Subjects tend to falsely recognize new faces (or words) when they are constructed by recombining parts of previously studied faces (or words). The experiments that I will present test whether the same underlying mechanisms give rise to these errors for the two stimulus types. In addition, they show that patterns of attention during the study phase play a central role in producing these recognition errors. I‰ll briefly sketch out a theoretical explanation for miscombination errors for faces and words, in which familiarity and intentional recollection both play roles.