Evalie Rehor¹ (), Jorge Otero-Millan¹; ¹UC Berkeley

The eyes are never still, even during attempted fixation. Fixational eye movements (microsaccades, drift, and tremor) are typically studied during sustained fixation at central locations. While microsaccades have been characterized during natural viewing, fixational drift during free viewing remains comparatively unexplored. To investigate how fixational drift differs between sustained fixation and free viewing, we recorded binocular eye movements from head-fixed participants under both conditions (n = 5). Fixation targets were presented across a range of eccentricities matching those reached during free viewing. Eye position was measured using dual Purkinje imaging (DPI) eye tracking within the OpenIris framework. To quantify the systematic components of drift, we fit a linear regression model of drift velocity onto eye position. The slope of the regression measures the position-dependent drift velocity bias (arcmin/s per degree of eccentricity), and the intercept of the regression measures the overall position-independent drift velocity bias. Across subjects, drift velocity exhibited a significant position dependence in the vertical direction in both conditions (fixation: -0.34 ± 0.07 arcmin/s/deg; free viewing: -0.71 ± 0.06 arcmin/s/deg), and in the horizontal direction during free viewing (-.69 ± 0.18 arcmin/s/deg). Negative slope indicates that drift velocity goes opposite the current position. Vertical drift intercepts were significantly negative (fixation: -16.4 ± 4.7 arcmin/s; free viewing: -21.5 ± 4.6 arcmin/s). When comparing conditions, vertical drift velocity dependence on eye position was stronger during free viewing (p = .02), and vertical intercepts were more negative during free viewing (p = .03). Horizontal slopes and intercepts showed no difference across conditions. These findings show that drift exhibits systematic velocity biases during both sustained fixation and free viewing, with a stronger vertical bias during free viewing.

Acknowledgements: Grant Support: NIH Grant T32-EY007043. Thank you to Ravi Srinivasan for assistance in building the DPI eye tracker.