The dominant assumption about the causal architecture of the mind is, that it is composed of a stable set of components that contribute independently to relevant observables that are employed to measure cognitive activity. This view has been called component-dominant dynamics. An alternative has been proposed, according to which the different components are not independent, but fundamentally interdependent, and are not stable basic properties of the mind, but rather an emergent feature of the mind given a particular task context. This view has been called interaction-dominant dynamics. In this paper, we review evidence for interaction-dominant dynamics as the causal architecture of the mind. We point out, that such an architecture is consistent with problems of convergence in research on the level of results and theorizing. Moreover, we point out that if interaction-dominant dynamics as the causal architecture of the mind were to be true, this would naturally lead to (some degree of) problems with generalization and replicability in sciences of the mind and brain, and would probably warrant changes in the scientific practice with regard to study-design and data analysis.