On the correlated nature of evoked brain activity: biophysical and statistical considerations.

The contribution of EP research to the mind-brain problem is considered. The view is promoted that the scalp EP represents a macroscopic rescaling of patterns of activity in cortical columns, which are held by many workers to be the functional units of the neo-cortex. Implicit in re-scaling operations though is a loss of degrees of freedom. However, in reducing the dimensionality of the EP to a comparable order to that of the molar phenomena of behavior, it is argued that this loss may be regarded constructively as potentially facilitating the successful forging of mind-brain correlations. Various techniques are reviewed for extracting the dimensions of scalp activity, including those based on electric field theory which are argued to have high face validity. The multivariate approach is also surveyed. Principal components analysis, in particular, is evaluated using simulated data. The limitations of the method are pointed up. Caution is advised in the design of a PCA, and in the interpretation of factors. The use of discriminant analysis in combination with MANOVA is suggested as a more appropriate multivariate strategy.