Detection and identification of crowded mirror-image letters in normal peripheral vision

Abstract

Performance for discriminating single mirror-image letters in peripheral vision can be as good as that in central vision, provided that letter size is scaled appropriately [Higgins, K. E., Arditi, A., & Knoblauch, K. (1996). Detection and identification of mirror-image letter pairs in central and peripheral vision. Vision Research, 36, 331-337]. In this study, we asked whether or not there is a reduction in performance for discriminating mirror-image letters when the letters are flanked closely by other letters, compared with unflanked (single) letters; and if so, whether or not this effect is greater in peripheral than in central vision. We compared contrast thresholds for detecting and identifying mirror-image letters "b" and "d" for a range of letter separations, at the fovea and 10 degrees eccentricity, for letters that were scaled in size. For comparison, thresholds were also determined for a pair of non-mirror-image letters "o" and "x". Our principal finding is that there is an additional loss in sensitivity for identifying mirror-image letters ("bd"), compared with non-mirror-image letters ("ox"), when the letters are flanked closely by other letters. The effect is greater in peripheral than central vision. An auxiliary experiment comparing thresholds for letters "d" and "q" vs. "b" and "d" shows that the additional loss in sensitivity for identifying crowded mirror-image letters cannot be attributed to the similarity in letter features between the two letters, but instead, is specific to the axis of symmetry. Our results suggest that in the presence of proximal objects, there is a specific loss in sensitivity for processing broad-band left-right mirror images in peripheral vision.

Figure 1

Contrast thresholds for detecting (unfilled symbols) and identifying (filled symbols) mirror-image letters “b” and “d” (bowtie symbols), and non -mirror-image letters “o” and “x” (circular symbols), are plotted as a function of letter separation. Data were obtained at the fovea. For each observer, all measurements (detection and identification) were obtained for the same letter size (see text for details). The letter sizes used were 0.117° for observers CV and VL and 0.146° for observer SC. The rightmost data points plotted in each panel represent thresholds obtained for single (unflanked) letters. Data for each observer are shown in separate panels, with the averaged data (AVE) shown in the lower right panel. Error bars represent ± 1 SEM.

Figure 2

Similar to Figure 1, with the exception that data plotted here were obtained at 10° eccentricity. Letter sizes used were 1.169° for observer CV, 1.079° for VL and 1.461° for SC.

Figure 3

Threshold ratios between identifying and detecting (I/ D ratios) letters of the pair “bd” (bowtie symbols), “ox” (circular symbols) and “dq” (hourglass symbols), averaged across three observers, are plotted as a function of letter separation. Note the difference in scale on the ordinate of the three panels. (A) Data obtained at the fovea for “bd” vs. “ox”. (B) Data obtained at 10° eccentricity for “bd” vs. “ox”. (C) Data obtained at 10° eccentricity for “bd” vs. “dq” (observers were different from those whose data are shown in panel B). Error bars represent ± 1 SEM.