Survivor interaction contrasts for errored response times: Non-parametric contrasts for serial and parallel systems

Systems Factorial Technology (SFT) is a theoretically-derived methodology that allows for strong inferences to be made about the underlying processing architecture (e.g., whether processing occurs in a pooled, coactive fashion or independently, in serial or in parallel). Measures of mental architecture using SFT have been restricted to the use of error-free response times. In this paper, through formal proofs and demonstrations, we extended the measure of architecture, the survivor interaction contrast (SIC), to response times conditioned on whether they are correct or incorrect. We show that so long as an ordering relation (between stimulus conditions of different difficulty) is preserved, unique conditional SIC predictions are found for several classes of processing models. We further prove that this ordering relation holds for the popular Wiener diffusion model for both correct and error RTs but fails under some instantiations of a Poisson counter model.