Changing faces: visual and non-visual coding processes in face recognition.

Two experiments examined the effect of recognition accuracy and latency of changing the view of faces between presentation and test. In Expt 1, all the faces were unfamiliar to the subjects. Faces at test were either unchanged, or changed in angle (e.g. full face to 3/4), expression (e.g. smiling to unsmiling) or both. Unchanged faces were recognized more quickly and accurately than faces with a change in angle or expression which were in turn better than faces with both changed. In Expt 2, half the faces were highly familiar to the subjects, and at test unfamiliar and familiar faces were either unchanged or changed in both angle and expression. Unfamiliar faces were recognized more slowly and less accurately if changed at test, while familiar faces were recognized more slowly though no less accurately if change (though performance was effectively at ceiling). Familiar faces were recognized more quickly and accurately than unfamiliar, though false positive rates and rejection latencies were similar for familiars and unfamiliars. The results are discussed in terms of the combination of information from "pictorial', "structural', and "semantic' and "name' codes.