Hemispheric asymmetries in feature integration during visual word recognition

Abstract

Although the definitive source of the left hemisphere's superiority for visual word recognition remains illusive, some argue that the left (LH) and right (RH) hemispheres engage different strategies during early perceptual processes involved in stimulus encoding. In particular, it is proposed that the LH treats a word as a unitary perceptual group whereas the RH processes the letters comprising a word as a series of individual perceptual units. The present study investigated support for this processing distinction by examining hemispheric strategies for temporal integration using Prinzmetal and Millis-Wright's (1984) feature-binding paradigm. A total of 20 participants identified the colour and identity of a target letter, presented within a three-letter word (e.g., ART) or nonword (e.g., HRF), directed to their left or right visual field. Errors were classified on the basis of whether they involved substitution of a colour present within the stimulus but at a different location (ON error), or the substitution of a colour not present within the stimulus (OFF error). As anticipated, for word stimuli there was a higher proportion of OFF errors associated with trials directed to the RH, consistent with the notion that the LH treats words as single perceptual units and is hence biased toward miscombination of perceptual information present within the stimulus. The pattern of ON errors across stimulus type provided clear evidence of RH sequential encoding effects, with the number of errors increasing markedly across the ordinal position of the letters comprising the stimulus string. As such, these data provide new evidence that the LH's advantage for visual word recognition arises, at least in part, from the ability to encode verbal stimuli as single perceptual units.

Acknowledgements

Isabel Arend and Robert Ward are supported by grant C501417 from the Biotechnology and Biological Sciences Research Council. Many thanks are extended to Michael Peters and an anonymous reviewer for their helpful comments on an earlier version of this paper.