Two stage rationality under risk: Experimental results and perspectives

Assuming that choices are made on grounds of unobservable heuristics and/or sets of routines, rational choices are nevertheless represented by use of choice functionals, the maximum of which on a given possibility set designates the individual’s choice. This paper argues that, in descriptive models, this does not imply at all a ‘strong’ (as opposed to bounded) type of individual rationality unless the functional is in some sense “universal”. Making use of such a criterion, the paper shows that it is possible to design an experimental protocol which allows distinguishing between these two modes of rational behavior. It shows further that decisions under risk appear as two-level rational ones, which is an experimental support to some of Herbert Simon’s views. Finally, it is shown that adaptiveness of rational behavior may be limited, in simple cases like choosing between static lotteries, to coefficients’ changes in a rank-dependent model.