Fear generalization of implicit conditioned facial features - Behavioral and magnetoencephalographic correlates

Acquired fear responses often generalize from conditioned stimuli (CS) towards perceptually similar, but harmless generalization stimuli (GS). Knowledge on similarities between CS and GS may be explicit or implicit. Employing behavioral measures and whole-head magnetoencephalography, we here investigated the neurocognitive mechanisms underpinning implicit fear generalization. Twenty-nine participants underwent a classical conditioning procedure in which 32 different faces were either paired with an aversive scream (16 CS+) or remained unpaired (16 CS-). CS+ and CS- faces systematically differed from each other regarding their ratio of eye distance and mouth width. High versus low values on this "threat-related feature (TF)" implicitly predicted the presence or absence of the aversive scream. In pre- and post-conditioning phases, all CS and 32 novel GS faces were presented. 16 GS+ ​faces shared the TF of the 16 CS+ ​faces, while 16 ​GS- faces shared the TF of the 16 CS- faces. Behavioral tests confirmed that participants were fully unaware of TF-US contingencies. CS+ ​compared to CS- faces revealed higher unpleasantness, arousal and US-expectancy ratings. A generalization of these behavioral fear responses to GS+ ​compared to GS- faces was observed by trend only. Source-estimations of event-related fields showed stronger neural responses to both CS+ and GS+ ​compared to CS- and GS- in anterior temporal (<100 ​ms) and temporo-occipital (<150 ​ms; 553-587 ​ms) ventral brain regions. Reverse effects were found in dorsal frontal areas (<100 ​ms; 173-203 ​ms; 257-290 ​ms). Neural data also revealed selectively enhanced responses to CS+ ​but not GS+ ​stimuli in occipital regions (110-167 ​ms; 330-413 ​ms), indicating perceptual discrimination. Our data suggest that the prioritized perceptual analysis of threat-associated conditioned faces in ventral networks rapidly generalizes to novel faces sharing threat-related features. This generalization process occurs in absence of contingency awareness and may thus contribute to implicit attentional biases. The coexisting perceptual discrimination suggests that fear generalization is not a mere consequence of insufficient stimulus discrimination but rather an active, integrative process.