Atypical Asymmetry and Sensory Processing in the Autistic Brain May Cause Impaired Language Development and Social Interactions

Autism, sometimes referred to as autism spectrum disorders or ASD, is a neurodevelopmental disorder with a threefold set of qualifying characteristics. Impairments in social interactions and communication, and repetitive and stereotyped patterns of behavior, interests, and activities must all manifest in order for a diagnosis of autism to be made (1). Deficits in sensory processing and integration may be a source of these classic social problems (2, 3). Anatomical studies, especially since the advent of structural MRI scanning, have shown numerous abnormalities in the developing and adult autistic brain (4), although consensus on these is yet to be reached. The left hemisphere of typically-developing humans is more developed than the right. However, there is abundant evidence to indicate atypical hemispheric asymmetry in multiple cortical and subcortical areas in the autistic brain (5, 6). In this review, the current state of studies in visual motion processing in autism will first be presented and discussed. This will be followed by discussion of the newest studies on bimodal sensory processing and brain symmetry, along with studies on the correlation of symmetry and communicative deficits in the autistic brain.

Visual Motion: Over- or Underrepresented?
Perception of visual motion can be essential to identifying shape and depth of objects, including other humans. Studies of perception of biological motion have shown this feature to be impaired in autistic children, which may in turn lead to impaired social recognition and interaction (7).

Studies have shown increased activation in human middle temporal cortex area (MT) (8), an area well accepted as the magnocellular/dorsal stream region where much visual motion processing takes place (9). In a functional MRI study of visual motion perception, Takarae et al. (10) used a motion after-effect paradigm to examine MT activation during passive viewing of moving stimuli. They found that the autistic group showed increased MT activation as compared to typically-developing (TD) controls, reported longer motion after-effects, and had a slower decay of MT hemodynamic response, which is a sign of longer experience of motion after-effect. They concluded that the reduced response decay, and consequently the prolonged after-effect, resulted from abnormally functioning or weakened inhibitory interneurons in the area. Their psychophysical studies showed no difference in threshold levels, so the authors speculated that increased MT activity indicated that individuals with autism are more sensitive to visual motion above their absolute detection threshold.

In an apparent contradiction to this study, Koldewyn et al. also conducted an fMRI and psychophysical study on autism, and found deficits in not only biological motion perception, but also in more basic motion perception (11). Their findings are supported by previous studies that also report deficits in dorsal stream visual processing (11). Koldewyn et al. scanned the brains of adolescent autistic and TD participants while presenting stimuli featuring motion coherence (dots moving left or right), and biological motion coherence (a point-light man walking left or right). They performed psychophysical analyses to generate perceptual threshold data. In direct contradiction to the Takarae study, this study’s psychophysical findings showed higher thresholds in both types of motion perception (p<0.04 in both cases), and decreased activation of MT. They also found that TD subjects responded bilaterally in MT, while autistic subjects’ responses were right-lateralized. Finally, in support of and as a supplement to previous studies, they found that autistic subjects did not show the bilateral parietal response to biological motion that the TD subjects showed (7). However, they did not go as far as designing a task to test the negative effect that impaired biological motion perception is supposed to have on social interactions. The authors reached the conclusion that some individuals with autism may have a selective deficit in visual motion processing, probably due to weak or inefficient connections along the dorsal/magnocellular path. These conclusions seem to be at odds with the above conclusion that inhibitory interneurons are weakened in the dorsal stream path, resulting in an overactivation of MT in response to visual motion.

How can the contradictory data of these two studies be resolved? The simplest answer is to consider internal errors in methodology or calculation algorithms, especially for the psychophysical data. Another answer lies in the type of task: while the first involved only passive viewing, the second involved a forced-choice task that may have involved more attention in autistic subjects than was properly controlled for. Previous studies on motion aftereffects have shown that visual or auditory attention to another task will significantly decrease activation in MT in comparison to passive viewing (12). Finally, differences in subjects’ developmental stages may have had an effect on activation in MT. As we shall see later, the developmental process of autistic children and adolescents is highly atypical and has yet to be fully characterized.

Brain Activation in Visual Motion Processing is Right-Lateralized
Although the dependability of the previous study’s functional data has been called into question, it is just the first in a series of studies pointing to this conclusion. In yet another fMRI study, Carmean et al. considered the question of the extent of involvement of the fusiform gyrus (FG) in processing human faces, hypothesizing a deficit in complex object processing preceding face processing (13). Their task involved stimulus matching in face, complex object, and abstract control categories. The autistic group showed activation only in the right FG in response to faces, which was weaker than the bilateral activation in the TD group (p<0.01). In response to complex objects, the autistic group showed bilateral parahippocampal gyrus (PG) activation (p<0.005). This bilateral PG activation again appeared in the object-face contrast, while the TD group showed only right PG activation (p<0.005). The autistic brain’s right-restricted FG response points to deficits caused by or reflected in decreased leftward asymmetry. The complex object stimuli consisted of images of objects such as trains and cars, and the PG activation observed in complex objects over faces and control may cause or be caused by the interest that autistic individuals show in such mechanical devices. The authors concluded that further studies are needed in complex object processing to discern its workings in the autistic brain, which is indeed a recurring theme in autism research.

Visual Domination in Visual-Auditory Integration
While visual perception is perhaps the most important sense we have, auditory perception is equally important in social interaction as it pertains to communication and language. Evidence of reduced long-distance connectivity in autism may suggest impaired sensory integration, especially across sensory modalities (14). Lack of proper integration across modalities results in fragmented information received by higher-level processes, such as attention and language. In order to study bimodal sensory processing and integration, Grenesko et al. looked at facilitation effects, which in typical adults manifest as reinforced information that is received through multiple sensory modalities (15). Theirs was a behavioral and fMRI study with a forced-choice task between high or low as presented in visual (high or low dot), auditory (high or low tone), and bimodal (dot and tone) conditions. Their expectations for reduced facilitation effects in autistic subjects were not fulfilled: both groups undifferentially experienced faster reaction times and greater accuracy in the bimodal condition. The authors observed that, in the autistic group, the facilitation statistics were biased by the poor accuracy (as only correct responses were included in analysis) and high variability in the auditory modality.

Closer examination of their response time graphs reveals that the autistic group’s visual and bimodal response times look very similar, as compared to the TD group’s response times in those conditions. This trend was not reported as significant, which is also attributable to only correct responses being used for this analysis. Other ways of perhaps amplifying this trend include adding subjects to the study, or having subjects verbalize their responses. In the Stroop task, verbalized responses enhance differentiation in response times (16), though this was not done because of the authors’ intention of taking the study and task into fMRI, where keeping still is important. This possible trend would suggest that some of the autistic subjects (because variability within the group was high) depended mostly on visual stimuli, even in the bimodal condition. What would have been interesting to see is whether those with better auditory perception would also have had more typical social behaviors, as determined by standardized diagnostic tests for autism spectrum disorders.

More concrete evidence for visual domination comes from Westerfield et al., in their simply-titled study, “Auditory-visual integration in autism” (17). The authors had subjects perform a behavioral task very similar to the Stroop task, by naming the category of the image presented (animal or instrument) while ignoring the auditory cue, which could either match or mismatch with the visual stimulus. Two variations were tested: spatial and temporal, in which spatial or temporal proximity of stimuli varied. While controls were negatively affected by bimodal incompatibilities, in a typical Stroop-like effect, autistic subjects were unaffected by incompatibilities in the spatial condition, demonstrating their ability to discard auditory information, whether or not it is congruent with visual information. This again reflects the apparent dominance of visual information in the autistic brain. In the temporal condition, controls were affected by incompatibilities only when proximity was 50 ms or less, whereas autistic subjects were affected by auditory tones only when proximity was 150 ms or more. The authors suggested that integration of auditory and visual information in autism is not a bottom-up process, but rather a top-down process that occurs later in the processing stream as an explanation for the temporal results. This explanation also accounts for the dominance of visual information: auditory information is integrated into the autistic person’s global percept too late for it to interfere with or enhance visual information, but does interfere with the production of a response when presented after a delay.

Alternative or complementary evidence for visual domination in the autistic brain comes from a study done by Martien et al. (18). This study began with the theory that synchronization of EEG oscillations across systems is a mechanism for binding and integrating cognitive processes (19). They took resting EEG measurements from, unfortunately, only one autistic adult and a matched TD control in two sensory states: eyes open and eyes closed. In the eyes open state, as compared to the control, the autistic subject lacked interhemispheric beta coherence (12+ Hz) between the temporal lobes and between temporal and frontal lobes bilaterally. The autistic also displayed increased delta (1-4 Hz) coherence between parietal and frontal areas and between the right posterior temporal and left hemisphere, which is not normally seen in awake, behaving adults, and which the authors suggested was a “compensatory reorganization,” likely for language processing in autistic individuals. The autistic and control showed no differential oscillatory activation in the eyes closed state. Clearly, this study can be regarded as little more than a pilot, but the potential support a follow-up study would provide to theories of visual domination and its effects on language processing is important. Viewing of the surroundings, even in the absence of a task, is enough to disrupt a highly active synchronization between temporal lobes — including auditory and language processing areas, as well as complex object processing, and between temporal lobes and frontal areas, where planning and movement coordination take place, again possibly inhibiting language processing and production. This study takes questions of polysensory integration and turns them into questions of polysensory interference, and the possible effects on higher-level cognitive processes. It also broaches a functional asymmetry problem of coordinating the right hemisphere with the underactive left hemisphere of the autistic brain, and supports previous evidence that long-range neural connectivity in the autistic brain is impaired (20).

Atypical Asymmetry Correlates with Social and Communicative Impairments
Observations of the autistic brain extend to anatomical and developmental anomalies. Autistic children experience increased initial neuronal growth and development over typical children (21). This overgrowth is checked by a cessation earlier than that in typical children, but whether the adult or adolescent autistic brain is larger or smaller, and in what specific regions this differentiation occurs, is still debated (22, 23). In addressing this question, Wallace et al. conducted a thorough structural MRI study comparing specific brain regions of autistic (mean age = 16.14) and typical (mean age = 17.19) adolescents (24). Their data show reduced gray and white matter volume in the adolescent brain. They also found decreased leftward asymmetry in the frontal gray and white matter in the autistic brain, consistent with recent studies (25). Importantly, they found that social impairment symptoms, as evaluated by parental and expert diagnostics, correlated positively with reduced leftward asymmetry.

A stronger correlation between typical symmetry and typical behavior, one that even implies causation, comes from a longitudinal study of language ability and early amygdale development. Previous studies suggest that amygdale size is associated with the ability to distinguish emotions in others, and is even predictive of the level of social impairment in children with disorders like autism (26). Ortiz-Mantilla et al. performed structural MRI scans on typically developing six- and 12-month-old infants after observing that no studies of the amygdale in autism in infants had been conducted (27). Cognitive and language assessments were also given to these children at six, 12, and 24 months of age. At six, but not at 12 months, left amygdalae were larger than right. Conversely, at 12, but not at six months, right hippocampi were larger than left. The right-lateralized autistic brain may experience a deviation between these stages, where amygdalae even out in size, and hippocampi differentiate. Correlations also emerged between the size of the left amygdala at six months of age and expressive language abilities at 24 months. As the authors concluded, this predictive correlation occurs so early in development, when emotional content of language interaction has such a profound effect on language and social development, it could even reflect causation of autistic social and communicative characteristics.

Auditory Information May Not Be Processed Quickly Enough for Typical Language Ability to Manifest
Evidence of atypical asymmetry-caused sensory deficits having a negative or even partially causative effect on language ability has been mostly inferred. The results of one study seem to link all three aspects together. Schmidt et al. (28) performed a magnetoencephalography (MEG) study of rapid temporal processing (29) (RTP) in children with autism. Two tones were presented to subjects in rapid succession, separated by a range of less than 850 ms, in a forced-choice task (did tones proceed high-low or low-high?), and recorded MEG activity at 100 ms post-stimulus. In the autistic group, increased language ability as measured standardized tests positively correlated with the presence of rapid temporal processing at a separation of 150 ms. This is the same time frame found by Westerfield et al. in their analysis of auditory-visual integration, and was seen as evidence that auditory processing is delayed when present in conjunction with visual processing (18). Schmidt et al. also found the autistic group lacked typical asymmetry in the location of auditory cortex, and that increased asymmetry positively correlated with increased language ability. The authors suggested that early auditory processing in the left hemisphere determines the level of language ability, and that language acuity is mediated by neural responses in the left hemisphere to tones. Children with autism have been shown to have atypical left-brain development at a very early age, which combined with atypical and delayed auditory processing may result in the diminished language development characteristic of autism (30). This may act in conjunction with or result in reduced leftward hemispheric volume and reduced functional integration between left and right hemispheres in the autistic brain.

Conclusions
In the autistic brain, visual motion, whether over- or underrepresented, is right-lateralized and interferes with auditory processing and system integration. The atypical asymmetry of the autistic brain has effects on sensory processes, and may result in the classic social and communicative impairments of individuals with autism. Future directions in this field must include studies asking whether atypical sensory perception causes the social and communicative impairments of autism. Longitudinal studies of sensory perception in infants with autism will produce information on the predictive abilities of sensory perception have on social development, and determine the extent of causation.

The study of autism in human subjects is unavoidably complicated: issues of availability of subjects and severity of condition invariably arise. While studying severely impaired autistic subjects would yield the most differential information from typically developing controls, these individuals are also the most difficult to work with. The development of adequate animal models has yet to be achieved, since behavioral criteria cannot be enough — how can reduced leftward asymmetry and specific language deficits be represented in an animal that does not typically have these features?

Nevertheless, headway is being made with rat and mouse models of autism (31, 32, 33). Genetic studies in homeobox genes — genes responsible for anatomical layout in the developing animal — as well as other studies in myelination and asymmetrical development causes would be facilitated with a dependable animal model. Treatment and therapy options guided more towards sensory training have been shown to be effective in the past (34). However, with current and developing information about the details of sensory deficits in autism, therapies can be modified to target these specific deficits, with the aim of improving quality and prognosis of social interactions in individuals with autism.

References
1. American Psychological Association, Diagnostic and statistical manual of mental disorders DSM-IV (American Psychological Association, Washington DC, ed. 4th, 1994).
2. G. T. Baranek, F. J. David, M. D. Poe, W. L. Stone, L. R. Watson, J. Child Psychol. Psychiatry 47, 591 (2006).
3. S. Dakin, U. Frith, Neuron 48, 497 (2005).
4. E. Courchesne et al., Neurology 57, 245 (2001).
5. D. C. Rojas, S. L. Camou, M. L. Reite, S. J. Rogers, J. Autism Dev. Disord. 35, 479 (2005).
6. H. C. Hazlett, M. D. Poe, G. Gerig, R. G. Smith, J. Piven, Biol. Psychiatry 59, 1 (2006).
7. R. Blake, L. M. Turner, M. J. Smoski, S. L. Pozdol, W. L. Stone, Psychol. Science 14, 151 (2003).
8. F. Castelli, C. Frith, F. Happé, U. Frith, Brain 125, 1839 (2002).
9. M. Livingstone, D. Hubel, Science 240, 740 (1998).
10. Y. Takarae, N. J. Minshew, B. Luna, J. Sweeney, Poster Presentation at 37th Annual Meeting of the Society for Neuroscience, San Diego, CA, 2007.
12. J. Spencer et al., Neuroreport 11, 2765 (2000).
13. R. A. Berman, C. L. Colby, Cognitive Brain Research 14, 64 (2002).
14. V. Carmean, B. A. Corbett, C. Carter, Poster Presentation at 37th Annual Meeting of the Society for Neuroscience, San Diego, CA, 2007.
15. R. A. Müller, Ment. Dev. Disab. Res. Rev. 13, 85 (2007).
16. E. L. Grenesko et al., Poster Presentation at 37th Annual Meeting of the Society for Neuroscience, San Diego, CA, 2007.
17. G. M. Redding, D. A. Gerjets, Percept. Mot. Skills. 45, 11 (1977).
18. M. A. Westerfield, P. T. Lai, G. L. Smith, A. Lincoln, J. Townsend, Poster Presentation at 37th Annual Meeting of the Society for Neuroscience, San Diego, CA, 2007.
19. K. Martien et al., Poster Presentation at 37th Annual Meeting of the Society for Neuroscience, San Diego, CA, 2007.
20. W. Singer, Neuron 24, 49 (1999).
21. S. Baron-Cohen, M. K. Belmonte, Annu. Rev. Neurosci. 28, 109 (2005).
22. J. E. Lainhart et al., J. Am. Acad. Child Adolesc. Psychiatry 36, 282 (1997).
23. J. Piven et al., Am. J. Psyc. 152, 1145 (1995).
24. G. M. McAlonan et al., Brain 128, 268 (2005).
25. G. L. Wallace, J. A. Silvers, A. Martin, J. N. Giedd, Poster Presentation at 37th Annual Meeting of the Society for Neuroscience, San Diego, CA, 2007.
26. L. De Fossé et al., Annals of Neurology 56, 757 (2004).
27. G. Munson et al., Arch Gen Psychiatry 63, 686 (2006).
28. S. C. Ortiz-Mantilla, M.S. Choe, J. Flax, P. E. Grant, A. A. Benasich, Poster Presentation at 37th Annual Meeting of the Society for Neuroscience, San Diego, CA, 2007.
29. G. L. Schmidt, L. G. Blaskey, M. M. Rey, S. E. Levy, T. P. L. Roberts, Poster Presentation at 37th Annual Meeting of the Society for Neuroscience, San Diego, CA, 2007.
30. J. E. Oram Cardy, E. J. Flagg, W. Roberts, J. Brian, T. P. L. Roberts, NeuroReport 16, 329 (2004).
31. L. Kanner, Nervous Child 2, 217 (1943).
32. P. M. Rodier, J. L. Ingram, B. Tisdale, V. J. Croog, Reprod. Toxicol. 11, 417 (1997).
33. S. S. Moy et al., Behav Brain Res. 176, 4 (2007).
34. V. J. Bolivar, S. R. Walters, J. L. Phoenix. Behav. Brain Res. 176, 21 (2007).
35. S. Bettison. J. Autism Dev. Disord. 26, 361 (1996).