The growth of learning during non-differential reinforcement.

Abstract

Sixty-four white rats were trained in a conventional 2-window Lashley jumping stand for a period of 16 days. For half the animals (Group I) all jumps were rewarded. The other half (Group II) had 2 punished trials (locked cards) randomly interspersed among their rewarded trials. Subjects of both groups were further subdivided on the basis of whether they were given 0, 4, 8, or 16 days of experience with horizontally- and vertically-striped cards. All rats in both groups were trained on a simultaneous discrimination between the horizontally- and vertically-striped cards. Two analyses of variance were computed, 1 for total errors and 1 for initial errors. Both major variables[long dash]punishment and experience with striped cards[long dash]were significant beyond the .01 level. These results fail to support predictions based on either the Hull or Bitterman theory of learning. Spence''s theory accounts adequately for the data, but this expt. did not provide a crucial test for his formulation.