Cosmos Editorial
We are taking a look back at stories from Cosmos Magazine in print. In December 2020, James Bradley explored how we test intelligence in fish, and how we treat this diverse, enigmatic community.
Earlier this year, while on a field trip in the Cocos Islands, I took some time out and went snorkelling. I was in a shallow channel between two islets, and the tide was running, so my only real option was to let the current carry me, which it did, and quickly, sending me shooting over an expanse of broken coral and sand. Although I had seen fish earlier in the day, there weren’t that many about in the water I was moving through, but after a few minutes a trevally came angling in towards me. It was a striking animal: its silvery, streamlined body some 70 or 80 centimetres long, with a vivid blue stripe running along its spine and back along its middle, and it approached me quickly, seemingly without fear. At the last moment it arced outward and swooped around me, before turning back to circle me again, and then again, a process it kept up for perhaps another 10 or 12 minutes as the tide carried me further down the channel into the lagoon.
For a time, I was worried it might be thinking of attacking me – certainly there was an edge of aggression to the way it kept circling – but what most struck me about it was its air of purpose, the sense I was being monitored and observed. There was no question this fish was there, a living presence with its own intentions and agenda.
Anybody who spends time in the water will have had similar encounters. But despite them, fish are generally dismissed as creatures of little wit or feeling, and almost never subjects of moral concern. When we do think of them, it is as food, or less commonly, pets of a purely ornamental kind. Pescatarians who regard slaughtering a cow or a pig or a chicken as unutterably cruel happily consume fish as if they are little different to vegetables. Even our everyday language erases their particularity: we speak of one fish or many fish, as if they are so interchangeable it is not worth according them a plural form. Yet these assumptions elide a world of astonishing complexity.
Fish first evolved over half a billion years ago, and have endured because they are supremely well-adapted to their environments. They are also extraordinarily diverse, the 34,000 identified species of fish making up fully 60% of all vertebrate species – more than mammals, birds and reptiles combined. Fish range in size from the minute Paedocypris progenetica, which is found in the peat swamps and blackwater streams of Sumatra and Bintan, and measures a mere 7.9 millimetres in length, to the immense whale shark, Rhincodon typus, which grows to 13 metres, and can weigh well over 20 tonnes (I once swam with one whose tail was taller than I am). They are found in the icy waters of the polar oceans and the blood-warm waters of the tropics, on the mudflats and intertidal zones of mangroves and more than eight kilometres below the surface in the darkness and bone-liquefying pressure of the Mariana Trench.
But this remarkable diversity is only the tip of the iceberg. Over the past two decades, researchers have amassed an impressive body of evidence that fish not only think and feel, but exhibit complex social behaviours and sophisticated cognitive abilities, are capable of learning, problem solving and tool use, and possess culture and even the sort of self-awareness previously assumed to be restricted to primates, dolphins, elephants and a few species of birds. These discoveries demand we rethink not just our assumptions about the cognitive capacities of our finned cousins, but challenge our ideas about what intelligence is – and how it can be tested and identified.
One of the leading figures in this emerging space is Macquarie University behavioural ecologist, Culum Brown. Brown’s early research was on rainbowfish, small river fish native to Australia, New Guinea and Indonesia. A popular aquarium species, rainbowfish possess that peculiarly piscine combination of nervousness and glassy regard that tends to lead humans to dismiss the idea they might be intelligent.
Yet Brown’s work revealed rainbowfish have complex social lives and hierarchies, are capable of learning to avoid dangers such as predators and traps, and – perhaps most surprisingly – of passing these techniques on to other rainbowfish. Nor are these abilities rudimentary: experiments show rainbowfish learn to associate signals with food three times as fast as rats and twice as fast as dogs.
Brown’s early research also added to the growing body of evidence that, contrary to the old joke about goldfish, many fish have excellent long-term mem – ories. Indeed when tested again almost a year later, rainbowfish responded as if no time had passed.
Similar abilities have been observed in many other species of fish: tilapia taught to associate a signal with netting, for instance, remembered the signal and responded accordingly 75 days later, while gobies, which form highly detailed mental maps of the tidal pools in which they live, recall the location of neighbouring pools for at least 40 days after being removed from their original home.
Even more importantly, though, Brown’s re – search demonstrates that not only are fish capable of remembering, they are capable of learning from each other through observation and interaction, meaning information can be passed between individuals and, even more significantly, between generations.
At one level these discoveries should not come as a surprise. As Brown points out, “we’ve known about social learning and cultural transmission in animals for 60 years. Everybody looked for it in chimps first, because they’re so much like us. But since then the search has cascaded through nearly all the animal taxa, to the point where I think it would be fair to say social learning and cultural traditions are present in almost all animals.”
Perhaps unsurprisingly, many of the known examples of social learning in fish relate to food acquisition. Archerfish learn their famous ability to shoot insects from the air by firing water from their mouths after observing the hunting techniques of older fish. Similarly tigerfish in Schroda Dam in South Africa have learned to prey on swallows that fly close to the water’s surface by leaping out of the water to grab them as they pass, an adaptation unique to this particular population, and one that has spread socially between the fish.
But fish also learn foraging and migration routes from each other. French grunts and bluehead wrasse have both been shown to pass information about migration routes from one generation to the next, the older fish teaching the younger the best paths to take. But as with human culture, this form of cultural transmission is highly vulnerable to disruption, with studies showing this knowledge can be quickly lost if the individuals that possess it are removed from the group.
Like other social animals, fish also possess individual personalities and affinities, and can form attachments to each other. A recent study by Brown’s lab discovered that the shy Port Jackson sharks ubiquitous on the eastern Australian coast have well-established social networks and seek out the company of individuals of the same age and sex. In other words, the sharks prefer to socialise with their peers, as we do.
Nor are social behaviours restricted to recognition and attachment. Fish make choices about mate selection on the basis of their observations of relative status within group hierarchies. They also often employ cooperative behaviours, especially when hunting – yellowtail kingfish (known in the US as amberjacks) off the Californian coast have been observed using U-shaped formations to separate out and trap groups of prey fish, behaviour that resembles the hunting techniques of mammals such as wolves and dolphins.
Co-operative behaviour also extends into other aspects of the lives of many fish, in particular species of cichlids that collaborate to protect and raise young. They are also quite capable of punishing fish that do not behave appropriately. When approaching potential predators, sticklebacks use a distinctive stop-start swimming motion to share the risk by taking turns at the front. But if an individual is reluctant to take the lead, or cheats by hanging back, its schoolmates will refuse to cooperate with it in future, meaning the sticklebacks recall the identity of malingerers and remember they are not to be relied upon.
Although these sorts of behaviours have only been observed in a relatively small number of species, it is likely they are widespread – as Brown emphasises, the practical obstacles to detailed study of fish means our understanding of their lives is extremely limited. But these examples make it clear many fish inhabit social worlds at least as rich as those of mammals and birds.
Comparing the intelligence of different species is notoriously difficult, but there is no question managing such social complexity is cognitively demanding, and suggests many fish possess considerably higher levels of cognition than is often assumed. Yet what of other behaviours that are often regarded as markers of high intelligence such as tool use?
The generally accepted definition of tool use requires an animal to grasp something and use it to manipulate another object or organism. For fish, which lack grasping appendages, such behaviour is effectively impossible, a problem compounded by the physics of the underwater environment, which makes it difficult to strike objects together or engage in fine control, both actions basic to many forms of tool use on land. Yet despite this, many fish engage in behaviours that seem to closely resemble tool use. A number of species of wrasse and tuskfish have been observed using rocks as anvils to crush sea urchins or break open shellfish. Likewise, a group of Atlantic cod in an aquaculture facility in Norway recently took to stealing food from an automatic dispenser after they realised tags attached to their bodies could be used to activate it.
To Brown, the refusal to treat such innovations as tool use says more about our definitions than it does about the behaviour. Instead, he argues, the emphasis upon grasping is fundamentally misconceived, and a reflection of the emphasis upon behaviours observed in apes and monkeys.
“What’s interesting is that when [primatologist] Jane Goodall and others came up with a definition of tool use it had nothing to do with appendages; instead it was all about using an object to achieve a goal. But then the whole concept got hijacked by the primatol – ogists, who substituted a primatecentric definition.”
Brown contends Goodall’s original definition was correct, and the most important element in tool use is the intention of the animal, its desire to achieve a goal that might elude it without the tool.
This emphasis upon intentionality and the ma – nipulation of the environment opens up a broader approach to our understanding of the evolutionary origins of such abilities. “There’s obviously a lot of overlap between building nests and using tools,” Brown says. “Both of them are clearly about manip – ulating the environment in a way that enhances your fitness, so you’re either more reproductively success – ful, you’re getting more food, or you’re safer from predators and other environmental stresses.”
This shift in approach is particularly relevant to fish, at least 9000 species of which build nests. Some – times these structures are relatively simple: certain species of wrasse, for instance, create mucous co – coons in which to shelter while they sleep, and some eels, gourami and bettas, such as the Siamese fight – ing fish, use oral secretions to create raft-like bubble nests in which they secrete their eggs. But many species also build structures to protect eggs and young or to provide shelter while they sleep. Sometimes, as with the mounds of coral that triggerfish create to hide within, these structures can seem primitive or haphazard. But as Brown observes, appearances can be deceptive. “It isn’t simple or boring behaviour.
It’s often seriously sophisticated. If you’re building an igloo out of coral rubble that’s going to protect you it has to be structurally complex enough that it won’t fall down around your ears.
In 2006, scientist Redouan Bshary noticed groupers in the Red Sea approaching moray eels and performing a distinctive head-shake, or shimmying motion, to attract a moray’s attention. That done, the tropical odd couple of grouper and moray would head off together in search of prey.
Fast-moving predators, groupers excel at catching fish in open water, but often lose prey when it takes shelter in holes or under rocks. The long, sinuous bodies of morays, on the other hand, are perfectly adapted to catching fish in crevices, but often lose them if they flee into open water. Working together, the groupers and morays in combination were able to cut off all routes of escape for their prey, making them far more effective than either in isolation. Bshary later observed groupers teaming up with humphead wrasse as well, using the much-larger humpheads’ ability to suck prey from crevices with their powerful jaws to complement the groupers’ open-water skills.
Interspecies cooperation of this sort is incredibly rare, and has only been observed in a handful of species of mammals and birds. Yet Bshary quickly realised the groupers also seemed to be engaging in another, even more surprising behaviour: in cases where they trapped prey somewhere inaccessible, the groupers would stop and perform an unusual headstand over its location, seemingly pointing in an effort to attract the attention of morays.
As anybody who has tried to direct the attention of a dog or a cat by pointing at an object knows, referential gestures that seem natural to humans are nothing of the sort. Indeed, while there is some evidence Australian magpies can understand them, and some dogs are able to learn to respond correctly, the only non-human species known for certain to understand referential gestures are chimpanzees and ravens. Even human children only become capable of using and understanding referential gestures around their first birthday.
As the roll call of creatures capable of using them suggests, referential gestures seem to be associated with high intelligence. As a result, biologists have developed a five-fold test to determine whether behaviours are genuinely referential. The gesture must be directed towards an object; it must be communicative rather than mechanically effective; it must be directed at a recipient or potential recipient, whose response must be voluntary; and – perhaps most importantly – it must exhibit the hallmarks of intentionality.
With this in mind, Australian scientist Alex Vail set out to establish whether coral trout – close relatives of the groupers that live on the Great Barrier Reef and engage in similar behaviour – met all five criteria. And as quickly became clear, they do. In other words, the groupers and the trout don’t just look like they’re pointing, they are pointing.
Vail, who now works as a cameraman on major natural history projects such as Blue Planet 2 and Netflix’s Our Planet, admits he was not hugely surprised by the results: having grown up at the Lizard Island Research Station on the Great Barrier Reef, he has been diving with coral trout all his life and has no doubt they are highly intelligent.
Yet he’s also cautious about overstating the scientific implications of the referential behaviour, emphasising that, like tool use (which it resembles in many ways), it is possible to understand it as an evolved response to ecological need rather than a sign of generalised intelligence. “Animals know what they need to do to succeed in their environment,” he says. “But that can be because they’re responding to environmental stimuli, or they can be doing it with a bit more understanding of their actions. In the end we went with the more cautious interpretation because it was more defensible, but even interpreting that way we had a hard time getting the paper published, because most of the people reviewing it were primatologists who aren’t keen on the idea fish might be able to do the same things as their beloved primates.”
The tension between primatologists and researchers studying the behaviour of fish has come to a head over a study of cleaner wrasse. Small, highly social fish native to coral reefs from the Red Sea to the Pacific, cleaner wrasse maintain “cleaning stations”, which other fish visit in order to have the wrasse remove dead skin and parasites. These stations are extremely popular, attracting large numbers of regular clients as well as more casual visitors.
Yet while the wrasse are skilled at their job, they are not entirely trustworthy, and will sometimes seek to supplement their diet with a mouthful of live skin and mucus.
Unsurprisingly the wrasses’ clients often respond badly. So the wrasse choose their victims carefully, only ever biting non-predatory species that are unlikely to retaliate by attacking or eating them. They also distinguish between regular clients and new clients, giving priority to newcomers and rarely nipping them, presumably to avoid scaring off a fish which might have the potential to become a regular client.
But the sneakiness of the wrasse doesn’t stop there. Not only are they less likely to bite clients if they know other fish are watching, they often attempt to manipulate fish that respond badly to being nipped, chasing after them and rubbing their backs and pelvic fins to mollify them. They even seem to understand the relationships between different client species, and if pursued by an irritated client will swim toward a predator of that species to prevent them chasing them further.
This Machiavellian behaviour suggests cleaner wrasse possess the capacity to attribute mental states to other fish and respond accordingly, a concept known as “theory of mind”, which implies self-awareness previously only observed in a handful of primates, dolphins, elephants and a few species of bird.
With this in mind a team led by Japanese scientist Masanori Kohda decided to test whether wrasse were capable of passing the mirror test – long regarded as proof an animal is self-aware – by recognising their reflection. Placed in a tank with a mirror, the wrasse initially reacted with the sort of aggression they would usually direct at a rival, but subsequently abandoned these behaviours in favour of unusual activities such as swimming toward the mirror upside-down, before finally settling down into non-aggressive postures to gaze at their reflections. Interestingly this sequence of behaviours is almost exactly the same as that observed in other species that have passed the mirror test, although spaced over several days rather than a few minutes or hours.
Kohda’s team then removed the wrasse from the tank, anaesthetised them and placed a mark on their faces or necks before returning them to the water. Once again the wrasse responded similarly to other animals that have passed the test, first assuming pos – tures that allow them to examine the mark, and then attempting to remove it by scraping the affected area against the side of the tank or a rock.
Many scientists have been highly critical of the study, chief among them the mirror test’s inventor, evolutionary psychologist Gordon Gallup, who argues the behaviour of the wrasse is ambiguous, and cannot be taken as evidence the fish recognise them – selves or possess the self-awareness that implies. Barbara Reiss, whose work with dolphins showed they were capable of passing the test is similarly sceptical, suggesting the fact the mark resembled a parasite, which the wrasse are hyper-evolved to de – tect, may have skewed the results.
Culum Brown is not impressed. “The mirror test has been the gold standard for individual recognition for 50 years. When primates passed people said, ‘Oh yes, of course’. Then when a dolphin passed they said, ‘I guess that makes sense’. And when some corvids passed they said, ‘Okay, maybe; corvids are pretty smart’. But when a fish passes, suddenly the test must be broken. It’s kind of mind-boggling, but it just goes to show that even scientists have these massive biases.”
Alex Jordan, an evolutionary biologist and leader of the Comparative Evolution of Social Behaviour Research Group at the Max Planck Institute of An – imal Behaviour in Konstanz, Germany, and senior author of the Kohda paper, is even less diplomatic. “If passing the mirror test is evidence of self-aware – ness in chimps and elephants and dolphins and all the other animals you pay $10 to see in the zoo, then if a fish passes you either have to accept it’s self aware or – pardon my French – your test is f**ked.”
So are the wrasse self-aware? Culum Brown laughs and says that he has no doubt. “The social complexity of the cleaner and client relationship clearly illustrates the wrasse are extremely sophisti – cated, so is it really surprising that they’re capable of self-recognition as well as recognising all those other different species? Probably not. Probably it’s part of a generalised social intelligence thing that includes self-recognition.”
Jordan is more circumspect, arguing scientists who see self-recognition as clear evidence of selfawareness are over-interpreting the results of a flawed test. “The mark test isn’t unequivocally telling us that an animal has theory of mind.” Instead he says it is possible the wrasse learn by a process of association that the mirror reflects their world and thus how to use it as a tool. “That would be a very interesting and amazing finding, and extremely cognitively complex. But it wouldn’t involve self-awareness, and all the other behaviours we observed could flow from that.”
What would it mean for a fish to be self aware? What might it be like to be a fish? A century ago the German biologist Jakob von Uexküll coined the term umwelt – literally “surrounding world” – as a way of capturing the way different organisms inhabit different realities, each defined by their particular sensory world.
The world of a fish, its umwelt, is radically different from our own. Fish inhabit a liquid environment, a weightless world of currents and eddies, and like birds in the air, float in three dimensions rather than the two-dimensional world ground-based animals like ourselves take for granted (although their awareness of depth is extremely acute, presumably because of the potentially fatal consequences of misjudging pressure).
Their sensory worlds are also far richer than ours. Although the basic physiology of fish and human eyes is similar, many fish possess extra receptors granting them tetrachromatic vision, and allowing them to see wavelengths we cannot. In some – especially freshwater fish – this allows them to see into the infrared, making it easier for them to see in the muddy, redshifted waters of rivers and lakes. Others are able to see ultraviolet light and some can detect polarised light. Some even have visual worlds that change across the course of their lives: the eyes of salmon, for instance, are calibrated to see blues better when they are in the ocean and reds better when in freshwater.
Most fish also have excellent hearing: the American shad is able to hear sounds up to 180,000 Hz, nine times the range of human hearing, presumably so they can detect the ultrasonic vocalisations of the dolphins that predate them, while other fish use infrasound to assist with migration, allowing them to hear the subterranean rumble of the tides and the movement of water as it breaks and flows against underwater terrain and the shore.
Alongside hearing and vision, fish also use chemoreception to smell chemical traces in the water around them. In most cases this is done using cells in the nostrils, although some species, like catfish, have receptors spread across their entire bodies, allowing them to taste anything they touch.
These chemoreceptors are often extraordinarily acute. Sometimes they are used for hunting. Sharks, for instance, can detect blood in amounts as low as one part in a million, allowing them to scent prey over long distances; an indication of the importance of this ability can be found in the fact that in great whites an astonishing 14% of their total brain mass is devoted to olfaction. Chemoreceptors also play an important part in migration: salmon learn the smell of stream in which they are born; later they follow the threads of that scent back to the same stretch of river.
Some fish also inhabit sensory dimensions entirely alien to our human experience of the world, possessing magnetoreceptors that allow them to follow the lines of force generated by Earth’s magnetic fields or the ability to sense minute electrical currents. Even more significantly fish are able to sense changes in pressure and water movement using specialised cells along their lateral lines. Known as neuromasts, these cells resemble tiny hairs encased in gel, and assist with schooling behaviours, allowing fish to respond almost instantaneously to the movements of other fish, even in the dark.
Can we even start to comprehend such a radically different way of being? After all, many of us struggle to imagine our way across lines of gender or culture. What must it be like to exist in a world where magnetic fields have dimension? To be able to see the direction of light, or extend the boundaries of your body’s perception by registering changes in pressure, or tiny movements of water? Or to be part of a school, moving in unison, one body amongst many?
Faced with the problem of such radical subjectivity, many philosophers have argued our minds simply cannot comprehend such fundamentally different ways of being in the world. Ludwig Wittgenstein suggested that “if a lion could talk we could not understand him”; likewise Thomas Nagel argued consciousness was essentially subjective, meaning that not only is it impossible to reduce the experience of consciousness to a purely material explanation, but – in an echo of Uexhüll – that there is no such thing as objective experience.
Yet even if we accept we can never truly know, perhaps there are ways of at least beginning to intuit these other ways of being. Fish fall for many of the same visual illusions as humans, suggesting similarities in our visual processing, and open-source software now exists that allows ecologists to simulate the visual experience of other organisms. Similarly, we share behaviours: sociality, curiosity, memory, even culture. Perhaps it could be as simple as actually beginning to look, to try to see them for themselves.
But perhaps the real question is not whether we can imagine their worlds, but what it might mean for us to try. How might that change the way we see them? How might that change us?
The philosopher Donna Haraway writes of the importance of making kin with other species, of recognising our connectedness with the non-human world. Making kin is not necessarily about recognising similarity, but about acknowledging difference, strangeness: kin are, in Haraway’s words, “unfamiliar … uncanny, haunting”, and making kin demands we reimagine our selves and our place in the world.
As Culum Brown suggests, mightn’t consciousness be better understood as something multi-variate, a multi-dimensional space shaped by the sensory and cognitive abilities of different species? A function, in other words, of its umwelt? As Alex Jordan puts it, “there are many, many paths to many, many different places. To argue there’s just one developmental scale of cognition and that’s the way a fish or a bacterium or a plant or any other thing interacts with the world is just narcissism.”
Recognising fish for what they are might shift our perspective in other ways as well. Trapped in our humanocentric viewpoint, we tend to conceive of intelligence and culture as relatively recent developments, ways of being in the world that only appeared with the advent of hominids like ourselves. Yet the jawed fish that exist today first appeared in the Silurian period: their existence stretches back hundreds of millions of years. Imagining intelligence spread across such oceans of time cannot help but alter our understanding of our own significance and relationship to other species.
It also demands we think again about our attitudes to fish, why it is so difficult for us to think of them as creatures with their own minds and ways of being in the world. There is no question that it is at least partly because of their otherness and unfamiliarity. But might it not also be because thinking of them as conscious is just too confronting?
In the absence of detailed records, we cannot know for certain how many fish humans kill a year, but studies suggest they number in the trillions. Most of these fish die slow and agonising deaths, suffocating over several hours after they are hauled aboard. Fish that are live-gutted tend not to suffer as long – only 25 to 65 minutes on average – although as the term “live-gutted” suggests, the relative brevity of their distress comes at a price.
In this context, studies arguing fish cannot feel pain seem not just misguided and anachronistic, but self-serving. Studies make it clear we routinely downplay the mental abilities and capacity for suffering of land-based animals we eat; is it so unlikely the same psychological strategy is at work with fish? Perhaps the time has come to reconsider our tendency to emphasise intelligence rather than the capacity for suffering as the basis of moral concern.
We need to recognise the urgency of this challenge. Alex Vail chose to work on coral trout on the Great Barrier Reef instead of the groupers in which the pointing behaviour was first observed because in the six years between Redouan Bshary’s initial observations and Vail’s study the grouper population in the Red Sea almost entirely disappeared.
This collapse is only a footnote to a catastrophe of planetary proportions: a 2015 report by the WWF estimated that between 1970 and 2010 the global fish population dropped by a half, with populations of fish species utilised for food falling by almost three-quarters. Warming waters are radically reshaping habitats and migration patterns, while ocean acidification is disrupting food chains, and affecting the bodies and brains of many species.
Overfishing is even destroying the capacity of some species to feed and migrate, as the removal of the largest – and therefore oldest – individuals leads to the loss of knowledge passed down over thousands of generations.
“Effectively you’re destroying animal cultures which are unlikely to be ever the same again,” says Brown, who describes this process as “cultural genocide” and believes its consequences have probably been underestimated. Yet studies also suggest that fish populations can be restored through the creation of marine reserves, and better regulation of fishing, and there is still a narrow window within which to save at least some of the world’s coral reefs.
Perhaps we will never truly understand the lives of fish; maybe they are too different, too strange, too fishy. But maybe by recognising their particularity, their individuality, their capacity to think and feel and learn, we might begin to see not just them, but ourselves, differently. Perhaps we might, as Haraway says, begin to make kin with them. Maybe by saving them we might save ourselves.
This article was supported by a grant from the Copyright Agency’s Cultural Fund
The Ultramarine project – focussing on research and innovation in our marine environments – is supported by Minderoo Foundation.