Fantastic Beasts and How to Rank Them

The relative plausibility of impossible beings tells you a lot about how the mind works.
Illustration by Jon Klassen

Consider the yeti. Reputed to live in the mountainous regions of Tibet, Bhutan, and Nepal. Also known by the alias Abominable Snowman. Overgrown, in both senses: eight or ten or twelve feet tall; shaggy. Shy. Possibly a remnant of an otherwise extinct species. More possibly an elaborate hoax, or an inextinguishable hope. Closely related to the Australian Yowie, the Canadian Nuk-luk, the Missouri Momo, the Louisiana Swamp Ape, and Bigfoot. O.K., then: on a scale not of zero to ten but of, say, leprechaun to zombie, how likely do you think it is that the yeti exists?

One of the strangest things about the human mind is that it can reason about unreasonable things. It is possible, for example, to calculate the speed at which the sleigh would have to travel for Santa Claus to deliver all those gifts on Christmas Eve. It is possible to assess the ratio of a dragon’s wings to its body to determine if it could fly. And it is possible to decide that a yeti is more likely to exist than a leprechaun, even if you think that the likelihood of either of them existing is precisely zero.

In fact, it is not only possible; it is fun. Take the following list of supernatural beings:

__ Angels
__ Demons
__ Dragons
__ Pixies
__ Ghosts
__ Harpies
__ Elves
__ Mermaids
__ Loch Ness monster
__ Leviathan
__ Giants
__ Pegasus
__ Centaurs
__ Unicorns
__ Tooth fairy
__ Phoenix
__ Werewolves
__ Vampires
__ Genies
__ Zombies

Never mind, for now, whether or not you actually believe in any of these creatures. We are interested here not in whether they are real but in to what extent they seem as if they could be. Your job, accordingly, is to rank them in order of plausibility, from most likely (No. 1) to least likely (No. 20). Better still, if you are in the mood for a party game this Halloween season, try having a lot of people rank them collectively. I guarantee that this will produce a surprising amount of concord—who among us could rank the tooth fairy above the Leviathan?—as well as a huge amount of impassioned disagreement. The Loch Ness monster will turn out to have a Johnnie Cochran-level defense attorney. Good friends of yours will say withering things about mermaids.

What’s odd about this exercise is that everyone knows that “impossible” is an absolute condition. “Possible versus impossible” is not like “tall versus short.” Tall and short exist on a gradient, and when we adjudge the Empire State Building taller than LeBron James and LeBron James taller than Meryl Streep, we are reflecting facts about the world we live in. But possibility and impossibility are binary, and when we adjudge the yeti more probable than the leprechaun we aren’t reflecting facts about the world we live in; we aren’t reflecting the world we live in at all. So how, exactly, are we drawing these distinctions? And what does it say about our own wildly implausible, unmistakably real selves that we are able to do so?

In the fourth century B.C., several hundred years after the advent of harpies and some two millennia before the emergence of dementors, Aristotle sat down to do some thinking about supernatural occurrences in literature. On the whole, he was not a fan; in his Poetics, he mostly discouraged would-be fabulists from messing around with them. But he did allow that, if forced to choose, writers “should prefer a probable impossibility to an unconvincing possibility.” Better for Odysseus to return safely to Ithaca with the aid of ghosts, gods, sea nymphs, and a leather bag containing the wind than for his wife, Penelope, to get bored with waiting for him, grow interested in metalworking, and abandon domestic life for a career as a blacksmith.

As that suggests, for a possible thing to seem plausible it must be reasonably consistent with our prior experience. But what makes an impossible thing seem plausible? In a convoluted passage in the Poetics, Aristotle tells us that if an impossible thing would “necessarily” require something else to occur along with it, you should put that second thing in your story, too, because then your readers will be more likely to believe the first one. In other words, even something that is factually impossible can be logically possible, and how closely that logic is followed will affect how plausible a supernatural being seems.

There’s a reason Aristotle addressed this advice to writers and artists. Unlike most of us, they have practical motives for wondering how best to make imaginary things seem convincing, a problem that must be solved as much for “Vanity Fair” as for “A Wrinkle in Time.” Accordingly, creative types have done an unusual amount of thinking about plausible impossibility. In the seventeen-nineties, for instance, Samuel Taylor Coleridge set out to write a series of poems about “persons and characters supernatural.” To do so, he knew, he had to make the fantastical seem credible—“to procure for these shadows of imagination,” he wrote, in a soon to be famous phrase, a “willing suspension of disbelief.”

Coleridge was excellent at inducing a suspension of disbelief. That’s why we are as gripped by “The Rime of the Ancient Mariner” as the wedding guest within the poem who can’t tear himself away from the sailor’s tale—even though the tale itself is an outrageous one involving a magical albatross, a terrible curse, and a ship crewed by ghosts. Yet Coleridge was vague about explaining how exactly he did it. His only advice for making impossible things seem believable was to give them “a semblance of truth.”

“Now imagine how good that would look completely sweated through on the Twenty-third Street subway platform.”

A little more than a hundred years later, a very different kind of artist got somewhat more specific. Although Walt Disney is best remembered today for his Magic Kingdom, his chief contribution to the art of animation was not his extraordinary imagination but his extraordinary realism. “We cannot do the fantastic things, based on the real, unless we first know the real,” he once wrote, by way of explaining why, in 1929, he began driving his animators to a studio in downtown Los Angeles for night classes in life drawing. In short order, the cartoons emerging from his workshop started exhibiting a quality that we have since come to take for granted but was revolutionary at the time: all those talking mice, singing lions, dancing puppets, and marching brooms began obeying the laws of physics.

It was Disney, for instance, who introduced to the cartoon universe one of the fundamental elements of the real one: gravity. Even those of his characters who could fly could fall, and, when they did, their knees, jowls, hair, and clothes responded as our human ones do when we thump to the ground. Other laws of nature applied, too. Witches on broomsticks got buffeted by the wind. Goofy, attached by his feet to the top of a roller-coaster track and by his neck to the cars, didn’t just get longer as the ride started plunging downhill; he also got skinnier, which is to say that his volume remained constant. To Disney, these concessions to reality were crucial to achieving what he called, in an echo of Aristotle, the “plausible impossible.” Any story based on “the fantastic, the unreal, the imaginative,” he understood, needed “a foundation of fact.”

Taken together, Disney’s foundation of fact and Coleridge’s semblance of truth suggest a good starting place for any Unified Theory of the Plausibility of Supernatural Beings: the more closely such creatures hew to the real world, the more likely we are to deem them believable. But the real world is enormous, wildly heterogeneous, extraordinarily complicated, and, itself, often surpassingly strange. So if, indeed, the most plausible supernatural creatures are those which most resemble reality, the question becomes: which part?

The obvious candidate, at first glance, is the animal kingdom. Supernatural creatures are, after all, creatures, and we infer from them, or impose upon them, all kinds of biological characteristics. Like their natural counterparts, they can be organized by taxon (cervid, like the white stag; caprid, like the faun; bovine, like the Minotaur; feline, like the sphinx), or by habitat (alpine, like yetis; woodland, like satyrs; cave-dwelling, like dragons; aquatic, like mermaids). Given this tendency to situate unnatural beings in the natural world, it seems conceivable that our judgments about their plausibility might reflect how well they conform to the constraints of modern biology.

If that’s the case, our friend the yeti should rank very high on the believability scale. So, too, should giants, elves, unicorns, ogres, imps, sea monsters, and pixies. By the same token, this biological theory would deal a credibility blow to angels, demons, fairies, vampires, and werewolves, plus all those creatures assembled, as by an insane taxidermist, from the separate parts of real species: mermaids, griffins, centaurs, chimeras, sphinxes. It would also undermine the plausibility of fire-breathing dragons, there being no analogue in nature to a Zippo. In fact, biological limitations cast doubt on dragons in another way as well, since four legs plus two wings is not a naturally occurring configuration—a bummer also for harpies, griffins, gargoyles, and Pegasus.

If you couldn’t make it through that paragraph without starting to formulate an objection, you already know the first problem with this theory: it invites a lot of quibbling over what is and isn’t biologically feasible. As defenders of the supernatural will be quick to point out, many arthropods have six limbs; squids, skunks, bombardier beetles, and plenty of other real creatures spew strange things; nature sometimes contrives to recombine old animals in new ways (see the half-striped zedonk—part zebra, part donkey—or the recent emergence of the coywolf: part coyote, part wolf); and, considering the many kinds of metamorphoses exhibited by animals—tadpole to frog, caterpillar to butterfly, baby-faced to bearded—how far-fetched is it, really, for a bat to turn into a man?

Indeed, some fantastical creatures seem positively ordinary compared with the more byzantine products of four billion years of evolution. Consider the giant oarfish, a thirty-six-foot-long behemoth with a silver body, a bright-red mane, and a tendency to hang out in the ocean vertically, like a shiny piscine telephone pole. Or consider the blue glaucus, an inch-long hermaphroditic sea slug capable of killing a Portuguese man-of-war—a beast three hundred times its size—and then storing its poison for later use, including on humans.

Given so much natural extravagance, it’s not surprising that the real and the unreal are sometimes mistaken for each other. In 1735, when Carl Linnaeus organized all the species in the world into one vast taxonomy, he included a section on “Animalia Paradoxa”: creatures, common in folklore and myth or attested to by far-flung explorers, that he felt compelled to itemize yet deemed unlikely to exist. Among these were the manticore (head of a man, body of a lion, spiky tail), the lamia (head of a man, breasts of a woman, body of a scaly cow), and the Scythian lamb (like a regular lamb, except it grows out of a stalk in the ground)—but also, arrestingly, the antelope and the pelican. Conversely, a contributor to “This American Life” once recounted the experience of asking a group of strangers at a party, in all sincerity, whether unicorns were endangered or extinct. One sympathizes. Consider the giraffe. Consider the kangaroo.

On top of all this, the biological theory of plausibility also suffers from a graver problem: its predictive powers are faulty. By its logic, many creatures that we find highly believable should instead rank near the bottom of the list. Angels, for instance, are physiologically unlikely: in addition to being able to fly (fine for birds, unheard of in hominids), they manifest a particularly extreme version of the limb problem, since, per various sources, they have not just two but in some cases hundreds of wings. Demons present the same basic difficulties, as do fairies, and ghosts defy pretty much every biological principle: among other problems, they have no substance, require no sustenance, and do not decay or die. Yet given that seven out of ten Americans believe in angels, six out of ten believe in demons, and almost half believe in ghosts, it seems safe to assume that, on the scale of plausibility, such creatures outrank giants and unicorns.

So much for biology as the basis of our unified theory. But we can resolve at least some of these problems by modifying our hypothesis slightly. Perhaps we don’t care how much supernatural creatures resemble the animal kingdom in general; perhaps we only care how much they resemble us. This mirror theory of plausibility would still account for the high ranking of yetis, which, aside from not existing, are not so different from Homo sapiens. (Back in 2004, when scientists discovered an extinct species of an unusually small hominid on an island in Indonesia, a senior editor at Nature took the occasion to speculate that stories about yetis might reflect an extinct Himalayan species on the other end of the size spectrum.) The mirror theory would also explain the perceived plausibility of angels and demons, which, as presented in myth and literature, resemble exaggerated humans in our best and worst incarnations: moral giants and moral elves. And it would explain why vampires and werewolves, which should rank low on the list, what with the impossibility of radical metamorphosis, generally rank quite high. When they are not busy sprouting wings and fur, after all, such creatures look nearly indistinguishable from us.

On the other hand, this theory leads us quickly into ontological problems: are we humans more like mermaids, or more like ghosts? Worse, like the biological theory of plausibility, it fails to account for some of our intuitions about supernatural beings. Why, for example, would a centaur, which is fifty per cent human, strike us as less plausible than a unicorn, which is zero per cent human? And what are we to make of natural-born humans who are able to do supernatural things, à la Shakespeare’s Prospero, Hermione Granger, or that menace of Camelot Morgan le Fay?

This last category of being opens up a whole new can of orms. Magical creatures exist in a universe of magical powers, which themselves range wildly in probability and are not evenly distributed among the population. To understand our intuitions about plausibility, then, we need to look beyond entities to actions. For supernatural creatures, as for the rest of us, it might be that what matters most is not what we are but what we do.

What do supernatural creatures do? In many cases, not much. Somewhat strangely, not every magical being has magical powers. Some, like Santa Claus and the tooth fairy, mostly just have chores. Others merely hang around looking unusual; the yeti and Nessie just lurk; the Leviathan lurks, too, largerly; the record is mixed on giants, which in some accounts live on clouds but in most are just enormous and crabby. Wraiths only scare people, centaurs only awe people, and unicorns, aside from some healing properties in their horns, akin to the antibiotics in frog skin, only attract virgins—which, power-wise, puts them at the same level as boy bands. For these and many other supernatural creatures, their supernaturalness inheres chiefly in the fact (or the non-fact) of their existence.

Others, however, can do flatly impossible things. Fairies, by most accounts, can turn invisible, tell the future, and shape-shift. Ghosts can shrink, expand, time-travel, and walk through walls. Vampires can command the dead, summon storms, control lesser animals like bats and wolves, and—barring certain interventions with stakes or sunlight—live forever. Various other entities can, through their own powers or via potions, amulets, and spells, likewise achieve the unachievable: levitate, teleport, transmogrify, read minds, talk to animals, and, by occult means, charm, confuse, possess, haunt, hex, heal, or kill.

Like supernatural creatures, such powers can be ranked in terms of plausibility. Which seems more likely to work: Harry Potter’s apparating ability or Obi-Wan Kenobi’s Jedi mind trick? If you ask me, it’s obviously the mind trick, with its real-life analogies of charisma and hypnosis, not to mention its failure to defy any major laws of physics. On the other hand, apparating—vanishing from one place and appearing in another—strikes me as more plausible than time travel, possibly because we have many ways to move through space but only one way to move through time.

You can play this game forever, with any given set of magical powers. Controlling the elements, for instance, seems considerably harder than controlling an animal (unless, perhaps, it is a cat)—but, if you are going to try to control the elements, summoning a breeze seems easier than turning night to day. If you’re going to work magic on your own body, becoming invisible seems more plausible than transmogrifying, perhaps because of the abundance of everyday ways to conceal ourselves. Yet, if transmogrification is going to occur, I’d wager that it is easier to turn oneself into a wolf than one’s enemy into a toad.

As it happens, intuitions like these are broadly shared—a fact we know because, speaking of implausible things, two cognitive scientists at the Massachusetts Institute of Technology have shown it. Normally, Tomer Ullman studies our commonsense beliefs about physics and psychology, while his colleague John McCoy studies judgment and decision-making. Together, however, they figured that looking at how we reason about supernatural powers might shed light on how we reason about the real world. To that end, in 2015 they asked two hundred people, ranging in age from eighteen to eighty-three, to rank ten magic spells in order of difficulty. Since amphibians in magic have roughly the same status as rodents in science, all the spells featured things a sorcerer could do to a frog: conjure it into existence, conjure it out of existence, teleport it, levitate it, change its color, double its size, turn it into two frogs, turn it into a mouse, turn it to stone, and turn it invisible.

The results help explain why I am dubious about apparating. Over all, the subjects felt that spells were more difficult when they violated “more fundamental principles of intuitive physics.” What makes a principle of physics fundamental, in this case, is how early in our cognitive development we acquire it. For instance, we are born with an understanding of object permanence, and the two spells that violated it, by conjuring a frog into or out of existence, were ranked the most difficult. Similarly, we learn in infancy that objects have what developmental psychologists call “kind-identity”—they stay themselves—which may explain why the next-hardest spell involved turning a frog into a mouse. The two easiest spells, by comparison, entailed changing a frog’s color and levitating it, results that reflect our awareness that both color and location are transient rather than fixed features of the physical world.

To further plumb our intuitions about supernatural powers, McCoy and Ullman ran a second study, which asked the same questions but changed one of two things: either the target of the spell (Is it harder to conjure a frog or a cow?) or the extent of its power (Is it harder to levitate a frog one foot or a hundred feet?). Resoundingly: a cow; a hundred feet. These findings are striking, since levitating something ninety-nine extra feet does not violate any additional principles of physics. Nor does conjuring a cow instead of a frog. So why would those variants seem more challenging?

“I haven’t always been a cowboy, you know.”

Happily, two other cognitive scientists, Andrew Shtulman and Caitlin Morgan, of Occidental College, have addressed that question. (Full disclosure: my sister, a cognitive scientist at M.I.T., was Shtulman’s postdoctoral adviser, and has worked with McCoy and Ullman.) Last year, Shtulman and Morgan gave people pairs of magic spells and asked them to determine which one in each pair was more difficult. In every pair, both spells violated the same fundamental principle of physics, biology, or psychology, but each varied in how much it violated a secondary one. For instance, physics dictates that you can’t walk through anything solid, no matter what it’s made of, but also that materials differ with respect to properties like density and hardness. So which seems more difficult: walking through a wall made of stone or a wall made of wood?

Overwhelmingly, the subjects chose stone. They also determined that it would be harder to levitate a bowling ball than a basketball, and harder to grow an eye than a toe. Since levitation is categorically impossible, it shouldn’t matter that heavier objects, like bowling balls and cows, are harder to lift. But, as Aristotle understood, it does. According to Shtulman and Morgan, that’s because our understanding of causation—our sense of which things make other things happen—is not a series of separate if-then statements but a vast interconnected web, which continues to govern our intuitions even when one particular strand snaps. “Severing one link in a causal network,” they write, “still leaves the rest of the network intact.” And the more links you sever, the more powerful—or, put differently, the less probable—your magic seems.

Perhaps, then, the solution we seek is mathematical: tally up all the fundamental principles violated by a supernatural creature and its powers and—voilà, we’ll know where it stands in the hierarchy of likelihood. Call this the parsimony theory of plausibility: the fewer laws something violates, the more credible it will seem. The yeti, for instance, doesn’t really violate any natural laws at all. Vampires, by contrast, violate everything from the fact that things of substance cast shadows to Meteorology 101.

This parsimony theory is simple, elegant, and, unfortunately, wrong. If it were correct, we’d all find gnomes, whose only distinguishing characteristics are diminutiveness, avarice, and a preference for living underground, considerably more plausible than ghosts. Yet ghosts, despite their utter disregard for biology and physics, persist in seeming highly believable. Part of that might be explained by our existential condition: most of us feel that we have a core self, separate and separable from our body, and most of us find it hard to accept that we will someday cease to exist.

Part of it, however, might be explained by one final theory of supernatural plausibility. Consider a defense that my sister once mounted on behalf of the likelihood of fairies. Small impossible things, she contended, are more believable than large impossible things, because they could more easily exist without us noticing them. That argument isn’t based on our beliefs about physics or biology; it’s based on epistemology. From infancy on, we are extraordinarily sensitive to patterns of evidence (in fact, that’s how we acquire many of our beliefs about physics and biology), so it seems reasonable to think that evidence also determines our judgments about fantastical beings.

Of course, it also seems unreasonable to think that, since it’s unclear how we would find evidence for the existence of nonexistent creatures. In its absence, we can make do, as my sister did, with a good reason for why we haven’t found it, a strategy that lends plausibility not only to fairies in their tininess but also to ghosts and other creatures capable of vanishing. (It also gives us a reason, finally, to object to the yeti: if it existed, we should have found proof by now.)

Alternatively, we can accept attestation as a form of evidence—which, across domains, we do all the time, since many of our convictions about the world concern things we ourselves will never observe. Our sensitivity to attestation explains why culture has such a potent influence on our intuitions about the supernatural, which wouldn’t be the case if those intuitions were governed chiefly by biology or physics. It is why one community is more likely to believe in fairies and another in zombies, and why, with churches peddling a more palliative version of Christianity, demons have declined in plausibility vis-à-vis angels. And it is why, if you’re European-American, you’re more likely to believe in a vampire than in the coffin-dwelling, night-roaming, life-force-sucking Chinese jiangshi, even though, on the basis of their characteristics, there is not, so to speak, a lot of daylight between them.

Patterns of evidence, a grasp of biology, theories of physics: as it turns out, we need all of these to account for our intuitions about supernatural beings, just as we need all of them to explain any other complex cultural phenomenon, from a tennis match to a bar fight to a bluegrass band. That might seem like a lot of intellectual firepower for parsing the distinctions between fairies and mermaids, but the ability to think about nonexistent things isn’t just handy for playing parlor games on Halloween. It is utterly fundamental to who we are. Studying that ability helps us learn about ourselves; exercising it helps us learn about the world. A three-year-old talking about an imaginary friend can illuminate the workings of the human mind. A thirty-year-old conducting a thought experiment about twins, one of whom is launched into space at birth and one of whom remains behind, can illuminate the workings of the universe. As for those of us who are no longer toddlers and will never be Einstein: we use our ability to think about things that aren’t real all the time, in ways both everyday and momentous. It is what we are doing when we watch movies, write novels, weigh two different job offers, consider whether to have children.

As that last example suggests, perhaps the most extraordinary thing about this ability is that we can use it to nudge the impossible into the realm of the real. We stare at the sky, watch a seagull bob on a thermal, build wax wings and then fixed wings and then Apollo XI. We dream of black Presidents and female scientists; we dream, still, of self-driving cars, a cure for cancer, peace in the Middle East. These last things are interestingly like dragons and also interestingly unlike dragons, in ways that suggest that we may be wise, after all, to treat impossibility as something other than an absolute condition. Alone among all the creatures in the world, we can think about fantastical things and, at least some of the time, bring them into being.

Yet, in the end, what’s most remarkable is not that our fantasies contain so much reality; it is that our reality contains so much fantasy. Most of us understand that our perceptual systems, far from passively reflecting the world around us, actively sort, select, distort, ignore, and alter a huge amount of information in order to construct reality as we experience it. But reality as we experience it also departs from actual reality in deeper ways. In actual reality, space and time are inseparable, and neither one behaves anything like the way we perceive it; nor does light, and nor does gravity, and, in all likelihood, nor does consciousness. Yet all the while we go on experiencing space like a map we can walk on, time like a conveyor belt we travel on, ourselves as brimming with agency, our lives as mattering urgently.

That world, the one we inhabit every day of our lives, is a yeti—a fantastical thing constructed out of bits and pieces of reality plus the magic wand of the mind. If we could hand it over to some superior being for consideration, it might not even rank very high on the scale of plausibility. Then again, plausibility itself might not rank very high on the scale of qualities we prize. Better, perhaps, to know that what we feel in our happiest moments has some truth to it: life is magical. ♦