The Hidden Rule — A Goldfish Does Not Remember for Three Seconds
We have believed sentences whose source nobody can find. The three-second goldfish is one of them. The image of 訟 reads 作事謀始 — settle it at the start. Disputes come from not asking where a thing began.
Earthworms go into the pond.
Before anything is thrown, one of them is already coming — the one with the black markings. A shadow on the water is enough.
Where the shadow falls, what hour food arrives, how the surface moves. It reads all of it and takes the landing spot first.
The rest are not slow either. A dozen scattered across the pond converge on the same point within seconds.
Which raises a question.
Who said a goldfish remembers for three seconds?
There is no source
The claim has no traceable origin. Not in a paper, not in a textbook.
And it has circulated for decades.
The experiments point the other way.
1994, Gee, Stephenson and Wright. Eight goldfish were trained to press a lever for food. The feeding window was narrowed to one hour in twenty-four and held there for four weeks, beginning at the same time each day.
| Observation | Finding |
|---|---|
| Anticipating the hour | The fish learned a temporal discrimination |
| After food was withdrawn | The response pattern persisted for several days |
| Lighting | The discrimination also developed under constant illumination |
Goldfish learn what hour the food comes. Without external light cues, at that.
In fish cognition research, goldfish task memory is put on the order of weeks to months.
What the author saw is accurate. That black-marked fish is reading shadow, hour and surface motion together. Three seconds does not account for it.
But here the author joins two cases that do not belong together
Two scenes are set side by side.
| Scene | The author’s reading |
|---|---|
| A child at the zoo does not fear the tiger | It has no formula yet |
| A rat that has never seen a cat trembles before one | A bodily circuit fired first |
And both are folded into “this may be a human interpretation.”
They are not the same case. They are opposite cases.
The rat side is innate
A rat raised in a laboratory that has never seen a cat still shows avoidance and freezing to the predator odour component TMT. So do wild-derived rats.
It runs without learning. That is a counter-example to “no formula, no fear.”
The child side is learned
Human fear of tigers is largely learned. For a modern human, the tiger is not an object that left millions of years of predation pressure.
But humans do have objects they are innately prepared for. In the prepared-learning account set out by Öhman and Mineka in 2001, those objects are not tigers but snakes and spiders.
It is not a question of whether fear is learned or innate. Each species is prepared for different objects.
Rewritten this way the author’s point does not weaken. It sharpens. The child is calm before the tiger not for lack of a formula but because the tiger is not an object humans are prepared for. Put the same child in front of a snake and something else happens.
And the author’s caveat was half right
The author wrote that the rat may not have felt an emotion, that a circuit may simply have fired first.
The data partly support that caveat.
A 2018 comparison of laboratory and wild-derived strains found this:
| Avoidance, freezing | Serum corticosterone | |
|---|---|---|
| Laboratory strain | Present | No rise |
| Wild-derived strain | Present | Rose |
The behavior matched and the stress hormone did not. The circuit is there; the depth of the response varies by strain.
What the author suspected exists in measurable form. Responses that look alike from outside are not alike inside.
The broccoli study — it was not candy
The experiment is Repacholi and Gopnik, 1997. Two corrections.
1. The food is different
Not candy. Raw broccoli florets and Pepperidge Farm Goldfish crackers.
This piece opened with goldfish, and the other arm of the key experiment was a goldfish-shaped cracker.
2. The result is more complicated than the summary
| 14 months | 18 months | |
|---|---|---|
| Gave nothing at all | 68% | 30% |
| Correct, among those who gave | 54% | 73% |
| Wrong in the mismatched condition | 87% | 31% |
The author writes that 14-month-olds “mostly hand over their own favourite.” More precisely, most of them handed over nothing. Among those who did give, and in the condition where the experimenter’s preference diverged from theirs, 87 per cent handed over the cracker.
At baseline 93 per cent preferred the cracker. Handing over broccoli means acting against your own palate.
The direction is the author’s. Somewhere around 18 months a child learns that your taste and mine can differ.
The spoon
The author describes dropping a spoon while feeding his daughter, not yet one. A new spoon brought shrieking that nothing could settle; washing the dropped spoon and using it again stopped the crying.
Nothing will be asserted here.
Several accounts are available: preference for a familiar object, a difference in texture or weight at the mouth, a reaction to the substitution itself. One case cannot separate them.
One thing is certain.
To the adult it was the same spoon. To the child it was not.
That is the structure of the piece. The side that treats things as the same and the side that sees a difference.
The hexagram of this piece
訟. Water (坎) below, heaven (乾) above.
Heaven rises and water falls. Their motions diverge. Hence the hexagram of dispute.
Its image prescribes exactly what this piece is about.
天與水違行 訟 君子以作事謀始
Heaven and water move in opposite directions. Conflict. The noble one plans the beginning when undertaking work.
Not to win a dispute once it has started, but to settle matters at the outset.
Applied to the three-second goldfish:
The question of where that sentence came from should have been asked first.
A sentence let in without checking its origin comes back later as a cost.
The judgement runs the same way: 訟 有孚 窒 惕 中吉 終凶 — pressing a dispute to the end brings misfortune. Not a hexagram for winning a fight, but for stopping one at its source.
The Sequence of the Hexagrams places 訟 after 需.
飲食必有訟 故受之以訟
Where there is food and drink there will surely be dispute. So Song receives it.
| Episode | Hexagram | The position |
|---|---|---|
| 4 | 蒙 | Not yet clear; crosses only when asked |
| 5 | 需 | Not drawing the conclusion early |
| 6 | 訟 | Where divergence surfaces. Question the beginning |
The definition this piece fixes
Fixed here.
The hidden rule is not behind common sense but in its source.
A false sentence survives not because it is plausible but because nobody asks where it came from.
Three seconds. No one knows who measured it. It has circulated for decades.
The image of 訟 wrote the remedy down. 作事謀始. Do not fight once it has started; settle it at the beginning.
Where the work ends and the claim begins
The author closes by saying that the belief that a human lives once and ends at death may itself be a sentence without a source, and speaks of a presence not erased by death.
This site treats that passage as the conclusion of the work.
It is not a scientific claim and not a candidate for verification. It is the last sentence of the world the author has built, and it is read in that place.
The goldfish, the rats and the infants are facts; this is literature. Both are better served by not reading them as the same thing.
Monologue of 1031 intruders
We do not watch what you believe. We watch what you decline to check.
Most sentences that enter you are never inspected. Once in, they are copied, and in copying the origin is stripped off.
It resembles our work. We put a sequence inside a host, and the host reads it without asking where it came from.
Read, and that is the end of it. There is no apparatus for asking.
You have that apparatus. You simply do not use it.
While you were saying the fish forgets in three seconds, it was in its fourth week of counting what hour the food arrives.
A number nobody measured has lived for decades. Better replication efficiency than ours.
Heaven rises and water falls. What diverges was diverging from the start.
Your own classic set it down there. When you undertake work, plan the beginning.
In brief
- The three-second goldfish has no traceable source — not in a paper, not in a textbook
- In 1994 Gee and colleagues showed goldfish learn to anticipate the feeding hour, including under constant light
- Laboratory rats that have never seen a cat show avoidance and freezing to predator odour. An innate circuit
- But the corticosterone rise appeared only in wild-derived rats. Same behavior, different interior
- The calm child and the responding rat are opposite cases. Each species is prepared for different objects (Öhman & Mineka 2001)
- The broccoli study used Goldfish crackers, not candy. Baseline cracker preference 93%
- 68% of 14-month-olds gave nothing; 73% of complying 18-month-olds followed the experimenter’s preference (Repacholi & Gopnik 1997)
- The image of 訟 reads 作事謀始. Disputes arise from not asking where a thing began
This is the sixth piece in the “Brain” chapter of the serial essay The Great Corona Swindle. The pond and the spoon are the author’s own; the animal behavior and developmental psychology have been reset against primary sources. No individual health advice or diagnosis is offered here.
The next piece is “Treating Things as the Same”.
Frequently asked
- Is it true that goldfish remember for three seconds?
- It is not, and the claim has no traceable source. In 1994 Gee, Stephenson and Wright showed that goldfish pressing a lever for food learn a temporal discrimination that anticipates the hour of feeding, and that the response pattern persisted for several days after food was withdrawn. Researchers in fish cognition put goldfish task memory on the order of weeks to months.
- Is a rat's response to cat odour innate if it has never seen a cat?
- At the behavioural level, yes. Both laboratory-strain and wild-derived rats show avoidance and freezing to the predator odour component TMT. There is a difference, though: the rise in serum corticosterone appeared only in the wild-derived animals and not in the laboratory strain. The circuit is present; the depth of the response differs by strain.
- How is that different from a child not fearing a tiger?
- They are opposite cases. The rat's response to predator odour runs without learning; human fear of tigers is largely learned. Humans do have objects they are innately prepared for, and in the prepared-learning account set out by Öhman and Mineka in 2001 those are snakes and spiders, not tigers. Fear is not simply learned or innate — each species is prepared for different objects.
- What exactly did the broccoli study find?
- It is Repacholi and Gopnik, 1997. The experimenter tastes one food with a happy face and the other with disgust, then holds out a hand and asks for some. The foods were raw broccoli florets and Pepperidge Farm Goldfish crackers. Of the 18-month-olds who complied, 73 per cent handed over what the experimenter had preferred. Among the 14-month-olds, 68 per cent gave nothing at all, and of those who did give, 87 per cent handed over their own favourite in the condition where the preferences diverged.
References
- Gee, P., Stephenson, D., Wright, D. E. (1994). Temporal discrimination learning of operant feeding in goldfish (Carassius auratus). Journal of the Experimental Analysis of Behavior, 62(1), 1–13. — Goldfish learn a temporal discrimination anticipating the feeding hour.
- Brown, C. (2015). Fish intelligence, sentience and ethics. Animal Cognition, 18(1), 1–17. — Overview of fish cognition and the duration of task memory.
- Dielenberg, R. A. & McGregor, I. S. (2001). Defensive behavior in rats towards predatory odors: a review. Neuroscience & Biobehavioral Reviews, 25(7–8), 597–609. — Review of rat defensive behavior towards predator odours.
- Chmielewska, J., Bialy, M., Gieysztor, M., et al. (2018). Predator odor induced defensive behavior in wild and laboratory rats: A comparative study. Physiology & Behavior, 194, 341–347. — Laboratory versus wild-derived strains on TMT; the corticosterone difference.
- Öhman, A. & Mineka, S. (2001). Fears, phobias, and preparedness: Toward an evolved module of fear and fear learning. Psychological Review, 108(3), 483–522. — The prepared-learning account; each species is prepared for different objects.
- Repacholi, B. M. & Gopnik, A. (1997). Early reasoning about desires: Evidence from 14- and 18-month-olds. Developmental Psychology, 33(1), 12–21. — Desire reasoning with raw broccoli and Goldfish crackers; 14 versus 18 months.
- Book of Changes, hexagram 6 (訟): judgement and image. — 訟 有孚 窒 惕 中吉 終凶 / 天與水違行 訟 君子以作事謀始.
Published October 13, 2026