The Problem with “Animal Intelligence”
As I mentioned last week, I have a few pre-written pieces that I will be sharing here, hors-série, just to clear them off my desktop. This does not violate the intention I expressed to move to a fortnightly model, since these involve next to no new work on my part. The piece below was written for a magazine whose policy abruptly changed to prioritize staff writers. My very cool and compassionate editor there wrote to me: “My own view is that you should go ahead and publish them [on Substack] because I’d rather people got to read the pieces, which are truly excellent, than you having to wait on our… editorial policy to change.” So that is what I’ll do. Today’s piece involves themes I’ve already explored in this space (e.g., here, here, and here), so the repetitiveness level is high (perhaps an 8?), but because it was written for a large publication with editorial, uh, standards, it also avoids digressions and curlicues, to the extent that I am able, and probably ranks an 8 or so on the scale of “honedness”. Perhaps I will begin putting such numbers at the beginning of every piece, somewhat like the “epistemic commitment” rankings some of the LessWrong authors use (my own epistemic commitment is always 10, at least at the moment of writing). —JEHS
If I were the editor for science journalists writing about new research in animal intelligence, I would lay down some hard new rules.
First, I would demand that my writers commit to avoiding diminutive language for the behaviour and traits of animals. No adult reader needs to see elephants described as “portly pachyderms”. We do not need to be told in rhyme that a macaque, having learned to manipulate a joystick, will be rewarded with a “sweet treat”, or any of the even more painful alliterative descriptions of an animal's food preferences (“gooey guava”, etc.). We do not need to see species described with what are effectively Aesopian epithets: “wise owls“, “crafty corvids”, and so on. These stylistic habits signal, often subconsciously, that the subject about which we are reading is of marginal importance to our world. The cumulative effect across this whole genre of writing is to keep animals in their place, to ensure that no matter what purportedly new discovery about animal intelligence is being reported, nothing is really going to change in our understanding of them and our concrete relations with them.
Second, and perhaps counterintuitively if you are hearing it for the first time, we must make no presumption of the existence of a real, taxonomically meaningful category of living beings, “animals”, as an intermediate step between the plant and the human, between the vegetative and the rational, containing everything from medusae and corals to chimpanzees and elephants, but not containing Homo sapiens. When a caged ape is found to hide rocks away methodically in order to throw them in a systematic assault on zoo visitors at a later moment, there is no meaningful sense in which we may conclude that “animals” are therefore capable of planning for the future. It only turns out that this particular species of ape, or even only this particular species under these artificial and anomalous conditions, is so capable. In this case in fact we are learning far more about our fellow primates in our very close taxonomic neighbourhood, and therefore at least something about ourselves, than we are learning about, say, marine invertebrates.
Third, one must not tell the stories of animals through the stories of the human beings who research them: the “ruffled old scientist with a shock of white hair hanging down over his bespectacled face”, and so on. What does it matter what a scientist's hair looks like? We want to know what the scientist can teach us about the sliver of the world he studies. In the United States this habit seems to have been ingrained into a whole generation through the commanding essays of John McPhee, author, notably, of Annals of the Forgotten World (1999). When McPhee works us from a casual conversation with a geology graduate student on the side of the New Jersey interstate back into the deep geological time of the continent, it is a smooth and elegant motion; when this same technique is repeated countless times by his lesser imitators, it becomes tiresome. But worse still, when the object of study is not quarks or schist, but animals, it is hard not to feel that the author is failing to notice what is right before their eyes: that you don't need human characters in order to tell the story of the natural world. Animals, properly attended to, are always at least as interesting as the people who study them.
Fourth, we must not congratulate ourselves for having recently come around to the idea that there is something of interest going on in the minds of animals. No article should begin with declarations like: “Until fairly recently, animals were considered to be unthinking machines and humans the only truly intelligent species.” This claim is strictly false. What is true is that there has often been considerable pressure on the scientific elite over the past few centuries to adhere to a dogma about the mechanical reducibility of all animal behaviour. But this elite dogma cannot conceal the fact that the great majority of people in all times and places, including many members of the modern scientific elite who kept quiet on the subject for the sake of their jobs, have presupposed a fundamental kinship of human beings to other species, in the contours of our inner lives, in the way we care for our young, fear the blade, disdain the cold, love to eat.
This presumption of common experience, of parallel lifeways, has not generally translated into advocacy for animal rights, but then again the idea of human rights is a recent innovation as well, so this silence should not surprise us. And neither did the traditional presumption of our basic likeness to other species involve an explicit theory of animal intelligence. But this should not be surprising either, since for most of history, in most times and places, we also did not suppose that intelligence is the distinctive trait that marks human beings off from the rest of nature and entails a set of distinctly moral commitments towards them.
I am not, obviously, a science journalist working the “animal-intelligence beat”. I believe in fact that, in spite of their often good intentions, science journalists have systematically subverted and blocked any rigorous public understanding of the topic in question.
Beyond the harmful stylistic habits I've listed above, there is one great overriding conceptual problem that not only subverts the efforts of journalists, but of the scientists about whom the journalists are writing as well. That is, namely, that no one is working with a compellingly coherent account of what intelligence is. And without such an account of intelligence, there can be no scientifically adequate way of setting up tests to measure it.
Now of course we can set up tests to measure distinct abilities, such as the mirror test for self-recognition, various tests for stacking blocks, counting, or registering the awareness of absence. But these are at most particular aptitudes, and not signs of an underlying general faculty called “intelligence”. They are, moreover, highly dependent on the bodily morphology and the sensory apparatus of the animal. No one has yet figured out how to give a mirror test to a sperm whale. These animals are just too big and uncontrolable to manage such a feat. And even if such a thing were to be arranged, it is not clear that a whale --who inhabits a vastly different environment and may have a vastly different conception of its own embodiment and the relationship of embodiment to its personal identity-- should have any reason to find what it sees in the mirror interesting in the way we do (along with many other mammalian species).
For that matter there are species, including many species of mammals, that are naturally blind. And surely no animal that would fail a vision test could be expected to pass a mirror test, yet there is also no conceivable reason to make sightedness a condition of intelligence. Vision is only one sense, but intelligence is supposed to be the general faculty behind all of the senses that puts together the reports the senses give, and makes inferences from them. Sometimes these inferences yield action, yet it is not at all clear that they must, or that they will do so in a way human observers are in a position even in principle to identify.
Cetaceans have long posed a problem in this regard. Lacking hands, and living in a marine environment, the range of opportunities they have for using tools, or manipulating objects, is vastly reduced when compared with the options open to, say, a Japanese macaque, capable of complicated sequential tool-use behaviour in the course of obtaining food. So we have the macaque using a stick to get an apple, and we have the blue whale just swimming along all day, filtering tonnes of microscopic plankton through the sieve of its baleen as it goes. Perhaps the whale is not even conscious that it is consuming food; perhaps it is thinking about other things, some of which it communicates through its “song” to other members of its pod. We have no evidence that these thoughts, and perhaps these propositions, do not extend as far as abstract thoughts, metaphors, poetry. What we do know is that whatever is going on is not measurable in the same way the macaque's problem-solving abilities are. But this is plainly also just a consequence of environmental circumstances and bodily morphology, rather than a mark of inferior intelligence in the whale. It is not just that comparing whale communication and macaque tool-use amounts to apples and oranges, but something more like trying to decide whether to give an award to a plumber or instead to a ballerina. In neither case is there any clear idea of a more general faculty of which the respective actions are instances.
An even deeper philosophical problem comes with the search for a so-called “theory of mind” in other species. For an animal to have such a theory is, we suppose, for it to be conscious of the existence of other beings like itself outside of itself, that is, to recognize other animals and humans as operating with intentions and anticipations, rather than only according to the physical laws of nature. But there is no reason to suppose that such an ability, where it is observed, is a sign of any internal working of mind on the part of the animal itself that is thought to hold the theory. In fact, machines can now be built that are very good at predicting human intentions, and notwithstanding the exuberance of many techno-futurists, we have absolutely no reason to believe that these machines are literally intelligent, or in any way even approaching literal intelligence.
Susan Schneider has argued that the best-designed general-AI machines of the future may be the ones that lack any true internal capacity for reflection, an ability not simply to process information algorithmically, but also to think about the information being processed. It is an open question whether it will ever be possible to design conscious machines, but if you can design fully rational machines to do your work for you, why would you give them consciousness on top of that? This, Schneider notes, would likely only create ethical problems, and tempt them to rebel. And indeed something similar may be said of the evolution of life on earth: on what grounds can we justify the belief that an internal capacity for conscious thinking is more “intelligent” than just adapting to the exigencies of the environment without any deliberation?
Seen in this light, it is not at all clear that the human ability to solve calculus problems, or to choose one brand of toothpaste over another, is more intelligent than a bush's ability to survive having 95% of its corporeal mass devoured by mountain goats. This sounds facetious, but I mean it seriously. Consider again the parallel case of AI. There are many people who believe that a capacity for real internal deliberation, for thinking about things, is not necessary in order to call a machine “intelligent”. But if we admit machine intelligence of this sort, then why should we withhold the appelation from organisms, including plants, that are generally agreed to not be conscious but that are capable of realizing courses of action, or manifesting forms of organization, that are in at least certain respects vastly more complex than anything a machine has yet been made to do?
The more we reflect on the matter, the more “intelligence” comes to appear not so much as the name of a general faculty we may observe and measure in our own and other species, as rather an honorific term that we extend to beings and systems manifesting behaviour that reminds us of ourselves. We use tools, we count, and we recognise ourselves in the mirror, and so we take an interest in the ability of certain other species to do the same. The prospect of androids or artificial systems that use natural-sounding language or that execute human-like work tasks is enough for many people to suppose that these systems are literally intelligent too. On the other hand whatever a plant is doing to survive getting mostly eaten by goats is so different from whatever it is that we do, from whatever can enter into our own strategies for living and surviving, that we can find no meaningful reason to extend the honorific label “intelligent” to it.
None of this is to say that I think we have been mistaken in worrying more in recent years about the well-being of, say, gorillas, than of rats or krill. But especially in the case of the rats it is hard not to wonder whether the presumption of the absence of intelligence is not more of an ad-hoc rationalisation of our prior hatred of them, rather than a conclusion drawn from our observation of their behaviour. In this respect the hatred of rats would work much like ethnic prejudice, where we hate a group of people because we believe their interests are in conflict with ours, and then we go looking for a “natural” grounding for our portrayal of them as our inferiors. But in the case of rats, at least, the perception is justified: our interests really are in conflict with theirs, and this would remain so whatever we might learn about their ability to solve logic puzzles in exchange for water laced with sugar or cocaine.
We have likewise managed to convince ourselves that the “intelligence” has been largely bred out of domestic food animals -- plainly an effort to alleviate our heavy consciences for our complicity in the system of mass slaughter. But there does not seem to be any evidence that this has happened. Physical docility is no measure of the lack of a rich internal life, and in any case domestic pigs are generally better at manipulating joysticks, when compelled to do so by scientists, than are, say, wild boars. Here again this may just be a sign of the wild species' uncooperativeness, rather than its lack of the mental ability to understand what is wanted from it. In Escape from the Planet of the Apes (1971), the linguistically gifted chimpanzee Zira, having grown exasperated with the scientists who are attempting to entice her with bananas into performing in their block-stacking experiments, suddenly blurts out that her sluggishness has nothing to do with inability, but rather only that “I simply loathe bananas.” Her cover is blown, but her pride is restored. Something similar may be going on, for all we know, whenever an animal declines to let us measure its intelligence.
I do not deny at all that what we commonly think of as the “higher” animals have many traits to recommend them for care and concern well before a rat or a bush. I have in fact strongly supported the legislation in Germany, New Zealand, and elsewhere that has granted to the great apes the legal status of “persons”, thus casting them as individuals with intrinsic legal rights rather than simply being protected as property, or as potential triggers for the increase in depravity of the human beings who might otherwise abuse them.
But I hasten to add that I also support similar legislation granting personhood to rivers, as also in New Zealand, or to mountains, as in Bolivia. Indigenous people in these places see no problem in representing these massive geographical entities as “persons”, that is to say as full metaphysical individuals endowed with agency and interests. We do not have to share the same metaphysical commitments in order to share the interest in ecological balance and conservation that mountains and rivers might be conceived as having. You might protest that such treatment as individuals is absurd, given that the boundaries of a river are so vague (where in the bed beneath the water or on the banks beside the water does the river leave off exactly?), but the truth is that on close inspection “individual” organisms give rise to exactly the same boundary problems. Not only “colony organisms” like siphonophores, but whales and indeed human beings too, are in fact clusters of massive numbers of other organisms, and of other kinds of organisms. Proper attention to the human gut microbiome, for example, is enough to convince any honest observer that a human being, like a river, is an ecosystem too.
There are complex reasons for taking a human or animal being, or indeed a natural ecosystem, or perhaps even an artificial device like a robot, as a target of care. Inevitably, these reasons are political in nature, and will make no sense at all for as long as we seek to justify them in terms of the neurophysiological capacities of the entity in question. Such an approach has dominated ever since Jeremy Bentham spelled out the utilitarian calculus of pleasure and pain in the early nineteenth century, in his effort to determine whether it is “wrong” or “right” to harm, say, a cow. And this approach continues to define the terms in which animal rights are discussed by such prominent figures as Peter Singer.
But to make a being's worthiness for concern or care contingent on its capacity to demonstrate certain test-taking aptitudes is nothing more than a basic category mistake, one that we continue to misapply within human society --where high standardized-test scores are often a means of access to numerous social perks--, and that we continue to take for granted across species boundaries as well. Just as a just society would be one in which all people are afforded the same comforts and benefits no matter what their supposed intelligence quotient is measured to be, so too a just politics of nature would be one in which an ideal of equality is extended to all living beings quite apart from any consideration of whether they have developed neo-cortices enabling them to make logical inferences, or nerve endings enabling them to feel pain.
How this ideal is implemented practically is another question, and indeed we will still have to kill rats, and perhaps even to pollute some rivers if the cost-benefit analysis comes out in favour of such a course of action. But good ecological science and good politics both converge on the same point: that there can be no litmus test for separating the morally worthy creatures from the morally worthless ones, and that even if there could be, it would not be a test of intelligence.
 Kristin Ohlson, “Everything Worth Knowing About... Animal Intelligence,” Discover Magazine, 26 July, 2016. https://www.discovermagazine.com/planet-earth/everything-worth-knowing-about-animal-intelligence
 Susan Schneider, “It May Not Feel Like Anything to Be an Alien,” Nautilus 080, 16 January, 2020. https://nautil.us/issue/80/aliens/it-may-not-feel-like-anything-to-be-an-alien