Abstract
Recent studies have shown that visual working memory capacity is increased when more meaningful features can be extracted from a stimulus, even from perceptually identical stimuli (e.g., perceived vs. not-perceived Mooney faces; Asp et al., 2021). But is extracting more features the only benefit of meaningful stimuli? In a series of experiments, we investigated whether memory differs for the exact same single feature dimension (color) when this feature is present in recognizable objects vs. not-recognizable shapes. Participants (N=30) were presented with an array of either 4 colored real-world objects or unrecognizable scrambled versions of these objects for 1s. All stimuli were randomly colored by rotating them in hue space (Brady et al., 2013), so object identity and associated additional features provided no information about their colors. After a 1s delay, participants performed a 2AFC, indicating which of two colors were presented on a probed object during encoding (the target or a maximally distinct foil color). Memory performance was better when colors appeared on realistic objects relative to the unrecognizable images (p<0.001). We replicated this finding using upright objects vs. less recognizable upside-down images (N=30; p<0.001). Next, we added a spatial cue during the 2AFC task that participants could rely on to retrieve color information, and still found a real-world object benefit for color (N=30; p<0.05), showing this benefit holds even with retrieval cues other than object identity are available. Overall, our results demonstrate that visual working memory capacity even for simple visual features like color is not fixed. Contextual information, like how meaningful the stimuli is that the color is embedded in, changes how well we remember simple visual features. Thus, the benefit of realistic objects on visual working memory extends beyond additional extracted features: recognizing an object also scaffolds your memory for individual features of that object.