An Unconscious Bias test you can do at home

This is your unconscious bias...

Predicting the future is hard, and one of the reasons it’s hard is the fact we all have a natural unconscious bias to assume elements of the present will define the future. The influence of unconscious biases is often a bit tricky to explain to people, so here’s a little unconscious bias test you can do with your friends to help explain what behavioural psychologists (and mind reading magicians) mean by ‘unconscious bias’.

Nobody likes hearing they’ve got any sort of bias. They don’t hear the word “cognitive” or “unconscious” they just hear the word “bias”. And the term bias has a lot of inherently negative connotations for most people. It conjures ideas of racism, sexism, homophobia, nepotism and so on. But of course, bias doesn’t mean something negative at all. It’s a reference to asymmetry. A tendency to favour one side over another. You might argue that preferring beer to wine is a bias. However, because of the negativity associated with the term we have lots of words like preference or taste to express the exact same concept in a less negative way.

Anyway, explaining the role of bias is a thinking problem, so here’s a simple thinking tool to help people get to grips with it.


The Superheroes vs Engines Unconcsious Bias test

Here’s a way you can explain unconscious bias to people without them feeling on some level you’re criticising them or calling them a dick. It works like this:

SET-UP: If the people you’re talking to include a comic book junkie or a car mechanic, this might not work so well, so try to exclude anyone who you know has a really in-depth knowledge of either topic. It’s a test for general cognitive bias, so picking average people works best.


Okay, once you’re ready, it’s time to play…


You say “Okay, I want to demonstrate how unconscious factors in the way you process information affect the way to respond to questions. Everyone try this one: I’m thinking of comic book character with green skin, super strong, bad temper, keeps getting into fights. Can you guess which one?”

Then listen to the answers. The chances are they will all say “The Hulk”.


You say “Why didn’t any of you guess She Hulk?”

Then listen to the answers. They will all tend to revert to the same response “Who?” “That’s the most famous one” or “I don’t know any others, I’m not a comic book geek” and so on.

NB: If someone does, by chance, say She Hulk, you can respond with a different green skinned, super strong character like “The Green Goblin” or “Drax The Destroyer” or “Gamorra” or “The Lizard”.

Understanding the results:

Unconscious bias #1 – the Hierarchy of Prevalence:
There are around 80 green skinned comic book characters in popular titles, most with super strength and most of them get into a lot of fights. However there is a hierarchy of prevalence at work. Some characters are more dominant than others in terms of their frequency of making appearances and the dominance of their role as a central protagonist or an incidental character. So the fact most people guess “The Hulk” is down to the fact we’re exposed to a lot more images and mentions of “The Hulk” than any other character fitting that same category of comic book character. This demonstrates we have an unconscious tendency to answer a different question from the one that’s actually being asked, based on prevalence:

I.e. We hear “I’m thinking of a green skinned character, can you guess which one?” but our brain interprets that as “Name the most popular green skinned character.”

Unconscious bias #2 – influencing biases : The majority of them will say Hulk. They’re less likely to mention She Hulk because there are fewer female characters with strength who like fighting in the comic book world. That fact demonstrates a common gender bias in entertainment, the tendency for female characters to fit a certain kind of female gender role and male characters to fit a male role. Males tend to fulfil roles (cops, soldiers, politicians etc.), women tend to be other kinds of ancillary role (doctors, nurses, teachers, secretaries etc.). Although there are many female superheroes, they are massively outnumbered by the make gender. We all understand this gender bias, and it influences the effect of the hierarchy of prevalence. By the same token, they might not consider villains or obscure characters to be comic book characters, which of course, doesn’t make sense but remember, they’re not answering the question you asked, they’re answering their own question which might have precluded certain types of character because they aren’t big name superheroes.


You say: “Okay, I’ve got another one. I’m thinking of a component in an engine, made of aluminium. Can you guess which one?”

The chances are you’ll get far fewer answers. In fact, when I did it with a room of about 60 people, none of them offered a guess. This is important.

Understanding the result:

Unconscious bias #3 – shame loading:
Again, here we see the affect of the hierarchy of prevalence at work – or rather – a lack of it. The components within a car engine are a subject that has nothing like the common cultural exposure of a comic book character that’s featured in decades of high profile media output. Without a hierarchy of prevalence to shape our answer, we’re faced with the prospect of demonstrating our knowledge of the topic itself – which in this case, feels like a very serious engineering topic.

What happens here is what psychologists call “Shame loading”. It’s very simple – nobody feels their intellectual status is challenged by not knowing about comics. After all, so what? They are consumed by children, they’re not considered adult art or literature, they’re not for cultural elites like Opera or Shakespeare (and so on). Comics are considered pop culture and low brow. A car engine (and cars in general) on the other hand, is something we all rely upon. It’s something that is ingrained into our culture in various different ways, as a tool, as a vital part of the economy, as a status symbol. We all have to confront paying for getting our engines fixed at some point. You can’t ignore care engines in the way you can ignore comic books.

It’s a more serious topic. Which means for many people, they feel less comfortable admitting they know nothing about their car engine even if they have no problem whatsoever admitting they’ve never read a comic book in their lives. This loss of comfort is a shame reaction, the sense that we should know something, which means the fact we don’t is an admission of a weakness or failing on our part. We want to avoid feeling shame, so we tend to avoid responding to questions that might make us experience it.

(SIDE EXPERIMENT: If you’re interested in shame loading questions, probably the best example is asking people if they masturbate. Studies in many countries have shown that almost everyone does, upwards of 95% of men and 89% of women. Moreover, all children touch themselves for pleasure during their early years. More than that, it’s questionable that the <5% of men and <11% of women who answer they don’t masturbate in surveys are answering truthfully, given how normal a part of behaviour for humans it is. However, because talking about masturbation is a cultural taboo in many countries, it is heavily shame loaded and only a minority of people are happy answering the question openly in public. More than that, the fact someone is asking you about it makes you feel uncomfortable, there is a shame loading factor in giving people personal information because we fear the risk that they might share that information without your consent, causing you social embarrassment. It’s shame loading gold although, be warned, it can get you into some very weird and very short conversations if you try it).


You say “Okay, here’s the last experiment… I’m going to ask three quick questions” (by now they’ll be getting bored with it, so telling them it’s the last one will make them less irritated with you) “I’m thinking of a comic book character. Can you guess which one?”

It doesn’t matter what they say. They’ll all throw out names. Wait until they’ve finished and say “It was The Hulk.” This will throw them off balance a bit and mess with their heads, which is all part of the misdirection that helps illustrate the power of their unconscious thought processes.

Then you say “Great. Okay, now I’m thinking of a car engine part, do you think you can guess which one?”

Again, it doesn’t matter what they say. What you’ll notice is, again, fewer answers or answers that aren’t actually car engine parts, but other parts of the car. You answer “It was the engine block”.

Then ask “Can you tell which question you felt more confident answering?”

They’ll mostly answer the first, or say both were equal, but very few will rank the engine over the comic book character.

Understanding the result:

Unconscious bias #5 – internal odds reckoning and categorisation:
Before we make decisions, we have a natural tendency to make a calculation about our chances of the decision yielding the desired outcome. If we think we have no chance of winning, we’ll often choose not to play the game. This changes if the risk is matched with a high reward, like gambling or playing the lottery, or if the odds feel fairly even of a win or a loss. However the social context is important, because our understanding of risk is influenced by shame loading.

For example: There’s no shame in not winning the lottery or losing at a hand of poker. In those examples, there is only one winner and many losers, so the shame of being a loser is low. However, if you are under confident about how many losers there will be, the shame becomes a lot more influential. If you get the answer wrong and are in the minority, it’s akin to coming last in a race or failing an exam, you feel shame because you’re the only or worst loser.

There are over 5000 characters in Marvel comics alone. Take DC comics into account and that number grows to over 15,000. Add in all the independent comics and the number swells again to more like 16,000. You’ve got a 1/16000 (0.0000625) chance of getting the comic character right, versus a 1/2000 (0.0005) chance of getting the car part right (engines rarely have more than 2000 components, if it’s a complex engine from a cutting edge racing car).

So the engine question guess has eight times better odds of a correct answer than the comic book question, but most people assume the comic book question is easier. However, as we’ve noted, the hierarchy of prevalence means we’re more confident we can guess a comic book character over an engine component. The shame loading is also different. Again, who cares if you know nothing about comic book characters?

However, what’s really important here is the differential between the odds of guessing each answer correctly. Most people assume the odds are reversed, that engines are far more complex than they actually are and that comic books are much less complex than they actually are. We categorise topics without any logical reliance on data or knowledge about them. In this case, we assume automotive engineering is complex and pop culture is simple.

You can test this with a variation on the theme. Replace the engine question with trying to guess an Egyptian God (twice as likely to guess right over a comic book character) or a chemical element (135 times more likely to guess right) or a President of America (372 times more likely) and you’ll see precisely the same effect take place. In this case, the way you categorise academic topics like history, chemistry and politics means you’ll be less comfortable making the guess, just like with the engine.

Of course, again, the prevalence of hierarchy changes the results again, people with favour famous Egyptian Gods, popular elements and well known Presidents over lesser known ones. These questions also have similar shame loadings. But where they differ is in the probability of guessing correctly. Guessing the President is much easier than the chemical element, guessing the Egyptian God almost impossible by comparison, however all of the questions will appear to be harder compared to guessing the comic book character, which remains the statistically hardest one to guess right.


When it’s all over, you can reveal to people that you have manipulated the questions in order to achieve a certain kind of result. At this point they might well question your conclusions and argue over the reasons why they answered they way they did. It’s a fascinating topic, but also one that makes people uneasy (it’s a bit like you’re a magician) so here’s the icing on the cake.

Tell them all their answers were illogical!

i.e. The only actual question you have asked is: “Can you guess which one?”

Logical answer: “yes” / “no” / “maybe”.

The rest of the test was based on statements, not questions. They interpreted the questions unconsciously into their own internal question, they didn’t actually answer the question that was asked.

Okay, that seems like a trick, but actually, the notion of a trick question is in itself exploiting an unconscious bias towards hearing a different question in our heads from the one that’s actually being asked. Our brains are very well adapted to working out problems, and we also focus on making predictions about the outcomes of our decisions. Which means that, people assume that even if they answer “yes” they’ll still have to guess the comic book character, so they skip straight to the next predictable question, which is not doing what you asked them to do. You asked them to do A, they did B. That’s not logical.

Secondly, by giving an indication that they think they can, might or can’t answer the question you are bound to ask next, they are effectively increasing the shame loading on themselves by either setting themselves up to fail, or revealing a lack of confidence, or looking over confident. All of which mean when they guess, if they guess wrong, they’ll feel even worse about it.

At this point people might think you’re a bit of a dick, but it’s interesting to show how behaviour that feels perfectly logical can be unconscious and illogical. If they still don’t get it, remind them that you weren’t even asking them a whole bunch of stupid questions in a guessing game, it was (as you said before you started) an experiment to demonstrate an element of how their brains work, and so as a result, the answers were irrelevant anyway, you would always achieve the same result even if they guessed right. The game appears rigged, but it’s not a game, it’s a demonstration. They forgot it was a test and played it like a game.

Which is another unconscious bias at work. In this case, it’s the fact that the people answering the questions are unconsciously being competitive with each other. They can’t help it, it’s just how our brains work. Everyone loves a winner… right?

In this case, that’s something you might not want to tell them, because on some level they’ll unconsciously feel negative towards you if you make them feel like they’re losers.