Abstract
Humans exhibit surprisingly fine control of their eye movements. Numerous recent studies, drawing from a range of visual tasks, have revealed oculomotor precision far superior to what previously assumed (e.g., Intoy & Rucci, Nature Communications, 2020). Because of technical challenges, reliable high-precision measurements have traditionally required strict head immobilization. During natural viewing, however, eye movements co-occur with head movements and both contribute to shape the luminance stream entering the eyes. The need, therefore, emerges for extending high-resolution oculomotor measurements to normal head-free viewing. A well-known exception to the head-immobilization requirement is provided by the seminal work by Steinman and colleagues, who developed a coil-based approach (Revolving Field Monitor; Steinman, 1995) to precisely record eye movements while the head is free to move normally. Building on this pioneering work, here we present a system that integrates a custom scleral coil eye-tracker with a motion capture system to simultaneously track eyes, head, and hand movements with high-precision. Eye-tracking is achieved by means of three, orthogonal, oscillating, magnetic fields, which are continuously tuned to maintain high uniformity in the central 1 m3 region. Voltages induced by the field are collected by coils embedded in silicon annuli placed on the observer’s sclera and on a tightly fitting helmet. A passive motion-capture system, selected to minimally interfere with the magnetic fields, tracks head translations, hand movements, and all rigid objects involved in experiments. The integration of these two systems enables precise localization of the line of sight in the scene and reconstruction of the visual input to the retina. We report measurements of head and eye movements with arc-minute resolution in a variety of everyday tasks such as needle threading, sorting small objects, and reading fine print. Our results confirm a remarkably fine degree of oculomotor control as the observer interacts normally with the scene.