Early Investigations Leading to Neurofeedback.
Joe Kamiya, Ph.D.
The aims of this chapter are (1) to describe some of the early work which led to the field of research and therapeutic application initially referred to as Electroencephalographic (EEG) Biofeedback, but is now more commonly named neurofeedback; and (2) to address the question of what internal behaviors or private experiences are involved in learning to produce changes in specific EEG measures with the aid of neurofeedback.
The main use of the human EEG is in neurology clinics to help detect abnormal brain functions. My research with the EEG was not aimed at developing methods for improving diagnostic procedures or for treating persons with neurological or other problems, or for any other practical applications. It was instead conducted to pursue questions concerning relationships between the EEG of persons and their consciousness, also known variously as subjective experience, private experience, mental states or mental activity, internal behavior, mind.
I began my interests in EEG research in the sleep laboratory of Nathaniel Kleitman and his student and research assistant William Dement, at the University of Chicago. It was from that laboratory that Aserinsky and Kleitman (1953) and Dement and Kleitman (1957) published the pioneering papers showing that dreaming during sleep usually was accompanied by specific changes in the sleeping person’s EEG and eye movements as monitored by the electro-oculogram (EOG). Their papers did much to put private experience on the scientific map. Kleitman generously offered me the use of his laboratory to do some studies of my own. Dement taught me the technology of EEG and EOG recording of sleeping subjects. I did a study on other physiological concomitants of drowsiness and sleep (Kamiya, 1961). My student and colleague Johann Stoyva joined me in the laboratory. In addition, in response to confusion and disagreements in the field on the problem of how to interpret the occasional fact that reports of dreaming would occur despite the absence of their EEG and EOG indicators, and the absence of dream reports when the EEG and EOG indicate dreaming had occurred, we published an analysis of the logic of the relations between verbal reports and physiological indicators as convergent indicators of private events like dreaming. (Stoyva, J. and Kamiya, J. 1968). The problem is worth mentioning here because it arises in connection with the validity of the evidence of any sort of private experience, not just dreams.
In the course of preparing a subject for all night recording with EEG and EOG electrodes, I always ran a test of the EEG on the polygraph for a minute while the subject was still awake to make sure that all the electrode contacts with the scalp provided clean traces. It was during these tests that I noticed the irregularly timed appearance and disappearance of the EEG alpha rhythms. I wondered if they were related to any features of the consciousness of the person. How this interest led to the development of methods for studying that possibility, and eventually to the adoption of the method by others in the treatment of neurological disorders, is described in what follows.
The alpha rhythm’s intermittent time course of appearance and disappearance catches the attention of nearly all persons who first view their own record of a few minutes of their waking, eyes closed EEG. Among the records of several hundred persons I have seen over the years, at the sleep laboratory as well as elsewhere, there are wide variations in the characteristics of the trains of alpha rhythms. Some may last for but a fraction of a second, with only two or three waves appearing before they disappear while some last for as long as 10 to 20 seconds with only brief near-interruptions or none at all. And the waves of different trains of alpha vary in size, from barely noticeable in the record to roughly ten times in height (amplitude). The durations of the absence of alpha are also very irregular. The moments without alpha consist of a mixture of frequencies, both lower and higher than that of alpha waves, with many frequency components occurring simultaneously but with varying proportions of representation in the total spectrum, resulting in a much more jumbled incoherent appearance than the smooth waves of alpha activity.
There are also wide individual differences in the extent to which the alpha rhythm dominates the record in persons at rest, awake, with eyes closed. Most persons’ EEGs I pre-tested for my experiments showed alpha activity from about 15% to 80% of the total time, and for my experiments I chose people who varied in percent of time of the presence of alpha from about 25% to 70%. This restriction of the range of alpha risks the applicability of conclusions to the population at large, but was chosen for the reason that the experimental procedures would have been more difficult to administer reliably with persons whose percent time alpha scores were very low or high. I was depending on my visual inspection of the raw EEG to make decisions of whether alpha was present or absent in the early experiments. It was possible with most subjects to make reliable visual judgments of Alpha rhythm present vs absent, but to improve reliabiltiy a filter with a pass band of 6 to 12 Hz was ussedReliability was improved later with electronic devices.
The question that motivated me was whether there were subjective concomitants associated with the moments when alpha rhythms were present as opposed to when they were absent. Might there be a difference in the feel or mental activity between the two EEG states in the relatively short term alternations between the two that occur several times a minute? Considering what is known about conditions affecting alpha would help provide hints toward an answer. First, when drowsiness occurs, there would be a drop in alpha dominance in the graphic record. Sleep onset will produce a complete loss of alpha. Moving toward the other end of the continuum from sleep to maximum wakefulness, alpha dominance is usually substantially reduced by intense mental concentration, hyper vigilance, fear, anxiety, and distraction of attention by external disturbances. Providing data at several points along this continuum were Adrian and Matthews (1934). They were among the earliest investigators of EEG alpha activity. When the eyes which have been closed are then opened, there is an immediate drop in the dominance of alpha. In a personal communication with an early researcher in the field, Geoffrey Blundell, now deceased, I learned W. Grey Walter and colleagues experimented informally in the 1930s and 1940s and discovered that whether their eyes were open or closed, if they imagined seeing objects that were close to them, alpha activity was diminished. Mulholland (1968) pointed to a neurophysiological explanation of this result. He showed that alpha activity can be very much diminished by the oculomotor system. Binocular convergence, which clearly is increased in persons inspecting close-up visual stimuli, greatly diminishes alpha dominance. Thus it is possible that if people happen to have thoughts with visual images, they may be converging their eyes toward each other to some degree.
Another observation, reported by Mulholland and Evans (1965) and by Dewan (1967), is that rotation of the eyes upward increases the alpha activity. I have observed this also, but it is not a simple relationship because it does not always occur, and I found (Kamiya 1968) that the relationship between eye position and alpha activity is not a hard wired one; it can be disrupted or eliminated, at least under certain conditions and for perhaps an hour. The extent to which binocular convergence and upward eye orientation accounts for all the EEG alpha levels, especially with eyes closed, is uncertain. However, there is no doubt that a substantial portion of alpha suppression can be accounted for by some kinds of mental activity involving active visual imagery, in addition to the suppression seen in anxiety, distraction of attention or other mental states.
Since suppression of the alpha wave activity of awake persons resting with eyes closed occurs with drowsiness and sleep, on the one hand, and with active mental processes involving visual imagery, anxiety and distracted attention, on the other, the presence of alpha activity would be reflective of a state or states which involve intermediate levels of mental alertness and mental activity. However, typically the spontaneous comings and goings of alpha in persons resting with eyes closed are so rapid that I wondered whether those more transient states could be related to mental states as well. One might think that asking a person to give a running account of his private experiences while his EEG was being recorded might be a good way to answer the question of whether the presence of alpha felt different from its absence. However, the rate of alternation from alpha to non-alpha is so rapid, probably from around 6 to 20 times per minute in many individuals, it would be unlikely for a verbal report to be able to track the pace of whatever subjective experiences might be related to the presence of alpha as opposed to its absence. Moreover, the act of speaking would cause much interference of the EEG by the electrical activity of the cranial muscles, and the noise caused by head movements disturbing the scalp-electrode interface would further complicate such recordings. Finally the brain activity of speaking would also generate EEG activity which could become a source of ineradicable electrical noise in the recorded EEG. So, EEG recording during speech would not work for the question raised here.
My familiarity with the work in another laboratory at the University of Chicago became of great help in facing the question of the subjective differences between alpha and its absence. Operant conditioning of discriminative responses was being investigated with animals. If a specific behavior, e.g., a lever press of a hungry animal is reinforced (rewarded) with food only when the animal can see an external signal, like a light, then over time the response will gradually occur only when the signal is present. The light is named a discriminative stimulus because it has come to be discriminated as the stimulus for the reinforced behavior.
Because I was interested in relating a subjective experience to physiological activities which presumably underlay it, or which was at least concomitant with the physiological activities, I became interested in considering the feasibility of operant discrimination training in which the discriminative stimulus was internal to the person rather than the external location in ordinary discrimination training. There clearly are many such internal discriminative stimuli that all persons have learned early in life. Parents teach children to attend to bowel and bladder sensations as the discriminative stimuli controlling all phases of toilet behavior. The continuous stream of internal stimuli associated with voluntary behavior, eg., walking, are not only essential for the coordinated guidance of the behavior, but become identifiable as components of a sequence of actions which collectively become the discriminative stimuli identified and named walking.
Raising the question of the subjective experience associated with EEG alpha does not imply that the alpha rhythm itself as an electrical process might have stimulus properties. Instead, the question being raised is whether the alpha activity might be reflective of a complex system of brain activity that produces a difference in the subjective experience of the person. Nor does it imply that the subjective quality must be describable verbally. It could be that a person can be trained to discern that there is a difference in experience associated with the presence of alpha activity as opposed to its absence, but he or she could be quite inarticulate in describing the difference. If the person could not be trained to discern any difference between the two states with discrimination training, then one can assume there are no quality differences to report.
The procedure I used was borrowed directly from operant discrimination training, with a training method I had used in a pilot study to get persons to attend to the relative position of their two feet in the course of their walking. I did not tell the participants anything about foot position being involved. I requested that they walk continuously in a circle of about 20 ft. in diameter during the entire course of the experimental session. I told them they were to say either A or B whenever I rang a single ding of a bell, and that I would tell them whether their response was correct. I rang the bell about 3 or 4 times per min, each time being careful to ring it at exactly the moment the person planted either the left or the right foot forward in the course of his or her walking. “A” responses were correct if the bell had rung when the left foot had just been planted, and “B” responses were correct if the right foot had just been planted at the bell ring. Since it is clear that the person would be able to do the task without error if knowing that foot position was the key to learning, what had to be learned by the person was to direct his or her attention to foot position when the bell rang. This directing of attention would be making an internal “observing response” corresponding to the external observing response trained in pigeons by Wyckoff (1952). In his experiment the birds were trained to look over a barrier to locate the discriminative stimulus to a previously trained response.
In the case of any subjective experience that might be associated with alpha on vs. alpha off it seemed like a direct copy of the method of bell ding prompting of a verbal response just described would be useful, with alpha present replacing left foot forward, and alpha absent replacing right foot forward. However, in contrast with the foot position problem, it was not known whether there are any reliable discernible differences in sensations or feeling associated with the two EEG states, especially because the two states were quite transient. Therefore, the task could have been impossible even if the appropriate attention were directed to the task.
The successful execution of the idea for the EEG experiment would depend on my being able reliably to discern alpha present from alpha absent in the EEG record for the trials. I found that I could quite reliably judge from the raw record in the first few participants I tested. However, to assure reliability in future cases and to increase my reliability overall, I had the laboratory technician who was skilled in electronics build an electronic filter with a pass band of 8 to 12 hz -- the standard definition of the frequency of alpha, to eliminate all other EEG frequencies which could confuse me in administering the trials.
For the experiment I used university students as volunteer trainees in an “experiment on learning and brain waves” Electrodes were placed over the left occiput with the reference electrode placed on the left ear. The trainees sat in a dark room with eyes closed while the EEG equipment and I were in an adjacent sound deadened, totally dark room.. Communication with the trainee was by intercom. They were told that at random intervals a single ding of a bell would sound, sometimes when they were showing a particular pattern of EEG, and at other times the bell would be sounded in the absence of the EEG pattern. We wanted to know if they would be able to detect a difference in feel between the two EEG states, and that we would be giving them a series of trials to find out. I told them that each time I rang the bell, they were to make a guess which of the two brain states they were in at the time, by saying A or B, and that if they said A when the pattern of EEG was present at the time of the bell, I would say “Correct!”, and if they said B when the pattern was not there, I would say “Correct! The alternation of trials between alpha present and alpha absent for bell dings on the two EEG states was randomized, and only about five trials were given with the bell dings each minute, so that many more alternations in EEG state occurred than were tested by the bell. The bell was usually sounded only after about 1 or 2 seconds of each state had been observed, on the assumption that any persistence of subjective states in the transition between the EEG states could confuse the trainee if the bell ding occurred too soon after the start of each state.
The results were clear. In about 50 to 500 trials six trainees learned to make nearly 100% correct identifications of the alpha and non-alpha states. There was some agreement among the trainees that the non-alpha state was a more active mental state, especially having to do with concentration, trying too hard, or related to attempts at visualization of objects or persons in the room. There was little agreement on the subjective qualities associated with the presence of alpha, except that it was a calmer, more blank or vacant state of mind, more passive. One trainee, curiously the most accurate in his responses, was the most inarticulate about how the feel of the two states differed. After I gave him extra training sessions to help him continue his explorations for any sensations related to alpha activity, he mentioned that he thought there was a difference in the feel of the eyes when alpha was present. I decided to monitor his eye position with special monitoring electrodes for detecting vertical rotations of the standing corneal-retinal dipole. I found that indeed when his closed eyes were rotated upward, he showed an increase in alpha waves, and when his eyes returned to level position or turned downward, the alpha disappeared. With this result, one could reasonably infer that the trainee had learned merely to discriminate his eye position. From his lack of mention of eye position in explaining his successful discrimination of alpha vs its absence, it seemed he had not been intentionally varying his eye position for the trials to help him do better with A and B responses, but he was able to sense a subjective concomitant of eye position.