Date of Award

2013-01-01

Degree Name

Doctor of Philosophy

Department

Psychology

Advisor(s)

Stephen L. Crites

Abstract

The evaluative oddball is an implicit measure for detecting (concealed) attitudes. In evaluative oddball studies, low occurrence stimuli such as negative pictures are presented among high occurrence context stimuli such as positive pictures. Late positive potentials (LPPs) of the event-related potential (ERP) are larger to evaluatively incongruent stimuli such as negatives compared to congruent stimuli such as positives with the context (e.g., positives). In prior evaluative oddball paradigms, this evaluative congruity effect of the LPP was reduced when participants concealed compared to truthfully reported attitudes. Because prior evaluative oddballs have been focused on the group level analysis, it has been overlooked how this reduced evaluative congruity effect for concealing might affect the individual level classification of attitudes. The evaluative oddball paradigm was extended by using a Bayesian classification approach to determine whether the evaluative oddball task influenced classification of concealment trials. Fifty-two participants performed an evaluative oddball task in which they either counted or made key presses to evaluatively incongruent pictures. Altogether, evaluatively incongruent pictures were correctly classified at 76.9%. There was no difference between counting (19/25 or 76%) and key pressing (19/27 or 70%) classification rates of participants' attitudes during concealment trials. During truth telling, there was a non-significant effect for a larger percentage of participants' attitudes classified for key pressing (24/27 or 89%) than counting (18/25 or 72%). These data suggest that this Concealed Attitude Test (CAT) is promising for detecting (concealed) attitudes at the individual level using an LPP-based Bayesian approach.

Language

en

Provenance

Received from ProQuest

File Size

84 pages

File Format

application/pdf

Rights Holder

David R. Herring

Download

Share

COinS