|
| Matching |
| Tuesday, June 1, 2004 |
| 10:30 AM–11:50 AM |
| Commonwealth |
| Area: EAB |
| Chair: Kelly M. Banna (Auburn University) |
| |
| Rapid Acquisition in Concurrent Chains and the Provenance of Matching |
| Domain: Applied Research |
| RANDOLPH C. GRACE (University of Canterbury), Darren R. Christensen (University of Canterbury), Mark E. Berg (University of Canterbury) |
| |
| Abstract: One of the recurring controversies in choice research is the origin of matching. Over four decades of research using concurrent-schedules procedures has failed to resolve this question. We suggest that recent studies on acquisition in concurrent chains using rapid-alternation procedures might help to resolve this issue. This talk will review evidence that pigeons’ response allocation in concurrent chains can adjust to rapidly-changing contingences; that sensitivity to current- and prior-session contingencies depends on whether or not there is a predictive relationship between those contingencies; and most importantly, that response allocation conforming to both generalized matching and categorical discrimination can be obtained in some cases. All of these results can be explained by a decision model which assumes that preference for an alternative increases after a relatively short delay to reinforcement, and decreases after a relatively long delay to reinforcement. According to the model, generalized matching and temporal discounting are derived phenomena, resulting as the aggregate of a ‘winner take all’ decision process. |
| |
| Bias in the Generalized Matching Law is Not Independent of Reinforcement |
| Domain: Applied Research |
| JAMES S. MACDONALL (Fordham University) |
| |
| Abstract: The generalized matching law says that behavior allocation is a function of reinforcer allocation, plus a constant bias. The bias accounts for preferences that are not related to reinforcement, such as, color, and response preferences. The generalized matching law is based on ratios of reinforcer rates at the alternatives consequently differences in overall rates of reinforcers cancel out in the reinforcer ratio. Recently, MacDonall (1999, 2000) showed that concurrent performances could be constructed from two pairs of schedules; one member of each pair arranges reinforcers for staying at the alternative and the other member arranges reinforcers for changing over. This procedure allows rates of reinforcement to be varied independently at the alternatives. Rats were exposed to two pairs of stay and switch schedules, which comprise a concurrent schedule, while the rates of reinforcement at each alternative and the reinforcer allocation varied. The generalized matching law described the results of both arrangements, but a bias was consistently obtained when rates of reinforcement were different at the alternatives. These results indicate that, contrary to the original interpretation, bias is also a function of reinforcement. |
| |
| Application of the Matching Relation to Choice Behavior in the Bluegill Sunfish (Lepomis macrochirus) |
| Domain: Applied Research |
| KELLY M. BANNA (Auburn University), Ashley Akins (Auburn University), James Cardwell (Auburn University), Chauncey Holder (Auburn University), M. Christopher Newland (Auburn University) |
| |
| Abstract: Bluegill Sunfish (Lepomis macrochirus) were trained to lever press on concurrent schedules for food reinforcement. Overall rate of reinforcement was held constant, while the ratio of scheduled reinforcement rates varied among 1:8, 1:4, 1:1, 4:1, and 8:1. Response and time ratios varied as a function of scheduled reinforcer ratio and matching functions generated using both time and reinforcer ratios showed undermatching. These results have been replicated both within and across individual fish and are consistent with those seen in other species (e.g., non-human primates, pigeons, rats). Furthermore, the temporal pattern of behavior change during transitions between reinforcement schedules is similar to that seen in rats and squirrel monkeys. Taken together, these data suggest a high degree of continuity in concurrent schedule behavior along the phylogenetic scale. |
| |
|
| |