Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

Testing the predictive processing model of placebo hypoalgesia using multivariate hierarchical models: evidence for precision weighted effects of expectations and Bayesian updating of expectations accounting for volatility

Arthur S. Courtin¹, Jesper Fischer Ehmsen¹, Micah G. Allen, Francesca Fardo¹; ¹Aarhus University

Presenter: Arthur S. Courtin

Placebo hypoalgesia is often explained by predictive processing theories, in which perception arises from a form of (approximate) Bayesian integration of expectations (prior) and sensory evidence (likelihood). However, few studies have formally tested this model and uncertainty remains regarding its implementation in the nervous system. Here, we use a probabilistic pain learning task and computational modelling to test a series of hypotheses about how healthy volunteers form and update expectations about the painfulness of upcoming thermal stimuli and how these expectations shape the way they perceive these stimuli. Of note, our models jointly account for all response types collected during the task, constituting a first step towards a comprehensive computational model of pain perception. Our results support the full Bayesian predictive processing model, in which 1) the update of expectations is calibrated on different sources of uncertainty (posterior belief variance, sequence volatility), which are tracked continuously by the agent, and 2) the effect of expectations on perception is proportional to the relative uncertainty of predictions and sensory evidence (precision weighting).

Topic Area: Predictive Processing & Cognitive Control

Extended Abstract: Full Text PDF