Wednesday, September 7, 2011

Two Roads Diverging? Scientific Realism and Rational Choice

Hampsher-Monk and Hindmoor (2010; here, gated) argue that game theoretic models may be instrumentalist, structuralist, or realist. Whether a model is realist or not hinges on whether or not "interpretive" evidence, that is, evidence about how the actors in question think about an issue, is useful as evidence in favor of game theoretic models.

In instrumentalist models, the truth of assumptions is irrelevant because a model's utility is in its predictions. In structuralist models, actors are forced into their responses by external structures, so how actors think and process is irrelevant. In realist models, assumptions need to be true because models are both descriptive as well as explanatory.

It seems to me, however, that interpretive evidence should also be useful in structural models because a structural model may also be realist. That is, in arguing that structures determine behavior, it makes no claim about reasoning regarding behavior. In some sense, it's really a realist argument that says that reason doesn't matter for behavior, so we should be able to see some disjunction between reasoning and behavior across different actors. So interpretive evidence would be useful here.

It does strike me that they are correct, however, about instrumental models and interpretive evidence. It also suggests that a realist can legitimately argue that some factors aren't relevant as part of their explanations, but this strikes me as different than assuming that something isn't relevant a priori. Is this correct?

But this hinges on a bigger question: So, can a scientific realist make simplifying assumptions (and know that he or she is making assumptions) and still be a realist? Or does being a realist mean you don't have the ability (luxury?) of making assumptions?

No comments:

Post a Comment