Pain intensity ratings are subject to various cognitive modulations – yet the mechanisms underlying this influence are still not understood. In a conditioning protocol, pain-related expectations were induced through pairing predefined movements with a noxious or innocuous stimulus in either a predictable or unpredictable fashion. Healthy volunteers (N = 37) categorized the stimuli as either painful or nonpainful and rated its perceived intensity. Using a Hierarchical Drift Diffusion model based on the categorization data, we found that an a priori decision-making bias evolved toward the expected sensations (P < .001). In particular, our findings suggest that differences in both the amount of decision-making bias (P = .004) and the speed of sensory processing predict pain intensity ratings (P < .001). As such, changes in pain ratings could be based in either of these processes, which may require a different approach when targeted as part of psychological pain treatment.

Changes in reported pain levels were linked to two distinct mechanisms, suggesting that increased pain reports could be attributed to either enhanced sensory processing or biased inferences. Our results might contribute to the development of person-tailored treatments based on the identification of latent mechanisms using computational models.