The Word That Follows You Home

You learn a new word on a Tuesday. By Friday it is in a podcast you did not choose, a magazine headline at the dentist, a stranger’s offhand remark behind you on the escalator. You buy a black Subaru and the entire city seems to have been quietly tiled in black Subarus while you were not looking. You get a diagnosis from your doctor, an obscure one with a long Latin tail, and suddenly your cousin has it, your neighbor has it, a celebrity disclosed it last week on a morning show.

The instinct, especially late at night, is to read it as a sign. Something out there is speaking to you. The universe, the algorithm, the dead. At minimum, a coincidence so dense it must mean something.

It does mean something. Just not what you think.

The phenomenon has a name, and the name is a small accident of history. In 1994, a man named Terry Mullen wrote a letter to the St. Paul Pioneer Press describing an experience that had been bothering him. He had heard the phrase Baader-Meinhof, the name of a violent West German militant faction from the 1970s, twice in twenty-four hours, after a lifetime of not hearing it at all. The newspaper printed his note in its conversational column. Readers wrote back with their own versions: a forgotten high school friend’s name, a brand of mustard, a kind of moth. The column kept the label, half-joking, and the label escaped into the wider language. ¹

This is how one of the most common cognitive experiences in modern life ended up named after a group of people almost no one was thinking about.

A clinical name for a domestic feeling

Psychologists already had a more technical term for what Mullen had described, though it had been waiting for a wider audience. In 2005, the Stanford linguist Arnold Zwicky published a short essay on his blog and in academic notes calling it the frequency illusion. ² Zwicky was interested in how language behaves when people start paying attention to it. He noticed his students would learn a new construction in his syntax class and then claim, indignantly, that the construction had suddenly appeared everywhere, as if the language itself had reorganized to spite them.

Zwicky’s diagnosis was deflating in the best way. The construction had been there all along. The student had not. What had changed was not the world but the student’s relationship to it: a new filter, freshly installed, was now flagging instances that had previously passed through unremarked.

The frequency illusion, in Zwicky’s account, runs on two cooperating mechanisms. The first is selective attention. The second is confirmation bias. Neither is exotic. Both are old companions of human cognition, well-documented in the experimental literature since the middle of the twentieth century. What is striking is how cleanly they combine to produce the specific, almost spooky sensation of being followed by a word.

The narrow door of consciousness

To understand selective attention, it helps to consider the sheer volume of information that the nervous system intercepts in any given moment. Estimates vary, but a frequently cited figure from the neuroscientist Manfred Zimmermann puts the bandwidth of the human sensory system at roughly eleven million bits per second. ³ The retinas alone are responsible for the lion’s share. The skin, the cochlea, the proprioceptive sensors in your joints, the chemoreceptors in your nose and tongue, all contribute their own torrents.

Conscious awareness, by contrast, processes something like forty to fifty bits per second. The ratio is on the order of two hundred thousand to one. Almost everything your body detects, you never know about. The hum of the refrigerator stops registering. The pressure of the chair against your back vanishes the moment you stand. The advertisement on the bus you walked past this morning is gone, untraceable, even though your eyes recorded it.

This is not a flaw. It is the only way the system can function. A creature that consciously processed every sensory input would be paralyzed. So the brain runs a kind of permanent triage, deciding what is signal and what is noise. The deciders are diffuse: networks involving the thalamus, the prefrontal cortex, the basal ganglia, and a brainstem system known as the reticular activating system, which in older textbooks was sometimes called the “gatekeeper” of consciousness, though the metaphor flatters its tidiness. ⁴

What the gatekeeper attends to is not random. It is shaped, moment by moment, by what is relevant. Salience, in the cognitive science term of art. Threat is salient. Novelty is salient. Anything tagged as personally meaningful, even minutes ago, suddenly becomes salient too.

This is why, the day after you buy your black Subaru, your visual system starts surfacing black Subarus for you. You did not consciously decide to look for them. You did not even ask. But the filter has been adjusted. A category that was previously irrelevant has been promoted, and the gatekeeper now waves it through.

The mind as pattern detector

On its own, selective attention would only make you notice the new thing more often. It would not, by itself, produce the eerie sense that the new thing is multiplying. For that, you need the second mechanism: confirmation bias.

The psychologist Peter Wason coined the term in 1960, in a series of experiments where he asked participants to identify a rule governing sequences of numbers. ⁵ His subjects, almost universally, generated hypotheses and then tested them by searching for confirming examples. They rarely tried to disprove their guesses. When confirmations arrived, they took them as evidence. When non-confirmations arrived, they barely registered them at all.

This asymmetry is not laziness, exactly. It is a deep architectural feature of human cognition. The brain is a prediction machine, and predictions are validated, energetically, by matches rather than mismatches. A hit feels like understanding. A miss feels like nothing.

Apply this to the new word, the new car, the new diagnosis. Once selective attention has tagged the category, every instance of it lights up in memory like a small fluorescent flag. Every non-instance, that is, every car that is not a black Subaru, every page of text that does not contain the new word, is filed away as background, unremarked, unremembered. The hits accumulate. The misses evaporate. By the end of the week, the hits look like a pattern.

This is why the Baader-Meinhof phenomenon has the texture of revelation rather than mere noticing. It is not that you saw the word twice. It is that you cannot remember a single moment in the past week when you didn’t see it. The misses left no trace, so the hits have no competition.

Heuristics and the availability trap

The deeper machinery underneath was charted, more than anyone else, by Daniel Kahneman and Amos Tversky in a series of papers beginning in the early 1970s. Kahneman, who would receive the Nobel Prize in Economic Sciences in 2002 for the work, gave it the name we now use casually: heuristics. ⁶ Mental shortcuts. Inferential rules of thumb that the brain deploys when the cost of careful reasoning outweighs its benefit, which is most of the time.

One heuristic, particularly relevant to the frequency illusion, is the availability heuristic. The rule is short: if examples come to mind easily, the thing they represent must be common. If they come to mind with effort, it must be rare. The shortcut is usually serviceable. Common things really do appear more often, and a lifetime of exposure tends to make common things easier to recall.

But the heuristic misfires whenever something becomes easy to recall for reasons other than frequency. Vividness misfires it. So does emotional charge. So does recency. A plane crash, however statistically unusual, becomes vivid the moment it dominates a news cycle, and for weeks afterward people overestimate the danger of flying. A shark attack on a coastal beach can empty boardwalks across the country. The image is sticky. The brain mistakes the stickiness for evidence.

The new word in your life is sticky in exactly this sense. It is novel, recently encoded, and tagged with the emotional weight of having just been learned. Each time you encounter it again, the encoding deepens, and the next encounter feels even more remarkable. The availability heuristic, doing what it was built to do, looks at how easily the word comes to mind and concludes, sensibly enough, that the word must be everywhere. The conclusion is wrong. The mechanism is working perfectly.

What the experiments actually show

The frequency illusion has been studied directly, though less extensively than its ubiquity might suggest. The clearest demonstrations come from work on what researchers call illusory truth and recognition memory, in which participants are exposed to obscure words or factual claims and then asked, later, to estimate how often the items appeared in a subsequent reading task. The pattern is consistent across studies: participants reliably overestimate the frequency of items they have recently learned. The overestimation persists even when the actual frequency is zero, that is, even when the items did not reappear at all. ⁷

Related work on attentional capture and priming has shown that recently encoded stimuli are detected faster, fixated on longer by the eye, and remembered with greater confidence than control stimuli, even when objective exposure is held constant. ⁸ The work is technical, but the upshot is intuitive. The brain is not a tape recorder. It does not count. It estimates, and its estimates are shaped by how loudly each instance announced itself on the way in.

This is the engine, then, behind the experience that Terry Mullen tried to describe in his letter. He had not heard Baader-Meinhof more often that week. He had heard it more loudly. The first hearing had primed the second. The second had felt impossibly coincidental. Coincidence, in turn, is one of the most attention-capturing experiences the mind can have, which is why he sat down to write a newspaper about it.

The illusion as a feature, not a bug

It is tempting, having taken the illusion apart, to feel slightly embarrassed by it. The brain has been caught cheating. The universe was not, in fact, speaking to anyone. The new word was always there, and the only thing that changed was the listener.

But this framing misses what the mechanism is for. Selective attention and confirmation bias are not bugs in the system. They are the system. They are what allow a finite organism, with finite metabolic resources and finite working memory, to learn anything at all in an environment of catastrophic informational excess.

Consider what it means to become an expert at anything. A radiologist looking at a chest X-ray sees nodules that a layperson would never detect, not because the radiologist’s eyes are sharper but because years of training have reweighted the salience filter to flag certain shapes and shadows as meaningful. A jazz musician hears chord substitutions in a recording that another listener experiences only as warmth. A naturalist walks through what sounds, to the rest of us, like a silent forest and identifies six bird species by ear in under a minute. The forest was never silent. The naturalist’s filter has simply been retuned.

Every one of these experts almost certainly went through their own version of the Baader-Meinhof phase, somewhere in the early years of training. The new pattern, once learned, follows them home. It populates their dreams. It seems to appear in places where, in fact, it had simply been overlooked. The illusion is the noisy cousin of learning itself. It is what learning feels like when the filter is changing faster than the world.

This is the gentle reversal the phenomenon offers, once you stop being unsettled by it. The Baader-Meinhof effect is not a sign that the universe is sending you a message. It is a sign that your attention is doing one of the things it is best at: building a model of what matters to you and then enforcing that model against the firehose of everything else. The price of admission is a brief illusion of frequency. The reward is a world that is, for the first time, legible in a category that used to be invisible.

What it costs to see

There is, however, a darker companion to all of this, and the frequency illusion shares its anatomy with some less benign cognitive patterns. Conspiracy theories, in their structure, are frequency illusions with the brakes off. Once a person believes in a hidden pattern, the same machinery that surfaces black Subarus surfaces evidence of the conspiracy: news items, coincidences, ambiguous statements, all promoted in salience and accumulated without the corrective weight of the misses. Health anxiety operates the same way. Someone who has just read about a rare disease begins to notice symptoms in themselves that, a week earlier, would not have registered as anything at all.

The cure, to the extent there is one, is not to stop noticing. The filter cannot be turned off, and most of the time you would not want it to be. The cure is to develop a small, almost clerical habit of asking, when the universe seems to be repeating itself: what would I have noticed if I were not looking for this? It is a humble question. It tends to puncture the spell.

It also, in its small way, restores the world to something closer to its actual texture. The world is not coordinating with you. It is not multiplying things for your benefit or your warning. It is doing what it has always done, which is to be wildly more abundant than your attention can hold. You are simply, briefly, listening in a new key.

Next time a word follows you home, then, try to notice the noticing itself. Try to feel the small click as your filter rotates. The word was there yesterday too, and last month, and last year, drifting past in the background of conversations you were having about other things. Reality has not changed. Your aperture has. The universe was not waiting for you to catch up. It was just always this loud, and you, finally, are hearing it.

Sources

1. Pacific Standard, “There’s a Name for That: The Baader-Meinhof Phenomenon,” 2013. — https://psmag.com/social-justice/theres-a-name-for-that-the-baader-meinhof-phenomenon-59670
2. Arnold Zwicky, “Just Between Dr. Language and I,” Language Log, 2005. — https://itre.cis.upenn.edu/~myl/languagelog/archives/002386.html
3. Manfred Zimmermann, “The Nervous System in the Context of Information Theory,” in Human Physiology (Schmidt & Thews, eds.), Springer, 1989. — https://link.springer.com/chapter/10.1007/978-3-642-73831-9_7
4. Steven Petersen and Michael Posner, “The Attention System of the Human Brain: 20 Years After,” Annual Review of Neuroscience, 2012. — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3294328/
5. Peter Wason, “On the Failure to Eliminate Hypotheses in a Conceptual Task,” Quarterly Journal of Experimental Psychology, 1960. — https://www.tandfonline.com/doi/abs/10.1080/17470216008416717
6. Daniel Kahneman, Thinking, Fast and Slow, Farrar, Straus and Giroux, 2011. — https://us.macmillan.com/books/9780374533557/thinkingfastandslow
7. Amos Tversky and Daniel Kahneman, “Availability: A Heuristic for Judging Frequency and Probability,” Cognitive Psychology, 1973. — https://www.sciencedirect.com/science/article/abs/pii/0010028573900339
8. Jane Raymond, “Attentional modulation of visual processing,” Annual Review of Neuroscience, 2003. — https://www.annualreviews.org/doi/10.1146/annurev.neuro.26.041002.131039