Vernier acuity predicted from changes in the light distribution of the retinal image.

If two thin bars of different luminance are placed side by side, their joint spatial position in a Vernier alignment task is determined simply by their relative luminances. The threshold luminance contrast difference required to produce a just detectable change in spatial position corresponds to a spatial shift of 5-20 arcsec in the centroid of the retinal light distribution, depending upon contrast relative to the background. This technique may be used to measure acuity with a display that has a spatial resolution considerably worse than the Vernier offset threshold. We have also extended the centroid technique to components that differ both in wavelength and luminance. Colour was found to make no essential difference to the task. Taking into account the spread of light in the retinal image, the manifest contrast thresholds are equivalent to threshold intensity increments between adjacent foveal receptors of less than 1% comparable to the values reported by Hecht and Mintz for dark line detection.