Abstract
Letter recognition is usually thought to rely on the shape and arrangement of distinctive pattern features such as line segments and curves. In findings to be reported, we have found that high levels of letter-recognition accuracy are possible when low-pass filtering reduces the spatial bandwidths of letters to levels not expected to support adequate recognition of letter shape. We addressed this apparent discrepancy by testing the hypothesis that the human visual system relies increasingly on grayscale coding (contrast coding) for letter recognition when spatial resolution is severely limited. The hypothesis predicts that as spatial resolution for rendering letters decreases, subjects will rely more on grayscale variations, therefore requiring a larger gap between contrast thresholds for letter detection and letter recognition. We measured contrast thresholds for detecting and recognizing single letters (Courier, 1°) drawn at random from the 26 letters of the English alphabet. The letters were low-pass filtered (blurred) with a third-order Butterworth filter with bandwidths (defined as the frequency at half amplitude) of 0.9, 1.2, 2, and 3.5 cycles per letter. Threshoolds were also measured for unfiltered letters. Data from seven normally-sighted subjects showed that differences in contrast thresholds between detection and recognition increased substantially with decreasing bandwidth. The ratio of recognition to detection thresholds increased from 1.5 for the unfiltered letters to 8.8 for the most blurred letters (0.9 c/letter). These findings support the hypothesized increased reliance on grayscale information for letter recognition when spatial resolution is reduced.
This work was supported by NIH grant R01 EY002934.