Suzette Astley, Ph.D.
The context dependency of conditioned reinforcement: Can differential conditioned reinforcing value be specific to a complex of three cues?
An experiment tested differential implications of the hypotheses that conditioned reinforcing properties are acquired by component stimuli versus by patterns of cues.
On training trials, red and green keylights were differentially paired with response-independent food for fifteen adult pigeons. Test consisted of choice trials in which a peck to one of the side keys produced red and a peck to the other side key produced green. Trials were given in each of two houselight contexts. Whether reward followed the colored keylights depended upon the trial type (training or test) and the houselight context (A or B). In Context A, food followed red but not green on training trials. On test trials in Context A, food followed green and not red. The trial type-keylight combinations rewarded in Context B were the opposite of those in Context A; i.e., food followed green on training trials and red on test trials. There were 3 times as many training as test trials in each context. The red and green keylights lasted 3 and 30 seconds for Groups 3 and 30 respectively.
Pigeons in Group 3 came to choose green most often in Context A and red most often in Context B. Pigeons in Group 30 came to choose red most often in Context A and green most often in Context B. The results were interpreted as evidence that the short-term memory of differential trial antecedents persisted throughout the keylight stimuli only for Group 3, and prevented generalization between training and test trials for Group 3. Thus, the choice of the pigeons in Group 3 was determined by the relative probabilities of reward on test trials within a context. Choice for Group 30 was determined by the proportion of all trials followed by reward. The hypothesis that conditioned reinforcement is a property of cue patterns was supported.
Ph.D., Psychology, Kansas State University, 1984