DECISION MAKING: INTERACTIONS OF STIMULUS DIMENSIONS IN DIRECT TESTS OF THE ELIMINATION-BY-ASPECTS MODEL

ROBERT ARTHER HOWELLS, Purdue University

Abstract

Five experiments were conducted to determine whether dimensional interaction type, stimulus discriminability, and judgment type are factors that affect the applicability of the Elimination-By-Aspects (EBA) model. Three experiments represented attempts to classify dimensional interactions produced by four stimulus sets (PPIH = psychophysical, integral; PPSH = psychophysical, separable; HOIH = higher-order, integral; HOSH = higher-order, separable) on the basis of opposite outcomes for similarity scaling, reaction time (RT) facilitation, and selective attention. Experiment I employed a scaling procedure and confirmed the hypothesis that the metrics underlying similarity judgments in the integral and separable sets are, respectively, Euclidean and city-block. Experiment II confirmed the hypothesis that RT facilitation occurs only when interactions are integral. When RT was measured in a classification task, PPIH and HOIH exhibited facilitation for test conditions where stimulus dimensions were correlated, relative to control conditions where dimensions were orthogonal. Sets PPSH and HOSH did not exhibit facilitation. Experiment III confirmed the hypothesis that selective attention occurs only when interactions are separable. When RT was measured in a filtering task, PPSH and HOSH exhibited selective attention to uncorrelated irrelevant dimensions since RTs were similar for test conditions, where both relevant and irrelevant dimensions varied, and unidimensional controls. PPIH and HOIH exhibited no selective attention since RTs were greater for test conditions than for unidimensional controls. Taken together, these findings suggest that interactions produced by sets PPIH and HOIH were integral, while those of PPSH and HOSH were separable. Experiments IV and V tested the hypothesis that dimensional interaction type and stimulus discriminability are factors that affect the applicability of the EBA model. Experiment IV employed a behavioral test of the model in psychophysical judgments where interaction type and discriminability were orthogonally manipulated. The model was appropriate only when interactions were separable and discriminability was high. Experiment V was similar but involved preferential judgments. The model was appropriate when interactions were separable and discriminability was either high or low. These findings suggest that dimensional interaction type, judgment type, and stimulus discriminability are factors that interactively influence the applicability of the EBA model.

Degree

Ph.D.

Subject Area

Psychology|Experiments

COinS