Stimulus generalization of voluntary responses in humans following discrimination training
Shurtleff, Donald Alan, 1929
MetadataShow full item record
The relationship between stimulus discrimination and stimulus generalization has been explored within several different experimental settings. The purpose of the current study was to provide a methodology which would allow for the investigation of the effects of stimulus discrimination training on stimulus generalization in human subjects. In order to evaluate this method as a research technique two features of discrimination training were explored and related to performance in a subsequent test for stimulus discrimination. These were: (1) the amount of learning of a differential discrimination and, (2) the degree of physical separation between the discriminated pair. Ninety-three male and female subjects recruited from the Introductory and Experimental Psychology courses at Boston University participated in this experiment. The stimuli used in this study were selected from the auditory flutter continuum and differed with respect to frequency of auditory pulses per second. This dimension for which control of attendant physical properties, e.g., duration and intensity, is possible. Second, it is a dimension with which the subjects have had little, if any, nonexperimental experience. During discrimination training the onset of a yellow pilot light followed the response to a standard auditory flutter value (S+). No light was presented if he made the response in the presence of a second stimulus (S-). The subject was instructed to make a lever response only in the presence of the stimulus with which the light was associated. For groups differing in amount of discrimination training, the light was paired with 7.0 pulses per second, while no light was presented to responses in the presence of 7.4 pulses per second. In groups for which the S+ to S- separation was varied the light was again paired with 7.0 pulses per second and no light with one of the following: 6.2, 6.6, 7.4, 7.8, or 10.0 pulses per second. In discrimination training the subject was presented with progressively more S- than S+ trials thereby minimizing the expectancy of a fixed ratio of presentations of S+ and S-. Such an expectancy could be a confounding factor in the analysis of the number of responses made during the test for generalization. This procedure, also, permitted a smooth transition from discrimination training to the generalization period. The test for generalization, which was carried out in extinction, followed immediately after the termination of any given stimulus discrimination procedure. Two changes were introduced during the generalization period: (1) in addition to S+ and S-, stimuli from 6.2 - 7.8 pulses per second (in units of .2 pulses per second) were introduced and, (2) the light was no longer paired with responses made to the standard (7.0 pulses per second). The subject was not told of these changes in procedure. The results indicated that as the amount of discrimination learning increased postdiscrimination gradients were systematically reduced to S- and adjacent stimuli, while the point of maximum response frequency was shifted to stimuli to the nondiscriminated sode pf S+. There was a tendency for postdiscrimination gradients to be reduced to the discriminated side of S+ as the S+ to S- separation was decreased. The particular form of the postdiscrimination gradient seemed to depend upon the direction of S- variations, i.e., in units slower than the S+ or faster than the S+. This latter feature was related to the underlying discriminability of the stimuli used to test for generalization. Performance during stimulus discrimination tended to reflect the magnitude of separation between the discriminated pair. [TRUNCATED]
Thesis (Ph.D.)--Boston University.