A model of threshold and suprathreshold vision in relation to variation of object size is proposed. An important factor in suprathreshold vision is the signal-contrast loss in optical and neural components; in threshold vision, an important additional factor is the manner in which the brain acts on retinal noise. The signal-contrast loss has been measured in terms of a suprathreshold signal-transfer function for circular signals. In threshold vision, the noise processes are shown to predominate and the signal-contrast losses to be insignificant. Therefore signal-transfer functions cannot be deduced solely from threshold measurements; indeed, peak performance in suprathreshold vision is reached at object size one-tenth that at threshold. In suprathreshold vision, performance is shown to depend on a balance between optical unsharpness and neural sharpening (lateral inhibition). Contrast loss due to neural unsharpness (summation) appears to be insignificant in foveal vision and only partially significant in peripheral vision. Hence neural properties such as transfer functions, receptive-field sizes, and the spatial extent of neural interactions cannot be deduced from over-all measurements on the eye without correction for optical unsharpness.
GEORGE A. HAY and M. SUSAN CHESTERS, "Signal-Transfer Functions in Threshold and Suprathreshold Vision," J. Opt. Soc. Am. 62, 990-998 (1972)