Some Statistical Properties of the Visual Evoked Potential in Man and Their Application as a Criterion of Normality

The average visual evoked potential in man is found to be characterized by six major potential waves or deflections. A computer-aided visual inspection technique is used to extract and measure the 12 parameters corresponding to the amplitude and latency of the six peaks from the responses of 50 subjects. It is then demonstrated that each of these parameters is approximately individually Gaussian distributed and their means and variances are found. With this information, a simple statistical model for the normal visual evoked potential is constructed. It is next assumed that the same 12 parameters are jointly Gaussian distributed and the correlation matrix is computed. A principal components analysis is then applied to this matrix and it is found that the 12 correlated visual evoked potential parameters can be transformed into a reduced set of four uncorrelated components accounting for approximately 76 percent of the total variance of the original set. The four components are then rotated using a varimax technique and the correlation between these components and the original 12 parameters is found to reveal some interesting aspects of the visual evoked potential (VEP) pattern behavior.