Purchase this article with an account.
Srimant Tripathy, Syed Shafiullah, Michael Cox; Influence of Correspondence Noise on Dmax for Low Coherence Random-dot Kinematogram Stimuli. Journal of Vision 2014;14(10):467. doi: https://doi.org/10.1167/14.10.467.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Correspondence noise is a major factor limiting performance for detecting motion in random-dot kinematogram stimili, whether performance is measured as threshold coherence (Barlow, H.B. & Tripathy, S.P., Journal of Neuroscience, 17, 7954-7966, 1997) or as the upper limit for spatial displacement (Dmax) in fully-coherent two-frame kinematograms (Tripathy, S.P., Shafiullah, S.N. & Cox, M.J., PLOS ONE, 7:10, article e42995, 2012). The current study extends the earlier Dmax study to stimuli with low coherence-levels (approximately 30% coherence). Psychophysics: The dot density of 2-frame kinematograms was varied over the range 0.1-26.7 dots/deg2. For each tested dot-density in this range, dot displacement was varied and performance was measured as the proportion of trials with direction (left/right) of motion correctly identified. Dmax was the dot displacement that yielded 75% correct direction identification. Averaged over three observers, Dmax ranged from 42 arcmin at the lowest dot-density to 32 arcmin at the highest dot-density. Modelling: The stimuli used for the psychophysical experiments were presented to a model consisting of Reichardt detectors that randomly tiled the stimulus plane. The radius of the two catchment areas of each detector were scaled with the size of dot-displacement in the stimulus (Tripathy et al, 2012).In the earlier study a single scaling factor could account for Dmax across the range of dot densities tested using 100% coherent stimuli. In contrast, when the coherence level was 30%, the scaling factor was found to decrease monotonically with dot density.Conclusion: The results suggested that when stimuli were noisy, either the catchment areas of the motion detectors shrank monotonically, or the motion detecting system relied on a different set of smaller detectors, as dot density increased. These results complement recent physiological findings that adding visual noise causes the receptive fields of macaque MT neurons to shrink (e.g. Kumano, H. & Uka, T. Journal of Neurophysiology, 108, 215-226, 2012).
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only