In psychology, the take-the-best heuristic^{[page needed]} is a heuristic (a simple strategy for decision-making) which decides between two alternatives by choosing based on the first cue that discriminates them, where cues are ordered by cue validity (highest to lowest). In the original formulation, the cues were assumed to have binary values (that is, either yes or no) or have an unknown value. The logic of the heuristic is that it bases its choice on the best cue (reason) only and ignores the rest.

Psychologists Gerd Gigerenzer and Daniel Goldstein discovered that the heuristic did surprisingly well at making accurate inferences in real-world environments, such as inferring which of two cities is larger. The heuristic has since been modified and applied to domains from medicine, artificial intelligence, and political forecasting.^{[page needed][page needed]} The heuristic has been used to accurately model how experts, such as airport customs officers^{[page needed]} and professional burglars, make decisions;^{[page needed]} the model often makes better predictions of human behavior than more complex models that assume experts integrate all available cues.^{[page needed]}

Theories of decision making typically assume that all relevant reasons (features or cues) are searched and integrated into a final decision. Yet under uncertainty (as opposed to risk), the relevant cues are typically not all known, nor are their precise weights and the correlations between cues. In these situations, relying only on the best cue available may be a reasonable alternative that allows for fast, frugal, and accurate decisions. This is the logic of a class of heuristics known as "one-reason decision making," which includes take-the-best.^{[page needed]} Consider cues with binary values (0, 1), where 1 indicates the cue value that is associated with a higher criterion value. The task is to infer which of two alternatives has the higher criterion value. An example is which of two NBA teams will win the game, based on cues such as home match and who won the last match. The take-the-best heuristic entails three steps to make such an inference:^{[page needed]}

Search rule: Look through cues in the order of their validity.

Stopping rule: Stop search when the first cue is found where the values of the two alternatives differ.

Decision rule: Predict that the alternative with the higher cue value has the higher value on the outcome variable.

The validity ${\textstyle v}$ of a cue is given by ${\textstyle v=C/(C+W)}$, where ${\textstyle C}$ is the number of correct inferences when a cue discriminates, and ${\textstyle W}$ is the number of wrong inferences, all estimated from samples.

Consider the task to infer which object, ${\textstyle A}$ or ${\textstyle B}$, has a higher value on a numerical criterion. As an example imagine someone having to judge whether the German city of Cologne has a larger population than the other German city of Stuttgart. This judgment or inference has to be based on information provided by binary cues, like "Is the city a state capital?". From a formal point of view, the task is a categorization: A pair ${\textstyle \left(A,B\right)}$ is to be categorized as ${\textstyle X_{A}>X_{B}}$ or ${\textstyle X_{B}>X_{A}}$ (where ${\textstyle X}$ denotes the criterion), based on cue information.