Learning to Identify Human Faces L. R. Betts, P. J. Bennett, and B. B. Sekuler Abstract The current experiment investigates the mechanisms that underlie the perceptual learning of spatial frequency information in facial stimuli. In particular, the authors were interested in determining whether training on the first day with faces that contain a specific range of spatial frequency transferred to the same faces that contain a different range of spatial frequency information on the second day of testing. The results indicate that the amount of transfer between different spatial frequency conditions was very limited, and that in some cases, the performance in the second condition actually dropped below pre-training levels. Furthermore, extensive experience with all pass faces improved observers’ abilities to recognize low pass (2-7 cycles per image), medium pass (8-31 cycles per image), and high pass faces (32-127 cycles per image), but a substantial portion of this improvement was due to familiarity with the task demands.