Timing in simple conditioning and occasion setting: a neural network approach.

We present a neural network model of Pavlovian conditioning in which a timing mechanism, by which a CS can predict when the US is presented, activates an architecture in which a stimulus acts as a simple CS and/or as an occasion setter. In the model, stimuli evoke multiple traces of different duration and amplitude, peaking at different times after CS presentation [Grossberg and Schmajuk, 1989. Neural Netw. 2, 79-102]. These traces compete to become associated directly and indirectly (through hidden units) with the US [Schmajuk and DiCarlo, 1992. Psychol. Rev. 99, 268-305]. The output of the system predicts the value, moment, and duration of presentation of reinforcement. Importantly, in contrast to the model by Schmajuk and DiCarlo [Schmajuk and DiCarlo, 1992. Psychol. Rev. 99, 268-305], in the present model a stimulus may assume different roles (simple CS, occasion setter, or both) at different time moments. Moreover, while in the Schmajuk and DiCarlo model [Schmajuk and DiCarlo, 1992. Psychol. Rev. 99, 268-305], competition between CSs is purely associative, in the present model competition between CSs is both associative and temporal. CSs compete to predict not only the presence and the intensity of the US, but also its temporal characteristics: time of presentation and duration. The model is able to address both the temporal and associative properties of simple conditioning, compound conditioning, and occasion setting.