DISCRIMINATION OF BEHAVIORAL STATE IN CAT UTILIZING LONG-TERM EEG FREQUENCY ANALYSIS

A hybrid system is described which makes feasible a continuous, long-term analysis of the EEG. A broad-band frequency analyzer is interfaced with a digital voltmeter which in turn drives a typewriter and card-punch machine. Data are thus generated in a form directly compatible with the input devices of a large, electronic computer used to extract relevant information by multivariate statistical analysis. Spontaneous variations in the frequency spectrum of the EEG, as the result of changes in behavioral state, may be adequately represented as points on a two-dimensional surface. The location of a point corresponding to the spectrum during a given 1 min epoch allows an objective discrimination to be made between the states of slow sleep, fast sleep and alert. Minutes which are transitional between these three well-defined states can also be represented. This classification is in excellent agreement with independent behavioral assessment, is stable over long periods of time (1-2 months) and is consistent from cat to cat. The reliability and stability of this technique should provide a means of quantitative assessment of drug effects on the EEG over long periods of time. © 1969.