 1986. Management o f Neurogenic C o m m u n i c a t i o n Disorders Associated w i t h Dementia. In R. Chapey (Ed.), Language Intervention Strategies i n A d u l t A p h a s i a . Second E d . Baltimore: W i l l i a m s and W i M n s . Pp. 463. 5. M a r t i n , A . and Fedio, P. 1983. W o r d Production and Comprehension i n A l z h e i m e r ' s Disease: T h e B r e a k d o w n of Semantic Knowledge. B r a i n and Language. 19, pp. 135. 6. B i r r e n , J.E., Woods, A . M . and W i l l i a m s , M . V . 1980. B e h a v i o r a l S l o w i n g w i t h A g e : Causes, Organization, and Consequences. In L.W. P o o n (Ed.), A g i n g i n the 1980's: Psychological Issues. Washington D.C.: A m e r i c a n Psychological Association. Pp. 303. 7. Craik, F.I.M. and B y r d , M . 1982. A g i n g and Cognitive Deficits; the R o l e o f Attentional Resources. In F.I.M. C r a i k and S.E. Trehub (Eds.), A g i n g and Cognitive Processes. N e w Y o r k : P l e n u m Press. Pp. 205. 8. B y r d , M . 1984. A g e Differences i n the Retrieval o f Information M e m o r y . Experimental A g i n g Research. 10 (1), pp. 31.  from  Semantic  9. S p e c i a l thanks to H a r i Garudadri for the implementation of this program. 10. Special thanks to Dr. L. Beattie, M e d i c a l Director, R o b y n Lawrence, C l i n i c Coordinator, and a l l the staff at the A l z h e i m e r ' s C l i n i c for providing subjects. 11. Special thanks to K a t h y Fuller for collecting data f r o m the control subjects. 12. Special thanks V a l l e y v i e w Hospital.  to Dr. S. H o l l i d a y , D r . L i s h m a n ,  and the c l i n i c a l research staff of  72  Bibliography A p p e l l , J . , Kertesz, A., and Fisman, M . 1982. A Study of Language Functioning i n A l z h e i m e r Patients. B r a i n and Language. 17, 73-91. Battig, W.F., and Motague, W.E. 1969. Category N o r m s for V e r b a l Items i n 56 Categories: A Replication and Extension of the Connecticut Category Norms. Journal of Experimental Psychology M o n o g r a p h . 80 (3), 1-46. Bayles,K.A. 1986. Management of Neurogenic Communication Disorders Associated with Dementia. In R. Chapey (Ed.), Language Intervention Strategies i n A d u l t Aphasia. Second E d . Baltimore: W i l l i a m s & W i l k i n s . Pp. 462-473. Bayles, K.A. and Kaszniak, A . W . 1987. Communication and C o g n i t i o n i n N o r m a l A g i n g and Dementia. Boston: Little, B r o w n and Company. Bayles, K.A. and Tomoeda, C K . and Language. 19, 98-114.  1983. Confrontation N a m i n g Impairment i n Dementia. B r a i n  Birren, J.E., Woods, A . M . and W i l l i a m s , M . V . 1980. Behavioral S l o w i n g w i t h A g e : Causes, Organization, and Consequences. In L.W. P o o n (Ed.), A g i n g i n the 1980's: P s y c h o l o g i c a l Issues. Washington D . C : A m e r i c a n Psychological Association. Pp. 293-308. B o w l e s , N.L. and Poon, L.W. 1985. A g i n g Journal of Gerontology. 40 (1), 71-77.  and Retrieval of Words  B o w l e s , N.L. and Poon, L.W. 1981. The Effects of A g e Experimental A g i n g Research. 7 (4), 417-425.  i n Semantic  on Speed  Memory.  o f L e x i c a l Access.  Burke, D.M. and Y e e , P.L. 1984. Semantic Prirning D u r i n g Sentence Processing b y Y o u n g and Older Adults. Developmental Psychology. 20 (5), 903-910. B y r d , M . 1984. A g e Differences i n the Retrieval of Information f r o m Semantic Experimental A g i n g Research. 10 (1), 29-33.  Memory.  Craik, F.I.M. 1983. O n the transfer of information f r o m temporary to permanent memory. Philosophical Transactions of the R o y a l Society of L o n d o n . B 302, 341-359. Craik, F.I.M. and B y r d , M . 1982. A g i n g and Cognitive Deficits; the R o l e of Attentional Resources. In F.I.M. C r a i k and S.E. Trehub (Eds.), A g i n g and Cognitive Processes. N e w Y o r k : P l e n u m Press. Pp. 191-211. Craik, F.I.M., and Lockhart, R.S. 1972. L e v e l s of Processing: A Framework for Research. Journal of V e r b a l Learning and V e r b a l Behaviuor. 11, 671-684.  Memory  Foster, N.L., Chase, T.N., Fedio, P., Patronas, N.J., Brooks, R.A., and D i C h i r o , G . 1983. A l z h e i m e r ' s Disease: F o c a l Cortical Changes S h o w n by Positron E m i s s i o n Tomography. Neurology. 33, 961-965. G e r w i t h , L., Shindler, A . G . , and Hier, D.B. 1984. A l t e r e d Patterns of W o r d Associations i n Dementia and Aphasia. B r a i n and Language. 21, 307-317. Hannigan, M.L., Shelton, T.S., Franks, J.J., and Bransford, J.D. 1980. The effects of episodic and semantic memory on the identification of sentences masked b y white noise. M e m o r y & Cognition. 8 (3), 278-284.  73 Hier, D.B., Hagenlocker, K., and Shindler, A . G . 1985. Language Disintegration i n Dementia: Effects of E t i o l o g y and Severity. B r a i n and Language. 25, 117-133. Kahneman, D. 1973. Attention and Effort. E n g e l w o o d C l i f f s , N.J.: Prentice-Hall. Lorsbach, T.C. and Simpson, G.B. 1984. A g e Differences i n the Rate o f Processing i n Shortterm Memory.Journal of Gerontology. 39 (3), 315-321. L a w s o n , J.S., and Barker, M . G . 1968. The assessment of nominal dysphasia i n dementia: the use of reaction time measures. British Journal of M e d i c a l Psychology. 41, 411-414. M a r t i n , A., Brouwers, P., C o x , C , and Fedio, P. 1985. O n the Nature of the V e r b a l M e m o r y D e f i c i t i n A l z h e i m e r ' s Disease. B r a i n and Language. 2 5 , 323-341. M a r t i n , A . and Fedio, P. 1983. W o r d Production and Comprehension i n A l z h e i m e r ' s Disease: The B r e a k d o w n of Semantic Knowledge. B r a i n and Language. 19, 124-141. M a s u r , D.M. 1986. Semantic M e m o r y i n N o r m a l A g i n g and Dementia. Dissertation Abstracts International. 47 (03), 1305-B. M i l l e r , E. 1981. The Nature of the Cognitive Deficit i n Senile Dementia. In N.E. M i l l e r and G.D.Cohen (Eds.), C l i n i c a l Aspects of A l z h e i m e r ' s Disease and Senile Dementia ( A g i n g , Vol.15). N e w Y o r k : R a v e n Press. Pp. 103-120. M i l l e r , E . 1978. Retrieval f r o m long-term memory i n presenile dementia: two tests of an hypothesis. B r i t i s h Journal of S o c i a l and C l i n i c a l Psychology. 17, 143-148. M i t c h e l l , D.B., Hunt, R.R., and Schmitt, F.A. 1986. The Generation Effect and Reality M o n i t o r i n g : Evidence from Dementia and N o r m a l A g i n g . Journal of Gerontology. 41 (1), 79-84. Nebes, R.D., M a d d e n , D J . 1988. Different Patterns of C o g n i t i v e S l o w i n g A l z h e i m e r ' s Disease and N o r m a l A g i n g . Psychology and A g i n g . 3 (1), 102-104.  Produced  by  Nebes, R.D., M a r t i n , D.C., and H o r n , L.C. 1984. Sparimg of Semantic M e m o r y i n A l z h e i m e r ' s Disease. Journal o f A b n o r m a l Psychology. 93 (3), 321-330. Ober, B.A., Dronkers, N.F., K o s s , E., Delis, D.C., and Friedland, R.P. 1986. Retrieval f r o m Semantic M e m o r y i n Alzheimer-type Dementia. Journal of C l i n i c a l and Experimental Neuropsychology. 8 (1), 75-82. Obler, L.K. 1983. Language and B r a i n Dysfunction in Dementia. In S.J. Segalowitz Language Functions and B r a i n Organization. N e w Y o r k : A c a d e m i c Press. Pp. 267-282.  (Ed.),  Obler, L.K. and Albert, M . Language Across Adulthood. Source unknown. Rosen, W . G . 1980. V e r b a l Neuropsychology. 2 (2), 135-146.  Fluency  in  Aging  and  Dementia.  Journal  of  Clinical  Schwartz, M.F., M a r i n , O.S., and Saffran, E.M. 1979. Dissociations of Language Function i n Dementia: A Case Study. B r a i n and Language. 7, 277-306. Semple, S.A., Smith, C M . , and Swash, M . 1982. The A l z h e i m e r Disease Syndrome. In S. C o r k i n , K.L. Davis, J.H. G r o w d o n , E. U s d i n and R.J. Wurtman (Eds.), A l z h e i m e r ' s Disease: A Report of Progress i n Research. N e w Y o r k : R a v e n Press. Pp. 93-107. W a l s h , D.A., and B a l d w i n , M . 1977. Developmemtal Psychology, 13, 509-514.  Age  Differences  i n Integrated  Semantic  Memory.  74 Appendix 1 Subject Information  1A Pilot Subjects Subject N u m b e r  Diagnosis  Severity  Complicating Conditions  1  AD  Moderate  Nutritional status  2  AD  Mild  Possible lacunar infarct  3  Parkinson's Disease with dementia  Moderate  Diagnosis  Severity  Complicating Conditions  AD  Mild  A l c o h o l abuse Possible M I D  AD  Mild  AD  Mild  AD  Moderate  Depression  AD  Mild  Epilepsy  IB AD Patients Subject N u m b e r  4 6 7 8 10  75 IC Dementia Patients Diagnosis  Severity  Complicating Conditions  Dementia, etiology unknown  Mild  Depression Non-fluent aphasia  9  M e m o r y deficit  Unknown  Depression  11  Unknown  Unknown  Depression  12  Dementia, etiology unknown  Moderate  A l c o h o l abuse  13  Dementia, etiology unknown  Unknown  Depression  Subject N u m b e r  76 Appendix 2 Stimulus Pairs  Category  Correct Decision Exemplar  Friut  Blue Lavender Apple Mango  Sport  Football Volleyball  Colour  Musical Instrument  Piano Harmonica  Bird  Sparrow Chicken  Toy  Ball Yoyo  Vegetable  Carrot Cabbage  Flower  Tulip Lilac  Tree  Oak Poplar  Fish  Trout Minnow  Metal  Iron Platinum  Four-footed Animal  Dog Beaver  Incorrect Decision Exemplar  Generation Letter  Soccer Wrestling  R T  Copper Walnut  O L  Com Pink  B R  Hockey Celery  V  Guitar Mouse  R P  Banana Mercury  D J  Milk Zebra  P S  Potato Drum  R D  Tomato Lily  M W  Donkey Eagle  S H  Spruce Carnation  N  Birch Tuna  C R  o  S  77 Appendix 3 Trial Lists Mixed Trials Condition Practice L i s t C Fruit , Banana Sport , Tennis Bird , Lion Four-footed A n i m a l , Lettuce Vegetable , P i g Colour , Y M u s i c a l Instrument , T Fruit , P Bird , C Metal , G Practice L i s t D Tree , Pine F l o w e r , Bass F i s h , Brass Sport , G o l f Metal , Aluminum Four-footed A n i m a l , B Vegetable , C Flower , P Tree , F Toy , T Listll Metal, N F i s h , Trout Vegetable, Zebra Toy, B a l l Vegetable, P Tree, Tomato Colour, Soccer Bird, P Flower, T u l i p L i s t 12 Fruit, M a n g o Colour, T Fish, D o n k e y Four-footed A n i m a l , T u n a Vegetable, M i l k Bird, Chicken Fruit, O Metal, S M e t a l , Iron  L i s t 13 Fruit, A p p l e Colour, B l u e Toy, Y o y o Fish, H Bird, R Tree, L i l y F i s h , Eagle Tree, W Four-footed A n i m a l , Beaver L i s t 14 Four-footed A n i m a l , C Colour, Wrestling M e t a l , Carnation Vegetable, Cabbage Flower, R Sport, B Fruit, L Colour, Lavender Metal, Platinum L i s t 15 Fish, M i n n o w Tree, O a k M u s i c a l Instrument, Piano Four-footed A n i m a l , R Sport, P i n k B i r d , Guitar T o y , Banana Colour, R Flower, D L i s t 16 Fish, S Four-footed A n i m a l , B i r c h Fruit, Copper Sport, V o l l e y b a l l M u s i c a l Instrument, O Toy, D Tree, M Tree, Poplar B i r d , Sparrow  L i s t 17 Flower, Lilac Four-footed A n i m a l , D o g Fruit, Walnut Sport, C o r n M u s i c a l Instrument, Celery Toy, J M u s i c a l Instrument, V Sport, F o o t b a l l Vegetable, S L i s t 18 M u s i c a l Instrument, Vegetable, Carrot M u s i c a l Instrument, Bird, Mouse Toy, Mercury Flower, Potato Flower, D r u m M e t a l , Spruce Sport, R  Harmonica Hockey  Blocked Trials Condition Practice L i s t A Fruit, Banana Sport, Tennis Bird, Lion Four-Footed A i i i m a l , Lettuce Vegetable, P i g Tree, Pine F l o w e r , Bass F i s h , Brass Sport, G o l f Metal, Aluminum Practice L i s t B Colour, Y M u s i c a l Instrument, T Fruit, P Bird, C Metal, G Four-footed A n i m a l , B Vegetable, C Flower, P Tree, F Toy, T  List 1 Toy, Y o y o Flower, Lilac F i s h , Trout Sport, F o o t b a l l Four-footed A n i m a l , D o g Fruit, A p p l e Vegetable, M i l k Tree, Tomato Fruit, Copper M e t a l , Spruce M u s i c a l Instrument, H o c k e y Colour, Soccer List 2 Colour, R Sport, R M u s i c a l Instrument, Toy, D Tree, W Metal, S  V  List 3 Colour, Wrestling Sport, C o r n B i r d , Guitar T o y , Banana Flower, D r u m Fish, Donkey Colour, B l u e Fruit, M a n g o B i r d , Sparrow Vegetable, Cabbage Tree, O a k M e t a l , Platinum List 4 Colour, T Fruit, O Fruit, L Bird, R Fish, H Four-footed A n i m a l , C  List 5 Four-footed A n i m a l , Beaver M e t a l , Iron Colour, Lavender M u s i c a l Instrument, Piano Flower, T u l i p Fish, M i n n o w Four-footed A n i m a l , B i r c h M e t a l , Carnation Tree, L i l y Toy, M e r c u r y Bird, Mouse Fruit, Walnut List 6 Sport, B Toy, J Flower, R Metal, N Vegetable, P Fish, S List 7 Sport, V o l l e y b a l l M u s i c a l Instrument, Harmonica Bird, Chicken Toy, B a l l Sport, P i n k M u s i c a l Instrument, Celery Vegetable, Zebra Vegetable, Carrot Tree, Poplar F l o w e r , Potato F i s h , Eagle Four-footed A n i m a l , T u n a List 8 M u s i c a l Instrument, O Bird, P Four-footed A n i m a l , R Tree, M Flower, D Vegetable, S  82  Appendix 4 Instructions  The f o l l o w i n g instructions were read to each subject by the experimenter:  The purpose o f this study is to investigate memory f o r words i n A l z h e i m e r ' s Disease. W e do not k n o w i f y o u have A l z h e i m e r ' s Disease yet. I want to see i f the questions that I w i l l be asking can help us tell w h i c h patients have A l z h e i m e r ' s Disease and w h i c h do not. T h i s w i l l take about an hour to complete. Some o f the questions w i l l be hard, but that is necessary so that I can f i n d what k i n d o f problems y o u are having. I want y o u to try and answer as q u i c k l y as possible, but try to give the correct answer as well. T h i s study is to investigate memory f o r words. I want to see h o w q u i c k l y y o u can retrieve words f o r m memory. Y o u w i l l see a w o r d o n the screen f o r a f e w seconds. Sometimes y o u w i l l see another w o r d w i t h it (the example  Colour: Green was shown to the subject). W h e n y o u see t w o  words together, tell me yes or no, f o r example, is there a colour called green? Sometimes y o u w i l l see a w o r d and a letter (the example  Bird: B was shown to the subject).  W h e n y o u see this, tell me, f o r example, a bird that starts with B. W e w i l l practice some words on paper, then w e w i l l do t w o practice lists on the computer and eight more lists on the computer. Y o u w i l l have a two-minute break between each list. D o you have any questions?  The f o l l o w i n g instructions were read only to subjects i n the blocked trials condition:  E a c h group o f questions w i l l be o f the same type, and y o u can use this to help y o u answer. F o r example, one group w i l l be t w o words, l i k e this (the example  Colour: Green was shown). The  next group w i l l be a word f o l l o w e d b y a letter, l i k e this (the example Remember, i n one group they are a l l the same type o f answer.  Bird: B was shown).  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0097650/manifest

Comment

Related Items