Modeling Reaction Time: An investigation on input-strength variability in two-alternative, forced-choice tasks

We study the effects of input variability on choice reaction time (choice-RT) models under the two-alternative, forced choice paradigm. We test the well known and widely used (by choice-RT models) assumption that slower errors are produced under higher input-strength variability conditions (between blocks of trials). Two experiments, featuring a ‘numerosity task’, are presented. Analysis of the data from 46 participants, considering the difference between mean error RT and mean correct RT, across input-strength variability conditions, produced weak support for the hypothesis. Further analyses of complete RT distributions, per participant, strongly disfavors the hypothesis’ predictions. We argue that the evidence presented by this work is enough to render more detailed investigation on choice-RT models. 1 Partially supported by a Summer ’03 Fellowship awarded to the first author by both the Institute for Mathematical Sciences and the School of Social Sciences. 2 Send correspondence to Institute for Mathematical Behavioral Sciences. 3161 SSPA, University of California, Irvine, 92697-5100. Phone (949) 824-3766; FAX (949) 824-1670; E-mail [email protected]. Preprint submitted to Elsevier Science 22 February 2004