David's comments about when rational choice is and isn't useful got me thinking about how to schematize good and bad occasions for using rational choice. I'll try to lay out the situations in which I find it most helpful. And it can be VERY helpful in combination with thoughtful cultural analysis. The two are not mutually exclusive at all.

First, rational choice NEVER has off-the-shelf explanations (have explanations, will find problem to explain). It would be absurd to imagine that any theory is arrogant enough to offer an explanation in advance of a problem!

Second, of course, if one has to infer preferences from behavior (this is guessing at cause from effect), one is creating tautology and doing bad social science.
When one MUST do this for lack of independent information about preferences, one must be prepared to amend the assumptions, the information, the predictions, the model as soon as better information arrives. But doing this can still be helpful because it forces one to be imaginative about an actor's preferences and the constraints s/he faces, and one may discover conflicts and difficulties in the situation just by trying to think through it. We may be doing that with Noda and his taxes.

Third, it is always harder to attempt making point predictions about a single individual actor rather than about a class of actors all facing similar situations.
The point prediction requires that one really have a lot of psychological insight about that particular individual.
Generalizing about a class of people (legislators in this body, merchants who sell food made with subsidized corn syrup, rent-seekers in the timber industry, farmers in an irrigation system) within which there is still room for individual exceptions is a lot safer.

Fourth, rational choice can be quite helpful for working through mysteries when you can be fairly certain about the preferences but find that people with those preferences do not seem to be taking action to pursue them (as compared to situations where you see behavior but do not know about the
preferences and have to invent them). I offer my
favorite example, which
pits good rational choice against bad cultural analysis. Development advisors flying in to Tanzania found two valleys, one with good tree cover and prosperous agriculture, and another lacking both. The snotty advisors told the people in the bad valley they should plant trees (prevent soil erosion, retain water, provide leaf litter as fertilizer, etc). The people nodded yes, we should plant trees on the hillsides around our fields. But they didn't. Next wave of advisors, same advice, same response. So these people said they wanted trees but wouldn't plant them.
Advisors concluded there was a cultural resistance to progress, laziness, lack of understanding of relationship between investment and results, etc. Lots of unflattering conclusions, these people aren't "ready" for modernization.
(And of course we know what David McClelland would have advised -- create focus groups and read Horatio Alger stories to these people to give them entrepreneurial spirit. Makes me vomit.) More thoughtful researchers arrived and actually began talking to the people more, asking them why, learning about the situation. well, it turns out they lived in a kleptocracy run by the landlords, and they didn't own the hillsides themselves, and anticipated that if they planted trees and cared for them to grow them to maturity (a multi-year investment of effort), the local elite would just harvest the trees and sell them for personal income. The investment required would be a donation of labor to people they already hated. They would not be able to capture the benefit of this effort.
These people were rational (and materialistic) egoists par excellence actually. They knew far better than the fly-in-fly-out advisors about progress, effort, investment, results, and probabilities. The remedy would lie not in tree-planting, but in institutional reforms they were too politically weak to accomplish on their own. Hence stasis in a sub-optimal
trap. What rational choice offered here was simply
the dictum that one
should not throw out the assumption that these people are rational and have good knowledge until you've tried everything else. Question other things first. Soak and poke. Above all, investigate rather than reaching premature conclusions.

For me, rational choice's biggest contributions are
two: 1) the collective action insight, which you cannot arrive at without assumptions of rationality, and which explains regrettably much in our lives; and
2) the reluctance to shut down an investigation by declaring that an actor simply must not be rational.
So I defend it as an approach. But note that far from being an alternative to sound empirical investigation and sensitive ethnnography, it requires those things.

For Noda, rational choice offers no quickie answers.
Instead, we have to keep considering alternatives, and at some point get enough information
from him to figure it out. The quick answer (this is
often what people
resort to on North Korea or Iran too) that the actors are just nuts, or nasty, or uninformed, or suicidal, is a surrender rather than an answer.

Best,

Meg McKean