ArticlePDF AvailableLiterature Review

Abstract

Fish are one of the most highly utilised vertebrate taxa by humans; they are harvested from wild stocks as part of global fishing industries, grown under intensive aquaculture conditions, are the most common pet and are widely used for scientific research. But fish are seldom afforded the same level of compassion or welfare as warm-blooded vertebrates. Part of the problem is the large gap between people's perception of fish intelligence and the scientific reality. This is an important issue because public perception guides government policy. The perception of an animal's intelligence often drives our decision whether or not to include them in our moral circle. From a welfare perspective, most researchers would suggest that if an animal is sentient, then it can most likely suffer and should therefore be offered some form of formal protection. There has been a debate about fish welfare for decades which centres on the question of whether they are sentient or conscious. The implications for affording the same level of protection to fish as other vertebrates are great, not least because of fishing-related industries. Here, I review the current state of knowledge of fish cognition starting with their sensory perception and moving on to cognition. The review reveals that fish perception and cognitive abilities often match or exceed other vertebrates. A review of the evidence for pain perception strongly suggests that fish experience pain in a manner similar to the rest of the vertebrates. Although scientists cannot provide a definitive answer on the level of consciousness for any non-human vertebrate, the extensive evidence of fish behavioural and cognitive sophistication and pain perception suggests that best practice would be to lend fish the same level of protection as any other vertebrate.
REVIEW
Fish intelligence, sentience and ethics
Culum Brown
Received: 13 November 2013 / Revised: 7 May 2014 / Accepted: 19 May 2014
ÓSpringer-Verlag Berlin Heidelberg 2014
Abstract Fish are one of the most highly utilised ver-
tebrate taxa by humans; they are harvested from wild
stocks as part of global fishing industries, grown under
intensive aquaculture conditions, are the most common
pet and are widely used for scientific research. But fish
are seldom afforded the same level of compassion or
welfare as warm-blooded vertebrates. Part of the problem
is the large gap between people’s perception of fish
intelligence and the scientific reality. This is an impor-
tant issue because public perception guides government
policy. The perception of an animal’s intelligence often
drives our decision whether or not to include them in our
moral circle. From a welfare perspective, most
researchers would suggest that if an animal is sentient,
then it can most likely suffer and should therefore be
offered some form of formal protection. There has been
a debate about fish welfare for decades which centres on
the question of whether they are sentient or conscious.
The implications for affording the same level of pro-
tection to fish as other vertebrates are great, not least
because of fishing-related industries. Here, I review the
current state of knowledge of fish cognition starting with
their sensory perception and moving on to cognition. The
review reveals that fish perception and cognitive abilities
often match or exceed other vertebrates. A review of the
evidence for pain perception strongly suggests that fish
experience pain in a manner similar to the rest of the
vertebrates. Although scientists cannot provide a defini-
tive answer on the level of consciousness for any non-
human vertebrate, the extensive evidence of fish
behavioural and cognitive sophistication and pain per-
ception suggests that best practice would be to lend fish
the same level of protection as any other vertebrate.
Keywords Fish cognition Sentience Welfare Pain
Intelligence Ethics
Introduction
Most people rarely think about fish other than as food or as
pets. However, fish are the most consumed animal in terms
of numbers, the most numerous kind of pet, and are second
only to mice in terms of the numbers used in scientific
research (Huntingford et al. 2006; Iwama 2007; Sneddon
2013). There are more species of fish than all other ver-
tebrates combined (ca 32,000 known species). Despite the
fact that humans and fish interact on many levels and in
greater numbers than mammals or birds, fish have not
roused much public concern regarding their welfare. Per-
haps the primary reason for this lack of consideration
relates to the public perception of their level of intelligence
and ultimately of whether they are conscious. All evidence
suggests that fish are, in fact, far more intelligent than we
give them credit. Recent reviews of fish cognition suggest
fish show a rich array of sophisticated behaviours. For
example, they have excellent long-term memories, develop
complex traditions, show signs of Machiavellian intelli-
gence, cooperate with and recognise one another and are
even capable of tool use (Bshary et al. 2002; Brown et al.
2011a,b). Emerging evidence also suggests that, despite
appearances, the fish brain is also more similar to our own
than we previously thought. There is every reason to
believe that they might also be conscious and thus capable
of suffering.
C. Brown (&)
Department of Biological Sciences, Macquarie University,
Sydney 2109, Australia
e-mail: Culum.Brown@mq.edu.au
123
Anim Cogn
DOI 10.1007/s10071-014-0761-0
The objective of this review is to bridge the gap between
public perception of fish cognition and scientific reality
with a view to informing the fish ethics/welfare debate.
This is particularly pertinent given the general belief that
welfare depends to a large degree on an animal’s cognitive
ability (Duncan and Petherick 1991; but see Dawkins
2001). A greater understanding of animal cognition will
likely influence our view of animal welfare. I do not
attempt to provide detailed reviews of fish perception and
cognition as such work would require many volumes (see
Brown et al. 2011a). I simply selectively refer to this work
to illustrate the depth and scope of fish perception and
cognitive abilities. Armed with this information, the public
is in a far better position to make a judgement about fish
consciousness and thus whether they should be afforded
greater protection from cruelty. The review is largely
restricted to bony fish (teleosts) since we know far too little
about the cognitive abilities of sharks and rays (elasmo-
branchs). Moreover, we know next to nothing about the
vast majority of fish species, and while it is hard to gen-
eralise between such diverse groups, we must assume that
the capabilities that have been revealed in model taxa are
likely to be exemplary of teleosts as a whole.
Here, I use the ‘feelings’’-based approach to animal
welfare and ethics which focuses on animal suffering
(Appleby and Sandøe 2002). Suffering refers to prolonged
periods of unpleasant mental states such as pain and fear. I
acknowledge that the wild animals are frequently subjected
to both pleasant and unpleasant experiences, but from a
welfare and ethics perspective we wish to prevent extended
period of suffering wherever possible. All of this ultimately
depends on whether fish are capable of conscious, sub-
jective experiences, and if we as scientists are capable of
interpreting such experiences, both of which are generally
controversial topics (Dawkins 1998).
Science, public opinion and fish welfare
The fact that we rarely come into contact with fish in their
natural environment has implications with respect to our
attitudes towards them. Fish live in an environment that we
do not really understand and rarely visit, thus few of us
ever see fish behaving naturally. The vast majority of
people only ever see fish on their plate, and while lots of
people keep fish as pets, fish seldom have the opportunity
to express their natural behavioural patterns in captivity.
This is in stark contrast with terrestrial animals such as
dogs, cats and horses that have lived closely with people
for tens of thousands of years and with whom we are very
familiar. Whales, dolphins and seals, while living in the
same aquatic media as fishes, are mammals, and thus, we
seem far more capable of extending our circle of morality
to include them. However, because fish are also phyloge-
netically distant to humans in comparisons with mammals,
we find it very difficult to empathise with them. We cannot
hear them vocalise, and they lack recognisable facial
expressions both of which are primary cues for human
empathy. Because we are not familiar with them, we do not
notice behavioural signs indicative of poor welfare.
Our perception of an animal’s intelligence largely
defines the way we treat them when we interact with them
(Kirkwood and Hubrecht 2001). Their level of intelligence
defines whether or not we think they are sentient beings
and thus worthy of inclusion in our ‘‘moral circle’ (Lund
et al. 2007). Science has the capacity to inform this debate
and plays an important role in developing an accepted
agreement about the standards in animal welfare. Science
can also shift belief systems, albeit slowly. Changes in the
belief system then trigger changes in legislation. For
example, the UK’s Animals (Scientific Procedures) Act
1986 prohibits experiments on vertebrates except under
licence based on an assumption that all vertebrates have the
capacity for conscious feelings. A ban on all great ape
research in the EU was later declared (Directive 2010/63/
EU) presumably because it is commonly believed that
these animals have superior cognitive powers and thus a
greater capacity for suffering. This perception that cogni-
tion and consciousness are related is interesting consider-
ing that consciousness may have arisen long before
cognitive complexity (Dawkins 2001). Given the common
poor perception of fish intelligence, it is hardly surprising
that the ethical treatment of fishes lags far behind their
vertebrate cousins.
The UK Cruelty to Cattle Act 1822 banned the improper
treatment of cows and sheep and was one of the world’s
first attempt at animal welfare legislation. The 1835 Cru-
elty to Animals Act included dogs and explicitly outlawed
bear baiting and cockfighting. This was later repealed by
the Cruelty to Animal Act of 1876, which was perhaps the
first to formally protect animals from pain and suffering
during scientific research, and gave special protection to
cats, dogs and horses. Subsequently, primates became the
subject of great empathy, but our compassion has gradually
spread to cover elephants, dolphins and so on to recently
include other farm animals in an industrial setting (e.g.
pigs, chickens). The recent push for protecting the welfare
of farm animals is a good example of how public opinion
has shifted and resulted in significant changes in legisla-
tion. The move to protect primates, elephants and dolphins
has perhaps arisen from our greater understanding of the
cognitive capacities of these animals; the result of over
50 years of intensive research. Such a sea change has yet to
occur for fishes, and fish are still not afforded appropriate
protection in most countries. This is interesting given that
they are vertebrates and technically (though clearly not in
Anim Cogn
123
practice) are afforded the same legal protection in many
developed countries.
Why is legislation rarely applied to fishes? I suspect that
there are two explanations. The first relates back to public
perception. Until scientific findings relating to fish intelli-
gence filter through to the public, and the public believe
that fish are conscious beings capable of suffering, there
will be a lack of political or social will to promote fish
welfare. Secondly, and this is a far more practical issue, the
ramifications for such animal welfare legislation, should it
be applied to fishes, is perhaps too daunting to consider.
Certainly, the fishing industry, including aquaculture and
recreational angling, would have to drastically alter its
approach. These interest groups are also powerful lobby-
ists, and the commercial value of these practices is sig-
nificant. Indeed this explains why fish are not covered by
the Animal Welfare Act in the USA. It is interesting to
note, however, that the UK is starting to move in this
direction with Aquatic Animal Health Regulations 2009
requiring fish farms to keep records on mortalities and
stock movements. With limited resources for animal wel-
fare and limited knowledge about fish cognition, welfare
priorities often fall well short of protecting fish.
Evolution and biological complexity
Given the complexity of the environment in which many
fish live (social and physical), it should not be surprising
that fish have a high degree of behavioural plasticity and
compare favourably to humans and other terrestrial verte-
brates across a range of intelligence tests (Bshary et al.
2002; Brown et al. 2011a,b). Whether or not fish are
worthy of human consideration, and thereby afforded some
protection, is informed by scientific knowledge regarding
their mental capacities. Once this information is at hand,
we can have an informed debate about fish–human inter-
actions and create moral guidelines to direct animal welfare
legislation.
Sadly, the general poor perception of fish intelligence is
not restricted to the broader public but it is also deeply
imbedded in scientific dogma and can be traced back to the
very beginnings of scientific investigation in the western
world. The scientific study of animal brains and behaviour
(and by proxy intelligence) has been dogged by the deep
rooted notion that the evolution of vertebrates follows a
linear progression from inferior to superior forms, culmi-
nating in humans at the apex. This anthropocentric story
tells us that along this path, we see increasing complexity
and advances in brain and behaviour, leading to greater
intelligence and behavioural plasticity at each evolutionary
stage. Accordingly, the ‘primitive’ vertebrates such as
fishes are supposed to have simple brains with a few neural
circuits that control elementary forms of behaviour. The
more recent or ‘advanced groups such as humans and
other primates, so this discredited but still commonly
believed theory argues, have evolved more complex brain
circuits providing sophisticated and flexible cognitive
capabilities including emotions, feelings and conscious-
ness. Romanes’ Animal Intelligence (1882) provides a
classic example of this kind of thinking where he tries to
show that animals become more advanced (humanlike) as
we move up the evolutionary scale. More recent examples
include Thorpe’s Learning and Instinct in Animals (1956)
and even Mcphail (1982) remains loyal to ‘the scale of
nature’ structure even as he explicitly rejects the idea of
cognitive differentiation among vertebrates.
In 1859, Darwin published The Origin of Species, which
totally revolutionised evolutionary biology. But despite
Darwin’s theoretical insights, the tiered and directed per-
spective of evolution continues to be taught at school
150 years later. In reality, evolution is a random rather than
directed process. Rather than representing evolution as a
linear or tiered progression of increasing complexity, we
now view the vertebrates as a highly diverse radiation each
with their own specialisations. Each group represents a
more or less parallel line of evolution radiating from a
common ancestor, and each species is specifically tuned to
match the niche it occupies. In short, every species is
unique. Following this reasoning, biological or cognitive
complexity is not defined by how closely animals are
related to humans but rather by the niche they occupy and
the problems they commonly face during their everyday
existence.
While there is no doubt that fishes are the most ancient
vertebrates, they are only ‘primitive’ in the sense that fish
of some description have been on earth for in excess of 500
million years and that all other vertebrates evolved from
some common fishlike ancestor around 360 million years
ago. But fish have not been standing in an evolutionary
backwater all that time, rather they have been evolving and
adapting to new environments which has led to the massive
diversity that we see today. There has been ample time for
fishes to evolve complex and diverse behavioural patterns
as well as the cognitive hardware that goes with it to match
the diversity of ecological niches they occupy. Moreover,
the vast majority of fish species on the planet are relatively
new arrivals. While the Devonian period (410 million years
ago) is often referred to as The Age of Fishes, most of the
contemporary species are from the Percomorpha which
showed a spate of massive speciation just 50 mya and
reached peak diversity around 15 mya. To put that in
perspective, the family Hominidae also evolved around the
same time. Thus, most fish species are no more ‘‘primitive’
than we are. And, despite apparent differences between fish
and humans, evolution tends to be highly conservative;
Anim Cogn
123
thus, many human traits are identical to or derived from our
fishlike ancestors.
Defining intelligence, consciousness and sentience
Most animal ethics practitioners draw a line in the sand
when it comes to which animals should be afforded pro-
tection from cruelty. For most people, this is whether or not
the animal in sentient or conscious. Sentience is not an easy
thing to define or measure, and the meaning is constantly
debated by scientists and philosophers alike, but it might be
summed up in an ethical context as the ability to experi-
ence pleasure and pain (i.e. subjective, perceptual experi-
ences; Dawkins 1998; Appleby and Sandøe 2002).
Consciousness is a far broader concept and includes sen-
tience, intelligence and self-awareness. Consciousness
might be broadly described as an awareness of internal and
external stimuli, having a sense of self and some under-
standing of ones place in the world (Chandroo et al. 2004;
Bekoff and Sherman 2004). When philosophers discuss
consciousness, they are nearly always referring to higher-
order consciousness (thinking about one’s actions and
thinking about thinking). Most people agree, however, that
suffering is a conscious experience albeit at a low level of
consciousness (phenomenal consciousness; Block 1991).
Many have argued that since we cannot measure mental
states, feelings or perceptual experiences, the science of
animal welfare is on far safer ground if we measure sub-
jective behavioural responses that imply consciousness
(Dawkins 2001). Measures such as attention and percep-
tion, self-recognition, theory of mind and episodic memory
are often the subject of experimental examination and thus
may be used as experimental evidence for consciousness.
It is widely argued that an animal that displays a high
degree of cognitive complexity (intelligence) is likely to be
sentient (Kirkwood and Hubrecht 2001; Huntingford et al.
2006, but see Bekoff 2006 re-suffering). The rationale
suggests that consciousness is an emergent property of
advanced cognitive capacities and associated complex
neural circuitry (Macphail 1998). Moreover, if we assume
that animals are conscious, then understanding their cog-
nitive capacities is likely to inform us about the sorts of
stimuli that induce unpleasant feelings (Nicol 1996). For
example, a limited capacity for memory would suggest that
an animal is unlikely to make associations between certain
events and negative stimuli or their consequences. This in
turn has profound implications for welfare considerations.
Cognition is well defined, and the definition is broadly
accepted. Animal cognition is the process by which ani-
mals acquire, process, store and act on information gath-
ered from the environment (Shettleworth 2010, p. 4). Some
would suggest that intelligence is about learning from past
experiences and applying this knowledge to solve novel
problems, in other words intelligence is about behavioural
flexibility (Roth and Dicke 2005). Most scientists would
probably agree that holistic intelligence relies on underly-
ing cognitive processes. Here, I will use intelligence as a
synonym for cognitive complexity.
One of the common problems we face is trying to be
objective when measuring intelligence. A common
approach in comparative cognition is to examine the
problem-solving skills of animals in novel contexts, typi-
cally in controlled laboratory conditions. Some animals,
however, do not adjust well to captivity and captive-reared
animals are commonly a poor exemplar of their wild
cousins. Moreover, one must devise a test that is equally
applicable to a wide range of animals, and the apparatus
must be one with which the subjects are motivated to and
can physically engage with (Shettleworth 2010, Chapter 1).
To do anything to the contrary is to not provide a level
playing field. Bitterman (1965,1975) carried out a famous
series of experiments comparing the cognitive performance
of fish and a range of other vertebrates in laboratory-based
studies. He made some attempt to design the various tasks
with the ecology (and hence evolutionary background) of
each animal in mind, but in doing so highlighted the
problems of comparing the cognitive ability of such vastly
different animals. Our understanding of fish behaviour is
now considerably advanced and more appropriate tests
have been applied to a wide range of species. The results
suggest that Bitterman’s early findings are certainly out-
dated. Some classic examples of suitable tasks for com-
parative cognition include spatial learning, classical and
operant conditioning and reversal learning. The use of
social learning is also commonly viewed as a sign of
intelligence. Copying social role models is very common
behaviour in children and can lead to rapid acquisition of
novel skills. Moreover, social learning can lead to the
development of cultural traditions. Social learning allows
cumulative information gathering over multiple genera-
tions and explains many of the advances in human tech-
nology. For example, the modern motor car could never
have been invented from scratch, but rather represents
cumulative knowledge gained and applied over more than a
hundred years. There are also forms of social intelligence,
for example where individuals deliberately manipulate the
behaviour of others by deception or reconciliation (i.e.
Machiavellian intelligence). Indeed, the social intelligence
hypothesis is often touted as an explanation for the evo-
lution of the large primate brain (Whiten and Byrne 1997).
Tool use is also often associated with intelligence, and
among primates and birds seem to be associated with rapid
problem-solving, innovation and large executive brain
areas (Reader 2003). The development of tools is seen as
the turning point in human evolution. An animal might also
Anim Cogn
123
be considered intelligent if it can form categories and
generalise, for example by partitioning objects into groups
based on similarity or dissimilarity. Finally, the use of
numbers and making calculations based on addition and
subtraction is also considered a sign of intelligence (for a
comprehensive review of comparative cognition, see
Shettleworth 2010). Although these tasks are widely used
for comparative studies, there is still an inherent difficulty
of comparing results across taxa. While not all of these
behaviours have been investigated in fishes, a great many
of them have and are discussed at length below. In general,
fish compare well to the rest of the vertebrates in most tasks
(Bshary et al. 2002; Brown et al. 2011a,b).
One of the basic canons of comparative cognition is that
one should never invoke higher-order explanations for
behaviour if they might be explained by simpler ones
(Morgan’s Canon). While this is strictly applied by many
who study comparative cognition in fish, it is often not
meaningfully applied by those studying humans and pri-
mates. Heyes (1993,1994), for example, points out that
many of the higher cognitive skills claimed for primates,
such as theory of mind, have never been rigorously proven;
for a nice example of this, see Coussi-Korbel (1994). Thus,
once more we see that comparative psychologists are not
playing on an even field. Part of the explanation for this
disparity is the fact that there are basically two camps in
comparative cognition research. The first seeks to explain
the evolution of human intelligence by examining our
closest ancestors. The second looks for broader patterns or
generalisations across all vertebrate taxa in relation to the
correlation between environmental factors and the evolu-
tion of intelligence, typically by looking at specific areas of
the brain and related behaviours (e.g. caching behaviour in
birds and the hippocampus). It is reasonably simple to
understand why the former seldom conform to Morgan’s
canon; if humans solve problems in this way, then it is
reasonable to assume that our closest relatives might also.
To ignore this possibility is to ignore the most basic rules
of the evolutionary process; the difference between closely
related species is one of degree not of kind. Thus, such
higher-order cognitive process is often attributed to pri-
mates without rigorous testing, but this seldom occurs in
fish because it would not be accepted by the scientific
community due to an underlying assumption of simplicity
rather than complexity.
Fishes sensory perception
The study of comparative cognition always begins with
information acquisition, and thus, there is a strong under-
lying need to understand the sensory perception capabili-
ties of the study animal. Part of this relates to avoiding
anthropomorphism, for example by assuming that animals
literally see the world as we do. However, it also relates to
designing tests that are both appropriate and applicable to
the study species. So in order to get inside an animal’s
head, to try to understand why it does what it does, one
must first understand how it views the world around it. Fish
occupy an incredible array of habitats, and there are a lot of
species, so it is very difficult to generalise. Nevertheless,
the environment in which a fish lives really shapes the
senses it relies on. This theme will crop up repeatedly in
what follows. Water differs from air in a wide range of
biologically important ways, and it has ramifications for
understanding the evolution of fish senses, behaviour and
cognition.
Vision
Most fish have standard vertebrate eyes and those living in
shallow water habitats, such as coral reefs, have the full
spectrum of light available to them. Since the early
experiments conducted by von Frisch (1913), we have
known that most species are at least tetrachromatic which
means they can differentiate between colours better than
we can. The common goldfish for example has cones that
absorb at 400, 450, 530 and 620 nm (Neumeyer 1982).
Fishes’ visual acuity (i.e. how clearly they see objects) is
just as good as ours (Douglas and Hawryshyn 1990).
Work in cichlids, guppies and sticklebacks all show that
ultraviolet (UV) is important for mate choice and species
recognition. The detection of UV can also occur in some
parts of the life cycle and then be reduced or lost entirely.
Brown trout, for example, can see in the UV when they are
young and occupy relatively shallow streams, but then lose
this ability as they grow and occupy deeper waters
(Deutschlander et al. 2001). There is some suggestion that
some species of fish use UV as a discrete channel for
communication since their predators are unable to see in
this part of the visual spectrum (Siebeck et al. 2010). An
equal array of fish can see polarised light, including damsel
fish, cichlids, salmonids and goldfish (Kamermans and
Hawryshyn 2011). In these species of fish, polarised light
can be used for three primary purposes: (1) increased
contrast when foraging on small prey such as zooplankton,
(2) visual communication and (3) spatial orientation.
The fact that fish have such good vision means it can be
difficult to present them with realistic stimuli using stan-
dard video equipment. Nevertheless, fish do respond to
video playback. In fact, video playback is often a useful
tool to break down various elements of complex cues
(Rosenthal 1999). For example, female swordtails are
attracted to the long swords that adorn the male caudal fin,
but it is difficult to isolate the rest of the male’s charac-
teristics from the sword. Rosenthal and Evans (1998)
Anim Cogn
123
showed manipulated video clips of the same male with a
tail, without a tail but of equal total length and a dancing
tail on its own to receptive females. Females were not
interested in a dancing sword on its own and did not dif-
ferentiate between males with and without tails. Thus, the
sword tail most likely evolved as a cheap means for males
to fake larger body sizes preferred by females.
Interestingly, fish are also known to fall for optical
illusions which suggest that they not only take in visual
information but complex processing also occur at the level
of perceptual organisation (see Agrillo et al. 2013 for a
review). For example, both goldfish and redtail splitfin
perceive illusory contours in a manner similar to primates
(Wyzisk and Neumeyer 2007; Sovrano and Bisazza 2008,
2009). Even sharks see optical illusions (Fuss et al. 2014).
The fact that these fish species are highly divergent sug-
gests that this ability is widespread in fish and vertebrates
generally. Rather than seeing what is actually there, the
vertebrate brain makes assumptions based on preconcep-
tions which are likely based on a mix of prior experience
and neural circuitry (Kandel et al. 2000). Thus, the fish
brain, like the primate brain, seems to examine objects as a
whole rather than paying attention to particular parts. If
certain parts of an object are missing, the brain fills them
in.
Olfaction
Smell (olfaction) and taste (gustation) in fishes work much
in the same way as in humans. The chemosensory ability of
most fishes is very highly developed, and they use this
information for a wide variety of behaviours including
feeding, predator recognition, mate choice and navigation.
The smelling ability of sharks is about 10,000 times more
sensitive than ours. Of the five primary gustatory catego-
ries, human identify and respond to (sweet, sour, salty,
umami, and bitter), fish probably only respond to two of
these (sour and bitter) in a biologically meaningful way
(for a review see Hara 1994).
Because fish are submersed in water, there is no
restriction as to the location of their taste buds. The vari-
ability in location and density of taste buds tells us a lot
about the environment in which the fish lives and its
feeding habits. Cyprinids, for example, may have as many
as 300 taste buds per mm
2
, and density is associated with
the fishes’ feeding habits (Gomahr et al. 1992). Some
classic examples serve to illustrate how important chemo-
sensory abilities are to fishes include response to alarm
substance, the use of the major histocompatibility complex
(MHC) for kinship and mate recognition, and imprinting on
natal stream chemical cues in salmonid homing behaviour.
von Frisch (1941) discovered that minnows showed an
innate fright response when one of their conspecifics was
harmed. He named the mysterious chemical substance
Schreckstoff (literally ‘fright stuff’’), and we now know
that this substance is held in special skin cells (club cells)
in all ostariophysan fishes. The substance plays a crucial
role in enabling naı
¨ve fish to recognise dangerous predators
via classical conditioning. When a conspecific is hurt or
killed by a predator, the alarm substance is released into the
water resulting in anti-predator responses in the surround-
ing fish. When this occurs, fish are capable of pairing the
appearance of a predator with an appropriate biological
response (e.g. schooling or crypsis) (for a review, see
Brown et al. 2011b). Recent evidence has shown that
embryonic fish are capable of distinguishing between pre-
dators (Oulton et al. 2013) and that the pairing between
alarm substances and predator cues during embryonic
development enhances anti-predator behaviour once the
eggs hatch (Nelson et al. 2013). This learning mechanism
affords a high degree of behavioural plasticity which is
necessary in a reasonably unpredictable, natural
environment.
As we saw with the swordtail example above, female
fish can be very choosey about whom to mate with. They
can rely on a range of visual cues to determine the likely
fitness of a male (such as size or colour) but they can also
rely on chemical cues to figure out how genetically com-
patible males might be. Studies using sticklebacks have
shown that females prefer to mate with males that have a
different set of MHC alleles to themselves. MHC allelic
diversity determines how effective an individual’s immune
system is likely to be. Female sticklebacks choose males
that optimise rather than maximise MHC diversity in her
resulting offspring (Milinski 2003). Studies conducted on
humans have yielded similar results (Thornhill et al. 2003).
Last and not least, one of the greatest examples of ani-
mal migrations is the annual return of salmonids to their
home streams. Many salmonids breed in cool upland rivers
where their eggs are laid in gravel beds. As the eggs
develop and hatch, the larvae and emerging young imprint
on the chemical signature of their home stream. After
spending various lengths of time in freshwater, they
migrate to the ocean where they may spend several years
growing to adulthood. Once mature, the fish return to their
home streams by following chemical plumes from the river
and matching it to the template laid down during the
imprinting stage (Dittman and Quinn 1996).
Hearing
Anyone who has ever been diving will know that the
underwater world is full of pops, creaks, snaps and
crackles. As in terrestrial systems, there is a dawn and dusk
chorus underwater associated with peaks in daily activity.
Sound travels far greater distances (less attenuated) in
Anim Cogn
123
water and at much greater speed (4.59faster) than in the
air. Moreover, air is readily compressed, so terrestrial
vertebrates rely on membrane vibration to detect sound. In
water, however, particle motion is the key to sound dis-
persal and detection. Thus, most aquatic animals rely on
this vector to detect noise (see Popper and Lu 2000 for a
review). This has a number of implications for the evolu-
tion of hearing in aquatic organisms.
Once again there is a large amount of hearing ability
variability in teleost fishes, which largely depends on the
environment the fish lives in, but scientists generally rec-
ognise hearing specialists and generalists. Specialists such
as carp, catfishes, mormyrids (mildly electric fishes from
Africa) detect a broad frequency range of the pressure
component of sound, whereas generalists may only detect
the particle motion component over low frequencies (e.g.
perch and salmonids) (Amoser and Ladich 2005). Spe-
cialists generally come from relatively quiet environments
(e.g. deep water, still lakes) whereas the reverse is true for
generalists.
As well as having ears in a conventional sense (the inner
ear is essentially conserved in all vertebrates), fishes also
have a specialised system known as the lateral line which
gathers vibration information from all over the body and
sends it to the brain via the inner ear to be processed. This
gives them a high degree of sensitivity, which is one of the
ways fish in schools are so tightly coordinated. Some fish,
including carp, use their swim bladder to detect sound and
it this vibratory information which passes to the inner ear
through a series of ear bones very similar to our own.
Otiliths (ear bones) inside the fish’s head also respond
directly to vibration.
One generally ‘asks’ what a fish can hear by conduct-
ing behavioural experiments, but to date we have specific
information about the hearing ability of fewer than 100
species. Some fishlike perch and salmon are capable of
detecting infrasound (\35 Hz), whereas others such as
herring can hear sound above 180 kHz. Although histori-
cally there has been debate about whether fish can detect
the location of a sound, recent evidence suggests that they
can but they likely rely on different mechanisms than ter-
restrial vertebrates (Popper and Lu 2000). This certainly
makes sense given the biological importance of localisation
and the manner in which sound travels in air and water.
Fish use sound in a similar way to terrestrial animals,
include mate choice, navigation, territoriality and so on.
They communicate to one another using sound in a variety
of ways. It is thought that perhaps as many as 50 % of fish
species make some kind of meaningful noise for commu-
nication. Male cod and haddock for example have drum-
ming muscles that vibrate on their swim bladder (Hawkins
and Amorim 2000). Females choose males based on their
drumming ability. Similarly, minnows shout at one another
during aggressive interactions (Johnston and Johnson
2000). Many reef fish use the sound of the reef as a cue for
navigation and settlement (Simpson et al. 2004).
Other senses
Unlike mammals, many fishes are also capable of detecting
and creating electric currents in the water. Some fish such
as the elephant nose fish and knife fish have developed a
system similar to sonar where they send out bursts of
electricity and detect changes in the electric field around
them in order to find prey, navigate in their environment
and communicate with their fellows. These fish are both
electrogenic (they produce electricity) and electroreceptive
(they detect and respond to electricity). Others, including
sharks, rays and catfish, sense the tiny amounts of elec-
tricity produced when the neurons in their prey fire.
Strongly electric fish, such as the electric eel, can produce
enough current to stun their prey.
Many fishes can also detect and use the earth’s magnetic
field for both large- and small-scale navigation (magneto-
ception) as do many birds (Mouritsen and Ritz 2005).
Sharks and other cartilaginous fishes have very highly
developed magnetoception and possess a specialised organ
(ampullae of Lorenzi) which detects changes in electric
potential. This system is also involved in electroreceptivity
as mentioned above. Despite the fact that the idea that fish
could detect the earth’s magnetic field and use it for long-
range migration was first raised 120 years ago, the first
experimental evidence did not appear until about 1980.
There is still a considerable debate about how it works and
what it is used for (for a review see Walker et al. 1997).
Cerebral lateralisation
Not so long ago cerebral lateralisation was thought to be a
unique human trait because it is so closely associated with
higher-order cognitive processes in humans such as lan-
guage production. But we now realise that brain hemi-
sphere specialisation is widespread among vertebrates and
perhaps even invertebrates (Rogers 2002; Vallortigara et al.
2011). Fish, like humans, prefer to use one side of their
brain over the other when analysing particular sources of
information (for a review see Bisazza and Brown 2011).
However, the pattern of laterality often varies between
species, between populations of the same species and even
between individuals (Bisazza et al. 2000; Brown et al.
2004; Irving and Brown 2013). For example, Sarasins
minnows look at a familiar individuals with their left eye
and use their right eye to view unfamiliar individuals
(Sovrano 2004). In this species, the right eye is commonly
reserved for looking out for predators or other potentially
threatening objects. This preference for using one eye or
Anim Cogn
123
the other in social contexts also plays a role in the location
rainbowfish prefer to adopt within a school (Bibost and
Brown 2013). Individuals that prefer to use their left eye to
view school mates tend to prefer positions on the right side
of the school, while the reverse is true for those who prefer
to view school mates with the right eye. In this way, a
school might be comprised of an optimal number of fish
with various degrees of laterality. Such a school would
show faster responses to predators and prey on the
periphery of the school while simultaneously rapidly
responding to the behaviour of their school mates. This
idea that laterality enhances multitasking is a common
theory as to why laterality evolved in vertebrates particu-
larly in social contexts (Vallortigara and Rogers 2005;
Brown 2005). The study of laterality is yet another way to
illustrate that fish can perform multiple complex tasks
simultaneously and provides a non-invasive method to
understand which hemisphere fish use when attending to
various stimuli. Moreover, it is increasingly apparent that
laterality varies with emotive content of stimuli and thus
provides a method of studying emotional responses in fish
(Brown and Bibost 2014a).
Fish cognition
So now we have a better understanding of how fish per-
ceive the world. Clearly their senses are just as developed,
and in some aspects better developed than our own. The
question is what do they do with all the information they
are taking in? In the following section, I will briefly outline
types of cognitive abilities of which fish are capable. For
comprehensive reviews on fish cognition, see Brown et al.
(2011a,b) and Bshary et al. (2002).
Learning and memory
Perhaps the best place to begin is with fish memory.
Anyone who has ever kept fish will know that they
remember the hand that feeds them, the time of day and
even the location where the food is likely to appear.
Associating the latter two together is a phenomenon known
as time-place learning. Klausewitz (1960), for example,
reports a situation where a person hand fed fish in a lake.
The fish gradually habituated to the feeder and readily
accepted food when he was present. Even after a 6-month
break the fish recognised him and fed within 3 min and
refused to accept food from others. Time-place learning has
been demonstrated in a number of fish species. A typical
approach is to feed the fish at one end of an aquarium in the
evening and the other end in the morning. Each day the
location of the fish is recorded just prior to feeding. If the
fish show anticipatory behaviour by congregating at the
feeding end, then they have learnt the task. Poeciliids and
galaxiids can learn this task in around two weeks (Brown
unpublished data; Reebs 1999). Golden shiners and angel
fish (Pterophyllum scalare) take 3–4 weeks (Reebs 1996;
Gomez-Laplaza and Morgan 2005). By comparison, rats
take about 19 days to learn this task (Means et al. 2000)
and garden warblers learn slightly more complex tasks
involving four locations and four time periods in just
11 days (Biebach et al. 1989).
It is quite common to see variability in learning ability
between fish species and sometimes within species, par-
ticularly when fish are collected from different locations.
Some of this variability is due to genetics (as a result of
natural selection operating in differing environments),
while some is due to the varying experiences fishes have
during ontogeny. Both mechanisms operate to shape
behaviour to match the prevailing environmental condi-
tions. Typically, one finds that the learning ability, and
behaviour generally, is shaped to suit the environment in
which the fish live. We have examined the effects of
rearing environment on learning using classical or Pav-
lovian conditioning. In Pavlovian conditioning, an animal
effectively learns that one previously meaningless stimulus
(e.g. a bell) is associated with or predicts a biologically
important event (e.g. the arrival of food). This is a funda-
mental form of learning for most animals and was
famously shown in Pavlov’s dogs. In our experiments, fish
had to learn that when a light was turned on food would
shortly be delivered down a feeding tube. Wild rainbowfish
showed rapid learning over 7 days (14 trials) and while a
captive-reared population was initially slower to learn, they
eventually caught up (Brown and Bibost 2014b). By
comparison, rats take around 40 trials to learn an associa-
tion between a tone and food (Bouton and Peck 1989).
Pavlov’s dogs learned the association between a novel
odour and the administration of acid in the mouth in 20
trials, whereas in the classically cited study, a buzzer pre-
dicted the arrival of food ‘after a single combination’
(Pavlov 1927, p. 27).
Spatial learning has been widely studied in fishes (for a
review see Odling-Smee and Braithwaite 2003) and is for
the most part comparable to any other vertebrates. For
example, fish, birds, monkeys and humans can use basic
geometry to recall the location of an object when in an
oblong-shaped room (see Cheng and Newcombe 2005 for a
review). If one of the walls is painted and thus can act as a
salient cue, then adult humans will go to the correct loca-
tion, but toddlers still rely on geometry (Hermer and
Spelke 1994). In fact children cannot use featural cues until
they are about 6 years old. In fishes trained with both
features and geometric cues, fish can use both but rely on
geometry once the features are removed, thus features do
not overshadow geometry during the learning process
Anim Cogn
123
(Sovrano et al. 2002,2003; Vargas et al. 2004). Thus, the
fishes’ use of geometric cues is very similar to birds and
rats and exceeds that of toddlers. One classic example of
complex spatial learning in fish is exhibited by rock-pool
dwelling gobies. A number of studies have shown that
these small fish can also return to their home pool even
after being displaced by 30 m (e.g. White and Brown
2013). Moreover, when disturbed, these gobies can leap
into neighbouring rock pools (Aronson 1956). Even after
being removed from their home pools for 40 days, the fish
could still remember the location of surrounding pools.
This astonishing ability makes use of a cognitive map that
is built-up during the high tide when the fish are free to
roam over the rock platform and which can obviously be
maintained for extended periods of time in the absence of
reinforcement. Thus, gobies and a range of other species
have impressive long-term memories. The use of cognitive
maps is a well-known concept in human navigation, but the
term was first coined by Tolman (1948) to describe aspects
of navigation in rats. It is now known that this sophisticated
mental representation of the spatial relationships between
environmental features is widespread in animal navigation
(Collett and Graham 2004). Clark’s nutcrackers are often
considered to be the animal champions of long-term spatial
memory and can retrieve cached food hordes at least
285 days after they were stashed but a fair degree of for-
getting starts to set in after 183 days (Balda and Kamil
1992).
Animals often form long-term memories of negative
experiences. For example when dogs are trained to avoid
an electric shock, they maintain the avoidance behaviour
for up to 650 trials after the shock has been removed
(Solomon and Wynne 1954). Fish can also learn to avoid
aversive stimuli rapidly and retain the information for
extensive periods. Fish can show one trial learning, for
example pike that have been hooked often show hook
shyness for over a year (Beukemaj 1970). Similarly, rain-
bowfish taught to swim through a hole in a net that trav-
elled down the length of their aquarium took just five runs
to figure out the location of the escape route. When tested
almost a year later, they still recalled how to escape the net
even though they had not seen the net in the intervening
period (Brown 2001). This is remarkable for a fish that only
lives for 2 years in the wild. Moreover, the more fish
present in the group the faster they learn (Brown and
Warburton 1999).
Social learning and traditions
There are widespread examples of social learning in ani-
mals. Perhaps the most celebrated is the unique use of tools
among various populations of chimpanzees (Yamamoto
et al. 2013). Social learning occurs when information
passes from one individual to another by observation or
interaction. This can lead to a transfer of information
through generations (vertical transmission) resulting in
cultural traditions. In the UK, blue and great tits learned
how to open the lids on milk bottles to access the cream on
top and the behaviour rapidly spread across the country via
social learning (Fisher and Hinde 1949). Over the years my
colleagues and I have conducted a large number of
experiments examining social learning in fishes (see Brown
and Laland 2011 for a review). For example, we showed
that hatchery-reared salmon could be taught to recognise
novel live prey by pairing them with fish that already
recognised the prey. The naı
¨ve observers not only learn to
eat the new food but they can also learn where to forage for
it (Brown and Laland 2002a). We have also trained guppies
to use particular routes in order to locate a foraging patch.
In this instance, a school of fish (demonstrators) were
trained to swim through a random door. Gradually one of
the demonstrators was removed from the group and
replaced with a naı
¨ve fish (observers). This continued until
all the original demonstrators had been removed and the
shoals still remained faithful to the original foraging route
(Laland and Williams 1997). In some contexts, however,
the tradition can break down in the absence of demon-
strators (Brown and Laland 2002b), and this can lead to
rapid loss of social traditions.
Social traditions are likely to be particularly strong in
long-lived species and are thought to be responsible for
the migration routes of several species of fish, including
cod (Ferno
¨et al. 2011). It has been postulated that the
recent shifts in cod spawning grounds is the result of the
systematic removal of older, knowledgeable individuals
by commercial fishing (Ferno
¨et al. 2011). Multiple
experiments have shown us how this might work in
practice. During the day, French grunts are found hiding
among sea urchins where they are protected from pre-
dation by the urchin spines. Shortly after sunset, they
migrate to feeding patches often associated with seagrass
beds. The path they use between the resting and foraging
locations is highly temporally stable. Helfman and
Schultz (1984) transplanted fish between schools and
observed what path the transplanted fish took on their
sunset foraging run. Those individuals that were trans-
planted into schools where some of the original school
remained followed the rest of the group (the ‘natives’’)
to their traditional foraging ground. Those that were
transplanted to new locations, where the original school
had been removed, moved off in a heading similar to
that they would have used if they were still at their
home location. Obviously, the latter failed to find a
foraging patch. Thus, it is quite apparent that fishes are
capable of developing cultural traditions that are similar
to some of those seen in birds and primates.
Anim Cogn
123
Individual and kin recognition
Social groups of animals are seldom made up of a random
selection of individuals, and this is also the case with fish.
Usually fish form relatively stable groups and become
familiar with the individuals within that group. Similarly
horses and dogs not only recognise individuals of their own
species (Bonanni et al. 2011) but are capable of recognis-
ing individual humans based on a number of different cues
(Proops and McComb 2012) (recall the early example in
fish). In humans, this capacity for individual recognition
develops between 6 and 10 months of age (Fagan and
Singer 1983). Guppies can remember the identities of up to
15 individuals with little difficulty (Griffiths and Magurran
1997). In relatively small shoals, familiarity takes about
12 days to develop and may be maintained even after
5 weeks of isolation (Bhat and Magurran 2006). When
given a choice, fish will prefer to shoal with familiar rather
than unfamiliar individuals (Magurran et al. 1994).
Remembering who’s who is important as it enables one to
predict how each individual is going to behave in any given
situation. Shoals comprised of familiar individuals, for
example, and are better at avoiding predators than groups
comprised of complete strangers (Chivers et al. 1995).
Social learning is also enhanced in groups of familiar fish
(Lachlan et al. 1998), and territorial cichlids show lower
levels of aggression towards familiar than unfamiliar
neighbours (dear enemy effect; Frostman and Sherman
2004). However, after many generations in captivity, fish
may lose this preference for familiarity (Kydd and Brown
2009) which will have follow-on consequences for a range
of social behaviours. Fish may also be able to recognise kin
using both visual and chemical information (Arnold 2000;
Gerlach et al. 2008), but in many cases this appears to be
facilitated through familiarity mechanisms (Frommen et al.
2007).
Self-recognition
Many have argued that to be conscious animals must be
capable of recognising the self. The typical test employed
in this context is the mirror self-recognition test (MSR:
Parker et al. 1994). In the classic example, a chimp has a
small sticky dot attached to its forehead and then is given
access to a mirror. In some cases, the chimp uses the mirror
to study its reflection and then removes the dot from its
forehead (as opposed to trying to remove the dot from the
mirror; Gallup 1970). This suggests that the animal
understands that it is looking at a reflection of itself and not
an unknown conspecific. Most vertebrates, however, fail
this test and the ability to recognise a mirror image of
oneself seems to be largely limited to humans, great apes
and dolphins (Reiss and Marino 2001). A typical response
towards a mirror image by most animals is fear or
aggression towards the unknown conspecific. Such
responses are commonly observed in fishes. Great debate
continues about whether or not the MSR tells us anything
about an animal’s ability to have more abstract levels of
self-awareness. However, it may be argued that self-rec-
ognition may not necessarily occur through solely through
vision, and in many cases, olfactory recognition is perhaps
more appropriate. This is certainly the case in fishes. Fish
seldom (if ever) see their reflection so they are unlikely to
have evolved visual self-recognition, but chemical cues
play a very important role in aquatic ecosystems. There is
compelling evidence that fish are capable of self-recogni-
tion using chemical cues. Male cichlids, for example,
preferred their own odour to those from unrelated males or
brothers (Thu
¨nken et al. 2009).
Social intelligence: cooperation and reconciliation
Some have argued that the stable social groups one
observes in primates have led to an increase in cognition
over evolutionary time as members of the group attempt to
outsmart one another to gain access to mating opportunities
(Whiten and Byrne 1997). Submissive behaviours such as
appeasement by low-ranking individuals might be one
method to manipulate higher-ranked individuals. Similar
observations have been made in cichlid groups which also
exist in extended family groups with a number of helpers
(Taborsky 1984) as well as in anemonefish (Fricke 1974).
Moreover, Siamese fighting fish are able to judge their
position in a hierarchy by watching third parties interact
with one another, a process known as eavesdropping
(Oliveira et al. 1998). Female guppies can use the same
method to judge male quality and may switch their prior
mate choice decision based on this information (Dugatkin
and Godin 1992). Similar sorts of behaviour are probably
common in social groups where social hierarchies exist
(e.g. chickens, Hogue et al. 1996; macaques, Silk 1999).
Living in complex social groups thus requires a whole
new set of cognitive capacities to keep track of social
relationships. Social intelligence has been mostly studied in
primates (Byrne and Whiten 1989), but studies with dogs
show that they are also highly socially skilful. Indeed there
is evidence that dogs are better at reading social cues from
humans than any primate, a skill which likely occurred
early in the domestication process and co-opted the very
well-developed social skills ordinarily reserved for social-
ising with pack members (Miklo
´si et al. 2004). This
example serves to illustrate how important the environment
is to shaping cognition and has little to do with how closely
related animals are to humans.
Most fish live in social groups for most of their lives,
and it should come as no surprise that they also display all
Anim Cogn
123
sorts of behaviours that are indicative of social intelligence
(Bshary et al. 2014). Fish often provide the best examples
of cooperation in the animal kingdom (e.g. Shettleworth
2010, chapter 12). For example, fish tend to cooperate with
one another when they are doing dangerous deeds like
inspecting predators. If a pair of fish inspects a predator,
they glide back and forth as they advance towards the
predator each taking it in turn to lead. If a partner should
defect or cheat in any way, perhaps by hanging back, the
other fish will refuse to cooperate with that individual on
future encounters (Milinski 1990). This shows that the fish
not only recall the identity of the defector but they also
assign a social tag to them and punish them on future
encounters.
Perhaps the best known example of cooperation between
fishes is that of the cleaner wrasse and its clients (see a
review by Bshary 2011). Cleaner wrasse occupy cleaning
stations on coral outcrops and remove parasites and dead
skin from the surface of client fish. They have a large
number of regular customers, and they recognise them all
individually. The clients present themselves and perform a
‘clean me’’ stance which signals to the cleaner that they
require a good service. Of course there are many stations a
client can potentially visit so it is very important that the
cleaner does a good job to keep up its reputation. If the
cleaner should accidentally bite the client, then the client
will rapidly swim away. But the cleaner has a mode of
reconciliation; they chase after the distraught client and
give them a back rub, thus enticing them to come again
(Bshary and Wurth 2001). Cleaner fish can be selective
about who they serve and they appear to classify their
clients according to their residential status (local or tran-
sient) and their foraging tendencies (predator or non-
predator). If there are a number of clients waiting to be
serviced, cleaner fish will prioritise transients, knowing
that the locals have nowhere else to go (Tebbich et al.
2002). Cleaners are also partial to cheating and occasion-
ally they nip the skin of clients to obtain a cheap meal. It is
perhaps not surprising that they are far less inclined to
perform this behaviour if the client is a predator. Thus, the
cleaner wrasse not only recognise individual fish, but can
also categorise different fish species based on their preda-
tory tendencies. This ability to generalise is widespread in
fishes. To date, however, categorisation based on ‘‘same-
ness’’, another classic sign of intelligence in animals, has
not been formally tested in fishes.
Another fascinating example of cooperation in fish
occurs between groupers and moray eels. In this case, the
grouper approaches the moray eel and signals its intention
to go hunting. In response, the eel emerges from its cavity
and follows the grouper to the proposed hunting site. Upon
arrival, the eel fossicks among the coral while the grouper
cruises over the surface. Any fish that are scared by the eel
are eaten by the grouper and vice versa. In this way, the
foraging efficiency of both fish is greatly enhanced (Bshary
et al. 2006). Cooperation in general is rare (Stevens et al.
2005) but cooperation between species like this is extre-
mely rare and is reminiscent of the relationship between
working dogs and humans.
Building and tool use
Some have argued that nest building is a precursor or even
equivalent to tool use (Hansell and Ruxton 2008)asit
involves complex manipulation of external objects. More-
over, nest building is often associated with parental care
which is widespread among cichlids and gobies. What is
not clear is the extent to which building involves cognition
as many nest building behaviours appear to have strong
innate underpinnings. Nevertheless, building is reasonably
rare among vertebrates other than birds, although of course
beavers are a key exception. It is worth keeping in mind
that there are more species of fish builders than there are
mammal species. At least 9,000 species of fish build a nest
of some sort, either for laying eggs, or for shelter from
predators. The nests vary tremendously in the materials
used and their ultimate shape and function. Male gouramis
build nests out of bubbles in which they entice females to
lay their eggs. Various species of wrasse produce mucus
cocoons which protect them from predators and parasites
while they sleep (Grutter et al. 2011). The cutlips minnow
collects around 300 identical pebbles and builds a mound
35 cm wide and 10 cm high, and tilefish make coral
mounds up to 1 m high. The jawfish gathers together small
rocks from the sand floor to build a wall in front of its
burrow. These fishy masons search meticulously for rocks
that fit together like a puzzle and leave a hole just big
enough for it to slip through. Perhaps the most impressive
builder is the rockmover wrasse. Each night it gathers up
large bits of coral to make a house, in which it spends the
night sleeping and abandons the next morning. Unlike the
numerous studies of bird nest building behaviour, there
have been very few relating to building behaviour in fish
other than sticklebacks (O
¨stlund-Nilsson and Holmlund
2003).
Not so long ago, tool use was one in a long list of skills
that was supposed to be unique to humans; the theory was
that tool use required an order of cognitive sophistication
that no other species could match and is often cited as a key
moment in our evolutionary history. Most famously Jane
Goodall showed that primates and a number of other ver-
tebrates also use tools (reviewed in van Lawick-Goodall
1970), although tool use in macaques had been identified
much earlier (Carpenter 1887). Since then it has been
revealed that all sorts of animals use tools and fish are no
exception (for reviews see Seed and Byrne 2010; Brown
Anim Cogn
123
2012). Primates are still the main players, but tool use is
also reasonably common in birds, particularly corvids,
where New Caledonian crows are the star tool users (Hunt
1996). Among fishes, a number of species of wrasse use
rocks to crush sea urchins so as to access the meat inside.
Wrasse also commonly use anvils to break open shellfish
(Jones et al. 2011). An examination of tool use in the
wrasse family reveals that tool use has either evolved
multiple times or is a common behaviour (Brown 2012).
Interestingly, wrasse also have a larger than expected brain
size, a theme that is also apparent in both tool-using birds
and primates (Lefebvre et al. 2002; Reader and Laland
2002). Other examples of tool use in fish include cichlids
and catfish which often glue their eggs to leaves and small
rocks and then carry them around when their nest is
threatened. Numerous fish, most notably the archerfish,
squirt water from their mouths in order to dislodge prey
items. Archerfish are perhaps the most well-studied group,
and there is good evidence that they learn to compensate
for the refraction as light passes from water to air (Dill
1977).
Numerical competency
The ability to conduct quantitative analyses is a funda-
mental ability in humans. For a long time, it was thought
that humans were the only species capable of counting, but
recent evidence is emerging that perhaps other species can
also do this. Six-month-old infants can discriminate among
small collections of objects (Antell and Keating 1983). Both
five-month-old infants and monkeys can perform simple
arithmetic with small numbers (Wynn 1992; Sulkowski and
Hauser 2001). Similarly, young chickens are capable of
simple arithmetic (Rugani et al. 2009). In humans, there
appears to be two different systems for extracting numerical
information that operate over different ends of the numer-
ical spectrum (Trick and Pylyshyn 1994) although this is
still controversial. The first is a system for estimating large
quantities, referred to as the analogue magnitude system
(Nieder and Dehaene 2009). It has no upper limits but is
subject to Weber’s law as it relates to the ability to differ-
entiate between two quantities. When the ratio is small (i.e.
there is small relative difference between quantities), it is
hard to distinguish between them, but as the ratio increases
(large relative difference) it becomes increasingly simple to
differentiate between them. Such a system is important for
making rapid choices regarding quantities. For example, if
you had to choose between fighting 20 men or 10 men in a
battle, you would rapidly head towards the fight with the
best odds. The second system is basically an object-tracking
system that allows us to follow individual objects (Whalen
et al. 1999). This system is highly accurate but is limited in
that we can usually only keep track of four elements at any
one time. It allows us to instantly report how many objects
we can see without having to count them in series. Dogs, for
example, can spontaneously assess large magnitudes based
on noise, and when in small groups use as object-tracking
system when choosing to attack if their group outnumbers
the invading group by 1 individual (i.e. 2v1, 3v2 or 4v3;
Bonanni et al. 2011). Similar observations have been made
in lions and monkeys (McComb et al. 1994; Kitchen 2004).
Recent research has shown that guppies use the same
methods of quantitative analysis as humans (Agrillo et al.
2012).
Examining counting ability in fishes is relatively simple
and it makes use of the fact that many species school. The
fundamental principal of schooling is that there is safety in
numbers, so the bigger the group the safer you are. Thus,
one can present fish with schools of differing sizes (and
ratios) and their choice of schools tells us about their ability
to differentiate between the two groups. Using this tech-
nique, we have learned that angel fish, for example, use both
numerical systems to keep track of quantities (Go
´mez-
Laplaza and Gerlai 2011a,b). When using the analogue
magnitude system, angel fish could discriminate between
large shoals when the ratio was just below 2:1. When
choosing between small shoals fish could reliably chose 4
versus 1, 3 versus 1, 2 versus 1 and 3 versus 2 individuals,
but could not choose between 4 versus 3, 5 versus 4 and 6
versus 5. In contrast, both guppies and humans can differ-
entiate 3 versus 4 just as easily as 1 versus 4 (Agrillo et al.
2012). Thus, there appears to be some small variability
between fish species, but this likely depends on their
schooling tendencies. There are some problems associated
with these methods in that it is not entirely clear what cues
the fish is using when choosing a school. For example, it
might make a judgement on the overall amount of move-
ment or the total area of fish. However, many of these issues
can overcome by training the fish to respond to inanimate
objects in which surface area and so on can be controlled.
Fish pain perception and consciousness
One of the more controversial questions in fish biology,
and animal ethics generally, is whether fishes feel pain. It is
worth stating from the very outset that there is absolutely
no doubt that fish have all the hardware associated with
pain perception (nociception) and that the application of
analgesics reduces the symptoms (Sneddon 2003). Indeed
if one examines the pain receptors in fish you would find
remarkable resemblance to those in humans. The pain
receptors in all vertebrates are conserved and derived from
an early fishlike ancestor. Although the A-delta and C-fibre
ratios differ somewhat, many of the fish A-delta fibres have
the same characteristics as mammalian C-fibres. So the
Anim Cogn
123
only controversial question at the moment is how do fish
respond to pain in a cognitive sense?
Rose et al. (2014) suggests that pain is an entirely
emotional experience completely separated from the
detection of painful stimuli (nociception). I suggest, how-
ever, that it is unreasonable to separate the physical
detection of pain from the emotional or cognitive response
to it since they are so clearly part of an integrated system
that has evolved to reduce the chances of injury (Rollin
1989; Broom 2001). Thus, pain perception and processing
of nociception stimuli clearly go hand in hand to maximise
an animal’s chances of survival. As Dawkins (2001) sug-
gests, consciousness is a Darwinian adaptation and thus
clearly provides animals with some advantage. Even the
simplest emotional response to pain, fear, is widespread
among vertebrates, and rightly so given that the primary
outcome of this integrated system is to protect the animal
from future harm. Thus, a basic response to pain is to avoid
making contact with noxious stimuli in future. Fear-based
conditioning is one of the foundations of comparative
psychology (Skinner 1938). Moreover, the study of pred-
ator avoidance behaviour is arguably the most common
research topic in modern behavioural ecology. Fish rapidly
learn to associate certain objects, smells and contexts with
potential harm and avoid such things in future (for a
reviews of anti-predator behaviour in fishes see Kelley and
Magurran 2003; Kelley and Brown 2011).
One of the more interesting forms of evidence that fish
respond to pain in a cognitive sense is the fact that pain
appears to distract fish and prevents them from carrying out
other tasks or paying attention to external stimuli. When in
pain, fish suffer from attention deficits. For example, a fish
that has been injected with acetic acid loses fear of novel
objects (neophobia) presumably because the cognitive
experience of pain is dominant over or overshadows other
processes (Sneddon et al. 2003). Similarly, the anti-pred-
ator responses of rainbow trout are disrupted and dominant
fish become less aggressive if they are suffering pain
(Ashley et al. 2009). It is perhaps not surprising given the
diversity of fish behaviour that behavioural indicators of
pain vary from species to species (Reilly et al. 2008). Fish
are also willing to pay a cost to access pain relief (Sneddon
et al. unpublished data).
There are effectively two opposing lines of arguments
regarding emotional/cognitive response to pain. The first
proposed by Rose (2002); Rose et al. (2014) suggests that fish
do not have the cognitive complexity (or hardware) to
respond to pain in an emotional sense. The argument is based
on the fact that fish lack a neocortex and are thus not capable
of consciousness and therefore do not respond to pain as
humans do. The second argument favoured by Sneddon,
Braithwaite, Huntingford, Chandaroo and others suggests
that fish are in fact highly cognitive beings and show all the
signs of responding to pain in a similar way that we do. They
point to experimental evidence that suggests that fish show
fear and other behavioural responses to pain that are not
dissimilar to our own (Sneddon 2003, Braithwaite and
Huntingford 2004; Chandroo et al. 2004). All of the science
discussed in this paper offers strong evidence of fish as
cognitive beings and most likely sentient, a fact Rose must
deny in order to reach his conclusion. More to the point,
however, it would be impossible for fish to survive as the
cognitively and behaviourally complex animals they are
without a capacity to feel pain. Feeling pain and responding
appropriately (e.g. by avoidance) is clearly critical to sur-
vival and may even precede the evolution of higher-order
cognitive processing (Dawkins 2001).
Despite the fact that Rose (2002) is strongly critical of
anthropomorphism, his central argument is both anthropo-
centric and anthropomorphic: fish lack a humanlike neo-
cortex and thus do not respond to pain in a meaningful
conscious sense. In fact this would rule out pain perception in
most vertebrates (e.g. most mammals, all birds, reptiles and
fish). But this argument is contingent on Rose’s erroneous
belief that the neocortex is the centre of consciousness in
humans and that fish lack any comparable structure. On the
first point, recent analysis of the human neocortex has
revealed that it has not undergone any special, radical evo-
lution in humans, and there is no reason to suspect it is any
more involved in consciousness than any other part of the
human brain (Barton and Venditti 2013 and references
therein). As Damasio (1999) points out, core consciousness
in humans depends critically on the activity of a great many
phylogenetically ancient brain areas as well. Modern
approaches to neuropsychology tend to look at conscious-
ness as an emergent property of a highly complex neural
network that cannot be pinned to any particular location or
structure. Even the ‘dynamic core hypothesis’ of con-
sciousness specifically states that the location of the core is
not found any particular location within the brain, but rather
depends on neuronal functional connectivity that is inde-
pendent from anatomical proximity (Tononi and Edelman
1998). In addition, many scientists believe that there are
multiple levels of consciousness (primary, secondary, and
tertiary), and the sorts of processing associated with fear and
pain are almost certainly associated with primary-process
consciousness that are likely widespread among vertebrates
(Panksepp 2005).
On the second point, the more we find out about the brain
structure of fishes, the more we realise they effectively have
analogous structures and functions to other vertebrates
(Broglio et al. 2011; Demski 2013). The fish forebrain is
dominated by eversion of pallial masses rather than evagi-
nation as in the other vertebrates, thus the topology is dif-
ferent which has hindered comparative studies.
Nevertheless, over the last three decades, it has become
Anim Cogn
123
evident that the teleost brain is in fact very similar to that of
the rest of the vertebrates. For example, the teleost dorsal
telencephalon represents the pallium and the ventral telen-
cephalon is analogous to the subpallium and that these areas
are highly connected to the rest of the brain (Rink and
Wullimann 2004). The main divisions in the telencephalic
dorsal pallium in fish are homologous to the tetrapod hip-
pocampus, amygdala and neocortex (Broglio et al. 2011).
We have known for a long time that areas of the telenceph-
alon are involved in emotion, particularly the medium pal-
lium, based on a data ranging from gene expression to
behaviour. Similarly, the fish cerebellum is greatly involved
in fear conditioning as it is in mammals (Sacchetti et al. 2009
for a review). Moreover, fMRI studies have shown that when
suffering from pain, there is significant activity in the fish
forebrain which is highly reminiscent of that observed in
humans (see Sneddon 2011).
Conclusions
Fish have very good memories, live in complex social
communities where they keep track of individuals and can
learn from one another; a process that leads to the develop-
ment of stable cultural traditions. They recognise themselves
and others. They cooperate with one another and show signs
of Machiavellian intelligence such as cooperation and rec-
onciliation. They build complex structures, are capable of
tool use and use the same methods for keeping track of
quantities as we do. For the most part, their primary senses
are just as good, and in many cases better, than our own.
When comparing their behaviour to primates, one finds very
few differences with the exception, perhaps, of the ability for
imitation (Bshary et al. 2002). One must conclude, therefore,
that the level of cognitive complexity displayed by fishes is
on a par with most other vertebrates, and that if any animals
are sentient then one must conclude that fish are too. While
their brain evolutionary and developmental trajectory differs
from other vertebrates, it is evident that there are many
analogous structures that perform similar functions. This
body of evidence strongly suggests that they are sentient and
the evidence that they are capably of feeling pain in a manner
similar to humans is gradually mounting. I submit that there
are compelling reasons to include fish in our ‘moral circle’
and afford them the protection they deserve.
References
Agrillo C, Piffer L, Bisazza A, Butterworth B (2012) Evidence for
two numerical systems that are similar in humans and guppies.
PLoS One 7(2):e31923
Agrillo C, Petrazzini MEM, Dadda M (2013) Illusionary patterns are
fishy for fish, too. Front Neural Circuits 7:137
Amoser S, Ladich F (2005) Are hearing sensitivities of freshwater fish
adapted to the ambient noise in their habitats? J Exp Biol
208:3533–3542
Antell SE, Keating DP (1983) Perception of numerical invariance in
neonates. Child Dev 54:695–701
Appleby MC, Sandøe P (2002) Philosophical debate on the nature of
well-being: implications for animal welfare. Anim Welf
11:283–294
Arnold KE (2000) Kin recognition in rainbowfish (Melanotaenia
eachamensis): sex, sibs and shoaling. Behav Ecol Sociobiol
48:385–391
Aronson LR (1956) Further studies on orientation and jumping
behavior in the goby fish, Bathygobius soporator. Anat Rec
125:606
Ashley PJ, Ringrose S, Edwards KL, Wallington E, McCrohan CR,
Sneddon LU (2009) Effect of noxious stimulation upon
antipredator responses and dominance status in rainbow trout.
Anim Behav 77:403–410
Balda RP, Kamil AC (1992) Long-term spatial memory in clark’s
nutcracker, Nucifraga columbiana. Anim Behav 44:761–769
Barton RA, Venditti C (2013) Human frontal lobes are not relatively
large. Proc Natl Acad Sci 110:9001–9006
Bekoff M (2006) Animal passions and beastly virtues: reflections on
redecorating nature. Temple University Press, Philadelphia
Bekoff M, Sherman PW (2004) Reflections on animal selves. Trends
Ecol Evol 19:176–180
Beukemaj JJ (1970) Acquired hook-avoidance in the pike Esox lucius
L. fished with artificial and natural baits. J Fish Biol 2:155–160
Bhat A, Magurran AE (2006) Benefits of familiarity persist after
prolonged isolation in guppies. J Fish Biol 68:759–766
Bibost A-L, Brown C (2013) Laterality enhances schooling position
in rainbowfish, Melaotaenia spp. PLoS One 8(11):e80907
Biebach H, Gordijn M, Krebs JR (1989) Time-and-place learning by
garden warblers, Sylvia borin. Anim Behav 37:353–360
Bisazza A, Brown C (2011) Lateralization of cognitive functions in
fish. In: Brown C, Krause J, Laland KN (eds) Fish cognition and
behavior. Wiley, Oxford, pp 298–324
Bisazza A, Cantalupo C, Capocchiano M, Vallortigara G (2000)
Population lateralisation and social behaviour: a study with 16
species of fish. Laterality 5:269–284
Bitterman ME (1965) The evolution of intelligence. Sci Am
212:92–100
Bitterman ME (1975) The comparative analysis of learning. Science
188:699–709
Block N (1991) Evidence against epiphenomenonalism. Behav Brain
Sci 14:670–672
Bonanni R, Natoli E, Cafazzo S, Valsecchi P (2011) Free-ranging
dogs assess the quantity of opponents in intergroup conflicts.
Anim Cogn 14:103–115
Bouton ME, Peck CA (1989) Context effects of conditioning,
extinction, and reinstatement in an appetitive conditioning
preparation. Anim Learn Behav 17:188–198
Braithwaite VAB, Huntingford FA (2004) Fish and welfare: do fish
have the capacity for pain perception and suffering? Anim Welf
13:s87–s92
Broglio C, Go
´mez A, Dura
´n E, Salas C, Rodrı
´guez F (2011) Brain and
cognition in teleost fish. In: Brown C, Krause J, Laland K (eds)
Fish cognition and behavior. Wiley, Oxford, pp 325–358
Broom DM (2001) Evolution of pain. R Soc Med Int Congr Symp Ser
246:17–25
Brown C (2001) Familiarity with the test environment improves
escape responses in the crimson spotted rainbowfish, Melano-
taenia duboulayi. Anim Cogn 4:109–113
Anim Cogn
123
Brown C (2005) Cerebral lateralisation; social constraints and
coordinated antipredator responses. Behav Brain Sci 28:591–592
Brown C (2012) Tool use in fishes. Fish Fish 13:105–115
Brown C, Bibost A-L (2014a) Laterality is linked to personality in the
black-lined rainbowfish, Melanotaenia nigrans. Behav Ecol
Sociobiol 68:999–1005
Brown C, Bibost L-A (2014b) Laterality influences cognitive
performance in rainbowfish Melanotaenia duboulayi. Anim
Cogn. doi:10.1007/s10071-014-0734-3
Brown C, Laland K (2002a) Social enhancement and social inhibition
of foraging behaviour in hatchery-reared Atlantic salmon. J Fish
Biol 61:987–998
Brown C, Laland KN (2002b) Social learning of a novel avoidance
task in the guppy: conformity and social release. Anim Behav
64:41–47
Brown C, Laland K (2011) Social learning in fishes. In: Brown C,
Krause J, Laland K (eds) Fish cognition and behavior. Wiley,
Oxford, pp 240–257
Brown C, Warburton K (1999) Social mechanisms enhance escape
responses in shoals of rainbowfish, Melanotaenia duboulayi.
Environ Biol Fish 56:455–459
Brown C, Gardner C, Braithwaite VA (2004) Population variation in
lateralised eye use in the poeciliid Brachyraphis episcopi. Proc R
Soc Lond B 271:S455–S457
Brown C, Laland K, Krause J (2011a) Fish cognition and behavior.
In: Brown C, Krause J, Laland K (eds) Fish cognition and
behaviour. Wiley, Oxford, pp 1–9
Brown GE, Ferrari MCO, Chivers DP (2011b) Learning about danger:
chemical alarm cues and threat-sensitive assessment of predation
risk by fishes. In: Brown C, Krause J, Laland K (eds) Fish
cognition and behavior. Wiley, Oxford, pp 59–80
Bshary R (2011) Machiavellian intelligence in fishes. In: Brown C,
Krause J, Laland K (eds) Fish cognition and behavior. Wiley,
Cambridge, pp 277–297
Bshary R, Wurth M (2001) Cleaner fish Labroides dimidiatus
manipulate client reef fish by providing tactile stimulation. Proc
R Soc Lond B 268:1495–1501
Bshary R, Wickler W, Fricke H (2002) Fish cognition: a primate’s
eye view. Anim Cogn 5:1–13
Bshary R, Hohner A, Ait-el-Djoudi K, Fricke H (2006) Interspecific
communicative and coordinated hunting between groupers and
giant moray eels in the Red Sea. PLoS Biol 4:e431
Bshary R, Gingins S, Vail AL (2014) Social cognition in fish. Trends
Cogn Sci. doi:10.1016/j.tics.2014.04.005
Byrne R, Whiten A (1989) Machiavellian intelligence II: Extensions
and evaluations. Cambridge University Press
Carpenter A (1887) Monkeys opening oysters. Nature 36:53
Chandroo KP, Yue S, Moccia RD (2004) An evaluation of current
perspectives on consciousness and pain in fishes. Fish Fish
5:281–295
Cheng K, Newcombe NS (2005) Is there a geometric module for
spatial orientation? Squaring theory and evidence. Psychon Bull
Rev 12:1–23
Chivers DP, Brown GE, Smith RJF (1995) Familiarity and shoal
cohesion in fathead minnows (Pimephales promelas)—implica-
tions for antipredator behavior. Can J Zool 73:955–960
Collett TS, Graham P (2004) Animal navigation: path integration,
visual landmarks and cognitive maps. Curr Biol 14:R475–R477
Coussi-Korbel S (1994) Learning to outwit a competitor in manga-
beys (Cercocebus torquatus torquatus). J Comp Psychol
108:164–171
Damasio A (1999) The feeling of what happens. Heinemann, London
Dawkins MS (1998) Evolution and animal welfare. Q Rev Biol
73:305–328
Dawkins MS (2001) Who needs consciousness? Anim Welf 10(Suppl
1):19–29
Demski LS (2013) The pallium and mind/behavior relationships in
Teleost fishes. Brain Behav Evol 82:31–44
Deutschlander ME, Greaves DK, Haimberger T, Hawryshyn CW
(2001) Functional mapping of UV photosensitivity during
metamorphic transitions in a salmonid fish, Oncorhynchus
mykiss. J Exp Biol 204:2401–2413
Dill LM (1977) Refraction and the spitting behavior of the archerfish
(Toxotes chatareus). Behav Ecol Sociobiol 2:169–184
Dittman A, Quinn T (1996) Homing in Pacific salmon: mechanisms
and ecological basis. J Exp Biol 199:83–91
Douglas R, Hawryshyn C (1990) Behavioural studies of fish vision:
an analysis of visual capabilities. In: Douglas R, Djamgoz M
(eds) The visual system of fish. Springer, Netherlands,
pp 373–418
Dugatkin LA, Godin J-GJ (1992) Reversal of female mate choice by
copying in the guppy Poecilia reticulata. Proc R Soc Lond B
249:179–184
Duncan IJH, Petherick JC (1991) The implications of cognitive
processes for animal welfare. J Anim Sci 69:5017–5022
Fagan JF, Singer LT (1983) Infant recognition memory as a measure
of intelligence. Adv Infancy Res 2:31–78
Ferno
¨A, Huse G, Jakobsen PJ, Kristiansen TS (2011) The role of fish
learning skills in fisheries and aquaculture. In: Brown C, Krause
J, Laland KN (eds) Fish cognition and behaviour. Blackwell,
Oxford, pp 278–310
Fisher J, Hinde RA (1949) The opening of milk bottles by birds. Br
Birds 42:347–358
Fricke H (1974) O
¨ko-Ethologie des monogamen Anemonenfisches
Amphiprion bicinctus. Z Tierpsychol 36:429–512
Frommen JG, Luz C, Bakker TCM (2007) Kin discrimination in
sticklebacks is mediated by social learning rather than innate
recognition. Ethology 113:276–282
Frostman P, Sherman PT (2004) Behavioral response to familiar and
unfamiliar neighbors in a territorial cichlid, Neolamprologus
pulcher. Ichthyol Res 51:283–285
Fuss T, Belackmann H, Schleussel V (2014) The brain creates
illusions not just for us: sharks (Chiloscyllium griseum) can ‘‘see
the magic’ as well. Front Neural Circuits 8:24
Gallup GG (1970) Chimpanzees: self recognition. Science 167:86–87
Gerlach G, Hodgins-Davis A, Avolio C, Schunter C (2008) Kin
recognition in zebrafish: a 24-hour window for olfactory
imprinting. Proc R Soc Lond B 275:2165–2170
Gomahr A, Palzenberger M, Kotrschal K (1992) Density and
distribution of external taste buds in cyprinids. Environ Biol
Fish 33:125–134
Go
´mez-Laplaza L, Gerlai R (2011a) Can angelfish (Pterophyllum
scalare) count? Discrimination between different shoal sizes
follows Weber’s law. Anim Cogn 14:1–9
Go
´mez-Laplaza L, Gerlai R (2011b) Spontaneous discrimination of
small quantities: shoaling preferences in angelfish (Pterophyllum
scalare). Anim Cogn 14:565–574
Gomez-Laplaza LM, Morgan E (2005) Time-place learning in the
cichlid angelfish, Pterophyllum scalare. Behav Process
70:177–181
Griffiths SW, Magurran AE (1997) Familiarity in schooling fish; how
long does it take to acquire. Anim Behav 53:945–949
Grutter AS, Rumney JG, Sinclair-Taylor T, Waldie P, Franklin CE
(2011) Fish mucous cocoons: the ‘mosquito nets’ of the sea. Biol
Lett 7:292–294
Hansell M, Ruxton GD (2008) Setting tool use within the context of
animal construction behaviour. Trends Ecol Evol 23:73–78
Hara TJ (1994) Olfaction and gustation in fish: an overview. Acta
Physiol Scand 152:207–217
Hawkins A, Amorim MC (2000) Spawning sounds of the male
haddock, Melanogrammus aeglefinus. Environ Biol Fish
59:29–41
Anim Cogn
123
Helfman GS, Schultz ET (1984) Social transmission of behavioural
traditions in a coral reef fish. Anim Behav 32:379–384
Hermer L, Spelke ES (1994) A geometric process for spatial
reorientation in young children. Nature 370:57–59
Heyes CM (1993) Imitation, culture and cognition. Anim Behav
46:999–1010
Heyes CM (1994) Social learning in animals: categories and
mechanisms. Biol Rev 69:207–231
Hogue ME, Beaugrand JP, Lague
¨PC (1996) Coherent use of
information by hens observing their former dominant defeating
or being defeated by a stranger. Behav Process 38:241–252
Hunt GR (1996) Manufacture and use of hook-tools by New
Caledonian crows. Nature 379:249–251
Huntingford FA, Adams C, Braithwaite VA, Kadri S, Pottinger TG,
Sandøe P, Turnbull JF (2006) Current issues in fish welfare.
J Fish Biol 68:332–372
Irving E, Brown C (2013) Examining the link between personality
and laterality in a feral guppy Poecilia reticulata population.
J Fish Biol 83:311–325
Iwama GK (2007) The welfare of fish. Dis Aquat Org 75:155–158
Johnston CE, Johnson DL (2000) Sound production in Pimephales
notatus (Rafinesque) (Cyprinidae). Copeia 2000:567–571
Jones AM, Brown C, Gardner S (2011) Tool use in the tuskfish
Choerodon schoenleinii? Coral Reefs 30:865
Kamermans M, Hawryshyn CW (2011) Teleost polarisation vision:
how it might work and what it might be good for. Philos Trans R
Soc Lond B 366:742–756
Kandel ER, Schwartz JH, Jessell TM (2000) Principles of neural
science. McGraw-Hill, New York
Kelley JL, Brown C (2011) Predation risk and decision making in
poeciliid prey. In: Evans JP, Pilastro A, Schlupp I (eds) Ecology
and evolution of poeciliid fishes. University of Chicago Press,
pp 174–184
Kelley JL, Magurran AE (2003) Learned predator recognition and
anti-predator responses in fishes. Fish Fish 4:216–226
Kirkwood KK, Hubrecht R (2001) Animal consciousness, cognition
and welfare. Anim Welf 10(Supplement 1):5–17
Kitchen DM (2004) Alpha male black howler monkey responses to
loud calls: effect of numeric odds, male companion behaviour
and reproductive investment. Anim Behav 67:125–139
Klausewitz W (1960) Ein bemerkenswerter Za
¨hmungsversuch an
freilebenden Fischen. Natur Volk 90:91–96
Kydd E, Brown C (2009) Loss of shoaling preference for familiar
individuals in captive-reared crimson spotted rainbowfish Mel-
anotaenia duboulayi. J Fish Biol 74:2187–2195
Lachlan RF, Crooks L, Laland KN (1998) Who follows whom?
Shoaling preferences and social learning of foraging information
in guppies. Anim Behav 56:181–190
Laland K, Williams K (1997) Shoaling generates social learning of
foraging information in guppies. Anim Behav 53:1161–1169
Lefebvre L, Nicolakakis N, Boire D (2002) Tools and brains in birds.
Behavior 139:939–973
Lund V, Mejdell CM, Rocklingsberg H, Anthony R, Hastein T (2007)
Expanding the moral circle: farmed fish as objects of moral
concern. Dis Aquat Org 75:109–118
Macphail EM (1998) The evolution of consciousness. Oxford
University Press, Oxford
Magurran AE, Seghers BH, Shaw PW, Carvalho GR (1994)
Schooling preferences for familiar fish in the guppy, Poecilia
reticulata. J Fish Biol 45:401–406
McComb K, Packer C, Pusey A (1994) Roaring and numerical
assessment in contests between groups of female lions, Panthera
leo. Anim Behav 47:379–387
Mcphail EM (1982) Brian and intelligence in vertebrates. Oxford
University Press, Oxford
Means LW, Ginn SR, Arolfo MP, Pence JD (2000) Breakfast in the
nook and dinner in the dining room: time-of-day discrimination
in rats. Behav Process 49:21–33
Miklo
´si A
´, Topa
´l J, Csa
´nyi V (2004) Comparative social cognition:
what can dogs teach us? Anim Behav 67:995–1004
Milinski M (1990) On cooperation in sticklebacks. Anim Behav
40:1190–1191
Milinski M (2003) The function of mate choice in sticklebacks:
optimizing MHC genetics. J Fish Biol 63:1–16
Mouritsen H, Ritz T (2005) Magnetoreception and its use in bird
navigation. Curr Opin Neurobiol 15:406–414
Nelson A, Alemadi S, Wisenden B (2013) Learned recognition of
novel predator odour by convict cichlid embryos. Behav Ecol
Sociobiol 67:1269–1273
Neumeyer C (1982) Wavelength discrimination in the goldfish.
J Comp Phys A 171:639–649
Nicol CJ (1996) Farm animal cognition. Anim Sci 62:375–391
Nieder A, Dehaene S (2009) Representation of number in the brain.
Ann Rev Neurosci 32:185–208
Odling-Smee L, Braithwaite VA (2003) The role of learning in fish
orientation. Fish Fish 4:235–246
Oliveira RF, McGregor PK, Latruffe C (1998) Know thine enemy:
fighting fish gather information from observing conspecific
interactions. Proc R Soc Lond B 265:1045–1049
O
¨stlund-Nilsson S, Holmlund M (2003) The artistic three-spined
stickleback (Gasterosteous aculeatus). Behav Ecol Sociobiol
53:214–220
Oulton LJ, Haviland V, Brown C (2013) Predator recognition in
rainbowfish, Melanotaenia duboulayi, embryos. PLoS One
8(10):e76061
Panksepp J (2005) Affective consciousness: core emotional feelings
in animals and humans. Conscious Cogn 14:30–80
Parker ST, Mitchell RW, Boccia ML (1994) Self-awareness in
animals and humans: developmental perspectives. Cambridge
University Press, New York
Pavlov IP (1927) Conditioned reflexes. Oxford UniversityPress, Oxford
Popper AN, Lu Z (2000) Structure–function relationships in fish
otolith organs. Fish Res 46:15–25
Proops L, McComb K (2012) Cross-modal individual recognition in
domestic horses (Equus caballus) extends to familiar humans.
Proc R Soc Lond B 279:3131–3138
Reader S (2003) Innovation and social learning: individual variation
and brain evolution. Anim Biol 53:147–158
Reader SM, Laland KN (2002) Social intelligence, innovation and
enhanced brain size in primates. Proc Natl Acad Sci
99:4436–4441
Reebs S (1996) Time–place learning in golden shiners (Pisces:
Cyprinidae). Behav Process 36:253–262
Reebs S (1999) Time-place learning based on food but not on
predation risk in a fish, the inanga (Galaxias maculatus).
Ethology 105:361–371
Reilly SC, Quinn JP, Cossins AR, Sneddon LU (2008) Behavioural
analysis of a nociceptive event in fish: comparisons between
three species demonstrate specific responses. Appl Anim Behav
Sci 114:248–259
Reiss D, Marino L (2001) Mirror self-recognition in the bottlenose
dolphin: a case of cognitive convergence. Proc Natl Acad Sci
98:5937–5942
Rink E, Wullimann MF (2004) Connections of the ventral telenceph-
alon (subpallium) in the zebrafish (Danio rerio). Brain Res
1011:206–220
Rogers LJ (2002) Lateralization in vertebrates: its early evolution,
general pattern, and development. Adv Study Behav 31:107–161
Rollin BE (1989) The unheeded cry: animal consciousness animal
pain and science. Oxford University Press, Oxford, UK
Anim Cogn
123
Romanes GJ (1882) Animal intelligence. Appleton and Co., New
York
Rose JD (2002) The neurobehavioral nature of fishes and the question
of awareness and pain. Rev Fish Sci 10:1–38
Rose JD, Arlinghaus R, Cooke SJ, Diggles BK, Sawynok W, Stevens
ED, Wynne CDL (2014) Can fish really feel pain? Fish Fish
15:97–133
Rosenthal G (1999) Using video playback to study sexual commu-
nication. Environ Biol Fish 56:307–316
Rosenthal GG, Evans CS (1998) Female preference for swords in
Xiphophorus helleri reflects a bias for large apparent size. Proc
Natl Acad Sci 95:4431–4436
Roth G, Dicke U (2005) Evolution of the brain and intelligence.
Trends Cogn Sci 9:250–257
Rugani R, Fontanari L, Simoni E, Regolin L, Vallortigara G (2009)
Arithmetic in newborn chicks. Proc R Soc Lond B
276:2451–2460
Sacchetti B, Scelfo B, Strata P (2009) Cerebellum and emotional
behaviour. Neuroscience 162:756–762
Seed A, Byrne R (2010) Animal tool-use. Curr Biol 20:R1032–R1039
Shettleworth, SJ (2010) Cognition, evolution, and behaviour, 2nd edn.
Kindle Edition. Oxford University Press, Oxford
Siebeck UE, Parker AN, Prenger D, Mathger LM, Wallis G (2010)
Covert face recognition in a species of reef fish. Curr Biol
20:407–410
Silk JB (1999) Male bonnet macaques use information about third-
party rank relationships to recruit allies. Anim Behav 58:45–51
Simpson SD, Meekan MG, McCauley RD, Jeffs A (2004) Attraction
of settlement-stage coral reed fishes to reef noise. Mar Ecol Prog
Ser 276:263–268
Skinner BF (1938) The behavior of organisms: an experimental
analysis. Appleton-Century-Crofts, New York
Sneddon LU (2003) The evidence for pain in fish: the use of morphine
as an analgesic. Appl Anim Behav Sci 83:153–162
Sneddon LU (2011) Pain perception in fish: evidence and implications
for the use of fish. J Conscious Stud 18:209–229
Sneddon LU (2013) Cognition and welfare. In: Brown C, Krause J,
Laland K (eds) Fish cognition and behaviour. Wiley, Cambridge,
pp 405–434
Sneddon LU, Braithwaite VA, Gentle MJ (2003) Novel object test:
examining nociception and fear in the rainbow trout. J Pain
4:431–440
Solomon RL, Wynne LC (1954) Traumatic avoidance learning: the
principles of anxiety conservation and partial irreversibility.
Psychol Rev 61:353–385
Sovrano VA (2004) Visual lateralization in response to familiar and
unfamiliar stimuli in fish. Behav Brain Res 152:385–391
Sovrano VA, Bisazza A (2008) Recognition of partly occluded
objects by fish. Anim Cogn 11:161–166
Sovrano VA, Bisazza A (2009) Perception of subjective contours in
fish. Perception 38:579–590
Sovrano VA, Bisazza A, Vallortigara G (2002) Modularity and spatial
reorientation in a simple mind: encoding of geometric and
nongeometric properties of a spatial environment by fish.
Cognition 85:B51–B59
Sovrano VA, Bisazza A, Vallortigara G (2003) Modularity as a fish
(Xenotoca eiseni) views it: conjoining geometric and nongeo-
metric information for spatial reorientation. J Exp Psychol Anim
Behav Process 29:199–210
Stevens JR, Cushman FA, Hauser MD (2005) Evolving the psycho-
logical mechanisms for cooperation. Ann Rev Ecol Evol Syst
36:499–518
Sulkowski GM, Hauser MD (2001) Can rhesus monkeys spontane-
ously subtract? Cognition 79:239–262
Taborsky M (1984) Broodcare helpers in the cichlid fish, Lamprol-
ogus brichardi: their costs and benefits. Anim Behav
32:1236–1252
Tebbich S, Bshary R, Grutter A (2002) Cleaner fish Labroides
dimidiatus recognise familiar clients. Anim Cogn 5:139–145
Thornhill R, Gangestad SW, Miller R, Scheyd G, McCollough JK,
Franklin M (2003) Major histocompatibility complex genes,
symmetry, and body scent attractiveness in men and women.
Behav Ecol 14:668–678
Thorpe WH (1956) Learning and instinct in animals. Methuen,
London
Thu
¨nken T, Waltschyk N, Bakker T, Kullmann H (2009) Olfactory
self-recognition in a cichlid fish. Anim Cogn 12:717–724
Tolman EC (1948) Cognitive maps in rats and men. Psychol Rev
55:189–208
Tononi G, Edelman G (1998) Consciousness and complexity. Science
282:1846–1851
Trick LM, Pylyshyn ZW (1994) Why are small and large numbers
enumerated differently: a limited-capacity preattentive stage in
vision. Psychol Rev 101:80–102
Vallortigara G, Rogers LJ (2005) Survival with an asymmetrical
brain: advantages and disadvantages of cerebral lateralization.
Behav Brain Sci 28:575–633
Vallortigara G, Chiandetti C, Sovrano VA (2011) Brain symmetry
(animal). Wiley Interdiscip Rev Cogn Sci 2:146–157
van Lawick-Goodall J (1970) Tool-using primates and other verte-
brates. In: Lehrman D, Hinde R, Shaw E (eds) Advances in the
study of behavior, vol 3. Academic Press, New York,
pp 195–249
Vargas JP, Lo
´pez JC, Salas C, Thinus-Blanc C (2004) Encoding of
geometric and featural spatial information by goldfish (Carassius
auratus). J Comp Psychol 118(2):206–216
von Frisch K (1913) Weitere Untersuchungen u
¨ber den Farbensinn
der Fische. Zool Jahrb Abt allg Zool Physiol Tiere 34:43–68
von Frisch K (1941) U
¨ber einen Schreckstoff der Fischhaut und seine
biologische Bedeutung. Z vergl Physiol 29:46–145
Walker MM, Diebel CE, Haugh CV, Pankhurst PM, Montgomery JC,
Green CR (1997) Structure and function of the vertebrate
magnetic sense. Nature 390:371–376
Whalen J, Gallistel CR, Gelman R (1999) Nonverbal counting in
humans: the psychophysics of number representation. Psychol
Sci 10:130–137
White G, Brown C (2013) Site fidelity and homing behaviour in
intertidal fishes. Mar Biol 160:1365–1372
Whiten A, Byrne RW (1997) Machiavellian intelligence II: exten-
sions and evaluations. Cambridge University Press, Cambridge
Wynn K (1992) Addition and subtraction by human infants. Nature
358:749–750
Wyzisk K, Neumeyer C (2007) Perception of illusory surfaces and
contours in goldfish. Vis Neurosci 24:291–298
Yamamoto S, Humle T, Tanaka M (2013) Basis for cumulative
cultural evolution in chimpanzees: social learning of a more
efficient tool-use technique. PLoS One 8(1):e55768
Anim Cogn
123
... The teleost fishes are the largest and most varied vertebrate group, containing thousands of species with a wide range of different lifehistories (Sowersby et al., 2022), ecological niches (Sowersby et al., 2021) habitat preferences (Helfman et al., 2009), and cognitive abilities (reviewed in: Bshary et al., 2002;Bshary, 2011;Oliveira, 2013;Bshary and Brown, 2014;Brown, 2015;Sneddon and Brown 2020). Fishes have well developed visual capabilities (Rosa Salva et al., 2014) and over the last 15 years several studies have tested and documented the capacity for different fish species to visually recognize and distinguish between faces (e.g., Siebeck et al., 2010;Kohda et al., 2015;Wang and Takeuchi, 2017;Saeki et al., 2018;Kawasaka et al., 2019;Sogawa et al., 2023Sogawa et al., , 2024. ...
... Yet despite the face inversion effect occurring in other animals, the ability of non-human animals to holistically process faces has remained controversial (Burke and Sulikowski, 2013). Chimps (Parr et al., 1999;Tomonaga, 1999;Parr et al., 2000;Tate et al., 2006), spider monkeys (Ateles sp.), rhesus macaque (Macaca mulatta; Parr, 2011), sheep (Ovis aries) and the budgerigar (Melopsittacus undulatus) are examples of other animals that have demonstrated the face inversion effect (Brown andDooling, 1992, 1993;Kendrick et al., 1995). However, certain species, including crows and some monkeys so far have not (Parr, 2011;Brecht et al., 2017). ...
Article
Full-text available
The face is the most important area on the human body for visually differentiating between individuals. When encountering another person, humans initially gaze at and perceive the face holistically, utilizing first-order relational information and specific neural systems. Information such as identity and emotional state are then obtained from the face by distinguishing between small inter-individual differences, i.e., second-order relational information. Similar patterns and mechanisms underlying individual face recognition have been documented in primates, other social mammals, birds, and more recently in some fishes. Like humans, fish are capable of rapidly (<0.5 s) and accurately recognizing multiple familiar conspecifics by individual-specific variation in the face. Fish can also recognize faces from various distances and angles, providing evidence for mental representation of faces in this large and diverse vertebrate group. One species, the cleaner fish, has even demonstrated mirror self-recognition (MSR) via self-face recognition, strengthening the claim that non-human animals are capable of having mental images and concepts of faces. Here, we review the evidence for individual face recognition in fishes and speculate that face identification neural networks are both similar and widespread across vertebrates. Furthermore, we hypothesize that first-and second-order face recognition in vertebrates originated in bony fishes in the Paleozoic era ~450 Mya, when social systems first evolved, increasing the importance of individual recognition.
... Fish lack facial and bodily expressions akin to those of mammals; they cannot "scream" when injured or threatened, making it harder for humans to empathize. This message from Braithwaite's (2010) book remains valid today, yet substantial evidence demonstrates that fish are sentient beings capable of experiencing pain and suffering (Braithwaite et al., 2007;Sneddon, 2013Sneddon, , 2019Brown, 2014). They also exhibit complex cognitive abilities, such as nest-building (Bessa et al., 2022), tool use (Brown, 2012), long-term memory (Csányi et al., 1989;Triki and Bshary, 2020), and self-recognition (Kohda et al., 2023). ...
... Indeed, behaviour is influenced by a wide array of factors, including cognitive processes, social interactions and environmental variables, and although our aim in this Commentary is to highlight the physiological nature of behaviour, we acknowledge that the complexity and multidimensionality of behaviour cannot be entirely explained by its physiological underpinnings alone. For example, although social behaviours have a physiological basis that may either enhance or constrain social interactions (Seebacher and Krause, 2017), an animal's decision to interact with others will also be determined by sentient thoughts (Brown, 2015) such as avoidance of certain individuals because of agonistic interactionsand influenced by its motivational state. The latter is discussed in more detail in the following section. ...
Article
The role of behaviour in animal physiology is much debated, with researchers divided between the traditional view that separates physiology and behaviour, and a progressive perspective that sees behaviour as a physiological effector. We advocate for the latter, and in this Commentary, we argue that behaviour is inherently a physiological process. To do so, we outline the physiological basis for behaviour and draw parallels with recognised physiological processes. We also emphasise the importance of precise language that is shared across biological disciplines, as clear communication is foundational in integrating behaviour into physiology. Our goal with this Commentary is to set the stage for a debate and persuade readers of the merits of including behaviour within the domain of animal physiology. We argue that recognising behaviour as a physiological process is crucial for advancing a unified understanding of physiology, especially in the context of anthropogenic impacts.
... Sentience is the capacity to have feelings, such as feelings of pain, pleasure, hunger, thirst, warmth, joy, comfort and excitement. There is substantial scientific evidence for sentience in fish, cephalopods and decapod crustaceans (Birch et al. 2021;Brown 2014). For example, studies reporting the presence of pain receptors and connections to the brain, reactions to painkillers, behavioural responses to pain or fear, individual preferences and associative learning. ...
Chapter
Full-text available
There is growing recognition that global food system transformation is needed, but aquaculture is often overlooked. Aquaculture is the fastest growing animal food production sector of recent decades, increasing 5%–7% annually. This chapter reveals how intensive systems are increasing, with more and more wild fishes used to feed farmed carnivorous species. It reveals how aquatic food production often damages and pollutes the environment and causes suffering to aquatic animals, since they are sentient beings. Each year, an estimated 1.1–2.2 trillion fishes suffer from the inhumane capture and killing practices of commercial fisheries, and a further 78–171 billion fishes are farmed. In intensive farms, animals are subject to high stocking densities, barren environments, stressful handling and transport and inhumane slaughter. Many aquaculture innovations focus on ways to further intensify production and do not properly address environmental and animal welfare issues. The authors recommend that, in future, aquaculture should refocus on regenerative, high welfare farming of species that are low in the food chain and have a positive effect on the environment. Aquatic foods are interconnected with the rest of the food system and must be part of global food system transformation.
... Growing public expectations for humane treatment of fish-whether destined for consumption or used in scientific research-are likely to extend to more public settings like zoos and aquariums. Moreover, our understanding of fish needs has expanded substantially over the past few decades, now addressing more than basic health maintenance (Brown, 2015). There is a marked shift towards giving priority to the mental wellbeing of fishes (Kristiansen et al., 2020;Sneddon and Brown, 2020), as highlighted in the five-domains model of animal welfare, which encompasses nutrition, environment, health, and behavior as factors influencing mental state (Mellor and Beausoleil, 2015). ...
... Indeed, animal welfare, including that of fishes, has been the subject of increasing interest and attention over the years (Figure 1b), and research has highlighted the need for data-driven improvements (Oldfield and Bonano 2023). The philosophical aspects of the topic-in particular the relationship between animal welfare and consciousness, sentience, emotional experiences and pain-have been long debated (Chandroo, Yue, and Moccia 2004;Yue Cottee 2012;Jones 2013;Brown 2015;Sneddon et al. 2014;Lavery and Mason 2023). While academically fascinating, this debate is divisive and can be difficult to tackle empirically (Hart 2023). ...
Article
Full-text available
Billions of fishes are kept in captivity for research and food production world‐wide, with a strong impetus for maintaining high welfare standards. Accordingly, the importance of empirical research into the welfare and husbandry of captive fishes is increasingly acknowledged in both science and aquaculture, alongside growing public and governmental interest. Physical enrichment can have an important influence on welfare in of captive fishes, but many questions remain. Here, we summarise the current state of research and outline knowledge gaps in the area of physical enrichment, which is a fundamental aspect to improving welfare of captive fishes. To explore the level of research interest this area across time we conducted a series of surveys, using the number of papers published per year as a metric. These surveys highlight that work on fish welfare, while representing a relatively low proportion of fish research overall, is increasing rapidly. For species that are of aquaculture importance or used commonly as laboratory subjects, we show a positive relationship between general research interest and number of welfare‐related papers. However, for many, particularly relatively less studied, species the proportion of papers on enrichment remains low, with a slower increase compared to welfare‐related papers in general. In terms of common metrics used to quantify fish welfare, there is a reliance on growth and behaviour, with scope for inclusion and combination of a more comprehensive range of reproducible measures. We finish by highlighting recent progress, promising areas for future research and suggestions for advances in this area.
... Ultimately, adaptation to high stocking densities encourages behavioural plasticity [80,81] as part of the different habituation strategies to a challenging condition [12,82]. However, increasing experimental evidence points to complex cognitive abilities and behaviours of fish [83,84], which may differ depending on the type and intensity of stress [85,86] and, more importantly, on fish species. Notably, this enables some fish species to become habituated [20,38,87,88] or not [37,89,90] to specific stress stimuli. ...
Article
Full-text available
A confinement stress test with 75% tank space reduction and behavioural monitoring through tri-axial accelerometers externally attached to the operculum was designed. This procedure was validated by demonstrating the less pronounced stress response in gilthead sea bream than in European sea bass (950–1200 g). Our study aimed to assess habituation to high stocking densities with such procedure in gilthead sea bream. Animals (420–450 g) were reared (June–August) in a flow-through system at two stocking densities (CTRL: 10–15 kg/m3; HD: 18–24 kg/m3), with natural photoperiod and temperature (21–29 °C), and oxygen levels at 5.2–4.2 (CTRL) and 4.2–3.2 ppm (HD). At the end, blood and muscle were sampled for haematology and transcriptomic analyses, and external tissue damage was assessed by image-based scoring. Four days later, fish underwent a 45 min confinement stress test over two consecutive days. HD fish showed reduced feed intake, growth rates and haematopoietic activity. Muscle transcriptome changes indicated a shift from systemic to local growth regulation and a primed muscle regeneration over protein accretion in HD animals with slight external injuries. After stress testing, HD fish exhibited a decreased recovery time in activity and respiration rates, which was shorter after a second stressor exposure, confirming habituation to high densities.
... Projections indicate that this trajectory will continue, with production expected to reach approximately 111 million tons by 2032, which represents a global surge of 17% compared to 2022 (1). Therefore, it is urgent to take into account the welfare conditions of farmed fish, as they are sentient beings capable of feeling pain and suffering (2)(3)(4)(5)(6)(7). Furthermore, fish frequently face challenges in aquaculture conditions, including spatial constraints, unnatural groupings, barren surroundings, and an array of artificial stressors (8,9). ...
Article
Full-text available
An adequate understanding of fish behaviors and their interaction with farm-specific environments is pivotal for enhancing fish welfare in aquaculture. The fair-fish database aims to provide a consistent overview of the welfare of farmed fish. This platform consolidates ethological knowledge into profiles of farmed aquatic species. Its WelfareCheck profiles are organized around welfare indicators, with each criterion receiving classifications (no findings, unclear, low, medium, and high) regarding the likelihood and potential for individuals of a given species to experience good welfare in aquaculture systems, along with the associated certainty level. These criteria include home range, depth range, migration patterns, reproduction, aggregation patterns, aggressive behavior, substrate needs, stress responses, malformations, and slaughtering protocols. We investigated which of these 10 criteria are most relevant to the overall welfare of a species, considering the likelihood, potential, and certainty of good welfare in aquaculture. To achieve this, we reviewed and recorded the high classifications across each criterion and dimension from all published WelfareCheck profiles. To further investigate knowledge gaps across the criteria, we also recorded classifications marked as unclear and no findings. These were then compared across the criteria to assess the frequency of such classifications. While no significant differences were found between the criteria regarding the likelihood that the surveyed species meet their basic welfare needs, criteria related to reproduction, slaughter practices, and substrate needs demonstrated a high potential for better welfare outcomes. Moreover, reproduction and migration patterns exhibited high certainty in the available literature. Based on these findings, we conclude that improving the reproduction of farmed aquatic species, considering their natural needs and behavior, could be an effective and reliable approach to improving welfare. However, we also found a low certainty of information on aggression and an absence or conflicting data on home range, aggregation patterns, stress, and malformations. This highlights an urgent need for research in these areas, which are fundamental for developing more accurate assessments and recommendations for farmed aquatic species.