Dr. Johnson has reviewed several issues in the operant control of EEG alpha rhythm as well as the interactions among cortical electrical activity, general physiological reactivity and subjective experience. Our presentation seeks to expand and clarify somewhat the nature of the interaction between alpha and high levels of arousal/activation associated with fear.
As many others, we were encouraged by the early reports that positive subjective experience was linked to alpha production. Therefore, Orne & Paskewitz began a collaboration to explore the apparent potential of alpha feedback as a means for the individual to gain direct control over neurophysiological arousal and therefore subjective tension-anxiety. Initially we replicated the early findings both to confirm them and to gain experience with the general phenomenon of brain wave feedback.
As Dr. Johnson reported earlier we found that, although subjects did show increasing alpha density with training, the effects seemed to be a function of the presence of ambient light or the use of lights to signal the presence and absence of alpha. Subjects in total darkness began with a spontaneously high level of alpha density which was immediately depressed by visual stimuli such as the feedback signals. Subjects then increased their alpha production, typically by learning to observe the feedback light's change in color in a passive manner by not focusing their eyes on the light.
Further, experimentation carried out by using tones as the feed-
107
108 M. ORNE AND S. WILSON
back signals and varying the presence or absence of ambient light, as reported by Paskewitz and Orne in 1973, clearly documented that visuomotor activity was of primary importance in the results of the usual alpha feedback training experiment. It seemed that the apparent augmentation of alpha density occurred only when alpha had previously been depressed, primarily by visuomotor activity, from the individual's optimum nondrowsy resting baseline obtained in total darkness. Thus, learning to increase alpha from these depressed levels seemed to involve the individual's gradually learning to ignore the stimulus that had been responsible for alpha suppression in the first place. What is more, subjects trained in the dark did not significantly increase their alpha density over that found in an optimal baseline in total darkness, even after six days of alpha increment training.
Thus, we had begun to recognize that the original assumptions about the control of alpha density by a subject's conscious intent, effort or learning were valid only in conditions which somehow had depressed an individual's moderate to high baseline alpha density. Therefore, we reconceptualized the effects of alpha feedback training as an experience which taught the subject to augment alpha under circumstances which would ordinarily reduce the amount of alpha in the EEG (Paskewitz, Lynch, Orne, & Costello, 1970). The visuomotor system seemed to be the overriding factor determining alpha levels in those circumstances in which the person could see visual patterns (or even in a totally dark room attempt to see them) a point which among those interested in feedback training, only Mulholland (1968, 1972) and Peper (Peper & Mulholland, 1970) had emphasized.
With the assumption that different mechanisms might be involved in blocking alpha it seemed to follow that different skills might be necessary to learn to augment alpha activity. We assumed that what was learned during alpha feedback training would depend on the nature of the primary stimulus which was depressing alpha density. Because of the disproportionate significance of visual effects for alpha density, if the individual were trained in the presence of light, he would first need to learn to disregard visual stimuli before the alpha blocking effect due to other factors would be affected. In order to focus on other potential blockers, in particular arousal/activation, we chose to avoid the presence of ambient light in subsequent studies of the function.
In a complex experiment on alpha and activation (Orne & Paskewitz, 1974) subjects came initially to the laboratory to participate in a simple alpha feedback study. Every effort was made to produce a comfortable and relaxing environment so that the initial baselines and feedback alpha
109 ALPHA, BIOFEEDBACK, AND AROUSAL/ACTIVATION
densities were obtained from a minimally aroused subject little affected by fear, anxiety or an active visual system. At the conclusion of this initial session subjects with greater than 25% resting eyes closed alpha density in the dark were invited to return for a second session that would involve electric shock which would "not cause tissue damage but would range from mildly uncomfortable to quite painful."
Of the 22 eligible subjects ten were persuaded to continue. During the second session the same sensory electrodes for EEG, electrooculogram, electrodermal activity and heart rate were attached; finally, the experimenter ostentatiously attached two large silver electrodes and a ground over the right calf and explained that these were for the shock stimuli. Subjects were purposefully not informed when the shocks would be given since it was felt that this ambiguity would maximize anxiety. The lights were turned off and the second day's baselines obtained.
Four five-minute feedback trials preceded any further instructions regarding shock. Only after these were over were the shock instructions given. These explained that during the next part of the experiment he would from time to time be in jeopardy of receiving electric shocks ranging from barely perceptible to extremely painful. These jeopardy periods would be signaled by a third tone different from the feedback tones. This tone would be on only when he wasn't producing alpha. Since these jeopardy periods would be the only time that he would be shocked, he could, simply by turning on alpha, turn off the jeopardy tone and prevent himself from being shocked. Thus, it was explained that the more alpha he could produce, the less the likelihood of being shocked.
Subsequent to these shock instructions five five-minute feedback trials, each divided into ten half-minute segments, were given. Five of these segments were in fact jeopardy segments during which the third tone was always present simultaneously with the no-alpha tone. During the other five segments only the usual alpha, no-alpha tones were present. This same procedure was repeated on a third visit to the laboratory.
The findings were as puzzling as they were complex. First, one would expect that a subject anticipating an experiment which includes painful shock would be anxious and have lower alpha density than that on the first day. However, the anticipated drop in initial alpha baselines did not appear; instead, there was a non-significant rise from the first day mean of 35. 2 seconds per minute. We expected that the initial feedback trials would permit the subject to augment reduced baseline levels; however, since alpha densities were not lower it was hardly surprising, therefore, that we also failed to observe a significant increment in alpha
110 M. ORNE AND S. WILSON
density during the first four feedback trials. Finally, we had expected that following the shock instructions the increased apprehension would again decrease alpha. Most striking, however, was the total absence of alpha blocking following the shock instructions. Further, even during the
111 ALPHA, BIOFEEDBACK, AND AROUSAL/ACTIVATION
jeopardy periods themselves, there was only a very small and transient drop in alpha density, which disappeared by the third feedback trial.
In interpreting these data, the first possibility we considered was whether the shock manipulations were unsuccessful in making the subjects anxious and/or aroused. However, post-experimental inquiries as well as physiological data, notably heart rate and electrodermal response, clearly substantiated that subjects were indeed aroused and anxious as shown in Figure 1.
For example, the heart rate baseline during the first shock session was clearly and significantly higher than that during the first day. Further, it dropped during the first four feedback periods in the precise manner we had anticipated alpha density would rise. Again when shock instructions were given an instantaneous and dramatic increase in heart rate took place, well over ten beats per minute. On both days heart rate was significantly higher during jeopardy periods than during nonjeopardy periods. The electrodermal activity showed a closely analogous pattern of change.
Thus it appeared that neither the apprehension about the shock session, which might have been reflected in the second session baseline, nor the more intense concern following the shock instructions, nor even the acute fear of being shocked, resulted in the anticipated sharp drop in alpha density. The expected relationship between high levels of activation and reduced alpha density did not materialize.
Although this study appears to have a number of ramifications, care must be taken in generalizing from it in view of the lack of additional relevant controls that in retrospect would have been helpful in understanding the results and the special nature of the volunteer sample. In particular, more data are needed comparing the effect of specific stimuli and tasks on subjects' alpha density in the presence as well as the absence of feedback. However, the present data do clearly indicate the lack of a necessary relationship between alpha density and the subject's level of arousal, fear, anxiety or apprehension. Clearly, this finding challenges the widely accepted view concerning the inverted U-shaped relationship between level of arousal and alpha density. It is this view which in large part suggested that high alpha density would index a particular state of the organism characterized by relaxation without drowsiness. It is perhaps now worth examining more carefully the data upon which this theory is based.
The empirical observations which led to the inverted U hypothesis on the relationship between alpha density and arousal included: First, as
112 M. ORNE AND S. WILSON
an individual becomes drowsy and approaches stage one sleep alpha density is dramatically reduced and tends to approach zero; second, that if an individual is confronted with a stimulus evoking sudden arousal such as a loud noise, a period of alpha blocking is observed; third, some studies have demonstrated a moderate relationship between anxiety levels and alpha baselines. What is different about the conditions in the experiment just reported that failed to demonstrate a drop in alpha density associated with a highly arousing stimulus -- the threat of shock -- when compared with circumstances in which previous observations were made?
One major difference between our studies and others reported in the literature is that we used dark adapted individuals in a situation totally devoid of light. We emphasize the importance of dark adaptation since it is not the presence of light but rather the attempt to see which is responsible for alpha blocking. 1 Second, as one considers the nature of the stimuli which have been used to block alpha, it can be seen that they are all of a nature to also cause an orienting response which has a strong visual component. Certainly in dim light and, with subjects not dark adapted, even in total darkness this visual component of the orienting response might well serve to mediate the alpha blocking which has in the past been ascribed to the effect of arousal/activation. However, early studies did not systematically and purposefully vary the level of arousal without using stimuli that are prone to produce a concurrent visual orienting response.
Third, one must consider the relationship of alpha density to low arousal. These data appear considerably more satisfactory. Thus, all subjects show a dramatic drop in alpha density with approaching sleep. For example, Paskewitz and Orne (1973) have demonstrated a remarkably high association between periods of unusually low alpha density within a subject's record and the presence of independently assessed slow rolling eye movements -- well known precursors of sleep.
It is tempting, therefore, to accept these data as documenting the relationship between low arousal and the absence of alpha. Here too, however, caution is needed. The drop in alpha density may not be a function of low arousal at all but rather it may be an incidental manifestation of the active processes associated with sleep.
For example, if one examines night time sleep records, there are periods when individuals show a great deal of arousal. Notably, REM is associated not only with the rapid eye movements which give the sleep state its name but also with penile erection and marked variation in heart rate, suggesting a heightened level of arousal. Nonetheless, during these periods there is usually no alpha or only a very small increase in alpha.
113 ALPHA, BIOFEEDBACK, AND AROUSAL/ACTIVATION
This absence of any change is especially striking when one considers the amount of mentation associated with dreaming and the level of autonomic arousal as indexed by penile erection and cardiovascular changes.
A similar paradox is associated with the periods of GSR storms during stage four sleep first described by Johnson's group. While this fascinating phenomenon does not seem to be accompanied by changes in other physiological parameters such as heart rate and respiration the presence of spontaneous GSR tends to be one of the better indices of arousal and has been closely linked to the level of activation of the reticular formation. Again, we are unaware of any evidence suggesting that alpha density normally increases during such periods.
Thus, during what may be considered the extremes of activation within sleep, both resembling, as it were, major fragments of waking physiological arousal, no considerable emergence of increased alpha density has been documented. Rather, we see an apparent clear separation of processes signaling increased neurophysiological arousal from cortical alpha production. It seems to follow that, were the progressive drop in alpha density seen as an individual falls asleep primarily the result of the associated low level of activation, one ought to find differences in alpha density during sleep itself as the widely variable levels of activation manifest themselves. The absence of such changes strongly suggests that the decrease in alpha density seen as an individual approaches sleep reflects an active inhibition of alpha by the mechanisms associated with sleep rather than a low level of arousal per se.
In considering the basic logic underlying the use of alpha feedback for the control of arousal one final point which has been important in justifying alpha feedback is Kamiya's casual observation that subjects could learn to identify the presence or absence of alpha in their EEG. This clearly suggested that, potentially at least, some form of cortical representation of alpha existed. While perhaps not directly related to arousal, this point was a crucial link in previous arguments justifying the use of alpha feedback training to bring about a special state of consciousness.
In attempting to replicate this study it became clear to us that temporal cues could easily serve to mediate the apparent ability of the individual to recognize the presence or absence of alpha. Thus, if an individual has high alpha density and the experimenter intends to have the next trial be during a nonalpha event, the inter-trial interval will tend to be considerably longer than if he were looking for an alpha event. It follows that subtle temporal cues, which in casual experimentation might well escape the notice of both experimenter and subject, could have accounted for
114 M. ORNE AND S. WILSON
Kamiya's findings.
Controlling for temporal cues by using a modified signal detection procedure, we were unable to observe any increase in the subject's skill in identifying the presence or absence of alpha. 2 Pavloski, Cott, and Black (1975) and Legewie (1975) have reported similar difficulties. While it is of course possible that it is necessary to train individuals with longer windows than those which were used in these studies, it would seem essential that more carefully controlled positive observations be obtained before it is justified to assume that the simple presence of alpha has cortical representation.
To summarize our observations thus far: One, subjects do not appear to learn to increase their alpha density above their optimal resting baseline through feedback; two, visuomotor activity is of prime importance in depressing optimal alpha density and in subsequently learning to enhance alpha; three, high levels of alpha density can be present even during the very high arousal and subjective fear; further, the absence of alpha during sleep seems equally unrelated to arousal levels; and four, subjects may not be able to discriminate directly between alpha and nonalpha waking states. In sum, the views that alpha production is closely related to subjective experience, has specific cortical representation and alone reflects the level of arousal cannot be justified with presently available data.
Given these observations it would seem that the entire basis justifying the potential benefits of alpha feedback training is lacking and accordingly one might well choose to dismiss this entire line of inquiry. However, throughout our efforts to understand alpha feedback we have become increasingly aware of the need to also understand the underlying processes and have been forced to reevaluate issues that had been assumed to be resolved by previous work in order to reconcile the conflicting reports in the literature. Of the several issues that arise, the single most important factor, which has been essentially ignored in the reported work to date, relates to systematic individual differences in the dynamics of the alpha response. Such differences will be discussed more extensively in a paper in progress by Wilson, Orne, and Paskewitz, but it is possible to explore a few of them here.
As you will recall in regard to the shock study I discussed earlier, no group mean alpha density changes were found either in the baselines or during the anticipation of being shocked nor even when electric shock was imminent. However, striking individual differences are masked when the data are grouped in means. For example, consider the Day 2 base-
115 ALPHA, BIOFEEDBACK, AND AROUSAL/ACTIVATION
lines. Of the ten subjects there are two who during the initial baseline on Day 2 decreased alpha by 16 seconds per minute below Day 1 levels while two others increased alpha by 16 seconds per minute. Again after shock instructions, one subject shows an alpha density decrease of 70% while two subjects show increases of 40% over that day's baseline levels.
These findings that some subjects increase, others decrease, while still others show no change in response to the identical stimulus are consistent across Day 2 and 3 and appear to represent different patterns typical of the individual. In this study the alpha response to arousal was investigated in the context of experimentally induced fear and arousal. Analogous observations have been reported by Lemere (1936), Simonova (1968) and Birbaumer (1970) in nonfeedback settings. In each of these instances different kinds of alpha density responses to intense arousal were noted. It seemed important to establish whether these differences in alpha dynamics require intense arousal or represent a more general pattern of response to some other stimuli which alter alpha density.
An earlier study by Paskewitz and Orne had explored the differences between counting backward by one and counting backward by seven across days in a feedback setting. The interpretation of the seemingly consistent drop in alpha density associated with counting backward by seven -- but not with counting backward by one -- was confounded by the combination of the two tasks, counting backward by seven and attempting to augment alpha. In order to explore the nature of potential differences in basic alpha density dynamics, an experiment was performed without feedback on changes in alpha density during cognitive task performance. 3 Eleven subjects, while in total darkness, on three different days performed serial subtractions by several different numbers as well as a descending subtraction task. For the latter the person began by subtracting nine from a three digit number, then eight from the remainder, then seven from that remainder and so on until reaching two, when he began again with nine, eight, seven, etc., until told to stop. We have found performance of this task to be extremely demanding for our subjects, as opposed to counting backwards by one, which presents no difficulty.
Two different major patterns of alpha response to such a task emerge when one examines the data for individual differences in subjects well adapted to the laboratory and performing the task on the second day. For example, during subtraction by one, one group increases alpha and another group decreases alpha relative to their own baseline. In order to evaluate the consistency of such response styles we examined the individual's response during a much more difficult descending subtraction task. Thus
116 M. ORNE AND S. WILSON
subjects were divided on the basis of whether they increased or decreased alpha density while counting backward by one in order to see if this split also differentiated them when they performed the very difficult descending subtraction task. Remarkable similarities are seen. Subjects who block alpha while counting backward by one also do so during descending subtraction. What is more, those who increase their alpha density while performing the simple task increase it during the difficult one as well. This phenomenon can be seen clearly in Figure 2.
This figure shows the mean percent left hemisphere alpha density in two groups, four alpha blockers (dotted lines) and seven alpha augmenters (solid lines), defined by alpha changes, either decrease or increase relative to baseline, during subtraction by one on Day 2. As you can see these two groups have comparable resting alpha density both during the initial baseline and the rests preceding the tasks. Since the two groups were defined by the direction of their alpha response during subtraction by one it is hardly surprising that the difference in alpha density between them is significant during this task.
These individual differences in reaction are not related to the individual's success or speed in counting backward but one perhaps could attribute the result to the general ease with which everyone can perform the task. One might then expect that everyone would block alpha during the difficult descending subtraction task. However, on the second day this task evokes mean alpha changes during trial one very similar to those during subtraction by ones (t = 1. 82, p = .05). Indeed, the split becomes even more significant on the second trial (t = 3. 09, p < . 01).
117 ALPHA, BIOFEEDBACK, AND AROUSAL/ACTIVATION
The stability of the direction of change depends partially on the individual's characteristic amount of phasic blocking. Those showing little change in alpha during performance of a task tend to oscillate between increasing and decreasing alpha during tasks in an apparently random fashion. However, the consistency of an individual's alpha augmenting or blocking response is further demonstrated by the continued significant differences between these two groups, separated by alpha changes with subtraction by one on Day 2, during Day 3 descending subtraction. The t for the differences between the two groups on Day 3 descending subtraction trial one is 3. 02 and on trial two is 3. 58, thus demonstrating the stability of the differences first seen on Day 2. Thus, perhaps contrary to expectations that there might be little blocking to the rote task and a uniform strong blocking effect from the descending subtraction task, there is a remarkable within individuals consistency in the direction of alpha change during various tasks across days. These kinds of differences, although observed by others in the past, have tended to be ignored because they are masked either by the presence of light or by novelty on the first day of testing and were therefore considered to be random variation. However, our data suggest that the individual's style of alpha density response to task situations is consistent across tasks and across days. Further, if one splits subjects on the basis of performance on Day 2 significant differences in alpha blocking in the Day 1 data become apparent.
Clearly, these kinds of data are not reflected in current theories regarding alpha, activation, behavior and subjective experience outside of the feedback context. It is equally clear that we cannot expect to apply alpha feedback to obtain predictable results unless these powerful systematic individual differences are understood and integrated into our general set of assumptions about alpha phenomena during feedback.
What may we then consider to be established conclusions regarding the relationship between activation/arousal and alpha density? Perhaps first we must recognize that, contrary to our initial naive hopes, we cannot assume that high alpha density is uniformly accompanied by moderate physiological arousal or subjective calm. It is now clear that a number of different mechanisms influence alpha density and interact to determine an individual's tonic levels and phasic changes in alpha. Second, visuomotor activity is of primary importance in determining alpha density and in learning to augment alpha, in the presence of light. It would appear that feedback training carried out in light requires the development of different skills and may have very different subjective and objective results than that carried out in total darkness. Third, while in early work we could not get people to exceed baseline alpha levels, it is now clear that some subjects do exceed their baseline level -- but in response to activation! Fourth,
118 M. ORNE AND S. WILSON
clearly the widely accepted inverted U-shaped function hypothesized to relate alpha density to activation needs to be reevaluated. Fifth, very important systematic individual differences in alpha dynamics must be accounted for in further explorations of the relationships between cortical electrical activity, subjective experience and behavior as well as in alpha feedback training research. The individual differences reported here may help to direct further efforts at understanding results such as those of Travis, Kondo, and Knott (1975) who found that only about 60% of subjects feel the attempt to enhance alpha in the dark is relaxing or pleasant.
In sum, our findings suggest that the proposal to use alpha feedback training as a means of helping individuals to self-regulate arousal in the face of stimuli eliciting activation/arousal cannot be expected to be uniformly useful. However, we have begun to recognize that totally opposite alpha dynamics in the face of high arousal may exist in the population and that some individuals do respond to activation with drops in alpha density, even in total darkness. If we apply alpha feedback specifically to these subjects results in line with early optimistic reports by others may yet be obtained.
REFERENCES
Birbaumer, N. Praeventive alpha-inhibition and angst. Studia Psychologica, 1970, 12, 179-187.
Broughton, R. J. Sleep disorders: Disorders of arousal? Science, 1968, 159, 1070-1078.
Jasper, H. H. Cortical excitatory state and variability in human brain. Science, 1936, 83, 259-260.
Legewie, H. Subjective correlates of EEG feedback: Discrimination, learning or superstition? Paper presented in the symposium "EEG Alpha Learning: State of the Art" at 1975 meeting of the Biofeedback Research Society.
Lemere, F. The significance of individual differences in the Berger rhythm. Brain, 1936, 59, 366-375.
Mulholland, T. The concept of attention and the electroencephalographic alpha rhythm. In C. R. Evans & T. B. Mulholland (Eds. ), Attention in neurophysiology. New York: Appleton-Century-Crofts, 1969.
Mulholland, T. Occipital alpha revisited. Psychological Bulletin, 1972, 78, 176-182.
Orne, M. T., & Paskewitz, D. Aversive situational effects on alpha feedback training. Science, 1974, 186, 458-460.
119 ALPHA, BIOFEEDBACK, AND AROUSAL/ACTIVATION
Paskewitz, D. A. A hybrid circuit to indicate the presence of alpha activity. Psychophysiology, 1970, 8, 107-112.
Paskewitz, D. A., Lynch, J. J., Orne, M. T., & Costello, J. G. The feedback control of alpha activity: Conditioning or disinhibition? Psychophysiology, 1970, 6, 637-638.
Paskewitz, D. A., & Orne, M. T. Visual effects of alpha feedback training. Science, 1973, 181, 360-363.
Pavloski, R. , Cott, A., & Black, A. H. Paper presented at the Annual Meeting of the Eastern Psychological Association, April 1975.
Peper, E., & Mulholland, T. B. Methodological and theoretical problems in the voluntary control of electroencephalographic occipital alpha by the subject. Kybernetic, 1970, 7, 10-13.
Simonova, P. V. Basic (alpha) EEG rhythm as electrographic manifestation of preventive inhibition of brain structures. Progress in Brain Research, 1968, 22, 138-147.
Stennett, R. G. The relationship of alpha amplitude to the level of palmar conductance. Electroencephalography and Clinical Neurophysiology, 1957, 9, 131-138.
Travis, T. X., Kondo, C. Y. , & Knott, J. R. Subjective aspects of alpha enhancement. British Journal of Psychiatry, 1975, 127, 122-126.
NOTES
1. In an unpublished study subjects were told to signal by pushing a button when they had found a faint pinpoint of light which would be turned on sometime after a signal was given. The light was not actually turned on until some thirty seconds after subjects were told to begin looking for it. However, alpha blocking took place as soon as subjects were signaled to begin looking for the light source. Incidentally, this effect is quite specific to visual attention and is not seen in an analogous situation using auditory stimuli.
2. We wish to thank Anthony L. Van Campen who served as Biomedical Engineer for this attempted replication.
3. Different aspects of this study were conducted by Frederick J. Evans, Betsy E. Lawrence and Anthony L. Van Campen.
The line of research reported here would not have been possible without the close collaboration of David A. Paskewitz who designed the equipment, ran the subjects, and supervised the analysis of all but the most recent
120 M. ORNE AND S. WILSON
studies. This later work was carried out in collaboration with Frederick J. Evans, Betsy E. Lawrence, Emily Carota Orne, and Anthony L. Van Campen. We would also like to express our appreciation to them and to William M. Waid for helpful comments and suggestions in the preparation of this manuscript.
The substantive work upon which this paper is based was supported in part by grant #MH 19156-04 from the National Institute of Mental Health, the Advanced Research Projects Agency of the Department of Defense and was monitored by the Office of Naval Research under contract N0001470-C-0350 to the San Diego State College Foundation, and by a grant from the Institute for Experimental Psychiatry.