Abstract
Recognizing low-contrast patterns off the point of fixation is more difficult than what would be expected on the basis of retinal receptor density or size of projection onto the primary visual cortex (Strasburger et al. 1994, Eur. J. Neurosci. 6, 1583–1588). Nonetheless, estimates of the cortical magnification factor predict reasonably well, across the visual field, both recognition performance for single, high contrast letters and grating detection contrast sensitivity. After a review of the M-scaling concept, a non-linear descriptive model is presented which predicts single-character recognition at arbitrary contrast across the visual field. The model is a simplified version from the one presented previously, based on extensive additional data (Gothe, unpublished thesis; in total 19 sizes, 65 visual field positions, ∼ 160,000 responses, in 17 young subjects). It states a hyperbolic relationship between letter size and log recognition contrast sensitivity, the asymptotes of which depend linearly on retinal eccentricity. Solved for letter size, it represents a generalization of M scaling which at high contrast is reduced to conventional M scaling. The prediction of percent correct recognition is by the psychometric function. A comparison to Gabor detection contrast sensitivity is drawn in the same subjects. The relationships to Levi's E2 concept and to Bijl's account of receptive field overlap are discussed. With the provided empirical parameters, the model predicts the proportion of correct recognition performance for singly presented characters of any contrast, of any size and at any position in the visual field.