Mind of the Raven: Investigations and Adventures With Wolf-Birds

by Bernd Heinrich

  Some months later, the manuscript was again rejected. A reviewer claimed that I had made a “clear dichotomy between learning and genetic programming—a dichotomy that is twenty years out of date.” Of course, he was partially correct! There is no clear dichotomy. But I had tried to create one. That was the experiment. About twenty years earlier, I had published several papers on learning and innate behavior in bees, making a point to show the relationship between genetic programming and learning. That a reviewer would now suppose I’d think that there is a “clear dichotomy” in real life seemed odd. The point of this experiment was that I managed to force a small chink in the armor, to drive a wedge into a mechanism, much as an experimental physiologist might ligate a blood vessel to find out what organ it might supply. The point of the experiment was that the effects of both genetic programming and learning could be minimized, to see if any behavior remained. I felt I was being taken to task for the study’s strength rather than its weaknesses. Yet another reviewer used such phrases as “incredible leap of faith” and “matters of the heart.” I felt that these, and several other best-not-repeated comments, were more emotional than rational. Had I violated a taboo? It might seem I suffered from attachments to my birds, thus reading human motives into them. Perhaps. But I was instead mostly wondering the opposite: what unconscious drives would move someone to reject that which they find new and unfamiliar, a very conspicuous raven behavior. Ultimately, the paper was rejected five times. Part of the price of doing what you feel is really rewarding and novel is sometimes the necessity to endure harsh criticism.

  Several years after my paper was published, I got another group of six ravens and repeated the experiments, but that time in greater detail and when the birds were only nine to ten months old. I built a new aviary in which I could insert an opaque partition to separate two sections, so I could test each bird in isolation and counter the criticism of “social learning.” Since string shyness had been a big problem, I had also habituated the birds to string by tying several strands tautly between branches and onto the vertical cage walls so the birds could see string but not still pull and step. The results were essentially identical, except that five of the six birds in isolation pulled up meat much more quickly, all within four to eight minutes after contacting it. However, they first tried several alternate methods, including pecking, yanking, and twisting the string, before doing the pull-up. That is, the younger birds were overtly experimental. As in the first group, one bird only flew at the meat and never pulled string.

  In the future, when I have another group of birds, I will give them the task of trying to access food that is suspended below them but that they can get only by pulling down on a string that is above the perch. I predict this counterintuitive task will not be performed by naive ravens without lengthy learning trials, if at all. Those who already pull string up learn this trick quickly, as expected in what psychologists call transfer learning.

  Any phenomenon can potentially be explained by several alternative hypotheses at the same time. The scientist then seeks to disprove each one. If all the likely alternatives are disproved but one, that one is generally considered to be the most likely answer—until new data come along that provide a better explanation. The currently accepted answer ultimately depends on the alternatives with which one begins. If insight is denied as a possibility from the very beginning, then it can never become an alternative hypothesis, and hence it can never be the best hypothesis that remains.

  It is hardly to be expected that the human animal would be qualitatively different from all others. The psychologists who have studied learning in rats and pigeons have assumed (and found) similarities across species. If that is anthropomorphizing, I’m all for it. There is no evidence to suggest that humans have some new or different mysterious vital essence that other animals lack. Indeed, the raison d’être for studying animals is the unspoken assumption that results can be extrapolated to humans. Otherwise, the agencies that award research grants would not have spent untold millions of dollars on rats.

  At the most fundamental level, learning, consciousness, insight, and all such correlates as problem-solving and intelligence, are simply the firing of neurons. Neurons are components of intelligence and insight, but you can’t probe specific neurons and say, “There it is. Insight!!” You can no more critically define insight by examining neurons than you can discover the structure of the Maine coastline by examining the grains of sand on the beaches with ever finer detail. The relevant patterns can sometimes best be seen by stepping back and looking from a new, unfamiliar vantage point.

  I often see “intelligence” in my ravens in the stupid things they do. One time early in November 1992, I surprised a group of them in a fir thicket. They were noisy and raucous around a long-dry cow scapula. Gathering around a dry bone is “goofy.” A chickadee wouldn’t do it, or a blue jay, or a crow. These birds would not be so foolish as ravens. But then, few species of birds except ravens (and some parrots) would end up pecking airplane wings, pulling off windshield wipers, swiping golf balls, and not incidentally, getting and opening food from Dunkin’ Donut dumpsters, or sealed black garbage bags, or from a string, or sliding on their bellies in the snow, or doing barrel-rolls in the air when returning alone to the roost at night.

  Doing foolish things like stealing windshield wipers and dancing around a dry cow scapula is, like play, one of the costs of being bright. It’s a little like the “intelligence” of the immune system. Our immune system produces thousands of different kinds of molecules that most of the time do nothing useful. It may seem like a huge waste to produce them at all. If by chance, one of these odd, seemingly senseless molecules neutralizes a specific unanticipated invading pathogen, then this one is recognized and remembered by the body, to be replicated in huge numbers. That is, the body “learns” through selection. Neural networks work in the same way, but they pretty much have to be present all the time, barring some exceptions. Those that are used or rewarded are activated and strengthened, taking preference over others. From our own experience, we all know that we can try out what “works” within our minds even before we try it out physically. When we want to reach an apple over our heads, we can evaluate, by mental projection of our limbs, the feasibility of trying to reach it by hand. Or we can try to reach it by jumping up and grabbing, getting a chair to stand on, bringing a ladder, swinging a stick, calling the fire department, throwing rocks at it, hurling sticks, shooting the branch off with a shotgun. There are endless possibilities that we evaluate and discard in milliseconds. We may quickly come upon one that rewards us mentally, and when we do, we continue to run the scenario through our mind before actually trying it. If we had a raven’s mind? We’d be forced to try more possibilities overtly, and we’d have a great deal fewer and less elaborate possibilities to choose from.

  This raven had been caching food in snow, and is carrying food to cache in its throat pouch.

  TWENTY-SEVEN

  Brains and Brain Volume

  ALL VERTEBRATE ANIMAL BRAINS consist of fore-, mid-, and hindbrains. These divisions have different functions, with the hind- and midbrains responsible mainly for integrating and processing sensory information and organizing movement and attention. The forebrain is the locus of conscious activity, playing important roles in sensation, learning, memory, and mood. There is relatively little variation among species in hind- and midbrains, but fore-brains vary greatly, and in such animals with large brains as humans, it is the forebrain that accounts for the large brain size.

  In general, the greater the average brain volume of a species, the more information the animals can handle. Large animals require bigger overall brain size than small animals simply to control their bodies; and in general, brain size increases proportionally to body mass or volume. If brain size is greater than what would be predicted by body size alone, that is called the “residual” factor, and is a measure of the brains’ “encephalization.” Humans are some of the most encephalized animals in the
world, second only to some species of dolphins.

  Some birds also have high cephalization. In the 1940s, the Swiss zoologist Adolphe Portman compiled data on brain volume in birds, reporting that the corvids as a whole, which include ravens, crows, jays, magpies, and nutcrackers, had one of the highest encephalization indexes, scoring a 15. The raven scored a 19, the highest member of the corvid group and therefore the highest of any bird. All other passerine, or perching, birds ranged from 4 to 8.

  1A Bluejay. 1B Raven. 1C American Crow.

  The potential information-processing power of the brain is presumably related to the number of units, or neurons, it contains, and the complexity of their interactions. Brain volume is closely correlated with the number of neurons, and the complexity of interconnections of neurons is independent of brain size and of species. Encephalization is thus probably a fairly objective measure of behavioral flexibility. We intuitively infer that intelligence is correlated to brain size, and this inference is generally supported by a variety of criteria. It is also true, however, that we can’t credibly claim that one species is more intelligent than another unless we specify intelligent with respect to what, since each animal lives in a different world of its own sensory inputs and decoding mechanisms of those inputs.

  Primates live largely in a visual world, and in general have large areas of the brain devoted to processing visual information. Humans, in contrast to other primates, additionally have large brain areas committed to auditory processing, speech, and language. The large fore-brains of some dolphins and killer and sperm whales are thought to be devoted to echolocation. Echolocation alone, however, does not explain their huge brains, because bats and some other whales and dolphins echolocate superbly with very small brains. Birds’ encephalization could be necessary to coordinate flight, yet such insects as dragonflies manage exquisite flight (and walking) coordination of their four wings (and six legs) operating independently all with a brain smaller than a pinhead. Why would a raven need a large brain to coordinate two wings?

  Brain tissue is metabolically as active and hence as expensive as muscle, and it is active day and night. Our brain accounts for only about 1.5 percent of our body weight, but it demands about 20 percent of our energy supply. At any one time, this energy is used mainly by those neurons that are active, and we can determine the regions of the brain where neural activity is concentrated using a modern technique called PET (positron emission tomography) imaging. PET scans provide pictures of the regions of the brain where neural activity is most intense, moment by moment. For example, different areas light up when we hear, see, speak, or generate words. When we see an object, a specific area of our brain indicates neural activity. When we later think of that object, the same area lights up. That suggests to me that thinking involves in some way the same or some of the same neurons that are involved in processing and storing incoming information, indicating a suspiciously close link to memory.

  Animals have evolved to minimize energy use whenever possible. Large, metabolically expensive brains would only have developed and be maintained for very compelling reasons. As already mentioned, we can infer that sensory processing and motor coordination alone do not explain why dolphins, humans, and some birds have such large brains. One suggestion that neurobiologists have made is that the often limited stimulus load that the animal accepts from the environment is considerably less demanding, in terms of number of neurons required, than what is done with the stimuli—how large stores of memories are projected and manipulated. Different animals in effect not only see different worlds (because they have different sensors and different sensitivities), they also handle the incoming information in different ways, to create different worlds in their heads. For example, a bat and a dolphin both live in a world where pressure waves and vibrations of air and water, respectively, are highly important to their survival. A bat uses pressure waves as information to plot an interception with flying insects. A dolphin, however, could to use them not only to intercept prey but also to plot the ocean floor, to navigate over thousands of miles, to distinguish individuals in a herd, and maybe even to discern the moods of other dolphins (a popular theory), and to track individual dolphins for mutual interactions.

  An animal may extract enormous amounts of information from the environment, organize it, and give it meaning. Before the animal encounters the patterns of vibrations or pressure waves, for example, there is no sound. The animal’s sensors detect these vibrations or pressure waves, making them stimuli; the brain then interprets these stimuli to perceive them as sounds, and to manipulate or sort them to create “stories” and scenarios out of them. Since the brain creates or specifies the animal’s unique world, it is difficult to apply the same intelligence tests across species. Perhaps the only objective criterion is brain volume itself.

  In June 1988, on a canoe trip down the Noatak River in Alaska, my companions and I found the remains of a dead raven behind a trapper’s cabin, the only human structure we saw along four hundred miles of river. At the time, I saved the bird’s skull as a curiosity, or perhaps as a memento of the trip. Later that year, one of my five tame ravens unexpectedly died. I had nicknamed this bird the “cretin” because I had the subjective impression it was incredibly dumb relative to the other birds. The cretin’s skull showed an injury, possibly a peck, which could have resulted in abnormal development, and possibly death. I saved the skull, and on a whim compared it with the Alaskan raven’s skull. Both skulls were almost identical in length, but the Alaskan’s had a strikingly larger and more rounded brain case. I next weighed a known volume of sugar and filled the brain cavities with sugar through the foramen magnum, then weighed the sugar to determine brain volume. The Noatak raven’s brain volume was 18 cubic centimeters, while the cretin’s was only 11.8 cc. I had never before seen two similarly sized skulls of the same species with such an enormous difference in brain volume. With numbers like that, you immediately wonder whether the cretin’s brain was abnormally small, or the other’s abnormally large.

  Pursuing the possibilities, I called my friend and ornithological colleague Fran James at Florida State University in Tallahassee, who steered me to Phil Angle of the Smithsonian Institution in Washington, D.C., Phil loaned me a boxful of raven skulls from their collections. I measured brain volumes as before, determining that my Noatak raven skull was close in volume to the others from Alaska. That is, it was not abnormally large. The Alaska ravens, with an average volume of 17 cc, had a higher brain volume than the ravens from the western United States, with mean volume 13.1 cc (see Table 27.1). Brain volume of the Maine ravens overlapped both, with a mean of 15.5 cc. Alaskan ravens are larger than western ravens, and perhaps their larger brain volumes can be attributed to their larger body mass.

  The northern raven’s absolute brain volume of about 17 cc is twice that of an American crow and nine times that of the common or rock pigeon. Both crow and pigeon weigh about 400 grams versus the raven’s 1,200 to 1,400 grams. In contrast, a domestic chicken (Rhode Island Red) weighed for comparison had a pea-sized brain volume of 3.1 cc, even though its body weighed twice that of a raven (see Table 27.2). These numbers reinforce general prejudices that pigeons and chickens don’t come close to ravens in intelligence, but crows probably do.

  Yet brain mass in any one individual is not constant. Recent research with some birds shows that the mass of the hippocampus, the portion of the brain devoted to the specific functions of singing and food caching, increases and decreases seasonally with use. Birds can grow and shrink brain tissue as needed, thereby avoiding expensive maintenance of tissues not in use.

  There has been much discussion of why some animals have large brains. Most of this discussion has centered on human brain evolution, but the same ideas probably apply generally. One thing stands out: The large brain of hominids appeared rather suddenly. Furthermore, it did not evolve uniformly in all of the hominids. For millions of years, our ancestors, such as Australopithecus afarensis, walked bipedally and had an esse
ntially modern human form, but had a brain volume just marginally larger than a chimpanzee’s. For millions of years, that smaller brain sufficed. Suddenly (in evolutionary terms), a small brain wasn’t good enough anymore for one small group of ancestral hominids. Something changed for them alone. What was it? The one correlation we have is that as hominids became meat-eaters, they became larger in body size, and brainier. The others who remained largely vegetarian remained small-brained. Is this change in diet a clue?

  The diet connection is strong, but interpretations of it differ. The prominent anthropological argument acknowledges the diet connection and attempts to explain it by saying that meat from large animals—large amounts of high-powered concentrated food—was needed to power that metabolically costly brain, so we turned to scavenging and hunting. I think that particular explanation is backwards, because it assumes a large brain is a good thing. It isn’t, necessarily. Both ravens and hawks are meat-eaters, but hawks are small-brained, and they are very effective and successful predators. Diet alone therefore does not explain raven brain evolution. Another explanation is that a protein diet enabled an expensive brain, and that some strong selective pressure, such as sociality, then drove the evolution of increased encephalization.

  Much recent research in mammals has converged to indicate that perhaps the major driving force behind the evolution of increased brain size is social complexity. In turn, social complexity increases inordinately when individual recognition becomes possible and the animal tracks not just others, but myriad specific others. Ravens, like other corvids, and like dolphins and most primates, are highly social. As I have indicated (Chapter 14), they apparently recognize one another. Furthermore, not only do nonbreeding subadult ravens form coalitions against breeding adults, but adults may cooperate in pairs and perhaps in coalitions of pairs (Chapters 9 and 10). Ravens also are exposed to interactions with dangerous carnivores as they attempt to get meat from them (Chapters 19, 20, and 21). All of these interactions require instant reactions or choices that can be made much more quickly and safely in the head, rather than overtly. In short, they may require consciousness, the ability to examine, evaluate, and make mental choices before committing to action.