Abstract
A moving observer must be able to judge her direction of motion accurately even when moving objects are in the scene. Psychophysical results (Royden & Hildreth, 1996) show that moving objects cause biases in human heading judgments that depend on the position and direction of motion of the object. Objects moving horizontally cause a heading bias in the same direction as the object's motion. Objects moving in depth, however, sometimes cause a bias in the direction toward the object's own focus of expansion (a direction opposite the object's horizontal component of motion). What can account for these seemingly conflicting results? The present study tested a model for heading computation that is based on the motion-opponent properties of neurons in the Middle Temporal area (MT) of primate visual cortex (Royden, 1997). The purpose was to determine whether moving objects would cause biases in the computed heading direction similar to those seen with human subjects. This model was tested using conditions identical to those used in Royden and Hildreth, which simulated observer motion toward 2 frontoparallel planes. The scene also contained a moving object (10 × 10 degrees in size) which moved either horizontally with respect to the observer or in depth toward the observer. When the object moved in depth, the direction of motion relative to the observer was either 1 deg or 10 deg to the right of the center of the scene, while the simulated observer motion was 6 deg to the right of center. Simulations were run for each of the different object positions tested in the psychophysical studies. The results show that the model behaves similarly to human observers. For each of the object motion conditions the model shows biases in the same direction and magnitude as seen for human observers. This suggests that motion-opponent properties of MT neurons may account for these biases.
Supported by NSF Grant #IBN-0196068.