Abstract:
Prolonged exposure to contingencies makes the likely time and location of subsequent reinforcers predictable, to the extent that these events occur systematically across time and location, and provided such events can be recalled. Reinforcers that are unusual in light of previous experience ("surprising" reinforcers) often have a greater impact on behavior than reinforcers that are expected (Davison & Baum, 2000; Maki, 1979). We investigated the effects of memory on discrimination of relationships between events, how surprising reinforcers guide behavior, and whether surprising reinforcers have a greater effect in stable rather than rapidly changing environments. Within a session, reinforcers occurred on a fixed-ratio1 schedule, either for the same response, or for the response that had not just produced the last reinforcer. After training, reinforcers that were inconsistent with this pattern were introduced occasionally. As the inter trial interval length decreased, discrimination of the contingencies increased. Surprising reinforcers were generally followed by a shift in choice towards the response that would next produce a reinforce on the basis of what reinforcers had come to signal in baseline in conditions of high discrimination. These findings are not consistent with a response-strengthening account of reinforcer control but suggest instead that behavior may come under control of patterns of reinforcer delivery that are relatively extended in time.