Julian Ramos
← All projects Attention · Notifications

Interruptibility Detection

Recognizing opportune moments for notifications from smartphone activity and transitions across devices.

When is a good moment to interrupt?

Notifications can arrive in the middle of an activity, forcing people to switch attention at an inconvenient moment. The challenge is to recognize a natural breakpoint from the signals a phone or watch can observe, rather than relying on a fixed delivery schedule.

We developed Attelia to detect breakpoints in smartphone interaction and defer notifications until those moments. Attelia II extended this approach to transitions in physical activity and signals shared across devices, so opportunities could be detected both during device use and while a device was carried or worn.

Attelia II notification timeline showing breakpoints during device use and transitions between physical activities
We developed a machine learning model to detect breakpoints from smartphone interaction events, including taps, scrolling, text changes, and app/window transitions. Attelia II combines this with physical-activity transition detection to time notifications.

Combining signals across devices

A breakpoint on one device may not tell the whole story: someone can stop using their phone while continuing another activity. Attelia II therefore combines UI-based breakpoint detection with physical-activity transitions, such as stopping a walk.

Phones and watches share their detected breakpoints and whether they are actively being used. A combination layer brings these signals together to decide when to release a deferred notification, extending detection beyond a single device or type of activity.

Attelia II layered model: physical-activity and UI-based breakpoint detectors share results across devices, feeding a combined breakpoint decision
Physical-activity and UI-based detectors share breakpoint signals across devices. A combination layer uses them to make the final notification-timing decision. Attelia II, Figure 4.

Evaluating the experience

The related studies examined cognitive load and perceived mental effort around notifications. They connect sensing and system design with evaluation of the resulting user experience.

This line of research also connects to receptivity in mobile health: whether someone is available to receive, understand, and act on a recommendation.

Full list →