Glossary
Heuristic
What is a Heuristic?
A heuristic is a shortcut for making a judgment or decision without examining every relevant detail. A traveler might judge the risk of flying by how easily plane-crash stories come to mind. Someone choosing a restaurant might stop searching as soon as one meets their requirements. Both simplify a decision. Whether the shortcut works depends on the task and the information it uses.
Examples of heuristics
In their influential 1974 paper, Amos Tversky and Daniel Kahneman described three judgment shortcuts:
- Representativeness: judging probability by resemblance to a category or prototype.
- Availability: judging frequency or likelihood by how easily examples come to mind.
- Anchoring and adjustment: starting with a reference value and adjusting from it.
These shortcuts can produce systematic errors. A vivid news story, for example, can be easier to recall than a more common event that receives little attention.
Heuristic thinking and problem solving
Heuristic thinking, or heuristic reasoning, means using a simplifying rule to reach a judgment. To think heuristically is to use such a shortcut. For example, someone choosing a restaurant could check only places within walking distance and select the first that meets their price and dietary requirements. This solves the immediate problem without comparing every restaurant, but it does not guarantee the best possible choice.
Are heuristics always bad?
No. Gerd Gigerenzer and Wolfgang Gaissmaier's review examines how simple rules can work well in environments that suit them. Ignoring information does not always reduce accuracy. It can also reduce the cost of searching and comparing.
This connects to Herbert Simon's ideas about bounded rationality and satisficing: people have limited time and information, and may look for an acceptable option rather than an impossible-to-verify best option.
Why heuristics matter
The useful question is which shortcut fits which situation. Memorizing a list of biases will not answer it. A decision rule needs to be evaluated against the actual task, the available information, and the cost of being wrong.