Remembering the lost neuroscience of pharmaco-EEG

To the electrophysiologist accustomed, in his animal experiments, to probing with his electrode into the nervous system and even into the nerve cell itself, it remains a matter of astonishment that so many of the brain’s secrets escape across the wall of the skull to electrodes fixed to the scalp of man. That they indeed do so is testimony to the fact that the brain’s electrical activity is a most sensitive indicator of its function. In this fact lies the attraction of electroencephalography for its devotees.

Mary A. B. Brazier (1)

I am known for my defense of electroconvulsive therapy (ECT) during the decades when professional leaders and the laity denounced and legislated against its use. The worldwide resurrection of ECT in the past decade is a heartening endorsement of those efforts, a story well told by professional historians (2, 3). Researchers are marked by their enthusiasms, and in my life I have followed leads in clinical psychopharmacology, opiates and their antagonists, the physiology and tolerance to cannabis, and for two decades, the psychopathology of catatonia and melancholia. In this essay I relate my 30-year enthusiasm for pharmaco-EEG – the science of the effects of psychoactive substances on brain electrical activity (4–8).

After public schooling in New York City and graduation in medicine from the New York University College of Medicine in 1945, I followed a conventional course of internship, military service as an Army physician, residencies in neurology and psychiatry, to certification in neurology (1952), psychoanalysis (1953) and psychiatry (1954). In my fifth year of residency training at Hillside Hospital (a community psychiatric hospital in Long Island, New York) I was introduced to insulin coma (ICT) and ECT and established the hospital’s EEG service to study these treatments. When I introduced chlorpromazine and imipramine to the hospital in the mid-1950s, we were equipped to determine their EEG profiles as well as their clinical effects.

Quantifying EEG rhythms was a challenge. It was fortuitous that the digital computer revolution was applied to EEG in 1960 (9) and for three decades my associates and I applied quantitative methods to study the effects of drugs on brain functions in patients and volunteers. This essay describes that experience, mirroring the rise and the fall of pharmaco-EEG, a neuroscience that was discarded well before its full potential was achieved.

EEG and the ECT process

The hours of encouraging the recollections of the hospitalized mentally ill in psychotherapy sessions, each of 50 min by the clock, stimulated little interest. How often could one listen to strange beliefs and petty thoughts that were repeated again and again, hour after hour, like a well worn phonograph record? Summarizing these stories for a rabbinical teacher-analyst brought little insight and less benefit to the patients. One teacher, Sidney Tarachow proposed a study of a psychoanalytic hypothesis that absence of a parent (by death or divorce) during childhood determined the form of expression of an adult neurosis, whether obsessive (one parent) or hysteric (two parents). A review of the HH records found no support for this fancy (10). A detailed case report about the panic attacks and paranoia of a homosexual man is another marker of this psychoanalytic period. The patient was helped little but the enthusiastic interpretations of the association of paranoia and homosexuality was a cause célèbre in the HH grand rounds schedule (11).

By contrast, my patients’ aberrant moods and psychotic thoughts were quickly relieved with ECT, allowing their return to home and work. What occasioned the rapid change in behavior? Brain functions were surely involved, more so than the psychodynamics of the unconscious mind. Electroencephalography was a new science, with less than a quarter century of experience; it promised an objective measure of brain functions.

With a fellowship from the National Foundation for Infantile Paralysis I trained in EEG at Mount Sinai Hospital with Hans Strauss, the author of a leading textbook (12). I established the EEG facility with the support of the Dazian Foundation for the equipment and the hospital supported the technician. We quickly confirmed reports that the EEG changed as patients improved with ECT and ICT (13) But could we relate the EEG changes to clinical behavior?

A 1954 National Institute of Mental Health research grant (M-927) supported the EEG studies of ECT. Within a few years I received additional support for EEG studies of the newly introduced psychoactive drugs.

EEG during an ECT course

The central event in each treatment is the induction of a grand mal seizure that is clearly expressed by extensive changes in the scalp recorded EEG. Seizure events are stereotyped so that we are now able to examine the recording of the seizure and decide whether the induction was effective or not. With repeated treatments the inter-seizure recordings change. Frequencies slow, amplitudes increase, and sharp spikes and slow wave bursts and runs appear. Figure 1 shows the patterns with electrical induced seizures, and Fig. 2 shows parallel recordings with seizures induced using the inhalant flurothyl (Indoklon). The sequences are similar but the EEG reflects differences that were explored clinically. Since effective treatments could be induced chemically as well as electrically, we concluded that nothing about the electricity is specific for the treatment benefits. As electric inductions are much easier to administer, flurothyl was quickly phased out (14) [This experience cautioned me to doubt the recent enthusiasm for TMS, VNS and DBS as replacements for ECT; none elicit seizures more readily than the present method of electric induction and the induction of a seizure is essential to clinical benefits (15)].

Figure 1.  (a) Serial interseizure EEG records of a 60-year-old depressed man exhibiting progressive slowing of frequencies, occasional spikes and increased amplitudes during course of ECT. The rhythms return almost to pretreatment levels after 2 weeks, with greater amounts of alpha frequencies. (b) Serial interseizure EEG records of 44-year-old depressed woman treated with flurothyl (Indoklon) induced seizures. The changes in EEG are parallel to patient in 1a, with greater degree of slowing and higher amplitudes.

Figure 2.  (a) Effect of intravenous LSD-25 on EEG and the reversal by chlorpromazine in a psychotic male patient, age 29. Note frequency analyzer trace. (b) Frequency analysis of the same records. By varying doses of LSD and CPZ we assayed the relative potencies of the agents.

We systematically recorded the inter-seizure EEG 24–30 h after a treatment once each week, documenting the changes during the treatment course. The rhythms return to normal within a month after the last seizure.

The conventional practice in examining EEG records was to rapidly turn pages containing eight lines of EEG from different regions of the head and to estimate symmetry, wiggle shapes and amplitude changes. These descriptions were little more than Rorschach impressions of low reliability. To quantify the changes, we hand-measured the width (frequency) and the height (amplitude) of each wave in six artifact-free 10-s epochs using a ruler and calipers. This procedure was tedious and at best, bracketed degrees of change as ‘high’, ‘intermediate’ and ‘low’. Independent clinical assessments rated behavior as ‘much improved’, ‘improved’ and ‘unimproved’. Even with this simple quantification, we were able to conclude that the greater the slowing of EEG rhythms, the greater the increase in amplitudes and the earlier their appearance, the greater the induced behavioral change. Slowing of EEG rhythms was necessary for clinical improvement in ECT (16).

But not all the patients whose rhythms slowed improved clinically. The physiologic changes in the EEG were necessary but not sufficient to assure recovery. What determined who improved and who did not? The patients were of many psychopathologies, deemed unsuitable for psychotherapy or after they had failed months of such treatment. Their diagnoses (DSM-I) varied widely. The labels were not helpful in predicting who would improve and those who would not, although the older patients were more responsive than the younger. Patients with depressive and manic moods, psychosis and catatonia all remitted. We needed a hypothesis to test.

The denial hypothesis

Enthusiasm for psychodynamic thinking was at its peak among American psychiatrists in the 1950s. Edwin Weinstein, a psychoanalyst and neurologist, following the notions of the time, disdained clinical diagnosis and explained the benefits of ECT by the psychological defense of ‘denial of illness’. (17) Psychiatric symptoms resulted from continuing subconscious irritation of memories of childhood’s traumatic events. The memories are out of immediate awareness but dissipate when ‘understanding’ comes with free association of thoughts and discussions of dream reports with a skilled guide. In examining patients brain-damaged by tumor, stroke or trauma, Weinstein found such patients were relieved of their anxieties by the psychologic defense of denial, a defense that was encouraged when consciousness was partially clouded by an injection of amobarbital. Could ‘denial’ explain the improvement seen in patients during a course of ECT? Was the EEG slowing evidence of an obtunded consciousness that enhanced denial and explained the benefit of the treatment?

We examined the patients before and after ECT, recording their speech and measuring the extent of denial in the content. We expected denial language to increase with clinical recovery. Although the scores increased with treatments in some patients, we could not relate the denial defense to recovery (16). The denial hypothesis failed.

These writings present a child-like image, a religious fervor for psychological explanations of mental illness that pervaded the professions and society. We were so innocent and so naïve to accept such simple explanations of psychiatric illnesses!

LSD, ECT and chlorpromazine

By 1953 LSD had become a popular research tool, exciting interest in psychiatry and also among a hippie populace the hailed the experience as a path to nirvana. We administered LSD to our patients late in the ECT course, when the EEG was filled with slow waves, and were surprised by the immediate potency of LSD to inhibit the EEG effects of ECT (18).

Incidentally, we offered our psychiatry residents LSD experiences under conditions of EEG recording. Almost all volunteered and except for one paranoid response in which the subject refused his lunch as ‘poisoned’, all ‘enjoyed’ their experience and encouraged others to volunteer. In my personal LSD experience my EEG rhythms and heart rate changed characteristically, but I failed the nirvana of imagery and insight promised by hippie enthusiasts.

Chlorpromazine

Our next interest was in chlorpromazine (CPZ), elaborating both its EEG and its clinical effects. Henry Brill, the commissioner for mental health in New York, had launched studies of chlorpromazine in the state’s research centers. In1954 Herman Denber, Anthony Sainz, Sidney Merlis and Nathan Kline, newly appointed research directors at state psychiatric hospitals, enthusiastically reported reductions in psychosis rating scale scores, numbers of broken windows, fire-setting and harm to the staff with its use.

I offered CPZ to psychotic patients referred for ICT. Within a few weeks, stories of the reduction in agitation, excitement and delusions were whispered in the halls and then shouted in the dining halls. Nurses enthusiastically recommended patients for the new treatment. Our first experiences were not without difficulty, however, as three of the first seven patients developed jaundice, a finding that was experienced at a few other centers. The sponsors suggested that a contamination from the batch made in South America was the culprit; we accepted that conclusion as jaundice disappeared.

We began dosing at 50 mg and increased to 1200 mg/day. At this dose we effectively reduced psychosis and motor excitement in 80% of our patients. Rigidity, posturing and tremors occurred in some patients, marking the upper limit of dosing. We settled on procyclidine (Kemadrin) as an effective prophylactic for motor rigidity.

Chlorpromazine slowed the EEG frequencies by increasing the theta (4–7 Hz) frequencies and sharply decreasing the beta (13–25 Hz) frequencies. Seizure burst activity was induced at high dosages. The changes were easily differentiated from those of ECT, LSD and amobarbital. The interactions of these psychoactive agents were clearly seen in the EEG changes. Figure 2a shows the effect of LSD and the blockade of this effect by CPZ; Fig. 2b shows the actual recording and the superimposed frequency analyzer trace.

Imipramine

The EEG profile of imipramine (IMI) and its effect on the interseizure EEG differed from that of CPZ and LSD. IMI increased both the percent time of the theta and the high beta frequencies and inhibited the appearance of the alpha (8–12 Hz) frequencies. In patients with post-ECT EEG slowing, IMI blocked the slow waves and enhanced the fast waves, an effect that mimicked the effects of the anticholinergic agents atropine and diethazine. The clinical effects of IMI also differed from CPZ.

I presented the findings with CPZ and IMI at the 1958 CINP meeting in Rome on the same program with Turan Itil, then from Erlangen Germany (Fig. 3). We reported the same effects and we were excited to realize that we could exchange our slides to present each other’s findings. The science of pharmaco-EEG was launched that day and our association was active for more than 30 years (19, 20).

Figure 3.  Turan Itil, CINP 1961. Itil developed the EEG classification of psychotropic drugs by EEG criteria and mentored Sevket Akpinar (Ankara), Werner Herrmann (Berlin), Masami Saito (Kobe) and Bernd Saletu (Vienna) – each went on to establish independent pharmaco-EEG laboratories.

The EEG and clinical patterns of IMI mimicked the effects of the anticholinergic agents diethazine and atropine. Dry mouth and faster heart rates were additional consistent findings. We concluded IMI was anticholinergic in the CNS. At a meeting on the pharmacology of imipramine in Montreal in 1959 this interpretation was challenged by pharmacologists whose preclinical studies did not evince any anticholinergic activity (21). In time the anticholinergic effects of IMI was supported; indeed, these effects are the principal basis for its clinical discard.

IMI withdrawal syndrome

In our IMI studies we discovered its withdrawal syndrome. In a RCT of CPZ-IMI-PLO (see below), dosing stopped suddenly at the end of 6 weeks of treatment. Within 48 h, the IMI patients complained of malaise, nausea, vomiting, anorexia, and diffuse aches and pains. They were afebrile. The syndrome was relieved with IMI dosing, an early example of withdrawal in TCA agents (22).

Clinical psychopharmacology studies

Hillside Hospital was a unique setting for clinical and EEG research. The hospital’s leadership flaunted the banner of psychodynamic psychotherapy, proudly offering daily individual and frequent group therapy sessions for each patient. An orthodox accredited psychoanalyst supervised each resident. In the training hierarchy, young, attractive and intelligent patients were prized as learning opportunities while psychotic, aggressive and the elderly were shunted to the ECT service. When chlorpromazine was introduced in the fall of 1954, it was considered an experimental ‘biological’ treatment, of little interest to the trainees or their teachers, allowing the members of the Department of Experimental Psychiatry to control the experimental treatments. This attitude encouraged unique clinical and experimental trials.

Reserpine

A 1953 New York Times report of experiments with Rauwolfia alkaloids in India stimulated my interest and that of Nathan Kline (23). I was having little success in treating a severely psychotic young woman with ICT and after I described the reports and the note that Ciba was producing a purified extract to her industrialist father with connections in Europe, he returned with a box of 25 ampoules, 5 mg each, of reserpine. Alas, injections neither helped his daughter nor reduced anxiety in our patients but they did elicit dystonia and other side-effects to discourage further clinical use (24).

Chlorpromazine-insulin coma RCT

The rapid efficacy of CPZ stimulated our random controlled trial of patients referred for ICT to either 50 comas or 3 months of daily oral dosing with CPZ. Dosing of both treatments were optimized by a research clinician. After experience with 60 patients we reported equivalent efficacy in the discharge improvement ratings and the incidence and severity of complications. Seizures were induced in 15% of the ICT and 9% of the CPZ treatment groups. Prolonged coma occurred in 10% of the ICT group. We concluded that CPZ was safer, easier to administer, at much less cost with similar outcomes to ICT and that ‘… neither treatment altered the basic schizophrenic process, nor is there any evidence that there is a greater specificity of either form of therapy for schizophrenic illnesses’ (25). The HH Medical Board closed the ICT facility in 1959.

Years later I was asked to consult for the film A Beautiful Mind, the story of the treatment with ICT of the 1994 Nobel Economics prize winner John Nash. Reviewing the ICT literature led me to re-assess its mode of action. Since 20% of patients developed grand mal seizures, it is parsimonious to consider that the induction of seizures is the essential therapeutic element, and not any inherent characteristic of insulin or coma (26). ICT is an inefficient method of seizure induction. At the same time, I mentored the Harvard College student Deborah Doroshow in her history of science thesis. She described the impact of ICT as encouraging more humane environments in psychiatric facilities (27)

Chlorpromazine, imipramine, placebo RCT

By 1958 we were puzzled for whom the different new agents might be useful. Donald Klein had joined our research team and at one point Max Hamilton visited us. He believed he could prescribe these treatments on clinical criteria alone, and challenged our doubts and inability to use psychiatric diagnosis for treatment selection. After he had seen our patients, we agreed that the DSM diagnoses were unreliable. Was there a better way?

We planned an RCT of patients referred for medication regardless of their clinical diagnosis. Patients were randomly assigned to 6-week courses of either CPZ (with procyclidine), IMI, or placebo. Dosing was according to fixed schedules up to 1200 mg/day for CPZ and 300 mg/day for IMI. In a sample of 150 patients we confirmed the antipsychotic benefits of CPZ and the antidepressant benefits of IMI. We recorded an antidepressant benefit for CPZ (28). Phobias in adolescents were relieved by IMI while psychotic adolescents became more aggressive (29, 30). We identified treatment response as a useful pharmacologic dissection of clinical states, a finding that was used by Richard Abrams and Michael Taylor (31–34) and by Donald Klein (35) to verify clinical diagnoses. Decades later, Taylor and I used treatment response to validate our arguments for the independent classification of catatonia (36, 37) and of melancholia (38–40).

Our HH research group included five psychologists and we carried out extensive neuropsychologic tests. Social attitude (California F Scale), figure-ground embedded figures, Rorschach, tachistoscopic presentation of ambiguous figures, psycholinguistic measures of speech samples, and simultaneous tactile and auditory tests were explored, finding different effects for CPZ and IMI, that also differed from similar tests in patients treated with ECT (16, 41). Although CPZ and IMI had similar chemical structures, they were distinguishable by their effects in EEG, neuropsychology and behavior.

Controversies

Seeking other agents that might influence the EEG rhythms after ECT, we turned to anticholinergic drugs. We examined atropine and experimental anticholinergic deliriants being investigated in Parkinsonism (42, 43). Diethazine was one such experimental agent introduced by Herman Denber. Before ECT, diethazine increased EEG amplitudes and both slow-wave and fast-wave activity, signs that accompanied the induced delirium. The patterns differed from those induced by LSD. (In later studies we separately classified LSD as a hallucinogen, diethazine and other anticholinergic agents as deliriants.) In post-ECT patients who were no longer depressed, the high-voltage EEG slow-waves induced by the treatment were quickly replaced with low voltage fast frequencies. Simultaneously the patients complained of un-ease, sadness, guilt and paranoia. This dramatic switch in behavior lasted for one to 3 h and as post-ECT slowing returned, so did the relief in behavior (Fig. 4).

Figure 4.  EEG slow-wave activity induced by ECT is inhibited by injection of the anticholinergic agent diethazine. The same effects were found with atropine and scopolamine, the anti-parkinson agent procyclidine, and experimental hallucinogens of the JB-series. These observations encouraged the cholinergic hypothesis of the action of ECT.

The EEG-behavior association–dissociation controversy

Comparisons of human and animal physiology are the bedrock of medical science. But examinations of psychoactive substances in animals poorly predict their effects in man. The neurotransmitter effects of psychoactive agents, widely invoked as the basis for their psychoactive effects, are universally measured in animals, but the findings are not verified in human studies. Discrepancies between human and animal trials plagued the pharmaco-EEG studies in the 1960s when pharmacologists, unable to find associations between EEG measures and behavior for known psychoactive agents, claimed that the EEG effects were dissociated from the behavioral effects. In studies of anticholinergic drugs in dogs Abraham Wikler reported ‘sleep EEG’ records with high voltage burst activity when the animals were restless with running motor movements (44, 45). The same findings were reported in rabbits, cats and monkeys. Yet, human trials of the same agents found a remarkable association of the EEG changes with human behavior. An explanation appeared in a dialog between pharmacologists and electrophysiologists at a 1966 meeting of the CINP in Washington (46).

Anticholinergic drugs induce deliria in animals and man. The EEG shows high voltage slowing and very fast frequencies, and in both animals and man the sensorium is clouded and motor movements become purposeless. The subjects are not ‘asleep’ and their EEG is distinguished from that of normal sleep by the preponderance of fast frequencies and lack of patterned sleep stages. Animals are neither able to carry out normal commands nor to make their usual responses to sensory cues, while man does respond, although often with errors. Quantitative EEG methods had not been applied in the animal studies. The dissociation of EEG and behavior reported by pharmacologists resulted from their limited range of observations – limiting measures of behavior 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. Their observations were also flawed by their disregard of species specificity, in assuming the equation that pharmacologic observations in animals are predictive of the effects in humans. They are not.

Pharmaco-EEG studies

Drug effects by EEG criteria

We lacked adequate clinical criteria for selection of treatments for our patients. Nor did we have useful methods to measure the effects of the treatments on behavior or physiology. Salivary and pupillary measures, the blood pressure response to Mecholyl and epinephrine (Funkenstein Test), time to induced nystagmus after intravenous amobarbital (Shagass sedation threshold), hand writing analyses and reaction time tests for motor function, and an array of neuropsychological tests were all studied to determine differences in drug effects and as predictors for treatment selection and outcome verification. An even greater array of animal trial measures filled a burgeoning literature. Turan Itil and I contributed to this cacophony by classifying psychoactive agents by the differences in their EEG patterns in man (47–50).

In 1961, participants in a symposium on EEG and behavior at the Third World Congress of Psychiatry in Montreal asked: Was there a predictable relationship between the changes in EEG patterns and behaviors altered by psychoactive drugs? Three themes emerged. Patients who failed to show their characteristic EEG changes showed poor clinical responses despite seemingly adequate doses of medication; psychotropic drugs affected the EEG in characteristic ways in responsive patients; and frequency analysis accurately reflected the subtle effects of these agents (51). The human EEG reflected changes in brain chemistry and physiology that were relevant to changes in mood and thought, the basis for the hypothesis of the association of EEG and behavior in man.

At first we thought drug patterns were specific to each compound. With better quantitative methods we separated antipsychotic, antidepressant, anxiolytic, psychostimulant, hallucinogenic and deliriant drug classes by their EEG criteria. These analyses, first done in patients and later in normal adult volunteers, predicted whether the agents were clinically psychoactive, their clinical targets and effective dosing schedules. This model became the basis for international collaborations and biennial meetings of the International Pharmaco-EEG Group (IPEG) in which the investigators shared their methods and experiences (Fig. 5).

Figure 5.  (a) Members attending the Symposium ‘Computerized EEG Analysis’ held in Kronberg/Taunus, Germany April 8–10, 1974. (b) Identifying names of participants.

The pharmaco-EEG profiles of almost all new psychoactive entities were studied in many laboratories throughout the world, using patients for the early assessments and normal volunteers later. We assigned compounds to drug classes, predicted their clinical application, and suggested initial dosing schedules. Studies of doxepin, mianserin, flutroline and 6-azamianserin are examples of our experience.

Doxepin.  Preclinical pharmacologic data suggested that doxepin (Sinequan) would have anxiolytic properties and it was recommended to the clinic to compete with meprobamate and barbiturates. It was a poor anxiolytic, however, and marketing efforts were desultory. At a Pfizer-sponsored investigators’ meeting Herman Denber, Turan Itil and I examined available human EEG recordings. We found the patterns similar to those of TCA antidepressants and not to the known anxiolytics. We challenged the managers to undertake clinical trials in depressed patients. They did, found doxepin to be an active antidepressant, and embarked on marketing under the name Sinequan that was successful for many years. Our EEG study of doxepin confirmed its antidepressant profile (52).

Mianserin.  By 1970, Organon pharmacologists supported quantitative EEG methods in phase-2 clinical assessments of their new psychotropic drugs. Both Itil in St. Louis and I in New York studied different putative compounds made by their chemists. Itil found the EEG patterns of GB-94 (mianserin) to be similar to those of amitriptyline, predicted its antidepressant potential, and recommended clinical trials (53).

Organon pharmacologists scoffed at his prediction. They found GB-94 an active serotonin and histamine antagonist and predicted anti-asthmatic and anti-migraine activity. But Jack Vossenaar, director of medical research, gambled on Itil’s prediction and supported clinical trials. These trials quickly demonstrated mianserin to be an active clinical antidepressant. Its effective dosage was 1/5 that of known tricyclic antidepressant drugs, with less cardiotoxicity, making it valuable in the clinic (54). It was successfully marketed as an antidepressant in Europe under the name Tolvon until replaced by newer entities.

6-azamianserin (mirtazapine).  Mirtazapine, chemically related to mianserin, is also a racemic mixture. In preclinical studies, the dextro-enantiomer was found to be active and the leavo-not. We examined the EEG profiles of both enantiomers and recorded equal EEG changes that were most similar to the mianserin profile (55). Clinical trials for each enantiomer found both to be ‘clinically effective’ although neither differed from placebo at the doses tested. The racemic mixture was marketed as an antidepressant Remeron in the1990s.

Flutroline.  Pharmacologic studies in the dog reported that a single 1-mg dose of flutroline inhibited the vomiting induced by apomorphine for as long as 1 week. Extrapolated to man, pharmacologists enthused that flutroline would be an ideal antipsychotic, requiring a single oral dose each week. Alas, our clinical trials failed to elicit an antipsychotic effect, even at multiple dosing schedules. Nor did the EEG measures show any measurable change. The preclinical prediction of small doses being effective for days or weeks was untenable and studies of the drug ended (56).

Extended studies

Over the three decades of our activity, we determined the EEG profiles of many agents – of the sedative drugs triflubazam, bromazepam, brontizalam; putative antihistaminics, psychostimulants and phenytoin; and assorted peptides. In a study of acetylsalicylic acid (Aspirin) single doses less than 3.0 g were inactive; those higher than 3.6 g elicited EEG and behavior changes characteristic of soporifics (57). Phenytoin patterns mimicked those of antidepressant drugs (58).

The effects of opioids and their antagonists were cataloged and their interactions described (59). The EEG, physiology and behavior effects of cannabis produced at government farms in Mississippi were compared to those of Greek produced hashish (60). Itil expanded the list of identified drugs to hormones and cognitive enhancers. Of particular clinical and theoretic interest were the antidepressant effects of a synthetic male hormone and the anxiolytic effects of an anti-male hormone (61). The widespread use of pharmaco-EEG methods led the Federal Health Office in Germany to convene a special panel of experts in 1979 to set guidelines for these assessments and their reporting (62).

EEG quantification

A principal interest in our pharmaco-EEG studies was how best to measure the induced effects. Electroencephalographers commonly scan multi-channel recordings on printout pages, each page showing 10 s of the ongoing rhythms. Recordings are usually 20 min in length offering 180–200 pages to be visually scanned and interpreted. Pages are quickly turned for a kaleidoscope image of waveforms (bursts, runs, spikes, symmetry), estimates of dominant EEG frequencies and amplitudes, and focal differences or asymmetries. Eye and head movement artifacts are de-selected at the whim of the reviewer. Such visual estimates are neither reliable nor quantitative. In my first attempts at quantification of the EEG in ECT I measured each wave using a ruler and calipers for the record of one channel of 60 s of artifact free recording and coded the degree of change. But such work was tedious, made imprecise by fatigue and boredom.

In 1957, I read of an electronic analog frequency analyzer that could replace hand measurements, built by George Ulett in St. Louis on designs made by Grey Walter of the UK Burden Neurological Institute. Walter had designed the analyzer to measure the severity of head injuries of British soldiers during the war. Ulett and his engineer Robert Loeffel agreed to build such a device for us. With NIMH funding, this device arrived at HH in 1959. Although unstable, requiring delicate nursing and calibration each day, for a time it yielded more objective measures of drug effects (See trace in Fig. 2a).

My friendship with Ulett continued and when he was asked by the Governor of Missouri to re-organize the Missouri state mental hospital system, he obtained a newly developed building on the grounds of the St. Louis State Hospital for a research institute. Over the Christmas holidays of 1961 he visited my family in New York, seducing me by the opportunity to start a new research center, with free rein in appointment, an affiliation with Washington University, and a good state budget. George was a master story-teller and skilled magician, captivating my children with sleight-of-hand tricks. We arrived in St. Louis in July 1962 and over the next year equipped the building, appointed scientists, developed research protocols and by early 1963 we were well on the way with both laboratory and clinical studies. Much support came from Jonathan O. Cole of the NIMH Psychopharmacology Center. He supported our programs in clinical psychopharmacology and encouraged our interest in EEG quantification. The first appointee was the Librarian Nina Matheson (who went on to head the Welch Library of Johns Hopkins University), followed quickly by Turan Itil from Germany and Sam Gershon from Australia. The WU Computer Center sent us Donald M. Shapiro who developed the digital computer programs that became the centerpiece of pharmaco-EEG.

The first digital computer analysis of an EEG sample was demonstrated by MIT scientists at a meeting at UCLA in 1960 (9). I attended this inauguration of the UCLA Brain Research Institute and this report encouraged me to ask the computer scientists at Washington University to help us apply these new methods to our studies. We first used an IBM 1710 digital computer (with card reader, card punch for data entry) in 1964 and soon had Fortran computer programs for amplitude, period and power spectrum analyses. The first data analyses were limited to minutes of recording, but by the 1970s, continuous quantification of epochs became feasible using the more sophisticated IBM 1800 system. The science of pharmaco-EEG is founded on this quantitative technology (Fig. 6).

Figure 6.  EEG recording and IBM-1800 analysis system established at New York Medical College in 1967 and moved to Stony Brook in 1973. EEG analysis system (a), and computer analysis system of recorded tapes (b). Peter Irwin was my associate and laboratory director from 1969 to 1983.

My pharmaco-EEG studies end

By the 1970s we had examined a long list of psychoactive compounds in a well-designed quantitative methodology based on trials in patients and in normal volunteers. A catalog of EEG and drug correlations showed the predictive merits of the system. But our methods were expensive and time-consuming and increasingly met ethical hurdles.

Studies in patients became difficult. Patients were admitted to hospitals after having taken many medications whose effects persisted for days and weeks after their prescription ended. Hospital stays were shortened, allowing little chance to dissipate these medication effects. Questions of ‘voluntary consent’ became increasingly strident, both from administrators and the patients themselves.

We thought we had solved the subject problem by studying normal human volunteers. How much preclinical information was needed to justify human experiments in volunteers? How much could volunteers be paid before the fair price became an unethical challenge? Could prisoners, children, and the elderly ‘volunteer’ for studies with substances with unknown effects? While subjects agreed to resist taking psychoactive substances, including common foodstuffs (caffeine and theobromine) and smoking (nicotine and cannabis), how does one patrol such use before laboratory experiments?

For a time, pharmaceutical company research directors were enthusiastic about pharmaco-EEG studies as these offered valuable guides to clinical trials. These interests waned as human studies became more expensive. Some companies developed in-house animal laboratories for EEG studies but these faltered, mainly on species-specificity issues and the risks in maintaining primates in laboratory settings. Physician research directors were replaced by marketing executives whose interests were focused on those studies that would support marketing approval from the government’s licensing authority in each country and not on better understanding of the pharmacology of the agents.

At first, presentations at national and international meetings were enthusiastic affairs with much active interchange. The rapid shift of programs to industry sponsored infomercials rather than academic research experiences removed this stimulus for continued research. The final blow was the NIMH decision to end the ECDEU system of evaluation that ended my independent research studies (6).

Renewed interest in convulsive therapy and psychopathology

My interest in pharmaco-EEG also faltered by competing interests in ECT. The ECT research program at New York Medical College that began in 1968 successfully answered questions regarding electrode placement and multiple treatments. An NIMH sponsored conference on the psychobiology of ECT in 1972 rekindled my interest in ECT mechanism (63). The 1974 legislative mandate to prohibit ECT in California led the American Psychiatric Association to establish a Task Force on ECT in 1975 (64). I signed the commission’s report but I was chagrined by the number of recommendations that were based on majority opinions rather than on reliable evidence. As a response, I published my text Convulsive Therapy: Theory and Practice in 1979 to offer my image of what was known (16). I summarized the known data and my extensive studies. The missing information encouraged my report on continuation ECT (65) and the launching of the NIMH-supported collaborative studies of continuation treatments that occupied my interest from 1994 to 2006 (66, 67). In 1980 I assumed responsibility for the ECT Service at University Hospital at Stony Brook, giving me the opportunity to answer clinical research questions (68, 69). Establishing the quarterly journal Convulsive Therapy in 1985 was another distraction. The competing interest in ECT trumped the struggles to continue pharmaco-EEG research and I closed my EEG laboratory in 1985.

My interest in psychopathology was aroused during my clinical service at Stony Brook. I was intrigued by catatonia, a clinical syndrome tied to the diagnosis of schizophrenia. But many patients did not meet schizophrenia criteria. Also, they responded to benzodiazepines and to ECT, two treatments not recommended for schizophrenia. With Michael Taylor, we set out to document the broad nature of catatonia and to recommend that it was an identifiable and treatable syndrome that warranted a home of its own in the psychiatric classification (36, 37). A historical review of how catatonia came to be seen as a type of schizophrenia documented Kraepelin’s fallacy (70). A few years later we examined ‘melancholia’ and again urged that this readily identifiable and treatable syndrome deserved a home of its own in the classification (38–40).

L’envoi

Pharmaco-EEG, indeed electroencephalography, has limited interest among neuroscientists. The enthusiasm for this quantitative science is muted, surviving in a few isolated centers, as the promises of genetics and brain-imaging are the dominant enthusiasms. This shift is sad, since pharmaco-EEG is a quantitative science that yields direct images of ongoing biochemical and neurophysiologic brain events that are intimately connected to human behavior. It is a non-invasive procedure, reflecting the moment-to-moment changes in brain function and vigilance. It may be studied for hours without risk to the subjects. Quantification algorithms and equipment are inexpensive and readily available. While we lack understanding of what the changes in electrical rhythms tell us about cellular physiology and biochemistry, we do see relationships with the aberrances in mood, thought, and recollection that are of interest to neurologists and psychiatrists. It is a valuable science that has much to teach us about psychiatric illness and treatment of the severe mentally ill. I found the study very rewarding and am left with memories of exciting experiments, dedicated co-workers, and a large community of enthusiastic scientists seeking to better understand the relation between brain and behavior.