Training Effects on the Use of Simple Heuristics in Threat Assessment


  1. Bryant, D.J.
Corporate Authors
Defence R&D Canada - Toronto, Toronto ONT (CAN)
Previous research has examined the use of fast and frugal heuristics for threat classification with probabilistic cues. In all previous studies subjects learned to underlying relationships of cues to friend/foe classification through trial-and-error learning. Explicit training, however, is theoretically interesting because it potentially involves the integration of deliberate cognitive learning of the task’s underlying cue structure with implicit, procedural learning of specific cue-criterion probabilities. In the experiment, subjects learned to classify contacts in a simulated naval warfare environment and then were tested on sets of contacts that were designed to contrast predictions of several heuristics, including the Take-the-Best-for-Classification (TTB-C) and Pros Rule developed specifically for the threat classification task, as well as a Bayesian strategy based on computation of the conditional probabilities of friend or foe classification given the particular pattern of cues. The interpretability of the results was limited by the large proportion of subjects who exhibited uninterpretable patterns of responding. In contrast to previous experiments, very few subjects employed TTB-C, although more did use the less frugal Pros Rule. The experiment did yield the novel finding that some subjects can, given explicit training, employ a Bayesian strategy for this task.

Il y a un résumé en français ici.

Threat assessment;heuristics;training
Report Number
DRDC-TORONTO-TR-2005-229 — Technical Report
Date of publication
26 Aug 2005
Number of Pages
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: