The first section of this paper comprises brief analyses of conceptual change theory, as originally formulated by Posner, et al., and the generative model of learning. Both models fail to explain exactly how students' acquire new concepts; in particular, in the former model, the process of learning appears simply to be taken for granted. These models recapitulate the learning paradox (Bereiter, 1985), which arises when claims are made about novel thought despite the lack of an 'intrinsic generative mechanism' that makes this thought possible (Juckes, 1991). The second section reviews Stenhouse's (1986) suggestion for a solution to the problem of how conceptual change takes place. Stenhouse adopts Wittgenstein's (1963) concept of the 'language game' and recommends that science teachers use children's 'own language games' to facilitate change. The main purpose of the third section is to promote a neuroscientific theoretical basis for a resolution of the learning paradox, and, in particular, to show how this 'non- psychological' theory predicts the successful use of analogies in promoting conceptual change (e.g., Brown, 1992; Duit, 1991). While aspects of Stenhouse's framework are problematic, several parallels between the latter and the former theory are briefly examined.