Purchase this article with an account.
Jean-Baptiste Bernard, Daniel Coates, Susana Chung; Target and flanker perception are related in crowded letter identification. Journal of Vision 2011;11(11):1141. https://doi.org/10.1167/11.11.1141.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Crowding refers to the inability to recognize an object in the presence of other objects. One account of crowding is that it results from an inappropriate integration of features from the target and its flankers. There is evidence that the inappropriate integration is partly due to feature migration – features from the flankers are mislocalized from their veridical locations and are incorrectly considered as part of the target. This theory predicts that when several objects are present together, parts of each object could migrate, causing a simultaneous failure to recognize different objects. To test this prediction, we measured letter identification accuracy in seven normally-sighted observers. On each trial, three lowercase letters rendered at normal letter spacing were presented at 10° below fixation. Observers' task was to name the three letters. We analyzed the trials with respect to the presence of four basic letter structures (ascenders, descenders, round and oblique segments) in the stimuli and observers' responses. Across 44100 trials, the accuracy of identifying the flanked middle letters was 64% when both flankers were correctly identified, 36% when only one flanker was correctly identified and 20% when neither flanker was correctly identified. Further, for a pair of incorrectly identified adjacent letters, if the response to one of the letters did not include a specific letter structure that was present in the stimulus, this structure was found in the response to the adjacent letter in 59% of the trials. In contrast, when the response letter contained the correct letter structure, this structure was present in the response to the adjacent letter in only 19% of the trials. These results are consistent with the feature migration account of inappropriate feature integration and suggest that the identification of different objects in a crowded scene consists of an attempt to assemble features that are ambiguously localized.
This PDF is available to Subscribers Only