Appeared in Neuropsychobiology 12: 173-178 (1984)

Dedication

Dieter Bente was a pioneer in developing the science of pharmaco-electroencephalography. His studies with Turan Itil in Erlangen laid the groundwork for the classifications of psychotropic drugs using EEG methods. His laboratory of clinical electrophysiology in Berlin provided much of our knowledge of the role of vigilance in the relations between drug effects and psychological performance tasks. His death in 1983 is a loss to the science, and this history and prospect is dedicated to his memory.

To study the relationships between brain function and the abnormal behaviour of the mentally ill, Hans Berger developed the electroencephalograph in 1929. He recorded rhythmic oscillations in electric potentials from electrodes on the scalp of both normal volunteers and psychiatric patients. He reported rapid changes with attention, eye closure and eye opening, calculations, and after drugs. Recordings in patients with schizophrenia and melancholia were found similar to those of normal subjects, but 'abnormal' records were found in patients acutely ill with dementia paralytica, epilepsy, and brain damage.

In his third paper, Berger [1931] reported that the amplitude of the primary or alpha waves increased several times after the subcutaneous administration of 30 mg cocaine, at a time when the pupils were dilated, pulse rate increased, and psychic processes enhanced. After the combination of 20 mg morphine and 1 mg scopolamine in an agitated psychiatric patient, he reported a distinct loss of EEG alpha activity with desynchronization of the record. With chloroform anaesthesia, EEG amplitudes progressively decreased as narcosis deepened, and then increased again as narcosis waned. He concluded that,

'... the alpha-waves indeed originate from cerebral cortex and that they represent concomitant phenomena of the material processes in the cerebral cortex which are connected with processes of consciousness'. ........... 'If centrally active poisons which elicit clear alterations in the course of psychical events produce such evident findings in the EEG, then one also ought to expect that those diseases of the brain which are associated with disturbances of psychophysical processes should produce similar changes' [Gloor, 1969, p. 116].

,... when using these drugs it became apparent that there was mutual correspondence between the changes in the alpha-waves of the EEG and the effects on mental processes; whenever a significant action of the drug upon the latter failed to occur, visible alterations of the alpha-waves of the EEG were also lacking' [Gloor, 1969, p. 210].

In the excitement induced by scopolamine, he found that the frequency of EEG beta-waves increased; and when behavioural sedation occurred, as after large doses, the mean frequency of the alpha-waves slowed. Different anaesthetics elicited different EEG patterns. After oral coffee and parenteral caffeine, the EEG was desynchronised. In schizophrenic patients, the EEG records were normal before insulin coma, only to show a marked enhancement in very slow EEG waves during coma. Upon awakening, the subject's records were again normal. From these observations, he defined an association between changes in EEG variables and in behaviour [Berger, 1938]:

'... the cerebral cortex functions as a whole as far as mental processes are concerned and that precisely the beta-waves, and not the alpha-waves, are the concomitant phenomena of mental activity! I believe that the alpha-waves of the EEG of man are concomitant phenomena of the automatic physiological cortical processes and that certain beta-waves with a length of 11-24 milliseconds (40-90 Hz) represent material concomitants of the processes of consciousness' [Gloor, 1969, p. 319].

With the confirmation by Adrian and Matthews [1934] that rhythmic potentials could indeed be recorded from the human scalp, more scientists looked to the EEG to study brain functions. Descriptions of EEG changes with ageing, psychological state, and mental performance, as well as after psychotropic drugs and psychiatric treatments became commonplace. Electroencephalograms were recorded with enthusiasm, and publication lists expanded rapidly, so that a bibliography of citations alone, published by Brazier in 1950, included more than 2,700 citations in 163 pages, with an index of more than 1, 100 authors!

The electroencephalograph became a diagnostic tool for studies of brain lesions in strokes, cerebral tumours, and epilepsy. Some authors sought relationships between EEG patterns and mental diseases, but these efforts were largely unfruitful. Indeed, their failure led to a gradual disuse of the EEG in psychiatric hospitals. These efforts were premature, however. Shagass [1956] showed that there was a relationship between mental state and the EEG response to a set dose of intravenous amobarbital. Later, he found relationships with the averaged evoked response [Shagass, 1972]. Recent studies of hemispheric asymmetry suggest relationships between EEG patterns and the diagnoses of depression and schizophrenia. There is also much interest in sleep EEG parameters and clinical diagnosis, particularly in the diagnosis of major depressive disorders. Berger's faith may yet be fulfilled as the methods of behavioural assessment improve, allowing better correlations to be made.

The discovery that reserpine and chlorpromazine were useful in psychiatric treatment encouraged an unparalleled enthusiasm in clinical psychiatry in the 1950s. Therapists could now help their patients, safely and effectively, without isolation rooms, wet-packs, camisoles, insulin coma, leucotomy, or intensive electroshock. The proper usage of the drugs and the prevention of complications became matters of intense clinical and research interest. Reserpine and chlorpromazine were soon followed by many other antipsychotic, anxiolytic, and antidepressant substances, thereby launching our present era of pharmacopsychiatry.

The EEG effects of chlorpromazine were reported concurrently by Verdeaux and Marty [1954] in Paris, and by Bente and Itil [1954] in Erlangen. Chlorpromazine increased slow waves and decreased fast activity in the EEG. These observations were quickly confirmed, and at the CINP meeting in Rome in 1958 the figures were found to be so similar that different reporters could use one another's slides with little change in their reports.

By the end of the 1950s, various classifications of the chemicals were proposed. Some were based on chemical structure; some on clinical characteristics; some on pharmacologic and physiologic effects; and a few on biochemical tests. The EEG patterns provided another useful classification, which at first seemed to be drug-specific, but later was seen as class-related [Bente, 1961; Itil, 19611. Later elaborations identified EEG patterns of anxiolytics, antipsychotics, antidepressants, and even hallucinogens and deliriants [Fink, 1968; Fink and Itil, 1968a, b].

In 1961, a symposium on EEG and behaviour was organised at the Third World Congress of Psychiatry meeting in Montreal. Participants included Bente and Itil (Erlangen), Verdeaux (Paris), Hajnsek (Zagreb), Lechner (Graz), Schneider (Colmar), Ulett (St. Louis), and Fink (New York) with Herbert Jasper (Montreal) as discussant. Three themes emerged - that patients who failed to show EEG changes showed poor clinical responses; that psychotropic drugs affected the EEG in characteristic ways in responsive patients; and that quantitative methods, particularly frequency analysis, were useful in defining the subtle effects of these agents [Fink, 1963a].

The development of a behaviour-related classification of substances rooted in physiology implied a close association between brain electric potentials and human behaviour, encouraging us to think that perhaps we could predict drug effects from their EEG patterns, to select the proper drug treatment for each psychiatric patient, and even to understand why some patients failed to improve. But pharmacologic studies of the effects of anticholinergic drugs provided an impediment to such ideas. After such substances, when the EEG exhibited the salient characteristics of a 'sleep EEG' with high-voltage burst activity, animals appeared restless with running motor movements, suggesting a 'dissociation' between the EEG pattern (similar to that seen in sleep) and the behaviour (running) [Wikler, 1952, 1954]. Similar observations were soon reported in rabbits, cats, and monkeys, and images of a dissociation between EEG and behaviour dominated the thinking of electrophysiologists and pharmacologists. If there was a 'dissociation' between EEG and behaviour, it was logical to assume that EEG studies had little to offer neuroscientists.

This discrepancy in the findings in animals and in man led to a dialogue between pharmacologists and electrophysiologists, culminating at the 1966 meeting of the CINP in Washington [Bradley and Fink, 1968]. Animals given large doses of anticholinergic drugs develop a delirium, in which the EEG shows both high-voltage EEG slowing and excessive amounts of very fast frequencies. Under these conditions, the animals are restless, exhibiting excessive motor movements, and their sensorium is clouded. They are neither able to carry out normal commands nor to make their usual responses to sensory cues. Their EEG is easily distinguished from that of normal sleep, and the 'dissociation' reported by pharmacologists was seen to result from their limited observations - limiting behaviour to motor functions only, and limiting the EEG to visual measures of the superficial similarity between the EEG of normal sleep to that occurring in delirium. This argument coloured much of the interest in the effects of drugs on EEG and behaviour in the 1960s, and it is only in the past decade that pharmacologists have again sought relationships between EEG changes after drugs and changes in behaviour in animals, particularly in primates.

The science of pharmaco-EEG depends on a methodology which encourages a quantitative approach to the measurement of EEG and behaviour. The earliest EEG records were scored by hand, using calipers and rulers to measure each wave. By 1943 Grey Walter, under the demands of war, developed an analog frequency analyser to quantify the brain changes resulting from head trauma. Post-war, Ulett used a frequency analyser to study the changes resulting from electroconvulsive therapy and the interactions with atropine. Shagass [1956] used quantitative methods to study the sedation threshold, and when we obtained our analyser in 1958, we applied it to the study of the effects of chlorpromazine and imipramine and, thereafter, to a wide range of other psychotropic drugs [Fink, 1963b].

Drohocki [1956] introduced a simple amplitude integrator for EEG analysis which was used to evaluate drug effects by Goldstein and Beck [1965]. This technique was inherently more stable than frequency analysis, but the

method was limited to a single characteristic of the EEG, a characteristic that was of little utility in evaluating drug effects.

In October, 1960, at the dedication of the UCLA Brain Research Institute, a symposium on 'Computer Techniques in EEG Analysis' demonstrated the power of digital computers for the extensive 'number-crunching' which EEG analysis required [Brazier, 1961]. Presentations on averaging techniques, pattern recognition, and frequency analysis highlighted the many ways in which digital computer methods could be used to quantify brain electrical activity. Burch [1959] had developed the mathematics of period analysis for the EEG, so that it was possible for Itil, Shapiro, and Ulett in my laboratory in St. Louis to develop the programs and the methods for a digital computer analysis of an EEG signal using an IBM 1820, and later, an IBM 1800.

Other algorithms for EEG analysis were studied, including interval analysis, Hjorth parameters, sensory evoked potentials, and the sleep EEG, yet harmonic analysis remains our dominant data reduction method. At first, period analysis was the most efficient program available, but as the speed of power spectral density analysis programs improved, especially with the development of the Fast Fourier transform, and as digital computers increased their speed of calculation and enlarged their storage capacity, power spectral density analysis programs became as efficient as period analysis and are now the dominant program in the field.

These advances in data reduction would not have been possible were it not for important electronic developments which increased the ability to record and store EEG signals. Sensitive and stable amplifiers, FM-modulated tape recording, and sensitive frequency filters made it possible to record and replay signals reliably. But as the quality of our recording and analytic methods improved, we became less secure in the data of our studies in psychiatric patients. Such subjects were more variable from session to session; they were less dependable for control of vigilance; and their exposure to psychotropic drugs led to unpredictable withdrawal phenomena and unequal degrees of tolerance to drug effects which interfered with our analyses for days, weeks, and even months after their last dosing. More and more we studied the effects of substances in normal volunteers, gradually replacing our interest in the effects of psychotropic drugs in patients. Studies in volunteers led to techniques to monitor vigilance during the experiments, and to define variables, such as pre-dosing EEG characteristics, sex, and handedness of the subjects, which affected the stability of our measures.

While these methods were used to develop EEG profiles and to estimate the persistence of drug effects on drug withdrawal and the interaction of drugs, most interest was generated by the prediction of the clinical activity of putative psychoactive substances from their EEG activity. Predictions of the clinical activity of fenfluramine, cyclazocine, and doxepin during their preclinical studies set the stage for the prediction by Itil et al. [1972] of the antidepressant activity of mianserin. This compound had failed to affect the usual pharmacologic screens as a conventional antidepressant substance, so that the suggestion of antidepressant activity was met with scepticism. Clinical trials clearly demonstrated antidepressant activity. Following this report, the use of pharmaco-EEG methods in the development of new substances became widespread and in 1980, the Federal Health Office in Germany convened a special panel of experts to set guidelines for these assessments and their reporting [Stille et al., 1982].

In prediction studies, threshold doses of substances are used. The sensitivity of the EEG to drug effects led to comparisons of the relative potency of compounds and of different formulations of compounds, to suggest guidelines for clinical dosing patterns. These methods have been applied to studies of benzodiazepines, antidepressants, and antipsychotic agents; and, most interestingly, to determine the duration of action of different formulations of substances in the search for those which were safe and long-acting.

In the assessment of substances not ordinarily seen as psychotropic, pharmaco-EEG analysis has found CNS effects that previously had only been suggested by reports of side effects. These include studies of hormones, antihistamines, beta-adrenergic antagonists used as aiitihypertensive agents, aspirin, and phenytoin.

It is in these applications that pharmaco-EEG has achieved its greatest usage and success. Yet, our methods remain expensive and time-consuming, and subject to the risks of human experimentation. There is need for an animal model to replace these human trials, but no satisfactory model has been forthcoming. In part, this failure lies in our aim - to predict and measure the behavioural effects of new substances in the mentally ill. As we have been unable to find an animal species with a brain pharmacology which is so similar to that of man that new substances would have the same metabolism and toxicity, we have also been hampered by our inability to define CNS differences among our patients which could be related to our preferred usage of psychotropic drugs. An animal model predictive of a human psychiatric illness remains a hope of the future; for the present, our tedious, expensive, and riskful methods in man remain the best available predictors of the behavioural activity of psychoactive substances.

The science of pharmaco-EEG is still young. The awareness of the EEG effects of psychotropic substances clearly goes back to Berger's report of 1931. There were many reports of drug effects, but for clinical psychiatry, the first interesting application was probably Shagass' use of the beta EEG response with amobarbital to separate patients into different psychopathic classes. An innovative application was also suggested by Roth [1951] when he used the slow-wave response to pentothal as a predictor of a patient's improvement with electroconvulsive therapy, finding that those who showed a large degree of EEG delta slowing after the first treatment achieved better outcomes than those who failed to show such slowing.

Demonstrations of EEG effects for chlorpromazine and reserpine, and the separation of their effects from those of imipramine and chlordiazepoxide, provided the impetus to develop a systematic science. Bente, Itil, and Künkel in Germany; Borenstein, Cahn, Schneider, and Verdeaux in France; Burch, Denber, Fink, Goldstein, Merlis, Monroe, Pfeiffer, Shagass, and Ulett in the United States; Hajnsek in Yugoslavia; Roubicek, Matousek, Volavka, and Vojtechovsky in Czechoslovakia, and Rinaldi in Italy are a few of the early investigators who contributed to the development of this science. The World Congress of Psychiatry meeting in 1961 and the CINP sessions in 1966 were important milestones in the development. The biennial symposia in Europe, organised by Schenk [1973], Dolce and Künkel (1974) [published in 1975], Matejcek and Schenk [1975], Matejcek (1978) [published in 1979], and Herrmann (1980) [published in 19821 made us aware of the growth of the discipline. The review volume edited by Itil [1974], and the monographs by Itil [1964], Matousek [1967], Saletu [1976], Herrmann [1980], and Matejcek [1981] showed that the science had come of age. During meetings in Berlin, organised by Stille and Herrmann to develop guidelines for pharmaco-EEG analysis, IPEG was established. The first formal meeting was organised by Künkel in 1982 in Hanover. This was followed by the designation of the Karger-published journal Neuropsychobiology as the official journal in 1983, and the second meeting of IPEG in Vienna organised by Petsche in 1984.

There is much work to be done if the promise of these early years is to be fulfilled. The methods of pharmaco-EEG are rooted in defined technical and statistical procedures which allow the careful assessment of threshold effects of drugs. These must be used with precision if the predictions are to be supported and believed. Few centers for the assessment of new drugs exist, more are needed, and their establishment by government, by industry, and by universities is to be encouraged.

The theory of association of EEG and behaviour is couched in broad terms. We have not yet defined the characteristics of the EEG which are associated with important aspects of mood and cognition. Berger first drew our attention to the role of beta activity in thought processes, and of slow waves in vigilance. Wikler [1954] focused our attention on additional relationships. The definitions of brain change and of behaviour are now more refined, and it may be the time to again experimentally examine the relationships between changes in frequency bands and specific aspects of behaviour.

While much progress has been made in the quantification of the EEG, the measurement of behavioural change remains less precise. Qualitative rating scales are the principal tool, even though many attempts have been made to use psycholinguistic measures, reaction time and other performance tests, and memory, perceptual, and motor tests.

There is a need to expand the initial aims of Berger to define the relations between psychopathology and the electrical characteristics of the brain. Some work has been done with evoked potentials and the contingent negative variation, but these techniques, based on averaging of repeated signals, are subject to adaptation and are inherently insensitive to the subtle changes in brain function which accompany mood and cognition. There is an important clue in the studies of Shagass and Roth, which demonstrated that while patients with psychopathology exhibit resting EEG patterns which are indistinguishable from those of normal subjects, they differ in their response to chemical stimulation. Psychoactive drugs are sensitive probes for stimulation, particularly the use of those substances which have predictable effects on abnormal behaviour. That is what Shagass attempted with the sedation threshold, and to some extent he was successful. But neither amobarbital nor pentothal would now be considered specific psychoactive agents. Perhaps trials based on the effects of haloperidol or fluphenazine for schizophrenic subjects; or imipramine, mianserin, or phenelzine for depressed patients would be more successful in identifying subpopulations of the mentally ill. The ongoing studies in Berlin of maprotiline and chlorimipramine are in this class. And, the clinical challenges of sodium lactate infusions in patients with anxiety attacks warrant EEG correlations.

An important observation at the 1961 meeting in Montreal was the sensitivity of the EEG to change in those patients who had a good clinical prognosis, and its failure to change in those who had a poor prognosis. Therapy-resistant psychoses and depressions remain important clinical issues in research. Quantitative methods to define a minimal and a moderate EEG change should be useful in predicting the clinical results of treatments. If such correlations were established, it would do much to provide better guidelines to the duration and dosing schedules of treatments of patients, and for some treatments, guidelines for adequate maintenance therapies.

There are times when patients under psychiatric treatment show a dramatic change in behaviour, occasionally a worsening. This may result from toxicity, and serial EEG records can reflect this development and its resolution. Such records are ordinarily 'read' by visual, non-quantitative means; under these conditions, shifts in frequency spectrum are easily missed. The techniques of pharmaco-EEG are particularly useful in evaluating the course of treatment of patients with lithium, lithium and antipsychotic agents, and convulsive therapy.

There has been much progress in our discipline. From a descriptive art without a theoretic foundation, it has developed into a quantitative science. The relations of brain function and human interpersonal behaviour remain an exciting challenge for neuroscience. Grey Walter [1956], in one of his more exuberant moods, invited us to join in this search: 'The history of science is one long series of tournaments, of champions and challengers, of pledges and surrenders and rewards beyond compare in chivalry. The conquests and captures of knights errant of natural history, the feats of backroom squires and laboratory heroes, are also legendary. But they are less ephemeral than those of romance, because the scene of the scientist adventurers is an expanding one and their purpose endless and unchanging. The Round Table of Science, moreover, has room for all who can find their way to it.'

Adrian, E.D.; Matthews, B.H.C.: The Berger rhythm. Potential changes from the occipital lobes in man. Brain 57.. 355-385 (1934).

Bente, D.: Electroencephalographische Gesichtspunkte zur Klassifikation neuro- und thymoleptischer Pharmaka. Medna exp. 5: 337-346 (1961).

Bente, D.; Itil, T.: Zur Wirkung des Phenothiazinkörpers Megaphen auf das menschliche Hirnstrombild. Arzneimittel-Forsch. 4: 418-423 (1954).

Berger, H.: Über das Elektroenkephalogramm des Menschen. Arch. Psychiat. NervKrankh. 87..527-570 (1929).

Berger, H.: Über das Elektroenkephalogramm des Menschen. Dritte Mitteilung. Arch. Psychiat. NervKrankh. 94: 16-60 (193 1).

Berger, H.: Über das Elektroenkephalogramm des Menschen. Achte Mitteilung. Arch. Psychiat. NervKrankh. 101: 452-469 (1933).

Berger, H.: Über das Elektroenkephalogramm des Menschen. Vierzehnte Mitteilung. Arch. Psychiat. NervKrank. 106: 577-584 (1938).

Bradley, P.; Fink, M. (eds.): Anticholinergic drugs and brain functions in animals and man. Prog. Brain Res. 28 (1968).

Brazier, M.A.B.: Bibliography of electroencephalography. Int. Fed. Electroenceph. clin. Neurophysiol. (1950).

Brazier, M.A.B. (ed.): Computer techniques in EEG analysis. Electroenceph. clin. Neurophysiol., suppl. 20 (1961).

Burch, N.: Automatic analysis of the electroencephalogram: a review and classification of systems. Electroenceph. clin. Neurophysiol. 11..827-834 (1959).

Dolce, G.; Künkel, H. (eds.): CEAN computerised EEG analysis (Fischer, Stuttgart 1975)

Drohocki, Z.: An electronic integrator for the automatic measurement of average tension in the EEG. Electroenceph. clin. Neurophysiol. 8: 706-707 (1956).

Fink, M.: EEG and human psychopharmacology. Abstr. Proc. Symp. 3rd World Congr. Psychiatry, Montreal 1961. Electroenceph. clin. Neurophysiol. 15: 133-137 (1963a).

Fink, M.: Quantitative electroencephalography in human psychopharmacology. II. Drug patterns; in Glaser, EEG and behaviour, pp. 177-179 (Basic Books, New York 1963b).

Fink, M.: EEG classification of psychoactive compounds in man: review and theory of behavioural associations; in Efron, Cole, Levine, Wittenborn, Psychopharmacology: a review of progress 1957-1967, pp. 497-507 (U.S.G.P.O., Washington 1968).

Fink, M.; Itil, T.: Neurophysiology of the phantastica: EEG and behavioural relations in man; in Efron, Cole, Levine, Wittenborn, Psychopharmacology: a review of progress 1957-1967, pp. 123 I1239 (U.S.G.P.O., Washington 1968a).

Fink, M.; Itil, T.: EEG and human psychopharmacology. IV. Clinical antidepressants; in Efron, Cole, Levine, Wittenborn, Psychopharmacology: a review of progress 1957-1967, pp. 671-682 (U.S.G.P.O., Washington 1968b).

Gloor, P. (trans. & ed.): Hans Berger on the electroencephalogram of man. Electroenceph. clin. Neurophysiol., suppl. 28 (1969) (350 pp.).

Goldstein, L.; Beck, R.A.: Amplitude analysis of the electroencephalogram. Int. Rev. Neurobiol. 8: 265~312 (1965).

Herrmann, W.M.: Entwicklung und kritische Bewertung eines objektiven Verfahrens zur elektroenzephalographischen Klassifizierung von Psychopharmaka. Universitätsklinikum Charlottenburg der Freien Universität Berlin (1 9 80).

Herrmann, W.M.: Electroencephalography in drug research (Fischer, Stuttgart 1982).

Itil, T.M.: Elektroencephalographische Befunde zur Klassifikation neuro- und thymoleptischer Medikamente. Medna exp. 5: 347363 (1961).

Itil, T.M.: Elektroencephalographische Studien bei Psychosen und psychotropen Medikamenten (Ahmet Sait Matbaasi, Istanbul 1964).

Itil, T.M. (ed.): Psychotropic drugs and the human EEG. Mod. Probl. Pharmacopsychiat., vol. 8 (Karger, Basel 1974).

Itil, T.; Polvan, N.; Hsu, W.: Clinical and EEG effects of GB-94, a tetracyclic antidepressant: EEG model in the discovery of a new psychotropic drug. Curr. ther. Res. 14: 395-413 (1972).

Matejcek, M. (ed.): Latest advances in pharmacoelectroencephalography. Pharmakopsychiatr. Neuro-Psychopharmakol. 12: 1245 (1979).

Matejcek, M.: EEG und Vigilanz. Institut fair Philosophie, P5dagogik und Psychologie, Stuttgart (1981).

Matejcek, M.; Schenk, G.K.: Quantitative analysis of the EEG (AEG-Telefunken, Konstanz 1975).

Matousek, M.: Automatic analysis in clinical electroencephalography. Psychiatric Research Institute, Prague (1967).

Roth, M.: Changes in the EEG under barbiturate anaesthesia produced by electroconvulsive treatment and their significance for the theory of ECT action. Electroenceph. clin. Neurophysiol. 3: 261-280 (1951).

Saletu, B.: Psychopharmaka, Gehirntätigkeit und Schlaf (Karger, Basel 1976).

Schenk, G.K. (ed.): Die Quantifizierung des Elektroencephalogramms (AEG-Telefunken, Konstanz 1973).

Shagass, C.: Sedation threshold. A neurophysiological tool for psychosomatic research. Psychosom. Med. 18: 410-419 (1956). Shagass, C.: Evoked brain potential in psychiatry (Plenum Press, New York 1972).

Stille, G.; Herrmann, W.; Bente, D.; Fink, M.; Itil, T.; Koella, W.P.; Kubicki, S.; Künkel, H.; Kugler, J.; Matejcek, M.; Petsche, H.: Guidelines for pharmaco-EEG studies in man. Pharmacopsychiatry 15: 107-108 (1982).

Verdeaux, G.; Marty, R.: Action sur l’électroencephalogramme de substances pharmacodynamiques d'intérêt clinique. Revue neurol. 91: 405-427 (1954).

Walter, W.G.: An automatic low frequency analyser. Elec. Eng. 16: 9-13 (1943).

Walter, W.G.: The curve of the snowflake (Norton, New York 1956).

Wikler, A.: Pharmacologic dissociation of behaviour and EEG 'sleep patterns' in dogs: morphine, n-allylnormorphine and atropine. Proc. Soc. exp. Biol. Med. 79: 261-264 (1952).

Wikler, A.: Clinical and electroencephalographic studies on the effects of mescaline, n-allylnormorpine and morphine in man. J. nerv. ment. Dis. 120.. 157-175 (1954).