The paper contrasts to completely different approaches on the relations of reason and nature: the one assumed in the context of the studies on 'bounded rationality' and the one proposed in McDowell's Mind and World. As a result of this contrast two different models of the relation of reason and nature appear: what may be called respectively a top-down and a botton-up model. Drawing on Kant's own proposals in his 'Schematism of Pure Reason' I will develop some criticism of McDowell's bottom up approach, defending the need of a normative non-conceptual level on which basis the conceptual would be developed. I will draw some connections between these proposal and some suggestions made by Peacocke, arguing finally that the resulting position would seem to be compatible with the bounded rationality approach.