New evidence helps explain why social situations can feel treacherous for people high in social anxiety: even a friendly face may not look quite so friendly.
A smile is supposed to be the simplest signal in the social world. Yet for people prone to social anxiety, that signal can blur into doubt. New research in PLOS One finds that after a mild social stressor, people with high trait social anxiety grew worse at recognizing happy expressions, while neutral faces remained especially hard to read.
The finding dovetails with cognitive models of social anxiety that predict a bias toward interpreting ambiguous cues negatively. It also underscores how fragile the social reward system can feel when anxiety creeps in, subtly shifting how we see others and, in turn, how we behave.
Why faces can mislead anxious minds
Facial expressions prime our decisions to approach or avoid. A scowl hints at conflict, a neutral stare invites guesswork, and a genuine smile typically smooths the path to connection. For people high in social anxiety, that interpretive machinery often tilts toward threat—especially when cues are ambiguous. Prior work, including a 2023 meta-analysis, has shown that people with social anxiety disorder struggle most with neutral and, unexpectedly, happy faces, sometimes reading them as mocking or untrustworthy.
That pattern matters because misreading a smile can short-circuit social reward. When a positive cue fails to register, people may avert their gaze, withdraw, or brace for rejection—behaviors that maintain anxiety over time.
The new study
What the researchers did
In the new PLOS One paper, University of Ottawa researchers Chloé Lacombe, Kim Dubé, and Catherine Collin recruited 68 undergraduates who scored high on trait social anxiety; 52 participants’ data were included in the final analyses. Each participant completed two facial emotion recognition tasks—one before and one after a brief, socially evaluative conversation designed to raise state anxiety.
Across 120 trials, participants viewed standardized images from the Karolinska Directed Emotional Faces database showing happy, angry, or neutral expressions. After each 100-millisecond face presentation, participants used six sliders (anger, surprise, fear, sadness, happy, disgust) to rate what they perceived. The team analyzed three outcomes: accuracy (did the participant identify the target emotion), intensity (how strongly it was perceived), and saliency (how purely the target emotion appeared versus being mixed with others).
To gauge whether the conversation raised anxiety, participants rated their state anxiety at multiple points using the Subjective Units of Distress Scale.
What they found
First, the manipulation check: the five-minute discussion did not significantly increase state anxiety on average. Even so, performance shifted in telling ways. After the conversation, participants became significantly worse at recognizing happy faces. Neutral faces were consistently difficult before and after; they were often mistaken for sadness or other negative expressions. Angry faces were frequently confused with disgust—two cues that both imply potential social threat.
Participants also rated their own conversation performance more harshly than the confederate’s, echoing a well-documented self-evaluation bias in social anxiety. Notably, happy expressions remained easier overall than neutral or angry ones; the key change was a post-discussion dip in recognizing happiness accurately.
“This study highlights that positively-valenced expressions are met with increased uncertainty particularly when experiencing elevations in state anxiety.” — Lacombe, Dubé, and Collin, PLOS One (2025)
Why it matters
Social anxiety is often framed as a fear of negative evaluation, but the new results suggest an additional, quieter mechanism: moments of heightened tension can erode confidence in positive cues. If a warm expression reads as ambiguous—or worse, as disapproval—people may disengage, deny themselves social reward, and reinforce expectations of rejection.
These perceptual shifts line up with decades of theory. When anxiety narrows attention and fuels hypervigilance, ambiguous cues loom large. The brain, primed for threat, fills gaps with caution. Over time, that bias can become self-fulfilling: less eye contact and fewer social bids mean fewer opportunities to gather corrective feedback that smiles are genuine and interactions are safe.
Caveats and context
The study was carefully controlled but has important limitations. The social stressor—a five-minute evaluative conversation—did not reliably raise state anxiety, which likely muted effects. The sample was relatively small, drawn from a university population, and there was no low–social-anxiety comparison group. The stimuli featured standardized, primarily white faces, which may limit generalizability across cultures and contexts.
Even with those constraints, the pattern is consistent with broader evidence: anxious states can degrade emotion recognition, and in socially anxious individuals, uncertainty often clusters around neutral and positive expressions. The take-home is less about any single stressor and more about how fragile the perception of positivity can be when anxiety enters the frame.
What comes next
Future studies can sharpen the picture by using longer or more potent mood-induction procedures, adding multi-item state anxiety measures, and including comparison groups. Tasks that track gaze patterns may also reveal whether people with social anxiety sample less information from the eye and mouth regions—critical zones for reading genuine happiness.
Therapeutically, the work adds weight to approaches that help people pause and consider alternative interpretations of ambiguous faces, and that gently retrain attention toward prosocial cues. Restoring trust in a smile is not a trivial goal; it may be one of the most direct ways to reopen access to everyday social reward.