Pain perception is modulated by expectations and learning processes, but the influence of uncertainty in this relationship is not well established. We aimed to examine the relationship between uncertainty, pain learning and perception using hierarchical Bayesian modeling. In an aversive learning task, fifty participants learned contingencies between auditory cues and painful stimulations under changing levels of uncertainty to create periods of stability and volatility. Model-free analysis of our data suggested unexpected trials resulted in reduced accuracy and greater response times. In unexpected trials, high pain perception was reduced, while low pain perception was increased, in line with documented effects of expectations on pain perception. Computational model fitting revealed participants’ learning was best described by a two-level hierarchical gaussian filter model, suggesting participants adapted their beliefs at multiple levels during the task. Uncertainty influenced pain perception in opposite patterns for high and low pain stimulations: high pain perception was greater under high levels of uncertainty, while there was a non-significant trend for low pain perception to be reduced. Analyses of individual differences suggested depressive symptoms were associated with a reduced learning rate throughout the task. These results shed light on processes involved in pain learning in changing environments. They also suggest a possible relationship between learning alterations and psychological traits commonly found in chronic pain, such as depressive symptoms.

This article explores the influence of varying levels of uncertainty on pain perception and learning. Findings reveal that uncertainty modulates pain perception differently depending on pain intensity and that pain learning is influenced by psychological traits. These results contribute to our understanding of pain modulation in dynamic environments.