Neutral faces, threatened brains: how social anxiety reshapes face processing in childhood
- In 10–15-year-olds with social anxiety disorder (SAD, n=53), clinical controls with specific phobia (SP, n=41), and healthy controls (HC, n=61), neutral child faces were judged more negatively by the SAD group than by healthy peers – a subjective interpretive bias present already in childhood.
- At the neural level, SAD youth showed reduced N170 amplitudes to all neutral faces relative to both control groups, independent of the surrounding social context – a structural-encoding signature that distinguishes anxious from non-anxious and from phobic children.
- Social context cues shifted both ratings and early neural responses (more negative ratings for negatively framed faces) in every group, meaning context sensitivity itself was intact and not driven by psychopathology.
- Only the younger SAD children (10–12 years) showed elevated Late Positive Potential (LPP) amplitudes versus younger healthy controls – hinting that early childhood may be a sensitive window for sustained attentional engagement with ambiguous faces.
Social anxiety disorder is among the earliest-emerging anxiety conditions, yet most neurobiological work has been conducted in adults, leaving the developmental origins of its maintaining mechanisms underspecified. This German multi-site study (Leipzig, Freiburg, Potsdam) addresses that gap directly by recording continuous EEG while children and adolescents viewed neutral child faces embedded in short verbal vignettes of negative, neutral, or positive valence. The three-group design – SAD, specific phobia, and healthy controls – is the methodological strength here: by including an anxious-but-not-socially-anxious clinical comparison, the authors can separate a general anxiety signal from a disorder-specific one.
The central result is the blunted N170. The N170 is a face-sensitive event-related potential reflecting early structural encoding of faces, typically peaking around 170 ms over occipitotemporal sites. The SAD group's reduced N170 to neutral faces, present across all context conditions and absent in the phobia group, suggests an alteration in the perceptual front end of face processing that is specific to social fear rather than to anxiety in general. This challenges the intuitive hypertvigilance narrative: rather than an amplified early response, socially anxious youth showed an attenuated one, consistent with accounts of avoidant or disengaged early encoding that may later cascade into biased interpretation.
The subjective data complete the picture. SAD youth rated all faces – even neutral and positively framed ones – as more negative, the hallmark interpretive bias of the disorder. Critically, contextual framing moved ratings and early potentials in every group equally; context processing was not itself disordered. The pathology lies in the baseline valence assigned to ambiguous social stimuli, not in an inability to use context.
The age-moderated LPP finding is the most developmentally provocative. The LPP indexes sustained, motivated attention to emotionally salient stimuli. Its elevation only in the youngest SAD children, and not in adolescents, raises the possibility that sustained engagement with ambiguous faces is an early feature that is later replaced or masked by avoidance – a maturational shift that cross-sectional data can only hint at. For clinicians, this reframes social anxiety not as a single static bias but as a moving target across development.
Why early structural encoding matters
The N170 result locates part of the SAD mechanism upstream of conscious appraisal. If the very encoding of faces is altered, interventions targeting only top-down reappraisal may miss a perceptual contribution. Attention-training and exposure paradigms that re-engage early face processing deserve renewed scrutiny in youth samples.
From mechanism to the clinic
This is a mechanistic, not a treatment study, and it does not tell us which children will remit. But it offers a developmentally anchored marker: a face-specific neural attenuation that is already detectable by age 10 and that separates social fear from phobia. Clinicians should treat early, ambiguous-stimulus interpretive bias as a tractable target and remain alert to age as a moderator of how that bias presents.
Socially anxious children did not over-respond to faces – they under-encoded them, suggesting the mechanism begins at perception, not appraisal.
Cross-sectional design cannot establish whether the N170 attenuation is a cause, consequence, or correlate of SAD; the age effect on LPP is inferred from group comparisons, not longitudinal follow-up. Stimuli were neutral child faces only, limiting generalisation to adult or emotional faces. Sample sizes per cell are modest for ERP age-moderation analyses, and the cohort was drawn from a single national context.