Beyond ‘what’ and ‘how many’:

Memory is a must for thinking about objects. We frequently reason about objects even when we lack direct perceptual evidence of their existence, as when we saccade from one visual location to another, experience darkness, or observe occlusion. In all of these cases, object representations must be stored in memory in order to support even the most basic of computations—computations such as deciding whether an object is the same as one seen a moment earlier, or determining whether a hidden object can still be obtained. The richness of object representations as a case study in cognitive science is revealed by the diversity of chapters in this book. Our co-authors explore issues ranging from cross-species comparisons of object representations (Jordan & Brannon, Chapter 3) to implicit versus explicit knowledge of objects (Gjersoe & Hood, Chapter 13), to the representational consequences when tenets of object-hood are violated (Baillargeon et al., Chapter 12; Rosenberg & Carey, Chapter 7). Here, we offer the suggestion that none of these inquiries into the nature of object representations would be possible in the absence of working memory. In what follows, we explore the architecture of the working memory system that supports object representation throughout development. Specifically, we suggest that although recent research has addressed the question of how many objects working memory can represent—both in adults and in infants—much less work has asked how this memory capacity is affected by the nature of the object representations themselves. The question of how the quantity and the quality of object representations interact has been particularly underexplored from a developmental perspective. Here, we offer some data from our work with infants that may serve as a first step toward filling this gap. We proceed in three parts. First, we review