Negative effects of redundant targets

The visual system can encode many stimuli simultaneously, but there are limits to howwell multiple objects can be identified in parallel. At the extreme, some objects mighthave to be identified serially. The redundant target paradigm is one tool fordistinguishing specific parallel and serial models. It compares responses to displayscontaining one target versus displays containing two targets. The typical result is apositive redundant target effect: faster correct responses to two targets, as predicted bymany parallel models. Here we generalize three standard models to account for responseaccuracy as well as speed. Surprisingly, two models predict a reversal of the redundanttarget effect (slower responses to two targets than to one target): the generalized standardserial model, and a specific form of a fixed-capacity parallel model. To test that prediction,we measured performance for three different judgments of written words: colordetection, lexical decision, and semantic categorization. The color task yielded positiveredundant target effects, which reject the standard serial model. The semantic taskyielded consistently negative effects, which are consistent with either the standard serialmodel or some limited-capacity parallel models. Thus, redundant targets can havenegative effects, and they demonstrate limits that impair simultaneous recognition of twowords.