Invited Talk @ Human-Automated Vehicle Interaction Design (002673), Hongik University

The talk was titled "Reading the user's mind with an eye tracker," and covered pupil tracking together with cases from my own user studies.

On using the eye as an input device: what separates a glance meant as a selection from an ordinary one — dwell-based selection, gaze path selection, and how head angle shifts the field of view. On reading the eye as an output device: fixation and saccade, areas of interest and heat maps, and the three families of measure — position, numerosity, latency. The history ran from the nineteenth-century devices that strapped a drum to the eyelid, through corneal reflection and Dual Purkinje Imaging, to today's P-CR.

The last section is what you only learn after running the studies: resolution and sample rate must be at least twice the quantity you intend to measure, participants need screening in advance, lighting and electromagnetic interference corrupt readings, and there is a threshold that decides whether gaze belongs in degrees or in millimetres. The evidence came from Winklick, Secrets of Gosu, ICUC, and Lattice Menu.