Computational and neural mechanisms of statistical pain learning

Journal article

The ability to learn from experience and update beliefs about the environment is essential for adaptive behaviour—a process often modelled as reinforcement learning. Pain and reward are fundamental drivers of learning, yet how they independently or jointly shape behaviour under varying levels of environmental uncertainty remains unclear. Using a hierarchical Bayesian model, we investigated how pain, monetary reward, and their combination shaped learning across stable and volatile contexts. Uncertainty modulated learning across all conditions. Pain increased response time, suggesting heightened vigilance—particularly in the pain-avoidance group, where response times were correlated with fear of pain. Despite differences in behavioural adaptations, learning rates remained comparable across motivational contexts, suggesting that motivational influence on learning engaged mechanisms beyond belief-updating speed. By leveraging a biologically plausible model, we show that vigilance is a key factor underlying motivational differences in learning, providing insight into adaptive behaviours in aversive and rewarding environments.M.Sc

Authors: Mancini, F; Zhang, S; Seymour, B

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Nature Research
• Journal: Nature Communications
• Volume: 13
• Issue: 1
• Pages: 6613-6613
• Article number: 6613
• Publication date: 2022-11-03
• DOI: 10.1038/s41467-022-34283-9
• EISSN: 2041-1723
• ISSN: 2041-1723
• Language: English
• Keywords: Neuroscience; Computer science; Bayes' theorem; Posterior parietal cortex; Sensory system; Mathematics; Inference; Statistical inference; Psychology; Probabilistic logic; Bayesian probability; Stimulus (psychology); Bayesian inference; Statistical model; Statistics; Cognitive psychology; Artificial intelligence