Disentangling the effects of object position and motion on heading judgments in the presence of a moving object

Abstract

As we move about the world, our retinal image of the world undergoes a lawful transformation, called optic flow (Gibson, 1950). Specifically, when we move along a straight path without rotating our eyes, head and body, the optic flow takes a radial pattern, in which all the elements in the environment originate from a static point called the focus of expansion (FOE) that lies in the direction of self-motion. It has been shown that the direction of self-motion is perceived readily from optic flow when traveling along a straight line without making eye movements (Warren, Morris, & Kalish, 1988), when making eye movements (L. Li & Warren, 2000) and when traveling along a curved path (L. Li & Cheng, 2011), with errors usually within 2 degrees. This ability has been demonstrated using rigid worlds in which all motion is due to self-motion, but we frequently travel through worlds where other objects are moving, too, be they oncoming pedestrians or cars in the street. A moving object would add an extra motion component locally to the retinal flow generated by self-motion, disturbing the coherent pattern and potentially making heading judgments based on the global flow field less reliable. Previous research has found that a moving object can bias heading perception only when it occludes (or almost occludes) the focus of expansion (FOE) in the background optic flow, with the direction of bias depending on whether the moving object approaches or is at a fixed distance from the moving observer (Royden & Hildreth, 1996; Warren & Saunders, 1995). However, the effect of object motion on heading perception was confounded with object position in previous studies. In the current study, we examined whether the occlusion of the background FOE is a prerequisite for a moving object to induce heading bias and whether strong visual cues can be used to segment out the moving object prior to estimating the heading. In Experiment 1, the display simulated forward observer motion through a 3D random dot cloud in the presence of an object that remained at a fixed position on the screen but with inside dots moving in different patterns. We found that an object far away from the background FOE elicited the same amount of heading bias as an object occluding the FOE, and the direction of heading bias was determined by the direction of object motion when dots within the object moved in a coherent way and by the object position when dots within the object moved randomly. In Experiments 2 and 3, we introduced color contrast and stereoscopic depth to the displays. The results showed that the human visual system used neither of these cues to segment the object from the global flow field prior to heading estimation.