Abstract
Large performance benefits in face processing for familiar compared to unfamiliar faces are often interpreted as reflecting qualitative changes in processing. However, directly testing this claim has proven difficult because of variation in tasks, stimuli and performance scales used to measure unfamiliar and familiar face recognition. Here we introduce a new method to compare familiar and unfamiliar face identity processing abilities, using the same task, and the same faces across experimental conditions. Familiarity was manipulated naturally, based on participants' viewing history for one of two popular TV shows, such that characters from one show were familiar, and the other unfamiliar. Prior exposure to characters led to improved identity matching performance for actors on that show, confirming effective manipulation of familiarity, but our method avoided floor and ceiling effects which have made previous research hard to interpret. Across two experiments, we report strong and reliable associations between viewers' accuracy on familiar and unfamiliar face matching. We conclude that despite the considerable accuracy differences typically observed between tests of familiar and unfamiliar faces, they appear to share underlying processes. We discuss these results in terms of the debate about the degree to which quantitative or qualitative differences best characterise the relationship between familiar and unfamiliar face processing.