Using computer-generated faces in experimental psychology: The role of realism and exposure

Abstract

In psychology, researchers often rely on face photographs to study face perception. However, finding suitable face stimuli for experiments is often challenging. Computer-generated (CG) faces have emerged as a potential solution due to their flexibility and controllability. However, it has been suggested that these stimuli are evaluated and processed differently from real faces, and their suitability as alternatives in experimental settings remains unclear. To address this, two studies were conducted to examine the impact of CG faces' realism and observers’ self-reported exposure to CG faces on faces appraisals (Study 1) and processing (Study 2).In Study 1 (n = 97), we assessed perceptions of both real and CG faces. Findings indicated that participants generally viewed CG faces less favourably, especially when these faces lacked realism. This trend was particularly pronounced among individuals less exposed to digital characters.In Study 2 (n = 33), we examined the recognition accuracy of these faces in a memory task. The data revealed that CG faces, especially those less realistic, were less accurately recognized. However, this discrepancy was primarily observed among individuals with limited exposure to digital characters, while those more familiar with such characters showed no significant difference in recognition.Overall, this work confirmed that, to date, CG faces are not an adequate alternative to real faces and that researchers should be cautious when using these stimuli in experiments involving face processing. However, as digital graphics improve and as digital characters become more commonplace in daily life, the perceptual gap between CG and real faces may diminish.

Keywords

• Computer-generated faces
• Face perception
• Uncanny valley
• Realism
• Exposure