Discrete versus Continuous Models of Response Preparation: A Reaction-Time Analysis
David E. Meyer Steven Yantis Allen Osman J. E. Keith Smith University of Michigan
Results are reported from a response-priming procedure designed to investigate how partial information is accumulated by rapid mental processes during the performance of reaction-time tasks. When a delay occurs between a priming stimulus and a subsequent test stimulus, the results reveal that subjects typically use the prime to prepare for the response required by the test stimulus, producing faster reaction times. Depending on the amount of delay, observed reaction-time distributions are mixtures of a few underlying component distributions. Such an outcome, which complements some of our other findings ( Meyer & Irwin, 1982), provides further tests for continuous (e.g., cascade and random-walk) models versus discrete (e.g., stage) models of information processing and response preparation. It suggests that response preparation involves only a small number of discrete states. This is consistent with stage models of preparatory processes and casts doubt on cascade models.
Since the advent of the information-processing revolution in psychology, cognitive psychologists have developed an increasingly rich conception of the mental operations that take place between the presentation of a stimulus and the production of a subsequent response. Most current models of human information processing include a number of major components, for example, encoding, memory retrieval, decision, motor programming, and so forth. Yet these models do not