1 Introduction

In his book, “A Philosophy for the Science of Animal Consciousness” that discusses many studies of consciousness in animals, Dr. Walter Veit advocates an approach toward the study of such consciousness that takes a biological view inspired by Darwinian theory. This differs markedly from the ways in which more conventional philosophers may view consciousness. The issue of animal consciousness has been traditionally explored in a top-down manner, thus from an anthropocentric viewpoint (e.g. Gallup, 1970; Suddendorf & Butler, 2013; Byrne, 2016; de Waal, 2019; Morin, 2022). Eschewing anthropocentrism, Dr. Veit recommends that animal consciousness should be investigated using a bottom-up approach. Such an exploration of the consciousness of animals on the lower rungs of the intelligence ladder constitutes an extension of the approach of Griffin (Griffin, 1976). I agree that such a framework might advance the study of animal consciousness, and would likely align with the views of cognitive ethologists who generally favour the Griffin approach that was, in all probability, initially introduced by Romanes (Romanes, 1884).

It is here that the pathological complexity thesis will offer us a useful framework for accommodating the criticism commonly raised by cognitive ethologists that the animal’s ecological life-history context is being ignored within comparative psychology. (p32, Line 3–6 of Veit, 2023)

Self-awareness is an important term in considering animal consciousness (Morin, 2022). When seeking to establish that animals are self-aware, confirmation of mirror self-recognition (MSR) has been regarded as essential since the time of the pioneering studies of Prof. Gallup (Gallup, 1970, 1982; Kohda et al., 2019, 2023). Based on animal self-awareness confirmed by MSR, several hypotheses on the origin and evolution thereof have been advanced. These include the Big Bang theory (Gallup, 1970; Gallup & Anderson, 2020) and the Gradualist view hypothesis (de Waal, 2016, 2019). Both underpin the study of animal self-awareness. The latter hypothesis was treated affirmatively by Veit (Veit, 2023). MSR has been documented in large-brained animals including chimpanzees, dolphins, elephants, and magpies (Gallup, 1970; Reiss & Marino, 2001; Plotnick et al., 2006; Prior et al., 2008) and, recently, in a small reef fish, the cleaner wrasse (Labroides dimidiatus) (Kohda et al., 2019, 2022, 2023; Kobayashi et al., 2024).

MSR by a small-brained fish, phylogenetically very distant from the human, will greatly impact the traditional view of animal self-awareness (de Waal, 2019; Gallup & Anderson, 2020; Morin, 2023). Indeed, the fish may serve as the model species for study of MSR because, first, it passes perfectly the virtual mark test; second, the cognitive basis in practicing MSR is cleared (in contrast to the situation in other MSR animals); and third, the fish passes the “photograph mark-test” that excludes alternatives to a mentalistic view (Gallup & Anderson, 2020). I will refer to the two principal studies of cleaner wrasse MSR published after Dr. Veit’s book appeared. These serve as good case studies from the viewpoint of cognitive ethologists and suggest a new explanation of how self-awareness originated and evolved in vertebrates.

2 False-negatives in MSR Studies

A mark test is used to confirm animal MSR. After an animal views its mirror image for some time, a mark is placed on a body part that can be seen by the animal only in a mirror (for example, on the head or throat), and the animal is exposed to the mirror again. If the animal touches or scrapes the mark on the body only after seeing it in the mirror, the animal is regarded as passing the mark test and thus exhibiting MSR (Gallup, 1970). It recognises that the new mark is on self body, meaning that it recognises the mirror image as itself. However, false-negatives are common. Some animals may not touch/scrape marks although they recognise the marks as new. Certain animals that failed the test, including pigs, rhesus monkeys and many other monkeys (Rajala et al., 2010; Kohda et al., 2022), are considered to lack MSR because no false-negatives have ever been proven (Kohda et al., 2022).

The cleaner wrasse seeks and then collects small ectoparasites from the bodies of other fish in wild. It might be predicted that if a wrasse bore a coloured mark that resembled an ectoparasite, the wrasse would seek to remove it. Indeed, when such a mark was placed on a region of the body directly visible to the wrasse (thus, a mirror was not required), the wrasse sought to remove the mark by scraping its body on the aquarium bottom (Kohda et al., 2019). The wrasse was very eager to remove a mark resembling a parasite, thus an “ecologically relevant mark”. The pass rate in this test was 97% (37/38 of fish tested by Kohda et al., 2019, 2022, 2023), while much less when “meaningless colour marks” were used in previous successful studies. Specifically, no wrasse tested in all 10 trials sought to remove meaningless marks of blue or green, suggesting that these were not recognised as parasites or as otherwise harmful (Kohda et al., 2022). If the mark test of wrasse employed only meaningless marks, we would come to the completely incorrect conclusion that cleaner wrasses lack MSR. When animals fail mark tests that employ meaningless marks, it cannot be concluded that MSR is lacking. To the best of my knowledge, almost all MSR tests to date, with the exception of the wrasse tests, employed meaningless, thus not ecologically relevant, marks. Many MSR studies, therefore, would have come to the incorrect conclusions, spreading confusion (Kohda et al., 2022, 2023).

Animal species that pass the mark test do so at rather low rates, thus 30–40% for chimpanzees, elephants, and magpies (Povinelli et al., 1993; Plotnik et al., 2006; Prior et al., 2008). Such low rates suggest that the mark test employing meaningless mark is inadequate. It is illogical to state that, within the same species, animals that pass the test have self-awsreness and the others do not. It will be plausible that individuals that are more curious and/or sensitive than others at the moment will be more likely to explore a mark. We predict that if “ecologically relevant marks” can be seen in a mirror (Kohda et al., 2019), future studies of mirror mark-test may show that many animals exhibit MSR (Kohda et al., 2022; Kohda & Sogawa, 2025).

3 The Cognitive Basis of MSR in the Cleaner Wrasse: Self-face Recognition Via Photograph Self-Recognition (PSR)

It is generally accepted that animals exhibiting MSR proceed through three behavioural stages before passing the mark test. Such animals include chimpanzees, dolphins, elephants, magpies, house crows, and cleaner wrasse (Gallup, 1970; Reiss & Marino, 2001; Plotnik et al., 2006; Prior et al., 2008; Buniyaadi et al., 2020; Kohda et al., 2019). First, they exhibit social behaviours when viewing the mirror reflection, suggesting that they initially recognise the image as an unknown conspecific. Then, they begin to examine the mirror image by testing how their movements match those of the image. Finally, they exhibit self-directed behaviours. For example, they may examine otherwise invisible body parts. These three distinctive stages appear to be consistent across all vertebrates from fish to birds and then to mammals including the great apes (Gallup, 1970; Plotnik et al., 2006; Kohda et al., 2019, 2023: Vanhooland et al., 2023). This suggests that vertebrate MSR is based on common cognitive processing (Kohda et al., 2019, 2023).

Although MSR has been documented in some animals, the cognitive basis thereof remains unknown (de Waal, 2019; Morin, 2022), but is only clear in humans, who identify their mirror image via self-face recognition after attainment of MSR (Keenan et al., 2003). During self-recognition in a mirror and in portraits, we employ a mental image of the self-face, similar to how we visually identify all of our familiars using mental images of their faces (Keenan et al., 2003).

Recently, face-recognition that is similar to humans has been documented in some primate and non-primate mammals, and birds (Leopold & Rhodes, 2010; Kohda et al., 2024), and also in a variety of fish (e.g. Tanganyikan daffodil cichlids, discus fish, medaka, guppy, and stickleback). Such fish distinguish familiar fish from others via face-recognition (Kohda et al., 2024). Cleaner wrasse also visually distinguish familiar fish from others via face-recognition (Kohda et al., 2023). Fish that recognise each other live in a stable social group with a dominance hierarchy and/or territory, or as a sexual pair (Kohda et al., 2024). Several fish species likely identify many familiar individuals via face recognition, and this will be true individual recognition (TIR) (Saeki et al., 2018).

After establishing MSR, a cleaner wrasse does not attack its own photograph but is strongly aggressive when shown photographs of unknown fish (Fig. 3 in Kohda et al., 2023). When the face of a familiar wrasse was digitally added to the body of another strange fish, this photograph was not attcked, but a photograph with its own body but the face of a stranger was attacked. Thus, self-recognition of cleaner fish was exclusively face-based, as in humans (Kohda et al., 2023).

Self-awareness has been divided into three levels by Morin (2006). One of the three is “Private self-awareness” (PrSA), and it has a mental representation (image) of the self in an animal with a “mind”. To demonstrate this, it is essential to exclude the use of kinesthetic visual-matching during MSR (Mitchell 1993, 1997). MSR must, rather, be based on a mental image of oneself. A “photograph self-recognition” (PSR) test is conducted. If an animal recognises its own photograph, it has an inner mental image of the self, thus PrSA (Kohda et al., 2023). If animals engage in kinesthetic visual matching, they do not recognise motionless portraits as themselves. Wrasses capable of MSR pass the photograph mark-test. When a portrait with an ecological mark on the throat was presented, the wrasse scraped the throat area of the mark after viewing the portrait despite the fact that its own throat was not marked! This constitutes compelling evidence that the fish identified a portrait as itself because it had a mental image of its own face. Cleaner wrasse thus exhibit PrSA (Kohda et al., 2023; Kobayashi et al., 2024). This is the best evidence adduced over more than 50 years of MSR studies. I believe that the PSR test, or similar tests, will reveal PrSA in many MSR-capable animals.

4 Criticism of the Big Bang and Gradualist View Hypotheses, and a New Proposal

Mark tests have been performed on many primates (Gallup, 1982; Keenan et al., 2003; Sudderndorf & Bulter, 2013). It is largely accepted that the great apes passed but gibbons and many monkeys failed it (Gallup, 1970; Sudderndorf & Bulter, 2013; Gallup & Anderson, 2020). It was thus assumed that PrSA evolved only in the ancestors of the great apes (the Big Bang hypothesis) and that monkeys and non-primate animals more distant from humans lacked MSR (Suddendorf & Bulter 2013; Gallup & Anderson, 2020). However, probable false-negatives were not excluded.

de Waal accepted the results of mark tests performed on elephants, dolphins and magpies, and suggested their MSR will show the presence of self-awareness (Fig. 2.2a in Veit, 2023) (de Waal, 2019). He advanced a “parallel evolution” hypothesis on the evolution of their self-awareness because he, too, regarded that the MSR capacity evolved in the groups of great apes, elephant, dolphins and corvids, and the other groups would lack the capacity of MSR (de Waal, 2019). Moreover, de Waal advanced a “Gradualist view” hypothesis on the evolution of animal self-awareness based on three main observations. First, some animals passed the mark test by combining multimodal and tactile perception. Second, a number of animals used mirrors to look for hidden sources such as food. Third, some species that failed the mark test recognised their own mirror images as uncanny conspecific that always mimics own movements but not a real strange individual (de Waal, 2019). Considering these capacities, he suggested that self-awareness evolved from a simple to a more complex form related to MSR.

First, I argue that MSR capacity is a cognitive skill, not self-awareness per se. The gradualist model confuses such self-awareness with certain skills or related phenomena mediated via self-awareness. Skills achieved with the aid of self-awareness per se contain such as MSR, the use of a mirror as a tool, theory of mind, empathy or cooperation. If any such skills are viewed as equivalent to self-awareness, some species will exhibit many skills and others less, and animal self-awareness does then seem to have evolved gradually (Fig. 2.2b in Veit, 2023). de Waal implies that ‘apparent’ self-awareness (the sum of various skills) is related to brain size. However, these skills will vary by the environments in which respective species lives. For example, primates living in larger groups exhibit more social intelligence and larger brains than smaller group size, and a correlation between group size and brain capacity is apparent (Dunber, 2009), probably because of the summed variety of social skills.

Similarly, in relation to the parallel evolution hypothesis, Byrne suggested that only four clades, the hominid great apes, dolphins, elephants, and corvids, all of which have relatively large brains and live in highly organized societies with exhibition of empathy, exhibit mentalizing ability, which is equivalent to theory of mind (Byrne, 2016). The mentalizing ability will be a skill, and thus will be related to the brain sizes. The sizes of brain regions in bony fish reflect the major sensory perceptions, such as visual, olfactory or auditory one, required by their lifestyles (Yamamoto, 2017). In each case, animals increase the cognitive skills by which they recognise and interact with their social and physical environments. Thus, brain size will be related to the sum of skills but not to PrSA per se, as revealed in the studies on cleaner wrasse (Kohda et al., 2023; Kobayashi et al., 2024). In any case, it would be strange if a form of self-awareness simpler than that of cleaner wrasse with a tiny brain, was applicable to dogs, cats and pigs, with larger brains, all of which are much closer to humans in the gradualist model (Fig. 2.2b in Veit, 2023). Thus, the gradualist view hypothesis will not be a valid model of self-awareness evolution.

Turning to the second plank of the hypothesis, it is widely assumed that the ability to use a mirror to gain information is not equivalent to recognising the self in a mirror image (e.g. de Waal, 2019). Many animals can use a mirror as a tool but fail the mark test. Examples include the lesser ape, the African gray parrot, the pig, the sea lion, the New Caledonian crow, and so on (see de Waal, 2019; Kohda et al., 2025). Due to the restricted numbers of animals capable of MSR, it is generally accepted that MSR requires more sophisticated cognitive ability than recognition of reflections of other objects (de Waal, 2016, 2019). However, this may be incorrect when the mark test false-negatives are considered. When using a mirror as a tool, animals will see both their own image and whatever they seek, but they do not react aggressively to their own reflections. Given the false-negative, it would be reasonable to suggest that the animals start using mirrors after they recognise the self-image as the self. Indeed, several species that use mirrors as tools, including domestic pigs, rhesus monkeys, manta rays, corvids and domestic fowl, are thought to lack MSR, but perform contingency-testing behaviours (examining the own mirror-image before self-directed behaviour) when presented with a mirror (e.g. Broom et al., 2009; Rajala et al., 2010; Ari & D’Agostino, 2016; Vanhooland et al., 2023; Hillemacher et al., 2023). This strongly suggests that these animals developed an MSR capacity before using a mirror as a tool (Kohda & Sogawa, 2025).

We have clearly shown that the cleaner wrasse passes the mark test using a cognitive human-like feature (Kohda et al., 2019, 2022, 2023), and we have overcome a major issue of the mark test, the false-negatives. We predict that the cognitive basis of MSR after establishment of it, i.e. a mental image of the self-face, will be found in other MSR animals. This will not support either of the two hypotheses introduced above and will encourage re-evaluation of how self-awareness evolved in vertebrates. The MSR of the cleaner wrasse suggests that self-awareness is not restricted by either brain size or the phylogenetic distance from the human. Bony fish are the ancestors of tetrapods, suggesting that not only the four clades of large-brained animals but also many other tetrapods will have PrSA. This will be confirmed using an approach devoid of false-negatives (Kohda et al., 2025). We would advance a “self-awareness homology” hypothesis: The evolutionary origin of self-awareness in vertebrates will be traced back to the bony fish at latest that were in existence no later than Devonian period (Kohda et al., 2025).

Studies of animal MSR that commenced in 1970 are only now eschewing top-down anthropocentrism and favouring a bottom-up approach. Cleaner wrasse achieve MSR/PSR by accessing mental representations of their own faces (Kohda et al., 2023). The anthropocentric hypotheses (Suddendorf & Bulter, 2013; Byrne, 2016; Gallup & Anderson, 2020; de Waal, 2019) that aimed to explain the evolution of self-awareness limited in great apes and large brained animal groups with empathy will be thus incorrect (Kohda et al., 2023, 2025). Improved and/or new methods will reveal that self-awareness is likely not restricted to vertebrates but would be also present in invertebrates including cephalopods, crustaceans and insects, as some pioneering works have already suggested (e.g. Robinson, 2023). The bottom-up approach in studies of MSR and thus consciousness in animals will bring on profound implications. For this purpose, Griffin’s perspectives (Griffin, 1976) of the study of cognitive abilities of animals should be re-evaluated as stated by Veit (2023).