Item

Abstract

Optic flow is generated by observer motion relative to stationary objects, by movement of objects relative to a stationary observer, and by combinations of those situations. To determine the relative contributions of object and self motion to the observed optic flow, the nervous system can use vestibular signals. An object’s speed relative to earth is given by the difference between its speed relative to the head and the head’s speed relative to the earth. The variance of the difference is the sum of the component variances. In contrast, if observers estimate self-motion from optic flow and vestibular signals, and assume a stationary visual scene, visual and vestibular estimates may be combined in a weighted average to yield more precise selfmotion estimates. So depending on whether the subject reports object motion or self-motion, the two-modality variance is predicted to be respectively higher or lower than the component variances. To test these predictions and the influence of the visual scene upon them, we measured speed-discrimination thresholds for fore-aft translations. There were two single-modality conditions, Visual and Vestibular, and two multi-modality conditions, Self-motion and Object motion. In the Visual, Vestibular, and Self-motion conditions, observers indicated if the movement was faster or slower than a standard. In the Object-motion condition, observers indicated if the object appeared to move with or against the self-motion. The experiment was run for two visual scenes, random-dots and ground plane with columns. In both scenes, multi-modal object-motion thresholds were, as predicted, higher than single-modality thresholds, and multimodal self-motion thresholds were, as predicted, generally lower than single- modality thresholds. For the ground plane scene, the self-motion thresholds could be consistently predicted by the weighted average of the single modalities, which was not the case for the random-dots condition.