Perceiving path from optic flow

Abstract

We have previously shown that when traveling on a circular path, observers did not rely on flow lines or vector normals to perceive the path trajectory of their traveling from optic flow. Furthermore, they could not perceive path rotation from the change of heading when retinal flow did not contain rotation. Here we further investigated this topic. In Experiment 1, the display simulated an observer traveling on a circular path (curvature: ±0.017 m⁻¹, ±0.026 m⁻¹, & ±0.035 m⁻¹) over a textured ground or a textured ground with 20 posts (depth range: 5–20 m). The simulated observer gaze direction was pointed to a target (1) on the path at 30° away from the initial heading, (2) at 15° outside of the path, (3) at 15° inside of the path, or (4) along the Z-axis of the simulated environment. Consistent with our previous findings, path performance was not accurate and was affected by path curvature for all viewing conditions. Furthermore, path performance was similar for the two display conditions, indicating that reference objects do not help path perception. In Experiment 2, instead of pointing to the target, the simulated observer gaze direction pointed to heading thus the target moved on the screen rendering pursuit eye movements. Compared with data from Experiment 1, path performance was more accurate and less affected by path curvature, indicating that extra-retinal signals improved path perception. In Experiment 3, observers were asked to estimate path rotation using a 2AFC task for the four viewing conditions with the textured ground display from Experiment 1. Path rotation estimation errors were highly correlated with path perception errors observed in Experiment 1. We conclude that observers estimate path curvature using their perceived path rotation and then perceive path relative to their heading. Extra-retinal information helps observers accurately estimate path rotation and thus contributes to accurate path perception.