In recent years researchers from a variety of cognitive science disciplines have begun to challenge some of the core assumptions of the dominant theoretical framework of cognitivism including the representation-computational view of cognition, the sense-model-plan-act understanding of cognitive architecture, and the use of a formal task description strategy for investigating the organisation of internal mental processes. Challenges to these assumptions are illustrated using empirical findings and theoretical arguments from the fields such as situated robotics, dynamical systems approaches to cognition, situated action and distributed cognition research, and sociohistorical studies of cognitive development. Several shared themes are extracted from the findings in these research programmes including: a focus on agent-environment systems as the primary unit of analysis; an attention to agent-environment interaction dynamics; a vision of the cognizer's internal mechanisms as essentially reactive and decentralised in nature; and a tendency for mutual definitions of agent, environment, and activity. It is argued that, taken together, these themes signal the emergence of a new approach to cognition called embodied, situated, and distributed interactionism. This interactionist alternative has many resonances with the dynamical systems approach to cognition. However, this approach does not provide a theory of the implementing substrate sufficient for an interactionist theoretical framework. It is suggested that such a theory can be found in a view of animals as autonomous systems coupled with a portrayal of the nervous system as a regulatory, coordinative, and integrative bodily subsystem. Although a number of recent simulations show connectionism's promise as a computational technique in simulating the role of the nervous system from an interactionist perspective, this embodied connectionist framework does not lend itself to understanding the advanced 'representation hungry' cognition we witness in much human behaviour. It is argued that this problem can be solved by understanding advanced cognition as the re-use of basic perception-action skills and structures that this feat is enabled by a general education within a social symbol-using environment.