The main motivation behind reductionist theories is to explain how different descriptions, using very different concepts (those, for example, of physics, chemistry and biology) can apply truly to the same hunk of matter. The classic forms of reduction in the philosophy of science are translation reduction (Carnap) and nomological reduction (Nagel). Both these are generally judged to be untenable. It is in rejecting these that both Davidson and Fodor regard their approaches as non-reductive. This tends to lead to talk of supervenience as a replacement for reduction. This doctrine seems to leave it mysterious how or why one level should supervene on another. There is a third of reduction, corresponding to the idea that a higher level need be ‘nothing but’ the constituents of the lower level, which I call the a priori sufficiency ( or logical adequacy) of the base. This accommodates Fodor (and Davidson) but leaves us in need of an account of the predicates/properties of the higher levels. I claim that these cannot be taken in a strictly realist fashion, but are ‘ways of seeing’ the base. As this would apply to the higher level science of psychology itself, we are faced with a choice between a vicious regress and dualism. The rest of Part II constitutes a defense of this claim.