Stopping Speed as State, Not Trait: Exploring Within-Animal Varying Stopping Speeds in a Multi-Session Stop-Signal Task

Being able to reactively stop ongoing movements is important for safe navigation through the environment. Reactive stopping is typically studied using the stop-signal task, where participants are occasionally instructed to stop initiated movements. The speed of stopping, also referred to as the stop-signal reaction time (SSRT), is not observable because successful stopping lacks a response, but can be estimated. Researchers most often acquire one session of data per participant to estimate the speed of stopping, but sometimes more sessions of data are acquired to maximize the signal-to-noise ratio, for example when the task is combined with neural recordings such as electrophysiology. However, it is unknown whether the estimated stopping speed is a fixed trait or a state that can vary under identical experimental conditions. In this study, we investigate whether a separately estimated SSRT for each acquired session is statistically meaningful compared to estimating an across-session SSRT, by collecting many sessions in which male rats performed a stop-signal task. Results revealed that within-animal stopping speeds meaningfully changed from session to session and were not following a trend over time (e.g., due to task learning). Single-session SSRT estimates with lower reliabilities were associated with higher go trial response time variabilities, lower skewness levels of the go trial response time distribution, and lower stop accuracies. We also explored which factors explained changing SSRTs, and showed that motivation, shared motor dynamics, and attention could play a role. In conclusion, we encourage researchers to treat SSRTs as state-like variables when collecting multi-session stop-signal task data, as our results have convincingly shown that stopping speeds are far from trait-like under identical experimental conditions. This session-by-session approach will help future research in which neural signatures of reactive stopping need to be extracted in a time-precise manner, because time-locking stop-related neural activity to session-specific SSRTs is expected to capture the signature more precisely as opposed to an across-session SSRT.