CYBER PSYCHOLOGY & BEHAVIOR Volume 5, Number 3, 2002 Mary Ann Liebert, Inc.

Atypical Face Gaze in Autism

CHERYL TREPAGNIER, Ph.D., MARC M. SEBRECHTS, Ph.D., and REBECCA PETERSON, M.A.

ABSTRACT An eye-tracking study of face and object recognition was conducted to clarify the character of face gaze in autistic spectrum disorders. Experimental participants were a group of individuals diagnosed with Aspergers disorder or high-functioning autistic disorder according to their medical records and confirmed by the Autism Diagnostic InterviewRevised (ADI-R). Controls were selected on the basis of age, gender, and educational level to be comparable to the experimental group. In order to maintain attentional focus, stereoscopic images were presented in a virtual reality (VR) headset in which the eye-tracking system was installed. Preliminary analyses show impairment in face recognition, in contrast with equivalent and even superior performance in object recognition among participants with autism-related diagnoses, relative to controls. Experimental participants displayed less fixation on the central face than did control-group participants. The findings, within the limitations of the small number of subjects and technical difficulties encountered in utilizing the helmet-mounted display, suggest an impairment in face processing on the part of the individuals in the experimental group. This is consistent with the hypothesis of disruption in the first months of life, a period that may be critical to typical social and cognitive development, and has important implications for selection of appropriate targets of intervention. INTRODUCTION pacity for social connection,2 impaired ability to shift attention,3 inability to represent and manipulate complex information,4 inability to conceive of others mental states,5 an impairment of central coherence, the ability to extract organizing principles,6 and more multifaceted cognitive deficits.7 Research has increasingly pointed to reductions in the connectivity among brain regions that typically work as integrated systems.810 It is important to note, however, that functional and structural brain differences can reflect as well as cause atypical behavior.11 Gaze-disruption hypothesis A hypothesis proposed by the first author12 suggests that disruption of social attention dur-

is a behaviorally defined condition, diagnosed by severe impairments in social and communicative domains, and the presence of repetitive behaviors and/or restrictive, narrow interests.1 Aspergers disorder, considered a less severe variant of autistic disorder, is similar to high-functioning autistic disorder, such that there remain unresolved questions about whether they should be treated as distinct diagnostic categories. For convenience, autistic disorder and Aspergers disorder will be referred to here as autistic spectrum disorder, or simply, autism. Accounts of the core psychobiological deficit in autism have included impairment in the caUTISTIC DISORDER

Department of Psychology, The Catholic University of America, Washington, D.C.

213

214

TREPAGNIER ET AL.

ing the early months of life due to underlying neuropsychiatric disorder may be sufficient to bring about autistic development. There is increasing evidence of elevated incidence of neuropsychiatric disorder of the anxiety disorders/depressive disorders type among individuals with autism and their first-degree relatives (prior to the birth of the disabled individual). Very early onset or precursor symptoms may disrupt infant social interaction by impairing the infants ability to tolerate the overwhelming excitement of social stimuli.13 Individuals born blind because of cataracts, whose vision is restored by surgery even as early as age 2 months, have a persistent impairment in configurational face processing.14 In the presence of intact vision, impaired configurational face processing may be the marker for a failure to engage in very early social interaction, experience that is critical to normal social, cognitive, and emotional development. Face attention and interpretation By 1 year of age, typically developing infants routinely monitor their adults facial expressions, and modify their own actions and responses accordingly. In contrast, children and many adolescents and adults with autism do not.1517 Children with autistic disorder did not look at faces to discern intention,16 and adults with Aspergers were more likely to look at someones face when they themselves were talking to that person, rather than looking at someone who was speaking to them.18 A recent study of children with autism seen at the Yale Child Study Center over a period of several years found that all were impaired in face recognition relative to controls.19 In functional imaging studies, the fusiform gyrus of controls was activated during performance of face-recognition tasks; this was not the case for individuals with autism carrying out the same tasks.20 In a pilot study using eye-tracking during a face recognition task, a high-functioning young adult with autism spent much less time than controls looking at eyes and the central face. 21 This study of face and object recognition extends those preliminary findings and makes two predictions: that individuals with autistic

spectrum disorders would be impaired in face recognition, but not in object recognition or recognition of inverted faces (in which configurational information is not preserved), relative to controls; and that persons with autistic spectrum disorder would spend less of their time than controls fixating on the central face region (from the eyebrows down to and including the mouth, with width determined by the outer corners of the eyes).

MATERIALS AND METHODS Participants were individuals with autism spectrum diagnoses according to their medical records and confirmed by the Autism Diagnostic InterviewRevised (ADI-R),22 carried out by the first author, and a group of controls of similar age. Verbal intelligence and nonverbal intelligence were not controlled for; however, all experimental participants were very high-functioning, including one individual with a graduate degree and a successful professional career, and successful students and college-bound adolescents. Participants were asked to look at and remember a series of images of faces and objects presented stereoscopically in a virtual reality (VR) head-mounted display equipped with an ISCAN eye-tracking system, and then to view subsequently a longer series of images and indicate which they had already seen. Four blocks of trials were presented sequentially, each with eight acquisition trials (images to learn) followed by 16 recognition trials (images to evaluate as already seen or not). In each acquisition set, participants were shown four faces and four objects. The acquisition images of the third and fourth blocks included inverted as well as upright images. Each image was presented for 4 sec, with a 4-sec interstimulus interval. Each recognition set included eight previously presented images and eight foils. Participants were asked to indicate by saying yes or no whether or not they had seen the image before. Recognition stimuli remained present until the participant responded. Point of regard (POR) was sampled at 60 Hz, and a videotaped recording was acquired, at

ATYPICAL FACE GAZE IN AUTISM

215
TABLE 1. RECOGNITION ACCURACY Faces Block 1 Block 2 Block 3 Block 4 All upright All inverted
ap bp

approximately 30 frames/sec, showing a cursor representing the PORs superimposed on a two-dimensional image of the stimulus. Temporal and spatial thresholds were set to define fixation as a gaze of at least 100 msec in length, within approximately 1 degree of visual angle of the mean location of the points. Further constraints were introduced into the fixationdetection algorithm to reduce the effects of spurious points and dropped tracking. Fixation locations were examined with respect to predefined regions of the face stimuli. Data are reported here from five individuals with autism spectrum disorder (one female, mean age 18.4 years), and six controls (two females, mean age 19.5 years). These were individuals who responded to a call for subjects disturbed via the Autism Society of America chapter newsletters and local physicians treating individuals with autism. Individuals with severe disorder self-excluded when they or their parents were informed that the study involved wearing a headset that covered much of the front of the face. Other candidates and some subsets of the data from the participants were excluded for the following reasons: failure to meet criterion for autism on the social subscale of the ADI-R (one participant); technical problems in the data-acquisition program (some of the data from two participants); error (some of the data from one participant); familiarity with one or more of the people whose images were used (exclusion of three individuals; exclusion of half the data from one participant); and age (one participant).

Objects n.s. n.s. A > Ca A > Ca n.s. A > Cb

n.s. C > Aa C > Aa n.s. C > Ab n.s.

< 0.05, one-tailed. < 0.1. n.s., not significant; C, control group; A, autistic group.

tism participants were slower overall than controls to make the shift from their starting point to active examination of the image (p < 0.05, one-tailed). Where differences reached significance, recognition accuracy predictions were confirmed (Table 1). Results for allocation of gaze to the central face region for 4 sec of exposure (Table 2) are consistent with predictions. The same comparison for the initial second out of the four did not reach significance. DISCUSSION When between-group differences in recognition accuracy reached significance, they were in the expected direction, attesting to the experimental groups impairment in face recognition. At the same time, their relatively superior performance in recognizing objects does not support an underlying visual perceptual deficit or short-term visual memory impairment. Foils for faces were selected to resemble the corresponding target images; for example, for a target female with light skin and long, straight blonde hair, there was a foil who was also a light-complexioned female with long, straight blonde hair (i.e., the most salient features, such as hair length and color, and skin color, were not helpful). Foils for objects such as a stapler or a chair were another stapler and another chair, with the difference residing in features of color or detail. The superior performance of the autism participants with objects, along with their reduced success with faces, points to a face-recognition impairment, possibly an impairment in processing configurational information.

RESULTS Participants initial fixations occurred in the general area at which they had been looking as the image appeared. Since difficulty in shifting attention is expected to the autism population,3 data visualizations for all files (up to 96 per participant) were reviewed to determine start time of the first fixation after shift. Examination revealed cases of artifact appended to the beginning of several files, resulting from technical difficulties in managing the buffer size of the Onyx workstation. Verifiable artifacts were removed, and as expected, au-

216
TABLE 2. PERCENT OF FIXATIONS ON CENTRAL FACE R EGION

TREPAGNIER ET AL.

Blocks 1 and 2 1 and 2 3 and 4 3 and 4

a Upright

Task Acquisition (16 images) Recognition (32) Acquisitiona (8) Recognition (32) images only.

Autistic 50% 46% 55% 57%

Control 73% 78% 85% 88%

t scores (p < 0.05, one-tailed) t(7) = 2.14 t(6) = 2.16 t(9) = 2.60 t(8) = 3.01

Because of difficulties in achieving calibration of the VR helmet-mounted eye-tracking device, and the additional barrier it represented to enrollment of individuals with autistic spectrum disorders, use of the headset-mounted system was discontinued. The study is about to resume using a desk-top eye-tracking device and stimuli presented on a large monitor. An Integraph computer is now being used to acquire the data, and the program has been thoroughly tested to assure that no buffermanagement artifacts occur. It remains to be seen whether the nonimmersive, nonthreedimensional presentation of images will evoke similar responses.

ACKNOWLEDGMENTS Partial support for this work was provided by the Rehabilitation Engineering Research Center on Telerehabilitation, which is funded by the National Institute on Disability and Rehabilitation Research of the U.S. Department of Education under grant no. H133E980025. REFERENCES
1. Filipek, P.A., Accardo, P.J., Ashwal, S., et al. (2000). Practice parameter, screening and diagnosis of autism, report of the Quality Standards Subcommittee of the American Academy of Neurology and the Child Neurology Society. Neurology 55:468479. 2. Fein, D., Pennington, B., Markowitz, P., et al. (1986). Toward a neuropsychological model of infantile autism, are the social deficits primary? Journal of the American Academy of Child Psychiatry 25:198212. 3. Mundy, P., Sigman, M., Kasari, C. (1990). A longitudinal study of joint attention and language development in autistic children. Journal of Autism and Developmental Disorders 20:115128. 4. Minshew, N.J., Goldstein, G., Siegel, D.J. (1997). Neuropsychologic functioning in autism, profile of a complex information processing disorder. Journal of the International Neuropsychological Society (JINS) 3: 303316. 5. Baron-Cohen, S., Jolliffe, T., Mortimore, C., et al. (1997). Another advanced test of theory of mind, evidence from very high functioning adults with autism or asperger syndrome. Journal of Child Psychology and Psychiatry 38:813822. 6. Happe, F., Briskman, J., Frith, U. (2001). Exploring the cognitive phenotype of autism: weak central coherence in parents and siblings of children with autism: I. Experimental tests. Journal of Child Psychology and Psychiatry 42:299307. 7. Frith, U. (1996). Cognitive explanations of autism. Acta Paediatrica Supplement 416:6368.

CONCLUSION The results of this study must be regarded as provisional, in view of the technical problems that taint the initial second of some of the files, the calibration difficulties encountered, and the small number of participants. If these results are borne out, they indicate a relationship between the face-recognition difficulties of autism and maladaptive allocation of face gaze. This suggests that intervention directed at increasing attention to the informative area of the face, and teaching how to interpret faceborne information, may be beneficial, all the more so if provided early in development. These findings are also consistent with the hypothesis of failure to acquire configurational face-processing skills, pointing to very early disruption in the infants availability to social experience.

ATYPICAL FACE GAZE IN AUTISM

8. Waterhouse, L., Fein, D., Modahl, C. (1996). Neurofunctional mechanisms in autism. Psychological Review 103:457489. 9. Aylward, E.H., Minshew, N.J., Goldstein, G., et al. (1999). MRI volumes of amygdala and hippocampus in nonmentally retarded autistic adolescents and adults. Neurology 53:21452150. 10. Just, M. (2001). Brain activation in language comprehension and executive planning in autism: distribution of activation and patterns of functional connectivity. Presented at the Annual Meeting of the NICHD/NIDCD Collaborative Programs of Excellence in Autism (CPEA), Yale University, New Haven, CT, April 30 to May 2. 11. Byl, N.N., & Melnick, M. (1997). The neural consequences of repetition: clinical implications of a learning hypothesis. Journal of Hand Therapy 10:160174. 12. Trepagnier, C. (1996). A possible origin for the social and communicative deficits of autism. Focus on Autism and Other Developmental Disabilities. 11:170182. 13. Kalmanson, B. (1992). Diagnosis and treatment of infants and young children with pervasive developmental disorders. Zero to Three 13:2126. 14. Legrand, R., Mondlock, C.J., Maurer, D., et al. (2001). Early visual experience and face processing. Developmental Science 4:233251. 15. Klin, A., Volkmar, F.R., & Sparrow, S.S. (1992). Autistic social dysfunction: some limitations of the theory of mind hypothesis. Journal of Child Psychology and Psychiatry 33:861876. 16. Phillips, W., Baron-Cohen, S., & Rutter, M. (1992). The role of eye contact in goal detection: evidence from normal infants and children with autism or mental handicap. Development and Psychopathology 4:375383. 17. Parritz, R.H., Mangelsdorf, S., & Gunnar, M.R. (1992). Control, social referencing, and the infants

217
appraisal of threat. In: Feinman S, ed. Social referencing and the social construction of reality in infancy. New York: Plenum Press, pp. 209228. Tantam, D., Holmes, D., Cordess, C. (1993). Nonverbal expression in autism of the Asperger type. Journal of Autism and Developmental Disorders 23: 111134. Klin, A., Sparrow, S.S., de Bildt, A., et al. (1999). A normed study of face recognition in autism and related disorders. Journal of Autism and Developmental Disorders 29:499508. Schultz, R.T., Gauthier, I., Klin, A., et al. (2000). Abnormal ventral temporal cortical activity during face discrimination among individuals with autism and Asperger syndrome. Archives of General Psychiatry 57:331340. Trepagnier, C., Gupta, V., Sebrechts, M.M., et al. (2000). How does he look?tracking autistic gaze. In: Proceedings of the RESNA 2000 Annual Conference pp. 2830. Lord, C., Rutter, M., & Le Couteur, A. (1994). Autism Diagnostic InterviewRevised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. Journal of Autism and Developmental Disorders 24: 659685.

18.

19.

20.

21.

22.

Address reprint requests to: Cheryl Trepagnier, Ph.D. Rehabilitation Engineering Service National Rehabilitation Hospital 102 Irving St. NW Washington, DC 20010 E-mail: Cheryl.trepagnier@medstar.net