A neuroimaging study suggests the brain uses a shared code for gender across images of faces, bodies and objects, with a small cluster in the right middle temporal gyrus emerging as a possible universal gender signal. The work, described as a reviewed preprint, has been published in eLife.
Humans are remarkably adept at inferring gender from visual cues, doing so rapidly and with little conscious effort. Faces are especially informative, but people can also read gender from body shape, posture, clothing, and even the objects associated with men or women.
Researchers asked whether the brain represents gender in a way that transcends the type of image shown. While gender processing from faces has been studied, less was known about bodies and objects and whether these sources converge on a common neural representation.
Wenjie Liu and colleagues from the Institute of Psychology of the Chinese Academy of Sciences used functional magnetic resonance imaging to monitor brain activity as participants viewed images of male and female faces, bodies, and gender-associated objects. They sought to find brain regions that encode gender regardless of the image type, aiming to understand how diverse visual signals are converted into an abstract social category.
Twenty-two healthy adults aged 19 to 28 took part in the main experiment, with an average age of 22; eleven identified as men.
During the scanning, participants viewed six image types in 12-second blocks: male and female faces, male and female bodies, and objects linked to each gender. Each type comprised 30 distinct images. Faces displayed Chinese male and female actors; bodies appeared as full torsos in underwear; objects were chosen for their gender associations from a prior study. Each run lasted 5 minutes 30 seconds and included three presentations of each condition in random order, across six runs (the first two participants completed seven runs).
To ensure attention, participants pressed a button when the same image appeared twice in a row, performing the task with about 97 per cent accuracy; they were not required to judge gender explicitly.
Using a machine-learning approach, the researchers asked whether patterns of brain activity could distinguish male from female images within a given brain region. The sites carrying gender information varied by image type. For faces, gender signals appeared in several visual areas on both sides of the brain—the cuneus, calcarine cortex, lingual gyrus, superior and middle occipital gyri, and middle temporal gyrus—as well as in the right superior parietal lobule and the left fusiform gyrus.
For bodies, gender information emerged across a large portion of the occipitotemporal cortex, with clusters in the superior parietal lobule, inferior parietal lobule, and postcentral gyrus on both sides of the brain. For objects, the relevant regions included the cuneus, calcarine cortex, lingual gyrus, fusiform gyrus, and several occipital and inferior temporal regions, plus the right middle temporal gyrus.
The team then tested cross-decoding: could a classifier trained on faces predict the gender of bodies or objects, and vice versa? The only region where cross-type decoding succeeded for all three pairings was a small cluster in the right middle temporal gyrus, supported by a second analysis that pointed to the same area. This region, however, did not code gender in isolation; it also carried information about whether the viewed stimulus was a face, a body or an object.
To probe the nature of the information used, the researchers compared the right middle temporal gyrus activity with an artificial neural network trained to classify images by gender. The brain region’s responses most closely resembled the network’s middle layers, suggesting reliance on mid-level visual features such as contours, textures, shapes and spatial arrangements that lie between basic properties and whole-object recognition.
Further, the team explored how gender-related regions interacted. The connectivity patterns were moderately similar when participants viewed faces or bodies, but not when viewing objects.
“Together, these findings identify a category-general neural hub for gender perception and illuminate how mid-level visual features and distributed networks support the abstraction of gender across heterogeneous visual inputs,” the authors concluded.
The study advances understanding of the neural basis of social perception, but eLife’s reviewers described the evidence as incomplete, citing conceptual concerns, weak statistical methods, and potential confounds such as differences in image size. They advised tempering claims that the identified region holds an abstract representation of gender, and the preprint had not been revised in response at the time of review.
The researchers acknowledge several limitations: the small sample of 22 young adults, and the fact that participants did not need to attend to gender during the task. Because the faces depicted Chinese actors and the gender associations of objects are culturally influenced, the results may not generalise to other populations.
The paper, titled A shared neural code for gender across faces, bodies, and objects in the human brain, is authored by Wenjie Liu, Xiqian Lu, Yuhui Cheng, Ruidi Wang, Xinquan Lu, Xiangyong Yuan, and Yi Jiang.
