Abstract
When field luminance rises or falls quickly, the temporal course of the change must be difficult, if not impossible, for perception to follow. This proposition is formally rooted in the well-known studies of two-pulse interactions at threshold; it is supported by a study by Crawford that deals with spatially coextensive, suprathreshold luminance changes.1 The present experiment measured direction-reversal thresholds and assessed observers' sensitivities to a sharp alteration in the temporal course of a luminance change. Luminance was incremented or decremented uniformly over the entire extent of the field (7.5° diameter). The magnitude of change was constant at 75 cd/m2 (50% of base luminance). The rate of change was maximum with the steepest ramp (time: 6.6 ms), and it was at minimum with the shallowest ramp (time: 280 ms). The test stimulus was a "zigzag" produced by a ramp disfigured by a step in the opposite direction to that of the ramp; the step always occurred at the half-height of the ramp. by means of a 2AFC stair-case procedure, the magnitude of the step was varied until the observer could discriminate between a smooth ramp and a zigzag (about 79% correct rate): (1) With ramp time 20 ms or less, the mean magnitude of the reversal at threshold was 78.5 cd/m2. (2) As ramp time became longer than 40 ms, reversal thresholds quickly fell, to a minimum of 6 cd/m2. (3) Thresholds were usually sign-dependent (lower with negative-going ramps) but not with either the steepest or the shallowest ramp. Apparently, linear systems analysis can account for some, but not all, of these results.