LAUSR

Based on our expectations about material properties we can implicitly predict an object’s future states, e.g. a wine glass falling down will break when it hits the ground. How these expectations affect relatively low level perceptual decisions, however, has not been systematically studied previously. To seek an answer to this question we conducted a behavioral experiment using animations of various familiar objects (e.g. key, wine glass etc.) freely falling and hitting the ground. During a training session participants first built expectations about the dynamic properties of those objects. Half of the participants (N=28) built expectations consistent with our daily lives (e.g. a key bounces rigidly), whereas the other half learned an anomalous behavior (e.g. a key wobbles). This was followed by experimental sessions, in which expectations were unmet in 20% of the trials. In both training and experimental sessions, participants’ task was to report whether the objects broke or not upon hitting the ground. Critically a specific object always remained intact or broke, only the manner with which it did so differed. For example, a key could wobble or remain rigid, but it never broke. We found that participants’ reaction times were longer when expectations were unmet even when those expectations were anomalous and learned during the training session. Furthermore, we found an interplay between long-term and newly learned expectations, which could be predicted by a Bayesian updating approach. Overall, our results show that expectations about material properties can have an impact on relatively low-level perceptual decision making processes.