Attachment theory (Bowlby, 1969/1982) posits the existence of internal working models as a foundational feature of human bonds. Radical embodied approaches instead suggest that cognition requires no computation or representation, favoring a cognition situated in a body in an environmental context with affordances for action (Chemero, 2009; Barrett, 2011; Wilson and Golonka, 2013; Casasanto and Lupyan, 2015). We explore whether embodied approaches to social soothing, interpersonal warmth, separation distress, and support seeking could replace representational constructs such as internal working models with a view of relationship cognition anchored in the resources afforded to the individual by their brain, body, and environment in interaction. We review the neurobiological bases for social attachments and relationships and attempt to delineate how these systems overlap or don’t with more basic physiological systems in ways that support or contradict a radical embodied explanation. We suggest that many effects might be the result of the fact that relationship cognition depends on and emerges out of the action of neural systems that regulate several clearly physically grounded systems. For example, the neuropeptide oxytocin appears to be central to attachment and pair-bond behavior (Carter and Keverne, 2002) and is implicated in social thermoregulation more broadly, being necessary for maintaining a warm body temperature (for a review, see IJzerman et al., 2015b). Finally, we discuss the most challenging issues around taking a radically embodied perspective on social relationships. We find the most crucial challenge in individual differences in support seeking and responses to social contact, which have long been thought to be a function of representational structures in the mind (e.g., Baldwin, 1995). Together we entertain the thought to explain such individual differences without mediating representations or computations, but in the end propose a hybrid model of radical embodiment and internal representations.