Stopping timed actions.

Abstract

The ability to inhibit ongoing responses that suddenly become inappropriate is essential for safe and effective interaction with an ever-changing and often unpredictable world. This ability is quantified by the stop-signal reaction time (SSRT), the completion time of an inhibitory process triggered by a signal to stop responding. Because SSRT cannot be directly observed, it must be inferred based on a model in which inhibitory (''stop'') and response (''go'') processes race with each other to control behavior. Inhibitory control is usually studied in the context of choice responses, but there has been increasing recent interest in what is often a key component of skilled behaviour, stopping a response that is timed to coincide with an anticipated event. We show that SSRT measurement via the commonly used independent race model fails for anticipated responses because the event signaling the need to stop changes the perception of the passage of time, due to the long known ''filled-interval illusion''. We propose a computational cognitive model of anticipated response inhibition that takes account of the distortion of time perception caused by the filled-interval illusion. We show that this model produces valid estimates of not only SSRT, but also another key process that determines inhibitory ability, lapses in attention. Our new model and accompanying Bayesian estimation procedures provide a solid basis for the burgeoning study of timed-action control.

Comparing the speed of action initiation and action inhibition

“…By matching experimental conditions as closely as possible between cancelling an impending action (i.e., "stopping") and generating an action (i.e., "going"), we found that the time for these two processes is comparable, both around 290 ms (with a ~33 ms delay caused by visual display, which was not accounted for in our calculations), suggesting that stopping and going can occur equally rapidly. This result is echoed by previous evidence from two independent research fields showing that simple reaction time (Luce, 1991;Welford, 1980) and the time to stop an action (He et al, 2021;Leunissen et al, 2017;Logan and Cowan, 1984;Matzke et al, 2021) are both around 200 -250 ms, and both can reduce to around 150 ms triggered by an unexpected event (Carlsen et al, 2004;Haith et al, 2016;Wessel and Aron, 2017). Our finding questions the consensus view from previous experimental, computational, and theoretical work that a rapid, dedicated inhibition mechanism exists to act like an "emergency brake" on response initiation and prevent an unwanted response to be produced (Aron et al, 2014;Boucher et al, 2007;Dunovan et al, 2015;Logan and Cowan, 1984;Slater-Hammel, 1960;Verbruggen et al, 2019;Wiecki and Frank, 2013).…”

“…By contrast, the need to cancel a response has high urgency because it has to be done before the response is initiated and the time needed to abort a response ranges from 180 ms to 270 ms Leunissen et al, 2017;Logan and Cowan, 1984;Matzke et al, 2021). In the stop-signal paradigm, therefore, there is a marked asymmetry in urgency between the requirements to initiate a response and the potential requirement to cancel one.…”

“…In stop-signal tasks, the delay in action initiation may be further exacerbated, with reaction times averaging up to around 450 ms (e.g., Leunissen et al, 2017) – much longer than the typical reactions times of 200-250 ms to respond to a go cue (Luce, 1991; Welford, 1980). By contrast, the need to cancel a response has high urgency because it has to be done before the response is initiated and the time needed to abort a response ranges from 180 ms to 270 ms (He et al, 2021; Leunissen et al, 2017; Logan and Cowan, 1984; Matzke et al, 2021). In the stop-signal paradigm, therefore, there is a marked asymmetry in urgency between the requirements to initiate a response and the potential requirement to cancel one.…”

OSARI, an Open-Source Anticipated Response Inhibition task

“…Verbruggen and Logan 2018 for a more comprehensive explanation 30 ). In a recent study, evidence for violations of the assumption of context independence were reported in several existing ARI datasets (Matzke et al, submitted 37 ), including data collected with OSARI. In the same study, a Bayesian approach was modified to estimate parametric stop-signal race models (BEESTS), accounting for ARI performance and the associated context independence violations.…”

Cognitive process modeling addresses context independence violations in the ABCD Study stop-signal task

“…It is the sum of a normal distribution (with mean, , and standard deviation, ) and an exponential distribution (with mean ), where the latter produces the positive skew characteristic of RT distributions. Parametric models such as BEESTS do not need to assume context independence in order to provide valid SSRT estimates when they are fit to only stop trial data (Matzke et al, 2021), but usually do so in order to leverage information provided by go trials to make estimation more efficient. Although estimating parameters from go and stop trials separately could, in principle, address context independence violations in ABCD, such estimates are unlikely to be sufficiently reliable because the design has relatively few stop trials.…”