After reading Youssef Cohen’s Radicals, Reformers, and Reactionaries for my summer course, I was reminded of what we mean when we assume rationality in rational choice models. We assume rationality in decision making, that individuals will attempt to maximize their utility given their preferences and the constraints that exist. There are a few important components in rational choice models, however, that we do not make any assumptions and judgments about.
First, in terms of preferences, we merely require that they are transitive. We do not judge their quality. If the actors prefer to buy a bright orange couch for their home over an expensive, designer made couch, as long as those preferences are transitive, we do not make any judgment of the quality of their choices. Further, in terms of beliefs, we do not have to assume that there are rational and have some basis in reality. If someone thinks that praying will make them more likely to get an orange couch instead of going to the store and buying one, this type of belief can be taken into account in a rational choice model.
The point is: someone can make a rational choice based on seemingly bizarre preferences and irrational beliefs as long as they believe the result of that choice will result in utility maximization.
Not judging preferences and allowing unrealistic beliefs to be tied into the model allows rational choice models to explain a large variety of behavior, from voting for a bill in a legislature to suicide bombing. Rational choice models may not be able to adequately explain where bizarre preferences come from, or why unrealistic beliefs are formed, but it can explain why individuals make the choices they do when both of those conditions are met.
Explain, perhaps, but not necessarily predict. The problem comes in determining ex ante what potentially “bizarre” preferences are and how they might be weighed by actors.
One of the biggest difficulties in formal modeling lies in the effort to provide quantification to things that are fundamentally undetectable and immeasureable. The doctrine of “revealed preferences” helps with explanation, but is risky with regards to prediction.