Abstract
When perceptual systems adapt they may be sacrificing accuracy for precision. By allocating the existing coding apparatus so as to tune its range to the locally experienced stimulus space, adaptation processes can enhance the discriminability of likely stimuli. For example, the tight coupling of perception and action during locomotion provides an important informational substrate for improving perceptual discrimination. We have found that the perceptual discrimination of optic flow speed is automatically tuned to walking speed such that perceived flow rates are reduced during walking. However, contrary to threshold elevation accounts of this phenomenon, perceptual discrimination is actually reliably enhanced for flow speeds appropriate to walking. To measure this we used a 2IFC speed discrimination task administered in a wide-area virtual reality apparatus that allowed us full control over the visual speeds presented during natural walking. In two experiments it was found that discrimination performance deteriorated dramatically for visual speeds less than 40% of walking speed, but showed striking and consistent improvement for visual speeds near to and above walking speed. It is as if the entire coding space available for visual motion is dynamically reallocated when walking — leaving slow visual motions undetectable, but enhancing the discrimination of self-motion-relevant visual speeds. Although many vivid demonstrations of perceptual aftereffects seem to show that perceptual systems are susceptible to error and bias, techniques that look for the gains in perceptual precision and sensitivity may be more revealing.
Swarthmore faculty research grant