Abstract
Purpose :
Visual span refers to the number of letters recognized reliably without moving the eyes. Reduced size of the visual span limits reading speed in patients with central vision loss (Legge et al., 2001). Typically, visual span is measured with a trigram task (identifying random strings of three letters). The middle letters of the trigrams are thought to represent crowded interior letters of words. In daily reading, words often contain more than three letters, and their interior letters may suffer more crowding than the middle letters of the trigrams. Here we investigate how visual span changes as a function of string length.
Methods :
Five normally-sighted subjects recognized letters presented at 9 various positions left and right of the midline at the fovea and 10° eccentricity in the inferior visual field. Stimulus contained 1, 2, 3, 4, 5 or 6 random-selected letters. Print sizes were larger than the critical print size at each eccentricity. To minimize nonvisual cognitive influences such as memory demand, partial report (i.e. recognizing one letter on each trial) was used.
Results :
We found that visual-span size was at the ceiling for single letter recognition, and decreased linearly with increasing string length starting from two letters. The slope of the best fitted line was -2.82±0.32(SE) bits per letter at the foveal and -3.20±0.23 at the 10° eccentricity. By comparing the results with our previous studies, we found that visual span measured with trigrams was larger for partial report than for full report (e.g., Yu et al., 2007; 2010). The difference was small at the fovea (~3%) and larger at the 10° eccentricity (~20%), indicating that nonvisual cognitive influences on letter recognition increase with increasing eccentricity.
Conclusions :
In summary, to reflect the level of crowding in the printed text, we may need to use longer letter strings to measure visual span in both central and peripheral vision. Partial report should be used to minimize the task-specific cognitive influences.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.