Bjoern Joerges¹, Marlene Wessels², Heiko Hecht², Laurence R. Harris¹; ¹Center for Vision Research, York University, ²Johannes Gutenberg-Universität Mainz

It is well known that humans do not fully use the acceleration of a target to estimate the time-to-contact of a disappearing object. Rather, they rely on the object’s speed at the point of disappearance to extrapolate its motion, which leads to an overestimation of time-to-contact when the target is under the influence of a positive acceleration. It has further been observed that this overestimation is attenuated or even fully absent when the object moves under the influence of Earth gravity, indicating that humans use a strong Earth gravity prior to make such judgements. In this project, we investigated whether this is also the case for self-motion. We immersed 8 participants in a 3D virtual environment consisting of tall buildings and a road and then visually simulated them as either falling down (towards the ground) or “falling up” (towards the simulated ceiling). To avoid confounds, an identical stretch of road was used both as the ceiling and the ground. Participants started out from rest and accelerated at +9.8 m/s2 either upwards or downwards in randomly interleaved trials. After experiencing such optic flow for 0.8 to 1.4s, the display blacked out and participants were asked to indicate by button press when they either hit the ceiling with their head (-1g condition) or the ground with their feet (1g condition). Participants pressed the button significantly earlier when falling down than when falling up, indicating that they relied more on acceleration-based cues when predicting their own downwards motion than for upwards motion. This shows that – just like when judging the motion of objects – humans rely on a strong Earth gravity prior when assessing their self-motion.