Social pulse: The effects of mediated heartbeat communication on social connectedness, liking and pro-social behaviour.



Master thesis

Master Thesis Human-Technology Interaction. Supervised by Wijnand A. IJsselsteijn & Daan van Bel. Eindhoven: Eindhoven University of Technology


  • van Gennip, D.


Short intro

Tim Burton’s movie “Corpse bride” (2005) tells the tale of a bride killed during the night before her wedding. Unable to find rest she remains to dwell on Earth. She falls in love with a young man rehearsing his vows in the forest the night before his own wedding. The bride takes his vows as meant for her and gladly accepts the offer. As their relationship is off to a peculiar start she wonders if she can love him like any other living girl, since her heart cannot beat for him.

The doubt about true love without a beating heart is a small element in the storyline but it points at the special role the heart takes in the human view of social relationships and emotion. In this thesis I will argue why mediated intimacy is valuable but hampered by our current means of communication. Using heartbeats as a communication medium could relieve some of the problems associated with mediated communication and lack of social connectedness with other people. This topic resonates with the philosophy of affective computing (Picard, 1997), and more specifically it relates to what Bell, Brooke, Churchill and Paulos (2003, p. 2) describe as intimacy through technology; the technology enabling the mediation of intimate signals to another person. Joris Janssen labelled it physiosocial technology in his doctoral thesis (2012), and his work provides a backdrop for the work reported here. This thesis aims to address the question whether mediated heartbeat communication can prove its worth in dyadic social interactions that are of a longer duration than previous work has covered (Aron & Aron).

Three-page summary

People are social beings. We desire to have meaningful, intimate relationships with other people. Most of this social fabric is woven in face-to-face communications. Because in face-to-face communication there is no obstruction in conveying disclosure of signals, it is often held as the highest form of communication. It offers the ultimate experience in verbal and non-verbal affective information for all parties involved. However such affective qualities are less well conveyed in forms of communication where partners are in different locations. Currently available technologies for mediated communication are often considered suboptimal for expressing intimacy which leaves room for improvements.

This topic resonates with the philosophy of affective computing and more specifically intimacy through technology; the technology enabling the mediation of intimate signals to another person. This thesis builds on the work of Joris Janssen and others, and follows the idea that one way to make people feel closer is to communicate something about them that is closely related to emotions, namely physiology. Emotions have both a cognitive and a physiological component thus any emotional experience is re$ected in physiological responses; think sweat and fear, or a raging heartbeat for something exciting. The heart and heartbeats are often regarded as a very emotional part of our bodies, so the core idea is that communicating this signal can tell us something about the person who the heartbeat belongs to. Normally these signals can only be experienced if one is very close to the physique of the source, which perhaps underlines the intimate character of such cues. Using heartbeats as a communication medium could thus relieve some of the problems associated with mediated communication and lack of social connectedness with other people.

The study reported in this work builds on earlier efforts that show heartbeat communication can indeed be seen as an intimate cue. First of all, giving people false feedback of their own heartbeat can alter impression of an external artefact, because they may attribute the perceived change to this externality (the so-called Valins effect). Janssen et al. (2010) have shown that attribution of an artificial heartbeat to another person made people keep more distance to this person (or virtual avatar in some of the studies), a sign they try to compensate for an increase in apparent perceived intimacy. The main hypothesis is that if physiology-based communication is regarded as a meaningful form of disclosure it should affect mutual appraisal in similar fashion to other forms of (non-)verbal disclosure. Previous studies indeed seem to confirm this hypothesis although the duration in those studies was in the order of minutes which is short in comparison to social relations. Perhaps the effects found thus far are limited to the short timeframe as a kind of novelty effect.

This study tries to shine a light on the question whether mediated heartbeat communication can also show effects on perceptions of disclosure, social connectedness, liking and pro-social behaviour when people are interacting for a longer period of time. More specifically, this study assesses the merits of using mediated heartbeat communication in a text-based conversation. This study used a mediated text-based chat format through which people will get to know each other. We conducted a laboratory experiment using a 2 (personal topics of conversation versus small talk) x 2 (attributed to other versus non-attributed heartbeat cue) between-subjects design.

Participants were randomly assigned to one of the four conditions and participated in one condition only. More personal, sensitive topics of conversation were hypothesised to increase perceived disclosure, social connectedness and closeness – similar in reasoning to the expected effects for heartbeat communication. Participants were matched with one of the two female confederates. Couples would not meet before they engaged in a text-based chat session for fifteen minutes. The participant was assigned to ask predetermined questions while the confederate disclosed her answers. This asymmetric setup allows for scripting the conversation, hereby increasing control over the discussed content. The audio stimulus heard was the same across conditions, namely an artificially constructed but realistic heartbeat sound. Dependent variables of the study are self-report measures on disclosure, social connectedness, closeness, liking and pro-social behaviour.

Results did not confirm the expectations. While the manipulations appear to have registered as intended, the expected beneficial effects of more personal conversation and heartbeat communication were mostly absent. Because self-reports alone may be subject to cognitive biases, a measure of pro-social behaviour was included as well. Results on this Dictator game fell short of expectations, as the majority of participant decided to split the endowment equally. In short this means for all predictions the null hypotheses cannot be refuted. Thus in this study no general effects were found for mediated heartbeat communication nor for different levels of interpersonal closeness (except for a partial measure on relationship quality). This also means there is no cumulative effect of both manipulations, at least not within the limitations of this study.

Limitations include some of the design choices made. The rigid structure of using a confederate to disclose scripted answers can be seen as a negative in$uence for natural conversation. All participants within the same interpersonal closeness condition were exposed to approximately the same content but quite a few participants expressed their discontent at the imposed limits to their input. Participants were (mostly) strangers at the beginning but the expectation that marginal gains in closeness would thus be higher seems incorrect. Perhaps the lack of a pre-chat real life introduction kept participant too far removed from each other, hereby outdoing the supposed benefit of limiting pre-study impressions. There is a possibility people do not have absolute evaluations for their experiences but rather those are relative to other experiences. This would imply one measure such as taken in a between-subjects design as was used here cannot capture the relative value people attach to an experience.

It may be that initially the impact of hearing someone’s heartbeat can be higher and then wane off or drown amid other social cues. Since no discernible effects were found after the 15 minutes of chatting this novelty effect could be a reasonable explanation. Because no intermediate measures were taken it cannot be determined whether there would have been effects at the beginning. In that sense the current methodology was insufficient. However, if this would be true a novelty effect would go against the anticipated, positive effect of familiarity or such an advantage does not pay off within a quarter of an hour.

The heartbeat stimulus employed was not based on a live signal from the conversation partner. Although having a constant stimulus across conditions has methodological benefits, there are potential benefits to a live signal. Ecological validity would the prime argument as the stimulus would broadcast any cardiovascular response. It would allow for interpretation of the changes in light of what happens between participants, in this study the ongoing conversation. Some of the participants gave feedback indicating they were looking for such effects. Although the stimulus incorporated more variability than in previous studies it was still perceived as quite monotonous, perhaps to the detriment of perceived realism and believability of the mediation. Most participants admitted to not really paying attention to the heartbeat stimulus which may explain the modest strength of the manipulation. People felt the stimulus was a “background sound,” and for some it may have escaped their attention as it seemed to contain little information.

It appears for heartbeat communication to be a valuable source of information it must lend itself to interpretation. Without interpretation it is likely to carry less meaning. Such interpretation hinges on context and familiarity with the source of the heartbeat signal. This study illustrates that without such familiarity and a clearly interpretable relation between the in$uence (i.e., the action, thought, or emotion) and the state of this signal (i.e., heart rate [variability]), there is no easily discerned effect of its communication.

From the outset this work has leaned more towards a focus on the value that can be offered by mediated heartbeat communication. While this work aims to help the development of intimate technologies, the investigation of relationship formation between strangers is only a precursor to intimacy. Intimate acts are often nuanced, having acquired meaning through development of mutual interactions. If one thing can be taken from the results of this study it is that just adding a (mediated) heartbeat signal does not work wonders, as it may over time become a background noise. Having purpose and context helps its interpretation, and therefore practical value in aiding bringing people closer together through technology. However, as has been noted in the discussion, the connection itself as a means to maintain rapport may also be what lends intimacy rather than the content. Follow-up studies could opt for a different methodology to study the same effects in a different way, and try to alleviate some of the limitations of this study.