Evaluating the Independent Race Model for the Stop Signal Paradigm: Context Independence is Violated at Short Stop Signal Delays

Abstract

The main theoretical vehicle that explains performance in the stop signal paradigm is the Independent Race Model (Logan & Cowan, 1984; Logan, Van Zandt, Verbruggen, & Wagenmakers, 2014), which suggests that a go process begins at go stimulus onset and races independently against a stop process that begins at stop signal onset. The race model predicts that only the fastest go processes will escape inhibition when the delay between going and stopping (stop signal delay, SSD) is short. In a large set of data, I looked for evidence that observed stop-failure RT was longer than what is predicted by the race model at short SSDs. Evidence of violations at short SSDs was widespread, with violations in fast and slow subjects, fast and slow conditions, saccades and manual responses, visual and auditory stop signals, and simple and stimulus selective stopping. The violations were particularly large in subjects with slower go RT and in stimulus selective stopping. This delineation identifies and evaluates methods for quantifying the violations of independence, reveals the prevalence of the violation, and evaluates the effect of common manipulations on the violation. Implications of the violation and potential adjustments to the race model to account for the violations are discussed.