The work reported in this thesis involved the exploration of 12 year old children's conceptions of inheritance. Results derived from interviews and from videotaped recordings of small group discussions, indicate that children's conceptions of inheritance are well developed before they are formally taught these notions in school. A series of open-ended problem solving tasks were designed to elicit student's conceptions and to facilitate group discussion. Results suggest that children's conceptions of inherited characteristics are heterogeneous and 'organized' in a highly flexible way. A research model, based on the notion of conceptual ecosystems, was developed to provide a framework for data analysis. Features of conceptual ecosystems are described. It is suggested that such systems are characterised by their 'openness', adaptiveness and resilience. Results suggest that the heterogeneity, flexibility and fluctuating character of such ecosystems confer on conceptions the ability both to transform well and to resist change well. The study describes how these characteristics of resilience and adaptiveness are displayed in the cognitive and social interactions of individual students. A theory of conceptual change is advanced which considers learning as a series of continuous qualitative changes made by the learner to existing personal conceptions. The significance of these 'metatransitions' is discussed in the light of existing teaching and learning strategies. It is suggested that conceptual change can be facilitated by helping students to make their existing conceptions explicit, co-active and interactive within conceptual ecosystems. The social and cognitive consequences of conceptual conflict, disagreement and consensus are described. It is proposed that an ecosystemic view of children's conceptions may help explain and overcome the difficulties experienced by students when they try to reconcile scientific concepts with their existing conceptions.