FinEcon wrote:It is my understanding that game theory does not provide an explanation of the sub optimal player, the irrational player or the aloof player.
In a one-shot game there is only one round and some of these games (ultimatum game) will not be played as predicted by economics in academic tests.
The issue is that while you (and the study authors) may think they are "one shot" games the players may not see it that way. They may see it as part of their continuing, ongoing interaction with other people. Your one shot game does not represent the last time they'll ever be involved in negotiating a settlement with another person, and in that larger context, punishing an unfair negotiating partner may be near optimal behavior.
As for misjudging probabilities and rules it's clear that the very mathematical game theory models are gross simplifications of reality. That doesn't necessarily mean they aren't useful ways of analyzing a situation, it just means that their results won't perfectly match real observations. If all they need, though, is the addition of some statistical noise, then they're pretty good models--and that would be the case if people are equally likely to under- or over- judge a probability, or misunderstand rules in ways that cancel out over large numbers.
Even if there are behavioral biases that cause people to systematically err in one direction, say always over-estimtating a probability, the simple model can still wind up being your starting point in whatever correct model you wind up coming up with. The academic way to discuss this is to talk about what percentage of the observed behavior is explained by the model--so maybe 80% of what you observe is explained by the simple model, and the 20% that varies from it can be explained by some 2nd level behavioral or other analysis.
No-one really doubts that the simple abstractions used in economic models are not quite like real people--the question really is whether they are close enough to what real people do that, despite some error, they overall do predict a substantial portion of observed behavior. If they do, then the simple models will only be discarded when some other model comes along that predicts even more of the behavior--even though, in the meantime, we know that the model we've got is less than 100%. There is always room for improvement in any science--the soft sciences generally have much more room to improve than the hard ones, but it's true across the board that all scientific models are ultimately over-simplifications--even the theory of gravity.
[url=http://www.efficientmarket.ca/]Canadian Mutual Fund, ETF, and Self-Directed RRSP Advice[/url]