“Intelligent animals” like chimpanzees and dolphins, possess an ability to recognize themselves in a mirror. Mirror self-recognition (MSR) via mental image of the self provides a background of the animal’s private self-awareness or “mind”. However, implications remain controversial since an alternative process, such as checking the synchronicity of the mirror-image has not been ruled out. To show animals’ mental self-image is used in MSR, no studies, so far, have taken place. 

 

Previous studies revealed that the small cleaner fish (Labroides dimidiatus) have MSR-capacity. What mental image of self do they have? They have individually different facial color patterns and use them to recognize familiar individuals. Similarly, we recognize ourselves by the mental image of our own or known people's faces in the mirror. Therefore, we assumed the fish would also recognize mirror-image as self via mental image of self-face. We tested this hypothesis using a self-photograph as the self motionless. 

 

A total of ten cleaner fish were exposed to a mirror for a week; a mark-test confirmed that all fish had MSR-capacity. Before exposure, these fish frequently attacked photo models of themselves and unknown strangers. After passing the mark-test, they still strongly attacked unknown fish photographs, but not their own. Results indicate that cleaner fish may recognize self-photographs as themselves. 

 

They were shown two types of composite photographs of their own face/stranger body and stranger face/own body. The formers were not attacked like the self-photograph. In contrast, the latter were attacked as frequently as the photograph of stranger fish. This strongly suggests that cleaner fish with MSR-capacity recognize their own facial characteristics in photographs. Since the recognition of motionless photographs does not involve a kinesthetic-visual matching process, we provide evidence that the MSR mechanism of cleaner fish is likely based on a mental image of their own face. That is, cleaner fish have internal mental images like humans. 

 

However, fish may consider the self-photograph to be a “super-familiar” fish, and they should not attack it. To reject this possibility, we tested whether cleaner fish identified self-photographs as themselves using the “self-photograph mark-test.” We prepared eight new cleaner fish with MSR-capacity. Focal fish were shown their self-photograph with a parasite-like mark on the throat. Since these fish would have mental images of self, we predicted that they would scrape their own throat to remove the mark (no mark was on their throat) when they saw the photograph. 

 

As predicted, six out of eight fish scraped their own throats on the bottom substrates when they saw the photographs, but no fish did so when they saw the unmarked self- or familiar-neighbor photographs with marks. So, we concluded that cleaner fish recognize the self-photographs as themselves; neither self-face nor mark itself triggered throat scraping, and they had no opportunity for associative learning. 

 

Thus, cleaner fish likely consider the mark an ectoparasite, which will induce their motivation to remove it. Indeed, after scraping their throat, they immediately looked at their reflection in the mirror or at the photographs to see if they successfully removed the mark. This recognition might be an example of meta self-awareness. Meta self-awareness is the next level of awareness in which one is aware of awareness. We are further exploring this hypothesis in ongoing studies. Cleaner fish have a mental image of their self-face, probably together with motivations, aims, and intentions. Thus, they may have private self-awareness, or “mind.” 

 

In social groups, animals need to readily recall previous interactions with other individuals, and having mental images of conspecific faces would facilitate the rapid identification of recently encountered individuals. We show that fish can distinguish between the faces of self, familiar, and unknown fish. The ability to recognize faces and to adjust behaviors accordingly (e.g., friendly or aggressive) suggests that cleaner fish can distinguish the self and others without any language. 

 

Our study implies that intelligence and self-awareness may not be directly relevant either to brain size or phylogenetic proximity to humans. We believe that our study will be a milestone in animal cognitive studies and challenge the anthropocentrism that has persisted since Descartes.