A study published in Neuron
last week, conducted at Ïã¸ÛÁùºÏ²Ê's Wellcome Trust Centre for Neuroimaging, shows that the
conventional view of how our brains learn has to be revisited.
Previously, it
was thought that the dopaminergic inputs that the ventral striatum receives
from mid-brain structures signal a reward prediction error which facilitates
learning from rewards. However, the study conducted by Miriam Klein-Flugge,
Tim Behrens and colleagues, found that in situations where
learning does not depend on rewards, the ventral striatal signal flexibly
adapts and instead reflects a behaviourally relevant teaching signal, while the
mid-brain still encodes the classic reward prediction error.