Subtle prejudice

Many people sincerely believe they are not prejudiced. Research shows, though, that subtle prejudice is quite common. Furthermore, getting rid of it is very difficult.


There are all sorts of prejudice, on the basis of race, religion, gender, sexuality, age, occupation and various other characteristics. Some people are overtly prejudiced, for example expressing antagonism towards gays or refusing to say hello to neighbours of another nationality. However, many types of prejudice are less acceptable than before. In some circles, racist language will alienate listeners.

But can you be prejudiced even if you think you’re not? There’s a way to find out whether you have this sort of subtle, unconscious prejudice. Take the Implicit Association Test (IAT). But let me warn you. If you like to think of yourself as unbiased, be prepared for a possible shock when you take the IAT.


Actually, there are many versions of the IAT, covering race, gender, age and other dimensions. Let’s say you take the age IAT. You look at faces of people and, as quickly as possible, classify them as young or old. You look at words and, as quickly as possible, classify them as good or bad. Then you look at interwoven sequences of faces and words, and classify them the same way. The online IAT calculates how long it takes you to do these tasks. If it takes you longer to respond to certain sequences, the implication is that you have an implicit bias against old people. It’s very common, and worldwide.

If you want to learn more about subtle prejudice, read the book Blindspot by Mahzarin Banaji and Anthony Greenwald. It’s about the IAT and much more. Greenwald developed the IAT; Banaji was his student and has collaborated with him for many years.


The disturbing finding is that most people who believe they are unprejudiced are actually biased, unconsciously. Greenwald and Banaji admit they are too, even though they wish they weren’t.

Banaji is from India and was dismayed to discover, via the IAT, that she had an unconscious bias against non-whites. The IAT reveals that many women hold, unconsciously, the gender stereotype of male = work, female = family. Many US blacks have an unconscious bias against blacks, and so forth.

Mahzarin Banaji
Mahzarin Banaji

How could this be? The two-minds model of human mental functioning  helps explain. According to a standard view in psychology, humans have two minds, in other words two independent mental systems. One is the rational mind, used for much conscious thinking. It is slow, careful and requires effort. The other system is the intuitive mind, which operates rapidly, automatically and usually out of consciousness.

When you meet someone new, your intuitive mind rapidly classifies the person in terms of categories, including sex, ethnicity, age and social status. This happens out of conscious awareness. This automatic classification process can also result in judgements and changes in behaviour.

The IAT, by requiring rapid responses, taps into the intuitive mind and its biases. Conscious effort by the rational mind is required to overcome automatic responses, and the more effort that is required, the slower your response. The slower response is detectable and used to assess your level of bias, perhaps none but perhaps more than you expect.

As Blindspot‘s subtitle – Hidden Biases of Good People – suggests, Banaji and Greenwald are concerned about people who want to be unbiased, and who usually believe they are unbiased, but who actually have subtle prejudices.

You might want to dismiss your results on the IAT as not reflecting anything serious. Countering this, Banaji and Greenwald report on a large amount of psychological research showing that IAT scores correlate with prejudicial behaviours. But the bias in behaviour is subtle, so people – “good people” in Banaji and Greenwald’s terms – do not realise it is occurring.

Some experiments involved white students being approached by an interviewer, either a black or a white woman, on a neutral topic. A videocamera covertly recorded each interaction. Nonverbal indicators of the subject’s friendliness or discomfort were judged by independent assessors. A typical finding is that most white US subjects respond more positively to white interviewers.

African American woman taking an interview of a woman
African American woman taking an interview of a woman

Another experiment involved leaving a stamped, addressed and unsealed envelope in an airport phone box. On looking inside at the contents, a photo of the writer is hard to miss. White subjects are more likely to seal and post the envelope if the writer is white rather than black. The differences in response are not all that great, but enough to show a correlation between IAT scores and behaviour.

Dozens of clever studies of this sort show that subtle bias is common and that it affects behaviour, often with drastic effects. Banaji and Greenwald say that today subtle, unconscious bias may be the prime source of unfair discrimination in the US, overshadowing overt prejudice in its effects. Subtle bias is all the more potent because it is unconscious and hence unrecognised.

One of the issues addressed by Banaji and Greenwald is support in the US for measures to make up for historical disadvantage. Although a majority of whites disavow racial prejudice, a majority also thinks the playing field is now level, so extra support for disadvantaged groups is not needed. This can be interpreted as the expression of unconscious prejudice.

Because of the attention to the rights of minorities in recent decades, many US whites now believe they are suffering from adverse discrimination. Because their own prejudice is unconscious, their views about social injustice are opposite to what is shown by the evidence. Banaji and Greenwald provide a detailed appendix about black disadvantage in the US, giving ample evidence that it is real and significant.

Anthony Greenwald
Anthony Greenwald

Why unconscious prejudice exists

You might imagine that good intentions, or personal experience, would be enough to overcome unconscious bias. How could a career woman like Banaji hold an unconscious stereotype of female being associated with family rather than career? Why would a capable older person unconsciously accept negative stereotyping about old people? The answer is in cultural stereotypes that we imbibe from many sources.

In studying history, we read a lot about leaders, especially heads of state. Most of them are men, and a large proportion are white, at least for US people studying US history.

Mount Rushmore
No black or female faces on Mount Rushmore

In US television news, most reporting is about the US, and hardly any is about sub-Saharan Africa. As a result, it is easy for the intuitive mind to become attuned to white faces and not easily differentiate black ones.

In Hollywood productions, there are embedded assumptions about gender, age, social class, crime and a host of other social dimensions. Though not all shows have the same implicit message, the overall effect of watching thousands of hours of Hollywood entertainment is to implant associations and assumptions into many viewers’ minds.

The result is that stereotypes become deeply embedded. Think “surgeon” or “soldier” and the mental image of a white man may be automatic.

The messages of culture, including media, buildings, textbooks, ceremonies and much else, shape the unconscious, intuitive mind. The IAT probes the prejudices of this mind.

What to do

In Blindspot, Banaji and Greenwald devote considerable attention to the challenge of overcoming unconscious prejudice. They assume that “good people” would prefer to be unbiased and, if they knew about their unconscious prejudices, would like to change them.

However, changing the intuitive mind is not easy, especially when culture is continually fostering and reinforcing unconscious stereotypes. Banaji and Greenwald are not optimistic about trying to change intuitive responses. So they pursue a different approach: avoiding or sidestepping unconscious prejudice.

A good example is marking essays without knowing the identities of the students. In this way, unconscious biases about gender, ethnicity, appearance and personality are avoided. The same approach is used by editors of some scholarly journals when soliciting peer reviews: the author’s name and other identifying information are removed from the paper. Though reviewers sometimes can figure out the identity of the writer, the intention is to reduce the influence of conscious or unconscious bias.


However, this sort of blinding has limited applicability. In the workplace, a person’s gender and ethnicity cannot easily be disguised.

An initial step is for more people to become aware of their own unconscious biases. It can be disconcerting, even distressing, for a consciously fair-minded person to discover prejudice lurking in their intuitive mind, shaping their behaviour without revealing any trace to consciousness. Accepting that this can occur, though, can unleash creativity and energy in redressing culturally induced blindspots.

Recognising that it is difficult for individuals to overcome their own unconscious biases, there may be better prospects for groups of concerned individuals working together with a shared intent of developing ways of reducing prejudice.

A radical alternative

Prejudice is damaging psychologically, and also when it affects people’s job opportunities, promotions, housing options and other material aspects of life. It seems difficult to deal with hidden prejudice because judgements by others are so central to success in life.

This can be seen as a flaw in the connected systems of meritocracy and inequality. The usual idea of meritocracy is that people rise or fall in their station in life depending on their capacities and achievements: those who do better on tests get better grades and advance further through formal education; those who are more qualified and experienced get better jobs; those who do better work are promoted; those who generate new ideas receive recognition. Inequality in society is justified by belief in the merit principle: those who earn more money or become famous deserve their privileges because of what they’ve done.

The problem is that meritocracy is a myth. People often get ahead because of having wealthy parents, good luck, social advantage such as knowing the right people – and prejudice. Inequality due to birth, luck, connections or prejudice is simply unfair, and not justified by merit.


One way to nullify the impact of prejudice (and the role of birth, luck and connections) is to move to a more egalitarian society. At the workplace, instead of having huge pay differentials, most workers could receive the same pay. Or, more generally, everyone in society could be guaranteed the same basic income. With such a system, the impact of prejudice on income would be greatly reduced.

The existence of subtle prejudice can be used to motivate thought experiments. If we imagine that unconscious bias is widespread and virtually impossible to overcome at the psychological level, then the question becomes, how should society be organised so this unconscious bias has minimal consequences? There are some radical possibilities. For example, instead of electing politicians, members of decision-making groups could be chosen randomly.

Banaji and Greenwald say that figuring out how to overcome unconscious bias “may require some thought.” That thought needs to include radical alternatives.

Brian Martin

Thanks to Emma Barkus, Anu Bissoonauth-Bedford, Trent Brown, Rae Campbell, Kathy Flynn, Xiaoping Gao and Alfie Herrero de Haro for useful comments on drafts.