Incorrect responses in the response time interaction contrast

摘要

Townsend and Nozawa (1995) derived predictions for response time interaction contrasts that distinguish several classes of cognitive architectures (serial, parallel, coactive, exhaustive, self-terminating) in double factorial experiments. Their original theorems were limited to experimental tasks with ceiling accuracy. In this theoretical note I investigate systems factorial technology (SFT) within two canonical classes of models generating incorrect responses, namely, models with independent racers for informed responses and guesses, and models with mutually exclusive separate states for informed responding and guessing. I derive generalized interaction contrasts under these two model classes; these turn out to be related to the Kaplan-Meier and Aalen-Johansen estimators known from survival analysis. I discuss the limitations of the SFT approach if the incorrect responses arise from the component processes, and propose an alternative experimental setup that varies the temporal onset of the stimulus components. I demonstrate that with onset delay, SFT methodology can be generalized to non-perfect accuracy, and I point out the consequences for response time experimentation. (C) 2019 Elsevier Inc. All rights reserved.