Discriminations based on differential responding.
Item
-
Title
-
Discriminations based on differential responding.
-
Identifier
-
AAI9000727
-
identifier
-
9000727
-
Creator
-
Ramirez, Elizabeth.
-
Contributor
-
Adviser: Brett K. Cole
-
Date
-
1989
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Psychology, Experimental
-
Abstract
-
In a conditional discrimination paradigm, a pigeon's choice of a red side key was reinforced if the animal had completed a single peck (FR 1) requirement on the center key; choice of a green side key was reinforced when the animal completed a probabilistic ({dollar}p{dollar}), or random ratio (RR), requirement on the center key. Eight values of {dollar}p{dollar} were investigated, from {dollar}p{dollar} =.025 to {dollar}p{dollar} = 1.0. With FR 1 and {dollar}p{dollar} responding as discriminative antecedents to choice, a gradual decline was obtained in overall accuracy of choice, eventually to a chance level, with increases in {dollar}p{dollar} value. The values of {dollar}p{dollar} sampled variations of behavior differences between components. Discriminability of the component run length (RL) values of the {dollar}p{dollar} requirement were analyzed. Although accuracy as a function of RL value decreased with increases in {dollar}p{dollar} value, the effect was small, and for some animals, differential responding to RL {dollar}>{dollar} 1 was not eliminated. The systematic decrease in overall accuracy was accounted for in terms of an increasing frequency of abolishment trials that resulted from the increasing overlap, with increases in {dollar}p{dollar} value, between the single peck (FR 1) requirement, where red choices were reinforced, and the single peck (RL 1) from the {dollar}p{dollar} requirement, where green choices were reinforced. The RL discriminability data showed an increase in errors at a given RL {dollar}>{dollar} 1, with increases in {dollar}p{dollar}. This was viewed, first, as a 'context' effect, since at each value of p, any RL was embedded in a different distribution of other RL values. Second, increasing the {dollar}p{dollar} value not only increased the frequency of possible non-differential reinforcement, it also represented a decrease in the opportunity for differential responding. These reinforcement based changes were viewed as responsible for the abolishment of discriminative performance on single peck trials, and for the incidental reinforcement of position based choices. Such abolishment also accounted for the generalization of position preference to RL {dollar}>{dollar} 1 trials, producing an error gradient wherein RL values closest to a single peck produced the most errors.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
