Advancing neurodynamic understanding of Bayesian models of chronic pain: Evidence, perspectives, and new insights

Chronic pain is a prevalent and complex condition that significantly impacts individuals' quality of life. Understanding its neurobiology and clinical phenomenology requires the integration of cognitive-behavioral and sensory-motor frameworks. Traditional models inadequately address the multilevel and transnosological aspects of chronic pain, particularly the interplay between interoception and predictive coding. Recent neurocomputational approaches present the theoretical basis for a Bayesian model of chronic pain that unifies interoception and predictive processing. However, providing empirical evidence for this model remains challenging. In this regard, we present a comprehensive review demonstrating the applications of neurocomputational models to understand the complex pain experience under four key frameworks: (i) likelihood and the altered interoceptive processing; (ii) priors and the pain-related expectations; (iii) kinesiophobia shaped by anticipatory capability; and (iv) the heart-brain axis and its role in decision-making. Empirical evidence supporting these frameworks may guide clinical interventions and improve patient outcomes by targeting the specific maladaptive processes that underlie chronic pain.