Stimulus Generalization as a Function of Discrimination Learning with and without Errors

Abstract

After discrimination learning with and without errors, three groups of pigeons were tested for generalization to the angularity (negative stimulus) dimension. For half the subjects from each group, each angle was superimposed on a light of 555-nanometer wavelength, the positive stimulus during training; for the other half, the angles were on a black background. When tested to the angularity alone, the group which had made errors (pecked at the negative stimulus) during discrimination training showed a typical incremental (inverted) gradient of responding to the negative stimulus dimension. The control group and the one learning the discrimination without errors showed no differential control by the negative stimulus dimension. When each angle was superimposed on the positive stimulus during testing, a decremental gradient with maximum response strength to the negative stimulus was evidenced by the group learning with and the one group learning without errors.