Does “Item-Specific” Cognitive Control Operate at the Item Level?

People learn and retrieve cognitive control settings (e.g., attentional focus) associated with stimulus and contextual features. It has been theorized that control adjustments occur at the item level (e.g., for a specific picture) and the category level (i.e., for the overarching category represented by the picture), but evidence is lacking for the former. We aimed to determine whether control can truly operate at the item level. In Experiments 1–3, we manipulated item-specific proportion congruencies in a picture–word Stroop task while holding category-specific proportion congruencies constant at 50% congruent. One item in each animal category (e.g., Dog 1, Fish 1) was mostly congruent (MC) and one item (e.g., Dog 2, Fish 2) was mostly incongruent (MI). Item-level control (i.e., larger Stroop effect for MC items compared to MI items) was observed in Experiment 1, but neither Experiment 2 nor 3 replicated this finding. Experiments 4a and 4b used MC and MI categories, with each comprising both MC and MI items, allowing us to potentially index both levels of control. However, the findings indicated that control operated only at the category level and not the item level. Using novel stimuli, Experiment 5 showed Stroop effects differed between items that shared a response but were visually/conceptually dissimilar. This finding suggests that applying item-level control may be difficult when items within a category are visually/conceptually similar (as in Experiments 1–4). Collectively, our findings provided little evidence for item-level control; instead, the findings suggest control primarily operates at the category level in the picture–word Stroop task.