Atypical Perception in Autism: Characterizing Inference in Basic Perceptual Processing

Background: Perceptual atypicalities are a widely acknowledged but poorly understood feature of autism. Prevailing models, formulized in Bayesian terms, suggest that reduced top-down influences underlie atypicalities in perception. Within this framework, changes in bottom-up factors, if they exist at all, are considered quantitative, mostly involving changes in noise levels. Thus, testing has been generally limited to examining top-down factors without much recourse to potential underlying constraints of sensory processing.

Objectives: We have recently demonstrated violation of Weber’s law in autism, in the visual and tactic modalities, suggesting a modality-independent mechanism of abnormal stimulus encoding. Specifically, JNDs in individuals with ASD are not scaled with intensities, indicating deficits in the low-level calibration mechanism. Here, we examined whether this modulated mechanism of stimulus encoding is associated with the reduced effects of priors (e.g., anchoring, range effects). The typical pattern of increased biases for noisier measurements may not be evident in autism, where scalar variability (Weber ratio) does not seem to hold.

Methods: In Experiment 1, we extended our recent findings by demonstrating the violation of Weber’s law in autism also in the auditory domain, while manipulating the effect of perceptual anchoring. We examined discrimination thresholds for empty tones intervals using three standards duration. JNDs were compared between the different duration and between two conditions: (1) blocked presentation, in which an anchoring is created for each standard (2) random presentation, in which anchoring is weaker. Experiment 2, was design to directly examine the relationship between the precision of the sensory input and the utilization of priors. The height-width illusion was used to manipulate perceptual bias (i.e., higher rectangles are perceived narrower than lower ones), and level of contour blurriness (i.e., precision of the measurement). Greater biases were expected in neurotypicals (TD) for the noisier measurements.

Preliminary Results: In Experiment 1, 17/17 of our TD showed the expected increase of JNDs with increasing duration, and the effect of anchoring, resulting in higher JNDs for random presentation. Interestingly, 4/7 of our ASD did not show any scaling of JNDs with magnitude but a sub-group of 3/7 showed adherence to Weber’s law. Perceptual anchoring was smaller in this group and did not increase with duration to the same extent as in TD. Experiment 2 demonstrated substantial increase in the amount of the height-width illusion as a function of blurriness for TD, while ASD tended to show the illusion only when the input became noisier, and JNDs did not scaled with the increase in blurriness.

Conclusions: We replicated the violation of Weber’s law in the auditory modality (but only for a sub-group of our ASD participants), and demonstrated that individuals with autism exhibit utilization of priors under certain conditions. Clear susceptibility to perceptual illusions is seen in ASD; however, when precision in the sensory input is high, the otherwise mandatory prior effects do not bias perception.