Models of Conditioning and Reinforcement Learning
The hippocampus is involved in the timing of when the US is expected to arrive, which makes it important for reinforcement learning and attention to motivationally relevant cues, whereas the cerebellum is involved in the timing of the response itself. Stephen Grossberg set out to provide a unified mechanism for both classical and operant conditioning in the framework of his earlier articles on spatial pattern learning. Differential Hebbian learning rules evolved into the temporal difference rules that are widely used in models that partially explain responses of dopamine neurons to unexpected rewards or unexpected absences of rewards. Results on the interactions of electrical potentials, transmitters, and second messengers have been incorporated into many network models of associative learning. The first neural networks for Pavlovian conditioning were developed in the 1960s within the framework of all-or-none neuronal models.