Terms, proofs and refinement

We give a simple account of the connection between lambda terms and natural deduction proofs, showing how the terms can be rearranged into a form close to conventional proofs, and also to less conventional "top down" proofs. Creating proofs interactively by refinement can be seen as just keying in a lambda expression one symbol at a time in response to prompts from the machine. The aim is to convey some basic ideas to the uninitiated without technical or pragmatic detail.<<ETX>>