Comparison of stimulus generalization following variable-ratio and variable-interval training.

Abstract

Seventeen pairs of pigeons were trained to key peck on variable-ratio (VR) and variable-interval (VI) reinforcement schedules and then tested for generalization along the wavelength dimension. The pairs were yoked such that when the lead bird earned a reinforcement on the VR schedule the next response by the other (VI) bird was also reinforced. In accord with prediction the VR Ss responded at higher rates and generated flatter wavelength generalization gradients than did their VI partners. Both of these effects were strongest at the beginning of generalization testing in extinction but decreased in magnitude during the course of the test.