Relatively few studies have examined the time course by which unfamiliar faces accrue familiarity, and those that have rely upon ‘internal feature advantage’ for familiar faces, requiring that central or external facial features are selectively ablated (Young et al 1985, Perception 14). In the present study, a procedure was used that required observers to seek a pre-cued face from among a number of alternatives, akin to finding a face in a crowd. Two factors were examined across 450 trials divided into blocks of 50 trials, run in counterbalanced order with 24 observers: viewpoint (the use of front and 30 deg yaw face stimuli at learning and search); and search target familiarity, being either unfamiliar (changed from trial-to-trial), learned (re-used from trial-to-trial), or familiar a priori (famous). In common with several existing studies, we found that the learned face condition yielded significantly decreased search time relative to unfamiliar faces (p<0.05) and eliminated viewpoint effects, a hallmark of familiarization. However, performance did not approach that of the a priori familiar faces, suggesting an incomplete transition to a familiar representation across the number of trials used. Surprisingly, once generic practice effects were removed, residual performance differences showed that search times for unfamiliar and learned conditions diverged sharply within the first few trials and that no further separation occurred, indicating that a small number of brief exposures to a new face are sufficient to yield a significant search performance advantage, but that a longer period of learning would be required to encroach upon familiar face performance.
