Figure 3
From: Defensive freezing and its relation to approach–avoidance decision-making under threat
Heart rate and computational modeling results. Panel (a) depicts the average trial time course of baseline corrected heart rate (HR) as a function of the varying shock amounts, with the gray shaded area representing the time window of interest used in all analyses. The vertical dashed line represents the earliest possible target movement onset. Model comparison of various freeze models (b) showed that the best fitting model included the subjective value (i.e., potential money and shock amounts), bradycardia, the action context, and the interaction between the subjective value and action context parameters (indicated with a red asterisk). Additionally, on the subject level, stronger threat-related bradycardia (i.e., high vs. low threat) was related to the integration of action context (AC) and subjective value (SV) for approach–avoidance choices (c), but not with AC or SV alone (d,e). Simulations show that higher values of this interaction parameter led to a diminished effect of subjective value on choice in situations incongruent with passive/active response tendencies (f–h; for brevity, only simulations of passive ‘participants’ with a positive SV:AC interaction are shown here, see Supplementary Information for the full results). Model AIC scores are plotted as the difference from the best fitting model (ΔAIC); lower AIC scores indicate better fit. The blue dashed line (b) represents the reference model’s fit and the red asterisk indicates the overall winning model. Solid vs. dashed regression lines in (c–e) reflect significant versus non-significant relationships. Error bars in (f–h) represent one standard error of the mean (SEM). SV = subjective value, AC = action context, Fr = umbrella term for freeze indices: All models were fitted with either bradycardia or body sway as freeze index, see Supplementary Information; colons (:) denote interactions.
