Attention-based visual routines: sprites.

A central role of visual attention is to generate object descriptions that are not available from early vision. Simple examples are counting elements in a display or deciding whether a dot is inside or outside a closed contour (Ullman, Cognition 18 (1984) 97). We are interested in the high-level descriptions of dynamic patterns - the motions that characterize familiar objects undergoing stereotypical action - such as a pencil bouncing on a table top, a butterfly in flight, or a closing door. We examine whether the perception of these action patterns is mediated by attention as a high-level animation or 'sprite'. We have studied the discrimination of displays made up of simple, rigidly linked sets of points in motion: either pairs of points in orbiting motion or 11 points in biological motion mimicking human walking. We find that discrimination of even the simplest dynamic patterns demands attention.

Limits of attentive tracking reveal temporal properties of attention.

The maximum speed for attentive tracking of targets was measured in three types of (radial) motion displays: ambiguous motion where only attentive tracking produced an impression of direction, apparent motion, and continuous motion. The upper limit for tracking (about 50 deg s-1) was an order of magnitude lower than the maximum speed at which motion can be perceived for some of these stimuli. In all cases but one, the ultimate limit appeared to be one of temporal frequency, 4-8 Hz, not retinal speed or rotation rate. It was argued that this rate reflects the temporal resolution of attention, the maximum rate at which events can be individuated from those that precede or follow them. In one condition, evidence was also found for a speed limit to attentive tracking, a maximum rate at which attention could follow a path around the display.

Contrast perception across changes in luminance and spatial frequency.

The interaction of the effects of luminance and spatial frequency on perception of suprathreshold contrast was studied with use of a contrast-matching paradigm. Four subjects matched the appearance of Gabor patches at different luminances and spatial frequencies. The contrast of a 1-octave Gabor test patch at one of five frequencies [1-16 cycles/degree (c/deg) in 1-octave steps] and at one of seven mean luminance levels (0.5-50 cd/m2 in 1/3-log-unit steps) was matched, by the method of adjustment to a standard patch of 3 c/deg at 50 cd/m2 at a nominal contrast of 0.3. For each block of trials the spatial frequency of the test patch was randomly changed (three repetitions at each frequency per block) while the luminance was fixed. The subject regularly shifted fixation between the two targets in response to a metronome tone every 1.5 s. Contrast constancy was demonstrated across the entire luminance range tested for all but the two highest frequencies. For 8 c/deg the perceived test contrast was reduced only when the luminance was less than 2 cd/m2. For 16 c/deg, perceived contrast decreased linearly (with a slope of -1/2 on a log scale) with decreases in luminance across the entire luminance range. As at threshold, reduction in luminance across the levels commonly available on a CRT display has only minimal effects on low-frequency suprathreshold contrast perception. However, the apparent contrast of high-frequency features, in binocular free-viewing conditions, is rapidly reduced with a local reduction in screen luminance. This effect has important implications for visual models used in image-quality analysis.