Simulating Mother–Child Interaction: Exploring Two Varieties of a Non-linear Dynamic Systems Approach

In theoretical analyses of early mother–child interaction, it has been argued that interaction should be studied in its flow over time, and that the behaviour of each interactant is likely to be non-linearly determined. The mother–child dyad can be seen as a non-linear dynamic system, the development of which is determined by the mutual relations between its elements. The present study is based on the idea that computer simulations can be used to find out what kind of empirical implication these ideas have. Accordingly, we describe two non-linear dynamic systems-based models for simulating mother–child interaction, i.e. a connectionist network model and a logistic growth model. Three determinants of the nature of the interaction, i.e. the child’s irritability, the mother’s sensitive responsiveness, and the intensity of an external stressor bothering the child, are varied systematically. Although the results of both simulations differed considerably, they shared the fact that small changes in stressor intensity produced abrupt changes from one type of interaction to another. In addition, increasing stressor intensity sometimes had the paradoxical effect of resulting in less, rather than more, distress on the side of the child. Though irritability and responsiveness were varied in a less fine-grained way than stressor intensity, the results suggest that similarly small differences in these dimensions at different parts of the dimension’s scale range have differentially strong effects on the nature of the interaction. It is concluded that these simulations help us to specify the nature of empirically researchable phenomena that are to be expected, given the assumptions listed above. Further elaboration of the models and comparison with longitudinal empirical data is needed to answer further theoretical and practical questions. Copyright © 2000 John Wiley & Sons, Ltd.