Home > Attention, Memory > Own-race Bias: Why Some People Might Look The Same to You

Own-race Bias: Why Some People Might Look The Same to You

As someone who is a fan of true crime podcasts, I couldn’t help but to binge-listen to the episodes of Wrongful Convictions–a podcast by the Innocence Project detailing stories of people who were convicted for crimes they did not commit. In one of the stories, a person was convicted due to the misidentification by one of the victims of the crime. This eyewitness testimony was enough to carry a 30-year-old sentence in prison.

Pattern recognition of faces of other races can vary according to the cross-race bias.
PC: The Guardian

There are several cognitive errors that could make of this eyewitness testimony (and others) unreliable. Daniel Schacter described in the Seven Sins of Memory (2001) different cognitive errors that the memory is sensitive to, including memory biases. In the case of the person wrongfully convicted, a cross-race or own-race bias could have influenced the misidentification. An own-race bias refers to the tendency of being more accurate at recognizing faces of your own race than faces of another race (Malpass & Kravits, 1969).  The bias is not exclusive to the context of eyewitness testimonies and the criminal justice system, however; you can stumble upon the own-race bias during a trivial day. For example, have you ever had a feeling that people who don’t belong to your racial category look “all the same” to you? Or have you wondered why you are very good at recognizing faces of your own racial profile yet can’t make the same accurate distinctions cross-racially? Or maybe, have you ever confused two people from outside your race because you couldn’t distinguish certain individual characteristics to make them apart? Then you’re in the right place to learn about this cross-race phenomenon! In this blog, we discuss possible mechanisms behind the cross-race bias.

Recognizing faces analytically rather than holistically by looking at individual features.
PC: Wrongful Convictions Blog

Previous studies have reported that the cross-race bias has a strong effect on eyewitness testimony (Meissner and Brigham, 2001). Pattern recognition and selective attention play a role in the categorization of these faces (Hugenberg et al., 2010; Ho & Pezdeck, 2015; Rossion & Michel, 2011), and they play an active role when making recognition mistakes. The own-race bias suggests that when exposed to faces from outside our race, we are not as efficient at recognizing the features that individualize them.

Pattern Recognition and Own Race Bias

Pattern recognition is the process of assigning meaning to information once it is perceived. When we pattern recognize faces, we do so holistically rather than analytically. This means that when we look at faces we look at them as a whole (holistically) rather than looking at individual facial features such as a nose, eyes, and eyelashes (analytically). Evidence from Michel, Corneille, & Rossion (2010) suggests that the holistic processing of faces varies depending on the race of the face we are trying to pattern recognize. In their study, White participants were asked to identify faces of White individuals, Asian individuals, and racially ambiguous individuals. Faces that looked the most similar to faces with White individuals were processed more holistically when compared to the identification of faces categorized as belonging to another race. The results from Michel et al. suggest that pattern recognition differs depending on whether we are looking at faces of our own or different race, because we fail to process the information holistically. So… we are not that good at recognizing face from another race due to failure to recognize a face as a whole.

But not everything is pattern recognition’s fault

PC: Keenetrial.com. Motivation to recognize faces from other races can vary.

Processing stages also play a role in the cross-race bias. Processing stages are the different steps involved in remembering information. They are divided into storage (transference of information to long-term memory), retrieval (the search of information previously stored), and encoding (information is converted into a mental representation). The most relevant processing stage to the cross-race bias is encoding: when we see a face, we need to convert it into a mental representation of the face in order to assign meaning to it (pattern recognize) later on. When we think that a particular face is important or relevant to encode, then selective attention is paid to the individualizing facial structure of the individual. According to Hugenberg et al. (2010), this process involved in the discrimination of faces is applied to faces of the same race of people making observations. Conversively, when the identity of the face is perceived as less relevant to encode, selective attention is directed toward category-diagnostic facial features such as race, and less directed toward individualizing facial features. Due to the different ways of encoding the information, the Categorization Individuation Model (Hugenberg et al., 2010) proposes that same-race faces are better recognized than other-race faces, because the motivation to attend to a face might differ.

Motivation to recognize faces

Yes, pattern recognition matters when recognizing faces, as well as the initial encoding of the faces. But motivation matters too! Findings by Michel, Corneille & Rossion (2010) suggest that there is a social aspect in the bias that is important to incorporate when talking about it. In response to the need of the incorporation of social aspects, Hugenberg et al. (2010) proposed the Categorization Individuation Model (2010) as previously mentioned, which provides theoretical insights that can explain the cross-race bias. This model proposes that face processing involves an attention component, in the sense that when we are processing a face, we categorize whether the identity of the face is important to encode or not. But the model also considers a more social aspect: the motivation to attend to a particular face. Situational cues can influence the motivation to redirect attention to a face of the other-race. This idea is related to the contact hypothesis described by Allport (1954), which refers to the idea that constant exposure to an outside group can decrease prejudice. In the context of the own-race bias, exposure to people from outside our race can facilitate holistic pattern recognition of faces, and allow us to recognize faces from outside our race more efficiently (Chiroro, 1995).

Eyewitnesses and other blues

To illustrate how the CIM and the results from Michel, Corneille & Rossion play a role in the cross-race bias, let’s revisit an eyewitness example. In the case of misidentification during an eyewitness testimony, if the eyewitness is a white person who witnessed a black person committing a crime, then there is a chance of the cross-race bias influencing the accuracy of the testimony. When given the opportunity to identify the perpetrator of the crime, the witness might not be able to accurately identify the person and someone else could be mistakenly identified as the perpetrator.

…to recognize faces from other races.
PC: Memegenerator

According to the results from Michel (2010)’s study, for example, the eyewitness could have seen the face of the perpetrator and looked at their face in a less holistic way which is uncharacteristic of the way we pattern recognize faces. Similarly, the CIM can expand on this idea of deficits during the encoding stage, by suggesting that at the moment of the crime scene, the identity of the face wasn’t considered very relevant (for example, attention could have been redirected to what was actually happening during the scene), and attention was directed toward race rather than toward individualizing features of the face, such as shape of nose, eye color, and mouth shape. In this case, the situational cue makes the motivation to direct attention toward the face of the perpetuator less likely.


The cross race bias is an example of how powerful cognition can be, and how cognitive biases affect the way we interact with the world. It’s not only seen in the context of true crime. For example, think back of the last time you confused two or three members of another race with each other. In a place like Colby, a predominantly white institution, being surrounded by people of your own race is not hard if you’re a white person. So as a white person, when someone sees a face of the same race, they are more likely to pattern recognize it holistically, and to pay attention to individualizing features besides race, than when they see a person of another race. Just encoding a face differently can determine how we pattern recognize faces and how we interact and identify other individuals. Hence contributing to the idea that this outgroup “looks the same.”


Arnold, M. M. (2013). Monitoring and meta-metacognition in the own-race bias. Acta Psychologica,144(2), 380-389. doi:10.1016/j.actpsy.2013.07.007

Chiroro, P., & Valentine, T. (1995). An Investigation of the Contact Hypothesis of the Own-race Bias in Face Recognition. The Quarterly Journal of Experimental Psychology Section A,48(4), 879-894. doi:10.1080/146407495084014

Ho, M. R., & Pezdek, K. (2015). Postencoding cognitive processes in the cross-race effect: Categorization and individuation during face recognition. Psychonomic Bulletin & Review,23(3), 771-780. doi:10.3758/s13423-015-0945-x

Hugenberg, K., Young, S. G., Bernstein, M. J., & Sacco, D. F. (2010). The categorization-individuation model: An integrative account of the other-race recognition deficit. Psychological Review,117(4), 1168-1187. doi:10.1037/a0020463

Malpass, R. S., & Kravitz, J. (1969). Recognition for faces of own and other race. Journal of Personality and Social Psychology,13(4), 330-334. doi:10.1037/h0028434

Meissner, C. A., & Brigham, J. C. (2001). Thirty years of investigating the own-race bias in memory for faces: A meta-analytic review. Psychology, Public Policy, and Law,7(1), 3-35. doi:10.1037/1076-8971.7.1.3

Michel, C., Corneille, O., & Rossion, B. (2010). Categorization of face race modulates holistic face processing. Journal of Vision,6(6), 435-435. doi:10.1167/6.6.435

  1. May 1st, 2017 at 21:45 | #1

    Thank you for writing such a great blog! I really enjoyed learning more about own-race bias, and especially liked how you connected this bias to the Colby community. I agree that it is really important to remember that, while many Colby students would like to think otherwise, the fact that we are surrounded by mostly white people makes us inherently better at recognizing the faces of other white people. Your blog post reminded me of the study we read in class by Payne (2005), which found that when quickly flashed a person’s face (either white or black) and a picture of a weapon or a tool, that people were more likely to see a tool as a weapon when flashed a picture of a black person’s face than a white person’s face. I thought of this when you were talking about pattern recognition, because this study shows the role top-down processing plays. As you mentioned, and was talked about in class, this is a major reason in why eyewitness testimonies can often not be trusted: what we see is not completely veridical and can be influenced by our stereotypes.

  2. Vianny Lugo Aracena
    May 6th, 2017 at 08:12 | #2

    @Mollie Rich

    Thanks for reading the post!

    And yes, what you’re saying about how the fact that we are mostly surrounded by white people makes us better at recognizing better, it ties back to the contact effect that we talked about during class. Being more exposed to, or having more contact to a particular race will facilitate pattern recognition. It’s interesting to look at this bias from the perspective of a person of color attending to a place like Colby, the bias is usually less strong since they are constantly surrounded by and in touch with non-poc.

  3. yipeilo
    May 8th, 2017 at 16:42 | #3

    I really enjoyed reading this post! Own-race bias is truly an issue in eyewitness testimony that involves lots of cognitive errors. I have learned the seven sins of memory in the context of the eyewitness testimony, but I did not know about the Categorization individual Model before I read the post. It is interesting that people use different ways, analytically and holistically, while encoding the information. 

    As an Asian international student, I experienced this own-race bias for the first few weeks after I arrived Colby, but the bias decreases significantly as time passes and as I interact with more people with various race background. The own-race bias makes me think of one article we read in our cognitive class, in which the author discusses several explanations for own-age bias in face recognition (Harrison & Hole, 2009). Similar explanations might also work here. For example, people usually have more exposure with others who have the same race as them, the more practice makes them face experts on their own race. The paper also talks about how motivation can influence the accuracy of face recognition, for example, trainee teachers are better at recognizing children’s’ faces because they are more interested in and paying more attention to children. I wonder if interacting and practicing recognizing faces in different races could reduce this own-race bias.

    Harrison, V., & Hole, G. J. (2009). Evidence for a contact-based explanation of the own-age bias in face recognition. Psychonomic Bulletin & Review,16(2), 264-269. doi:10.3758/pbr.16.2.264

  4. May 11th, 2017 at 17:27 | #4

    This was a very interesting post, and will change my approach to the way I look at people.

    While reading your post, I started thinking about the difference between holistic and distributed models and how they relate to your post. The holistic model says that knowledge is represented holistically, which means that you have concepts for objects. On the other hand, the distributed model says that instead of having a concept for an object, we instead represent it as a bundle of features. Applying these models to how we recognize faces helped me make sense of why certain types of facial recognition techniques lead to the highest errors. You talked about how, with individuals of our own race, we process them by using the distributed model, which involves representing them as a bundle of features. We can then combine these features so that each individual will be more recognizable. You noted that, when viewing faces of individuals of different races, you talked about how we would use holistic ways of recognizing the face. This would lead us to process the face more holistically, which means we assign it as a concept (for example, Asian) and not as an individual. This helps explain why we have higher errors when viewing different races. It makes me wonder if people raised in multi-racial families are better at recognizing individual features in different races.

You must be logged in to post a comment.