LAUSR

This article provides an interoceptive active inference (IAI) account of social anxiety disorder (SAD). Through a neurocognitive framework, we argue that the cognitive and behavioural profile of SAD is best conceived of as a form of maladaptive IAI produced by a negatively biased self-model that cannot reconcile inconsistent tendencies to approach and avoid social interaction. Anticipated future social interactions produce interoceptive prediction error (bodily states of arousal). These interoceptive states are transcribed and experienced as states of distress due to the influence of inconsistent and unstable self-models across a hierarchy of interrelated systems involved in emotional, interoceptive and affective processing. We highlight the role of the insula cortex, in concert with the striatum, amygdala and dorsal anterior cingulate in the generation and reduction of interoceptive prediction errors as well as the resolution of social approach-avoidance conflict. The novelty of our account is a shift in explanatory priority from the representation of the social world in SAD to the representation of the SAD self. In particular, we show how a high-level conceptual self-model of social vulnerability and inadequacy fails to minimize prediction errors produced by a basic drive for social affiliation combined with strong avoidant tendencies. The result is a cascade of interoceptive prediction errors whose attempted minimization through action (i.e. active inference) yields the symptom profile of SAD. We conclude this article by proposing testable hypotheses to further investigate the neurocognitive traits of the SAD self with respect to IAI.