Search the atlas or filter by field of impact. Each page follows the same structure, so browsing stays fast even when the concept gets slippery.
Ambiguity bias is preferring a known risk over an unknown one, even when the unknown option may have equal or better expected value. Uncertainty about the probabilities themselves feels like extra danger beyond the outcomes at stake.
Anchoring is the pull of an initial number, price, or estimate on later judgment — even when the starting point is arbitrary. We adjust away from the anchor, but usually not far enough.
The anecdotal fallacy is treating one vivid story as evidence of a general rule while ignoring how representative the story is. An anecdote can make a possibility emotionally real; it cannot by itself establish frequency, cause, or typical outcome.
Anthropomorphism is interpreting an animal, object, system, or natural event through human intentions, emotions, or social traits.
Attentional bias is repeatedly noticing information that fits a current concern, goal, fear, or desire while giving less attention to other relevant signals.
Authority bias is giving a claim extra weight because it comes from a prestigious, powerful, or credentialed person, beyond what their relevant evidence and expertise justify. Authority becomes a substitute for checking the claim itself.
Automation bias is giving an automated recommendation more trust than its reliability and the surrounding evidence justify.
The availability heuristic is judging frequency, danger, or importance by how easily examples come to mind. What is vivid, recent, emotional, or repeatedly covered by media feels common — even when the base rate says otherwise.
The bandwagon effect is adopting a belief, product, or action because many other people appear to have adopted it. Popularity becomes a shortcut for truth, quality, or safety — and each new adopter makes the signal look stronger.
Belief bias is judging an argument's logic by whether its conclusion feels believable. A valid argument can be rejected because its conclusion seems false, while an invalid argument can be accepted because its conclusion matches what we already think.
Choice-supportive bias is revising memory so the option we chose looks better and the options we rejected look worse than they did at the time. The decision becomes evidence of its own quality.
The clustering illusion is seeing meaningful runs, clusters, or hot spots in data that random variation can produce on its own. Randomness is lumpy; we expect it to look more evenly mixed than it does.
Confabulation is producing a confident, coherent account to fill a gap in memory or explanation without intending to deceive.
Confirmation bias is the tendency to treat evidence as a lawyer for what we already believe rather than a test of whether we are right. It shapes what we search for, notice, remember, and count as convincing — often without feeling like bias at all.
The conjunction fallacy is judging a detailed story as more likely than a broader category that includes it. A description can feel highly representative, but 'A and B' can never be more probable than 'A' alone.
Conservatism cognitive bias refers to the tendency of individuals to insufficiently revise their beliefs when presented with new evidence. This bias falls under the broader category of information overload, as people tend to give disproportionate weight to their prior knowledge or beliefs and do not adequately adjust them with fresh information. The phenomenon is closely associated with the psychological difficulty of abandoning previously held concepts and the innate preference for consistency.
The continued influence effect is using misinformation in reasoning after learning that it was false. A correction can update a stated belief while the original claim continues to supply the causal story that explains what happened.
The cross-race effect, also known as own-race bias, refers to the tendency for people to more easily recognize faces of their own race compared to faces of other races. This cognitive bias is prevalent in facial recognition tasks and has significant implications in various fields, such as criminal justice, psychological research, and social interactions.
The Delmore effect is the tendency to set clear, detailed, and articulate goals for the lower-priority areas of our lives, while our most important ambitions remain vague and unarticulated. Named after the poet Delmore Schwartz, it describes why people often have precise plans for minor tasks but only hazy intentions for the goals that matter most to them.
The Dunning–Kruger effect is the debated finding that low performers can overestimate their ability, partly because the knowledge needed to perform a task also helps a person judge performance. The popular caricature — ignorant people are always supremely confident and experts always humble — goes far beyond the evidence.
Egocentric bias is overweighting your own perspective, effort, and role because it is more available to you than everyone else's.
The empathy gap is failing to appreciate how strongly a different emotional or bodily state will shape our own or another person's choices. Calm selves underestimate hunger, pain, fear, craving, anger, and stress; hot selves underestimate what calm reflection will want.
A bad bet can become harder to leave precisely because so much has already gone into it. Escalation of commitment is the tendency to add time, money, or reputation to a failing course in order to justify an earlier decision, rather than deciding from the remaining costs and expected benefits.
The false-consensus effect is overestimating how many other people share your view, preference, or behavior.
The frequency illusion, often called the Baader–Meinhof phenomenon, is the sense that something suddenly appears everywhere after we first notice it. Attention changes what enters awareness, and the new sightings are mistaken for evidence that the world changed.
The fundamental attribution error is the tendency to explain other people's behavior by character while underweighting the situation they were in. We see the person clearly and the pressures, incentives, and missing information around them only dimly.
Hindsight bias is seeing an outcome as obvious after it happens and remembering our earlier uncertainty as smaller than it was. Once we know the answer, it becomes hard to reconstruct a world in which the answer was still unknown.
Having a choice can feel like having influence, even when chance or outside conditions determine the result. The illusion of control is the tendency to overestimate how much one’s actions affect an outcome that is largely random or externally constrained.
The illusion of validity is feeling confident in a judgment because the evidence forms a coherent story, even when it predicts poorly.
Illusory correlation is seeing a meaningful relationship between two things because their co-occurrences are vivid, even when the full data do not support one.
Illusory superiority is rating your own ability or character as better than most people's on a trait where that cannot be true for everyone.
The illusory truth effect is finding a repeated claim more believable because familiarity feels like evidence.
Implicit associations are automatic links between social categories and concepts that can be measured in rapid-response tasks, whether or not a person endorses those links explicitly.
Information bias is seeking more information even when it will not change the decision. More data feels like progress and protection from regret, but it can add delay, noise, and a false sense of rigor without adding decision value.
Insensitivity to sample size is treating a striking result from a small sample as though it were as stable as the same result from a large sample.
A smaller option can look better when it is judged on its own, then lose as soon as it is placed beside a clearly superior alternative. The less-is-better effect is this preference reversal: people overweight an easy-to-evaluate cue and underweight total value in separate evaluation.
Leveling and sharpening are complementary memory changes: retellings lose qualifying detail while a few salient details become more central or exaggerated.
Loss aversion is the tendency for a loss to hurt more than an equivalent gain feels good. It makes keeping what we have psychologically compelling, even when a neutral comparison of future outcomes would favor a change.
The masked-man fallacy is wrongly treating a difference in what someone knows about two descriptions as proof that the descriptions refer to different things.
Murphy's Law is a useful engineering maxim, not a cognitive bias: if a failure mode is possible and left unguarded across enough opportunities, it may eventually occur.
Naïve realism is experiencing our own view as a direct reading of reality and explaining disagreement as ignorance, irrationality, or bias in the other person. We forget that our own attention, values, incentives, and prior beliefs also shape what looks obvious.
Negativity bias is giving bad news, criticism, threats, and losses more attention and weight than equally strong good news. One sharp negative signal can dominate a much larger body of positive or neutral evidence.
Neglect of probability is reacting to the size or vividness of a possible outcome while giving too little weight to how likely it is.
Not Invented Here is discounting a useful external idea, tool, or solution because it came from outside the team or organization. Internal ownership feels like evidence of quality; external origin feels like a reason for suspicion.
Observer expectancy is when an evaluator's expectations quietly shape what participants do, what gets noticed, or how ambiguous results are scored.
Occam's razor is a reasoning principle, not a bias: when explanations fit the evidence equally well, prefer the one that adds fewer unsupported assumptions.
Omission bias is judging harm caused by action as worse than equal harm caused by inaction. Doing nothing feels morally cleaner because the causal role is less visible, even when the decision to refrain was itself a choice.
Optimism bias is expecting better outcomes for ourselves than for similar other people. We accept the general risk — layoffs, illness, delays, losses — while quietly treating our own case as the exception.
Outcome bias is treating a decision as good because it worked out, or bad because it did not, even when the result depended heavily on luck.
Overconfidence is believing a judgment is more accurate, a skill stronger, or an outcome more controllable than the evidence warrants. Its most useful form is overprecision: confidence intervals that are far too narrow for how uncertain the world really is.
Pareidolia is perceiving a familiar pattern—often a face, voice, or figure—in ambiguous or random input.
The picture-superiority effect is remembering pictures more readily than equivalent words in many memory tasks. A concrete image can supply a distinctive visual trace as well as a verbal meaning, giving memory more than one route back to the item.
Prejudice is a negative or hostile evaluation of people because of their group membership. Unlike a stereotype, which is a generalized belief, prejudice includes an evaluative attitude that can shape attention, trust, and treatment before individual evidence is considered.
The primacy effect is remembering or overweighting what came first in a sequence. Early items receive more attention and rehearsal, and early impressions can become the frame through which later information is interpreted.
Pro-innovation bias is treating a new technology or idea as inherently beneficial while discounting its trade-offs, failure modes, and distribution of costs.
Projection bias is assuming that future preferences will resemble the preferences we feel right now. Hunger, excitement, loneliness, motivation, and fear masquerade as stable information about what our future selves will want.
The recency effect is remembering or overweighting what came last because it is still active in short-term memory. The newest information can feel most representative even when it is only the latest item in a longer record.
The representativeness heuristic is the tendency to judge the probability that something belongs to a category by how closely it resembles the typical member of that category, rather than by how likely it actually is. Because similarity is easier to assess than probability, we substitute one for the other — and ignore base rates, sample sizes, and chance in the process.
Selective perception is noticing, interpreting, and remembering information through a filter shaped by what you already expect or believe.
Self-serving bias is claiming personal credit for success while assigning failure to circumstances, luck, or other people. The same evidence gets two attribution rules, both of which protect a flattering view of the self.
When a list is long enough to be difficult, its middle is often the least memorable territory. In immediate recall, people commonly recover more items from the beginning and end than from the center—a serial-position pattern rather than proof that the middle was unimportant.
Status quo bias is preferring the current arrangement because it is current, even when a fresh comparison would favor change. The costs of changing are vivid and immediate; the costs of staying put are familiar, dispersed, and easy to leave uncounted.
Stereotyping is using a broad belief about a group to infer traits, ability, motives, or behavior in a particular person. A category replaces individual evidence, often with more confidence than the evidence can support.
The streetlight effect is the tendency to search for answers where searching is easiest rather than where the answer is most likely to be — like the drunkard looking for his keys under the streetlight because 'that's where the light is.' In research and analytics, we measure what is measurable and quietly redefine the question to match.
The subadditivity effect is assigning a lower probability to a whole event than the probabilities assigned to its mutually exclusive parts add up to.
The sunk-cost fallacy is continuing a course of action because of time, money, or effort already spent, even though that investment cannot be recovered. The past feels like a reason to continue; economically, only the value and cost from this moment forward should decide.
Survivorship bias is drawing conclusions from the winners we can see while missing the failures that vanished from view. The visible sample feels like the whole population, so success looks more common — and more easily copied — than it really is.
A change is judged against its background, not in isolation. Adding one candle is obvious in a dark room but barely registers in a brightly lit one. The Weber–Fechner tradition describes this diminishing sensitivity: as a stimulus grows, a larger absolute change is often needed for people to notice a difference.