1 - (SYM 97) Fmri Evidence for the Prediction Error-like Processing of Absent Danger in the Reward and Salience Pathway and Its Relation to Self-reported Relief

We feel relieved when an imminent threat is unexpectedly averted. Accumulating evidence suggests that this relief might be an emotional correlate of the theoretical prediction error (PE) driving the learning of safety during exposure therapy. Accordingly, increasing evidence points to similarities between the neural processing of absent threat and the well-established reward PE. Yet, how exactly omissions of threat are processed by the human brain, and how this is related to the subjective feelings of relief remains unclear.

To answer these questions, 31 healthy volunteers performed the previously validated Expectancy Violation Assessment (EVA) task within an MRI scanner. On each trial, participants were presented with probability and intensity instructions of an upcoming aversive electrical stimulation to the wrist, time-locked by a countdown clock. Most trials, however, did not contain the electrical stimulation and therefore constituted a violation of expectancies. We measured ratings of relief-pleasantness, omission-induced fMRI responses and changes in skin conductivity during all omitted stimulations. We predicted that unexpected omissions of threat would elicit reward PE-like activations in the Nucleus Accumbens (NAC), Ventral Tegmental Area (VTA), ventromedial Prefrontal Cortex (vmPFC) and Ventral Putamen (VP); meaning that the signal would increase as a function of instructed probability and intensity; but that completely predicted outcomes (0% and 100% trials) would elicit equivalent fMRI activation. Additionally, we predicted that this activity would be related to subjective relief.

We found that unexpected omissions of threat elicited greater fMRI activations than fully expected omissions in the VTA and VP. However, none of the regions univocally satisfied the PE-requirements. Still, we found that VTA activity increased with increasing intensity and at trend-level probability (p=.055) of omitted threat; and that the activity of both VTA and VP was positively related to subjective relief-pleasantness on a trial-by-trial level.

Together, our findings provide additional support for an overlap in the neural processing of absent danger and rewards in humans.