When a person encounters someone new, they may quickly internally question and categorize the individual based on their own mental representation of that category.

For example, one might think that the guy resembles a yoga instructor, due to his calm demeanor, flexible behavior, healthy lifestyle, and outward positivity.

Her strategy uses the representativeness heuristic—the process whereby judgments of likelihood are based on comparing the similarity between people or events to an existing mental prototype—what one thinks is the most relevant example of the group.

In this case, the shortcut was valid: he is a yoga teacher in this town filled with yogis. Here, most of the members’ traits tend to cluster around the mental image.

In another situation, perhaps she’s asked to judge whether a fashionable woman who has her nose buried into a wine book is a Master Sommelier or a janitor. The use of the representativeness heuristic would lead her to guess Master Sommelier.

However, this strong sense of resemblance leads her to ignore a particular source of information, such as the base-rate likelihood—knowledge about the relative frequency of a category member in the population.

Considering that only 26 women have earned the title of Master Sommelier, there are far more female janitors. In this sense, the sheer mental image does not align with the actual probability of her being a Master Sommelier, and her judgment is incorrect.

In addition, representativeness is at play in assessments between cause and effect. For instance, in astrology, the central belief is that the sign someone’s born under influences their personality.

Upon closer inspection, the personality profiles are based on the principle that “like goes with like.” Thus, a resemblance exists between a sign’s characteristics and the personality traits thought to characterize those within a particular sign. However, multiple studies have indicated that this relationship is also invalid.

In the end, just because something appears to be representative of a group does not make the assessment valid.