Two 9-Second Glances: What a Cat Learns Before You’ve Finished a Sentence
Thirty-one cats sat in front of a screen at Azabu University in Kanagawa, Japan, and needed only about four trials, two 9-second looks at a single image, to link it to a word that meant absolutely nothing. That’s roughly 18 seconds of exposure per word-picture pair. A cat wired together a sound and a picture faster than it takes most people to read this paragraph twice.
The cats, 20 males and 11 females, came from three cat cafés and eight private homes near the university. Saho Takagi, with co-authors Koyasu, Nagasawa, and Kikusui, published the work as “Rapid formation of picture-word association in cats” in Scientific Reports in 2024. The made-up words were “keraru,” paired with a growing and shrinking creature resembling a unicorn, and “parumo,” paired with an expanding red-faced sun. Neither word exists in Japanese or English. That was the design: no cat could have picked either one up anywhere except that room.
For comparison, the paper notes that human infants tested with the same kind of setup usually get at least four 20-second trials per pair, more than four times the total exposure the cats needed to get there.
How You Catch a Cat Thinking
You can’t ask a cat what it just learned, so the researchers borrowed a method from infant psychology called the switch task. A cat first watches an image paired with a word until it stops reacting to it, habituated, in the jargon, meaning the brain has filed the pairing as old news. Then researchers swap it: same word, wrong picture now. A cat that had genuinely formed an association notices something is off and stares. One that hadn’t looks away at the same rate as before.
In the test phase, cats spent 33% more time looking at the screen on average when the word and picture no longer matched, evidence that an expectation had been violated. The researchers ran a second version of the setup with electronic beeps instead of human speech, and that version produced no such reaction. So the effect showed up only when a human voice was doing the talking.
Takagi has described watching some cats’ pupils dilate during the switched trials, and said it was striking, even charming, how seriously the animals seemed to take part. That’s her impression from being in the room, not a measured result, but it lines up with what the gaze data show.
Why Would a Cat Even Bother?
Dogs were bred for thousands of years to work off human words and gestures; understanding a command was, quite literally, the job description. Cats were never bred for that. Deliberate cat breeding only began in the last few centuries, and mostly for looks, not obedience. So where does this kind of listening speed come from in an animal nobody selected for it?
Takagi’s proposed answer is that it may be a leftover of self-domestication: cats moved in near humans and each other on their own terms, and tracking who is who and what goes with what may have been useful currency long before anyone offered a reward for getting it right. That’s her interpretation, not a settled conclusion, but it recasts whatever’s happening on a cat café floor as an old habit rather than a new trick.
The Myth This Kills
The standing assumption is that cats simply don’t process human words, that whatever a dog can do with language, a cat is constitutionally unequipped for. The Azabu data contradict that directly: ordinary house cats and café cats, with no training and no reward on offer, formed a working link between a sound and a picture inside a few trials. Nobody handed them treats for getting “keraru” right. They just noticed.
What the study does not support is the opposite overcorrection, crowning cats the sharper listeners. Dog owners report their dogs know an average of 89 words, and the most “gifted” dogs can pick up as many as 12 new words in a single session. Those dogs are being tested on retrieval: hear the word, go get the object. The Azabu cats were never asked to fetch anything. They were tested on pure association, whether a sound and a picture had quietly linked up with no reward attached.
Shany Dror, a cognitive researcher at the University of Vienna not involved in the study, has said the data don’t actually support a direct comparison between cats and dogs on this. Two different tasks were never running the same race.
What Nobody Knows Yet
The study’s own authors say they can’t yet explain why cats form these links this quickly, and that answering it will take work from both an evolutionary and a developmental angle. There are gaps built into the design itself, too. Brittany Florkiewicz, commenting on the study, has pointed out that the original infant version of this test used one-syllable words spoken in an unfamiliar voice, while the cats heard three-syllable made-up words in exaggerated speech from their own caregivers, a real mismatch that makes the two experiments harder to line up than they first appear. Coverage of the study by LiveScience cautioned against reading the cat-toddler comparison too literally.
Nobody has tested how long a cat holds onto a link like “parumo means sun,” or whether that same cat, hearing the word in its own living room, would glance at an actual object instead of a screen. And with a study mixing cat-café cats against household cats, whether a busier, more social environment changes any of this is still a question waiting for its experiment, not an answer anyone can quote yet.