Sex on Six Legs: Lessons on Life, Love, and Language from the Insect World - Marlene Zuk (2011)

Chapter 1. If You're So Smart, Why Aren't You Rich?

Learning on Six Legs

THE FAMOUS eighteenth-century naturalist Jean-Henri Fabre meticulously examined the mason bees of his native France, marveling at the tiny clay cells they constructed as cradles for their helpless larvae. When the young bee is ready to emerge, under the normal scheme of things it scissors its way out of the clay with its mandibles and squeezes through the opening. But Fabre, like calculating scientists before and after him, tested the ability of the bees to think by examining how they were able to overcome various manipulations of their chambers. First, he thwarted a bee by removing part of the clay and replacing it with a piece of paper. Undaunted, it sliced through the paper with the same motions it used for the thicker material. Then Fabre presented the young bee with not one, but two barriers: the usual clay, and paper a half inch in front of it, so that the insect needed to repeat the motion it had performed on the original cell on the paper. This it was unable to do, instead tapping fruitlessly at the paper it was completely capable of cutting through. Two barriers are never found in nature, and the bee couldn't perform acts outside its repertoire; we now know that it lacks the kind of neurological GPS ("if a new roadblock appears, repeat steps A through G until you see air") necessary to adapt to altered circumstances. Fabre tut-tutted over the bee's ineptitude, noting, "The insect would have to repeat the act which it has just accomplished, the act which it is not intended to perform more than once in its life; it would, in short, have to make into a double act that which by nature is a single one; and the insect cannot do this, for the sole reason that it has not the wish to. The Mason-bee perishes for lack of the smallest gleam of intelligence."

Later scientists were equally condescending, noting with belittling superiority that although quite a few kinds of insects can perform remarkable tasks, they cannot learn from experience the way we humans can. In the late nineteenth century, the English physician David Douglas Cunningham was posted to the Indian Medical Service in Calcutta, where in addition to studying the pathology of infectious diseases he made detailed observations of the local flora and fauna, including the many large and easily observed insects. He was prepared to admit that some of the large wasps that provisioned their young with paralyzed caterpillars and other prey possessed something along the lines of what he termed intellect, given their complex behavior. But he was also fond of performing "practical jests" on the wasps. The females built mud nests on many objects, including the pipes in his study, and Lieutenant Colonel Cunningham enjoyed occasionally moving the pipe a foot or two from its original location while a wasp was out foraging for prey. He noted that it was then "amusing to observe the astonishment of its tenant when she returns to find her nest gone, and wanders round in perplexity until it is replaced and joyfully recognized." One could certainly wonder about how hard up for amusement one has to be before taking up playing jokes on wasps, but regardless, the same note of self-satisfaction creeps into Cunningham's writing that is seen in the writings of most of the early naturalists. Not being able to find something after it was moved, or being unable to recognize a novel feature in the environment must mean that insects, regardless of their awe-inspiring abilities to construct elaborate hives and find flowers miles away, are dimwitted at heart.

But in fact, it is turning out that here too our faith in our uniqueness may be misplaced, and that insects are capable of feats of intelligence that qualitatively, at least, may be quite similar to our own. This finding has many useful implications, from the construction of better computers and robots to a potential cure for brain damage. And it also challenges our ideas about what our own enormous brains might be for.

Six-Legged Smarts

THE LIKELIEST candidates for insect intelligence, or at least the first ones to be considered by naturalists, have always been the bees, wasps, and ants. Partly this is because we see them more—in our gardens and kitchens—and they seem to be doing things, such as finding food and taking it back to their nest or hive, that require something resembling reasoning. Partly it is because of the sociability of many species, since we use our own intelligence to interact with each other so much. And partly, I think, it has something to do with the way that such insects use objects in their environment, whether it is to build paper cells from chewed wood pulp or to remove pollen from flowers and cram it into the built-in shopping bags on a bee's leg. Animals that have possessions seem smarter, somehow, which may be a comment on our own valuing of material goods.

Fabre, Cunningham, and a host of other naturalists paid particular attention to the provisioning wasps and bees. These relatives of yellow jackets and honeybees do not live in social groups with a queen and workers. Instead, once she has mated, a single female searches for prey such as caterpillars or large toothsome spiders. After capturing the item, she stings it so that it is paralyzed but not dead, a kind of suspended animation refrigeration system. She lugs her victim to her nest, which may be a burrow in the soil or a custom-built cell on the surface of an object, as with the pipe-loving butts of Cunningham's "jests," and lays an egg on it. After the egg hatches, the young larva has a ready food supply that won't spoil. Depending on the species, the mother may return many times to add prey to supplement the larder or to lay more eggs in additional chambers.

While grisly in certain respects, the wasp's behavior undeniably requires two of the prerequisites for intelligence: learning and memory. The mother wasp has to remember where her burrow is, find the correct size and number of prey—in one species, the number of food items brought back to the nest is calibrated to the needs of the hungry waiting larvae—and go back to the correct place. All of this cannot be done purely by rote, because each nest is built anew, each cell provisioned separately, and each prey item puts up a different fight. The wasps seem to use landmarks to find their nests, like remembering where one's house is by recalling the location of the Starbucks at the corner, and if the landmarks are moved, the wasps fly around the area, like the agitated subjects of the jokes played by Cunningham. In their defense, incidentally, one wonders how most of us would do if we suddenly found the aforementioned coffee shop lifted in its entirety off of the block, between one latte and the next, and whether we too wouldn't mill around the area, unable to believe our eyes.

Even more impressive than the ability of these wasps is that of another species of wasp that exploits the provisioning kind. These parasites do not care about the wasp larvae waiting for their paralyzed meal, but about the caterpillars that are brought to the larder. Instead of going out and hunting down their own prey, the parasitic wasps capitalize on the food brought in by the hunters and lay their own eggs on the item. The problem is that only a very narrow window of opportunity to lay an egg on the caterpillar exists, which is during the time that the caterpillar is being dragged into the nest by the wasp that first captured it. So instead of trusting to luck to find a host at exactly the right moment, the parasitic wasp performs a reconnaissance mission, flying around areas where the provisioning wasps are likely to be digging their nests, an activity that takes quite a long time and is much more apparent than the brief provisioning period. Once a nest-building wasp is detected, the parasitic wasp remembers where the nest is located and keeps that nest site under surveillance, so that she can spot when provisioning occurs, often many days later. Then she slips in and hurriedly lays her own eggs on the caterpillar.

Yet another species of parasitic wasp lays its eggs on clusters of checkerspot butterfly eggs. The catch here is that the eggs can be successfully parasitized only for the few hours when the checkerspot babies have developed into first-stage larvae but have not yet broken out of the egg. The wasp circumvents this difficulty by learning where the eggs are ahead of time and then monitoring their progress until they are ready, with some individual wasps finding an egg cluster and then revisiting it for up to three weeks, a substantial portion of the wasp's lifetime.

The wasps and their relatives among the other social insects are not the only ones that can learn new things. The caterpillars and butterflies the wasps use as prey are also capable of learning, and they can also develop preferences for particular foods, depending on the type of plant on which their mother laid her eggs. Such food snobbery is of more than academic interest, since some pest caterpillars that eat crops, for example, the young of the familiar cabbage white butterfly, can learn to eat new varieties of cruciferous vegetables; planting broccoli in hopes of evading butterflies that grew up eating cauliflower is futile. Interestingly, not all kinds of butterflies can learn to go to one kind of plant rather than another; checkerspots, eastern swallowtails, and a species of Heliconius butterfly all seem to be relative dullards. You can rear them on one kind of plant, but if you try to train them to visit another kind when it's time to lay eggs, the mother butterflies just won't make the switch. Perhaps it's not stupidity so much as brand loyalty, like refusing to accept Pepsi instead of Coke even if the former is on sale.

Parents often swear that their children are born picky eaters, and that they cannot be taught to prefer healthy snacks. But grasshoppers and their relatives the locusts can be taught to determine the nutritional content of different plants and feed preferentially on the most nourishing ones. In the laboratory, grasshoppers can be fed little cubes of synthetic diet, kind of like the power gels consumed by marathon runners, and the contents of the cubes varied according to the experiment. In one study, groups of locusts were given food lacking either protein or digestible carbohydrates. The experimenters gave one food in a yellow tube and one in a green tube, alternating the association between subjects, and then let the insects feed on a balanced diet for a few days to make sure they didn't become malnourished. Then, the locusts were deprived of food for four hours, a rather long time between meals for the insects, which usually eat more or less nonstop. When the locusts were placed in a test chamber containing yellow and green tubes, but no food, they went to the color associated with the nutrient—either protein or carbs—they had been lacking. This feat is particularly impressive because it isn't just the grasshoppers having some holistic instinct for eating what is good for them, but a learned association between color and a nutritional deficit. Toddlers, take note. Admittedly, the researchers didn't try offering the insects a choice between the hopper equivalent of Twinkies and that of tofu, but then I am not sure quite how one would go about determining what insect junk food would be like.

Honeybees have long been known to navigate using landmarks and use information from each other to find food, as I discuss below and in another chapter, but a recently discovered ability deserves special mention: they can count. The ability to enumerate objects is considered one of those gold standards of intelligence by scientists, and several kinds of primates, some other mammals such as dolphins and dogs, and psychologist Irene Pepperberg's late African gray parrot, Alex, have been shown to do so. Still, you just don't think about insects in the same breath as you do arithmetic. But scientists Marie Dacke and Mandyam Srinivasan of the Australian National University in Canberra trained the bees to fly down a tunnel toward a food reward, using landmarks set along the walls and floor. To get to the food, the bees couldn't simply memorize the position of the landmarks, because the locations of the landmarks were shifted every 5 minutes. Instead, the bees had to learn that the food could be found at the base of landmark number 1, 2, 3, 4, or 5, depending on the individual experiment. Counting to four was mastered relatively easily, but getting to five proved challenging. Nonetheless, that the bees could generalize to a number at all, rather than simply flying until they saw an object in the same place it had been before, is an extraordinary accomplishment.

The bees' ability is exciting not only because it helps demolish that boundary of the backbone with regard to intelligence, but because being forced to design the experiments required to demonstrate counting in a creature so different from us makes us strip down our methods to their essentials. Finding out if your three-year-old can count is one thing. But how do you come up with a test for counting, or learning in general, when your subjects can't talk, walk on two legs, point to anything, or even get rewarded with something they want, the way most people can? If we can design ways to study animals with these limitations, maybe it will help us work more effectively to test humans with limited abilities, or even design computer programs that could substitute for the abilities that are lacking.

Figuring out exactly how to test insect intelligence in a way that is meaningful to them but also tells us something is challenging. Reuven Dukas, a biologist at McMaster University in Canada, has studied learning in a wide variety of insects and thinks we may only be scratching the surface of their abilities. After all, if insects don't learn something, he says, echoing teachers everywhere, "Is it because I'm not a good teacher or because the animal doesn't learn?" It's always hard to know what tasks an animal will be able to perform that we can then generalize to other species. Jan Wessnitzer and colleagues from the University of Edinburgh showed that my favorite insects, crickets, could relocate a particular spot on the floor using objects in a photo along the wall of their experimental arena as landmarks—the best navigational aid was a rather stark landscape that looked like a desert in the American Southwest. The training scheme they used consisted of a floor heated to an uncomfortable temperature except for a single cooler spot that the crickets presumably preferred to stand on. It was called, without comment, the Tennessee Williams paradigm.

The Face Is Familiar, but What about the Antennae?

LEARNING about food sources is one thing, since it is a natural behavior on the part of many insects, perhaps particularly honeybees. But scientists are now demonstrating that insects can be taught far more sophisticated tasks, sometimes having no apparent relation to their day-to-day requirements. Recently, for example, Shaowu Zhang and his colleagues trained honeybees to be extraordinarily discriminating in their decision making. The bees were given a reward if they chose a particular pattern on a card, but the "right" choice depended on whether it was morning or afternoon, whether the bees were out visiting flowers or returning to the hive, or a combination of both. The bees took a while to learn their task, but they eventually could make the distinctions, an impressive cognitive feat. Bees from laboratories in both Australia and Germany were tested, and in a happy blow for global diplomacy, turned out to be roughly equal at the task.

But remembering to choose one visual cue over another pales in comparison to another bee achievement: bees can learn to recognize individual human faces. Adrian Dyer at La Trobe University in Melbourne, Australia, and his colleagues there and at Cambridge University in England rewarded honeybees with a sip of sugar solution if they flew toward a particular image, a technique that has frequently been used by researchers. What was novel was the kind of image in his experiments: a black-and-white photograph of a man from a stock collection, compared with a photo of a different person, the same face upside down, and a drawing. Not all the bees got it right, but those that did could remember an individual face several days after their initial training. Dyer isn't suggesting that the bees actually "know" what they are looking at, or that they spend their days scrutinizing the people around them or developing an attachment to the beekeeper. They can't possibly undergo the same cognitive processes that we do when we recognize each other, given their limited nervous systems. Instead, Dyer believes that the ability is probably related to their skill at distinguishing one flower from another while foraging, something more useful in a bee's life. In other words, a bee that can tell a columbine from a daisy could use the same technique to tell a Roman-nosed individual from a snub-nosed one. Dyer went on to demonstrate that honeybees could discriminate among photographs of very similar natural scenes, with images of forests that differed only in the orientation of the branches, an ability that probably makes returning to the hive after a long foraging flight easier to accomplish.

Regardless of how or why they do it, the bees' capacity to learn to recognize human faces has some important implications. Facial recognition has always been one of those skills thought to require a large brain, and psychologists had even speculated that a special part of the human brain is devoted to just that task. But bees don't have any of the same brain components that humans and other vertebrates do, so such a specialized structure must not be necessary to accomplish the discrimination. As Mandyam Srinivasan said, "Sometimes I wonder what we are doing with two-kilogram brains."

In addition to further blurring those boundaries between human and insect, there are some practical uses for the discovery. Computerized facial recognition would be a boon to security and crime-fighting agencies, and studying the mechanisms behind the bees' ability might yield insights into how to create such programs. I was seized by the image of a chamber with a bee at airport security, for instance, scrutinizing the faces of passengers to look for matches with photos of known terrorists. Whether this would work better than some of the current efforts is an interesting question.

Some humans themselves cannot distinguish among human faces, a condition known as prosopagnosia, or face blindness, thought to be due to a genetic defect; one estimate claims that 2.5 percent of the population suffers from some form of it. Some people with prosopagnosia can distinguish individual animals, but not people; Jane Goodall is said to have this form of the disorder. Prosopagnosia can also be present to greater or lesser degrees, so that one can have the disorder under certain circumstances but not others. In severe cases, sufferers cannot recognize their own face in a photograph. It seems to be related to the inability to navigate in the environment, which means that bees might be particularly suitable for using as models for studying the disorder, since of course bees are superstars at locating food sources and remembering nest sites. At the moment, no one has worked out the mechanisms by which the bees learn faces, but if they are linked to the ways in which the bees orient in the wild, understanding the bees' abilities could help people overcome their own face blindness.

Each Ant, Teach Ant

WE—AND other animals—can learn things from objects in the world around us, like Dyer's bees or the wasps that remember the location of a rock near their nest. But most of us remember learning in school, from teachers. Insects may lack classrooms and textbooks, but increasing evidence suggests that they too can learn from, and act as, teachers.

In common use, the word teaching usually means the transfer of information from one individual to another. A boy sees his sister feed the dog under the table and promptly learns to get rid of his unwanted broccoli the same way. Under that definition, though, even casual observation of another animal doing something that the observer then does would qualify. You could learn to run away from fires by noting a crowd fleeing a burning building, for instance, but has the crowd actually taught you? Even Charles Darwin suggested that many animals, including insects, do this; bees, he pointed out, could follow another worker flying to a source of nectar. If crickets are placed in a container with other crickets that have been hiding under leaves from predatory spiders, they are more likely to find a shelter and hide themselves. But this kind of use of public information seems a bit too haphazard to be real teaching. Animal psychologists are more stringent in their definition and often require the behavior to happen only when a naive observer (one that doesn't know how to do the task being taught) is present. That means that although a young male white-crowned sparrow learns his song from his father, the father isn't teaching him, because the adult bird would sing whether or not his son were there. Teaching also has to help the observer while costing the teacher something, usually the time and effort required for the demonstration.

Finding an occurrence of this more narrowly defined behavior in nature has been daunting, and until very recently scientists had essentially no examples of real teaching by animals. Just within the last few years, however, researchers have found three cases of it—one in a bird, one in a mammal (the meerkat), and one, in credibly, in ants. People are often surprised by the selectivity of this group, suggesting that surely some other primate besides humans teaches in a natural setting. At least for the moment, the answer appears to be no, which says something about our anthropocentric desire to only see, or bestow, special qualities on those we think are closest to us. That teaching happens in ants and not monkeys or apes is unsettling for the same reason I love studying insects: it's all about getting to the same destination with different modes of transportation.

As anyone who has had to battle the brown ribbons of workers heading toward the sugar bowl knows, ants follow each other to get to food sources. It looks like they are just marching endlessly, one after the other, perhaps following the smell left behind by earlier foragers, but paying no more attention to each other than riders on the same subway train. Odor does play an important role in leading ants to food. But in at least one ant species, a single worker will actively recruit another ant to follow her to a food source or a new nest, or just to explore a new area, in a process called tandem running. The lead ant goes in front, while the follower keeps contact by tapping her with her antennae. If the follower gets behind, the leader waits for her to catch up, and spends time on the task that wouldn't be needed if the leader were alone, fulfilling the criteria outlined above. According to Ellouise Leadbeater and her Queen Mary University of London colleagues, who didn't do the research but study similar kinds of insect social behavior, "The intimate interaction between leader and follower in a pair of tandemly running ants at first sight bears all the hallmarks of a parent teaching a child to ride a bicycle." After being led, the following ant is able to find the target on her own, showing that she has indeed learned from the leader.

This is big news. As an accomplishment it may not rank with conveying the beauty of Shakespeare to a high school senior, but it means that even ants can respond to feedback from other individuals and modify their behavior so that they improve their performance. Feedback makes teaching different from so-called telling, where in effect one individual says, "Hey, there's a puddle of jam over in the north corner of the countertop, see you there," and then just takes off for the food. This behavior has therefore made scientists question how they define learning, teaching, and their prerequisites. Some researchers feel that because the ants don't improve the skills of those they teach, but simply lead their students along a path, the behavior doesn't really constitute teaching. But in a paper with the subtitle "Ants Are Sensitive Teachers," Thomas Richardson, who led the original project on tandem running, and his colleagues at the University of Bristol in the United Kingdom muse that the arguments over whether the ants are "really" teaching may just be "tracking our own understanding of what is special when humans teach.... We should thereby avoid succumbing to the understandable temptation to use the most exotic, extreme case, i.e., the human one, to define what is perhaps a relatively common phenomenon." In other words, once we find that ants do something like teaching, we should not redefine teaching so only humans can be said to do it. And if ants do teach, what other animals might be showing the same thing, if we only open our minds to see it?

Smarter Is as Smarter Does

THE GENIUS of ants notwithstanding, if the basic components of learning and even intelligence lie within a great many creatures, why then are our minds so different? Why do we talk about crows and raccoons and dolphins being intelligent, but chickens and cows as dumb? Is being smarter always better? And if it is, why haven't all animals evolved to be smarter?

The answers to these questions come from an unlikely source: the humble fruit fly. Now, I can usually sell people on crickets, and ladybugs, ants, and bees already get their own movies, toys, and children's songs. People are less than enthusiastic, though, about the possibility of a sparkling intellect lurking in the sesame seed-sized flies that buzz in clouds around decaying fruit. But in Tad Kawecki's laboratory at the University of Fribourg in Switzerland, fruit flies are contestants in an unending game of Jeopardy, insect style. And some of them are big winners.

The flies don't learn How the West Was Won or Celebrity Children, but they do have to master a category that might be called Distinctive Odors, by deciding whether to feed and then lay their eggs on a substance that smells like orange or one that smells like pineapple. One of the two offerings is infused with quinine, which tastes bitter, and the flies avoid that odor and fly over to the other area. Once the quinine is removed, some of the flies still remember to stay away from the place that had the nasty taste, showing they have truly learned the association. Then Kawecki takes the eggs that were laid in the tasty stuff, rears the adults that emerge, and repeats the whole experiment again and again. This means that only the genes from the flies that performed the discrimination correctly are passed on to the next generation. It's the same principle—artificial selection—that farmers have used for centuries to generate cows that give a lot of milk or corn that has large ears, but much faster and with an end product of faster-learning flies rather than county fair material.

Doing these experiments requires painstaking maintenance of the tiny flies in hundreds of jars held under exactly identical conditions—the same temperature, the same food, and in complete darkness. Most modern biology buildings have elaborate facilities for keeping the insects, but of course many scientists labor in less-than-ideal circumstances. As it happened, Kawecki used to work at the University of Basel, also in Switzerland, where his lab was in a crumbling fifteenth-century building in which the doctoral students used a former lecture hall of Friedrich Nietzsche for their office. Although charmingly located on the banks of the Rhine River and architecturally impressive, the building suffered from a variety of maintenance ills, many of which required the service people to enter the attic. The attic in turn was occupied by numerous pigeons and swifts, and one of the building maintenance workers complained so vociferously about the birds' lice and fleas he supposedly encountered in his effort to repair things that an exterminator was called in. While most people welcome the removal of insects from their homes, in a building where precious experimental flies are being kept, the situation is somewhat different. Kawecki and his colleagues made numerous panicky phone calls to the exterminators to make sure the process wouldn't decimate their subjects, and were assured that all would be well.

Unfortunately, as Kawecki puts it, "the only animals [the exterminator] knew about were cats and budgerigars," and the insecticide proved fatal to some of the carefully reared fruit flies. Luckily, the scientists had to stagger the breeding of the flies because they didn't have enough room to raise them all at once, so they did not lose all of their years of effort. But Kawecki remains nettled at the company, which never admitted any wrongdoing, instead suggesting "it was our fault, keeping those stupid flies rather than cats and budgerigars, as proper Swiss citizens do."

Despite these setbacks, one generation of flies led to another. Through the selective breeding process, the flies rapidly improved their ability to remember which substance was attractive and which was not, and after about twenty generations, Kawecki had flies that could go to the bug equivalent of Harvard or Princeton. Instead of taking three hours to learn which substance has quinine in it, the new and improved flies knocked the task out in less than an hour. What's more, they could generalize their ability to other tasks that required them to avoid or prefer one odor to another, and even to other stimuli besides odor, which means that the flies were not simply evolving better discrimination of pineapple versus orange, they were actually getting smarter.

Presumably, being able to detect good places to feed and lay eggs faster would also be useful in the real world, outside Kawecki's lab. So why don't flies show this brainiac capacity naturally? To put it another way, if the flies can get to be so smart, why aren't they rich, or at least more successful?

The answer seems to be that they don't live long enough. The life span of flies from the smarter lines averaged 15 percent shorter than their unselected relatives. Furthermore, the smarter females laid fewer eggs, an ominous characteristic from the standpoint of evolution, since it means fewer potential copies of genes in future generations. The decreased survival was particularly notable when food was in short supply, which gives a clue to the reason for the finding: learning is costly, and investing brain resources into intelligence may mean that you pay the price somewhere else. More brain, fewer eggs. The trade-off even occurs within the lifetime of a single fly. A group of flies that was trained to associate an odor with a mechanical shock and then deprived of food and water died 4 hours earlier than flies that were exposed to the smell or the shock but didn't have to go through the training, suggesting that something about the process of remembering the association drained the resources of the diligent flies.

Such trade-offs are common among living things, as I discuss in the chapter on personality. Animals that have many young also tend to have smaller babies, whereas species like us that give birth to one or a few offspring at a time generally produce relatively large ones. Here the trade-off seems to be that when natural selection gives a good learner, it takes away a long life. This could happen at two time scales. Within the lifetime of the fly, the energy a fly acquires could go either to helping it survive longer, or to nervous system machinery, but not both. It may be cheap to upgrade the memory in your laptop, but doing so in the brain is going to cost you.

Over many generations, a different process may be at work. Say that a gene makes a fly smart, but because most genes have more than one effect, it also makes the fly vulnerable to starvation, or maybe more susceptible to infections. If being smart is advantageous enough—in Kawecki's lab, it made the difference between reproducing or not—then the gene conferring it will persist in the population, even if it also has some downsides.

Of course, it's not as if all animals get to go to some primordial retail smorgasbord and shop for a certain number of abilities, with some picking learning, long legs, and a mean tennis serve, while others choose curly eyelashes and a talent for languages but end up dimwitted. Exactly which abilities end up having to trade off against which others is still a mystery. But Kawecki's work suggests that the ability to learn, and hence perhaps intelligence, exacts a high price. And that in turn could shed some light on our own evolution. Humans may well have given up some other abilities when we evolved our large brains. What's more, having to learn everything from infancy, rather than being born with our skills, makes our childhoods vulnerable to everything from hot stoves to saber-toothed tigers or their modern-day equivalents. The trade-off in our case must have been substantial, but scientists are still wondering about exactly what it was that we humans had to pay for our intellect.

Better Learning through Chemistry

ONE OF the wonderful things about using animals such as fruit flies and other insects to study learning is that they present a window into the brain. Exactly what happens in the body when you learn the capital of Mongolia, or how to get to the theater? We all have some vague idea that nerve cells send messages somewhere, that electrical impulses in the brain do ... something. And we can use complicated brain scans with colorful images of different centers of nerve activity, or detailed dissections, to try and figure out what that might be. But insects, unlike humans, let us alter a chemical here, or breed up offspring with a special mutation there, which means it is sometimes possible to pinpoint precisely what makes an individual able to perform a certain task. If one bug has gene variant A, and another bug's genes are exactly like it except for having gene variant B, and if the two differ in the time it takes them to find a food reward in a maze, then presto, we have a gene linked to learning.

In most cases, those carefully bred and engineered insects are fruit flies. In the chapter on personality I mention the "rover" and "sitter" flies, which exhibit genetically programmed differences in behavior. Kawecki and his colleagues, most notably Frederic Mery, examined these tendencies in light of their studies on learning. Each behavior is associated with a form of a single gene, and flies with the rover variant of the rover-sitter gene have better short-term, but worse long-term memory, something of a reversed Alzheimer's, where it is easier for sufferers to recall events of decades past than what they had for lunch. Sitters show the opposite pattern and can remember associations from several days ago, but not the fly equivalent of what they ate at an earlier meal. These different strengths and weaknesses make sense in the natural world of a fly; rovers are likely to move from one food source to another, so being able to quickly learn whether a given fruit is ripe or not is more important than remembering what happened in the more distant past. Together with Marla Sokolowski from the University of Toronto, who first discovered the rover-sitter dichotomy and has worked on its details for many years, the scientists then discovered that the differences in memory can be manipulated by increasing or decreasing the amount of an enzyme in the odor detection centers of the insects' brains. That enzyme may be the key to the trade-off between memory types, at least in flies, and suggests some interesting directions for similar studies in people.

Another set of experiments focused on a different chemical. Using a modified version of the Tennessee Williams paradigm, in which flies are placed into a chamber that heats up on one side when the flies move to it, a group of researchers from the University of Missouri recently demonstrated that serotonin, the same brain chemical that features so prominently in human depression and its treatment, is key to the tiny flies being able to learn to avoid the hot spots.

The ability to stick to a task after having been distracted—something many children with learning disabilities struggle to accomplish—is also controlled by a few nerve cells and chemicals. Flies tend to move toward a visual object, for example, a stripe on the end of their container. If you remove the goal and show them a "distracter" stripe somewhere else, they veer off for a short time but can still remember where the original stripe was located. Geneticists have bred flies with mutations in various genes that produce chemicals important in learning (as with many specialized strains of fruit flies, these have fanciful names such as dunce and ignorant), and it turns out that while some of the mutants can still perform the task of recalling their goal as well as normal flies, others cannot. The mutant amnesiac, for example, learns just fine in the first place, but forgets what it learned almost immediately. This distressing tendency can be attributed to a defect in a single neurochemical, one that is extremely similar to a chemical in the human nervous system. Being able to break down a behavior such as recovering after a distraction into components so fine that we can determine exactly which gene is responsible for which part of learning is possible only in insects, at least so far, but maybe someday we will be able to extend this kind of detailed understanding to our own learning difficulties. What's more, the prospect of altering or curing defects in memory with gene therapy in insects suggests that similar treatments may eventually substitute for drugs or surgery in humans, a solution that could have fewer side effects and be targeted more precisely than current approaches.

He Who Learns Last

WHICH came first, learning or instinct? Because humans rely so heavily on learning, we tend to think of it as an innovation, an evolutionary novelty that we alone have mastered. In effect, we like to think we invented invention. But centuries ago, naturalists believed that instincts, behaviors that are performed more or less the same way every time, arose after learning. Early animals, they claimed, had to learn things from scratch, and then after time, the repetition of a task was somehow impressed into the fiber of the organism so that eventually it became instinctive. This idea was particularly championed by Jean-Baptiste Lamarck, the French biologist whose ideas about the inheritance of acquired characteristics were first embraced by early evolutionists, including Charles Darwin, but later discredited. Knowing nothing about how genes and chromosomes could be passed from parents to offspring, Lamarck and his contemporaries reasoned that if, say, a horselike animal continually reached for its food at the top of a tree, its neck would become longer. This greater development would then somehow be passed onto its offspring, who in turn would develop even longer necks, eventually resulting in what we now call a giraffe.

Lamarck was quite interested in invertebrates, a subject not much studied by naturalists in the late eighteenth and early nineteenth centuries, and he was intrigued by the idea that they exhibited such fixed behaviors. It seemed reasonable to him and many of his contemporaries that doing something, whether following an odor trail or learning to count, could cause a permanent change in the body, and that such changes could be inherited. Of course, we now know that the genes cannot be influenced in exactly the way that Lamarck imagined. And it is likely that both learned and instinctive behaviors evolved together. Most behaviors, even in insects, are due to a combination of influences from the environment, and hence subject to learning, and influences from the genes, and hence instinctive, making the old argument somewhat moot.

Dan Papaj, a biologist at the University of Arizona, doesn't believe Lamarck himself was correct, but he does wonder if there aren't new ways in which learned behavior could influence evolution. He works with a variety of species, from butterflies to parasitic wasps, to see just how learning operates in nature. He points out that the idea that fixed behaviors could have arisen from something an ancestral insect learned to do is not as far-fetched as it might seem. Researchers in the fields of robotics and artificial intelligence are particularly interested in how changes in stimuli—that is, the response a computer gets when it executes an action—could then make the computer's actions more sophisticated. It would be amusing if the behaviors so derided by Fabre and Cunningham turned out to pave the way for better, and more flexible, computers.

Finally, social insects are well known for their genetically hard-wired altruism; honeybees can't help committing suicide when they sting an intruder in defense of the colony, because the stinging apparatus remains imbedded in the victim, tearing the innards of the bee asunder after the sting. But it has just come to light that ants, at least, can also choose to rescue their kin from harm even when the peril is novel. Elise Nowbahari and her colleagues in France and the United States took ants, partially submerged them in sand, and restrained them with a nylon filament so that their bindings were concealed under the surface. The scientists then re-leased either strangers or nest mates of the victim and watched the ants' behavior. If, and only if, the entrapped ant was from the same nest, the other ants hurried over, dug her out, and bit the snare away. Ants from foreign colonies, even though they were the same species, were left to struggle helplessly.

Such a complex sequence of behaviors pushes the boundaries of what we thought an insect could learn. And if the same ability applies to species other than ants, we might want to rethink those sticky traps that attract cockroaches and trap them, alive and kicking, on the surface. If the roaches become able to rescue their fellows by nibbling through the glue, you have to start wondering if they might then be capable of plotting revenge.