In this study, we merged methods from engineering control theory, machine learning, and human neuroimaging to critically test the putative role of the dorsal anterior cingulate cortex (dACC) in performance monitoring during an emotion regulation task. Healthy adult participants (n=75) underwent cued-recall of affective image stimuli with concurrent functional magnetic resonance imaging and psychophysiological response recording. During cued-recall, participants engaged in explicit self-regulation of their affective state toward defined affective goals. Established decoding methods measured affect processing from fMRI BOLD signals across the orthogonal affective dimensions of valence and arousal. We independently validated participants’ affective state representations via stimulus-dependent facial electromyography (valence) and electrodermal activity (arousal) responses. We then used the decoded affective signatures to test and compare four computational models of performance monitoring (i.e., error, predicted response outcome, action-value, and conflict) by their relative abilities to explain task-related dACC activation. We found that the dACC most plausibly encodes action-value for both valence and arousal processing. We confirmed that the dACC directly encodes affective arousal and also likely encodes recruitment of attention and regulation resources. Beyond its contribution to parsing the roles of the dACC in emotion regulation, this study introduced a novel analytical framework through which affect processing and regulation may be functionally dissociated, thereby permitting mechanistic analysis of real-world emotion regulation strategies, e.g., distraction and reappraisal, which are widely employed in cognitive behavioral therapy to address clinical deficits in emotion regulation.
bioRxiv Subject Collection: Neuroscience