Object-based attention sustains feature binding in working memory without requiring strategic control.

Maintaining feature bindings in visual working memory (VWM) is thought to require object-based attention: consuming this resource during retention selectively impairs binding memory while sparing memory for the constituent features. However, existing demonstrations have almost exclusively used long (~ 1,000 ms) encoding windows, which permit not only rapid, automatic binding but also a slower strategic-control process that elaborates and stabilizes the initial representations. It is therefore unclear whether object-based attention is needed to maintain bindings as such, or only those bindings that strategic control has stabilized. Across three experiments, participants performed a change-detection task on color-shape conjunctions or on the constituent colors and shapes, while a concurrent Duncan's object feature-report task manipulated the demand on object-based attention (reporting features from two different objects vs. two features of the same object). To progressively rule out strategic control while holding the attentional manipulation constant, the memory array was presented for 200 ms (Experiment 1) or 100 ms (Experiment 2) before the secondary task, and, in Experiment 3, the memory array and the secondary-task stimulus appeared simultaneously, leaving no post-encoding window in which object-based attention could act on the memory representations unopposed. Across all three experiments, consuming object-based attention selectively impaired binding memory relative to feature memory, and the magnitude of this selective cost was statistically indistinguishable across experiments. The dependence of VWM bindings on object-based attention is thus not a downstream consequence of strategic stabilization during extended encoding, but a property of binding representations themselves. OPEN PRACTICES STATEMENT: The data, JASP analysis files, experimental programs, and stimuli for all experiments are publicly available on the Open Science Framework at: https://osf.io/r68vu/ . None of the experiments were preregistered.