"This Week in Dognitive Science: Kanzi, Marc Hauser & Charlie Rose."
If dogs had a language it would be first person plural.
Originally published in slightly different form on August 13, 2010 at PsychologyToday.com.
What Animals Think?
It’s been an interesting week in dognitive science. Time Magazine’s cover story, “What Animals Think,” proclaimed, “Science is revealing just how smart other species can be.” Time’s senior science editor, former attorney Jeffrey Kluger, provides evidence (most of which has been questioned or even debunked by others in the field) showing that some animals are capable of grasping the fundamentals of human language (that’s controversial), others are using tools (that’s true), and that still others (baboons and pigeons) are supposedly capable of understanding abstract thinking about relations between relations (also not true). Meanwhile, in another part of the cognitive-science galaxy, far, far away, one of its leading lights, Marc Hauser, head of Harvard’s new Canine Cognition Lab, has taken a year-long leave of absence while under investigation for “scientific misconduct.” One of his key papers, on cognition in tamarin monkeys, has been withdrawn due to discrepancies between what’s stated in the study and what’s reportedly missing from a videotape. Several other of his papers are under review. The Harvard Crimson: “An article published in 2002 in the journal Cognition that suggested tamarin monkeys learned rules as human infants did, after an investigation by a Harvard committee concluded that his conclusions were not supported by his data. “Hauser’s lab has been under investigation for the past three years, and Hauser described the inquiry as ‘painful’ in a letter to his colleagues.” It’s not clear yet on what this means for the Canine Cognition Lab. Presumably, they’ll continue their work, which is still in its incipient stages. Interestingly, Hauser’s was the only voice of moderation in the Time Magazine article. “Hauser,” Kluger writes, “takes a more nuanced view [about the differences between human and animal cognition], arguing that people are possessed with what he calls humaniqueness, a suite of cognitive skills including the ability to recombine information to gain new understanding, a talent animals simply don’t have. All creatures may exist on a developmental continuum, [Hauser] argues, but the gap between us and the second-place finishers is so big it shows we truly are something special. Animals have a myopic intelligence, but they never experience the ah-hah moment a 2-year-old child gets.” I could be wrong, but I’m taking a wait-and-see attitude. I don’t like to think that Hauser has been deliberately dishonest.
Kanzi the Bonobo Who Reportedly Understands Human Language Charlie Rose had Time Magazine’s Jeffrey Kluger on his show this week to promote Time’s cover story on the new “science” on animal cognition. And Kluger brought along a videotape of Kanzi, the famous bonobo, who reportedly understands human language, and who supposedly once gave his handler, Sue Savage-Rumbaugh, the signs for fire and marshmallow, then built a fire, lit it with a match, and toasted the marshmallows. (If you click on the link and look at the video you'll notice that it’s heavily edited, possibly to give viewers the impression that Kanzi is doing all this without any coaching; I’m not saying he was coached, just that the editing seems a bit suspicious.) Meanwhile, back on the Charlie Rose video, Savage-Rumbaugh is seen wearing a welder’s mask to ensure that Kanzi can’t read any facial expressions that might tip him off to what she wants him to do. She sits on the floor across from Kanzi and asks him to perform a number of action tasks, each one involving 2 nouns, a verb, and a preposition. This was done to demonstrate how well Kanzi understands human language. And in each case, except one, Kanzi follows her instructions almost to the letter. Here’s my own transcript of the final segment of that tape, starting at (00:53) (The YouTube video I’ve linked to is a bit longer than the one seen on Charlie Rose.) “Kanzi, are you listening?” Kanzi shows no response. “Could you put some soap in the water?” Kanzi looks around at a number of objects on the floor. He sees the bottle of dish soap, picks it up, and squeezes it into a pitcher of water. “Good job. Thank you, thank you. Can you listen again?” Kanzi seems distracted, shakes his head slightly. Savage-Rumbaugh waits, says “Okay...” in a soft voice, then: “Could you pour a little Coke in the water? Could you pour some coke in the water?” Kanzi reaches for a can of Coke and begins pouring it into a pitcher of water. Savage-Rumbaugh reaches out her hand to stop him from pouring too much. “Thank you, that’s enough. You can … okay...” (There’s a jump cut at this point, indicating some missing footage.) Now here’s where it gets interesting. At (01:31), Savage-Rumbaugh says, “Kanzi, pour the Perrier water...” and then pauses ever-so slightly. During that pause, Kanzi reaches for a jelly jar, not the bottle of Perrier. Then Savage-Rumbaugh finishes her sentence, “...in the jelly.” It’s only then that Kanzi picks up the bottle of Perrier and pours it into the jelly jar that he’s already holding in his hand.
Oops! Kanzi Wasn’t Thinking After All! That’s where the video shown on Charlie Rose ended. Rose gave his patented smile of sheer amazement at Kanzi’s ability to understand human language. Neither he nor Kluger seemed to notice that during his last task Kanzi had acted out of sequence with Savage-Rumbaugh’s words. If he’d only been responding to her words, he would’ve reached for the Perrier bottle first, then waited to hear where he was supposed to pour it. But he didn’t. He reached for the jelly jar before hearing that he was supposed to pour the Perrier into it. Does this mean that Savage-Rumbaugh “cheated?” It’s hard to say. That pause seemed like a genuine break in her normal speech patterns, not part of a Clever Hans-like parlor trick. Does this mean Kanzi really isn’t able to attach word-like “meanings’ to objects and actions? Not necessarily. What it does mean is that—in this one instance, at least—something other than a linguistic ability was probably at play; the behavior had either been “rehearsed” so many times that Kanzi automatically reached for the jelly jar, or else it shows that animals can pick up mental images from us telepathically, something I wrote about in a recent article here. Other than sheer coincidence, there is no other explanation. Or is there?
When Two People “Click“
A recent study from Princeton showed that “when two people ‘click’ in conversation, they actually show startling similarities in their brainwave patterns. An article in New Scientist reports, “The [fMRI] scans showed that the listeners’ brain patterns tracked those of the storyteller almost exactly, though they trailed 1 to 3 seconds behind. But in some listeners, brain patterns even preceded those of the storyteller.” So was Kanzi, in effect, finishing his teacher’s sentence? Were his brain waves in-synch with hers, and is that why he reached for the jelly jar first? If so, is this an indicator that something like telepathy really was at work? It’s fun to speculate, but ultimately I don’t think it’s very significant compared to the fact that no one seems to have noticed this simple, yet glaring, glitch in Kanzi’s demonstration. And I think this gives us a hint at how cognitive scientists might be missing the big picture by not picking up on the smaller, telling details, even when they’re caught on videotape.
Looking Through the Wrong Lens Last year, when I heard that both Marc Hauser and Brian Hare were starting their own dognitive science labs (Hare’s is at Duke), I was intrigued, but not overly optimistic. Most dognitive scientists, with a few exceptions like Monique Udell and Alexandra Horowitz, don’t really get dogs. Many of them, including Hauser and Hare, started out studying apes and monkeys, and I think they tend to see dogs through the same lens. (In fact I’ve gotten reports from two dog trainers I know who’ve had a chance to witness the work going on at Harvard, and they’ve both indicated that the researchers are missing out on some key behaviors that a real dog person would pick up on immediately.) In my last article here I critiqued a recent study, “Domesticated dogs (Canis familiaris) react to what others can and cannot hear,” and one of my complaints was that these researchers don’t seem to understand how dogs really think. If they did, they would have changed how the study was set up, and may have come to some very different conclusions as a result. The truth is, dogs think quite differently than apes and monkeys do. In fact, there’s such a significant difference that it would be like finding a remote tribe whose only pronouns were all second personal plural—us, we, our, ours—and trying to understand their psychology as if this were just an odd evolutionary defect or regional idiosyncrasy, and not a clue to the fact that they might process information as a collective, rather than as separate individuals. I think dognitive scientists would do well to hire an old-time SchutzHund master, or a veteran police dog handler, or a sheep man or cattle rancher, or anyone else with a longstanding working knowledge of how dogs really think, and run the ideas for future studies by such a person before setting up their experiments. Because when a working dog is working, there is simply no scale that can measure his intelligence. It’s a force of nature. And remember: the true language of dogs is almost always spoken in the first person plural: “Let’s eat!” “Let’s go for a walk!” “Let’s play!”