Flicker contours for a square image of 3° visual angle, centered 6° on the temporal side of the fovea, the light sectored at a focus, are strikingly modified if the same illuminated area is arranged in four squares separated by a narrow opaque cross. The "cone" curves are made much steeper, and their abscissae of inflection (τ' are at higher intensities; Fmax. is not greatly changed, but alters less with change of light-time fraction in the flash cycle (tL). This modification is accompanied by a great enlargement of the scotopic segment of the duplex curves, consistent with the theory of the integrative relations of neural effects in the two groups of units involved. The changes are not consistent with the view that flicker end-points are determined by the activation of retinal cells with a fixed spatial distribution of invariable thresholds. At tL = 0.50 the 3° subdivided area gives very nearly the same contour as does a square 6° x 6°, with the same total perimeter of light-dark separation; the "edge effect" thus suggested is complicated by differences in the dependence of Fmax. and τ' upon tL.
When an image pattern is produced by a grid of light bars separated by equally broad opaque spaces (10° x 10° over-all, centered at the fovea), the photopic flicker contours are made very steep and their midpoints are situated at quite low intensities, while the "rod" contribution tends to be more completely fused with the "cone" than is found for fields not subdivided. However, instead of a progressive increase of τ' with tL the curves for tL = 0.75 and 0.90 lie respectively below that for tL = 0.25 and 0.50 for a field of four broader stripes (1.43°) and both are below tL = 0.25 for a field of seven narrower stripes (0.77°). These latter changes are discussed in terms of the participation of subsidiary phenomena involving so called "γ movement."
It is pointed out that since in these data σ1/Im is for each set of conditions a statistically constant quantity with a characteristic breadth of scatter σσ, it is possible to calculate a "coefficient of internal correlation" r which is a function of the conditions (as: image area, location, wave length of light, structure of image, light-time fraction) and which describes a property of any entire contour. The changes in r, as a function of the conditions of flicker excitation, reflect changes in the neural organization responsible for the liminal discrimination of flicker.
It is shown that as consequence of simple changes in the image field, three parameters, as of the probability summation, are required for the description of a simplex flicker contour—since each of these is independently modifiable as to its magnitude and in its dependence on the light-time fraction.
Subdivision of the image, with light sectored at a focus, produces in part only the changes in the flicker contour which we have earlier labelled the "pecten effect." In the latter, with light not sectored at a focus but with bar images moving across a field with inclined fixed opaque bars, the "cone" slope (dF/d log I) is sharply increased for tL > 0.50, but not below tL = 0.50, and the value of τ' is much less than it "should be." Consequently, the change in contrast brought about by the moving contact of light/dark borders is the significant factor in the "pecten effect," not simply pulsatile interruption of the light.
THEORY AND MEASUREMENT OF VISUAL MECHANISMS