Figure 6. Trace and Delay conditioning paradigms.

(A,B) Explanation of delay (A) and trace (B) conditioning. In delay conditioning, the cueing stimulus remains on until the reward appears. In trace conditioning, the cueing stimulus turns back off before the reward appears. (C,D) State spaces for delay-conditioning (C) and trace-conditioning (D). In delay conditioning, the presence of the (presumably salient) stimulus produces a single, observationally-defined state. In trace conditioning the absence of a salient stimulus produces a collection of equivalent states. (E) Simulations of trace vs. delay conditioning. Value learning at the CS state is slower under trace conditioning due to the intervening collection of equivalent states. Larger sets of equivalent states lead to slower value-growth of the CS state.