Related Topic(s):

MEDITATION, REST, AND SLEEP ONSET: A COMPARISON OF EEG AND RESPIRATION CHANGES

A chapter from the Textbook of Neurofeedback, EEG Biofeedback and Brain Self Regulation
edited by Rob Kall and Joe Kamiya

MEDITATION, REST, AND SLEEP ONSET: A COMPARISON OF EEG AND RESPIRATION CHANGES

Karen H. Naifeh, Ph.D.
University of California, San Francisco



BACKGROUND

Various forms of meditation have attracted increasing interest in the West because of the positive effect they are reputed to have on physical and mental health. In an age of high stress, a method for the "attainment of inner equilibrium under all circumstances" (Swami, 1926, p. 42) constitutes a compelling proposition for Westerners. Meditation is now recommended by some professionals as an adjunct to psychotherapy (Kutz, Borysenko & Benson, 1985), and has been included as a part of the treatment package for stress related disorders such as hypertension and asthma. It has also been shown to be efficacious in the treatment of anxiety disorders, phobias, insomnia, and addictions (for example, Boswell & Murray, 1979; Goldman, Domitor & Murray, 1979; Honsberger, 1973; Parker, Gilbert & Thoreson, 1978; Patel, 1975; Raskin Bali & Peeke, 1980; Shafii, Lavely & Jaffe, 1975; Sururt, Shapiro & Good, 1978; Thomas & Abbas, 1978; Woolfolk, Carr-Kaffeshan & McNulty, 1976; Zuroff & Schwartz, 1978; Shapiro & Giber, 19
78 (review); Delmonte, 1985 (review)).

A good deal of research on meditation has addressed the question of the mechanisms by which meditation exerts its therapeutic effects, and as a corollary has attempted to conceptualize meditation both psychologically and physiologically. Given the current emphasis, within psychiatric (and to some extent psychological ) thought, on forming a unitary concept of mental life which includes the various biological and psychological disciplines, it is important to try to integrate our knowledge of meditation in these areas. Because this study is primarily psychophysiological in nature, emphasis is placed on a review of psychophysiological research on meditation; however, relevant psychological material is reviewed and integrated. In this study, psychophysiologic conceptualizations of meditation as a relaxation technique and as a "holding on" to the hypnogogic state between wakefulness and sleep, together with research relevant to these conceptualizations, are reviewed. Additionally, their relevance to hypothesized mechanisms by which meditation could be therapeutically effective are explored.

This study investigates patterns of psychophysiologic measures not previously examined in order to further explore the hypothesis that meditation is a process of maintaining the normally transient hypnogogic state between wakefulness and sleep. The states associated with three different meditative traditions and a non-meditating control group will be studied; data will consist of electroencephalogram (EEG), forehead electromyogram (EMG), end-tidal carbon dioxide (CO2) tension, respiratory rate, and thoracic and abdominal respiratory patterns, collected during resting baseline, meditation (or relaxation in controls), and sleep onset. The primary objective is to clarify the relationship of EEG and respiratory variables to meditation, normal wakefulness, drowsiness, and sleep.

Definitions of Meditation
According to Davidson (1976, p. 346), meditation is "a term applied to a diverse group of practices having the common goal of producing in the short term desired mental states, and in the long term the promotion of personality growth and mental health (traditionally referred to as 'enlightenment')." The practices extend far back in recorded history and in the past were taught most often within the context of a spiritual discipline. Along with the large increase of interest in meditation in the West, there has been some dissociation of meditative practice per se from its connection with any religious discipline; as in the case of transcendental meditation, clinically standardized meditation (Carrington, 1978), and the relaxation response (Benson, 1975).

Scientifically, the most widely studied types of meditation are those in which the subject sits quietly, rather than the more active types such as tai chi, chanting meditations or dervish dancing. In the meditation practices involving quiet sitting, common elements are a quiet environment, a focus of attention, and a passive attitude. Shapiro (1982) described three broad groupings of attentional strategies in meditation: a focus on the whole field, as in mindfulness meditation; a focus on a specific object within the field, as in concentrative meditation, and a shifting back and forth between the two, or integrated meditation.


Briefly, the task during concentrative meditation is to focus attention on the specific object chosen; in Zen meditation this might be the breath; in TM a mantra, in other types of yoga meditation it might be the breath, energy flow within the body, or a mantra; in Tibetan meditation it might be a visual image. Other mental activity is perceived as a distraction , and when the mind wanders from the object of concentration the meditator passively disregards the intrusion, repeatedly refocusing on the object of choice. In mindfulness meditation, once the meditator learns to merely observe the contents of the mind in a detached fashion using a concentrative technique, the attention is then allowed to detach itself from the primary focus, and to scan in an open focus, shifting freely from one mental activity to the next. All are observed in a detached fashion, and eventually the meditator begins to recognize patterns and habits that dictate thought formation and dissolution (Kutz et al, 1985).

While these descriptions give a general flavor of certain types of meditation, it should be emphasized that the varieties of meditation are quite numerous. The term "meditation" will refer here to the quiet concentrative or mindfulness meditation techniques described above. Although a number of different meditation techniques are therefore being covered in this study under the rubric of "meditation," this does not imply that they are to be considered identical or equivalent, but simply that they are thought to have certain similarities.


Psychological Conceptualizations of Meditation
Efforts to understand meditation in psychological terms have accompanied the introduction and spread of meditation in the West. Early on, Freud (1930/1961) interpreted what he called the "oceanic" meditative experience as a reaction formation of omnipotence to infantile helplessness. Even Jung, who was at home with mystical philosophy and Eastern thought, felt that meditation was not useful to Westerners (Jung, 1943/1968). More recently, however, mental health professionals have written more favorably about the psychology of meditation. Delmonte (1987) has postulated that the repetitive nature of meditative foci such as breathing or repetition of a mantra helps to block or displace normal rational thoughts, thus producing a reduction in logical, clear mentation and an increase in "primary process" mentation, or what Kelly (1955) has called prelogical construing.
Today, according to Noy (1978), primary process thinking is considered to be an indispensable mode of mental functioning known for its intuitive flexibility and multidimensional treatment of psychic content. He termed it essential for combining data and feelings into internally meaningful schemes. The emotional receptivity during meditation is thought by Kutz and colleagues (1985) to loosen the defenses and allow the emergence of repressed material. They feel that meditation can provide a richer flow of primary process material than is normally available in psychotherapy or even psychoanalysis. Repressed thoughts frequently come into consciousness during meditation, according to Delmonte, due to the blockage of the kinds of thoughts that normally form a part of the repressive barrier. Goleman (1971) has referred to the slow, almost self-paced exposure to unpleasant repressed memories which can accompany the relaxed state induced by meditation, as a form of desensitization. The sense of detachment and the detached observation of mental contents which occur during meditation are thought to promote these processes.

In a similar vein, several authors have spoken of the reduced sensory input and monotonous single focus as encouraging the occurrence of adaptive regression (Maupin, 1962; Shafii, 1973; Tart, 1972). Delmonte (1987) adds that in such a state one may not be able to marshal the intellectual defenses of a more alert state. Kelly (1955) has hypothesized that hypnogogic reverie helps to integrate new experience via a temporary loosening of one's personal construct system; the evidence for a hypnogogic state during meditation is considerable and will be presented in detail later. Delmonte feels that loosening of personal construct systems and adaptive regression are similar concepts, and that meditation provides integration of new experiences via these mechanisms.


Psychophysiology of Meditation
Psychophysiologic aspects of meditation have also undergone fairly extensive investigation, especially in the last 20 years. The early research suggested that meditation is a unique physiologic state of consciousness, different from normal wakefulness and sleep. Early scientific studies of meditation indicated that the following physiological changes are associated with both Yoga (transcendental meditation, ananda marga meditation, and other variants) and Zen meditation:
1) the emergence of high amplitude, well modulated alpha waves in the EEG, and in some cases well-formed trains of theta waves or synchronous beta activity as well (Das & Gastaut 1955; Bagchi & Wenger 1957; Kasamatsu, et al 1957, Anand, et al 1961; Akishige 1970; Banquet 1973);
2) a large increase in basal skin resistance, with appearance of spontaneous GSR's also (Bagchi & Wenger, 1957; Hirai, 1974);
3) very low levels of EMG activity (Das & Gastaut, 1955);
4) decreased respiratory rate (Bagchi & Wenger, 1957; Hirai, 1974);
5) lowered oxygen consumption (Wallace & Benson, 1972; Hirai, 1974).


Meditation as Relaxation
Because of the findings of lowered somatic arousal, meditation began to be conceptualized psychophysiologically as a relaxation technique (Benson, 1975). Benson argued that meditation was not a unique state and that the physiological changes associated with meditation could be brought on by any passive relaxation technique; he coined the term relaxation response to describe these changes. His study with Beary (Beary & Benson, 1974) showed greater decreases in physiological measures of arousal when his subjects elicited the relaxation response as compared to when they were merely sitting quietly. Thus an issue which became the focus of later research was whether meditation was different physiologically from other clinical self regulation strategies such as progressive relaxation or self hypnosis.

A number of studies were conducted which showed no differences between meditation and self-hypnosis, EMG biofeedback, or progressive relaxation for the production of relaxation-related changes in skin resistance, heart rate, respiration rate, or systolic and diastolic blood pressure (Boswell & Murray, 1979; Cauthen & Prymak, 1977; Curtis & Wessburg, 1975-76; Lehrer et al, 1983; Morse et al, 1977; Travis et al, 1976; Walrath & Hamilton, 1975; Warrenburg et al, 1980). Thus in terms of lowering somatic arousal, meditation does not appear to differ from other relaxation or self-regulation techniques. Shapiro (1982) concluded from his review of the literature that meditation appears to lower somatic arousal, thereby creating a state of relaxation; and that the changes are greater than simply resting, but no greater than other self-regulation techniques.

Evidence for meditation producing greater reduction in somatic arousal than just sitting quietly and resting came from the following studies: A greater rise in basal skin resistance and decrease in respiration rate in meditating vs resting subjects was found by Elson and colleagues (1977). A reduction in adrenergic end-organ responsivity in meditating vs. resting subjects was reported by Hoffman and colleagues (1982), and replicated by Mills and colleagues (1990). A greater increase in blood flow was reported by Jevning and colleagues (1978). A greater rise in basal skin resistance in meditating vs. resting subjects was reported by Orme-Johnson (1973). A greater decrease in blood pressure was found by Parker and colleagues (1978). Greater decreases in muscle tension in meditating vs resting subjects were found by Malec & Sipprelle (1977) and by Morse and colleagues (1977). West (1979) found a progressively greater decrease in skin conductance during meditation over a six-month period, which was not found in controls. Greater decreases in oxygen consumption and respiratory tidal volume were also found (Dhanaraj & Singh, 1977). While all of these studies reported significant differences between meditators and resting controls for some measures, however, many of them also found no differences between those groups for other measures. For example, Elson and colleagues (1977), Dhanaraj and Singh (1977), Malec and Sipprelle (1977), and Morse and colleagues (1977) reported no difference in HR reduction, and Malec and Sipprelle (1977) and Morse and colleagues (1977) found no difference in reduction of electrodermal activity.

There were also studies which found that subjects who meditated showed no greater reduction in somatic arousal on any measure than a group of non-meditators who simply rested for an equivalent period of time: Boswell and Murray (1979) found no difference in the lowering of heart rate (HR) or electrodermal response (SCR); Bahrke and Morgan (1978) found no difference in lowering of HR and oxygen consumption and the raising of hand temperature (TEMP); Curtis and Wessburg (1975-76) found no difference in the lowering of HR, respiration rate (RR) and SCR; Holmes and colleagues (1983) found no difference in the lowering of HR, RR, SCR or blood pressure (BP); Lintel (1980) reported no difference in electrodermal response; Michaels and colleagues (1979) found no difference in the lowering of HR and BP; Puente & Bieman (1981) found no difference in the lowering of HR, RR, EMG or SCL; Routt (1977) found no difference in the lowering of HR, RR, SCR, and blood flow; Travis and colleagues (1976) found no difference in the lowering of HR and muscle tension (EMG); and Walrath and Hamilton (1977) found no difference in the lowering of HR, RR, and SCR.

In reviewing this literature with a focus on the question of whether meditation produced greater somatic arousal reduction than simply resting for an equivalent period of time, Holmes (1984) considered only studies which included a separate group of nonmeditators who rested, together with a meditation group who meditated. His conclusion was that meditation did not differ significantly from rest in terms of its ability to lower somatic arousal. He pointed out that across experiments there was no measure of arousal on which meditating subjects had reliably lower arousal than resting subjects, and within any study there was no consistent evidence across measures that meditators had a lower arousal level than resting subjects. Holmes' review caused great furor in the meditation research community. He was criticized for being negatively biased regarding meditation, for misstating and/or misinterpreting information, and his clinical conclusions for being overgeneralized and misleading (Benson & Friedman, 1985; Morrell, 1986; Shapiro, 1985; Suler, 1985; West, 1985). Close inspection of the original studies indicates that Holmes did not misstate information, but a negative bias toward meditation is apparent, and a misinterpretation of data did occur with respect to at least one study (Hoffman et al, 1982) as pointed out by Morrell (1986). Also, his conclusion that "the personal and professional use of meditation as an antidote for high somatic arousal is not justified by the existing research data" does not seem warranted, as the studies all demonstrated a decrease in somatic arousal produced by meditation, even though it could not be demonstrated to be consistently any greater than that produced by resting. Based on the studies he reviewed, Holmes' overall conclusion about the lack of evidence that meditation consistently lowers somatic arousal more than resting, however, does appear well founded.

In response to Holmes' review, Dillbeck and Orme-Johnson (1987) did a meta-analysis of the literature on transcendental meditation and somatic arousal. This type of analysis allowed them to estimate effect sizes across studies and thereby include even studies without a separate resting control group (Glass et al, 1981). They reported that TM was associated with a significantly larger effect size than rest for skin resistance, respiration rate, and plasma lactate. In addition, some of the studies Holmes did not consider in his review because subjects served as their own control actually compared subjects when they simply rested to when they meditated (Beary & Benson, 1974; West, 1977). Both of these studies showed a greater reduction in somatic arousal measures during meditation than during rest. Thus controversy still exists regarding this issue, with data both supporting and refuting meditation as producing a greater decrease in somatic arousal than does rest. Part of the problem appears to be a certain amount of emotional bias, both positive and negative. I must agree with Shapiro (1985) in calling for unbiased research, "neither nay-saying nor filled with hosannas."

In all of these studies meditation was conceptualized as a relaxation technique, the main question being whether the somatic arousal reduction brought on by the relaxation in meditation is more profound than the relaxation produced by simply resting. Shapiro (1982) pointed out in his review some of the difficulties with conceptualizing meditation as a relaxation technique. One is defining an independent variable by its dependent variables, which is tautological. Another is that there are many types of meditation techniques, some involving sitting quietly, some involving being physically active, some producing excitement and some producing quiescence. Shapiro therefore favored conceptualizing meditation in terms of attentional mechanisms, and defined meditation as "a family of techniques which have in common a conscious attempt to focus attention in a nonanalytical way and an attempt not to dwell on discursive, ruminating thought" (p.346).


Meditation as Maintaining the Sleep-Wake Transition
The EEG literature has been more mixed in its conceptualization of meditation. The earliest literature spoke of meditation simply as a "unique state of consciousness, different from wakefulness and sleep." Chief among the evidence for this statement were the findings of high amplitude, well modulated, slow alpha waves in the EEG, and in some cases well-formed trains of theta waves or synchronous beta activity (Das & Gastaut 1955; Bagchi & Wenger 1957; Kasamatsu, et al 1957, Anand, et al 1961; Akishige 1970; Banquet 1973). Glueck and Stroebel (1975) also found a spreading of slow alpha frontally over time as subjects continued to practice meditation. Hebert and Lehmann (1977) confirmed the presence of theta trains in approximately one third of the meditators studied, but none of the controls, which included 36 subjects who were studied as they fell asleep; and Fenwick and colleagues (1977) reported theta bursts in four of ten meditating subjects. Other studies have questioned the uniqueness of the meditative state as evidenced by EEG findings: Morse and colleagues (1977), using an "own control" design, reported that synchronous alpha in several EEG channels was present when subjects were relaxing or under self hypnosis as well as when they were meditating. Warrenburg and colleagues (1980) reported theta bursts in both meditators and progressive relaxation subjects, although there was a trend for meditators to produce more (in terms of frequency and duration) than the others. Tebecis (1975), in a controlled study comparing hypnosis, meditation and simple rest, found the greatest increase in theta power for the hypnosis group, followed by the meditation group. There were no significant differences between groups on alpha power. Corby and colleagues (1978) found significantly more theta activity during baseline but not during meditation in meditators as compared to controls who simply rested, and there were no significant differences regarding alpha power.

It is difficult to reach firm conclusions about the uniqueness of EEG patterns during meditation due to a number of methodological problems in the above studies. With the exception of the studies by Tebecis and by Corby and colleagues, none of them used quantitative analysis to demonstrate that statistically significant increases in alpha amplitude,slowing of alpha frequency, or any of the other above-mentioned changes occurred. However, neither Tebecis nor Corby and colleagues used spectral analysis, which separates the EEG into component frequencies, and the lumping together of all frequencies within a particular band would have masked any slowing within a band which might have occurred. Also, Banquet, in his study using spectral analysis, made a distinction between the typical mixed theta frequencies of drowsiness and a single predominant theta frequency appearing in some subjects during meditation; such a distinction would be impossible to make with the band pass filtering method used in many of these studies and therefore such information would be missed.

The study by Morse et al (1977) had each subject perform three self-regulation techniques (meditation, self-hypnosis, and progressive relaxation), which could have allowed for "contamination" of one technique by the others, especially as there was only a three minute "rest" period between techniques. That study also allowed an extremely short time period for the practice of meditation and the other two techniques (6-8 minutes each), so that the development of the meditative or other state might have been compromised. In the Fenwick study the presence of slow rolling eye movements was taken as pathognomonic of drowsiness and no other EEG features were considered if they were present; thus potentially useful data were not examined.

Still another methodological problem is that of self-selection when experienced meditators are studied (Delmonte, 1984). People with certain EEG characteristics might be more drawn to meditation, and stay with the practice longer than others with different EEG characteristics. The lack of random assignment might mean that such innate differences are being found, rather than any differences due to the practice of meditation. One study did use random assignment of subjects who had signed up to learn TM. Lukas (1973) found no increase in alpha production during meditation after three months. Vassiliadis (1973), in a longitudinal study, found no difference at pretest between prospective meditators and controls, but reported increases in occipital alpha and in theta activity in meditators as compared to controls as a function of the length of practice of TM. There were no quantitative analyses in either study, however.

Taken together, strong EEG evidence for meditation as a "unique" state of consciousness is scanty at best, although the evidence against its being unique has problems as well.. But the frequent presence of alpha waves together with mixed theta frequencies in EEG research on meditation has made a number of authors remark that meditation seems to be a "holding on" to the transitional state between unequivocal wakefulness and true sleep (Elson, et al, 1978; Fenwick et al, 1977; Warrenburg et al, 1980). Stage 1 sleep is normally only a short transitional stage between wakefulness and definite sleep (stage 2 sleep), and normally lasts from less than a minute to 5 or 6 minutes, being followed by definite (stage 2) sleep.

Several studies have indicated that during meditation the EEG displays activity of relaxed wakefulness and stage 1 sleep, whereas during simple rest or relaxation EEG activity of deeper sleep stages is more common. Travis and associates (1976) found that the amount of occipital alpha present during TM meditation did not change significantly, while in a group of control subjects relaxing, occipital alpha activity decreased by 50%; four of 16 meditators exhibited stage 1 sleep activity; 13 of sixteen controls exhibited stage 2 or deeper sleep activity. Fenwick and colleagues (1977) found that meditators exhibited EEG activity of late stage 1 or stage 2 sleep when they were simply resting, but showed EEG activity of relaxed wakefulness or early stage 1 sleep while meditating. Corby and colleagues (1978) reported mostly EEG alpha (relaxed wakefulness) and some stage 1 sleep activity during meditation, with less than 1% of subjects showing stage 2 sleep activity. Elson and colleagues (1977) defined "descending alpha theta" as EEG activity containing more than 50% alpha or a predominance of theta activity that is typical of stage 1 sleep, and which is followed by stage 2 sleep within 5 minutes; they defined "non-descending alpha-theta" as similar EEG activity not followed by stage 2 sleep. They found significantly more non-descending alpha theta while meditators meditated than while controls relaxed; the controls either fell asleep or showed complete arousal with beta waves present. Six of 11 controls were found to have stage 2 sleep present, while none of the meditators showed stage 2 or deeper sleep. Warrenburg and associates (1980) reported an increase in stage 2 sleep from 4.5% to 29% from day 1 to day 2 during progressive relaxation by experienced practitioners, while experienced meditators showed a change from 2.3% to 0.9% stage 2 sleep from day 1 to day 2 of meditation.

A few studies have reported different findings, however. Two studies (Jevning et al, 1977, 1978) reported equal amounts of stage 1 (20% and 22%) and deeper stages (8% and 5%) during meditation in meditators and during rest in controls, respectively. Two studies (Pagano et al, 1976; Younger et al, 1975) have reported evidence of real sleep during meditation; however, the study by Younger and colleagues included stage 1 sleep in their estimation of sleep time, so it is difficult to determine how much was stage 1 and how much deeper stages of sleep. Pagano and colleagues reported 23% of meditation time spent in stage 2 sleep and 17% of meditation time spent in stages 3 or 4. This is the only study to report such large amounts of definite sleep during meditation, and the reasons for the discrepancy are not clear. One factor may be the longer meditation time (40 minutes); it is possible that maintaining what is normally a brief transitional state is difficult enough that subjects cannot do it for more than a relatively short period of time, after which they move into the deeper stages of sleep. The Corby et al study used a 40 minute meditation time and reported very little stage 2 or deeper sleep; however, they studied ananda marga yoga meditation, which seemed to produce more somatic arousal, and which they hypothesized aided in subjects' ability to remain in the transitional state and not slip off into sleep. Despite disagreements between the studies as to actual depth of sleep achieved during meditation, all have reported a preponderance of the earliest transitional stages, in which periods of high EEG alpha production are mixed with periods of EEG stage 1 activity (Rechtschaffen & Kales, 1968) and the large majority of studies have reported very little presence of stage 2 or deeper sleep. Thus there appears to be strong evidence that meditation is usually associated with the prolonged presence of EEG activity which is normally found only briefly during the transition between wakefulness and sleep, and that simply relaxing is more often associated with EEG activity of definite sleep or with alert wakefulness. Thus while somatic measures often do not appear to provide a means of differentiation between meditation and simply resting, EEG measures do.


The findings of a slowing of frequency of alpha waves, an increase in their amplitude, and the presence of theta waves, the spreading of alpha and theta activity to frontal areas of the brain, and the presence of stage 1 sleep EEG activity all point to a lowering of cortical (or brain) arousal, which can be conceptualized as another aspect of deep relaxation. However, in controls who simply rested, the transitional EEG state frequently was replaced by definite sleep, or alert EEG activity returned. These findings thus point to differences between meditation and simple rest or relaxation. Somatically, meditation appears to be a relaxation phenomenon; however, meditation also appears to provide a way to prevent, or at least retard, the onset of sleep when somatic arousal is lowered. The EEG findings in most such studies suggest that a low arousal, hypnogogic state of wakeful consciousness is maintained. Such a state may promote derepression of conflictual material, primary process mentation , and adaptive regression as discussed previously (Delmonte, 1985; Kutz et al, 1987). The EEG findings can also be conceptualized as evidence for a decrease in cognitive processing. The occipital areas normally show alpha waves when a person's eyes are closed, as no visual input is being received. Alpha waves are much less usual in frontal areas where more complex cognitive processing takes place, as cognitive activity is apparently more or less continuous during normal wakefulness. The increased incidence of alpha waves in frontal and central areas during meditation found in some of the studies could therefore reflect decreases in the cognitive activity which normally originates from these areas. Kelly's (1955) notion of a loosening of cognitive constructs during the hypnogogic state is compatible with these EEG findings. Thus the conceptualization of meditation as a "holding-on" to the transitional state between wakefulness and sleep is useful in terms of its ability to integrate our present psychological and physiological knowledge concerning meditation. Studies which further explore the relationship of meditation to the more common manifestation of that state, normal sleep onset, and to such transitional states as occur in simple rest, would therefore be important. Directly comparing different types of meditation would further enhance our understanding of the varieties of meditative technique.

The EEG studies cited above indicate that meditation is associated with lowered cortical arousal as compared with normal wakefulness, but greater cortical arousal as compared with sleep. Given that these EEG findings provide the most clear cut understanding of the physiological differences between meditation and rest, and the relation of meditation to more familiar states of consciousness (sleep and wakefulness), it is important to explore other physiological variables which have also been shown to be sensitive to shifts in levels of arousal. Neurophysiologically, the EEG activity during meditation reflects a particular state of activation of the ascending reticular activating system (ARAS), which is now thought to be the neurophysiologic substrate for gradations of attention and consciousness (Lindsley, 1959). At the same time there is strong evidence that the ARAS and the respiratory centers of the brain stem are functionally interrelated.


Breathing and meditation
Attention to respiration is often a central part of meditative practice. Respiration is a central theme in the religious philosophies and in the actual meditative techniques of both Yoga and Zen: "The principal points of Zen practice are formulated to the harmony of body, breath and mind" (Akishige, 1968, p. 135-136); "Prana is our very life, the absolute force which is everywhere. The breath is the external manifestation of prana" (Satchidananda, 1970, p. 5). In the practice of Zazen, or Zen meditation, posture and breathing techniques are emphasized; Yoga has several variants, and breathing is central to many of them. In raja yoga, for example, which emphasizes meditation per se, passive attention to the breath is prescribed. It is possible that attention to the breath in these disciplines might bring about physiological changes, perhaps at the level of central neural control of respiration.


Definitions of variables used in studies of respiratory neurophysiology:
Respiratory variables other than rate are not commonly studied in psychological research; therefore a short description of the variables included in the discussions that follow may be helpful. Alveolar carbon dioxide tension is the proportion of carbon dioxide found in the gases of the alveoli, the gas exchange portion of the lungs. It is in approximate equilibrium with arterial blood CO2 levels, which in turn are closely regulated by the respiratory centers of the brain, not varying more than approximately 0.5 mm Hg in an individual. Alveolar CO2 tension is therefore a reflection of central nervous system ventilatory drive. Even though it is regulated very closely within an individual, there is a considerable range across inidviduals; the normal range is 35 to 41 mm Hg.

Usually alveolar CO2 tension is expressed as Pco2, where P stands for partial pressure. A partial pressure is the fraction of total pressure which is contributed by one gas in a mixture of gases. The total pressure for physiological systems is atmospheric pressure. Pressures are usually expressed in units of millmeters of mercury (mm Hg), also known as Torr. Tidal volume is the volume of air in a normal breath. Minute ventilation is the volume of air breathed in a minute. Hypercapnia is the term used to denote increased CO2 in inspired air.


Respiration-EEG Relationships During the Wake-Sleep Transition
A functional interrelationship between the ARAS and the respiratory centers has been postulated by Bulow & Ingvar (1961) and Bulow (1963), based on their studies of respiratory parameters during sleep and wakefulness. With the transition from wakefulness to sleep, several authors have reported concomitant changes in tidal volume, alveolar carbon dioxide tension, and the sensitivity of the respiratory centers to carbon dioxide (Robin, et al, 1958; Birchfield, et al, 1959; Bulow & Ingvar, 1961; Bulow, 1963). In all studies tidal volume was found to decrease, alveolar CO2 tension to rise, and the CO2 sensitivity of the respiratory centers to decrease. Bulow (1963), in a careful and extended study, reported that during the transition from wakefulness to sleep (stage 1 EEG activity), even transient reappearance of alpha in the EEG was always accompanied by an increase in ventilation and a decrease in alveolar CO2 tension, the converse occurring as alpha dropped out again.

In a study by Naifeh and Kamiya (1981), alveolar CO2 was recorded along with EEG during the transition from wakefulness to sleep. Their results replicated the findings of Bulow; they reported that a rise in alveolar CO2 tension was closely associated with the disappearance of alpha and the appearance of the mixed frequency theta activity characteristic of stage 1 sleep. No such changes occurred when subjects lay quietly awake for an equal time period. The study thus showed that changes in alveolar CO2 are linked to changes in the EEG indicative of wakefulness and sleep onset.

Direct evidence for influence of the ARAS on respiration was obtained by Hugelin & Cohen (1963), who showed (in cats) that stimulation of the rostral ARAS produced changes in respiration along with the appearance of a wakeful pattern in the EEG. There was an increase in the amplitude and duration of the inspiratory (phrenic nerve to the diaphragm) discharge and an increase in the frequency of the respiratory cycle (all of which would produce an increase in ventilation and reduce CO2). They obtained identical results by means of sensory stimulation, indicating that the respiratory changes seen with sensory stimulation are probably due to ARAS activation through the extra-lemniscal sensory system.

Further evidence of the functional interrelationship between ARAS and the respiratory centers is the fact that stimulation of the respiratory centers in a sleeping subject by introducing a high percentage of CO2 into his inspired air (6% or higher) always produces arousal (Reed & Kellogg, 1958; Bulow, 1963). Thus respiratory center stimulation produces ARAS activation, as well as the converse.



Respiration During Meditation
The above discussion provides ample evidence that during the transitional state between sleep and wakefulness, changes in certain respiratory variables occur which are closely tied to changes in EEG activity indicative of the level of ARAS activation; this evidence demonstrates the functional interrelationship between the ARAS and the respiratory centers. The respiratory variables previously shown to be closely tied to the basic states of consciousness - sleep and wakefulness - have been studied only relatively recently in the meditative state, however. Respiratory rate, the most frequently studied respiratory variable in earlier meditation research, has not been found to change consistently during the sleep-wake transition (Bulow, 1963; Naifeh and Kamiya, 1981).

Recent studies have begun to provide data on the more interesting respiratory variables. Alveolar CO2 tension was found to increase significantly during meditation, but also during relaxation in controls;while minute ventilation was found to decrease during meditation but not control relaxation by Wolkove and colleagues (1984). Unfortunately their study had one rather major methodological problem: the control group consisted of physicians and technicians from the hospital staff, all of whom were much more comfortable with the intrusive apparatus (tight-fitting respiratory masks) as well as the CO2 sensitivity test (which produces a quadrupling of ventilation within a one minute period) than the experimentally naive meditators. There is evidence that extraneous stimuli and novelty of the situation can artifactually raise measures of ventilation and CO2 sensitivity (Gilbert et.al., 1972). The larger decrease in ventilation reported for meditation could therefore be contaminated by a larger response during baseline in the meditators. Indeed, the meditators did have a higher baseline CO2 sensitivity than controls.

A study by Kesterson and Clinch (1989), using experimentally naive controls as well as meditators, reported a significant decrease in CO2 elimination during meditation but not control relaxation, thus providing indirect evidence for an increase in CO2 tension during meditation. Wilson and colleagues (1987) reported significant increases in arterial CO2 tension during both meditation and control relaxation, with a more marked increase during meditation. They did not perform analyses to determine whether the increase during meditation was significantly higher than during relaxation, however. Nevertheless, the findings of all three studies point to a rise in CO2 tension associated with meditation and hypothesize that it is due to a reduction in central neural drive to breathing.

While some of the studies included EEG monitoring of a few subjects in order to determine whether the occurrence of any changes in respiratory measures were due to subjects falling asleep, none of them looked for the kind of connection between EEG and respiration that has been described during the normal sleep-wake transition. A study by Badawi and colleagues (Badawi, Wallace, Orme-Johnson & Rouzere, 1984) studied EEG during breath suspension periods in TM meditation, finding an increase in alpha power and decrease in theta and delta power. In sleep onset, breath suspension, while rare, occurs in some subjects in association with disappearance of alpha and appearance of stage 1 activity; thus the findings of Badawi and colleagues seem to be the opposite of the EEG-respiration connections of normal sleep onset. This one study aside, given the evidence for the similarity of EEG activity in meditation and in the sleep-wake transition, one might expect a similar connection between EEG and respiration. A fi
nding of similar close connection between EEG activity and these key respiratory variables during meditation would add further support to the conceptualization of meditation as a "holding-on" to the sleep/wake transition. Alternatively, a lack of such a connection could be considered evidence for neurophysiological differences between meditation and the sleep-wake transition. Direct comparisons, in the same subjects, of the normal wake/sleep transition and meditation, would provide even clearer information as to whether these respiratory and EEG connections are the same in both states. Comparisons of subjects who meditate with others who simply rest for an equivalent period of time would provide information on whether meditation differs from simple rest in terms of these respiratory-EEG connections. Finally, comparisons of subjects from different meditative traditions would provide information on whether different meditation techniques are associated with different respiratory-EEG connections. The present research provides these comparisons.

The general hypothesis to be tested in this study is that meditation is a process of maintaining what is normally a transitional state between sleep and wakefulness, and that respiratory as well as EEG data will support that hypothesis. Further, the reason for similarities between meditators meditating and nonmeditators resting, in terms of somatic measures of arousal, is that both are entering the same transitional state; however, meditators have learned to maintain it, whereas nonmeditators either slip off into sleep or wake up again; hence the variability in reported results comparing meditation to relaxation or rest. Specific hypotheses which are tested in this study are the following:

1) Meditators of various disciplines will show significantly more relaxed awake EEG (alpha predominant) and early stage 1 EEG activity during meditation than will nonmeditators during relaxation. Conversely, nonmeditators will show significantly more late stage 1/stage 2 sleep EEG activity during relaxation than will meditators during meditation.
2) In both meditators and controls, alveolar Pco2 will increase significantly in association with EEG stage 1 activity, but remain similar to baseline in association with awake EEG activity (alpha), during meditation and relaxation respectively.
3) Changes in alveolar Pco2 and respiratory movements, as outlined in 2) and 3) above, will be of similar pattern and magnitude in meditation/relaxation and sleep onset for both meditators and nonmeditators.
4) Changes in respiratory rate will be small, and similar in meditators and nonmeditators in both meditation/relaxation and in sleep.
5) Meditators from different meditation disciplines will not differ significantly from each other on any of the above measures.

METHODS

This study used data collected from 1978 to 1980 as part of a National Institute of Mental Health grant in which meditators from three different disciplines and nonmeditating controls were studied during meditation (or rest) and during sleep onset. EEG, alveolar CO2 tension, and respiratory rate were obtained.

Subjects
Eighteen intermediate and advanced meditators with 3-15 (mean =7) years experience, from three different disciplines: tantric Yoga (N=4), Zen Buddhism (N=7) and Tibetan Buddhism (N=7); and control subjects (N=8) with no meditation experience, participated in the study. The Yoga group performed an eye-closed meditation in which they silently recited a mantra or attended to "energy flow" within their bodies. Breathing was not specifically focused on. The Zen group performed an eyes-partially-open meditation in which they attended passively to their breathing or focused on "koans," the Zen meditative parables. The Tibetan group, chosen for the study by the Rinpoche, performed an eyes-closed meditation in which they focused on a mental image. The Zen group had six males, one female (median age 40, range 25-56); the Tibetan group had 3 males, 4 females (median age 45, range 32-58); the Yoga group had 1 male, 3 females (one of whom was assistant to the Yoga master) (median age 38, range 32-54); the control
group, consisting of graduate students and local community volunteers, had 4 males, 4 females (median age 42, range 24-60). All subjects were Caucasian. To partially control for the problem of self-selection of


TABLE 1.
GENERAL CHARACTERISTICS OF SUBJECTS
Years Meditation Median Age Gender
Group Experience & range M F
(mean ± S.D.) (years)

Control 0 42 (24 - 60) 4 4

Tibetan 8.7±4.5 43 (32 - 58) 3 4

Yoga 7.3±5.1 38 (22 - 54) 1 3

Zen 5.9±2.9 40 (25 - 56) 6 1

meditators (Schwartz, Davidson, & Goleman, 1980), control subjects were chosen who expressed an interest in meditation but had not meditated. Informed consent was obtained after the nature of the procedures had been fully explained. Table 1 presents general characteristics of all groups.


Procedures
EEG was recorded from F3A1, F4A2, C3A1, C4A2, O1A1 and O2A2 using gold cup electrodes and Grass electrode paste.

Alveolar Pco2 was measured by means of a 4 mm I.D. tube, inserted 5 mm inside one nostril and taped above the upper lip, which led a portion of the subject's expired air to a Beckman LB2 CO2 analyzer. The analyzer was calibrated against standard gases before and after each session. This procedure has been reported to be quite comfortable. A computer program computed in real time peak, or alveolar Pco2 for each breath and stored the data for later analysis. This is the accepted method among respiratory physiologists for measuring alveolar Pco2 (Collier, Affeldt & Farr, 1955). Only breaths with a definite plateau were accepted as reflecting alveolar Pco2. All subjects breathed through the nose, their accustomed manner of breathing.

The experiments were conducted in a dimly lit, sound-attenuated chamber. Meditators sat in their preferred meditation position for a 5 -minute eyes-open baseline period and a 30-35 minute meditation period, followed by another eyes-open 5-minute baseline period. Control subjects sat in a comfortable chair for a 5-minute eyes-open baseline period and a 30-35 minute period in which they were instructed to close their eyes and allow themselves to relax. Another 5-minute eyes-open baseline period followed the relaxation period. These positions, although not identical, were used because, for the meditators, sitting in a chair to meditate would have interfered with their accustomed form of meditation, whereas asking controls to relax for 30 minutes while sitting cross-legged on the floor with no back support would be painful and not conducive to relaxation. The study had been designed to gather meditation data on 3 separate days for each subject; however, due to scheduling problems, attrition, loss of data
due to computer problems, and termination of the grant, 3 days of data were available for only 5 Tibetan meditators, the 4 Yoga meditators, 3 Zen meditators and 3 controls. Two days and in a few cases only 1 day of data were available for the remaining subjects.

During sleep sessions, subjects lay on a mattress in the same experimental chamber. Following a 5-minute eyes-open baseline, subjects were instructed to try to go to sleep. They were awakened and the experiment concluded at the end of 30 minutes. Meditation and sleep sessions were run at approximately the same time of day to control for the effects of circadian rhythms. Data were available for one sleep session on all subjects, with the exception of 1 control subject.

Data Reduction
For scoring the EEG we have modified the criteria of Rechtschaffen and Kales (1968) such that an epoch was scored as "Stage A" if it contained 75% or more of alpha or mixed alpha and beta activity. An epoch was scored as stage 1 if it contained 25% or more of low-voltage, mixed frequency (2-7 cps) activity. This was done to achieve a more conservative criterion for scoring EEG as "awake." Stage 1 activity was further broken down for the EEG analysis section into Category B (25% to 75% mixed theta frequency, the rest alpha) and Category C (75% to 100% mixed theta frequencies). For the EEG analyses Stage A was referred to as "Category A." Stage 2 sleep EEG activity was scored according to the usual Rechtschaffen and Kales criteria. This is the stage which is characterized by short 13-14 Hz bursts in the EEG record (sleep spindles), and is accepted a sign of definite sleep. The scorer did not have access to the CO2 data and was blind to type of session.

For the respiratory analyses, each meditation, relaxation and sleep session were divided according to EEG category. For these analyses category B and category C were combined and considered stage 1. All breaths in all portions of each session falling into Stage A or Stage 1 categories, except for those excluded due to movement artifact, were used in the analysis. In the sleep sessions, only those in which the subject entered stage 2 sleep were analyzed. When the data were collected, meditators were asked if their meditation seemed satisfactory; relaxation controls if they felt they had been able to relax. One run each on two meditators (Tibetan group) was excluded because they did not feel able to meditate satisfactorily. The first relaxation run on two control subjects was excluded because subjects felt unable to relax. Repeated measures analysis of variance was used for the statistical analyses of all data. Level of significance of all results was taken as 0.05 or lower in a two-tailed test.



RESULTS

EEG Changes
All meditation subjects displayed periods of high EEG alpha activity during meditation, as did all controls during relaxation. Six of the seven Tibetan meditators, four of the seven Zen meditators and all four Yoga meditators also displayed a shift from EEG alpha to stage 1 activity at least once during each meditation session. All eight controls displayed similar shifts during all relaxation sessions. None of the meditators entered EEG stage 2 or deeper during a meditation session. Two control subjects entered stage 2 during relaxation sessions; none entered stages 3,4 or REM.

In the sleep sessions all subjects initially displayed periods of high EEG alpha activity. Fourteen of the eighteen meditation subjects and seven of the eight controls also entered EEG sleep stages 1 and 2 during at least one sleep session. One control subject entered stage 3.

Quantitative analysis of the meditation/relaxation EEG's was carried out as described in the Data Analysis section. The percent of the total meditation or relaxation period spent in each EEG category described previously was computed for each meditation (or relaxation) session for each subject. Data from all runs on each subject were pooled. The percent time spent in each category by each group was compared by a two-way, repeated measures analysis of variance (Group (Control, Tibetan, Yoga, Zen) x EEG category (A, B, C)). Post-hoc analyses were carried out using Bartlett's test.

Results of the analysis are presented in Figure 1. Virtually all of the EEG
---------------------------------------------------------------------------------------------------
Insert Figure 1 about here
---------------------------------------------------------------------------------------------------
activity of all subjects fell into categories A, B, and C. There was a significant EEG category by Group interaction (p<0.003), indicating that the groups differed in the percentages of EEG activity in the different EEG categories. In the post hoc tests, the Zen group was found to spend significantly more time in category A (range for all comparisons was p<0.02 to p< 0.001) and significantly less time in category B (range for all comparisons was p<0.02 to p<0.009) than the other groups. The controls spent significantly more time in category C than the Tibetan and Zen groups (range for all comparisons was p<0.05 to p<0.001). Thus the Zen group spent more time in clear wakefulness than the other groups, while the control group spent more time at the "sleep" end of the sleep-wake transition. Two control subjects showed stage 2 sleep activity; one spent 13% of the session in stage 2 sleep, the other 22%.


Carbon Dioxide
Figure 2 presents pre-meditation baseline, meditation EEG stage A and stage 1 (top); and pre-sleep baseline, sleep EEG stage A and stage 1 alveolar
Pco2 values (bottom) for all groups.

For quantitative analyses, an ANOVA which compared only the meditation groups was performed first to determine if meditation disciplines differed from each other. In this ANOVA EEG Category included the pre-test Baseline, Stage A and Stage 1; while Baseline is not an EEG category, it was included as the point of comparison for the 2 actual EEG categories. The 3-way, repeated measures ANOVA (Condition (meditation session vs sleep session) X EEG category (Baseline, Stage A, Stage 1) X Group (Tibetan, Yoga, Zen)) indicated that in the meditators Pco2 was different for meditation than for sleep, and that it
--------------------------------------------------------------------------------------------------------
Insert Figure 2 about here
--------------------------------------------------------------------------------------------------------
changed differently across EEG categories in meditation versus sleep, but that there were no differences between meditation groups on these measures. Consequently, all meditators were combined into a single meditation group for subsequent analyses.

A question arises at this point about the reason for the discrepancy between the EEG results, where the Zen group differed from the other two meditation groups, and the Pco2 results for the different EEG categories, which were similar in all 3 meditation groups. In fact there is no discrepancy: the Zen group had a much lower percentage of stage 1 and higher percentage of stage A EEG activity, but while each type of activity was present, the "behavior" of Pco2 was the same in the Zen group as in the other 2 meditation groups.
The Condition X EEG category X Group ANOVA using the control group and the combined meditation group showed a significant effect of Condition (p < 0.001), indicating again that overall, the meditation/relaxation sessions were different from sleep sessions in terms of Pco2. Additionally, there was a significant Condition by EEG Category interaction (p < 0.008), indicating that for controls and meditators combined, Pco2 changes across EEG categories differed in meditation sessions versus sleep sessions. Finally, there was a significant Group by Condition by EEG Category interaction (p < 0.02), indicating that Pco2 changes across EEG categories varied in meditation sessions versus sleep sessions differently for meditators than for controls.

The post-hoc paired t-tests run subsequently revealed that during relaxation in controls, alveolar Pco2 was significantly elevated over baseline values only in association with stage 1 EEG activity (p <.01). It remained similar to baseline while the EEG was scored as stage A (p =0.23). This pattern of change in Pco2 was similar to the sleep sessions of all subjects: Pco2 was significantly elevated over pre-sleep baseline values, for both meditators and controls, only in association with EEG stage 1 (meditators p < 0.001; controls p < 0.02). Pco2 was similar to pre-sleep baseline values during EEG stage A in meditators (p = 0.11) and controls (p = 0.23). Thus control subjects relaxing showed the same coupled pattern of EEG and Pco2 changes as were found during normal sleep onset in all groups.

In contrast, during the meditation sessions, alveolar Pco2 was significantly increased in meditators over baseline values during EEG stage A (p < 0.0001), as well as during EEG stage 1 (p < 0.0001). Thus meditators meditating did not show the same coupled pattern of EEG and Pco2 changes as is found in normal sleep onset; for them, Pco2 increased while stage A was present and stayed up when stage 1 EEG activity appeared. These findings are not consistent with the hypothesis that there will be a close connection between the pattern of EEG and alveolar Pco2 changes during meditation which is similar to that of the sleep-wake transition. The finding that alveolar Pco2 increased significantly during meditation while the EEG showed a relaxed wakeful (stage A) pattern was not anticipated.

In summary, meditators meditating were found to differ significantly from controls relaxing and from themselves falling asleep in terms of the pattern of EEG and Pco2 changes which occurred. During the wake-sleep transition in all subjects, and during relaxation in controls, there was a significant increase in
Pco2 only in association with stage 1 (drowsy) EEG activity. During meditation, on the other hand, there was as large an increase in Pco2 in association with stage A as with stage 1.

Respiration rate
Table 6 presents respiration rate values for all groups, EEG categories, and conditions. As with alveolar Pco2 analyses, an ANOVA which compared only the meditation groups was performed first to determine if meditation disciplines differed from each other.



TABLE 6.
RESPIRATION RATE ACROSS EEG CATEGORIES
breaths/minute mean (± S.D.)
-----------Meditation---------- -------------Sleep-------------
Group N Baseline Stage A Stage 1 Baseline Stage A Stage 1

Control 8 12.6 11.5 10.9 12.5 11.0 13.2
(1.0) (1.3) (0.8) (0.8) (1.2) (0.9)

Tibetan 7 12.0 8.9 11.5 11.8 12.3 11.3
(1.1) (1.4) (1.5) (1.0) (0.6) (0.9)

Yoga 4 13.6 12.1 11.0 15.0 13.6 13.6
(1.1) (1.2) (0.9) (1.3) (1.7) (1.1)

Zen 7 11.8 8.2 9.2 12.3 11.9 13.5
(1.2) (1.4) (2.1) (1.2) (1.1) (1.1)


The findings indicate that in the meditators respiration rate was different for meditation than for sleep, but that it changed similarly across EEG categories in meditation and sleep, and there were no