Recognition Memory for Photographs of Faces

Recognition memory was compared for faces seen in positive, in negative, and in two different expressions. Pure pattern characteristics plus those properties that differ in smile and resting state cannot fully account for the accuracies obtained. But the characteristics that differ in smiling and nonsmiling expressions did play some role in recognition, which suggests that expressional variation offers a technique for investigating the properties by which faces are discriminated and remembered. Hochberg and Galper (1967) demonstrated that although subjects can recognize still photographs of faces with surprising accuracy, recognition is significantly poorer for inverted faces, a finding corroborated recently by Yin (1969). Galper (1970) found that recognition accuracy was also impaired when photographs of faces were presented in negative. These results indicate that recognition accuracy for faces is better than for patterns of identical geometrical complexity (i.e., the inverted or negative photographs) and that this superiority rests at least in part on information that becomes inaccessible when a photograph is inverted or its brightness relations reversed. More subtle (and more natural) transformations may also affect face recognition, to the extent that viewers rely on those characteristics that are altered by changes in expression. If viewers could not identify which of two different expressions of the same person was previously presented, this would be evidence that those changes in pattern associated with the difference in expression do not include the characteristics which mediate recognition. As a step toward determining the stimulus attributes in terms of which photographs of faces are perceived and remembered, therefore, the present experiment investigated the effect of an expressional variation (resting state versus smile) on recognition accuracy, and compared that effect with the accuracy obtained using previous procedures.