Detection and resolution of vanishing optotype letters in central and peripheral vision.

Previous studies of peripheral vision have shown that detection acuity is superior to resolution acuity for gratings over a range of contrasts, which is attributed to different limiting mechanisms (contrast insufficiency and neural undersampling) for the two tasks. To extend the analysis to letters in a way that avoided luminance cues, we used "vanishing optotype" characters, conveying second-order information, and constructed from tripole strokes having the same mean luminance as the surround. We measured the minimum letter size for detection and identification tasks for two different pairs of vanishing optotype characters (O vs. + and orthogonally oriented Landolt-C's) as a function of contrast in central and peripheral vision. Foveally there was no significant difference between detection acuity and resolution acuity for either pair of letters over a range of stimulus contrasts from 20% to 100%, indicating performance is contrast-limited for both tasks. The same result was obtained at 30° eccentricity in the peripheral field for the O vs. + letters, again indicating performance is contrast-limited for both tasks. However, resolution acuity for the Landolt-C letters was significantly worse than detection acuity in the periphery over the same range of contrasts, which suggests performance is limited by neural undersampling for these letters. All of our experimental results are explained by a model of neural sampling in which detection acuity is determined by the size of neural receptive fields relative to the dimensions of the tripole responsible for spatial contrast, whereas resolution acuity is determined by the spacing of receptive fields relative to the spacing between strokes responsible for letter form.