Categorizing by race: As automatic as reading?

i-eca0cf2af9fc3ac4445c7dff7d8aab70-research.gifThe Stroop Effect was originally just a language effect: we're slower identifying the color text is printed in when the words themselves name different colors. In the 81 years since the effect was first observed, it's been applied to a variety of very different phenomena. In general, the effect is explained by automatic processing: when a process is automatic, it conflicts with the desired goal and so slows processing. In fact, the Stroop Effect is so robust that researchers now use it to determine if a process is indeed automatic.

Much research has focused on the issue of whether racial bias is automatic, but a team led by Jerzy Karylowski wanted to know if racial categorization itself is automatic, so they turned to the Stroop task. Would you be slower to identify the color a person's name is printed in if it conflicts with their race, regardless of your racial bias?

The researchers created 288 different slides displaying the name of one of 24 well-known public figures. Half of the people named were white, and half were black. Each name was displayed in white, black, green, or blue, and displayed on a background corresonding to each of the other colors -- a total of 12 different displays for each name. 94 participants viewed the slides on a computer screen, and their task was the standard Stroop task: name the color each name was displayed in, as quickly as possible. Here are the results:


The green and blue text had no effect on reaction times, but participants consistently responded slower to white text when the person named in the text was black, and slower to black text when the person named was white. You can try it out for yourself with this graphic: go through the lists below, naming the colors instead of reading the text.


Did you have an easier time with the second list? If so, you're not alone.

Immediately after the experiment, the researchers also asked participants to try to recall as many as possible of the people named over the course of the experiment. Considering the fact that each of the 24 names appeared 12 times, you might expect a relatively high accuracy rate -- but in fact, participants on average could only recall 30 percent of the names. So race appears to be encoded even more rapidly and automatically than the names themselves.

Just as it is difficult to stop ourselves from reading words, even when our only job is to say what color they're printed in, so we also appear to involuntarily place people racial categories in the same circumstances. This process is so automatic that interferes with our ability to do the primary task.

Karylowski et al. offer an important caveat to their work: they point out that racial categorization is not "universal" or "natural." After all, racial categories themselves can change depending on social and historical context. As races become more mixed, who's to say where one race begins and another ends?

Karylowski, J.J, Motes, M.A., Curry, D., & Van Liempd, D. (2002). "In what font color is Bill Cosby's name written?": Automatic racial categorization in a Stroop task. North American Journal of Psychology, 4(1), 1-12.

More like this

All of you are probably familiar with color opponency, but just in case, I'll give you a quick refresher. I'll even start with the history. In the 19th century, there were two competing theories of color vision. The first was the Young-Helmholtz theory (sometimes called the trichromatic theory),…
A Witches' Bible states that "the sensitive is psychically aware of character qualities, or emotional or spiritual states, in the subject, and this awareness presents itself to him or her as visual phenomena." It's easy to dismiss such claims as pseudoscientific claptrap, yet there exist humans who…
You may have heard of the Stroop effect, and you may have even seen it demonstrated. But can the Stroop effect itself be manipulated? This short demo may show that it can. In case you're not familiar with the effect, it occurs when you try to say the color a word is printed in, while the word…
The Stroop effect is a well-documented phenomenon that shows how easily we can be distracted from a simple task. In the classic Stroop experiment, we are shown a word, such as GREEN, and asked to indicate the color it is printed in. When the meaning of the word itself conflicts with the word's…

The result is not too surprising for Americans, since race and racial awareness are such a big part of our history and our current condition. I wonder how the experiment would have gone using Asians or American Indians. I also wonder how it would have gone if the people had been unknowns instead of famous people who already have a race (that is, the term "black" or "white") associated with them.

By Mark Paris (not verified) on 19 May 2006 #permalink

This is quite curious. The previous small test here of the stroop effect (with the text and names of colours in various different colours) really showed a massive difference for me. I could immediately run down the list where "green" was in green text etc, but very noticeably slowed where "yellow" was in red, for example.

In this test, both lists came out the same speed - almost immediately. I wonder if there's two reasons for this.

One being that the names are so much longer than colours, so I'm able to read the colour of the pixels making up the text quicker than I can read the name, so I can say "white" immediately, before I've read past "Janet" in "Janet Jackson".

Second, could a cultural element affect my reading?. I'm from an area of Australia that's predominantly white - so much so that I could go weeks without seeing non-whites. One guess is that except for cases like this post where discussion of skin colour comes up, my mind hasn't learnt to automatically see a white person as "white", as there's no contrast to compare. To put it in perspective, it might be like the difference between seeing earth people as "earthlings" vs seeing people from off-world as "aliens" at some point in the future if we meet aliens, but as for now at a time when we don't see space aliens, we don't automatically read "earthling" when we see a person.

I also don't watch television or movies, so I don't see the people described in the text very much, although I do know who they all are.

By Hello World (not verified) on 19 May 2006 #permalink

Hello World: Good questions. I wish we had a good way to measure reaction time on Cognitive Daily, because this one would be another great experiment to try to replicate. I imagine there are going to be significant cultural differences on a test like this -- especially if you're not immediately familiar with the names on the test.

I also think the effect is not quite as pronounced as the classic Stroop effect, so it might be a little harder to notice in an uncontrolled setting.

Mark: I'm not sure how you could do the test with "unknowns" -- they need to be well-known enough that people associate them with a particular race. Otherwise, how would you conduct the experiment?

I believe that's a very small effect. A bit of searching suggests that a typical Stroop effect is on the order of 100 ms (MacLeod, 1991), while this experiment shows only a 12 ms effect. My introspective reaction is that, as Hello World says, I don't read the names fast enough for the Stroop effect to really occur. I can produce the text color while still fixating on the beginning of the name, a long time before my gaze moves over to the right size of the name. I suspect that if the text was initially just an outline, then after (say) 100 ms changed to another color, you might get a much larger effect.

Dave, I haven't read the paper; what's the p value, and do the authors report effect size compared to a typical Stroop effect size?

North American Journal of Psychology is not a very high impact journal... I suspect the reviewers were not very impressed either...

I agree with Hello World and Harlan - the effect definitely doesn't seem as big as the text color/color word Stroop Effect. I think another factor that makes the effect less prominent is that it will be sensitive to how strongly the person associates the individual with "blackness" or "whiteness." Have any other experiments tried to establish whether this effect applies to other visual property vs content of words situation as well? If arrows pointing left or right with the words left or right near them exhibit a Stroop Effect, we could try the same thing with names of prominent left and right wing political figures to see if the small effect we see here is specific to race or part of a general tendency to apply categories in response to reading a name.

Harlan, the p values range from .01 (comparing black and white to the other colors) to .001 (comparing black and white to each other), so the effect is definitely significant. As you've pointed out, it's not as large as classic Stroop, though.

I also suspect that awareness of the task has something to do with it -- once you're conscious that this is a test about racial identification, you can block it out -- again, something which indicates that it's not as robust as Stroop.

I agree with previous commenters: I didn't feel any difference between the two lists at all. Perhaps it's because of not being inmersed in American culture and not having so quick reactions to the the names and races of celebrities, though I know almost all of them.

Huh, well I'm quite American, but I was very easily able to ignore the names and just say the colors.

I personally think a more interesting test would be to have pictures of people with words on top of the pictures ("White", "Black", "Asian").

I wonder at this experiment, because like Jennifer, I've always found this test to be absurdly easy. I can read colors or names, either one, without any trouble; I can also switch back and forth easily.

This experiment is faulty in my mind. Your culture will affect your reaction to everything here--culture, not race. Beyond that, though, it will also depend on where you grew up regionally and what influences you as a unique person (e.g. new age music versus R&B). That has little to do with race and more to do with interest. The chances of your recognizing Oprah's race accurately and quickly if you watch her every day immediately skyrocket, and therefore affect the test greatly. These being common celebrities' names, too, are going to affect reactions, no matter one's race.

Also, going back to the regional setting, would this test be affected if one being tested had lived in multiple racial settings throughout his or her lifetime? What about those in cities, where multiple races are encountered daily? For instance, I have lived in areas that were predominantly black (I'm white); I've lived in mostly white areas, and I've even lived in areas where there is a large Hispanic base. What then? This would affect your perception of race, even to this test's level, I believe.

I'd be more interested in seeing a breakdown of each name, really. This is too broad of a test to cover the experiences and influences of a large group of people to know the deeply-rooted feelings of racism.

I wonder how the experiment would have gone using Asians or American Indians.

I think the critical difference is that it's no longer common to use color words in describing those ethnicities. 'White' and 'Black' are socially permissable, while 'Yellow' and 'Red' aren't.

By Matthew George (not verified) on 20 May 2006 #permalink

I think the effect observed is interesting but have difficulty in concluding that this type of bias is the result of implicit processing per se. As others have noted surly social determinants could have resulted in this finding, ie. the amount of importance someone places on belonging to a particular race. Furthermore the author points to the role of races becoming increasingly mixed, in this case then how might using famous people of mixed origin effect the result? I agree with the comment therefore that it would be a more interesting method to look at how the category race eg ASIAN and saying the colour effects time to say.

By Kevin Cassidy (not verified) on 31 May 2006 #permalink

Isn't Whitney Houston pretty mixed?

By Varda Lobanov (not verified) on 31 May 2006 #permalink

Agreeing with others: I didn't even read the names, since it was easier to see the colour without bothering. I really can't see that anyone would be thinking about the colour long enough to get through the name, understand who it is, and associate the skin colour.

I dunno. Are we missing an element of the method here? Were the people asked to read the name out loud and then idetify the colour of the word?

(Aside: I'm Australian too, and the term "race" just seems icky to me. I worked for a US company in Australia some time ago, and processed job applications. The company supplied the forms, and they included a question asking for the applicant's race. Quite a few folk were offended, and we had a *lot* of people answer "human").

By SmellyTerror (not verified) on 01 Jun 2006 #permalink