Interoception, the ability of the nervous system to sense and interpret internal bodily signals, plays a crucial role in aligning internal states with external environments. A review proposes an interoceptive computational framework to link mechanistic insights with clinical interventions for anxiety disorders. This framework is based on predictive processing theory, which suggests that mismatches between expected and actual sensory inputs can lead to interoceptive dysfunction, a factor in anxiety.

The review draws on evidence from computational psychiatry, neurophysiology, and clinical studies, modeling anxiety as a disorder of interoceptive prediction. It identifies that anxiety is influenced by hyperprecise threat priors and context rigidity, which increase interoceptive prediction errors, resulting in heightened defensive responses and cognitive biases.

By integrating computational modeling with targeted interventions, such as interoceptive exposure linked to Bayesian belief updating, the framework aims to enhance diagnostic accuracy and treatment outcomes. This multidimensional approach offers a pathway to precision psychiatry, emphasizing the need for further research on perturbed-prior experiments and hybrid interventions to improve personalized treatment in anxiety disorders.