Human Reading and the Curse of Dimensionality

Whereas optical character recognition (OCR) systems learn to classify single characters; people learn to classify long character strings in parallel, within a single fixation . This difference is surprising because high dimensionality is associated with poor classification learning. This paper suggests that the human reading system avoids these problems because the number of to-be-classified images is reduced by consistent and optimal eye fixation positions, and by character sequence regularities. An interesting difference exists between human reading and optical character recognition (OCR) systems. The input/output dimensionality of character classification in human reading is much greater than that for OCR systems (see Figure 1) . OCR systems classify one character at time; while the human reading system classifies as many as 8-13 characters per eye fixation (Rayner, 1979) and within a fixation, character category and sequence information is extracted in parallel (Blanchard, McConkie, Zola, and Wolverton, 1984; Reicher, 1969). OCR (Low Dbnensionality) I Dorothy lived In the .... I [Q] ... _ ................................. .. "D" ~ ................................. .. "0" o ~ "R" HUnlan Reading (High Dbnensionality) I Dorothy lived In the midst of the ..... I Dorothy lil Ilived In the I ....... I midst of the I .............. "DOROTHY LI" .. . .. )00 "LIVED IN THE" ... "MIDST OF THE" Figure 1: Character classification versus character sequence classification. This is an interesting difference because high dimensionality is associated with poor classification learning-the so-called curse of dimensionality (Denker, et ali 1987; Geman, Bienenstock, & Doursat, 1992). OCR systems are designed to classify single characters to minimize such problems. The fact that most people learn to read quite well even with the high dimensional inputs and outputs, implies that variance