Commentary
Comment on “Zamariola et al. (2018), Interoceptive Accuracy Scores are Problematic: Evidence from Simple Bivariate Correlations”—The empirical data base, the conceptual reasoning and the analysis behind this statement are misconceived and do not support the authors’ conclusions

https://doi.org/10.1016/j.biopsycho.2020.107870Get rights and content

Highlights

  • We comment on Zamariola et al.’s (2018) 4 criticisms of the heartbeat-counting task.

  • Interoceptive accuracy (IAcc) is essentially a ratio of reported/recorded heartbeats.

  • Spurious “simple bivariate correlations” by IAcc quintile rely on circular reasoning.

  • Heart rate correlates with IAcc because recorded heartbeats depend on heart rate.

  • Differences of IAcc between trials are confounded by recorded heart rate. We present data to refute this.

Abstract

A recent paper by Zamariola and colleagues is widely cited as an authority on the invalidity of the Heartbeat Counting Task as a measure of interoceptive accuracy. Given the widespread interest in this field, it is essential that papers about methods are conceptually sound. However, only one of the authors’ four criticisms appears substantiated - that people count too few heartbeats. Their arguments about “simple bivariate correlations” and their finding that interoceptive accuracy and heart rate correlate, depend on ‘spurious correlations’ arising from the overlooked point that interoceptive accuracy is a ratio. Moreover, scrutiny of the authors’ data shows that their fourth criticism (that interoceptive accuracy is lower on longer trials) is confounded by differences in mean heart rate between trials. We present data from our own labs to refute it. We draw the authors’ and editors’ attention to these issues and trust that they will reconsider these erroneous conclusions.

Introduction

Interoception, and in particular awareness of interoceptive states, has been the subject of a recent exponential growth in research interest, across various sub-disciplines of psychological sciences and social-affective neuroscience (Khalsa & Lapidus, 2016). To date, there are two standard interoceptive accuracy tasks that are widely used for measuring Interoceptive Accuracy (IAcc) in the cardioceptive domain. In the ‘Heartbeat Discrimination Task’ (Whitehead & Drescher, 1980) individuals report the perceived synchrony of their heartbeats with a series of external stimuli (usually auditory cues). During the ‘Heartbeat Counting Task’ (Schandry, 1981) participants count their own heartbeats for several short periods of time, between the onset and offset of a signal (e.g. a tone). Interoceptive accuracy is quantified as a heartbeat perception score which relates the number of counted (i.e. subjectively reported) heartbeats to the actual (i.e. objectively recorded) heartbeats, in the various counting intervals, with higher scores indicating greater accuracy.

For n counting trials, this formula is given by:IAcc=1-1n(recordedHBs-reportedHBs)/recordedHBs

(See Section 7 below for more detailed comment on the Heartbeat Counting Task.)

While it has been argued that each task has advantages and disadvantages, the Heartbeat Counting Task (HCT) has come under increasing criticism (Ring & Brener, 2018). In their recent article, Zamariola, and colleagues present arguments and empirical data to support their proposal that heartbeat counting is not a valid measure of IAcc (Zamariola, Maurage, Luminet, & Corneille, 2018). They illustrate their arguments with a data set of students, N = 572 (386 women, mean age = 22.2, SD = 4.3, publicly available at https://doi.org/10.5281/zenodo.1288323 that includes the actual/recorded and the reported heartbeats for each trial interval, from which interoceptive accuracy is calculated, according to the formula above. By way of contrast, we make publicly available at https://osf.io/axv4w/ a comparable data set of N = 394 (303 women, ages 18–59 years, mean = 22, SD = 5.8) from our labs, with trials of the same length, in the HCT in healthy subjects. (Our sample includes fewer men and a small number of older participants than Zamariola’s sample.)

Specifically, Zamariola and colleagues put forward the following four key criticisms. Firstly, they note that IAcc, as measured by the HCT, largely depends on only one “error type”, which is the under-reporting of heartbeats (i.e. most people count fewer than the true number of their heartbeats). Secondly, they propose that the total number of actual heartbeats and total number of reported heartbeats (summed across all the test trials) should correlate. In their data, they found only a weak overall positive correlation for the data set as taken a whole. Furthermore, they argue that the correlation between the total number of actual and the total number of reported heartbeats should increase in a linear fashion as IAcc rises, i.e. across IAcc quintile groups. They found no such rising linear trend. Thirdly, they suggest that a valid measure of IAcc should not be “structurally tied to heart condition” but they found that IAcc was negatively correlated with heart rate (HR). Finally, the authors report that, in their published data set, IAcc scores differed significantly across the three testing intervals of the Heartbeat Counting Task, with lower IAcc on the 45 s trial than on the 25 s or 35 s trial. They imply that this reflects a general (but presumably under-reported) tendency to poorer performance on longer trials of the HCT. We consider these criticisms in turn and provide critical and constructive evaluation of the authors’ claims. We also address some more general issues that relate to the implementation of the HCT task, data collection and data quality.

Section snippets

Under-reporting of heartbeats

The proposal in Zamariola’s paper that heartbeat counting should result in false positives (as in ‘Heartbeat Discrimination tasks’), is not self-evident. On the contrary, it seems reasonable that when people are asked to attend to their bodies and count their heartbeats they will miss some beats, as would be the case in most perceptual signal detection tasks that have substantial levels of noise. Were they to over-count, this would imply that they were hallucinating heartbeats that were never

Simple bivariate correlations” between actual and reported heartbeats are mathematically spurious

In their second argument, the authors state: “if IAcc scores reflect people’s ability to accurately perceive their inner states, a correlation between actual and reported heartbeats should be observed and this correlation should linearly increase with higher IAcc scores. This is not the case. The correlation between actual and reported heartbeats is low overall. And, even more critically, it is higher at average than higher in IAcc levels” (Zamariola et al., 2018, p. 13). In their data set, the

The relationship between IAcc and Heart Rate (HR)

The authors report a negative correlation between heart rate and IAcc in their data and propose that this undermines the validity of the test. Again, this relationship is a function of the ratio arithmetic that links these two variables.

We have already shown that for people who undercount heartbeats (who are in the great majority):IAccreportedHBsrecordedHBs

However:HeartRate=recordedHBstriallength*60

Rearranging this:recordedHBs=HeartRate*(triallength60)

Therefore:recordedHBs=k*HeartR

The authors report significant differences in IAcc scores across their three HCT trials

Zamariola and colleagues’ fourth criticism of the HCT follows from their finding that in their data “IAcc scores were significantly lower for 45 s interval than for 25 s [t(571) = 6.61, p < .001] and 35 s [t(571) = 7.68, p < .001].” They comment that: “as actual heart rate is quite constant across intervals, this variation is related to a reduction of reported bpm at the longest interval” (Zamariola et al., 2018, p.14). They imply that their finding is an example of a general tendency to

Further data anomalies

Zamariola and colleagues’ data was collected over four different experiments, using a Polar Watch RS800CX with software Polar ProTrainer5 to extract the actual number of heartbeats for calculations of the IAcc score. The authors do not comment in their Methods about the careful methodological protocols that must be followed when using the Polar Watches for accurate heartbeat counting. The most significant issue is that with any other heartbeat recording device (for example ECG) the researcher

Further comment on the heartbeat counting task

The Heartbeat Counting Task (Schandry, 1981) is very widely used to assess cardioceptive IAcc (e.g. Ainley, Maister, Brokfeld, Farmer, & Tsakiris, 2013; Garfinkel, Seth, Barrett, Suzuki, & Critchley, 2015; Herbert, Herbert et al., 2012; Herbert, Muth et al., 2012; Pollatos, Kirsch, & Schandry, 2005b; Schulz, Lass-Hennemann, Sütterlin, Schächinger, & Vögele, 2013). Critics, however, (e.g. Brener & Ring, 2016) suggest, for example, that participants may use knowledge of their heart rates and of

Summary: three of the authors’ four criticisms of the heartbeat counting task are not valid

To reprise, Zamariola and colleagues criticise heartbeat counting tasks on four grounds:

  • 1

    They note that the majority of people undercount on the HCT. This is not news. In our own data, only 3.8 % over-counted. This pattern has often been reported in the past and it remains an area to be investigated (Ring, Brener, Knapp, & Mailloux, 2015; Schandry & Specht, 1981; Schandry et al., 1993).

  • 2

    The authors’ second criticism is that they found a low overall Pearson correlation between total actual

Conclusion

We write to express our concern about: the soundness of the mathematical/logical arguments of the authors; significant aspects of the authors’ methodology that are omitted from the paper; the quality of the data collected; and the correctness of the analyses reported and illustrated.

Given the reach of this paper and its potential impact, as well as increasing interest in interoceptive accuracy across different sub-disciplines of psychological sciences, it is important for these concerns to be

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper."

References (58)

  • J.P. Kootz et al.

    Social interaction and sensory intake rejection cardiovascular correlates of attention in young adults and school-aged boys

    Journal of the American Academy of Child Psychiatry

    (1979)
  • S. Mai et al.

    Interoception is associated with heartbeat-evoked brain potentials (HEPs) in adolescents

    Biological Psychology

    (2018)
  • P. Montoya et al.

    Heartbeat evoked potentials (HEP): Topography and influence of cardiac awareness and focus of attention

    Electroencephalography and Clinical Neurophysiology

    (1993)
  • O. Pollatos et al.

    On the relationship between interoceptive awareness, emotional experience, and brain processes

    Cognitive Brain Research

    (2005)
  • O. Pollatos et al.

    Interoceptive awareness, anxiety and cardiovascular reactivity to isometric exercise

    International Journal of Psychophysiology

    (2007)
  • O. Pollatos et al.

    Brain structures mediating cardiovascular arousal and interoceptive awareness

    Brain Research

    (2007)
  • C. Ring et al.

    Effects of heartbeat feedback on beliefs about heart rate and heartbeat counting : A cautionary tale about interoceptive awareness

    Biological Psychology

    (2015)
  • A. Schulz et al.

    Cold pressor stress induces opposite effects on cardioceptive accuracy dependent on assessment paradigm

    Biological Psychology

    (2013)
  • G. Zamariola et al.

    Interoceptive accuracy scores are problematic: Evidence from simple bivariate correlations

    Biological Psychology

    (2018)
  • L.F. Barrett et al.

    Interoceptive sensitivity and self-reports of emotional experience

    Journal of Personality

    (2004)
  • J. Brener et al.

    Towards a psychophysics of interoceptive processes: The measurement of heartbeat detection

    Philosophical Transactions of the Royal Society B

    (2016)
  • O.G. Cameron

    Interoception: The inside story--A model for psychosomatic processes

    Psychosomatic Medicine

    (2001)
  • A.D. Craig

    Interoception: The sense of the physiological condition of the body

    Current Opinion in Neurobiology

    (2002)
  • H.D. Critchley et al.

    Cerebral correlates of autonomic cardiovascular arousal: A functional neuroimaging investigation in humans

    The Journal of Physiology

    (2000)
  • H.D. Critchley et al.

    Neural systems supporting interoceptive awareness

    Nature Neuroscience

    (2004)
  • C. Durlik et al.

    Being watched : The effect of social self-focus on interoceptive and exteroceptive somatosensory perception

    Conciousness and Cognition

    (2014)
  • E. Georgiou et al.

    Interaction of physical activity and interoception in children

    Frontiers in Psychology

    (2015)
  • J. Hart

    Normal resting pulse rate ranges

    Journal of Nursing Education and Practice

    (2015)
  • B.M. Herbert et al.

    Cardiac awareness and autonomic cardiac reactivity during emotional picture viewing and mental stress

    Psychophysiology

    (2010)
  • Cited by (62)

    View all citing articles on Scopus
    1

    Both authors contributed equally.

    View full text