Do other animals know they exist?

In 1970 chimpanzees given mirrors moved from treating the reflection as another animal to using it to inspect themselves, and then touched red dye marks they could not see without it.

One of three Asian elephants tested in 2006 touched a mark on its head that it could only see in a mirror, having first used the mirror to inspect itself. One animal of three is the shape of results in this literature: samples are small, the magpies were credited on two birds of five, and a species is often credited on the strength of one or two convincing performers.

Over the following decades the same result was reported for other great apes, for elephants, for magpies and for dolphins. Only the evidence for chimpanzees, and to a lesser degree orangutans, is accepted without dispute. Gallup, who devised the test, and James Anderson argued in 2018 that the elephants, dolphins and magpies had not given compelling, reproducible evidence, and a larger repeat of the magpie study in 2020 did not find it. The test itself became the standard evidence for self-recognition in non-human animals.

In 2019 a study reported that cleaner wrasse, a small reef fish, behaved towards mirrors the way the passing species do: first aggression, then unusual behaviour in front of the mirror, then attempts to remove a coloured mark from their own bodies by scraping.

At first the authors did not claim the fish is self-aware in the sense people mean. They put the dilemma the other way round, and titled the paper with it: if a fish can pass the mark test, what are the implications for testing self-awareness in animals at all.

The result is disputed. Frans de Waal, in a commentary published beside it, argued that the injected marks were probably irritating and that scraping is a routine behaviour for fish, so the fish fall short of self-recognition. The same group then repeated the test: all fourteen new fish scraped at a brown mark, and only when a mirror was present. By 2023 they were going further, arguing that the fish recognise their own faces in photographs through a mental image of the self, which in people is counted as a form of private self-awareness.

Dogs have not passed the mirror test; like cats, they learn to ignore their reflection. Given their own urine instead, since smell is their primary sense, they spent longer on it when another odour had been added, which the study’s author reads as recognising the smell as their own. Gallup and Anderson accept this olfactory version as a fair analogue of the mark test and still find no compelling, reproducible evidence of self-recognition in dogs. Species that failed may have failed the apparatus rather than the concept.

A test that cannot distinguish absence of self-recognition from indifference to a coloured spot is weaker evidence than its reputation, in both directions.

Recognising a body in a mirror is a claim about perception. Knowing that you exist, having a sense of your own past and future, is a much larger claim, and the test does not distinguish them.

Other evidence bears on the larger claim, including planning for future needs in corvids and apes, but it accumulates species by species and has no single decisive experiment.

Every method here reads behaviour. None of it reaches experience, in a fish or in a chimpanzee, for the same reason nothing reaches it in another person.

The practical consequence is that the burden of proof does the work. Requiring evidence of an inner life before granting one produces a different set of animals than requiring evidence of its absence, and nothing in the data chooses between those two policies.