Psychology ETDs

Publication Date

6-3-1975

Abstract

A procedure for steady-state generalization testing was developed and evaluated. Following a suggestion made by Farmer and Schoenfeld (1966), generalization gradients were constructed from responses occurring in the presence of test stimuli interpolated on a fixed interval schedule of reinforcement. Given appropriate modifications, the procedure was successful in maintaining generalization gradients for individual subjects. Evaluation of the steady-state procedure necessitated the adoption of a generic function. A search was conducted for a polynomial to approximate the response functions obtained for one of two groups of pigeons during one phase of the experiment. An initial ANOVA of the trend of orthogonal polynomials was followed by a least squares polynomial regression. It was concluded that a full quadratic equation sufficiently reproduced the response functions. The traditional linear function, implied by the literature, did not faithfully reproduce the data. The quadratic equation was applied to the daily response functions of all subjects and selected data from the literature. Dimensional stimulus control, an empirical phenomenon, was defined as the percentage of response variability that could be accounted for by the generic function. In addition, the form assumed by the generic function, when applied to daily response functions, was observed. Response functions obtained using the steady-state procedure agreed quite well with the response functions obtained using more traditional procedures. The agreement included not only the degree of control by the line tilt stimulus dimension but also the general form of the response gradients obtained across the stimulus dimension. More generally, the steady-state procedure was shown to maintain generalization gradients in such a manner that possible confounding by the reinforcement schedule was minimized. This was accomplished by isolating the test subinterval from the reinforcement subinterval and by removing all consequences from responses made in the presence of test stimuli. The degree of control exhibited by the maintained gradients appeared to be high and stable. The stability of the gradients contributes to the effectiveness of the procedure for within-subject designs. In addition, the procedure is structured in such a way that the consequences of responses to test stimuli need not be changed in order to initiate within-subject manipulations, including the establishment of a baseline of dimensional stimulus control.

Degree Name

Psychology

Level of Degree

Doctoral

Department Name

Psychology

First Committee Member (Chair)

Douglas Peter Ferraro

Second Committee Member

G. Robert Grice

Third Committee Member

Joseph Anthony Parsons

Fourth Committee Member

John Paul Gluck Jr.

Language

English

Document Type

Dissertation

Included in

Psychology Commons

Share

COinS