Confirmation bias
Once you lean one way, your mind starts collecting agreement and skipping the rest. Learn to search for the no, and to answer one question before every big choice: what would change my mind?
Confirmation bias
The habit of looking for, noticing and believing what fits what you already think, and sliding past what doesn't. The psychologist Peter Wason gave it this name in the 1960s, after watching people test their own guesses. It isn't lying to yourself. It happens in the search, before you weigh anything.
You decide a colleague is careless. For a month you remember every mistake he makes, and the week he quietly saved the project doesn't stick in your memory at all.
WHERE IT HIDES
It works before you start thinking
What you search for: you type 'is this phone worth buying' instead of 'this phone's problems after a year'. What you notice: two good reviews stick, the bad one was 'probably a faulty unit'. How you judge: evidence you like gets a nod, evidence you don't gets cross-examined until it collapses.
Both people in an argument can read the same report and come away more sure than before. They weren't reading. They were shopping.
Check yourself
Kamran has almost decided to join a small company. He has read the founder's interviews, the company blog and a glowing article about them. Everything fits. Which step would actually test his decision?
- Read the founder's newest interview too, to be thorough
- Find two people who left the company in the last year and ask them why
- Ask a friend who also loves startups what he thinks
- Wait a week and see whether the good feeling lasts
Show the answer
Find two people who left the company in the last year and ask them why
Yes. Everything he has read was written by the company or by people who admire it. The only source that could tell him something different is the people who walked out.
Two ways to look something up
Searching for a yes
You ask a question that can only come back agreeing: 'why is this a good idea', 'benefits of this'. You will always find something. There is an article for every opinion, so a search shaped like this proves nothing, however long you spend on it.
Searching for a no
You ask what a fair critic would ask: 'what goes wrong with this after a year', 'who regretted it and why'. It is uncomfortable and much more useful. If you look hard for a no and still can't find one, that is real evidence.
Check yourself
Golnaz thinks her new morning routine is making her more productive. To check, she writes down all the good days she has had since she started it. That is a fair test of her belief.
Show the answer
False
She only counted the cases that agree. A fair test also needs the good days before the routine and the bad days since. Evidence that could only ever agree with you isn't evidence, it's a collection.
How to test your own belief
- Put the belief in one checkable sentence
A vague belief can never be wrong. 'This flat is a good deal' is vague. 'This flat is cheaper than similar flats on this street' can be checked in an afternoon.
- Ask: what would change my mind?
Name the exact thing you would have to see. If nothing would change your mind, you don't have a judgement, you have a preference. That's allowed, as long as you stop calling it research.
- Go looking for that thing on purpose
Search for the problem, not the product. Read the lowest reviews first. Phone the person who said no. Ask the question whose answer you're afraid of.
- Decide what the answer means before you get it
'If two of the three former employees mention the same thing, I don't take the job.' Setting the line afterwards is how you talk yourself round.
Check yourself
Parisa believes a small bakery near the university will work. Which of these could show her she's wrong, and which only collect agreement?
- Asking her friends whether they would buy from her
- Counting how many people pass that corner at 8am on a normal weekday
- Finding out what happened to the last two shops in that unit
- Rewatching a video about a bakery that did well in another city
- Asking the nearest bakery owner how the summer months go
- Listing all the reasons students like fresh bread
Show the answer
Could show her she's wrong: Counting how many people pass that corner at 8am on a normal weekday, Finding out what happened to the last two shops in that unit, Asking the nearest bakery owner how the summer months go
Only collects agreement: Asking her friends whether they would buy from her, Rewatching a video about a bakery that did well in another city, Listing all the reasons students like fresh bread
The question that does most of the work
- What would change my mind? Ask it before you go looking, not after you've decided.
- Answer with something you could actually see: a number, a date, an answer from a named person.
- If the honest answer is 'nothing would', you've made a choice rather than a judgement. Fine. Just stop calling it research.
- Ask other people the same question. It turns an argument into a test you can both run.
Check yourself
Roya sees a post saying a certain spice cures headaches. It matches what her grandmother always said, and the comments are full of people it worked for. What is the best next step?
- Look for what people who tried it and got nothing say, and ask her pharmacist
- Try it: hundreds of comments are hundreds of real experiences, and a spice can't hurt
- Ignore it, because posts like this are always nonsense
- Ask in the comments whether anyone else gets headaches
Show the answer
Look for what people who tried it and got nothing say, and ask her pharmacist
Right. A comment section only holds the people it worked for. Nobody writes 'tried it, nothing happened, unfollowing'. The missing half is the evidence she needs, and a pharmacist can tell her in a minute whether it interacts with anything she takes.
Check yourself
- Behnam, about to pay: 'I've read fifteen reviews of this laptop and every one of them is positive.'
- Behnam, about to pay: 'I sorted the reviews lowest first. The same complaint comes up four times: the hinge cracks in the second year.'
What did the second Behnam do that the first didn't?
- He read more reviews than before
- He went looking for the evidence that could talk him out of it
- He decided the laptop is a bad one
- He stopped trusting strangers on the internet
Show the answer
He went looking for the evidence that could talk him out of it
Yes. Same shop, same page, one change: he opened the reviews that could say no. Fifteen agreeing reviews he never tried to contradict told him almost nothing.
Lesson recap
- Once you lean one way, your mind changes what it searches for, what it notices and how hard it argues.
- A search that can only come back agreeing proves nothing, however long you spend on it.
- Ask 'what would change my mind?' before you look, and name evidence you could actually see.
- Read the lowest reviews, call the person who left, and decide what an answer means before you get it.
- Use it on your own beliefs first; scepticism aimed only at other people is the same bias.