Decision Rules for Pictorial Threat Classification

PDF

Authors
  1. Bryant, D.J.
Corporate Authors
Defence R&D Canada - Toronto, Toronto ONT (CAN)
Abstract
Two experiments examined the use of heuristic and analytic decision strategies in a simulated threat assessment task. Subjects learned to classify targets as friend or foe on the bases of uncertain cues (i.e. characteristics that were probabilistically associated with classification of a target as friend or foe). Subjects were then asked to classify targets that contrasted predictions of several decision rules, including a simple heuristic called Take-the-Best-for-Classification (TTB-C) that uses a single cue to classify targets and the Bayesian classification strategy that is based on formal statistic models. Results of Experiment 1 indicated that the mode of presentation (Text versus picture) did not affect the tendency of subjects to use either decision strategy. Results of Experiment 2 indicated that exposure time of pictorial stimuli also did not affect the proportions of subjects employing TTB-C versus the Bayesian strategy. However, an unexpected but very large effect of the target set was observed in the second experiment. This effect may indicate that the interaction of the perceptual salience of cues with the diagnosticity of those cues is a predictor of strategy use. Future research will examine this possibility.

Il y a un résumé en français ici.

Keywords
decision making;classification;combat identification
Report Number
DRDC-TORONTO-TR-2009-126 — Technical Report
Date of publication
01 Jul 2009
Number of Pages
70
DSTKIM No
CA033065
CANDIS No
532353
Format(s):
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: