Meditation and somatic arousal reduction. A review of the experimental evidence.

The conceptual and methodological issues associated with research on the effects of meditation are reviewed. A summary of the research in which the somatic arousal of meditating subjects was compared to the somatic arousal of resting subjects did not reveal any consistent differences between meditating and resting subjects on measures of heart rate, electrodermal activity, respiration rate, systolic blood pressure, diastolic blood pressure, skin temperature, oxygen consumption, EMG activity, blood flow, or various biochemical factors. Similarly, a review of the research on the effects of meditation in controlling arousal in threatening situations did not reveal any consistent differences between meditating and nonmeditating (no-treatment, antimeditation, or relaxation) subjects. The implications of these findings for research and practice are discussed. Over the past 15 years there has been widespread interest in the practice of meditation, with the most publicized and popular technique being transcendental meditation (TM; Maharishi, 1963). It appears that many persons use meditation to reduce physiological and subjective arousal, and because of its purported effects on arousal, meditation is used to treat numerous disorders that stem from or involve hyperarousal. For example, meditation has been used to treat a variety of psychiatric disorders (Bloomfield, Cain, Jaffee, & Kory, 1975; Glueck & Stroebel, 1975), hypertension (Benson, Rosner, & Marzetta, 1973; Benson & Wallace, 1972; Blackwell et al., 1975; Michaels, Huber, & McCann, 1976; Simon, Oparil, & Kimball, 1977), asthma (Wilson, Honsberger, Chiu, & Novey, 1975), inflammation of the gums (Klemons, 1977), drug abuse (Benson & Wallace, 1972; Sham', Lavely, & Jaffe, 1974), alcohol abuse (Shafii, Lavely, &Jafte, 1975), insomnia (Miskiman, 1977a, 1977b), and stuttering (Mclntyre, Silverman, & Trotter, 1974). Furthermore, meditation has been suggested as an alternative to progressive muscle relaxation training (Boudreau, 1972). In view of the pervasive interest in the use of meditation to reduce arousal, it seemed important to carefully review the empirical research on the effects of meditation on somatic arousal, and therefore the present review was prepared. This review is divided into four sections. In the first section, conceptual and methodological issues and problems will be discussed, thus providing a perspective within which to evaluate the research. The second section will be focused on the question of whether subjects who meditate show lower somatic arousal while they are meditating than other subjects show while they are simply resting. The third section will be focused on the question of whether subjects who meditate show less somatic response to threat. In the fourth section overall conclusions will be drawn. Conceptual and Methodological Considerations Types of Investigations Case study. The research on meditation can be divided into three distinct groups. The first group contains case studies of the effects of meditation. Those accounts provide a rich source of hypotheses concerning the effects of meditation, but because they lack controls they cannot be used as empirical tests of the effects of meditation, and consequently they will not be considered here. Requests for reprints should be sent to David S. Holmes, Department of Psychology, University of Kansas, Fraser Hall, Lawrence, Kansas 66045. 1 It should be recognized that not all forms of meditation are designed to reduce arousal. Indeed, some types of meditation— such as Maulavi, the dancing practice of the "whirling dervishes"— involve considerable bodily activity and serve to increase arousal. The present review focuses on more passive forms of meditation that are designed to reduce arousal. For a discussion of the types of meditation, see Naranjo and Ornstein (1971). The influence of meditation on arousal has been examined in a number of previous reviews (Akishige, 1968; Davidson, 1976; Oellhorn & Kiely, 1972; Kanellakos & Lukas, 1974; Rigby, 1977; Schuman, 1980; Shapiro, 1980; Woolfolk, 1975). Unfortunately, the conclusions drawn in previous reviews generally cannot be accepted because the authors were selective in the investigations they cited; disregarded methodological problems in drawing conclusions from investigations; and/or indiscriminantly mixed results of case studies, uncontrolled investigations, and appropriately controlled experiments in drawing their conclusions. This manuscript focuses on the somatic responses associated with meditation, but is not intended to deny the wide variety of potentially important subjective experiences that have been atJanuary 1984 • American Psychologist Copyright 1984 by the American Psychological Association, Inc. Vol. 39, No. 1, !-10 Own-control. The second group contains investigations in which the own-control research design is employed (e.g., Bakker, 1977; Beary & Benson, 1974; Benson, Steinart, Greenwood, Klemchuk, & Peterson, 1975; Herbert, 1977;Janby, 1977;McCuaig, 1974; Wallace, 1970; Wallace & Benson, 1972; Wallace, Benson, & Wilson, 1971; West, 1977; Younger, Adriance, & Berger, 1975). In this type of investigation, subjects are first asked to simply sit quietly without meditating, then they are asked to meditate, and finally they usually are asked again to simply sit quietly. The preand postmeditation periods are used as "control" periods, and the levels of arousal evidenced during those periods are compared to the level of arousal evidenced during the meditation period. The results obtained with the own-control design have indicated consistently that subjects had lower arousal while meditating than they did before or after meditating, and those results have been generally accepted as evidence for the effectiveness of meditation for reducing arousal. It is important to recognize, however, that those investigations suffer from a serious limitation because they do not indicate whether meditation is more effective than other arousal-reducing strategies such as simply resting. Indeed, subjects who sit quietly and then rest may show the same decrease in arousal as subjects who sit quietly and then meditate. Because it is generally assumed that meditation results in different effects than does simply resting, and because those proposed differences have a variety of important theoretical and practical implications, it is essential to directly compare the effects of meditation to the effects of resting. Therefore, this review will not rely on data from investigations that employed the own-control approach. Experimental control. The third group contains investigations in which the experimental-control approach was used. With that approach, a group of subjects trained in meditation are asked to meditate, whereas another group of subjects not trained in meditation are asked to rest. Then the arousal levels of meditating subjects are compared to the arousal levels of resting subjects. This is generally the most tributed to meditation. Space limitations do not permit a consideration of all of the responses associated with meditation, and the somatic responses were selected for study because of their relevance for the treatment of the disorders noted previously. Readers interested in the research concerning the influence of meditation on subjective or cognitive responses should consult other reviews (e.g., Shapiro, 1980; Smith, 1975). 2 It might be noted that to overcome some of the limitations of the typical own-control procedure, a multiple-activity own-control procedure could be used in which, for example, subjects sat quietly, meditated, sat quietly, rested, sat quietly, rested, and sat quietly. Unfortunately, this procedure has not been used to study meditation, although an own-control with different activities on different days has been used (see the discussion of the research of Pagano, Rose, Stivers, & Warrenburg, 1976, in the next section). effective approach to studying the effects of meditation, and my review focuses on the results of experiments in which this approach was used. Before concluding this overview of approaches, I should note that some investigators have used a combination of the own-control design and the experimental-control design. In one case, the subjects participated in a sit/meditation/sit sequence on some days and in a sit/rest/sit sequence on other days, thus making it possible to compare meditation and resting within the same subjects (Pagano, Rose, Stivers, & Warrenburg, 1976). Unfortunately, if this approach did not yield differences between the meditation and rest days, it could be argued that on the rest days the meditatiors "slipped into" their meditative state. In another group of investigations, subjects' arousal levels were assessed while resting before they had been taught to meditate, and then three to four months later after the subjects had been taught to meditate their arousal levels were assessed while meditating (Jevning, Wilson, Smith, & Morton, 1978; Jevning, Pirkle, & Wilson, 1977; Jevning, Wilson, & Davidson, 1978; Jevning, Wilson, & VanderLaan, 1978). With that approach, however, the meditation versus rest comparison is confounded with factors such as time, history, laboratory experience, and so on, thus making the conclusions drawn from the data subject to alternative interpretations. Methodological Issues Although the experimental-control design is generally the best approach for studying the effects of meditation, many of the investigations in which that approach was used suffer from one or more potentially serious methodological problems, and some brief attention should be given to those problems before reviewing the data. Amount of training/experience with meditation. One potential problem revolves around the nature and amount of training and experience the meditating subjects had with meditation. Obviously, if the subjects in the meditation condition were not adequately trained or experienced with the technique, the comparison with the nonmeditating subjects would be meaningless. It has been asserted that almost anyone can learn the TM technique in only a few training sessions (Maharishi, 1963), but a number of investigations have revealed differences between novices and advanced meditators (cf. Jevning et al., 1977; Jevning, Wilson, & Davidson, 1978; Jevning, Wilson, Smith, & Morton, 1978; Jevning, Wilson, & VanderLann, 1978; Lang, Dehob, Meurer, & Kaufman, 1979). Certainly, if subjects with limited training or experience served in an experiment that did not reveal differences between meditating and resting subjects, the lack of a difference might be attributed to an ineffective manipulation of the independent January 1984 • American Psychologist variable (i.e., meditation). With regard to the training/ experience issue, it might be noted that in some investigations the meditators had as little as two or three weeks of experience (cf. Boswell & Murray, 1979; Parker, Gilbert, & Thoreson, 1978), whereas in others the subjects had between three and five years of experience or were qualified as teachers of TM (cf. Holmes, Solomon, Cappo, & Greenberg, 1983; Jevning et al., 1977; Jevning, Wilson, & Davidson, 1978; Jevning, Wilson, Smith, & Morton, 1978; Jevning, Wilson, & VanderLann, 1978). Subject selection. It is, of course, desirable to randomly assign subjects to conditions, but if subjects are randomly assigned it may be practically impossible to conduct and maintain the experimental manipulation (learning and practicing meditation) over the number of weeks, months, or years that may be necessary to assure that the meditators receive sufficient training and experience. Consequently, in the investigations that involved prolonged training or experiencej subjects were not randomly assigned to conditions. Instead, persons who earlier had voluntarily elected to become meditators and who had maintained the practice for some time were compared (a) to a matched group of nonmeditators, (b) to a randomly selected group of nonmeditators, or (c) to a group of nonmeditators who had recently elected to learn meditation but who had not yet learned or practiced the technique. Those procedures may introduce problems, because persons who elect to learn to meditate and who continue the practice for many years may differ in some ways from persons who do not elect to learn to meditate or who do not continue with meditation once it is learned. Analyses of data. The major statistical problem encountered in this body of research revolves around the failure of investigators to control for the influence of initial differences in arousal between meditating and nonmeditating subjects (i.e., the law of initial values; Lacey, 1956; Wilder, 1962). Initial differences between meditating and nonmeditating subjects can influence subsequent scores directly (e.g., in the absence of any change, subjects with lower initial arousal may appear to have decreased their arousal at subsequent times relative to subjects with higher initial arousal) and can inhibit the degree to which subjects can change their levels of arousal (e.g., ceiling and floor effects). The effects of initial differences are particularly pronounced with physiological measures, and it has been demonstrated that even nonreliable initial differences can create or obscure subsequent reliable differences (Kinsman & Staudenmayer, 1978). Simple difference scores are insufficient for controlling for the effects of initial differences, and therefore either covariance or residualization procedures must be employed (Benjamin, 1967; Cronbach & Furby, 1970). Unfortunately, these controls have been used only rarely. In the absence of those corrections, at a minimum it would be necessary to conduct a conditions (meditating subjects vs. resting subjects) by trials (premeditation/prerest period vs. meditation/ rest period) analysis of variance in which a conditions by trials interaction is predicted (e.g., greater reduction in arousal across trials for meditating than for resting subjects). Regrettably, in most cases the investigators simply compared the meditating and resting subjects during the meditation/rest period. Somatic Arousal During Meditation Having identified the types of investigations that have been conducted and the various potential methodological problems, in this section I will examine whether subjects who are meditating evidence lower levels of arousal than subjects who are resting. Table 1 contains a summary of the experimental findings concerning heart rate, electrodermal activity, respiration rate, blood pressure, EMG activity, skin temperature, oxygen consumption, and blood flow. The findings concerning biochemical factors could not be efficiently summarized in tabular form because so many substances have been examined, and therefore a summary of the experimental findings concerning biochemical factors is presented in a subsequent paragraph. The information in Table 1 and the paragraph on biochemical factors make it possible to examine the effects of meditation on any one response across experiments and to examine the effects of meditation across responses within any one experiment.