Journal of Autism and Developmental Disorders

, Volume 36, Issue 2, pp 199–210

Subjective Perceptual Distortions and Visual Dysfunction in Children with Autism


  • Rebecca A. O. Davis
    • Department of PsychologyIndiana University
    • Department of OtolaryngologyIndiana University School of Medicine
  • Marcia A. Bockbrader
    • Department of PsychologyIndiana University
  • Robin R. Murphy
    • Department of PsychologyIndiana University
  • William P. Hetrick
    • Department of PsychologyIndiana University
    • Department of PsychologyIndiana University
    • Department of PsychologyIndiana University

DOI: 10.1007/s10803-005-0055-0

Cite this article as:
Davis, R.A.O., Bockbrader, M.A., Murphy, R.R. et al. J Autism Dev Disord (2006) 36: 199. doi:10.1007/s10803-005-0055-0

Case reports and sensory inventories suggest that autism involves sensory processing anomalies. Behavioral tests indicate impaired motion and normal form perception in autism. The present study used first-person accounts to investigate perceptual anomalies and related subjective to psychophysical measures. Nine high-functioning children with autism and nine typically-developing children were given a questionnaire to assess the frequency of sensory anomalies, as well as psychophysical tests of visual perception. Results indicated that children with autism experience increased perceptual anomalies, deficits in trajectory discrimination consistent with dysfunction in the cortical dorsal pathway or in cerebellar midsagittal vermis, and high spatial frequency contrast impairments consistent with dysfunctional parvocellular processing. Subjective visual hypersensitivity was significantly related to greater deficits across vision tests.


Autism; visual system; contrast sensitivity; sensory profile; motion processing; subjective distortions.


Autism is a developmental disorder characterized by perseverative behaviors, communication deficits, and social dysfunction. Case reports (Stehli, 1991; Williams, 1994) suggest that these behaviors are often accompanied by sensory anomalies. For example, in the visual modality, some individuals with autism report hyperacute sensitivity for and detail (Stehli, 1991), while others indicate that moving stimuli are associated with pleasurable or comforting sensations (Grandin, 2000). Abnormal responses to sensory stimulation have been noted by many clinical observers (Gillberg & Colemen, 2000). Previous studies have used either sensory inventories or psychophysical tests to investigate sensory anomalies in autism spectrum (AS) disorders. The present study combines these approaches and investigates the relationship between self-reports of perceptual disturbance and performance on psychophysical tasks.

Sensory inventories, which are typically based on parental responses during clinical interviews or on questionnaires, support case reports of perceptual anomalies in AS disorders. For example, Ornitz and colleagues (1977) found that over 70% of autistic children showed either altered sensory sensitivity or non-responsiveness in the auditory, visual, tactile or vestibular modalities. Similarly, Kientz and Dunn (1997) reported that behaviors related to auditory, visual, gustatory, and tactile anomalies discriminated AS children from typically-developing (TD) children. Finally, Talay-Ongan and Wood (2000) found significantly greater sensory hyper- or hyposensitivity in AS children across the visual, auditory, tactile, gustatory, olfactory and vestibular domains. Each of these studies based their findings on parental responses.

In the visual modality, converging psychophysical evidence confirms case reports of normal to hyperacute color and form perception, but disordered motion perception. O’Riordan and Plaisted (2001) compared visual search performance between children with AS disorders and TD children matched on age- and non-verbal ability. They found that the AS group was significantly faster at identifying targets based on small differences in form or color, supporting hyperacute form and color perception in autism. Spencer and colleagues (2000) assessed ability to locate circular forms in coherent line patterns and ability to locate moving patches in coherent dot displays. Their AS group was matched to a TD group on verbal mental age. Spencer and colleagues found group differences on the motion, but not the form task. An additional study using coherent dot patterns replicated this motion perception deficit (Milne et al.,2002). Taken together, these results indicate that while choice of task influences psychophysical performance in AS disorders, perception of motion in random dot displays is consistently impaired.

While sensory disturbances are consistent with theories of autism, and sensory inventories and psychophysical tests provide evidence for sensory dysfunction in AS disorders, subjective anomalies and psychophysical impairments in AS disorders have not been well characterized. In particular, studies targeting subjective anomalies have used non-homogeneous samples of AS children, third person reports from parents or caregivers, and have lacked psychophysical measures to corroborate the presence of a sensory disturbance. In addition, psychophysical studies have not used a battery of tests that differentially examined the subcortical and cortical visual system pathways. The present study was designed to address these limitations.

To investigate subjective parental and self-reports of perceptual anomalies, the present study adapted an instrument used to study perceptual distortions in schizophrenia, the Structured Interview for Assessing Perceptual Anomalies (SIAPA; Bunney et al., 1999). The SIAPA assesses sensory sensitivity, flooding, and attentional dysregulation in the auditory, visual, tactile, gustatory, and olfactory modalities.

Tests of form and motion perception were used to assess function in the cortical visual pathways. Form matching and discrimination were used to target ventral pathways, which project from primary visual cortex through visual areas in the inferior temporal lobe (Ungerleider & Mishkin, 1982). Similar paradigms were used to parametrically evaluate coherent motion perception in dynamic dot displays. Primate studies indicate that these motion tasks activate dorsal visual pathways, which project from primary visual cortex through V5/MT and IP (Livingstone & Hubel, 1988; Newsome & Pare, 1988); brain activation studies (Dupont, Orban, de Bruyn, Verbruggen, & Mortelmans, 1994; Hautzel et al., 2001) and evidence from human stroke patients (Nawrot & Rizzo, 1998) suggest that the cerebellar vermis is also involved in these motion tasks.

Finally, contrast sensitivity tasks were used to assess processing in the subcortical visual pathways through the lateral geniculate nucleus. Primate studies indicate that the subcortical magnocellular system is sensitive to low spatial, high temporal frequency stimuli (Merigan, Byrne, & Maunsell, 1991a; Merigan & Maunsell, 1990), such as flickering gratings with widely-spaced lines. In contrast, the subcortical parvocellular system is sensitive to high spatial, low temporal frequency stimuli (Merigan, Katz, & Maunsell, 1991b), such as stationary gratings with narrowly-spaced lines. Therefore, flicker detection tasks with low spatial frequency stimuli were used to target magnocellular dysfunction, while pattern detection tasks with high spatial frequency stimuli were used to target parvocellular dysfunction.

Based on case reports of heightened sensory sensitivity in autism, as well as Talay-Ongan and Woods’ findings (2000), we expected to find increased reports of sensory disturbances in all modalities tested on the SIAPA. We also expected to find unimpaired form perception coupled with impaired motion perception in autism. Previous studies using our methods in Axis-I and Axis-II disorders have found that deficits present in discrimination conditions become exacerbated in delayed match-to-sample tasks (Brenner, Wilt, Lysaker, & O’Donnell, 2003; Farmer et al.,2000). Therefore, we expected motion deficits to increase between the discrimination and matching conditions. No previous studies have addressed contrast sensitivity in autism. However, we expected to find group differences in the pattern detection task, consistent with case reports of super-sensitivity for fine detail in autism (O’Riordan & Plaisted, 2001). Finally, given the hypothesized theoretical relationship between subjective anomalies and psychophysical performance, we expected psychophysical performance on tests of visual function to be related to distortions of visual sensitivity endorsed on the SIAPA.



Nine children with autism (10–18 years) and nine age-matched, typically-developing (TD) children (7–15 years) participated in the study (Table I provides participant characteristics). Autism is characterized by impairment in communication, and many individuals with autism never learn to communicate verbally. However, a primary inclusion criterion was linguistic facility to ensure full understanding of the SIAPA items. Therefore, only high-functioning children with autism were recruited (Table I).
Table I.

Subject Characteristics


Children with Autism (n=9)

Typically-Developing (n=9)


9 M

6 M, 3 F

Age in years

12.3 (2.4)

11.9 (2.7)


    Picture Completion*

12.4 (3.4)

16.3 (2.5)


12.7 (4.4)

15.1 (1.6)

*Significant difference between AS and TD groups (p<0.05).

Autism diagnoses were determined by a clinical psychologist (RRM), who administered the Autism Diagnostic Inventory-Revised (ADI-R; Lord, Rutter, & Le Couteur, 1994). All children with autism met criteria using the ADI-R diagnostic algorithm, and by DSM-IV criteria. No children in the study had expressive language developmental delays. All but two of the autistic children were medicated at the time of testing.

Children in both groups were screened to exclude neurological disorders, uncorrected vision problems, and drug or alcohol abuse. Informed consent was obtained from a parent or legal guardian before participation. Both groups received the Similarities and Picture Completion subtests of the Wechsler Intelligence Scale for Children, Third Edition (WISC-III; The Psychological Corporation 1991) to assess abstract verbal capabilities and visual ability to discriminate essential from nonessential information. While the groups did not differ significantly in their Similarities score, indicating matched verbal capacity, the children with autism scored significantly lower than the TD group on Picture Completion.

Structured Interview for Assessing Perceptual Anomalies-Child Version (SIAPA-CV)

The Structured Interview for Assessing Perceptual Anomalies (SIAPA; Bunney et al.,1999) contains 15-items which assess hypersensitivity, flooding, and attentional dysregulation in the auditory, visual, tactile, olfactory, and gustatory modalities. An interviewer rates the frequency of each item on a 5-point Likert scale, ranging from 0 (“never”) to 4 (“always”). Dependent measures included individual item scores, average scores for each sensory modality, and a sum of each of the SIAPA item scores.

Because some phrasing in the SIAPA was too advanced for children, several questions were reworded for the SIAPA-CV1.1. Changes involved single word replacements using a simpler word with the same meaning. To facilitate spontaneity, a sentence assessing global sensory experiences was asked before individual SIAPA-CV items in each modality (i.e., “Is there anything different about your (insert modality here) compared to your friends or brothers and sisters?”). Positive responses were noted under the appropriate anomaly (hypersensitivity, flooding or attention), and were marked as being spontaneously volunteered. In addition, participants who responded positively were asked to provide examples to substantiate their claim. This methodology was used in order to control for response bias.

Psychophysical Tests of Form and Motion Perception

Form and Motion paradigms were adapted from vision tests used to study schizotypal personality disorder (Farmer et al.,2000) and schizophrenia (Brenner et al.,2003). Stimuli were presented on a CRT monitor in a dark, quiet room at a viewing distance of 70 cm. To determine performance thresholds, an adaptive staircase procedure (Levitt, 1970) was used that modulated the amount of noise in the display. In this procedure, noise was introduced into the stimulus after every two consecutive correct responses and reduced after every incorrect response (maintaining a performance criterion of 70.71% correct responses). After a series of trials, the participant’s performance gradually converged at a threshold value. Thresholds were calculated from the noise levels of the final four reversals on each 50-trial staircase. The order of presentation for Form and Motion tasks was randomized across subjects.

In Form tasks, outlines of black letters were presented in a white, 75 × 75 pixel window (2.37 degrees of visual angle) centered on the screen. Visual noise was generated by an algorithm that inverted black pixels to white, and vice versa, at randomly selected positions in the window. Thresholds computed by the staircase were converted to percent of display noise, which was used as the dependent variable in statistical tests.

In Motion tasks, a pattern of one hundred 2.5×2.5 mm black dots was presented on a gray background centrally on the monitor. The dots moved across a 200×200 pixel window (6.31 degrees of visual angle) at a speed of 6.36 degrees of visual angle per second. In the no-noise (coherent motion) condition, all dots in the display moved in the same direction. Noise was added to the display by reducing the number of dots that moved with the coherent pattern. The staircase procedure determined the minimum number of dots needed in the pattern for the participant to see coherent motion. Coherence thresholds from the staircase were converted to noise thresholds (100-coherence level), which were used as dependent variables in statistical tests.

Discrimination Tasks

Form stimuli were capital letters (“D” or “U”) presented for 1000 ms (Fig. 1). After the stimulus presentation, a question-mark prompt appeared, indicating that participants should verbally identify the letter.
Fig. 1.

Stimuli used in psychophysical vision tests. Top left: Form stimulus. Noise was added by inverting random pixels in the display. Top right: Motion stimulus. Arrows represent the trajectory of each dot. Noise was added by decreasing the number of dots that moved coherently with the pattern. Bottom left: Pattern detection stimulus. High spatial frequency vertical grating. Bottom right: Flicker detection stimulus. Low spatial frequency vertical grating.

Motion discrimination tasks were completed for both short (220 ms) and long (1000 ms) stimulus durations to test the effect of stimulus duration on ability to integrate stimulus characteristics across time (this methodology was not used in the form tasks due to the static nature of stimuli). Trials began with the appearance of a white fixation point then the dynamic dot display appeared for the stimulus duration (Fig. 1). Coherent motion in the display was either to the right or to the left. Finally, a question-mark prompt appeared, indicating that the participant should respond by identifying the motion direction. Because children sometimes have trouble discriminating left from right, a couple of methods were employed to ensure correct coding of responses. First, a display was hung above the screen with a large “L” on the left-hand side and a large “R” on the right-hand side. In addition, all subjects were asked to identify the direction of motion by pointing to the left or right as well as by making a verbal response.

Matching Tasks

Form stimuli were capital letters (“D”, “U”, “C”, or “L”) displayed on the screen for 220 ms with a 1000 ms inter-stimulus-interval (ISI). After the presentation of both stimuli, a question-mark prompt appeared, indicating that participants should verbally identify whether the letters were the same or different.

Motion matching tests were completed for short (70 ms) and long (1000 ms) ISIs to provide a parametric evaluation of the effect of delay interval on ability to maintain a representation of stimulus characteristics. Trials began with a white fixation point then the first dot display appeared for 220 ms, followed by the ISI, and the second pattern of moving dots. Coherent patterns moved either left, right, up or down. Finally, a question-mark prompt appeared, indicating that the participant should make a verbal response identifying whether the motion directions in the two stimuli were the same or different.

Contrast Sensitivity Tests for Pattern and Flicker Detection

Contrast sensitivity is the inverse of the contrast threshold (the minimum difference in brightness needed to reliably detect a stimulus). Contrast stimuli were sinusoidally modulated, vertically oriented gratings produced by the Morphonome Image Psychophysics System (Tyler, 1995). Gratings were presented at a distance of 115 cm on a calibrated CRT monitor in a dark, quiet room. Stimuli had a Gaussian envelope that subtended 5.23 degrees of visual angle. The order of Contrast Sensitivity tasks was randomized across subjects, and these tasks were always presented after the Form and Motion block.

A Yes/No staircase was used to estimate thresholds (Tyler, 1991; Tyler, 1995) by varying contrast based on correct and incorrect responses. The staircase increased contrast for incorrect responses, making judgments easier, and decreased contrast for correct responses, making judgments harder. This method provided a 75% correct threshold level, which was calculated on the moving average of the last 16 trials in the sequence. For each contrast test, the threshold estimate produced by Morphonome was converted to log base-10 sensitivity (log10 sens=log10 (1/threshold)), which was used as the dependent measure in statistical analyses.

Pattern Detection

The pattern detection task used a high spatial frequency grating (13.4 cycles per degree (cpd)) as shown in Fig. 1. Two types of trials (grating-present, null) were presented in random order. Fifty percent of the trials were grating-present trials. In grating-present trials, the grating gradually faded in and out over a gray background (luminance=50.31 cd/m2). The duration of each trial was 493 ms, resulting in a temporal frequency of 2 Hz, with a cosine time envelope. In null trials, the gray background was displayed without modulation for 493 ms. In all trials, participants verbally reported whether or not a grating was present.

Flicker Detection

There were two flicker detection tasks (5 Hz flicker, 12.5 Hz flicker). Each used a low spatial frequency (0.5 cpd) grating as shown in Fig. 1. Because magnocellular and parvocellular functions overlap for stimuli with intermediate temporal or spatial frequency properties, only the high temporal frequency stimulus was interpreted as targeting magnocellular function.

Two types of trials (flicker-present, null) were presented in random order for both tasks. Seventy percent of the trials were flicker-present trials. In flicker-present trials, the low spatial frequency grating underwent one counter-phase modulation. (In a counter-phase modulation, three pictures are shown in rapid succession: the initial grating, a negative image of the initial grating, and the initial grating.) Gratings were presented over a gray background (luminance=50.31 cd/m2). Trial duration was 200 ms in the 5 Hz condition and 80 ms in the 12.5 Hz condition. In null trials, the grating was displayed without counter-phase modulation. In all trials, participants verbally reported whether or not a flicker was present.

Statistical Analysis

Three types of statistical analyses were used to test study hypotheses. Non-parametric Mann–Whitney U-tests were used to test group differences on the SIAPA-CV and the psychophysical tests due to the ordinality and non-normality of the data, respectively. For similar reasons, Spearman rho’s were computed to test correlational hypotheses. Partial least squares analysis (PLS; Bookstein, Sampson, Streissguth, & Barr, 1996; O’Donnell et al.,1999) was used to test the relationship between sets of variables (SIAPA-CV scores and vision test measures).


Structured Interview for Assessing Perceptual Anomalies-Child Version

Individuals with autism scored higher than typically-developing (TD) controls on all SIAPA-CV measures (Table II). Mann–Whitney U-Tests confirmed that Auditory (U=0.50, p<0.001), Visual (U=11.00, p<0.008), Tactile (U=9.50, p<0.004), Olfactory (U=10.00, p<0.006), Gustatory (U=13.00, p<0.014), and Total Score (U=0.50, p<0.001) measures of the SIAPA-CV interview all differed significantly between groups (Table II).
Table II.

Structured Interview for Assessing Perceptual Anomalies-Child Version (SIAPA-CV) Measures


Children with autism


Significance level


2.41 (0.83)

0.15 (0.34)



1.59 (1.21)

0.26 (0.36)



1.30 (0.87)

0.18 (0.24)



1.00 (0.80)

0.04 (0.11)



0.93 (0.94)

0.07 (0.15)


Average SIAPA-CV item score

1.44 (0.56)

0.14 (0.13)


All children with autism reported at least one sensory anomaly. However, the frequency of anomalies differed both across children and across modalities. Figure 2 shows the relative frequency of each type of anomaly on the SIAPA-CV for children with autism. Over 40% of children with autism reported sensory hypersensitivity across all five modalities, with auditory and tactile hyper-intensity as the most frequently endorsed items. Sensory flooding showed a similar profile, with the exception of the visual modality, which was not as frequently endorsed. Nearly all children reported auditory and visual attentional dysregulation; however, attentional dysregulation in the other three modalities was far less frequently endorsed.
Fig. 2.

Percentage of children with autism reporting sensory anomalies on the SIAPA-CV. Notably, only four questions had a SIAPA-CV response above 1 (rarely) from a TD child. Moreover, each of these four questions had only one such TD child response, and these responses were given by four different children.

Moreover, the types of anomalies reported were qualitatively different between groups. In particular, children with autism were bothered by everyday environmental sensory stimuli, while TD children tended to mention distressing or out-of-the-ordinary events. For example, in the auditory sensitivity category, one participant with autism reported, “I can always hear things from far away. I have keen hearing.... It bothers me when people tap their fingers in class.” In contrast, a TD child reported, “Sometimes when it’s really quiet I can hear everything.” Similarly, in the auditory flooding category, a participant with autism reported, “Yes, [I feel flooded or overwhelmed] by all loud noises. I’m not allowed in the band room at school. I never go near the band because of the loud instruments.” In contrast, a TD child reported, “It’s only happened once or twice, but [I feel flooded or inundated] when people are yelling, like when my parents are having a fight.”

Psychophysical Tests of Visual Perception

Discrimination Tests

Children with autism performed worse than TD children on motion discrimination tasks, although the difference was significant only for the 1000 ms condition, U=18.00, p<0.05 (effect size=0.69). There were no significant differences between groups on the form discrimination task. Figure 3 compares performance on these three measures between children with autism, TD children, and healthy adult norms.
Fig. 3.

Performance on form and motion discrimination tests. Error bars represent standard deviations. For comparison, adult norms are also shown (N=35, 18–35 years). *Significant difference between AS and TD groups (p<0.05).

Matching Tasks

There were no significant differences between these groups on the form and motion matching tests. Figure 4 shows relative performance on these three measures for children with autism, TD children, and healthy adult norms.
Fig. 4.

Performance on form and motion matching tests. Error bars represent standard deviations. In each test, the duration of the stimulus presentations was 220 ms. For comparison, adult norms are also shown (N=35, 18–35 years).

Contrast Sensitivity Tasks

Participants with autism showed similar contrast sensitivity profiles for flickering stimuli compared to TD children. Mann–Whitney U-tests comparing performance between groups on the 5 Hz and 12.5 Hz flicker tasks revealed no significant differences. However, children with autism had reduced contrast sensitivity for pattern detection. A Mann–Whitney U-test revealed that the reduction in sensitivity was significant, U=9.00, p<0.004. Figure 5 shows relative contrast sensitivity on these tasks for children with autism, TD children, and healthy adult norms.
Fig. 5.

Performance on contrast sensitivity measures. Error bars represent standard deviations. The pattern test assessed each participant’s ability to detect the gradual onset of a high spatial frequency grating. The flicker tests determined how well participants could detect a single, rapid (5 or 12.5 Hz) flicker of a low spatial frequency stimulus. For comparison, adult norms are also shown (Pattern: N=28, 18–43 years; Flicker: N=11, 19–43 years). * Denotes significant difference between AS and TD groups (p<0.05).

Relationship between Psychophysical Sensitivity and Subjective Visual Anomalies

Partial least squares analysis (PLS; Bookstein et al.,1996; O’Donnell et al.,1999) was used to determine whether there was an overall relationship between psychophysically measured visual sensitivity and subjectively reported visual sensitivity in the children with autism. This recently developed multivariate analysis can be used to test for relationships between two sets of measures in small samples. The first pair of latent variables obtained from the analysis indicated a significant relationship between the subjective and psychophysical measures. The psychophysical sensitivity latent variable was composed of all the dependent measures from the vision tests. The subjective visual sensitivity latent variable was composed of the visual sensitivity measure from the SIAPA-CV. The PLS analysis indicated an overall relationship between psychophysical sensitivity and frequency of sensitivity-related visual anomalies (p<0.0462, by permutation test). All psychophysical vision tests were negatively related to the subjective visual anomaly latent variable, indicating that worse performance on the psychophysical tests was associated with greater frequency of visual sensitivity anomalies. The vision tests that loaded most strongly on the psychophysical test latent variable were: Form Discrimination (−0.534), Pattern Detection (−0.460), 1000 ms duration Motion Discrimination (−0.392) and Form Matching (−0.381).



The results of this study suggest that perceptual anomalies occur more pervasively and frequently among children with autism than among typically-developing (TD) children. Within each sensory modality, there was a significant difference between groups in the reported frequency of anomalies. In addition, the total SIAPA-CV score differed significantly between groups. These findings concur with frequent clinical observations of sensory impairments in autism (Gillberg & Coleman, 2000).

Within the autistic group, a different profile of distortions emerged for each sensory modality. While all children in this group reported at least one sensory anomaly, the most frequent occurred in the auditory, visual, and tactile modalities. These findings are consistent with case studies (Grandin, 1992; Stehli, 1991) suggesting that auditory and tactile anomalies occur frequently in autism. In addition, these results highlight the importance of visual anomalies in the phenomenology of autism.

A different profile of distortions also emerged within the autistic group across categories of sensory anomalies (i.e., sensitivity, flooding, attentional dysregulation). All children with autism reported auditory attentional dysregulation and all but one reported visual attentional dysregulation. In addition, all of the children with autism reported sensory hypersensitivity in at least one modality, with most of these children reporting heightened sensitivity in several modalities. This is consistent with the findings of Talay-Ongan and Wood (2000) and Ornitz and colleagues (1977), who reported unusual sensory sensitivities in autism. We also found sensory flooding was frequently endorsed in all modalities except visual. The co-occurrence of sensitivity, flooding, and attentional dysregulation in particular children suggests that these three categories of distortions may not be mutually exclusive, in the sense that a child who is experiencing heightened sensory sensitivity or flooding may have difficulty regulating attention.

In summary, the present study’s findings strongly support the inclusion of the SIAPA-CV/SIAPA as clinical tools, which could complement diagnostic interviews by providing a standardized format for assessing specific sensory disturbances.

Psychophysical Tests of Visual Perception

Form and Motion Perception

Children with autism performed worse than TD children on motion discrimination, but did not differ from controls on form discrimination. In addition, children with autism had unimpaired performance compared to TD children on both matching paradigms. Because the criterion level of performance was held fixed across tests (at 70.71% correct), this pattern of results may imply a differential deficit for motion discrimination in autism (Chapman & Chapman, 1978).

The motion discrimination deficit only reached significance in the 1000 ms condition. This finding is consistent with the work of Milne and colleagues (2002), who found a deficit in high-functioning children with autism for a random dot stimulus with an equivalent duration. The lack of significant differences in the 220 ms condition may be due to our small sample size (effect size=0.58). The work of Spencer and colleagues (2000) supports this claim; they found a significant motion coherence deficit for children with autism using a stimulus that reversed direction every 330 ms. Alternatively, the difference in significance levels between the two motion discrimination conditions may highlight a peculiarity of motion processing specific to autism: Both TD children and adult norms show a performance gain with a longer stimulus duration, while autistic children do not, resulting in a larger separation between group means. To test this hypothesis, a parametric evaluation of the effect of stimulus duration on motion discrimination is warranted in autism.

The motion deficit only appeared in discrimination and not matching tests, suggesting that the deficit is specific to identifying, and not comparing, directions of moving dots. While the former task involves specific neural mechanisms in primates (Shadlen & Newsome, 2001), the latter task may be successfully completed using alternative compensatory strategies. For example, a person with autism who thinks “in pictures”, like Temple Grandin describes (1992), could perform well on the matching tasks by comparing two stored visual images. However, this strategy would not work for identification, where a participant’s response cannot be based on a comparison of two images.

The observed motion processing deficit in children with autism is consistent with degraded processing in dorsal cortical visual pathways (Newsome & Pare, 1988; Shadlen & Newsome, 2001) or in cerebellar vermis (Dupont et al.,1994; Nawrot & Rizzo, 1998). Either of these two interpretations is consistent with theories of autism. The dorsal stream interpretation implies disordered processing in parietal lobe regions, which are associated with spatial attention and integration. This theory is consistent with other studies that reported parietal abnormalities in autism (Belmonte & Yurgelun-Todd, 2003; Haas et al.,1996; Muller, Kleinhans, Kemmotsu, Pierce, & Courchesne, 2003; Townsend, Harris, & Courchesne, 1996; Townsend et al.,2001). Dysfunction in these regions could be responsible for a local processing bias, which is hypothesized to occur in autism (Mottron, Bellevile, & Menard, 1999). Alternatively, neuroimaging studies have indicated that the cerebellar vermis is activated during motion perception tasks like those used in the present study (Dupont et al.,1994; Nawrot & Rizzo, 1998). Abnormalities in the cerebellar vermis have been reported by imaging and anatomical studies in autism (Courchesne, 1997). Cerebellar function has been associated with stereotyped behavior and reduced exploration in autism (Pierce & Courchesne, 2001), as well as the modulation of brainstem visual, auditory, and somatosensory neural activity (Courchesne, 1997). These two lines of evidence taken together suggest that cerebellar dysfunction may be responsible for the motion processing deficit.

Previous studies have interpreted coherent motion perception impairments as evidence for subcortical magnocellular dysfunction. However, this interpretation is inappropriate here, because the moving dot patterns have both high and low spatial frequency information (Skottun, 2000).

Contrast Sensitivity Tasks

Compared to TD children, children with autism showed worse contrast performance for pattern, but not flicker, detection tasks. Because pattern stimuli targeted parvocellular mechanisms (Merigan et al.,1991b; Merigan & Maunsell, 1993) and flicker stimuli targeted magnocellular mechanisms (Merigan et al.,1991a; Merigan & Maunsell, 1990), this pattern of findings is consistent with intact magnocellular and disrupted parvocellular function in autism.

Parvocellular dysfunction is associated with reduced psychophysical sensitivity for fine detail, which appears to contradict case reports of visual flooding for colors and details (Stehli, 1991; Williams, 1994) and subjective reports of super-sensitivity by our own subjects (“I pick up the small details”). However, parvocellular dysfunction in the direction of overload, combined with extreme distractibility, could result in poor pattern detection performance and account for subjective super-sensitivity. This interpretation is consistent with the local bias theory of autism, which predicts that autistic children would perceive very fine details on the screen as information rather than perceiving the grating as a Gestalt. This problem would not be expected to occur in flicker detection tasks, because the flickering gratings had all high spatial frequency (fine detail) information removed.

Relationship between Psychophysical Sensitivity and Subjective Visual Anomalies

A preliminary partial least squares analysis found a significant overall relationship between psychophysically measured visual sensitivity and subjectively reported visual hypersensitivity in autism. Increased frequency of subjective visual hypersensitivity was related to impaired function across all vision tests. The visual hypersensitivity was related most strongly to psychophysical test requiring unimpaired processing of high spatial frequency information. This pattern of results provides further support for the connection between parvocellular dysfunction, subjective visual hyper-sensitivity, and a severe high spatial frequency processing bias in autism.

Limitations and Future Directions

These preliminary findings are based on a small number of children. However, the effect sizes and direction of these effects indicate meaningful group differences that are consistent with reported subjective and psychophysical anomalies in autism. Based on these results, further investigation into the subjective distortions and visual system dysfunction of children with autism is warranted.

A limitation of the study was the inclusion of only high-functioning children with autism. Therefore, these results may not generalize to all children with autism. However, this is the first study to investigate relationships between self-reported subjective experiences and psychophysical performance. The present study’s hypotheses necessitated that participants have sufficient verbal and cognitive abilities to reliably report subjective anomalies and behavioral responses. Future studies targeting lower-functioning children could use parental report on the SIAPA to replace self-report on the SIAPA-CV, although this sacrifices the benefits of first-person accounts. Similarly, physiological measures of sensory function (e.g., electroencephalography or BOLD activation) could replace performance measures on psychophysical tests, although this sacrifices task-relevant behavioral responses.

Furthermore, to eliminate attention and fatigue effects, only a few conditions were used to assess psychophysical sensitivity. However, conditions were chosen to target specific hypotheses of visual pathway function. The present study’s findings indicate that parametric evaluation of psychophysical performance is warranted, especially for motion and pattern detection conditions. Finally, the present study has only shown a relationship between psychophysical performance and subjective anomalies for the visual modality. Future studies should investigate whether a similar relationship exists in other modalities.


The revised instrument is available from the authors upon request.



This research was supported by a National Defense Science and Engineering Graduate Fellowship, NIMH grants (RO1 MH62150-01; RO3 MH63112-01) awarded to Brian F. O’Donnell, the Indiana University Cognitive Science Program for Marcia B. Bockbrader, and the Indiana University Honors College for Rebecca A.O. Davis. We thank all the mothers and children who volunteered their time and their participation. We are grateful to Evan Davis for technical assistance regarding the psychophysical paradigms and P. Poskozim and W. Clevenger for assistance in testing participants. We acknowledge K. O’Bryan for her inspiration and guidance and E. Davis and E. Wilt for their continued support.

Copyright information

© Springer Science+Business Media, Inc. 2006