impairment in shifting attention in autistic and cerebellar patients€¦ · impairment in shifting...

Behavioral Neuroscience1994, Vol. 108, No. 5, 848-865 Copyright 1994 by the American Psychological Association, Inc.

" 073.v?044/94/$:U)0

Impairment in Shifting Attention in Autistic and Cerebellar Patients

Eric Courchesne, Jeanne Townsend, Natacha A. Akshoomoff, Osamu Saitoh,Rachel Yeung-Courchesne, Alan J. Lincoln, Hector E. James,

Richard H. Haas, Laura Schreibman, and Lily Lau

MRI and autopsy evidence of early maldevelopment of cerebellar vermis and hemispheres inautism raise the question of how cerebellar maldevelopment contributes to the cognitive andsocial deficits characteristic of autism. Compared with normal controls, autistic patients andpatients with acquired cerebellar lesions were similarly impaired in a task requiring rapid andaccurate shifts of attention between auditory and visual stimuli. Neurophysiologic and behavioralevidence rules out motor dysfunction as the cause of this deficit. These findings are consistent withthe proposal that in autism cerebellar maldevelopment may contribute to an inability to executerapid attention shifts, which in turn undermines social and cognitive development, and also withthe proposal that the human cerebellum is involved in the coordination of rapid attention shifts in afashion analogous to its role in the coordination of movement.

The concept that the cerebellum might influence highermental operations in humans is not a new concept but one thatbegan to emerge more than 70 years ago. At that time,Beyerman (1917) observed that "cerebello-ataxic mental defi-ciency" (p. 651) occurred in some individuals who did not haveextracerebellar pathology. More than 40 years ago, Snider(1950, p. 219) anticipated recent theory and the presentresearch study by declaring that one should

adopt broader concepts of cerebellar functions [which] mustencompass cerebellar influences on the sensory and motor centersof the cerebrum as well as related influences on the diencephalic,mesencephalic, and medullary centers.... The cerebellum standsout as "the great modulator of neurologic function," and newhorizons of cerebellar action are introduced into neurology andpsychiatry [italics added].

Although the narrower concept that the cerebellum isprimarily linked to motor control and motor learning continuesto dominate contemporary views on the role of the cerebellum

Eric Courchesne and Jeanne Townsend, Neuropsychology ResearchLaboratory, Children's Hospital, San Diego, California, and Neurosci-ences Department, University of California, San Diego; Natacha A.Akshoomoff and Laura Schreibman, Psychology Department, Univer-sity of California, San Diego; Osamu Saitoh, Neuropsychology Re-search Laboratory, Children's Hospital, San Diego, California, andNational Center Hospital for Mental, Nervous, and Muscular Disor-ders, National Center of Neurology and Psychiatry, Tokyo, Japan;Rachel Yeung-Courchesne and Lily Lau, Neuropsychology ResearchLaboratory, Children's Hospital, San Diego, California; Alan J.Lincoln, Neuropsychology Research Laboratory, Children's Hospital,San Diego, California, and California School of Professional Psychol-ogy, San Diego; Hector E. James, Department of Surgery, Universityof California, San Diego; Richard H. Haas, Neurosciences Depart-ment, University of California, San Diego.

This research was supported by National Institute of Mental HealthGrant 1-RO1-MH-36840 and National Institute of Neurological andCommunicative Disorders and Stroke Grant 5 RO1-NS-19855.

Correspondence concerning this article should be addressed to EricCourchesne, Neuropsychology Research Laboratory, Children's Hos-pital Research Center, 3020 Children's Way, San Diego, California92123. Electronic mail may be sent to [email protected].

(Ghez, 1991; King, 1987; Llinas & Sotelo, 1992), Snider's(1950, 1967) historically original position has continued to beespoused by many who study human neurocognition. Forexample, Hamilton, Frick, Takahashi, and Hopping (1983)reported on 3 patients who had initially been diagnosed asapparently having psychiatric disorders; on closer examina-tion, Hamilton et al. found that they had cerebellar pathologyand memory loss, labile mood, poor concentration, impairedabstraction, and intolerance of environmental stimulation.Following Snider's view, Hamilton et al. discussed the possibil-ity that these symptoms may have been the result of thecerebellar pathology because the cerebellum has widespreadconnections with numerous nonmotor systems. Botez, Gravel,Attig, and Vezina (1985) reported on 1 epileptic patient whoexhibited visuospatial logical organizational difficulties duringphenytoin intoxication that improved after treatment of thiscondition. They raised the question of whether the cerebellumvia its frontal and parietal projections might play a role in"cognitive thought" (p. 1152). Leiner, Leiner, and Dow (1986,1989) published a theory linking the neocerebellum to theability to perform mental operations, most particularly lan-guage operations. A serendipitous finding from a recentpositron emission tomography (PET) study demonstrated thatmetabolic activation of regions within the right neocerebellumdoes indeed occur in normal subjects who are successfullyperforming certain language tasks (Petersen, Fox, Posner,Mintun, & Raichle, 1989). Fiez, Petersen, Cheney, and Raichle(1992) showed that damage to the cerebellum can significantlyhinder the performance of such language tasks.

Courchesne and colleagues, in a series of lectures andpapers over the last decade, have theorized that the neocerebel-lum may play a role in the coordination of attention andarousal systems (as well as a number of other nonmotorsystems to which it has reciprocal connections) that is analo-gous to the role it plays in motor control (Courchesne, 1985,1987, 1989a, 1989b; Courchesne, Townsend, et al., 1994;Courchesne, Yeung-Courchesne, Press, Hesselink, & Jerni-gan, 1988). They also proposed that damage to the neocerebel-lum could hinder the acquisition of normal intellectual and

ATTENTION IN AUTISTIC AND CEREBELLAR PATIENTS 849

social skills. Like the Hamilton et al. (1983) and Botez et al.(1985) conclusions, these conclusions were initially derivedfrom observations of patients, but in this case, autistic pa-tients. In the very first description of an autistic patient, it wasnoted that arousal, attention, and sensory responsiveness wereseverely impaired (Kanner, 1943), and in the first neurobio-logic theory of autism, disturbances in the functioning of thereticular activating system (RAS) were hypothesized to under-lie the characteristic social and cognitive deficits in thisdisorder (Rimland, 1964). The first evidence of neuropathol-ogy in autism was Purkinje neuron loss throughout the cerebel-lum of a single patient (Williams, Mauser, Purpura, DeLong, &Swisher, 1980); Purkinje neurons are key to cerebellar func-tion. No other pathology was observed in that patient. Thequestion, then, was whether this cerebellar finding in a singleautopsy case was anomalous and of no significance in explain-ing the attention, arousal, sensory, and social symptomatologyin autism.

The cerebellum has been shown in animals to have physio-anatomical connections with the RAS and attentional systems(Itoh & Mizuno, 1979; Kitano, Ishida, Ishikawa, & Murayama,1976; Moruzzi & Magoun, 1949; Nieuwenhuys, Voogd, & vanHuijzen, 1988; Sasaki, Jinnai, Gemba, Hashimoto, & Mizuno,1979; Sasaki, Matsuda, Kawaguchi, & Mizuno, 1972; Schmah-mann & Pandya, 1989; Snider, 1950; Steriade & Stoupel, 1960;Vilensky & Van Hoesen, 1981), and to have a modulatoryeffect on sensory responsiveness (Crispino & Bullock, 1984).The cerebellum also has physioanatomical connections withbrainstem, thalamic, and parietal systems implicated in shift-ing attention in studies of humans with focal lesions (Posner &Petersen, 1990; Posner, Walker, Freidrich, & Rafal, 1984;Rafal & Posner, 1987; Rafal, Posner, Freidman, Inhoff, &Bernstein, 1988). Therefore, substantial Purkinje neuron loss

throughout the cerebellum could very well be of great signifi-cance in explaining attention, arousal, and sensory dysfunction inautism.

We hypothesized, therefore, (a) that the cerebellum mightplay a role in the coordination of attention in a fashionanalogous to the role it plays in motor control and (b) that inautism, cerebellum maldevelopment is a consistent featurethat renders the child unable to adjust his or her mental focusof attention to follow the rapidly changing verbal, gestural,postural, tactile, and facial cues that signal changes in a streamof social information (Courchesne, 1985,1987,1989a, 1989b,1991; Courchesne et al., 1988; Courchesne, Chisum, &Townsend, in press; Courchesne, Townsend, et al., 1994). Suchcues signal the normal child to move his or her "spotlight ofattention" from one source of information (e.g., auditory) toanother (e.g., visual). This process involves disengaging atten-tion from one source and then moving and reengaging it onanother (i.e., inhibition of one source and enhancement ofanother; Posner et al., 1984; Posner & Petersen, 1990). Toselectively adjust the focus of attention, the nervous systemmust quickly and accurately alter the pattern of neuralresponsiveness to sensory signals—from an enhanced neuralresponse to certain stimuli (e.g., vocalizations) to an enhancedresponse to other stimuli (e.g., gestures) and from inhibitedneural response to some stimuli to inhibited response toothers.

Our earlier prediction that cerebellar pathology would be aconsistent feature of infantile autism has now been verified by16 quantitative magnetic resonance (MR) and autopsy reportsfrom nine independent research groups involving 240 autisticcases (see Table 1). Two distinctly different forms of cerebellarpathology have been found (Courchesne, Saitoh, et al., 1994;Courchesne, Townsend, & Saitoh, 1994). One form is hypopla-

Table 1Quantitative Magnetic Resonance (MR) and Autopsy Studies of the Cerebellum in Infantile Autism

Studies finding abnormalities Studies finding no abnormalities

Researchers Study Researchers Study

Williams, Hauser, Purpura,DeLong, & Swisher (1980) Autopsy

Bauman & Kemper (1985) AutopsyBauman & Kemper (1986) AutopsyRitvo et ai., 1986 AutopsyGaffney, Tsai, Kuperman & Min-

chin (1987) MRCourchesne, Yeung-Courchesne,

Press, Hesselink & Jernigan(1988) MR

Murakami, Courchesne, Press,Young-Courchesne, & Hes-selink (1989) MR

Bauman & Kemper (1990) AutopsyCiesielski, Allen, et al. (1990) MRArm, Bauman, & Kemper (1991) AutopsyBauman (1991) AutopsyPiven et al. (1992) MRa

Kleiman, Neff, & Rosman (1992) MRa

Courchesne, Saitoh, et al. (1994) MRSaitoh, Courchesne, Egaas, Linc-

oln, & Schreibman (in press) MRHashimoto et al. (in press) MR

Holttum, Minshew, Sanders, &Philips (1992)

MR

Garberetal. (1989)Garber & Ritvo (1992)

MRMR

aEvidence based on reanalyses of study data by Courchesne, Townsend, & Saitoh (1994).

850 COURCHESNE ET AL.

sia of the cerebellar vermis and hemispheres, which is presentin the great majority (roughly 89%) of autistic patients. At themacroscopic level, this is seen as a reduction in size; forexample, the midsagittal area of vermian lobules VI-VII wasshown to be 18% smaller in 78 autistic patients from fivedifferent studies than in 91 normal, healthy volunteers (seeFigure 1), and two other quantitative MR studies have re-ported evidence of a reduction in the size of the cerebellarhemispheres (Gaffney, Tsai, Kuperman, & Minchin, 1987;Murakami, Courchesne, Press, Yeung-Courchesne, & Hes-selink, 1989). This macroscopic size reduction is apparentlydue to a reduced number of cerebellar neurons. For example,recent quantitative microscopic analyses place the most severePurkinje neuron loss in posterior vermian lobules VI-VII andVIII-X (50%-60%) and posterior hemispheres (42%-56%)and the least in vermian lobules I-V (Arin et al., 1991; D. M.Arin, personal communication, February 1993). Onset ofcerebellar hypoplasia may be as early as the first year ofpostnatal life (Hashimoto, et al., in press). Substantial concor-dance exists among microscopic autopsy and macroscopic MR

studies as to the location of the hypoplasia (for reviews seeCourchesne, 1991; Courchesne, Townsend, et al., 1994). Asecond form of cerebellar pathology only recently identified ishyperplasia of the cerebellar vermis, in which vermian lobulesVI-VII were 31% larger than normal (see Figure 1); thispathology is present in roughly 11% of autistic patientsexamined (Courchesne, Saitoh, et al., 1994; Courchesne,Townsend, & Saitoh, 1994). Such pathology has not yet beenseen at autopsy. No abnormalities at autopsy have been foundin the brainstem, diencephalon, or cerebral cortex (Bauman,1991; Coleman, Romano, Lapham, & Simon, 1985; Williams etal., 1980), but new MR studies have reported evidence ofparietal volume loss in 43% of autistic patients examined(Courchesne, Press, & Yeung-Courchesne, 1993). Althoughreductions in neuron size in CA4 of the hippocampus andincreased cell packing density in several limbic structures havebeen reported in 4 cases by one laboratory (Bauman, 1991;Raymond, Bauman, & Kemper, 1989), no such autopsy abnor-malities were discerned by another laboratory (Williams et al.,1980), and hippocampal size was normal in 50 autistic patients

as_>

NORMALVermian Lobules VI-VII

AUTISMVermian Lobules VI-VII

D Courcnesnet

O Courchesne 2

X

A Raz

120-2 5 -1.5 -0.5 0.5 1.5 2.5 5 -1.5 -0.5 0.5 1.5 2.5

Expected Normal Value(z-scores)

Expected Normal Value(z-scores)

Figure 1. The midsagittal area of vermal lobules from each of 91 normal subjects (a) and 78 autisticpatients (b) from separate MR studies of the vermis (Courchesne, Yeung-Courchesne, Press, Hesselink,& Jernigan, 1988; Courchesne, Saitoh, et al., 1994; Kleiman, Neff, & Rosman, 1992; Piven et al., 1992;Raz, Torres, Spencer, White, & Acker, 1992). Graphs show the degree to which the size distributions ofvermian lobules VI-VII approximate a unimodal Gaussian curve of normal distribution. A perfectlynormal unimodal distribution of data forms a straight line when plotted against expected values on aGaussian curve. Data from autistic patients (b) show that the distribution of lobules VI-VII size deviatessignificantly (p < .001) from a normal distribution. Instead, the distribution is bimodal. Specifically, thereare two separate clusters of autistic patients: one with smaller than normal lobules VI-VII (88% of allpatients) and the second with enlarged lobules VI-VII (12% of all patients). Data in (a) from normalsubjects (Courchesne et al., 1988; Courchesne, Saitoh, et al., 1994; Raz et al., 1992) closely approximated astraight line for vermian lobules VI-VII. The line from these normal data is graphed for comparisonagainst the autism data (b). The arrows point to individual autistic and normal subjects who wererandomly selected to participate in the present study of shifting attention. Courchesne 1 = Courchesne etal., 1988; Courchesne 2 = Courchsene, Saitoh, et al., 1994.


as quantified by MR techniques (Saitoh, Courchesne, Egaas,Lincoln, & Schreibman, in press).

To test our theory that the cerebellum contributes to thedynamic, moment-to-moment control of the mental focus ofattention, in a preliminary study (Courchesne, Akshoomoff, &Ciesielski, 1990) we examined the ability of 10 adult autisticpatients to alternate attentional focus between auditory andvisual stimulation. The task was a distillation of the socialworld, in which cues presented at unpredictable time intervalsdirect patients to voluntarily initiate shifts of attention (seeFigure 2). We found that these autistic patients were unable toaccurately alternate attentional focus if required to do sorapidly (within < 2.5 s) but were accurate when given muchmore time (2.5 to 30 s) to shift their attention between auditoryand visual stimulation. They had no difficulty maintaining afixed focus of attention on one modality for many seconds. Toverify that this shifting attention deficit was due to cerebellarpathology, we tested patients with acquired cerebellar lesionsand other patients with acquired cerebral lesions in similarshift and focus attention tasks (Akshoomoff & Courchesne,1992, 1994). The cerebellar patients were significantly im-paired in their ability to make rapid and accurate voluntaryshifts of attention. Whereas cerebellar patients required sev-eral seconds to successfully shift attention, normal controlsubjects in our study required less than 1 s. This is consistentwith previous literature showing that normal adults mayrequire as little as 300 ms to shift attention (Sperling &Reeves, 1980; Weichselgartner & Sperling, 1987).

In the present study, a new group of autistic patientsperformed the identical shift attention task given to ourpatients with acquired cerebellar or cerebral lesions, and theirperformance was statistically compared with the published

data on these two groups of lesioned patients as well as withchronologic- and mental-age-matched normal volunteer con-trols. Because disengaging of attention has been proposed tobe impaired by parietal damage (Posner et al., 1984; Posner &Petersen, 1990) and some autistic patients have been reportedto have parietal damage (Courchesne, Press, & Yeung-Courchesne, 1993; Egaas, Courchesne, & Saitoh, 1994), weincluded among the autistic patients in the present experimentthose with and those without MR evidence of volume loss inposterior cerebral regions. In addition, we recorded event-related brain potentials (ERPs) from several autistic subjectsin order to assay attention-related brain responses to stimulithat immediately followed cues to shift attention. We nowreport the results from these tests.

Method

Study Groups

Patients With Autism

Subjects were 8 patients (mean age = 13.9 years) diagnosed withinfantile autism as denned by the revised third edition of the Diagnosticand Statistical Manual of Mental Disorders (DSM-IH-R; AmericanPsychiatric Association, 1987; see Table 2). All patients also passed thecriteria for infantile autism denned by the Autism Diagnostic Inter-view (LeCouteur, et al., 1989) and the Autism Diagnostic ObservationSchedule (Lord et al., 1989). In these patients, autism was notcomplicated by severe mental retardation, cerebral palsy, epilepsy, orother neurologic disease. DNA analyses demonstrated that all 8 werenegative for fragile-X. Table 3 lists additional patient characteristics.

In addition to these 8, another autistic patient (Wechsler AdultIntelligence Scale—Revised [WAIS-R] Verbal IQ = 74; Performance

SHIFT FOCUS ATTENTION

HIT MISS HIT FA

g RED1 . 1ilo

IGN

red g g. . 1 1 , 1 ,i i i ilo lo lo HI

tRED g g g REDi. . i.i i ,iM i ilo lo hi lo

t9 red gI . I iI I

lo lo

HIT

HIT HIT

f f9 RED RED 9 9I I I I I

1lo

II I Ilo lo lo hi

IGN

HIT HIT

f f

HIT

f

RED g g 9RED9 RED 91 1 1 1 1 1 I II I I 1

lo lo hi loIGN

i ilo lo

disengage engage

engage

Figure 2. Schematic drawing of the shift attention (A) and focus attention (B) tasks. Visual stimuli werered and green (g) flashes; auditory stimuli were 2-kHz (hi) and 1-kHz (lo) tone pips. HIT = correctlydetected target; MISS = a failure to respond to a target; FA = an erroneous response to a rare stimulusthat was in a modality to be ignored; IGN = a rare stimulus that was correctly ignored. In the example ofthe shift experiment, the subject pressed a button (arrow) to the first rare visual target stimulus (RED).This served as a cue to shift attention to the auditory stimuli, ignore (IGN) the rare visual stimuli, andrespond to the next auditory target (HI). The auditory target in turn served as a cue to shift attention backto the visual stimuli.


Table 2Matching Characteristics of Normal and Autistic Subjects

Subjectgroup

NormalAdolescent

6M, 2FChild

7M,3FAutistic adolescent

6M,2F

Age(years M ± SD)

13.8 ± 0.8

8.6 ± 1.7

13.9 ± 1.6

Block design(M ± SD)

11.0 ± 1.7

10.5 ± 2.8

11.8 ±3.9

Object assembly(M + SD)

10.5 ± 2.6

12.0 ± 3.2

12.1 ± 3.1

Note. The Wechsler Intelligence Scale for Children—Revised(WISC-R) was administered to children and adolescents. AverageWechsler subtest scores for the normal population = 10.0 ± 3.0.M = male; F = female.

IQ = 92) was tested in the shift and focus attention tasks describedfurther on. He was tested on six occasions, each separated by 1-2-weekintervals. Because he was an adult (26 years of age), his data arepresented separately and were not included in the statistical analyseswith the 8 younger autistic patients.

Measures of the cerebellum. All 8 had participated in an MR study(Courchesne, Saitoh, et al., 1994) which demonstrated that there aretwo distinctly different types of vermal pathology in autism: hypoplasiaof cerebellar vermian lobules VI-VII, which occurs in about 89% ofautistic cases, and hyperplasia of vermian lobules VI-VII, whichoccurs in about 11 % of autistic cases. Their MR data were also used ina meta-analysis of recent findings from other research laboratorieswhich demonstrated comparable findings of vermal pathology acrossfour research laboratories involving 78 total autistic patients (seeFigure 1; Courchesne, Saitoh, et al., 1994). The 8 autistic subjects wereselected strictly on the basis of age and availability for participation inthe present study. Seven of the 8 had vermal hypoplasia and 1 hadvermal hyperplasia (see values indicated by arrows in Figure IB).Vermian lobules VI-VII in the 8 patients differed significantly fromthose of the normal controls in the present study by 1.4 standard

deviations. Figure IB and Figure 2 show that these 8 are a representa-tive sample of the larger population of autistic patients studied byseveral different laboratories. In the present study, these vermismeasures serve as a convenient and reliable index of cerebellarabnormality in these autistic patients, but it should be reiterated herethat autopsy and MR studies show that the cerebellar hemispheres aswell as the vermis are abnormal in autism (Gaffney, Tsai, et al., 1987;Murakami et al., 1989).

Measures of posterior corpus callosum. A new MR study hasreported qualitative evidence of parietal volume loss in 43% of autisticpatients (Courchesne, Press, & Yeung-Courchesne, 1993). Followingup on this qualitative finding, another new quantitative and statisticalMR study found the posterior segments of the corpus callosum, whichinclude axons from parietal cortex, to be significantly reduced incross-sectional size in 41 autistic patients compared with 41 age- andgender-matched normal healthy volunteer controls (Egaas et al.,1994); frontal segments were nonsignificantly different between autis-tic and normal groups.

In the present study, this quantitative measure of the posteriorsegment of the corpus callosum was used as an objective MR index ofposterior cerebral abnormality. For 4 of the 8 autistic patients in thepresent study, the cross-sectional area of the posterior segment of thecorpus callosum was -2.6 standard deviations below the normal meanreported by Egaas et al. (1994); also, each of these 4 autistic patientshad areas that were smaller than even the autistic mean reported byEgaas et al. For the other 4 of the 8 autistic patients in the presentstudy, the cross-sectional area of the posterior segment of the corpuscallosum was —0.1 standard deviation below the normal mean re-ported by Egaas et al. (1994).

Normal Controls

Eighteen normal volunteer controls were tested (see Tables 2 and3); they had no history of substance abuse, special education, majormedical illness, psychiatric illness, or developmental disorder. MRscans of the cerebellar vermis were obtained from 7 of the normalcontrols. Figure 1A (see values indicated by arrows) shows that these 7normal subjects are a representative sample of the larger population ofnormal volunteer subjects studied by several different laboratories.

Table 3Additional Characteristics of Subjects

WISC-R „,. . .ClinicalSubjectgroup

NormalAdolescentChild

Autistic

Cerebellar lesions

PPVT-R(M ± SD)

118 ± 11.5111 ± 12.154 ± 13.4

97 ± 14.8

Verbal IQ(M ± SD)

114± 10.5108 ± 6.659 ± 13.7

97 ± 15.5

Performance IQ seizure(M ± SD) activity

107 ± 8.6 —111 ± 12.2 —89 ± 10.4 7/8: negative

96 ± 23.2 —

Medication

7/8:none;1/8: Trilafon and

Cogentin

Note. The Peabody Picture Vocabulary Test—Revised (PPVT-R) is a test of receptive language ability.Average PPVT-R and Wechsler Intelligence Scale for Children—Revised (WISC-R) IQ scores in thenormal population = 100 ± 15.0. Note that the large discrepancy between Verbal and Performance IQscores in this sample of autistic subjects was principally due to two very low Verbal subtest scores(Vocabulary and Comprehension: 2.8 and 1.4, respectively) and to two relatively normal Performancesubtest scores (Block Design and Object Assembly: 11.8 and 12.1, respectively). Such an extremediscrepancy has been suggested to be typical of patients with infantile autism (Bartak, Rutter, & Cox,1975; Lincoln, Allen, & Piacentini, 1994; Lincoln, Courchesne, Kilman, Elmasian, & Allen, 1988; Ohta,1987) and stands in contrast to the normally expected pattern of comparable scores across all subtests(e.g., in the normal adolescents, these four subtest scores were 11.6, 11.9, 11.0, and 10.5, respectively).Empty cells indicate that medical history was negative for abnormality or use of medication at time oftesting.


Mental-age-matched normal controls. Ten normal children (meanage = 8.6 years) were matched with the 8 autistic patients on mentalage, which was based on the Wechsler Intelligence Scale for Children—Revised (WISC-R) Verbal IQ; this IQ index reflects areas of verbaland cognitive weaknesses typical of autism (see Table 2; Bartak,Rutter, & Cox, 1975; Lincoln, Allen, & Piacentini, 1994; Lincoln,Courchesne, Kilman, Elmasian, & Allen, 1988; Ohta, 1987). Specifi-cally, the mean mental age of this normal control group was 9.3 years(chronologic age x VIQ/100), and that of the autistic adolescents was8.2 years. The data from these 10 controls are from Akshoomoff andCourchesne (1992).

Chronologic-age-matched normal controls. The 8 older normalcontrols (mean age = 13.8 years) were matched with 8 autistic patientson chronological age, gender, and performance on selected subtests ofthe WISC-R that reflect nonverbal cognitive strengths in autism (seeTable 2; Bartak et al., 1975; Lincoln et al., 1988,1994; Ohta, 1987).

Comparison Group: Patients With Acquired CerebellarLesions

The data from these patients are from Akshoomoff and Courchesne(1992). Neocerebellar damage was substantial in 5 of these 6 patients.Five of the 6 patients were children who had undergone surgicalresection of cerebellar astrocytomas, and, as in Akshoomoff andCourchesne (1992), they were age- and IQ-matched to the 10 normalchildren described above. The mean age at the time of diagnosis was5.2 years (SD = 1.2), and the mean age at the time of testing was 8.6years (SD = 1.8). Neurological symptoms associated with cerebellardamage had improved since the time of surgery but persisted in allpatients (e.g., dysmetria and gait difficulty). Two received focalirradiation of the posterior fossa, but 3 did not; none of the patientsreceived chemotherapy. Three patients had unilateral neocerebellarhemisphere lesions (including the ventral dentate nucleus in 1 case),and 1 patient had damaged superior cerebellar peduncles bilaterally,which thus eliminated the principal output pathways of the neocerebel-lum. The 5th patient had a small focal unilateral lesion and noindication of extensive damage to neocerebellar structures or inputand output pathways. The 6th patient was a 21-year-old man with anormal developmental history who exhibited signs of an idiopathiccerebellar degenerative disorder at 16 years of age (Akshoomoff,Courchesne, Press, & Iraqui, 1992). He had extensive parenchymalvolume loss in the cerebellum but relatively little cerebral corticalatrophy. Prior to his disability, this patient was an honors student inhigh school.

Protocol for Neurobehavioral Testing

Each subject participated in two experimental tasks—focus atten-tion and shift attention—in which visual and auditory stimuli wererandomly intermixed and rapidly presented. In each attention task,subjects were required to detect and press a button in response tospecified target stimuli (see Figure 2). Responses were continuouslydigitized, and computer algorithms were used to score behavioralperformance. Both the baseline focus task and the shift task requiredsubjects to make equally rapid detections and responses to successivetargets, but compared with the baseline task, the shift task additionallyrequired subjects to shift attention between sensory modalities tomake such successive correct detections.

We took several steps to ensure that test performance was notaffected by hand, head, and eye position and movements. Each subjectalways had the button in hand and placed the thumb of the preferredhand on the top of the button; to respond, the subject merely had tosqueeze down. Sounds were presented over headphones, so eachsubject did not have to move head or eyes to optimize sound detection.

Each subject was trained to maintain eye fixation on the video monitor,where visual stimuli were presented. Experimenters continuouslymonitored compliance during testing. In addition, compliance waselectrophysiologically monitored by placing silver-silver chloride elec-trodes on one outer canthus and above and below one eye in 4 autisticpatients and 2 cerebellar patients. These electrophysiologic recordingsshowed that the patients followed our instructions well and kept theireyes fixated on the location where the visual stimuli were presented.

Visual stimuli were red and green squares subtending 1.2° of visualangle and presented on a video monitor; auditory stimuli were 1- and2-kHz binaural tones presented over headphones. All visual andauditory stimuli were 50 ms in duration, and interstimulus intervalsvaried between 450 and 1,450 ms. Twenty-five percent of all stimuliwere target stimuli to which subjects were required to respond bypressing a button. Prior to beginning each experimental task, eachsubject was required to achieve at least a 75% overall correct responserate in practice trials; and once the experimental task was under way,performance data were included only if the overall correct responserate in any set sequence of 80 stimuli was at least 50%. A correctlydetected target was scored as a hit if the response occurred between200 and 1,400 ms after stimulus onset, and failure to respond to atarget during this time window was a miss. An erroneous response to anontarget stimulus was scored as & false alarm.

The focus attention task tested the subject's ability to continuouslymaintain a focus of attention and to detect rare target stimuli in onesensory modality (e.g., visual) while ignoring all stimuli in the othermodality (e.g., auditory; see Figure 2B). This served as a control, orbaseline, for the shift attention task. The focus attention task consistedof two conditions: a visual focus attention condition and an auditoryfocus attention condition. A minimum of five sets of 80 stimuli eachwere presented in each condition. To assess subjects' ability to rapidlymake two detections and responses in a row while continuouslymaintaining a focus of attention, we analyzed the hit and false alarmdata in the focus attention task at five intervals of elapsed timeimmediately following the onset of the last correctly detected target:0.4-2.5 s, 2.5-4.5 s, 4.5-6.5 s, 6.5-10.5 s, and 10.5-30.0 s.

In the shift attention task (see Figure 2A), subjects were presented aminimum of 10 stimulus sequences identical to those used in the focusattention task. The shift task, however, was different: Subjects wererequired to alternate attention between visual and auditory stimuli assignaled by the appearance of the rare target stimuli (see Figure 2A).Correct detection of a target (hit) in the attended modality served tosignal subjects to disengage their attention to stimuli in the currentmodality and to move and reengage their focus of attention as rapidlyas possible to stimuli in the other modality in order to detect the verynext target appearing in that modality (see Figure 2A). A failure tofully disengage and inhibit attention to stimuli in the same modalityfollowing a signal to do so would register as a false alarm (i.e., when asubject pressed the button to a target in the unattended modality aftera signal to shift attention away from that modality; see Figure 2A). Afailure to rapidly move and reengage attention in the oppositemodality would register as a miss and would result in a reducedpercentage of correct target detections in the time period immediatelyfollowing a signal to shift attention (see Figure 2A). Therefore, toassess the time course of disengaging, moving, and reengaging atten-tion, we analyzed hit and false alarm data in the shift attention task atthe same five intervals of elapsed time immediately following the onsetof each correctly detected target, as we did for the focus attention task.

Post hoc focus attention experiment. Analyses of these focus andshift experiments indicated significant group differences only in theshift experiment, not in the focus experiment (see Results section).Therefore we conducted a post hoc experiment that further challengedthe capacity of the autistic patients and normal children to performunder focused attention conditions. We accomplished this by increas-ing the difficulty of the stimulus discrimination during the auditory


focus attention task, (With our software, fine adjustments werepossible with auditory frequency, but not color.) The question waswhether a significant increase in attentional and perceptual demandsin the focus attention task would reveal group differences analogous tothose seen in the shift task.

Whereas the easy focus attention task described previously promptedvery nearly no false alarm responses to nontarget stimuli in theattended modality, in the difficult discrimination attention experimentdescribed next, subjects produced some false alarm responses to suchstimuli. Therefore, we calculated the percentage of hits using acorrection described by Swets (1986a, 1986b)—namely, correctedpercentage of hits (H'c) equals percentage of hits minus percentage offalse alarms.

Five of the normal children and 7 of the autistic patients wereavailable for this retesting. Using a tone separation of 30 Hz betweentarget (1030 Hz) and nontarget (1000 Hz) sounds, we tested eachnormal child across five sets of 80 stimuli under the auditory focusattention condition. If a normal subject's H'c performance levelexceeded 70%, the tone separation was reduced to 20 Hz, and ifperformance continued to be better than 70%, then the tone separa-tion was reduced to 15 Hz. Data collected from the focus attention setswith the smallest tone separation that produced 70% or better H'cwere used in statistical comparisons with autistic patients. This 70%performance criterion was met at the 30-Hz separation for 1 normalsubject, at 20 Hz for 3 normal subjects, and at 15 Hz for 1 normalsubject.

Autistic patients were tested in the same way. The 70% H'cperformance criterion was met at the 30-Hz separation for 3 autisticpatients and at the 15-Hz separation for 4 others. One of these autisticpatients far exceeded the performance of all normal and other autisticpatients by achieving 98% H'c at the 15-Hz level; the best any normalsubject did at this separation was 73%.

In sum, thresholds established during the calibration phase werebased on overall accuracy (H'c). For the test phase, the dependentvariable of interest was the difference between groups in accuracy duringthe shortest time interval, given comparable overall performance. It isimportant to note that between the two groups overall accuracy andthe range of pitch differences were comparable.

Event-related brain potentials. ERPs were recorded during theexperiments for 3 of the autistic patients, 2 of the cerebellar patients,and the normal control subjects with the same method used in aprevious study (Ciesielski, Courchesne, & Elmasian, 1990). The ERPeffects from 1 cerebellar patient and 1 normal child were fromAkshoomoff and Courchesne (1992).

Statistical Analysis

There were no significant modality by group effects for any of thecomparisons presented next, so all performance data values presentedwere collapsed across modalities. Autistic patients were statisticallycompared (a) with adolescent normal subjects who matched them ongender, chronological age, and measures of nonverbal IQ and (b) withnormal children who matched them on measures of verbal IQ (in areasof relative cognitive weakness in autism; see Tables 2 and 3).Akshoomoff and Courchesne (1992) previously reported that thecerebellar patients had significantly fewer correct target detections inthe 0.4-2.5-s time zone in the shift attention experiment than did theirchronological- and mental-age-matched normal subjects.

Results

Mental-Age-Matched Comparisons

Comparison with normal but much younger, mental-age-rnatched children provided a strong test of whether autistic

patients have significant deficits in shifting attention, particu-larly because target detection performance among the normalchildren, autistic patients, and cerebellar patients was belowceiling on the shift attention task (see Figure 3).

Autistic and cerebellar patients performed similarly to thenormal children in the focus attention task but were signifi-cantly impaired in their ability to accurately and rapidlyrespond to targets in the shift attention task; repeated mea-sures analysis of variance using a Huynh-Feldt correctionfactor showed a significant Group x Experiment x Timeinteraction, F(8,84) = 2.29, p < .029.

Focus Attention Task

Percentage of hits, percentage of false alarms, and reactiontimes (RT). In the focus attention task, accuracy (as indexedby percentage of hits and percentage of false alarms) andspeed (as indexed by RT) in detecting targets as a function ofelapsed time since the last correct target detection were notsignificantly different across normal children, autistic patients,and cerebellar patients: For percentage of hits, percentage offalse alarms, and reaction times, the Group x Time compari-son yielded F(8,84) < 0.99,p a .45 (see Figure 4 and Table 4).

SHIFT ATTENTION

100-

90-

I80-

* 70-

60-

50-

I- Normal Children

I- Autistic Patients

It- Cerebellar Patients

0.4-2.5 2.5-4.5 4.5-6.5 6.5-10.5 10.5-30

TIME SINCE LAST CUE (MC)

0.4-2.5 2.5-4.5 4.5-6.5 6.5-10.5 10.5-30

TIME SINCE LAST CUE (MC)

Figure 3. Mean percentage of hits (top) and mean percentage of falsealarms (bottom) for normal children, autistic patients, and cerebellarpatients in the shift attention task. All data were analyzed at fiveintervals of time (in seconds) elapsed since the onset of everyimmediately preceding cue that had been correctly detected; datavalues were collapsed across visual and auditory modalities. With lessthan 2.5 s between visual and auditory cues in the shift attention task,autistic and cerebellar patients were significantly impaired at targetdetection relative to normal subjects.


FOCUS ATTEfmON

100-

90-

80-

70-

Normal Children

Autistic Patients

Cerebellar Patients

0,4-2.5 2.5-4.5 4.5-6.5 6.5-10.5 10.5-30

TIME SINCE LAST TARGET (SEC)

10-

0.4-2.5 2.5-4.5 4.5-6.5 6.5-10.5 10.5-30

TIME SINCE LAST TARGET (sec)

Figure 4. Mean percentage of hits (top) and mean percentage of falsealarms (bottom) for normal children, autistic patients, and cerebellarpatients in the baseline focus attention task. All data were analyzed atfive intervals of time (in seconds) elapsed since the onset of everyimmediately preceding target that had been correctly detected; datavalues were collapsed across visual and auditory modalities. In thebaseline focus tasks when autistic and cerebellar patients did not haveto shift attention, their performance was similar to that of normalsubjects even when little time had elapsed (0.4-2.5 s) between twotargets in the same modality.

Post hoc focus attention experiment. In this experiment inwhich attentional and perceptual demands were increasedover the first, easier focus attention task, the performances ofnormal subjects and autistic patients were not significantly

different from each other: For H'c and reaction times, eachGroup x Time comparison yielded F(3, 30) < 0.98, p 2: .41.H'c was 75.8 ± 5% (with 3.4% false alarms) for normals and77.4 ± 7% (with 4.6% false alarms) for autistic patients; RTwas 673 ± 76 ms for normals and 632 ± 81 ms for autisticpatients. In the shortest time interval (0.4-2.5 s), H'c was64.3% for normals and 72.7% for autistic patients (see Figure5); RT was 707 ms for normals and 628 ms for autistic patients.

Shift Attention Task

Percentage of hits. By contrast, in the shift attention task,accuracy (percentage of hits) in target detection as a functionof elapsed time since the last correctly detected target wassignificantly different across groups: Group x Time, F(8,84) = 2.88, p < .007. When 2.5 s or less had elapsed since thelast correct target detection, autistic and cerebellar patientscorrectly detected only 58.9% and 59.2%, respectively, of thetargets, compared with 78% correct detection by normalchildren (Bonferroni corrected pairwise t tests of autistic vs.normal children and of cerebellar patients vs. normal childrenwere significant sAp < .05; see Figures 3 and 6). When moretime (2.5-30.0 s) had elapsed, however, autistic (86%) andcerebellar patients (90%) were much more accurate and didnot perform significantly differently from the normal children(91%) or from their own performance levels in the focusattention task (Figure 4).

Figure 7 shows that for a single autistic subject who wasretested on six separate occasions, this impairment in theability to accurately and rapidly shift attention was persistentand profound despite repeated training and experience inthese attention tasks. In comparison, his target detectionaccuracy in the focus task was excellent regardless of elapsedtime since the last correct target detection (see Figure 8).

Reaction times. Reaction times for all subject groups as afunction of time interval and attention task are presented inTable 4. In the shift attention task, all subject groups had fasterreaction times with longer intervals between targets, F(4,84) = 17.39, p < .0001, and the children with cerebellar lesionshad longer reaction times overall than did the age-matchedcontrol subjects, F(l, 21) = 4.74, p < .05. In the focus

Table 4Mean Reaction Times to Targets in the Shift and Focus Attention Tasks

Time since last cue

Subjectgroup


AutisticCerebellar lesions


AutisticCerebellar lesions

0.4-2.5(M

494598622740

438560514632

±SD)

±37± 116±81± 190

±41± 104± 103+ 143

2.5-4.5(M

450505477567

411497505562

±SD)Shift task

±54±87±68± 56

Focus task

± 19±62±78±87

4.5-6.5(M

464512481582

424570474586

+

+H-

-t-

±

+

-t-

±

±

SD)

57975994

32121110111

(s)6.5-10.5

(M±

438 ±530 ±464 ±600 ±

413 ±520 ±469 ±566 ±

SD)

381104995

298317869

10.5-30(M

424541490619

423525477622

±SD)

±30±94±62± 94

±35±65±86± 122


FOCUS ATTENTION - DIFFICULT DISCRIMINATION

100

90-

80-

70-

60-

50-

400,4-2.5 2.S-4.5 4.5-6.5 6.5 +

TIME SINCE LAST TARGET (sec)

Normal Children

Autistic Patients

100

90

80 -I

70-

60-

50-

40

......... -i

0.4-2.5 2.54.5 4.5-6.5 6.5 +TIME SINCE LAST TARGET (sec)

i

60

50

40-

2 20-|88

10-

0,4-2.5 2.5-4.5 4.5-6.5 6.5 +TIME SINCE LAST TARGET (sec)

Figure 5. Results from the post hoc experiment, a focus attention task with more difficult discrimination.Mean percentage of hits (bottom left), mean percentage of false alarms (bottom right), and H'c (correctedpercentage of hits; top) are shown for normal children and autistic patients as a function of time betweencorrect target detections.

attention task there was a tendency for reaction times in allgroups to be longer with shorter intervals between targets, F(4,84) = 2.07, p < .092, but there was no significant difference inreaction times between cerebellar lesion patients and age-matched control children, F(l, 21) = 1.82, p < .20. Autisticpatients had shorter reaction times than either the cerebellarlesion patients or the normal controls, but because the autisticpatients were also chronologically older, these statistical com-parisons were not performed. To further examine thesefindings, we performed a post hoc analysis on reaction timedata from the shortest time interval (0.4-2.5 s) and the nextlongest one (2.5-4.5 s) in the focus and shift tasks. The resultsare graphed in Figure 9. Compared with normal controlsubjects, autistic and cerebellar lesion patients showed asignificantly greater difference in reaction times at the twotime intervals in the two tasks, F(l, 21) = 4.71,p = .042. In the

focus attention task, all three groups were only slightly slower(by 2% to 13%) in correctly responding to targets in theshortest time interval relative to the longer one. However, inthe shift attention task, autistic and cerebellar patients weresubstantially slower (by 30% and 31%) when only 2.5 s or lesshad elapsed compared with when there was more time to shiftattention.

Percentage of false alarms. For each group, the false alarmrate was higher in the shift task than in the focus task, F(l,21) = 19.96, p < .0003. In the shift task, the percentages offalse alarms for autistic, cerebellar, and normal children were15.6%, 4.3%, and 8.7%, respectively; and in the focus task,they were 1.5%, 0.5%, and 0.4%, respectively. There were nosignificant effects of elapsed time since the last correctly detectedtarget, F(4,84) = I.l4,p > .34, and no interaction of subject group,time, and experimental task, F(8,84) = 0.91,p > .50.

ATTENTION IN AUTISTIC AND CEREBELLAR PATIENTS

%Hits (.4 - 2.5 sec)

857

«

1 "o

o* -10Q(0

-15

AA

2

Mean

-2

±1"

SHIFT ATTENTION

(3)1UU-

90-

80-

70-

-

50-

40-

30-

20-

10-

0-

P;,̂ ,/̂ !̂̂ %

ff^x^7*n /d' /

//*

0.4-1.2 1.2-2.5 2.5-4.5 4.5-6.5 6.5-10.5 10.5-30TIME SINCE LAST CUE (sac)

-V- 9-3-91 -O-- 9-24-91

•••©•• 9-13-91 -A- 10-8-91

-O~ 9-17-91 •••*•• 10-28-91

Figure 7. Shift attention task data for a single autistic patient who wastested on 6 separate days (dates of testing are given in inset). Therewas a persistent and profound impairment in his ability to accuratelyand rapidly shift attention despite repeated training and experience.Data values were collapsed across visual and auditory modalities. Thenumbers in parentheses indicate the number of sessions in which thispatient achieved a 100% hit rate in a specified time interval.

• Normal

• Autistic

A Cerebellar

x Cerebral

Figure 6. With less than 2.5 s to shift attention from one modality tothe other to detect two successive targets, autistic and cerebellarindividuals performed worse than the great majority of the 18individual normal controls and 3 cerebral lesioned patients. Meanequals the mean hit rate in this shortest time interval for each of thetwo normal age groups. The hit rate for each individual is plotted instandard deviations from the normal group mean for that person'schronological age. The 1 cerebellar individual who fell above normalwas the one who had the small focal lesion largely sparing cerebellarcortex and cerebellar input and output pathways (see Methodsection).

Parietal Abnormal Versus Normal Comparisons

With short intervals between targets, the 4 autistic patientswith MR evidence of parietal abnormality, indexed by reducedposterior corpus callosum, performed the shift attention taskwith slightly, but nonsignificantly, better speed and accuracythan the 4 autistic patients without MR evidence of parietalabnormality: t(6) = 0.58, p > .59 and / (6) = 0.64, p > .54.When only 2.5 s or less had elapsed since the last correct targetdetection, the former subgroup correctly detected only 62.1%(RT = 604 ms) of the targets compared with 55.9% (RT = 639rns) correct detection by the latter subgroup. The autisticsubgroup with a reduced posterior corpus callosum alsoshowed slightly less difference in response time at shortintervals in shift tasks compared with focus tasks, responding

slower in shift tasks by 26%; the autistic subgroup without anormal posterior corpus callosum was slower by 34%.

In addition, there was no systematic difference in false alarmrates for these patients as a function of parietal abnormality.Among the 4 autistic patients with MR evidence of parietalabnormality, 2 had elevated levels of false alarms (25% and

FOCUS ATTENTION

(S) (5) (3) (3) (8) (5)100-

90-

80

70-

j 50-I

' 40-

30-

20-

10-0

.-• "13

0.4-1.2 1.2-2.5 2.5-4.5 4.5-6.5 6.5-10.5 10.5-30TIME SINCE LAST TARGET (sec)

»"*-, 9-3.91 ...p... 9-24-91

~A- 9-13-91 ~Q~ 10-8-91

••<>• 9-17-91 ...?- 10-22-91

Figure 8. Focus attention task data for a single autistic patient whowas tested on 6 separate days (dates of testing are given in inset). Inthe focus attention task, in contrast to the shift attention task, histarget detection accuracy was excellent regardless of time elapsedsince the last correct target detection. Data values were collapsedacross visual and auditory modalities. The numbers in parenthesesindicate the number of sessions in which this patient achieved a 100%hit rate in a specified time interval.


Focus RT: 0.4-2.5 vs 2.5-4.5 sec intervals

Shift RT: 0.4-2.5 vs 2.5-4.5 sec intervals

Normal Normal Autistic CerebellarAdolescent Children Patients Patients

Figure 9. In the shift attention task, autistic and cerebellar patientswere substantially slower to detect targets when only 2.5 s or less hadelapsed compared with when there was more time to shift attention.The index of the speed of shifting and detecting targets is shown as thepercentage difference in reaction time (RT) score and was computedas follows: the difference between the RT to the short cue-to-targetdelay (0.4-2.5-s interval) and the RT to the long cue-to-target delay(2.5-4.5-s interval) divided by the latter RT (e.g., for autistic patients,the percentage difference in shift RT = 622 ms minus 477 ms, dividedby 477 ms multiplied by 100). In the shift task, then, the more positivethe percentage difference score, the poorer the ability to rapidly shiftand detect targets; on this score, autistic and cerebellar patients werepoorer than normal controls. In the focus task, the more positive thepercentage difference score, the poorer the ability to make two rapidtarget detections in a row while keeping a single constant focus ofattention; on this score, autistic and cerebellar patients were as goodas normal controls.

21%) and 2 did not (0% and 10%). Among the 4 without MRevidence of parietal abnormality, 2 had elevated levels of falsealarms (28% and 21%) and 2 did not (2% and 4%).

Chronologic-Age-Matched Comparisons

With one exception, all significant and nonsignificant find-ings from comparisons between autistic patients and normalchildren were also obtained from comparisons between autis-tic patients and their chronologic-age-matched normal con-trols. The single exception was that autistic patients hadsignificantly more false alarms in the shift task than theirchronologic-age-matched controls: 15.6% vs. 1.8%, F(l, 14) =6.57,p < .023.

P3b Responses

To verify that autistic and cerebellar patients had notmentally shifted their attention when they missed targets, werecorded the P3b component of the ERP for several subjects.Like control subjects, the autistic and cerebellar patientsexhibited a P3b response to correctly detected targets but notto missed stimuli that occurred 2.5 s or less following a cue (seeFigure 10; also see Figure 5 of Akshoomoff & Courchesne,1992).

Discussion

Our results indicate that like patients with acquired cerebel-lar damage, patients with autism are severely impaired in theirability to coordinate accurate and rapid voluntary shifts ofattention between sensory modalities. Within 2.5 s or lessfollowing a cue to shift attention to a different sensorymodality, both autistic and cerebellar patients frequentlyfailed to detect target information in the new focus (seeFigures 3,6, and 7), and had slower reaction times to those thatthey did correctly detect (see Figure 9). This shift deficit waspersistent and profound despite repeated training and experi-ence in these tasks in 1 autistic patient who was able to beretested on six separate occasions (see Figures 7 and 8). In theAkshoomoff and Courchesne (1992) study, 3 patients withcerebral cortical lesions did not show this extreme impairmentwhen performing the identical shift task. When given moretime (more than 2.5 s), autistic and cerebellar patients wereable to shift attention and detect targets nearly as accurately asnormal controls. Therefore, in these two patient groups, itappears that neural pathology did not prevent the execution ofshifts of attention, but rather seems to have created subopti-mal performance of such shifts.

The inability to accurately and rapidly shift attention in theautistic and cerebellar patients was not due to motor controlproblems. First, as noted above, these patients were notsignificantly impaired in responding to targets that occurred2.5 s or less after a correctly detected target as long as therewas not a requirement to shift attention.

Second, all of the reaction times for the autistic andcerebellar patients were within the 200-1,400-ms time windowallotted for responses to count, which eliminates the possibilitythat responses were so slow they were not counted. Also,relative to this broad time window allowed for responding totargets, there were relatively small absolute differences in RTsbetween the autistic patients and their chronologic- andmental-age-matched normal controls as well as between thecerebellar patients and their normal controls (see Table 4).

Third, the presence of the P3b can be a sign of covertattention, namely the recognition of rare target informationindependent of overt motor action (Courchesne, Hillyard, &Courchesne, 1977), and it is absent when a target stimulus isignored or missed (Ciesielski, Courchesne, & Elmasian, 1990;Squires, Hillyard, & Lindsay, 1973). In the present study and ina study by Akshoomoff and Courchesne (1992), autistic andcerebellar patients, like normal controls, exhibited a P3bresponse to correctly detected targets but not to missed stimulithat occurred 2.5 s or less following a cue. These findings


A.P3bI B. C.

HIT HIT FAIGN MISS

g|

1lo

g|

1lo

•— T-RED

|

g g gI 1 1

1 1lo lo

g red g

1 1 11 I

lo lo

g1

1HI

— U

RED

I1

lo

g|

1to

g1

1 1hi lo

1RED

|

1

lo

1red

|

1

lo

g1

HITIGN

Figure 10. Event-related potentials elicited during the shift attention task by correctly detected (A),correctly ignored (B), and missed visual rare (C) stimuli are shown for the autistic patient (solid line)whose data appear in Figures 7 and 8 and for the average of age-matched normal controls (dotted line).Potentials were recorded from the Pz scalp electrode located above parietal regions. Visual stimuli werered and green (g) flashes; auditory stimuli were 2-kHz (hi) and 1-kHz (lo) tone pips. HIT = correctlydetected target; MISS = a failure to respond to a target; FA = an erroneous response to a rare stimulusthat was in a modality to be ignored; IGN = a rare stimulus that was correctly ignored. In the example ofthe shift experiment, the subject pressed a button (arrow) to the first rare visual target stimulus (RED).This served as a cue to shift attention to the auditory stimuli, ignore (IGN) the rare visual stimuli, andrespond to the next auditory target (HI). The auditory target in turn served as a cue to shift attention backto the visual stimuli.

suggest that when these patients missed targets that rapidlyfollowed a cue, they were not covertly attending and thus hadnot fully shifted their attention to the new focus. As with thepreceding observations, this one also reflects a mental ratherthan a motor output error.

Possible Neural Bases for Shift Deficits in Autism

Contemporary models for the neural basis of shifting themental focus of selective attention do not include the cerebel-lum but do include numerous other structures such as parietalcortex, superior colliculus, pulvinar, and frontal cortex (e.g.,Posner & Petersen, 1990).

In autism, parietal lobe abnormalities have been reported tooccur in 43% of cases (Courchesne, Press, & Yeung-Courchesne, 1993). In nonautistic patients, parietal damagehas been suggested to result in deficits in disengaging visuospa-tial attention (Posner et al., 1984; Posner & Petersen, 1990). Inthe present study there were two subgroups of autistic pa-tients: those with significant reductions in the posterior seg-ment of the corpus callosum and those without. The formersubgroup performed slightly and nonsignificantly better (fasterand more accurately) than the latter subgroup. The presentstudy, therefore, does not provide evidence pointing to a clear

parietal (or posterior cortex) contribution to the shift attentiondeficits seen in the autistic patients.

The most likely explanation for an absence of parietal effectson shift performance in the present study is that parietaldamage may primarily affect visuospatial attention operations,which are not invoked by the present between-modality shiftattention task. Another possibility is that parietal circuitryinvolved in shifting attention operations may not hinge oninterhemisphere communication via the posterior segment ofthe corpus callosum. Still another possibility is that our corpuscallosum index of posterior cortical abnormality may not havebeen sensitive enough to characterize the full extent ofposterior cortical anatomic abnormality in our subjects.

In autism, neuroanatomic abnormalities have not beenfound in any other structure previously implicated in models ofshifting attention. Cerebellar pathology has, however, beenfound in 16 quantitative MR and autopsy reports from nineindependent research groups involving 240 autistic cases (seeTable 1 and Figure 1) and therefore appears to be a consistentfeature of autism as earlier proposed (Courchesne et al,1988). Because cerebellar abnormalities are common to autis-tic and cerebellar patient groups, the present results lead tothe suggestion that the cerebellar pathology in autistic patients


may contribute to their impairment in shifting attention. Thismay happen via cerebellar connections with brainstem, tha-lamic, and cortical systems that have been implicated invarious attentional operations, including moving and engagingattention (Rafal & Posner, 1987; Rafal et al., 1988; Posner &Petersen, 1990).

Role of the Cerebellum

Since the time of Galen, the cerebellum has traditionallybeen linked by medical science to motor control and motorlearning (Ghez, 1991; Oilman, Bloedel, & Lechtenberg, 1981;Glickstein & Yeo, 1990). However, by the turn of the lastcentury, the question of cerebellar involvement in mentalfunctioning had been raised (Beyerman, 1917). Then, in 1949,a short comment marked an important new branch in under-standing about the role of the cerebellum. While reporting thediscovery of the reticular activating system (RAS), Moruzziand Magoun (1949) stated, "Exploration of the overlyingcerebellum has revealed excitable points in its fastigial nuclei,the responses possibly being mediated by connections of the(cerebellar) roof nuclei with the brain stem reticular forma-tion" (p. 458).

Shortly thereafter, "sensory attention" deficits were re-ported in cats with lesions to vermian lobules VI-VII (Cham-bers & Sprague, 1955)—the same lobules first reported by MRtechniques to be abnormal in autism (Courchesne et al., 1988).Furthermore, data from Steriade and Stoupel (1960) sug-gested that the cerebellum may control levels of excitability inthe RAS. Much evidence has now accumulated that thecerebellum has physioanatomical connections with many sys-tems thought to be associated with attention, arousal, andsensory responsiveness (Itoh & Mizuno, 1979; Kitano et al.,1976; Moruzzi & Magoun, 1949; Nieuwenhuys et al., 1988;Sasaki et al., 1972,1979; Schmahmann & Pandya, 1989; Snider,1950; Steriade & Stoupel, 1960; Vilensky & Van Hoesen,1981).

More recently, Crispino and Bullock (1984) found thatstimulation of vermian lobules VI-VII can modulate brain-stem, thalamic, and cortical neural responses to auditory andvisual sensory input in nonbehaving rats. For example, theneural response of the rat superior colliculus to a visualstimulus is enhanced when the stimulus is preceded at anoptimum interval by stimulation of vermian lobules VI-VII(Crispino & Bullock, 1984); similar effects have been reportedfor neural responsiveness in the hippocampus following stimu-lation of vermian lobules VI-VII in cats (Newman & Reza,1979). Moreover, when background luminance is sufficient toobscure the rat colliculus response to a brief flash, thecolliculus response to the brief flash will emerge abovebackground noise levels if, once again, it is preceded atoptimum intervals by stimulation of vermian lobules VI-VII(Crispino & Bullock, 1984).

It seems possible, therefore, that within the domain ofsensory processing, the cerebellum anticipates upcoming eventsand the context in which they will occur and then performsoperations that optimize the neural signal-to-noise conditionsin whichever systems (e.g., brainstem, thalamic, cerebral, andhippocampal) will be involved in processing such events.

Selective attention may involve an analogous anticipatoryenhancement of signal (the desired, to-be-attended informa-tion) relative to noise (the information to be ignored). If so,perhaps vermian lobules VI-VII as well as other cerebellarregions perform analogous anticipatory or preparatory optimiz-ing operations on neural activity in structures known tomediate attention and arousal.

Such a function would be especially important duringconditions that require rapid, unpredictable, and frequentshifts in the focus of selective attention. In our paradigm, thecues to shift attention signaled the need to rapidly prepare todetect and respond to a different target-event-backgroundcontext. Under such demanding conditions, our cerebellar andautistic patients made many more errors than normal. How-ever, an important additional observation was that given moretime, their performance approached normal levels. Perhaps,then, cerebellar pathology does not eliminate attentionaloperations but instead results in an inability to rapidly imple-ment a new optimal pattern of selective neural responsiveness.(It is possible that suboptimal attentional functioning may verywell be sufficient to satisfactorily accomplish some tasks andthus could go unnoticed.)

A parsimonious suggestion is that the cerebellum providesan analogous function for all of the systems with which it isinterconnected, including the motor, attention, arousal, sen-sory, memory, limbic, hypothalamic, serotonergic, dopaminer-gic, and noradrenergic systems (for reviews, see Courchesne,1989b, 1991). That is, without the aid of the cerebellum, eachsystem continues to perform its prescribed specific function,but suboptimally. In his treatise on signs of human cerebellardysfunction, Gordon Holmes (1939) quoted a cerebellar pa-tient as saying, "The movements of my left [unaffected] armare done subconsciously, but I have to think out each move-ment of the right [affected] arm. I come to a dead stop inturning and have to think before I start again" (p. 22). So,cerebellar pathology does not eliminate voluntary motor ac-tion but instead makes motor action slow, inaccurate, andeffortful; the patient may often have to consciously "think"through each step in preparation for action and during theexecution of each action. In a parallel fashion, the evidencecited above suggests that cerebellar pathology does not elimi-nate voluntary shifts of attention but instead makes such shiftsslow and inaccurate. Similarly, cerebellar pathology appar-ently does not prevent relatively good perceptual judgments oftime intervals between any two stimuli but instead makes suchjudgments more variable. For instance, whereas normal sub-jects and cerebellar-lesioned patients differ from each other byas little as 1% in correctly tapping out a 550-ms time intervalbetween two tones, cerebellar patients show a significantincrease relative to normals in the standard deviations associ-ated with such performance or perceptual judgments (e.g.,standard deviations for perceptual judgments between twotones were ± 45.7 ms for cerebellar patients vs. ± 26.1 ms fornormals; Ivry & Keele, 1989). Likewise, cerebellar pathologyapparently does not eliminate the classically conditionednictitating membrane response or its engram but insteadproduces variability in response onset latency and amplitude(e.g., see Figure 16.13 from Welsh & Harvey, 1992). So, itappears that in motor, attentional, perceptual, and associa-


tion learning domains, cerebellar damage does not eliminatefunction but does increase suboptimal variability in responsethresholds, times, amplitudes, and effort.

As reviewed by Welsh and Harvey (1992), rooted in the workof Rolando (1809-1823) and Flourens (1824-1842) is theconcept that the cerebellum ensures "the optimal perfor-mance of the voluntary motor act" (p. 332). It may do so byadjusting the "excitability thresholds of motor nuclei" (p. 332)so that they respond in a manner proportional to motivationalsignificance, associative strength, and intensity of elicitingstimuli (Welsh & Harvey, 1992). A new variant on this theme isthe proposal that by monitoring sensory input, the cerebellumguides movement so as to optimize the quality of sensoryinformation obtained during exploration (Bower & Kassel,1990).

We suggest that these concepts be extended and proposethat the cerebellum optimally shifts excitability thresholds inneurons likely to be used in any sensorimotor or mental action.That is, the cerebellum adjusts responsiveness in whateverneural array or network is anticipated to be needed to attain aprescribed goal (the goal perhaps being prescribed by cerebralcortical or other subcortical systems). Optimal preparation ofneural networks needed to achieve such goals may require thecerebellum to implement a succession of precisely timed andselected changes in the pattern or level of neural activity indiverse networks. The patterns chosen and the time imple-mented hinge on the goals, current context, and anticipatedintervening events. As the actual sequence of events occurs,the cerebellum updates and adjusts patterns of responsiveness.Thus, the cerebellum is continuously providing the optimalneural conditions necessary for handling anticipated events.The cerebellum is no more an "attentional" structure or a"timing module" than it is a motor one. Rather, it is a mastercomputational system for providing the optimal context for thesmooth interdigitated, coordinated neural action of whateversystems are needed from moment to moment to achieve aspecified goal within the context of continuously fluctuatinginternal and external contexts.

Possible Role of Cerebellar Dysfunction in Social andCognitive Development

In normal development, social knowledge and many highercognitive, affective, and communicative functions spring fromearly infant-mother interactions (Bakeman & Adamson, 1984;Bruner, 1975; Trevarthen & Hubley, 1978; Tronick, 1982;Werner & Kaplan, 1963). Tronick (1982) wrote that "success-ful regulation of joint interchanges ... results in normal[cognitive, affective, and social] development" (p. 1), and "Thecrucial element is that the infant and mother... share the samedirectional tendencies or focus of attention during the interaction"(italics added; p. 4). Because during any normal joint inter-change the locus of information (objects, actions, sounds,expressions, internal experience, etc.) frequently, rapidly, andoften unpredictably changes, shifting attention is of fundamen-tal importance. By 12 to 15 months of age, infants normallyachieve the skill of coordinating their attention between asocial partner and objects of mutual interest, a major develop-mental milestone in social communication (Bakeman & Adam-

son, 1984; Bruner, 1975; Trevarthen & Hubley, 1978; Tronick,1982; Werner & Kaplan, 1963). To achieve this major mile-stone, an infant needs to do more than simply focus his or herattention on a single, captivating aspect of an object or person.He or she must follow the rapid and unpredictable ebb andflow of human social activity, such as words, gestures, touching,postures, facial expressions, and actions on objects. Theseactivities provide signals that direct the infant to shift his or herattention in order to follow the varying sources of social,emotional, and situational information. By being able tosmoothly, selectively, and rapidly shift his or her attention withthese signals, the infant is able to combine, as a single entity inmemory, the various and separate elements of a social situa-tion.

Tronick (1982) postulated that "failure to succeed in theregulatory process ... derails [the] infant... from the normaldevelopmental track" (p. 1). Presumably, brain damage thatcauses early and extreme developmental failure of this regula-tory system could produce severe developmental abnormality.

Infantile autism is a prime example of just such a failure ofnormal infant-mother joint attentional interactions. On thebasis of the evidence that autism involves prenatal or earlypostnatal cerebellar damage and that this cerebellar damagemight impair the capacity to control the direction of attention,it is reasonable to suggest that these deficits—present through-out infancy—could contribute to the deviation in social,affective, and cognitive development that characterizes pa-tients with autism. In the autistic infant, such attentionaldysfunction may interfere with the ability to continuouslyfollow the rapidly changing events that compose reciprocalsocial interactions. Much would be missed, and the fragmentscaught would lack "context or temporal continuity" (Cour-chesne, 1987, p. 314). Thus, the autistic infant's "knowledge ofthe [social] world would be made up of disconnected fragmentsof [gestural, facial, vocal, and emotional] information" (Cour-chesne, 1987, p. 314). In addition to failing to normallyapprehend information, the cerebellar-damaged autistic infantmight also be severely impaired in the ability to express in atimely fashion his or her own affective and cognitive reactionsto the information he or she does experience.1 According toTronick's (1982) model, such difficulties should seriouslyhinder the child's ability to engage in joint social interchanges,which in turn should lead to deficiencies in social knowledgeand communication, including imitation, turn-taking, symbolicplay, and the ability to exchange experiences and emotionswith others about topics of mutual interest. Each of thesedeficiencies is, in fact, highly characteristic of infantile autism(Curcio, 1978; Dawson & Lewy, 1989; Kanner, 1943; Landry &Loveland, 1988; Loveland & Landry, 1986; Mundy, Sigman, &Kasari, 1990; Mundy, Sigman, Ungerer, & Sherman, 1986;Sigman, Ungerer, Mundy, & Sherman, 1987; Wetherby &Pruning, 1984).

Thus, it is possible to construct a hypothetical outline of one

1 Brain damage sustained after normal development of affective andcommunicative reactions and knowledge would not necessarily lead totheir regression or loss, for example, just as hippocampal damagesustained after normal language development does not lead to loss ofalready acquired language knowledge.


path of maldevelopment in autism from Purkinje neuron loss,to deficits in the timely and accurate control of the direction ofattention and the expression of intention, to deficits in infant-mother joint social and affective interactions, to deficits inmore complex verbal and nonverbal communication abilities(Courchesne, 1985, 1987, 1989a, 1989b; Courchesne et al.,1988; Courchesne, Townsend, et al., 1994; Courchesne, Chisum,& Townsend, in press).

This suggestion does not preclude the possibility that addi-tional neurocognitive abnormalities contribute as well to thesesocial, affective, and cognitive deficits in autism. For example,as mentioned earlier, a new site of neuroanatomical abnormal-ity has been tentatively identified in a subset (43%) of autisticpatients—the parietal lobes (Courchesne, Press, & Yeung-Courchesne, et al., 1993). This neural abnormality has nowbeen linked to a striking ability to concentrate attention in one,narrow spatial location at the expense of awareness of sensoryevents at other locations (Townsend & Courchesne, 1994).Such an exaggerated narrowing of attention, we speculate, mayresult in a form of functional sensory neglect of events outsidesuch a narrowed focus of conscious attention. Along with thecerebellar findings, this parietal-behavioral finding may alsohelp explain why some autistic patients seem so often to misssocial signals. Uta Frith (1989) wrote, "Minimal stimuli, suchas a slightly raised eyebrow ..., may have profound communi-cative effects" (p. 141). Such important social cues may beentirely missed by autistic patients with parietal abnormalities.As another example, some have proposed a role for thehippocampus and the amygdala in the difficulty autistic pa-tients have in associating social stimuli with emotional states.However, recent studies indicate that such social-emotionalassociations can be established without the hippocampus andthe amygdala (Tranel & Damasio, 1993). Moreover, in a recentstudy Maurer (1988) reported that a patient with bilateraldysgenesis of the hippocampal formation and the amygdala didnot have autistic social, language, cognitive, emotional, orbehavioral features.

In addition to helping explain the principal social communi-cation deficits in autism, the impairment in the control of shiftsof attention may also be at the root of a wide variety ofcommonly observed cognitive abnormalities in autism, includ-ing stimulus overselectivity, uneven memory, insistence onsameness, perseveration, repetitive and ritualistic behaviors,narrowed interests, formation of peculiar associations, andpoor performance on executive function tasks requiring shift-ing of mental sets (Kanner, 1943; Lovaas, Koegel, & Schreib-man, 1979; Lovaas, Schreibman, Koegel, & Rehm, 1971;Schreibman & Lovaas, 1973). For example, stimulus overselec-tivity or "overselective attention" (long suggested to contrib-ute to social deficits in autism) is the tendency of the autisticchild to respond to only one stimulus element of a complexarray of stimuli composing an object, place, or episode (Lovaaset al., 1971, 1979; Schreibman & Lovaas, 1973). The fullexploration and apprehension of such complex stimulus arraysrequires the ability to voluntarily shift attention from elementto element smoothly, precisely, and effortlessly. The presentdata suggest that autistic patients lack this normal ability. Thefailure to attentively and fully explore in a timely fashionnonsocfa/ as well as social stimulus arrays would result in

incomplete memory for events, more so for transient andvariable events (e.g., social, language, or acoustic) and less sofor invariant or predictably repeatable events (e.g., calendarsor flushing toilets). What would be retained would not neces-sarily be stimulus elements that had causal relationships witheach other or spatial or temporal contiguity. For the autisticpatient, the normal coherence of various elements of nonsocialand social events would be lost, as would the predictivecorrelations between elements. Instead, disparate fragmentswould often be registered. They would compose a ratherimpoverished picture of the actual events and the context inwhich they occurred; imprecise, misleading, and possiblybizarre associations and predictions about relationships be-tween successive fragments could ensue. This condition, wesuggest, could provoke disjointed, disorganized, and unex-pected behavioral and emotional reactions by the autisticperson during interactions with the nonsocial environment andwith social partners. Furthermore, this condition could pro-mote a preference for repeatable, predictable, or invariantobjects, events, and activities over novelty, exploration, andsocial partners and social settings. That is, the preference forthe former situations would be expressed as an insistence onsameness, repetitive behavior, and narrowed interests, and thelatter situations would be avoided as confusing and distressing.

Finally, neurobiological data show that early damage in onesystem can alter neural organization in other systems withwhich it is interconnected (Courchesne, Chisum, & Townsend,in press). One wonders, therefore, whether the early cerebel-lar damage in autism might not contribute to abnormalorganization and functioning within the attention, arousal,memory, limbic, sensory, hypothalamic, serotonergic, dopamin-ergic, noradrenergic, and motor systems with which it isinterconnected (Bava, Manzoni, & Urbano, 1966; Chambers &Sprague, 1955; Crispino & Bullock, 1984; Haines & Dietrichs,1987; Ito, 1984; Itoh & Mizuno, 1979; Kitano et al., 1976;Moruzzi & Magoun, 1949; Newman & Reza, 1979; Nieuwen-huys et al., 1988; Saint-Cyr & Woodward, 1980a, 1980b; Sasakiet al., 1972,1979; Schmahmann & Pandya, 1989; Snider, 1950,1967; Steriade & Stoupel, 1960; Vilensky & Van Hoesen, 1981;Watson, 1978). In fact, many parallels exist between thesesystems and systems often posited to be abnormal in autism(Courchesne, 1987,1991; Courchesne et al., 1988).

Concluding Comments

For nearly a century, it has been recognized that whenattention is deficient, cognition suffers: William James (1890)stated that "an object once attended will remain in memory,whilst one inattentively allowed to pass will leave no tracesbehind" (p. 427). Autism is a disorder in which attention ismarkedly deficient. In contrast to the agility with which anormal child can control shifts of attention, the very firstpublished description of an autistic child stated that the childdisplayed "an abstraction of mind which made him perfectlyoblivious to everything about him . . . and [that] to get hisattention almost requires one to break down a mental barrierbetween his inner consciousness and the outside world"(Kanner, 1943, p. 218). Surprisingly, this mental barrierapparently involves damage to cerebellar structures that may


play a role in the coordination of voluntary shifts of attention.This and a host of animal and human behavioral and physioana-tomical data (see previous paragraph) strongly support theconcepts first introduced by Snider (1950, 1967) that thecerebellum is involved in a broad array of nonmotor functions,including higher mental functions, and that damage to it cancontribute to psychiatric disorder.

References

Akshootnoff, N. A., & Courchesne, E. (1992). A new role of thecerebellum in cognitive operations. Behavioral Neuroscience, 106,731-738.

Akshoomoff, N. A., & Courchesne, E. (in press). Intramodalityshifting attention in children with damage to the neocerebellum.Journal of Cognitive Neuroscience.

Akshoomoff, N. A., Courchesne, E., Press, O., & Iraqui, V. (1992).Contribution of the cerebellum to neuropsychology functioning:Evidence from a case of cerebellar degeneration disorder. Neuropsy-chohgia, 30, 315-328.

American Psychiatric Association. (1987). Diagnostic and statisticalmanual of mental disorders (3rd ed., rev.). Washington, DC: Author.

Arin, D. M., Bauman, M. L., & Kemper, T. L. (1991). The distributionof Purkinje cell loss in the cerebellum in autism. Neurology, 41,(Suppl 1) 307.

Bakeman, R., & Adamson, L. B. (1984). Coordinating attention topeople and objects in mother-infant and peer—infant interaction.Child Development, 55, 1278-1289.

Bartak, L., Rutter, M., & Cox, A. (1975). A comparative study ofinfantile autism and specific developmental receptive languagedisorder: I. The children. British Journal of Psychiatry, 126, 127-145.

Bauman, M. L. (1991). Microscopic neuroanatomic abnormalities inautism. Pediatrics, 87, 791-796.

Bauman, M. L., & Kemper, T. L. (1985). Histoanatomic observationsof the brain in early infantile autism. Neurology, 35, 866-874.

Bauman, M. L., & Kemper, T. L. (1986). Developmental cerebellarabnormalities: A consistent finding in early infantile autism. Neurol-ogy, 36, 190.

Bauman, M. L., & Kemper, T. L. (1990). Limbic and cerebellarabnormalities are also present in an autistic child of normalintelligence. Neurology, 40, 359.

Bava, A., Manzoni, T., & Urbano, A. (1966). Cerebellar influences onneuronal elements of thalamic somatosensory relay-nuclei. Archiviodi Szienze Biologiche, 50, 199.

Beyerman, W. (1917). Uber angeborene kleinhirnstorungen [Oncongenital cerebellar disorders]. Archiv Fiir Psychiatric, 57, 610-658.

Botez, M. I., Gravel, J., Attig, E., & Vezina, J. L. (1985). Reversiblechronic cerebellar ataxia after phenytoin intoxication: Possible roleof cerebellum in cognitive thought. Neurology, 35, 1152-1157.

Bower, J. M., & Kassel, J. (1990). Variability in tactile projectionpatterns to the cerebellar folia crus HA of the Norway rat. Journal ofComparative Neurology, 302, 768-778.

Bruner, J. (1975). The ontogenesis of speech acts. Journal of ChildLanguage, 2, 1-19.

Chambers, W. W., & Sprague, J. M. (1955). Functional localization inthe cerebellum. II. Somatotopic organization in cortex and nuclei.Archives of Neurological Psychiatry, 74, 653-680.

Ciesielski, K. T., Allen, P. S., Sinclair, B. D., Pabst, H. F., Yanossky,R., & Ludwig, R. (1990, November). Hypoplasia of cerebellarvermis in autism and childhood leukemia. In Proceedings of the 5thInternational Child Neurology Congress. Tokyo, Japan (p. 650). Basel,Switzerland: Karger.

Ciesielski, K. T., Courchesne, E., & Elmasian, R. (1990). Effects offocused selective attention tasks on event-related potentials in

autistic and normal individuals. Electroencephalography and ClinicalNeurophysiology, 75, 207-220.

Coleman, P. D., Romano, J., Lapham, L., & Simon, W. (1985). Cellcounts in cerebral cortex of an autistic patient. Journal of Autism andDevelopmental Disorders, 15, 245-255.

Courchesne, E. (1985, May 19-22). The missing ingredients in autism.Paper presented at the Conference on Brain and BehavioralDevelopment: Biosocial Dimension, Elridge, MD.

Courchesne, E. (1987). A neurophysiological view of autism. In E.Schopler & G. B. Mesibov (Eds.), Neurobiological issues in autism(pp. 285-324). New York: Plenum.

Courchesne, E. (1989a, July). The control of attention: A key cognitivedeficit in autism. Paper presented at the National Conference of theAutism Society of America, Seattle, WA.

Courchesne, E. (1989b). Neuroanatomical systems involved in infan-tile autism: The implications of cerebellar abnormalities. In G.Dawson (Ed.), Autism: New perspectives on diagnosis, nature andtreatment (pp. 119-143). New York: Guilford Press.

Courchesne, E. (1991). Neuroanatomic imaging in autism. Pediatrics,87, 781-790.

Courchesne, E., Akshoomoff, N. A., & Ciesielski, K. T. (1990). Shiftingattention abnormalities in autism: ERP and performance evidence.Poster presented at the meeting of the International Neuropsycho-logical Society, Orlando, FL.

Courchesne, E., Chisum, H., & Townsend, J. (in press). Neuralactivity-dependent brain changes in development: Implications forpsychopathology. In D. Cicchetti & B. Nurcombe (Eds.), Develop-ment and psychopathology.

Courchesne, E., Hillyard, S. A., & Courchesne, R. Y. (1977). P3 wavesto the discrimination of targets in homogeneous and heterogeneousstimulus sequences. Psychophysiology, 14, 590-597.

Courchesne, E., Press, G. A., & Yeung-Courchesne, R. (1993).Parietal lobe abnormalities detected on magnetic resonance imagesof patients with infantile autism. American Journal of Roentgenology,160, 387-393.

Courchesne, E., Saitoh, O., Yeung-Courchesne, R., Press, G. A.,Lincoln, A. J., Haas, R. H., & Schreibman, L. (1994). Abnormalityof cerebellar vermian lobules VI and VII in patients with infantileautism: Identification of hypoplastic and hyperplastic subgroups byMR imaging. American Journal of Roentgenology, 162, 123-130.

Courchesne, E., Townsend, J. P., Akshoomoff, N. A., Yeung-Courchesne, R., Press, G. A., Murakami, J. W., Lincoln, A. J.,James, H. E., Saitoh, O., Egaas, B., Haas, R. H., & Schreibman, L.(1994). A new finding: Impairment in shifting attention in autisticand cerebellar patients. In S. H. Broman & J. Grafman (Eds.),Atypical cognitive deficits in developmental disorders: Implications forbrain function, (pp. 101-137). Hillsdale, NJ: Erlbaum.

Courchesne, E., Townsend, J. P., & Saitoh, O. (1994). The brain ininfantile autism: Posterior fossa structures are abnormal. Neurology,44, 214-223.

Courchesne, E., Yeung-Courchesne, R., Press, G. A., Hesselink, J. R.,& Jernigan, T. L. (1988). Hypoplasia of cerebellar vermal lobules VIand VII in autism. New England Journal of Medicine, 318, 1349-1354.

Crispino, L., & Bullock, T. H. (1984). Cerebellum mediates modality-specific modulation of sensory responses of the midbrain andforebrain in rat. Proceedings of the National Academy of Sciences,USA, 81, 2917-2920.

Curcio, F. (1978). Sensorimotor functioning and communication inmute autistic children. Journal of Autism and Childhood Schizophre-nia, 8, 281-292.

Dawson, G., & Lewy, A. (1989). Arousal, attention, and the socioemo-tional impairments of individuals with autism. In G. Dawson (Ed.),Autism: New perspectives on diagnosis, nature and treatment (pp.49-74). New York: Guilford Press.


Egaas, B., Courchesne, E., & Saitoh, O. (in press). Reduced size ofcorpus callosum in autism. Archives of Neurology.

Fiez, J. A., Petersen, S. E,, Cheney, M. K., & Raichle, M. E. (1992).Impaired non-motor learning and error detection associated withcerebellar damage. Brain, 115, 155-178.

Frith, U. (1989). A new look at language and communication inautism. British Journal of Disorders of Communication, 24, 123-150.

Gaffney, G. R., Tsai, L. Y., Kuperman, S., & Minchin, S. (1987).Cerebellar structure in autism. American Journal of Diseases inChildren, 141, 1330-1332.

Garber, H. J., & Ritvo, E. R. (1992). Magnetic resonance imaging ofthe posterior fossa in autistic adults. American Journal of Psychiatry,149, 245-247.

Garber, H. J., Ritvo, E. R., Chui, L. C, Griswold, V. J., Kashanian, A.,& Oldendorf, W. H. (1989). A magnetic resonance imaging study ofautism: Normal fourth ventricle size and absence of pathology.American Journal of Psychiatry, 146, 532-535.

Ghez, C. (1991). The cerebellum. In E. R. Kandel, J. H. Schwartz, &T. M. Jessell (Eds.), Principles of neural sciences (3rd ed., pp.626-646). New York: Elsevier Science.

Oilman, S., Bloedel, J. R., & Lechtenberg, R. (1981). Disorders of thecerebellum. Philadelphia: F. A. Davis.

Glickstein, M., & Yeo, C. (1990). The cerebellum and motor learning.Journal of Cognitive Neuroscience, 2, 69-80.

Haines, D. E., & Dietrichs, E. (1987). On the organization ofinterconnections between the cerebellum and hypothalamus. InJ. E. King (Ed.), New concepts in cerebellar neurobiology (pp.113-149). New York: Alan R. Liss.

Hamilton, N. G., Frick, R. B., Takahashi, T., & Hopping, M. W.(1983). Psychiatric symptoms and cerebellar pathology. AmericanJournal of Psychiatry, 140, 1322-1326.

Hashimoto, T., Tayama, M., Murakawa, K., Yoshimoto, T., Miyazaki,M., Harada, M., & Kuroda, Y. (in press). Development of thebrainstem and cerebellum in autistic patients. Journal of Autism andDevelopmental Disorders.

Holmes, G. (1939). The cerebellum of man. Brain, 62, 1-30.Holttum, J. R., Minshew, N. J., Sanders, R. S., & Phillips, N. E. (1992).

Magnetic resonance imaging of the posterior fossa in autism.Biological Psychiatry. 32, 1091-1101.

Ito, M. (1984). The cerebellum and neural control. New York: RavenPress.

Itoh, K., & Mizuno, N. (1979). A cerebello-pulvinar projection in thecat as visualized by the use of anterograde transport of horseradishperoxidase. Brain Research, 171, 131-134.

Ivry, R., & Keele, S. W. (1989). Timing functions of the cerebellum.Journal of Cognitive Neuroscience, 1, 136-152.

James, W. (1890). The principles of psychology (Vol. 1). New York:Dover.

Kanner, L. (1943). Autistic disturbances of affective contact. NervousChild, 2, 217-250.

King, J. E. (1987). New concepts in cerebellar neurobiology. New York:Alan R. Liss.

Kitano, K., Ishida, Y., Ishikawa, T., & Murayama, S. (1976). Re-sponses of extralemniscal thalamic neurones to stimulation of thefastigial nucleus and influences of the cerebral cortex in the cat.Brain Research, 106, 172-175.

Kleiman, M. D., Neff, S., & Rosman, N. P. (1992). The brain ininfantile autism: Are posterior fossa structures abnormal? Neurol-ogy, 42, 753-760.

Landry, S. H., & Loveland, K. A. (1988). Communication behaviors inautism and developmental language delay. Journal of Child Psychol-ogy and Psychiatry and Allied Disciplines, 29, 621—634.

LeCouteur, A., Rutter, M., Lord, C., Rios, P., Robertson, S., Hold-grafer, M., & McLennan, J. (1989). Autism Diagnostic Interview: A

standardized investigator-based instrument. Journal of Autism andDevelopmental Disorders, 3, 363-387.

Leiner, H. C., Leiner, A. L., & Dow, R. S. (1986). Does the cerebellumcontribute to mental skills? Behavioral Neuroscience, 100, 443-454.

Leiner, H. C., Leiner, A. L., & Dow, R. S. (1989). Reappraising thecerebellum: What does the hindbrain contribute to the forebrain?Behavioral Neuroscience, 103, 998-1008.

Lincoln, A. J., Allen, M., & Piacentini, A. (1994). The assessment andinterpretation of intellectual abilities in people with autism. In E.Schopler & G. Mesibov (Eds.), Learning and cognition in autism.New York: Plenum Press.

Lincoln, A. J., Courchesne, E., Kilman, B. A., Elmasian, R., & Allen,M. (1988). A study of intellectual abilities in high-functioningpeople with autism. Journal of Autism and Developmental Disorders,18, 505-524.

Llinas, R., & Sotelo, C. (1992). The cerebellum revisited. New York:Springer-Verlag.

Lord, C., Rutter, M., Goode, S., Heemsbergen, J., Jordan, H.,Mawhood, L., & Schopler, E. (1989). Autism Diagnostic Observa-tion Schedule: A standardized observation of communicative andsocial behavior. Journal of Autism and Developmental Disorders, 2,185-212.

Lovaas, O. I., Koegel, R. L., & Schreibman, L. (1979). Stimulusoverselectivity in autism: A review of research. Psychological Bulle-tin, 86, 1236-1254.

Lovaas, O. L, Schreibman, L., Koegel, R. L., & Rehm, R. (1971).Selective responding by autistic children to multiple sensory input.Journal of Abnormal Psychology, 77, 211-222.

Loveland, K., & Landry, S. (1986). Joint attention and language inautism and developmental language delay. Journal of Autism andDevelopmental Disorders, 16, 335-349.

Maurer, R. (1988, March). A case of congenital amnestic syndrome: Acounter example to limbic system dysfunction in autism. Lecturepresented at the National Institute of Mental Health Workshop onCNS Localization Studies and Autism, Bethesda, MD.

Moruzzi, G., & Magoun, H. W. (1949). Brain stem reticular formationand activation of the EEG. Electroencephalography and ClinicalNeurophysiology, 1, 455 t̂73.

Mundy, P., Sigman, M., & Kasari, C. (1990). A longitudinal study ofjoint attention and language development in autistic children.Journal of Autism and Developmental Disorders, 20, 115-128.

Mundy, P., Sigman, M., Ungerer, J., & Sherman, T. (1986). Definingthe social deficits of autism: The contribution of non-verbal commu-nication measures. Journal of Child Psychology and Psychiatry andAllied Disciplines, 27, 657-669.

Murakami, J. W., Courchesne, E., Press, G. A., Yeung-Courchesne,R., & Hesselink, J. R. (1989). Reduced cerebellar hemisphere sizeand its relationship to vermal hypoplasia in autism. Archives ofNeurology, 46, 689-694.

Newman, P. P., & Reza, H. (1979). Functional relationships betweenthe hippocampus and the cerebellum: An electrophysiological studyof the cat. Journal of 'Physiology (London), 287, 405-426.

Nieuwenhuys, R., Voogd, J., & van Huijzen, C. (1988). The humancentral nervous system: A synopsis and atlas. Berlin: Springer-Verlag.

Ohta, M. (1987). Cognitive disorders of infantile autism: A studyemploying the WISC, spatial relationship conceptualization, andgesture imitations. Journal of Autism and Developmental Disorders,17, 45-62.

Petersen, S. E., Fox, P. T., Posner, M. I., Mintun, M., & Raichle, M. E.(1989). Positron emission tomographic studies of the processing ofsingle words. Journal of Cognitive Neuroscience, 1, 153-170.

Piven, J., Nehme, E., Simon, J., Barta, P., Pearlson, G., & Folstein,S. E. (1992). Magnetic resonance imaging in autism: Measurementof the cerebellum pons and fourth ventricle. Biological Psychiatry, 31,491-504.


Posner, M. I., & Petersen, S. (1990). The attention system of thehuman brain. Annual Review of Neuroscience, 13, 25-42.

Posner, M. I., Walker, J. A., Freidrich, F. A., & Rafal, R. D. (1984).Effects of parietal injury on covert orienting of attention. Journal ofNeuroscience, 4, 1863-1874.

Rafal, R. D., & Posner, M. I. (1987). Deficits in human visual spatialattention following thalamic lesions. Proceedings of the NationalAcademy of Science, 84, 7349-7353.

Rafal, R. D., Posner, M. I., Freidman, J. H., Inhoff, A. W., &Bernstein, E. (1988). Orienting of visual attention in progressivesupranuclear palsy. Brain, 111, 267-280.

Raymond, G., Bauman, M. L., & Kemper, T. (1989). The hippocam-pus in autism: Golgi analysis. Annals of Neurology, 26, 483-484.

Raz, N., Torres, I. J., Spencer, W. D., White, K., & Acker, J. D. (1992).Age-related regional differences in cerebellar vermis observed invivo.Archives of Neurology, 49, 412—416.

Rimland, B. (1964). Infantile autism: The syndrome and its implicationsfor a neural theory of behavior. Englewood Cliffs, NJ: Prentice-Hall.

Ritvo, E. R., Freeman, B. J., & Scheibel, A. B., Avong, T., Robinson,H., Guthrie, D., & Ritvo, A. (1986). Lower Purkinje cell counts inthe cerebella of four autistic subjects: Initial findings of theUCLA-NS AC autopsy research report. American Journal of Psychia-try, 143, 862-866.

Saint-Cyr, J. A., & Woodward, D. J. (1980a). Activation of mossy andclimbing fiber pathways to the cerebellar cortex by stimulation of thefornix in the rat. Experimental Brain Research, 40, 1-12.

Saint-Cyr, J. A., & Woodward, D. J. (1980b). A topographic analysis oflimbic and somatic inputs to the cerebellar cortex in the rat.Experimental Brain Research, 40, 13-22.

Saitoh, O., Courchesne, E., Egaas, B., Lincoln, A. J., & Schreibman, L.(in press). Cross-sectional area of the posterior hippocampus inautistic patients with cerebellar and corpus callosum abnormalities.Neurology.

Sasaki, K., Jinnai, K., Gemba, H., Hashimoto, S., & Mizuno, N. (1979).Projection of the cerebellar dentate nucleus onto the frontalassociation cortex in monkeys. Experimental Brain Research, 37,193-198.

Sasaki, K., Matsuda, Y., Kawaguchi, S., & Mizuno, N. (1972). On thecerebello-thalamo-cerebral pathway for the parietal cortex. Experi-mental Brain Research, 16, 89-103.

Schmahmann, J., & Pandya, D. N. (1989). Anatomical investigation ofprojections to the basis pont from posterior parietal associationcortices in rhesus monkey. Journal of Comparative Neurology, 289,53-73.

Schreibman, L., & Lovaas, O. I. (1973). Overselective response tosocial stimuli by autistic children. Journal of Abnormal Child Psychol-ogy, 1, 152-168.

Sigman, M., Ungerer, J. A., Mundy, P., & Sherman, T. (1987).Cognition in autistic children. In D. Cohen & A. M. Donnellan(Eds.), Handbook of autism and pervasive developmental disorders(pp. 103-120). New York: Wiley.

Snider, R. S. (1950). Recent contributions to the anatomy andphysiology of the cerebellum. Archives of Neurological Psychiatry, 64,196-219.

Snider, R. S. (1967). Functional alterations of cerebral sensory areasby the cerebellum. In C. A. Fox, & R. S. Snider (Eds.), Progress in

brain research, Vol. V. The cerebellum, Vol. 25, 322-332. Amsterdam:Elsevier.

Sperling, G., & Reeves, A. (1980). Measuring the reaction time of ashift of visual attention. In R. S. Nickerson (Ed.), Attention andperformance VIII. Hillsdale, NJ: Erlbaum.

Squires, K. C., Hillyard, S. A., & Lindsay, P. H. (1973). Vertexpotentials evoked during auditory signal detection: Relation todecision criteria. Perception & Psychophysics, 14, 265-272.

Steriade, M., & Stoupel, N. (1960). Contribution a 1'etude desrelations entre 1'aire auditive du cerebelet et 1'ecore cerebrale chezle chat [Contribution of the study of the relationships between theauditory regions of the cerebellum and the cerebral cortex in thecat]. Electroencephalic Clinical Neurophysiology, 12, 119-136.

Swets, J. A. (1986a). Form of empirical ROCs in discrimination anddiagnostic tasks: Implications for theory and measurement ofperformance. Psychological Bulletin, 99, 181-198.

Swets, J. A. (1986b). Indices of discrimination or diagnostic accuracy:Their ROCs and implied models. Psychological Bulletin, 99, 100-117.

Townsend, J., & Courchesne, E. (1994). Parietal damage and narrow"spotlight" spatial attention. Journal of Cognitive Neurosciences, 6,218-230.

Tranel, D., & Damasio, A. R. (1993). The covert learning of affectivevalence does not require structures in hippocampal system oramygdala. Journal of Cognitive Neurosciences, 5, 79-88.

Trevarthen, C., & Hubley, P. (1978). Secondary intersubjectivity. In A.Lock (Ed.), Confidence, confiding and acts of meaning in the first year:Action, gesture, and symbol. San Diego, CA: Academic Press.

Tronick, E. Z. (1982). Affectivity and sharing. In E. Z. Tronick (Ed.),Social interchange in infancy: Affect, cognition and communication(pp. 1-6). Baltimore, MD: University Park Press.

Vilensky, J. A., & Van Hoesen, G. W. (1981). Corticopontineprojections from the cingulate cortex in the rhesus monkey. BrainResearch, 205, 391-395.

Watson, P. J. (1978). Nonmotor functions of the cerebellum. Psychologi-cal Bulletin, 85, 944-967.

Weichselgartner, E., & Sperling, G. (1987). Dynamics of automaticand controlled visual attention. Science, 238, 778-780.

Welsh, J. P., & Harvey, J. A. (1992). The role of the cerebellum involuntary and reflexive movements: History and current status. In R.Llinas & C. Sotelo (Eds.), The cerebellum revisited (pp. 301-334).New York: Springer-Verlag.

Werner, H., & Kaplan, B. (1963). Symbol formation. New York: Wiley.Wetherby, A., & Pruning, C. (1984). Profiles of communicative and

cognitive-social abilities in autistic children. Journal of Speech andHearing Research, 27, 264-377.

Williams, R. S., Hauser, S. L., Purpura, D. P., DeLong, R., & Swisher,C. N. (1980). Autism and mental retardation: Neuropathologicalstudies performed in four retarded persons with autistic behavior.Archives of Neurology, 37, 749-753.

Received May 31,1993Revision received April 20,1994

Accepted April 20,1994

impairment in shifting attention in autistic and cerebellar patients€¦ · impairment in shifting...

Documents