MyTraces

Abstract

Most of the existing work concerning the analysis of emotional states and mobile phone interaction has been based on correlation analysis. In this paper, for the first time, we carry out a causality study to investigate the causal links between users’ emotional states and their interaction with mobile phones, which could provide valuable information to practitioners and researchers. The analysis is based on a dataset collected in-the-wild. We recorded 5,118 mood reports from 28 users over a period of 20 days. Our results show that users’ emotions have a causal impact on different aspects of mobile phone interaction. On the other hand, we can observe a causal impact of the use of specific applications, reflecting the external users’ context, such as socializing and traveling, on happiness and stress level. This study has profound implications for the design of interactive mobile systems since it identifies the dimensions that have causal effects on users’ interaction with mobile phones and vice versa. These findings might lead to the design of more effective computing systems and services that rely on the analysis of the emotional state of users, for example for marketing and digital health applications.

Wearable-Based Affect Recognition—A Review

“…According to Schimmack and Reisenzein [50], this split is justified by the observation that only the energetic arousal component is influenced by the sleep-wake cycle. Considering the wearable affect and stress recognition literature, a recent study conducted by Mehrotra et al [51] uses this three-dimensional emotion model (valence, tense arousal and energetic arousal) to investigate correlation and causation between emotional states and cell phone interaction.…”

Fusing Mobile Phone Sensing and Brain Imaging to Assess Depression in College Students

“…Depression has been linked to a variety of metrics available from smartphone sensing applications including amount of stationary time, GPS patterns, phone usage and conversation patterns, among others (Burns et al, 2011; Canzian and Musolesi, 2015; Saeb et al, 2015; Mehrotra et al, 2017; Wang et al, 2018). The higher amplitude circadian rhythms as measured by accelerometer are associated with reduced chances of major depressive disorder and other negative mental health outcomes (Lyall et al, 2018).…”

“…and depression, with further accuracy in prediction when adding context, such as if the individual is alone, with other people (particularly friends) or current physical exertion status (Burns et al, 2011). Self-reported happiness has been linked to decreased phone usage in the subsequent hour (Mehrotra et al, 2017). While several groups have started to characterize traits linked to depression, phone usage and circadian rhythms are the ones that are most prominent in the current literature.…”

Modeling interaction as a complex system

“…Untangling correlation and causation is a topic that has recently received increased attention across multiple disciplines, including HCI. To address it, researchers have taken various experimental and observational approaches, such as Tsapeli et al [39] and Mehrotra et al [26] in their analysis of the relationship between smartphone interaction and the emotional state of the user. Their findings indicate that the emotions of the smartphone user have a causal impact on different aspects of smartphone interaction.…”

“…Their findings indicate that the emotions of the smartphone user have a causal impact on different aspects of smartphone interaction. In their work, Mehrotra et al [26] first perform a correlation analysis to determine which variables have a significant relationship. For example, one of the tested correlations is the user's self-reported stress level and number of received notifications.…”