Understanding heart rate sharing: towards unpacking physiosocial space

Understanding Heart Rate Sharing: Towards Unpacking Physiosocial Space Petr Slovák1 , Joris H. Janssen2,3 , Geraldine Fitzpatrick1 Vienna University of Technology, Argentinierstrasse 8, Vienna, Austria 2 Philips Research, High Tech Compus 34, Eindhoven, The Netherlands 3 Eindhoven University of Technology, Den Dolech 2, Eindhoven, The Netherlands {petr,geraldine.fitzpatrick}@igw.tuwien.ac.at, joris.h.janssen@philips.com 1 ABSTRACT Advances in biosensing make it possible to include heart rate monitoring in applications and several studies have suggested that heart rate communication has potential for improving social connectedness. However, it is not known how people understand heart rate feedback, or what issues need to be taken into account when designing technologies including heart rate feedback. To explore this, we created a heart rate communication probe that was used in two qualitative in-lab studies and a two-week field trial in participants’ homes. Results show that heart rate feedback is a strong connectedness cue that affects the interaction in various ways, depending on a number of interrelated factors. In particular, we found two distinct categories of effects: heart rate as information and heart rate as connection. We propose two mechanisms that could explain these observations and draw out the implications they have for future use of heartbeat communication to support social connectedness or other aspects of social interaction. Author Keywords social connectedness; relationships; physiological signals ACM Classification Keywords H.5.3 Information Interfaces and Presentation: Group and Organization Interfaces INTRODUCTION Social connectedness is one of the most important predictors of well-being [22, 32], increasing happiness [3] and reducing the risk of depression [8]. Moreover, social connectedness also has strong effects on physical health, influencing, for instance, neuroendocrine and cardiovascular systems [29] and systolic blood pressure. This importance of social connectedness has sparked research on systems and technologies in HCI that can improve our sense of belonging and help reduce our feeling of loneliness, such as awareness systems for families and friends [7, 10, 24, 31] and devices supporting social connectedness for remote couples [12, 34]. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. CHI’12, May 5–10, 2012, Austin, Texas, USA. Copyright 2012 ACM 978-1-4503-1015-4/12/05...$10.00. Recent work has also pointed towards the potential of heartbeat communication for increasing social connectedness. Such studies are building on an increasing maturity of biosensors, which is becoming an area of increasing interest for the CHI community also in other contexts, e.g., [23, 11]. The imPulse system by Lotan and Croft [20] wirelessly transmits heartbeat rhythms to companion units, synchronizing light and vibrations with participants’ heartbeats. Similarly, Werner et al [35] proposed the ’united-pulse’ system for sharing heartbeats, using pulsing rings to measure and present heartbeats. Interviews suggested that participants from both studies thought that such systems can increase social connectedness. Finally, Janssen et. al. [15] showed that participants rate connectedness higher when led to believe that they hear another’s heartbeat. Moreover, participants adapted their nonverbal behavior towards the person they were hearing the heartbeat from in reaction to the felt increases in connectedness. Altogether, these findings suggest that sharing heartbeats could potentially be a powerful mechanism for awareness systems to increase social connectedness. Although previous work identifies the potential for heartbeat communication, it does not investigate how people make sense of the heartbeat feedback or provide an understanding of why the observed effects take place [15]. It is important to understand such mechanisms to aid design of potential systems for practical situations and to identify potential caveats and pitfalls of using heartbeat communication. Furthermore, only relatively short studies in laboratory environments have been conducted so far and we have little understanding of how people might make sense of heartbeat communication in more authentic social settings and over longer periods of time. In this paper, we present the results of a qualitative study aimed at understanding the effects of heartbeat communication in various contexts, using a technology probe. We first review HCI literature on social connectedness as well as additional psychology literature supporting the idea that heartbeat feedback might be a viable signal to support social connectedness. Next, we describe the technology probe and its deployment into the homes of couples, as well as the use of the same system in two qualitative in-lab studies. Focusing on data from interviews, we draw out the recurring patterns that appeared over both home deployment and laboratory contexts. In particular, we discuss two core categories of heart rate (HR) sharing effects that have emerged from our data: HR as information and HR as connection. We then propose a mechanism for each category, giving a plausible explanation for the observed effects, and discuss their implications for future work. The main contribution of this work is identifying how people themselves make sense of heart rate feedback within the context of different social situations and articulating the possible mechanisms underlying the observed increases in social connectedness during HR sharing. Moreover, our data extends prior work by suggesting how previously described effects of HR sharing, found in the lab, transfer to everyday contexts and to inferences of other people’s HR. By noting the key differences between two categories emerging from our data, i.e., HR as information and HR as connection, we point to future research directions investigating how HR feedback can be used to support social connectedness. BACKGROUND Previous research has used several ways to support social connectedness (or a related concept, intimacy) between people, especially focusing on families and remote couples. Several studies aimed to support the partners or family to get in touch with one another via various kinds of messaging systems. In addition to using generally available technologies such as videoconferencing (e.g., [1]) or exchange of SMS messages, more elaborate concepts were developed. For instance, Strong and Gaver [30] suggested systems that change the physical surrounding (e.g., release scent) when a remote partner interacts with another object. Similarly, Garnaes et al. [12] designed systems to allow the partners to send expressive and emotional messages using a personalised code language. Also, a number of systems have been developed to support tangible communication with the partner, (e.g., Secret Touch in [34]). Other studies aimed at providing awareness of other’s everyday activities, for example by linking physical objects (e.g., pillows in [28]). Similarly, technologies such as the Whereabouts Clock [7] increased the feelings of connectedness to other members of a family by providing coarse information about the other’s location (e.g., also [10]). People also often use tokens like an object linked to a shared experience, photo in their wallet or picture on their workdesk to reduce the felt distance to people they would like to be close to. The work referenced above draws on several mechanisms to support social connectedness. For example, some of the work (e.g., [1, 12, 30, 34]) is building on the observation that intentional exchange of messages with a loved one often increases feelings of connectedness [24, Chp. 7], as it implies we are being thought of and cared for. Other work builds on the link between awareness (i.e., making the activities of a remote other more visible in our life), and feelings of connectedness, as argued for example by Dey and Guzman [10]. In this case, it is not the intentionality of the act, but rather the background awareness of the other’s activities that matters. Yet other work, as discussed in the Introduction, makes use of physiological biosensors to communicate heart rate to increase social connectedness, leading to, e.g., higher levels self-reported measures of closeness and adaptation of nonverbal behaviour. For heart rate communication, it is how- ever yet unclear which mechanisms are behind the observed effects [15]. For example, one possible explanation might be that its effects are inherent to heartbeat itself, given that it is a well known bodily signal connected in Western culture to love, life and emotions. Another possible explanation is that the observed effects could stem from heart rate as a signal that provides background awareness of the existence of the other, which would correspond to a similar mechanism as in Dey et al.’s work [10]. There are, however, other plausible explanations. For example, studies suggest that sharing heart rate could be understood as an involuntary self-disclosure, which has been repeatedly linked to increases in social connectedness (e.g., [18, 34]). The underlying rationale comes in several steps. In general, heart rate is strongly related to emotions. In a review of 134 studies, Kreibig [17] shows that HR changes during both negative and positive emotions (also, e.g., [25]). Related literature also suggests that people tend to interpret (their own) heart rate information in terms of emotions. Building on a seminal paper by Valins [33], a number of studies show how providing a false heart rate feedback influences participants’ perceptions of the situation (see [26] for a review). For example, when the false heart rate was higher, participants judged themselves as being more aroused. While we know that people make such appraisals based on their own perceived heart rate, we do not know how people interpret heart rate information of another person, although it is plausible to assume that similar effects might arise. Taken together, if people can interpret heart rate as information about the emotional state of another, sharing heart rate can be seen as a form of emotional self-disclosure. Due to the strong relation between increases in self-disclosure and connectedness, this could be another mechanism for the observed effects. In summary, heart rate can play a role in social connectedness and controlled laboratory studies show that communicating heart rate can make people feel closer together [15]. However, we do not understand the mechanisms underlying the observed increases of closeness and if and how these play out in more authentic social situations. A better understanding of these mechanisms will be useful for designers in deciding which contexts HR sharing might be useful (e.g., synchronous communication or background sharing), who might benefit most (e.g., couples, family, friends, strangers), and what other effects it might bring. STUDY METHOD As there is little prior work on HR sharing in everyday contexts, we chose to focus on understanding people’s experiences with, and use of, HR sharing technology in their everyday life. To explore these issues we developed a technology probe for sharing heart rate and conducted qualitative studies of its use in two different settings. Such qualitative approaches are often argued as a good method when analyzing a novel technology in HCI (e.g., [14, 16]). In order to tap into the initial reactions of a larger number of people, we set up two social interaction scenarios in the laboratory for 13 pairs of participants. To understand reactions based on longer term use in a home setting, we also conducted 2 week deployments with 5 couples who used the probe at home. We go on to describe the probe, each of the studies and the analysis. Figure 1. Example of visual feedback mode – random ripples The technology for HR sharing Following the suggestions of Hutchinson et. al. [14], we aimed to design a relatively simple technology that would nevertheless give the participants a possibility to interact with it freely, and, hopefully, use it in unexpected ways. A heartbeat belt (Zephyr HxM BT) was paired with a standard laptop (HP EliteBook 8440P), connected to the heartbeat belt using Bluetooth radio. Each pair of participants received two such sets, with the pairing of belts to computers clearly marked. Participants wear the belt on their body around their chest. The system logged the HR and system status (e.g., receiving HR information or not) at 1 Hz. Furthermore, every time a setting was changed this was logged and the latest settings were saved. The laptop could present the HR information both visually and aurally. The two modes of feedback ran independently, so the participant could choose if (s)he wants to use both at the same time, or select just one. The aural feedback was implemented by increasing the volume of a sound linearly with increases in HR. We provided the participants with two preselected sounds of nature (soft rain and crackling of campfire), aiming for sounds that could create a natural, unobtrusive background, as well as blend together well. The system supported three types of visual feedback on the display of the laptop. The first type used water ripples displayed over an image, either along the horizon (where the position of the ripple indicated current HR), or at random locations (the number of ripples shown simultaneously corresponded to HR, see Figure 1). The second type consisted of an image that was linearly faded to black with decreases in HR. The last possibility was a white number on a black background, with history over the last minutes on the bottom. Participants in the home study could switch between the feedback modes at will, add and change the visualised pictures and sounds, and also specify ranges in which the feedback will be active. The in-lab participants were provided with a default settings, see below. Laboratory studies In the two short-term in-lab studies, we wanted to tap into the initial impressions of participants in two different contexts, specifically: watching a movie together and doing a negotiation task. In both cases, the researcher conducted a 20 to 30 minute interview with the pair of participants about their experiences during the task. Participants got feedback using the same technology probe system as in the home deployment. In the first in-lab study seven pairs of volunteers participated. Participants (6 men and 8 women) knew each other mostly as colleagues and friends, or in the case of one pair, romantic partners. All participants worked at a large electronics company and were over 18 years old. When the participants came into the laboratory, the system was explained to them. They put on the HR belts and were asked to watch two videos together (when sitting on the couch next to each other). Both videos were animated short pictures that have received various awards at recent film festivals. The first was chosen to be slightly scary and the second funny, to lighten the mood. The movie watching took approx 10 minutes. During this time, the participants could perceive each their own and other’s HR, first in the form of the aural (5 minutes) and then visual feedback (5 minutes, ripples along the horizon mode). In the second in-lab study, six new pairs of volunteers participated (6 men and 6 women). The participants were recruited from the same population as in the first in-lab study. In this study, the participants first watched an 8 minute long neutral nature movie and were told that they can use the time to observe how their HR reacts. The system presented HR feedback both in the visual (ripples on horizon) as well as aural form. Afterwards, the participants played a classical task, called the desert survival situation (as used, e.g., in [6]), in which they have to negotiate what items are most useful for surviving in the desert. This took about 20 minutes. During this task they could again see and hear their own and other’s HR feedback. Home deployment study Five couples participated in the home deployment studies. Household A consisted of a married couple in their late 20s, both with a post-graduate education and working at a university. Household B consisted of a couple in their late 20s, who have been together for 2 years. Both have a PhD degree, John is a physicist and Maddie a medical doctor. Household C consisted of a couple in their early 20s, who have been together for 1 year. Both were post-graduate students at a university, on topics of HCI and design. Household D was a family of four: the father works as a Senior Manager at a big electronics company, the mother is a kindergarden teacher. They have two sons, aged 14 and 16. Household E consisted of a couple in their 20s, who have been together for 5 years. Stacy has just finished her master’s degree in design, and Brad has a PhD in Chemical engineering, working at a research department of a medium sized company. The home deployment started with an hour long session in which the researcher visited the home of the participants and explained the working of the system. Participants were told that they could use it in any way they wanted including changing the feedback, to feel free to explore, and that we were interested in their experiences with it. Apart for asking the participants to try the system out during the first two days, we left the decision whether, when, where, and how long to use the system up to them. We also did not tell them anything about the fact that we were interested in social interaction and the effects of HR on this. Participants then tested the system and the researcher left when everything was clear and setup prop- Figure 2. The diary probe package erly. Participants also received a probe package with a diary to stimulate them in thinking about possibilities of the system and record experiences they had with it (see Figure 2). The diary contained various stickers and colourful small objects to subtly promote the playful behaviour we were interested in. However, we did not include any catch-phrases or tasks to see what uses of the system the participants come up with by themselves. Participants were encouraged to write something in the diary once a day. After a week of use, the researcher conducted an interim interview with the participants for about 30 to 40 minutes. We were interested in their initial experiences, asked them to tell us notable stories from the last week and went through their diaries together. We were also interested in any changes to the system the participants would like, and whenever possible, tried to provide them with it (see, e.g., the poker game described in the next section). Finally, after two weeks of use, the researcher conducted a final interview with the participants lasting about one hour on average. We were interested in the situations they used it in and whether it changed their interaction in any way. We tried to explore what the participants found as important during their experience with the system rather then strictly following a prepared interview guide. The issue of social connectedness was often brought up by the participants independently, in which case we probed more deeply into it. In other cases, it was one of the last questions, not to skew participants’ responses. Analysis We analysed the data by first transcribing the interviews and then conducted thematic analysis of the data. Two of the authors went through all the transcriptions, diaries and recordings of the interviews independently and identified concepts and issues that were present in the data. We used a bottom up approach, with the concepts arising directly from the data, not from a predefined framework. The findings were then compared and clustered. Where there were disagreements, we went back to the audio recordings to check our findings to resolve issues. Finally, clusters of data were structured and related to each other. USAGE AND REACTIONS TO THE TECH PROBE This section presents an overview of the observations concerning the design and usage of the system. Due to space constraints and the fact that the main concern of this paper is with the interpretation of HR feedback by the participants in Figure 3. Set-up for the poker game (photo courtesy of our participants, published with their consent) social settings, rather than the design of the feedback system, we keep this section brief. With regards to usage, our short-term in-lab participants were generally interested in the HR feedback, often comparing their HR with their partner’s and doing little experiments (e.g., “what happens when I stand up”). They had similar reactions to the design as the home participants did. All of our home participants were initially very excited about the probe and were curious to see how their heart reacted to various situations. This initial curiosity and engagement wore off however after few days and all our couples reported that they soon knew what to expect with their HR in everyday activities and it became boring afterwards. As Annie put it in the final interview, “. . . so that’s what I learned from that. It’s not very exciting to monitor your HR [in everyday situations] because it’s you know always between 70-100”. However, most of our participants still tried various ‘experiments’ throughout to explore their HR changes – emotionally intense movies (e.g., Clockwork Orange), darts, playing computer games, physical activity etc. One couple in particular (Household A) thought it would be interesting to use the the system playing poker and requested 4 additional laptops for one night when they had friends over for a poker game. They enjoyed the game immensely, and considered it the highlight of the study. With regards to the offered feedback modalities, participants had often their personal favourite choices, depending on their prefered usage and personal taste. We report some of the more generalisable observations below. Most participants found that HR feedback is not very informative in many everyday moments, and continuous monitoring requires attention that does not provide enough value. In this sense, the continuous audio feedback provided enough information, but was often obtrusive at moments “when nothing happened”. Instead, many would have preferred notification of sudden changes in HR together with the possibility for more precise readings available if they wanted to pay attention. Similarly, the visual feedback gave detailed information, but participants felt that it required too much attention if monitored continuously. Participants suggested notifications based e.g., on a short sound when reaching a given threshold, or changes in lighting in the room (i.e., ambient, peripheral and unobtrusive feedback). Some also suggested various history options, allowing for example a shared reflection of the day with the partner. Participants also noted that the laptop was often too heavy and big to carry around easily if they wanted to move it to a different room, and would prefer tablet-like devices to provide a more fluent movement of the feedback throughout the home. the information”. Similarly, many participants thought they would not be able to hold back on interpreting. We saw some differences in extent of usage of the system between households. Households A, B and D used it quite often, with the system running on 9, 8, and 10 days respectively (the average usage in minutes per participant was approximately 2200, 800 and 1200, respectively). The remaining Households C and E used the system less: Household C on 3 days, for approx 200 minutes altogether; Household E then on 5 days, for approximately 600 minutes altogether. The differences in usage between households A, B and D arose from various work related constraints, spending a few nights out of the appartment etc. Participants from households C and E were all quite busy with work related activities during the evenings. This might be one of the reasons why they did not experiment with the system so much, never tried it out in situations different from everyday routine, and did not use it much after the initial curiosity wore off. Need for context EFFECTS OF HEART RATE SHARING Our particular focus in the analysis was on how people used and interpreted HR data in various social situations. We present data from both the shorter-term in-lab studies and home deployments together as most of the concepts appeared consistently over both contexts. We identified two main categories in our data. The first is about interpreting HR as objective information about one’s own or someone else’s internal state. The second is about HR as a direct connection to the other, without the need for it to give any understandable information about the others’ state. The next two sections discuss the main results respectively under these two themes. HR as information A strong majority of participants from both the short-term inlab and longer-term in-home studies consistently related HR to emotions (or stress). They tended to interpret both their own and others’ HR information accordingly. However, the ways in which participants interpreted HR and the effects it could have on them were influenced by several interrelated factors: availability of context (e.g., non-verbal cues), how relevant emotions were in the situation, and the type of relation they had with the person they perceived the HR from. In the rest of this section, we provide an overview of these recurring patterns in our interviews. First of all, we found differences among our participants in the extent of interpretations they thought were possible from heart rate. Some participants were optimistic, expecting that HR “actually shows what you’re thinking, what is going on in your mind” (Erik, Lab 2). Others had lower expectations of HR interpretation. For instance, Abby (Lab 2) thought that “you can’t draw direct conclusions from it, but it is a measure of not being completely settled”. Even if the participants were unsure of what information they could get out of HR, just the presence of the feedback made them think about possible interpretations. For instance, Debbie (Lab 2) said: “I think this is a bit distracting, because even if you can’t interpret much, you’ll start thinking about it, just because you have Although most participants thought HR could be interpreted to a certain extent, they all acknowledged that there is a need for additional cues to guide the interpretation. As Marianne (Lab 1) put it, “you also have to know what that person is doing because otherwise [you’d think] oh he’s nervous but he’s just walking up the stairs.” Similarly, the general understanding among our participants was that while you might try to interpret HR even without sufficient additional cues, you will not get it right. Mary (Lab 1) said: “without the context you cannot really interpret it, you can only misinterpret it” (emphasis Mary’s). Having additional cues to interpret HR remained important, even after two weeks of usage. For example, Annie (Home A) talked about calling her parents on Skype while sharing HR: “In face-to-face I have the impression that [the HR feedback] is natural and everything [is] clear, but here when I have the video and the HR, then the interpretation might not be correct, they might think something else, they might misunderstand me ... I would expect such kind of effects ... Not that much in face-to-face though.” Emotional relevance of situation Having enough context to make the interpretation of HR plausible did not necessarily mean that the participants felt that HR is adding any new or interesting information. Our participants generally expected that interpretation of HR feedback could provide additional information only in situations that are emotionally relevant for them or their partner. For example, Abby (Lab 2) says “[It was ok to show HR to a friend] it’s nothing to be embarrassed about, it’s just a heart rate”. Nonetheless, a few minutes later she continues to say that “but I guess that when you’re in the middle of a social interaction that is not completely casual to you, you wouldn’t want the other person to see”. In particular, participants often mentioned that HR feedback would be most effective in situations where there is something to conceal or, more generally, where one wants to setup an impression. Participants usually added that they would not like to share their HR during such moments; that it feels “too personal”. For example, Margit (Lab 2) was thinking about HR feedback during an interaction with her boss: “[I wouldn’t like this] because sometimes, you can look like you’re not that stressed and in fact you are ... [with this] you just can’t hide anything, you don’t have a choice”. Furthermore, participants remarked that they could not control the HR. For instance, Mary (Lab 1) said “I think that this is why its weird. . . because it’s your own but you cannot control it.” In contrast, participants were happy to share HR during less emotionally relevant situations. As Margit (Lab 2) talks about the lab session, “we didn’t have any really personal things to do, it was just the movie and this task ... I think it’s not enough personal to use this information”. Similarly, Annie (Home A), said that she probably wouldn’t mind having the HR feedback on the door of her office during casual work days, as “I think it wouldn’t be very interesting for other people because it would be somehow constant [. . . ] but if it were interesting, then I would mind.” poker face, but there is everything” and continued that “it something new, nice additional thing for the game.” HR as connection Effect of relationship type Besides the emotional relevance of the situation, expectations of how informative HR could be also depended on the relationship between the interacting individuals. In particular, participants were often much more concerned about sharing their HR with someone they did not know well. This was common to both in-lab as well as home participants. For example, when we asked John (Home B) during the final interview to tell us with whom he would not be comfortable sharing his HR, he said: “People who you don’t want to tell every, all ins and outs immediately ... who you want to keep distance sometimes ... Sometimes you don’t want to express everything you have on your heart ... like metaphorically on your heart [laughs].” Similarly, Annie (Home A), said “If they are a complete foreigner to me and they are commenting on my physiological state ... I wouldn’t like that.” Sharing of HR was understood as implying a great openness in what you are willing to share. As John (Home B) puts it when asked if it would change anything in his interaction with close family: “ Yeah, not so much ... maybe when my mum is on the telephone and she would hear that, she could get really concerned, then she wants to know everything ... I’m willing to tell her, it’s no issue, but it also implies more or less that when you share your HR, you have more to tell. Has implications for the further communication.” However, interviews from the in-lab studies as well as the home deployment indicate that participants thought that HR often does not bring new information when the other person is well known. In particular, most participants felt that in a long term relationship, HR sharing (while collocated) usually does not add any new information into what you already should be able to understand from other social cues. Stacy (Lab 1) puts it “Yeah when I’m together with my boyfriend I can see like this if he’s relaxed or not so I don’t really need the HR visualized I can see it in his movement and his eyes if he really is relaxed and I’m not sure if this tool really adds something to that.” Similarly, Mary (Lab 1) was even more strict: “Its like sending text messages when you’re in the same room ... why not just talk to each other.” In line with this, none of our home participants expressed any problems or issues with sharing their heartbeat with their partner in everyday situations. The only situations suggested by our home participants, where HR might reveal something new in a long-term relationship, were tied to emotionally relevant contexts, in which one would like to conceal his/her internal state, such as a fight or white lies. All in all, these findings suggest that for many situations in which HR could be used as information, people prefer not to share it. An exception were games or generally fun activities, i.e., emotionally low-risk contexts. An example was the poker game at Annie’s and Michael’s place, where the HR feedback played a major role in the experience. Michael (Home A) told us: “People were quite enthusiastic about it. When we were receiving the new cards, everyone was looking there, just to check what’s there ... because you know, you have the HR as a connection came up as an aspect of HR communication independently for three of our home deployment couples and two pairs from the in-lab studies. In contrast to findings about HR as information, HR as connection seems to be based on the feeling of being connected to the other just by the fact that the partner’s HR is available. Hence, this is independent of the additional information the other’s HR could provide. The three couples felt that the HR can provide a particular form of a direct connection to the other person. For instance, Michael (Home A) came up with the idea of a living picture frame, where you would have the photo or another link to the person, together with his/her HR. Relating to this, Michael said: “I see it moving and it is somehow connected ... you can see that something is happening there ... much more directly connected to the other person. Always a bit different, when you see it, you don’t know what the others are doing, but you know that they are doing something.” In response, Annie, his wife, remarked: “[laughs] ... like that I’m existing, that I’m still alive”. Similarly, Maddie (Home B) thought that such feedback could help her be closer to her partner. “I think if you have something very important like an interview or something, then the other can see” John, her husband agrees and continues “John: It’s like that you can feel what the other person feels .. Maddie: That you sort of together you go through something.” In the same vein, Adrian (Home D) remarked that “[HR] would help to create a bond with them. That I can imagine would work. If you are far away, and you can hear the other’s heartbeat, I can imagine that would work. However, it does not work when someone is sitting next to you already, because you have contact with them anyway. You have much less contact with someone overseas, so there it could be beneficial ... it would create the feeling of being much closer to the other.” All three couples noted that this HR as connection is most relevant in situations in which they are physically apart (although the distance between them does not need to be large). For example, John (Home B) said during the interim interview: “I like the fireplace sound when Maddie is doing something in the kitchen. . . so then the computer is here [living room] and I hear this firesound ... yeah, it’s a little bit like somebody is present, that’s nice.” He again mentioned in another interview as one of the highlights of the technology probe experience. “[And what I liked most was] when I heard Maddie in the kitchen. It’s like communicating in a different way [...] there is at least some link ... It would be nice if you had something [at your work] and I would have it [at my work].” This could be also one of the reasons why this effect was not mentioned often by our in-lab participants, who spend all of the time collocated. The participants made it clear that for this effect of connection to take place, less context is better. They mentioned two types of reasons for this. First, to get context about the current activities of the other, you need to be already connected; either by physically being together, or for example talking on the phone. In such cases, the HR would not add to the feeling of connection that was already existing. Second, less context also reduces privacy concerns as it would be more difficult, if not impossible, to interpret HR without context. Michael (Home A) says: “ [just] HR is less intrusive, in a sense. If you have video its like sort of surveillance ... HR I think is less... in some situations it can be, but you need to have some extra information ... if you just have HR it doesn’t tell you anything ... just that someone is moving, or is agitated somehow”. Nevertheless, Michael still continues to say that “it’s a kind of presence information ... Like someone is living there [...] you might feel more connected to the person.” In sum, these findings show that although HR might be problematic to use as information in practical applications, most of the couples from the home deployment agree that HR communication could have value by providing a feeling of connection. As suggested by our participants, this could be for instance implemented in a picture frame or other unobtrusive display that can be put in the background and does not take a lot of attention. Especially because the viewers of such displays would not know the context in which the HR is displayed, participants did not express privacy concerns for this way of using HR communication, as long as only close family members can see it. DISCUSSION To gain insights into possible mechanisms underlying the effects of sharing HR, we deployed a technology probe for sharing HR in two in-lab studies and in a two week home deployment. Interviews with participants in the in-lab and home deployment studies showed that HR can provide information about another and that HR can create a connection to another. In this section, we propose two mechanisms that could help in understanding our results. For HR as information, we suggest that the observed effects could be explained in terms of impression management. Similarly, we propose that the effects of HR as a connection can be understood based on the idea that people perceive the source of HR feedback as a representation of a part of their body, even if it is physically remote. Finally, we discuss the implications these mechanisms would have for future work. HR as information Our results show that participants tended to interpret HR in terms of emotional states. In that light, participants often considered HR as an objective signal about one’s state that cannot be voluntarily controlled. This corresponds with findings of studies that gave participants false feedback on their own HR [26, 33]. These studies suggest that people tend to reappraise situations when there is a mismatch between their understanding of the situation and a change in HR. The results showed several interrelated factors that can influence the extent to which HR is informative for the participants: Participants thought that having context (e.g., additional nonverbal cues) could help by guiding the interpretation of the HR. This shows that our participants acknowledged the multi-facetedness of HR, where it’s changes can have different meaning depending on the context [11, 17]. HR was also thought to be more informative in situations in which emotions were relevant to the interaction or to one of the interactors. Additionally, the type of relationship between the interactors also influences if HR can be informative. Yet, in most of the cases where people thought the HR could be informative, they were not willing to share it. We think that it can prove useful to explain these findings by the idea that HR sharing undermines impression management. Concept of impression management draws on an idea that social interaction can be viewed as a form of role playing, as postulated in one of the most influential works on impression management by Goffman [13]. In this role play, each individual tries to set up an impression (e.g., that of an eager, hardworking, funny, or serious individual) for those he or she interacts with. Goffman uses a theater-based metaphor to explain the interplay between a presenter as well as his or her audience. The audience is usually expected to play along, helping the individual to maintain the plausibility of the situation (i.e., the role he or she tries to convey). See, e.g., Consolvo et. al. [9] for examples of how impression management has been already utilised in other HCI contexts. As HR was perceived by our participants as both uncontrollable and related to emotions, it is plausible that participants could feel that their impression management was undermined if others could use their HR to reappraise the impression they try to present. We argue that this would explain why HR sharing felt more informative in situations where (hiding) emotions is relevant, as well as the changes in effects depending on the relationship type: In situations where one wants to appear differently from what one actually feels (e.g., confident during a presentation, while actually feeling nervous), HR sharing might undermine this impression management. Other examples of such situations where we often try to conceal or otherwise manage our real emotions are meeting with a boss, an early date, or an argument with your partner, all of which were specifically mentioned during the interviews as situations where HR could be informative. In contrast, our participants were generally fine with sharing the HR in everyday situations that were not emotionally relevant – recall, for example, Annie’s quote about having her HR on her office door. In these situations, emotions are not important for keeping up appearances. Seeing HR as undermining impression management could also explain the fact that the type of relationship participants had with the person they were sharing HR with could influence how comfortable they felt about sharing HR. For instance, participants tended to feel more comfortable sharing their HR with their romantic partners than with strangers. This can be explained by the observation that when relationships become closer, the amount of self disclosure increases [18, 34], lowering the need to keep up appearances. Furthermore, our interviews suggest that participants feel that sharing HR implies a willingness to disclose emotional information, which might not be congruent with an impression one wants to give when interacting with people one is less close with (e.g., [13]). In general, the described threat to impression management would also explain the consistent uncom- fortableness with sharing HR feedback in situations where it could add new information, as HR sharing in this case precludes our ability to manage how others view us. It is also possible to link the assumed HR effects on impression management to the reported increases in closeness. First, sharing HR can be seen as a form of emotional self-disclosure in situations where emotions are a part of impression management. Emotional self-disclosures have been repeatedly shown to increase feelings of connectedness (e.g., [18, 15]). This is in line with the idea that people believe that HR can reveal emotional information which you would normally choose not to show, and that it therefore undermines impression management. Second, psychological literature on intimacy (i.e., felt connectedness) builds on equilibrium theory, that states that people try to keep a certain intimacy equilibrium with their interaction partners [4]. Whenever the interaction becomes too intimate, people will start to feel uncomfortable and try to reduce the intimacy in the interaction [2]. Altogether, in situations in which sharing HR can be regarded as a form of emotional self-disclosure (i.e., those where emotions play role in impression management), it seems that sharing HR leads to increases in closeness. However, as people cannot control HR, the self-disclosure is not controllable and may lead to interaction that feels too intimate, thence, uncomfortable. Moreover, people keep different equilibria with different people, depending also on their relationship status [4]. This is in line with the observations that HR feedback did not feel uncomfortable for long-term couples as opposed to strangers. Finally, we found that there were some interactions in which people did not feel uncomfortable sharing HR although it could undermine impression management. These were situations of low personal risk. An example of such a situation is an informal poker game, where part of the fun is in managing a particular impression, a “poker face”. Therefore, adding HR to a poker game can give new challenges and interesting dimensions to the impression management in the game. Similar reasoning holds for example also for watching movies in a social setting, also mentioned by our participants. HR as a connection Beside the finding that HR can be informative, our findings also suggested that some participants thought HR could be valuable as it provides a direct connection to another. In particular, three out of five couples in the home study remarked that sharing HR is a way of creating a connection to someone in a remote location. It is important to separate this from HR as information, as there are some fundamental differences between them. For example, with HR as information, participants consistently said that there was a need for context in order for the HR to be informative. In contrast, with HR as connection, participants suggested that less context was better. Instead of being able to interpret the HR, participants remarked that it was more important that the HR was available and coming from someone they feel close to. This suggests that the results of HR as information and HR as connection are based on separate mechanisms. We suggest that the effects of HR as connection could be explained by the idea that participants saw the HR as a part of the other. For example, participants used statements like “someone is living there”, “she is present here with me”, “direct connection” to talk about the HR feedback in this sense. This suggests that having the HR feedback can feel as if a part of the other is (physically) closer. Such a physical connection can be compared to tokens of loved ones that we sometimes keep close to us, like pictures or jewelry. If enhanced with HR, the token might be more potent as it is not just a static representation of the other, but it seems to create a continuous, open connection with the other. Leveraging HR feedback as a continuous open connection to another could contribute to previous work in HCI on supporting intimacy for remote couples or close family (e.g., [30, 34, 21]). In such projects, feelings of closeness were often created through either explicit actions that express thinking of the other [30], or by facilitating the feeling of awareness [10]. Both processes can lead to presence in absence. In this vein, sharing HR can be also seen as a way of creating presence in absence, as it might facilitate awareness and trigger actions that express thinking of the other. As long there is no context, HR is difficult to interpret and participants are less likely to be concerned with privacy issues of sharing their HR with their close ones. Implications, caveats and future work We identified two core categories of effects appearing during HR sharing: ‘HR as information’ and ‘HR as connection’. Previous sections then suggested two possible mechanisms, namely HR as threat to impression management and understanding the source of HR as representation of the body, that could explain how and why sharing HR leads to the observed effects. We will now discuss implications of these mechanisms and identify opportunities for future work. Our results suggest that, due to the effects on impression management, HR sharing might not be a viable way to improve closeness in workplace related situations, or everyday encounters of strangers, friends and collocated couples. However, our results do not imply that all HR sharing must be uncomfortable or ineffective. On the contrary, when impression management is not an issue, HR as information could be useful. For instance, in self-reflective situations in which people are learning or training their social skills (e.g., marital therapy), it could be important to understand the differences between impressions one puts up and the underlying feelings one has. Therefore, HR feedback could have useful effects in such situations, helping people to reflect on what is happening “here and now”. More generally, results of this study could inform also other contexts currently relevant for HCI, where biometric data are collected and displayed, such as projects related to Ambient Assisted Living, hospital environment etc. Another possibility is to use HR as information based on composite signals (i.e., combining physiological signals from different people together), which was also suggested by some of our participants. For example, composite signal based on synchronization of HRs can, for instance, indicate situational empathy [19]. Using such indicators and reflecting them back to people might improve their social interactions [5]. Moreover, if such feedback methods use composite signals, they might not necessarily affect the impression management in the same way. As such, sharing composite feedback of HR might be a viable alternative to using individual signals that could be explored further. Additionally, it seems that HR feedback might support feelings of closeness for remote couples. Several of our participants from the longer-term in-home study found that it brought them closer to another, without undermining impression management as it was not interpreted. We suggested that the effect could be based on understanding HR feedback as a representation of the other person, directly connected to him/her. We expect that using haptics to represent the HR might increase closeness even more than visual or aural presentations, as it further reduces physical distance to the feedback source. This suggests that HR sharing could serve as a basis for future systems enhancing feelings of social connectedness for remote couples. More generally, while many studies in HCI focus on extracting higher-level concepts from physiological signals automatically by machines or researchers (e.g., [11, 23]), this work explores another approach for using physiological signals in HCI, where the focus is on what the people themselves make of the provided feedback, and applying it in a social context. The utility of such approach is also supported by, e.g., recent work of Sanches at al. [27] on personal HR feedback, suggesting its usefulness for self-reflection; as well as social psychological research, e.g., showing that false HR feedback influences one’s perception of a social situation [36]. We recognise that this exploratory study is limited in several ways. For instance, we tested only a particular form of feedback design. Other design decisions (e.g., tactile or light-based feedback devices) might yield different effects. This study uses a combination of a longer-term in-home study and short-term in-lab explorations with friends and couples, mainly in everyday situations. Although, our results show that the core observations appeared consistently in both these contexts, further research is desirable to verify and replicate our findings with users interacting in different cultural and situational contexts, as well as with different relationship types (e.g., colleagues or strangers). Finally, this work focused on reaching a qualitative understanding of how users understand and make sense of HR feedback, providing several hypotheses on the processes behind the observed effects of HR sharing. Further work is now needed to test these hypotheses in more quantitatively oriented studies. For HR as information effects, such studies could for example draw on the large body of psychology literature in the social dilemma paradigm. Future research could also investigate if our findings generalize to other bodily signals. For example, we suggest that the effects in the HR as information category are based on the participant’s expectation that HR is interpretable in terms of emotions. Therefore, we would expect that these findings generalize to any bodily stimulus that participants could interpret in term of emotions, e.g., skin conductance. CONCLUSIONS This work has explored the possibilities of heart rate (HR) communication, providing a first step towards understanding the mechanisms behind physiosocial interaction. We support the suggestion that HR communication can have strong effects on intimacy, complementing previous quantitative studies by testing outside the lab in an everyday context, and by using HR sharing within a social context (whereas prior studies focused on highly structured tasks in the lab, with a selffeedback scenario without social interaction). Our data suggests that HR effects similar to those found in the lab transfer also to everyday contexts and, more importantly, to inferences of other people’s HR. Moreover, we extend previous work by drawing out how the perceived value and effect of sharing HR changes in various various social situations and identify two core categories of effects arising from HR sharing in our data: i.e., HR as information and HR as connection. Finally, we propose two mechanisms that could help understanding the interplay between factors influencing the effects of HR sharing and draw out their implication for further work. This work also connects to the recent interest in CHI community on bio-feedback technologies, suggesting additional opportunities for applications based on physiological signals, applied in social situations. We believe that such physiosocial technologies form a promising possibilities for future communication technologies, and we hope that this work provides a first step towards unpacking the physiosocial design space. Acknowledgements We thank our participants for their time and engagment in the studies. Many thanks to Joyce Westerink, Wijnand IJsselsteijn, Aaron Houssian, and our reviewers, for helpful comments. The first author also thanks Philips for their support in conducting this research. REFERENCES 1. M. G. Ames, J. Go, J. J. Kaye, and M. Spasojevic. Making love in the network closet: the benefits and work of family videochat. In CSCW ’10, page 145. ACM Press, 2010. 2. P. A. Andersen, L. K. Guerrero, D. B. Buller, and P. F. Jorgensen. An empirical comparison of three theories of nonverbal immediacy exchange. Human Communication Research, 24:501–535, 1998. 3. M. Argyle. The psychology of happiness. Methuen, London, 1987. 4. M. Argyle and J. Dean. Eye-contact, distance, and affiliation. Sociometry, 28:289–304, 1965. 5. M. Balaam, G. Fitzpatrick, J. Good, and E. Harris. Enhancing interactional synchrony with an ambient display. In CHI ’11, pages 867–876. ACM Press, 2011. 6. E. Bradner and G. Mark. Why distance matters: effects on cooperation, persuasion and deception. In CSCW ’02, page 226. ACM Press, 2002. 7. B. Brown, A. S. Taylor, S. Izadi, A. Sellen, J. J. Kaye, and R. Eardley. Locating family values: a field trial of the whereabouts clock. In Ubicomp ’07, pages 354–371. Springer-Verlag, 2007. 8. J. T. Cacioppo, M. E. Hughes, L. J. Waite, L. C. Hawkley, and R. A. Thisted. Loneliness as a specific risk factor for depressive symptoms: Cross-sectional and longtitudinal analyses. Psychology and Aging, 21:140–151, 2006. 9. S. Consolvo, D. W. McDonald, and J. A. Landay. Theory-driven design strategies for technologies that support behavior change in everyday life. In CHI ’09, pages 405–414, New York, USA, 2009. ACM. 10. A. K. Dey and E. de Guzman. From awareness to connectedness: the design and deployment of presence displays. In CHI ’06, page 899. ACM Press, 2006. 11. S. H. Fairclough. Fundamentals of physiological computing. Interacting with Computers, 21(1-2):133–145, 2009. 12. K. Garnaes, O. Grünberger, J. Kjeldskov, and M. B. Skov. Designing technologies for presence-in-absence. Personal Ubiquitous Computing, 11:403–408, 2007. 13. E. Goffman. The presentation of self in everyday life. Anchor Books, New York, 1959. 14. H. Hutchinson, H. Hansen, N. Roussel, B. Eiderbäck, W. Mackay, B. Westerlund, B. B. Bederson, A. Druin, C. Plaisant, M. Beaudouin-Lafon, S. Conversy, and H. Evans. Technology probes: inspiring design for and with families. In CHI ’03, pages 17–24. ACM Press. 15. J. H. Janssen, J. N. Bailenson, W. A. IJsselsteijn, and J. H. D. M. Westerink. Intimate heartbeats: Opportunities for affective communication technology. IEEE Transactions on Affective Computing, 1:72–80, 2010. 16. P. Klasnja, S. Consolvo, and W. Pratt. How to evaluate technologies for health behavior change in HCI research. In CHI ’11, pages 3063—-3072. ACM, 2011. 23. R. Mandryk, K. Inkpen, and T. Calvert. Using psychophysiological techniques to measure user experience with entertainment technologies. Behaviour & Information Technology, 25(2):141–158, 2006. 24. P. Markopoulos, B. De Ruyter, and W. Mackay. Awareness systems: Advances in theory, methodology, and design. Human-Computer Interaction Series. Springer, 2009. 25. H. Murakami and H. Ohira. Influence of attention manipulation on emotion and autonomic responses. Perceptual and Motor Skills, 105(1):299–308, 2007. 26. B. Parkinson. Emotional effects of false autonomic feedback. Psychological Bulletin, 98:471–494, 1985. 27. P. Sanches, K. Höök, E. Vaara, C. Weymann, M. Bylund, P. Ferreira, N. Peira, and M. Sjölinder. Mind the body!: designing a mobile stress management application encouraging personal reflection. In DIS ’10, pages 47–56. ACM Press, 2010. 28. T. Scherini, P. Melo, T. van Craenendonck, W. Zou, and M. Kaptein. Enhancing the sleeping quality of partners living apart. In DIS ’10, page 171. ACM Press, 2010. 29. A. Steptoe, N. Owen, S. R. Kunz-Ebrecht, and L. Brydon. Loneliness and neuroendocrine, cardiovascular, and inflammatory stress response in middle-aged men and women. Psychoneuroendocrinology, 29:593–611, 2004. 30. R. Strong and W. Gaver. Feather, Scent & Shaker: Supporting Simple Intimacy in Videos. In Extended Proceedings of CSCW’96, pages 29–30, 1996. 31. H. Tsujita, K. Tsukada, and S. Itiro. InPhase: evaluation of a communication system focused on ”happy coincidences” of daily behaviors. In CHI ’10, page 2481. ACM Press, 2010. 17. S. D. Kreibig. Autonomic nervous system activity in emotion: A review. Biological Psychology, 84:394–421, 2010. 32. G. E. Vaillant. Aging Well: Surprising Guideposts to a Happier Life from the Landmark Harvard Study of Adult Development. Little, Brown and Company, New York, 2003. 18. J. P. Laurenceau, L. M. Rivera, A. R. Schaffer, and P. R. Pietromonaco. Intimacy as an interpersonal process: Current status and future directions. In D. Mashek and A. Aron, editors, Handbook of closeness and intimacy, pages 61–78. Lawrence Erlbaum, Mahwah, NJ, 2004. 33. S. Valins. Cognitive effects of false heart-rate feedback. Journal of Personality and Social Psychology, 4:400–408, 1966. 19. R. W. Levenson and A. M. Ruef. Empathy: A physiological substrate. Journal of Personality and Social Psychology, 63(2):234–246, 1992. 20. G. Lotan and C. Croft. imPulse. In CHI ’07 Extended Abstracts CHI ’07, pages 1983–1988. ACM Press. 21. D. Lottridge, N. Masson, and W. Mackay. Sharing empty moments: design for remote couples. In CHI ’09, page 2329. ACM Press, 2009. 22. J. J. Lynch. The broken heart: The medical consequences of loneliness. Basic Books, New York, 1979. 34. F. Vetere, M. R. Gibbs, J. Kjeldskov, S. Howard, F. F. Mueller, S. Pedell, K. Mecoles, and M. Bunyan. Mediating intimacy: designing technologies to support strong-tie relationships. In CHI ’05, pages 471–480. ACM Press, 2005. 35. J. Werner, R. Wettach, and E. Hornecker. United-pulse: feeling your partner’s pulse. In MobileHCI ’08, pages 535–538. ACM Press, 2008. 36. J. Wild, D. M. Clark, A. Ehlers, and F. McManus. Perception of arousal in social anxiety: effects of false feedback during a social interaction. Journal of behavior therapy and experimental psychiatry, 39(2):102–16, 2008.