Abstract
The responses of motion mechanisms have been shown, both neurophysiologically and psychophysically, to depend not only on the direction of a moving stimulus, but also on contrast, coherence and speed. We examined how contrast, coherence and directional selectivity interact by measuring directional tuning psychophysically across a wide range of motion coherence and contrast levels.
A 2-IFC procedure was used; the ‘noise’ interval contained randomly moving dots and the ‘signal’ interval contained noise plus two superimposed fields of coherently moving dots. Observers reported which interval contained global motion. We tested 7 angular differences between the directions of the two dot fields (0°, 22.5°, 45°, 67.5°, 90°, 135°, and 180°). We obtained two types of thresholds for 3 subjects: (1) coherence thresholds, where the proportion of dots moving coherently in the ‘signal’ interval was varied to obtain a coherence threshold for several fixed contrasts (rms, 3.2 – 30.2%) and (2) contrast thresholds, where the contrast of the dots was varied to obtain a contrast detection threshold for several fixed coherences (20 – 100% coherence). Thresholds for a given contrast-coherence pairing did not depend on whether coherence was fixed and contrast varied, or vice versa.
To model our results, we assumed that the response of the mechanism mediating performance depended on the summation between responses to the two fields of dots, while the response to each field depended multiplicatively on contrast and coherence. Contrast and coherence gain parameters were fixed based on neurophysiological estimates. This very constrained model describes the data well, and estimates a directional tuning bandwidth of ~83° for directionally selective mechanisms across a wide range of contrasts and coherences.
The good fit of the model demonstrates that directional tuning is nearly invariant across a wide range of luminance contrasts and coherence levels, as predicted by models of contrast normalization.