Over the first year of life, infants gain conceptual skills which allow them to construe semantically related items as similar, even when they have few if any directly-perceived attributes in common. Moreover, this skill first encompasses only broad semantic categories, and only later extends to more subtle distinctions, when conceptual and perceptual similarity relations do not coincide. In this paper we suggest that a new mechanism must be added to the mix of possible bases for this observed developmental change. In agreement with many others, we suggest that infants’ earliest conceptual representations are organised with respect to certain especially useful or salient properties, regardless of whether such properties can be directly observed. However we suggest that in many cases this salience may itself be acquired, through domain-general learning mechanisms that are sensitive to the high-order coherent covariation of directly-observed stimulus properties across a breadth of experience. To support this argument we will describe simulations with a simple PDP model of semantic memory. When trained with backpropagation to complete queries about the properties of different objects, the model’s internal representations differentiate in a coarse-to-fine manner. As a consequence, different sets of properties come to be especially “salient” to the

-ls much of this book attests, a wealth of research provides evidence that tuman infants have a core capacity for representing objects and their motions. The environment contains a diversity of objects, however, with waried properties and behaviors. Objects such as pebbles and blocks are inert;:hev move or change only in response to an external force. Objects such as rutterflies and cars have internal mechanisms generating forces that can proptl them. Self-propelled objects can be further differentiated, according to *e nature and characteristic pattern of their motions and the circumstances that evoke theni. To navigate successfully in this diverse and changing environment, perceivers and actors must categorize the objects around them appropriately, determining what kind of thing each object is and how it is iikely to behave. Here we consider three general accounts of the development of this ability in humans. First, all human learning about objects may be supported by a;ingle core domain that identifies and tracks objects through space and time e.g., Flombaum et al., Chapter 6; but see Amso and fohnson, Chapter 9).-\s infants track pebbles and parrots, balls and cars, they may gradually learn that objects fall into general kinds with distinctive properties and behavior