Fig. 2: Subjective, physiological and neural responses during the extinction-training protocol.

Panel a: Mean scores and standard errors for US expectancy ratings, SCRs and startle blink magnitudes, respectively, in study 1 as a function of stimulus type (CS+ and CS−, as well as US and ITI in case of SCR and startle, respectively) with two trials per block except for the single reacquisition trial. During phases of preconditioning, acquisition training, and extinction training, each block included two startle probes for both CS+ and CS− and four startle probes during the ITI. Panel b: BOLD group activation during early extinction training. BOLD activation assessed by the t-contrast CS + <CS− (p < 0.001 uncorr., k = 20) is illustrated in blue/green, and BOLD activation assessed by the t-contrast CS + > CS− (p < 0.001 uncorr., k = 20) is illustrated in red/yellow. The bar plots show mean extracted beta values and standard errors in early and late extinction training and during return of fear test procedure (reinstatement). During early extinction training, the ventromedial prefrontal cortex (vmPFC) and left hippocampus (HC L) were significantly more strongly activated for CS− than for CS+ (left), whereas the dorsomedial prefrontal cortex (dmPFC) and anterior insula (aINS) were significantly more strongly activated for CS+ than for CS−. Asterisks mark significant interactions (Stimulus×Time) between early extinction training and return of fear test phase (p < 0.001). The VmPFC was significantly more strongly activated for CS− than CS+ in early extinction, which diminished after reinstatement administration. Activity in the right aINS decreased for both CSs after reinstatement administration compared to early extinction with a stronger decrease to the CS+. CS conditioned stimulus; US unconditioned stimulus; ITI intertrial interval; SCR skin conductance response.