Sensitive calibration and measurement procedures based on the amplification principle in motion perception.

We compare two types of sampled motion stimuli: ordinary periodic displays with modulation amplitude m(o=e) that translate 90 degrees between successive frames and amplifier sandwich displays. In sandwich displays, even-numbered frames are of one type, odd-numbered frames are of the same or different type, and (1) both types have the same period, (2) translate in a consistent direction 90 degrees between frames, and (3) even frames have modulation amplitude m(e), odd frames have modulation amplitude m(o). In both first-order motion (van Santen, J.P.H. & Sperling, G. (1984). Temporal covariance model of human motion perception. Journal of the Optical Society of America A, 1, 451-73) and second-order motion (Werkhoven, P., Sperling, G., & Chubb, C. (1993). Motion perception between dissimilar gratings: a single channel theory. Vision Research, 33, 463-85) the motion strength of amplifier sandwich displays is proportional to the product m(o)m(e) for a wide range of m(e). By setting m(e) to a large value, an amplifier sandwich stimulus with a very small value of m(o) can still produce visible motion. The amplification factor is the ratio of two threshold modulation amplitudes: ordinary circumflexm(o=e) over amplified circumflexm(o), circumflexm(o=e)/circumflexm(o). We find amplification factors of up to about 8x. Light adaptation and contrast gain control in early visual processing distort the representations of visual stimuli so that inputs to subsequent perceptual processes contain undesired distortion products or 'impurities'. Motion amplification is used to measure and thence to reduce these unwanted components in a stimulus to a small fraction of their threshold. Such stimuli are certifiably pure in the sense that the residual impurity is less than a specified value. Six applications are considered: (1) removing (first-order) luminance contamination from moving (second-order) texture gratings; (2) removing luminance contamination from moving chromatic gratings to produce pure isoluminant gratings; (3) removing distortion products in luminance-modulated (first-order) gratings - by iterative application, all significant distortion products can be removed; (4) removing second-order texture contamination from third-order motion displays; (5) removing feature bias from third-order motion displays; (6) and the same general principles are applied to texture-slant discrimination in which x,y spatial coordinates replace the x,t motion coordinates. In all applicable domains, the amplification principle provides a powerful assay method for the precise measurement of very weak stimuli, and thereby a means of producing visual displays of certifiable purity.