Does meaning matter? The impact of word knowledge on lateral masking

Background. Many older adults have lost the use of their foveae due to disease. Because of this, they must use peripheral retina to do tasks formerly done using the fovea, such as letter identification and reading. Several researchers have recently proposed that reading in the periphery is limited by lateral masking: the reduced perceptibility of a target when it is surrounded by flanking stimuli. None of these studies tested the perceptibility of letters within words, where the semantic relationship between the target letter and the flanking letters is likely to improve perceptibility. This study sought to determine whether the semantic relationship between the flanking stimuli and the target stimulus affects perceptibility of letters in the periphery. Methods. Young, normally-sighted observers identified single letters and the middle letters of three-letter words, trigrams, and letters flanked by x's (xax) presented for 150 ms at 10 degrees eccentricity on each of the primary meridia. In the first experiment, subjects identified letters presented alone and letters flanked by one x on either side. In the second experiment, subjects identified the middle letters of words, trigrams, and xax stimuli, which were either presented in blocks or randomly. In the third experiment, only the words and trigrams were used, and subjects were required to both identify the middle letter and simultaneously indicate whether the stimulus was a word or a trigram. Results. The first experiment showed that single letters were identified about 50% more accurately than letters flanked by x's. In the second and third experiments, letter identification was better when the target and flanking letters formed a word than when they formed a trigram and better along the horizontal meridian than along the vertical meridian; which is consistent with previous studies. In the third experiment, the subject's ability to identify the stimulus type had a significant impact on performance. When the stimuli were words, subjects identified the middle letter on 87% of trials when they correctly identified the stimulus and 40% of the trials when they did not; for the trigrams these percentages were 65% and 5%. Conclusions. When the flanking letters and the target letter formed a word, observers were better able to identify the target than when they did not form a word, and this pattern was mediated by the observers' knowledge of stimulus type. These data suggest that to draw conclusions regarding the impact of lateral masking on reading, the stimuli used should be words, not random-letter strings or other target/flanker combinations.