Biofeedback and Self-Regulation, Vol. 4, No. 4, 1979
Frontal Electromyographic Feedback Stress Attenuation and Generalization t William T. M c G o w a n University of South Carolina
Stephen N. Haynes 2 Southern lllinois University
C. Chrisman Wilson University of South Carolina
This study evaluated the effects o f one session of frontal electromyographic (EMG) feedback on (1) frontal EMG, (2) frontal EMG response to stress, (3) cardiovascular variables, and (4) cardiovascular responses to stress. Eighteen male and female undergraduate volunteers received either frontal E M G feedback or a relaxation instructions control procedure and were then exposed to a fear stimulus (visualization o f a feared situation) and a poststress adaptation period while several cardiovascular measures were monitored. In comparison to the control group, frontal E M G feedback significantly reduced resting levels o f frontal E M G and frontal EMG response to stress but had no significant effect on cardiovascular measures. The results o f this study suggest that one session o f frontal EMG feedback may attenuate response to stress but, within the paradigm utilized, may be confined to the specific muscle groups monitored. Additional areas o f needed research were noted including individual differences in generalization, the effects o f EMG feedback f r o m multiple sites sequentially and concomitantly, and the generalized effects f r o m symptom-specific sites.
'The authors would like to express their appreciation to Linda Gannon for her helpful comments on the manuscript. 2Address all correspondence to Stephen N. Haynes, Department of Psychology, Southern Illinois University, Carbondale, Illinois 62901. 323 0363-3586/79/1200-0323503.00/0 Q 1979 Plenum Publishing Corporation
324
McGowan, Haynes, and Wilson
The frontalis 3 area has been frequently selected as a recording site in electromyography (EMG) because of assumptions that it is a sensitive index of general level of relaxation and response to stress, and that it covaries with other indices of autonomic arousal (Stoyva & Budzynski, 1974; Goldstein, 1972). It has also been assumed that skeletal (particularly frontal) muscle tension plays a mediational role in physiological arousal (Wolpe, 1969; Jacobson, 1964). That is, changes in frontal EMG have been assumed to mediate changes in autonomically mediated responses. Decreases in frontal EMG level, for example, have been assumed to be associated with generalized decreases in sympathetically mediated responses such as blood pressure, heart rate, or peripheral vasomotor tone. Based on assumptions that frontal EMG is a sensitive indicator of response to stress and covaries with indices of autonomic arousal and that it can function as a mediator of physiological response to stress, frontal EMG feedback has been extensively employed as an intervention with a variety of psychophysiologic disorders such as headaches (Cox, Freundlich, & Meyer, 1975; Haynes, Griffin, Mooney, & Parise, 1975; Tasto & Hinkle, 1973; Wickramasekera, 1973), insomnia (Haynes, Sides, & Lockwood, 1977), asthma (Davis, Saunders, Creer, & Chai, !973), essential hypertension (Montgomery, Love, & Moeller, 1974), and generalized anxiety (Raskin, Johnson, & Rondestvedt, 1973). Although there are numerous methodological and procedural flaws in these studies (Blanchard & Young, 1974), results are consistent with the hypothesis that an intervention package involving frontal EMG feedback can be associated with clinically significant symptom reduction for some subjects. In contrast to the frequent clinical applications of frontal EMG feedback, there have been few laboratory investigations of its generalized effects on autonomically mediated physiological activity or on its ability to attenuate physiological responses to stress. Additional investigation of the specificity or stress attenuation effects of frontal EMG feedback may facilitate its appropriate clinical applications as well as delineating concomitant autonomically mediated physiological responses. If the effects of frontal EMG feedback are confined to the frontalis or nearby muscle groups, EMG feedback may be appropriate only with disorders such as muscle-contraction headaches, torticollis, tics, or neuromuscular disorders that involve relatively circumscribed muscle groups. If there are more generalized physiological effects of frontal EMG feedback, it may be an appropriate intervention strategy with psychophysiologic disorders involving physiological mechanisms other than skeletal muscle tension such as essential hypertension or insomnia. 3Thetermfrontalisis used in this paper with the understanding that musclegroups adjacent to the frontalisgroupsare involvedin "frontalis" EMG measuresand feedback.
EMG Feedback
325
The efficacy of frontal EMG feedback in facilitating reductions in frontal EMG level has been previously studied. Haynes, Moseley, and McGowan (1975) and Kondo, Canter, and Bean (1977), using different methodologies, found that frontal EMG feedback resulted in significantly greater reduction in frontal EMG activity than did no treatment or placebo treatment conditions. Alexander (1975) evaluated the effects of frontal EMG feedback on other muscle groups. He found that feedback-facilitated reductions in frontal EMG were not associated with reductions in EMG activity from the arm or leg. Results congruent with those of Alexander were reported by Shedivy and Kleinman (1977). Shedivy and Kleinman found insignificant correlations between the frontalis and semispinalis/splenius muscle groups during conditions of relaxation and tension. However, none of these studies evaluated the effect of frontal EMG feedback on autonomically mediated responses or on response to stress. The present paper reports on a study assessing (1) the effects of frontal EMG feedback on several cardiovascular measures (heart rate, peripheral blood flow, peripheral temperature), (2) the effects of frontal EMG feedback on frontal EMG activity during stress, and (3) the effects of frontal EMG feedback on cardiovascular activity during stress.
METHOD
Subjects Subjects were 18 male and female undergraduate students (ages 1 8 - 22) enrolled in introductory psychology courses. While in their classes, all potential subjects were administered the Fear Survey Schedule II (Geer, 1965). On this questionnaire, subjects indicated their level of fear (on a 5point scale) to a number of specific stimuli and situations. Questionnaires were later analyzed to identify fears frequently reported by subjects. Subjects who reported intense fears (5 on the 5-point Fear Rating Scale) on frequently reported items (e.g., falling, snakes, rats) were selected for the experiment. All subjects were fully informed of the experimental procedures prior to participation.
Stressor The most commonly used stressors in the psychophysiology literature involve physical discomfort, such as cold pressor or electric shock. In the present study visualization of a feared stimulus was used as the stressor be-
McGowan, Haynes, and Wilson
326
cause it was assumed to be a closer approximation to the "psychological" stressors normally encountered in the natural environment. Visualization of feared stimuli has been previously shown to elicit self-report and physiological indices of arousal (May & Johnson, 1973; Craig, 1968; Grossberg & Wilson, 1968). Each subject was presented with a prerecorded description of a feared stimulus that matched a high fear item on the Fear Survey Schedule II. Thus, only subjects reporting a high fear on one of the prerecorded fear descriptions were utilized in this study.
Group Assignment Prior to the experiment, subjects were matched on the basis of sex and type of fear reported for a high fear item and randomly assigned to either a frontal E M G feedback group or a relaxation instructions control group. For example, half of the subjects with a high fear of falling were in the feedback group and half were in the control group; half the subjects with a high fear of snakes were in the feedback group and half in the control group.
Target Responses and Recording Procedures Heart Rate. Heart rate (HR) was recorded on a Grass Model 7B polygraph and tachograph. Using Beckman electrode paste, silver-silver chloride electrodes were attached to the right wrist and left leg with a ground on the right leg. Heart rate was recorded in beats per minute. Blood-Volume Pulse. Blood-volume pulse (BVP) was monitored on a Grass Model 7B polygraph using a Grass Model P T T I photoelectric transducer attached to the distal segment of the right index finger. Because absolute measures of BVP are difficult to derive, BVP change scores throughout the experiment were calculated on the basis of the baseline levels. Mean "valley to p e a k " measures of pulse amplitude were taken during the first 60 seconds of baseline. Change scores for subsequent measurement periods were calculated on the basis of the baseline average. Skin Temperature. Skin temperature was monitored with a Biofeedback Technology Model 301 Feedback Thermometer and Yellow Springs Thermistor lightly taped to the ventral side of the distal segment of the left index finger. Temperature was recorded in degrees Fahrenheit. Frontal EMG. Frontal E M G was monitored with a Biofeedback Systems BIFS B-1 utilizing standard amplification filtration, band-pass and noise-cancellation techniques. The forehead was scrubbed with acetone and
EMG Feedback
327
three silver electrodes (2cm 2) were centered on the forehead 2.4 cm apart and 2.4 cm above the eyebrows. Frontal E M G levels were integrated across 64 seconds (,~V/mins) with a 20-second intertrial interval.
Procedure Preliminary Preparation. All subjects were administered the Fear Survey Schedule II while in their respective classes. After selection of subjects on the basis of reported fear, all subjects were contacted by phone and the procedures thoroughly explained; appointments were made with those who volunteered. Subjects were then matched as previously described and assigned to groups. During the experiment, subjects were seated in a climate-controlled, sound-attenuated experimental room (2.5 m x 3.0 m). Subjects sat in a comfortable recliner chair while a tape-recorded message reiterated the nature of the experiment, informed them of the measurement procedures, and reassured them of their safety. The experimenter then entered the room, answered any questions, attached the electrodes and transducers, and placed earphones over the subject's ears. The lights were dimmed and the experimenter left the room. Interaction between experimenter and subject was minimized to reduce the possibility of bias. Baseline. After the experimenter left the room, another taped message was played that instructed subjects to close their eyes, relax, remain as still as possible, and await further instructions. The purpose of this phase was to allow for adaptation to the experimental conditions and provide baseline data from which to evaluate subsequent conditions. The baseline phase lasted 14 minutes? Relaxation L Following the 14-minute baseline phase, a taped message was played that explained to the subjects that during the next phase of the experiment they were to attempt to become as relaxed as possible. Instructions to relax were identical for both groups except that subjects receiving frontal E M G feedback were given additional instructions about the functioning of the feedback tone (Haynes et al., 1975). Subjects in this group were informed that the tone would provide information about their degree of relaxation; they were not informed that the tone reflected frontal E M G activity. Subjects in the feedback group heard a soft variable-frequency tone through earphones whose pitch varied directly with their level of frontal E M G activity. The control group was instructed to become as
~Previous research in this laboratory suggested that 14 minutes of adaptation resulted in minimal variabilityand change across time in physiologicalmeasures.
328
McGowan, Haynes, and Wilson
relaxed as possible but received no further assistance. The intervention phase lasted 10 minutes? Visualization L Following the 10-minute relaxation phase, the variable frequency tone to the feedback group was stopped and all subjects heard a taped message explaining that they were to visualize, as clearly as possible, a scene derived from their responses on the Fear Survey Schedule but to remain as relaxed as possible while doing so. A tape-recorded description of the feared stimulus was then played to each subject. The taped description and visualization lasted approximately 20 seconds and subjects continued visualization for an additional 20 seconds? To illustrate, the following is the taped arousal message for subjects with a fear of heights: " I want you to imagine that you are looking out the open window of the top story of a tall building. As you are looking down at the street hundreds of feet below, you slip and begin to fall out the window. You barely catch yourself and you are dangling out the window while trying to hold on the windowsill. You look down and see the tiny people below and feel your grip on the window beginning to slip." Adaptation L Following the 40-second presentation and visualization of the stressor, subjects were instructed to stop imagining the scene and to continue to relax as much as possible. The adaptation phase lasted for 6 minutes. Replication. Immediately following Adaptation I, Relaxation, Visualization, and Adaptation conditions were repeated exactly as outlined in the previous sections. These phases are indicated as "Relaxation I I , " "Visualization I I , " and "Adaptation I I " on the tables and illustrations. Debriefing. After the last adaptation phase, a taped message informed the subjects that the experiment was over. The experimenter entered the room, removed electrodes and transducers, and answered any questions asked by the subject.
Data Sampling Physiological measures were recorded continuously throughout the session. For the purpose of statistical analysis, however, the data were sampled only at strategic points. To facilitate interpretation of results, datasampling periods for baseline, relaxation, visualization, and adaptation phases are presented in Table I. Heart rate and blood-volume pulses measures were derived from the average of 18 heartbeats at the points ~Previous research in this laboratory (Haynes, Moseley,& McGowan, 1975)suggested that 10 minutes of frontal EMG feedback was sufficient to effect significant decreases in frontalis EMG activity. ~Visualizationwas continued for an additional 20 seconds after scenedescriptionbecause pilot work with this technique suggestedthat some subjects required more than 20 seconds to generate a clear imageand demonstratephysiologicalarousal.
~
o
<
~
~
o
o~.~ o
~
+.~
~ o'-'n
o <
~.~ ~.~ <
> o
o'~
0
0 I
~,---t
<
0
m
¢~
r~ 0
•<
.<
<
.<
°
[.. .~ o
o~.
~
~'n
o
330
McGowan, Haynes, and Wilson
indicated in Table I and the 18th beat always coincided with the end of an E M G integration period. It should be reiterated that each subject was exposed to two Relaxation, Visualization, and Adaptation phases and the sampling points for the replication phases were identical to those of the initial phases.
RESULTS
Presented in Table II are the mean heart rate, temperature, and EMG measures across each phase for both groups. Because of the relative nature of blood-volume pulse data and their failure to demonstrate any significant effects, they were omitted from Table II. Data were analyzed by repeatedmeasures analysis of variance and slope (time series) analysis (Glass, Willson, & Gottman, 1975). 7
Group Differences in Baseline A between-groups analysis of variance was performed on all four dependent measures derived during baseline (sampling points 1 and 2). The analysis revealed no statistically significant differences between graups on any of the four measures.
Effect o f Frontal EMG Feedback on Frontal EMG A repeated-measures analysis of variance revealed significant differences between groups on frontalis E M G levels (F(7/112) = 2.22, p < .05) across phases. Post hoc analyses (Duncan-Kirk, 1968) revealed significant differences between groups during Visualization I (sampling point 6), following Relaxation II (sampling point 8), during Visualization II (sampling point 11), and following Adaptation II (sampling point 12). At those points, the mean frontal E M G level of the E M G feedback group was significantly lower than that of the Relaxation instructions control group. The mean frontal EMG level of the feedback group was lower than that of the control group at all other sampling points but the differences did not attain statistical significance. Figure 1 illustrates E M G activity for both groups across phases. A slope analysis addressed the question of whether the overall slopes of the two sets of plots were significantly different. An
~Slope analysis was carried out by a computer program based on the work of Gottman and Glass developedby the first author.
EMG Feedback
331
©
.,9, °
o
o~
._o o~
J
:Z
~=~
,..,._o 0
~
~ ,...o ,~
0
~
o=
332
McGowan, Haynes, and Wilson I
I
I
RELAXATION
I
I
BASELINE ~' ID Z
I
I
ADAPTATION
I
VISUALIZATION
I
81.
z
I
VISUALIZATION
I
I
I
RELAXATION
I
ADAPTATION
"IT
]I
8O" 79" 78
ffl 77' I~ 76' 75' ILl P <
74" 73"
¢
72, 71, 70'
a" .~ I
69" 68"
EMG FEEDBACK I =
67"
I
I
1
2
I
I
-
= NO EMG FEEDBACK
I 3
I
=
:
~
4
5
15
I
MEASUREMENT
I 7
I
I~
I ",ll
I
I 8
I
I
I
I
I
9
10
11
I
t 1'2
PERIOD
Fig. 1. Mean integrated frontalis EMG for the frontalis EMG feedback and relaxation instructions groups across all sampling points.
analysis of the slopes of the plots in Figure 1 also revealed significant differences between the two groups and confirmed the results of the repeated-measures ANOVA.
Effects o f Frontal EMG Feedback on Cardiovascular Variables To assess the effects of frontal E M G feedback on cardiovascular measures, the two groups were compared on heart rate, BVP, and finger temperature. Between-groups repeated-measures analysis of variance revealed no significant differences on measures of heart rate, finger temperature, or blood-volume pulse. Slope analysis of points over time for the cardiovascular measures also suggested that the feedback and control groups were not significantly different on any of the cardiovascular measures. For illustrative purposes, heart rate data are presented in Figure 2.
Stress Manipulation To evaluate the effectiveness of visualization of a feared stimulus as a stressor, separate repeated-measures ANOVAs were calculated between
EMG Feedback
333
I
E
BASELINE I
I .,. 12•_. 11-
<
9-
8"
-
T
~
54-
tU ~;
I
I
I
,
i
I
I
I
i /
I
I I
I I
~r
I
I
I
I
I
,
I
I
I
I
I~
ADAPTATION
]I
VISUALIZATION
I
/=,,,
'
I
RELAXATION
ADAPTATION
I
TI"
I
y
I
/-+-,,,'S-"
7"
E (9
i
I
RELAXATION I
~
I
VISUALIZATION
I
I
I
I I
3-
[
e
-" EMG
=
= NO
FEEDBACK
I
EMG FEEDBACK
I
I
I
m
I
I
1
2
[
I 3
I
I
I
I
4
5
6
I
MEASUREMENT
I 7
I
I 8
I
I
I
I
9
10
11
I
I 12
PERIOD
Fig. 2. Mean heart rate for the frontalis EMG feedback and relaxation instructions groups across all sampling points.
postrelaxation measures and measures at the termination of visualization for each group and for each dependent variable. Results suggested that the feared stimulus visualization was associated with a significant increase in frontal EMG (F(1/16) = 5.67, p < .05) and heart rate (F(1/16) = 8.66, p < .05) for the feedback group, significant increases in frontal EMG (F(1/16) = 5.15, p < .05) and heart rate (F(1/16) = 8.09, p < .05) for the relaxation instructions control group, and no significant changes in measures of peripheral vasomotor activity (finger temperature, BVP) for either group. Correlations among Dependent Measures
To further assess the relationship among EMG, finger temperature, HR, and BVP, the four dependent variables were subject to correlation analyses (Pearson). The correlation matrix, presented in Table III, was based on the eight sampling points common to all dependent variables (sampling points 1, 2, 3, 6, 7, 8, 11, and 12) for all subjects. Because the data were nonindependent (six data points from each subject), confidence levels are not presented. Examination of Table III, however, reveals that the correlations among the four dependent measures were small.
334
MeGowan, Haynes, and Wilson
Table lII. Matrix of Correlations among Dependent Variables for All Subjects
EMG Finger temperature Heart rate
EMG
Finger temperature
Heart rate
Blood-volume pulse
-
-.09 -
-.06 .03 -
-.17 .37 .07
DISCUSSION The functions of the study were to assess the effects of one session of frontalis E M G feedback on response to stress and on cardiovascular variables. The results of the study suggested that one session of frontal E M G feedback effected a reduction in the level of frontal E M G significantly more than instructions to relax and attenuated frontal E M G response to stress significantly more than instructions to relax; frontal E M G feedback and instructions to relax were not significantly different in their effect on cardiovascular measures during feedback or in response to stress. The finding that one session of frontal E M G feedback can facilitate a reduction in frontal E M G level more than a relaxation instructions control procedure is consistent with the finding of Haynes, Moseley, & M c G o w a n (1975), Alexander (1975), and Kondo et ah (1977). In the present study, however, it was also demonstrated that frontal E M G feedback could attenuate frontal E M G response to stress to a significantly greater degree than a relaxation instructions control procedure. Although the effects m a y be limited to the particular parameters utilized in this study (e.g., type of feedback, type o f stressors), the findings are consistent with the hypothesized stressattenuation qualities of frontal E M G feedback. Although stress stimuli were presented to subjects in the absence of the feedback tone, it is not clear whether observed differences between the two groups in frontal E M G response to stress were a function of voluntary control learned in the preceding training phases or a function of different prestress levels of frontal E M G activity. The slopes of frontal E M G activity between relaxation and stress conditions were not significantly different between the two groups (although the means were significantly different), suggesting that lower prestress levels m a y account for the difference between the two groups. It might be predicted f r o m the " l a w of initial values" (Sternbach, 1966), however, that the feedback group would demonstrate a greater increase in frontal E M G level because their prestress frontal level was lower. A factorial study with varying levels of prestress frontal E M G would shed additional light on the mechanisms involved. Regardless of etiological mechanisms, however, these findings provide
EMG Feedback
335
empirical support for the application of EMG feedback in the modification of disorders with specific muscle-function etiology. The finding that one session of frontal E M G feedback did not effect a reduction in cardiovascular activity greater than that produced by an instructions-only control condition suggests that the effects of frontal E M G feedback may be confined to the specific site being monitored. The finding of. specificity of effect is consistent with that reported by Alexander (1975) and suggests that caution should be exercised in the application of frontal E M G feedback to disorders without specific muscle-function etiologies. It is impossible, of course, to rule out the possibility that failure of the cardiovascular measures to covary with frontal E M G may have been a function of the particular experimental parameters utilized in this study. As suggested in a recent article by DeGood and Chisolm (1977), additional training sessions may have resulted in greater response generalization. DeGood and Chisolm found indications of response generalization (changes in heart and respiration rate) accompanying four sessions of frontal E M G feedback. If frontal E M G feedback does not have generalized effects on autonomically mediated responses, it is difficult to account for the successful application of frontal E M G feedback to disorders that do not have specific muscle-function etiologies. One possibility is that elements in the treatment packages, other than frontal E M G reduction, account for the clinical effects. Clinical effectiveness may be accounted for by expectancy, placebo, demand, or cognitive factors (Haynes, 1978). This hypothesis is supported by research with insomniacs, which has suggested that insomniacs may not be more physiologically aroused than noninsomniacs at bedtime and therefore biofeedback intervention programs that have as their aim the reduction of physiological arousal may be effective for a different reason from that originally hypothesized. In addition to a factorial study varying level o f prestress activation, additional research on the generalized effects of E M G feedback might address issues of (1) individual differences in generalized effects, (2) the effects of E M G feedback from multiple sites concomitantly, (3) the effects of E M G feedback from multiple sites sequentially, (4) the generalized effects of E M G feedback from other individual sites, (5) the generalized effects of E M G feedback from symptom-etiological sites (e.g., jaw muscles for bruxism, back of neck for tension headaches). REFERENCES
Alexander, A. B. An experimentaltest of assumptions relating to the use of electromyographic biofeedback as a general relaxation technique. Psychophysiology, 1975, 12, 656-662. Blanchard, E. B., & Young, L. D. Clinical applications of biofeedback training. Archives of General Psychiatry, 1974,30, 573-589.
336
McGowan, Haynes, and Wilson
Cox, D., Freundlich, A., & Meyer, R. Differential effectiveness of electromyographic feedback, verbal relaxation instruction and medication placebo with tension headaches. Journal of Consulting and ClinicalPsychology, 1975, 43, 892-898. Craig, K. Physiological arousal as a function of imagined, vicarious and direct stress experiences. Journal ofAbnormalPsychology, 1968, 73, 513-520. Davis, W. H., Saunders, D. R., Creer, T. L., & Chai, H. Relaxation training facilitated by biofeedback apparatus as a supplemental treatment in bronchial asthma. Journal of Psychosomatic Research, 1973, 17, 121-128. DeGood, D. E., & Chisolm, R. C. Multiple response comparison of parietal EEG and frontalis EMG biofeedback. Psychophysiology, 1977, 14, 258-265. Geer, J. H. The development of a scale to measure fear. Behaviour Research and Therapy, 1965, 3, 45-53. Glass, G. V., Willson, V. L., & Gottman, J. M. Design and analysis of time-series experiments. Boulder: Colorado Associated University Press, 1974. Goldstein, I. B. Electromyography: A measure of skeletal muscle response. In N . S . Greenfield & R. A. Sternback (Eds:), Handbook ofpsychophysiology. New York: Holt, Rinehart & Winston, 1972. Pp. 329-366. Grossberg, J., & Wilson, H. Physiological changes accompanying the visualization of fearful and neutral situations. Journal of Personality and Social Psychology, 1968, 10, 124-133. Haynes, S. N. Principles of behavioral assessment. New York: Gardner Press, 1978. Haynes, S. N., Gri.~fin, P., Mooney, D., & Parise, M. Electromyographic biofeedback and relaxation instructions in the treatment of muscle contraction headaches. Behavior Therapy, 1975, 6, 672-678. Haynes, S., Moseley, D., & McGowan, W. T. The comparative effectiveness of EMG biofeedback and relaxation instructions. Psychophysiology, 1975, 12, 26-35. Haynes', S. N., Sides, H., & Lockwood, G. Frontalis electromyographic feedback and relaxation instructions treatment of insomnia. Behavior Therapy, 1977, 4, 642-652. Jacobson, E. Anxiety and tension control: A physiological approach. Philadelphia: Lippincott, 1964. Kirk, R. Experimentaldesign. Belmont, Ca.: Brooks/Cole, 1968. Kondo, C. Y., Canter, A., & Bean, J. A. Intersession interval and reductions in frontalis EMG during biofeedback training. Psychophysiology, 1977, 14, 15-17. May, J. R., & Johnson, H. J. Physiological activity to internally elicited arousal and inhibitory thoughts. Journal of Abnormal Psychology, 1973, 82, 239-245. Montgomery, D. D., Love, W. A., & Moeller, J. A. Effects of electromyographic feedback and relaxation training on blood pressure in essential hypertensives. Paper presented at the fifth annual meeting of the Biofeedback Research Society, Denver, 1974. Tasto, D., & Hinkle, J. Muscle relaxation treatment for tension headaches. Behaviour Research and Therapy, 1973, 11, 347-349. Ruskin, M., Johnson, G., & Rondestvedt, J. W. Chronic anxiety treated by feedback-induced muscle relaxation: A pilot study. Archives of General Psychiatry, 1973, 28, 263-267. Shedivy, D. I., & Kleinman, K. M. Lack of correlation between frontalis EMG and either neck EMG or verbal ratings of tension. Psychophysiology, 1977, 14, 182-186. Sternbach, R. A. Principles ofpsychophysiology. New York: Academic Press, 1966. Stoyva, J., & Budzynski, T. Cultivated low arousal: An antistress response? In L. DiCara (Ed.), Limbic and autonomic nervous system research. New York: Plenum, 1974. Wickramasekera, I. The application of verbal instructions and EMG feedback training to the management of tension headaches: Preliminary observations. Headache, 1973, 13, 74-76. Wolpe, J. The practice of behavior therapy. New York: Pergamon Press, 1969. (Revision received May 21, 1979)