Choice reaction times for identifying the direction of first-order motion and different varieties of second-order motion

This study sought to quantify the temporal properties of the human visual system by measuring forced-choice reaction times for discriminating the drift direction of first-order motion (luminance-modulated noise) and a variety of second-order motion patterns (modulations of either the contrast, polarity, orientation or spatial length of a noise carrier) over a range of stimulus modulation depths. In general, reaction times for all types of second-order motion were slower than those for first-order motion. Specifically, reaction times were similar for modulations of image contrast, polarity and orientation but were markedly slower for modulations of spatial length. There was also a tendency for reaction times to decrease as stimulus modulation depth increased. The rate of this decrease was shallowest for first-order, luminance-defined patterns. For second-order motion reaction times decreased at a similar rate for contrast, polarity and orientation but this decrease was steepest for spatial length. However, when equated in terms of visibility (multiples of direction-discrimination threshold), the rate at which reaction times decreased as modulation depth increased became comparable for patterns defined by luminance, contrast, polarity and orientation. For patterns defined by spatial length, performance could not be equated in this manner. These findings demonstrate that the time taken to encode the direction of each pattern is not an invariant response metric. The results are consistent with psychophysical and electrophysiological evidence for longer response latencies for second-order motion and may reflect the additional processing stages (e.g. filter–rectify–filter) required for its extraction.