Race effects in facial recognition: the evolution

Abstract

Introduction: In our everyday life we see a large number of faces that share a set of basic elements, making it necessary to distinguish those that we already know from those we know for the first time. This is particularly important in crimes where victims need to identify their offenders, making facial recognition an important theme in Forensic Psychology [1]. Over the time, there are more and more studies that attach greater importance to the face, which is increasingly considered an important aspect for human interaction since it provides a large amount of information that should be analyzed efficiently [2]. It is known that we have more difficulties in recognizing a face from a different race of our own, and this variable has been designated as race bias, cross race effect, other race bias and own race bias.

This paper aims at a better understanding of race effects in face recognition from the 80’s until today as it is still an understudied but relevant subject to Forensic Psychology.

Materials and methods: For this study we included articles that focused on the effect of race on face recognition. To that end, studies were included from the 80’s until 2018, written in Portuguese, English and Spanish. Quantitative studies were included. Finally, the research was limited to articles published in peer-reviewed journals/journals limited to full-text articles. The articles were selected by reading the title and the respective abstract.

The included articles are available in full text in the EBSCOhost Web, B-On and SAGE Journals databases. The articles were found by combining the following descriptors: “other race bias”; “own race bias” and “eyewitness AND cross-race identifications”.

After applying the inclusion criteria, a total of 510 articles were found (already with the duplicates removed). Of the total number of articles 497 articles were excluded. These were excluded for the following reasons: they were not available for reading (n = 257); are not quantitative studies (n = 113); do not report the effect of race on face recognition (n = 77); were not relevant to forensic psychology (n = 50).

Results: In this review, 13 empirical articles were included, most of them being carried out between 2004 and 2016, reinforcing the importance given to this issue today. The absence of studies between 1990 and 2004 could be due to a possible lack of interest in the subject during this time. Most studies are distributed across Europe (n = 7) and North America (n = 5) and only one in Asia (Zhou, Pu, Young, & Tse, 2014). The total number of participants is 1493, with the lowest being 40 participants (Brigham, Maass, & Martinez, 1983, Yan, Andrews, Jenkins & Young, 2016) and the largest being 304 participants (Lawson & Dysart, 2014). Only five studies used an experimental method (38%) while the others used quantitative methods (62%).

After analysing the articles chosen, it was possible to underline some results. In the 80’s, it was believed there was no bias in facial recognition when it came to race [3], thus it was pointed out that own-race effect is no more common than the cross-race identifications [4]. However, in the 90’s it started to be implied that there was a better memory to identify own race faces than other races, which meant that other race bias would increase when the contact with other races would decrease [4]. In the new millennium, a greater number of variables, as well as the use of different methodologies and facial recognition technique begins to emerge. In a showup, observing an innocent person of the same race of the perpetrator suspect increased the identification of false lineups [5]. Young people have a better accuracy when discriminating their own race from other races, which is not influenced by interracial contact, unlike what happens with elderly [6]. Also, Caucasian people were found to make more mistakes in identifications, comparing to other races, and it’s more common to mistake identity and expressions of Caucasian than Asian faces [7]. Facial expressions began to be given more relevance in the last few years [7].

Discussion and conclusions: There are significant differences in how individuals of different races look at the same face and determine recognition points differing from each other. Some faces are better recognized than others, creating practical implications in white witnesses which will lead to a greater probability of identifying a non-familiar black face than an unfamiliar face white and thereby increasing the likelihood of “false positive” results when the lineups are shown [5]. There is still no significant methodology evolution to test the effect of race in facial recognition, being the most used one the grey-scale photography line up.