Social learning theory and the dynamics of interaction

The recent controversy between Bandura (1983) and Phillips and Orton (1983) about the causal relations involved in social interactions prompted a discussion of the proper role for formal models in the analysis of social interactions. I argue that (a) Bandura's attempt at formal modeling of patterns of causation is vague and misleading; (b) internal variables, such as expectation and self-monitoring, can easily be handled by formal models; (c) simple deterministic models can behave in unexpectedly complex ways, so cannot be ruled out in principle as explanations for social interaction; and (d) unaided verbal reasoning cannot hope to come to grips with the dynamics of even simple interacting systems.