Discover Magazine looks at what scientists have learned about goldfish memory, checked claim by claim.
51
claims in 33 statements
Published by the Pikaan team. This review is read-only, and reading it uses no credits.
Every claim in the source, in order. The summary under each claim is written by AI from its sources; open the sources to check it.
Claim 1 of 51Needs review
Factual claim · Statement 1 of 33
Every source that addresses the claim calls the three-second memory a myth: World Animal Rescue Network answers 'No' and says trained goldfish retain feeding cues and maze routes for months, IFLScience calls it a 'common misconception' and cites memories spanning weeks, months and even years, and Live Science says goldfish memories can last weeks, months or even years. Inkl notes the three-second figure was never based on any study, and Sciencing calls the belief 'entirely false'.
Pikaan found this exact sentence on the publisher’s page. Finding it does not show whether it supports or contradicts the claim.
Factual claim · Statement 2 of 33 · Strong evidence
Shortform and ScienceABC state that scientists have known since the 1950s-1960s, through controlled experiments, that goldfish memory is far longer than three seconds. Trueseeker and debunked.info likewise describe multiple studies over several decades contradicting the three-second claim, directly supporting the claim that this has been known for decades.
Factual claim · Statement 3 of 33 · Strong evidence
The evidence consistently reports that goldfish memory research has found retention spanning months, not just seconds. Sciencing notes goldfish retained maze maps after several months, Ripley's says their memories last several months, and A-Z Animals cites studies finding long-term memories lasting weeks or months. Live Science likewise reports memories lasting weeks, months, and even years, while notes the exact upper limit varies by task and study design.
Factual claim · Statement 4 of 33 · Good evidence
Independent research pages describe a recent surge in fish-cognition research and report goldfish abilities in object perception, numerical judgment comparable to trained humans/primates, spatial cognition, and memory-related brain regions analogous to human ones. Together these support the claim that goldfish share several cognitive abilities with humans, though the evidence speaks of specific abilities rather than blanket equivalence.
Factual claim · Good evidence
A recent surge in research on fish brain function has found that goldfish share many of the same cognitive abilities as humans.
Research at RIT describes goldfish work on visual object perception and states that goldfish may recognize objects from any orientation, called object constancy, just like humans. A study in the NIH paper found goldfish successfully distinguished rotated photographs of a turtle and frog, which the authors call evidence of object constancy, a process essential to human visual perception.
Factual claim
A recent surge in research on fish brain function has found that goldfish share many of the same cognitive abilities as humans.
The Wikipedia page says goldfish have social learning skills, and gov.scot reports fish social learning and shared vertebrate numerosity processing with humans. But Pond Informer says evidence for social learning in goldfish is contradictory, with one study finding that watching other goldfish slowed learning and describing their social intelligence as relatively low.
Pikaan found this exact sentence on the publisher’s page. Finding it does not show whether it supports or contradicts the claim.
Factual claim
A recent surge in research on fish brain function has found that goldfish share many of the same cognitive abilities as humans.
The sources found do not settle this claim.
No source links were saved with this assessment. Treat this answer as limited until it is checked again.
No source links were published for this claim.
Factual claim · Statement 5 of 33
The claim that the hippocampus is essential for mammalian memory is supported, and the Humane League and AMNH pages say fish lack a mammalian-style hippocampus. However, fish have hippocampal homologues such as the lateral pallium/dorsolateral telencephalon, non-mammalian vertebrates have a homologous structure, and one source describes the hippocampus as a medial pallium found in all vertebrates, so the claim is accurate only under a narrow anatomical definition and needs that context.
These links belong to the claim, but no exact sentence was found on their pages. Review the sources before relying on this result.
Factual claim · Statement 6 of 33 · Strong evidence
The AMNH page states that animals with higher-level brain function and a hippocampus have that structure as the part of the cortex that stores explicit memory, and a physiology source identifies the hippocampus as a key component for explicit-memory storage in higher vertebrates. A textbook quotation on the NIH page likewise says the hippocampus stores declarative information; the Weizmann page adds that long-term storage is distributed in the cortex even though the hippocampus is required for storage.
Factual claim · Strong evidence
In many animals with higher-level brain function, the hippocampus stores explicit memory.
Multiple independent sources define explicit memory exactly as the claim does: a type of long-term memory involving conscious recall of facts and events, including both semantic (facts) and episodic (firsthand personal experiences) memory. The claim's combination of "specific facts and firsthand experiences" matches the standard two-subtype description.
Quote · Statement 7 of 33 · Strong evidence
The Nahf pages state plainly that the hippocampus is absent in fish and that this 'led many people to believe goldfish don't have the ability to store memories at all', and the Saturday Evening Post likewise says fish 'were believed to have no memory' because their brains lack a hippocampus. The claim accurately describes this widely held misconception linking the missing hippocampus to a total absence of memory.
Factual claim
For years, many pet owners believed that the lack of a hippocampus meant goldfish didn’t have the ability to store memories at all.
The evidence establishes that the three-second goldfish memory claim is a myth and that its origins are uncertain or cultural, but it does not establish the specific causal claim that a belief about goldfish lacking memory storage led to the myth. The sources that address origins say there is no single origin and no scientific study behind the three-second figure, while another attributes its spread to media repetition.
These links belong to the claim, but no exact sentence was found on their pages. Review the sources before relying on this result.
Factual claim · Statement 8 of 33
The IFLScience article directly identifies Adelaide Sibeaux as a behavioral ecologist who researches animal navigation at the University of Oxford, matching the claim.
These links belong to the claim, but no exact sentence was found on their pages. Review the sources before relying on this result.
Factual claim
Adelaide Sibeaux is a behavioral ecologist.
The IFLScience article directly identifies Adelaide Sibeaux as a behavioral ecologist “who researches navigation in animals at the University of Oxford”. Other sources likewise describe her as an Oxford researcher working on fish navigation and spatial navigation studies.
These links belong to the claim, but no exact sentence was found on their pages. Review the sources before relying on this result.
Quote · Strong evidence
Adelaide Sibeaux is a behavioral ecologist.
Evidiant directly reports that researcher Adelaide Sibeaux attributes the misconception to people keeping fish in a circular bowl. A second source similarly says Sibeaux notes that public perception is shaped by how fish are kept, including tiny glass bowls.
Factual claim · Statement 9 of 33 · Strong evidence
Multiple independent sources establish that controlled experiments going back to the 1960s showed goldfish retaining trained tasks, objects, and feeding cues for weeks or months, with mid-1960s PNAS shock-avoidance work providing early evidence that they formed and stored memories. Live Science also notes the science backing longer goldfish memories has been around for more than 60 years. The claim that research as early as the 1960s proved goldfish have much longer memories than three seconds is therefore supported.
Factual claim · Statement 10 of 33 · Strong evidence
Multiple sources, including IFLScience and Nahf, state that in 1966 researchers at the University of Michigan ran a shock-avoidance study in which goldfish improved over multiple trials and days. A 1966 PNAS paper from Michigan's biochemistry and mental health research departments reports goldfish increasing avoidance responses across 20 trials and improving further in retraining trials three days later, which matches the claim's details.
Factual claim · Strong evidence
In 1966, scientists at the University of Michigan conducted a study showing that goldfish improved at a shock-avoidance exercise over the course of multiple trials and multiple days.
IFLScience and Nahf both describe a 1966 University of Michigan study in which goldfish improved at a shock-avoidance task over multiple trials and days, providing evidence of short-term memory formation. Another source confirms the study is real and identifies it as Davis and Agranoff's 1966 PNAS paper on stages of memory formation in goldfish.
Factual claim · Statement 11 of 33
Goldfish have genuinely been trained on memory- and perception-based tasks, and one overview states they can recognise individual faces, including those of their keepers. But the actual face-recognition experiments reported by the news outlets were run on archerfish, not goldfish: fish were trained to spit at a chosen human face for a food reward, and coverage of that 2016 study says 'fish can recognise patterns in the human face' and 'some fish can learn to recognise human faces'. The sentences that most closely mirror the claim read as restatements of the claim itself rather than independent confirmation of goldfish (as opposed to archerfish) face training.
Pikaan found this exact sentence on the publisher’s page. Finding it does not show whether it supports or contradicts the claim.
Factual claim · Strong evidence
Scientists have consistently succeeded in training goldfish on memory- and perception-related exercises, teaching them to recognize human faces.
Multiple independent outlets — The Independent, Discovery, NPR, NBC News and Reuters — report that researchers at Ben-Gurion University in Israel trained goldfish to operate a fish-operated vehicle, a robotic car/tank on wheels steered by the fish's movements. That matches the claim precisely.
Factual claim · Statement 12 of 33 · Strong evidence
The Nahf pages state directly that goldfish can connect events and objects at different points in time and that improved performance over trials and days, plus tasks such as recognizing faces or driving a robotic car, are evidence of this. An independent page also says fish have a lateral pallium responsible for connecting events at different points in time. This matches the claim's central point.
Factual claim · Good evidence
Goldfish skills in memory- and perception-related exercises provide evidence that goldfish are able to connect events and objects at different points in time.
Several pages state directly that connecting events and objects at different points in time is a key indicator of explicit memory. Independent definitions of explicit and episodic memory describe it as involving when and where events occurred, including time-stamped recall and spatial and temporal context, which supports the same point.
Factual claim · Statement 13 of 33
The claim is supported by a review finding a dramatic increase in fish cognition studies over the last decade, with 68% of identified studies published between 2010 and 2019. Other sources likewise describe considerable growth in fish cognition research over the last decade and report that 2010–2019 accounts for about 70% of studies on fish cognition.
These links belong to the claim, but no exact sentence was found on their pages. Review the sources before relying on this result.
Factual claim · Statement 14 of 33 · Strong evidence
A 2022 University of Oxford study led by Dr Adelaide Sibeaux, published in Proceedings of the Royal Society B, trained goldfish to swim a set distance and found they could accurately estimate distance traveled using optic flow. The Oxford University release states most goldfish accurately estimated the set distance, and other outlets confirm the study and its authorship.
Factual claim · Statement 15 of 33 · Strong evidence
The original study, published on PMC (nih.gov) and reported by the University of Oxford, describes training goldfish to swim a set distance in a narrow tank covered with a vertical striped pattern. Independent write-ups confirm the setup: a tank lined with vertical stripes used to train the fish to a target distance.
Factual claim · Strong evidence
Researchers trained goldfish to swim a set distance through a tank lined with vertical stripes.
The nazh.org page states the method directly: researchers trained goldfish to swim a set distance through a tank lined with vertical stripes and tracked how their swimming pattern changed when the stripe pattern was altered. The University of Oxford release confirms the team tested whether fish would swim the same distance when the background pattern was manipulated, and that altering the stripe frequency to 1 cm stripes made goldfish overestimate distance by 36 percent, as does the PMC paper.
Reported claim · Statement 16 of 33
The Discover Magazine sentence quoted on Merriam-Webster states exactly that researchers concluded goldfish navigate their surroundings and remember visual cues via optic flow. Other reports say scientists proved goldfish have good memories and can navigate, and that the Oxford study showed goldfish estimate distance using optic flow. A study summary likewise says goldfish can reproduce a learned distance and use optic flow information to do so.
This passage repeats the speaker’s own words. It shows the words were said, not that they are true.
Reported claim · Strong evidence
Researchers concluded that goldfish are able to navigate their surroundings and remember visual cues through a process known as optic flow.
Multiple independent reports of the 2022 University of Oxford study — the BBC, Earth.com, Phys.org's coverage and the study itself on PMC/NIH — state that researchers concluded goldfish estimate distance by using optic flow, i.e. The apparent motion of objects around them. The detail that goldfish track the motion of surrounding objects matches the study's conclusion as described in and.
Quote · Statement 17 of 33
Pikaan is not showing a result or sources for this claim until its wording is fixed.
No source links were published for this claim.
Underlying fact
Sibeaux stated that these results could position goldfish as a model system to investigate the evolution of the mechanisms that underpin spatial cognition in vertebrates.
Pikaan is not showing a result or sources for this claim until its wording is fixed.
No source links were published for this claim.
Factual claim · Statement 18 of 33 · Strong evidence
The Vet Desk says goldfish intelligence is not easily comparable to humans and that they do not show intelligence in the same way humans do, while the BBC report says goldfish appear to process optic-flow information differently from terrestrial animals including humans. Revlox likewise states that this does not mean goldfish experience the world exactly as humans do, and IFLScience notes fish lack a hippocampus. The evidence supports the cautious claim that goldfish may remember and process their surroundings differently from humans, not that they lack memory.
Factual claim · Strong evidence
Goldfish may not have the ability to remember and process their surroundings in the same way that humans do.
The evidence directly states that goldfish cognitive abilities are higher than previously believed. Says results suggest goldfish can perform cognitive functions at a much higher level than previously thought, says their cognitive faculties are more advanced than previously thought, and says numerous studies demonstrate goldfish are far more capable than previously thought.
Factual claim · Statement 19 of 33 · Good evidence
The NIH paper “Visual Perception of Photographs of Rotated 3D Objects in Goldfish” reports that six goldfish were presented with photographs of a plastic model turtle and frog and that all successfully distinguished between the turtle and frog photographs. Secondary summaries also describe this as a 2022 study showing goldfish can distinguish photographs of plastic turtles and plastic frogs.
Factual claim · Statement 20 of 33 · Strong evidence
The sources define object constancy as the ability to recognize an object despite changes in appearance caused by viewing it from different angles, and describe it as recognizing objects at different viewpoints. Other sources give the same definition: recognizing an object from vastly different perspectives or recognizing objects across different viewpoints.
Estimate · Strong evidence
The process of object constancy is the ability to recognize different objects from different angles.
The evidence directly reports that goldfish show object constancy and that this ability is treated as an indicator of intelligence, with the Nahf page saying it suggests a more complex memory system than previously thought. The NIH study reports successful object-constancy performance in goldfish, and the Springer/RIT-linked paper likewise discusses goldfish achieving visual object constancy; an independent university page says the studies make goldfish appear smarter than some people think.
Factual claim · Statement 21 of 33
Several sources state that fish do not have a hippocampus in the mammalian or structural sense. However, comparative neuroscience sources describe a homologous or analogous pallial region in fish, such as the lateral pallium or dorsolateral telencephalon, that performs hippocampus-like spatial memory functions. The claim is therefore accurate only with that structural qualifier; stated flatly, it omits important context.
These links belong to the claim, but no exact sentence was found on their pages. Review the sources before relying on this result.
Factual claim · Strong evidence
Fish don’t have a hippocampus.
The evidence states that the pallium is a shared vertebrate brain region and that teleost fish have a dorsal telencephalon/pallium with homologies to mammalian pallial regions. Fish pallium is everted and anatomically distinct, and a neocortex-like region remains debated, but those caveats do not contradict the claim that fish have a similar pallium region.
Quote
Fish don’t have a hippocampus.
The Nahf page states that the pallium helps goldfish connect events in time and space, and The Humane League page says the lateral pallium is responsible for connecting events at different points in time. NIH review sources similarly describe the hippocampal pallium of teleost fish as involved in associating temporally separate events and in navigating the spatial and temporal dimensions of relational memories.
These links belong to the claim, but no exact sentence was found on their pages. Review the sources before relying on this result.
Factual claim · Statement 22 of 33 · Strong evidence
Two independent pages state the claim almost word for word: Merriam-Webster's example sentence attributed to Discover Magazine and nahf.org's goldfish memory article both say research has split on which part of the pallium stores spatial memory. A 2023 goldfish lesion study from Ben-Gurion University backs this up, noting that lesion and electrophysiology studies differ on whether navigation sits in the lateral or medial pallium, and that its own results question any single-region account.
Factual claim · Statement 23 of 33 · Good evidence
The study itself, 'Lateral and medial telencephalic pallium lesions impair spatial memory in goldfish' (published 1 Nov 2023, per S22), lesioned goldfish in either the lateral or medial pallium and retested them, concluding that 'both the medial and lateral pallium are essential for the coding of spatial memory' and that this challenges the idea that one distinct region is responsible. Summaries on nahf.org repeat the same finding. One caveat the abstract records is that the lateral lesion group declined on the breadboard task but not the plus-maze, while the medial lesion affected both, so 'regardless of which part' holds for the study's overall conclusion rather than for every task.
Estimate · Statement 24 of 33 · Strong evidence
The Nahf article directly states that the study suggests goldfish may use both parts of the pallium for memory. An independent BGU source likewise reports that both the medial and lateral pallium are essential for spatial memory in goldfish. The claim is hedged as what the study suggests, which matches this evidence.
Factual claim
The study suggests that goldfish may use both parts of the pallium for memory.
The claim does not identify which study or which 'part' is meant. The retrieved evidence describes several unrelated studies challenging different assumptions, such as the striatum/motor-cortex division of labor and global versus limb-specific ANS responses, but no sentence ties a specific study to a prior assumption that only one part was involved.
These links belong to the claim, but no exact sentence was found on their pages. Review the sources before relying on this result.
Quote · Statement 25 of 33
The claim attributes a specific statement to Sibeaux, but none of the retrieved sentences report Sibeaux saying anything about sweeping change, humans, or global change; the only Sibeaux material in the source context concerns goldfish as a model for spatial cognition appear only as general background about change in the world and human cultures, not as anything Sibeaux said.
These links belong to the claim, but no exact sentence was found on their pages. Review the sources before relying on this result.
Quote · Statement 26 of 33
The sources found do not settle this claim.
No source links were saved with this assessment. Treat this answer as limited until it is checked again.
No source links were published for this claim.
Factual claim · Statement 27 of 33
Pikaan is not showing a result or sources for this claim until its wording is fixed.
No source links were published for this claim.
Factual claim · Statement 28 of 33
Pikaan is not showing a result or sources for this claim until its wording is fixed.
No source links were published for this claim.
Factual claim · Statement 29 of 33
Pikaan is not showing a result or sources for this claim until its wording is fixed.
No source links were published for this claim.
Factual claim · Statement 30 of 33
Pikaan is not showing a result or sources for this claim until its wording is fixed.
No source links were published for this claim.
Factual claim · Statement 31 of 33
Pikaan is not showing a result or sources for this claim until its wording is fixed.
No source links were published for this claim.
Factual claim · Statement 32 of 33
Pikaan is not showing a result or sources for this claim until its wording is fixed.
No source links were published for this claim.
Factual claim · Statement 33 of 33
A 2023 study on the bgu.ac.il page reports that medial lesions affected both tasks and concludes that both the medial and lateral pallium are essential for coding spatial memory, and the ScienceDirect page carries the same title. However, earlier NIH studies report that lateral, but not medial, pallium lesions produced the spatial-memory impairment, so the evidence is split by study and task.
Pikaan found this exact sentence on the publisher’s page. Finding it does not show whether it supports or contradicts the claim.
In this review
Claim 1 of 51
Check your own videos, articles and documents the same way: every claim, its state and its sources. 300 free credits to try Pikaan, valid for 3 days. No card needed.
