With momentum increasing in the use of social robots as long-term assistive and collaborative partners, humans developing social bonds with these artificial agents appears to be inevitable. In human-human dyads, social bonding plays a powerful role in regulating behaviours, emotions, and even health. If this is to extend to human-robot dyads, the phenomenology of such relationships (including their emergence and stability) must be better understood. In this paper, we discuss potential approaches towards operationalizing the phenomenon of social bonding between human-robot dyads. We will discuss a number of biobehavioural proxies of social bonding, moving away from existing approaches that use subjective, psychological measures, and instead grounding our approach in some of the evolutionary, neurobiological and physiological correlates of social bond formation in natural systems: (a) reductions in physiological stress (the ''social buffering'' phenomenon), (b) narrowing of spatial proximity between dyads, and (c) inter-dyad behavioural synchrony. We provide relevant evolutionary support for each proposed component, with suggestions and considerations for how they can be recorded in (real-time) human-robot interaction scenarios. With this, we aim to inspire more robust operationalisation of ''social bonding'' between human and artificial (robotic) agents.