2
There is a large body of work in the international relations literature that uses formal
game-theoretical methods to model strategic behavior based on this assumption. What is the role
of this literature in IR theory, and by what standards should it be evaluated? This essay begins
by observing that this literature is becoming increasingly bifurcated between the use of such
modeling to generate predictive hypotheses, which are then evaluated by empirical, usually
quantitative, analysis, and the use of such modeling to contribute to a generalized deductive
model of international politics based on the assumptions of rational choice theory. The first use
of modeling is based on an instrumentalist-empiricist philosophy of social science, the second on
a scientific-realist philosophy.
1
A third use of these methods, as heuristic models, based on a
Weberian philosophy of social science, is becoming increasingly marginalized.
The argument here is that this third use of formal game-theoretic modeling remains an
eminently useful tool in the study of international relations, that risks being displaced by the uses
of modeling noted above. Formal modeling can be useful both as a heuristic device, and
corollarily as a prescriptive device. This essay discusses these uses of game theory, and the
standards by which they can reasonably be evaluated. It argues that the use of evaluative
standards designed for other uses of formal modeling are inappropriate to the use of this
technique for heuristic purposes. In particular, it discusses situations in which statistical testing
of models is either irrelevant to heuristic models, or actively misleading, and looks at the dangers
of the mathematical complexity so often valued by rational choice theorists to heuristic models.
In short, it argues for an evaluative standard of usefulness, rather than one of precision, for the
heuristic use of game theory.
1
MacDonald 2003.
Convention