Despite the elemental relationship of letters to written words, letter acuity is well-known to be a poor predictor of several indices of reading performance. This is why direct measurement of reading acuity is necessary in prescription of effective reading aids in low vision rehabilitation.

This paper presents a) a general discussion of differences between reading and letter acuity stimuli, and cognitive demands in the two acuity tasks, that may account for the poor power of acuity for single letters to predict reading performance, and b) an outline of a sensor model of text processing that can account for text and optotype crowding phenomena. The model has a linear filtering stage consisting of an array of sensors localized in space and spatial frequency that produce parallel bandpass representations of the text stimulus, and a decision stage consisting of a letter template-matcher and an attentional mechanism that selects the bandpass representations on which to perform the match. It is suggested that crowding effects at the acuity limit occur because neighboring letters or contour elements effectively add noise that falls within the the receptive fields and pass bands of sensors processing the letter.