Purchase this article with an account.
ST L Chung, BS Tjan; Shift in Spatial Scale in Identifying Crowded Letters . Invest. Ophthalmol. Vis. Sci. 2002;43(13):4788.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: Crowding refers to the increased difficulty in identifying a letter flanked by other letters. Our earlier work (Tjan, 1996; Tjan, Chung & Legge, 2001) implies that the spatial-frequency characteristics of an ideal observer for letter identification is largely unaffected by crowding. The purpose of this study was to determine if the peak sensitivity of the human visual system shifts to a different spatial frequency when identifying flanked letters, compared with single letters. Methods: We measured contrast thresholds for identifying the middle target letters in trigrams, for a range of spatial frequencies, letter spacings and letter sizes. Letters making up each trigram were chosen randomly from all 26 letters of the Times-Roman alphabet, and were each digitally filtered with a set of 7 raised cosine log filters, with peak object spatial frequencies ranging from 0.88 to 7.07 c/letter. Bandwidth of the filters was 1 octave. All 3 letters of each trigram had the same object frequency and contrast. We tested 3 letter spacings (center-to-center separations): 0.8, 1 and 1.25X the x-height, as well as the unflanked condition (single letters). Four letter sizes, ranging from 0.2 to 0.8 log units larger than the acuity-letter, were tested. Results: Plots of contrast sensitivity vs. filter frequency exhibit spatial tuning, for all letter sizes and letter spacings tested. When expressed as retinal spatial frequencies (c/deg), the peak tuning frequency grows as the 0.7 power of the letter size, independent of letter spacing. The function relating peak tuning frequency and letter size shows an upward shift for the data obtained at 0.8X spacing, such that for the same letter size, peak tuning frequency occurs at a frequency that is about 0.4 octaves higher at the 0.8X spacing than for the other conditions. Conclusion: Contrary to the ideal-observer's characteristics, the most sensitive channel for letter identification in human observers is higher for flanked letters than for single letters. This finding suggests that the human visual system actively selects a higher spatial-frequency channel when identifying letters in the presence of nearby letters.
This PDF is available to Subscribers Only