L'optimisation est-elle un critère de rationalité individuelle ?

The contention was often made that the optimizing model of decision cannot by itself account for the whole of the pretheoretic concept of individual rationality. This paper raises a more drastic issue : are we sure that the optimizing model is one of rational decision making ? The first section analyzes suitable chosen examples of reasonable intransitivities. The second section appraises the assumptions that axiomatize the optimization criterion in the context of revealed preference theory. The third section shows that the optimization criterion involves the theorist in an infinite regress that may be " vicious ". Typical of the approach of this paper is the backwards and forwards motion between formal properties and normative viewpoints which can only be grasped informally. None of the arguments put forward here is a decisive one, but their conjunction seems to be impressive enough to cast a doubt about the tenet that optimization is a rationality criterion.