> I think they actually behave rationally, but all according to different fact-bases.
Rationality, in the economic sense, involves (among other things) actors acting with accurate and complete information about all consequences of their market choices. Actors acting with "different fact-bases" are necessarily acting with inaccurate and/or incomplete information (because if it was both complete and accurate, it wouldn't differ between actors), and therefore, economically speaking, irrationally.
>Rationality, in the economic sense, involves (among other things) actors acting with accurate and complete information about all consequences of their market choices.
No, it assumes that agents use all available information optimally. Complete information is not a requirement.
> No, it assumes that agents use all available information optimally.
No, it is defined to include perfect information; actors using all available information optimally within the limits of their cognitive ability and considering, also, the opportunity cost of devoting resources to informationa evaluation and decision-making is a weaker alternative ("bounded rationality") to pure rationality.
> Complete" and "perfect" information are not the same thing.
Perfect information, in the rational choice theory sense, is both complete in the sense that it covers all economic consequences that impact the experienced utilities of the decision maker positively or negatively out to an infinite time horizon, and free of errors over that entire scope, which is why I used "accurate and complete" as a less-formal equivalent of "perfect" upthread.
> In games like poker, players have perfect information, but not complete information.
In poker, players do not have perfect information in the rational choice theory sense (or the game theory sense, which is generally stated differently, but I think is actually equivalent.) Poker is, in fact, one of the canonical examples of a game featuring imperfect information.
This is impossible, because unless the information is also complete then the utility functions of opponents (and hence future payoffs) are unknown by the decision maker at the time of the decision.
>which is why I used "accurate and complete" as a less-formal equivalent of "perfect" upthread.
I suppose if you believe that the two are equivalent, there's not much point in arguing.
Well, yes, rational choice theory is based on premises which are known to be impossible, but which are analytically convenient and which produce models which are (at least, argued to be) useful approximations of real world behavior.
Rationality, in the economic sense, involves (among other things) actors acting with accurate and complete information about all consequences of their market choices. Actors acting with "different fact-bases" are necessarily acting with inaccurate and/or incomplete information (because if it was both complete and accurate, it wouldn't differ between actors), and therefore, economically speaking, irrationally.