INPUT‐OUTPUT RELATIONS *

Recent developments in averaging techniques have provided several interesting possibilities for electrophysiological investigation of sensory disorders. One common, conceptually simple approach has been to compare average responses evoked by some fixed repeated stimulus and then to decide, presumably on the basis of predetermined criteria, that a particular average is either normal or not normal. In our experience, if reasonable statistical precautions are observed in the design and conduct of such experiments, the procedure does not seem to provide as reliable an objective test for sensory deficits as has been suggested by other^.^^^ In general, we have found this method to be diagnostically less accurate than the more conventional behavioral techniques. A more fundamental reason for our rejecting this approach is that the model of nervous system organization it assumes is too static and too restricted to yield much information about patterns of sensory deficits associated with lesions of the central nervous system. This model appears to have sensory receptors connected by fixed pathways to sharply localized responsive centers in the brain. The expectation is that if any portion of such a system is damaged, sensation will be reduced and a correlated change will be found in the electrical output of responsive neural elements. Now it is true that in clinical situations one commonly recognizes, and to some extent localizes structural changes associated with sensory disturbances by noting an elevation of threshold or perhaps a decrease in magnitudes of sensation. However, this type of deficit is generally embedded in a much more extensive framework of disorganized behavior. What is of most interest to us, is that one often encounters a loss of differentiated sensory and motor function in association with brain lesions. Stimuli that normally lead to independent percepts will interfere with one another, and movements that are normally discrete will be accompanied by synkinetic activity. These phenomena may not often provide useful diagnostic information but they do illustrate a less rigid type of neural structure that may be profitably studied. One summary statement of the principle involved is as follows: A reduced nervous system wilI be reorganized in such a way that the relative independence of its component mechanisms will be altered, On the basis of these considerations we felt that a reasonable way to utilize evoked scalp potentials was to investigate their relation to sensory addition. The tentative operational concept we employed for this application was that the more closely the average response evoked by two