Autistic traits lower the flexibility of information sampling: insights from a two-stage decision model

Abstract

Information sampling can reduce uncertainty in future decisions but is often costly. To maximize expected gain, people need to balance sampling cost and information gain. Suboptimal information sampling is observed in many core symptoms of autism spectrum disorder (ASD) and may be a general impairment in ASD. Here we aimed to understand how autistic traits influence the optimality of information sampling and to identify the particularly affected cognitive processes. Healthy human adults of varying autistic traits performed a probabilistic inference task, where they drew samples sequentially to accumulate evidence for their inference and could stop sampling at any time. Each additional sample would increase the probability of correct inference but also incur a fixed cost. We manipulated the cost and evidence associated with each sample and compared participants’ performance to optimal choices. We found that, compared to their peers, participants with high autistic traits had a lower efficiency of winning rewards in the high cost and low evidence condition due to oversampling. Computational modeling of participants’ sampling choices revealed two decision stages: When the first stage does not reach the choice of stopping sampling, a second stage is probabilistically involved, which provides a second chance to stop sampling. We found that autistic traits did not influence the choices in either stage but specifically modulated the participant’s probability to recruit the second decision stage (“second-thought probability”). In particular, participants with high autistic traits were less flexible in adjusting their second-thought probabilities with the increase of cost. To conclude, high-levels of autistic traits are associated with suboptimal information sampling and such suboptimality implies a failure of adjusting the strategic complexity (with or without second thought) appropriately with sampling cost. Author Summary Children with autism can spend hours practicing lining up toys or learning all about cars or lighthouses. This kind of behaviors, we think, may reflect suboptimal information sampling strategies, that is, a failure to balance the gain of information with the cost (time, energy, or money) of information sampling. We hypothesized that suboptimal information sampling is a general characteristic of people with autism or high level of autistic traits. In our experiment, we tested how participants may adjust their sampling strategies with the change of sampling cost and information gain in the environment. Though all participants were healthy young adults who had similar IQs, those with high level of autistic traits had a lower efficiency of winning rewards due to oversampling under high sampling cost. Further computational modeling revealed a two-stage decision process, with the second stage being optional and serving as a second chance to stop sampling. Autistic traits did not influence decisions in either stage but lowered the strategic flexibility of using the second stage to reduce costly sampling. In other words, people should increase their “second thoughts” with the increase of sampling cost but those with high autistic traits fail to do it well.