Two-state versus continuous-state stimulus representations: A test based on attentional constraints

Abstract

New predictions of a class of two-state and a class of continuous-state attention models are described and then tested in an experiment requiring subjects to detect and locate a target occurring in one of four possible locations. The model predictions are tested by comparing the effect of focused- versus divided-attention instructions on the probability of correctly locating a target. The class of two-state models predicts that target localization performance will be better in the focused- than in the divided-attention condition, while the class of continuousstate models predicts that performance in the divided-attention condition will equal or exceed that in the focused attention condition. Results from an experiment testing these predictions favor the continuous-state models. © 1983 Psychonomic Society, Inc.