This article lays out an alternative view of that which enables human competence, which is framed by complexity theory and draws on connectionism, and situated cognition. It suggests that developing the implications of such a view can provide the basis of a more fully articulated theory of instruction and that such a reconceptualisation may benefit the practice of education. What educators are ultimately after, I am convinced, is that-which-enables-competent-action. It is the desire to equip our students with what they will need to navigate successfully the world that motivates a teacher's activity. But that motivation is mediated by cultural beliefs as to just what enables competent action. The traditional framework shapes and constrains what is seen as sensible and even possible in the practice of education in problematic ways. Education has long identified `intelligence' as the basic capacity for competence and `reasoning' as the activity that generates competence. Historically we have tended to see these two as reciprocal: intelligence is the ability to reason, especially to reason formally, and formal rationality is the expression of intelligence. This linkage has led to particular forms of schooling based on derived notions of learning, thought, memory, and perception. But newer conceptions of our humanity (and not a few older ones, such as those of Dewey, Piaget, Wittgenstein, Merleau-Ponty, Vygotsky, and James) lead us away from the traditional set of ideas and towards a conception both more diffuse and more complex of what `enablers' of our action might be. Where the traditional framework focused attention on the individual, more recent formulations blur the boundaries between self and biology on the one hand and between self and world on the other. As a consequence, we are asked to confront a wider range of phenomena when we conceive of what happens during learning. Educational practices and educational theory have been shaped by (and have shaped) a traditional view of the mind. The traditional image of mind as a logic machine, instituted by Descartes and endorsed by the rationalists of the Enlightenment, has reached its logical completion in the popular conception of the computer and in the activity of the field of artificial intelligence (Al). Disregarding Descartes' dualism, Al researchers attempt to build machines that, logically enough, think. They have failed. But their failure, though predictable in outline by those who attend to the successes and failures of education in enacting a logical image of students, is wryly instructive in detail. Logicist images of mind have failed because they are too rigid and because they presume a complete knowledge of context before effective action can be taken (Dreyfus & Dreyfus, 1987). These two factors are related. The rigidity of AI's `intelligence' is due to the logical image that constitutes its basis. Objects of knowledge are known by their necessary and sufficient features and are related by the implicative rules of logic. For example: Axiom A and Fact B, related Logically, yield Implication C. But to be certain of the answer one must have all the facts right, for any difference that might make a difference in action must be described before deduction and the ensuing activity occur. The representational scheme of logicism is too unwieldy to be of much use in a world characterised by limited knowledge and categories that function by `family resemblance' (Triche, 1994; Wittgenstein, 1953). The world that we act within is a very different place. We recognise the educational expression of this theoretical position in the assumption that education consists of the acquisition of learning `objects'--facts and procedural rules that can be understood as facts. Such a position makes sensible the back-to-basics movement in many countries which emphasises the possession of content, understood as facts and basic procedures prior to and usually in place of any ability to put a student's understanding to use. …