Chocolate dandy bars can make you fat: Sentence context, word recognition, and repetition blindness.

摘要

When a sentence such as She ran her best time yet in the rice last week is displayed rapidly one word at a time on a computer screen, viewers sometimes misread rice as race (Potter, Moryadas, Abrams, & Noel, 1993). However, because the effect of biasing context is determined using an immediate recall measure, it is not clear whether context actually affects perception of the words in the sentence, or whether the effect only arises during recall. Repetition blindness (RB) is a deficit in reporting the second of two identical or orthographically similar words, relative to non-similar words, in rapid displays. When the misreading paradigm is combined with an RB paradigm, it becomes possible to determine whether selection of a word by biasing sentence context represents a genuine perceptual effect.; In Experiment 1, when participants misread We can prevent forest hires if he hires your mother as We can prevent forest fires..., RB affecting the downstream word hires was absent, just as with actual display of We can prevent forest fires. In contrast, a correct reading of the first hires generally resulted in non-report of the second hires. Experiment 2 replicated this finding with non-identical critical words. Experiments 3 and 4 demonstrated that misreading can also cause RB. When subjects misread Eat a piece of birthday rake and care much later as Eat a piece of birthday cake..., RB for the downstream word care was as large as when the word cake was actually present in the visual input. Experiments 5 and 6 demonstrated that RB resulting from a misread word and standard orthographic RB interact with temporal parameters in the same fashion. In Experiment 7, a recognition task was used as a test of the alternative hypothesis that RB results from reconstructive memory processes rather than from perceptual processes. The size of the RB effect was similar to that found in Experiment 4 using a standard recall measure. Taken together, the results suggest that the effect of prior context in rapidly displayed sentences occurs during perception, rather than resulting from reconstructive processes applied during recall of the sentence.