Abstract
Theories of adaptive (and cognitive) control assume that selecting target information in the context of highly salient distractors depends on limited-capacity resources. Building on this assumption, theories of language-control associations propose that the opportunities afforded by bilingualism to engage such effortful selection, such as when speaking in a nondominant language, might improve domain-general adaptive control. The assumption that domain-general or language-specific selection is resource-demanding, however, has surprisingly little empirical support. Here, we tested that assumption by having unbalanced Italian-English bilinguals perform both an L1 Stroop task and an L2 picture-naming task simultaneously with an n-back task. Both tasks showed costs due to the load produced by the n-back task and distractor interference, with slower responses to incongruent (the word GREEN in the color red) and congruent stimuli (RED in red) than neutral ones (XXX in red) in the L1 Stroop task and to noncognate than cognate pictures (pictures with different/similar L1 and L2 names) in the L2 picture-naming task. However, neither task showed larger distractor interference with greater load, with Bayesian analyses favoring the absence of such interactions. These results suggest that domain-general and language-specific selection may occur reactively, i.e., only when the need arises, with no strong reliance on limited-capacity resources. Further, they invite a rethinking of both adaptive-control theories assuming resource-demanding selection and theories of language-control associations assuming that regularly engaging such selection would be conducive to domain-general benefits.