The research team, led by Yumnah T. Khan of the Autism Research Centre in the Department of Psychiatry, together with colleagues including Alex Tsompanidis, Carrie Allison, neonatologist Topun Austin, and Simon Baron-Cohen, tested 130 newborns, 67 males and 63 females, at a mean age of just 33 hours after birth. That extraordinarily tight window is the study’s methodological centrepiece. Infants tested at 33 hours of life have experienced almost none of the social world: no extended eye contact, no named gender, no toys, no language, and no deliberate socialisation. Any behavioural difference observed at this stage, the researchers argue, is far harder to attribute to learning than differences measured in older babies, toddlers, or children.

The experimental design was deliberately minimal. Each newborn was presented simultaneously with a video of a human face and a video of a non-social object, while gaze-tracking equipment recorded precisely where the infant looked and for how long. The stimuli were matched in their presentation, and the infants’ spontaneous looking behaviour served as the sole measure. Eye-tracking is the gold standard for interrogating early attention because newborns cannot follow verbal instructions, and their visual behaviour is one of the few channels through which the hidden architecture of their emerging cognition can be observed at all.

The results split into two revealing layers. When the researchers analysed percentage looking time — the proportion of total looking devoted to the face rather than the object — female newborns on average allocated a greater percentage of their attention to the face than male newborns did. The effect size was moderate, at Cohen’s d of 0.40, a magnitude that is meaningful in behavioural science without being dramatic. When the analysis shifted to absolute looking times, an equally nuanced pattern appeared: females showed a genuine preference, looking longer at the face than at the non-social object, whereas males showed no preference for either stimulus.

Crucially, however, the team found no statistically significant differences between the sexes in the absolute amount of time each group spent looking at the face, or at the object. That absence is what elevates the study above a simple headline. The sex difference is not that girls look more at faces in total, nor that boys look less; it is that girls distribute their attention differently between the two types of stimuli. Sex differences in this experiment live in the relative allocation of attention — in the internal comparison an infant appears to make between a social signal and a non-social one — rather than in raw visual appetite.

The significance of that distinction extends to how scientists interpret the causes of behavioural sex differences more broadly. Decades of research have documented on-average psychological differences between males and females, from interests and social orientation to the prevalence of particular neurodevelopmental conditions. But the scientific community remains divided over the engines of those differences. One camp emphasises gender socialisation: from birth, children are treated, spoken to, and encouraged differently depending on their sex, and any observed divergence could in principle be the residue of that immersion. The other camp allows for partial biological contributions, notably prenatal sex hormones such as testosterone, which differ systematically between male and female foetuses and are known to influence brain development.

Newborns are among the very few human populations capable of adjudicating between these accounts, at least in principle. Because infants at 33 hours have had negligible postnatal social experience, a reliable behavioural difference at this age suggests that something occurring before or at birth — prenatal hormones, genetic factors, or differences in maturational tempo — may contribute to the earliest organisation of social attention. The Cambridge team is careful, in its published abstract and plain-language summary, to frame prenatal sex hormones as one possible explanation rather than a proven cause. They also name two credible alternatives that the current design cannot fully exclude.

The first alternative concerns maturation. Male and female foetuses develop on slightly different schedules, and differences in the maturity of visual or attentional systems at birth could produce divergent looking patterns without any specifically social mechanism being involved. A less visually mature infant might, for instance, engage more diffusely with both stimuli, diluting any face preference. The second alternative concerns the stimuli themselves. Low-level visual properties — contrast, motion, complexity, colour, luminance — can drive infant looking independently of whether a stimulus is social. If the face video happened to possess visual features that interact differently with male and female early visual systems, the observed effect could reflect those properties rather than the face’s social meaning. The authors present these explanations not as disqualifiers but as honest boundary conditions on interpretation.

Context strengthens the new findings considerably. Previous research on neonatal sex differences in face preference is described by the team as extremely limited, with the small number of existing studies yielding mixed results, raising methodological concerns, and remaining unreplicated. The new work, with a sample of 130 neonates — large by the demanding standards of newborn research, where recruiting families within hours of birth is logistically fraught — represents one of the most rigorous attempts to date to re-evaluate the question. Conducted at the Rosie Hospital in Cambridge with NHS Research Ethics Committee approval and written informed maternal consent, the study benefits from a controlled protocol, objective eye-tracking measurement, and a sample size that gives the moderate effect observed some statistical credibility.

The longer-term stakes reach into psychiatry and neurodevelopmental medicine. Conditions such as autism are diagnosed more frequently in males than females, and the reasons for that imbalance remain unresolved, spanning biological, diagnostic, and social hypotheses. If sex differences in the earliest attentional orientation toward social stimuli are genuinely present from birth, they may constitute one thread in the developmental fabric that eventually produces sex-differentiated profiles of social behaviour and vulnerability. The authors explicitly note that findings of this kind may help explain why several psychiatric and neurodevelopmental conditions show sex differences in prevalence — a claim offered as a research direction rather than a conclusion, and one that will require longitudinal follow-up linking newborn gaze measures to later developmental outcomes.

For now, the study’s most durable contribution is conceptual. It demonstrates that the relative allocation of attention between social and non-social visual worlds differs on average between male and female humans at the very threshold of postnatal life, while showing that neither sex looks more overall. That pattern is subtle, and the authors resist triumphalism: the effect could reflect prenatal biology, maturational differences, stimulus properties, or some combination of all three. What the study removes from the table, however, is the comfortable assumption that attentional sex differences must be products of the social environment alone. At 33 hours old, most of the social environment has not yet begun. Whatever tips these newborn girls’ eyes toward faces a little more often, it appears to have been installed before the first feed — and scientists now have a sharper, better-powered window into when and perhaps how that installation happens.

Subject of Research: Sex differences in newborn face-looking preferences and their possible prenatal origins

Article Title: Sex differences in face looking preferences at birth

Article References: Sex differences in face looking preferences at birth. (n.d.). https://doi.org/10.1186/s13293-026-00976-9

Image Credits: AI Generated

DOI: 10.1186/s13293-026-00976-9

Keywords: sex differences, newborns, face preference, eye-tracking, attention, prenatal hormones, neonatal development, social cognition, neurodevelopment, autism, infancy, Biology of Sex Differences

The starting point for the argument is a striking feature of human social life. Anthropologist Robin Dunbar’s work on social networks shows that although people typically maintain 150 to 250 social ties, only three to five of these are genuinely close relationships, usually involving parents, children, partners, and best friends. Close relationships are qualitatively different from casual ones: they are enduring, emotionally significant bonds marked by mutual interdependence, frequent interaction, and deep cognitive and affective involvement of both self and other. Casual or instrumental ties, by contrast, are relatively transient, organized around specific goals or resources, and involve minimal emotional investment. Recent functional neuroimaging work by Roseman, Dunbar, and Arzy suggests that these two categories are handled by partly distinct neural circuitry. Thinking about close others activates the temporoparietal junction, precuneus, dorsomedial prefrontal cortex, and default mode network A, while casual ties recruit the medial temporal lobe, retrosplenial cortex, ventromedial prefrontal cortex, and default mode network C. Remarkably, brain responses to close relationships correlated strongly with responses to small spaces such as a room or household, whereas casual ties tracked with large spaces like a town or region. Reaction times were also significantly shorter when people thought about their close people, suggesting that close relationships are processed more rapidly, with stronger self-related and intuitive components.

Yet despite this growing neuroscientific picture, Winnette identifies a conceptual gap. Attachment theory, the dominant framework for early social development, explains how infants form bonds with caregivers and build internal working models, cognitive-affective representations of self and others that guide expectations, emotion regulation, and behavior in later relationships. Object relations theory similarly proposes that early interactions are internalized as enduring mental representations that serve as road maps for social life. But neither tradition, the article argues, explicitly explains how individuals come to recognize and classify relationships as belonging to qualitatively different types, that is, how we learn that some relationships are close and others are not. The neuroscience tells us that close and casual ties involve different cognitive processes, but the developmental mechanism by which these categories are acquired has remained insufficiently understood.

The proposed solution draws on a concept from memory research: the memory schema. As defined by Ghosh and Gilboa, a memory schema is a multidimensional, dynamic body of knowledge that emerges from integrating multiple interrelated experiences. Schemas encode commonalities across episodes, enabling abstraction and generalization, while discarding the details of individual events. They are inherently associative, linking perceptual, conceptual, and contextual elements, and they can incorporate affective components so that emotions become embedded within their structure. The perspective proposes that repeated, emotionally salient interactions with primary caregivers give rise not only to specific attachment bonds but also to a generalized, domain-general memory schema of relational closeness. Through countless interactions, the infant’s memory system abstracts recurring relational features such as predictability, proximity, emotional attunement, and interactional frequency, forming a multidimensional, affectively laden template of what constitutes a close versus a casual relationship.

Crucially, the developmental timing fits what memory science now knows about infancy. Declarative episodic memory, our ability to remember specific events in rich detail, is limited early in life and improves markedly around age seven or eight, which is why adults experience infantile amnesia. Yet infants are prodigious learners of generalized knowledge. Statistical learning, the ability to detect regularities in repetitive experience, is intensely active from the earliest months, and an fMRI study by Ellis and colleagues showed that the anterior hippocampus supports statistical learning from as early as three months of age, before the trisynaptic pathway supporting episodic memory matures. Alberini and Travaglia’s model of memory development proposes that hippocampal learning proceeds through a cascade of critical periods, with generalized semantic and statistical learning preceding detailed episodic learning. In other words, the infant brain is built first to extract categories and regularities, and only later to encode specific episodes. This hierarchical principle, Winnette argues, is exactly what would be needed to construct a categorical representation of relational closeness before the child can remember any particular close moment.

The building blocks of this schema are laid down through several well-documented learning processes. Newborns arrive with implicit memories of their biological mother’s sounds, smells, and tastes acquired in utero, and can distinguish her face from a stranger’s within hours of birth. Perceptual constancy allows infants to recognize the caregiver across changing conditions, and perceptual learning then builds increasingly nuanced knowledge on top of that stable foundation. Perceptual narrowing progressively tunes infants to the faces and voices of familiar people. Memory generalization, present from birth, preserves the core meaning of experiences while their details fade, and emotionally intense experiences are preferentially retained. Nim Tottenham’s research on early adversity further suggests that medial prefrontal-subcortical circuitry encodes affective memory schemas of early caregiving, so that even without episodic memories of traumatic events, implicit schemas of hostility or coldness can shape relationships for a lifetime. The same mechanism, applied to warm and responsive care, would encode a positive template of closeness.

Interactive social learning in infancy provides the raw material. Imitation, documented in newborns by Meltzoff and Moore and debated ever since, may function not merely as a learning tool for specific behaviors but as a process through which infants extract regularities across interactions, distinguishing familiar, stable partners from less predictable others. Parent-infant synchrony, studied extensively by Ruth Feldman, involves shared affective states, temporal coordination, and biological tuning, providing a structured, emotionally rich context in which infants encode core features of closeness such as mutual regulation and shared affect. Intersubjectivity, the capacity to share another person’s lived experience described by Daniel Stern and Colwyn Trevarthen, appears to be an innate program active within the first months of life, with infant-caregiver face-to-face exchanges statistically resembling informal adult conversation. Meanwhile, the medial prefrontal cortex, though anatomically immature, is partially active in infancy and responds preferentially to self-relevant social stimuli, including a mother’s smile, infant-directed speech, and affectionate touch, which has been shown to speed face recognition in four-month-olds.

The proposed framework thus complements rather than replaces attachment theory. Where attachment theory and object relations theory explain how individuals expect particular relationships to function, the new account explains how individuals come to organize and classify relationships more broadly, introducing cognitive-neuroscientific mechanisms of abstraction and generalization that link early repetitive interactions to the formation of relational categories. The author is careful to acknowledge limitations: the framework is conceptual rather than empirically tested, the construct of a closeness schema has not yet been operationalized, and the model may underrepresent individual differences in temperament and genetics, later-life plasticity, and the possibility that closeness exists on a continuum rather than as a categorical boundary. Future research directions include longitudinal designs, neurocognitive measures that test whether close and casual relationship stimuli differentially engage memory and affect systems, and cross-cultural studies of whether the close-casual distinction is universal.

The stakes of the proposal are considerable. Close relationships are essential for emotional well-being, and their absence or disruption is a primary determinant of loneliness, defined as a perceived discrepancy between desired and actual relational closeness. Loneliness and social isolation rank among the major issues in mental health, and research links early social learning to later relationship quality, loneliness, and social isolation. If Winnette is right, the capacity for intimacy rests on a foundation laid in the earliest weeks and months of life, when an infant’s memory system quietly distills thousands of small interactions into a single, powerful idea: that with this person, interactions are frequent, predictable, emotionally intense, and reliably the same when they return. That abstracted sense of closeness, encoded long before conscious recollection begins, may then serve as the cognitive-affective template on which every subsequent close relationship, from childhood friendships to adult partnerships, is built.

Subject of Research: The developmental emergence of a memory schema of relational closeness in infancy

Article Title: Perspective: Early Development of Memory Schema of Closeness in Social Relationships

Article References: Winnette, P. (2026). Perspective: Early Development of Memory Schema of Closeness in Social Relationships. Trends in Psychology. https://doi.org/10.1007/s43076-026-00538-3

Image Credits: AI Generated

DOI: 10.1007/s43076-026-00538-3

Keywords: memory schema, relational closeness, attachment theory, infancy, social neuroscience, statistical learning, internal working models, primary caregiver, perceptual narrowing, imitation, intersubjectivity, loneliness