Biomedical subjects
Publications and source records attributed to H S TERRACE.
Pigeons were trained to discriminate between 580 mk (S+) and 540 mmicro (S-) by procedures which resulted in either a large amount of responding to S-or in little or no responding to S-. A shift in the peak of a postdiscrimination generalization gradient, away from S-, was obtained only if the discrimination was learned with a large amount of responding to S-.
Chlorpromazine or imipramine disrupts a pigeon's performance on a discrimination between a vertical and horizontal line only if the discrimination was learned with errors. Errorless learning is obtained if training starts with an easy-to-learn discrimination of color and shifts progressively to the more difficult horizontal-vertical discrimination.
Responses to S- ("errors") are not a necessary condition for the formation of an operant discrimination of color. Errors do not occur if discrimination training begins early in conditioning and if S+ and S- initially differ with respect to brightness, duration and wavelength. After training starts, S-'s duration and brightness is progressively increased until S+ and S- differ only with respect to wavelength. Errors do occur if training starts after much conditioning in the presence of S+ has occurred or if S+ and S- differ only with respect to wavelength throughout training. Performance following discrimination learning without errors lacks three characteristics that are found following learning with errors. Only those birds that learned the discrimination with errors showed (1) "emotional" responses in the presence of S-, (2) an increase in the rate (or a decrease in the latency) of its response to S+, and (3) occasional bursts of responses to S-.
A procedure developed earlier (Terrace, 1963) successfully trained a red-green discrimination without the occurrence of any errors in 12 out of 12 cases. Errorless transfer from the red-green discrimination to a discrimination between a vertical and a horizontal line was accomplished by first superimposing the vertical and the horizontal lines on the red and green backgrounds, respectively, and then fading out the red and the green backgrounds. Superimposition of the two sets of stimuli without fading, or an abrupt transfer from the first to the second set of stimuli, resulted in the occurrence of errors during transfer. Superimposition, however, did result in some "incidental learning". Performance following acquisition of the vertical-horizontal discrimination with errors differed from performance following acquisition without errors. If the vertical-horizontal discrimination was learned with errors, the latency of the response to S+ was permanently shortened and errors occurred during subsequent testing on the red-green discrimination even though the red-green discrimination was originally acquired without errors. If the vertical-horizontal discrimination was learned without errors, the latency of the response to S+ was unaffected and no errors occurred during subsequent testing on the red-green discrimination.
Explore the source record for details and available documents.
Explore the source record for details and available documents.