A model of Learning (in particular Classical conditioning)

See sec 14.2.4 of Sutton-Barto

It corresponds to Temporal difference learning with Linear approximation of the value function (which in this context need not represent value, but intensity of response..)