The Surprising Nature of the Reaction Time Task

Abstract

At one level, the scientific enterprise engaged in by Guy Van Orden was about how to analyze reaction time data. At another level it was an attempt to understand the kind of system that one is dealing with in a reaction time (RT) experiment—the system that accords with the instructions that the experimenter gives, produces the responses to the particular class of stimuli that the experimenter presents, at latencies that the experimenter measures and analyzes. That there can be any question about the essential nature of the system under study is perhaps surprising given the long and influential history of RT research and the relative simplicity and transparency of the RT task. In this brief note we hope to show that the question is deserved and that on close inspection the nature of the RT task is, indeed, surprising.

Notes

¹ System identification is a technical term referring to an ideally formal process by which a dynamical model (linear or nonlinear) is crafted for a phenomenon of interest on the basis of available measures. Here, the phenomenon of interest is performance in the RT task. Van Orden's argument was that the required dynamical model is not attainable by measures that fail to incorporate (explicitly or implicitly) the multiple tasks and time scales that embed the individual occurrences of expeditious responding to a presented stimulus item (e.g., Van Orden, Holden, & Turvey, 2005; Van Orden, Hollis, & Wallot, 2012; Wijnants, Cox, Hasselman, Bosman, & Van Orden, 2012).