Group Performance in a Visual Signal-Detection Task

Abstract

Sixteen male subjects were run in groups of four on 720 trials of a 16-alternative, forced-choice, visual signal-detection task. The first and third blocks of 240 trials required individual judgments concerning target location. The second block of 240 trials required a group decision as to target location for two groups of subjects, and individual responses followed by a group decision for the other two groups. Comparisons were made among individual performance, real group performance, and the performance of two types of pseudogroup (derived from individual data) using an information-integration and a decision-threshold model. Group performance was better than individual performance, and was midway between the performances predicted by the two pseudogroup models, indicating that neither of the latter were adequate models for group performance. Real groups showed some characteristics both of independence and dependence among subjects, with some apparent use of logical decision rules in pooling individual information. Groups which made both individual and group responses on the second block of trials showed greater increases in sensitivity with practice than groups which made only group responses. It appeared that making individual responses had specific practice effects on subsequent individual performance, and these effects were greater in magnitude than the practice effects resulting from making group responses.