Related flanking letters (e.g., RO ROCK CK) are known to facilitate processing of a central target word compared with unrelated flankers (e.g., PA ROCK TH) in the flanking letters lexical decision task. According to one account, these effects are driven by pre-orthographic visual features rather than orthographic information per se. Here we tested this preorthographic account, by comparing flanker effects with letter stimuli and flanker effects with symbol stimuli using the same-different matching task. We found significant flanker effects for letter strings in RTs and accuracy rates to "same" responses, as well as in discriminability indices (d') in a signal detection analysis. Crucially, we failed to find any significant impact of flanker symbols for responses to symbol targets, in all our dependent measures. The results contradict a low-level feature integration account of flanking letter effects and support the conclusion that they reflect the spatial integration of orthographic information.
Massol et al. (Tue,) studied this question.