The multiplicity view (MV) is part of Newen’s person model theory (PMT) and claims that individuals apply multiple epistemic strategies to make sense of other people, namely simulation, theoretical inference, direct perception (DP) and primary interaction.[1] He thus interestingly argues against the view that there is something like a default strategy of social understanding. In the following, I will scrutinize MV and, in doing so, attempt to reach three goals: First, I reconstruct the main claims of MV and suggest that the development of such a pluralistic account of social cognition can be seen as contributing to the so-called “interactive turn” (Overgaard & Michael 2013; section “The multiplicity view”). MV has the potential to integrate bodily and interactive contexts, while also paying more attention to the phenomenology of social encounters. Second, I argue that current pluralistic depictions of social cognition – of which MV is a clear example – run the risk of operating under (often implicit) contradictory background assumptions. In the section “Multiplicity needs coherence”, I first show how and why different social cognitive mechanisms have been described under different sets of metaphysical assumptions. Since these assumptions are often contradictory, a coherent version of MV cannot simply claim to combine them. I then go on to argue that the concept of DP as an epistemic mechanism is either metaphysically incompatible with simulation and theorizing, empirically implausible, or – if it is re-formulated so that it fits a representationalist description – does not meet the goal of integrating embodiment and phenomenology anymore. I will thus claim that DP should be used as a phenomenological rather than epistemological concept. My third goal is then to suggest novel ways of adopting a pluralistic perspective on social cognition, while remaining in metaphysically coherent territory. Metzinger’s theory of first-, second-, and third-order embodiment (1-3E) is a conceptual framework that combines representationalist and non-representationalist levels of analysis in order to show how a specific phenomenal quality (e.g., phenomenal selfhood) can arise within an embodied system (Metzinger 2014). Metzinger claims that phenomenal properties are computationally grounded in a representation of one’s body (the “body model”, ibid., p. 273), which in turn is physically implemented by bodily and neural structures. I aim to apply this idea to the study of social understanding (section “1-3sE – Levels of social embodiment”). This application enables a more fine-grained depiction of different phenomenal qualities in social encounters and shows their putative relation to representational and physical counterparts. I ask which parts of the body model could potentially be shared and thus be exploited for a skillful navigation of an individual’s social environment. In a last step, I sketch the physical grounds of social cognition.