Anis Idrizovic¹, Ethan Hundley¹, Ruei-Jr Wu¹, Jannick P. Rolland¹, Michele Rucci¹; ¹University of Rochester

Precise measurement of eye movements is essential for understanding the mechanisms that govern visual perception. Recent work has shown that fine oculomotor behavior – the small fixational movements occurring between saccades – actively structures the spatiotemporal input to the retina, profoundly influencing spatial sensitivity, visual acuity, fine pattern vision, and even stereopsis. These findings emphasize the need for eye-trackers capable of resolving eye motion with arcminute precision. At the same time, many natural tasks require accurate tracking over tens of degrees of visual angle. The Dual Purkinje Image (DPI) method introduced by Cornsweet and Crane (1973) achieves arcminute-level precision by measuring the relative motion of reflections from the cornea (P1) and posterior lens surface (P4). Building on this principle and recent advances in digital imaging, we recently developed a digital DPI (dDPI) eye-tracker (Wu et al., J Vis, 2023) that detects sub-arcminute eye rotations within the central 15° of the visual field. To accommodate diverse experimental demands, we have now generalized this approach into a modular architecture that can be tuned to the desired tracking span. Here we present an implementation with two laterally offset illuminators and optimized magnification, extending the high precision afforded by the dDPI across a 40° range. The design supports two operating regimes: a central region where two P1 and two P4 reflections are visible and both Purkinje pairs contribute to tracking, and peripheral regions where two P1s and a single P4 support robust single-pair tracking following the classical DPI principle. We validated the system with an artificial eye mounted on a high-precision rotational stage and with human observers, confirming both the underlying optical principles and operational performance. The expanded tracking span enables oculomotor measurements with arcminute resolution across a broad eccentricity range, supporting reliable tracking of eye movements over large amplitudes and speeds.

Acknowledgements: Supported by NIH EY18363 and P30 EY001319.