Researchpomdptonic painchronic painreinforcement learning
Brain Models Injury-Induced Pain Via POMDP
8.3
Relevance ScoreMahajan, Dayan and Seymour publish Jan 22, 2026 in PLoS Computational Biology proposing a computational architecture that models injury and post-injury pain as a partially observable Markov decision process (POMDP), integrating inference and optimal control. Their simulations reproduce paradoxical behaviours like probing injured areas and show how information restriction can drive transitions from adaptive tonic pain to chronic pathological pain, suggesting targets for experimental and therapeutic follow-up.
