Overview
(This project was done in collaboration with Matthew Kersey, Jacob Alexander, and Rico Pesce)
To investigate if bio-signals such as heartrate could be used to replace other information lost in teleconferences, we designed PulseChat, a color-coded display of heartrate that is displayed alongside a teleconference environment. We designed a study to answer if having heartrates displayed would result in less workplace environments, with the reasoning that seeing someone else's heartrate would increase empathy and hence reduce confrontation between. Our investigation revealed a surprising result: we did not find a reduction in confrontation, but rather we found that seeing heartrates decreased the impression of likability and competence in the other person, despite many participants expressing that they felt heartrate was independent of those characteristics.
Key Contributions
- Designed and implemented front-end of Heart-rate display in React.
- Was primary designer of the study structure and wrote the script that the study facilitator reads out.
- Performed and participated in study, which consisted of 12 participants.
- Collected quotes from study, and performed thematic analysis on them.
- Participated in writeup and video presentation of project.
Design Process
With the sudden post-pandemic development of ubiquitous teleconferencing in the workplace on our minds, we began our project with the goal of combating the interpersonal information loss found in teleconferencing. Specifically, we wanted to recover lost emotional cues between teleconferencing participants. In an attempt to pinpoint a measurable and specific emotional cue to recover, we formulated the question: Will users be less confrontational if they have access to others’ heart rates in a workplace environment? With our question, we assume that users may miss body language cues such as rigid posture, hand wringing, or foot/leg movements that could imply anxiety, and we replace those cues with a new one, heart rate. We believe that heart rate has a direct correlation to perceived anxiety (i.e. users will associate a high heart rate with high anxiety and a low heart rate with low anxiety). Results from our post interview survey reinforce this belief.
Hence, we designed PulseChat, a device that measures a person’s pulse and displays the user’s heart beats per minute with a color coded system. The pulse data is then sent to a server implemented in Go and hosted on Amazon Web Services. The server then forwards the data to a web front-end implemented using React. The front-end consists of a number representing the heart rate being displayed at the center of the page with a colored background. If the heart rate is close to the person’s resting heart rate, then the screen background displays a green color. When the person’s heart-rate rises beyond normal range, the background color displays a yellowish orange color.
The PulseChat display next to a teleconferencing window.
We performed a Wizard of Oz style study with fake heart rates, which we enhanced after a pilot study, using our working prototype to see if the system had an effect on the user’s actions and impression of the person whose heart rate is being displayed. Our study also had an additional goal to get a general idea of how the system is perceived. Our study involved twelve participants, 7 of whom were female and 5 of whom were male. The ages of the participants ranged from 21 to 65, with a median age of 25. We recruited from the team’s social circle, but during the study we made sure that the study facilitator and the actor were unknown to the participants in order to avoid confounding variables. The participants were split into 3 groups of 4 people. One group acted as the control group, and conducted the interview without knowing the existence of PulseChat. The second group was shown the PulseChat front-end with the heart rate hovering around 60 beats per minute. The third group was shown the PulseChat frontend with the heart-rate varying in level. For each participant in this group, 6 out of 10 questions were designated as elevated heart rate and 4 were designated as non-elevated (resting) heart rate. Questions were divided into five contiguous pairs, each of which was randomly assigned a designation. This was done to imitate real heart rates, which would not be able to change quickly enough to switch on each question. Which questions received corresponding elevated heartrates was also counterbalanced for each participants to reduce confounding variables.
Each participant was asked to interview another participant (the other participant was actually a confederate). The scenario the participants were given was as follows, “You work for a large company and your boss has tasked you with putting together a team of employees from other parts of the company to work on a very important project. You will be interviewing potential team members and deciding whether or not you would like them to join your team.” The participants then went on to ask the confederate 20 questions, which received canned responses from the confederate. Video was turned off to reduce bias. The questions were structured so that every other question was an introductory question and then there were two possible follow up questions that the participants had to choose from. Each of the pairs of follow up questions included one more confrontational question and one less confrontational question (the perceived level of confrontation of the question pairs was verified with a survey).
The list of questions asked by study participants to our study confederate. Under each numbered "main" question, are two follow up questions that participants choose to ask, with one question being more "confrontational" than the other.
The study script that the study facilitator reads to study participants.
After the interviews, the participants were walked through a post-interview survey by the interview facilitator in order to ensure understanding of all questions and to receive real-time feedback on their perceptions of the interviewee and (when applicable) the PulseChat system.
An excerpt of the Post-Interview survey given to study participants.
The results did not align with what our team expected. Initially, we suspected that participants who saw the changing heart rate would empathize with the confederate, ask them gentler questions, and be more charitable with how they viewed them afterwards. Our results contradicted this, despite some participants stating that changing heart rates does not imply that the other person is less competent or likable. Our findings have a few implications for future usage of heart rate information in a workplace context. First, people were unlikely to be comfortable with sharing heart rate data in a workplace setting. Secondly, heart rate data may not be very useful in a workplace setting due to people needing to appear competent by pretending to be calm even in high stress situations. Furthermore, it is not always clear how people should respond to seeing someone having a high heart rate. These reasons make implementation of heart rate sharing systems in a workplace setting unlikely in its current form.
Full writeup of this project is available here.