The essay presents with some lenght Sperber's thesis that "symbolic thought is necessarily built after a previous rational treatment". Although in general terms Sperber's position is accepted, a critique is presented concerning three themes: the fírst deals with the defícient recognition it gives to Turner and Lévi-Strauss; the second, the principle that "all representation can be invoked by any stimulus in determined conditions"; third, the basic epistemological problem: "similarity" and "contiguity", contrary to what Sperber apparently holds, aren't empirical concepts.