Application of phase detection and averaging techniques in computer analysis of EEG records in the cat

Abstract

Consistency of phase patterns in EEG records from the hippocampal system and midbrain reticular formation have been examined in the course of behavioral training in five cats with implated electrodes. These studies have involved novel application of computing techniques, including continuous measurement of phase and amplitude characteristics of single wave trains by digital filtering techniques, the use of cross-spectral analyses with calculation of complex transfer functions, and averaging of records during repeated behavioral performances with calculation of coherence functions in cross-spectral examinations. It was found that rhythmicity appeared in the computed average at the rate of the dominant EEG frequency as T-maze performance reached a high level, and was related to a reduction in scatter in phase patterns at high performance levels. Reversal of behavioral cues was associated with an immediate brief increase in rhythmicity of the average, then a decline followed by a progressive reemergence of rhymicity at the end of retraining. Continuous phase-and-amplitude examinations during delayed response performance indicated a rhythmic phase modulation on wave trains with an apparently single frequency appearing during periods of discrimination. Comparison of probability bounds in cross-spectral analysis with use of a polar coordinate display have indicated major differences in phase relations between different hippocampal regions between correct responses in certain circumstances, with consistency in different examinations. These findings are discussed in relation to a stochastic model of the cerebral system, with reference to the possible role of the wave process in handling and storage of information.