We use the term ‘social’ to refer to a wide range of phenomena at different levels of abstraction. From the weakest to the strongest sense, these are:

1. As commonality, so that it is enough for different individuals to have something in common for there to be a social phenomenon. Thus, one person dressing all in red at the park would not be a social phenomenon, but many people doing it would.
2. As plurality, so that a phenomenon is social if it essentially involves more than one individual. Thus, even the simplest of human interactions – a conversation, greeting an acquaintance on the street, etc. – counts as social in this sense.
3. As emergent dimension, so that social phenomena emerge from the interaction of individuals. Think of Lewis’ notion of convention and other similar views of institutions, where these are genuine phenomena even if grounded on individual actions, desires, beliefs, etc.
4. As a genuinely autonomous ontological category, irreducible to the individual.

There are ‘social’ phenomena of all these kinds, and it is very likely that most if not all of the social phenomena we care about as philosophers are complex enough to occur at more than one ontological level. This raises the question of what to do with complex social phenomena that do not fit neatly into a single one of the levels above. I could only identify the following three options: First, one might adopt a merely logical composition of the different levels, so that for example, one might think that all it means for something to be social is to be so in at least one of these four senses, or just in some of them, etc. This approach has the advantage of being very simple, but just like any disjunctive account, it is not very robust in so far as it does not explain what these different levels of sociality have in common. Thus, one might preferred a more structured picture of sociality. This can be of one of two kinds: either hierarchical or non-hierarchical. Thus, for example, one might think that one or more of the levels is more fundamental or more properly social than the others. In contrast, one might think that the different levels are not independent, but their relations do not form a hierarchical structure. For example, one might think that the commonality sense of “social” is too weak and conceive it at most as defeasible evidence that there is an underlying genuinely social phenomena that, for instance, made all those people dress the same way in those circumstances.