[Multivariate correspondence analysis of a visually screened EEG. Application to a study of psychotropic drugs...

Eighty-seven EEG recordings were obtained from 9 healthy volunteers, medical students who were part of a controlled, double-blind, randomized study. After a first control recording, they received orally, at 1 week interval, either 3 placebos or 3 compounds: yohimbine chlorhydrate (4 mg), propranolol (40 mg), a psychotropic compound in experimentation. The polygraphic recordings were obtained 2 or 6 h after placebos or active compounds. Four EEG channels were recorded simultaneously (T4-C4, T3-C3, P4-O2, P3-O1). The EEGs were sampled at 200 Hz, for 120 successive epochs, free of artifacts. The recordings of 5 min were then submitted to spectral analysis followed by data reduction providing characteristic spectral parameters. A descriptive questionnaire of EEG made of 100 items was filled out by two electroencephalographists for each of 87 recordings. The items, with dichotomic choices, described 4 EEG frequency bands: delta, theta, alpha, and fast frequencies (beta). For each frequency band the following was evaluated: frequency, amplitude, asymmetry, morphology, relative abundance, irregularity, lability of frequency, diffusion, dominant localization, EMG activity and superposition of rhythms. The number of periods of drowsiness in 5 min and fluctuations of rhythms were also ascertained in the same way. The items were coded on punch cards and submitted to a first analysis of correspondences. The 3 factorial planes were plotted with the projections of each item together with the centers of gravity of placebos and treatments. The first factor can be interpreted as a bipolar axis 'mental activation-diffused vigilance', the first pole being characterized by a 'beta cluster' and the second by alpha and theta superimposed clusters. The second factor can be also considered as a bipolar axis 'drowsiness -quiet wakefulness', the first pole being characterized by a 'theta cluster' and a larger 'delta cluster' containing a smaller 'alpha cluster' of labile alpha rhythm intermingled with slower rhythms. In respect to the barycentric centers of gravity, yohimbine and propranolol are opposed to the placebo center along the factor-two axis. This may be considered as an 'anti-sedation effect'. However, along the first factor axis, the propranolol center can be associated with a greater diffusion of vigilance than the yohimbine center.