Pranav Mahajan

This paper reconciles conflicting theories about dopamine signals in the tail of the striatum by proposing an entropy-regularised reinforcement learning model that integrates aversive and action prediction errors.

The paper proposes a computational framework to model injury recovery and pain transitions using a partially observable Markov decision process, explaining paradoxical behaviors and chronic pain development.

This paper introduces an entropy-regularised reinforcement-learning framework to address challenges in dopamine-mediated learning, showing it improves multi-reward learning and adapts to non-stationary environments.

This study identifies increased punishment sensitivity and specific neural activity patterns in chronic pain and fatigue patients during decision-making tasks.

This study shows that the Pavlovian fear system enables flexible and efficient exploratory decision-making by balancing safety and efficiency through uncertainty-based avoidance.

The paper models cognitive conflict in addiction by proposing an algorithmic framework that explains how drug-seeking persists despite negative outcomes.