Situations within the central nervous system are elusive when we try to render them in anything more abstract than approximate verbal descriptions. A description in terms of ionic movements through cell membranes, for example, might be useful when dealing with information transmission in sense cells but becomes unwieldy if applied to situations involving many neurons of the brain. Attempts at applying the formalism of a logical calculus, inspired by the success of binary algebra in dealing with switching networks, seem to impose a restriction on the working of neurons (e.g. by assuming discrete time) which is not justified by electrophysiology. Occasionally an insight may be gained by reasoning in terms of channel capacity, amount of information and redundancy, but this language again seems more successful in the analysis of peripheral sensory events than in the study of central processes.