Research in face recognition can broadly be divided into those projects concerned with front-end image processing and those projects concerned with memory for familiar people. The problem for image-processing projects is to derive a label from a particular image, perhaps through computation of some characteristic aspects of that image which are shared by known characteristics of a target. The problem for those studying memory for familiar people is to understand how we can know that this person before us is (for example) a colleague, a family member or a TV personality. These two areas of research have proceeded largely in parallel, with only limited mutual influence. However, empirical studies suggest that there is much to be gained from a greater interaction between these fields, and it is possible that progress can only be made through formal interaction.