Reading performance was also measured with text that was defined by differing amounts of color and luminance contrast, in order to determine the influence of color information on reading. Results indicated that when luminance contrast was well above threshold, varying the chromatic (color) contrast had little effect on reading performance. However, when luminance contrast was very low, near – threshold, chromatic contrast sustained reading rates of nearly 300 words per minute, almost as high as those found with high luminance contrasts. On the other hand, for some low vision observers, text defined by color contrast interfered with reading performance. Further investigations are planned to determine why the reading of some low vision observers is adversely affected by color contrast that is not accompanied by sufficient luminance contrast.
The National Institute on Aging, with Mars Perceptrix Inc., were working to develop an adjustable font software for users with low vision. The software allows the adjustment of many font parameters in real-time, so that the user can see the impact of parameter adjustments on legibility as it happens. The prototype software works, but evaluation using participants with low vision suggested that font legibility could not be significantly improved above and beyond that provided by highly evolved and readily available fonts such as TrueType Arial and TrueType Times New Roman (Arditi, in press).
By comparing the letter confusions observed under wide and narrow inter-letter spacing, they were able to determine that a large portion of the deterioration of legibility under narrow spacing conditions could be attributed to unique letter confusions which did not occur when inter-letter spacing was wider. The cause of these unique letter confusions was lateral interference (inhibition) from neighboring letters.
The human observers not only made mistakes in identifying closely packed small letters, but they also misjudged the number of letters in the string. Using interlaced four-letter and five-letter strings, Drs. Arditi and Liu demonstrated that human observers tended to mistake more five-letter strings for four-letter strings when the inter-letter spacing was narrowed. Typically, the observers would either omit one of the three letters in the middle of the five-letter string, or combine two neighboring letters into a new letter. The researchers used a computer simulation to demonstrate that an optical blur of the eye might have played an important role in this new aspect of the "crowding effect".
Visual acuity is measured as a function of the separation between the Landolt C and the flanking bars. The inhibitory effect of the flanking bars is demonstrated as reductions of visual acuity at certain separations. Drs. Liu and Arditi studied the effects of contrast polarity by measuring contour interaction between a black C with four white bars, and a white C with four black bars. They found that features of different contrast polarities were still engaged in inhibitory interaction, although the interaction appeared to be weaker than that observed with features of the same contrast polarity. Therefore, a simple linear receptive field model was not applicable to suprathreshold contour interaction.
Flanking bars that were parallel exerted the strongest inhibition at the narrowest separation between the Landolt C and the bars. At separations narrower than two gap widths, the inhibition caused by a pair of parallel bars was stronger than that observed when all four bars were present. The orthogonal bars only exerted moderate inhibition at wider separations. It appeared that when four flanking bars were present, the orthogonal bars alleviated the inhibitory effect of the parallel bars at narrower separations and enhanced it at wider separations.
Such a deficit could affect reading performance by interfering with the discriminability of letters having the same spatial frequency content but different spatial phase spectra (i.e., mirror image letters like "b" vs. "d”). However, they found that when letters were size-scaled to compensate for differences in contrast sensitivity, the relationship between detection and identification performance was the same in both the central and peripheral retinas. These results thus argue against the hypothesis that the poorer reading performance outside the fovea is, somehow, due to reduced letter discriminability that might occur secondarily to a loss of peripheral-retina phase sensitivity.