QUANTITATIVE MODELS OF ATTENTION AND RESPONSE PROCESSES IN SHAPE IDENTIFICATION TASKS

Two classes of attention models are applied to the identification of a n item in a cluttered field: a serial model, exemplified by a two-proc ess filter/graded-distribution model, and a parallel model, exemplifie d by a single-process zoom-lens model. The two attention models genera te response-time distributions and response probabilities when the ser ial attention model is linked with a serial counting model and when th e parallel attention model is linked to a parallel counting model. Eac h response counter is assumed to accumulate its counts independently o f other response counters. It is shown that the two compound models gi ve rise to equivalent response-time equations. The serial counting mod el is then modified by allowing the response counters to interact duri ng processing. When the predictions of the independent and interactive counter models of response evocation are compared with data from a ta rget/flanker identification task, both models predict the obtained con vergence of response times obtained from the Incompatible and Compatib le flanker displays, but the interactive model appeared to give a bett er account of response times obtained from the Neutral flanker display s. (C) 1994 Academic Press, Inc.