Link

Daniel Chen Queen's University, Kingston, Ontario, Canada
Roel VertegaalQueen's University, Kingston, Ontario, Canada

Conference on Human Factors in Computing Systems archive
Extended abstracts of the 2004 conference on Human factors and computing systems table of contents
Vienna, Austria
SESSION: Late breaking result papers table of contents
Pages: 1513 - 1516
Year of Publication: 2004
ISBN:1-58113-703-6

Abstract:
Today's user is surrounded by mobile appliances that continuously disrupt activities through instant message, email and phone call notifications. In this paper, we present a system that regulates notifications by such devices dynamically on the basis of direct measures of the user's mental load. We discuss a prototype Physiologically Attentive User Interface (PAUI) that measures mental load using Heart Rate Variability (HRV) signals, and motor activity using electroencephalogram (EEG) analysis. The PAUI uses this information to distinguish between 4 attentional states of the user: at rest, moving, thinking and busy. We discuss an example PAUI application in the automated regulation of notifications in a mobile cell phone appliance.

My Discussion:
Short as this late-breaing results paper is, it carries in it a substantial hurdle to taking it seriously. While the work done to model the state of attention of the user based on ECG (heartbeat) and EEG (brainwave) measurements reads as sound, the idea that we need to stick electrodes to our bodies just to be able to handle our disruptive mobile phones and computers just seems too much. However, contemplating that that this measuring equipment could be shrunk to a wearable state like a watch or jewelry, it is good that the groundwork is being done to decide whether the effort to do this shrinking and productization is actually ever going to be worth it. There are fields of work that are so vital yet interrupt-driven (stockbrokers, nuclear-plant crisis managers) that the workers could be persuaded to wear some monitoring equipment if it would mean their tools would be less disruptive and more effective at giving them the kind of information they need when they need it, and at no other time. On that basis, the background and pointers for further study given in this paper are valuable, the idea that ignoring an interrupt is an input about attention itself is accurate, and the initial data they have about the accuracy of the modelling, even if based on a tiny sample, is heartening.