Social referencing is the process of looking to another person for emotional cues when you encounter something unfamiliar or uncertain, then using those cues to decide what to do. It is one of the earliest and most important social skills humans develop, typically appearing in recognizable form between about 10 and 12 months of age. While it is most studied in infants glancing at a parent’s face before approaching a strange toy or a sudden noise, the same basic process persists throughout life and even shows up in other species. The research on how it works, when it goes wrong, and what shapes it is richer and stranger than a simple “baby looks at mom” story suggests.
How It Develops in the First Year
Babies do not come into the world ready to social-reference. The skill emerges gradually. A landmark study tracking infants across three age groups found clear shifts in how children used parental cues. Infants between six and nine months looked at their parents but showed no particular preference for looking at the parent’s face versus other parts of the body. They also tended to look more when the parent was expressing positive emotion. Older infants, between 10 and 13 months, became more selective: they zeroed in on the parent’s face specifically, and they began looking more when the parent displayed fear toward a stimulus. Only infants older than 14 months consistently waited to touch a new toy until after they had checked the parent’s reaction first.
A related study found that by 10 months, infants adjust their referencing based on whether the other person is actually paying attention to them. Seven-month-olds looked toward an experimenter after a startling event regardless of whether that person was attending to them. Ten-month-olds, though, increased their looks only when the experimenter was already looking their way. That shift suggests that somewhere around 10 months, babies begin to understand that the other person’s attention matters for whether their emotional signals are relevant.
Voice Carries More Weight Than the Face
Most people picture social referencing as a baby scanning a parent’s facial expression, but research consistently shows that vocal cues are at least as powerful, and often more so. In a visual cliff experiment where 12-month-olds encountered an apparent drop-off, infants who heard their mother’s voice alone crossed the cliff faster than infants who could only see her facial expression. When voice and face were combined, infants looked at the mother more, but voice alone was enough to guide behavior effectively.
The type of emotional message matters too. Positive vocal signals, like an encouraging tone of voice, appear to be more frequent and more effective in shaping infant behavior than fearful vocal signals. That finding runs counter to what some earlier studies had predicted, suggesting that encouragement may be a more reliable channel for guiding babies than warnings are.
This makes intuitive sense if you think about the situations where social referencing happens. A baby encountering something new is often not looking directly at the parent; they are looking at the novel object or event. Voice reaches them regardless of where their gaze is pointed. Facial expressions require the baby to already be looking in the right direction, which adds a step to the process.
What Babies Do With Ambiguity
Social referencing is most active when the situation is genuinely uncertain. In experiments using a visual cliff, where a transparent surface creates the illusion of a steep drop, researchers varied the apparent depth. When the cliff looked obviously deep, infants crossed much more slowly and spent longer looking at their mother’s face before deciding what to do. When the cliff was shallow and less ambiguous, infants crossed faster and their initial looks toward their mother were shorter. The first look to the parent in the deep condition was notably prolonged, suggesting the infant was actively seeking information rather than just reflexively glancing over.
This distinction between information-seeking looks and comfort-seeking looks is one of the more nuanced aspects of the research. Not every glance at a caregiver during a stressful moment is social referencing in the strict sense. Sometimes a baby looks at the parent because they want reassurance, not because they are trying to read the parent’s emotional verdict on the situation. Researchers distinguish these by looking at the timing and duration of the glances and whether the baby subsequently adjusts their behavior toward the novel stimulus based on what they saw.
Attachment and Temperament Shape the Pattern
Not all infants reference their caregivers in the same way, and the differences connect to the child’s attachment style and temperament. In one study, insecure-resistant infants, those who tend to be clingy and distressed during separations, referenced their caregivers most frequently and started doing so early in an encounter with something new. Insecure-avoidant infants, who tend to seem indifferent during separations, referenced the least. Securely attached infants fell in between. The finding suggests that social referencing is not just about gathering information; it also reflects the broader relationship pattern between parent and child.
Temperament adds another layer. A broad temperament factor predicted the information-seeking component of social referencing in one study, but the expected links between attachment classification and social referencing behavior were not consistently confirmed. In other words, a baby’s innate disposition toward wariness or boldness affects how much they seek out emotional cues from others, and this effect is at least partly independent of the attachment relationship. Disentangling temperament from attachment remains one of the trickier problems in the field.
How Parental Anxiety Gets Transmitted
One of the most practically important findings in social referencing research is that the process can serve as a pipeline for passing anxiety from parent to child. When mothers were instructed to interact with a stranger in a socially anxious manner, their infants subsequently became more fearful and avoidant of that stranger compared to infants whose mothers had interacted normally. The infants were not simply reacting to the stranger being unfamiliar; they were picking up on the emotional quality of their mother’s interaction and applying it to their own behavior.
This is not limited to momentary experimental effects. In a prospective study following families over time, toddlers whose parents had a lifetime history of social anxiety disorder showed more fear and avoidance in social referencing situations than toddlers of parents without anxiety disorders. Parents with other types of anxiety disorders but not social anxiety did not produce the same effect, which suggests something specific about social anxiety, perhaps the way it manifests in interpersonal signals, makes it especially transmittable through the social referencing channel.
The brain imaging side of this story fills in some of the mechanism. Adults with social anxiety disorder show heightened amygdala responses when learning from another person’s fearful facial expressions. In a study comparing people with social anxiety to healthy controls, those with the disorder had significantly stronger amygdala activation when processing stimuli that had been paired with fearful faces and showed a trend toward reduced responses to stimuli paired with happy or neutral faces. The amygdala, in effect, was turning up the volume on threat signals received from other people and turning it down on reassurance. That neural bias could explain why socially anxious individuals are both more attentive to others’ fear signals and more likely to transmit anxiety through social referencing.
Social Referencing and Autism
Differences in social referencing are among the earliest detectable signs that a child may be on the autism spectrum. A systematic review of the literature found that studies consistently identified deficits in social referencing among individuals with autism, with seven studies specifically reporting either a failure to engage in spontaneous looking toward others or a failure to adjust behavior based on the emotional signals received in ambiguous situations.
Research on infant siblings of children already diagnosed with autism has been especially revealing. Compared to both typically developing infants and high-risk siblings who did not go on to develop autism, infants who were later diagnosed engaged in slower information-seeking looks, meaning they took longer to look toward the adult for emotional guidance when faced with uncertainty. This slowness in seeking out social information may be one of the earliest behavioral markers of the condition.
Interestingly, the high-risk siblings who did not go on to develop autism themselves also showed some differences. They had difficulty regulating their behavior based on the adults’ emotional signals, even though their information-seeking was intact. This pattern, where the looking is fine but the behavioral adjustment is impaired, was present in high-risk infants regardless of whether they were later diagnosed. Researchers have proposed that this may reflect an endophenotype for autism, a genetically linked trait that is more common in relatives of affected individuals even when those relatives do not meet diagnostic criteria.
One study using eye-tracking during a videogame task found that children with autism did not differ significantly from typically developing children in the percentage of time they spent looking at various areas of interest, with one possible exception. That finding complicates the picture somewhat and suggests that the social referencing deficits in autism may be more about the quality or timing of looking rather than the raw quantity of visual attention directed at other people.
Dogs Read Their Owners Too
Social referencing is not unique to humans. Dogs provide the most-studied example in another species. When confronted with an unfamiliar or mildly threatening object, most dogs look back at a nearby person, and their subsequent behavior changes depending on what emotional signal they receive. In one experiment, dogs that received a positive emotional message from their owner spent more time approaching and staying near the object, while dogs given a negative message took longer to approach and interact with it.
Whether dogs are truly engaging in social referencing or simply learning through observational conditioning is debated. A separate study found that while most dogs, about 83%, did look referentially toward the owner after encountering a strange object, the differences between dogs in positive-signal and negative-signal groups were modest when it came to behavioral regulation. However, a strong observational conditioning effect was present: dogs in the positive group moved closer to the object and dogs in the negative group moved away, mirroring their owner’s behavior. The distinction matters because social referencing implies the dog is extracting emotional information and applying it to the situation, while observational conditioning implies the dog is simply copying what the person did.
The signals from the owner specifically seem to matter more than those from a stranger. When the owner acted as the informant, dogs that received positive messages looked at the owner more frequently and spent more time near the object. When a stranger delivered the same positive or negative signals, dogs still looked at them but did not adjust their behavior as clearly. That owner-specificity is consistent with the attachment-based framework seen in human infant research.
Great Apes and the Limits of Comparison
Great apes show a related but distinct ability. In a series of experiments, apes were able to distinguish between a human experimenter’s happy and disgusted reactions to hidden food and chose accordingly. More impressively, in a follow-up condition where the apes saw the experimenter eat something after reacting positively to one container and negatively to another, they correctly chose the container the experimenter had reacted negatively to, inferring that the preferred food had already been eaten. This suggests apes understand both the directedness of emotional expressions, meaning they know what the expression is about, and their valence, meaning they know whether it is positive or negative.
Whether this constitutes social referencing in the same sense as an infant checking a parent’s face is an open question. Apes in these studies were using emotional information instrumentally to get food, which is somewhat different from an infant using a parent’s emotional reaction to calibrate their own feelings about a new situation. The ape studies show that the cognitive raw materials for social referencing, reading emotion direction and valence from another’s face, are present in our closest relatives, even if the full social referencing package may be distinctively human or at least distinctively developed in humans.
When Peers Replace Parents
Most social referencing research focuses on the infant-caregiver relationship, but children also attempt to read emotional cues from other children. The results are messier. In a study where 12-, 18-, and 24-month-olds watched a peer express positive, negative, or neutral emotions toward toys, the responses were age-dependent and somewhat counterintuitive. Twelve-month-olds decreased their play with toys toward which a peer had expressed any emotion, whether positive or negative, compared to toys the peer had simply looked at neutrally. Eighteen-month-olds showed no systematic response. And 24-month-olds increased their play after watching a peer display negative emotion toward a toy.
The 24-month-old finding, where children approached a toy more after seeing another child react negatively to it, is puzzling at first but may reflect a developing sense of autonomy or competitive interest. If another child does not want something, that might make it more available or interesting. It also suggests that peer social referencing operates under different rules than caregiver social referencing, where negative signals reliably inhibit approach behavior. The social dynamics between equals are not the same as those between an infant and a trusted authority figure.
Cultural Variation in Emotional Display
The emotional signals that feed social referencing are not culturally uniform. A study comparing infants from urban Germany and rural Ecuador found that three-month-old German infants showed higher intensities of positive affect in response to intense positive stimulation than Kichwa infants from Ecuador. The researchers linked these differences to cultural display rules, where the German cultural milieu is associated with a preference for high levels of positive emotional expression. If caregivers in one culture express emotions more intensely or differently than caregivers in another, the raw material that infants use for social referencing will differ accordingly.
This has implications that researchers are still working through. Social referencing studies have been conducted overwhelmingly in Western, industrialized populations. Whether the developmental timeline, the relative importance of voice versus face, the links to attachment, and the role in anxiety transmission all hold across widely different cultural contexts is an area where the evidence is thin. Given that caregiving practices, emotional expressiveness norms, and even the physical proximity between infants and caregivers vary dramatically across cultures, it would be surprising if social referencing looked exactly the same everywhere.
Social Referencing Beyond Infancy
Although the term is most associated with infant development, the underlying process persists into adulthood in forms that are easy to overlook because they are so automatic. You glance at a friend’s face when the waiter sets down an unfamiliar dish. You read the room before deciding whether to laugh at an off-color joke. You watch how your colleagues react to a new policy before forming your own opinion. These are all adult versions of the same basic process: using other people’s emotional reactions as data for navigating uncertain situations.
The amygdala research on social anxiety disorder described earlier is technically a study of adults, and it demonstrates that the neural machinery for social referencing does not switch off after infancy. It continues to operate, and individual differences in how that machinery is calibrated, whether from genetics, early experience, or clinical conditions, continue to matter for how people process and respond to emotional signals from those around them. The infant who showed slower information-seeking in the lab may grow into the adult whose amygdala over-responds to fearful faces, though tracing those developmental paths in full remains one of the field’s major unfinished projects.