A fish behaved as though it had noticed a mark on its own throat, and the argument that followed is about what the test measures at all. · published 12 October 2026
Marked fish held themselves so the mark faced the mirror and then scraped that spot on the substrate, doing neither with an invisible mark or without a mirror - though whether that is self-recognition, a parasite-removal reflex aimed at an apparent neighbour, or a response to a mark they could feel as well as see, is argued in print.
These are the working notes, unedited. Everything the video
states as fact comes from here, and nothing else may be stated as fact.
The sections on limits and on what the finding does not show are
part of the brief, not an afterthought — see
the method.
The test being used. In the mark test, an animal is marked somewhere it cannot see directly, then given a mirror. Touching the mark on its own body is scored as recognising the reflection as itself. Great apes pass. Most species do not.
The study. Masanori Kohda and colleagues, “If a fish can pass the mark test, what are the implications for consciousness and self-awareness testing in animals?”, PLOS Biology, 2019. The subject was the cleaner wrasse, Labroides dimidiatus - a small reef fish that makes its living picking parasites off other fish.
What they did. A coloured elastomer was injected under the skin, leaving a visible mark, in some cases on the throat. The fish were then given a mirror. The controls were careful: a colourless sham mark, an invisible mark, and a condition with no mirror at all.
What they found. Marked fish spent a long time at the mirror and held themselves so the marked side faced it. They then scraped the marked spot against the substrate - throat-scraping after a throat mark. They did not do this with an invisible mark, and they did not do it without a mirror. The authors argued this meets the criteria of the mark test.
The objection, from the primer published alongside it. Frans de Waal argued the result falls short of mirror self-recognition, for two reasons worth stating precisely.
First, the behaviour is ambiguous. The whole case rests on one stereotyped act, self-scraping, which is a fixed action pattern in many fish. And this species, of all species, makes a living reacting to parasites on other fish - so a wrasse seeing a marked fish in a mirror may be responding to an apparently parasitised neighbour rather than to itself. De Waal argued that a behaviour never seen in the absence of a mirror would be far stronger evidence.
Second, the mark is felt as well as seen. An injection under the skin is likely to be irritating. The sham mark controls for touch alone, but not for the combination of a strong physical sensation with a visual mark. De Waal called this a “Felt Mark Test”, noting that rhesus macaques show self-directed behaviour at a mirror only after that kind of combined stimulation - which would place the wrasse nearer monkeys than apes.
What he proposed instead. Not a verdict but a change of scale: drop the pass/fail split and treat mirror understanding as graded, with intermediate levels such as using a mirror as a tool or simply not treating the reflection as a stranger. Kohda’s group has since run further experiments answering the methodological points, and has itself argued that either the criteria for self-awareness or the mark test needs rethinking.
Limits
Few fish, one species, and a mark delivered by injection. The sample size was among the published criticisms. The central dispute is not really about this experiment’s execution - it is about whether a single scraping behaviour can carry the weight the mark test puts on it.
What this does not show
Not “fish are self-aware”, and not “the study was debunked”. Two groups of serious people agree on what the fish did and disagree about what it shows, which is the normal state of a good result and is usually lost in the retelling.