Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2020.04.30.070227v1?rss=1 Authors: Kikumoto, A., Mayr, U. Abstract: Prominent theories of action control propose that conjunctive representations, which integrate task-relevant features in a nonlinear manner, are critical for successful action control. Thus, in order to stop an initiated action, which is a key aspect of self-control, conjunctive representations should be the primary target of the stopping process. We tested this hypothesis by combining a rule-based action selection task with the stop-signal paradigm. Participants selected actions based on abstract stimulus-response rules and occasionally received a stop-signal as a prompt to halt the intended action. Using time-resolved representational similarity analysis of the EEG signal, we decoded both orthogonal, constituent action-relevant representations (rules, stimuli, and responses) and conjunctions of these features in a time-resolved manner and on the level of single trials. In Exp. 1, where a short stop-signal interval (100 ms) ensured high stopping success, simple action-relevant features and their nonlinear conjunction were robustly expressed in the EEG signal. Importantly, the conjunctive representation was selectively suppressed on stop trials. In Exp. 2, an adaptive staircase targeting a stopping success of 50%. Here, conjunctions were selectively suppressed on successful stop trials compared to both go and failed stop trials. Moreover, the strength of conjunctive representations at the time of the stop-signal uniquely predicted stopping failures. Combined, these results clarify that the stopping process does not just target motor output representations. Rather, conjunctive representations seem to be critical for selecting a specific action and for that reason also need to be suppressed by the stopping process in order to cancel intended actions. Copy rights belong to original authors. Visit the link for more info