Seeing Patterns and Causes That Are Not There

Start with the broadest habit. Apophenia is the tendency to perceive meaningful connections between unrelated things. Klaus Conrad coined the term in 1958 for the first stages of delusional thinking in psychosis, and psychology later widened it to everyday cases such as faces in clouds.

Small samples of random data inevitably contain streaks and clusters, and people read them as non-random. The page ties the effect to the 1985 hot hand study of basketball shooting by Gilovich, Vallone and Tversky, which found runs in real shooting records that matched chance.

After noticing a word or product once, it seems to turn up everywhere. The page explains this as selective attention plus confirmation bias, with no real rise in how often the thing occurs. The informal name Baader-Meinhof phenomenon was coined in 1994 by Terry Mullen in a letter to a newspaper.

Here the false pattern is a relationship between two things. The term goes back to Chapman in 1967, and the Chapmans' work found that clinicians and lay people reported links between Rorschach responses and diagnoses that were not in the data. The idea is also used to explain how stereotypes form.

People rate vague, general personality descriptions as highly accurate for themselves. Bertram Forer showed this in 1948 with generic statements drawn from horoscopes, and Paul Meehl named it the Barnum effect in 1956. It is used to explain the pull of astrology and fortune telling.

Reading human intentions into non-human things is another way of finding a pattern that is not there. The page treats it as a deeply rooted tendency rather than a simple error, using the same machinery that infers other people's minds. The first empirical study was the 1944 work of Fritz Heider and Marianne Simmel.

A coherent story feels more probable than a plain fact. In the 1983 Linda problem of Tversky and Kahneman, most people judged it more likely that Linda was a bank teller active in the feminist movement than that she was a bank teller, which cannot be true. They attributed it to the representativeness heuristic.

The last two stops are about data rather than perception. Selection bias arises when the individuals or data chosen for analysis are not representative of the population meant to be studied, so a result can look sound in the sample and fail to generalize. Survivorship bias is a commonly cited special case.

Selection can manufacture a relationship that does not exist. Joseph Berkson described in 1946 how studying only hospital patients can make two unrelated diseases look negatively associated, because either one raises the chance of admission. It is a standard warning in epidemiology about data from a filtered subgroup.