Recently I’ve been reading up on the use of rational choice models for analysing political behaviour in international politics. Rational choice theory appears enticingly powerful for its representing and explaining the array of possible decisions and strategies available to actors. Despite the sorts of cognitive biases and errors which we now know lead decision-making often quite far from the assumptions of rational choice theory, we might nevertheless see it as capable of solving some of the most intractable problems and of rendering social wholes into analysable component parts. The ability to determine what a utility-maximising actor should do in a given situation or interaction permits us not only to make predictions about actor behaviour in situations where we might expect actors to conform adequately to rationality assumptions, it also tells us what we should do if we would like to conform to those assumptions – for example, in making economic or diplomatic choices. Not only that, but it offers the power to explain both through nomothetic generalisations deduced as hypotheses from the model and also through teleological statements about the intentions of actors, viz., that some action was the most efficient means of realising some intended outcome. As the rational choice theorist knows, the answer to one ancient question can be easily found so long as we assume that the other side of the road is the best place to be, and that the best transport option available to the chicken is to cross.
However, rational choice theory has also proved quite unpopular amongst sceptics who assert that rationality assumptions fail to accurately and adequately describe any real human being in any real human situation, and who question the value of many rational choice analyses on the grounds that they substitute obscure sophistry for truly rigorous and empirically grounded scholarship. Others may be less sceptical but still find that, pace Milton Friedman, while models may offer sound predictions despite being constructed out of many unrealistic assumptions, they prefer to develop scientific theories which do more than predict but also, ideally, describe how things really are – that carve reality at its joints. To these sceptics I offer two additional, if less common, uses for rational choice theory that often get missed in the debate.
1: ‘Weber, Wittgenstein, Winch’
While it is nothing new, for economists, to recognise that they must take into account a full and rich range of values and tastes as guiding decisions beyond the narrow goal of financial gain, we can not only infer decisions based upon preferences but preferences, qua norms, based upon decisions. Obviously this method must be conducted with considerable humility, given the infinite possible belief-value schemes  which could render some decision best, but informal or even formal methods of comparison, such as Bayesian networks, could allow a scientist to find the most probable background for understanding – for this is a hermeneutic method – why some actor behaved as they did. This method of studying social action owes much to Weber, though it bears notable similarities to Geertz’s influential interpretive methodology, and has received a more structural formulation in the Wittgenstein-inspired thesis argued by Peter Winch in his book ‘The Idea of a Social Science and its Relation to Philosophy’, in which he holds that social scientists are to explicate the rules, norms, or conventions according to which some action is appropriate or warranted. While this kind of methodology still suffers from many of the limitations of the rationality assumptions – limitations which might be transcended by a ‘logic of habit’ or what Weber called ‘traditional rationality’, for example – it nevertheless offers an appealing instrument to quivering math-phobes, godless quallies, or anyone else for whom the formality and predictive bent of game-theoretic modelling renders it less appealing of a method than one aimed at cultural exploration.
2: ‘All the fun is in the error’
Another use for ideal-typical models based upon rational choice assumptions is as a way of exploring the limitations of our prior theories while locating new areas for research. Here ideal-types, to spare you the Weber quotation, are designed as simulations of a simplified reality operating according to assumptions which we believe are adequately true. If we find that these models predict outcomes that diverge strongly from reality, we would not say that the model is falsified – how can you falsify a simulation? – but rather that it is incomplete as a representation of what is really happening, in significant ways. There is something in the error that we are failing to account for. Rather than abandon our prior theories, then, we instead look for the additional factors that produce the error.
 The belief that some combination of beliefs and desires explains why some rational actor did something is typically expressed (by philosophers, anyway) in what is often called the axiom of folk psychology: ‘if an actor desires X and believes that Y is the best way to achieve X, then, all things being equal, that actor will Y’.