Unconscious bias: your mental armpits

September 19, 2021 • 1:00 pm

Unconscious bias, also known as “implicit bias,” is the proposition that people (usually white people) have an unrealized degree of bias against people of other races or genders, and it is this unconscious bigotry that contributes to racism and sexism and their harmful effects.

One’s unconscious bias (UB) is measured via an implicit association test (IAT), characterized by Wikipedia as

“a controversial assessment intended to detect subconscious associations between mental representations of objects (concepts) in memory. Its best-known application is the assessment of implicit stereotypes held by test subjects, such as associations between particular racial categories and stereotypes about those groups.”

Both UB and IATs are controversial ideas for several reasons; I’ll give just a few:

1.)  If you are biased or bigoted, some say, it’s unlikely that it’s unconscious. (NOTE: See my comment below, I am willing to believe that some people are biased but absolutely have no awareness of it.)

2.) Most important, IATs, which supposedly measure your UB and then are used to promote implicit bias training is a scheme that doesn’t seem to work. While taking such a test may make you chastened or feel better temporarily, tests have shown that they have virtually no effect on either longer term bias or, more important, how you treat people who have been subjects of bigotry. In fact, the tests have been said to increase bias by alienating people by telling some of them that they’re bigots.

3.) The results are easy to fake. I took one some time ago, and while the results showed I was not prejudiced against blacks, and I tried to answer quickly and honestly, upon reviewing the test I could see that certain answers would either raise or lower your bigotry score.

You can see further critiques of these concepts and the tests here and here, while can find a list of articles critical of IATs and the idea of IB here.  Wikipedia also lists the problems in a section about “criticism and controversy” of IATs.

This is not meant to suggest, of course, that people aren’t biased, nor that organizations shouldn’t take steps to reduce bias and discrimination. Of course they should. It’s just that the evidence for the IATs and UB training actually doing anything to reduce bias is thin and unconvincing. Both the Guardian and Scientific American articles suggest more sustained, complex, and permanent ways to reduce bias in the workplace.

But of course when people perceive a problem, or manufacture one, like “the need to train people out of unconscious bias”, people come forward to profit, and now there is an industry of administering IATs to ferret out UB, followed by training to root out your UB.

Here’s one example sent by a reader (click on the screenshot to read). It’s part of a website advertising a firm that ferrets out bias and gives (paid) training to many professional scientific societies and organizations.

The title is unfortunate:

The unfortunate comparison from the head of the organization:

I love the “boots and sandals” analogy for understanding how to respond when called on your privilege. (If you don’t know about this, check it out here.)

It inspired me to come up with my own analogy that I’ve found useful in talking about unconscious biases.

Unconscious biases are like armpits.

  •  Everyone has them.
  •  It is not your fault that you have them.
  • Most of the time you are not aware of them.
  • Sometimes other people might point out to you that your armpits are offensive.
  • When that happens, it is your responsibility to clean them up.

The difference, of course, is that we know that armpits really exist.

The rationale goes on, using an evolutionary analogy asserting that we’re evolved to be biased. Well, we might have evolved to be xenophobic, but whether this leads to racism today is questionable. If this were the case, we’d be fearful and bigoted against people of every ethnic group, while bias is said to apply mostly to whites and their attitudes towards blacks and Hispanics. But there may be an element of tribalism that remains in our genomes. But even if this is the case, that doesn’t show that the biases are unconscious nor that the kind of testing and training that so many people are merchandising eliminate this atavistic “xenophobia”. In fact, they don’t seem to be:

Being good at pattern recognition is a valuable survival skill. When our ancestors confronted a large predator – a saber-toothed tiger, say – an individual’s chance of survival was heightened if they did not have to stand there thinking, “Hmmm, it’s big and furry and has stripes and these really big teeth and enormous claws and maybe I should consider leaving the scene. . . .” Those who recognized in an instant that the thing approaching them was a threat lived to produce the folks who produced us. Our brains are so good a this, we recognize patterns even where there are none to be found – just ask the people who find images of Elvis or Jesus or Kurt Cobain seared into their breakfast toast.

Preference for what is familiar is another survival skill. When it comes to avoiding predators, poisons, and other dangers, sticking with what you know has advantages. As you would imagine, it definitely gets in the way of making any diverse group of people truly inclusive. The flip side of preferring the familiar is a reaction ranging from mild discomfort to fear in the face of the unfamiliar. We prefer what we know well, even when sticking with the familiar doesn’t get us what we say we want, or what will bring us happiness. This is how life coaches can make a living nudging people out of their “comfort zones” to make real changes in their lives.

Read the Guardian and Sci. Am. articles to look at some better fixes for bigotry.

UPDATE: A friend read this and told me I lack a “conclusion.” I thought it was implicit in what I wrote, but perhaps it’s not. The most important lesson is to not automatically buy into implicit association tests as a measure of your bias, and especially not to fall for companies who try to sell you or your scientific organization “training” and “meeting monitors” based on the concept of pervasive implicit bias.

Too many scientific societies are now paying lots of money for this kind of training and even hiring people to walk around at meetings trying to sniff out offensive statements and interactions. Societies of course do need a procedure for investigating malfeasance, but now they are acting like a bunch of helicopter parents who must monitor the interaction of full-grown scientists assumed to be ridden with sexism and racism.

31 thoughts on “Unconscious bias: your mental armpits

  1. I just read this newsletter on substack, very good and free.

    And this particular one deals with the issue of hard work vs innate talent. And innate talent would be totally anathema to blank-slatism and also to the whole Kendi notion of equal outcomes, etc. and that unequal outcomes can only be due to racism tout court.

    This particular issue begins with a discussion of Emma Raducanu win at 2021 US Open….


    1. Good article. It would seem obvious that certain inherited traits give one an advantage in particular endeavors, but that it takes hard work and training to take advantage of those traits. No matter how much I might have wanted it I would never have had much chance of playing in the NBA with my height of just short of 5’8″.

      I’m not sure how you’re relating this to the argument at hand though. Is it that universal equality is an unattainable goal, therefore attempts to provide equal access are futile?

      One thing that I feel needs reminding is that discrimination is not evil in itself, it’s doing it based on false premises that’s bad.

    2. Interesting article.
      I’d say that anybody having or having worked with small children knows that innate ability plays a role.
      Of course, to reach the top in any field, talent is rarely sufficient, hard work is needed too.

  2. In this post I read

    “Both UB and IATs are controversial ideas for several reasons”

    “The difference, of course, is that we know that armpits really exist.”

    I am surprised by the idea that the concept UB is controversial. I think I remember a post on this website in which Pr Coyne discussed the case of people who will state that they have Jewish friends, will claim to not be antisemites, who look sincere and yet who give out evidence that they are to some extent antisemitic.

    Did I misunderstand that post of a while ago ?

    1. I suspect that they know they are antisemitic. You can know something in your heart and not admit it, you know.

      But I’ll back off a bit on the claim that implicit bias doesn’t exist. I’ll just claim that measuring it and then training people on the basis of your score is a useless endeavor. Also, the Di Angelo-ian claim that virtually all whites are unconsciously biased is a claim I don’t support.

      I’ve changed the post a bit to reflect this. It’s the training, combined with the unsupported assertion by some that every white person is unconsciously biased, that I object to.

      1. I can imagine a person being biased and not knowing it, but i guess that assertion can neither be proved not disproved.

        I’m looking through my memories to see if at some points during my life I went “Oh my God, I was biased against such and such people and completely anaware of it.”

  3. I taught a botany class. Quickly picked out two interested, good students, the 40+, oddly dressed woman and an unusual guy who sat in the front row and asked lots of questions. Good questions kept coming from a part of the room that had what I think of as “generic female students” — similar hair styles, similar faces, similar dress. Generally good students, people who do the work but are not all that interested. Not until the midterm did I realize that one of those “generics” was perhaps the top student in the class. (All the students in the class were white.) Unconscious bias can run in unexpected directions.

    1. Let’s go back to a few months prior to that. Supposed somebody had asked you “When you see a [standard description of the generic female students you talk about] what kind of student do you expect her to be ? What would you have answered ?

    2. In your defense, perhaps it just proved that prowess in botany doesn’t correlate well with how people look. Until the incident, you overestimated your skill in spotting intellectual ability based on a person’s look. 😉

  4. Unconscious bias cuts both ways, including assuming all whites are unconsciously biased. ’tain’t necessarily so.

    1. Yes, that’s part of the tactic. By saying one can have unacknowledged bias, they can say that all white people are racist, even if they do not outwardly express it.

    2. I don’t think there’s anything unconscious about that particular bias. I think it’s a conscious judgement and it’s somewhat akin to the concept of original sin in Christianity.

  5. Armpit is an interesting analogy in this context, since the rest of the world makes fun of America’s intense personal hygiene, and goal of an odorless body (thanks soap companies!). In other parts of the world, people don’t actually mind if they smell a little bit, and would be offended if I, based on my American values, told them to clean up. I

    1. I hope that was TIC, if not it is a perfect example of unconscious bias. We are clean and they are dirty. Perfect.
      Do you seriously claim USians are more obsessed with cleanliness and bodily hygiene than say East Asians (Japanese, Thai, etc.)?

  6. On a lighter side sometimes UB can be broadsided by a (fashionable?) trend. I speak of tattoos.
    Once a ‘common” practice for sailors (still is for all I know) criminal types, bad guys, macho goons and lowlife varieties. It can be tribal and worn like a badge stigmatising the wearer in short.
    So ubiquitous they have become it is just plain dumb to use them as a marker for any category.
    High, low, in-between, any gender you care to mention in any socio-economic group.
    Although some tattoos seem to make the wearer unapproachable and downright scary and not a prospect for the sales job you want to fill.
    Unless it’s telemarketing or such like…

  7. “Hiring people to walk around at meetings trying to sniff out offensive statements and interactions”.

    I have a great idea: such societies should also sell insurance against violations of the conference code of conduct. Sort of like liability insurance for car drivers. This allows drivers (society members) to drive on public roads (attend conferences) with confidence that if an accident happens (a pecksniff takes offence) the other driver (offended person) will have some recourse for compensation.

    Don’t ask me what the premium or the deductible should be. Admittedly this also sounds a bit like a protection racket.

  8. Given the manifold defects of the IAT, I advise the use of a superior psychometric tool, the Ducking Stool Test (DST). In this test, the subject is strapped into a chair and lowered into a river. If the suspect survives, this is evidence of guilt; if the suspect drowns, this is evidence of innocence.

    The Ducking Stool was so popular in 16th/17th century England that one was reportedly kept constantly on hand at a bridge in Cambridge. The method was employed both for judicial investigation and for punishment and rehabilitation. Its use for the latter purpose was celebrated by a poet of the time as follows: “Cool grows the fever of the breast, and surging passions seek to rest. The lesson ex cathedra taught here balance in the scale of thought. Then say if ever Socratic school such lesson taught as ducking-stool.

  9. Wow I really love the armpits-and-UB analogy because it works on so many levels. I think the really important thing that armpits and UB have in common is that they are both spandrels (in the Gouldian sense): non-functional and unintended consequences of selection acting on other functional traits of humans (the attachment of arms to the body, or the tendency to recognize and favor group members based on appearance and language). It’s such a great analogy because as Jerry says it is hard to demonstrate the extent to which UB exists in any particular person (like finding the armpits on a double amputee).

    Although one can decorate a spandrel with sculpture, or talk about how pattern recognition leads to a tendency toward bias, or dress up a smelly armpit (I urge you not to google the phrase “bedazzled armpit hair extensions”), none of these things exist specifically for those purposes, and from that point of view none of them are things on their own (they are just consequences of other things).

    I hasten to add this isn’t an argument against personal reflection on possible bias, or a naturalist fallacy argument in favor of bigotry. I catch myself reflecting on those things often. I just love the analogy.

    1. In the 80s, Stephen J. Gould gave a public lecture in Auckland (NZ) on the subject of spandrels. I took a bunch of my Biology pupils(17yo) along for a listen. It was a wonderful occasion with Stephen’s presentation a marvel of clarity and excitement. I am currently re-reading his ‘The Mismeasure of Man’, which is a trenchant investigation of the role of both conscious and unconscious bias in science.

      1. Yes Gould was a great speaker and a commanding presence. I remember his visit to my graduate program in the late 1980s – it was my first experience of one of those Great Man events. No groupies, but just about everything else.

        OTOH, I thought the criticism of “Mismeasure” was convincing. Gould mischaracterized much of the existing evidence at the time on variation in human intelligence, and made his own errors in analyzing and interpreting that variation. In hindsight, I find pretty much everything Gould wrote to be stirring and inspirational but a lot of it was also wrong.

        Maybe that’s not a terrible legacy?

  10. “Unconscious biases are like armpits: everyone has them.”

    You might as well stop the discussion right there. This implies that one is unconsciously biased no matter what one does. (Let’s leave surgery out of the analogy.) As far as I’m concerned, there’s no call to start the conversation there. It is like dividing by zero in a calculation. Once you’ve done it, the rest is a complete waste of time.

  11. But we know we have unconscious biases at the moment we notice ourselves having them. I remember a long time ago having a conversation with a man and being surprised at how intelligent he was and then afterwards wondering why I was surprised and realising that it was because he was black. I would be surprised if this was something unique to me or something rare in humans.

    In fact I was under the impression that the whole idea of kin selection was that we would act altruistically towards individuals who we more genetically related to us. How would that work if we didn’t have unconscious biases?

  12. This is the sort of thing where I think scientism might be fairly used. Not that unconscious don’t exist – clearly they do and are important in our cognition (think ethics panels in hospitals and the need for ethicists because panelists kept favouring people who resembled themselves) – but that it’s using science to diagnose and solve a complex social problem.

    The difficulty I think is what they’re doing is well-intentioned, for a good cause, and has *some* science behind it. To be critical with that combination puts one into the category of “racist opposing science-based anti-racism measures”. After all (rhetorically-speaking), why would *anyone* oppose trying to combat racism unless they wanted to continue to benefit from the racism?

  13. But of course when people perceive a problem, or manufacture one, like “the need to train people out of unconscious bias”, people come forward to profit, and now there is an industry of administering IATs to ferret out UB, followed by training to root out your UB.

    But a weakness is ‘intersectionality’. There can be many competing biases to explain ‘outcomes’ but they may require different training to ‘overcome’. And then you need a clear vison of what ‘outcome’ would be regarded as unbiased. A 50%/50% female/male split might be inappropriate (or not). A population demographic representative split of black/white might be inappropriate (or not).

    If you want a thought experiment to practice on ask yourself “Do the number of vegetarian black women models in pornography represent an appropriate outcome?”. You can pick your own categories and activities, but real life is not as binary as some would insist.

  14. To have a valid test, it has to: 1.) provide consistent results, 2.) be predictive.

    There is no evidence that IAT’s produce consistent results across time in subjects. There is no evidence they predict anything about a person (even their future score on an IAT). Further, because they don’t predict anything, its unclear what “training” would train away.

    On the subject of bias, there is a plenty of literature on how people are able to use reason to override bias. If people are truly troubled by bias, then the only real solution is critical thinking skills, and teaching people to argue and debate.

  15. IMO resume studies, the different results of blind auditions, etc. are all pretty convincing evidence that unconscious bias exists. However I don’t think it’s worth obsessing over as a personal problem. If there are folks who are obviously biased, yes training them to be less biased will probably help. For most normal people, however, I don’t think training is going to do much. When you’ve got normal people trying their best to do an objective evaluation, I think the standard methods of ‘blinding and peer review’ are our best bet. When possible, (1) remove the irrelevant factors so evaluators don’t consider them and (2) gather several independent views on candidates or a body of work, so that individual biases will hopefully ‘wash out.’ It’s not perfect – nothing is – but that’s the point of blinding and peer review; to create a system which works given imperfect evaluators. After all, that’s the entire point of peer review, right? We get other people to check our work because we recognize that we can be pretty biased when it comes to checking it ourselves.

  16. I am glad that pattern recognition was mentioned as an important factor here. My son and I were discussing the subject, and the best we could come up with is the following hypothetical-
    Every day you have to pass a group of strangers. Every once in a while, one of the strangers punches you in the shoulder without warning. In this hypothetical world, you cannot just avoid the crowd. You go out there every day. You notice pretty quickly that every time you get punched, the person doing it has red hair. Of course, most of the people with red hair are perfectly friendly and do not harm you. But still, every time you pass one of them you tense up. Because you are human.
    When some of the nice gingers notice your reaction, their feelings are hurt. That is also a reasonable reaction. But their ire should be directed at the people who keep punching you.
    The inner city shopkeeper from Bangladesh, who gets robbed regularly, cannot help but notice the shared physical characteristics and mannerisms of those who rob him. He will surely notice when a stranger enters the store looking and acting that way.
    It is a fact that treating every person you meet with respect and giving them the benefit of the doubt is a worthy aspiration.
    Training someone to not flinch when they pass a ginger is just not as effective a solution as training the few violent gingers to stop punching people.
    Even if we exclude pattern recognition as a survival strategy against violence, everyone is still going to notice less dangerous patterns, as long as they keep presenting themselves.

Leave a Reply