Color spaces of color-normal and color-abnormal observers reconstructed from response times and dissimilarity ratings.

With multidimensional scaling analysis, color spaces were reconstructed from reaction times (RTs) required to make same-different judgements of pairs of 15 equiluminant colors and from dissimilarity ratings between them. In addition to normal trichromats, observers with red-green color deficiency were tested. Two main purposes were served by this study: (1) to compare spatial representations of colors derived from discriminative RTs with those derived from dissimilarity measures and (2) to examine whether the task may selectively affect the dimension reflecting, in color-abnormal spaces, the deficient red-green mechanism. Contradicting our hypothesis of lower dimensionality of RT spaces, as compared with rating spaces, no consistent differences in solution dimensionality were found. However, configurations derived from the two measures diverged. The rating procedure yielded the most logical results for recovering color space. The RT configuration revealed contraction in the tritanopic direction, indicating longer color processing when the short-wavelength mechanism is involved, and in addition, for color-abnormal observers, clustering in the protanopic and deuteranopic directions, indicating even longer processing by the deficient red-green mechanism. This finding implies that RTs are suitable for detecting temporal differences in color processing but, for that very same reason, rather ill-suited for reconstructing color spaces.

Trends in European visual science: 1978-1997.

European visual science during the past 20 years--with the glaring exception of ophthalmological research--is reflected in the annual programmes of the European Conference on Visual Perception. The number of countries that have participated has increased with time, and the spectrum of topics that have been presented has broadened. The number of multiple-author papers has increased dramatically, as has the number of papers by authors from different institutions and different countries.

Effect of luminance on color perception of protanopes.

Small-field color-naming performance of two protanopes over a 4-log luminance range was impoverished in comparison with that of normal trichromats, and was more strongly affected by changes in luminance. At 200 cd/m2 responses to mid-spectral lights were dominated by 'yellow'; with lowering luminance, 'green' and 'red' were increasingly used. In the color spaces derived from these data the first two dimensions for trichromats are red-green and yellow-blue: those of the protanopes appear to be brightness and 'red-blue'. In the protanopes' color space the greater separation of stimuli at 0.2 cd/m2 suggests that with low luminance their color discrimination improves.

Motion extrapolation and velocity transposition.

A study of the effect of the size of a moving target and the extent of its visible motion on motion extrapolation is reported. Targets (a horizontal pair of dots separated by either 0.2 or 0.8 deg) moved across a 10 deg rectilinear path and were then occluded. Observers pressed a key when they thought the leading dot of a hidden target had reached a randomly specified position (0-12 deg from the point of occlusion). In experiment 1, in agreement with velocity-transposition predictions, at moderate (5 deg s-1) and rapid (10 deg s-1) velocities extrapolation times were longer for large targets than for small ones. At slow velocity (2.5 deg s-1) this effect was reversed. In experiment 2 the effect of target size at moderate velocity was found for a short (2.5 deg) visible path. However, the extrapolation time increased with shorter (2.5 deg versus 10 deg) paths. A proposed account of these effects suggests that the visual system performs a spatiotemporal scaling, according to the velocity-transposition principle, not only of visible motion but also of extrapolated motion.

Dynamic complexity of visuo-motor coordination: an extension of Bernstein's conception of the degrees-of-freedom problem.

Extending Bernstein's spatial conception of the degrees-of-freedom problem in the human motor system, we introduce a method developed from the theory of non-linear dynamics that allows one to quantify the spatio-temporal, i.e. dynamic, complexity of visuo-motor coordination. The correlation dimension D is used to measure the effective number of dynamic degrees of freedom in the coordination that a subject uses when performing a visuo-motor tracking task. The validity of the estimator employed is demonstrated. Visuo-motor coordination had a low-dimensional (mean D-SD=6.07 -0.82) dynamic structure, which was consistent with deterministic chaos rather than with pure stochastic noise. D correlated with tracking performance, P. Both D and P were closely related to the degree of visuo-motor compatibility that the task presented to the subject. However, for short periods of training P increased, but D did not. As these seemingly contradictory results suggest, our dynamic conception of the degrees-of-freedom problem may reveal far more intricate visuo-motor interactions than Bernstein could identify on the basis of his spatial analyses of bodily movement patterns and by the methods of evaluation that were available to him at the time.

Varying the strength of the Munker-White effect by stereoscopic viewing.

In the Munker-White effect grey target bars appear lighter when they are flanked by white bars, and darker when they are flanked by black bars. It is shown that the effect is enhanced if the patterns are presented stereoscopically so that the grey bars appear either behind the grating, in which case they are seen as a rectangle that is occluded by the white bars of the grating, or in front of the grating, so that they form a transparent rectangle. These results are explained in terms of object perception: contrast enhances differences between an object and its surroundings, whereas assimilation reduces differences within an object.

Simulated fovea of the human retina: psychophysical data confirming the model's ability to accurately predict resolution.

A series of psychophysical tests were designed to determine whether a computer simulation of the human retina could accurately predict the geometry of various stimuli that were optimally resolved for human foveal vision. Stimuli were used that were of the order of the grain of the cone mosaic, i.e., of the order of 2 x 2'. In the first set of experiments, resolution was tested using a two-bar stimulus. In one experimental series the gap between the two bars was varied, and in a second series the gap was kept constant and the width of the bars varied. In a second set of experiments, various block letters and a number of series for each letter were used; in each experimental series a single parameter was systematically varied. The same stimuli were also used as inputs for the computer simulation. When proper controls were used, the psychophysical data and computer simulation gave remarkably comparable results. Care was taken to differentiate between simple detection of a pattern, and resolution, which involved proper identification of the image.

Changes in sensitivity of the dark-adapted eye during concurrent light adaptation of the other eye.

Thresholds for detection of light by a dark-adapted test eye were measured while the other, non-test eye was either similarly dark adapted or while it was exposed to an intense red adapting field. An interocular effect that depends on the retinal location of the test was found: compared to the threshold during binocular dark adaptation, sensitivity decreased during contralateral light adaptation when the test was presented to the foveola and up to 4 deg above it; but sensitivity increased when the test was between 7 and 12 deg, showing a reversal at 5 deg.

Counterphase dichoptic flicker is seen as its own second harmonic.

If a patternless field is modulated sinusoidally in time so that the luminance change in one eye is in counterphase to that in the other, the resulting flicker appears faster than if the modulation to both eyes has the same phase. If observers set the frequency and the amplitude of a comparison in-phase field so that it matches a neighbouring counterphase field, modulated at, say, 2.5 x its threshold, they set the frequency to twice the counterphase frequency, and the amplitude to a value that is, for a given frequency, a constant ratio of the modulation of the counterphase field. Counterphase stimulation thus appears to cause an internal second-harmonic signal. However, it is not possible to cancel this by adding a second harmonic component to the stimulus.

A reduction in stimulus duration can improve wavelength discriminations mediated by short-wave cones.

Virtually all visual discriminations become less accurate when either the luminance or the duration of the stimulus is reduced. An exception is found for wavelength discriminations near 460 nm, where an increase in either luminance or duration can cause the threshold to rise. For flashes of 100 msec or less, the critical variable is the total energy of the flash (i.e. the product of retinal illuminance and flash duration), and wavelength discrimination is optimal at an intermediate value; higher stimulus energy causes discrimination to deteriorate. To explain these findings we suppose that discrimination in this region of the spectrum is mediated by a channel that draws opposed signals from the short-wavelength cones and from some combination of the middle- and long-wavelength cones, and that high stimulus energies cause saturation of this channel.

A multiple-channel model for grating detection.

The multiple-channel model that was proposed in 1974 by MacLeod and Rosenfeld to describe the sensitivity of the visual system to complex gratings has been applied to a variety of complex gratings. The model predicts correctly the contrast sensitivity at both high and low spatial frequencies to square-wave gratings, square-wave with missing fundamental, sawtooth-wave, and trapezoid-wave gratings of different ramp widths. It provides a physiologically reasonable explanation of the detection of luminance gradients without requiring special gradient detectors. The model resembles the threshold behavior of the visual system in that it is sensitive to the relative phase of frequency components at low frequencies, but not at high frequencies.