Nonlinear dynamic systems provide a powerful new framework for analyzing development and other forms of change, but two research approaches have limited research to date. Data-driven approaches have focused on describing specific phenomena, especially actions, and have used dynamic concepts mostly as loose metaphors instead of building explicit models. Model-driven approaches have focused on the rich hypotheses in developmental theory to generate and explore formal models of change processes, mostly involving cognition and language. They have neglected the need for careful research to measure the growth patterns to be explained. The difficulties of doing dynamic research on cognitive, language, and socioemotional development stem in large part from the absence of well constructed scales for assessing behaviors other than actions. Construction of such scales is facilitated by combining scores on carefully analyzed tasks and by assessing scale properties across different assessment conditions, so as to separate growth properties from scale anomalies. With good scales, models and data can be used in dynamic interaction to generate powerful explanations of development, as illustrated by models of hierarchical growth and predator-prey relations in cognitive and brain development.

原文地址：http://gseweb.harvard.edu/~ddl/articlesCopy/RoseFischer-Models.1998.pdf