Post-Decision Wagering Shows That ASD Subjects Appear to Be Unusually Confident in Their Choices During a Motion Discrimination Task

Background: Several studies have found that individuals with autism spectrum disorder (ASD) exhibit impairments in discriminating the direction of motion in a visual display. Last year, we reported similar results, but also showed that ASD subjects performed as well as typically developing (TD) subjects when required to delay their response. These findings challenged the idea that the impairments in motion processing of ASD-diagnosed individuals are due to abnormalities in the early stages of visual processing, and instead suggest that the impairments are related to abnormalities in the decision process underlying the behavioral response. Objectives: To investigate this alternative explanation, we have now designed a task that probes the subjects' confidence in their judgments by asking them to wager on their choices in the motion discrimination task. Methods: Subjects (ASD and TD children, aged 10 to 18) performed a 2-AFC visual motion discrimination task with post-decision wagering. On each trial, subjects viewed a stochastic motion stimulus for 750 ms (8º diameter patch, 8º above a fixated spot, variable strength rightward or leftward motion), and indicated their motion direction judgment by pressing either the left or right button on a response pad. They then were asked to place a bet, either high or low, on the correctness of their choice, by pressing either the upper or lower button on the same response pad. We ran two versions of the task that were identical in all respects except that the payoff values were adjusted so that the optimal betting strategy changed. In one version, the best strategy was to always wager high, whereas in the second version, the best strategy was to bet low when the correct answer was uncertain. Subjects received visual and auditory feedback about their choices after each trial and received a monetary reward for total points won. Results: As we found previously, ASD and TD subjects had similar thresholds for discriminating the direction of motion when required to delay their response. However, their betting strategies were very different. TD subjects tended to bet high when the motion coherence was high and bet low when the motion coherence was low, possibly reflecting a tendency to be risk averse. AD subjects, in contrast, tended to bet high regardless of motion coherence, and thus tended to not take the quality of the sensory stimulus into account when setting their wager. Moreover, when the payoff values were changed in the second version of the task, TD subjects modified their betting strategy as expected, increasing their frequency of low bets. ASD subjects wagered the same way in both versions of the task. Conclusions: These results show that ASD subjects appear to be unusually confident in their choices, even when the sensory evidence is weak and the correct choice is uncertain. These findings support the idea that impairments in sensory discrimination observed in ASD are due, at least in part, to abnormalities in the decision-making process, perhaps reflecting an inability to accurately monitor the quality of signals prior to generating a behavioral response.