Visuocortical, Autonomic, and Behavioral Dynamics During the Generalization of Acquired Social Fear

When acquiring fear through aversive conditioning, biological organisms tend to generalize the acquired fear response to stimuli that resemble the conditioned threat cue (i.e. the CS+). Applied to interpersonal interactions, the generalization of a CS+ identity to similar faces is the basis of identifying potentially dangerous conspecifics in such an environment. The visuocortical correlates of fear generalization have been extensively studied using simple visual stimuli such as shapes and Gabor patches, but little work has examined complex stimuli such as faces. The present study investigated how neural and psychophysiological responses generalize across facial identities during Pavlovian aversive conditioning. Subjects were presented seven flickering faces varying along a similarity gradient, while measuring electroencephalography (EEG) and pupil diameter. One of these faces (CS+) was repeatedly paired with a loud noise (US). Various measures of affect-biased attention, including the steady-state visual evoked potential (ssVEP), endogenous alpha power, pupil diameter, and self-reported valence and arousal were collected. Trials were separated into early and late periods to analyze the effect of contingency learning on conditioned responses. Data were fit to a priori weights via nonparametric frequentist and Bayesian methods to examine the shape of the observed tuning functions. We found that across all faces, occipitoparietal ssVEP power fit a sharpening pattern resembling a Difference-of-Gaussians. Contrary to expectations, fit to this sharpening pattern was greatest during early trials and declined later in acquisition. In contrast, alpha responses during late trials showed a broad generalization pattern in occipitotemporal regions associated with processing of facial features. Lastly, pupil and evaluative responses exhibited shallow generalization during early trials, and this tuning function narrowed as learning progressed. These results indicate that (1) conditioned generalization of faces prompts similar changes as observed with simple visual cues; (2) that visuocortical, autonomic, and behavioral indices change at different temporal rates.