Figure 1.

State-space representation for contingency degradation: during the initial phase (left panel) the animal learns that action 1 (A1) earns outcome 1 (O1) and action 2 (A2) earns outcome 2 (O2). During degradation (right panel), the animal learns that in addition to A1–O1 and A2–O2, taking any “other action” (OA, anything other than pressing the levers, e.g., sitting, sniffing, exploring, et cetera) also earns O1, which serves to degrade the contingency between A1 and O1. It is posited that the animal partitions the initial learning contingencies (A1–O1, A2–O2) and the degradation contingencies (A1–O1, A2–O2, OA–O1) into two states, state 1 (S1) and state 2 (S2), respectively.