|Michael Davison completed his BSC (Hons) at Bristol University, UK, and his PhD at Otago University, NZ. He has been a lecturer at Otago University, University College London, and Auckland University, where he is now a full professor and Director of the Experimental Analysis of Behaviour Research Unit. He is a Fellow of the Royal Society of NZ (from which he has been awarded a silver medal for research), and of the Association for Behavior Analysis International, and he recently served as Associate Editor of JEAB. He also holds an appointment at the NZ National Research Centre for Growth and Development and at The Liggins Institute. He served as International Representative on the ABA Executive. His real interest in life is growing herbs.|
Two recent empirical results from the study of contingent food delivery on local choice may have implications for our basic understanding of reinforcement. First, in the within-sessions (Davison & Baum, 2000) procedure, and also in second-order schedules, brief stimuli that are paired with reinforcer delivery produce a preference pulse following the stimulus presentation; but so do brief stimuli that have never been paired with food delivery; and if stimuli that are paired with food delivery follow responses on the lower reinforcer-rate alternative, they are followed by a preference pulse on the other alternative. These results argue against the notion of conditional reinforcement and suggest that stimuli that signal higher subsequent conditional probabilities of food for a certain activity at a location are followed by increases in that activity at that location. Second, when food delivery itself signals subsequent higher conditional probabilities of food, a preference pulse follows to the alternative that gave food; but when food delivery signals a lower subsequent conditional probability of food, the following preference pulse is to the other alternative. If a contingent event that increases subsequent responding is a reinforcer, then sometimes an event paired with food is a reinforcer, sometimes it is not; sometimes, the reinforcer reinforces, and sometimes it does not. This unsatisfactory state of affairs is not alleviated by asserting that reinforcers and conditional reinforcers have discriminative properties additional to their reinforcing properties and that sometimes the discriminative properties may ablate the reinforcing properties. Rather, these results make us confront the possibility that we have been wrong about reinforcement for 100 years. Perhaps stimuli (including reinforcers themselves) simply signal to an animal where to look for more of the same, or where to avoid for fewer of the same. Perhaps a reinforcer, as we know it, is simply a punctate hedonic event (hedon) with some current valance, and perhaps behavior simply follows the signposts provided by discriminative stimuli. Such an approach has many theoretical and practical implications.