Testing race models for the stop signal paradigm: the copula approach

• Applicants: Professor Dr. Hans Colonius; Professorin Dr. Adele Diederich
• Subject Area: General, Cognitive and Mathematical Psychology ; Mathematics
• Term: since 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 466720487

The ability to suppress unwanted or inappropriate actions and impulses (response inhibition) is a crucial component of flexible, goal-directed behavior. The stop-signal task is an essential tool for studying response inhibition in psychology and neuroscience, among several other disciplines. In this task, participants typically perform a go task (e.g., press left when an arrow pointing to the left appears, and right when an arrow pointing to the right appears), but on a minority of the trials, a stop signal (e.g. an acoustic stimulus) appears after a variable stop-signal delay, instructing participants to suppress the imminent go response. A well-known major challenge for modelling the stop-signal task is that the stopping process, considered as an essential measure of inhibition, is not directly observable. To address this problem, so-called (horse) race models have been developed where performance is conceived as a race between a ``go process” triggered by presentation of the go stimulus and a ``stop process” triggered by presentation of the stop signal. When the stop process finishes before the go process, the response is inhibited; otherwise, it is executed. Unlike the latency of the go responses, response-inhibition latency cannot be observed directly because successful response inhibition results in the absence of an observable response. Independent race models make two assumptions: first, that the times taken by both processes are stochastically independent random variables. Second, they postulate that the distribution of processing speed of the go signal does not change, e.g., speed up or slow down when a stop signal is present, and is thus identical with the distribution of latencies observed when no stop signal is presented (``context invariance”). Both parametric and nonparametric versions of the independent race model make predictions that can be tested against the data obtained in the stop signal task. When observed data deviate from the model predictions, the validity of the derived measures of inhibition, like mean stop signal processing time, is dubious. Up to now it seems not possible to find out which one of the two assumptions does not hold, however. Both context and stochastic independence constitute essential features of the postulated inhibition process and they may bias measures of inhibition to different degrees. This project will generalize the independent race model by introducing explicit assumptions of dependence. Based on the statistical theory of copulas, we will assess, via simulation and analysis of generalized race models, the separate effects of context and stochastic dependence on model predictions and measures of response inhibition. Our findings will help users (a) gauge how much they can rely on their measurements under the independent race model, (b) assess how valid their conclusions about the mechanism of the stopping process can be, and (c) make available possible alternatives to the independent ra

• DFG Programme: Research Grants