Blog Dec 06 2016
This is how you communicate with a fellow intelligence: you hurt it, and keep on hurting it, until you can distinguish the speech from the screams.
—Blindsight
Believe it or not, the above quote was inspired by some real-world research on language and dolphins.
Admittedly the real-life inspiration was somewhat less grotesque: scientists taught a couple of dolphins how to respond to a certain stimulus (if you see a red circle, push the yellow button with your nose—that sort of thing) then put them in different tanks but still let them talk to each other. Show the stimulus to one, but put the response panel in the tank with the other. Let them talk. If the dolphin in the response tank goes and pushes the correct button, you can conclude that the two of them communicated that information vocally: you can infer the presence of language. What’s more, you’ve recorded the vocalizations that carried that information, so you’ve made a start at understanding said language.
As I recall, the scientists rewarded correct responses with fish snacks. Blindsight’s scientists didn’t know how to reward their captive aliens—they didn’t even know what the damn things ate—so they used a stick instead of a carrot, zapped the scramblers with painful microwaves for incorrect responses. It was more dramatic, and more in keeping with the angsty nihilism of the overall story. But the principle was the same: ask one being the question, let the other being answer it, analyze the information they exchanged to let them do that.
I’ve forgotten whether those dolphins ever passed the talk test. I’m guessing they did—a failed experiment would hardly make the cut for a show about The Incredible Smartness of Dolphins—but then, wouldn’t we have made more progress by now? Wouldn’t we at least know the dolphin words for “red circle” and “yellow button” and (given that we are talking about dolphins) “casual indiscriminate sex”? Why is it that—while dolphins seem able to learn a fair number of our words—we’ve so far failed to learn a single one of theirs?
I was always a sucker for the dolphins-as-fellow-sapients shtick. It’s what got me into marine mammalogy in the first place. One of my very first stories—written way back in high school—concerned a scientist living in an increasingly fundamentalist society, fighting funding cuts and social hostility over his attempts to crack the dolphin language because the very concept of a nonhuman intelligence was considered sacrilegious. (In the end he does get shut down, his quest to talk to the dolphins a complete failure—but the last scene shows his dolphins in a tank quietly conversing in their own language. They’ve decided to keep their smarts to themselves, you see. They know when they’re ahead.)
Anyway. I watched all those Nova documentaries, devoured all the neurological arguments for dolphin intelligence (Tursiops brains are 20% larger than ours! Their neocortices are more intricately folded, have greater surface area!). I read John Lilly’s books, embraced his claims that dolphins had a “digital language,” followed his Navy-funded experiments in which people sloshed around immersed to the waist in special human-dolphin habitats.
By the time I started my M.Sc. (on harbor porpoises—one of the bottlenosed dolphin’s stupider cousins), I’d grown significantly more skeptical. Decades of research had failed to yield any breakthroughs. Lilly had gone completely off the rails, seemed to be spending all his time dropping acid in isolation tanks and claiming that aliens from “Galactic Coincidence Control” were throwing car accidents at him. Even science fiction was cooling to the idea; those few books still featuring sapient dolphins (Foster’s Cachalot, Brin’s Uplift series) presented them as artificially enhanced, not the natural-born geniuses we’d once assumed.
We still knew cetaceans were damn smart, make no mistake. Certain killer whale foraging strategies are acts of tactical genius; dolphins successfully grasp the rudiments of language when taught. Then again, so do sea lions—and the fact that you can be taught to use a tool in captivity does not mean that your species has already invented that tool on their own. The expanded area of the dolphin neocortex didn’t look quite so superhuman when you factored in the fact that neuron density was lower than in us talking apes.
So I passed through grad school disabused of the notion that dolphins were our intellectual siblings in the sea. They were smart, but not that smart. They could learn language, but they didn’t have one. And while I continued to believe that we smug bastards routinely underestimate the cognitive capacities of other species, I grudgingly accepted that we were still probably the smartest game in town. It was a drag—especially considering how goddamned stupid we seem to be most of the time—but that was where the data pointed. (I even wrote another story1 about cetacean language—better-informed, and a lot more cynical—in which we ultimately did figure out the language of killer whales, only to discover that they were complete assholes who based their society on child slavery, and were only too willing to sell their kids to the Vancouver Aquarium if the price was right.)
But now. Now, Vyacheslav Ryabov—in the St. Petersburg Polytechnical University Journal: Physics and Mathematics—claims that dolphins have a language after all2. He says they speak in sentences of up to five words, maybe more. The pop-sci press was all over it3, and why not? I can’t be the only one who’s been waiting decades for this.
Now that it’s happened, I don’t quite believe it.
It wasn’t a controlled experiment, for one thing. No trained dolphins responding to signals that mean “ball” or “big” or “green”. Ryabov just eavesdropped on a couple of untrained dolphins—“Yana” and “Yasha”—as they chatted in a cement tank. We’re told these dolphins have lived in this tank for twenty years, and have “normal hearing.” We’re not told what “normal” is or how it’s measured, but concrete is an acoustic reflector; it’s fair to wonder how “normal” conditions really are when you take creatures whose primary sensory modality is sound, and lock them in an echo chamber for two decades.
Leaving that aside, Ryabov recorded Yana and Yasha exchanging fifty unique “noncoherent pulses” in “packs” of up to five pulses each. Each dolphin listened to the other without interruption, waiting until the other had finished speaking before responding in turn. Based on this Ryabov concludes that “most likely, each pulse . . . is a word of the dolphin’s spoken language, and a pulse pack is a sentence.” He goes on to compare these dolphin “words” with their human equivalents. (Dolphins words are much shorter than human words, for one thing—only about 0.25msec—because their wider frequency range means that all the phonemes in a “word” can be stacked on top of each other and pronounced simultaneously. Every word, no matter how long, can be spoken in the time it takes to pronounce a single syllable. Cool.)
I find this plausible. I do not find it remotely compelling. For one thing, it doesn’t pass the “tortured scrambler” test: while the pulses are structured, there’s no way of knowing what actual information—if any—is being conveyed. If Yana consistently did something whenever Yasha emitted a specific pulse sequence—say, swam to the bottom of the tank and nosed the drain—you could reasonably infer that the sequence provoked the behavior, that it was a request of some kind: Dude, do me a favor and go poke that grill. Language. But all we have here is two creatures taking turns making noises at each other, and dolphins are hardly the only creatures to do that. If you don’t believe me, head on over to YouTube and check out the talking cats.
Nobody denies that dolphins communicate. Lots of species do. Nature is full of animals who identify themselves with signature whistles, emit alarm calls that distinguish between different kinds of predator, use specific sounds to point out food sources or solicit sex. Killer whale pods have their own unique dialects. Honeybees communicate precise information about the distance, bearing, and quality of food sources by waggling their asses at each other. The world is rife with the exchange of information; but that’s not grammar, or syntax. It’s not language. The mere existence of structured pulses doesn’t suggest what Ryabov says it does.
He does buttress his point by invoking various cool things that dolphins can do: they can learn grammar if they have to, they can recognize images on TV screens (responding to a televised image of a trainer’s hand-signal the same way they would if the trainer was there in the flesh, for example). But all his examples are cadged from other studies; there’s nothing in Ryabov’s results to suggest a structured language as we understand the term, nothing to “indirectly confirm the hypothesis that each NP in the natural spoken language of the dolphin is a word with a specific meaning,” as he puts it.
It’s a tempting interpretation, I admit. This turn-by-turn exchange of sounds certainly seems like a conversation. You could even argue that a lack of correlated behaviors—the fact that Yana never did nose the drain, that neither pressed any buttons nor got any fish—suggests that if they were talking, they were talking about something that wasn’t in their immediate environment. Maybe they were talking in abstracts. You can’t prove they weren’t.
Then again, you can say all that about YouTube’s talking cats, too.
So for now at least, I have to turn my back on the claim that my life-long adolescent dream—the belief that actually shaped my career—has finally been vindicated. Maybe Ryabov’s on to something; but maybe isn’t good enough. It’s a sad corollary to the very principle of empiricism: the more you want something to be true, the less you can afford to believe it is.
But all is not lost. Take long-finned pilot whales, for example. They were never on anyone’s short list for Humanity’s Intellectual Equals—the Navy loved them for mine-sweeping, but they never got anywhere near the love that bottlenosed dolphins and killer whales soaked up—and yet, just a couple of years ago, we learned that their neocortices contain nearly twice as many neurons as ours do.
Maybe we’ve just been looking at the wrong species.
1 http://www.rifters.com/real/shorts/WattsChanner_Bulk_Food.pdf
2 http://www.sciencedirect.com/science/article/pii/S2405722316301177
3 See: <http://www.cnn.com/2016/09/13/europe/dolphin-language-conversation-research/>, <http://www.csmonitor.com/Science/2016/0912/Do-dolphins-use-language>, and <http://www.nationalgeographic.com.au/animals/dolphins-recorded-having-a-conversation-not-so-fast.aspx>
