Pergamon 004249@(93)RiM14-X Yrkion Res. Vol. 34, No.12, pp. 1585-1594, 2994 Copyright 0 1994 Etsevier Science Ltd Printed in Great Britain. All rights reserved 0042-6989/94 57.00 + 0.00 Timing of Visual Events for Motion Discrimination JtfRI ALLIK,* ALEKSANDER PULVER* Received 12 May 1993; in revised form 21 October 1993 The observer’s task was to identify the temporal order of the two adjacent luminance excursions one of which was a step-function and the other a linear increase in luminance starting from zero and reaching various final ~pR~de A after some period of thne t). The intense delay, At, between these two transitions was determined at which they appeared isochronous. The point of the subjective equality (PSE) depended on both ramp parameters, the rise-time duration zyxwvutsrqponmlkjihgfedcbaZYXWV D and its amplitude A. AR of the data can be accounted for by supposing that judgements about the temporal order are based on the comparison of a simple attribute of these luminance excursions, the time momeuts when the iumioous energy concentrated on low temporal frequencies exceeds some level. The perceived temporal order, which was experienced as a Ieftward or ~gh~ard ~placement of the whole pattern, was determined by the sequence in which Iow-freqnency portion of these two luminance excursions reached the threshold value. The implications of this simple contrast detection explanation for theories of motion analysis are discussed. Temporal order perception Motion analysis Timing of visual events Perceptual latency Visual masking INTRODUCTION In many cases, the human observer can easily tell which of two events occurred first. The capacity to make deliberate discrimination of temporal order of two per- ceptual events drops to chance when the difference between their onsets is about 40-50msec (Exner, 1875; Hirsh & Sherrick, 1961; Allan, 1975; Westheimer & McKee, 1977b). On the other hand, the subject can easily ~stin~ish sounds, flashes, and vibrations that differ only in the order in which two component stimuli occurred at a fraction of the interstimulus time interval at which he or she can explicitly specify their order (Efron, 1973). Analogously, the temporal order of two adjacent flashes can be correctly identified when one is delayed by as little as 3-5 msec (Sweet, 1953; Westheimer & McKee, 1977b). The observer’s sensitivity to the difference between two asynchronies is even better than the detection of temporal order per se. The smallest difference in the asynchrony of two flashes which can be reliably detected is about 2 msec (Burr, 1979; McKee, 1981). Two spatially proximate asynchronous flashes create a clear impression of motion which direction depends on the order of appearing of two flashes. Motion can be seen even though two stimuli could not be spatially resolved when they are exposed simul- taneously (Exner, 1876; Thorson, Lange & Biederman- Thorson, 1969), Evidently, the observer’s exquisite *Department of Psychology, University of Tartu, Tartu, Estonia EE-2400. timing accuracy of visual events at small spatial separ- ation is a consequence of the requirement to resolve visual motion. The prediction of the temporai order of two identical visual events appears to be an elementary task logically at least: if the temporal asynchrony between two visual events is long enough their order can be determined on the basis of their succession. The situation becomes more complicated, however, when events are not completely identical. There is no more a single and intuitively transparent rule how to predict the apparent perceptual order of two arbitrary luminance excursions. The mov- ing target leaves behind a translating pattern of changes in the luminance flux across the retinal surface. Such a translation produces typi~lly almost identical but de- layed luminance variations at two neighbouring sites along the motion path. Following this line of reasoning, Reichardt (1957) proposed an attractively simple delay- and-multiply scheme for the detection of motion. Ac- cording to this scheme, the basic operation for the motion detection is the multiplication of a signal re- ceived from one spatial location with a delayed signal from another adjacent spatial location. Although the response of the bilocal analyzers is maximized when two time-varying input signals have identical form and their relative delay corresponds to the internal delay, motion can be perceived when two signals have different forms. In the present study, we investigated the ability to discriminate temporal order of two luminance tran- sitions varied in the duration of their onset time. The onset time defined as the time needed for the amplitude 1.585