Faithful readers of this blog will already know that almost nothing in sociology frustrates me more than the “structure vs. agency” debate. The main reason for this frustration, I think, is the radical underdefinition of both terms (and probably the vs. connecting them too). Today, while reading for my comparative/historical methods class, I was faced with five articles and chapters that talked at great length about structure and agency and their relative importance in historical explanations. I was not elated, to put it mildly. What did each author mean by structure and agency? I won’t offer an analysis of each piece (or even mention them, as they often offered no definition of either term) but I want to lay out four things they could have meant by the word agency, each of which might make sense in some context, then offer a few examples:

So, first, agency as free-will. This one is pretty straightforward, I think, but more common than you might expect. I think it all goes back to Durkheim and his laying out of the sociological agenda in Rules. I have a page number somewhere of where he defines his project as examining the constraints of social facts vs. the free-will of human beings. If I recall correctly, Durkheim just uses the term free-will. Later authors invoke agency, but they often mean something more philosophical: the sacred power of an individual to be an individual, have goals that are somehow uncaused by social forces, and re-make the world in order to achieve those goals. I think this particular definition of agency is particularly useless for sociology, even sociology not intent on scientism. For all our fighting to show how various aspects of social life are constructed, up to and including the individual and individualism (cf. Marx, Foucault), we should not let an essentialized notion of individuals slip in the back so easily.

Second, agency as the explanatory residual of structure. This one is sometimes a useful rephrasing of the first without awkward philosophizing. We can posit that race, class and gender structure social life in particular ways that explain much of what happens, but that people still have “agency” which explains why there is not exact conformity. Hence, agency is the residual of structure – it accounts for that which is not accounted for. This notion of agency is pretty wimpy, on the whole, and can be systematically eliminated by the positing of other structures that account for the rest of the variation (much as statisticians might say that the residual is partially explained in terms of omitted variables). This definition does not, however, easily lend itself to an analysis of how agency alters structure unless you think of structures as being alterable within their own logic (i.e. that a structure can change without anything discontinuous happening, that repetition can lead to, or always already is, difference).

Ok, third, agency as a mode of action. This is the way that Callon defines agency in his discussion of the way economics creates “calculative agencies”. What Callon means is that a particular kind of knowledge (economics, marketing, accounting, etc.) makes possible certain modes of action (ways of trying to alter the world) by framing the world into calculable bits and providing the necessary tools and frameworks to make the calculations. It’s also what Callon meant when he investigated the agency of scallops in a now classic piece in early Actor-Network Theory. The scallops had agency because they had ways of acting that altered the outcome of a process. Mitchell often uses a similar definition of agency when he talks about the need for historians, and sociologists, to give agency to non-humans (e.g. “Can the Mosquito Speak?”).

The fourth is a variation or reinterpretation of the third: agency as an analytical counterfactual. This one is my own creation, I think, so its explanation is tentative. History and historical sociology (amongst other disciplines) are preoccupied with contingency – to what extent did things have to be the way they are? Models that emphasize structure suppose that things very much had to be as we observed them. Models that emphasize agency suggest that things could have been otherwise. Thus, agency is the possibility of having acting differently in a way that changes an outcome. Thus, endowing something or someone (a class, an organization, an individual, a mosquito or a scallop) with agency is equivalent to creating a counterfactual: if the mosquitoes had not traveled up the Nile, if the scallops had attached to the scientists’ nets, if the working class had not supported FDR, etc. Make sense so far?

Now, another interpretation of agency-as-counterfactual is that it’s also agency as possible site of intervention. If a scholar is interested in figuring out what the state could do in an economic crisis, then that scholar rates to emphasize the agency of the state in explanations of past economic crises. The question becomes, “What could a state do differently next time?” and thus the analysis focuses on counterfactuals of the form “What would have happened if the state had done things differently?” There is nothing particularly wrong with this focus – indeed, such a purpose justifies criteria for giving certain things or people agency and not others. So, to respond to Mitchell, we don’t give agency to the mosquito because we aren’t writing the book to mosquitoes, so we don’t care what the mosquitoes could have done differently. But, to make that justification work, you have to be clear in your definition of agency, and thus your choice of counterfactuals*, and thus your choice of what perspective the book is written for.

All this is a long way of saying: Sociologists of the world, define your bothering terms. And while you are at it, could you perhaps specify what the claims you are trying to make are for? Because “a better understanding of society” rarely helps anyone in the abstract, and will likely not make your theoretical choices clearer or more justifiable while a more specific intervention might do the trick.

* Not all counterfactuals involve agency, I suppose. You could invoke one based on chance – the outcome of a coin-flip, etc. Or that’s giving agency to nature. Whatever trips your trigger.