The goal of this paper is to argue for the importance of ‘meta-theories’, or "theories about how to build theories", in action research. Meta-theories express the fundamental assumptions about the world that underlie a theory and influence our ability to combine knowledge and communicate. Action research was originally based on "field theory", a meta-theory developed by Kurt Lewin but largely abandoned by his followers. One of the few meta-theories to have emerged from action research is "action science", which developed by Chris Argyris and Donald Schön. Although there appears to be little similarity between the two meta-theories (i.e. field theory and action science), this paper argues that they actually complement each other. It suggests that integrating field theory and action science into a unified meta-theory can provide action researchers and practitioners with concepts that can enable them see their "behavioral world" in ways expand the range of possible goals, actions, and relationships.