Abstract
Retinal image motion provides rich information about both the motion and depth of objects. When a stationary observer moves their eyes, the retinal image motion of a moving object is the resultant of eye movement and object motion. Thus, to infer object motion in the world, a coordinate transformation (CT) needs to be performed that subtracts the retinal motion component contributed by eye movements. In a different viewing context, in which the observer translates laterally and counter-rotates their eyes to maintain visual fixation on a world-fixed target, the depth of a stationary object can be estimated from the ratio between its retinal velocity and eye velocity (depth from motion parallax, MP; Nawrot & Stroyan 2009). Estimating the motion or depth of an object under these two viewing contexts (CT and MP) requires different computations: summation vs. division of retinal and eye velocity signals. It remains unclear how the same signals, retinal motion and eye movements, mediate the perception of diverse sensory variables under different viewing contexts. We asked human subjects to estimate the direction of motion of an object during actual or simulated pursuit eye movements, and we examined whether the subjects’ belief about the viewing context influences their perception of object motion. Preliminary results from three subjects show opposite biases in perceived direction of object motion when optic flow simulated the two different viewing contexts (CT and MP) described above. We found that a purely rotational optic flow field elicits a percept of object motion consistent with CT, whereas a flow field simulating the combination of observer translation and rotation triggers an opposite perceived direction after parsing out the component contributed by MP. Our study demonstrates that the interaction between retinal motion and eye movements can lead to very different percepts, depending on the interpretation of the viewing context.