These tools will no longer be maintained as of December 31, 2024. Archived website can be found here. PubMed4Hh GitHub repository can be found here. Contact NLM Customer Service if you have questions.


Pubmed for Handhelds

PUBMED FOR HANDHELDS

Search MEDLINE/PubMed

  • Title: Imaging the changing role of feedback during learning in decision-making.
    Author: Sailer U, Robinson S, Fischmeister FP, Moser E, Kryspin-Exner I, Bauer H.
    Journal: Neuroimage; 2007 Oct 01; 37(4):1474-86. PubMed ID: 17698371.
    Abstract:
    Learning from the outcome of decisions can be expected not only to change future decisions, but also our reaction to future outcomes. Using functional magnetic resonance imaging we investigated the neural responses of healthy subjects to feedback about choice outcomes before and after learning a response strategy which led to correct choices only. The task was designed so that losses were unavoidable even when all the choices made were correct. Subjects showed a distinct pattern of learning starting with an initial exploratory phase in which hypotheses about the correct strategy were generated and tested, followed by a phase of rapid strategy acquisition before reaching a final phase of proficiency. Neural activation was more pronounced during feedback processing in the exploratory phase than in the proficiency phase in a distributed network encompassing prefrontal and parietal areas as well as the striatum. These areas are involved in working memory processes, the management of uncertainty and the establishment of stimulus-outcome contingencies. Reduced activation during feedback processing following learning was not only observed within subjects across learning phases, but also between subjects with different learning speeds. Thus, controlled and automatic processing are characterised by differing amounts of activation in identical task-relevant areas. Furthermore, whereas the same brain regions coded for gains and losses, the activation following gains changed to a larger extent with learning than following losses. This suggests that positive prediction errors are more sensitive to increased reward predictability than are negative prediction errors.
    [Abstract] [Full Text] [Related] [New Search]