The preference to hold infants on the left rather than the right side of the body, as reported in human populations ¹ , reflects socio-emotional processes and potentially facilitates mother–infant relationships ^2,​3,​4 . Recent findings support the hypothesis that a left-cradling bias arises from the right hemisphere advantage for social processing ^5,​6,​7 , for example, visual recognition of infant facial expressions ⁸ . The position of an infant on the mother’s left side may optimize maternal monitoring, by directing sensory information predominantly to the mother’s right hemisphere ^3,9 . Right hemispheric superiority underlies the perceptual lateralization in social behaviours of many phylogenetically diverse taxa ^5,10 . More specifically, a variety of non-human mammals have shown a right hemisphere dominant role in the monitoring of conspecifics ^10,11 . Considerable evidence supports the idea that lateralized mother–infant interactions may not be restricted to the primate lineage. The paucity of studies on non-primate species restricts our understanding of how widespread and consistent lateralization is in mother–infant interactions among mammals. Here, we investigate the lateral biases in the behaviour of mothers and infants from a diverse range of mammal species.

Both mother and infant potentially impact on lateralization ¹² ; therefore, we aimed to differentiate their roles in asymmetrical positioning and to assess their lateral preferences separately. To record the choice of lateral position in mother–infant pairs, we studied ‘follower-type’ species, in which females and their dependent infants usually move side-by-side on parallel paths (Supplementary Fig. 1). Species with laterally placed eyes and relatively little binocular overlap were chosen because in such animals the expression of brain lateralization through one-sided behavioural biases is especially pronounced ¹³ . We employed a continuous focal animal sampling method ¹⁴ to record spontaneous mother–infant reunions after short-term spatial separations. The choice of lateral position was recorded when one pair member approached the other from behind and positioned itself on either the left or the right side (see Methods for details). This sampling method allowed for a straightforward analysis of individual- and population-level lateralization in a natural setting. Data consisted of 10,905 lateral position choices for 175 individually identified mother–infant pairs.

Hemispheric specialization for perception of conspecifics can be reflected by the physical positioning of individuals within their social environment ^{15,​16,​17,​18} . Therefore, a lateral position preference was used as a behavioural marker of social lateralization in infants and mothers. Infant lateral preferences at both the individual and population level were investigated in feral horses (Equus ferus caballus) living in natural social conditions and species living in the wild: Pacific walrus (Odobenus rosmarus divergens), Siberian tundra reindeer (Rangifer tarandus sibiricus), saiga antelope (Saiga tatarica tatarica), muskox (Ovibos moschatus), eastern grey kangaroo (Macropus giganteus) and red kangaroo (M. rufus). From one to five types of mother–infant behaviour were investigated for each species, with ‘slow travelling’ studied in all species. In slowly travelling pairs, the majority of infants (74–90%) preferred to keep mother on one side (left or right) rather than the other (Supplementary Table 1). Population-level analyses based on scores from an individual laterality index (see Methods) revealed a preference to keep mother on the left side, compared with the right side, in infants of all species (Fig. 1). An additional method of population-level lateralization testing that incorporates single observations per individual (see Methods) was used to increase the variety of studied species and behaviours. Based on this method, lateralization in an infant’s position near the mother was investigated in wild argali (Ovis ammon) and southern right whales (Eubalaena australis).

A significant population-level bias for keeping mother on the left side was found regardless of the species, behavioural type (with only the exception of resting in reindeer, see Supplementary Tables 1 and 2), or type of analysis (multiple versus single observations). Importantly, a left-side bias existed in infants not only in routine behaviours, but also in stressful situations, such as when fleeing. Our results indicated that a lateral bias reflects the infants’ preference to position the mother in their left hemispace (left visual field) rather than the preference for a particular side of the mother’s body. For example, Pacific walrus calves preferred to keep mother on the left side when resting both in a co-directed position (z = 2.16, P = 0.029) and in a counter-directed position (z = 3.06, P = 0.002) relative to the mother. That is, the leftward bias emerged regardless of which side of the mother’s body was exposed to the infant. The same was true for foals and saiga calves showing a significant preference to keep mother on the left when approaching her for suckling from different sides of her body (see Supplementary Tables 1 and 2). Factors such as an infant’s sex (tested in horses, grey kangaroos and red kangaroos) or age class (tested in walrus, horses and muskox) had no significant effect on lateralization (Supplementary Table 3).

The overall evidence for marine and terrestrial mammal species points to a common pattern of lateralization in an infant’s perception of mother (Fig. 1, Supplementary Tables 4–12). A meta-analytic approach was used for interspecies comparison. First, the meta-analysis based on individual lateral preferences revealed a uniform infant preference to keep mother on the left side, as indicated in the forest plot (Supplementary Fig. 2). Second, the meta-analysis based on single observations per individual showed lateralized spatial positioning to be consistent between infants of 11 non-human mammals approaching mothers (overall meta estimate = 0.74, P < 0.001; Fig. 2) and for human children approaching adults ¹⁵ . Similar to mammals in the present study, human children have been investigated unobtrusively in a natural setting that facilitates comparison with our results. It is important to note that the control condition, that is, choosing a position relative to stationary inanimate objects, did not elicit any lateralization either in children ¹⁵ or feral horse foals (see Supplementary Table 2). One-sided preferences in spatial positioning relative to a stimulus reflect the differential use of the lateral visual field of the left and right eye, and its underlying hemispheric specialization, as proved by a wide range of studies on vertebrates ^13,18,19 . Within this context, the spatial preferences demonstrated here arise from the dominant role of the right hemisphere in an infant’s perception of mother. A left eye/right hemisphere advantage for the socio-emotional processing has been repeatedly reported for both human and non-human primates ^5,11 . For example, great apes preferentially keep conspecifics on their left side compared with their right side ¹⁶ , much like non-primate infants in the present study. Our findings support the theory that a right lateralized ‘social brain’ in primates, including humans, stems from earlier forms of lateralization common to vertebrates ²⁰ . The preference to keep a parent on the left side when begging for food observed in five out of six juvenile Australian magpies (Gymnorhina tibicen) ²¹ may imply that the lateralization of parent–offspring interactions is not restricted to the mammalian lineage.

It is reasonable to suggest that a prevalent left-sided bias may confer advantages for an infant. To test this hypothesis, we analysed the behaviour of infants within the equal time intervals that they spent in different lateral positions (see Methods). First, we found that feral horse foals initiated more bonding behaviour (positive social contacts) with mother when they kept her on the left side, rather than the right side (paired t-test: t ₁₈ = 4.97, P < 0.001, n = 19). Second, we showed that the frequency of spatial separations between the pair members (when the infant was left behind the mother) was lower when the infant kept its mother on the left side before the separation, both in feral horses (paired t-test: t ₁₉ = 3.65, P = 0.002, n = 20) and Pacific walrus (paired t-test: t ₁₀ = 6.25, P < 0.001, n = 11). Thus, when perceiving a mother predominantly via the left eye–right hemisphere system, infants initiated more bonding behaviour and maintained the spatial proximity to mother more successfully than when perceiving her predominantly via the right eye–left hemisphere system (Supplementary Fig. 4). This may derive from a more general specialization of the right hemisphere for visuospatial processing, resulting in the attentional bias for the left visual hemifield ²² . Facilitation of bonding with mother and the maintenance of spatial proximity are obviously beneficial for an infant’s survival ²³ . From the point of view of selective pressures, our findings indicate that it is advantageous for an infant to be left-lateralized.

We further demonstrate that the left-side bias in infants is not a specific response to mother, but derives from a more general lateralization in social behaviour. For example, all muskox calves studied showed individual preferences for a particular lateral position when approaching other familiar calves of the same age class (z = 3.18, P < 0.001, n = 12). A population-level preference to keep age mates to the left rather that the right side was found (W = 76, P < 0.001). Analyses based on single observations per individual pair indicated similar left-side biases in infant–infant interactions of saiga antelope (z = 2.59, P = 0.008, n = 18), feral horses (z = 2.18, P = 0.027, n = 29), beluga ²⁴ and orca ²⁵ . These results are consistent with the lateralization in human children when approaching a peer ¹⁵ . Lateral preferences in positioning towards conspecifics previously shown in interactions between adult mammals ^16,17 were found to occur also in interactions between immature individuals.

In contrast to the infants’ lateral preferences, lateralization in a mother’s choice of position was behaviour-type-specific (Supplementary Tables 13 and 14). In feral horses, the numbers of lateralized and non-lateralized mares were compared, with significantly more mares showing no side preference for slow travelling (z = −2.62, P = 0.007, n = 21) and resting (z = −2.58, P = 0.007, n = 15). No population-level biases were found in these behaviours (slow travelling: W = −48, P = 0.383; resting: W = −59, P = 0.097). Conversely, when fleeing, the majority of mares were lateralized (z = 3.18, P < 0.001, n = 12) with a population-level preference for keeping foals on the left side (W = 76, P < 0.001). A similar pattern was observed in wild eastern grey kangaroos. In slowly travelling pairs, the significant majority of females had no preference for a particular lateral position relative to an infant (z = −2.02, P = 0.039, n = 12), and there was no population-level bias (one-sample t-test: t ₁₁ = 1.67, P = 0.123, n = 12). However, analysis based on single observations per individual showed that, when fleeing, grey kangaroo mothers kept their infants predominantly on the left side (z = 3.21, P < 0.001, n = 28). Our results indicate that maternal lateralization was not pronounced in routine non-threatening circumstances, but emerged in stressful, potentially threatening situations. A left-sided bias has also been suggested for orca mothers when stressed by a vessel approach ²⁵ . We can assume that a left-side bias in mothers is related to the increased need to monitor the infant in an unsafe environment. The left eye–right hemisphere system provides higher accuracy and speed for many types of social responses compared with the right eye–left hemisphere system ^5,10,18 . Therefore, when keeping an infant in the left visual field, the mother may derive optimal control of the infant state. This is generally consistent with human studies that have demonstrated that a left-cradling bias is most pronounced during the first weeks of an infant’s life, when maternal visual monitoring of the infant’s state is most critical ⁶ .

Our results demonstrate that humans are one of many species showing strongly lateralized mother–infant spatial relations. The remarkably consistent occurrence of lateralization in interactions between an infant and its mother indicate that this is likely to be a pervasive mammalian feature of ancient evolutionary origin. Previous studies have shown maternal lateralization for holding an infant ^1,6,26 in human and non-human primates. In this study we provide the first direct demonstration of lateralization in an infant’s active choice of spatial position near the mother in a diverse range of mammal species. Our findings suggest that sensory lateralization facilitates mother–infant bonding. We empirically demonstrate the advantages of lateralized infant behaviour in both marine and terrestrial mammals in support of the hypothesis that lateralization contributes to biological fitness ^19,20,27 .