Decision makers are often confronted with complex tasks which cannot be solved by an individual alone, but require collaboration in the form of a coalition. Previous literature argues that instability, in terms of the re-organization of a coalition with respect to its members over time, is detrimental to performance. Other lines of research, such as the dynamic capabilities framework, challenge this view. Our objective is to understand the effects of instability on the performance of coalitions which are formed to solve complex tasks. In order to do so, we adapt the NK-model to the context of human decision-making in coalitions, and introduce an auction-based mechanism for autonomous coalition formation and a learning mechanism for human agents. Preliminary results suggest that re-organizing innovative and well-performing teams is beneficial, but that this is true only in certain situations.