Sven Collette, Wolfgang M Pauli ... John O'Doherty
The human brain is capable of implementing inverse reinforcement learning, where an observer infers the hidden reward structure of a decision problem solely through observing another individual take actions.
Gail M Rosenbaum, Hannah L Grassie, Catherine A Hartley
Relative to children and adults, adolescents placed greater weight on negative prediction errors during learning and these age-varying learning idiosyncrasies biased subsequent memory for information associated with valenced outcomes.
fMRI evidence for off-task replay predicts subsequent replanning behavior in humans, suggesting that learning from simulated experience during replay helps update past policies in reinforcement learning.
Yumeya Yamamori, Oliver J Robinson, Jonathan P Roiser
A computational and translational approach to measuring anxiety-related behaviour was validated in two large independent samples, presenting opportunities for cross-species anxiety research and potential implications for anxiolytic drug development.
John P Grogan, Demitra Tsivos ... Elizabeth J Coulthard
Memory over 24 hours was impaired in Parkinson's patients off, rather than on, dopaminergic medication during reinforcement learning, whereas dopamine did not affect positive and negative reinforcement, in contrast to previous studies.
Dimensions for reinforcement learning reduced by dynamic organization of cerebellar climbing fiber response and synchrony in multiple functional components.
Challenging a popular theory in neuroeconomics, a computational cognitive study provides evidence against divisive normalization, a supposedly canonical neural computation, in favor of an alternative account, range normalization, in the context of value learning.
Computational modeling suggests that feedback between striatal cholinergic neurons and spiny neurons dynamically adjusts learning rates to optimize behavior in a variable world.
Maria Katharina Eckstein, Sarah L Master ... Anne GE Collins
Context factors such as experimental task and model parameterization can significantly impact modeling results, such that across tasks, the same participants show different fitted parameter values, and parameters capture different cognitive processes across tasks.
Humans perseverate on previously instructed goals in a novel multigoal reinforcement learning task, and do this to a greater extent for punishment avoidance goals.