High trait anxiety is associated with improved state inference

摘要

In aversive contexts, anxiety has previously been associated with an increased probability of fear relapse. It remains unclear whether this is due to faster learning or context-specific learning. We investigated whether high trait anxious (TA) individuals learn gradually or switch between multiple environmental states (i.e. contexts) in a probabilistic aversive learning task. Across three experiments, participants' behaviour indicated a positive relationship between switch steepness and TA. We followed this behavioural finding by developing a novel state inference model which fit the data of high TA individuals better than gradual learning models (Rescorla-Wagner, Pearce-Hall). These results provide behavioural and computational evidence that high TA individuals have a tendency to represent the environment as multiple states which serves as novel explanation for higher relapse rates in anxiety.