- Email: [email protected]
Human Pavlovian decelerative cardiac conditioning based on a respiratory-induced cardiac deceleration as an unconditional reflex
HUMAN BASED
PAVLOVIAN
DECELERATIVE
CARDIAC
ON A RESPIRATORY-INDUCED
DECELERATION
JOHN J. FUREDY
CARDIAC
AS AN UNCONDITIONAL
and CONSTANTINE
CONDITIONING
REFLEX
X. POULOS
Ull~l~~rsif~ of Toron#o,Torontu, ~~ltar~o, Cannda
Accepted for publication 29 August 1974
This study was directed at examining the feasibility of using a respiratory-induced cardiac decelerative reflex as an unconditioned response (UCR) in a Pavlovian conditioning paradigm. Experiment I assessed the reflexive features of the cardiac response under a respiratory procedure which involved 3 set of exhalation, 4 SW of inhalation, and then 30 set of breathholding (BH). The results indicated that the BH onset aspect of this respiratory cycle involved a phasic, large-magnitude cardiac deceleration (27 beats/min) with short latency, fast recruitment, and no indication of habituation over trials. Experiment 11 examined the feasibility of using BH onset as an unconditioned stimulus (UCS) in a Pavlovian paradigm by presenting an auditory conditional stimulus (CS) 1 set prior to the instructed BH aspect of the respiratory cycle. A separate control condition involved presenting the CS 10 set after BH onset which constituted a ‘backward’ conditioning control. The results indicated that the conditioning group displayed a significant cardiac deceleration on CS-alone test trials (under normal breathing) in contrast to the absence of such an effect for the control group: In brief, the study suggested that a decelerative conditioned response (CR) could be established by using a respiratory-induced cardiac deceleration as a Pavlovian UCS. Potential clinical applications of such a decelerative CR were noted.
I. Introduction The area of human classical cardiac conditioning has produced confusing and paradoxical data with regard to the topography of the conditioned response (CR). Human cardiac conditioning has by and large relied on using aversive events such as shock as the unconditional stimulus (UCS) which elicit cardiac acceleration as the unconditional response (UCR). However, contrary to the classical conditioning proviso that the CR be topographi~lly similar to the UCR, the human cardiac CR based on aversive UCSs has been variously identified as biphasic (Zeaman, Deanne and Wegner, 1954), decelerative (Wood and Obrist, 1964), and accelerative (cf. Zeaman and Smith, 1965). Another source of difficulty is that, due to ethical considerations, the stressful stimuli typically used as UCSs have been relatively weak and UCR habituation is often extensive. This study is unusual in that it involved an instructed respiration cycle 165
which terminated with breatli-holding (BH) as a nonaversive UCS. As will be shown in more detail below, this UCS elicits a phasic cardiac deceleration as the UCR, the CR is of similar form, i.e. decelerative, and there is little or no UCR habituation over repeated trials. Finally, in addition to its potential use for resolving theoretical problems in human cardiac conditioning, the preparation based on the BH UCS also has potential clinical use. Specifically, there is a class of cardiac dysfunctions which consist of inappropriately large cardiac accelerations to sudden stress. One potential form of behavioural control of such dysfunctions is that which relies on operant-feedback methods, and, with the current enthusiasm over biofeedback, most attention has been focused on this form of control. However, it would seem feasible and desirable also to examine the possibilit}~ of achieving behavioural control through Pavlovian methods, provided that a clearly deceierative CR was available. For example, if the desired modification involved a decrease in the heart rate (HR), then a subject-generated CS. to which a decelerative CR has been previously established, could be utilized to decrease HR whenever wanted or needed. The BH UCS preparation to be described in this paper could provide just such a potential clinical tool.
2. Experiment I: the BH-induced UCR The data to be reported in experiment I were gathered originally with ultimate operant rather than Pavlovian behavioural control in mind. Specifically, we were dissatisfied with the magnitude and reliability of noise- and shockelicited cardiac accelerations (which were to be used as the target responses for later studies on operant modification) and desired to increase such accelerations by controlled respiration (cf. Laird and Fenz. 1971). The method of respiratory control chosen was one where subjects exhaled, inhaled, and then held their breath for 30 sec. The last phase of this instructed cycle provided the UCS for the elicited deceleration, and the issues to be reported in experiment I concern such features of this BH-induced UCR as magnitude, latency, topography and degree of habituation.
The subjects were ten undergraduates from the University of Toronto, who were paid $2.00 for participation. The apparatus consisted of a hot face box which has been detailed elsewhere (Furedy and Klajner, 1972). Briefly, the subject put his face into the box which has an internal temperature of 110°F (80% humidity, i.e. a little cooler than a steam bath). inside the box were labelled indicator lights which were used to indicate to the subject when he should exhale, inhale, hold, and resume normal breathing. The HR was picked up through electrodes placed either on the forearms or on the sternum (for those subjects on which forearm attachments provided an inadequate signal) and recorded through a physiograph (E & M) tachometer. Respiration
Human Pnulovion deceleratiue
cardiac conditioning
167
was picked up by a bellows pneumograph placed around the chest, and also recorded through the physiograph. All stimuli were delivered automatically by a system of relays, timers, and a paper-tape recorder. The subject and experimenter were in separate rooms, but both visual and auditory contact was maintained through a one-way screen and a microphone arrangement. Prior to the experimental session, each subject was familiarized with the cued-breathing condition which involved the following sequence (indicated by the lights in the box): 3 set of even exhalation, 4 set of even inhalation, and, finally, 30 set of breath-holding (see also the abscissa of fig. I). The immediately following experimental session involved 24 presentations of the cued-breathing trial with rest periods of normal breathing of 40-60 set (varied unsystematically) between the termination of one trial and the initiation of the next trial. In addition, during 18 of the 24 breath-holding periods, a loud-noise stimulus (0.3 set I 15 dB white noise) was presented at various intervals during the breath-hold cycle (i.e. at 2, 10 or 25 set following BH onset). The purpose of presenting the loud noise was to study cardiac acceleration under various respiration conditions, but these noise-induced acceleratory data will not be reported here, as they are relevant only to the original operant-control interest mentioned above. 2.2. Results and discussion A general picture of the BH-induced UCR was obtained second-by-second during all of the six controlled-respiration
Fig. 1. Mean
second-by-second cardiac rate (in BsPM) during controlled-respiration
by measuring HR trials where no
of ten subjects, trials.
2 set prior
to and
168
J. J. Fkredv and C. X. POI//OF
loud-noise stimulus had been presented. Fig. I presents these results. The three clear trends are: (a) the inspiration-induced acceleration: (b) the BHinduced deceleration; and (c) the mean return to baseline following the phasic BH UCR. Moreover, inspection of individual responses indicated that, because of the averaging procedures used for fig. I, the latency and recruitment characteristics of the BH UCR were actually obscured by the relatively dramatic representation depicted in fig. I. For example, an individual’s second-by-second cardiac rate (in BsPM) from -2 to IO set following the onset of a respiratory trial was : 69,68 (baseline before exhalation), 65,66, 68 (during exhalation), 70, 78, 82, 88, 89 (during inhalation), 88, 62, 63 (following hold). Such individual BH UCRs, then, showed not only short latency (within one or two beats of BH), but also fast recruitment, with most of the decelerative response occurring suddenly between a pair of beats. To obtain more systematic evidence regarding these latency and recruitment characteristics of the BH UCR, we sought to employ the maximum number of trials as well as ensuring that at least some of the relevant data were based on an independent verification that the instructed BH had, in fact, occurred at the time instructed to the subject by the experimenter. The BH UCR itself (though not its recovery) could potentially be examined on I8 of the 24 respiratory-control trials (i.e. all those trials except where the acceleration-inducing loud-noise stimulus occurred 2 set following BH). On approximately 40”/, of these trials it was possible to exactly verify the point of actual (in contrast to instructed) BH onset. On other trials, unavoidable ‘peak clipping’ (i.e. the respiration-recording pen going off scale) precluded this precise specification. On 40”/) of the trials, data involving five beats following actual BH onset wereexamined, and the mean maximum deceleration occurring between any pair of beats after hold onset was 20.01 BsPM. The latency of these precipitous single beat-to-beat decelerations was short, as indicated by the fact that 80% of them occurred within the three beats after BH onset. That the recruitment of the BH UCR was fast is indicated by the fact that the single beat-to-beat deceleration value of 20.01 BsPM accounted for more than 70”/, of the overall beat-to-beat deceleration (27.3 BsPM) during the five-beat interval following BH onset. To maximize the number of analysable trials, and to avoid problems arising from the above unsystematic elimination of data on the grounds of peak clipping, we then examined 100 “/, of the potentially examinable I8 trials. Also, for ease of record reading, and to be consistent with the data depicted in fig. I, this was a second-by-second rather than a beat-by-beat analysis. Fig. 2 shows the latency (bottom panel) and recruitment (top panel) characteristic of the BH UCR. Inspection of fig. 2 suggests that the latency and recruitment aspects emerging from the beat-by-beat examination also emerged from the second-by-second analysis. Specifically, as shown in the bottom decelerations panel of fig. 2, over 607: of the largest single second-to-second
Human Paolocian decelerative
cardiac conditionhe
169
43-
IO -
n
0 I2
3
4
5
6
SECONDS AFTER ONSET ON WHICH LARGEST SEC.- TO - SEC. DECELERATION OCCURRED
Fig. 2. Latency (bottom characteristics
panel) and recruitment and development-over-trials of the BH-elicited UR. (For explanation, see text.)
(top panel)
within the 2 set interval after instructed BH onset, indicating a short latency of UCR onset. Again, the fast-recruitment characteristic is evident in the top panel of fig. 2, which shows that the maximum deceleration for a single second-to-second interval was about 70”/, of the overall maximum deceleration of the 6 set interval after instructed BH onset. Finally, as indicated by the data shown in the top panel of fig. 2, there was no sign of any habituation over trials of this BH UCR, the mean magnitude of which remained over 28 BsPM. These data, obtained with a relatively small sample of subjects, indicated quite clearly that the obtained cardiac deceleration had the following desiderata for a UCR: large magnitude, short latency, fast recruitment and total lack of habituation. In all these respects, moreover, the BH UCR is clearly superior to UCRs elicited by aversive shock or noise UCSs within the ethical intensity limits for human experimentation. We should note, of course, that it has been previously reported that a number of respiratory manoeuvres (e.g. sustained 6 set exhalation following I set inhalation, or a 6 cycle/min breathing rate; cf. Laird and Fenz, 1971) can produce a large and fast cardiac deceleration. Accordingly, the BH aspect of the deceleration may not be uniquely critical. However, to our knowledge, the Pavlovian-control-related aspects of this occurred
170
BH-induced reflex ation over repeated emphasized. On the data on the question Data relevant to this
J. J. Fkredy md C. X. Po~dos
latency, precipitousness, magnitude, and lack of habitutrials - has not been previously clearly documented or other hand, the present experiment provided no direct of whether this response can be classically conditioned. issue were therefore collected in the next experiment.
3. Experiment II: Pavlovian conditioning of BH-induced cardiac deceleration In the absence of previous conditioning studies with this response, the choice of the interstimulus interval (ISI) to be employed was complicated by the fact that the onset of breath-holding was indeterminate by as much as half a second. In addition, any ISI that was long enough to allow observation of the CR during CS-UCS training trials (i.e. about 5 set) could result in the conditioning of an opposing accelerative response, since inhalation preceding BH would produce cardiac acceleration (see fig. I). With these considerations in mind, the IS1 chosen was I set with (instructed) BH being taken as UCS onset. This relatively short IS1 necessitated the use of CS-alone test trials to assess conditioning, an assessment which was carried out in a paradigm employing an 80% schedule of reinforcement (per cent CS-UCS training trials of total number of trials). Finally, to control for nonassociative factors, a group was run which received the same number and types of stimuli as the experimental group, but for whom the CS followed the UCS by IO sec. a ‘backward’ conditioning arrangement which is known to produce little or no conditioning in other response systems. 3.1. Method Twenty-four subjects were used in experiment II. The apparatus used in experiment 1 was again employed, but with the following modifications: (a) the temperature and humidity were those of the air-conditioned experimental room; and (b) earphones delivered a 72 dB, 1000 Hz tone {as the CS). The high-heat conditions of experiment 1 were eliminated after data gathered from several pilot subjects indicated that the BH UCR’s characteristics were unaffected by whether the box was hot or at normal temper~~ture. All subjects in this experiment were familiarized with the cued-breathing condition, as in experiment 1. In addition, it was emphasized to the subjects that during the time that cue lights were on (‘exhale’, ‘inhale’, and ‘hold’), their breathing patterns should be completely determined by the lights. On the other hand, once the ‘hold’ light went off, the subject was to ‘catch his breath’ as he normally would after breath-holding, and then resume and maintain his normal breathing cycle. For the experimental group of I2 subjects, 32 complete exhale-inhale-hold trials were presented, where the 3 set tone CS was initiated I set before the onset of the hold light. To assess decelerative conditioning, four trials were given where the tone CS was presented alone and these were interspersed
Human Pavlovian
decelerative
cardiac conditioning
171
among the 32 conditioning trials. The control group of 12 subjects received exactly the same procedures, except that on the 32 exhale-inhale-hold trials, the CS occurred 10 set following hold onset in a ‘backward’ UCS-CS arrangement. It will be noted that the instructions were designed to ensure that the subject would not hold his breath on these CS-alone test trials, but would continue to breathe normally. Respiration was continuously monitored to assess the subject’s compliance. Of the 12 experimental subjects there was one occasion where deviation from normal breathing occurred on a CS-alone test trial; on this occasion another test trial was given. In the case of the one other subject, when abnormal breathing occurred for the second time following a CS-alone test trial, the experiment was terminated, and this subject’s data were not used. To equate attention to the CSs in the two groups, all subjects were informed before the experiment that a tone might come on ‘just before the “hold” signal, just after the “hold” signal, or a long way from the “hold” signal’. They were instructed to note the percentage of these possible tone-‘hold’ relationships, and to provide this information at the end of the experiment. That this attention was equalized by the instructions is indicated by the fact that identification of the appropriate tone-hold relationship was made by all subjects in both groups. 3.2. Results and discussion The top panel of fig. 3 presents the second-to-second mean HR for the 5 set preceding CS onset (pre-CS baseline) and following CS onset (post-CS responding) on CS-alone test trials for the conditioning group; the bottom panel of fig. 3 presents the same data for the control group. Mixed-design analysis of variance (cf. Winer, 1962) applied to these data yielded a significant interaction between the pre-post and groups (conditioning versus controls) factors, F( I, 22) = 6.38, p < 0.05, and separate analyses of this interaction indicated that while post-CS HR in the experimental group was significantly lower than pre-CS HR, F( I, II) = 13.07, p < 0.005, this pre-post difference did not approach significance in the control group. Evidence for decelerative cardiac conditioning was also obtained at below the 0.005 level of significance when the mean maximum deceleration over the 5 set interval in the 12 experimental subjects (7.7 BsPM) was found to be more than double that in the control subjects (3.1 BsPm). Considering the relatively small sample of subjects involved, the reliability of the effect attained in this first attempt to condition the BH UCR deserves noting. On the other hand, as would be expected from any preliminary study, the data did present some problems. The first of these can be seen in fig. 3, where there was an obvious difference in pre-CS base HR between the two groups, a difference presumably due to sampling error. However, in terms of the law of initial values, the lower pre-CS baseline for the conditioning group would
172
J. J. Fweriy untl C. X. Pdos 63 62-
/’ 61 -
/’
/’
-A_
MEAN
l-9
-b/-WE-CS
‘\..
BASELINE 5
‘\
IJ
d
60 59 -
767574 73CONTROL G D. PRE-CS o--d POST -CS -
7271
I 1
;
;
i
;
ONE- SECOND INTERVALS
Fig. 3. Mean second-by-second cardiac rate of conditioning (N = 12) and control (N = 12) group 5 set before CS onset (pre-CS baseline) and 5 set following CS onset (post-CS rate).
operate to minimize the magnitude of the decelerative response for that group in comparison to the control group. Accordingly, the direction of this pre-CS baseline difference is such that it does not constitute a serious problem. A greater problem of interpretation is presented by the within-sessions data which, for the four CS-alone test trials in the conditioning group, yielded a U-shaped function (mean maximum deceleration within 5 set was 8.8, 8.5, 5.3 and 8.3 BsPM for the four test trials, respectively) rather than a monotonic increase to some asymptote. Speculations to account for this outcome are possible, but it seems more fruitful to emphasize that four test trials per subject are probably not sufficient to assess adequately within-session conditioning effects. Future studies will need to provide more trials, perhaps by repeated sessions with each subject. More generally, it would be surprising indeed if we have stumbled upon the parameters which are capable of producing optimal decelerative conditioning in the first small-scale study of this kind. Indeed, it is possible that the instructed BH itself is not the best way of producing the UCR, and that a method like tilting the subject’s head downward will produce as suitable a UCR while permitting a more precise temporal identification of the actual UCS. In addition, especially in terms of the potential clinical significance of this form of conditioning that was raised in the introduction, it would be essential to determine whether the BH-induced cardiac deceleration is accom-
Haman Parllocian deceleraticc
cardiac conditioning
173
panied by such undesirable responses as an increase in blood pressure. Nevertheless, the present data are sufficiently clear to suggest that the use of nonhabituating deceleration-inducing UC% merit considerable further investigation in the Pavlovian conditioning of human cardiac behaviour. Acknowledgement This research was supported
by Medical
Research
Council
Grant
MA-4889.
References Furedy, J. J. and Klajner, F. (1972). Unconfounded autonomic indices of the aversiveness of signaled and unsignaled shocks. Journal of Experimental Psychology, 93,313-3 18. Laird, C. S. and Fenz, W. D. (1971). Effects of respiration and heart rate in an aversive classical conditioning situation. Canadian Journal ofPsychology, 25, 396411. Winer, B. J. (1962). StatisticalPrinciples in Experimental Design. McGraw-Hill: New York, 302-318. Wood, D. M. and Obrist, P. A. (1964). Effects of controlled and uncontrolled respiration on the conditioned heart rate response in humans. Journal of Experimental Psychology, 68,221-229. Zeaman, D.. Deane, C. and Wegner, N. (1954). Amplitude and latency characteristics of the conditioned heart response: Journal ofPq,chology, 38, 235-250. Zeaman, D. and Smith. R. W. (1965). Review of some recent findings in human cardiac conditioning. In: Prokasy, W.‘F. (Ed.) Classical Conditioning: A Symposium. AppletonCentury-Crofts: New York.
PAVLOVIAN
DECELERATIVE
CARDIAC
ON A RESPIRATORY-INDUCED
DECELERATION
JOHN J. FUREDY
CARDIAC
AS AN UNCONDITIONAL
and CONSTANTINE
CONDITIONING
REFLEX
X. POULOS
Ull~l~~rsif~ of Toron#o,Torontu, ~~ltar~o, Cannda
Accepted for publication 29 August 1974
This study was directed at examining the feasibility of using a respiratory-induced cardiac decelerative reflex as an unconditioned response (UCR) in a Pavlovian conditioning paradigm. Experiment I assessed the reflexive features of the cardiac response under a respiratory procedure which involved 3 set of exhalation, 4 SW of inhalation, and then 30 set of breathholding (BH). The results indicated that the BH onset aspect of this respiratory cycle involved a phasic, large-magnitude cardiac deceleration (27 beats/min) with short latency, fast recruitment, and no indication of habituation over trials. Experiment 11 examined the feasibility of using BH onset as an unconditioned stimulus (UCS) in a Pavlovian paradigm by presenting an auditory conditional stimulus (CS) 1 set prior to the instructed BH aspect of the respiratory cycle. A separate control condition involved presenting the CS 10 set after BH onset which constituted a ‘backward’ conditioning control. The results indicated that the conditioning group displayed a significant cardiac deceleration on CS-alone test trials (under normal breathing) in contrast to the absence of such an effect for the control group: In brief, the study suggested that a decelerative conditioned response (CR) could be established by using a respiratory-induced cardiac deceleration as a Pavlovian UCS. Potential clinical applications of such a decelerative CR were noted.
I. Introduction The area of human classical cardiac conditioning has produced confusing and paradoxical data with regard to the topography of the conditioned response (CR). Human cardiac conditioning has by and large relied on using aversive events such as shock as the unconditional stimulus (UCS) which elicit cardiac acceleration as the unconditional response (UCR). However, contrary to the classical conditioning proviso that the CR be topographi~lly similar to the UCR, the human cardiac CR based on aversive UCSs has been variously identified as biphasic (Zeaman, Deanne and Wegner, 1954), decelerative (Wood and Obrist, 1964), and accelerative (cf. Zeaman and Smith, 1965). Another source of difficulty is that, due to ethical considerations, the stressful stimuli typically used as UCSs have been relatively weak and UCR habituation is often extensive. This study is unusual in that it involved an instructed respiration cycle 165
which terminated with breatli-holding (BH) as a nonaversive UCS. As will be shown in more detail below, this UCS elicits a phasic cardiac deceleration as the UCR, the CR is of similar form, i.e. decelerative, and there is little or no UCR habituation over repeated trials. Finally, in addition to its potential use for resolving theoretical problems in human cardiac conditioning, the preparation based on the BH UCS also has potential clinical use. Specifically, there is a class of cardiac dysfunctions which consist of inappropriately large cardiac accelerations to sudden stress. One potential form of behavioural control of such dysfunctions is that which relies on operant-feedback methods, and, with the current enthusiasm over biofeedback, most attention has been focused on this form of control. However, it would seem feasible and desirable also to examine the possibilit}~ of achieving behavioural control through Pavlovian methods, provided that a clearly deceierative CR was available. For example, if the desired modification involved a decrease in the heart rate (HR), then a subject-generated CS. to which a decelerative CR has been previously established, could be utilized to decrease HR whenever wanted or needed. The BH UCS preparation to be described in this paper could provide just such a potential clinical tool.
2. Experiment I: the BH-induced UCR The data to be reported in experiment I were gathered originally with ultimate operant rather than Pavlovian behavioural control in mind. Specifically, we were dissatisfied with the magnitude and reliability of noise- and shockelicited cardiac accelerations (which were to be used as the target responses for later studies on operant modification) and desired to increase such accelerations by controlled respiration (cf. Laird and Fenz. 1971). The method of respiratory control chosen was one where subjects exhaled, inhaled, and then held their breath for 30 sec. The last phase of this instructed cycle provided the UCS for the elicited deceleration, and the issues to be reported in experiment I concern such features of this BH-induced UCR as magnitude, latency, topography and degree of habituation.
The subjects were ten undergraduates from the University of Toronto, who were paid $2.00 for participation. The apparatus consisted of a hot face box which has been detailed elsewhere (Furedy and Klajner, 1972). Briefly, the subject put his face into the box which has an internal temperature of 110°F (80% humidity, i.e. a little cooler than a steam bath). inside the box were labelled indicator lights which were used to indicate to the subject when he should exhale, inhale, hold, and resume normal breathing. The HR was picked up through electrodes placed either on the forearms or on the sternum (for those subjects on which forearm attachments provided an inadequate signal) and recorded through a physiograph (E & M) tachometer. Respiration
Human Pnulovion deceleratiue
cardiac conditioning
167
was picked up by a bellows pneumograph placed around the chest, and also recorded through the physiograph. All stimuli were delivered automatically by a system of relays, timers, and a paper-tape recorder. The subject and experimenter were in separate rooms, but both visual and auditory contact was maintained through a one-way screen and a microphone arrangement. Prior to the experimental session, each subject was familiarized with the cued-breathing condition which involved the following sequence (indicated by the lights in the box): 3 set of even exhalation, 4 set of even inhalation, and, finally, 30 set of breath-holding (see also the abscissa of fig. I). The immediately following experimental session involved 24 presentations of the cued-breathing trial with rest periods of normal breathing of 40-60 set (varied unsystematically) between the termination of one trial and the initiation of the next trial. In addition, during 18 of the 24 breath-holding periods, a loud-noise stimulus (0.3 set I 15 dB white noise) was presented at various intervals during the breath-hold cycle (i.e. at 2, 10 or 25 set following BH onset). The purpose of presenting the loud noise was to study cardiac acceleration under various respiration conditions, but these noise-induced acceleratory data will not be reported here, as they are relevant only to the original operant-control interest mentioned above. 2.2. Results and discussion A general picture of the BH-induced UCR was obtained second-by-second during all of the six controlled-respiration
Fig. 1. Mean
second-by-second cardiac rate (in BsPM) during controlled-respiration
by measuring HR trials where no
of ten subjects, trials.
2 set prior
to and
168
J. J. Fkredv and C. X. POI//OF
loud-noise stimulus had been presented. Fig. I presents these results. The three clear trends are: (a) the inspiration-induced acceleration: (b) the BHinduced deceleration; and (c) the mean return to baseline following the phasic BH UCR. Moreover, inspection of individual responses indicated that, because of the averaging procedures used for fig. I, the latency and recruitment characteristics of the BH UCR were actually obscured by the relatively dramatic representation depicted in fig. I. For example, an individual’s second-by-second cardiac rate (in BsPM) from -2 to IO set following the onset of a respiratory trial was : 69,68 (baseline before exhalation), 65,66, 68 (during exhalation), 70, 78, 82, 88, 89 (during inhalation), 88, 62, 63 (following hold). Such individual BH UCRs, then, showed not only short latency (within one or two beats of BH), but also fast recruitment, with most of the decelerative response occurring suddenly between a pair of beats. To obtain more systematic evidence regarding these latency and recruitment characteristics of the BH UCR, we sought to employ the maximum number of trials as well as ensuring that at least some of the relevant data were based on an independent verification that the instructed BH had, in fact, occurred at the time instructed to the subject by the experimenter. The BH UCR itself (though not its recovery) could potentially be examined on I8 of the 24 respiratory-control trials (i.e. all those trials except where the acceleration-inducing loud-noise stimulus occurred 2 set following BH). On approximately 40”/, of these trials it was possible to exactly verify the point of actual (in contrast to instructed) BH onset. On other trials, unavoidable ‘peak clipping’ (i.e. the respiration-recording pen going off scale) precluded this precise specification. On 40”/) of the trials, data involving five beats following actual BH onset wereexamined, and the mean maximum deceleration occurring between any pair of beats after hold onset was 20.01 BsPM. The latency of these precipitous single beat-to-beat decelerations was short, as indicated by the fact that 80% of them occurred within the three beats after BH onset. That the recruitment of the BH UCR was fast is indicated by the fact that the single beat-to-beat deceleration value of 20.01 BsPM accounted for more than 70”/, of the overall beat-to-beat deceleration (27.3 BsPM) during the five-beat interval following BH onset. To maximize the number of analysable trials, and to avoid problems arising from the above unsystematic elimination of data on the grounds of peak clipping, we then examined 100 “/, of the potentially examinable I8 trials. Also, for ease of record reading, and to be consistent with the data depicted in fig. I, this was a second-by-second rather than a beat-by-beat analysis. Fig. 2 shows the latency (bottom panel) and recruitment (top panel) characteristic of the BH UCR. Inspection of fig. 2 suggests that the latency and recruitment aspects emerging from the beat-by-beat examination also emerged from the second-by-second analysis. Specifically, as shown in the bottom decelerations panel of fig. 2, over 607: of the largest single second-to-second
Human Paolocian decelerative
cardiac conditionhe
169
43-
IO -
n
0 I2
3
4
5
6
SECONDS AFTER ONSET ON WHICH LARGEST SEC.- TO - SEC. DECELERATION OCCURRED
Fig. 2. Latency (bottom characteristics
panel) and recruitment and development-over-trials of the BH-elicited UR. (For explanation, see text.)
(top panel)
within the 2 set interval after instructed BH onset, indicating a short latency of UCR onset. Again, the fast-recruitment characteristic is evident in the top panel of fig. 2, which shows that the maximum deceleration for a single second-to-second interval was about 70”/, of the overall maximum deceleration of the 6 set interval after instructed BH onset. Finally, as indicated by the data shown in the top panel of fig. 2, there was no sign of any habituation over trials of this BH UCR, the mean magnitude of which remained over 28 BsPM. These data, obtained with a relatively small sample of subjects, indicated quite clearly that the obtained cardiac deceleration had the following desiderata for a UCR: large magnitude, short latency, fast recruitment and total lack of habituation. In all these respects, moreover, the BH UCR is clearly superior to UCRs elicited by aversive shock or noise UCSs within the ethical intensity limits for human experimentation. We should note, of course, that it has been previously reported that a number of respiratory manoeuvres (e.g. sustained 6 set exhalation following I set inhalation, or a 6 cycle/min breathing rate; cf. Laird and Fenz, 1971) can produce a large and fast cardiac deceleration. Accordingly, the BH aspect of the deceleration may not be uniquely critical. However, to our knowledge, the Pavlovian-control-related aspects of this occurred
170
BH-induced reflex ation over repeated emphasized. On the data on the question Data relevant to this
J. J. Fkredy md C. X. Po~dos
latency, precipitousness, magnitude, and lack of habitutrials - has not been previously clearly documented or other hand, the present experiment provided no direct of whether this response can be classically conditioned. issue were therefore collected in the next experiment.
3. Experiment II: Pavlovian conditioning of BH-induced cardiac deceleration In the absence of previous conditioning studies with this response, the choice of the interstimulus interval (ISI) to be employed was complicated by the fact that the onset of breath-holding was indeterminate by as much as half a second. In addition, any ISI that was long enough to allow observation of the CR during CS-UCS training trials (i.e. about 5 set) could result in the conditioning of an opposing accelerative response, since inhalation preceding BH would produce cardiac acceleration (see fig. I). With these considerations in mind, the IS1 chosen was I set with (instructed) BH being taken as UCS onset. This relatively short IS1 necessitated the use of CS-alone test trials to assess conditioning, an assessment which was carried out in a paradigm employing an 80% schedule of reinforcement (per cent CS-UCS training trials of total number of trials). Finally, to control for nonassociative factors, a group was run which received the same number and types of stimuli as the experimental group, but for whom the CS followed the UCS by IO sec. a ‘backward’ conditioning arrangement which is known to produce little or no conditioning in other response systems. 3.1. Method Twenty-four subjects were used in experiment II. The apparatus used in experiment 1 was again employed, but with the following modifications: (a) the temperature and humidity were those of the air-conditioned experimental room; and (b) earphones delivered a 72 dB, 1000 Hz tone {as the CS). The high-heat conditions of experiment 1 were eliminated after data gathered from several pilot subjects indicated that the BH UCR’s characteristics were unaffected by whether the box was hot or at normal temper~~ture. All subjects in this experiment were familiarized with the cued-breathing condition, as in experiment 1. In addition, it was emphasized to the subjects that during the time that cue lights were on (‘exhale’, ‘inhale’, and ‘hold’), their breathing patterns should be completely determined by the lights. On the other hand, once the ‘hold’ light went off, the subject was to ‘catch his breath’ as he normally would after breath-holding, and then resume and maintain his normal breathing cycle. For the experimental group of I2 subjects, 32 complete exhale-inhale-hold trials were presented, where the 3 set tone CS was initiated I set before the onset of the hold light. To assess decelerative conditioning, four trials were given where the tone CS was presented alone and these were interspersed
Human Pavlovian
decelerative
cardiac conditioning
171
among the 32 conditioning trials. The control group of 12 subjects received exactly the same procedures, except that on the 32 exhale-inhale-hold trials, the CS occurred 10 set following hold onset in a ‘backward’ UCS-CS arrangement. It will be noted that the instructions were designed to ensure that the subject would not hold his breath on these CS-alone test trials, but would continue to breathe normally. Respiration was continuously monitored to assess the subject’s compliance. Of the 12 experimental subjects there was one occasion where deviation from normal breathing occurred on a CS-alone test trial; on this occasion another test trial was given. In the case of the one other subject, when abnormal breathing occurred for the second time following a CS-alone test trial, the experiment was terminated, and this subject’s data were not used. To equate attention to the CSs in the two groups, all subjects were informed before the experiment that a tone might come on ‘just before the “hold” signal, just after the “hold” signal, or a long way from the “hold” signal’. They were instructed to note the percentage of these possible tone-‘hold’ relationships, and to provide this information at the end of the experiment. That this attention was equalized by the instructions is indicated by the fact that identification of the appropriate tone-hold relationship was made by all subjects in both groups. 3.2. Results and discussion The top panel of fig. 3 presents the second-to-second mean HR for the 5 set preceding CS onset (pre-CS baseline) and following CS onset (post-CS responding) on CS-alone test trials for the conditioning group; the bottom panel of fig. 3 presents the same data for the control group. Mixed-design analysis of variance (cf. Winer, 1962) applied to these data yielded a significant interaction between the pre-post and groups (conditioning versus controls) factors, F( I, 22) = 6.38, p < 0.05, and separate analyses of this interaction indicated that while post-CS HR in the experimental group was significantly lower than pre-CS HR, F( I, II) = 13.07, p < 0.005, this pre-post difference did not approach significance in the control group. Evidence for decelerative cardiac conditioning was also obtained at below the 0.005 level of significance when the mean maximum deceleration over the 5 set interval in the 12 experimental subjects (7.7 BsPM) was found to be more than double that in the control subjects (3.1 BsPm). Considering the relatively small sample of subjects involved, the reliability of the effect attained in this first attempt to condition the BH UCR deserves noting. On the other hand, as would be expected from any preliminary study, the data did present some problems. The first of these can be seen in fig. 3, where there was an obvious difference in pre-CS base HR between the two groups, a difference presumably due to sampling error. However, in terms of the law of initial values, the lower pre-CS baseline for the conditioning group would
172
J. J. Fweriy untl C. X. Pdos 63 62-
/’ 61 -
/’
/’
-A_
MEAN
l-9
-b/-WE-CS
‘\..
BASELINE 5
‘\
IJ
d
60 59 -
767574 73CONTROL G D. PRE-CS o--d POST -CS -
7271
I 1
;
;
i
;
ONE- SECOND INTERVALS
Fig. 3. Mean second-by-second cardiac rate of conditioning (N = 12) and control (N = 12) group 5 set before CS onset (pre-CS baseline) and 5 set following CS onset (post-CS rate).
operate to minimize the magnitude of the decelerative response for that group in comparison to the control group. Accordingly, the direction of this pre-CS baseline difference is such that it does not constitute a serious problem. A greater problem of interpretation is presented by the within-sessions data which, for the four CS-alone test trials in the conditioning group, yielded a U-shaped function (mean maximum deceleration within 5 set was 8.8, 8.5, 5.3 and 8.3 BsPM for the four test trials, respectively) rather than a monotonic increase to some asymptote. Speculations to account for this outcome are possible, but it seems more fruitful to emphasize that four test trials per subject are probably not sufficient to assess adequately within-session conditioning effects. Future studies will need to provide more trials, perhaps by repeated sessions with each subject. More generally, it would be surprising indeed if we have stumbled upon the parameters which are capable of producing optimal decelerative conditioning in the first small-scale study of this kind. Indeed, it is possible that the instructed BH itself is not the best way of producing the UCR, and that a method like tilting the subject’s head downward will produce as suitable a UCR while permitting a more precise temporal identification of the actual UCS. In addition, especially in terms of the potential clinical significance of this form of conditioning that was raised in the introduction, it would be essential to determine whether the BH-induced cardiac deceleration is accom-
Haman Parllocian deceleraticc
cardiac conditioning
173
panied by such undesirable responses as an increase in blood pressure. Nevertheless, the present data are sufficiently clear to suggest that the use of nonhabituating deceleration-inducing UC% merit considerable further investigation in the Pavlovian conditioning of human cardiac behaviour. Acknowledgement This research was supported
by Medical
Research
Council
Grant
MA-4889.
References Furedy, J. J. and Klajner, F. (1972). Unconfounded autonomic indices of the aversiveness of signaled and unsignaled shocks. Journal of Experimental Psychology, 93,313-3 18. Laird, C. S. and Fenz, W. D. (1971). Effects of respiration and heart rate in an aversive classical conditioning situation. Canadian Journal ofPsychology, 25, 396411. Winer, B. J. (1962). StatisticalPrinciples in Experimental Design. McGraw-Hill: New York, 302-318. Wood, D. M. and Obrist, P. A. (1964). Effects of controlled and uncontrolled respiration on the conditioned heart rate response in humans. Journal of Experimental Psychology, 68,221-229. Zeaman, D.. Deane, C. and Wegner, N. (1954). Amplitude and latency characteristics of the conditioned heart response: Journal ofPq,chology, 38, 235-250. Zeaman, D. and Smith. R. W. (1965). Review of some recent findings in human cardiac conditioning. In: Prokasy, W.‘F. (Ed.) Classical Conditioning: A Symposium. AppletonCentury-Crofts: New York.