PSYCHOPHYSIOL~GY Copyright 0 1991 by The Society for Psychophysiological Research, Inc.
Vol. 28, No. I Printed in U.S.A.
Weak Sensory Stimuli Induce a Phase Sensitive Bradycardia J. RICHARD JENNINGS, MAURITSW. VAN DER MOLEN, RIEKJ.M. SOMSEN, AND KAY BROCK University of Pittsburgh and University of Amsterdam
ABSTRACT We attempted to demonstrate that significant perceptual stimuli would induce different degrees of heart rate deceleration depending on when (phase) in the cardiac cycle they occurred. Relative to previous work, we concurrently examined a number of factors that might alter the amplitude of such a cardiac cycle time effect. Stimulus intensity and presence or absence of a speeded response were manipulated. Liminal stimuli and a perceptual rather than motor set were expected to maximize any cardiac cycle time effect. Respiratory phase, length of average interbeat interval, and number of trials were also investigated. Twenty-four college aged, male volunteers were randomly separated into equal groups receiving instructions either to judge which of two weak visual stimuli occurred or to execute a speeded, discriminative response to the stimuli. Discriminative stimuli were presented at either 0, 150, 250, 350, or 500 ms after the R-wave of the electrocardiogram. Stimuli were presented with an intensity that had yielded either 63% or 90% correct detections in a prior psychophysical assessment. A phase dependent deceleration occurred after both intensities of stimuli. Poststimulus deceleration was greater for stimuli in early to mid cycle as suggested by earlier work. As expected, this result was clear when the stimuli were presented during the expiratory phase of respiration. Neither perceptnal/motor set nor stimulus intensity altered the phase sensitive deceleration. Thus, phase sensitive deceleration was confirmed using demanding sensory stimuli and an improved representational technique. DESCRIPTORS Cardiac cycle time effect, Threshold stimuli, Heart rate deceleration, Phase sensitivity, Respiration.
The Laceys (e.g., 1974) have suggested that the and empirical grounds. Both Somsen, Jennings, and processing of a psychologically significant stimulus van der Molen (1988) and Velden, Barry, and Wolk elicits an immediate lengthening of cardiac inter- (1987) observed that the prior representations of beat interval. If the stimulus occurs in the second phase dependent stimulus induced effects may have quarter of the interval, e.g. 300-400 ms after the confounded anticipatory and not phase dependent R-wave of the EKG, greater deceleration is elicited effectswith any true phase dependent changes. We than if it occurs very early or late in the interval. (Jennings, van der Molen, & Terezis, 1987; JenBecause the size of the deceleration depends on the nings, van der Molen, Somsen, & Terezis, 1990) time of stimulus occurrence within the cardiac in- have also performed a number ofexperiments varyterbeat interval, such an effectcan be called a phase ing the phase timing ofstimuli, i.e., when the stimdependent effect or a cardiac Cycle time effect. The ulus occurred in the interbeat interval. In these exLaceYs (e-g.9 1974) termed the Phase dependent de- periments response timing, i.e. when the response celeratory change, primary bradycardia. occurred in the interbeat interval, appeared to show Recently, the reality of such Phase dependent a phase dependent influence on heartbeat timing, effects has been questioned on both methodological but stimulus timing did not-i.e., bradycardia was not evident. Both of these experiments, however, used a speed accuracy tradeoff task in We like to thank NIMH grant MH 40418 for which volunteers were paid to respond to stimuli its support of the work described and NWO grant 02-65020 for its support of Dr. van der Molen in related work. with different reaction speeds. This emphasis on Address requests for reprints to: J. Richard Jennings, motor performance may have created a 'motor WPIC, Room El 329, 381 1 O'Hara Street, Pittsburgh, PA (Garner, 1962) that could have been incompatible with stimulus induced cardiac deceleration. 15213. 1
2
Jennings, van der Molen, Somsen, and Brock
In sum, demonstrations of primary bradycardia to date can be faulted because procedures were used that were likely to obscure phase dependent effects. Frequently, reaction time paradigms were used in which response-related acceleration as well as anticipatory deceleratory effects can induce changes similar to primary bradycardia (Jennings et al., 1987). In other cases, the influence of respiratory phase or length of average interbeat interval has been ignored (see discussion in Coles, Pellegrini, & Wilson, 1982). Finally, representation techniques have often obscured rather than illuminated phase dependent effects. The purpose of the current experiment was to test whether or not phase dependent cardiac deceleration would be elicited by liminal stimuli. Discriminative stimuli were presented at five phases of the cardiac cycle varying between 0 and 500 ms after the R-wave of the electrocardiogram. Decelerative change was expected to be maximal for phases in the second quarter of the interbeat interval. Liminal stimuli were used to increase the likelihood of eliciting a poststimulus deceleration. Poststimulus cardiac deceleration has generally been relatively larger for liminal than for moderate intensity stimuli (Jackson, 1974). In the current experiment we varied stimulus intensity between a level just above the psychophysical threshold and a significantly more intense level. We further instructed one group of subjects to closely assess the stimulus and then give us a choice response-a perceptual set group; another group was instructed to make a choice response as quickly as possible in order to receive incentive pay for adequate speed of response-a motor set group. Low intensity stimuli and perceptual set were expected to enhance any phase dependent poststimulus deceleration. Conceptually, these expectations are based on the Graham and Clifton (1966) integration of the thinking of Sokolov and the Laceys. Cardiac deceleration is viewed as a component of the orienting response. Perceptual identification of a novel or significant liminal stimulus is presumed to take longer than comparable analysis of a more intense stimulus. A similar effect might be expected when instructions emphasize perceptual analysis rather than motor speed. Three other factors were examined in our attempt to define conditions favorable to the identification of primary bradycardia. First, stimuli were classified as occurring during either respiratory inspiration or expiration. Primary bradycardia is expected to occur during expiration (Coles et al., 1982). Second, volunteers with longer average interbeat intervals were separated from those with shorter interbeat intervals. Lacey and Lacey (1 974)
Vol. 28, No. 1
suggested that primary bradycardia would be more evident in individuals with longer interbeat intervals. The rationale for these two factors is primarily physiological. Central nervous system activity related to the perceptual detection of a significant stimulus is presumed to induce vagal activation. The efficacy of this activation is then modulated by the current state of cardiorespiratory control. Long interbeat intervals will permit different neurotransmitter or membrane kinetics in the cardiac pacemaker relative to short interbeat intervals (Lacey & Lacey, 1974; Somsen, Molenaar, van der Molen, & Jennings, in press). Inspiratory phase will inhibit vagal effects on the pacemaker whereas expiratory phase will permit it (Coles et al., 1982). The third factor examined was methodological rather than physiological. A large number of trials were given so that we could estimate the trial block length adequate for observing primary bradycardia. Finally, the results are graphically represented and analyzed using an old phasedelta phase procedure that we have recently described (Jennings, van der Molen, Somsen, & Ridderinkhof, in press). We avoid artifactual phase dependency by presenting the results as a function of real time relative to the stimulus rather than as a function of ordinal interbeat interval. This analytic technique also permits us to separate anticipatory effects from poststimulus changes. Thus, an analytic technique is used that is designed to avoid the representation and interpretive problems raised by Somsen et al. (1988) and Velden et al. (1987). Method Subjects
Twenty-four male volunteers aged 18-30 were recruited from the University of Pittsburgh campus to participate in the study. All volunteers were screened to ensure good health and absence of drug use. Experimental Task The volunteers were asked to identify one of two figures presented by a seven-segment light-emitting diode (LED) situated on a display board and to press one of two microswitch response keys corresponding to the figure identified. The display board, a 66 X 66 cm plexiglass sheet painted a flat black color, was mounted on a microphone stand and placed 1 18.5 cm in front of the semireclining volunteer in a soundproof chamber. The illumination of four segments of the seven-segment LED display formed the figures, which appeared as either "[" or "I". The 1.5 X 3 cm figure appeared in the center of the display board. Two intensity levels on the LED display were used corresponding to the volunteers' near threshold level (63% detection) and slightly brighter than the threshold (90% detection). The intensity of the figure was controlled
January, 1991
Phase Sensitive Primary Bradycardia
by altering the DC component of a modulated DC voltage (Nygaard & Frumkes, 1982). The modulation frequency was 1 KHz. A set of microswitch response keys was placed on the left or right arm of the chair, depending upon the handedness of the volunteer. Each trial began with the I-s illumination ofa warning signal which was a horizontal bar in the center of the LED display. The intensity of the warning bar was relatively bright and constant regardless of the intensity of the stimulus. The time of warning signal presentation was determined by the occurrence of the Rwave of the volunteer’s electrocardiogram and a preselected cycle delay time. During the following interbeat interval at the preselected cycle delay time, the discriminative stimulus appeared for 50 ms at the a p propriate intensity level. Timing of the stimulus as well as physiological data collection were controlled by a DEC 1173 MINC computer. After the volunteer’s response, a 2-s delay was followed by a 1-s presentation of feedback to the volunteer. A 1-3 s randomly vanable intertrial interval followed the termination of the feedback illumination. The five cardiac cycle delays were 0, 150,250, 350, and 500 ms. The cardiac cycle delay was randomly selected for each trial. Delays were equally probable, i.e., p=.2. At the end of the intertrial interval, the computer compared the elapsed time since the last Rwave and the delay randomly assigned for that trial. If the elapsed time was greater than the assigned delay, then the warning signal was presented at the assigned delay time after the next heartbeat. If the elapsed time was less than the assigned delay, then the warning stimulus was presented at the assigned delay time during the current interbeat interval. The discriminative stimulus appeared at the same assigned delay during the interbeat interval following that of the warning stimulus. With brief foreperiods, foreperiod length and delay (phase) time can become confounded unless warning and discriminative stimuli have the same delay time (Jennings et al., 1987). The 24 volunteers were divided randomly into two groups of 12. Both groups viewed the same discriminative stimuli, but one group performed a speeded motor response. Volunteers in this motor set group were required to press the microswitch response key (with their dominant hand) identifying the stimulus within 400-600 ms. Reaction times were measured using the clock on the DEC MINC computer. Feedback on incentive pay earned was based on both correctness of response and speed for this group. Volunteers in the other group did not perform under the speed condition, although for computer programming ease a time limit of 2000 ms was set. The feedback display for this perceptual set group was based solely on accuracy of choice. The intensity level of the feedback display was the same as that of the warning signal. Feedback served to maintain volunteers’ motivation and was presented on the same LED display as the discriminative stimulus. The volunteers received a bonus depending on their performance in addition to an hourly pay. The motor set group received $1.00 for
3
each set of 50 trials in which at least 50% of their responses were within the speed target and accurate, $SO for sets with between 25%and 50% on speed and accurate, and $.25 for sets with less than 25% on speed and accurate. The perceptual set group received S 1.OO for 80% or more correct discriminations within a 50trial set, $ S O for between 50% and 80% correct, and $.25 if less than 50% correct.
Procedure On the first day, the volunteers were trained to perform the task and the two intensity levels were established using a staircase method following Levitt (1 97 1, see below). Electrodes for physiological recording were attached, but no data were collected. First, a sample version of the experimental task was administered to orient the volunteers to the experiment. The perceptual set group received instructions emphasizing accuracy of identification and the motor set group received instructions placing an additional emphasis on the response time requirement. A 25-trial practice task was then administered with clearly visible stimuli. If any volunteer needed more practice in order to understand the task, a second 25-trial task was given. Each volunteer’s psychophysical sensitivity was then determined. The volunteers were adapted to almost complete darkness in the soundproof chamber for 10 min. A flashlight positioned on the floor was the only light source. The batteries were replaced each day for each volunteer to maintain this level. One of the two stimuli was randomly chosen by flipping a coin to serve as the constant test stimulus presented at varying intensities. The volunteers were instructed to press one microswitch response key if they detected the stimulus, and the other key if they could not see the stimulus. The warning bar appeared 1 s before the test stimulus. The volunteers were told that there was no time criterion in making their response, but a response was required after each stimulus in order for the procedure to continue. Four consecutive test stimuli were presented at the same intensity level. If 3 or 4 of these were detected, the intensity level decreased. When the volunteer detected 1 or 0, the intensity level increased. If 2 stimuli were detected, the same intensity level was used on the next 4 figures. The starting level for all volunteers was 1100 digital converted voltage units. The initial step size in increasing or decreasing intensity was 10. When the volunteer made a changeover (from detecting 3 or 4 to detecting 1 or 0 or vice versa), the step size decreased to 5. At the next changeover, the step size decreased to 2 and finally to 1. After the last changeover, 15 sets of 4 test stimuli were presented increasing or decreasing intensity by a step size of 1. The 63% correct detection level to be used as the low intensity value in the experimental task was computed by recording the average of the higher and lower intensities when a changeover occurred and then averaging all of those values. The higher intensity level to be used in the experimental task was computed by
4
Jennings, van der Molen, Somsen, and Brock
adding two standard deviations of those values to the threshold value. This value approximated 90% correct detection. Volunteers varied in the number of test stimuli required to determine their threshold level. The duration of the procedure ranged from 15 to 25 min. After the staircase procedure, volunteers were offered a break and reminded of the original task instructions. During this time, they remained in the chamber with the light level constant. After the break the volunteers received two 25-trial sets of the performance task with the brighter intensity stimuli and two 25-trial sets at the lower level. Within a set, the intensity level of the stimulus was constant. At the end of the session, light levels were slowly adjusted upward until the volunteer could comfortably leave the chamber. The practice day was followed by three experimental days. The experimental session began with the application of the physiological equipment. A 10-min baseline period preceded the performance task. During the baseline, the display board was unplugged and the physiological signals were recorded. The flashlight was positioned on the floor while the volunteer adjusted to the environmental light level. The volunteers received one 15-trial set at each intensity level established on the training day to verify the values. If the volunteer reported the stimulus extremely easy or difficult to detect during either set (substantiated by his performance), the staircase procedure was repeated. However, the starting level was the threshold value determined on the practice day. This shortened the time required to assess a more accurate threshold for that day. The new values were verified in two additional 15-trial sets. Once the threshold levels had been established, the volunteer received 6 sets of 50 trials of the performance task. Three of the sets presented the stimuli at the 63% level and 3 of the sets used the brighter level. The order of the sets was randomized within a day by flipping a coin. The volunteer was informed which level would be used before each set. If at any time during the study the volunteer was exposed to outside light, he spent 10 min readjusting to the darkened chamber. Physiological Measures Measurements of heart rate, peripheral vascular a o tivity, electromyography, blood pressure, and respiration were taken. Heart rate was measured as interbeat intervals (IBI) between R-waves of the electrocardiogram using external inputs to the Minnesota Cardiograph Model 304 Band the computer clock with an accuracy of 1 ms. Six interbeat intervals were recorded for each discriminative stimulus presentation3 before the stimulus, the interbeat interval of the discriminative stimulus, and 2 following the stimulus. For purposes of comparability,baseline data were collected using the same computer program, although the volunteers did not actually see or respond to any LEDs. Two electrodes in conjunction with four impedance bands of cardiograph electrode tape were used to obtain the electrocardiogram.A modified lead I1 place-
Vol. 28, No. 1
ment was used with one Beckman electrode placed caudal to the right pectoral muscle and one placed in the V5 position below the heart. Peripheral vascular activity was monitored with the pulse obtained from the impedance cardiograph and with a photoplethysmograph attached to the thumbnail of the nondominant hand of the volunteer. The first impedance band was placed around the base of the neck with the second band 2-3 cm above the first. The third impedance band was placed around the xiphoid process (caudal termination of sternum) and the fourth impedance band was placed 3-4 cm below the third. The third band served as ground for both impedance and electrocardiogram measures. A blood pressure cuff was also placed on the dominant arm and a heart sounds microphone positioned over the apex of the heart. Vascular results from these measures are reported elsewhere because they do not relate to the issues addressed in this report. Respiration was measured using a Yellow Springs thermistor placed at the opening of the right nostril. The respiration signal was amplified and conditioned by a Grass differentiator, the first derivative was taken and a positive signal (a square wave) was produced during inspiratory slope; the signal was at ground during expiration and expiratory pause. This signal was sent to a Coulbourn bipolar comparator for a finer adjustment of the threshold. Then the signal was sent to the digital input of the computer permitting a trialby-trial recording of respiratory phase coded as inspiration or expiration at the time of the discriminative stimulus.
Results Performance
Stimulus intensity and the perceptual/motor set instructions yielded the expected results for reaction time and proportion correct. Table 1 shows the results for both variables classified by the stimulus intensity and the instruction set groups-perceptual or motor set. Both reaction time and proportion correct were examined with an analysis of variance with factors of instructional set, phase of stimulus presentation, and stimulus intensity. Low intensity stimuli were associated with both a lower percentage of correct detections as well as slower detections. As instructed, motor set volunteers produced more rapid detections. Surprisingly, the motor set volunteers also produced more accurate responses than the perceptual set volunteers. Furthermore, the overall level of correct choices was higher than anticipated from our psychophysical adjustments. The effect of discrimination task with performance incentives was clearly different from the effect of the detection task used to determine thresholds. The statistical results for reaction time showed main effects of perceptual/motor set (F(1/22)=
January, 1991
Phase Sensitive Primary Bradycardia
5
Table 1
Reaction time and proportion correct as a function of instructional set and stimulus intensity Perceptual Set (11-12)
Motor Set (n=12)
Stimulus Intensity
Reaction Time (ms)
Proportion
Reaction Time (ms)
Proportion
Correct
Near Threshold (63% detection) Above Threshold (90% detection)
748
.90
496
.96
709
.93
482
.97
Correct
6 1.9) and stimulus intensity (F(1/22)= 12.9). Similarly, for proportion correct, there were significant main effects for set (F(1/22)=6.6) and intensity (F(1/22)=23.6). In these and all analyses a rejection region of p6.2, H-F adj.=.88) within expiration. No significant intensity effects were found in the inspiration results. Phase dependent influences on interbeat interval. The presence of clear deceleration at the time of the stimulus permits us to ask the central question: Does the processing of weak perceptual stimuli induce a short latency cardiac deceleration that depends upon when in the cardiac cycle the stimuli occur? Phase sensitive changes in interbeat interval were assessed with a recently developed graphical and statistical procedure (Jennings et al., in press). The graphical procedure plots the location of heartbeats in real time relative to stimulus presentation on the x axis. On the y axis differences from a base interbeat interval are plotted at the time of the heartbeat completing the interval. Figure 2 presents such a plot of the expiration data using the initial prestimulus interval (third interval prior to the stimulus) as the base interval. Separate beat sequences are plotted for each phase or cycle time relative to stimulus timing. For example, for the beat sequence centered on a phase time of 0 msstimulus occurring on the R-wave-the beat ending the last prestimulus interval is plotted at exactly 0 on the x axis. Correspondingly, for the beat sequence centered on a phase time of 500 ms, the
VoI. 28, No. 1
beat ending the last prestimulus interval is plotted at - 500 on the x axis. Note how for the prestimulus beat, the 0-ms phase point is on the extreme right of the cluster of points and the 500-ms point is on the extreme left. This ordinal relationship is maintained as the other beats from these sequences are plotted. The results are averaged across intensity condition and perceptual/motor set grouping, but presented in separate figures for respiratory inspiration and expiration. Linear functions have been fit with a least square procedure across the points for the different phase values at each heartbeat. Within each cluster of points (corresponding to a particular beat), the difference scores show an orderly relationship with phase timing. Linear and higher order trends can be fit to the data and tested against the experimental variables with an analysis of variance (Winer, 1971). We are primarily interested in the trend across phase position. Trends can be estimated across phase position for each beat and then the equivalence of these trends can be tested with the interaction of Heartbeat and the trend for Phase position. Such an analysis was performed with factors for percep tual/motor set, heartbeat, respiratory inspiration versus expiration, stimulus intensity, and phase of stimulus presentation. Linear and higher order trends were examined for the phase factor. The interactions of Heartbeat and the trends for Phase were the primary focus of the analysis. Phase Dependent Change--Expiration 60
7
40
M -I
m
-e -
20 --
E
0 --
0 0
c
?
-20.-
“
] -40 -2000
I
:
I
-1000
0
mo
;
2000
2
Do
Time from stimulus (rns)
Figure 2. Old phasedelta phase plot showing beat timing from discriminative stimulus presentation and interbeat interval values plotted at the beat that completes the interval. The results are those from the expiratory phase of respiration. A straight line has been fit to the interbeat interval values associated with the different phase timing for beats sharing the same ordinal position relative to the stimulus. Values that are from the same ordinal beat have the same symbol in the plot.
Phase Sensitive Primary Bradycardia
January, 1991
As suggested by Figure 2, a phase dependent poststimulus deceleration was supported by a significant Heartbeat X Phase interaction. The Heartbeat X linear trend in Phase interaction was significant (F(4/88)=78.1, H-F adj.=.42). Neither perceptual/motor set nor stimulus intensity interacted with phase, i.e., the slopes relating phase timing to degree of interbeat interval change were not altered by stimulus intensity or instructional set. Recall that the warning stimulus was also presented at specific phases within the cardiac cycle. If the warning stimulus induced phase dependent deceleration, then this should be evident as a change between the two initial beats plotted in Figure 2. It was not evident. Respiration and phase dependency. The form of the phase dependent response of interbeat interval was also influenced significantly by respiration. Respiration interacted significantly with the linear trend in Phase X Heartbeat interaction (F(4/88)= 7.0, H-F adj. =.37). Figure 3 presents the results for inspiration for comparison with the expiration results of Figure 2. Follow-up comparisons were done to show the primary loci of the phase dependent changes. These were done separately within the inspiratory data and expiratory data. The linear trend in Phase X Heartbeat interaction was significant for both inspiration (F(4/88)=24.0, H-F adj. = .5) and expiration (F(4/88)=57.2, H-F adj.=.80).
60 h
A
:
v
e+*
50.-
' 7
b
M
-I m
\
40.-
d'
;
30-
c
8
d
\ 'c \ b
20--
C 8)
b c
B
A
10 --
'c 0 O T
/I
7
The figures suggest that the trends change at the point of stimulus presentation for the expiration data but earlier for the inspiration data. Beat-tobeat comparisons of linear trend were done to support the above description and follow up the significant linear trend in Phase X Heartbeat interactions. Following Keselman and Keselman (1988), a Bonferroni correction was applied and significance was set at p
Vol. 28, No. I Printed in U.S.A.
Weak Sensory Stimuli Induce a Phase Sensitive Bradycardia J. RICHARD JENNINGS, MAURITSW. VAN DER MOLEN, RIEKJ.M. SOMSEN, AND KAY BROCK University of Pittsburgh and University of Amsterdam
ABSTRACT We attempted to demonstrate that significant perceptual stimuli would induce different degrees of heart rate deceleration depending on when (phase) in the cardiac cycle they occurred. Relative to previous work, we concurrently examined a number of factors that might alter the amplitude of such a cardiac cycle time effect. Stimulus intensity and presence or absence of a speeded response were manipulated. Liminal stimuli and a perceptual rather than motor set were expected to maximize any cardiac cycle time effect. Respiratory phase, length of average interbeat interval, and number of trials were also investigated. Twenty-four college aged, male volunteers were randomly separated into equal groups receiving instructions either to judge which of two weak visual stimuli occurred or to execute a speeded, discriminative response to the stimuli. Discriminative stimuli were presented at either 0, 150, 250, 350, or 500 ms after the R-wave of the electrocardiogram. Stimuli were presented with an intensity that had yielded either 63% or 90% correct detections in a prior psychophysical assessment. A phase dependent deceleration occurred after both intensities of stimuli. Poststimulus deceleration was greater for stimuli in early to mid cycle as suggested by earlier work. As expected, this result was clear when the stimuli were presented during the expiratory phase of respiration. Neither perceptnal/motor set nor stimulus intensity altered the phase sensitive deceleration. Thus, phase sensitive deceleration was confirmed using demanding sensory stimuli and an improved representational technique. DESCRIPTORS Cardiac cycle time effect, Threshold stimuli, Heart rate deceleration, Phase sensitivity, Respiration.
The Laceys (e.g., 1974) have suggested that the and empirical grounds. Both Somsen, Jennings, and processing of a psychologically significant stimulus van der Molen (1988) and Velden, Barry, and Wolk elicits an immediate lengthening of cardiac inter- (1987) observed that the prior representations of beat interval. If the stimulus occurs in the second phase dependent stimulus induced effects may have quarter of the interval, e.g. 300-400 ms after the confounded anticipatory and not phase dependent R-wave of the EKG, greater deceleration is elicited effectswith any true phase dependent changes. We than if it occurs very early or late in the interval. (Jennings, van der Molen, & Terezis, 1987; JenBecause the size of the deceleration depends on the nings, van der Molen, Somsen, & Terezis, 1990) time of stimulus occurrence within the cardiac in- have also performed a number ofexperiments varyterbeat interval, such an effectcan be called a phase ing the phase timing ofstimuli, i.e., when the stimdependent effect or a cardiac Cycle time effect. The ulus occurred in the interbeat interval. In these exLaceYs (e-g.9 1974) termed the Phase dependent de- periments response timing, i.e. when the response celeratory change, primary bradycardia. occurred in the interbeat interval, appeared to show Recently, the reality of such Phase dependent a phase dependent influence on heartbeat timing, effects has been questioned on both methodological but stimulus timing did not-i.e., bradycardia was not evident. Both of these experiments, however, used a speed accuracy tradeoff task in We like to thank NIMH grant MH 40418 for which volunteers were paid to respond to stimuli its support of the work described and NWO grant 02-65020 for its support of Dr. van der Molen in related work. with different reaction speeds. This emphasis on Address requests for reprints to: J. Richard Jennings, motor performance may have created a 'motor WPIC, Room El 329, 381 1 O'Hara Street, Pittsburgh, PA (Garner, 1962) that could have been incompatible with stimulus induced cardiac deceleration. 15213. 1
2
Jennings, van der Molen, Somsen, and Brock
In sum, demonstrations of primary bradycardia to date can be faulted because procedures were used that were likely to obscure phase dependent effects. Frequently, reaction time paradigms were used in which response-related acceleration as well as anticipatory deceleratory effects can induce changes similar to primary bradycardia (Jennings et al., 1987). In other cases, the influence of respiratory phase or length of average interbeat interval has been ignored (see discussion in Coles, Pellegrini, & Wilson, 1982). Finally, representation techniques have often obscured rather than illuminated phase dependent effects. The purpose of the current experiment was to test whether or not phase dependent cardiac deceleration would be elicited by liminal stimuli. Discriminative stimuli were presented at five phases of the cardiac cycle varying between 0 and 500 ms after the R-wave of the electrocardiogram. Decelerative change was expected to be maximal for phases in the second quarter of the interbeat interval. Liminal stimuli were used to increase the likelihood of eliciting a poststimulus deceleration. Poststimulus cardiac deceleration has generally been relatively larger for liminal than for moderate intensity stimuli (Jackson, 1974). In the current experiment we varied stimulus intensity between a level just above the psychophysical threshold and a significantly more intense level. We further instructed one group of subjects to closely assess the stimulus and then give us a choice response-a perceptual set group; another group was instructed to make a choice response as quickly as possible in order to receive incentive pay for adequate speed of response-a motor set group. Low intensity stimuli and perceptual set were expected to enhance any phase dependent poststimulus deceleration. Conceptually, these expectations are based on the Graham and Clifton (1966) integration of the thinking of Sokolov and the Laceys. Cardiac deceleration is viewed as a component of the orienting response. Perceptual identification of a novel or significant liminal stimulus is presumed to take longer than comparable analysis of a more intense stimulus. A similar effect might be expected when instructions emphasize perceptual analysis rather than motor speed. Three other factors were examined in our attempt to define conditions favorable to the identification of primary bradycardia. First, stimuli were classified as occurring during either respiratory inspiration or expiration. Primary bradycardia is expected to occur during expiration (Coles et al., 1982). Second, volunteers with longer average interbeat intervals were separated from those with shorter interbeat intervals. Lacey and Lacey (1 974)
Vol. 28, No. 1
suggested that primary bradycardia would be more evident in individuals with longer interbeat intervals. The rationale for these two factors is primarily physiological. Central nervous system activity related to the perceptual detection of a significant stimulus is presumed to induce vagal activation. The efficacy of this activation is then modulated by the current state of cardiorespiratory control. Long interbeat intervals will permit different neurotransmitter or membrane kinetics in the cardiac pacemaker relative to short interbeat intervals (Lacey & Lacey, 1974; Somsen, Molenaar, van der Molen, & Jennings, in press). Inspiratory phase will inhibit vagal effects on the pacemaker whereas expiratory phase will permit it (Coles et al., 1982). The third factor examined was methodological rather than physiological. A large number of trials were given so that we could estimate the trial block length adequate for observing primary bradycardia. Finally, the results are graphically represented and analyzed using an old phasedelta phase procedure that we have recently described (Jennings, van der Molen, Somsen, & Ridderinkhof, in press). We avoid artifactual phase dependency by presenting the results as a function of real time relative to the stimulus rather than as a function of ordinal interbeat interval. This analytic technique also permits us to separate anticipatory effects from poststimulus changes. Thus, an analytic technique is used that is designed to avoid the representation and interpretive problems raised by Somsen et al. (1988) and Velden et al. (1987). Method Subjects
Twenty-four male volunteers aged 18-30 were recruited from the University of Pittsburgh campus to participate in the study. All volunteers were screened to ensure good health and absence of drug use. Experimental Task The volunteers were asked to identify one of two figures presented by a seven-segment light-emitting diode (LED) situated on a display board and to press one of two microswitch response keys corresponding to the figure identified. The display board, a 66 X 66 cm plexiglass sheet painted a flat black color, was mounted on a microphone stand and placed 1 18.5 cm in front of the semireclining volunteer in a soundproof chamber. The illumination of four segments of the seven-segment LED display formed the figures, which appeared as either "[" or "I". The 1.5 X 3 cm figure appeared in the center of the display board. Two intensity levels on the LED display were used corresponding to the volunteers' near threshold level (63% detection) and slightly brighter than the threshold (90% detection). The intensity of the figure was controlled
January, 1991
Phase Sensitive Primary Bradycardia
by altering the DC component of a modulated DC voltage (Nygaard & Frumkes, 1982). The modulation frequency was 1 KHz. A set of microswitch response keys was placed on the left or right arm of the chair, depending upon the handedness of the volunteer. Each trial began with the I-s illumination ofa warning signal which was a horizontal bar in the center of the LED display. The intensity of the warning bar was relatively bright and constant regardless of the intensity of the stimulus. The time of warning signal presentation was determined by the occurrence of the Rwave of the volunteer’s electrocardiogram and a preselected cycle delay time. During the following interbeat interval at the preselected cycle delay time, the discriminative stimulus appeared for 50 ms at the a p propriate intensity level. Timing of the stimulus as well as physiological data collection were controlled by a DEC 1173 MINC computer. After the volunteer’s response, a 2-s delay was followed by a 1-s presentation of feedback to the volunteer. A 1-3 s randomly vanable intertrial interval followed the termination of the feedback illumination. The five cardiac cycle delays were 0, 150,250, 350, and 500 ms. The cardiac cycle delay was randomly selected for each trial. Delays were equally probable, i.e., p=.2. At the end of the intertrial interval, the computer compared the elapsed time since the last Rwave and the delay randomly assigned for that trial. If the elapsed time was greater than the assigned delay, then the warning signal was presented at the assigned delay time after the next heartbeat. If the elapsed time was less than the assigned delay, then the warning stimulus was presented at the assigned delay time during the current interbeat interval. The discriminative stimulus appeared at the same assigned delay during the interbeat interval following that of the warning stimulus. With brief foreperiods, foreperiod length and delay (phase) time can become confounded unless warning and discriminative stimuli have the same delay time (Jennings et al., 1987). The 24 volunteers were divided randomly into two groups of 12. Both groups viewed the same discriminative stimuli, but one group performed a speeded motor response. Volunteers in this motor set group were required to press the microswitch response key (with their dominant hand) identifying the stimulus within 400-600 ms. Reaction times were measured using the clock on the DEC MINC computer. Feedback on incentive pay earned was based on both correctness of response and speed for this group. Volunteers in the other group did not perform under the speed condition, although for computer programming ease a time limit of 2000 ms was set. The feedback display for this perceptual set group was based solely on accuracy of choice. The intensity level of the feedback display was the same as that of the warning signal. Feedback served to maintain volunteers’ motivation and was presented on the same LED display as the discriminative stimulus. The volunteers received a bonus depending on their performance in addition to an hourly pay. The motor set group received $1.00 for
3
each set of 50 trials in which at least 50% of their responses were within the speed target and accurate, $SO for sets with between 25%and 50% on speed and accurate, and $.25 for sets with less than 25% on speed and accurate. The perceptual set group received S 1.OO for 80% or more correct discriminations within a 50trial set, $ S O for between 50% and 80% correct, and $.25 if less than 50% correct.
Procedure On the first day, the volunteers were trained to perform the task and the two intensity levels were established using a staircase method following Levitt (1 97 1, see below). Electrodes for physiological recording were attached, but no data were collected. First, a sample version of the experimental task was administered to orient the volunteers to the experiment. The perceptual set group received instructions emphasizing accuracy of identification and the motor set group received instructions placing an additional emphasis on the response time requirement. A 25-trial practice task was then administered with clearly visible stimuli. If any volunteer needed more practice in order to understand the task, a second 25-trial task was given. Each volunteer’s psychophysical sensitivity was then determined. The volunteers were adapted to almost complete darkness in the soundproof chamber for 10 min. A flashlight positioned on the floor was the only light source. The batteries were replaced each day for each volunteer to maintain this level. One of the two stimuli was randomly chosen by flipping a coin to serve as the constant test stimulus presented at varying intensities. The volunteers were instructed to press one microswitch response key if they detected the stimulus, and the other key if they could not see the stimulus. The warning bar appeared 1 s before the test stimulus. The volunteers were told that there was no time criterion in making their response, but a response was required after each stimulus in order for the procedure to continue. Four consecutive test stimuli were presented at the same intensity level. If 3 or 4 of these were detected, the intensity level decreased. When the volunteer detected 1 or 0, the intensity level increased. If 2 stimuli were detected, the same intensity level was used on the next 4 figures. The starting level for all volunteers was 1100 digital converted voltage units. The initial step size in increasing or decreasing intensity was 10. When the volunteer made a changeover (from detecting 3 or 4 to detecting 1 or 0 or vice versa), the step size decreased to 5. At the next changeover, the step size decreased to 2 and finally to 1. After the last changeover, 15 sets of 4 test stimuli were presented increasing or decreasing intensity by a step size of 1. The 63% correct detection level to be used as the low intensity value in the experimental task was computed by recording the average of the higher and lower intensities when a changeover occurred and then averaging all of those values. The higher intensity level to be used in the experimental task was computed by
4
Jennings, van der Molen, Somsen, and Brock
adding two standard deviations of those values to the threshold value. This value approximated 90% correct detection. Volunteers varied in the number of test stimuli required to determine their threshold level. The duration of the procedure ranged from 15 to 25 min. After the staircase procedure, volunteers were offered a break and reminded of the original task instructions. During this time, they remained in the chamber with the light level constant. After the break the volunteers received two 25-trial sets of the performance task with the brighter intensity stimuli and two 25-trial sets at the lower level. Within a set, the intensity level of the stimulus was constant. At the end of the session, light levels were slowly adjusted upward until the volunteer could comfortably leave the chamber. The practice day was followed by three experimental days. The experimental session began with the application of the physiological equipment. A 10-min baseline period preceded the performance task. During the baseline, the display board was unplugged and the physiological signals were recorded. The flashlight was positioned on the floor while the volunteer adjusted to the environmental light level. The volunteers received one 15-trial set at each intensity level established on the training day to verify the values. If the volunteer reported the stimulus extremely easy or difficult to detect during either set (substantiated by his performance), the staircase procedure was repeated. However, the starting level was the threshold value determined on the practice day. This shortened the time required to assess a more accurate threshold for that day. The new values were verified in two additional 15-trial sets. Once the threshold levels had been established, the volunteer received 6 sets of 50 trials of the performance task. Three of the sets presented the stimuli at the 63% level and 3 of the sets used the brighter level. The order of the sets was randomized within a day by flipping a coin. The volunteer was informed which level would be used before each set. If at any time during the study the volunteer was exposed to outside light, he spent 10 min readjusting to the darkened chamber. Physiological Measures Measurements of heart rate, peripheral vascular a o tivity, electromyography, blood pressure, and respiration were taken. Heart rate was measured as interbeat intervals (IBI) between R-waves of the electrocardiogram using external inputs to the Minnesota Cardiograph Model 304 Band the computer clock with an accuracy of 1 ms. Six interbeat intervals were recorded for each discriminative stimulus presentation3 before the stimulus, the interbeat interval of the discriminative stimulus, and 2 following the stimulus. For purposes of comparability,baseline data were collected using the same computer program, although the volunteers did not actually see or respond to any LEDs. Two electrodes in conjunction with four impedance bands of cardiograph electrode tape were used to obtain the electrocardiogram.A modified lead I1 place-
Vol. 28, No. 1
ment was used with one Beckman electrode placed caudal to the right pectoral muscle and one placed in the V5 position below the heart. Peripheral vascular activity was monitored with the pulse obtained from the impedance cardiograph and with a photoplethysmograph attached to the thumbnail of the nondominant hand of the volunteer. The first impedance band was placed around the base of the neck with the second band 2-3 cm above the first. The third impedance band was placed around the xiphoid process (caudal termination of sternum) and the fourth impedance band was placed 3-4 cm below the third. The third band served as ground for both impedance and electrocardiogram measures. A blood pressure cuff was also placed on the dominant arm and a heart sounds microphone positioned over the apex of the heart. Vascular results from these measures are reported elsewhere because they do not relate to the issues addressed in this report. Respiration was measured using a Yellow Springs thermistor placed at the opening of the right nostril. The respiration signal was amplified and conditioned by a Grass differentiator, the first derivative was taken and a positive signal (a square wave) was produced during inspiratory slope; the signal was at ground during expiration and expiratory pause. This signal was sent to a Coulbourn bipolar comparator for a finer adjustment of the threshold. Then the signal was sent to the digital input of the computer permitting a trialby-trial recording of respiratory phase coded as inspiration or expiration at the time of the discriminative stimulus.
Results Performance
Stimulus intensity and the perceptual/motor set instructions yielded the expected results for reaction time and proportion correct. Table 1 shows the results for both variables classified by the stimulus intensity and the instruction set groups-perceptual or motor set. Both reaction time and proportion correct were examined with an analysis of variance with factors of instructional set, phase of stimulus presentation, and stimulus intensity. Low intensity stimuli were associated with both a lower percentage of correct detections as well as slower detections. As instructed, motor set volunteers produced more rapid detections. Surprisingly, the motor set volunteers also produced more accurate responses than the perceptual set volunteers. Furthermore, the overall level of correct choices was higher than anticipated from our psychophysical adjustments. The effect of discrimination task with performance incentives was clearly different from the effect of the detection task used to determine thresholds. The statistical results for reaction time showed main effects of perceptual/motor set (F(1/22)=
January, 1991
Phase Sensitive Primary Bradycardia
5
Table 1
Reaction time and proportion correct as a function of instructional set and stimulus intensity Perceptual Set (11-12)
Motor Set (n=12)
Stimulus Intensity
Reaction Time (ms)
Proportion
Reaction Time (ms)
Proportion
Correct
Near Threshold (63% detection) Above Threshold (90% detection)
748
.90
496
.96
709
.93
482
.97
Correct
6 1.9) and stimulus intensity (F(1/22)= 12.9). Similarly, for proportion correct, there were significant main effects for set (F(1/22)=6.6) and intensity (F(1/22)=23.6). In these and all analyses a rejection region of p6.2, H-F adj.=.88) within expiration. No significant intensity effects were found in the inspiration results. Phase dependent influences on interbeat interval. The presence of clear deceleration at the time of the stimulus permits us to ask the central question: Does the processing of weak perceptual stimuli induce a short latency cardiac deceleration that depends upon when in the cardiac cycle the stimuli occur? Phase sensitive changes in interbeat interval were assessed with a recently developed graphical and statistical procedure (Jennings et al., in press). The graphical procedure plots the location of heartbeats in real time relative to stimulus presentation on the x axis. On the y axis differences from a base interbeat interval are plotted at the time of the heartbeat completing the interval. Figure 2 presents such a plot of the expiration data using the initial prestimulus interval (third interval prior to the stimulus) as the base interval. Separate beat sequences are plotted for each phase or cycle time relative to stimulus timing. For example, for the beat sequence centered on a phase time of 0 msstimulus occurring on the R-wave-the beat ending the last prestimulus interval is plotted at exactly 0 on the x axis. Correspondingly, for the beat sequence centered on a phase time of 500 ms, the
VoI. 28, No. 1
beat ending the last prestimulus interval is plotted at - 500 on the x axis. Note how for the prestimulus beat, the 0-ms phase point is on the extreme right of the cluster of points and the 500-ms point is on the extreme left. This ordinal relationship is maintained as the other beats from these sequences are plotted. The results are averaged across intensity condition and perceptual/motor set grouping, but presented in separate figures for respiratory inspiration and expiration. Linear functions have been fit with a least square procedure across the points for the different phase values at each heartbeat. Within each cluster of points (corresponding to a particular beat), the difference scores show an orderly relationship with phase timing. Linear and higher order trends can be fit to the data and tested against the experimental variables with an analysis of variance (Winer, 1971). We are primarily interested in the trend across phase position. Trends can be estimated across phase position for each beat and then the equivalence of these trends can be tested with the interaction of Heartbeat and the trend for Phase position. Such an analysis was performed with factors for percep tual/motor set, heartbeat, respiratory inspiration versus expiration, stimulus intensity, and phase of stimulus presentation. Linear and higher order trends were examined for the phase factor. The interactions of Heartbeat and the trends for Phase were the primary focus of the analysis. Phase Dependent Change--Expiration 60
7
40
M -I
m
-e -
20 --
E
0 --
0 0
c
?
-20.-
“
] -40 -2000
I
:
I
-1000
0
mo
;
2000
2
Do
Time from stimulus (rns)
Figure 2. Old phasedelta phase plot showing beat timing from discriminative stimulus presentation and interbeat interval values plotted at the beat that completes the interval. The results are those from the expiratory phase of respiration. A straight line has been fit to the interbeat interval values associated with the different phase timing for beats sharing the same ordinal position relative to the stimulus. Values that are from the same ordinal beat have the same symbol in the plot.
Phase Sensitive Primary Bradycardia
January, 1991
As suggested by Figure 2, a phase dependent poststimulus deceleration was supported by a significant Heartbeat X Phase interaction. The Heartbeat X linear trend in Phase interaction was significant (F(4/88)=78.1, H-F adj.=.42). Neither perceptual/motor set nor stimulus intensity interacted with phase, i.e., the slopes relating phase timing to degree of interbeat interval change were not altered by stimulus intensity or instructional set. Recall that the warning stimulus was also presented at specific phases within the cardiac cycle. If the warning stimulus induced phase dependent deceleration, then this should be evident as a change between the two initial beats plotted in Figure 2. It was not evident. Respiration and phase dependency. The form of the phase dependent response of interbeat interval was also influenced significantly by respiration. Respiration interacted significantly with the linear trend in Phase X Heartbeat interaction (F(4/88)= 7.0, H-F adj. =.37). Figure 3 presents the results for inspiration for comparison with the expiration results of Figure 2. Follow-up comparisons were done to show the primary loci of the phase dependent changes. These were done separately within the inspiratory data and expiratory data. The linear trend in Phase X Heartbeat interaction was significant for both inspiration (F(4/88)=24.0, H-F adj. = .5) and expiration (F(4/88)=57.2, H-F adj.=.80).
60 h
A
:
v
e+*
50.-
' 7
b
M
-I m
\
40.-
d'
;
30-
c
8
d
\ 'c \ b
20--
C 8)
b c
B
A
10 --
'c 0 O T
/I
7
The figures suggest that the trends change at the point of stimulus presentation for the expiration data but earlier for the inspiration data. Beat-tobeat comparisons of linear trend were done to support the above description and follow up the significant linear trend in Phase X Heartbeat interactions. Following Keselman and Keselman (1988), a Bonferroni correction was applied and significance was set at p
