20 Cognitive Biases With ExamplesThe harder task is catching your own
Most bias lists are excellent at naming other people's mistakes. The useful test is whether a name can interrupt the same shortcut while the judgment is still yours.
Judgment is always working with less time and evidence than it wants. Heuristics make that possible; the same economy can also make an error repeatable. Tversky and Kahneman's landmark work on availability, representativeness, and anchoring describes systematic tendencies, not a license to diagnose whoever disagrees with you.
A bias name earns its keep only when it changes what you inspect next. Confirmation bias should send you toward disconfirming evidence. Anchoring should make you estimate before seeing the opening number. Read each example as a mirror and each question as the part to carry into the next review, estimate, or postmortem.
confirmation bias
seeking, interpreting, or remembering evidence in ways that protect an existing belief
Example: Committed to a redesign, a product lead records praise from beta users but treats every complaint as an onboarding problem.
Check: What evidence would change my mind, and did I look for it?
anchoring
letting an initial value pull later estimates toward it, even when that value is arbitrary
Example: A vendor opens at $120,000. Its $82,000 revision feels modest before procurement has compared a single competing quote.
Check: What would I estimate if I had not seen the first number?
availability heuristic
estimating frequency or probability from how easily examples come to mind
Example: After one widely shared account takeover, a team overfunds that scenario while ignoring a quieter stream of routine access failures.
Check: Am I using a representative sample or the example I can recall fastest?
hindsight bias
remembering an outcome as more foreseeable after learning that it occurred
Example: Once an outage begins, a warning in last week's logs suddenly looks decisive, although nobody escalated it at the time.
Check: What did the record show people actually predicted beforehand?
sunk cost fallacy
continuing because of irrecoverable past investment rather than expected future value
Example: A team spends a fourth sprint rescuing a migration because abandoning three completed sprints feels like waste.
Check: If this project arrived today with no history, would I fund its next step?
loss aversion
giving a prospective loss more psychological weight than a comparable gain
Example: A manager rejects a trial that could save ten hours a week because the possibility of losing two hours during setup dominates the upside.
Check: Would I judge the same outcome differently if it were framed as a gain?
framing effect
changing a judgment when equivalent information is presented in different terms
Example: A treatment described as saving 90 of 100 patients receives a warmer response than the same treatment described as losing 10.
Check: Does my choice survive when the same facts are stated in the opposite frame?
status quo bias
treating the current or default option as preferable without comparing it afresh
Example: A company renews an expensive analytics contract because renewal is automatic and replacement requires an owner.
Check: If neither option were already in place, which one would I choose?
gambler's fallacy
expecting independent random events to compensate for a recent streak
Example: After a fair coin lands heads five times, a player bets heavily on tails because it feels due.
Check: Are these events independent, or can earlier outcomes actually affect the next one?
recency bias
giving the latest observations more influence than the full record warrants
Example: One polished demo dominates a quarterly review despite eleven weeks of missed handoffs and weak documentation.
Check: What conclusion appears when I score the whole period consistently?
halo effect
allowing one favorable impression to color judgments of unrelated qualities
Example: A fluent presenter is assumed to have a sound implementation plan before anyone inspects the dependencies.
Check: Which specific evidence supports each trait I am assigning?
horn effect
allowing one unfavorable impression to contaminate a broader evaluation
Example: A typo on the first page makes a reviewer read an otherwise careful proposal as careless.
Check: Would this flaw matter as much if the rest of the work came from someone else?
fundamental attribution error
overweighting disposition and underweighting circumstances when explaining another person's behavior
Example: A support agent is called inattentive for a slow reply; the reviewer never sees the queue outage that hid the ticket.
Check: What situational constraint could produce the same behavior?
in-group bias
evaluating members of one's own group more favorably than comparable outsiders
Example: Interviewers describe an alumnus's vague answer as promising but call the same answer evasive from another candidate.
Check: Would I score this evidence the same way without the shared affiliation?
projection bias
assuming other people, or one's future self, will share one's current preferences
Example: A founder who enjoys dense dashboards assumes customers want every metric visible at once.
Check: Did the people affected express this preference, or did I supply it for them?
just-world hypothesis
assuming outcomes generally reflect what people deserve
Example: After a layoff, colleagues search for faults in the person dismissed rather than accept that the selection may have been arbitrary.
Check: Am I explaining the outcome with evidence or protecting my belief that the system is fair?
Dunning-Kruger effect
miscalibration that can occur when the knowledge needed to perform a task is also needed to assess that performance
Example: After one SQL course, a developer rates a query production-ready; review reveals missing indexes, locks, and rollback handling.
Check: What independent feedback or objective test calibrates this self-assessment?
selection bias
drawing a conclusion from a sample produced by a nonrepresentative selection process
Example: A cancellation survey reaches only customers who still open company emails, then reports that former users remain highly engaged.
Check: Who had no chance, or little chance, to appear in this sample?
survivorship bias
studying the cases that remain visible while overlooking those filtered out by failure or disappearance
Example: A founder copies habits from ten enduring startups without examining the hundreds that used the same habits and closed.
Check: Where are the failures, departures, or missing cases?
base rate neglect
underweighting how common an outcome is when vivid case-specific information is available
Example: A compelling fraud story drives a risk estimate even though the historical rate for this transaction class is tiny.
Check: What was the outcome rate before I learned the details of this case?
Research basis: the APA defines a heuristic as an efficient strategy that does not guarantee a correct answer. See the APA Dictionary of Psychology, Tversky and Kahneman's 1974 paper, and the original Dunning-Kruger study. The examples above are illustrations, not diagnostic tests.
What the list is for
Twenty labels can make a postmortem sound perceptive. Twenty checks can change the decision before the postmortem exists. That is the point of these examples: not to explain judgment away, but to give it another pass while the evidence can still matter.
Names for the next decision
A working vocabulary makes the hidden shortcut discussable. Segue groups these checks into decision, social, memory, and probability patterns, with a full entry behind every term in the list.
Study cognitive biases