Three experiments are reported in which the effects of viewpoint on the rec
ognition of distinctive and typical faces were explored. Specifically we in
vestigated whether generalization across views would be better for distinct
ive faces than for typical faces. In Experiment 1 the time to match differe
nt Views of the same typical faces and the same distinctive faces was depen
dent on the difference between the views shown. In contrast, the accuracy a
nd latency of correct responses on trials in which two different faces were
presented were independent of viewpoint if the faces were distinctive but
were view-dependent if the faces were typical. In Experiment 2 we tested pa
rticipants' recognition memory for unfamiliar faces that had been studied a
t a single three-quarter view Participants were presented with all face vie
ws during test. Finally, in Experiment 3, participants were tested on their
recognition of unfamiliar faces that had been studied at all views. In bot
h Experiments 2 and 3 we found an effect of distinctiveness and viewpoint b
ut no interaction between these factors. The results are discussed in terms
of a model of face representation based on inter-item similarity in which
the representations are view specific.