Biological Psychology 5 (1977) 221-231 O North-Holland Publishing Company
INSTRUMENTAL CONDITIONING OF HUMAN HEART RATE DURING FREE AND CONTROLLED RESPIRATION D.H. V A N D E R C A R * ' t , M.A. FELDSTEIN, and H. SOLOMON Department o f Physiological Psychology, Rockefeller University, NYC, New York 10021, U.S.A. Accepted for publication 31 March 1977
The effect of respiratory constraint on heart rate control was assessed in a biofeedback situation with feedback consisting of changes in both illumination level and intensity of a prerecorded baby's cry. Subjects were reinforced for alternately increasing and decreasing heart rate on each of seven days during which respiration was unconstrained (Phase 1). This phase was followed by eight days when respiration was constrained during training sessions with a control respirator (Phase 2). Seven additional days of training followed in the unconstrained situation (Phase 3). Results indicate that the control of heart rate in biofeedback situations is very closely related to respiratory and other somatic activity. The implication of these findings for the field of visceral control is discussed.
1. Introduction In recent years, many investigators have reported success in operantly conditioning heart rate in human subjects (Brener and Hothersall, 1966; Engel and Hansen, 1966; Engel and Chism, 1967; Brener, 1974). Although these results have been encouraging, a basic question remains as to whether the changes recorded represent true operant conditioning o f autonomic responses or whether the heart rate changes are mediated by the operant conditioning of skeletal responses (Katkin and Murray, 1968; Crider, Schwartz and Schnidman, 1969; Katldn, Murray and Lachman, 1969). Another question that has been raised is whether biofeedback training leads to the acquisition o f control over heart rate or whether the biofeedback experiment only assesses a subject's pre-experimental ability to control heart rate (Levenson, 1976). The present study was designed to investigate the acquisition of heart rate con-
* Now at University of South Florida. "~ Address requests for reprints to: David H. VanDercar, Ph.D., Dept. of Psychology, University of South Florida, Tampa, Florida 33620. 221
222
D.H. VanDercaret al. / Instrumental conditioning of human heart rate
trol, the extent to which subjects can establish control over heart rate when provided with feedback, and the extent to which somatic responses such as muscular activity, and rate and depth of respiration are involved in these heart rate changes. In order to answer these questions, the present study examined heart rate, electro° myographic activity, and respiratory responses during multiple feedback training sessions. In some of these sessions the subjects were able to control respiration rate and depth, whereas in other sessions the subjects were passively respirated and unable to influence respiration rate or tidal volume. Although demand and control respirators are routinely used in medical settings, their use for controlling respiration in biofeedback research is unique.
2. Method 2.1. Subjects
Six normal female subjects between the ages of 18 and 30 years were paid $ 3.00 per hour for participating in this study. All subjects were given brief examinations by a physician and found to be free of obvious respiratory or cardiovascular disorders. 2.2. Apparatus
Electrocardiogram (ECG), heart rate, electromyogram (EMG), respiration rate, depth of respiration, and the direction of change attempted during trial periods were recorded on a Grass model 7 polygraph. In addition, integrated EMG was converted to pulses by coupling the output of the polygraph to a voltage-to-frequency converter which, together with heart rate, were recorded on a printout counter each minute during baseline and conditioning periods. The ECG was recorded from a lead II configuration with Beckman electrodes and telemetered via a Narco BioSystems model E2 transmitter. Respiration rate and depth of respiration were monitored with a chest bellows placed around the chest and connected to a Grass PT 5 pressure transducer. The EMG was recorded and telemetered from the sternocleidomastoid muscle of the neck and displayed as both raw EMG and integrated activity. EMG recordings were not taken from: (a) frontalis muscles because of potential contamination from eye movement responses associated with attending to visual cues, or (b) chin muscles, which would have been adversely affected by the respirator mask. All recording equipment, together with BRS circuitry for programming trials, was located outside a sound-attenuated chamber (Industrial Acoustics Corporation). A modified dimmer switch coupled to the output of the Grass cardiotachometer was used to provide continuous visual and auditory heart rate feedback to the subjects. Feedback consisted of a tape recording of a baby's cry and a 100 watt light
D.H. VanDercaret al. /Instrumental conditioning of human heart rate
223
bulb, both of which varied in intensity as a function of the subject's immediate heart rate. The feedback circuitry was arranged so that the experimenter could select either an increase or decrease in heart rate as the correct response for producing an attenuation of the intensity of both the baby's cry and the light. The use of an ecologically significant stimulus (i.e., crying) was intended to facilitate training by providing reinforcement as well as informational content to the subject. In the second phase of this experiment a Bennett MA1 respirator was used to respirate subjects at constant rate and tidal volume. The Bennett respirator can be used in either a demand or control mode. In the demand mode the subject initiates respiration but the respirator provides a fixed volume of air. In the control mode, which was used in the present study, the experimenter can set respiration rate, tidal volume, and inspiration-expiration ratio. During the present study the inspirationexpiration ratio was fixed at 1 : 1 for all subjects, but the respiration rate and tidal volume were adjusted by each subject. Within several adaptation sessions the subjects stabilized at values approximating their norms for weight and age (Consolazio, Johnson and Pecoris, 1963). 2.3. Procedure
Subjects were conditioned while reclining on a comfortable cot located within a sound-attenuated chamber. After all leads had been attached for recording heart late, EMG, and respiration, subjects were informed that the purpose of this study was to determine whether they could voluntarily control their heart rate. Each subject was instructed to remain relaxed and not to attempt to influence heart rate by changing breathing or engaging in muscular activity. They were also told to pay close attention to the feedback which would be provided, and to engage in any 'mental activity' which helped in controlling heart rate. In addition to feedback, the subjects were presented with either a red or green cue light and were told that, whenever the red light was on, they were to attempt to lower their heart rate, and conversely, increase their heart rate when the green light was presented. They were further informed that, as they succeeded in changing their heart rate in the appropriate direction, the intensity of both the baby's cry and the light would decrease. 2.3.1. Phase 1
Each subject received seven days of conditioning, during which no attempt was made to control respiration other than through instructions. Each daily session lasted 20 min and began with a 4-min baseline period during which the cue lights and feedback were absent. The subsequent 16 min consisted of four 4-min periods during which time the red and green lights were alternately presented. On each successive day of conditioning, the cue light which began a session was alternated. 2.3.2. Phase 2
Following Phase 1, all subjects underwent several sessions designed to acclimate
1 2 3
1 2 3
1 2 3
1 2 3
1 2 3
Subject 2
Subject 3
Subject 4
Subject 5
Subject 6
2.5 0.7 3.0
9.1 4.1 7.6
3.1 0.81 0.8
2.3 0.12 1.5
4.0 3.1 1.6
6.2 0.4 1.9
0.05 >0.05
0.05
. 0.05 >0.05
HR-EMG
Table 1 Difference scores (d) and their correlations (r) be tw e e n increase and decrease periods for heart rate (HR), e l e c t r o m y o g r a p h i c activity (EMG), respiration rate (RR), d ept h (RD), and pressure (RP), during norma l (Phase 1 and 3) and constrained (Phase 2) conditions
bo to
D.H. VanDercar et al. /Instrumental conditioning o f human heart rate
225
them to being passively respirated by a Bennett MA1 respirator. All subjects were individually fitted with a face mask with adjustments being made to insure that air did not escape during inspiration and expiration. With the aid of the subjects, adjustments were made in both the rate and depth of respiration. In this manner, parameter values were obtained which permitted the subjects to be respirated for periods of 30 min without reporting discomfort. In order to assess whether the subjects were engaging in maneuvers which might affect respiration, the Grass PT5 pressure transducer was attached via a tube to the face mask. This measure allowed the experimenter to monitor pressure changes during respiration and proved to be a sensitive measure of whether the subject was panting, closing the glottis, or assisting the respirator during either inspiration or expiration. Following these adaptation sessions, each subject received eight additional days of conditioning identical to Phase 1, with the exception that respiration rate and tidal volume were carefully controlled. 2.3.3. Phase 3
Seven additional days of conditioning were given without the use of the respirator, in a manner identical to that of Phase 1. 2.4. Data quantification and analysis
Heart rate and integrated EMG were recorded and printed out each minute during conditioning. The amount of reinforcement, i.e., intensity of the light and the bably's cry together with the respiration, was recorded on the polygraph. Respiration, was recorded on the polygraph. Respiration rate was measured by counting the number of complete respiration cycles occurring each minute. Depth of respiration during Phases I and 3, and respiratory pressure during Phase 2, were obtained every 30 sec by measuring the maximum pen excursion associated with pressure changes either at the pneumograph or at the face mask. These results are summarized in table 1. Statistical analyses were based on difference scores, obtained by subtracting the average value for each response during daily decrease periods from those obtained during increase periods. Correlations were calculated by pairing response measures, e.g., heart rate-respiration rate, obtained during each 4-min period of daily sessions. Unless otherwise specified, all correlations and tests of significance were based upon results obtained from individual subjects.
3. Results 3.1. Phase 1
Mean heart rate for the six subjects during Phase 1 was 74.2 (standard deviation s.d. = 10.9). This compares with mean heart rates of 73.8 (s.d. = 12.7) and 74.1 (s.d. = 12.4) during Phases 2 and 3 respectively.
D.H. VanDercar et al. / Instrumental conditioning o f human heart rate
226 e
~
~
2
I
Phase
~
Phase 2 O|e
Phase 3
period
0-/
~ o
\
=.'~-
i/ ~"-
~-4-
P~ I-6-6--
\
\
P-'¢ /
Constrainedmspiration
I
I I I I I I I I I 2 3 4 6 e r
I
2
I
J ,,a
/
,decreaseperiod
Normal respiration I
~
I
3 4
I
I
6 6
I
?
Normalrespiration e
I I I I I I I I 2 s 4 6 e
r
Days of Conditioning with Normal and Constrained Respiration Fig. 1. Mean heart rate changes from individual baselines during daily sessions in which six subjects were trained to both increase and decrease heart rate during conditions of constrained (Phase 2) and normal (Phases 1 and 3) respiration.
Fig. I displays the average daily heart rate differences from baseline during increase and decrease periods for all six subjects. Repeated measures analyses of variance conducted upon these data revealed that during Phase 1 subjects were able to maintain highly significant heart rate differences between increase and decrease periods (F (1,5) = 16.32, p < 0.01). Although fig. 1 indicates that these heart rate changes were bidirectional, statistically significant differences from baseline were obtained during increase periods (F (1,5) = 6.73, p < 0.05) but not during decrease periods (/7 (1,5) = 4.15, p > 0.05). Table 1 lists the average difference scores and the results of dependent t-tests for heart rate, respiration rate, depth or pressure of respiration, and EMG for each subject during Phases 1, 2, and 3. In addition, the average daily correlations between heart rate and each of the other responses measured are given. Examination of the heart rate columns at the far left of table 1 reveals that during Phase 1, when the subjects were not passively respirated, they all were able to maintain significantly faster heart rates during increase periods than during decrease periods. Although subjects were instructed not to change their respiration, significantly higher respiration rates were found during increase periods than during decrease periods for all subjects. Significant differences in depth of respiration were also found in four of the six subjects. Unlike respiration rate, these differences were not in the same direction for all subjects. Subjects 1 and 5 had larger depth of respiration during increase periods, whereas subjects 3 and 4 manifested larger depth of respiration during
D.H. VanDercaret al. /Instrumental conditioning of human heart rate
227
decrease periods. Of the four subjects who were monitored for EMG, none showed significant differences. The correlation between heart rate and the other physiological measures recorded is shown on the right side of table 1. The relationship between heart and respiration rates was found to be significantly positively correlated in five of the subjects (1, 2, 4, 5, 6). Significant positive correlations were also found between heart rate and depth of respiration in the two subjects who had higher depth of respiration during increase periods (subjects 1 and 5). 3.2. Phase 2
During Phase 2, each subject received eight additional days of training while being passively respirated. Respiration rate and depth of respiration (tidal volume) were controlled by the Bennett MA1 respirator and were not monitored. Instead, the pressure changes during inspiration and expiration were monitored at the face mask. Although small changes were observed within sessions, these pressure variations were determined not to be due to "panting" or closing the glottis but appear rather to be small differences in the extent to which the subject assisted the respirator. As such, the respirator pressure data from Phase 2 represent subjectproduced variations in interpulmonary pressure. Examination of fig. 1 indicates that the ability of subjects to maintain bidirectional heart rate control while being passively respirated was considerably less than in the Phase 1 condition. Separate repeated measures analyses of variance conducted across subjects during Phase 2 revealed no significant heart rate differences from baseline during either the increase (F (1,5) = 0.85, p > 0.05) or decrease periods (F (1,5) = 1.05, p > 0.05). Additionally, heart rate during increase periods was found not to be significantly different from decrease periods (F (1,5) = 4.42, p > 0.05). From table 1 it can be seen that the heart rate difference scores of each subject was reduced during Phase 2; only subjects 2 and 5 maintained significant heart rate control. Reliable pressure changes were found in four of the six subjects (1, 2, 3, 6). In each of these subjects, pressure changes were greater during decrease periods than during increase ones. Unlike the results in Phase 1, significant EMG differences were found in two of the four subjects monitored in Phase 2 (subjects 5 and 6). In both cases, EMG was greater during increase than during decrease periods. Thus, of the two subjects capable of maintaining heart rate control during Phase 2, subject 2 had significant differences in respiratory pressure and subject 5 had significant correlations between heart rate and pressure changes and between heart rate and EMG. 3.3. Phase 3
In Phase 3 all subjects received seven additional days of conditioning identical to that described for Phase 1. During Phase 3 subjects were again instructed not to change their respiration and to remain as relaxed as possible.
228
D.H. VanDercar et al. / Instrumental conditioning o f human heart rate
Examination of fig. 1 indicates that during this phase, subjects regained to a large extent their ability to control their heart rate. Repeated measures analysis of variance conducted upon these data revealed that during Phase 3 subjects were able to maintain significant heart rate differences between increase and decrease periods (F (1,5) -- 8.70, p < 0.05). Although fig. 1 indicates that these heart rate changes were bidirectional, statistically significant differences from baseline were only obtained during increase periods (F (1,5) = 12.45, p < 0.025) and not during decrease periods (F (1,5) = 0.40, p > 0.05). From table 1 it can be seen that heart rate differences, although somewhat smaller than during Phase 1, were significant in all but subject 6. Reliable differences in respiration rate were found in four of the six subjects (2, 3, 4, 5). Subject 3's results, unlike those in Phase 1, showed higher respiration rates during decrease than during increase periods. Significant differences in depth of respiration were found in two of the six subjects. Consistent with Phase 1, depth of respiration was greater during decrease periods for subject 4, whereas subject 5 had larger depth of respiration during increase periods. Significant correlations between heart rate and respiration rate were found in subjects 2, 3, and 5. Subject 3's correlations was -0.54, which was consistent with her respiration rate differences, which were greater during decrease than increase periods. Only subject 5 had a significant heart ratedepth of respiration correlation. No significant heart rate-EMG correlations were found.
4. Discussion
In the first part of this study we confirmed that human subjects can exert bidirectional control of heart rate if they are given appropriate feedback. However, by careful monitoring of respiratory variables, such as rate and depth, we were able to show that these bidirectional changes in heart rate were closely related to changes in respiration. Similar findings have previously been reported (i.e., Levenson, 1976). Without claiming at this stage that all of the heart rate variability in our situation can be ascribed to the influence of these respiratory variables, it is evident that the concept of somatic-visceral coupling can account for a reasonable amount of heart rate variance. These correlations do not necessarily prove that the changes in somatic responses cause those in heart rate; both changes could be the result of some central process (Obrist, Webb, Sutterer and Howard, 1970). However, if the somatic responses do cause the heart rate changes, the ability of the subjects to control their heart rate in the appropriate direction should be diminished or disappear when the subjects are constrained to breathe at a constant rate and tidal volume. When subjects are so constrained, their ability to control heart rate disappears in most instances. In the two cases where heart rate control was still demonstrated, we were able to show significant changes in respiratory pressure in one subject, and significant correlations between heart rate-EMG and heart rate-respiratory pres-
D.H. VanDercar et al. / Instrumental conditioning of human heart rate
229
sure in the other. Because of the relatively large number of sessions run under both constrained and unconstrained respiratory conditions, it would appear that critics of work demonstrating cardiac-somatic relationships cannot claim that had more training been carried out, heart rate specificity would have been observed. Quite clearly, the addition of respiratory constraints to a biofeedback situation does lend some ambiguity to the interpretation of results. For example it could be claimed that subjects cannot attend so well when their attention is focused on maintaining appropriate coordination with the respirator. This problem was anticipated in the design of our experimental procedure, and we took pains to circumvent it by adapting subjects to the respirator until we were satisfied that they could deal adequately with this task. Subjects' comments on the procedure supported our belief that, after some initial difficulty, they were able to feel at ease in this situation. Another possibility of confounding could occur if setting the depths and rates of the respirator led to hyperventilation. However, the findings that baseline heart rates did not differ significantly among Phases 1,2, and 3 strongly suggests that the failure of subjects to show heart rate control while being respirated was not due to respiratory alkalosis. During Phase 2, when subjects were not capable of changing rate or depth of respiration, it appeared as if they switched to other strategies (table 1). Thus, during this phase, four subjects varied their respiration pressure by resisting inspiration during decrease periods more than during increase periods. In addition, significant differences in the EMG were noted for the first time in two of the four subjects who were monitored. In a recent paper, Levenson (1976) argued that biofeedback training does not lead to the acquisition of control over heart rate, but only assesses a subject's preexperimental ability to control heart rate. Our results confirm those of Levenson insofar as we observed differential heart rate responding during the first four minutes of heart rate increase and the first four minutes of heart rate decrease training. Moreover, the greatest degree of bidirectional heart rate control was found during the first session and was thereafter attenuated. In order to demonstrate that the subjects' ability to control their heart rate was not simply diminishing with time, due to factors such as boredom or lack of motivation, we retrained subjects in the unconstrained situation, after they were trained on the respirator. In general, it is quite clear that they were still able to control their heart rate when given appropriate feedback and that, therefore, their relative inability to exert heart rate control during respiratory constraint was not due to motivational factors. However, it is interesting to note that their ability to control their heart rate was diminished during this phase when compared with the first phase of training. We suggest that this is due to the learning by the subjects of different patterns of respiration induced in part by the controlled respiration procedure. Many studies have monitored respiration and/or EMG and have claimed that
230
D.H. VanDercar et al. /Instrumental conditioning of human heart rate
such somatic responses do not explain the ability'of human subjects to control their heart rate (Engel and Hansen, 1966; Engel and Chism, 1967; Brener and HothersaU, 1966; Sroufe, 1969). More recently, researchers utilizing paced respiration have demonstrated that even moderate changes in respiration can have pronounced effects on cardiac activity (Sroufe, 1971; Obrist, Galosy, Lawler, Gaebelein, Howard and Shanks, 1975). The present study, which more rigorously controlled the rate and depth of respiration, supports these findings and suggests that respiration rate, respiration depth, and EMG activity may interact in a complex fashion to influence heart rate control. In the present study, bidirectional changes in heart rate consistently occurred when subjects were capable of changing rate and depth of respiration. In assessing the results of this training it is important that the changes occurring in one of the directions, perhaps decreases, cannot simply be attributed to a change in baseline heart rate. During the present study, the direction of required heart rate changes was alternated every four minutes within each session. Control over heart rate was consistent across trials of the same type throughout sessions as well as across sessions. Moreover, it made little difference whether the subjects were required to increase or decrease heart rate during the first trial of each session. Additionally, the heart rate changes that did occur were consistently related to changes in respiration and/or EMG. While it is conceivable that it may have been easier for subjects to either increase or decrease heart rate because of changes in baseline associated with reclining in a room over a period of time, it is likely that these heart rate baseline changes themselves were related to EMG and respiratory changes. Interestingly, when multiple baselines have been assessed within sessions during biofeedback experiments, reliable increases and decreases in heart rate have been observed (e.g., Levenson, 1976). As far as the clinical application of biofeedback procedures is concerned, it may not make any difference that somatic-visceral coupling can be demonstrated when the researcher is attempting to control a visceral event (Miller and Dworkin, 1974). However, our evidence that specific mediators play a role in the control of heart rate during biofeedback training suggests that clinicians may want to focus upon those mediators as well, when attempting to alleviate such symptoms as tachycardia and arrythmia. Although the control of heart rate has already proven useful clinically (Engel, 1973), many biofeedback studies have demonstrated statistically rather than clinically significant changes (Blanchard and Young, 1973). By focusing upon heart rate and its mediators, future clinical studies may be able to augment the heart rate changes that have been observed to date.
Acknowledgement This research was conducted at the Rockefeller University and was supported by USPHS grants GM01789, MH 13189, MH 19183.
D.H. VanDercar et al. /Instrumental conditioning o f human heart rate
231
The writers would like to thank Dr. N.E. Miller for his assistance and support, as well as Dr. N. Schneiderman for his help in reviewing the manuscript.
References Blanchard, E.B. and Young, L.D. (1973). Self-control of cardiac functioning: A promise as yet unfulfilled. Psychological Bulletin, 79, 145 - 163. Brener, J. (1974). A general model of voluntary control applied to the phenomena of learned cardiovascular change. In P.A. Obrist, A.H. Black, J. Brener and L.V. Dicara (Eds.), Cardiovascular psychophysiology: Current issues in response mechanisms, biofeedback and methodology. Aldine: Chicago, 365-391. Brener, J. and Hothersall, D. (1966). Heart rate control under conditions of augmented sensory feedback. Psychophysiology, 3, 23-28. Consolazio, C.F., Johnson, R.E. and Pecoris, L.J. (1963). Physiological measurements of metabolic function in man. McGraw-Hill: New York, 194-199. Crider, A., Schwartz, G.E. and Schnidman, S. (1969). On the criteria for instrumental autonomic conditioning: A reply to Katkin and Murray. Psychological Bulletin, 71,455-461. Engel, B.T. (1973). Clinical application of operant conditioning techniques in the control of cardiac arrythmias. Seminars in Psychiatry, 5,433-438. Engel, B.T. and Chism, R. (1967). Operant conditioning of heart rate speeding. Psyehophysiology, 3,418-426. Engel, B.T. and Hansen, S.P. (1966). Operant conditioning of heart rate slowing. Psychophysiology, 3,176-188. Katkin, E.S. and Murray, E.N. (1968). Instrumental conditioning of autonomically mediated behavior: Theoretical and methodological issues. Psychological Bulletin, 70, 52-68. Katkin, E.S., Murray, E.N. and Lachman, R. (1969). Concerning instrumental autonomic conditioning: A rejoinder. Psychological Bulletin, 71,426-466. Levenson, R.W. (1976). Feedback effects and respiratory involvement in voluntary control of heart rate. Psychophysiology, 13,108-114. Miller, N.E. and Dworkin, B.R. (1974). Visceral learning: Recent difficulties with curarized rats and significant problems for human research. In P.A. Obrist, A.H. Black, J. Brener and L.V. Dicara (Eds.), Cardiovascular psychophysiology: Current issues in response mechanisms, biofeedback and methodology. Aldine: Chicago, 312-331, Obrist, P.A., Webb, R.A. Sutterer, J.R. and Howard, J.L. (1970). The cardiac-somatic relationship: Some reformulations. Psychophysiology, 6, 569-587. Obrist, P.A., Galosy, R.A., Lawler, J.E., Gaebelein, C.J., Howard, J.L. and Shanks, E.M. (1975). Operant conditioning of heart rate: Somatic correlates. Psychophysiology, 12,445-455. Shapiro, D., Tursky, B. and Schwartz, G.E. (1970). Differentiation of heart rate and Systolicblood pressure in man by operant conditioning. Psychosomatic Medicine, 32,417-423. Sroufe, L.A. (1969). Learned stabilization of cardiac rate with respiration experimentally controlled. Journal of Experimental Psychology, 81, 391-393. Sroufe, L.A. (1971). Effects of depth and rate of breathing on heart rate and heart rate variability. Psychophysiology, 8,648-655.
INSTRUMENTAL CONDITIONING OF HUMAN HEART RATE DURING FREE AND CONTROLLED RESPIRATION D.H. V A N D E R C A R * ' t , M.A. FELDSTEIN, and H. SOLOMON Department o f Physiological Psychology, Rockefeller University, NYC, New York 10021, U.S.A. Accepted for publication 31 March 1977
The effect of respiratory constraint on heart rate control was assessed in a biofeedback situation with feedback consisting of changes in both illumination level and intensity of a prerecorded baby's cry. Subjects were reinforced for alternately increasing and decreasing heart rate on each of seven days during which respiration was unconstrained (Phase 1). This phase was followed by eight days when respiration was constrained during training sessions with a control respirator (Phase 2). Seven additional days of training followed in the unconstrained situation (Phase 3). Results indicate that the control of heart rate in biofeedback situations is very closely related to respiratory and other somatic activity. The implication of these findings for the field of visceral control is discussed.
1. Introduction In recent years, many investigators have reported success in operantly conditioning heart rate in human subjects (Brener and Hothersall, 1966; Engel and Hansen, 1966; Engel and Chism, 1967; Brener, 1974). Although these results have been encouraging, a basic question remains as to whether the changes recorded represent true operant conditioning o f autonomic responses or whether the heart rate changes are mediated by the operant conditioning of skeletal responses (Katkin and Murray, 1968; Crider, Schwartz and Schnidman, 1969; Katldn, Murray and Lachman, 1969). Another question that has been raised is whether biofeedback training leads to the acquisition o f control over heart rate or whether the biofeedback experiment only assesses a subject's pre-experimental ability to control heart rate (Levenson, 1976). The present study was designed to investigate the acquisition of heart rate con-
* Now at University of South Florida. "~ Address requests for reprints to: David H. VanDercar, Ph.D., Dept. of Psychology, University of South Florida, Tampa, Florida 33620. 221
222
D.H. VanDercaret al. / Instrumental conditioning of human heart rate
trol, the extent to which subjects can establish control over heart rate when provided with feedback, and the extent to which somatic responses such as muscular activity, and rate and depth of respiration are involved in these heart rate changes. In order to answer these questions, the present study examined heart rate, electro° myographic activity, and respiratory responses during multiple feedback training sessions. In some of these sessions the subjects were able to control respiration rate and depth, whereas in other sessions the subjects were passively respirated and unable to influence respiration rate or tidal volume. Although demand and control respirators are routinely used in medical settings, their use for controlling respiration in biofeedback research is unique.
2. Method 2.1. Subjects
Six normal female subjects between the ages of 18 and 30 years were paid $ 3.00 per hour for participating in this study. All subjects were given brief examinations by a physician and found to be free of obvious respiratory or cardiovascular disorders. 2.2. Apparatus
Electrocardiogram (ECG), heart rate, electromyogram (EMG), respiration rate, depth of respiration, and the direction of change attempted during trial periods were recorded on a Grass model 7 polygraph. In addition, integrated EMG was converted to pulses by coupling the output of the polygraph to a voltage-to-frequency converter which, together with heart rate, were recorded on a printout counter each minute during baseline and conditioning periods. The ECG was recorded from a lead II configuration with Beckman electrodes and telemetered via a Narco BioSystems model E2 transmitter. Respiration rate and depth of respiration were monitored with a chest bellows placed around the chest and connected to a Grass PT 5 pressure transducer. The EMG was recorded and telemetered from the sternocleidomastoid muscle of the neck and displayed as both raw EMG and integrated activity. EMG recordings were not taken from: (a) frontalis muscles because of potential contamination from eye movement responses associated with attending to visual cues, or (b) chin muscles, which would have been adversely affected by the respirator mask. All recording equipment, together with BRS circuitry for programming trials, was located outside a sound-attenuated chamber (Industrial Acoustics Corporation). A modified dimmer switch coupled to the output of the Grass cardiotachometer was used to provide continuous visual and auditory heart rate feedback to the subjects. Feedback consisted of a tape recording of a baby's cry and a 100 watt light
D.H. VanDercaret al. /Instrumental conditioning of human heart rate
223
bulb, both of which varied in intensity as a function of the subject's immediate heart rate. The feedback circuitry was arranged so that the experimenter could select either an increase or decrease in heart rate as the correct response for producing an attenuation of the intensity of both the baby's cry and the light. The use of an ecologically significant stimulus (i.e., crying) was intended to facilitate training by providing reinforcement as well as informational content to the subject. In the second phase of this experiment a Bennett MA1 respirator was used to respirate subjects at constant rate and tidal volume. The Bennett respirator can be used in either a demand or control mode. In the demand mode the subject initiates respiration but the respirator provides a fixed volume of air. In the control mode, which was used in the present study, the experimenter can set respiration rate, tidal volume, and inspiration-expiration ratio. During the present study the inspirationexpiration ratio was fixed at 1 : 1 for all subjects, but the respiration rate and tidal volume were adjusted by each subject. Within several adaptation sessions the subjects stabilized at values approximating their norms for weight and age (Consolazio, Johnson and Pecoris, 1963). 2.3. Procedure
Subjects were conditioned while reclining on a comfortable cot located within a sound-attenuated chamber. After all leads had been attached for recording heart late, EMG, and respiration, subjects were informed that the purpose of this study was to determine whether they could voluntarily control their heart rate. Each subject was instructed to remain relaxed and not to attempt to influence heart rate by changing breathing or engaging in muscular activity. They were also told to pay close attention to the feedback which would be provided, and to engage in any 'mental activity' which helped in controlling heart rate. In addition to feedback, the subjects were presented with either a red or green cue light and were told that, whenever the red light was on, they were to attempt to lower their heart rate, and conversely, increase their heart rate when the green light was presented. They were further informed that, as they succeeded in changing their heart rate in the appropriate direction, the intensity of both the baby's cry and the light would decrease. 2.3.1. Phase 1
Each subject received seven days of conditioning, during which no attempt was made to control respiration other than through instructions. Each daily session lasted 20 min and began with a 4-min baseline period during which the cue lights and feedback were absent. The subsequent 16 min consisted of four 4-min periods during which time the red and green lights were alternately presented. On each successive day of conditioning, the cue light which began a session was alternated. 2.3.2. Phase 2
Following Phase 1, all subjects underwent several sessions designed to acclimate
1 2 3
1 2 3
1 2 3
1 2 3
1 2 3
Subject 2
Subject 3
Subject 4
Subject 5
Subject 6
2.5 0.7 3.0
9.1 4.1 7.6
3.1 0.81 0.8
2.3 0.12 1.5
4.0 3.1 1.6
6.2 0.4 1.9
0.05 >0.05
0.05
. 0.05 >0.05
HR-EMG
Table 1 Difference scores (d) and their correlations (r) be tw e e n increase and decrease periods for heart rate (HR), e l e c t r o m y o g r a p h i c activity (EMG), respiration rate (RR), d ept h (RD), and pressure (RP), during norma l (Phase 1 and 3) and constrained (Phase 2) conditions
bo to
D.H. VanDercar et al. /Instrumental conditioning o f human heart rate
225
them to being passively respirated by a Bennett MA1 respirator. All subjects were individually fitted with a face mask with adjustments being made to insure that air did not escape during inspiration and expiration. With the aid of the subjects, adjustments were made in both the rate and depth of respiration. In this manner, parameter values were obtained which permitted the subjects to be respirated for periods of 30 min without reporting discomfort. In order to assess whether the subjects were engaging in maneuvers which might affect respiration, the Grass PT5 pressure transducer was attached via a tube to the face mask. This measure allowed the experimenter to monitor pressure changes during respiration and proved to be a sensitive measure of whether the subject was panting, closing the glottis, or assisting the respirator during either inspiration or expiration. Following these adaptation sessions, each subject received eight additional days of conditioning identical to Phase 1, with the exception that respiration rate and tidal volume were carefully controlled. 2.3.3. Phase 3
Seven additional days of conditioning were given without the use of the respirator, in a manner identical to that of Phase 1. 2.4. Data quantification and analysis
Heart rate and integrated EMG were recorded and printed out each minute during conditioning. The amount of reinforcement, i.e., intensity of the light and the bably's cry together with the respiration, was recorded on the polygraph. Respiration, was recorded on the polygraph. Respiration rate was measured by counting the number of complete respiration cycles occurring each minute. Depth of respiration during Phases I and 3, and respiratory pressure during Phase 2, were obtained every 30 sec by measuring the maximum pen excursion associated with pressure changes either at the pneumograph or at the face mask. These results are summarized in table 1. Statistical analyses were based on difference scores, obtained by subtracting the average value for each response during daily decrease periods from those obtained during increase periods. Correlations were calculated by pairing response measures, e.g., heart rate-respiration rate, obtained during each 4-min period of daily sessions. Unless otherwise specified, all correlations and tests of significance were based upon results obtained from individual subjects.
3. Results 3.1. Phase 1
Mean heart rate for the six subjects during Phase 1 was 74.2 (standard deviation s.d. = 10.9). This compares with mean heart rates of 73.8 (s.d. = 12.7) and 74.1 (s.d. = 12.4) during Phases 2 and 3 respectively.
D.H. VanDercar et al. / Instrumental conditioning o f human heart rate
226 e
~
~
2
I
Phase
~
Phase 2 O|e
Phase 3
period
0-/
~ o
\
=.'~-
i/ ~"-
~-4-
P~ I-6-6--
\
\
P-'¢ /
Constrainedmspiration
I
I I I I I I I I I 2 3 4 6 e r
I
2
I
J ,,a
/
,decreaseperiod
Normal respiration I
~
I
3 4
I
I
6 6
I
?
Normalrespiration e
I I I I I I I I 2 s 4 6 e
r
Days of Conditioning with Normal and Constrained Respiration Fig. 1. Mean heart rate changes from individual baselines during daily sessions in which six subjects were trained to both increase and decrease heart rate during conditions of constrained (Phase 2) and normal (Phases 1 and 3) respiration.
Fig. I displays the average daily heart rate differences from baseline during increase and decrease periods for all six subjects. Repeated measures analyses of variance conducted upon these data revealed that during Phase 1 subjects were able to maintain highly significant heart rate differences between increase and decrease periods (F (1,5) = 16.32, p < 0.01). Although fig. 1 indicates that these heart rate changes were bidirectional, statistically significant differences from baseline were obtained during increase periods (F (1,5) = 6.73, p < 0.05) but not during decrease periods (/7 (1,5) = 4.15, p > 0.05). Table 1 lists the average difference scores and the results of dependent t-tests for heart rate, respiration rate, depth or pressure of respiration, and EMG for each subject during Phases 1, 2, and 3. In addition, the average daily correlations between heart rate and each of the other responses measured are given. Examination of the heart rate columns at the far left of table 1 reveals that during Phase 1, when the subjects were not passively respirated, they all were able to maintain significantly faster heart rates during increase periods than during decrease periods. Although subjects were instructed not to change their respiration, significantly higher respiration rates were found during increase periods than during decrease periods for all subjects. Significant differences in depth of respiration were also found in four of the six subjects. Unlike respiration rate, these differences were not in the same direction for all subjects. Subjects 1 and 5 had larger depth of respiration during increase periods, whereas subjects 3 and 4 manifested larger depth of respiration during
D.H. VanDercaret al. /Instrumental conditioning of human heart rate
227
decrease periods. Of the four subjects who were monitored for EMG, none showed significant differences. The correlation between heart rate and the other physiological measures recorded is shown on the right side of table 1. The relationship between heart and respiration rates was found to be significantly positively correlated in five of the subjects (1, 2, 4, 5, 6). Significant positive correlations were also found between heart rate and depth of respiration in the two subjects who had higher depth of respiration during increase periods (subjects 1 and 5). 3.2. Phase 2
During Phase 2, each subject received eight additional days of training while being passively respirated. Respiration rate and depth of respiration (tidal volume) were controlled by the Bennett MA1 respirator and were not monitored. Instead, the pressure changes during inspiration and expiration were monitored at the face mask. Although small changes were observed within sessions, these pressure variations were determined not to be due to "panting" or closing the glottis but appear rather to be small differences in the extent to which the subject assisted the respirator. As such, the respirator pressure data from Phase 2 represent subjectproduced variations in interpulmonary pressure. Examination of fig. 1 indicates that the ability of subjects to maintain bidirectional heart rate control while being passively respirated was considerably less than in the Phase 1 condition. Separate repeated measures analyses of variance conducted across subjects during Phase 2 revealed no significant heart rate differences from baseline during either the increase (F (1,5) = 0.85, p > 0.05) or decrease periods (F (1,5) = 1.05, p > 0.05). Additionally, heart rate during increase periods was found not to be significantly different from decrease periods (F (1,5) = 4.42, p > 0.05). From table 1 it can be seen that the heart rate difference scores of each subject was reduced during Phase 2; only subjects 2 and 5 maintained significant heart rate control. Reliable pressure changes were found in four of the six subjects (1, 2, 3, 6). In each of these subjects, pressure changes were greater during decrease periods than during increase ones. Unlike the results in Phase 1, significant EMG differences were found in two of the four subjects monitored in Phase 2 (subjects 5 and 6). In both cases, EMG was greater during increase than during decrease periods. Thus, of the two subjects capable of maintaining heart rate control during Phase 2, subject 2 had significant differences in respiratory pressure and subject 5 had significant correlations between heart rate and pressure changes and between heart rate and EMG. 3.3. Phase 3
In Phase 3 all subjects received seven additional days of conditioning identical to that described for Phase 1. During Phase 3 subjects were again instructed not to change their respiration and to remain as relaxed as possible.
228
D.H. VanDercar et al. / Instrumental conditioning o f human heart rate
Examination of fig. 1 indicates that during this phase, subjects regained to a large extent their ability to control their heart rate. Repeated measures analysis of variance conducted upon these data revealed that during Phase 3 subjects were able to maintain significant heart rate differences between increase and decrease periods (F (1,5) -- 8.70, p < 0.05). Although fig. 1 indicates that these heart rate changes were bidirectional, statistically significant differences from baseline were only obtained during increase periods (F (1,5) = 12.45, p < 0.025) and not during decrease periods (F (1,5) = 0.40, p > 0.05). From table 1 it can be seen that heart rate differences, although somewhat smaller than during Phase 1, were significant in all but subject 6. Reliable differences in respiration rate were found in four of the six subjects (2, 3, 4, 5). Subject 3's results, unlike those in Phase 1, showed higher respiration rates during decrease than during increase periods. Significant differences in depth of respiration were found in two of the six subjects. Consistent with Phase 1, depth of respiration was greater during decrease periods for subject 4, whereas subject 5 had larger depth of respiration during increase periods. Significant correlations between heart rate and respiration rate were found in subjects 2, 3, and 5. Subject 3's correlations was -0.54, which was consistent with her respiration rate differences, which were greater during decrease than increase periods. Only subject 5 had a significant heart ratedepth of respiration correlation. No significant heart rate-EMG correlations were found.
4. Discussion
In the first part of this study we confirmed that human subjects can exert bidirectional control of heart rate if they are given appropriate feedback. However, by careful monitoring of respiratory variables, such as rate and depth, we were able to show that these bidirectional changes in heart rate were closely related to changes in respiration. Similar findings have previously been reported (i.e., Levenson, 1976). Without claiming at this stage that all of the heart rate variability in our situation can be ascribed to the influence of these respiratory variables, it is evident that the concept of somatic-visceral coupling can account for a reasonable amount of heart rate variance. These correlations do not necessarily prove that the changes in somatic responses cause those in heart rate; both changes could be the result of some central process (Obrist, Webb, Sutterer and Howard, 1970). However, if the somatic responses do cause the heart rate changes, the ability of the subjects to control their heart rate in the appropriate direction should be diminished or disappear when the subjects are constrained to breathe at a constant rate and tidal volume. When subjects are so constrained, their ability to control heart rate disappears in most instances. In the two cases where heart rate control was still demonstrated, we were able to show significant changes in respiratory pressure in one subject, and significant correlations between heart rate-EMG and heart rate-respiratory pres-
D.H. VanDercar et al. / Instrumental conditioning of human heart rate
229
sure in the other. Because of the relatively large number of sessions run under both constrained and unconstrained respiratory conditions, it would appear that critics of work demonstrating cardiac-somatic relationships cannot claim that had more training been carried out, heart rate specificity would have been observed. Quite clearly, the addition of respiratory constraints to a biofeedback situation does lend some ambiguity to the interpretation of results. For example it could be claimed that subjects cannot attend so well when their attention is focused on maintaining appropriate coordination with the respirator. This problem was anticipated in the design of our experimental procedure, and we took pains to circumvent it by adapting subjects to the respirator until we were satisfied that they could deal adequately with this task. Subjects' comments on the procedure supported our belief that, after some initial difficulty, they were able to feel at ease in this situation. Another possibility of confounding could occur if setting the depths and rates of the respirator led to hyperventilation. However, the findings that baseline heart rates did not differ significantly among Phases 1,2, and 3 strongly suggests that the failure of subjects to show heart rate control while being respirated was not due to respiratory alkalosis. During Phase 2, when subjects were not capable of changing rate or depth of respiration, it appeared as if they switched to other strategies (table 1). Thus, during this phase, four subjects varied their respiration pressure by resisting inspiration during decrease periods more than during increase periods. In addition, significant differences in the EMG were noted for the first time in two of the four subjects who were monitored. In a recent paper, Levenson (1976) argued that biofeedback training does not lead to the acquisition of control over heart rate, but only assesses a subject's preexperimental ability to control heart rate. Our results confirm those of Levenson insofar as we observed differential heart rate responding during the first four minutes of heart rate increase and the first four minutes of heart rate decrease training. Moreover, the greatest degree of bidirectional heart rate control was found during the first session and was thereafter attenuated. In order to demonstrate that the subjects' ability to control their heart rate was not simply diminishing with time, due to factors such as boredom or lack of motivation, we retrained subjects in the unconstrained situation, after they were trained on the respirator. In general, it is quite clear that they were still able to control their heart rate when given appropriate feedback and that, therefore, their relative inability to exert heart rate control during respiratory constraint was not due to motivational factors. However, it is interesting to note that their ability to control their heart rate was diminished during this phase when compared with the first phase of training. We suggest that this is due to the learning by the subjects of different patterns of respiration induced in part by the controlled respiration procedure. Many studies have monitored respiration and/or EMG and have claimed that
230
D.H. VanDercar et al. /Instrumental conditioning of human heart rate
such somatic responses do not explain the ability'of human subjects to control their heart rate (Engel and Hansen, 1966; Engel and Chism, 1967; Brener and HothersaU, 1966; Sroufe, 1969). More recently, researchers utilizing paced respiration have demonstrated that even moderate changes in respiration can have pronounced effects on cardiac activity (Sroufe, 1971; Obrist, Galosy, Lawler, Gaebelein, Howard and Shanks, 1975). The present study, which more rigorously controlled the rate and depth of respiration, supports these findings and suggests that respiration rate, respiration depth, and EMG activity may interact in a complex fashion to influence heart rate control. In the present study, bidirectional changes in heart rate consistently occurred when subjects were capable of changing rate and depth of respiration. In assessing the results of this training it is important that the changes occurring in one of the directions, perhaps decreases, cannot simply be attributed to a change in baseline heart rate. During the present study, the direction of required heart rate changes was alternated every four minutes within each session. Control over heart rate was consistent across trials of the same type throughout sessions as well as across sessions. Moreover, it made little difference whether the subjects were required to increase or decrease heart rate during the first trial of each session. Additionally, the heart rate changes that did occur were consistently related to changes in respiration and/or EMG. While it is conceivable that it may have been easier for subjects to either increase or decrease heart rate because of changes in baseline associated with reclining in a room over a period of time, it is likely that these heart rate baseline changes themselves were related to EMG and respiratory changes. Interestingly, when multiple baselines have been assessed within sessions during biofeedback experiments, reliable increases and decreases in heart rate have been observed (e.g., Levenson, 1976). As far as the clinical application of biofeedback procedures is concerned, it may not make any difference that somatic-visceral coupling can be demonstrated when the researcher is attempting to control a visceral event (Miller and Dworkin, 1974). However, our evidence that specific mediators play a role in the control of heart rate during biofeedback training suggests that clinicians may want to focus upon those mediators as well, when attempting to alleviate such symptoms as tachycardia and arrythmia. Although the control of heart rate has already proven useful clinically (Engel, 1973), many biofeedback studies have demonstrated statistically rather than clinically significant changes (Blanchard and Young, 1973). By focusing upon heart rate and its mediators, future clinical studies may be able to augment the heart rate changes that have been observed to date.
Acknowledgement This research was conducted at the Rockefeller University and was supported by USPHS grants GM01789, MH 13189, MH 19183.
D.H. VanDercar et al. /Instrumental conditioning o f human heart rate
231
The writers would like to thank Dr. N.E. Miller for his assistance and support, as well as Dr. N. Schneiderman for his help in reviewing the manuscript.
References Blanchard, E.B. and Young, L.D. (1973). Self-control of cardiac functioning: A promise as yet unfulfilled. Psychological Bulletin, 79, 145 - 163. Brener, J. (1974). A general model of voluntary control applied to the phenomena of learned cardiovascular change. In P.A. Obrist, A.H. Black, J. Brener and L.V. Dicara (Eds.), Cardiovascular psychophysiology: Current issues in response mechanisms, biofeedback and methodology. Aldine: Chicago, 365-391. Brener, J. and Hothersall, D. (1966). Heart rate control under conditions of augmented sensory feedback. Psychophysiology, 3, 23-28. Consolazio, C.F., Johnson, R.E. and Pecoris, L.J. (1963). Physiological measurements of metabolic function in man. McGraw-Hill: New York, 194-199. Crider, A., Schwartz, G.E. and Schnidman, S. (1969). On the criteria for instrumental autonomic conditioning: A reply to Katkin and Murray. Psychological Bulletin, 71,455-461. Engel, B.T. (1973). Clinical application of operant conditioning techniques in the control of cardiac arrythmias. Seminars in Psychiatry, 5,433-438. Engel, B.T. and Chism, R. (1967). Operant conditioning of heart rate speeding. Psyehophysiology, 3,418-426. Engel, B.T. and Hansen, S.P. (1966). Operant conditioning of heart rate slowing. Psychophysiology, 3,176-188. Katkin, E.S. and Murray, E.N. (1968). Instrumental conditioning of autonomically mediated behavior: Theoretical and methodological issues. Psychological Bulletin, 70, 52-68. Katkin, E.S., Murray, E.N. and Lachman, R. (1969). Concerning instrumental autonomic conditioning: A rejoinder. Psychological Bulletin, 71,426-466. Levenson, R.W. (1976). Feedback effects and respiratory involvement in voluntary control of heart rate. Psychophysiology, 13,108-114. Miller, N.E. and Dworkin, B.R. (1974). Visceral learning: Recent difficulties with curarized rats and significant problems for human research. In P.A. Obrist, A.H. Black, J. Brener and L.V. Dicara (Eds.), Cardiovascular psychophysiology: Current issues in response mechanisms, biofeedback and methodology. Aldine: Chicago, 312-331, Obrist, P.A., Webb, R.A. Sutterer, J.R. and Howard, J.L. (1970). The cardiac-somatic relationship: Some reformulations. Psychophysiology, 6, 569-587. Obrist, P.A., Galosy, R.A., Lawler, J.E., Gaebelein, C.J., Howard, J.L. and Shanks, E.M. (1975). Operant conditioning of heart rate: Somatic correlates. Psychophysiology, 12,445-455. Shapiro, D., Tursky, B. and Schwartz, G.E. (1970). Differentiation of heart rate and Systolicblood pressure in man by operant conditioning. Psychosomatic Medicine, 32,417-423. Sroufe, L.A. (1969). Learned stabilization of cardiac rate with respiration experimentally controlled. Journal of Experimental Psychology, 81, 391-393. Sroufe, L.A. (1971). Effects of depth and rate of breathing on heart rate and heart rate variability. Psychophysiology, 8,648-655.
