Traditional theories of cognition concentrate on the problem of stability, how it is that people perform the same cognitive act over and over despite different task pressures and varying circumstances. We suggest a different approach that focuses on the flexible and the inventive. From a dynamic systems perspective, intelligence is the adaptive flexibility that integrates the stability of past experience with the idiosyncrasies of the moment. In this paper we demonstrate the value of this perspective in the context of children's novel noun generalizations. We describe D-ALA, a dynamic systems model that embodies the principles of the account and use this model to explain the findings of three different experiments that show how young children flexibly and smartly adapt to novel circumstances in the task of learning new words.
Flexibility and Variability 3Flexibility and variability: Essential to human cognition and the study of human cognition Traditional theories of cognition concentrate on the problem of stability, how it is that people perform the same cognitive act over and over despite varying tasks and the vagaries of the moment. How is it, these theories ask, that we understand people, frogs, and trees to all be living things? How is it that we understand the frog in the pond, the frog in the jar, and the frog in the comic strip to all be the same kind of thing -a frog?Traditional theory explains these stabilities by positing stable knowledge structures; one constant internal representation is activated each time we understand some thing to be a living thing, another that is activated each time we understand something to be a frog. (Keil, 1994) This focus on stability as the principle phenomenon to be explained has led to a set of well-accepted practices in the study of cognitive development. These are:1. Concentrate on findings that are generalizable across tasks because it is the pattern of performance that transcends specific tasks that is most likely to be the result of, and thus informative about, the underlying representational constants.2. Seek tasks that minimize variability. By reducing variability, we can "see through" performance to the knowledge representations underneath.3. Emphasize optimal performance and the best conditions.Performance that comes closest to our theoretical definitions of the underlying knowledge (or to adult standards) present the clearest view These "good research practices" direct us to the stable, the general, the optimal. Using these methods, researchers of cognitive development have learned a lot about early cognitive competencies (e.g., Keil, 1989;Spelke, 1990;Mandler, 1992; Gelman, Coley, & Gottfried, 1994;Kemler Nelson, Frankenfield, Morris, & Blair, 2000;Bloom, 2000; Booth, Huang, & Waxman, 2005;Lavin & Hall, 2001). Nonetheless, this approach may be fundamentally flawed, both theoretically and empirically (see, Thelen & Smith, 1994;Smith, 2005). The theoretical failure lies in the concentration on the stable and the generalizable at the expense of the flex...