Two-way generalization peak shift after two-key training in the pigeon

Abstract

Thirteen pigeons were reinforced on a 1-min variable-interval schedule (VI 1 min) for pecking at a key illuminated by a 550-nm light. All birds were tested for generalization to 550 nm and 10 other wavelengths, following three variations of training as follows: Group 1 received only single-stimulus training to 550 nm before the test. Group 2 received successive single-key discrimination training between 550 nm as S+ and 559 nm as S− (extinction) before the test. Group 3 received successive training, but 559 nm was not associated with extinction; rather, in the presence of 559 nm, the birds were reinforced on VI 1 min for pecking a second key. In agreement with previous results, Group 1 yielded generalization gradients peaked at 550 nm, and Group 2 produced peaks shifted away from S−. For Group 3 a separate gradient was obtained for responses to each key, and each of these gradients was shifted away from the wavelength that signaled reinforcement on the other key. The results suggest that peak shift can result from the interaction of stimulus control from each of two “excitatory” stimuli, if each controls a response that interferes with the response controlled by the other. A general description of the conditions for peak shift and peak-shift failure is briefly suggested.