Abstract
Visual working memory capacity is greater for meaningful objects (e.g., trees) than both simple features (e.g., colors) and visually-matched meaningless stimuli (e.g., scrambled trees; e.g., Brady & Störmer, 2020). Here, we tested whether adding labels to ambiguous shape stimuli during encoding would improve subsequent memory performance for objects from the same category. Participants were shown three ambiguous shapes made from objects of the same category (e.g., three watches) for 1000ms, followed by a short delay period (1000ms) and a 2-AFC within-category discrimination test. In Experiment 1 (N=200), the shapes were black silhouettes of familiar objects that have been shown to be more recognizable when labels are provided (Koutstaal et al., 2003). In Experiment 2 (N=200), the shapes were warped images of real-world objects that were distorted just beyond recognition using the diffeomorphic scrambling technique (Stojanoski & Cusack, 2014). On each trial, half of the participants were prompted to “remember these shapes” and the other half were shown a category label (“remember these trees”) that could help participants recognize the shapes. In order to ensure participants did not rely on verbal strategies to perform the memory task itself, they rehearsed four digits out loud throughout each trial. Although we predicted the group that received the category labels would have better memory performance than the group that did not receive labels, memory performance (d’) did not differ between groups in Experiment 1 (p=.37; BF01=4.35) or Experiment 2 (p=.60; BF01=5.72). This suggests that providing labels to support recognition of ambiguous shape stimuli during encoding does not allow them to be memorized with enough additional detail to improve subsequent within-category discriminability. In future work, we will test whether providing category labels improves memory performance for a within-category discrimination test when the memory set consists of more distinct objects (from different categories).