From the Lexicon to Expectations About Kinds: A Role for Associative Learning.

Abstract

In the novel noun generalization task, 2 1/2-year-old children display generalized expectations about how solid and nonsolid things are named, extending names for never-before-encountered solids by shape and for never-before-encountered nonsolids by material. This distinction between solids and nonsolids has been interpreted in terms of an ontological distinction between objects and substances. Nine simulations and behavioral experiments tested the hypothesis that these expectations arise from the correlations characterizing early learned noun categories. In the simulation studies, connectionist networks were trained on noun vocabularies modeled after those of children. These networks formed generalized expectations about solids and nonsolids that match children's performances in the novel noun generalization task in the very different languages of English and Japanese. The simulations also generate new predictions supported by new experiments with children. Implications are discussed in terms of children's development of distinctions between kinds of categories and in terms of the nature of this knowledge.Concepts are hypothetical constructs, theoretical devices hypothesized to explain data, what people do, and what people say. The question of whether a particular theory can explain children's concepts is therefore semantically strange because strictly speaking this question asks about an explanation of an explanation. We begin with this reminder because the goal of the research reported here is to understand the role of associative processes in children's systematic attention to the shape of solid things and to the material of nonsolid things in the task of forming new lexical categories. These attentional biases have been interpreted in terms of children's concepts about the ontological kinds of object and substance (e.g., Dickinson, 1988;Imai & Gentner, 1997;Soja, Carey, & Spelke, 1992;Subrahmanyam, Landau, & Gelman, 1999). These concepts and the notion of a psychological ontology are theoretical constructs offered within the framework that posits propositional representations.In the simulations and experiments reported here, we show how abstract distinctions about different kinds of categories may be made through associative learning and the patterns of correlations between the perceptual properties of things and words. The theoretical explanation we offer and the experiments we report do not explain the hypothetical constructs of object and substance nor do they fully replace these constructs by explaining all of the data that have been subsumed under them. Nonetheless, the results do show how simple associative processes may create abstract distinctions about different kinds and, in so doing, these processes may play a crucial role in children's category and lexical learning.

Connectionist modeling of developmental changes in infancy: Approaches, challenges, and contributions.

“…The implication of this prediction was that visual similarity of nonsolid items of the same material can support generalization, but vocabulary exposure is needed for shape-based generalization to emerge. Colunga and Smith (2005) administered the noun generalization task to a sample of younger children and found empirical support for this prediction: Children who had not yet learned many words could generalize the names of nonsolid items by material.…”

“…For example, Rakison and Lupyan's (2008) simulation showed that long-term experience with objects in the real world can constrain the feature correlations that children can learn in the lab. Similarly, Colunga and Smith (2005) have shown that linguistic experience, as evidenced by vocabulary size, can constrain the manner in which children generalize novel nouns such that they learn to use different properties to generalize about different types of objects. Both simulations highlight the emergence of complex constraints on behavior from domain-general associative learning as long-term accumulated weights affect the moment-to-moment behaviors.…”

“…A final example of a connectionist model that contributed to theory building through the instantiation of the hypothesized mechanism is the work of Colunga and Smith (2005) on novel noun generalization for solids and nonsolids based on shape and material, respectively. The theoretical account for the development of generalization emphasized the role of the statistics of the language to which children are exposed: the correlations between solidity, shape, and material for nouns in the language were theorized to drive performance in the novel noun generalization task.…”

“…An example of an empirically supported developmental prediction was generated by Colunga and Smith's (2005) simulation of a noun generalization task. The model predicted that children who are in the initial stage of word learning would extend labels of nonsolid test items but not of solid test items.…”

The Name Pathway, an Elementary Cognitive Model of Names as Sensory Representations

“…A large body of empirical and theoretical literature on word acquisition exists (Colunga and Smith, 2005;Regier et al, 2005;Garagnani et al, 2008;Cummings et al, 2009;Frank et al, 2009;Mayor and Plunkett, 2010). Many proposals have advocated a role for symbols and names as sensory representations (Harnad, 1990;Humphreys et al, 1999;Feldman and Narayanan, 2004;Sheridan, 2005).…”