── ── Cognitive bias
Confirmation Bias
Confirmation Bias is the tendency to seek, notice, and overweight evidence that supports what we already believe, while discounting or ignoring evidence that contradicts it. It operates unconsciously and affects experts and novices alike. The countermeasure is to deliberately hunt for what would prove you wrong, rather than gathering reasons you're right.
How it works
The bias works at three stages: we search in places likely to confirm us, we interpret ambiguous data as supporting our view, and we remember confirming cases more vividly than disconfirming ones. None of this feels like bias from the inside — it feels like being right.
The discipline is to invert the question. Instead of asking 'what supports my belief?', ask 'what evidence would change my mind, and have I genuinely looked for it?' Actively seeking disconfirmation is the only reliable counter.
When to use it
- Reading early customer feedback that conveniently validates your product vision
- Evaluating a strategy you're already emotionally committed to
- Reviewing data that supports a decision you've publicly announced
- Hiring or investing when you 'just have a good feeling'
When not to use it
Not a mandate to second-guess every settled conviction indefinitely — past a reasonable search for disconfirmation, you must still decide and act.
Worked example
Wason's selection task
Psychologist Peter Wason's experiments in the 1960s showed that when people are given a rule to test, they overwhelmingly look for cases that confirm it and rarely seek the cases that would falsify it — even though only the falsifying test is logically decisive. The studies demonstrated that the instinct to confirm rather than disprove is deep and systematic, not a failing of the careless.
Why it matters for founders
Founders survive on conviction, which is precisely what makes confirmation bias so dangerous: the early signals that feel like validation are often the ones you went looking for. The discipline of seeking the churned user and the unconvinced buyer is what keeps belief honest. deciqAI's agents actively probe for disconfirming evidence before recommending a path, so a decision rests on what survived scrutiny rather than what flattered the plan.
Install this skill (free, MIT)
npx skills add deciqAI/knowledge-skillsFAQ
If it's unconscious, how can I counter it?
By building disconfirmation into your process: assign someone to argue the opposite, seek out users who churned rather than only fans, and write down in advance what evidence would change your mind.
Isn't some confidence in your beliefs necessary?
Yes — the goal isn't endless doubt but calibrated belief. You earn confidence by having genuinely looked for contrary evidence and not found it, not by avoiding the search.