The effects of multiple-exemplar training and stimulus variability on generalization to line graphs.
Item
-
Title
-
The effects of multiple-exemplar training and stimulus variability on generalization to line graphs.
-
Identifier
-
AAI9830752
-
identifier
-
9830752
-
Creator
-
Pluchino, Sal Vincent.
-
Contributor
-
Adviser: Lanny Fields
-
Date
-
1998
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Psychology, Behavioral | Psychology, Experimental
-
Abstract
-
Behavior analysts often assess the effectiveness of interventions by visual inspection of line graphs. The dissertation identified factors that influence accurate graph reading and generalization to novel graphs. Each stimulus was a line graph that had a left and a right panel called A and B, respectively. Each panel contained ten data points that were distributed around a regression line. The regression line in panel A always had a {dollar}0\sp\circ{dollar} slope. The regression line in panel B had slopes that varied from {dollar}0\sp\circ{dollar} to {dollar}18\sp\circ{dollar} in {dollar}2\sp\circ{dollar} increments. The dispersion of data points around the regression line in both panels was either 3 or 6 standard deviations (SD). The slope of the regression line in panel B defined each stimulus.;College students were trained to discriminate between {dollar}0\sp\circ{dollar} (S-) and {dollar}18\sp\circ{dollar} (S+) stimuli presented in 3 SD form. Single-exemplar training was conducted with one S+ and one S-. Multiple-exemplar was training was conducted with forty-eight {dollar}18\sp\circ{dollar} S+'s and forty-eight {dollar}0\sp\circ{dollar} S-'s. The forty-eight S+'s differed by the location of data points, as did the forty-eight S-'s. Single-exemplar overtraining was conducted with one S+ and one S-, but the total number of trials matched that used in multiple-exemplar training.;Generalization tests were conducted using {dollar}0\sp\circ{dollar} through {dollar}18\sp\circ{dollar} stimuli in 3 SD and 6 SD form. Discriminative control by the S+ and S{dollar}-{dollar} was disrupted after training or overtraining with single exemplars but was maintained after multiple-exemplar training. Disruption resulted from both decrements in S+ responding and increments in S{dollar}-{dollar} responding. More pronounced disruptions of stimulus control were seen when these stimuli were presented in 6 SD form than in 3 SD form.;A signal detection analysis of performance during 3 SD tests showed there was no bias regardless of prior training regimen. During 6 SD tests, there were substantial biases after single-exemplar and multiple-exemplars training but no bias after single-exemplar overtraining.;Slopes of generalization gradients were steeper for test stimuli presented in 3 SD form than 6 SD form. Slopes of generalization gradients were steepest after multiple-exemplar training.;Multiple-exemplar training was the only procedure that resulted in the maintenance of discriminative control and the induction of dimensional control during generalization tests.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
