The Asymmetric Mixed-Category Advantage in Visual Working Memory: An Attentional, Not Perceptual (Face-Specific) Account

The mixed-category advantage in visual working memory refers to improved memory for an image in a display containing two different categories relative to a display containing only one category (Cohen et al., 2014). Jiang, Remington, et al. (2016) found that this advantage characterizes mainly faces and suggested that face-only displays suffer from enhanced interference due to the unique configural nature of faces. Faces, however, possess social and emotional significance that may bias attention toward them in mixed-category displays at the expense of their counterpart category. Consequently, the counterpart category may suffer from little/no advantage or even an inversed effect. Using a change-detection task, we showed that a category that demonstrated a mixed-category disadvantage when paired with faces demonstrated a mixed-category advantage when paired with other nonfacial categories. Furthermore, manipulating the likelihood of testing a specific category (i.e., changing its task relevance) in mixed-category trials altered its advantaged/disadvantaged status, suggesting that the effect may be mediated by attention. Finally, to control for perceptual exposure factors, a sequential presentation experimental version was conducted. Whereas faces showed a typical mixed-category advantage, this pattern was again modulated (yielding an advantage for a nonfacial category) when inserting a task-relevance manipulation. Taken together, our findings support a central resource allocation account, according to which the asymmetric mixed-category effect likely stems from an attentional bias to one of the two categories. This attentional bias is not necessarily spatial in its nature, and it presumably affects processing stages subsequent to the initial perceptual encoding phase in working memory. Public Significance StatementThe present study investigated working memory (WM) for real-world images. People tend to better remember images taken from different categories than from the same category; however, this pattern characterizes only certain types of stimuli (e.g., faces). Here, we show that the beneficial memory for faces is not a category-specific effect (e.g., due to their holistic perception). Rather, it is an outcome of an attentional bias to faces, on the expense of memory for other categories. When measuring WM capacity, then, one needs to take into account potential biases that may occur at the encoding or the retention phase of WM.