Face gender perception influences automatic neural face-sex categorization

Technical Summary: Face Gender Perception Influences Automatic Neural Face-Sex Categorization

Problem Statement
Face-sex perception is typically characterized as a rapid, automatic, and binary categorization process. However, evidence suggests that face gender exists along a graded continuum rather than a strict binary. A central unresolved question in visual perception and social psychology is whether and how graded face gender information modulates the automatic neural processing of face-sex. Previous research has yielded divergent findings, with some studies supporting categorical mechanisms (stepwise neural activation) and others suggesting graded representations (linear modulation). A significant limitation in prior literature is the reliance on laboratory-modified stimuli (e.g., morphed faces, homogenized low-level cues) which may not reflect ecological validity. Furthermore, it remains unclear whether observed neural sensitivities to face-sex stem from intrinsic detection mechanisms or higher-level categorization processes operating on graded perceptual representations.

Methodology
The study employed a two-experiment design utilizing natural, unmanipulated face images to bridge behavioral and neural measures.

• Stimuli: The research utilized a "New set" of 68 natural face portraits (34 male, 34 female) characterized by diverse viewpoints, expressions, and lighting conditions. These were presented in both intact and phase-scrambled versions. Phase scrambling preserved low-level visual properties (Fourier amplitude, color, luminance) while disrupting structural and shape cues necessary for item recognition.
• Experiment 1 (Behavioral): Thirty-four participants performed a fast face-sex classification task (167 ms exposure) followed by a continuous gender rating task. Participants rated the masculinity/femininity of each intact face on a scale from -100 to +100. This allowed the researchers to characterize the stimuli based on both classification accuracy/speed and perceived gender distinctiveness.
• Experiment 2 (Neural - FT-EEG): Thirty-two participants underwent electroencephalography (EEG) while viewing Fast Periodic Visual Stimulation (FPVS) sequences. Faces (either male or female) were presented as periodic oddballs at 1 Hz, embedded within a base stream of non-face stimuli (plants, animals, objects) presented at 6 Hz. Crucially, male and female faces were randomly intermixed within the same sequence to prevent exogenous attentional biases. Participants performed an orthogonal target detection task to ensure sustained attention. The EEG data were analyzed using frequency-tagging to isolate face-selective neural responses at 1 Hz (and harmonics) from general visual responses at 6 Hz.

Key Results

• Behavioral Findings (Exp 1):

  • Classification: Participants achieved high accuracy (97.6%) for intact faces, while scrambled faces were classified at chance level (50.1%), confirming that face-sex categorization relies on structural cues.
  • Gender Distinctiveness: Ratings revealed a graded continuum; faces were not perceived as strictly binary but distributed along a feminine-masculine scale. Female faces were rated as more gender-distinctive (more extreme) than male faces.
  • Reaction Times: Faster classification times correlated with higher gender distinctiveness (more masculine or feminine ratings), particularly for male faces.
  • Participant Sex: Female participants responded significantly faster than males, but no "own-gender bias" (superior performance for one's own sex) was observed in classification speed or accuracy.

• Neural Findings (Exp 2):

  • Face-Sex Selectivity: A robust face-selective response was observed at 1 Hz for intact faces but not for scrambled faces, confirming the response was specific to face categorization rather than low-level visual features.
  • Categorical Asymmetry: Male faces elicited significantly stronger face-selective neural amplitudes than female faces. This asymmetry persisted despite male faces showing higher variability in behavioral gender ratings, suggesting a more homogeneous neural representation for the male category.
  • Modulation by Graded Gender: Neural activity was modulated by the degree of gender distinctiveness. Faces rated as more gender-distinctive (either highly masculine or highly feminine) elicited enhanced activity, specifically showing a stronger right occipital-temporal (rOT) dominance.
  • Participant Sex Effects: While participant sex did not modulate face-selective responses, female participants exhibited significantly higher amplitudes in the general visual response (6 Hz), suggesting a general task advantage or higher sensory sensitivity rather than a face-specific effect.

Key Contributions

1. Integration of Categorical and Graded Representations: The study demonstrates that automatic neural face-sex categorization is not purely binary. While the neural system exhibits rapid categorical selectivity (distinguishing male from female), this process is simultaneously modulated by the graded distinctiveness of gender cues.
2. Ecological Validity: By using natural, unmanipulated images, the research confirms that graded gender perception and its neural correlates exist in real-world visual contexts, extending findings previously limited to morphed or artificial stimuli.
3. Neural Mechanisms of Distinctiveness: The findings link behavioral distinctiveness to neural efficiency, showing that faces with clearer gender cues (more distinct) drive stronger and more consistent phase-locked neural activity, particularly in the right hemisphere.
4. Homogeneity of Male Categorization: The results support the hypothesis that male faces may form a more homogeneous neural category than female faces, evidenced by stronger 1-Hz amplitudes despite higher behavioral rating variability.

Significance and Claims
The authors claim that their findings support a "multilevel account" of face perception. In this framework, the extraction of face-sex as a rapid, categorical signal coexists with continuous, graded perceptual representations of gender-related facial cues. The study establishes that naturalistic face stimuli can be used to probe fine-grained face-sex representations, offering a more realistic model of how the brain processes gender in daily life.

Crucially, the paper posits that the automatic neural processing of face-sex is not rigid; it is sensitive to the "graded gender distinctiveness" of the stimulus. The stronger neural response to male faces is interpreted as evidence of a more homogeneous category representation, while the modulation by distinctiveness suggests that the brain's face-selective system is tuned to the clarity of gender cues. The absence of an own-gender bias in these automatic processes challenges the notion that social familiarity drives the initial neural categorization of sex, suggesting instead that these mechanisms are largely universal and driven by perceptual structure.