Impact of Emotion Analytics on Emotion Perception & Memory
An increasing number of systems promise to provide ’emotion analytics’ about how one feels. We explored algorithmic authority, people’s tendency to trust machine-based interpretations, in the context of these ‘emotion sensing’ systems. To understand how machine feedback influences emotional self-judgements, we experimentally compared three system framings: Positive (‘alert and engaged’), Negative (‘stressed’), and Control (no framing) in a mixed-methods study with 64 participants. Despite participants reporting active strategies to understand their internal states and test feedback, perceived emotions were significantly influenced by framings. Participants sometimes overrode their own assessments, believing the system had access to privileged or more accurate information about their emotions. Experimental evidence suggests that these effects are negatively biased; participants reported greater anxiety for the negative algorithm framing compared to benefits gained from positive framing. We explore design implications for personal informatics including risks of trusting systems that seemingly ‘unlock’ hidden aspects of the self, versus opportunities for emotion-monitoring systems to utilize these framing effects.
This project is still in progress. Please email if you are interested in collaborating.
EmotiCal (Emotion Calendaring) – A Prospective Tool for Emotional Well-Being
While current personal informatics systems provide rich records of our pasts, they typically do not convert these records into actionable future plans. To address this, we designed EmotiCal, an emotion forecasting system, to improve users’ everyday moods and promote well-being. Motivated by initial surveys and user interviews, EmotiCal analyzes logs of users’ past moods and mood triggers to generate a 2-day forecast predicting users’ potential future moods. EmotiCal encourages users to change these predictions by recommending enjoyable activities that are personally-tailored to the user and connect to higher mood ratings. A three-week field study with 60 participants evaluated the effectiveness of EmotiCal against a mood-monitoring only intervention and do-nothing control. Activity recommendations and visualizations for future mood, as prompted by EmotiCal, improved logfile mood ratings, pre-post changes in self-awareness and reported frequency and success of engaging activities to improve mood. However, we saw no significant difference between intervention conditions for changes in PANAS score. This paper presents design implications for both emotion regulation systems and general tools for actionable personal informatics to support behavior change.
A recent set of interviews with 23 EmotiCal users is looking at how users determine trust in system vs. self judgements about their mood.
Change of Heart – Emotional Reflection to Promote Behavior Change
Preventable behaviors contribute to many life threatening health problems. Behavior change systems have been deployed to modify these, but such systems typically draw on traditional behavioral theories that overlook affect. This project examines the importance of emotional reflection for behavior change, and useful strategies to harness its benefits.
Here is an abstract of a Change of Heart study (from CHI’15).
Echo – A Reflective Tool for Emotional Well-being
We have developed a personal microblogging Android and iPhone app called Echo that allows you to record and reflect on past experiences. We call this practice, Technology Mediated Reflection (TMR). We are interested in how TMR changes people’s lives and influences well-being. Our research explores the intersection of human memory and emotion. If you are interested in learning more and possibly participating, please click here for our Echo website. Also, here is a recent sidebar about our work in a special issue of the International Journal of Design (featured on page 12).
You can view a presentation of my (dusty 2011) findings from study 1 here.
Here is an abstract of an Echo study (from CHI’13).