Abstract
Attention plays a key role in binding features during perception. Is attention also necessary to maintain bindings in visual working memory? Luck & Vogel (1997) proposed that visual working memory stores features in integrated object representations, with no additional effort required to maintain feature bindings. In contrast, Wheeler and Treisman (2002) proposed that attention is necessary to maintain bindings but not individual features in working memory. The present study was designed to distinguish between these alternative hypotheses.
Experiment 1 used a color-orientation change detection task and investigated whether memory for bindings would be worse than memory for features when attention was divided among multiple items in the test array. Memory for bindings was found to be statistically indistinguishable from memory for features despite the presence of multiple test-array items, suggesting that focused attention is no more important for bindings than for individual features.
In Experiment 2, subjects performed an attention-demanding visual search task during the delay interval of the change-detection task. If the maintenance of bindings is uniquely dependent on the continued engagement of attention, then performance of the search task during the delay interval should greatly impair memory for feature bindings but should not impair memory for the individual features. The interposed search task produced a small but significant decrement in change-detection performance, but the decrement was nearly equivalent for feature memory and binding memory. Thus, the maintenance of bindings in visual working memory does not depend on the continued engagement of attention, consistent with the integrated object hypothesis.