Intact and Impaired Memory Functions in Autism

Intact and Impaired Memory Functions in
Autism
Loisa Bennetto and Bruee F. Pennington

University of Denver
Sally J. Rogers
University of Colorado Health Sciences Center
BENKETTO. LOISA: PENNINGTON, BRUCE F . and ROGERS, SALLY J Intact and Impaired Memory
Functions in Autism. CHILD DEVELOPMENT, 1996, 67, 1816-1835. This study examined memory'
functions m individuals witb autism. Based on previous evidence of executive function (EF)
deficits, we hypothesized that subjects with autism would demonstrate a pattern of intact and
impaired memory functions similar to that found in other groups witb EF deficits, sucb as patients
witb frontal lobe patbology. We compared tbe performance of bigb-functioning cbildren and
adolescents witb autism (n = 19) and clinical comparison subjects (n = 19) matched on sex, CA,
and VIQ on measures of memory and EF. The group witb autism perfonned significantly worse
tban comparison subjects on measures of tempoial order memory, source memory, supraspan
free recall, working memory, and EF, but not on sbort- and long-term recognibon, cued recall,
or new learning ability, consistent with tbe predictions of tbe EF tbeory. Tbe cognitive measures
were significantly moTe intercorrelated m tbe autism group tban tlie comparison group, consistent
witb a limit m central cognition.
Several independent studies have found could explain many of the behavioral symp-
evidence of poor performance on measures toms found in autism, including rigid and
of executive function in children, adoles- inflexible behavior, perseveration, inappro-
cents, and adults with autism (e.g., Hughes, priate responding in social situations, inabil-
Russell. & Robbins, 1994; McEvoy, Rogers, it>' to learn by experience and to adapt to
& Pennington. 1993; Ozonoff, Pennington, changing envuonments, lack of initiative,
& Rogers, 1991; Prior & Hoffmann, 1990; and concreteness in thought processes (Da-
Rumsey & Hamburger, 1988). The tasks masio & Maurer, 1978; Rumsey, 1983; Rut-
used included standard clinical tasks such as ter, 1983).
the Wisconsin Card Sorting Test, Tower of
Hanoi. Trail Making Test, Rey-Osterrieth Executive function impairment is usu-
Compiex Figure, and Milner Maze, as well ally associated with patients with frontal
as more experimentally based tasks such as lobe lesions. It should be noted, however,
Hughes and Russell's (1993) object retrieval that diffuse brain damage (Heaton, Chelune,
task. These findings have led some research- Talley, Kay, & Curtiss, 1993; Levin et al.,
ers to propose a theory of autism based on 1994) or damage to subcortical structures
deficits in executive function (Hughes & connected to the frontal lobes (Cummmgs,
Russell, 1993; Ozonoff et a!., 1991). "Execu- 1993) can produce EF impairment. Damage
tive function" (EF) is an umbrella term used to attentional control systems, such as the
by psychologists to refer to goal-directed, fu- cerebellum, may also compromise perfor-
ture-oriented behaviors that involve plan- mance on EF measures (Courchesne et al.,
ning flexible strategy' employment, inliibi- 1994). Thus, while patients with frontal le-
tion and organized search (Welsh & sions provide a good illustration oi EF lm-
Penkington, 1988). Deficits in these skills pairment, not all EF impairment is associ-
This research was supported by an NIMH grant MH10470 (NRSA) to tbe first author The
second author was supported by tbe following NIMH grants- MH00419 (RSDA), MH38820
(MERIT award) and MH45916, as well as by an NICHD Center Grant (HD 27802). We gratefully
acknowledge the children and families who participated in this study. We also like to thank
Sally Ozonoff for sbaimg ber executive function data with us, Ralph J Roberts for bis helpiul
comments on this study, and Ellen Braaten for her help in data collection. Address correspon-
dence to Loisa Bennetto, University' of Denver, Department of Psychology, 21oo S. Race Street,
Denver, CO 80208.
[ChMDevelopinent, 1996,67,1816-1833 © 1996 by the Society for Research in Child Development, Inc
All rights resers'ed 0009-3920,'96/6704-0027§01.001
Bennetto, Pennington, and Rogers 1817
ated with frontal lobe patholog5'. So the age and consolidation of declarative in-
finding of EF impairment in autism does not formation, patients with frontal lesions
necessarily imply localized frontal lobe pa- exhibit deficits in remembering certain
thology. kinds of contextual or spatiotemporal infor-
In addition to showing deficits on E F mation (Schacter, 1987). In particular, pa-
tasks, patients with frontal lesions also dem- tients with frontal lesions have shown con-
onstrate a well-characterized pattern of sistent impairment on measures of memory
memory deficits (Shimamura, Janowsky, & for temporal order and source memory,
Squire, 1991). Therefore, in the current which we will review below. Shimamura et
study, we evaluated the performance of chil- al. (1991) argue that these kinds of tasks uti-
dren and adolescents with autism in the do- lize a memory system that allows an individ-
main of memory. The cognitive neuropsy- ual to access, organize, and manipulate
chology of memory is better developed, both memories, thus allowing consideration of
in terms of theories and tasks, than is the their spatial, temporal, or semantic context.
cognitive neuropsychology of executive Moscovitch (1994) has coined the term
functions. Memory theorists distinguish sev- "working with memory" to refer to these ex-
eral different types of memory, have devel- ecutive processes utilized by the memory
oped experimental tasks for each lype (and system. In contrast to their poor performance
its components), and have demonstrated dif- on temporal order and source memory, pa-
ferent profiles of memory impairment across tients with frontal lesions are typically not
patient groups with different etiologies and impaired on learning new information or
lesion locations (Squire, 1987). Thus, the recognition memory', which rely primarily
area of memory provides both well-articu- on effective storage and consolidation of de-
lated theories and fine-grained measures clarative information.
for further examining neurocogniilve defi- On tasks of temporal order memory, sev-
cits in autism. eral studies have found that patients with
In the current study, we present a com- frontal lesions have marked impairments
prehensive assessment of memor;^ perfor- (e.g., Eslinger & Crattan, 1991; Milner,
mance in autism, and examine whether indi- Corsi, & Leonard, 1991; Milner, Petrides, &
viduals with autism demonstrate a pattern of Smith, 1983; Shimamura, Janowsky, &
memory performance similar to patients Squire, 1990). Memory for temporal order is
with frontal lesions. Similar patterns of often assessed with the Corsi task (cited in
memory performance would demonstrate an Milner, 1971), in which subjects are shown
additional link between the cognitive defi- a series of verbal or nonverbal stimuli, and
cits of individuals with autism and patients periodically asked which of a pair of stimuli
with frontal lesions, and would be consistent was presented more recently. In addition,
with an EF theory of autism. However, a recognition trials, in which one of the stim-
fundamental limitation to an E F theory of uli is new, are interspersed with the tempo-
autism is that "executive functions" com- ral order trials. Thus, the task provides a
prise a complex set of abilities, which limits within-subjects comparison of temporal or-
the usefulness of this construct as a core der memory and recognition memory. Com-
deficit of the disorder (Sigman, 1994). One pared to patients vrith temporal lesions and
component process that has been proposed control subjects, paitients with frontal lesions
to be involved in I?F tasks is working mem- exhibit poor temporal order memory, de-
ory (e.g., Cohen & Serv-an-Schreiber, 1992). spite intact performance on the recognition
In the discussion, we will address the impli- trials.
cations pf our results for a working memory Patients with frontal lesions may show
model of autism. deficits on temporaJ order memory tasks be-
cause memory for temporal order requires
MEMOBY FUNCTIONS IN PATIENTS WITH the subject to retain more than the content
FRONTAL LESIONS of the memory. Memory for temporal order
also requires organization of distinct memo-
Frontal lobe pathology is associated ries and retention of their temporal relation-
with a specific profile of memory dysfunc- ship to each other, Impaired processing of
tion, which is distinct from the pattern of temporal variables such as time and order
deficits associated with damage to the me- has been cited as a characteristic deficit in
dial temporal lobes or diencephalic rnidline. patients with frontsil lesions (e.g., Crafman,
While traditional amnesic patients generally 1989), as well as frontally lesioned animals'
perform poorly on tasks that require the stor- (e.g., Fuster, 1985).
1818 Child Development
On tests of source memory, patients tional memory" (Goldman-Rakic, 1987), be-
with frontal lesions also show impairment cause subjects must hold a representation of
(e.g., Janowsky, Shimamura, & Squire, 1989,- relevant infoi-mation on line over a delay,
Schacter, 1987). The phenomenon of and use this information to generate an up-
"source ainnesia" was first described by coming action. Goldman-Rakic (1987) has
Schacter, Harbluk, and McLachlan (1984) to established the role of the dorsolateral pre-
refer to cases in which art individual was frontal cortex in spatial working memory,
able to remember sonietliing without being through a series of lesion aad single-unit re-
able to remember where or wh^n the infor- cording studies in nonhuman priniates.
matioo was otiginaily learned. This effect WorkiHg mernory has also been shown to
has been shown witli experimental tasks in rely on the prefrontal cortex iii hurnan popu-
vi'hich subjects are taught fictitious state- lations. Gold, Bermaii, Randolph, Goldberg,
mentSj, and then later tested for retention of and Weinberger (1991) have developed a
the information and its source. An inability spatial working memory task for adults,
to utilize the source of learned infprriiation wHieh is cGnceptually siiHilar to the delayed
has also been shown on a variety of multilist alternation tasks used with infants and pri-
Ieamiiig tasks. Patients with firgntal impair- mates, Usiflg regional cerebral blood flow
ments liai?e-sJjGwn an ex^.ggerkted tendency ( R C B F ) aad positron emission tomography
to intriide early list items oiito successive (PET): ftiey have demoijstrated the role of
lists (Cermak, Butters, "& 'Sloreines, 1974; the doisalateral and orbital prefrontal corti-
^'loscovitch, 1992; Stuss et al:, l#&a),. These ces in subserving working meraory in hu-
intrusioijis are essentially source errOrSv^ mans. FiHally, Case (1992) has argued for the
Source memory is similar to memor)' for roie-of the frontal lobes in the execuLtion of
teniporal order because it relies on memory verbal working memory (Counting Span) as
for opntext, rather than mempry for factual well as spatial werkin^ niemory in ehildren.
information. In fact, source amtiesia has basesdôia results: from fiEG studies/Though
)>een dBstioristrated' to be qnielated to mea- tibere apgears to be eviden*3e for the role of
ptires of fact recall (Graik,: Moms, Morris, & the frotital cortices in verbal working mem-
Loewen, 1990; Scfeactfer, Harbluk, & ory, howeyer, 'there have be.en no studies of
McLaehlan, 1984). A recent deyelopiiiental paJti<pnts with frontal lesioriLS on this type of
ptudif suggested a relation between source task.
entors and axecutivefuiietloB deficits in cbil-
|lren (Ryliash & Golila, tOM). TMis study While patients with frontal lesions show
also found fhat source errors were unrelated impairment on certain types of memory
to fact recall or digit span. tasks, they are typically not impaired on the
tasks that involve learning new material,
Patient? with frontal lesions have also cued recall, or recognition. Such types of
been shown to perform poorly on some ty-pes mensoiy bave been shown to be both theo-
of free recalltests, despite good performance retiealiy aiid empirically distinct froni "fron-
on qompai"£Lbl:e tests of ciied irecall and recog- tal" mei-aorj-- tasks. Schacter (1987) distin-
nitioii. Specifieallf 5 some studies have dem- goisfees between "contextual infoitnatidn,"
onstrated that patiertts with frontal lesions Wliich refers to wken and where a specific
are inapaired in the free recaJi of maltiple event occurred, and "item inforniation,"
trials of words (jfetter. Poser, Freernan, & whieh refers to what occurred. Considerable
Markowitsch, 19Sfi; Shiiioaijiura et al., 1991). emgirica] evidence stigg:ests that contextna!
Cither studies, however, haT.<e failed to find meaiory ts dissociable from menjorj' for item
clâr impainrtents (see Schacter, 1987, for a infoirinatian in bodi huriQan and animal sub-
review). Cognitive deSclts in fluency, orga- (sefe Schaeter, 1987, for a review).
pizat:ion, and; self-rnonitoririg that result from
frontal damagfe iliay affect self-initiated stra- EVIDENCE FOE MEMORY DEFICITS IN AUTISM
tegic retriei«a| sMlls, as well as the ability to The results on mempry dysfunction in
suppress pistentially interierio.g itenis vAen autism are equivocal. Several studies of
doing free recall tasfcs. meraoiy in autism have proposed that autism
Finally, recent work with both humans is best characterized as an amnesic disorder
and prirnates: suggests a strong tie betvveen (BoHcher & Warririgton, 1976; DeLong,
the frontal iotes aind woriking memorj'. 1992). These theories arose and have been
Working memory refers to the siniultaneOus suppGiied to soine degree by neuropatholog-
profceasirig and storage of information during ical stridies of autism that have nncovered
the perfprinance of a lange of cogriitive histplqgical abxiornaalities in the hippo-
tasks. It has been described as "representa- and related structures (Bauman &
Kemper, 1985; Raymond, Bauman, & Kem- to both comparison groups. Ameli, Courch-
per, 1989). Further support has come from esne, Lincoln, Kaufman, and Grillon (1988)
evidence of abnormal social behavior and found that the performance of individuals
memory loss in monkeys following neona- with autism on visual memory tasks was
tal ablation of the limbic system (Bacheva- compromised by inflexible cognitive strat-
lier, 1991). Several neuropsychological stud- egies.
ies of individuals with autism have also re-
ported evidence consistent with a traditional In summary, while some studies have
amnesic theory of autism. Boucher and War- found support for impaired memory, others
rington (1976) found impairments in free re- have found evidence of relatively intact
call and recognition memory. Boucher memory function. Furthermore, some ofthe
(1981) found impairments in recall when re- observations of memory dysfunctions in au-
trieval cues had to be encoded at mput. tism are consistent widi a deficit in execu-
tive processes.
In contrast, other studies of subjects In this study, we will attempt to better
with autism have reported intact rote mem- characterize the nature ofthe memory deficit
ory skills relative to both retarded and nor- in autism. If individuals with autism mainly
mal control subjects (Bartak, Rutter, & Gox, have EF impairments, then their profile of
1975; Hermelin & O'Connor, 1975; Prior & memory performance ought to be similar to
Chen, 1976). Rumsey and Hamburger (1988) that exhibited by patients with frontal le-
examined immediate recall and delayed sions. Specifically, subjects with autism
memory function in a group of high- ought to be impaired on measures of tempo-
functioning men with autism using the ral order memory, source memory, and
Wechsler Memory Scale Paragraph and De- working memory, but not impaired on mea-
sign tests. They found no significant differ- sures of short- and long-term recognition,
ences compai-ed to control subjects, sug- cued recall, or new learning ability. This is
gesting that autism does not a;ffect the the main hypothesis of the current study.
consolidation, storage, and retrieval pro-
cesses that are mediated by the temporal Method
lobes. Boucher (1978) examined echoic
memory capacity in children witlli autism SUBJECTS
and found no differences compared to nor-
mal age-matched control subjects. More re- Two groups of subjects participated in
cently, Minshew and Goldstein (1993) exam- the present study: a group of individuals
ined the performance of high-functioning with autism (n = 19), and a clinical compari-
individuals with autism on the California son group of individuals without autism
Verbal Learning Test. They found no consis- (n = 19). Subjects were drawn from a sample
tent evidence of poor recall or recognition of individuals who participated in a research
in autism. Their data did, however, suggest project 3 years ago at our laboratory (Ozonoff
that subjects with autism may have less effi- et al., 1991). The present autistic group com-
cient mechanisms for organizing infor- prised all autistic subjects from the original
mation. study who were still living in the area (19 of
23). The present comparison group com-
Several other studies have also found prised 14 of the original 20 subjects, plus
evidence of poor organization affecting the five new clinical comparison subjects re-
performance of subjects with autism on cruited from our laboratory. All comparison
memory tasks. Ozonoff et al. (1991) found subjects had nonautistic learning disorders,
deficits in a group of high-functioning chil- including borderline intellectual function-
dren with autism on a supraspan free recall ing (n, = 2), dyslexia (n = 13), attention
task, the Bushke Selective Reminding Task. deficit hyperactivity disorder (n = 2), ge-
This finding may be explained by difficulties netic disorder (n = 1), and unspecified
children with autism show in organizing in- learning disorder (n = 1). The autistic group
formation, as well as in initiating verbal re- was matched with the comparison group on
sponses. Boucher (1981) examined memory sex, chronological age (CA), and Verbal IQ
for recent events in children with autism (VIQ); the groups were not different on
compai-ed to typical children matched on handedness, socioeconomic status (SES;
age, and children with mental retardation Hollingshead, 1975), Performance IQ (PIQ)'
matched on age and abilitj- leveL She found or FuIl-Scale IQ (FSIQ), although they were'
that the ability of children with autism to not explicitly matched on these variables.
recall the order or context in which an event Each subject had obtained an IQ esti-
occurred was significantly inferior compared mate (prorated from five subtests of the
TABLE 1
DESCRIPTIVE CHABACTERISTICS OF THE SAMPLE
COMPARISON
CROUP GHOCP
M (SD) M (SD) t
13.95 (3.3) 15.23 (2.6) .76
82.32 (15.2) 88.26 (11.4) -1.36
98.11 (1.5,9) 97.58 (14.3) .11
FSIO 88-89 (11.1) 91.74 (12.1) -.75
Reading'skill 90.63 (16.7) 84.26 (21.4) 1.02
C\RS 34.89 (4.1) 18.31 (3.0) 12.33***
SES 45.53 (10.0) 45.31 (15.2) .05
Sex(M:F) 17'.2 17:2
Handedness (R:L) 17:2 16^3^
]\JOXE. TV" = 19 for botli the aTidstic group and the comparison group.
*** p < .001.
WISC-R: Informa-tion, Vocabulary, Similari- The autistic group comprised 19 indi-

ties Block Design, and Object Assembly) viduals between the ages of 11 and 24 (M -
within the past 3 years. To ensure that the 15.95, SD = 3.25). All subjects in this group
IQ estimates remained valid, two subtests met the DSM-III-R criteria for Autistic Dis-
from the WlSC-III (Vocabulary and Block order (n = 11) or Pervasive Developmental
Design) were administered to all subjects. Disorder Not Otherwise Specified
There were no significant group differences (PDDJSiOS; n = 8). In addition, all 19 sub-
on age-scaled scores for Vocabulary, t(36) = jects in the autistic group fell in the autistic
- 1 47 N S or Block Design, t(36) = - .20, range on the Childhood Autism Rating Scale
N S Furthermore, paired t tests performed (CARS; Schopler, Reichler, & Renner,
within each group indicated that there was 1986). Since this study was designed to ex-
no significant difference between the old amine deficits in high-functioning individu-
(WISC-R) and new (WISC-III) scores for ei- ais with autism, the subjects with PDDNOS
ther Vocabulary, i(lS) = - . 1 6 , N.S., or were retained, with the rationale that any
Block Design, t(l8) = -1.82, N.S. Since the bias they might introduce would be a con-
prior IQ scores were considered a more sta- ser\'ative one against finding the hypothe-
ble estimate of intellectual functioning be- sized differences.
cause they relied on more subtests, they xhe comparison group consisted of 19
were used in subsequent analyses. individuals without autism between the
I
autism (score
son group vyas 91.i4 (aJJ j
several tasks involved reading, the Letter-
Word Identification subtest of the Wood-
cock-Johnson Revised Tests of Achievement Executive Function Measures
(Woodcock & Johnson, 1989) was admmis- Wisconsin Card Sorting Test (WCST).—
tered to all subjects. All subjects in both This task, which was originally developed
groups demonstrated sufficient reading abil- by Grant and Berg (1948 , measures concep-
itv to complete tlie study tasks. On this read- tual problem-solving abilities, mcluc^ng the
ing measure, the mean standard score, of the ability to modify incorrect strategies flexibly
autistic group was 90.63 (SD = 16.72), and and the ability to inhibit prepotent but incor-
of Ae c l p ^ - i s o n group was 84.26 (SD = rect responses The WCST was a t a ^
21.37). In addition, we required that all sub- tered and scored according to the standard
fects were cooperative and attentive enough Heaton (Heaton et al 1993) instructions.
to understandthe testing instructions and to Three dependent variables were used in this
complete all the tasks. No subjects were ex- study: number of categories completed
eluded for this reason. Table 1 contains de- number of perseverative responses PR), and
scriptive characteristics of the sample. number of failures to maintain set (FTMb).
Tower of Hanoi.—Researchers have this task were selected from tlie Peabody
used this disk-transfer task to study the plan- Picture Vocabulary Test (Dunn, 1965) and
ning and problem-solving capacities of nor- the Expressive One-Word Picture Vocabu-
mal and impaired children and adults (e.g., lary Test (Gardner, 1979). The subject
Borys, Spitz, & Dorans, 1982; Levin et al., viewed each card for 3 sec before the exam-
1994; Welsh, 1991). It was administered and iner turned to the next card. Periodicdly, a
scored according to the system described by yellow response card appeared, with a ques-
Borys et al. (1982). The dependent variable tion mark between the two drawings. When-
in this task was a "planning efficiency" ever this happened, the subject was in-
score, which reflected the number of moves structed to point to the picture seen more
a subject required to solve the test problems. recently. On temporal order cards, both
drawings had appeared before. On recogni-
Memory Measures tion cards, one of llie drawings was new, and
Temporal order versus recognition the subject merely had to point to the picture
mem,oru.—This task provided a within- seen before. Since no stimulus item ap-
subject comparison of memory for temporal peared more than one time on an inspection
order and recognition memory for both ver- card, judgment for relative recency was al-
bal and nonverbal stimuli. It was adapted ways for a trial-unique pair.
from the original Corsi Memory Task (Mil-
ner et al., 1991). In this task, subjects were One hundred of the cards were white
shown series of" verbal and nonverbal stim- inspection cards; the remaining 90 were yel-
uli, and then were asked to judge which of low response cards. Forty-six response cards
two stimuli had been presented more re- tested temporal order discrimination; 29
cently (for temporal order trials) or which tested recognition memory. The remaining
stimulus they had seen before (for recogni- cards were "dummy" trials, which included
tion trials). several easy recency and recognition trials
at the beginning of the deck that were de-
We did not use the original version of signed to help the subject fam.iliarize him-
this task for several reasons. First, liie origi- self or herself with the task. The first several
nal version was designed for use wilii adults. cards in the deck were inspection cards; re-
Because the present study examined perfor- sponse cards and iaspection cards alternated
mance in children, stimuli needed to be thereafter. Recognition response cards oc-
modified appropriately for age and ability curred following 16, 32, and 64 intervening
level (e.g., verbal IQ, reading skill). Second, items. Three levels of temporal order mem-
ceiling and floor effects have compromised ory were tested (i.e., 4 vs. 32, 4 vs. 64, 4 vs.
previous studies of the Corsi Memory Test. 128).
For example, in Milner et al. (1991), normal
comparison subjects performed at a mean The procedure for the Concrete Words
recognition accuracy of 94% for concrete task was similar to that of the Representa-
words, and 96% for representational draw- tional Drawings task. Two concrete nouns
ings, irhe adaptation of this task (which is appeared on each card. Nouns chosen for
described below) was piloted with children this task were high on a rating of concrete-
and adults to ensure that performance on ness, low on a rating of ambiguity, and had
temporal order and recognition conditions a mean age of acquisition under 7 years (Gil-
yielded equivalent true score variances hooly & Logie, 1980; Toglia & Battig, 1978).
(Chapman & Chapman, 1978) and was not Finally, words that appeared elsewhere in
compromised by ceiling or floor effects. the test battery (e.g., on a different memory
Without these precautions, the argument task) were excluded to prevent a possible
from differential group performance to a dif- priming effect. The subject was instructed
ferential deficit wtûld be tenuous at best. to read aloud the words on the inspection
Pilot data indicated that normal subjects per- cards. On the response cards, the subject
formed at approximately 75% correct for was instructed to point to the appropriate
both temporaJ order and recognition trials of word.
the Concrete Words task.
Prior to the task administration, each
There were two tasks in the test series: subject performed a verbal practice task,
Representational Drawings and Concrete which was designed to teach the concepts of
Words. The Representational Drawings task temporal order and recognition memory, as
consisted of a series of 190 cards (8V2 x 11 well as the task instructions. Each experi-
inches). Each caird contained two drawings, mental task required approximately 10 min
placed one above the other. Drawings from to complete. There were two dependent
variables for this task: the percent of correct both parents and children participating in
responses for temporal order, and the per- the study. The first author tested 74% ofthe
cent of correct responses for recognition. subjeets in the autistic group, and 79% of
the subjects in the comparison group. The
California Verbal Lea.rning Test—Chil- remainder of the subjects were tested by a
dren's Version (CVLT; Delis, Kramer, Kap- trained masters-level research assistant,
lan, 6- Ober, 1986).—This task measured blind to the diagnoses of the subîects and
learning and retention of verbal informa- the hypotheses of the experiment.
tion. It provided information on verbal
learning processes, strategies, and pattems Measures in the experimental battery
cjif errors. In paiiiciilar, the C\TLT yielded were presented in a fixed order for all sub-
information on free recall, cued recall, and jects. First, Letter-Word identification was
recognition jiiemory, as well as learning administered to screen for reading ability.
rates and retention over botb a short and Following this were Temporal Order versus
long delay.. Furdiermore, analysis of error Recogiiition Memorj-' (Words), the C\TLT
patterns on the C\'LT allows fbr an examina- learning and immediate recall trials. Block
tion of source memory deficits. Desigii, Counting Span, Digit Span, the
Btanford^Binet IwteMi^nce Scale, CW;T delay trials. Temporal Order versus
Fourth Edition—Memory for Digits Test RecQgiiition Memory for Pictures, Sentence
Span, and Vocabu'larj'. Though this in-
(ThomcMke,Hage«, ir Saitler.y 1986).—This creased the risk of order effects, it was nec-
is a standard,iz;ed test of short-teim auditory essary because of the large number of verbal
memory for digits (Digits Forward and Dig- rnemoiY tasks in th:e battery (e.gi, to control
its Reversed). for priniiiig effects). The frill testing batterj-
Working Memory—Sentence Span.— took approximately 2 hours; the attention
This task, was adapted for use with children span and cooperation of all siibjects were
(Siegel 6? Ryan, 1989) from the procedure sufficient that testing was conducted in one
developed by Daneirian and Carpenter session, wilh tinae for several breaks. Sub-
(198G). This task requires the subject to pro- jects were paid $25 for tbeii participation in
cess veAal information oia-line and to store ife stpdy., Although iill subjects were in-
tlie results of this processing for later recall. formeti that they could discontinue testing
TThe experilmenter read simple setitences at any tiroe, none chose to do so.
that were missiag the last word, and in-
structed the suBject to supply the missing
woM. A.t the end of a set of such sentences, Results
the experimenter asked the suljject to recall PBELIMINARY ANALYSES
the siipplied words in the order the sen- \l] <J<^pcndent measures were examined
tences' were presented. Set sizes ranged foi \T()iutions of the required assumptions.
from t\vo to six senterices, with three trials Foi scvorul variables (i.e., intrusion and rec-
at each set size. The degendeat variable was o(Ti,]iht.m eirors on the CVLT"), the distribu-
the total iim-nber of trials correet tion of ,^cores was significantly skewed and
' Working Memor§—Counting Span.— ncmphrimetric analyses were consequently
This task was designed by Case, Kurland, inf'd. OI all i i e variables examined,
and Goldberg (1982) arid has bieen used in was only one main effect of tester,
a stu!dy of children with learning disorders within the number that would be
(Siegel & Byan, •1989)..In this task, the ex- EctJ bv chance. In addition, there was
perimenter in sti-ncted the subject to count not J '.iKnificant difference in the number of
iiloiid: the yellow? (Jots iiitersjiersed with blue erl' with autism and coinparison sub-
dots, ail arranged randomly on an 8% x 11 t^ lasted by each examiner, x^(l' ^ ~ 38)
inch card. After the subject counted the yel- .J \, N.S. Consequently, data were col-
low d:Qts on each of a setof cards, the experi- lapsed flcross the tester condition for all sub-
menter aisked tlae suitgect to recall, in order, l
the number of ye^llow dots that appeared on I)till e a laige nuniber of statistical tests
each caircl. Tlid3re,were,fiye set sizes (two to viOiV performed, significant results may
six caids) with three trials at each, size. The ha^'^' tsipitalized on chance, and tlie overall
dependent varialile was fhe total number of pi obabi Iity of a Type I error likely exceeded
trials correct. 5% Hu w'Hvvr., the primary hypothesis of this
siui\\ predicted a specific pattern of results
PROCEDURE aci[)h'> 1 langc of tasks, which included both
All subjects were tested at our labora- lmpaiifHl. and intact performances. Setting
tory. Informed consent was obtained from the ,n-i^(^ptable alplia too low would reduce
the power of being able to detect a group To test the strength ofthe pattern of pre-
difference on the tasks for which intact per- dicted results, a paired t test was computed
formance was predicted. To lower capital- for these two composite variables. This test
ization on chance and reduce the number indicated that there was a significant differ-
of statistical comparisons, specific analyses ence between the predicted-lower and pre-
were planned a priori. In addition, we report dicted-same variables, f(18) = -3.68, p =
exact p values for all analyses for which p .002. Since this test described the size ofthe
was less than .05, and caution the reader to difference, but tells less about the pattern
interjjret the results conservatively. of individual performance, a sign test was
conducted for each composite variable. For
PROFILE OF INTACT AND IMPAIRED 17 of 19 subjects with autism, the score on
MEMORY FUNCTIONS the predicted-lower variable was lower than
the score on the predicted-same variable,
Based on the literature on frontal lobe yielding an observed significance level of
functioning, we predicted a specific profile p = .0007.
of results on the mernory tests in this study.
Specifically, v/e predicted that subjects witii These omnibus analyses suggest that
autism would perform worse than compari- the overall predicted pattern of results is
son subjects on tasks of temporal order mem- present in this sample. Further analyses
ory, source memory, and working memory. were conducted to describe this pattern in
In contrast, we predicted there would be no more detail.
group differences on tasks of cued recall,
short-term recognition, and long-tenn recog- EXECUTIVE FUNCTCON MEASURES
nition. To evaluate the fit of the predicted
pattern to our data, an omnibus test was con- Independent group t tests were per-
ducted, using all variables for which there formed on the executive function measures.
were strong a priori hv-potheses based on Performance data on these and other depen-
studies of patients with frontal lesions, or dent measures are presented in Table 2. On
studies of frontally lesioned animals. Spe- the WCST, the autistic group made more
cifically, we predicted that subjects with au- perseverative responses, t(36) = 3.68, p =
tism would perform worse than comparison .001, and completed fewer categories, t(36)
subjects on the two temporal order memorj' = -2.79, p = .009, than the comparison
tasks and the two working memory tasks, group. In contrast, there was not a group dif-
and would have a higher number of intru- ference in the mean number of failures to
sions and false positives on the CVLT. We maintain set. On the Tower of Hanoi, the
predicted no group differences on the two autistic group had a worse total efficiency
recognition memory tasks, and the cued re- score than the comparison group, ^(35) =
call and recognition trials of the CVLT. -7.45, p < .001. Thus, as discussed else-
where (Ozonoff & McEvoy, 1994), this group
Two composite scores were computed of high-functioning persons with autism had
for each subject with autism; one based on longitudinally stable deficits on these two
tasks that were predicted to be worse than executive function measures.
comparison subjects (predicted-lower), and
one based on tasks that were predicted to MEMORY MEASURE.S
show no difference (predicted-same). The
composites were computed as follows. For Temporal Order versus Recognition
each subject with autism, each component Memory
variable was converted to a standardized ef- Because previous studies suggest that
fect size, based on the mean and standard different hemispheres may mediate memory
deviation ofthe comparison group [(score - for verbal information and pictures (e.g.,
Me)/SD J . By using standardized effect size Milner et al., 1991), performances on the
scores, we have accounted for the mean per- concrete word (Word) and representational
formance ofthe comparison group. Next, the drawing (Picture) tasks were analyzed sepa-
two composite scores were computed for rately. Both tasks were analyzed via a 2 x 2
each subject by averaging the predicted- mixed-model ANOVA, with task (temporal
lower effect sizes and predicted-same effect order, recognition) as the within-subjects
sizes. The mean effect size was - 3 . 3 3 factor and group as the between-subjects
(SD = 4.35) for the predicted-lower vari- factor.
ables, and - . 3 5 (SD = 1.10) for liie pre-
dicted-same variables, revealing a differ- The ANOVA for the Word tasks yielded
ence of nearly 3 SDs between l i e two a group X task interaction, F(l, 35) = 4.96,
composites. p = .03. Post hoc analyses showed that this
interaction was attributable to the fact that
TABLE 2
PEHrORM.4SCE DATA AND G B O U P DIFFERENCES
AUTISTIC COMPARISON
GROUP GROUP
M (SD) M (SD)
Executive function;
WCST categories 3.26 (2.51) 5.11 {1.41}**
WCST perseverative responses 60.37 (47.93) 17.16 (17.85)**
WCST FTMS 68 (1.25) 1.05 (1.35)
Tower of Hanoi 23.58 (9.70) 42.89 (5.32)***
Temporal order memory:"
Temporal order-pictures 61.05 (12.89) 69.68 (9.71)*
Recognition-pictures 79.53 (15.50) 86.11 (8.93)
Temporal order-words 64.28 (10.71) 73.26 (11.56)*
Recognition-words 74 83 (10.44) 75.89 (9.00)
Digit span.
Digits forward 5.68 (2.71) 6.05 (1.96)
Digits reversed 4.68 (2.36) 5.26 (2.32)
Total span 10.37 (4.67) 11.32 (3.64)
Standard age score'' 43.00 (9.81) 45.00 (6.32)
Working memory;
Sentence span 3.53 (2.07) 6.37 (2.29)***
Counting span 4.53 (3.95) 7.53 (2.50)**
" Percent correct (chance = 50%)
'' Standard age scare (mean = 50, SD = 8)
* p < .05.
** p < .01
*** p < .001
snbjects with antism performed differen- deficits are less likely to be artifacts of test-
tially worse than comparison subjects on the ing error. In contrast, for the Picture task,
Temporal Order condition compared to the comparison subjects' mean level of perfor-
Recognition condition, as predicted. Spe- mance on the Temporal Order condition was
cifically, analyses indicated that the groups lower than performance on the Recognition
were different on the Temporal Order task, condition. Though the variance on both con-
t(36) = -2.45, p = .02, but not on the Rec- ditions is similar, it is still worth considering
ognition task, ^(36) = - .33, p = .74. that this pattern may reduce the capability
of detecting a true differential deficit.
The corresponding ANOVA for the Pic-
ture tasks yielded a main effect of group, California Verbal Learning Test (CVLT),
F(l, 35) = 5.86, p = .02, and a main effect Children's Version
of task, F(l, 35) = 56.34, p < .001. There A series of t tests were performed to
was not a group x task interaction. compare the performance ofthe autistic and
comparison groups on the free recall and
Notably, performance on these tasks cued recall scores of the CVLT (see Table
does not appear to be compromised by a re- 3). The autistic group recalled fewer items
stricted range of scores, mean level of per- than the comparison group on the final three
formance for all subjects was well above learning trials of List A: Trial 3, t(36) =
chance (50% correct) and below ceiling -2.22, p = .03; Trial 4, t{36) = -2.25, p =
(100% correct) for all four tasks. From the .03; and Trial 5, t(36) = -2.08, p = .04.
Word task, one can infer from comparison
Consequently, their total recall score for List
subjects' performance that the Temporal Or-
A was low-er than that ofthe comparison sub-
der and Recognition conditions of the Word
jects, t(36) = -2.10, p = .04. There were
task both had an accuracy level midway be-
tween chance aiid perfect performance (i.e., no group differences on the first two learn-
75% correct). Tasks with a mean level of per- ing trials of List A. Similarly, there were no
formance in this range tend to be more reli- group differences on the free recall trial of
able and yield observed-score variances the interference list. List B.
more similar to true-score variances (Chap- Since the autistic group learned fewer
man & Chapman, 1978). Thus, differential items than the comparison group on the
TABLE 3
CVLT PEEFORMAXCE
AUTISTIC COMPARISON
GROUP GROUP
M (SD) M (SD)
List A.
Nnmher correct.
Trial 1 5.58 (2.78) 6 84 (2.48)
Trial 2 8.00 (3.40) 9..53 (2.76)
Trial 3 9.21 (3.87) 11.58 (2.59)*
Trial 4 9.16 (4.62) 11.95 (2.80)*
Trial 5 9.68 (4.08) 12.00 (2.60)*
Total (1-5) 41.63 (17.42) 51.90 (12.28)*
List B;
Number correct 6.16 (2.8.5) 6.21 (2.10)
List A short delay;
Percent recalled from List A, Trial 5;
Free 81.15 (.32) 91.44 (.14)
Cued 86.85 (.30) 93.14 (.16)
List A long delay;
Percent recalled from List A. Trial 5;
Free 90.04 (.26) 93.94 (.14)
Cued 88.83 (.29) 96.83 (.15)
Total perseverations 5.79 (6.29) 3.47 (2.78)
Total intrusions 19.32 (31.69) 2.21 (2.55)*
Delayed recognition;
Cowect (of 15) 13.16 (3.08) 13.89 (2.28)
False positives 5.26 (7.62) .32 (.58)**
Type;
B (shared categorj') 1.21 (1.48) .16 (.50)*
B (semantic, similar) 1.05 (1 58) 0*
B (nonshared category) 47 (1.02) 0
Semantic. Similar to A 1.37 (1.80) .16 (.38)*
Unrelated 79 (1.48) 0
Phonetically similar 37 (1.01) 0
* p < .05.
** p < 01.
*** p < .001
learning trials, retention scores for each of als. An analysis of these error types allowed
the delay trials were calculated relative to for us to examine the degree to which sub-
each subject's performance on the final re- jects were infiuenced by an inappropriate
call trial of List A. There were no group dif- context (i.e., a source other than the current
ferences on the percent of information re- list) in generating responses. The autistic
tained across the short or long delays for and comparison groups were compared on
either free or cued recall. After a short delay, the total number of perseverations and intru-
the subjects with autism remembered an av- sions made across trials on the CVLT. Per-
erage of 81% on the free recall trial and 87% severations were defined as repetitions of
when they received cues. The comparison the same word within a given trial; intru-
subjects remembered 91% on the free recall sions were defined as responses that were
trial and 93% on the cued trial. After a 30- not on the target list. Note that the term
min delay, the subjects with autism lecalled "perseveration" used here does not carry the
90% on the free recall trial and 89% when same meaning as perseverative responses on
they received cues. The comparison subjects EF tasks like the WCST; an increased num-
recalled 94% on the free recall trial and 97% ber of perseverations on the CVLT reflects
when they received cues. not only a perseverative response style, but
also a severely impaired short-term memor>',
The format ofthe CVLT allows for cod- since subjects are repeating the same
ing of recall errors that might interfere with word(s) within a severed second response pe-
performance on the cued and free recall tri- riod. There was no group difference on the
total number of perseverations across all re- the total score (Thorndike et al., 1986); mean
call trials. In contrast, the subjects with au- performance levels of the autistic and com-
tism made more intrusioD errors than the parison groups on this score indicated that
comparison subjects across all recall trials both groups were within the normal range
t{36} = 2.35, p = .03. when compared to other children their age.
An analysis of the type of intrusion er- Working Memory Span Tasks
rors rtiade by the subjeets with autism and The mean score of the autistic group
th^ coinparison subjects yielded several in- was lower than that of the comparison group
teresting results. AH intrasion en-ors were on both the Sentence Span task, t(36) =
scored as fallioig into one of four exclusive -4.Q2, p < .001, and the Counting Span
catjegories. Source errors were deifined as task, tim) = -2.80, p = .008. Because ver-
items that appeared on a previous list. Se- bal short-term niemory is thought to be one
mantic intrusions were Words that were se- of the requisite abilities fpr performance on
mantically siiiiilar to one of the categories these types of dual tasks (e.g., Baddeley,
on the target list, excluding source errors. 1986), Pearson correlation coefficients were
Phonetic intrusions were those intrusions coniputed between Digit Span Forward and
that were phonetically similar to one of the each working memorj' task: Counting Span,
words on the target list (e.g., "bat" for "hat"), r(38) = .60, p < .001; Sentence Span, r(38)
and were not scorable as source or semantic = .36, p = .03. Analyses of covariance were
errors. Uia,related intrusions <;oBiprised all performed on both working memorj- tasks,
other intrusiciias. Majtin-Whitnisy tests \yere with forward digit span as the covariate. The
perfonned to examine group differences on group difference remained for both the Sen-
these error tv'pes. There was a group differ- tence Span task, F(l, 37) = 16.62, p <. 001,
ence on the number of source errors made and tlie Counting Spao task, F(i, 37) = 9.87,
(V = 81.5, p = ,dO6), with subjects with au- p = .003.
tism making more source errqrs than com-
parison subjects. Though the autistic group
RELATIONS AMONG DEPENDENT MEASURES
made more total semantic intrusions than
the comparison group, this difference did Pearson correlation coefficients were
not reach statistiieal significancfe. There were computed to assess the relations between
no group clifEei-eHees on the number of pho- FSIQ, short-term declarative memory, exec-
netic or unrelated intrusion errors. utive function, working memory span, mem-
ory for temporal order, and a composite
Oi;i the delayed recognition trial of the score from the CVLT that reflects the ability
CVLT, the t3i¥o poups did not differ on the to inhibit irrelevant responses in recall.
number of correct items endorsed, but the Short-term declarative men]or\' was mea-
autistic group maije more false positive er- sured by forward digit span. Composite
rors than the comparison group (U = 85, p = scores were made froni the two e.\ecutive
.005). Furthermore, the subjects with autism function tasks, the two working memory
and the comparison siibj'ects differed on the span tasks, the temporal order memory tasks
types of distractor itertls diey endorsed. (Word and Picture), and the recognition
Cornpared to the comparison subjects, the merrior).' tasks (Word and Picture). The inhi-
subjeets with autism endorsed more items bitioH composite score firom the CVLT was
that were semanticaliy slmiJar to words on computed by transforming the total number
the target list. These weeds were either of ifltrusions and false positives into z scores.
items from List B that shared the same se- The sum df these two numbers was then re-
manti!c category as List A (L' = 93.5, p = standardized. The correlation analyses were
.01), a similar semantic categary with List A performed separately for the autistic and
{U — 114.(), p •= .(iS):, • ei* were new words corftparison groups. The higher scores ofthe
that were serBaatieally sirnilar to a category comparison group on the EF tasks, and the
on List A (17 = 105;0,:jj = .63). In'coiiitrast, fact that adult-level performanee on the
the groups did not diiffer on semantically un- WQST is reportedly achieved by age 10
rdated false positives. (Welsh, Pennington, & Croisser, 1991),
raises the possibility that the comparison
Digit Span group's performance level is close to ceiling
The groups did not differ on the raw on the standard EF tasks.
score of the Stanford-Binet Digits Forward
taslc. Digits Reversed task, or the total score The intercorrelation matrix is presented
of tliis test (see Table 2). A standard age in Table 4. To facilitate comparison, all vari-
scoire was also calculated for each subject on ables were recoded so that a positive score
TABLE 4
COHHELATtONS .4MOXG DEPENDENT VAHIABLES
Verbal Temporal
FSIQ STM EF WM Order Recognition
Intercorrelation Matrix, Auinstic Group Only
Verbal STM 25
EF .78*** .28
WM 73*** .43* .61**
Temporal order 21 .53* .55* .35
Recognition 18 72** .46 .52* .53*
CVLT-Inhibition 65** .28 .79*** .63** .28 .61*
1 Tcorrelation Matrix, Comparison Gn
Verbal STM -.07
EF 23 .54**
WM 45* .49* .63**
Temporal order 03 .12 .34 .19
Recognition 77*** -.11 01 .24 .05
CVLT-Inhibition 39 -.15 .19 .47* .10 .17
* p < .05
*' p < 01.
*** p < .001.
reflected good performance. Below we sum- p = .01). In all but the last case (FSIQ and
marize the correlations that were significant recognition memory), the correlations were
at p < .05 or lower. In the autistic group, higher in the autiistic group.
working memory was related to FSIQ, ver-
bal STM, executive function, recognition Discnssion
memory, and the CVLT inhibition compos-
ite. Executive function was also related to This study examined memory functions
FSIQ, temporal order memory, and the in a sample of high-functioning individuals
CVLT inhibition composite. The CVLT in- with autism and age- and IQ-matched clini-
hibition composite was also related to FSIQ cal comparison subjects. The results of the
and recognition memory. Recogniti on mem- study help to clarify the nature of memorv'
ory was also related to temporal order mem- functions in autism, and suggest important
ory and verbal STM. Finally, verbal STM implications for tlie existence of an impair-
was also related to temporal order memory. ment in general cognitive processes in this
In the comparison group, working memory disorder.
was related to FSIQ, verbal STM, executive Subjects with autism displayed a pre-
function, and the CVLT inhibition compos- dicted pattern of performance across a vari-
ite. Executive function was also related to ety of memory tasks. They demonstrated a
verbal STM, and FSIQ was correlated with profile of intact and impaired functioning
recognition memory. similar to that found in patients with execu-
Ofthe 21 correlations produced by this tive function deficits, such as patients with
matrix, 13 were significant at p < .05 or frontal lobe pathotogv-.
lower in the autistic group, compared to six Subjects with autism performed worse
in the comparison group; both proportions than comparison subjects on temporal order
are greater than would be expected by memory' for verbal information. Their intact
chance (e.g.., 1/21). The correlations in the performance on tlie recognition portion of
two groups were compared via Fisher's Z the task indicates that their impairment on
transformation (Cohen & Cohen, 1983). Four temporal order memory was not the result of
of the 21 correlations were significantly dif- a generalized STM deficit. Since this task
ferent between the two groups, executive was carefully constructed so that attentional
function and FSIQ (Z = 2.25, p = .005), demands and item difficulty were consistent
executive function and CVLT inhibition across the two conditions, we are more con-
(Z = 2.44, p = .007), verbal STM and recog- fident that this finding reflects a differential
nition memor>' (Z = 2.75, p = .003), and deficit, rather than just differential perfor-
FSIQ and recognition memory (Z == -2.33, mance.
There was not, however, a group x task primed the appropriate semantic categories,
interaction on the representational drawing but the subjects with autism failed to use
portion of this task. There are several possi- the context of the current task to deselect
ble explanations for this null result. First, inappropriate responses. For example, when
using a similar version of this task, Milner asked to recall words from the target list that
et al. (1991) found a deficit on temporal order were "things to eat," one subject responded
for concrete words in patients iwth both left with "apples, oranges, bananas, plums,
and right frontal lesions, but found a deficit grapes, lemons, coconut, pineapples, ber-
in temporal order for representational draw- ries." Only "bananas" and "grapes" were on
ings only in patients with right frontal le- the target list.
sions. Thus, the Temporal Order for Words
task rpav be more sensitive in detecting iiti- Finally, subjects with autism demon-
pairment, regardless of lesion loeation. Sec- strated intact performance on Digit Span, a
ond, performance on picture recognition was measure of auditory short-term memory.
relatively good in both groups. This finding There were no group differences on either
is consistent %vith previous findings tliat rec- the Digits Forward or Digits Backward por-
ognition memory for pictures develops early tions of this task. This may seem surprising,
and reaches high levels of accuracy at a since Digits Backward is a more demanding
young sige (Brown & Scott, 1971). The fact task, which may draw upon more organiza-
that there was a significant effect of task tional and on-line processing resources than
(with recognition memory better than tem- Digits Forward. However, the additional
poral order memorv) in both groups may processing demands may not ba significant
have made it mpre difBcult to detect a group enough to affect performance in subjects
X task interaction. witii relatively intact STM. Moreover., per-
formance.s on tiiese two subtasks are highly
On the CVLT, subjects with autism con-elated in the normaj population (Thorn-
demonstrated a flatter learning curve on the dike et aL, 1986).
List A learning trials than comparison sub-
jects. While the two groups' performances ^ Thus, across a variety of memory tasks,
were similar on early learning trials, subjects suDjects with a.uiism demonstrated a consis-
with ûtism were unable to increase their tent pattern of memory functions that is simi-
span On the last two trials. This result may lar tc that observed in patients with frontal
reflect the descreased abiiitj- of subjects with lesioiis. This pattern included impairment
autism to use strategic organization or plan- on temporal order memory for verbal infor-
ning in recall This poor performance on su- matiDn, supraspan verbal learning, and the
praspan learning is consistent with previous ability to maintain the appropriate context of
studies of the performance of children with the ipfquriaaon they had learned. In addi-
autisin on the Bushke Selective Reminding tion, subjects with autism showed an impair-
Test (Ozonoff et ai., 1991) and ttie CVLT meat in two standard tasks of EF. The re-
(Minshew & Goldstein, 1993). sults Qf this study suggest that there is a
The pattern of errors made by children similaaitj' between individuals with autism
with autism on both recall and recognition and pfttients with frontal lesions on memory,
trials ofthe CYLT is consistent with a deficit as i*ell as EF tasks.
in source meitaory. On recall trials, the sub- IH addition, the pattern of intact perfor-
jects with autism made more intrusion errors mance of subjects with autism does not sup-
than comparison subjects. The nature of port ike existence of traditional amnesic
their intrusion errors reflected a failure to deficits in autism. Subjects with autism dem-
use the contejct ofthe current list to constrain onstrated intact functioning on standard de-
responses:' they recalled more words that clarajtive imemor>' fiinctions that are typically
had appeared on a previous but no longer compromised in classic amnesia. In particu-
correct list, aistd tended to recall words that lar, they were not different from comparison
were semantically similar to the target list. subjects on tasks of standard rote memorj',
Similarly, subjects with autism endorsed verbal long-term memory, or recognition
more false positive items than comparison merobry, either for verbal or pictorial infor-
subjects on the recognition trial. They matiOB. These results are consistent with
tended to endorse false positive items that boA previous neuropsychologieal studies of
yi'ere semajatieallv similar to the target list, autipi (e.g., Minshew & Goldstein, 1993),
but tljiey did aot endorse unrelated false pos- and Seliajvioral obserxâtions of often very
itive it^ms. This paftern of errors suggests goo4 rote memory in individuals with
that previous exposure to target words autisHi.
Although these results point to similari- WORKING MEMORY IN AUTISM
ties between individuals with autism and We propose that the pattern of impair-
patients with frontal lesions, and provide ments found in this study may reflect a more
support for an EF theory of autism, there are general deficit in Working Memory (WM).
several caveats to consider. First, as men- The construct of WM was first proposed by
tioned in the infroduction, deficits on EF Baddeley and Hitch (1974) to refer to the
and other tasks sensitive to frontal damage simultaneous processing and storage of in-
do not necessarily indicate the existence of formation during complex cognitive tasks.
frontal lobe patholog>'. Thus, even though Wliile working memory has been a very use-
we found that individuals with autism are ful construct in understanding performance
most severely impaired on tasks that are sen- on EF and other complex tasks, it too is be-
sitive to frontal lesions, we do not necessar- set with multiple and sometimes inconsis-
ily expect localized damage or dysfunction tent theoretical definitions. Currently, there
in the frontal lobes. Second, though patients are several different ways to define and mea-
with frontal lesions exhibit many social and sure this construct in humans (e.g., Bad-
cognitive deficits, the fact remains that they deley, 1986; Case, 1985; Halford, Maybery,
are not autistic. A review of tbe effects of O'Hare, & Grant, 1994), and even more in
early frontal lesions in children indicates the animal literabire.
that such individuals tend to present as con-
duct disordered rather than autistic (Pen- As an initial examination of possible
nington & Bennetto, 1993). Third, defi^cits in WM deficits in autism, the current study as-
executive functions are not specific to au- sessed the performance of subjects with au-
tism. Impairments on EF tasks have been tism on two verbal v/orking memory span
found across a rarige of neuropsychiatric and tasks. We found that subjects with autism
developmental disorders, including early- performed worse than comparison subjects
treated PKU (Diamond, Ciaramit£iro, Don- on both tasks, which provides preliminary
ner, Djali, & Robinson, 1994; Welsh, Pen- evidence for a deficit in WM. As mentioned
nington, Ozonoff, Rouse, & McCabe, 1990), above, however, WM is a complex construct;
Attention Deficit Hyperactivitv' Disorder the WM span tasks in this study were de-
(Grodzinsky 6c Diamond, 1992; Pennington, signed to measure concurrent storage and
Groisser, & Welsh, 1993), Tourette syn- processing of verbal information, and do not
drome (Bomstein, 1990), and Fragile (X) include control tasks to allow for a more
syndrome in females (Mazzocco, Hagerman, careful assessment of individual compo-
Cronister-Silverman, & Pennington, 1992). nents such as capacity or processing effi-
Thus, an impairment in EF cannot be neces- ciency, or other factors such as counting time
sary and sufficient to cause the spectrum of and articulation speed. Thus, there are sev-
symptoms associated with autism, if this im- eral alternative explanations that could ac-
pairment is of similar severity and kind as count for these results. For example, on the
those found in these other disorders. Sentence Span task, it is possible that the
subjects with autism had more difficulty fill-
Finally, tlie subjects in this study have ing in the missing word because of deficits
participated in other studies in this area in understanding relevance (e.g.. Frith &
(Ozonoff & McEvoy, 1994; Ozonoff et al., Happe, 1994). This increased processing
1991). Thus, it is important to address the load could leave less processing space avail-
potential threat this poses to the v,alidit>- of able for storage or the execution of otber ef-
our findings. As is common in low-i ticidence fortful processes in working memory. Al-
disorders, the same subject group is often though no subject from either group failed
used for multiple studies, which conse- to find a missing v/ord on these tasks, future
quently increases the risk of Type I error. research on nonverbal WM in autism is im-
Since it is usually not possible for a lab to portant. In addition to affecting differential
recruit independent samples for each study, processing demands, difficulty with finding
it is critical that these findings be replicated the relevant word could have increased the
with £in independent group before strong response times of subjects with autism.
conclusions are drawn. Longer response ti'mes on this, or the Count-
ing Span task, could mean that subjects with
To understand more clearly the nature autism performed worse simply because
of these niemory and E F deficits, and ad-
dress some of the preceding caveats, we they bad to hold onto the information over a
need to examine the component processes longer delay (Towse & Hitch, 1995). Future
that might be involved in these neujocogni- studies that measure or control for response
tive tasks. speed are needed to sort out these possibili-
ties. Nonetheless, these findings provide determine whether WM is particularly
preliminary evidence of a WM deficit and stressed in the types of social cognition tasks
indicate the importance of further examina- children with autism fail. As a preliminary
tion of possible subcomponents of EF in look at the reia;tionship betvveen WM and
autism. other symptoms of autism, we performed a
series of correlations between our cognitive
It has been proposed that WM permits variables and subjects' ratings on the CARS.
one to solve problems that are transient, con- In the autistic group, the CARS score was
text-specific, and require the integration of related to VIQ. f<19) = - . 8 5 , p < .001;
information over space or time (Pennington, FSIQ. r(19) = - .54, p = .02; WM Sentence
1994). From this perspective, the results of Span,r(19) = - . 7 2 , p = .001; and tire CVLT
the present study shed light on other cogni- inhibition composite, r(19) = — .68, p =
tive symptoms found in ûtism. In other ar- .002. Ie ali cases, a higher CARS score (i.e.,
eas of' research, empirical links have been iiiqi-e symptoms endorsed) was related to
found between WM and discourse processes worse performance on the task. There were
(e.g.. Carpenter & Just, 1989), narrative gen- no si gnificant correlations in the comparison
eration (e.g., Dennis, 1991; Grafman, 1989), group. The relationship between the CARS
learning (e.g., Baddeley, 19S6), aiid reason- and 10 is not surprising.^ since low verbal
ing and novel problem solving (e.g.. Carpen- abilities is one of the fiaiimarks of autism.
ter, Just, & Shell, 1990; Kyllonen & ChristaL Similarly, correlations with WM Sentence
199u), ail of wliicfa. ai-e deficient in autism. Span and CVLT inhibition could be re-
fleetiug the verbal deficit, Koaetheless,
How does this model relate to social im- these coraelations suggest that cognitive
pairment, die roie feature of individuals constructs such m WM or the inhibition of
With autism^ I h e r e has been a tendenev in inajppropriate context m:ay be related to
autism research to explam social impairment some of tlie symptoms of autism.
in teims of a specific deficit (see Baion-
Cohen, Tagei-FIu&beig, & Cohen, 1993, for Thus, an impairment in WM could ex-
a reMe\\) Howe\eT, givea nliat is known pla;in what appeal's to be a more specific im-
about the relationship beiween cognitive pairrneiitin social cognition. Efieetive social
9.nd social de-% elopment, one must consider interactioa depends upon the on-line inte-
whether a social impaiiment could be ex- gration' of constantly changing, coiitext-
plained m teims of impaimient of a general speciHc information. In fact, the subtlet>' and
cogniti^'e process Fr>e, Zelazo and Palfai coHiDiexity of Ae cues that must be pro-
(1992) ha\e demoiiStrated simila. age- cesse<3 in a social iateraclion, as well as the
jrelate'd changes la both tlieoiT of nimd and selection of appropiiate responses, likely
cognitive tasls is Inch escli stared a coni- placfe a '&T greater load on WM than do the
mon logical stiucture They suggest tlial a tasks described ill this study.
developmental chaage m cogmtne com-
plexit} underlies tbe developmental sliifts Therefore, we suggest that the primary
in children s abilities, regiirdiess of the task deficit in autism may be one of general im-
ônfent. Ru!.3ell. Mauthiier, Sharpe, and pairment rather than specific modular im-
Tidsviell (19911 ha\e shown that cfaildien pairment. Deficits in central cognitive pro-
i«itl-i autism are miable to pass stialegie de- cessing have been suggested by others to
account for some of the social impairments
peotiOD tasks not because the\ dtt unable to of autism (e^.g., Courchesne etal., 1994; Frith
repres<3nt anotlier person s mental state, but & Hicppe, l&Q^Buglies et:al., 1994). In the
jsecause they he\e le?" executi'.e control preseat study, there is additional evidence
o\fei then heha.\ loi and are iiJiabie to inhibit for a deieit in a central;cognitive process:
inappropriate responding In addition, e-êc- there was a stroiiger pattern of intercorrela-
utive funchoii aud v.oik:mg memoir' tasks tions among the dependent variables in the
lia\e been ahowi to be related to othei areas aiitiAie group coinpar:e4 *o the comparison
9f sociil abilit\ m dtitism McE\ov et al. gro«|>.:'®tiier stadies; of elnldren with autism
(1993* have shorn ii a relatton5.hip beL«een have toHttid a similar patt<sm of stronger rela-
pxecutive lunrtlon tasks and joint attention tionshtps in the autistic compared to control
/E in pieschooleri miiJri t groups'Bistween cognitive vaiiables (Ozo-
ndff-eta!., 1991; Yirmiya, Sigman, Kasari, &
Tlie current study did not examine so- Muodji?, 1992) as well as between cognitive
cial deficits ill autisin; tbiis we can only variaiSes and nenroflliysiologic indices
speculate on thfeJeiatioBstip: between gen- (Mii'iSilae-w, GoHsteitt,' Dqmbrowsld, Pan-
eral cognitive'grQGesses, siicb as WM, and chalingam, & Pettegrev/, 1993). A deficit in
social'tasks. FiiitJier research is necessai-y to
a central cognitive process would limit de- developmental disorders may be less severe,
velopment in many domains and permit less or manifest at a later stage. A second possi-
differentiation of domains than is found in bility is that the WM deficits m autism are
normal development. secondary to soroe other cause. All of the
evidence for an executive deficit in autism
So far, we have discussed the possibility has come from cross-sectional studies.
of a general impairment in WM to account Hence, this evidence is only correlational,
for the cognitive deficits found in tbis study, and thus tells us littie about the causal rela-
as well as some of the social deficits in au- tionship between WM and other deficits in
tism. However, there are several challenges autism. Even though these studies have con-
to this model. First, many of the tasks em- trolled for IQ and clinical status, it could still
ployed in this study were verbal in nature. be the case that growing up autistic has
Individuals with autism are typically weak unique effects on the development of WM.
in language skills, and thus poor perfor- Social interactions provide unrelenting prac-
mance on these tasks could be an artifact of tice in shifting cognitive set and distinguish-
poor verbal abilities (e.g., a deficit in seman- ing different contexts. A third possibility is
tic encoding). However, the subjects with that there are two deficits involved m au-
autism in this study were matched with com- tism: a general deficit in WM, and a specific
parison subjects on verbal IQ to ensure that deficit in some otheT area of development,
their ability to understand verbal informa- perhaps related to human sociability. For ex-
tion was not significantly worse than com- ample, there is evidence of an imitation
parison subjects. In addition, there is evi- deficit in autism (DeMeyer et al., 1972; Rog-
dence from this study and others (e.g., Ameli ers, Bennetto, McEvoy, & Pennington, in
et al., 1988; Ramondo & Milech, 1984; press). An early deficit in imitation would
Tager-Flusberg, 1991) that indi^dduals with prevent the infant from entering into the so-
autism are able to encode semantic infor- cial "dance." A general deficit in WM would
mation. make later understanding ofthe social world
difficult, since social interaction is transient,
Second, to recall or recognize a piece context-specific, and requires the integration
of information, subjects must be attending of information from diverse sources. Further
sufBcieutly to the to-be-remembered infor- research is needed to distinguish these com-
mation during the encoding stage. There is peting possibilities.
evidence that some individuals with autism
have an impairment in shifting attention rap-
idly from one stimulus or modality to an- References
other {Courchesne et al., 1994), which could
affect their performance on the metnor>' and Ameli, R., Courchesne, E., Lincoln, A., Kaufman,
EF tasks used in this study. Some types of A. S., & Gnlloa, C (1988). Visual memory
memory (e.g.. Temporal Order) were mea- processes in high-functionmg individuals
sured with tasks that controlled, within sub- with autism. Journal of Autism and Develop-
jects, the attentional demands needed for mental Disorders, 18, 601-615.
perfoimance. Future research in this area Bachevalier, J. (199]). .\n animal model for child-
would benefit from more control tasks to en- hood autism. Memory loss and socioemo-
sure that differences in attentional demands tional disturbances following neonatal dam-
are not biasing results. age to the limbic system m monkeys. In C. A.
Tamminga & S C. Schulz (Eds.), Advances
Finally, before we can be confident that in neuropstfchiatry and psychopharmacology
any cognitive deficit is primary to autism, we (Vol. I, pp. 129-140). New York: Raven.
need to establish the causal relation be- Baddeley, A. D. (1986). Working memory. New-
tween the deficit and the also present social York. Oxford University Press
deficits of aytism. For the model we have Baddeley, A. D., & Hitch, G, J. (1974) Working
proposed above, this relation has not yet memor>'. In G. H. Bower (Ed.), The psychol-
been established empirically. We ofier sev- ogy of learning and motivation (Vol. 8, pp.
eral alternative explanations for further re- 47-89). New York: Academic Press
search. The first possibility is that individu- Baron-Cohen, S., Tager-Flusberg, H., & Cohen,
als with autism have a primary deficit in WM D. J. (Eds.). (1993). Understanding other
that is so severe and occurs early enough in minds: Perspectives from autisin. New York.
development Ithat it interferes witli the in- Oxford Universiiiy Press.
fant's ability' to engage in other develop-
Bartak, L., Rutter, M., & Cox, A. (1975). A compar-
mental tasks such as imitation and intersub-
jectivity. The WM deficits in odier ative study of infantile autism and specific de-
velopmental receptive language disorder. I.
The children. British Journal of Psychiatry, ioral sciences (2d ed.). Hillsdale, NJ:
126,127-145. Erlbaum.
Bauman, M. L., & Kemper, T. L. (1985). Histoana- Courchesne, E., Townsend, J. P., AkshoomofF,
tomic observations of the brain in early infan- N- A., Yeung-Courchesiie, R., Press, G. A.,
tile autism. Neurology^ 35, 866-874. Murakami, J. W., Lincoln, A. J., James, H. E.,
Borys, S. V., Spitz, H. H., & Dorans, B. A. (1982). Saitoh, O., Egaas, B., Haas, R. H., & Schreib-
Tower of Hanoi performance of retarded inaR, L. (1994). A new finding: Impaiiment
young adults and lionretarded children as a in sMfting attention in autistic and cerebellar
function of solution length and goal state. patients. In S. H. Broman & J. Grafnian (Eds.),
Journal of Experimentdl Child Psychology, Atypical cogftitive deficits in developmental
33, 87-110. disorders: Implications for brain function.
Boucher, J. (1978). Echoic memory- capacitv' in au- Hillsdale, NJ: Erlbaum.
tistic children. Jonrnal of Child Psycfiology Craik., V. L M., Monis, L. W., Morris, R. G., &
and Psychiatry, 19,161-166. Loewen, E. R. (1990). Relations between
Boucher, J. (1981). Memory for recent events in source amnesia and frontal lobe functioning
autistic children. Journal ofAuMsm and De- in older adults. Psychokigy and Aging, 5,
velopmental, Disorders, 11, 293-.301. 148-151.
Boucher. J., & Warriiigton, E. K. (1976). Memory CumimiJags, | . L. (1993). Frontal-subcortical cir-
deficits in early infautitile autism. Some simi- cuits and human behavior. Archives of Neti-
larities to the amaesic syndrome. Britishjour- rologv, 50, 87.3-880.
VMlofPstijchology, 67, 73-87. Damasio, A. R., & Maurer, R. G. (1978). A neuro-
Bomstein, R. A. (19^0). Neuropsychological per- logical model for childhood autism Archives
formance in children with Tourette syn- oftfeumlogii, 35, 777-786.
drome. Psychiatry Research, 33, 73-81. nemiMi, M., & Carpenter, P. A. (1980). Individ-
Brown, A. L., & Scott, S. S. (1971). Recognition ual clifferences in working memory and read-
inemory for pictures in preschool children. ing, journal of Verbal Learning and Verbal
Journal of Experimental ChUd Psychology, Behaoior, 19, 450-466.
U. 401-412. Delis, D. C , Kramer, J. H., Kaplan, E., & Ober,
Carpenter, P. A.. & Just, M. A. (1989). The role of B. A. (1986). The California Verbal Learning
working inemory in language comprehension. Test-^-research edition. New York: Psycho-
In D. Itlahr & K. Kotovsky (Eds.), Complex logical Corp.
inforniation proce.ssing: The impact of Her- DeLong, G. R. (1992). Autism, amnesia, hippo-
hert A. Simon (pp. 31-68). Hillsdale, NJ: campus, and learning. Neuroscience and Bio-
Erlbaum. behdmoral Retnews, 16, 63—70.
Carpenter, P. A., Just, M. A., & Shell, P. (1990). DeMeyei-., M. K., Alpem, G. D., Barton, S., De-
What one intelligence test measures- A theo- Meyer, W. E., Churchill, D. W., Hingtgen,
retical account of the processing in the Raven J, K,, Brj'son, C. Q., Fontius, W., & Kimberlin,
Praêssive Matrices Test Psychological Re- C. (19T2). Imitation in autistic, early schizo-
view, 97, 404-4.31. phrenic, and non-psychotic children. Jounial
Case, R. (1985). Intellectual development: Birth of Attiism and Childhood Schizophrenia, 2,
to adulthood. New York; Academic Press. 26i-2S7.
Case, R. (1992). The role of the frontal lobes m Denais, Ki. (1991). Frontal lobe function in child-
the regulation, of cognitive development. hoocl artd adolescence: A heuristic for assess-
Brain and Cognition, 20, 51-73. jflg attention regulation, executive control,
Case, R., Kurland, M., & Coldberg, J. (1982). Op- and the intentional states important for social
erational eflBciency and the growth of short- diseoarse. Developmental Neuropsychology,
tenh memory S^pan. Journal of Experimental 7, 327-358.
Child Psychology, 33, 386-404. Diamond, A., Ciaramitaro, V., Donner, E., Djali,
Cermak, L. S., Butters, N., & Moreines, J. (1974). S., & Robinson, M. (1994). An animal model
Some analyses of the verbal encoding deficit of early-treated PKU! Journal of Neurosci-
m alcoholic Korsakofif patients. Brain and ence, 14, 3072-3082.
Language, 1, 141-150. Dunn, L. M. (1965). Peabody Picture Vocabulary
Chapman, L. J., & Chapman, ]. P. (1978). The Test, Circle Pines, MN: American Guidance
measurement of differentia! deficit./ournaZ of Service.
Psychiatric Research, 14, 303-3ll. Eslinger, P. J., & Grattan, L. M. (1991). Serial posi-
Cohen, J. D., & Servan-Schreiber, D. (1992). Con- tion dysfunction after frontal lobe lesion.
text, cortex, and dopamme: A connectionist Jmnial of Cliniaaland Experimental Neuro-
approach to behavior and biology in schizo- psuehMogy, 13, 44
jihreriia. Psychological Review, 99, 45-77. Fritij, U.., & Happe, F. (1994). Autism: Beyond
Cohen, J., & Cohen, P. (1983). Applied multiple •'theon' of mind." Cognition, 50, 115-132.
regres.sioni correlation analysis for the behav- Fr>'e D.. Zelazo. P. D., & Palfai, T. (1992). The
cognitive basis of theory of mind. Manuscript Janowsky, J. S., Shimamura, A. P., & Squire, L. R.
under review. (1989). Source memorj' impairment in pa-
Fuster, J. M. (1985). The prefrontal cortex, media- tients with frontal lobe lesions. Neuropsycho-
tor of cross-temporal contingencies. Hum,an logia, 27, 1043-1056.
Neurohiology, 4, 169-179. Jetter, W., Poser, U., Freeman, R. B., & Marko-
Gardner, M. F. (1979). Expressive one-word pic- witsch, H. J. (1986). A verbal long term mem-
ture vocabvlary test. Novato, CA. Academic ory deficit in frontal lobe damaged patients.
Therapy Publications. Cortex, 22, 229-242.
Gilhooly, K. J., & Logie, R. H. (1980). Age of ac- Kyllonen, P. C , & Christal, R. E. (1990). Reason-
quisition, imagerj', concreteness, familiaritj' ing abilitj' is (little more than) working mem-
and ambiguity measures for 1944 vi'ords. Be- ory capacitj'?! Intelligence, 14, 389-433.
havior Research Methods and Instrumenta- Levin, H. S., Mendelsohn, D., Lilly, M. A.,
tion, 12, 395-427. Fletcher, J. M., Culhane, K. A., Chapman,
Gold, J., Berman, K. F., Randolph, C , Goldberg, S. B., Harward, H., Kusnerik, L., Bruce, D.,
T., & Weinberger, D. R. (1991). PET valida- & Eisenberg, H. M. (1994). Tower of London
tion and clinical application of a novel pre- performance in relation to magnetic reso-
frontal task. Neuropsychology, 13, 81. nance imaging following closed head injury
Goldm.m-Rakic, P. S. (1987). Circuitry of primate m children. Neuropsychology, 8, 171-179
prefrontal cortex and regulation of behavior Mazzocco, M. M. M., Hagerman, R. J., Cronister-
by representational memory'. In F. Plum Silverman, A., & Pennington, B. E. (1992).
(Ed.), The handbook of physiology: Section Specific frontal lobe deficits among women
1. The nervous system: Vol. 5. Higher func- with the fragile X gene. American Academy
tiijns of the brain (pp. 373-417). .Bethesda, of Child and Adolescent Psychiatry. 31,
MD: American Fhysiological Society. 1141-1148.
Grafman, J. (1989). Plans, actions, and mental sets. McEvoy, R. E., Rogers, S. J., & Pennington, R. F.
Managerial knowledge units in the frontal (1993). Executive function and social commu-
lobes. In E. Perecman (Ed.), Integrating the- nication deficits m young autistic children.
ory and practice in clinical neuropsychology Journal of Child Psychology and Psychiatry
(pp. 93-138). Hillsdale, NJ: Erlbaum. 34, 563-578.
Grant, D. A., & Ber^, E. A. (1948). A hehavioral Milner, B. (1971). Interhemispheric differences in
analysis of degree of reinforcement and ease the localization of psychological processes m
of shifting to new responses in a Weigletype man. British Medical Bulletin, 127,272-277.
card sorting problem. Journal of Experimen- Milner, B., Corsi, P., & Leonard, G. (1991). Fron-
tal Psychology, 32, 404-411. tal-lobe contribution to recency judgments.
Grodziiisky, G. M., & Diamond, A. (1992). Frontal Neuropsychologia, 29, 601-618.
lobe functioning in boys with attention-deficit Milner, B., Petrides, M., & Smith, M. L. (1985).
hyperactivity disorder. Developmental Neu- Frontal lobes and the temporal organization
ropsychology, 8, 427-445. of memory. Human Neurobiology, 4,137-142.
Halfprd, G. S., Maybery, M. T., O'Hare, A W., & Minshew, M. J., & Goldstein, G. (1993). s autism
Grant, P. (1994). The development oFmemory an amnesic disorder? E\ddence from the Cali-
and processing capacity. Child Development fornia Verbal Learning Test. Neurop.sychol-
65, 1338-1356. ogy, 7, 209-216.
Heaton, E. K., Chelune, G. J.. Talley, J L.. Kay, Minshew, N. J, Goldstein, G., Dombrowski, S M.,
G. G., & Curtiss, G. (1993). Wisconsin Card Panchahngam, K., & Pettegrew, J. W. (1993),
Sorting Test manual. Revised and expanded A preliminary 31P MRS study of antism: Evi-
Odessa, FL: Psychological Assessment Re- dence for undersynthesis and increased deg-
sources. radation of brain membranes. Biological Psy-
Hermelin. B., & O'Connor, N (1975). The recall chiatry, 33, 762-773.
of digits by normal deaf and autistic children Moscovdtch, M. (1992) Memory' and working-
British Journal of Psychology, 66, 203-209. with-memory: A component process model
Hollingshead, A. B. (1975). Two-Factor Index of based on modules and central system./oumaZ
Social Status. New Haven, CT: Yale Uni- of Cognitive Neuroscience, 4, 257-267.
versity. Moscôvitch, M. (1994). Cognitive resources and
Hughes, C., & Russell, J. (1993). Autistic chil- dual-task interference effects at retrieval in
dren's difficulty with mental disengagement normal people: The role of the frontal lobes
from an object: Its implications for theories and medial temporal cortex. Neuropsychol-
of autism. Developmental Psychology 29 ogy, 8, 524-534.
498-510. Ozonoif; t>., & Mcbvoy, R. E. (1994). A longitudi-
Hughes, C , Russell, J., & Robbins, T. W. (1994). nal study of executive function and theory of
Evidence for executive dysfunction m autism. mind development m autism. Development
Neuropsychologia, 32, 477-492. and Psychopathology, 6, 415-432.
Ozonoff, S., Pennington, B. F., & Rogers, S. J. Rybash, J. M., & Colilla, J. L. (1994). Source mem-
(1991). Executive fnnction deficits in high- ory deficits and frontal lobe fiinctioning in
functioning autistic individuals: Relationship children. Developmental Neuropsychology,
to theory of mind. Journal of Child Psychol- 10, 67-73.
ogy and Psychiatry, i2, 1081-1105. Schacter, D. L. (1987). Memory', amnesia, and
Pennington, B. F. (1994). Tlie working memory frontal lobe dysfunction. Psychobiology, 15,
function of the prefrontaJ cortices: Implica- 21-36.
iions for developmental and individual differ- Schacter, D. L., Harbluk, J. L., & McLachlan,
ences in cognition. In M. M. Haifli, J. B. Ben- D. R. (1984). Retrieval without recollection:
son, H. J^ Roberts, & B. F. PenniHgton (Eds.), Aa experimental analysis of source amnesia.
Future oriented processes in development Journal of Verbal Learning and Verbal Be-
(pp. 243-292). Chicago: University of Chi- havior, 23, 593-611.
cago Press. Schopier, E., Reichler, R. J., & Renner, B. R.
Penninpton, B. F., & Bennetto, L. (1993). Main (1986). The childhood autism rating scale
effpcts or transactions in the neuropsychology {CARS) for diagnostic .screening and classifi-
of condtict disorder? Commentary on "'The cation of autism. New York: Irvington.
neuropsychology of condtict disorder." De- Shimamtiia, A. P., Janowsky, J. S., & Squire,
velopment and Psgchopathology, S, 153-164. L. H. (1990). Memory for the temporal order
Penninôn, B. F., Groisser, D., & Welsh, M. C. of event.s in patients witli frontal lobe lesions
(1993). Contrasting deficits in attention deficit and amnesic patients. Neuropsychologia, 28,
hyperactivity disorder versus readhig disabil- S03-814.
ity. Developmental Psychology, 29,^11-523. Shimarmira, A. P., Janowsky, J. S., & Squire, L. R.
Prior, M. R., & Chen, C. S. (19t6). Short-terai and (1961). What is the role of frontal lobe damage
serial memory in autistic, retarded, and noi- in memory disorders? In H. S. Lerin, H. M.
mal children. Journal of Autisra and Child- Eisenberg, & A. L. Benton (Eds.), Frontal
hood Schisophrenia, 6, 121—131. lobe- function and injury (pp. 173.-195). New
Prior. M., & tioffmaim, W. (1990). Brief report: York: Oxford University Press.
Neuropsychological testing of autistic chil- Siegel, L. S., & Ryan, E. B. (1989). The develop-
dren through an exploration with frontal lobe ment of working memory in nonnally achiev-
tests. fournM of Auti-sm and Developmental ing arid subtyfpes of learning disabled chil-
Disortfere, 30., 581-590. dxen. Child Development, 60, 973-980.
Ramondo, N., & Milecti, D. (1984). The nature Sigman, M. (1994). '^Vhat aie the core deficits m
aad specificity ofthe language coding deficit aiifism? In S. H. Broman & J. Grafman (Eds.),
in autistic children. British Journal of Psy- At-gpical cognitive d^cits in developmental
chology, 75, 95-103. disorder.?.- Impllcaiions for brain function
Ra>Tnond, G., Baumaa, M., & Kemper, T. (1989). Hillsdale, NJ: Eribauro. "
The hippocampus in autism: Goigi analysis. Squire, L. R. (1987). Memory and brain. New
Annals of Neurology, 26, 483-484. York: Oxford University Press.
Rogers, S. J., Bennetto, L., McEvoy, R. E., & Pen- Stuss. D. T., Kaplan, E. F., Benson, D. F., Weir,
nington, B. F. (in press). Imitation and panto- W. S., Chiulli, S., & Sarazin, F. F. (1982). Evi-
mime in high-fmactioniiig adolescents with dence for the involvement of orbitofrontal
autisrri spectrum disorders. Child Devel- cortex in memory functions: An interference
optnent. effect Journal of Comparative and Physio-
Rumsev, J. M. (1985). Conceptual problem- logical Psychaldgij, 96, 913-925.
solving in highly verbal, nonretarded autistic Tager-Flusberg, H. (1991). Semantic processing
men. Journal of Autism, and Developmental in the free recall of autistic childien: Further
Disorders, 15, 23-36. evidence for a cognitive deficit. British Jour-
Rumsey, J. M.,fieHamburger, S. D. (1988). Neuro- nal of Defvelopmental Psychology, 9, 417-
psychological findings in high-functioning 4.30.
men with infantile autism, residual state. Thonidike, R. L., Hagen, E. P., & Sattler, J. M.
Journal of dUnical mid Experimental Neuro- (1986). Stanford-Binet Intelligence Scale (4th
psychoUgy,m, 201-221. ed.). Chicago: Riverside.
Russell, J., Mauthner, N., Sharpe, S., & Tidswell, Toglia, M. P.. & Battig, W. F. (1978). Handbook
rr. (l991). The '•windows task' as a measure of of semantic iDord norms Hillsdale, NJ:
strategic deception in preschoolers and autis- Erlbaum.
tic sub^jects. British. Journal of Developmental Towse, J. Nl, & Hitch, G. J. (199.5). Is there a rela-
Fs^chologij, 9, 331-349. tionship between task demand and storage
Rutter, M. (1983)., Cognitive deficits in the patho- soace in tests of working memory capacity?
genesis of autism. Journal of Child Psychol- Quarterly Journal of Experimental Psychol-
ogy and Psychiatry, 24, 513-531. om, 48, 108-124.
Welsh, M. C. (1991). Rule-guided behavior and Welsh, M. C , Peanington, B. F., Ozonoff, S.,
self-monitoring on the Tower of Hanoi disk- Rouse, B., & McCabe, E. R. B. (1990). Neuro-
transfer task. Cognitive Development, 6, psychology of early-treated phenylketonuria.
^^-'''6. Specific executive function deficits. Child De-
Welsh, M. C , & Pennington, B F. (1988). Assess- velopment, 61, 1697-1713.
mg frontal lobe fiinctioning in children: Woodcock, R. W., 6: Johnson, M. B. (1989). Wood-
Views from developmental psychology. De- cock-Johnson Psycho-Educational Battery—
velopmental Neuropsychology, 4, 199-230. Revised. Allen, TX: DLM Teaching Re-
Welsh, M. C , Pennington, B. F., &c Groisser, sources.
D. B. (1991). A normative-developmental Yirmiya, N., Sigman, M. D., Kasari, C , 65: Mundy.
study ofesecutive function: A window of pre- P. (1992). Empathy and cognition m high-
frontal function in children. Developmental functioning children with autism. Child De-
Neuropsychology, 7, 131-149. velopment, 63, 150-160.

Intact and Impaired Memory Functions in Autism

Uploaded by

Copyright:

Available Formats

Intact and Impaired Memory Functions in Autism

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Intact and Impaired Memory Functions in Autism

Uploaded by

Copyright:

Available Formats

Intact and Impaired Memory Functions in

Loisa Bennetto and Bruee F. Pennington

WISC-R: Informa-tion, Vocabulary, Similari- The autistic group comprised 19 indi-

You might also like