Abstract
A key question in the development of working memory is whether children's capacity increases in the number of objects or features. Prior research (Forsberg et al., 2022) indicated that object-based capacity limits predominate, with relatively small improvements in feature retention. However, the procedure in that study included only two features per object, aside from location. In contrast, we tested memory for pairs of four-featured objects with children aged 8-9 years or 10-12 years and young adults (total N = 101). Quantitative modeling of the results showed that feature richness plays a much larger role in working memory development than previously thought. Each object was a stylized arrow with a color, an orientation, an emblem on the arrowhead, and a shape of the stalk of the arrow, an analogy to many real-world objects (e.g., a shirt with a color, current orientation, pocket emblem, and shape of the shirt tail). The analyses separated errors due to attention lapses in which an object was not encoded and errors due to incomplete retention of features within an object. Our findings suggest a two-step model of working memory encoding in which objects are first encoded (with some lapses or unencoded objects), followed by encoding of the multiple features of these objects independently. With these feature-rich stimuli, we showed feature encoding to be a major source of developmental improvement, along with object encoding. These findings challenge prior assumptions and provide new insights into how working memory for realistic, complex objects may develop.