关键词: 多图像优势, 面孔表征, 面孔匹配, 面孔识别

Abstract:

Presenting multiple face images of the same person has been shown to enhance a person's ability to recognize faces, which is known as multiple image advantage (MIA). However, MIA appears selective as evidenced by it manifesting in delayed face matching tasks but being absent in non-delayed face matching tasks. Two hypotheses are proposed to account for this selectivity. In one hypothesis, the delay after multiple face images are presented facilitates face representation formation, which strengthens participants’ face recognition. In the other hypothesis, it is proposed that the quantity of image information and task difficulty are key variables. That is, the more face images presented, the more information is processed, which improves recognition performance. Additionally, delayed face matching tasks are more difficult than non−delayed ones. Together, these yield MIA in delayed face matching tasks but the ceiling effect in non−delayed tasks.

In the current study, we conducted two experiments to investigate the cognitive mechanisms underlying MIA in face identity recognition. Experiment 1A adopted a face-matching paradigm where university students (N = 81) judged whether a target face matched a set of one, two, or three study faces presented simultaneously (non-delayed condition) or sequentially (delayed condition). In the simultaneous condition, target and study faces were displayed concurrently, allowing direct perceptual comparison; in the sequential condition, study faces were presented for 5, 000 ms followed by a 500 ms blank screen and the target face, requiring memory-based matching. In Experiment 1B, the simultaneous matching task was modified to enforce a mandatory 5, 000 ms viewing period for studying faces before allowing responses, which ensured identical exposure duration across both simultaneous and sequential conditions. This eliminated potential confounds arising from differences in processing time. Both experiments utilized unfamiliar face images, counterbalanced across tasks and participants, with familiarity screening applied post-test to exclude recognizable faces. In Experiment 2, we repeated Experiment 1 but inverted all faces to disrupt the integration processes of facial representation in the delay time.

The results of Experiment 1 show that: (1) MIA appears only when the images were presented sequentially as participants' discriminability improved when the number of images increased and (2) When presenting three face images, participants' discriminability under the sequential condition was higher than under the simultaneous condition. After controlling for duration of faces presented in Experiment 1B, results remain unchanged. For Experiment 2, (3) regardless of whether faces were single or multiple, participants' discriminability under the sequential condition was lower than under the simultaneous condition, and (4) no MIA appeared in either task. Altogether, these findings suggest that MIA in facial identity recognition in delayed face matching tasks is derived from the formation of facial representation that happens in memory.

Key words: multiple image advantage, facial representation, face matching, face recognition