Breast Feeding

Abstract

Breast feeding is a form of feeding behavior: the nourishing of an infant with milk directly from the breast, which cognitive psychology studies at three levels. This article develops that account—how the suckling infant drives a neuroendocrine loop in which oxytocin ejects milk and promotes maternal calm and bonding, how breastfeeding is associated with small gains in later intelligence and the hard problem of separating that effect from confounding, and how a mother’s breastfeeding self-efficacy—her belief that she can succeed—predicts whether she starts and continues. Breast feeding is therefore a behavior governed by hormones, by learning, and by emotion, and a natural experiment in how early nutrition bears on cognitive development. Three demonstrations make its confounding, the suckling–hormone loop, and the self-efficacy–duration relationship concrete.

Keywords: breast feeding, cognitive development, oxytocin, breastfeeding self-efficacy

Breast feeding is the heading the National Library of Medicine uses for feeding an infant at the breast, the biological default against which every other mode of infant feeding is compared. It looks like a purely nutritional act, but it sits at the intersection of three psychological literatures: a neuroendocrine reflex that links milk release to maternal emotional state, a decades-long argument about whether breastfed children end up cognitively advantaged and why, and a body of health-psychology work on the beliefs and supports that determine whether a mother breastfeeds at all. Those three threads—hormone, cognition, and decision—are what make breast feeding a problem for cognitive psychology rather than only for pediatrics, and they are the structure this article follows.

What Breast Feeding Is

Within the Medical Subject Headings, Breast Feeding denotes feeding an infant with milk directly from the mother’s breast, as distinct from bottle feeding expressed milk or formula. The major health bodies recommend it as the normative mode of infant nutrition: the American Academy of Pediatrics advises exclusive breastfeeding for about the first six months, with continued breastfeeding alongside complementary foods thereafter (Meek & Noble, 2022). That recommendation rests on a large epidemiological literature linking breastfeeding to lower infant mortality and morbidity and to a range of longer-term outcomes, synthesized in the 2016 Lancet series on breastfeeding in the twenty-first century (Victora et al., 2016).

For cognitive psychology the interest is narrower and sharper. Breast feeding is a behavior whose physiology is governed by a psychologically sensitive reflex, whose association with later cognition has driven one of the most instructive confounding debates in developmental science, and whose uptake depends on a mother’s beliefs and circumstances. The sections that follow take those three in turn, after first locating breast feeding within the MeSH tree of feeding behaviors.

Types of Breast Feeding

In the Medical Subject Headings, Breast Feeding sits beneath the parent heading feeding behavior and has one narrower descriptor of its own. The table below lists that child heading; MeSH is an indexing vocabulary, not a theory of lactation, so the classification groups headings for literature retrieval rather than asserting a single developmental taxonomy, and the clinical categories a mother meets—exclusive, predominant, and partial breastfeeding—are orthogonal distinctions of degree rather than separate MeSH descriptors.

Table 1

Narrower MeSH descriptor under Breast Feeding.

Cutting across that formal tree is the clinical gradient of exclusivity. Exclusive breastfeeding—the infant receiving only breast milk and no other food or drink—is the category most health recommendations target, and the question of how long it should optimally continue was settled for policy by a Cochrane review favoring about six months (Kramer & Kakuma, 2012). Predominant and partial breastfeeding describe lesser degrees, and the dose–response relationships between exclusivity, duration, and later outcomes are exactly what the cognitive-development literature below tries to exploit.

Breast Feeding and Child Cognitive Development

The claim that breastfed children are cognitively advantaged is old, widely believed, and genuinely hard to evaluate. Observational studies repeatedly find that children who were breastfed score a few points higher on later intelligence tests, and a classic meta-analysis put the advantage at roughly three IQ points after adjustment for covariates—but it also reported a substantially larger unadjusted gap, which is the heart of the problem (Anderson, Johnstone, & Remley, 1999). Mothers who breastfeed differ systematically from those who do not: on average they have higher intelligence, more education, and more cognitively stimulating homes, and each of those factors independently raises a child’s test scores. The raw correlation therefore mixes any causal effect of breast milk with the effects of the kind of mother who breastfeeds.

Two kinds of evidence try to break that confound. The first is a cluster-randomized trial—the PROBIT study in Belarus—which randomized maternity hospitals to a breastfeeding-promotion intervention and so manipulated breastfeeding rates experimentally; its children showed a meaningful advantage on verbal IQ at age six, the strongest causal evidence available because randomization balances the confounders (Kramer et al., 2008). The second is careful meta-analysis and cohort work: a systematic review pooling many studies again found a small advantage of a few IQ points that persisted after adjustment (Horta, Loret de Mola, & Victora, 2015), and a prospective cohort found that longer breastfeeding duration predicted higher receptive language and verbal IQ in a dose-dependent way (Belfort et al., 2013).

The most recent and most pointed work presses the confounding question directly. Analyzing the UK Millennium Cohort, one study showed that the apparent advantage of longer breastfeeding on cognitive scores up to age fourteen shrank sharply—though it did not entirely vanish—once maternal cognitive ability and socioeconomic position were fully adjusted for, implying that most, but perhaps not all, of the observed benefit is confounding (Pereyra-Elías, Quigley, & Carson, 2022). The demonstration below varies the degree of confounding adjustment, tracing how the breastfed–versus–non-breastfed gap attenuates from its raw size toward the small residual the best-controlled studies report.

The breastfed–non-breastfed IQ gap under confounding adjustment

The raw advantage is about 5.3 IQ points. Drag the slider to apply statistical adjustment for maternal intelligence and home environment; full adjustment removes about 41% of the gap, leaving the small residual the best-controlled studies report.

The Suckling–Hormone Loop and Mother–Infant Bonding

Beneath the developmental statistics is a neuroendocrine loop that makes breast feeding a two-way psychological exchange rather than a one-way delivery of nutrients. When the infant suckles, afferent signals from the nipple reach the hypothalamus and trigger the posterior pituitary to release two hormones: oxytocin, which contracts the cells around the milk glands and ejects milk into the ducts, and prolactin, which sustains the ongoing synthesis of milk for the next feed (McNeilly, Robinson, Houston, & Howie, 1983). The milk-ejection, or let-down, reflex that oxytocin drives was shown by Michael and Niles Newton in the 1940s to be psychologically sensitive—distraction and distress could inhibit it, while the sight or sound of the infant could restore it (Newton & Newton, 1948). The physiology of this secretory machinery, from the differentiation of the milk-producing cells to their hormonal control, is now well characterized (Pang & Hartmann, 2007). Figure 1 traces the positive-feedback loop that results: suckling releases the two hormones, milk is delivered, and the satisfied infant continues to suckle, sustaining the cycle until the feed ends.

What makes the loop psychological rather than merely mechanical is oxytocin’s second role. Beyond ejecting milk, oxytocin exerts broad anti-stress and calming effects—lowering blood pressure and cortisol and promoting affiliative, approach-oriented behavior—so that the same hormone that empties the breast also quiets the mother and supports her attachment to the infant (Uvnäs-Moberg, 1998). A systematic review of maternal plasma oxytocin during breastfeeding confirmed the pulsatile release pattern across studies and mapped how it tracks the timing of a feed, giving the behavioral claims a measurable physiological anchor (Uvnäs-Moberg et al., 2020). The loop is vulnerable at its emotional end: maternal stress around the time of birth is associated with a delayed onset of copious milk production—impaired lactogenesis—implicating the same oxytocin-sensitive pathway in establishing lactation and not only in running it (Dewey, 2001). The demonstration below lets suckling intensity vary and traces its effect on the oxytocin and prolactin responses and the calm-and-bonding state they produce.

Suckling drives oxytocin, prolactin, and maternal calm

Afferent signals from the suckling infant reach the hypothalamus and release two hormones: oxytocin ejects milk and calms the mother, while prolactin sustains synthesis for the next feed. Vary suckling intensity and watch each response, and the calm-and-bonding state oxytocin produces, track it.

Breastfeeding Self-Efficacy and the Decision to Continue

Whether a mother breastfeeds at all, and for how long, is less a matter of physiology than of psychology and circumstance—and the single construct that organizes much of this literature is breastfeeding self-efficacy, a mother’s confidence in her ability to breastfeed. Adapting Bandura’s general theory of self-efficacy to lactation, the Breastfeeding Self-Efficacy Scale operationalizes that confidence as a measurable score, and it predicts the outcomes that matter: mothers with higher self-efficacy are more likely to initiate breastfeeding, to breastfeed exclusively, and to persist through the difficult early weeks when most discontinuation occurs (Dennis, 2003). Self-efficacy is not merely a correlate; it is modifiable, which is what makes it a clinical lever. Because confidence is built from mastery experiences, vicarious learning, verbal persuasion, and physiological feedback, interventions that arrange early successful feeds, model technique, and reframe normal difficulties can raise self-efficacy and, with it, breastfeeding duration.

This reframes breastfeeding continuation as a problem of belief and learning rather than of will or anatomy. A mother who interprets an ordinary early setback—a painful latch, an unsettled night—as evidence that she cannot produce enough milk is at high risk of stopping, whereas the same event read as a surmountable, expected hurdle predicts persistence. The demonstration below maps a self-efficacy score onto the modeled probability that a mother is still breastfeeding exclusively at successive postpartum milestones, making the steepness of that relationship visible.

Self-efficacy and the probability of continued breastfeeding

A mother’s score on the Breastfeeding Self-Efficacy Scale (range 14–70) maps onto the modeled probability that she is still breastfeeding exclusively through the difficult early weeks. The relationship is steep in the middle: small gains in confidence near the center shift continuation the most.

Worked Example

Consider how confounding inflates the apparent cognitive benefit of breast feeding, using the logic behind the first demonstration. Suppose an observational study finds that breastfed children score, on average, 5.3 IQ points higher than non-breastfed children before any statistical adjustment—close to the unadjusted figure reported in the classic meta-analysis (Anderson et al., 1999). This raw gap is not the causal effect of breast milk; it is the causal effect plus the contribution of every maternal characteristic that both raises the chance of breastfeeding and independently raises the child’s score.

Now adjust. Suppose full adjustment for maternal intelligence and home environment removes about 41% of the raw gap, the attenuation needed to carry the unadjusted 5.3 points down to the adjusted 3.2 the same meta-analysis reported. With no adjustment the estimated advantage is the full 5.3 points; adjusting halfway (removing about 20% of the gap) leaves 4.2 points; adjusting fully leaves 3.2 points. If a more complete set of confounders—as in the Millennium Cohort analysis—removes still more, the residual shrinks toward one or two points and, in some outcomes, loses significance (Pereyra-Elías et al., 2022). The arithmetic makes the interpretive point precise: a plausible-sounding claim that breastfeeding adds five IQ points can be two-fifths an artifact of who breastfeeds, and the honest estimate of the causal effect is small—real enough to survive a randomized trial, but far smaller than the raw correlation advertises.

Discussion

Breast feeding is a clarifying case for cognitive psychology because it braids together three phenomena that are usually studied apart: a hormonally mediated reflex sensitive to emotion, a developmental association shot through with confounding, and a health behavior driven by self-efficacy. The hormone end shows learning and emotion commanding an autonomic output; the development end is a master class in why correlation is not causation; and the decision end shows a belief construct predicting and changing real behavior. Each is ordinary cognition operating on an unusually consequential act.

Table 2

Three psychological levels of breast feeding and what each explains.

Note. Each level is drawn from a section above; together they portray breast feeding as a behavior governed by hormones, confounded in its developmental signature, and driven by belief. The randomized evidence confirms a small but real cognitive effect survives the confounding (Kramer et al., 2008).

The account has open tensions. The causal size of the cognitive effect remains contested, with careful cohort analyses pushing the residual toward the small end; whether oxytocin’s calming effects are a cause or a correlate of successful breastfeeding is not fully resolved; and self-efficacy, though predictive, is itself partly a product of the circumstances—support, pain, supply—that it is invoked to explain.

Current Directions

Contemporary work has moved from asserting that breastfeeding benefits cognition to estimating how much of that benefit is real once confounding is taken seriously. The sharpest recent analyses use large national birth cohorts and rich covariate sets to show that the association with later cognitive scores attenuates substantially—though not always to zero—under full adjustment for maternal cognitive ability and social position, reframing the open question as the size, not the existence, of a causal effect (Pereyra-Elías et al., 2022). This is the methodological frontier: the field now treats the raw correlation as a starting point to be explained away, and asks what residual survives the most complete adjustment and the few randomized designs available.

On the physiological side, the search is for a measurable marker of the behavioral claims about bonding and calm, and the systematic-review mapping of plasma oxytocin across the course of a feed is a step toward that anchor (Uvnäs-Moberg et al., 2020). On the public-health side, policy continues to recommend exclusive breastfeeding while acknowledging the structural supports—parental leave, workplace accommodation, clinical counseling—that determine whether recommendations translate into behavior (Meek & Noble, 2022). The common thread is a move from demonstrating associations to quantifying mechanisms and removing the barriers between intention and act.

Cognitive Implications

Read as cognition, breast feeding illustrates three themes that reach beyond infant nutrition. First, it is a vivid reminder that an observed association between an early exposure and a cognitive outcome can be mostly confounding: the breastfeeding–IQ literature is a standing lesson in why developmental psychology leans so hard on randomized and quasi-experimental designs, and why a few controlled IQ points are more credible than a larger raw correlation (Anderson et al., 1999; Kramer et al., 2008; Horta et al., 2015). Second, it shows a hormonal system doubling as an emotional one: oxytocin’s shared role in milk ejection and in calm and bonding is a clean example of a single mechanism linking a bodily function to affect and social behavior (Uvnäs-Moberg, 1998). Third, it demonstrates the predictive and practical power of a self-efficacy account of health behavior: a measurable belief forecasts who will persist and offers a modifiable target, so that the psychology of confidence, not only the physiology of milk, determines how long breastfeeding lasts (Dennis, 2003). In each, a homely act turns out to be a window onto how emotion, confounding, and belief shape behavior and its measurement.

Common Misconceptions

Breastfeeding has been proven to make children substantially smarter.

Much of the raw IQ advantage reflects confounding by maternal intelligence and home environment; the best-controlled studies find a small residual effect of a few points, not the larger gap that unadjusted comparisons suggest (Anderson et al., 1999; Pereyra-Elías et al., 2022).

Breast feeding is purely nutrition—the mother’s emotional state is irrelevant.

Milk release depends on the oxytocin-driven let-down reflex, which stress can inhibit, and oxytocin in turn calms the mother and supports bonding, so feeding and emotion are coupled (Newton & Newton, 1948; Uvnäs-Moberg, 1998).

Whether a mother keeps breastfeeding is mostly about her milk supply.

A mother’s breastfeeding self-efficacy—her confidence that she can succeed—is among the strongest predictors of initiation and duration, and it is modifiable through support and early mastery (Dennis, 2003).

Glossary

Breast feeding.

Feeding an infant with milk directly from the mother’s breast, the biological default mode of infant nutrition and the parent category for breast milk expression.

Breastfeeding self-efficacy.

A mother’s confidence in her ability to breastfeed, measured by the Breastfeeding Self-Efficacy Scale; a modifiable belief that predicts whether she initiates and how long she continues.

Cluster-randomized trial.

An experiment that randomizes intact groups—here maternity hospitals, as in PROBIT—rather than individuals, used when the intervention (breastfeeding promotion) must be delivered at the group level.

Cohort study.

An observational design that follows a group over time, relating an earlier exposure such as breastfeeding duration to a later outcome such as verbal IQ, with statistical adjustment for measured confounders.

Confounding.

The distortion of an association when a third factor—here maternal intelligence and home environment—independently influences both the exposure and the outcome, inflating the apparent effect of breastfeeding on cognition.

Dose–response relationship.

A pattern in which more of an exposure predicts more of an outcome; longer breastfeeding duration predicting higher cognitive scores is cited as evidence that the association is not purely an artifact.

Exclusive breastfeeding.

Feeding the infant only breast milk, with no other food or drink; the category most health recommendations target, generally advised for about the first six months.

Lactogenesis.

The onset and establishment of copious milk production after birth; its timing is sensitive to maternal stress, implicating oxytocin-related pathways in starting lactation as well as maintaining it.

Let-down reflex.

The milk-ejection reflex; the release of milk into the ducts triggered by oxytocin in response to suckling or a conditioned infant cue, and inhibited by acute stress.

Meta-analysis.

A statistical synthesis pooling the results of many studies into a single estimate; the breastfeeding–intelligence literature is summarized by meta-analyses reporting a small adjusted advantage.

Oxytocin.

The posterior-pituitary hormone released in pulses during suckling that drives milk ejection and also has anti-stress, calming effects that support mother–infant bonding.

PROBIT trial.

The Promotion of Breastfeeding Intervention Trial, a cluster-randomized study in Belarus that manipulated breastfeeding rates and provided the strongest causal evidence of a cognitive effect.

Prolactin.

The pituitary hormone, also released in response to suckling, that drives the ongoing synthesis of milk, as distinct from oxytocin’s role in ejecting already-made milk.

Suckling.

The infant’s rhythmic drawing at the breast, whose afferent signals to the hypothalamus trigger the release of oxytocin and prolactin—the input that drives the feeding loop.

Key Researchers

Cindy-Lee Dennis

. Professor and Canada Research Chair in Perinatal Community Health at the University of Toronto; she developed the Breastfeeding Self-Efficacy Scale, which operationalized maternal confidence as a measurable, modifiable predictor of breastfeeding initiation and duration. Wikipedia · Wikidata · Faculty page · Google Scholar

Bernardo Lessa Horta

. Professor of epidemiology at the Federal University of Pelotas, Brazil; he led the systematic review and the Pelotas cohort analysis quantifying the association between breastfeeding and later intelligence, central to the modern estimate of its size. Wikidata · Faculty page · Google Scholar

Michael S. Kramer

. Emeritus professor of pediatrics and of epidemiology at McGill University; he designed and led the PROBIT cluster-randomized trial, the single strongest piece of causal evidence that breastfeeding affects child cognitive development. Faculty page · Google Scholar

Niles Newton

. Psychologist at Northwestern University Medical School (1923–1993); with Michael Newton she first demonstrated experimentally that the human let-down reflex is a conditionable, emotion-sensitive response that distraction and stress can inhibit, founding the psychological study of lactation. Biography

Kerstin Uvnäs-Moberg

. Professor emerita of physiology at the Swedish University of Agricultural Sciences; her research established oxytocin’s anti-stress and calming effects and their role in breastfeeding, linking the hormone of milk ejection to maternal emotional state and bonding. Wikipedia · Wikidata

Cesar G. Victora

. Emeritus professor of epidemiology at the Federal University of Pelotas, Brazil; he led the 2016 Lancet breastfeeding series and the Pelotas birth-cohort work on the lifelong effects of breastfeeding, including its relationship to adult intelligence and attainment. ORCID · Wikipedia · Google Scholar

Frequently Asked Questions

What is breast feeding?

It is feeding an infant with milk directly from the mother’s breast—the sense in which the National Library of Medicine uses the term, and the biological default against which bottle feeding and formula are compared. Major health bodies recommend exclusive breastfeeding for about the first six months (Meek & Noble, 2022).

Does breastfeeding actually make children smarter?

Breastfed children do score slightly higher on later intelligence tests, but much of that raw gap reflects the fact that mothers who breastfeed tend to have higher intelligence and more stimulating homes. After adjustment and in randomized evidence, a small advantage of a few IQ points remains (Anderson et al., 1999; Kramer et al., 2008).

Why is the cognitive effect so hard to measure?

Because breastfeeding is not randomly assigned: the same maternal characteristics that make breastfeeding more likely also independently raise children’s scores, so the correlation confounds cause with selection. Cohort studies that adjust for these factors see the effect shrink substantially (Pereyra-Elías et al., 2022).

What does oxytocin do during breastfeeding?

Released in pulses when the infant suckles, oxytocin ejects milk by contracting the cells around the milk glands, and it also lowers stress and promotes calm and bonding—so the hormone of milk release is also a hormone of maternal affiliation (McNeilly et al., 1983; Uvnäs-Moberg et al., 2020).

Can stress affect breastfeeding?

Yes. Acute stress can inhibit the let-down reflex by opposing oxytocin, and perinatal stress is associated with a delayed onset of full milk production, so a mother’s emotional state bears directly on lactation (Newton & Newton, 1948; Dewey, 2001).

What is breastfeeding self-efficacy and why does it matter?

It is a mother’s confidence in her ability to breastfeed, measured by the Breastfeeding Self-Efficacy Scale. It is one of the strongest modifiable predictors of whether she initiates breastfeeding and how long she continues, which makes it a practical target for support (Dennis, 2003).

How long should a mother breastfeed exclusively?

A Cochrane review comparing exclusive breastfeeding for about six months with shorter exclusivity supported six months as the policy default, and health bodies recommend continued breastfeeding alongside complementary foods thereafter (Kramer & Kakuma, 2012; Meek & Noble, 2022).

Does how long a child is breastfed matter for cognition?

Prospective cohort work finds a dose–response pattern in which longer breastfeeding duration predicts higher receptive language and verbal intelligence, consistent with the pooled meta-analytic estimate of a small advantage (Belfort et al., 2013; Horta et al., 2015).

References