The knowledge of human knowledge claims a place of its own in economics. Beyond the walls of our discipline, spectacular progress is taking place in the field of empirical research into human knowledge -the so-called "cognitive sciences". In the light of such advances, the old and new classicals' axiom that nothing scientific can be said beyond the axioms of substantive rationality now looks very much like the protective belt of a degenerating programme. On the other hand, criticisms and alternative programmes will hardly be effective so long as their arguments are purely negative or are drawn from armchair introspection. In the present study I wish to outline a pattern of human knowledge emerging from cognitive research that may be called "constructivist", and to point out the restrictions it sets on economic analysis. It is also my argument that such a pattern is consistent with the present non- or post-Walrasian trends in economic theory, and that it may provide them with firmer cognitive foundations.