Definition
A heuristic (from Greek heuriskein, "to discover") is a mental shortcut — a simplified procedure for answering a question or making a judgment when the full computation is too costly or complex. Heuristics reduce the cognitive effort required for judgment by substituting a manageable proxy for the actual target.
In cognitive psychology, heuristics are primarily studied through the heuristics-and-biases program developed by Daniel Kahneman and Amos Tversky, who documented three canonical heuristics: availability (ease of recall substitutes for frequency), representativeness (similarity to a prototype substitutes for probability), and affect (emotional response substitutes for assessments of risk and benefit). Each heuristic produces reliable, efficient judgments in familiar domains — and predictable, systematic biases when the proxy diverges from the target.
Why it matters
Heuristics as substitution
All three canonical heuristics share a common structure: a hard question is replaced by an easier one. Kahneman calls this heuristic substitution. The substitution is automatic and unconscious — the person answering the proxy question believes they are answering the target question.
| Hard target question | Easier heuristic question substituted | |---|---| | How frequent is X? | How easily can I recall examples of X? (Availability) | | How probable is X given category Y? | How similar is X to the prototype of Y? (Representativeness) | | What is the benefit / risk of X? | How do I feel about X? (Affect) |
The substitution is not foolish — in many domains, the proxy question is genuinely correlated with the target. Ease of recall often tracks frequency; prototype similarity often tracks probability; emotional responses often track genuine value. The failure mode occurs when the correlation breaks down.
When heuristics work and when they fail
Familiar domains with regular feedback: when experience has calibrated the proxy to the target, heuristics produce reliable, fast judgments that match what a slower, more careful analysis would produce. Expert pattern recognition is essentially a validated heuristic.
Unfamiliar domains or systematic biases in exposure: when the proxy diverges from the target because of asymmetric exposure (vivid events are over-recalled regardless of frequency), emotional activation (affect tags reflect media framing rather than actuarial risk), or prototype mismatch (the category's prototype reflects stereotype rather than statistical distribution), heuristics produce systematic errors.