Psychology ETDs

Publication Date

12-9-1976

Abstract

A recent formulation has incorporated encoding variability in a stimulus-sampling framework. This conceptualization suggests that an active sample of "encoding operators" encodes each stimulus. On subsequent presentations of the stimulus each operator is checked for a tag/no tag representing the use of that operator in encoding the stimulus. These assumptions produce binomial distributions of noise and signal + noise underlying recognition performance. The present experiment tested the forms of the underlying distributions in the context change situation. Three amounts of context change between training and testing were investigated, with results indicating that recognition performance decreased as amount of context change increased. Most importantly, attempts to fit binomial distributions of noise and signal + noise to the recognition data were totally unsuccessful. The data were better fit by normal distributions, which suggests that attempts to specify globally the nature of encoding processes produces distributions of noise and signal + noise that do not allow inferences about the processes they represent.

Degree Name

Psychology

Level of Degree

Masters

Department Name

Psychology

First Committee Member (Chair)

Henry C. Ellis

Second Committee Member

Peder Johnson

Third Committee Member

Carol Conrad

Document Type

Thesis

Download

Included in

Psychology Commons

Share

COinS