I think Meg McKean has done an excellent job reminding us of the fundamentals of rational choice, and especially that culturally influenced preferences can indeed be plugged into a rational choice model, so there need not be any trade-off between them.

"If we start out with a model of what we think people want, and cannot predict their behavior, rational choice analysis HELPS us because the failed predictions indicate that our initial assumptions were wrong (about the utilities preferred by the actors or about the actual options available or about the real payoffs that the individual actors see as consequences of those
options) and send us back to the drawing board."

This is right as far as the model goes, although it is also possible that the rationality assumption itself could be flawed.
As Daniel Kahneman has shown us, the choices we make often deviate from rational choice in predictable ways.
For example, although people are normally risk averse, they often become risk seeking when trying to avoid loss (but not when seeking gains). The difference can be illustrated with the claim made during this discussion that TEPCO executives were only thinking about maximizing company welfare during their tenure and not its long term welfare. While this could be true, it is not a necessary assumption: executives might have taken risks to avoid corporate losses, all the time believing that this is in the company's long-term interest. They could have concluded by realizing that they and their families would suffer long after they left the company if they did something clearly wrong or irresponsible while leading TEPCO.
This could also reflect institutional constraints or perhaps culture at some level, and of course these could be built into a rational choice model. The main problem with rational choice is not the model itself, but the facile assumptions it encourages us to fall into.

"So for decades we all thought Arabs didn't want democracy because they did diddly squat to get it. Now we know that "institutional constraints" can include monstrous dictators who have armies of hit men and a vast network of underground jails for dissenters, and this can silence a population that really wants democracy for forty years. what did it take for us to realize our conclusions were absolutely wrong? Arabs making different choices about their behavior, making it impossible to imagine any longer that they did not want democracy."

While this could be true to a significant extent, we certainly do not "know" this to be true. Nothing that has happened in the Arab world in the past two years tells us much about the preferences of populations in these countries in the past. Moreover, this doesn't even tell us much about the extent to which Arab populations "want" democracy today versus something else. Obviously some do, but overthrowing a dictatorship and wanting democracy are, when we turn off the instantaneous media and pause to think for a moment, can obviously be very different things. People can also want democracy as a means rather than as an
ends: for promoting economic development, getting a foreign power off your back, creating an Islamic state, overthrowing and possibly repressing a minority, etc.

Again, the problem with rational choice is not the model itself, but how easily we project certain values into the model.
The model offers a ready-made narrative that makes this look plausible, but that does not mean that it is.

Paul Midford
Norwegian University for Science and Technology