Olfaction scaffolds the developing human from neonate to adolescent and beyond

Abstract

The impact of the olfactory sense is regularly apparent across development. The fetus is bathed in amniotic fluid (AF) that conveys the mother's chemical ecology. Transnatal olfactory continuity between the odours of AF and milk assists in the transition to nursing. At the same time, odours emanating from the mammary areas provoke appetitive responses in newborns. Odours experienced from the mother's diet during breastfeeding, and from practices such as pre-mastication, may assist in the dietary transition at weaning. In parallel, infants are attracted to and recognize their mother's odours; later, children are able to recognize other kin and peers based on their odours. Familiar odours, such as those of the mother, regulate the child's emotions, and scaffold perception and learning through non-olfactory senses. During juvenility and adolescence, individuals become more sensitive to some bodily odours, while the timing of adolescence itself has been speculated to draw from the chemical ecology of the family unit. Odours learnt early in life and within the family niche continue to influence preferences as mate choice becomes relevant. Olfaction thus appears significant in turning on, sustaining and, in cases when mother odour is altered, disturbing adaptive reciprocity between offspring and carer during the multiple transitions of development between birth and adolescence.

This article is part of the Theo Murphy meeting issue ‘Olfactory communication in humans’.

1. Introduction

Like other mammalian offspring, human infants thrive through a predictable sequence of developmental transitions: embryonic and fetal growth; birth, breastfeeding and attachment; diversification in sociality and sustenance (weaning); motor autonomy and wariness of novelty; puberty and risk-taking; and dispersal, social enculturation through affiliation networks, and initiation of mate choice. Each transition comes with its particular timing, tensions and threats to offspring viability (e.g. [1–3]), requiring physiological, perceptual–cognitive and behavioural co-adaptations in the dependent infant and the investing parents.

This paper aims to review the adaptive contribution of olfaction in alleviating the challenges raised by these developmental transitions. It will describe how the fetal environment primes the growing offspring to their forthcoming environment, where neonates will need to discern the mother promptly in order to ingest colostrum/milk and reach physiological stability. Next, it will survey the nursing niche, where infant responsiveness to odours can assist in self-regulation and managing the uncertainties of emerging social and dietary novelty. Finally, parent-to-child olfactory communication will be considered in the context of expanding affiliative networks within the family and beyond. In all this, we aim to provide an overview of empirical research on parent-to-infant odour exchanges, identify gaps in current understanding and suggest new directions for future research.

2. From the prenatal to the postnatal niche: the birth transition

(a). Interacting physiologies: materno-fetal odorant transfers and transnatal olfactory continuity

An infant's olfactory preferences have their origins in the prenatal period. Nasal chemoreception begins functioning during the last gestational trimester [4], bathed in an amniotic pool that is permeated by odorous compounds that are regulated by the mother's genetic, immune and physiological constitution, modulated by her stress and health, and paced by her dietary, cosmetic or addictive inclinations. Odorous metabolites pass easily into amniotic fluid (AF), and such transplacental penetration can be so stark that the newborn's body odour is occasionally pungent [5,6]. Neonates favour the odour signature of AF [7], particularly their own AF [8], and react to odours experienced during gestation (e.g. anise [9]; alcohol [10]). Regular gestational exposure to strongly odorized foodstuffs (garlic, carrots, fish, cheese, green vegetables) influences the progeny's preferences for related odorants over periods that can last several months or even years [11–13].

The amniotic environment provides the fetus with an olfactory repertoire that prepares it for the outside world by virtue of transnatal olfactory continuity (TOC). Thus, amniotic and milk odours are equivalently attractive to neonates up to postnatal day 3, at which point, conspecific milk odour becomes more appealing [14]. Such initially undifferentiated responses between AF and colostrum have been found in the newborns of other species, suggesting a pan-mammalian convergence in TOC from both compositional and perceptual points of view [6]. Many of the maternal dietary odorants that infiltrated AF will permeate colostrum/milk, as will remnants of the mother's chemical ecology (tobacco smoke, cosmetics), and odour-active compounds deriving from her normal metabolism and lactogenic process, such as steroids or fatty acids (reviewed in [15,16]). Nurslings are sensitive to these odour changes in milk [17–21].

TOC represents a maternal sensory information transfer system that impinges both on the stimuli and on the receptive system of the offspring. Indeed, maternal physiology not only conveys odorant metabolites to fetuses, but also attunes the conceptus' olfaction to detect a range of odorants that will occur on her body surface and in her milk. In animal models, exposure to an odorant in utero induces epigenetic changes in olfactory receptor expression and orients neurogenesis and synaptic organization in the olfactory bulbs, eventually tuning olfactory sensitivity in newborns (e.g. [22,23]). Prenatal olfactory experience is then reinforced through postnatal reconsolidation, thereby facilitating newborn responsiveness to that odorant [24,25]. Thus, under normal mammalian circumstances, the mother, via odorants transferred in AF, designs the offspring sensors (fetal chemoreception), the medium (odour cues) and the message (familiarity between the prenatal and postnatal niches) (reviewed in [15,26]).

The fitness value of the TOC is best demonstrated with reference to the consequences of its disruption. Several non-human examples show that drastic odour mismatches between the prenatal and birth environments result in altered nipple grasping [24,25], increased stress levels [27] and even lethality [4]. In humans, one such perinatal odour mismatch is created by feeding neonates non-human milks or artificially engineered formulae. When infants are breastfed from birth, they show a preference for milk over AF in a paired-choice odour test run on day 4. By contrast, when exclusively fed cow-based formulae, same-age infants turn more to AF odour than to the reinforcing formula odour [14,28], indicating a differential path of preference development as a function of a progressive change based on TOC versus a saltational change of it. Another such mismatch is created when the AF odour is washed away right after birth. AF odour elicits a positive orientation response in newborns [7,29,30], and infants’ own spreading of AF on the breast facilitates their motivated responses. When the neonate's AF covering is left intact for at least 12 h after birth, infants evince better weight gain, revealing more optimal feeding responses [31,32]. Along the same lines, recreating the prenatal odour environment postnatally facilitates the neonates' adaptive responses. For example, providing AF odour to term or preterm newborns eases self-regulatory responses in reducing fussing and crying (e.g. [29,33–35]). In summary, prenatal odours not only guide the first directional actions, but also promote neonates’ energy allocation to anabolism and growth as opposed to catabolic wasting during a period of great metabolic vulnerability (see below).

(b). Birth and the rapid learning of the mother's body surface odours

The normal birth process represents both the last AF odour encoding and an upsurge of novel perceptual experience for the fetus as it becomes a newborn. The physiological/sensory upheavals of labour affect the brain, with rising catecholamines coinciding with high arousal levels [36]. Not only does labour set a last sensory update of the amniotic ‘smellscape’ [37], but it promotes neonatal learning of odours, as found in the rat [25], and inferred in human neonates. When exposed to an odorant for 30 min after a Caesarean section made before/after labour engagement, and then re-exposed to that odour 1–5 days later, only those neonates who were subjected to contractions preferred the familiar odour compared to a novel odour [38]. Thus, labour-related events mediate high arousal states during the first few postnatal hours when the brain appears, especially receptive to incoming stimulation. Human neonates exposed to an odorant for 30 min during the first postnatal hour go on to display a preference for the familiar odorant 2–3 days later, unlike those exposed later (after 12 h postpartum) [39]. Likewise, 4-day-olds mouth more to their mother's milk odour (than to another mother's milk odour) when they have been in contact with the mother's skin right after birth [40]. Thus, the birth process itself creates a neurosensory context that engages fast learning of odours associated with the mother's body.

Aversive perinatal odorants might also contribute to newborn performance. Odorous steroids (e.g. androstenone) or conjugates of acidic or thiol compounds occur in AF, milk and axillary sweat, where they convey salient odour notes, and are aversive to newborns (perhaps inherently so) when administered in pure form [41,42]. In AF/milk, the aversive value of these odorants may be attenuated when combined with positively valenced compounds, and indeed, such a blend of positive and negative constituents leads to an attention-capturing contrast effect [43], perhaps optimizing the learning of the odour qualities of AF or milk [41]. Axillary odours containing the above compounds elicit crying in 2–4-day-old newborns [44], although these odours become secondarily acceptable after pairing with maternal care: 2-week-olds orient to maternal axillary odours [45].

(c). Odour communication during nursing

Human nipples constitute an evolved multisensory trap that concentrates conspicuous tactile, gustatory, olfactory and visual cues, and function as vital interfaces between lactating females and neonates [46,47]. Darwin foresaw that a natural scent might drive the newborn to the nipple [48], but corroborating evidence awaited another century. The test consisted of presenting odorous cotton pads hanging over each side of the face of supine infants. When so exposed to one pad impregnated with the mother's breast odour against a clean control pad, 17 of 20 breast-fed infants (aged 2–7 days) turned their nose longer to the former stimulus, indicating attraction [49]. A later study verified the specificity of mammary odour for neonates: 2-week-old infants bottle-fed from birth turned longer to an unfamiliar lactating mother's breast odour than to the odour of their familiar formula [50]. Similarly, 2-week-old bottle-fed infants facing the breast odour of an unfamiliar lactating woman, against either (i) the breast odour of a non-parturient woman, or (ii) the axillary odour of that same woman, oriented more to the odour of the lactating breast [51]. As these (formula-fed) infants had never engaged with their mother's breast for feeding, this is consistent with the evidence that women emit a more attractive odour from the (lactating) mammary area than from the axilla. Finally, when laid prone on the mother's torso within an hour of birth, newborns crawl to the breast [52,53], with breast odour possibly driving directional actions [54]. Likewise, when left prone on a mattress, infants are swifter to approach a pad scented with their mother's breast odour than a scentless pad [55]; and when presented with mother's breast odour under the nose, infants display more rooting responses (than to control stimuli), and produce more efficient arm- and footsteps [56]. Finally, in different contexts, breast/milk odour can provide comfort, actualized in infants' reduced motor output [57], delayed onset of crying [58] and attenuated expressions of stress and pain [59,60].

The source of the breast's attractive and reinforcing odorants is unclear. The human areolar-nipple structure harbours skin glands of all types (eccrine, apocrine and sebaceous). Human nipples bear sebaceous glands at their distal end, that open into milk ducts as well as onto the nipple tip surface [61]. The areolae are dotted with Montgomery's glands [62,63], which are coalesced sebaceous and milk glands [64] that give off a whitish fluid during lactation [47,65]. When 3-day-old infants were exposed to their mother's entire breast, or isolated areola, or isolated nipple, or drops of milk [58], they responded alike to the odour of these different conditions, suggesting overlapping or equivalent attractive potencies in underlying mammary substrates. However, the odour of Montgomerian secretions, when presented separately, elicited a typical respiration pattern and more mouthing responses than milk, sebum and various controls [66]. Montgomerian odour may thus play a special role in the human infant's attraction to, and coordinated action upon, the lactating breast [47].

The most obvious contributor to breast odour is colostrum/milk. Infants born at term [67,68] or preterm [69] react to colostrum/milk odour by positive head-turning and appetitive facial–oral responses [28,70], even before they have been directly exposed to the breast. The odour of the mother's milk, compared to a familiar formula feed, increases the efficacy of nutritive sucking during a regular formula feed [71], and affects the pattern of non-nutritive sucking in premature infants [69,72]. Colostrum, milk or lactating breast odours further elicit cortical activation (as assessed by electroencephalography (EEG) or near-infrared spectroscopy [73–75]), and milk versus formula odours give rise to different patterns of cortical activation in infants' orbitofrontal regions, irrespective of their prior experience with formula [76]. Thus, human lacteal secretions are olfactorily detectable to infants aged from 2 months pre-birth to at least 2 months post-birth, and they affect infants’ arousal, attraction and appetitive responses.

The chemical nature of behaviourally active human milk odorant(s) remains unknown. Chemo-analytical attempts report various odorants in human milk (e.g. [77,78]), but their methodological diversity leaves us short of a comprehensive view [79]. The extraction, separation and identification of milk volatiles is challenging because of their low concentration and instability, yet behaviourally active milk compounds can be characterized, as shown by work in the European rabbit. A single component of fresh rabbit milk, 2-methyl-but-2-enal (2MB2), was as effective as whole rabbit milk odour in eliciting pups' oral grasping. Occurring in milk from varying rabbit genotypes and ecologies, being highly selective in releasing oral responses in other mammalian newborns, and requiring no prenatal/postnatal exposure to become functionally specified, 2MB2 was designated a ‘mammary pheromone’ [80]. However, there is no evidence to date that milk-based predisposed chemosignalling generalizes to other mammalian nursing systems [81–83].

It seems then that the relative constancy of the mammary chemical signature drives infants' continued attraction/appetitive responses to breastfeeding, and this is further evidenced in settings where the breast chemosignature is altered experimentally or physiologically. For example, the mother's diet or physical exercise can modify the odour of milk and transitorily affect the offspring's sucking behaviour [19,84], and nipples with alien odorants are rejected [85,86]. In undiagnosed cases of unilateral malignant tumour of the breast, nurslings have been reported to refuse the affected breast while accepting the healthy one [87]. Thus, infants can disengage their appetitive or consummatory responses following unacceptable fluctuations in their mothers' breast/milk chemosignature.

In summary, mammalian newborn attraction to the maternal body, mammary areas and nipples appears overdetermined. Convergent processes of tactile, visual and in particular odour signalling, all work to optimize the infant's attentional, integrative and motor responses. First, fetal olfactory memory biases neonates to sense the chemicals that post-parturient mothers present on their body. Second, odours in the lactation niche, some aligned with fetal experience, others novel, favour an infant's rapid learning of the idiosyncratic odour signatures of the mother. Any odour sensed at the breast may then be promptly acquired as a signal that reinforces interaction and, as such, elicits positive attraction [88,89]. Third, in addition to such opportunistic odour cues learned in amnio or in lacto, unconditional odour signals conveyed in mammary secretions may operate in humans as they do in other mammals, but evidence is lacking so far [90,91]. Human mammary odorants are indeed effective in eliciting appetitive social responses in infants before direct exposure to the breast or conspecific milk. Such specialized, species-specific signals need now to be chemically characterized and behaviourally assayed in humans [92].

3. Odour-based maternal weaning strategies

A growing infant's energetic/nutritional needs must be balanced against the continuation of the mother's investment in other offspring, among many other duties. Accordingly, exclusive offspring sustenance from human milk must be replaced by the local diet. Across human history, this weaning transition has been, and under harsh conditions continues to be, another period of high infant vulnerability owing to the new wave of challenges and pathogens brought in with non-milk foods [93,94]. One of several important challenges of weaning relates to confronting infants with multisensory ingestive novelty without provoking rejection, and mammalian females rely on multiple, non-exclusive olfactory strategies prior to and during weaning to boost gradual acceptance of non-milk foods. First, as already mentioned, human fetuses are primed to flavours from the pregnant mother's diet, and retain them postnatally for months. Second, fetal familiarization extends as maternal dietary flavours pass into milk. Such early odour experience favours the emergence of human infants' selective responsiveness to foods [12,13,95,96]. That odour cues positively associated with human milk support infant acceptance of novel feeding contexts (bottle: [97]) or of novel foods [11] attests to the strength of these initial maternal olfactory effects. Third, beyond experience of dominant flavour qualities in amnio or in lacto, early and prolonged exposure to chemosensory variety induces weanlings to tolerate ingesting more of a food that is a priori repulsive because it is unusual. A diversified maternal diet renders her milk variable in flavour, thus exposing the suckling to a tonus of ever changing, low-intensity chemosensory fluctuations. The infant's daily exposure to such flavour variety increases later tolerance for flavour novelty, further widening the repertoire of accepted flavours [98,99]. Upon first contact with non-milk foods (e.g. at 5–6 months of age), such chemosensory variety experience may influence acceptance of novel foods at least during childhood (up to 6 years; [100]).

Fourth, another pan-mammalian solution to olfactorily cue safe foods relies on an offspring's attraction to the mother's mouth or breath during ingestion (e.g. [101–103]). Such flavour-charged mouth or breath directs offspring multisensory scrutiny towards the eating mother, inducing attention and observational learning of palatable foodstuffs (e.g. [104,105]). This incidental ‘maternal demonstrator’ effect can be so powerful that it induces offspring to adopt atypical or maladaptive ingestive habits (e.g. kitten eating bananas: [106]). There is circumstantial evidence that human infants and young children want to taste foods following carers' oral food odours. Such mother-induced odour learning may be secondary to intentionally giving infants premasticated foods, a commonplace practice (e.g. [107–109]) which makes non-milk foods more digestible, and exposes infants to pre-treated highly odorous foodstuffs [110], whose novelty may be attenuated by the carer's added saliva and other oral odour substrates (labial sebaceous glands, breath). But, so far, nothing is known about whether human maternal saliva channels chemosensory information to offspring as it does in other species (e.g. [111,112]).

In other mammals, additional olfaction-based weaning strategies imply switch-like processes based on specific, unconditional chemosignals. Pheromones emitted in milk (rabbit: [80]) or in breath (murine rodents: [113,114]) tag as attractive any co-occurring odorant. The appetitive mammary pheromone of the rabbit is interesting in that context because its concentration in milk declines in parallel with its decreasing reactogenic potency for pups in the week preceding complete weaning [115], literally turning off milk-feeding. Another such odour-based ‘weaning gadget’ has been described in the lactating female rat, whose (unknown) caecal ‘pheromone’ attracts offspring to her faeces [116]. In many mammals, infants are coprophagic of maternal faeces (e.g. [117]), thereby taking in information on mother's dietary composition, as well as safe-tested microbiota (e.g. [118]; but see [119]). However, such unconditional odour-based biological switch processes seem absent from human weaning; but, at least in some human groups, efficient weaning-switch processes have been devised culturally by adulterating the breast with unfamiliar, irritating or disgust-eliciting odorants/flavours (e.g. [120]).

In summary, comparative evidence indicates that human mothers might familiarize their offspring with a range of odorants, and habituate them to cope with low-level environmental novelty, by presenting them with variable odour cues in milk or foods. Food odour-based familiarization, which sometimes appears imprinting-like, is achieved through multiple, redundant processes, some operating pre-functionally (perinatal learning), and others working alongside postnatal opportunities and constraints. Thus, human mothers shape draft versions of the food environment that their offspring will later face directly. In addition to these psychobiological facilitators of the weaning process, human societies have developed additional abrupt or progressive weaning strategies matching their own sociocultural settings (e.g. [121,122]).

4. Development of odour-based social cognition

(a). From discriminating parts to recognizing whole individuals

Nursing-related odour experience may kick off discriminative processes that initiate the recognition of distinct classes of conspecifics. Odour signatures in AF, milk, maternal skin or sweat potentially convey nested odour traits or states characteristic of multiple socio-cognitive levels: (i) species, (ii) classes of conspecifics, and (iii) individuals. Informative level (i) is exemplified in infants' differential treatment of human breast/milk odour and odours of heterospecific milk (e.g. bovine milk) [66,67,70]. Informative level (ii) allows categorizing classes of conspecifics: i.e. lactating women versus non-lactating women or males [50], or possibly early versus late lactational stage among lactating women (as in mice, [123]). Finally, newborns discriminate idiosyncratic odour traits of the mother: breast-fed newborns turn their head more to their mother's milk odour than to another woman's milk odour [14], and similar results arise in 6-week-old infants with breast odour [67].

Selectiveness for the mother's odour increases with age and suckling experience: 2-day-olds respond randomly, whereas greater than 6-day-olds turn longer to their mother's breast odour [49], while motor activity change reveals such differentiation from day 2 [57]. Finally, the mother's neck [57] and axillae [45,124,125] may also emit informative odours. Oral sources (lips, breath, saliva), head (scalp, hair, ears, tears, neck), hands and other odour sources await further testing. Mothers might thus be sensed as olfactory mosaics with the possibility that some cues work as time-givers because of their regular contingency with different affordances or social configurations (e.g. breast odour with sucking, neck with upright carrying, axilla with arm-carrying, face with kissing-vocally interacting, etc.). Additionally, mothers may convey redundant identity cues stemming from different body areas (e.g. skin and milk, [40]).

It is not clear at what level newborns recognize their mother. Do they orient to her breast scent because it carries inherently attractive chemosignals emitted by any lactating female, because they anticipate the recurrence of a rewarding experience, and/or because they view the mother as a unique individual? Tests on newborn olfactory recognition are so far equivocal because they have used odour stimuli from donors differing in both familiarity and relatedness [126]. The critical test for odour-based individual recognition would oppose the odours of donor individuals who are genetically equivalent and equally familiar to the tested subject (e.g. kin of equivalent exposure). Another paradigm to gauge individual recognition relies on artificial odorants, which can, when associated with maternal care, promptly release liking and wanting responses in babies [88,89,127], with a final attractive potency which equates to that of a natural odour [89]. Using easily controllable synthetic odorants constitutes a suitable way to understand stimulus-, subject- and development-related processes that convert an ‘emotionally neutral’ odorant into a meaningful cue.

Beyond early odour-based recognition of individuals, olfaction may boost social learning through other senses. Maternal odours appear to modulate early visual processing. When exposed to their mother's breast, neonates open their eyes more during the inhalation of a corresponding odour [58]. Thus, breast odour can mobilize vision and touch to approach the pigmented/warm areola of the breast. Indeed, synchronous olfactory–visual inputs recruit more oro-motor actions than each of those inputs independently [58], presumably facilitating both latching and attention to the contiguous mother's face. Overall, from the start of postnatal life, maternal odours, so far mostly investigated unimodally, may expedite the growth of multisensory social cognition [128].

Olfactory recognition and discrimination of parts of an individual, as described above, may pave the way to the representation of individuals as whole agents. Other investigations of the impact of odours in early social recognition have altered a conspecific's typical odour and documented the impact on social representations. In squirrel monkey neonates, olfactory and visual cues interact early to form what we think is a maternal representation; when the odour cue is altered, the visual representation is disturbed, degrading recognition [129,130]. Human mothers often alter their olfactory presence with artificial scents and, although we know infants easily learn synthetic odorants made contingent with the mother as familiarity cues [88,89,127], virtually no data exist on how they affect infants' social (re)cognition. Infants certainly integrate odour with traits detected in other modalities when performing recognition within multisensory scenes (reviewed in [128]). Indeed, when 4-month-olds view a female face versus a car, they look at the face (particularly the eyes) longer than at the car in the presence of the mother's odour [131]. At another level, the mother's body odour enhances a face-selective EEG response over the right occipito-temporal cortex in the infant brain [132]. Thus, in her physical absence, the mother's odour triggers face-selective behavioural and neural processes in infants. However, the specificity of the mother's body odour remains to be ascertained against another mother's/father's odour, or against any arbitrary intensity-matched odorant.

(b). Social diversification and olfactory recognition of conspecifics

Beyond infanthood, when toddlers can voice or otherwise indicate their choices, evidence for odour-based individual identification should become clearer. When children aged 3.5–5 years took an olfactory recognition test based on t-shirts from their mother versus an unfamiliar woman, 18 of 26 chose the mother's t-shirt in greater than 60% of the trials, but only eight in a statistically significant manner [57]. In another study, 3–5-year-olds had to select their mother's t-shirt among five others: only 6 of 19 succeeded [133]. A further test [134] assessed whether 6–15-year-old children could identify their mother's or father's t-shirt (relative to a t-shirt worn by a sex-matched unrelated participant). The father's odour was identified by daughters and sons alike. The mother's odour was chosen at random across the sample, but correctly by only the older group when the participants were split into 6–8 versus 9+ year-olds, illustrating how data processing can affect outcomes [135]. A last within-family odour recognition study found children's identification of their parents to be unreliable (7–18-year-old English sample; [136]). Finally, older daughters (11–21 years) recognize their mothers' neck odour, but not her axillary odour [137]. This inconsistency in recognition performance of parents’ body odour by their offspring is quite surprising. It may be related to contrasts in methods (instructions, context/social setting, nature of odour, collection and conservation, odour–distractor ratio), among other sources of variation pertaining to parental factors (nature/intensity of odour, prevalence of perfumes), child factors (sensitivity, attentional demand of tests: distractibility, boredom, fatigue) or both (attachment-related proxemics). Perhaps, the mother's olfactory presence is dominated by artificial scents, on which children might rely more to recognize the mother than on cues originating from her natural skin (but see [138]). Indeed, 5-year-olds express accurate recognition of their mother's perfume and greater desire to wear it as a scent [139]. Olfactory recognition of parents, especially mothers, is perhaps no longer functionally relevant when children begin to escape the family and engage with same-age groups, and there may be no strong pressure in children's everyday life to recognize parents by olfaction alone when more reliable distal (vision, audition) cues are available.

Other studies have examined children's olfactory recognition of siblings and peers. Three 8-year-olds could tell apart the t-shirts of their own full siblings from those of unrelated age-mates [140]. Among sibships sharing different degrees of consanguinity (full (0.5), half- (0.25), step-siblings (0)), 4–11-year-old children correctly identified only the odour of full siblings; either genetic makeup affects body odours more than sharing the same environment or the degree of relatedness is confounded with social experience (proximity, familiarity) that translates into greater odour awareness. Thus, children can recognize sibling odour, but the evidence for true individual recognition remains weak as tests often contrast two donors, related versus unrelated (but cf. [134]), which prevents confirmation of whether an odour characterizes an individual or a higher level category (e.g. familiarity, gender, age). Additional data on odour-based social recognition concern schoolchildren in whom genetic effects are minimized and familiarity effects maximized. For example, 4–5-year-olds can identify classmates from their neck odour, with girls succeeding in 69% of tests (36 out of 52) and boys in only 33% (21 out of 62) [141]. Otherwise, when 9-year-old classmates were asked to recognize the t-shirts worn by six different odour donors (i.e. self, most liked peers (same/opposite sex), least liked peer (same-sex) and mere acquaintances (same/opposite sex)), they could identify the donors better than chance [142]. The same-sex peers were more accurately identified than opposite-sex peers, in line with preferential same-sex affiliations at this age [143]. Thus, children's peer recognition appears to vary under the joint constraints of the gender of the smeller, the gender of the donor and their mutual familiarity/relatedness. The odour cues used in this latter recognition task are unclear as, for the sake of ecological validity, children's natural body odour was not separated from artificial scents. Both natural and artificial olfactory signatures are recognizable to children, as shown by their identification of t-shirts from unrelated donors [144], categorization of gender based on the perfumedness of body odour [142] or reliance on mother's perfume [139]. Thus, children's odour-based social cognition is a particularly interesting area to analyse the developmental dynamics of biology–culture interactions. Conclusions around whether children are better able to use olfactory cues to recognize peers than parents await a direct test of that question, comparing the different odour donors within one group of children.

(c). Children's use of parents’ odours in socio-emotional and cognitive contexts

Beyond mere recognition, body odours may subconsciously drive differentiated social behaviour. Newborns and young infants turn towards familiar body odour sources, consistent with the general trend of attraction to the familiar. The same trend appears for artificial odorants associated with the lactating mother [88,127], the reinforcing effects of which carry over to later object choices in novel contexts [127,145] and could influence later social selectivity. Such memories of the sensory features acquired in contingency with the carer might elicit differential affective treatment of conspecifics at later ages.

Children can draw from the emotional content of social odours. For example, the axillary odour of stressed adults augments startle responses as potently in prepubertal children as in adults [146], and children may thus be able to detect, monitor and remember adults' emotional states. Children also monitor extraneous odorants associated with emotion-arousing contexts such as alcohol or tobacco (e.g. [147,148]). In summary, children can single out the odours of individuals or categories of individuals, but might also tag such odours as cues to emotions transmitted or induced by them. Such odours may mediate strong discriminative treatment (affiliation/rejection; favouritism) between sibling or extra familial group members (e.g. [149–151]).

Children's self-regulation of negative emotions may constitute another context to confirm the long-term impact of parental odours, and perhaps even provide evidence of positive imprinting. Adolescents and adults often report that the highly pleasurable odour memories of conspecifics, especially of the mother, trace back into childhood [152,153]. In principle, subject to individual variations, maternal odours may provide cues of physical proximity and corresponding affordances, such as feelings of security, homeliness, reliance or trust. This informative content of maternal odours appears general among other mammals, where separated offspring are systematically soothed by the mere delivery of the mother's odour (e.g. [154]). Children and young adults often seek the body odour of familiar and/or related individuals in adverse situations (stress, anxiety, separation). The fact that they also rely on their own odours retained on an ‘attachment object’ suggests that familiarity is decisively soothing (e.g. [155]).

Other paradigms have gauged effects of maternal odours on children's socio-emotional functioning. One particularly interesting approach found that children aged 13 years with autistic spectrum disorders (ASD; but not those without) demonstrated enhanced automatic imitation in the presence of their own mother's axillary odour [156,157], indicating that children with ASD have greater attendance to social odours. Finally, through its buffering effects and the provision of an olfactory secure base, maternal odour may be beneficial in reducing fear, optimizing attention and learning, and easing response to novelty (see [158] for similar effects with familiar odorants). More investigation is needed into the emotional balancing and trust-enhancing effects of parent-related odorants, relying on behavioural markers of interpersonal trust and compliance (e.g. following behaviour, contact seeking, joint attention, eye contact, smiling, lexical content) in mutual infant–parent attention or in joint attention paradigms.

5. Homeostatic potency of maternal odours

Maternal odours are usually concurrent with the mother's presence. But the mother's odours (unlike her appearance, touch, warmth, sounds, etc.) can persist in the offspring's immediate environment in her physical absence, as an effluvium or on an object. They can thus prevent or accelerate recovery from the negative effects of separation, novelty, aggression or pain, and support the establishment of basic homeostatic processes in infants.

In line with the buffering effects mentioned above, restituting maternal odour to separated infants reduces the activation of the hypothalamo-pituitary-adrenal (HPA) axis and related behavioural and endocrine manifestations. Thus, cortisol release induced by acute pain inflicted upon separated neonates is tempered by the administration of human milk odour [159], or of dodecalactone, alleged to resemble milk odour [160]. This effect is stronger when the odour arises from own mother's milk than another mother's milk or formula, pointing to the involvement of a familiar, individual-specific chemical signature. Similar effects have been obtained in separated premature neonates who, with or without a pain challenge, evinced lower salivary cortisol when exposed to the odour of own mother's milk (against formula odour: [59,161]), an early response that can be interpreted in relation to the TOC. Replication in older infants and children awaits, although non-human studies find that maternal (and sometimes paternal) stimuli buffer stress only in pre- and post-weaning individuals (e.g. [162]). Similar processes were noted in 7-month-old infants looking at happy versus fearful faces during EEG recording. While exposed to own mother's t-shirt odour, the typical brain response to the fear stimulus did not occur, whereas it clearly appeared in the control contexts (another mother's odour or no odour) [163]. The social buffering effect of maternal stimuli on HPA activation decreases in adolescents compared to children [164], but we do not know whether this also occurs for maternal odours. Do odour stimuli from other social partners (age-mates) become potent buffering agents? By adulthood, a partner's odour (e.g. on clothes) can provide comfort and attachment in their absence, although some individuals report using their mother's odour (e.g. [165,166]). Clearly, further research is needed on the coping-aid function of maternal odours in the face of distress caused by separation and/or pain.

Maternal odour has also been shown to induce soothing and engage sleep in various mammalian infants (e.g. rats [167], cats [168], chimpanzees [169]), including human infants [57,170]. This fact has been translated into practice in exposing hospitalized infants/children to a cloth carrying maternal odour with the goal of aiding them to cope with separation anxiety in unfamiliar settings (e.g. [171,172]). But parent–infant separation arises regularly with sleep, at least in European-American cultures where sleeping apart prevails [173]. To cope with this recurrent transition, infants frequently rely on odorous ‘sleep-aids’ (pieces of cloth, fluffy objects or their own hands; e.g. [174,175]). The mother's odour appears to be effective in her absence, and thus may be an efficient regulator of calm and sleep in infants left alone. One experiment [173] explored this hypothesis longitudinally in infants aged 3, 6, 9 and 12 months who slept alone with a t-shirt containing the mother's odour, and did not find that the t-shirt was privileged in inducing soothing, but (suboptimally) only renewed the odour stimulus every month.

Maternal odour could also influence the structure of sleep and sleep-dependent cognition, as olfaction is clearly functional during infant sleep, especially in the active sleep stage (equivalent to adult rapid-eye-movement sleep) (e.g. [68,174]). When co-sleeping, mothers and infants mostly face each other, thereby exchanging body odours as well as non-odorant volatiles from breath (CO₂, NO), an exchange thought to stimulate the sleeping infants' respiration and awakenings [176,177]. So far, there is little paediatric research on whether and how information from social odours is acquired, integrated or consolidated during sleep (as shown in adults, e.g. [178,179]). Considering that (i) the newborn brain is receptive to odour information for at least 50% of its sleeping time (totalling 70% of the 24 h cycle), (ii) olfactory memory and its multisensory and hedonic connections are sleep-dependent [180], and (iii) learning and consolidation function well in sleeping infants when external interference is reduced [181,182], the fact that information co-occurring with mother's body odour can be acquired and updated in somno is a promising research topic in infants and children.

6. Parental olfactory influences in juvenility and adolescence

(a). Olfactory psychobiology of adolescence

Juvenility to adolescence is a period of increased developmental plasticity. Having benefited from family resources for somatic and psychological growth, juveniles' interests shift from relative neophobia to novelty seeking in all domains; their social interests shift from parents to peers, increasingly befriending opposite-sex peers; and their psychobiology enters the reproductive phase. This transition imposes novel constraints within the familial group, with increased risks of interpersonal conflict, inbreeding and precocious pregnancy [183]. Evolved strategies should have emerged to curb these risks towards fitness in modulating interpersonal attractions within families, regulating sexual maturation of offspring and somehow canalizing the ontogeny of mate choice. What roles could olfaction play in these strategies?

The advent of visible and non-visible secondary sexual characters advertise pubertal changes, when hypothalamo-pituitary-gonadal activation precipitates menarche or spermarche, and boosts all types of skin glands and body excretions into divergent body odours in females and males. In parallel, non-visible changes occur in olfactory sensitivity and reactivity, especially towards adult body odours and components therefrom. For instance, odour thresholds towards the odorants 2-methyl-3-sulfanyl butanol, androstenone and androstadienone, all occurring in axillary sweat, increase during puberty in males but not females [184–186]; but androstenone thresholds tend to decrease through puberty in female participants [184]. When asked to hedonically evaluate androstenone, younger participants rated it as smelling bad more frequently than older ones, and females more so than males. This is in line with the notion that a high sensitivity to androstenone comes along with a more unpleasant perception of its odour [187]. Also late pubescent subjects (15 year-olds) become more sensitive than prepubescent subjects to musky-urinous and sulfurous odorants conveyed in axillary sweat, saliva or sexual discharges [186,188], with pubescent females being more sensitive than their male counterparts [186]. To shed some coherent light on this topic, however, psychophysical research is necessary together with ecologically valid investigations in the same subjects, as developmental changes in the sensitivity to individual body odour constituents may result in different perceptions of odours. Thus, late pubescent children express much stronger aversion than prepubescents (8 year-olds) to the odour of t-shirts worn by unfamiliar young adults [189]. Within families, pubescent girls and boys tend to avoid the odour of fathers’ t-shirts (6–15-year-old Canadian sample; [134] or to clearly reject it [136]). However, Czech postpubescent girls report a liking for adult male odours [153] and indeed androstenone was shown to become attractive to females as a function of their association with sexual experience [190]. Thus, body odours from adults tend to evoke intense dislike before/during puberty and to become attractive in later adolescence. Post-menarcheal variations of olfaction during the fertile phase of the ovarian cycle [191] may also contribute to intermittently attenuate this repulsion.

(b). Olfaction and pubertal timing

The menarche milestone, easier to objectivate than spermarche, has attracted competing theories exploring the mechanisms of its onset and calibration during infancy and childhood [192]. Among multiple, complexly interactive drivers (heritability, nutrition, population density, urban lifestyle, socio-economic status, matrimonial regimen, stress, psychosocial development, exposure to endocrine-disrupting chemicals), some speculate that the chemosensory context inherent to the early developmental ecology may influence menarcheal onset. These speculations rely on epidemiological studies relating family variables and reproductive maturation in female offspring (e.g. [192–196]). First, family stability (presence of biological father) and lower stress levels are thought to provide developmental niches that delay reproductive maturity. Second, father absence and the presence of (an) unfamiliar adult male(s), with possibly co-occurring higher stress levels, would engender environments that translate into accelerated reproductive maturity in female offspring. The proximate mechanisms have been hypothesized to depend on the ‘pheromonal climate’ of their family group [193,196]. In the ‘father present’ family environments, the chemosphere would tend to extend childhood, viz. delay the onset of menarche, following mechanisms akin to the inhibition of neuroendocrine processes controlling oestrus or pubertal timing by chemical cues from the dominant female or older familiar siblings in primates (e.g. [197–199]). By contrast, the chemosphere from the ‘father absent’ familial groups would tend to shorten childhood by accelerating pubertal onset. The Vandenbergh effect, defined as pubertal acceleration by unfamiliar adult males' odour, is suggested to function here as it does in other mammals (e.g. [200]).

While these hypotheses may be consistent with the non-human literature, they are problematic to put to the test in humans because: (i) human studies on priming pheromones mediating socio-ecological conditions into neuroendocrine responses [201–203] have been so far unsuccessful in chemically identifying and functionally validating any candidate compounds responsible for the so-called pheromonal effects (e.g. [92,204–206]); (ii) the likelihood of olfaction dependence of human menarcheal timing, although enticing, appears dauntingly complex, contingent on multisensory events (particularly touch) and mitigated by multilevelled, interactive internal and external causes [192]; and (iii) the olfactory priming of female puberty in other mammals occurs after exposure in early development, often in synergy with exposure to stress, meaning that human studies would have to engage in longitudinal designs to measure events 10–15 years before they are translated into recordable physiological events. Thus, even overlooking the challenge of determining which human-produced compounds to measure in the household atmosphere, it would be difficult to assess the differential ‘pheromonal climate’ hypothesis of menarcheal timing. As a noteworthy aside, the ‘pheromone climate’ notion should be parsimoniously referred to as ‘odour climate’, as domestic effluvia are composed of thousands of biological volatiles emitted by humans (e.g. [207–209]) plus thousands of artificial volatiles [210,211], the latter being probably attended to by children as much as the former as potential cues to the affective climate of the family group (e.g. [147,148]). Perhaps a methodological leap will be possible if atmospheric chemists venture to sample familial environments contrasting in, e.g. social composition, affective stability, conflict or stress (cf. [212,213]).

(c). Developmental calibration of mate affiliation

Early olfactory experience within the familial group may also influence adolescents' reproductive behaviour by calibrating social preferences along which future mates will be selected. Is there a possibility of positive reproductive imprinting in human offspring as in other mammalian offspring (e.g. [214–217])? Such imprinting-like effects of early odour experience have been shown in the ingestive domain [127,218,219], and the perinatal and weaning phases are suggested to be sensitive periods for chemosensory learning in humans [12,13,220,221].

No parallel evidence is currently available for such positive olfactory imprinting effects in human mate selection, although body odour is reported to influence seduction and sexual interaction, especially in females (at least in Western samples of young adults; [222–224]). One study on adult response to human leucocyte antigen (HLA)-covarying odour cues [225] hints in that direction, however. Jacob et al. [226] asked women to rate six male odour donors after they had been HLA-typed and the number of allelic matches specified between the male donors, the women and the women's parents. The body odour of men who bore a low, intermediate level of HLA dissimilarity with the donor was preferred. More important for our argument, these women preferred the body odour of the males whose HLA-type matched that of their father, but not that of their mother. Thus, if fathers' HLA-type covaries with their body odour, a logical assumption would be that young adult females are more attracted to males sharing some odour cues with their own father. However, as mentioned above, negative appraisals of fathers’ body odour by prepubertal and pubertal daughters do not predict such an outcome [134,136]. One possible explanation is that a shift occurs somewhere during development, when a father's body odour changes from being perceived as non-repulsive or even attractive instead of somewhat repulsive. The developmental process involved may be posterior to pubertal perceptual changes, perhaps involving a reversal in olfactory incentives linked with experience of sexual reward [190]. Alternatively, it might be that daughters like odours that are similar (i.e. matched in HLA) but not identical to those of their father, in the same way as women are attracted to faces that resemble their father rather than being attracted to their father himself (e.g. [227]).

In summary, an odour-mediated aversion towards the opposite-sex parent is suggested in prepubertal girls that reverses after puberty to orient preference towards cues of this same parent. This positive imprinting-like process between daughters and fathers is an area open to further investigation. Extant data do not suggest any symmetric pattern for boys and their mothers [134]. This positive imprinting phenomenon goes in parallel with the Westermarck effect [228], a negative imprinting process actualized in later sexual disinterest between individuals who lived in physical proximity during their first 5 years of life [229]. This effect is hypothesized to: (i) depend partly on an odour-based process, (ii) operate during an early sensitive period, (iii) arise when adult-like sexual interests emerge, and (iv) be more potent among females [230]. Thus, similar processes may underlie two different types of imprinting-like phenomena that may be consequential for the avoidance of inbreeding in mate selection. It may be added that the father's odour may have a special status in these processes as it appears more recognizable to children than the mother's odour [134,136], probably owing to its perceptual saliency in terms of intensity and/or quality.

7. Conclusion and prospects

We would flag up four headline conclusions.

• (i)Chemocommunication tracks specific demands of early life-history stages. Note the perinatal and pubertal periods in particular. Newborns' keen olfactory sensitivity appears somehow synchronized with maternal chemo-emissions. Attractive prenatal odorants coat the maternal body areas near to the neonate's nose. The post-parturient's body odour is also influenced by intensified seborrhoea and chemo-emissions from the breast, leading to a probable early lactation-specific odour signature that may scaffold breastfeeding initiation. Adolescence clusters together changes in sensitivity, hedonic valuation and the psychological salience of parental body odours, as well as own body odour production and nascent attraction to others (perhaps canalized by earlier experience of parental odours).
• (ii)Offspring detect multiple informative cues in body odours. Neonates sense maternal odours and may create an odour map that relates different maternal body regions to their reward value, recognizing familiarity/individuality and lactational status. Later, children appear able to use adults' or age-mates’ body odours to detect familiarity, kin, gender, friends or foes, emotional states, perfumedness and atypical odour cues caused by illness. The informative and related chemical contents of all this chemocommunication are wide open to empirical investigation.
• (iii)The offspring's perception of social odours draws from general and specialized perceptual mechanisms. Domain-general perceptual mechanisms (familiarization, conditioning) trace the sensory regularities that pace typical human development. For instance, odour familiarity provides TOC that supports breastfeeding. Suckling also facilitates neonatal learning of the mother's odour after birth, potentially during sensitive windows. Alongside this, domain-specific processes may detect invariant odorant(s) of high survival value. The mammary structure may emit such inherently attractive chemosignals, the perceptual failure of which may compromise neonatal viability [47]. Well documented in other mammals [46,90,231], neonatal response to such specialized signals (pheromones) is a mother-to-infant chemosignalling option that needs to be fully explored in humans.
• (iv)Parental chemomessages have far-reaching outcomes. Existing data raise the possibility of social imprinting in human infancy, but this phenomenon needs to be addressed properly as it has begun to be in the food domain. Further, chemosignals nested in paternal (maternal?) odours have been conjectured to prime children's endocrine functions, contributing to the regulation of pubertal onset. The functional viability of such hypothetical pheromonal processes needs now to be assessed in humans.

The developmental study of social olfaction can serve to further illuminate important theoretical issues, such as the following.

• (a)Olfactory contributions to social cognition. Although audio-visual communication usually prevails in our species, odour may be more impactful early in life, when the audio-visual mode is still maturing. Further research in this area could unveil unexpected functions of olfaction in human cognition. For instance, social odours may pave the way to appreciating individuals as single entities, despite incessant shifts in vocality and visual appearance (posture, orientation, gestures, clothing). There may well be other unexpected functions of olfaction in human cognition, and focused studies may be able to demonstrate how early olfaction permeates the development of multiple non-olfactory cognitive domains.
• (b)Emotional state-dependent odour signalling. The maternal olfactory profile constitutes a safe haven for offspring, although one that can be vulnerable to maternal emotional perturbations (anxiety, depression, fear). Understanding whether such odour cues of perturbed safety occur is a key issue within mother-to-infant communication, with far-reaching consequences for the offspring's sensitivity to emotional contagion, and the development of their ‘landscape of fear’.
• (c)Sniffing behaviour. Questionnaires or interviews have been the principal methodology to understand how offspring engage in olfactory investigation (sniffing) of conspecifics (e.g. [153,232–234]). While these can uniquely capture elusive behaviours or intimate feelings, they focus participants' attention on a particular feature of conspecifics and are prone to reconstructive bias and social desirability effects. So far, we are missing ethologically valid behavioural studies of children's social odour-seeking behaviours. Thus, innovative research designs and devices are needed to objectively record sniffing behaviour in social contexts.
• (d)Biology–culture interactions. Infants are born in culturally constructed olfactory niches: mothers' scents are shaped by local practices (washing, perfuming) and odour-bearing rituals are enacted on offspring (e.g. [235]). Thus, natural and cultural systems of olfactory signs operate simultaneously and it is interesting to gauge whether they do so in synergy or in competition. Effects of this extended maternal odourtype have rarely been considered within early-life transitions.
• (e)Generalizability of research. Finally, some methodological prospects are warranted to improve species-wide generalizability of results. First, research on mother-to-infant chemocommunication should involve bigger samples than those typically studied so far, with a better distribution across ages and with psychobiologically defined age slices. For example, among studies on peripubertal olfactory functioning, rare are those that consider physiological markers of puberty. Second, the studied phenomena should be extended to non-western, educated, industrialized, rich, and democratic (WEIRD) societies [236] to better incorporate the wide range of parental care practices (distal versus proximal care systems; different reliance on olfaction) and how this affects chemocommunication. Relatedly, attention to infants and children afflicted with definitive (i.e. congenital anosmia) or incidental (e.g. enlarged adenoids) olfactory deprivations, or with atypical hypo- or hypersensitivity to odours (ASD, blindness), may be helpful to understand what odours do during development, in the same way as (the very limited number of) studies of adults with olfactory impairments have helped further our understanding of the functions of olfaction [237]. The characterization of odorant–response patterns that are robust across individuals and cultures is required in order to identify species-specific phenomena.

Data accessibility

This article does not contain any additional data.

Authors' contributions

B.S. drafted the article and T.K.S., H.L., K.D. and R.S. provided critical revisions. All authors approved final version of the article

Competing interests

We declare we have no competing interests.

Funding

This research was funded by the CNRS, French Research Agency (under project no. ANR-15-CE21-0009-01 ‘Milkodor’ to B.S.), and the Regional Council of Burgundy-Franche-Comté.