TrueSeeker · Verified claim report Case 4c1d989ff3 · 2026-07-30

§ Claim under review

Goldfish have a memory span of only three seconds.

Circulating claim, as submitted.

Verdict

False

Confidence

High
§

Summary

The claim that goldfish have a three-second memory is false. Peer-reviewed research, including a 1994 University of Plymouth study by Phil Gee published in the Journal of the Experimental Analysis of Behavior, showed goldfish can be trained to press a lever for food available only at a specific time each day, something impossible with a three-second memory. Broader research summarized by outlets such as Live Science, Discover, and BBC Wildlife indicates goldfish can retain learned information for weeks to several months. The origin of the "three seconds" figure appears to be folklore, not science. Goldfish memory is limited compared with mammals, but it is vastly longer than three seconds.

§

The readings

key figures from the evidence
1 hour per day

Feeding window goldfish learned to target in Gee (1994)

§

Why this verdict

The specific quantitative claim (three seconds) is contradicted by peer-reviewed experimental evidence, most notably Gee (1994), in which goldfish learned and retained a time-of-day feeding association that would be impossible with a three-second memory. Multiple independent secondary sources summarizing decades of research converge on retention periods of weeks to months. No scientific source supports the three-second figure.
§

Evidence

Multiple studies over several decades show goldfish can retain learned associations for far longer than three seconds. Studies have repeatedly shown that their memories can actually span several months . Countless studies have shown that goldfish have memories spanning weeks, months, and even years, and that they are also exceptional problem solvers . In the 1994 Plymouth experiment, eight subjects were trained to operate a lever that reinforced each press with food, and the period during which responses were reinforced was then progressively reduced until it was 1 hour in every 24 , and the fish learned to concentrate their lever pressing around the correct feeding time, demonstrating they could remember a time-of-day association.

§

Findings

What's accurate 2

  • Goldfish do have measurable memory (the myth's underlying premise that they can remember at all is correct).
  • Goldfish memory is finite and not perfect.

What's misleading 2

  • Exaggeration / fabrication of magnitude: the specific "three second" figure has no scientific basis. Experimental evidence shows retention over days, weeks, and months, not seconds.
  • Species generalization: the claim applies a single, unsupported number to all goldfish across all memory tasks.

? What's uncertain 2

  • The exact origin of the "three seconds" figure is not clearly documented; it appears to be folk belief rather than a misinterpreted study.
  • Upper bounds of goldfish memory vary by task, individual, and study design; sources report ranges from weeks up to several months or longer.
Distortion flags exaggeration
§

Sources

5 of 5 linked to records
[1]

Gee, P. et al. (1994), "Temporal Discrimination Learning of Operant Feeding in Goldfish (Carassius auratus)," Journal of the Experimental Analysis of Behavior

primary peer-reviewed journal
https://onlinelibrary.wiley.com/doi/abs/10.1901/jeab.1994.62-1 ↗
[2]

Live Science, "Do goldfish really have a 3-second memory?"

secondary quality science journalism
https://www.livescience.com/goldfish-memory.html ↗
[3]

Discover Magazine, "Goldfish May Have a Longer Memory Span Than Just Three Seconds"

secondary science journalism
https://www.discovermagazine.com/goldfish-may-have-longer-memories-than-just-three-seconds-46062 ↗
[4]

IFLScience, "Do Goldfish Have A Three-Second Memory?"

secondary science journalism
https://www.iflscience.com/do-goldfish-really-have-a-three-second-memory-75017 ↗
[5]

Discover Wildlife (BBC Wildlife), Ellen Husain

secondary expert commentary
https://www.discoverwildlife.com/animal-facts/fish/goldfish-memory ↗
How links are chosen. A source is linked only when the address comes from the investigation's own retrieval or from a registry lookup (PubMed, Crossref) that matches the citation's title and year. Author lists shown as registry-verified come from the registry record, not from the report text. Citations that cannot be matched are labeled, never guessed.
This is one case on the record See the full case, browse the archive, and search every checked claim on TrueSeeker Open on trueseeker.com →