Frequency Illusion: Why a Word You Just Learned Is Suddenly Everywhere

Two ordinary processes produce the illusion, and neither requires anything in the world to change. The first is the attention filter that decides what, out of everything passing a person in a day, gets noticed. The second is the one-sided way memory counts what has been noticed. Zwicky named only the first in 2005; later writers added the second, and the pair are now the standard account.

Attention selects on relevance, and relevance is set by what is already in memory. Dichotic listening experiments show the filter at work. In Moray's 1959 study, people who repeated aloud a message played to one ear could report almost nothing of a second message played to the other, yet about a third noticed their own name spoken in the ignored ear; Wood and Cowan's 1995 replication found 34.6 percent. A name is a stimulus with a strong entry in memory. A word that has no entry has nothing to be relevant to and goes the way of the ignored channel; learning the word creates the entry. The word's frequency in the world is unchanged, but the chance that an occurrence gets through the filter has gone up, which is what the demo above tries to catch.

The second process is the tally. Once a word is being noticed, each encounter is logged as evidence that it is common. The far larger number of pages, conversations and forecasts in which the word does not appear are not logged, because nobody notices an absence. Nickerson's 1998 review of confirmation bias describes this asymmetry in many settings: people attend to and remember instances that fit a hypothesis more readily than those that do not. The impression of a surge is built from hits alone and grows without a miss ever being counted.

Zwicky described the illusion for people who study language, who are prone to it because noticing usage is their job. His 2005 example was a research group convinced that a new use of the word all was common among young Californians; when the students recorded speech, it turned out to be rare. His remedy still applies: your impressions are not to be trusted, you have to check the facts.

No experiment measures the frequency illusion as such. There is a coinage, and a literature on each of its parts. On August 7, 2005, the Stanford linguist Arnold Zwicky published a Language Log post titled "Just between Dr. Language and I," about a usage that commentators believed was recent and common when it was neither. In it he named two illusions. The recency illusion is the belief that things you have noticed only recently are in fact recent. The frequency illusion is its companion: once you have noticed a phenomenon, you think it happens a whole lot, even all the time, because your estimate of frequency is skewed by your noticing nearly every occurrence that comes past you. He called both selective attention effects.

The experience already had a popular name. In October 1994 a reader of the St. Paul Pioneer Press had described it in the paper's Bulletin Board column and dubbed it the Baader-Meinhof phenomenon, after a German militant group whose name he had met twice in a day. The two names were treated as synonyms in the years that followed, and in March 2019 the Oxford English Dictionary added both, defining the frequency illusion as a quirk of perception whereby a phenomenon to which one is newly alert suddenly seems ubiquitous. Ben Zimmer marked the entry on Language Log that month and traced the term back to Zwicky's post.

The evidence for the mechanism comes from older literatures. Cherry's 1953 experiments on listening to one voice among several established that an unattended message is barely processed. Moray's 1959 follow-up showed that the listener's own name could break through, and Wood and Cowan confirmed it in 1995 with about a third of participants. Nickerson's 1998 review collected decades of work on counting confirming instances and overlooking the rest. The frequency illusion is what those findings feel like from the inside, when the relevant thing is a word you learned on Tuesday.

The illusion appears whenever something becomes relevant and is then met again. The rows below are illustrations, not findings from the sources cited on this page.

The illusion is harmless when it is about words. It matters when a sense of frequency becomes a claim about the world, and the counter is the one Zwicky gave: check.

• Date the noticing. If the sightings began the week you learned the term, your filter changed that week; the world would have had to change on the same schedule for the surge to be real.
• Count the misses. For a day, make a note of the occasions on which the thing could have appeared and did not. The tally in memory has only one column; a second column is the correction.
• Measure the base rate. A word's frequency over time, a product's sales share, a condition's prevalence: each has a number, and the number settles what an impression cannot.
• Expect coincidences from frequent things. Something that occurs a few times a day will often occur twice in an hour. A feeling of pattern is not a count.
• Use it deliberately. A new label in memory is what makes vocabulary stick and lets a trained eye see what a novice misses.

The frequency illusion sits among several effects of attention and tallying, and it is often conflated with them.

• It is not a different effect from the Baader-Meinhof phenomenon; the two are one experience under two names, the popular one from a 1994 newspaper letter and the scholarly one from Zwicky's 2005 post. That page treats the name and the experience; this one treats the mechanism.
• It is not the recency illusion, Zwicky's other coinage from the same post, the belief that something you have only recently noticed is itself recent. The frequency illusion is about how common a thing seems; the recency illusion is about how new it seems. They travel together.
• It is not confirmation bias as a whole, though the one-sided tally is an instance of it. Confirmation bias is the broad preference for evidence that fits a belief; the frequency illusion is what that preference does to an estimate of how often something happens.
• It is not the availability heuristic, in which frequency is judged by how easily examples come to mind. Availability is the rule; the frequency illusion is one way examples become easy to retrieve.