Cognitive bias
A cognitive bias is a systematic pattern of deviation from sound judgement.
Human judgement does not err at random. It leans in consistent, nameable directions, and those leanings are cognitive biases.
A cognitive bias is a systematic pattern of deviation from sound judgement, a way of thinking that reliably departs from what logic or evidence would dictate. Because biases are systematic rather than random, they can be identified, predicted, and sometimes corrected.
Systematic, not random
The crucial word is systematic. If people simply made scattered mistakes, there would be little to study, the errors would cancel out, and no pattern would emerge. Cognitive biases are different: large numbers of people err in the same direction in the same situations. Loss aversion, confirmation bias, overconfidence, and anchoring are not occasional slips but reliable tendencies, which is what makes them both scientifically tractable and practically dangerous.
Where they come from
Most biases are the by-products of mental shortcuts that usually work. The mind evolved to decide quickly under uncertainty, and the heuristics that make it fast also make it predictably wrong in conditions those heuristics were not built for. A bias is often a sensible rule applied outside its proper range, which is why biases are so persistent: they are bound up with the very efficiency that makes human cognition workable.
Knowing is not enough
A sobering finding is that simply learning about a bias does little to remove it. Biases operate largely below awareness, and people readily see them in others while missing them in themselves. Correcting for bias usually requires changing the situation rather than exhorting the individual: structured decision processes, outside checks, data instead of impression, designed defaults. Willpower and awareness are weak defences against a bias; better-designed decisions are strong ones.
Cognitive biases are not signs of stupidity but features of normal minds, and they appear in experts and novices alike. Taking them seriously means designing decisions, in business, policy, and personal life, on the assumption that judgement will lean predictably astray, and building in the checks that catch it.