Sensor-Mediated Empathy: A Mixed Methods Investigation of Social Biosensing

This dissertation investigates social biosensing: the sharing of physiological information between people, particularly through the lens of empathy and related interpersonal psychological processes. Self-tracking is the the collection and sharing of physiological information about oneself and has been a focus in academic research, but a next step in the use of this data, particularly in an era of communication at a distance, is sharing it with others. Research prototypes and already in-the-wild devices and applications designed for social biosensing are indicative of the shift in attention beyond the individual. Using a mixed quantitative and qualitative research methodology, this dissertation describes three research studies which build on one another to fill in gaps present in the current understanding of the underlying effects of social biosensing. This work seeks to provide answers to questions including: What is the measurable effect of social biosensing on one's accuracy in predicting another's internal states, what impact does social biosensing have on measures of closeness and humanization compared to other information types, and how do people form meanings of unfamiliar biosensory information in a social world. Chapters 1 and 2 cover the introduction and background to this emerging area of research as well as the mixed methods used in the covered studies. The subsequent three chapters detail results from original research, Studies 1, 2, and 3, respectively, conducted in three primary areas related to social biosensing: empathic accuracy, empathic precursors and interpersonal perceptions, and finally intersubjective meaning-making. Studies 1 and 2 report on the first ever experimental studies quantitatively assessing valence and arousal empathic accuracy with social biosensing. Study 3 uniquely implements the concept of intersubjectivity in an analysis of physically separated friends' understanding of social biosensing. Chapter 6 concludes with a broader discussion of these findings across studies, opportunities for continued research, and implications for theorists, researchers, and developers of social biosensing systems.

In aggregate, the results of this work suggest different strengths for biosensory information (in the form of electrodermal activity or EDA) compared to descriptive narration information for observers interpretations of a target's inner states. Narration information resulted in observers more accurately inferring the target's valence extremes - for positive events in Study 1, and negative events in Study 2, perhaps because it is closest to forms of typical communication through speech and text. However, the observers who received narration information did not perform well relative to observers in other conditions when inferring a target's arousal in Study 2, actually performing significantly worse than observers who received no additional information when inferring the target's relaxed events. Observers who received EDA information on the other hand significantly outperformed narration and no information observers when interpreting the target's most relaxed events, suggesting that EDA information, a physiological signal with ties to emotional arousal, may help observers infer this other dimension of emotion in a target person. Qualitative findings were more abstract than these specific quantitative metrics, coalescing instead around EDA providing a sense of the target’s existence for observers, and a reminder of the target’s unique perspective as different from their own. Regardless of accuracy outcomes, the mere presence of the target's EDA as a moving graph seemed to be enough to indicate the target’s existence and experience. Findings from Study 3, which further investigated the meaning-making process around EDA, indicated the relevance of both ecological and radical intersubjectivity approaches such that meanings can be formed through projecting one's understanding of one's own data onto another's, and also through dialogue about this new data type with close others. Furthermore, a set of temporal typologies emerged from analyses: "broadeners" who expanded their initially limited range of emotional sources of EDA, "stablers" who began and kept more abstract definitions from the start, and "puzzlers" who experienced growing uncertainty and skepticism regarding the meaning of EDA.

This dissertation demonstrates the utility of EDA as well as significant possible risks when used in social settings. EDA can serve the important function of conveying awareness and closeness with others and even communicate states like relaxation that are not typically served by language. However, flawed assumptions regarding EDA's scientific authority and direct ties to emotions can lead to misguided conclusions about the subjective states of others or even of oneself. Data about our bodies is simultaneously intimate and foreign, and depending on how it's wielded can serve to nourish and strengthen social connection in the face of widening technological isolation and division, or contribute to the harms, misinterpretations, and miscommunications already rampant in an increasingly mediated world.