Attelia: Reducing User’s Cognitive Load due to Interruptive Notifications on Smart Phones Tadashi Okoshi * , Julian Ramos ‡ , Hiroki Nozaki † , Jin Nakazawa † , Anind K. Dey ‡ and Hideyuki Tokuda † * Graduate School of Media and Governance, Keio University slash@ht.sfc.keio.ac.jp † Faculty of Environment and Information Studies, Keio University chacha@ht.sfc.keio.ac.jp, jin@ht.sfc.keio.ac.jp, hxt@ht.sfc.keio.ac.jp ‡ Human Computer Interaction Institute, Carnegie Mellon University ingenia@andrew.cmu.edu, anind@cs.cmu.edu Abstract—In today’s ubiquitous computing environment where the number of devices, applications and web services are ever increasing, human attention is the new bottleneck in computing. To minimize user cognitive load, we propose Attelia, a novel middleware that identifies breakpoints in user interaction and delivers notifications at these moments. Attelia works in real- time and uses only the mobile devices that users naturally use and wear, without any modifications to applications, and without any dedicated psycho-physiological sensors. Our evaluation proved the effectiveness of Attelia. A controlled user study showed that notifications at detected breakpoint timing resulted in 46% lower cognitive load compared to randomly-timed notifications. Furthermore, our “in-the-wild” user study with 30 participants for 16 days further validated Attelia’s value, with a 33% decrease in cognitive load compared to randomly-timed notifications. I. I NTRODUCTION The amount of information available for consumption has grown by orders of magnitude while the amount of attention users have has remained constant. This has, in part, driven users to multi-task more and use notifications on their comput- ing devices, often resulting in greater numbers of interruptions. The number of networked computing devices belonging to users, along with those embedded in the environment, such as home, office, or urban space, have also been increasing quickly. Users own, carry and interact with (even simultaneously) [1], an increasing number of mobile networked devices [2]. On each device, the number of applications, services, and commu- nication channels is increasing as well, being driven by both technological progress and market trends, such as maturing Web middleware and flexible cloud platform services for rapid service deployment, or global application markets such as the “AppStore”. Moreover, the advent of social networking services in addition to conventional communication channels such as email and SMS increases the number of people that users communicate with daily. Given this background of information overload, the limited resource of human attention is the new bottleneck [3] in computing. In this paper, we focus on interruption overload, a form of distraction caused by the excessive number and in- appropriate delivery of notifications from computing systems. Typical notification systems deliver notifications immediately after they are available, and this has been shown to negatively affect users’ work productivity [4], [5], [6], [7]. One possible solution is to defer notifications until the user’s natural break- point [8], the boundary between two adjacent units of users activity, which can lower the impact on users cognitive load caused by the interruption. In this paper, we particularly focus on user’s “mobile experience” on the phone while they are actively using their own devices, and demonstrate our ability to detect breakpoints, towards the realization of user-attention-aware adaptive notifi- cations. Our system, Attelia, (1) works on smartphone devices, (2) is applicable to situations of user mobility and use of a wide variety of applications, (3) performs real-time detection of breakpoints to support real-time adaptation, and (4) does not require the use of dedicated external psycho-physiological sensors. A controlled user study with 37 participants showed that providing notifications at detected breakpoints resulted in 46% lower cognitive load, compared with conventional notifications presented to users at “random” times, for users who showed higher sensitivity to notification timings. Furthermore, our “in- the-wild” user study for 16 days with 30 participants showed similar results. Providing notifications at detected breakpoints resulted in 33% lower cognitive load for the users who showed higher sensitivity in their subjective evaluations. Also, users’ response time to the notification was faster by 13% than using conventional (or random) notification timings. The contribution of this paper is two-fold. First, we present the design and implementation of our novel middleware for real-time breakpoint detection on smartphones that does not require the use of dedicated external psycho-physiological sen- sors. Second, we present the results from both a controlled and an “in-the-wild” field user study that resulted in significantly lower cognitive load when the breakpoint detection system was used to defer notifications. In the remainder of this paper, we describe the interruption overload problem caused by notifications from computing sys- tems in Section II. Next we define the requirements for adap- tive notification scheduling on smartphones after we introduce recent trends in notifications in Section III. We then present our design approach for Attelia in Section IV, and describe the Attelia system architecture in Section V. Section VI describes our controlled user study with 37 participants and Section VII reports on a follow-up field study for 16 days with 30 users. In Section VIII we discuss further research opportunities based on the findings from our user studies. Section IX describes