In studying the brain, two levels of investigation emerge naturally. One of these concerns itself with properties of nerve cells, their numbers, patterns of firing, interconnexions, and so forth. The other considers the whole nervous system in what one may call 'macroscopic' terms. Thus it discusses 'stimulus', 'response', 'decision', etc. At this latter level, the nervous system operates with considerable unity. The individual nerve cells must therefore be linked in a well-integrated manner and the general nature of this integration has been recognized, especially by neurophysiologists such as Sherrington, to present a problem of central importance for our understanding of the brain. In previously published work, I have put forward a theory of how this unification of neural activity might be achieved and of a possible molecular biological basis of the necessary neural organization. In this talk I restrict myself to the first of these and thus give an account of what might be called the basic logic of the unification. I also indicate briefly how a simple hypothesis about the basis of memory would fit into such a theory.