Modulation of the sensory and affective dimensions of pain by expectations and uncertainty: a Bayesian modeling approach

Abstract

The perception of pain is sensitive to various mental processes such as expectation about the nociceptive stimulus or individual differences in the affective and cognitive evaluations of pain. A promising approach to the investigation of the neurocomputational principles of pain perception is the theory of predictive processing which offers a general framework to understand perception and cognition. Rethinking the brain as a Bayesian inference organ, this theory has been recently applied to the experience of pain and its modulation with convincing results. We adapted an existing pain cueing paradigm to collect pain intensity and unpleasantness ratings from fifty-four healthy subjects with experience in meditation practice, while manipulating expectations and uncertainty about impending electrical stimulations. Using state-of-the-art statistical modeling, we modeled the generation of trial-wise pain ratings as a Bayesian inference process integrating probability distributions over sensory information, cue-based expectations and a trait-like, individual prior about pain experience. Notably, we extended previous hierarchical Bayesian models of pain by accounting for both the sensory and the affective components of pain. As predicted, individual psychological scores of pain catastrophizing and cognitive defusion showed opposite correlation patterns with the model parameters representing their computational counterparts. Furthermore, lifetime meditation practice was strongly and inversely correlated with the weight of short-term expectations in the perceptual process, as well as with a trait-like prior influencing the affective dimension of pain. We conclude that this approach offers a promising avenue for a principled investigation of pain perception mechanisms and, more specifically, of the effect of contemplative practices on the latter.