Biology
Dopamine transients encode reward prediction errors independent of learning rates
A. Mah, C. E. Golden, et al.
Biological models tie dopamine to reward prediction errors (RPEs) scaled by learning rates. Research conducted by Andrew Mah, Carla E.M. Golden, and Christine M. Constantinople shows that in a volatile, semi-observable-state task rats adjust initiation speed and use higher learning rates after state transitions, approximating Bayesian belief updates. Crucially, nucleus accumbens core dopamine encodes RPEs but not learning rates, pointing to dopamine-independent mechanisms for dynamic learning rates.
Related Publications
Explore these studies to deepen your understanding of the subject.
