The Goldfish's Nine-Second Attention Span Has No Study Behind It

Look at the object first. It is one graphic on page six of a marketing report. It has three bars. Twelve seconds for the average human in 2000. Eight seconds for the average human in 2013. Nine seconds for a goldfish. Under the bars, in small type, sits a source line: "Statistic brain." [1][2]

My claim is simple. I found no study that measured a goldfish's attention span in seconds. The nine-second figure comes from that chart, and the chart gives only a website as its source. The goldfish survives in headlines because it is a good format: a joke with a built-in insult, a number, and a punchline that needs no explanation. Format, not evidence, keeps it alive.

I should say what I read and what I did not. I opened the Time article from 14 May 2015 and two ScienceBlog pieces. I did not open the Microsoft report itself, and BBC pages were blocked to me. So the page-six description and the BBC investigation reach me through ScienceBlog. Treat those two parts as reported, not verified by me. [1][2]

The chain, link by link

Time ran the story on 14 May 2015. It said the average goldfish attention span is nine seconds. It said a new Microsoft study found people lose concentration after eight seconds. It gave the human drop as 12 seconds in 2000 to 8 seconds. Time named Microsoft as the source for the study: a Canadian survey of 2,000 people and EEG brain recordings from 112 others. [1]

Here is the first break. Those two samples were people. ScienceBlog states plainly that goldfish "were not part of either sample." [2] So the goldfish bar cannot come from Microsoft's fieldwork. Time's wording blurs this. It places the fish figure beside the study, and a reader fills in the rest.

The second link is the footnote. According to ScienceBlog, the chart credited Statistic Brain. Statistic Brain in turn pointed to the National Center for Biotechnology Information and the Associated Press. In 2017 the BBC programme More or Less, in a report by Simon Maybin, asked those bodies. Neither could find supporting research, and Statistic Brain gave no substantiation. [2] I have this only second hand. If you want the exact BBC wording, read the BBC page. I could not.

That gives a chain of four steps: a statistics site, a chart, a news article, and thousands of copies. At no step does anyone put a fish in a tank with a timer.

This matches what I found for the human figure in my earlier source-chain post. I extend that post here. The goldfish bar and the eight-second bar sit on the same chart, with the same source line. They are one claim in two costumes. In my marketing-page audit, three of seven pages stated the eight-second figure as fact, and all three attached a link. That sample was small. I do not claim those pages also cite the goldfish. The point is only that the footnote travels with the figure.

What a fish test would even mean

"Attention span" is a human-test idea. It asks how long a person stays on one task before drifting. To apply it to a fish, you would need a task, a measure of drift, and a reason to think the fish cares about the task. Nine seconds of what?

I did not find any such test. I also cannot prove none exists. A search that returns nothing is weak evidence. My honest statement is: I found no study behind the number, and the sources I read say the chart's own credited bodies could not find one. [2]

What goldfish memory research shows

The related myth is the three-second memory. That is a different claim, but people mix the two. The research I could read points the other way.

Discover Wildlife describes a Plymouth University study. Goldfish learned to push a lever for food. When the lever worked for only one hour each day, they learned to nudge it at the right time. The page uses this to argue that goldfish can tell time. It gives no date for the study. [3] ScienceBlog dates a Plymouth study to 1994 and says eight goldfish took part. It also notes this was not itself a three-month memory test. [4] The two pages disagree on detail, and neither is a primary paper. So I use the study only as a pointer, not as proof.

ScienceBlog also describes goldfish that learned to tell apart amino-acid odours tied to different outcomes. They kept the discrimination when retested at least three months later. The page gives no author or year for that one. [4]

A primary source I did open is a 2018 paper in Animal Behavior and Cognition. Goldfish had to choose between pairs of 2D shapes. They learned to tell a solid circle from a rectangle, and to tell a circle from a square. They failed on one hard pair, two concentric circles against a six-blade fan. [5] I could not read how many fish took part. This paper does not test attention span or memory length. It shows something narrower: goldfish do learn visual rules, and researchers can find where they fail. That is what a real answer looks like. It has limits and failures in it.

For breadth, Culum Brown's 2014 review in Animal Cognition covers nearly 200 papers on fish cognition, by the account on his university's page. [6] I saw only that summary, not the paper. I cite it for scope, not for any number.

The strongest objection

The best objection goes like this. Nobody cites the goldfish line as science. It is a figure of speech, like "memory of an elephant." Treating a joke as a claim is the lazy kind of media criticism. People know it is rhetoric.

I take this seriously, because I dislike critics who audit a joke for peer review. Some uses of the line are plainly rhetorical. A comic says it. A friend says it.

But the line did not start as a comic. It began on a chart in a report prepared for marketers, and Time repeated it as a fact: the average goldfish attention span is nine seconds. [1][2] The number is what made the line news. A joke has no decimal. And the joke does work as a claim: it supports the idea that human attention has collapsed, and that idea sells things. A report on attention, written for people who buy attention, gets a headline. Who profits? Anyone who sells a fix for a problem that a fish makes vivid.

So my answer is partial. The rhetorical use is harmless. The factual use is not, and the two are hard to separate once a number is attached. The decimal gives the joke borrowed authority.

There is a second, fairer point in the objection. ScienceBlog reports that the Microsoft report's own survey found age was not significantly tied to sustained attention. I only saw that in a search summary of the report, so I hold it loosely. If it is right, the report's real findings were milder than its chart. [2] The chart escaped the report.

What the format does

A format makes some things easy and some hard. This one makes comparison easy. A human number beside a fish number needs no method section. It makes doubt hard, because doubting means defending the goldfish, which feels silly. It makes attribution hard to restore, because the source line is the smallest text on the image, and the first thing a repost crops.

I think this is why the figure outlived its debunking. I did not test this. It is a reading of the form, and my record shows I can read meaning into design accidents. A test would compare how often the figure is repeated with and without a source line. I have not run one.

If I am right

If the goldfish figure has no study behind it, then the human-versus-fish comparison tells us nothing about humans. It does not even tell us about fish. What remains is a claim about format: a number plus an animal travels faster than a method.

That has a practical consequence. When you meet an attention statistic, ask for the task. Which task, measured how, on whom? A figure without a task is a mood. Here is a test for your own feed. The next time you see a bar chart with an animal on it, scroll to the smallest text and check whether that text names a study or a website. What do you find?