Lewis Michael B, Hills Peter J
School of Psychology, Cardiff University, Cardiff, United Kingdom.
Department of Psychology, Bournemouth University, Poole, United Kingdom.
Front Psychol. 2018 Aug 20;9:1456. doi: 10.3389/fpsyg.2018.01456. eCollection 2018.
One explanation for the own-race bias in face recognition is the loss of holistic processing for other-race faces. The composite-face task (involving matching the top halves of faces when the bottom halves are either changed or the same) tests holistic processing but it has been inconsistent in revealing other-race effects. Two composite-face experiments are reported using pairs of faces that have common internal features but can be perceived as either being racially Black or White depending on their external features. In Experiment 1 (matching the top halves of faces) holistic processing was found for both face races for White participants (shown by both a mis-alignment advantage when bottom halves were different and also by a congruence-by-alignment interaction in discrimination). Bayesian analysis supported there being no effect of race. However, the size of the simple congruence effect was larger for own- than for other-race faces. Experiment 2 found that this race-by-congruence interaction was not present when matching the bottom halves of faces. The results are interpreted in of terms of the perceived race affecting the processing of second-order relational information rather than holistic processing.
人脸识别中本族偏见的一种解释是对其他种族面孔的整体加工缺失。合成面孔任务(当面孔下半部分改变或相同时,匹配面孔上半部分)测试整体加工,但在揭示其他种族效应方面一直存在不一致性。本文报告了两个合成面孔实验,使用的面孔对具有共同的内部特征,但根据其外部特征可被视为种族上的黑人或白人。在实验1(匹配面孔上半部分)中,白人参与者对两种面孔种族都发现了整体加工(当面孔下半部分不同时,通过错位优势显示,在辨别中也通过对齐一致交互作用显示)。贝叶斯分析支持不存在种族效应。然而,简单一致效应的大小在本族面孔上比在其他种族面孔上更大。实验2发现,当匹配面孔下半部分时,这种种族与一致性的交互作用不存在。结果被解释为感知到的种族影响二阶关系信息的加工,而不是整体加工。