Flanker tasks based on congruency manipulation are biased measures of selective attention in perceptual load studies

Flanker tasks based on a flanker-target congruency manipulation are widely used in perceptual load studies to investigate under what circumstances task-irrelevant flankers may be processed. An implicit assumption underlying the congruency manipulation is that the three types of flankers (congruent, incongruent, and neutral) attract attention homogeneously. However, in the present study, we provide evidence to demonstrate that this assumption is wrong: We discovered that incongruent/congruent flankers attracted more attention than the neutral flanker did. To avoid this attentional bias induced by the flanker-target congruency manipulation, we developed a new flanker paradigm in which the extent of flanker processing was evaluated by comparing the threshold stimulus exposure durations (TSEDs) for successfully performing a target identification task when a task-irrelevant flanker was presented versus when the flanker was absent. The flanker was processed if the TSED was longer when the flanker was present than when it was absent. This new paradigm provides an unbiased measure of selective attention when neutral flankers are used. The present data, obtained with neutral flankers in the new paradigm, were consistent with the dilution theory of selective attention, but inconsistent with the perceptual load theory of selective attention.