Threshold exposure duration as a function of luminance for visual recognition

Threshold exposure duration was measured at set luminances for recognition at successively higher criteria of 1, 2, 3, 4, and 5 correct letters in nonsense words. Two sizes of letters were used and triads of Landolt “Cs” were compared with letters. Tachistoscopic procedures were the typical ascending seriesin the method of limits and pre- and postexposure light adaptation. Obtained log t vs log I functions were two-part straight line curves both describable by the equation, log t + log I^k = log C, but with k and C abruptly changing at a certain I. Reciprocity (k = 1) held only for low Is and the low criterion; as criterion was raised at low Is k became less negative (about ?.60 at five correct) for both sizesand for “Cs.” At higher Is, k shifted from ?.25 to ?.40 with increasing criterion for large letters, but variedaround ?.45 at all criteria for small. The lower the criterion, the lower the / and t at which k and C abruptly changed for large letters, but only t was lower for small. Changes in t with acuity requirement change, and from one criterion level to the next, were not uniform over all Is. Results may reflect obscuring procedural factors or basic properties of visual recognition processes. Some interpretations in terms of procedural variables are offered. It appears more likely that visual recognition involves central processes for which time is needed in excess of that for summation of luminances for mere detection by sensory processes, as expressed in Bloch s’ reciprocity law.