Item

Abstract

The anterior cingulate cortex (ACC) plays a central role in monitoring ongoing behavioral performance, for example by activating during tasks that evoke response conflict (i.e., the simultaneous engagement of two incompatible actions, such as going and stopping). This has been classically studied using the stop signal task in which an external cue instructs the subject to stop ongoing movement preparation. However, in this task, there is no observable behavioral event demarcating the occurrence of response conflict because the subject never actually moves. Thus, it is unknown precisely when and how ACC neuronal activity changes during response conflict, but the predominant view is that ACC neurons respond only after conflict occurs. Here, we introduce a new paradigm to observe response conflict with precise timing in headfixed rats on a treadmill. Rats learned that when presented a NoGo stimulus, the correct response was to remain immobile, but if they released a pre-potent running response, they could still correctly respond by stopping before crossing a distance threshold. On a subset of trials, rats committed these near-mistake (NM) movements. At the peak velocity of the NM movement, competing Go and NoGo actions were simultaneously engaged; thus, peak velocity provides an unambiguous time on each trial to mark the maximal conflict for neuronal activity alignment. Moreover, peak velocity can be used as a proxy for conflict magnitude on each trial because a larger incorrect movement needs more conflicting response to stop. We recorded 478 ACC single units from 3 rats and assessed neuronal activity changes in a 400 ms to 1400 ms window around NM peak velocity. Trials were divided into tertiles of peak velocities representing none-to-low, medium, or large conflict. Out of 478 units, 209 significantly scaled their firing rate with NM movement size (31% significantly positively correlated and 18% negatively with movement size). These units did not scale firing rate during running to a Go stimulus (when there is no conflict) and are therefore not simply correlated with speed. We also characterized population responses with demixed PCA using conflict magnitudes as conditions. The method isolates the components of the population activity that specifically scale with conflict. We again observed scaling with NM magnitude. The scaling is detectable as early as 400ms before velocity peak and persisting until after conflict was resolved (1400ms after velocity peak when the rat had stopped). Our findings provide strong evidence for ACC activity correlating with the degree to which incompatible actions compete. In contrast to current thinking that ACC only monitors past performance, we show that ACC neurons respond during conflict resolution and may inhibit actions.