Different processes underlie the detection of second-order motion at low and high temporal frequencies

Abstract

The aim of this project was to determine whether first- and second-order motion stimuli are detected by the same mechanism. We began by measuring the temporal contrast sensitivity function (CSF) for the directional discrimination of 0.3 c deg$^{-1}$ drifting spatial beat patterns and luminance modulated gratings. The CSF for beat patterns was bimodal, with sensitivity maximal near 1-2 Hz and 10-12 Hz. In contrast, the CSF for luminance gratings had a single peak near 10 Hz. Separate adaptation experiments were then done using all permutations of beat patterns and luminance gratings as the test and adaptation stimuli. In general, sensitivity to a test pattern of fixed temporal frequency (2, 4, 8 or 16 Hz) was measured both before and after adaptation to patterns whose temporal frequency varied over a wide range. The rationale for these experiments was that if first- and second-order stimuli are processed by the same mechanisms, the adaptation tuning curves should all be similar. Our results show that this is the case at high temporal frequencies (> 4 Hz), but not at low temporal frequencies. The post-adaptation sensitivity functions for test patterns with temporal frequencies of 8 Hz and 16 Hz were bandpass, with maximal adaptation near 12 Hz, and showed evidence of beat-specific adaptation; for 2 Hz test patterns the sensitivity functions were lowpass, adaptation declining above 20 Hz, but there was no beat-specific adaptation. The shape of the CSFS and the results of the adaptation experiments show that separate processes mediate the detection of drifting beat patterns at low and high temporal frequencies. The results are consistent with the hypothesis that fast second-order motion is detected by Fourier-type mechanisms, preceded by a nonlinearity, and slow second-order motion is detected by a process involving a comparison of local luminance features.