Yong Hoon Chung¹ (), Jamal Williams², Viola Störmer¹; ¹Dartmouth College, ²University of California San Diego

The contents of visual working memory can incidentally guide attention towards matching features in the environment, and the magnitude of this effect scales with the strength of the memoranda (Williams et al., 2022). Other work has shown that working memory performance for colors is increased when they are part of meaningful objects (e.g., a blue shirt) compared to unrecognizable objects (Chung et al., 2023). Here, we test whether colors that are encoded as part of meaningful objects also guide attention towards a simple stimulus that matches this color, and whether the amount of guidance scales with memory strength. Participants were asked to remember colored silhouettes of real-world objects or scrambled versions of them and, after a short delay, indicate which color they saw using continuous reports. To assess attentional guidance, during the retention interval, a visual search display was presented and participants indicated the orientation of a single slanted line among vertical lines all embedded in separate colored circles. On half of the trials, all search colors were unrelated to the memory colors; on the remaining trials, one of the nontarget colors matched a color maintained in working memory. The difference in search response times between match and no-match conditions served as an index of attentional guidance. Across several experiments, we manipulated memory strength by changing encoding time, set size, and the recognizability (i.e., meaningfulness) of the stimulus. Consistent with previous research, we found that color memory was stronger with longer encoding, smaller set sizes, and for recognizable relative to unrecognizable objects. Importantly, the amount of guidance also scaled with memory strength. Broadly, this is consistent with recent studies suggesting that real-world contexts can enhance working memory for simple features and that interactions between working memory and attention depend on the representational fidelity of memories.