Sampling efficiency and internal noise for motion detection, discrimination, and summation.

By comparing real observers to an ideal observer, previous studies have found that the detection of static patterns is limited by internal noise and by imperfect sampling efficiency. We developed and applied ideal observer models for the detection, discrimination, and summation of oppositely drifting gratings in Gaussian white noise. The three tasks share a common source of internal noise. The sampling efficiencies were on the order of 1-2% except for much lower efficiency in direction discrimination for faster moving gratings. The efficiency of direction discrimination relative to detection systematically declines as the speed is increased from 1 to 6 Hz. These results suggest that observers use mismatched filters tuned to slow speeds regardless of the signal speed. Human visual motion sensing appears to use distorted representations of the incoming signals, and this distortion is a major limitation to visual performance.