In six experiments in which a binary classification task was used, letter and nonletter (geometrical shapes, pseudoletters, or rotated letters) targets were presented either in isolation or surrounded by a geometrical shape. The surrounding shape could be congruent or incongruent with the target. When the classification required a distinction between letters and nonletters, either explicitly (Experiments 1-3) or implicitly (Experiment 4), a negative congruence effect was obtained for letters, contrasting with a regular, positive congruence effect for nonletters. When no distinction was to be made, letters and nonletters invariably showed a positive congruence effect (Experiments 5 and 6). In particular, between Experiments 1-4 and Experiments 5 and 6, the occurrence of negative or positive congruence effects for the same stimuli depended on the task. Feature interaction, target selection, and response competition explanations were tested against a feature integration approach. The results are explained in terms of different feature integration strategies for letters and nonletters.
CONGRUENCE IN LETTERS AND SHAPES909 sults of even these earliest studies cannot be understood fully on the basis of feature interaction alone. For instance, in Estes (1982), the effect of target-flanker similarity was minimal at an intermediate degree of similarity, a result Estes ascribed to the mediation of higher order processes. These processes are likely to involve task-specific factors.In particular, we may consider the role of task-specific factors in attentional target selection. Distractor information competes for selection when the perceiver is allocating attention to the target. This effect, too, will be stronger the more similar in shape the nontarget is to the target. When the stimulus is complex, the task is difficult, and distractors competing for attention may sometimes have to be suppressed while a target is selected. This provides another possible source of negative congruence effects. Accordingly, the time it takes to classify a target is longer if it is surrounded by nontargets of similar shape than if it is surrounded by ones of different shape (Bavelier, Deruelle, & Proksch, 2000;Briand, 1994;van Leeuwen & Bakker, 1995). These effects contrast with those usually obtained in more simple displays, where the time it takes to classify a target is shorter when congruent distractors surround it than when it is surrounded by incongruent ones. The latter, positive congruence effects have been studied extensively and are generally attributed to competition of late response-related processes (Eriksen & Eriksen, 1974). According to the literature, these effects are characteristic of simple stimuli, because when the task is easy (Lavie, 1995), attention recruited for feature integration (Treisman & Gelade, 1980) spills over to irrelevant stimulus attributes. When these call for a response opposite to that of the target, response competition will result.In sum, the effects of irrelevant surrounding information are commonly ex...