Spatio-temporal dynamics of safety learning during fear acquisition and fear generalization.

Existing neural studies have predominantly focused on threat processing during fear acquisition and generalization, whereas the spatiotemporal dynamics of safety learning remain poorly understood. Here, we examined safety learning across early acquisition, late acquisition, and fear generalization, while characterizing early and late event-related potential (ERP) responses using a fear conditioning and generalization paradigm with multiple conditioning sets (N = 29). Behavioral ratings and reaction times indicated enhanced safety learning from early to late acquisition, together with a generalization gradient from threat to safety stimuli. Non-parametric cluster-based permutation tests (CBPT) revealed larger frontal N2 amplitudes to the safety cue than the threat cue across all experimental phases. During fear generalization, both the safety cue and the generalization stimuli (GS1-GS4) elicited larger occipital P2 amplitudes than the threat cue. The safety cue further elicited larger right frontocentral late positive potential (LPP) amplitudes than the threat cue and the generalization stimuli, as well as larger sustained anterior negativity (SAN) than the generalization stimuli. Descriptive source reconstruction suggested shifts in the estimated sources of the N2 from early to late acquisition, whereas source estimates of the P2, LPP, and SAN showed spatial overlap with medial prefrontal, limbic subcortical, occipitotemporal, medial temporal, medial parietal, and cerebellar regions. Together, these findings provide a preliminary characterization of the spatiotemporal dynamics of safety learning across fear acquisition and generalization and establish a normative framework for future investigations of altered safety learning in anxiety disorders.