Psychophysiology, 45 (2008), 671–677. Wiley Periodicals, Inc. Printed in the USA.

Copyright r 2008 Society for Psychophysiological Research

DOI: 10.1111/j.1469-8986.2008.00666.x

Interoceptive awareness in experienced meditators

SAHIB S. KHALSA,a DAVID RUDRAUF,a ANTONIO R. DAMASIO,b

RICHARD J. DAVIDSON,c ANTOINE LUTZ,c and DANIEL TRANELa
a
Department of Neurology and Neuroscience Program, University of Iowa, Iowa City, Iowa, USA
b
Brain and Creativity Institute, Department of Psychology and Neurology, University of Southern California, Los Angeles, California, USA
c
W. M. Keck Laboratory for Functional Brain Imaging and Behavior, Waisman Center, and Laboratory for Affective Neuroscience, Department of
Psychology, University of Wisconsin, Madison, Wisconsin, USA

Abstract
Attention to internal body sensations is practiced in most meditation traditions. Many traditions state that this practice
results in increased awareness of internal body sensations, but scientific studies evaluating this claim are lacking. We
predicted that experienced meditators would display performance superior to that of nonmeditators on heartbeat
detection, a standard noninvasive measure of resting interoceptive awareness. We compared two groups of meditators
(Tibetan Buddhist and Kundalini) to an age- and body mass index-matched group of nonmeditators. Contrary to our
prediction, we found no evidence that meditators were superior to nonmeditators in the heartbeat detection task, across
several sessions and respiratory modulation conditions. Compared to nonmeditators, however, meditators consistently
rated their interoceptive performance as superior and the difficulty of the task as easier. These results provide evidence
against the notion that practicing attention to internal body sensations, a core feature of meditation, enhances the
ability to sense the heartbeat at rest.
Descriptors: Meditation, Interoception, Heartbeat detection, Awareness, Respiration

Meditation is a form of mental training that has been practiced monly practiced under conditions of rest, the subjective experi-
for thousands of years that can be conceptualized as a family of ence of these interoceptive body sensations is also routinely
complex emotional and attentional regulatory training regimens modulated through manipulations of the breath and musculo-
developed for various ends, including the cultivation of well- skeletal posture, particularly during the practice of yoga exercises
being and emotional balance (Lutz, Dunne, & Davidson, 2007). (Arambula, Peper, Kawakami, & Gibney, 2001; Bhajan Y, 2000;
Although typically practiced in the context of spiritual traditions, Peng et al., 2004). Many traditions state that the practice of
there has been a notable increase in the therapeutic application of attending to interoceptive sensations results in enhanced aware-
meditation as a complement in alternative medicine. (Arias, ness of these sensations, and also assert that the meditation
Steinberg, Banga, & Trestman, 2006; Astin, Shapiro, Eisenberg, practice results in enhanced awareness of a variety of other in-
& Forys, 2003; Barnes, Powell-Griner, McFann, & Nahin, ternal events, such as the ongoing experience of thoughts and
2004). emotions (Kabat-Zinn, 1990; Kornfield, 1996; Nairn, 2000).
Most meditation traditions incorporate attention to internal The idea that a meditation practice would enhance inter-
body sensations as a component of the practice, particularly in oceptive awareness is certainly plausible, but there is no scientific
the beginning stages of instruction, possibly because the avail- evidence to support this claim. In the current study, we sought to
ability of these sensations from moment to moment makes them address this knowledge gap by studying interoceptive awareness
a convenient object to focus on. The most commonly attended in experienced meditators.
body sensations include the breath, the position of the joints Several methods for assessing interoceptive awareness have
(proprioception), the degree of muscle tension, and the heartbeat been described, including gastrointestinal distension (Holzl,
(Kabat-Zinn, 1990; Kornfield, 1996; Nairn, 2000; Selby, 1992). Erasmus, & Moltner, 1996), adrenergic stimulation (Cameron
Although attention to internal body sensations is most com- & Minoshima, 2002; Khalsa, Rudrauf, Sandesara, Olshansky, &
Tranel, in press), and heartbeat perception (Brener & Kluvitse,
We thank M. Ricard and A. Francis for assistance with participant 1988; Schandry, 1981; Whitehead, Drescher, & Heiman, 1977).
recruitment. The project was supported by NIH NCCAM F31AT003061 The latter, heartbeat perception, is considered the standard and
from the National Center For Complementary & Alternative Medicine preferred method for the noninvasive assessment of interoceptive
(NCCAM) (S.K.), by the Mind and Life Institute (S.K.), by NCCAM awareness, and factors modulating awareness of cardiac sensa-
U01AT002114-01A1 (A.L.), and by NIDA R01 DA022549 (D.T.). tions have been extensively studied (Brener, Liu, & Ring, 1993;
Address reprint requests to: Sahib S. Khalsa, B.S., Neuroscience
Graduate Program, Department of Neurology, University of Iowa,
Eichler & Katkin, 1994; Jones, 1994; Knapp, Ring, & Brener,
200 Hawkins Drive, Iowa City, IA 52242, USA. E-mail: sahib-khalsa 1997; Ring & Brener, 1992; Rouse, Jones, & Jones, 1988; Schan-
@[Link] dry, Bestler, & Montoya, 1993). Recently, functional neuro-
671
672 S.S. Khalsa et al.

imaging studies have demonstrated that heartbeat perception Table 1. Demographic Data for All Three Groups (Means
SD)
tasks activate a network of brain regions including the insula,
primary somatosensory cortex, and the anterior cingulate cortex Tibetan
(Craig, 2002; Critchley, Wiens, Rotshtein, Ohman, & Dolan, Nonmeditators Kundalini Buddhist
2004; Pollatos, Schandry, Auer, & Kaufmann, 2007). These Sex 4 M:13 F 5 M:12 F 7 M:6 F
brain regions are considered necessary for the representation and Age (years) 50.6
9.6 52.1
8.6 48.8
10.1
maintenance of the internal state of the organism (Craig, 2002; Body mass index 24.8
5.1 24.0
5.2 22.3
3.3
Meditation practice 0
0 29.3
6.4 24.7
8.4
Critchley et al., 2004; Pollatos et al., 2007) and for the conscious (years)
experience of emotion and feelings (Damasio et al., 2000), lend- Cumulative 0
0 17,660
9128 24,903
14,270
ing further support to the notion that heartbeat perception is a meditation
good index of interoception. practice (hours)
Although there are several techniques for assessing heartbeat
perception, the most commonly used methods are heartbeat de-
tection and heartbeat tracking. During heartbeat detection, sub- ness would be reflected through more accurate subjective ratings
jects determine whether an exteroceptive stimulus, such as a light of interoceptive task performance in both groups of meditators
or a tone, is contemporaneous with their heartbeat sensation than in nonmeditators.
(Brener & Kluvitse, 1988; Schneider, Ring, & Katkin, 1998;
Whitehead et al., 1977). Performance is indexed by the number of
correct responses reported by the subject (e.g., true positives and Methods
true negatives), which also allows measurement of individual and
group response accuracy. Subjects are then classified as ‘‘good Participants
heartbeat detectors’’ when their performance lies above chance Seventeen nonmeditators, 17 Kundalini meditators, and 13 Ti-
according to the binomial distribution (Katkin, Wiens, & Oh- betan Buddhist meditators participated in the study (Table 1).
man, 2001; Schneider et al., 1998; Wiens & Palmer, 2001). Dur- Meditators were selected according to three criteria: (1) a min-
ing heartbeat tracking, subjects silently count their heartbeats imum of 15 years of formal meditation practice, (2) a self re-
during brief, fixed time periods. Performance is indexed by a ported strong daily practice, and (3) having attended at least one
cardiac perception score, in which the number of counted heart- meditation retreat during the previous year. Nonmeditators were
beats is contrasted with the number of actual heartbeats. Subjects identified as individuals who had never attended a formal yoga or
are classified as ‘‘good heartbeat perceivers’’ when their scores meditation course and did not practice self-taught meditation.
fall above a predetermined level (Herbert, Ulbrich, & Schandry, All groups were matched with respect to age and body mass
2007). Heartbeat detection has been the more commonly utilized index. Any participant reporting a history of neurological or
measure, perhaps because it appears to suffer from less method- psychiatric disease was excluded from the study. Based on this
ological confounds than heartbeat tracking. Such confounds in- criterion, 1 Kundalini meditator and 2 nonmeditators were pre-
clude the lack of a statistical measure to evaluate individual cluded from study participation. This study was approved by the
performance, the possible influence of a priori knowledge about University of Iowa’s Institutional Review Board, and all partic-
average heart rate on the rate of counting (Phillips, Jones, Rieger, ipants provided informed consent prior to participation.
& Snell, 1999; Ring & Brener, 1996), and the insensitivity of
heartbeat tracking tasks to changes in heart rate (Windmann, Tasks
Schonecke, Frohlig, & Maldener, 1999). Consequently, we se- Participants performed two types of tasks: a pulse detection fa-
lected heartbeat detection as an index of interoceptive awareness. miliarization task and a heartbeat detection task. Each task uti-
We identified experienced meditators from two different med- lized identical stimuli but required a different attentional focus.
itation traditions that are extensively practiced within the United During pulse detection, participants took their nondominant
States: Tibetan Buddhism and Kundalini yoga. These traditions wrist pulse and were required to judge whether a train of ext-
were selected to examine whether the effects of the meditation eroceptive stimuli (800-Hz, 50-ms tones) were simultaneous or
practice on interoceptive awareness were consistent across tra- nonsimultaneous with pulse sensations. During heartbeat detec-
ditions, despite the fact that each tradition adopts slightly differ- tion participants were not allowed to take their pulse and were
ing approaches to the cultivation of interoceptive awareness. For required to judge whether the tones were simultaneous or non-
example, in Tibetan Buddhism interoceptive awareness is more simultaneous with perceived heartbeat sensations.
commonly cultivated while meditating under resting physiolog-
ical conditions, whereas in Kundalini yoga interoceptive aware- Tone Delivery
ness is more commonly cultivated during yoga exercises that elicit Tone delivery was triggered by each myocardial contraction, as
conditions of mild physiological arousal. measured (indirectly) from the R-wave of a lead II electrocar-
We hypothesized that the long-term practice of meditation diogram (MP100 acquisition unit, Biopac Systems, Inc.). During
leads to enhanced interoceptive awareness. On this basis, we simultaneous trials, tones were delivered at the same time as the
predicted that experienced meditators from both traditions participant’s own finger pulse, approximately 250–300 ms after
would display enhanced awareness of heartbeat sensations dur- the R-wave1 (corresponding to the R-wave to pulse interval, or
ing performance of a heartbeat detection task at rest. We further RPI). The finger pulse was measured with an infrared photo-
hypothesized that experienced meditators would display meta- 1
cognitive awareness of this enhancement, that is, knowledge of This delay, around 250 to 300 ms, has been shown to lead to the
perception by accurate heartbeat detectors that heartbeats and tones are
accurate self performance, based on the rationale that meditation ‘‘simultaneous’’ (Brener et al., 1993; Eichler & Katkin, 1994; Jones, 1994;
cultivates a monitoring of experience at levels beyond mere in- Knapp et al., 1997; Ring & Brener, 1992; Rouse et al., 1988; Schandry
teroceptive processing. We predicted that metacognitive aware- et al., 1993).
Meditation and interoception 673

Table 2. Cardiovascular Parameters for All Three Groups meditation instruction was given to the participants. We con-
(Means
SD) trolled for breathing patterns during heartbeat detection for two
related reasons: (1) spontaneous respiratory manipulations have
Nonmeditators Kundalini Tibetan Buddhist been observed to occur in subjects in the absence of an instruction
Resting heart rate: 63.3
10.0 67.0
8.7 65.6
14.9 to breathe normally and have been suggested as a potential
Visit 1 (bpm) strategy for maximizing heartbeat sensations (Jones, 1994; We-
Resting heart rate: 65.8
8.4 68.8
11.5 65.3
15.3 isz, Balazs, & Adam, 1988), and (2) several meditators were ob-
Visit 2 (bpm) served to spontaneously display Ujjai breathing during the
R-wave to pulse 0.269
0.017 0.261
0.013 0.266
0.016
interval: Visit 1 (s)
piloting phase of the study and this, in and of itself, could be a
R-wave to pulse 0.266
0.015 0.260
0.014 0.264
0.014 basis for enhanced interoceptive accuracy. One Kundalini med-
interval: Visit 2 (s) itator was unable to perform the Ujjai breathing due to a pre-
existing respiratory condition and was excluded from the
heartbeat detection analysis.

Subjective Ratings
plethysmograph (TSD123B) attached to the distal phalange of Prior to performing each task, participants were asked to predict
the fifth digit of the dominant hand. The RPI was measured for task accuracy (e.g., ‘‘How good do you think you will be at [task
each participant by calculating the average delay between the X]?’’) and difficulty (‘‘How hard do you think [task X] will be?’’).
peak of the R-wave and the foot of the systolic upstroke mea- Upon completion of each task participants were also asked to
sured during a 2-min resting period. Mean resting heart rate and estimate task accuracy (‘‘How good do you think you were at
RPI intervals are listed in Table 2.2 During nonsimultaneous [task X]?’’) and difficulty (‘‘How hard do you think [task X]
trials, tones were delivered 400 ms after the RPI, approximately was?’’). Accuracy ratings could range from 1 (very bad) to 5 (very
650–700 ms after the R-wave. Thus, tone delivery was tempo- good). Difficulty ratings could range from 1 (very hard) to 5
rally linked to each participant’s actual heartbeat during each (very easy). To familiarize participants with each task, all par-
trial. Trial order was randomized within and across each block. ticipants were instructed to sample tones from each trial type
Tones were presented through noise canceling headphones (Qui- (i.e., one simultaneous and one nonsimultaneous trial) for an
etComfort, Bose Inc., Framingham, MA). All participants per- unlimited period prior to performing each task. In addition, task-
formed the tasks in the supine position with their eyes closed, and related feedback was withheld from all participants until the
they were given an unlimited time to respond during each trial. conclusion of the study.

Procedure Accuracy Measures

The study involved two visits, spaced 1–14 days apart. At the Accuracy scores were calculated using A 0 5 [1/21((HR 
beginning of each visit, resting pulse and heart rate were mea- FP)(11HR  FP))/(4HR(1  FP))], a nonparametric signal de-
sured for 2 min. After each resting period, the pulse plethysmo- tection analog of d 0 ideal for signal detection conditions with low
graph was removed in order to prevent participants from trial numbers (Grier, 1971). In this formula, HR 5 hit rate and
deriving heartbeat information from the finger pulse sensation FP 5 false positive. Following methods commonly utilized in
during the subsequent tasks. During the first visit, participants heartbeat detection studies (Brener et al., 1993; Jones, O’Leary,
performed one block of pulse detection followed by two blocks of & Pipkin, 1984; Rouse et al., 1988), A 0 scores were normalized
heartbeat detection, in the same order. During the second visit using the following formula: 2arcsin (sqrt A 0 ), such that perfor-
participants repeated both blocks of heartbeat detection, in the mance ranged from 0 to p (chance 5 p/2). Participants were fur-
same order as before. All blocks consisted of 23 trials. Any par- ther classified as ‘‘good heartbeat detectors’’ if they displayed
ticipant not meeting the criterion for good pulse detection (  16 above chance performance during a block of testing, defined
out of 23 trials correct, po.05 per binomial test) during the first as  16 out of 23 trials correct, po.05 per binomial test, again
visit was excused from the study (1 nonmeditator was excluded following the approach of previous studies (Katkin et al., 2001;
based on this criteria). During the first heartbeat detection block Schneider et al., 1998; Wiens & Palmer, 2001). Because no
(HB1), participants were instructed to breathe normally. During differences were predicted between the two groups of meditators,
the second heartbeat detection block (HB2), participants were the overall analyses examined the three groups separately. All
instructed to practice a yogic breathing pattern: Ujjai breath, a univariate repeated measures ANOVA tests were assessed for
technique that involves symmetric long deep nostril breathing violations of the sphericity assumption, and when violated, were
against airway resistance (Brown & Gerbarg, 2005). Respiratory corrected with the Huynh–Feldt method. In these instances the
rate was measured during all tasks with a thoracic respiratory corrected p values are reported, along with the Huynh–Feldt
belt (RSP100C). Each participant’s respiratory patterns were e correction.
examined for compliance with the Ujjai breathing condition after
completion of the task. All participants demonstrated satisfac-
Results
tory respiratory patterns, indicating accurate performance of the
Ujjai breath. Aside from this breathing technique, no formal Participants
Both groups of meditators reported significantly more years of
2
These intervals are similar to those observed when measuring the meditation practice, F(2, 44) 5 120.5, po.001, and hours of
RPI from the finger pulse (Teng & Zhang, 2006). However, shorter in- cumulative meditation practice, F(2, 44) 5 31.9, po.001, than
tervals have been reported when measuring the RPI from the ear pulse,
possibly due to the shorter distance traveled by the pulse wave (de Boer, the nonmeditators. The groups did not differ with respect to age
Ring, Curlett, Ridley, & Carroll, 2007; de Boer, Ring, Wood, et al., F(2, 44) 5 .45, p 5 .64, or BMI F(2, 44) 5 1.07, p 5 .35, (Table 1).
2007). There were also no differences between the proportion of men
674 S.S. Khalsa et al.

and women in the nonmeditators and the Kundalini meditators,

w1 5 .15, p 5 .70, or the Tibetan Buddhist meditators w1 5 1.76,
p 5 .19.

Cardiovascular Parameters
A 3  2 repeated measures ANOVA did not reveal any group
differences in resting heart rate, F(2, 44) 5 0.42, p 5 .66. There
was no effect of visit on resting heart rate, F(1, 44) 5 2.18,
p 5 .15, and there were no Group  Visit interactions, F(2,
44) 5 0.86, p 5 .43. The groups also did not differ with respect to
the R-wave to pulse interval, F(2, 44) 5 1.34, p 5 .26. There was
no effect of visit on the R-wave to pulse interval, F(1, 44) 5 0.10,
p 5 .75, and there were no Group  Visit interactions, F(2,
44) 5 0.38, p 5 .69. Figure 1. Heartbeat detection and pulse detection accuracy across each
block for each group. HB1: heartbeat detection during normal breathing.
Accuracy Measures HB2: heartbeat detection during Ujjai breathing. NM: nonmeditators.
There were no group differences in accuracy on the pulse KM: Kundalini meditators. TB: Tibetan Buddhist meditators.
detection task, F (2, 44) 5 1.27, p 5 .29. A 3  2  2 ANOVA Performance accuracy could range from 0 to p, with chance
performance 5 p/2. Dotted line 5 chance, Error bars: SE.
was run on heartbeat detection accuracy with group (Kundalini,
Tibetan Buddhist, nonmeditators) as the between-subjects factor
and with block (Block 1 and Block 2) and visit (Visit 1 and 43) 5 4.1, p 5 .023, Zp 2 ¼ :16, with both groups of meditators
Visit 2) as within-subject factors. There were no main effects rating their heartbeat detection performance to be more accurate
for group, F(2, 43) 5 0.30, p 5 .74, Zp 2 ¼ :01, block, F(1, and the task to be less difficult than the nonmeditators (Figures 2
43) 5 0.69, p 5 .41, or visit, F(1, 43) 5 1.38, p 5 .25, and there and 3). There were no other significant main effects or interac-
were no significant interactions between group and visit, F(2, tions for ratings of heartbeat detection accuracy. For ratings of
43) 5 0.77, p 5 .47, block and visit, F(1, 43) 5 .00, p 5 .99, or heartbeat detection difficulty, there was a significant effect of
group and block and visit, F(2, 43) 5 1.03, p 5 .37, (Figure 1). visit, F(2, 1) 5 4.5, p 5 .039, Zp 2 ¼ :10, with all groups rating
The lack of group differences was not accounted for by group both tasks as less difficult on the second visit. There was a sig-
differences in response bias, defined as the tendency to favor one nificant interaction between group and block, F(2, 2) 5 4.7,
particular response type over another, for either pulse detection, p 5 .015, Zp 2 ¼ :18, with both groups of meditators rating each
F(2, 44) 5 0.1, p 5 .91, or heartbeat detection, F(2, 43) 5 1.3, block to be less difficult than the nonmeditators. There was also a
p 5 .28. significant interaction between block and visit, F(2, 1) 5 4.3,
There were also no differences in the proportion of meditators p 5 .045, Zp 2 ¼ :10, with all groups rating blocks from the sec-
classified as good heartbeat detectors (Table 3). This was equally ond visit to be easier than the first visit.
true when the definition of good heartbeat detection performance
was restricted to above chance performance on two out of two Post Analysis
visits or loosened to above chance performance on at least one Both groups of meditators displayed higher subjective ratings of
out of two visits, for both tasks (Table 3). heartbeat detection accuracy and lower subjective ratings of
heartbeat detection difficulty than nonmeditators. Because these
Subjective Ratings findings occurred in the absence of an actual difference in heart-
There were no group differences in the ratings of pulse detection beat detection accuracy, we examined the relationship between
accuracy, F(2, 44) 5 0.12, p 5 .89, or pulse detection difficulty, the objective accuracy scores and the subjective accuracy ratings
F(2, 44) 5 .76, p 5 .48. However, a 3  2 2 ANOVA with two provided by each participant after the pulse and heartbeat de-
repeated measures factors revealed significant group differences tection tasks. Because maximum objective accuracy was almost
in ratings of heartbeat detection accuracy, F(2, 43) 5 5.77, always reached for pulse detection, we assumed that the subjec-
p 5 .007, Zp 2 ¼ :21, and heartbeat detection difficulty, F(2, tive accuracy ratings for pulse detection represented the maxi-

Table 3. Percentage of Individuals Classified as Good Heartbeat Detectors for Each Block of Heartbeat Detectiona

No. of visits above chance

Good heartbeat detectors: HB1 Good heartbeat detectors: HB2

0/2 1/2 2/2 0/2 1/2 2/2

NM (n 5 17) 47% (8) 18% (3) 35% (6) 47% (8) 18% (3) 35% (6)
KM (n 5 17) 35% (6) 47% (8) 17% (3) 38% (6) 50% (8) 13% (2)
w21  1=2 visits w21 ¼ :49, p 5 .24 w21 ¼ :31, p 5 .29
w21 2=2 visits w21 ¼ 0, p 5 .49 w1 5 .14, p 5 .35
TB (n 5 13) 46% (6) 23% (3) 31% (4) 31% (4) 38% (5) 31% (4)
w21  1=2 visits w1 5 .002, p 5 .48 w1 5 .28, p 5 .30
w21 2=2 visits w1 5 .02, p 5 .44 w1 5 .11, p 5 .37
a
The numbers of individuals meeting each criterion are listed in parentheses. A Chi-square with one degree of freedom compares whether the proportion
of meditators classified as good heartbeat detectors differs from the nonmeditators. NM: nonmeditators. KM: Kundalini meditators. TB: Tibetan
Buddhist meditators. Only 16 Kundalini meditators completed testing in Block 2.
Meditation and interoception 675

Figure 2. Mean ratings of pulse and heartbeat detection accuracy across Figure 3. Mean ratings of pulse and heartbeat detection difficulty across
each block for each group. Accuracy ratings could range from 1 (very each block for each group. Difficulty ratings could range from 1 (very
bad) to 5 (very good). NM: nonmeditators. KM: Kundalini meditators. hard) to 5 (very easy). NM: nonmeditators. KM: Kundalini meditators.
TB: Tibetan Buddhist meditators. Error bars: SE. TB: Tibetan Buddhist meditators. Error bars: SE.

mum possible range for subjective accuracy ratings in general. on interoceptive awareness. We conducted a power analysis
We then normalized each participant’s subjective heartbeat de- based on the observed main effect to determine the sample size
tection rating by the global average subjective pulse detection required to achieve a statistically meaningful result for heartbeat
accuracy ratings measured from all participants using the fol- detection. We found that group sizes would need to be increased
lowing formula: (individual heartbeat detection accuracy rating- by one order of magnitude before reaching the threshold of sig-
global mean pulse detection accuracy rating)/global mean pulse nificance. Thus if the sample sizes were increased 10-fold, the
detection accuracy rating. We also normalized each group’s main effect would be as follows: F(2, 457) 5 3.21, p 5 .041,
heartbeat detection accuracy scores (A 0 ) by the mean pulse de- Zp 2 ¼ :01. Even if this were the case, the presumed effect size
tection accuracy scores using this same procedure. Finally, we suggests that the influence of meditative experience on resting
examined the relationship between these normalized heartbeat awareness of heartbeat sensations would be quite small.
detection ratings and normalized accuracy scores for each group Sex differences in heartbeat detection ability have sometimes
for congruency. We found that, on average, meditators’ subjec- been reported in the literature (Jones & Hollandsworth, 1981;
tive accuracy ratings appeared more congruent with objective Katkin, Blascovich, & Goldband, 1981; Whitehead et al., 1977;
accuracy scores than the nonmeditators (Table 4). There was a but see Ring & Brener, 1992; Rouse et al., 1988). There were no
nearly significant interaction between group and scale, F(2, significant group differences between the proportion of males
2) 5 2.6, p 5 .08, Zp 2 ¼ :1, in support of this dissociation. and females in the current study. However, the sample sizes may
not be large enough to test whether sex differences influenced the
Discussion findings. In any event, it seems unlikely that sex played a role in
The current findings do not support the hypothesis that expe- the current findings: We examined whether there were any sex
rienced meditators would display increased interoceptive aware- related differences in interoceptive accuracy and found that none
ness, as meditators did not differ from nonmeditators in of the results differed with respect to sex.
heartbeat detection accuracy. The lack of an effect of medita- Although we believe that the current finding is reliable, it is
tion on awareness of heartbeat sensations appears to be a reliable important to consider alternatives that might explain it. First, it is
finding. It occurs in two different groups of experienced med- possible that awareness of heartbeat sensations alone is a poor
itators, measured at two time points, and with two different re- index of the type of interoceptive sensations cultivated by the
spiratory manipulation strategies. These results are consistent practice of meditation. Although attention to body sensations
with recent findings by Nielsen and Kaszniak (2006), who re- such as the heartbeat is practiced at some point of the training in
ported a lack of significant differences between a group of Bud- all meditation traditions, attention is more commonly directed
dhist meditators and a group of nonmeditators on a single session toward breathing. Any enhancement of interoceptive sensations
of standard heartbeat detection. The Nielsen and Kaszniak study that results from the long-term practice of meditation might be
had a small sample size, did not include comparison subjects specific only to the bodily signals that are attended. Thus the
matched for age or body mass (Rouse et al., 1988), and was current results do not rule out the possibility that meditation
conducted in a small number of sessions, limiting statistical cultivates interoceptive awareness for other body signals such as
power. The present study, however, did not suffer from any of breathing. Second, attention to nonbodily signals is also fre-
these limitations and still did not reveal any effect of meditation quently practiced in all meditation traditions. For example, in the

Table 4. Normalized Objective Heartbeat Detection Accuracy and Subjective Ratings of Heartbeat Detection Accuracya

Nonmeditators Kundalini Tibetan Buddhist

Normalized objective heartbeat detection accuracy 0.45 (0.10) 0.40 (0.10) 0.46 (0.14)
Normalized subjective heartbeat detection accuracy 0.27 (0.08) 0.51 (0.09) 0.51 (0.10)
a
Means (SE), range 0–1.
676 S.S. Khalsa et al.

Tibetan Buddhist tradition, attention is commonly focused on heartbeat detection difficulty than nonmeditators. In the absence
complex mental imagery or external visual objects during med- of an actual difference in heartbeat detection accuracy these
itation (Lutz et al., 2007), and the awareness that develops during findings were surprising. The ratings differences did not appear
such attention training might not translate readily into an en- to be due to a general rating bias, as there were no such group
hancement of interoceptive awareness. Third, the current study differences in the ratings of the pulse detection task. All groups
only examined interoceptive awareness under resting conditions. displayed accurate pulse detection performance, rated their ac-
We chose this starting point because the meditation practice most curacy accordingly, and found the task to be easy. With respect
commonly occurs during those conditions. It is possible that in- to heartbeat detection, the nonmeditators’ ratings were near the
teroceptive awareness for heartbeat sensations is limited at rest bottom of each scale, suggesting that they found the task to be
by a physiological mechanism not amenable to voluntary mod- difficult and felt that their performance was very poor. The
ulation, even through a long-standing meditation practice. Of meditators’ ratings were near the middle of the scale, suggesting
note, none of the groups in the current study displayed heartbeat that they found the task to be neither easy nor difficult and felt
detection rates above 50%. These rates are consistent with those that their performance was neither good nor bad. These rating
routinely reported in the heartbeat detection literature (Brener & differences could be explained either by the nonmeditators un-
Kluvitse, 1988; Eichler & Katkin, 1994; Jones, 1994; Knapp et derestimating their performance or the meditators overestimat-
al., 1997; Ring & Brener, 1992; Wiens & Palmer, 2001), sug- ing their performance. Even though both groups displayed
gesting that it is difficult for most individuals to display aware- performance within the normally reported range on the heartbeat
ness of heartbeat sensations at rest. However, the current finding detection task, ‘‘normal’’ performance is quite often below
does not guarantee that meditation would not be associated with chance, given the fact that heartbeat detection is most commonly
enhanced interoceptive awareness under other physiological measured at rest. Thus it could be argued that either the med-
conditions. Indeed, visceral sensations do not dominate con- itators’ or the nonmeditators’ ratings were congruent with the
scious experience under resting conditions, but quickly develop literature. An analysis of the relationship between each group’s
when conditions such as exercise or stress signal deviations in the objective accuracy scores and subjective ratings of accuracy sug-
homeostatic state (Cameron & Minoshima, 2002; Khalsa et al., gests it is more likely that the nonmeditators were underestimat-
in press). Thus it is still possible that meditators would display ing their heartbeat detection performance and provides support,
increased interoceptive awareness under these conditions. Such albeit limited, for the notion that experienced meditators’ sub-
considerations argue for the development of new measures of jective perceptions of interoceptive states are more in tune with
interoceptive awareness that take into account nonhomeostatic their performance on interoceptive tasks.
physiological body states. Overall, the results of this study provide evidence against the
Both groups of meditators displayed higher subjective ratings notion that practicing attention to internal body sensations, a core
of heartbeat detection accuracy and lower subjective ratings of feature of meditation, enhances resting interoceptive awareness.

REFERENCES

Arambula, P., Peper, E., Kawakami, M., & Gibney, K. H. (2001). The Critchley, H. D., Wiens, S., Rotshtein, P., Ohman, A., & Dolan, R. J.
physiological correlates of Kundalini Yoga meditation: A study of a (2004). Neural systems supporting interoceptive awareness. Nature
yoga master. Applied Psychophysiology and Biofeedback, 26, 147–153. Neuroscience, 7, 189–195.
Arias, A. J., Steinberg, K., Banga, A., & Trestman, R. L. (2006). Sys- Damasio, A. R., Grabowski, T. J., Bechara, A., Damasio, H., Ponto, L. L.,
tematic review of the efficacy of meditation techniques as treatments Parvizi, J., et al. (2000). Subcortical and cortical brain activity during the
for medical illness. Journal of Alternative and Complementary Med- feeling of self-generated emotions. Nature Neuroscience, 3, 1049–1056.
icine, 12, 817–832. de Boer, D., Ring, C., Curlett, A. C., Ridley, M., & Carroll, D. (2007).
Astin, J. A., Shapiro, S. L., Eisenberg, D. M., & Forys, K. L. (2003). Mental stress-induced hemoconcentration and its recovery: A con-
Mind-body medicine: State of the science, implications for practice. trolled study of time course and mechanisms. Psychophysiology, 44,
Journal of the American Board of Family Practice, 16, 131–147. 161–169.
Barnes, P. M., Powell-Griner, E., McFann, K., & Nahin, R. L. (2004). de Boer, D., Ring, C., Wood, M., Ford, C., Jessney, N., McIntyre, D.,
Complementary and alternative medicine use among adults: United et al. (2007). Time course and mechanisms of mental stress-induced
States, 2002. Advance data from vital and health statistics; no. 343. changes and their recovery: Hematocrit, colloid osmotic pressure,
Hyattsville, MD: National Center for Health Statistics. whole blood viscosity, coagulation times, and hemodynamic activity.
Bhajan, Y., & Khalsa, G. S. (2000). Breathwalk: Breathing your way to a Psychophysiology, 44, 639–649.
revitalized body, mind and spirit. New York: Broadway Books. Eichler, S., & Katkin, E. S. (1994). The relationship between cardio-
Brener, J., & Kluvitse, C. (1988). Heartbeat detection: Judgments of the vascular reactivity and heartbeat detection. Psychophysiology, 31,
simultaneity of external stimuli and heartbeats. Psychophysiology, 25, 229–234.
554–561. Grier, J. B. (1971). Nonparametric indices for sensitivity and bias: Com-
Brener, J., Liu, X., & Ring, C. (1993). A method of constant stimuli for puting formulas. Psychological Bulletin, 75, 424–429.
examining heartbeat detection: Comparison with the Brener-Kluvitse Herbert, B. M., Ulbrich, P., & Schandry, R. (2007). Interoceptive sen-
and Whitehead methods. Psychophysiology, 30, 657–665. sitivity and physical effort: Implications for the self-control of phys-
Brown, R. P., & Gerbarg, P. L. (2005). Sudarshan Kriya yogic breathing ical load in everyday life. Psychophysiology, 44, 194–202.
in the treatment of stress, anxiety, and depression: Part IFNeuro- Holzl, R., Erasmus, L. P., & Moltner, A. (1996). Detection, discrimi-
physiologic model. Journal of Alternative and Complementary Med- nation and sensation of visceral stimuli. Biological Psychology, 42,
icine, 11, 189–201. 199–214.
Cameron, O. G., & Minoshima, S. (2002). Regional brain activation due Jones, G. E. (1994). Perception of visceral sensations: A review of recent
to pharmacologically induced adrenergic interoceptive stimulation in findings, methodologies, and future directions (Vol. 5) London: Jessica
humans. Psychosomatic Medicine, 64, 851–861. Kingsley Publishers.
Craig, A. D. (2002). How do you feel? Interoception: The sense of the Jones, G. E., & Hollandsworth, J. G. (1981). Heart rate discrimination
physiological condition of the body. Nature Reviews. Neuroscience, 3, before and after exercise-induced augmented cardiac activity. Psy-
655–666. chophysiology, 18, 252–257.
Meditation and interoception 677

Jones, G. E., O’Leary, R. T., & Pipkin, B. L. (1984). Comparison of the Pollatos, O., Schandry, R., Auer, D. P., & Kaufmann, C. (2007). Brain
Brener-Jones and Whitehead procedures for assessing cardiac aware- structures mediating cardiovascular arousal and interoceptive aware-
ness. Psychophysiology, 21, 143–148. ness. Brain Research, 1141, 178–187.
Kabat-Zinn, J. (1990). Full catastrophe living: Using the wisdom of your Ring, C., & Brener, J. (1992). The temporal locations of heartbeat sen-
body and mind to face stress, pain, and illness. New York: Delta Trade sations. Psychophysiology, 29, 535–545.
Paperbacks. Ring, C., & Brener, J. (1996). Influence of beliefs about heart rate
Katkin, E. S., Blascovich, J., & Goldband, S. (1981). Empirical assess- and actual heart rate on heartbeat counting. Psychophysiology, 33,
ment of visceral self-perception: Individual and sex differences in the 541–546.
acquisition of heartbeat discrimination. Journal of Personality and Rouse, C. H., Jones, G. E., & Jones, K. R. (1988). The effect of body
Social Psychology, 40, 1095–1101. composition and gender on cardiac awareness. Psychophysiology, 25,
Katkin, E. S., Wiens, S., & Ohman, A. (2001). Nonconscious fear con- 400–407.
ditioning, visceral perception, and the development of gut feelings. Schandry, R. (1981). Heart beat perception and emotional experience.
Psychological Science, 12, 366–370. Psychophysiology, 18, 483–488.
Khalsa, S. S., Rudrauf, D., Sandesara, C., Olshansky, B., & Tranel, D. Schandry, R., Bestler, M., & Montoya, P. (1993). On the relation be-
(in press), Bolus isoproterenol infusions provide a reliable method for tween cardiodynamics and heartbeat perception. Psychophysiology,
assessing interoceptive awareness. International Journal of Psycho- 30, 467–474.
physiology. Schneider, T. R., Ring, C., & Katkin, E. S. (1998). A test of the validity of
Knapp, K., Ring, C., & Brener, J. (1997). Sensitivity to mechanical the method of constant stimuli as an index of heartbeat detection.
stimuli and the role of general sensory and perceptual processes in Psychophysiology, 35, 86–89.
heartbeat detection. Psychophysiology, 34, 467–473. Selby, J. (1992). Kundalini awakening: a gentle guide to Chakra activation
Kornfield, J. (1996). Living dharma: Teachings of twelve Buddhist masters. and spiritual growth. New York: Bantam Books.
Boston: Shambala Publications. Teng, X. F., & Zhang, Y. T. (2006). The effect of applied sensor
Lutz, A., Dunne, J. D., & Davidson, R. J. (2007). Meditation and the contact force on pulse transit time. Physiological Measurement, 27,
neuroscience of consciousness: An introduction. New York: Cambridge 675–684.
University Press. Weisz, J., Balazs, L., & Adam, G. (1988). The influence of self-focused
Nairn, R. (2000). What is meditation? Buddhism for everyone. Boston: attention on heartbeat perception. Psychophysiology, 25, 193–199.
Shambhala Publications. Whitehead, W. E., Drescher, V. M., & Heiman, P. (1977). Relation of
Nielsen, L., & Kaszniak, A. W. (2006). Awareness of subtle emotional heart rate control to heartbeat perception. Biofeedback and Self-Reg-
feelings: A comparison of long-term meditators and nonmeditators. ulation, 2, 371–392.
Emotion, 6, 392–405. Wiens, S., & Palmer, S. N. (2001). Quadratic trend analysis and heartbeat
Peng, C. K., Henry, I. C., Mietus, J. E., Hausdorff, J. M., Khalsa, G., detection. Biological Psychology, 58, 159–175.
Benson, H., et al. (2004). Heart rate dynamics during three forms of Windmann, S., Schonecke, O. W., Frohlig, G., & Maldener, G. (1999).
meditation. International Journal of Cardiology, 95, 19–27. Dissociating beliefs about heart rates and actual heart rates in patients
Phillips, G. C., Jones, G. E., Rieger, E. J., & Snell, J. B. (1999). Effects with cardiac pacemakers. Psychophysiology, 36, 339–342.
of the presentation of false heart-rate feedback on the performance
of two common heartbeat-detection tasks. Psychophysiology, 36,
504–510. (Received November 14, 2007; Accepted December 13, 2007)