Your brain takes shortcuts. It has to — there is far more information than attention, so it guesses, fills gaps, and moves on. Most of the time those guesses are good enough. But some run so consistently in the wrong direction that they quietly shape your money, your judgments and your relationships. Here are seven of the most common cognitive biases, what the evidence actually says about each, and how to push back.
For the bigger picture of why these shortcuts exist and why awareness alone does not fix them, start with our guide to the cognitive biases that control your life. This piece is the field guide to specific ones.
In this article:
- 1. Confirmation bias
- 2. Availability heuristic
- 3. The sunk cost fallacy
- 4. Anchoring
- 5. The Dunning-Kruger effect
- 6. The halo effect
- 7. Negativity bias
- The one defense that works for all of them
- FAQ
1. Confirmation Bias
Confirmation bias is the tendency to seek, notice and remember information that fits what you already believe — and to scrutinise or forget what does not.
The important detail, from Raymond Nickerson’s definitive review of the field, is that this is not one behaviour but at least four that compound: biased search, biased interpretation, biased memory, and biased weighting of evidence you have already conceded. None of them requires dishonesty. You can run the whole sequence while sincerely believing you are following the evidence.
Counter it by asking, before you decide, “What would change my mind?” — and then going to look for that specific thing. We cover the mechanism and the debiasing evidence in depth in confirmation bias explained.
2. Availability Heuristic
We judge how likely something is by how easily examples come to mind. Vivid, recent or emotional events — plane crashes, shark attacks — feel more probable than dull, common ones that actually kill far more people.
The mechanism is fluency, not ignorance. An event that is easy to picture feels frequent, and news coverage systematically makes rare-but-dramatic events easy to picture. This is why perceived risk and actual risk drift apart in a predictable direction rather than randomly.
The counter is to ask for a base rate before you trust a vivid example: not “can I think of a case?” but “how often does this actually happen, out of how many?”
3. The Sunk Cost Fallacy
Money, time or effort already spent pulls us to spend more — finishing the bad film, staying in the dead-end project, holding the losing investment.
Economically the past cost is gone whatever you choose next, so it should carry no weight. Psychologically it carries a great deal, because abandoning the thing means admitting the earlier spend was wasted. The bias is really about protecting a self-image, which is why it interlocks with self-deception, and why it is one of the mechanisms that keeps people in situations they can see are costing them.
The clean test: “Knowing what I know now, and ignoring what I have already put in, would I start this today?”
4. Anchoring
The first number you see drags every later judgment toward it — even when the number is obviously arbitrary. It runs prices, salaries and negotiations.
Anchoring is worth singling out because it is one of the better-replicated effects in the field. The Many Labs replication project ran it across many samples and recovered the effect consistently, which is not true of every classic finding from this era. The direction is robust; the size of any single headline figure deserves more caution.
We break it down fully in anchoring bias: why the first number wins.
5. The Dunning-Kruger Effect
The Dunning-Kruger effect describes how the skills you need to be good at something are often the same skills you need to see that you are not good at it — so beginners can badly overrate themselves.
This one needs a caveat that almost never travels with the meme. A serious statistical critique argues that the famous chart can be reproduced from random data plus two ordinary phenomena: regression to the mean, and the fact that most people rate themselves slightly above average. The British Psychological Society’s summary of the dispute is a fair place to read both sides.
What survives the argument is narrower but still useful: self-assessment accuracy does correlate with actual skill, and the correlation is weaker than the popular version claims. The honest takeaway is not “confident people are incompetent.” It is that your own estimate of your ability is a poor instrument, so get feedback from people who actually know the domain.
6. The Halo Effect
One positive trait — attractiveness, confidence, a good opening line — bleeds into our judgment of unrelated traits, so we assume good-looking or charismatic people are also kinder or more competent.
It shapes hiring, dating and juries, and it is unusually hard to catch because the halo feels like observation rather than inference. It is also the engine behind why first impressions are so hard to undo: the first read colours every later data point instead of being corrected by it.
The counter is procedural, not attitudinal: judge each trait against its own evidence, and where the decision matters, structure it so the traits are assessed separately.
7. Negativity Bias
Bad is stronger than good. One criticism outweighs several compliments; one bad evening rewrites a good month.
This is not folk wisdom — it is the title and conclusion of a major review by Roy Baumeister and colleagues, “Bad Is Stronger Than Good”, which found the asymmetry across everyday events, major life events, close relationships, social networks and learning. Bad information is processed more thoroughly, and bad impressions form faster and resist correction better than good ones.
It probably kept our ancestors alive. It distorts modern relationships and self-image, where the costs of a missed threat are far lower than they used to be. Naming it helps a little; deliberately giving equal airtime to what went right helps more.
The One Defense That Works for All of Them
Knowing a bias by name barely helps in the moment. Biases run faster than reflection, and the research on debiasing is consistent about this: general instructions to “be objective” produce very little correction.
What works is changing the procedure rather than the intention:
- Slow the decision down when it is consequential, and only then.
- Seek the opposite on purpose. Write the two strongest pieces of evidence against your view before deciding.
- Ask for base rates rather than examples.
- Keep a short decision journal so reality can grade your past reasoning — the only feedback loop that actually calibrates you.
The same logic powers good mental models: a model is useful because it is a procedure you can run, not a fact you can know. If you want the wider tour of how prediction, memory and judgement misfire, our collection of psychology facts that change how you see people covers a lot more ground.
FAQ
Can you get rid of cognitive biases?
No. They are built into how a fast, energy-efficient brain works. You cannot delete them, but you can build procedures — checklists, base rates, deliberate delays — that catch them before they cost you.
Does being smart protect you from bias?
Barely, and in some conditions it makes things worse, because a sharper mind builds better arguments for whatever it already believes. Reasoning ability determines the quality of the justification, not the direction of the search.
Which cognitive bias is most dangerous?
For most people, confirmation bias — because it filters all incoming evidence and therefore makes every other bias harder to notice.
Is the Dunning-Kruger effect real?
Partly. The strong popular version is disputed on statistical grounds, but the underlying point holds: people are poor judges of their own ability, and the poorest judges tend to be those with the least skill. Treat it as a reason to seek outside feedback, not as an insult to aim at other people.
Want one sharp, science-backed idea about the mind each week? Join the Mind After Dark newsletter — no noise, unsubscribe anytime.
