Abstract
It has been suggested that humans rely on an internal representation of earth gravity (9.81m/s) for a series of tasks such as catching. Furthermore, eye-movements seem to be partially guided by predictions about observed motion. Thus, the question arises whether knowledge about gravity is also used to guide eye-movements: If humans rely on a representation of earth gravity for the control of eye movements, earth-gravity-congruent motion should lead to improved visual pursuit. In a pre-registered experiment, we presented participants (n=6) with parabolic motion governed by six different gravities (−1/0.7/0.85/1/1.15/1.3g), two initial vertical velocities and two initial horizontal velocities in an immersive 3D environment. Participants were instructed to follow the target with their eyes. We tracked their gaze and computed the visual gain (velocity of the eyes divided by velocity of the target) as proxy for the quality of pursuit. An LMM analysis with gravity condition as fixed effect that allowed intercepts to vary per subject showed that the gain was lower for −1g than for 1g (by −0.134, SE = 0.068). This model was significantly better than a null model without gravity as fixed effect (p< 0.001), supporting our hypothesis. A comparison of 1g and the remaining gravity conditions revealed that 1.15g (by −0.028, SE=0.009) and 1.3g (by −0.039, SE=0.009) were associated with lower gains, while 0.7g (by 0.016, SE=0.009) and 0.85g (by 0.008, SE=0.009) were associated with higher gains. This model was again significantly better than a null model (p< 0.001), contradicting our hypothesis. Contrasting 1g with 0.7/0.85/1.15/1.3g is, however, a less specific test of our hypothesis because it adds possible confounds such as differences in curvature. This is largely circumvented in the −1g condition, which we therefore consider the stronger test of our hypothesis. Our data thus supports the hypothesis that internalized knowledge about earth gravity guides eye movements.