Interpersonal Synchronization of Breathing while Singing

When people sing together their cardiac and respiratory rhythms synchronise as previously shown by Müller and Lindenberger [1]. Inspired by the results of the study I asked myself a couple of questions:

1. To what extent do breathing patterns synchronise among a group of young high school students that have never sung songs together before?
2. Can singing together increase connectedness and generate uplifting peak experiences among young people in Norway?

The article presents the description of a workshop where high school students in Norway were taught to sing together in a choir. During the practice and a small final concert I measured their abdominal breathing patterns. This data was subject to some analysis to help me understand the degree of synchronization among the students. Finally, I discuss the basis for interpersonal synchronization and need for creating shared peak experiences among Norwegian youth.

The Experiment

Simula Research Laboratory invited high school students from Valler vidergående skole to a day long workshop in March 2019. I invited Ombeline Chardres, the choir conductor of Oslo Global Choir and a professional musician to teach students to sing together for the first time ever. Ombeline has considerable experience teaching music to both children and adults. I had great confidence that she would be able to make the children perform in a short time span of only two hours.

We prepared a repertoire of three relatively simple songs to be learnt in the span of two hours : Whoop Jamboree (Norwegian sailor song), We Will Rock You (Queen), and Let it be (The Beatles). Ombeline familiarized the students with song lyrics and divided its verses into different voice groups for the boys and girls. She essentially taught each group their respective part and then combined group efforts into entire song performances.

Figure 1. Learning the lyrics for choir singing

There were about 21 students of which 7 students (those who volunteered) were strapped only around the abdomen with the FLOW sensors developed by Sweetzpot. Typically, one may measure both ribcage and abdominal breathing using two sensors as shown in Figure 2. However, we only chose to record abdominal breathing as it was most relevant to singing where abdominal muscles are most used for phonation (production of speech sounds).

Figure 2. Example use of FLOW sensors around the chest and the abdomen

While Ombeline was going through the practice sessions for the songs, I was recording abdominal breathing patterns from the seven students using a Bluetooth Dongle as shown in Figure 3. The Bluetooth dongle concurrently connected to the 7 FLOW sensors and stored the data on a desktop computer.

Figure 3. Collection of breathing data using the Bluetooth Dongle

The breathing data is nothing but the forces exerted by the expansion and the contraction of the abdomen while singing. I recorded breathing data during parts of the two hour practice and during a final concert presented to the employees of Simula Research Laboratory as shown in Media 1 below.

Media 1. Concert Performance with the Young Students of Valler skole

Analysis

“Complex bodily rhythms are ubiquitous in living organisms” [2], and it becomes interesting to understand how these rhythms synchronise with each other and the signals from the environment. This can especially be true for team activities such as choir singing. For instance, a 90 second excerpt of choir singing, in Figure 4, shows how the 7 singers on several occasions inhale/exhale at the same time or with a small shift in time of just a few milliseconds.

Figure 4. 90 seconds of concert singing

Taking a close look at two singers in Figure 5: Mailen and Solveig we see that they inhale or exhale at approximately the same time on several occasions. One reason could be that both sang in unison as part of the sopranos in the choir and engaged their breathing muscles in the abdomen at the same time.

Figure 5. The breathing patterns of Mailen and Solveig

Can we quantify the degree of synchronization between two singers? To address this question I used metrics giving insight into how well two singers synchronize with each other. The metrics I used were:

1. Pearson’s correlation coefficient (-1 ≤ ρ ≤ 1)
2. Maximum information coefficient (0 ≤ MIC ≤ 1)

Given the breathing patterns from two people as input the Pearson’ coefficient ρ is a number between -1 and 1 that indicates the extent to which two variables are linearly related. The variables in our case is a vector of integer values proportional to the forces of expansion and contraction exerted by the abdomen over a period of time. The values are in millivolts of potential difference across a semiconductor strain gauge measured by the FLOW sensor.

The maximum information coefficient (MIC) [3] is a number between 0 and 1 and is the strength of the linear or non-linear association between two variables. MIC is able to find nonlinear relationships better than Pearson’s coefficient as shown in Figure 6. When two patterns are similar and have a nonlinear relationship instead of a linear relationship the MIC is systematically higher than the Pearson’s coefficient.

FIgure 6. Pearson’s coefficient and MIC score for different relationships

After some initial preprocessing of the data for seven singers we computed the correlation coefficients ρ and MIC between all pair of the singers as shown in Figure 7 for 90 seconds of singing in the concert setting.

Figure 7. Pearson’s coefficient and Maximum information coefficient between all singers.

The values of ρ and MIC were quite different between singers. For instance, Mailen and Stian showed a stronger nonlinear relationship (MIC = 0.34) compared to a weak linear relationship (ρ = 0.08). Who synchronised with who while singing in the concert? To answer this question we looked at the network of correlations between all the student singers. Given a threshold of 0.4 for both ρ and MIC, in Figure 8 we see how Markus was most in synchronization with Emma, Stian, and Johanna. Gjermund and Johanna had a strong synchronization. Markus and Gjermund with a MIC score of 0.35 fell just below the threshold and Mailen and Markus showed the least correlation.

Figure 8. (a) Pearson’s correlation coefficient network between singers (threshold=0.4) (b) MIC score network between singers (threshold = 0.4)

The MIC score performed much better than Pearson’s coefficient to discover some degree of nonlinear synchronization between the singers. What if the singers were breathing randomly (see Figure 9)? Would there be a strong interpersonal synchronization? The answer is no.

Figure 9. Random breathing in the same range of values as the original recordings

If the singers sing in random then then breathing is not correlated at all with MIC scores of 0.12 or 0.13 and Pearson’s coefficient at the order of 0.01 as shown in Figure 10. This base line shows how much humans consciously tend to fall in sync with each other. We are interlinked through our vital signs such as our breathing and we don’t even think about it!

Figure 10. No observed correlation when breathing is random

The entire source code for the analysis is available on Github.

Discussion

Can our vital signs such as breathing say something about how well a group of diverse people work together? This is the question that originally intrigued me. The question is of great relevance in a globalised world where people with very different backgrounds and ethnicities need to work together and address creative, industrial and societal challenges. We often perceive a good team to be working well when the interactions happen spontaneously and people feel excited about the same things. The shared experience can be perceived when you hang around such a team, it is scintillating and is often can be contagious. We made an attempt to measure the degree of synchronization in such a shared experience through breathing patterns of novice choir singers using metrics such as the Pearson’s coefficient and the Maximum Information Coefficient. The results show that people indeed synchronize but why?

Humans tend to fall in sync and imitate each other. The degree of synchronization may be high but it could also mean that everyone is equally excited about a task or just equally depressed sitting in a room with nothing to do. Many researchers have suggested the mirror neuron system [7] to explain interpersonal coupling but the underlying mechanism is still not well understood. Ramamoorthy et. al. [8] explains the reason to fall in sync as a form of continuous prediction and alignment. This results in the overall minimization of free energy [9] where we humans (adaptive agents) aim for a limited repertoire of states and hence minimize the long-term average of surprise associated with sensory exchanges with the world. This could also explain why our breathing patterns synchronise in order to minimize free energy when students sing together. Singing out of sync and in random would produce too many surprises for the sensory system and hence people prefer to be in sync with those around them.

In the student choir, Ombeline orchestrated the activities of the students who had no prior choir singing experience. From our experiments we obtained some signs of interpersonal synchronization as seen from their breathing patterns. Students were inhaling at similar moments of time and their exhales were often synchronised as well, possibly when singing in unison. The students reported an uplifting experience based on the overall workshop. Peak experiences in the Norwegian youth [4], especially in a group activities is of great importance to education. Norway is more of an individualistic rather than a collectivistic society where self-focused activities are highly valued (skiing alone for several kilometers in the forest). However, the education system must realize that individualism is not for everyone. Some great achievements are based on team effort where the members need to achieve a high level of synchronization. Many crave for such shared experiences. How can we create more such experiences? This little experiment gives us first signs that group activities orchestrated by a mentor could help create peak shared experiences among the Norwegian youth. I would be very happy to run the same experiment based on measurement of breathing in other activities : running/cycling/rowing together, programming together, sitting in a library together, dancing together, watching a movie together…and much more. Contact me if interested.

Acknowledgements

Special thanks to Pierre Bernabet, Freyja Jørgensen, Marianne Aasen, and Emmy Terese Lind for helping me organize and generate media content during the workshop.

Further reading

1. Müller, V., & Lindenberger, U. (2011). Cardiac and respiratory patterns synchronize between persons during choir singing. PloS one, 6(9), e24893.
2. Glass, L. (2001). Synchronization and rhythmic processes in physiology. Nature, 410(6825), 277.
3. Reshef, D. N., Reshef, Y. A., Finucane, H. K., Grossman, S. R., McVean, G., Turnbaugh, P. J., & Sabeti, P. C. (2011). Detecting novel associations in large data sets. science, 334(6062), 1518–1524.
4. Hoffman, E., Iversen, V., & Ortiz, F. A. (2010). Peak-experiences among Norwegian youth. Nordic Psychology.
5. Gadagkar, V., Puzerey, P. A., Chen, R., Baird-Daniel, E., Farhang, A. R., & Goldberg, J. H. (2016). Dopamine neurons encode performance error in singing birds. Science, 354(6317), 1278–1282.
6. Schmidt, R. C., Carello, C., & Turvey, M. T. (1990). Phase transitions and critical fluctuations in the visual coordination of rhythmic movements between people. Journal of experimental psychology: human perception and performance, 16(2), 227.
7. Rizzolatti, Giacomo, and Laila Craighero. “The mirror-neuron system.” Annu. Rev. Neurosci. 27 (2004): 169–192.
8. Koban, L., Ramamoorthy, A., & Konvalinka, I. (2019). Why do we fall into sync with others? Interpersonal synchronization and the brain’s optimization principle. Social neuroscience, 14(1), 1–9.
9. Friston, Karl. “The free-energy principle: a unified brain theory?.” Nature reviews neuroscience 11.2 (2010): 127.