Learning what to approach

Abstract

Most decisions share a common goal: maximize reward and minimize punishment. Achieving this goal requires learning which choices are likely to lead to favorable outcomes. Dopamine is essential for this process, enabling learning by signaling the difference between what we expect to get and what we actually get. Although all animals appear to use this dopamine prediction error circuit, some do so more than others, and this neural heterogeneity correlates with individual variability in behavior. In this issue of PLOS Biology, Lee and colleagues show that manipulating a simple task parameter can bias the animals’ behavioral strategy and modulate dopamine release, implying that how we learn is just as flexible as what we learn.