In the target article, Thomas proposes that infants’ innate, primitive understanding (i.e., “core knowledge”) of the social world includes categorization of interpersonal relationships as communal sharing, authority ranking, or equality-matching, in accordance with A. P. Fiske’s (Reference Fiske1992) Relational Models Theory. At two points, Thomas briefly notes the possibility that identification of attachment relationships may be even more fundamental. Here, we argue that this is likely the case and develop that proposal further. Attachment theory describes a well-documented set of psychological processes potentially underlying infants’ classification of relationships into these three models, with implications for future research on social-cognitive development.

Throughout the mammalian class, infants possess a suite of innate biobehavioral mechanisms for figuring out who is supposed to care for them – their attachment figures – and optimizing the care received (Feldman, Reference Feldman2017). Attachment figures provide comfort, protection, and reassurance in times of uncertainty or distress (safe haven) and enable confident exploration of and engagement with the world (secure base; Bowlby, Reference Bowlby1982). While it is not yet clear whether human infants define attachment figures categorically or along a continuum, they do so based on well-documented cues, and begin to show caregiver-specific proximity preference and separation distress at roughly six months of age – just as they begin to move around independently, and risk getting into trouble (Bowlby, Ainsworth, & Bretherton, Reference Bowlby, Ainsworth and Bretherton1992).

Fiske’s relationship models map to features of attachment bonds in striking ways. First, the maternal-infant bond is arguably the primordial communal “we are one” relationship. Much of Thomas’s evidence supporting infants’ ability to identify communal sharing relationships focuses on synchrony – temporal concordance of biological and social processes between two or more individuals. Synchrony is the most widely studied predictor of attachment formation, and a foundation for healthy development (Feldman, Reference Feldman2003; Reference Feldman2007; Reference Feldman2012). As Thomas notes, infants are not only sensitive to cues of synchrony (Powell & Spelke, Reference Powell and Spelke2013), but also observing others’ synchrony leads infants to expect that those individuals will be responsive to one another’s distress (Kudrnova et al., Reference Kudrnova, Spelke and Thomas2024) and will help and support one another (e.g., Jin & Baillargeon, Reference Jin and Baillargeon2017; Pun, Birch, & Baron, Reference Pun, Birch and Baron2021; Ting & Baillargeon, Reference Ting and Baillargeon2021). This suggests that infants may use synchrony as a cue to classify others’ relationships in attachment-related terms, as well as their own.

Moreover, infants’ attachment bonds are inherently authority-ranked. Attachment bonds are crucial for infant survival and development because the infant can defer to someone with more experience, resources, and physical capabilities – a useful “insurance policy” when exploring the environment. While we agree there is compelling evidence that infants can and do compare the relative rank (e.g., physical, material, and social) of those they observe interacting (e.g., Meng et al., Reference Meng, Nakawake, Nitta, Hashiya and Moriguchi2019; Pun, Birch, & Baron, Reference Pun, Birch and Baron2016), and expect those of higher rank to provide benefits to those of lower rank (Stavans & Baillargeon, Reference Stavans and Baillargeon2019), this is consistent with our proposition that infants understand social relationships first in terms of attachment bonds. Beyond simply identifying others’ relative ranks, infants expect these differentials to be stable over time (Bas & Sebastian Galles, Reference Bas and Sebastian-Galles2021; Mascaro & Csibra, Reference Mascaro and Csibra2012). This expectation maps onto how infants track what they can expect from their attachment figures – the extent to which they can be consistently relied upon for resources and support – and calibrate their own behavior accordingly (Ainsworth, Bell, & Stayton, Reference Ainsworth, Bell, Stayton and Schaffer1971; Shaver, Collins, & Clark, Reference Shaver, Collins, Clark, Fletcher and Fitness2014).

Infant-caregiver attachment bonds are also inherently unequal, with clearly defined roles of caregiver and care-receiver. If infants do indeed categorize others’ relationships in terms of attachment, equality-matching across interactions (or the lack thereof) may be one way to do so. With no prior information about a relationship, infants expect resources to be divided equally among interaction partners (e.g., Sloane, Baillargeon, & Premack, Reference Sloane, Baillargeon and Premack2012; Ziv & Sommerville, Reference Ziv and Sommerville2017). However, after observing initial interactions in which someone gives and someone receives, infants expect the giver to keep giving in future interactions (Tatone, Geraci, & Csibra, Reference Tatone, Geraci and Csibra2015). Perhaps infants assume by default that others’ relationships do not include attachment, and therefore should involve balanced give-and-take across interactions.

There is some evidence that infants generalize from their own attachment experiences to expectations for others’ relationships. While viewing a video designed to evoke expectancies for caregiver behavior in attachment-relevant situations, securely attached infants expected the “caregiver” character to be responsive, whereas insecure-avoidant and insecure-ambivalent infants expected the “caregiver” to be unresponsive (Johnson et al., Reference Johnson, Dweck, Chen, Stern, Ok and Barth2010). Our proposal raises a number of questions for future research. Is attachment-related core knowledge accessible in the first months of life, or only once the attachment system “turns on” and generates personal experience of attachment relationships? If the latter is true, we would expect greater homogeneity in infants’ expectations about others’ relationships prior to this point, and more diverse expectations and behaviors soon after infants are mobile.

Unfortunately, although mental models of attachment figures are central concepts in attachment theory, research explicitly examining infants’ mental representations of attachment bonds is surprisingly sparse (Sherman, Rice, & Cassidy, Reference Sherman, Rice and Cassidy2015; Luo, vanMarle, & Groh, Reference Luo, vanMarle and Groh2024). One study found that 4–12-month-old infants seem to expect adults to respond to infant crying – the most common cue of attachment distress – with care (Jin et al., Reference Jin, Houston, Baillargeon, Groh and Roisman2018). As others have proposed (Luo et al., Reference Luo, vanMarle and Groh2024), this suggests that mental representations of attachment bonds include behavioral scripts for and evaluations of attachment-relevant situations, which may be applied to understanding others’ relationships as well as their own. Importantly, however, the emphasis on individual differences in the study of both infant and adult attachment has resulted in a dearth of research on how mental representations of attachment figures form. While these mental representations are thought to be malleable, it is unclear how stable they are. Is one schema-violating experience enough to alter mental representations, or do changes emerge only with new, but consistent experiences? Does this process change across development? If mental representations of attachment bonds are indeed a lens through which infants interpret their social world, empirical work documenting how they form and change is crucial to understanding socioemotional development.