I got some snippy responses, well one in particular, to my post on the future of international relations theory based on a reading of the tea leaves over the last year or so. And it made me realize that there is a fundamental divide between me (and I hope others) and rationalists on the issue of assumptions. I thought I’d write about and get some feedback. I’m sure that there is a literature and debate on this somewhere else, but I blog about things that I don’t really have time to look into. Isn’t that the point? (Although I would appreciate it more responsible people pointed me in the right directions…..).
It seems for rationalists that assumptions are statements that one makes to make the building of theoretical models easier. It does not matter if they are true, only if they are useful. Assumptions in rationalism are just things you don’t touch. It is a synonym for elements of an argument that are not subjected to empirical analysis or testing. I guess this is a necessary evil to make formal models in particular work. Otherwise one can’t find equilibria and generate expectations of outcomes.
I think this is completely indefensible. Thinking of assumptions in this way indicates being closed off to information. Assumptions become presumptions. Rationalists essentially indicate that there are some things that they just aren’t going to think about. This is wholly unscientific. The scientific mindset is about questioning absolutely everything. That’s how the whole Enlightenment got started. There is nothing natural and true about the Church and the divine right of kings. Think, you stupid fools! Through the power of reason you will expose the falsehood of your assumptions.
In psychology the frequent use of assumptions is associated with a need for cognitive closure or a lack of “epistemic motivation.” Needless to say these are not things that any academic would like to be accused of. They amount to subjective bias and an unwillingness to think critically. These are the characteristics of ideologues not careful analysts.
When anyone starts a new research project, everything has to be up for grabs. Nothing can be decided in advance. Find a question first; the answer comes second. That is the very essence of social science. I might have a particular hunch based on a previous observation, but if I am not willing to rethink it then I have no business doing this for a living.
In general this leads to an indictment of the use of paradigms in international relations. Paradigms all revolve around assumptions. Oftentimes they are completely implicit and unconscious. For instance, I have argued that structural realists don’t recognize that they are making assumptions that decision-makers are fearful. This is necessary to make the whole thing go. If we don’t question our assumptions we are no better than some conservative at the Heritage Foundation who knows what he is going to find before he looks for it. Or a liberal at…. Well, I can’t think of any obvious counterpart.
We might agree that objectivity is generally impossible. But I can’t see any alternative but to strive for it while recognizing our cognitive limitations. This means being aware of our biases at all times and NOT ASSUMING ANYTHING. The irony of this position, it seems to me is that we get to positivism vis relativism. We must realize that if we are guided deductively by theory (assumptions), then we will have theory-laden (that is unrepresentative and biased) data which will inhibit our ability to see things as they are. So relativism teaches us how to be good positivists. It is a warning to question our assumptions. Isn’t that the point of critical theory, to emancipate us from our presumptions?
What about the counterargument that assumptions are useful even if not true? My point is that they can’t be useful if they aren’t true. It is like a house with a rotten foundation. It will come crashing down. For instance, I constantly encounter formal models on issues of bargaining or crisis behavior among states that sneak in little caveats like the assumption of risk neutrality or risk aversion. Some authors are honest that this is necessary to make the model tractable. Others try to make the claim that there are empirical reasons to make this assumption, which I just don’t buy. In any case, if it turns out to be false, the model completely falls apart. Fearon’s rationalist model of war rests entirely on the assumption of risk neutrality or risk aversion. If risk acceptance is allowed, then there is no overlap in the bargaining space. You don’t need any of that stuff about incentives to misrepresent one’s reservation price to explain war. Some just prefer it to peace, costly as it is.
Any theoretical model is only as good as its assumptions. So what Fearon offers is a model of political behavior among people whose decision-making actually conforms to the assumptions he makes. Nothing more and nothing less. And considering why we don’t have any empirical reason to think that this is how people systematically behave, I think the application is limited.
Don’t get me wrong though. I appreciate the fish in a barrel. Although not wanting to make assumptions, I remain open to being convinced otherwise.
There is already some great snarky replies from the game theorist camp (see Phil Arena on twitter), so let me try to constructively engage.
I think the disconnect here is over what it means to make an assumption. The role of assumptions in game theory and theory in general has been articulated much better than I can many times, but to take a crack: Assumptions are not things that we think are true, but simplifying premises we tentatively accept to see what conclusions result. This may sound wishy-washy, but it IS a necessary evil given 1) theorizing entails simplification, and 2) simplification entails making statements that are not universally true.
Assumptions are not just “statements that one makes to make the building of theoretical models easier”, they make theorizing possible.
If that sounds nuts, it might be better to respond with two questions. First, can you give an example of an assumption made in political science that is true? Second, can you give an example of theorizing (or research in general) which does not make assumptions?
Weber, Max. “‘Objectivity’ in Social Science and Social Policy.” In The Methodology of the Social Sciences, edited by Edward A. Shils and Henry A. Finch, 49–112. New York: The Free Press, 1949. Even though that’s a crappy translation — there’s a better one in the Whimster volume (https://www.amazon.com/Essential-Weber-Reader-Sam-Whimster/dp/0415244277) — it still covers the basics. To summarize: not all assumptions are subject to improvement given new information, because not all assumptions are hypotheses; some are ideal-typical. Oh, and one can’t do (social) science without ideal-types, argues Weber, because science is about explanation, not description.
I commented on this already, but it was via my iphone – technology. Anyway, there is a vast literature on this issue. In IR you might want to check out my Millennium piece one ‘Why Assuming the Can Opener Won’t work’ and Fred Chernoff’s responses (Fred is basically in the instrumentalist camp, as are many of the reflectivists, see PTJ above, so it’s no just a ‘rationalist vs the others’ issues). I’m yet to respond to Fred by the way, but I’ll have to at some point, if only because he’ll be upset if I don’t. :) Anyway, I differentiate between ‘abstractive’ assumptions (PTJs Ideal Types) and ‘entity’ assumptions; I think I’m the first (only?) to make this distinction but it is important. I also think PTJ and I have a different reading of Weber. So PTJ (I believe, could be wrong) reads Weber as suggesting that Ideal Types are basically made up by the analyst and bear no relationship to the object under study. I think this is a misreading of Weber and that in fact, Weber grounds his Ideal Types in concrete objects, meaning that although no ‘Ideal Types’ actually exist, they are grounded in the objects which are species of the type; and that’s what makes them useful. Without this we can’t make sense of their ultility. Patrick nicely raised the issue of explanation, but that’s a very deep ocean and I have no intention of getting my feet wet today. Suffice to say that the idea that ‘explanation’ is the goal (one of ) of science is very much a realist (scientific, not political) notion and sets itself against the more predictive, or instrumentalist versions. Worth noting, however, that prediction and explanation are not symmetrical; you can predict without being able to explain, and vice versa. And we don’t need an explanation in order to find things useful. If Patrick is moving into the terrain of explanation, I’ll be extremely pleased because it’s very difficult (some say impossible) to combine instrumentalism with explanation (which is why I think Fred is working on a book on explanation). Al interesting stuff. Welcome Brian (to realism that is).
This is one of those rare instances where I have something useful to contribute. All of the articles listed so far are quite good, but by far the most useful piece in parsing this issue for me is Paul MacDonald’s “Useful Fiction or Miracle Maker: The Competing Epistemological Foundations of Rational Choice Theory” APSR Vol. 97, No. 4 (Nov., 2003), pp. 551-565. The article does a great job outlining some of the inconsistencies in rational choice over the validity of assumptions as a result alternating and competing philosophies of science (empirical instrumentalism vs. scientific realism). Fantastic little article.
thanks eavanr…I remember reading it, but will go back and look again.
Brian,
I don’t think the divide as you’ve characterized it really exists today. Perhaps it did back in the 20th century but to dismiss the progress that the field has since made is a tad disingenuous. The imaginary “battle” I continue to see seems to stem from _assumptions_ about what “rationalists” are doing rather than bothering to take a look at what they are “actually” doing. For all the “we ought to be open and question everything!!!”, critics seem to remain remarkably close-minded when it comes to formal models–why?
It is wholly untrue that “rationalists” do not “touch” their assumptions. On the contrary, these assumptions are almost always under scrutiny; one basic exercise for a course in formal modeling is to break down an existing model, and consider what happens if one or more assumptions are relaxed. I’m not sure what assumptions you are referring to when you say they become presumptions – pray tell. The only assumption necessary in any formal model is that given several possible outcomes, individuals can rank their preferences over these outcomes. And these preferences must not cycle. Phil (Arena) has a blog that does great justice to clear up these misconceptions (I prefer to call them biases) and if you’re so inclined, I suggest you read the series of excellent posts he has on “rational choice” instead of making _assumptions_ about what “rationalism” entails.
Assumptions are inherent in every theory; it’d be well nigh impossible to study what we do otherwise. Reality is too complex not to simplify it in some way or another. The difference I continue to see between those you refer to as “rationalists” and others is that the former makes their assumptions explicit and clear rather than couching them in prose that too often obscures rather than clarify. Ironically, I think formal modelers are far more aware of their so-called biases than others, largely because they recognize that these are critical.
It never fails to amuse me that criticism about “rationlism”–which apparently means different things to different critics–almost always falls back on philosophy of science. While important, you cannot simply toss out words like epistemology and positivism to make a point. Why not talk about ontological assumptions, which we all make? I don’t ever see that. On that note, one reason I admire and read PTJ is his consistency of beliefs. His is a rare voice that actually gets into the “meat” instead of merely sprinkling a dash of “epistemological” salt to add flavor to an otherwise stale argument.
Your example of bargaining or crisis behavior harkens back to the 1990s. Fearon wrote that almost two decades ago–in 1995! Much has changed since and renders much of your criticism moot. The assumption of risk neutrality is almost always a (mathemetically) convenient one, and as you’ve mentioned (to your credit), is to help make analyses tractable. In most cases (though not all), relaxing this assumption does little to change the implications of the model. Models today are a lot richer and have moved beyond risk preferences. Accordingly, the work (read: math) required to work out the logical implications gets exponentially more complicated and forbidding but you cannot possibly fault these scholars for taking their assumptions seriously and seeing what happens if you drop them.
Also, bringing up risk preferences confirms that you are building your critique on work done in the 1990s, which suggests a heavy reliance on heretofore unstated _assumptions_ you’re making about this entire literature, no? Where’s the openness? You ask us to “Assume Nothing!” yet you assume a whole lot in this post.
I think the field benefits from the rich and myriad perspectives that scholars of different “orientations” (for lack of a better word) bring to bear on the complex nature of international interactions. And progress hinges on a willingness to engage constructively, not on a construction of artificial divides.