Ignorance

by Firestein, Stuart

  Finally a young graduate student in psychology, Oscar Pfungst, set up a series of experiments that revealed the method in the cleverness—and it was surprisingly subtle. Pfungst found that Hans could perform well no matter who his human interlocutor was—van Osten, strangers, or even Pfungst himself—as long as that person knew the answer. If the questioner was in the dark, so was Hans. Hans would also fail if he couldn’t see the person—for example, if they were separated by a partition or if Hans was wearing blinders. This led Pfungst to the realization that Hans had to be getting some cue from the person, and by careful observation—of the person, not the horse—he found that people would tense the muscles of their body and face at the beginning of Hans’s answer and release the tension when he arrived at the correct hoof tap. Hans was clever alright, just not at mathematics. Hans was reading very subtle changes in posture, expression, and attitude in his human collaborators. Most remarkable, Pfungst discovered that even after he knew he was providing these cues, he could not consciously prevent himself from doing so. If he knew the answer, he would involuntarily alter his demeanor in ways that the very clever Hans could observe.

  This realization changed the course of experimental psychology, and, for that matter, any field having to do with living organisms, forever. In large-scale drug testing, patients get either the real drug or a fake one, the placebo, and the administering doctor cannot know which one the patient is getting. Otherwise he or she would unconsciously cue the patient. Even the person supplying the drug to the doctor cannot know, because the doctor will figure it out and communicate that to the patient. And all of this can, and mostly does, happen unconsciously.

  The method that scientists use to control for these Clever Hans effects, as they have come to be known, is the “double blind.” That is, either the experimenter cannot know the correct answer or she cannot be available to the subject. The experimenter herself cannot be trusted to hide the correct answer because she will give it away involuntarily. This is all well and good for drug trials, but it creates a serious problem for the cognitive researcher. The very social cues that are believed to be critical to the complex task of communication, especially linguistic communication, are those that need to be ruthlessly removed from the experiment by a double-blind procedure. But removing the social aspects of the process destroys the experiment. This is the double bind of the double blind.

  It is true that the various language experiments with chimpanzees, and the few with dolphins, were ultimately failures in the sense that researchers never taught any of them to sit down and describe what life as a chimp or a dolphin is like. But they were groundbreaking nonetheless because they revived the subject of animal cognition as a scientific question. They led to the recognition of tool use, symbolic behavior, numerosity, empathy, and even self-awareness in animals previously thought of as mere behaving machines. Beginning in the late 1980s it once again became okay to investigate what was behind the behavior of animals, and to consider that it was more than just gears and levers.

  But how precisely do you ask questions about mental activity in an animal? Reiss and Pepperberg take two different approaches in the particulars, but they are fundamentally the same strategy, based on the same guiding principle. That principle is one that is difficult for many scientists to swallow, because it relaxes control, gets the experimenter out of the driver’s seat, and leaves it up to the subjects—dolphins, elephants, and parrots in this case—to produce the results. For both Reiss and Pepperberg, the key strategic leap was to give up worrying about the definition of consciousness or self-awareness, about what the thing (if there is a thing) was and how to produce it, and instead to provide an opportunity for an individual creature to simply show us whether it acted consciously. This may seem a subtle distinction, but it is what has allowed Reiss and Pepperberg to advance. This is an example of the question, the right question, asked in the right way, rather than the accumulation of more data, that allows a field to progress. As Reiss says, “Our only chance is to get these occasional glimpses of a mind at work.” It will not reveal itself in some cagey test. These have been tried before, and in each case some Cartesian-minded behavioral psychologist has been able to show that what looks like conscious, self-driven behavior can be just as easily replicated with simple stimulus-response schemes. No need to invoke consciousness to explain some apparently complex behavior, simple reward systems can do just as well. And you won’t see consciousness in a magnetic resonance image (MRI), because you can’t find it in human MRIs. There is no seat of consciousness, no bump on the head, and no deeper brain structure. It’s an “emergent phenomenon” that appears in some creatures and not in others. Who has it and why?

  I really like this idea of a “glimpse.” Like the notion of discovering, but even more humble; it is often all that is available. Engineering them, glimpses, is the subtlest kind of experiment that one can design. Facts don’t often stand still, and they are often only perceptible to peripheral vision. It’s hard to know (i.e., predict) where they will come from and when.

  For Reiss, the key is watching patiently and giving animals opportunities to demonstrate their abilities. She waits for the occasional “glimpse” of another mind at work, hoping that it will be recognizable. This notion of stealing a “glimpse” into the question is critical to Reiss’s approach. It is almost Zen-like. Looking directly at the thing causes it to disappear, and being too active creates only what you want to happen. She works to create instead an opportunity while remaining dispassionately focused. Do you see the paradox that is the life of this scientist?

  One practical problem with this strategy is that the numbers can be small. Indeed, often the data are anecdotal, they are a story, but with a little luck, one that perhaps helps to design an experiment. Reiss has a classic example with a dolphin subject that was part of her graduate work.

  As a first step to establishing a relationship with her subject animal, a female dolphin named Circe in an aquarium in southern France, Reiss took over feeding her. Feeding allowed her to establish some training parameters that would come in handy later. The most basic of these is something called holding station—it means simply to come stay here with me and focus on me. Reiss would hold her hand out, and Circe would come from wherever in the tank she was swimming and poke her head out of the water, for which she would get a fish. More precisely, she would a get a piece of fish because she was used to having her fish cut up into three sections—heads, bodies, and tails. Circe didn’t care for tails and would reject them—effectively training Reiss to only feed her heads and bodies. Now when Circe refused to hold station or to do some other task that was being requested of her, Reiss would not only withhold the food reward she would back off about 10 feet away from the tank for 5 minutes in what is known as a “time-out.” This time-out is not much different from what teachers use on misbehaving young children—it’s a kind of punishment because it means the offender, student or dolphin, cannot do anything to make it right or to get the reward. It’s a very effective strategy that has the added advantage of being physically painless.

  One day Reiss was conducting an experimental session and Circe was working away at performing some test, when Reiss inadvertently fed her a tail—the nonpreferred part of the fish—that had slipped into the bucket. Circe spit out the tail, promptly turned away, and swam back 10 feet from the station, turned around, and, the top half of her body out of the water, watched Reiss standing there bucket in hand but nothing to do. She had given Reiss a “time-out.”

  Or had she? For Reiss it was unmistakable. But it was once. It was an anecdote. Can an experiment be designed from this? Can an anecdote be turned into an experiment? As Reiss says, “While we are trying to learn about them, they are trying to learn about us, and that may be the most interesting part of the experiment.” The situation is fluid; there are too few controllable variables. And yet every so often there is a glimpse that reveals a mind at work—unmistakable but also unquantifiable. It is these glimpses t
hat Reiss works for. (In fact, Reiss staged a few more “mistaken” feedings of tails and each time Circe punished her with the same behavioral response—a time-out).

  So the problem is how to create a moment when you get a glimpse. This is a not uncommon theme in science. Discoveries don’t come every day. Even after you settle on the question, there is still the business about how you will pursue it.

  Currently, Reiss is using mirrors to get a glimpse of what animals think about themselves—if they think about themselves. By the age of 18 months all humans get the mirror illusion (that’s what it is after all), and since the pioneering work of Gordon Gallup Jr. in the late 1970s we know that chimpanzees and other great apes also figure it out. This is one of those experiments that seems painfully obvious—in retrospect. Once done, nobody can understand why it hadn’t been done decades earlier. Gallup, wondering whether chimpanzees, given the opportunity, might show self-recognition, simply provided some chimps with mirrors and observed their behavior. There was a clear evolution of behaviors from social (as if the image in the mirror were another chimp) to contingent (doing something and seeing whether the mirror guy imitates you—the classic Lucille Ball, Harpo Marx comedy routine) to self-directed (for example, using the mirror to inspect the inside of your mouth). So with the right opportunity the chimpanzee showed us that it can figure out the mirror. But the ultimate test is whether the chimpanzee “sees” this image as itself, in the sense of understanding it is itself. For this, Gallup devised what has famously come to be known as the mark test. Lightly anesthetized chimps were marked on their foreheads with a red dye and upon awakening were provided access to a mirror. When they got around to looking in the mirror, they noticed the red mark on their foreheads, touched it, and investigated it using the mirror. Indeed, the image in the mirror was herself. They are our cognitive cousins.

  But monkeys don’t get it, nor do dogs or cats or other very smart species. Is it just higher primates? Is there something specific about primate brains? Are we special after all? Do we just have to admit chimps and gorillas into the club to keep us special? Reiss had the mind of another smart species in mind; she wondered what dolphins might do with a mirror. Why? They have big brains, about the size of ours in relation to body size, but they are in every other way completely foreign. They live in the water and move easily in three dimensions, they have no hands, they have no facial expression except that frozen Mona Lisa smile, and in a host of other small and big ways they differ from the typical land mammal and in particular from primates. They are, as Reiss jokes, extraterrestrials (at least in the narrowest sense). Their last common ancestor with primates lived 60 million years ago. If they can do the mirror thing, then it is clearly not a primate-specific mental trick.

  The results of Reiss’s work are published and you can read the paper (Proceedings of the National Academy of Sciences, 2001), but the short answer is that dolphins do recognize themselves in a mirror, going through all the classic behaviors and phases of recognition seen in humans and chimps, including finally the mark test. And while the results in that paper are very striking, the real value in this work is the many questions it has generated because it has expanded the question set from what makes primates special to what the requirements are for a brain to develop a mind. What deeper commonalities are there between self-aware species? Why has self-awareness arisen in such different species? The demonstration of mirror self-recognition in dolphins generates more questions than it answers. In fact, all it does is generate questions. What a great experiment.

  But Reiss wasn’t finished yet. Along with Frans de Waal, her colleague and noted primatologist, she collaborated on a mirror self-recognition test of elephants. Elephants weren’t chosen out of the blue, so to speak, but because they showed some behavioral characteristics that de Waal and Reiss thought might predict that they would pass the mirror test. Once the exclusivity of the mirror self-recognition club had been broken by dolphins, it was reasonable to consider other characteristics than being a primate that would indicate strong sense of self. And, indeed, these large, and large-brained, animals needed little time to go through the now predictable behaviors leading to correct mirror use—social, contingent, self-directed—once a mirror, a very large, very strong mirror, was placed in their enclosure at the Bronx Zoo in New York City. Like the dolphins, the elephants were rapidly attracted to the mirror—perhaps because it was a new thing in an otherwise not so interesting environment that they were all too familiar with. Whatever the initial attraction, the appearance of oneself in the mirror was quickly recognized and tested by these large-brained mammals.

  Reiss thinks this testing may be part of the clue to determining what makes an animal mirror-self-conscious. Remember, these experiments were designed as an opportunity for animals to use their minds in ways we could interpret. What Reiss sees is that animals that come to recognize it’s them in the mirror are already highly inquisitive. They don’t just take note of things in their environment; all animals do that. And they don’t just learn cause and effect the way Skinner’s pigeons do; all animals learn from experience. These species, the mirror-savvy ones, “test contingencies,” as she likes to say. “They actively probe their environment looking for effects from causes they instigated. They are scientists.” Many animals check out the mirror when they first see one but rapidly dismiss it as having no effect. Some animals probe more deeply. Those are the ones that interest Reiss. Those are the animal scientists.

  Now the mirror-self-recognition fraternity in Reiss’s experiments included at least two nonprimate species that also had little in common with each other. So what was the common denominator? Is there one? Are there many? Has self-awareness evolved many times independently, like color vision? Is it perhaps not as rare as we once expected? Does it really indicate self-awareness, in the same way that we feel self-aware? That is, does it extend beyond seeing a version of yourself and simply taking advantage of the new information? Why would mirror-self-recognition evolve in the first place, even in humans?

  For Reiss the mirror is a kind of glimpse machine; it provokes the mind of the animal and provides a glimpse of what may be going in there. Can we locate the part of the brain that is active during mirror use? Is it just one place, or is it distributed? Will it be something that only certain brains possess? What happens when you distort the image in the mirror—can other animals “get” the funhouse mirror metaphor? Will they still recognize themselves in a distorted view? How far into the mind of another can a mirror take us?

  Reiss wants to try the mirror test on an octopus, a very visual, surprisingly intelligent invertebrate. What a Pandora’s box of questions that could raise. “Yes, indeed,” Reiss smiles.

  Dr. Irene Pepperberg used a technically different method to work with an African Grey Parrot named Alex, but with precisely the same purpose. She taught Alex a vocabulary of some 100 words to describe a variety of objects (blocks, fabric, food, etc.) and qualities (color, texture, number, etc.) and then allowed Alex to make use of these linguistic tools to manipulate his environment. Like Reiss, Pepperberg wanted to give another brain a chance to show what it was made of, in this case a bird brain. She has recently published a very accessible book on her work with Alex, so I won’t go into detail here. Rather I want to make the single point of how her strategy courts ignorance. The purpose of the training was not to show that a bird could produce utterances that appeared to be linguistic, because then we would just have to have a lot of maddeningly circular arguments with Noam Chomsky about what constitutes “true” language. Rather, all the training was to give Alex a chance to let us glimpse his brain. Finally, the purpose was to be able to frame better questions about higher brain function. Alex was a model system for consciousness. No, he wasn’t as complex as a human, but he turned out to have more abstract mental powers than you might have expected. Alex learned to count (really, not like Clever Hans) showing a sense of numerosity that he then applied to getting more of his favorite things; Alex coined n
ew words by combining existing phrases he knew to express new thoughts. Before Alex, who would have thought you could have a model system for consciousness? By breaking down our prejudice that consciousness is something made of whole cloth that only shows up in humans, Pepperberg’s work enables us to ask more detailed questions about what consciousness is and what it is made of and when and why it shows up. A talking parrot is not news; one who thinks about what it says is. Of course, that might be true for humans, too.

  Alex died in April 2010, suddenly, at the age of 35, a devastating blow to Pepperberg’s research. This is one of the great difficulties in research of this nature, where the subject is the limiting factor. In biochemistry, for example, it is the death of the researcher that interrupts the research—but they are easy to replace. Easier than a thinking parrot for sure.

  Irene carries on with new birds, but the amount of time and the intellectual and emotional effort dedicated to Alex will never be recovered.

  The take-home message from this case history is not just that scientists design an experimental strategy based on what they don’t know, but that the truly successful strategy is one that provides them even a glimpse of what’s on the other side of their ignorance and an opportunity to see if they can’t get the question to be bigger. That’s progress.

  2. EVERYTHING = 1(TUTTO = UNO)

  Why does there have to be a unified theory of everything before physicists will be content? There is no comparable drive in chemistry or biology. Are those fields fundamentally different or perhaps not as mature as physics? Physics is probably the most remarkable success story in science in its ability to explain everything—or almost everything. And perhaps there’s the rub: there seem to be just a few bits missing, and that’s sometimes worse than having whole swaths of ignorance. Physicists know a lot about how things work when they are very small and have almost no mass—this is the quantum world; and they know about the fundamental characteristics of the very big, the cosmologically big—this is relativistic physics. But they don’t know how to connect these two universes—this is the elusive unification. Of course, it may also be that the grand unification, if achieved, will immediately reveal more parts that are suddenly inexplicable—much like establishing the atom, the “indivisible” unit of mass, almost immediately gave rise to the recognition that there were constituent parts previously unimagined. Thus, even now there are the nefarious-sounding “dark matter” and “dark energy,” unseen (in the deepest sense of being undetectable directly) but making up the bulk of the universe, and that may or may not be part of the grand unification.