I taught the first session of International Security yesterday (see this post of the syllabus and the rationale behind it), and we spent a lot of time, as promised, on the role and promise of assumptions in theory-building and testing. I can go on at (too great a) length about these things, as I’m sure my students discovered, but it allowed for some good, in-depth discussions of a few critical points that I think are worth repeating here. Note that this isn’t an exhaustive outline of what we covered, but just some points I want to revisit. Below the break, of course…
First, with the rules of logic pretty much given to us, choosing our assumptions (or premises, if you prefer) well and creatively is our primary source of power when it comes to explaining our observations of the political world. If logic is the mechanistic part of what we do, the part that’s ideally out of our hands, the art comes in choosing assumptions that illuminate the answers to the questions that drove most of us into this line of work. Every model makes assumptions—whether the person proposing the theory admits it or not—and being cognizant of them, of what they tell us about the proper empirical domain of our theories, is part and parcel of being an effective and intellectually responsible social scientist. In fact, LFC and I have a brief (and, I think, pretty interesting) exchange at the bottom of this post from earlier in the week that illustrates just that.
Next, we talked about some of the more common assumptions made in formal theories of IR (and in game models more generally), one of which was the fixity of preferences. Those of us that use game theory will often tell you that, not only are we unconcerned where preferences “come from” before a given game or interaction begins, but also we assume that preferences are “fixed,” i.e. unchanging. One some level, this seems ridiculous (for example, I truly disliked green beans as a kid, but now, I’ll eat the mess out of them), but stating that our models require that we assume I can never change my mind about the taste of green beans is not what we mean when we assume that preferences are fixed.
What do we mean by that? It requires a definition, first, of preferences, which are just an actor’s subjective ranking of the possible outcomes of the interaction being studied: winning or losing a war, a range of possible bargains, losing or retaining office after any of those distributive outcomes, etc. No more, no less. The analyst chooses what the outcomes are (or the values of the outcome variable, if you prefer), we make assumptions about how players rank them, but these outcomes could be anything, at any point in the process of a political interaction. So when we talk about preferences, they’re defined relative to the outcomes of a specific interaction—outcomes that could be an end-state or a stopover on the way to one. We use preferences, then, to define what goals our actors pursue, and if we want to theorize about how they pursue those goals, we need to hold them fixed for the duration of the interaction in which we’re interested. So when we say that preferences are “fixed,” we simply mean that we need to hold them constant throughout the interaction under study in order to ensure that we can isolate the moving parts that we’re really interested in—the politics.
Does this mean that players can’t change their minds about, say, actions designed to reach those ends in the game? Not at all. But it all depends on what we choose to define as the interaction’s outcome (i.e., our outcome variable)…and that’s up to us, as analysts, when we specify our theories. In other words, what “preferences” are is in one sense arbitrary, because they can be over anything, but to understand the effect of some set of preferences on an outcome, we need to hold them constant so that the stuff we’re interested in—the politics—can be a moving part.
So, yes, we assume that preferences are fixed, but only with respect to the interaction we’re analyzing, and ”preferences” are just defined relative to that interaction as well. Whether someone changes from liberal to conservative, a warmonger to a pacifist, a green bean hater to a green bean enthusiast, are all interesting—and worthwhile—questions, and we can (if we choose) analyze all of them by thinking about them as alternatives on the way to satisfying some other end-state preference. Yes, that assumes they’re instrumental, and there are alternatives to that approach of explaining preference change, but it also implies a slightly different definition of “preferences” than I’m using here. My broader point, though, is just that assuming fixity of preferences is remarkably innocuous, because they’re just a ranking over a specific set of outcomes that we assume defines some actor’s model-specific goals.
I’m not saying anything new or controversial here, but it lets me make another point that I think is important to keep in mind (and that I lose sight of often enough myself): assumptions are not assertions. They’re made for the sake of argument, quite literally. To the extent they capture some element of the interaction we’re studying, then they allow us to make hypotheses conditional on them being satisfied, and—when we realize that assumptions are really the only input we have into the theory-building process—that’s powerful stuff.