You Are Here Page 10
KARL VON FRISCH
Edward Tolman appeared on the scene in experimental psychology in 1918, when he took up a post at the University of California at Berkeley that he kept for his entire career.1 The experiments he conducted using rats served as the beginning point of a modern cottage industry in using rats to understand how minds deal with space. Tolman began his work at a time when psychology in North America lay heavily in the shadow of a radical theory that seemed to deny the existence of mind itself. John Watson, the father of the behaviorist movement, had published a manifesto in which he claimed that all human behavior could be accounted for as sets of learned associations between stimuli and responses. Like Pavlov’s famous dogs, we were doomed to go through life responding more or less automatically to tolling bells and flashing lights, doing only what we had learned brought pleasure and avoided pain.
Though it seems incredibly brutal by today’s ethical standards in psychology, one of Watson’s most influential papers demonstrated that it was possible to train a young child to fear a white rat using simple conditioning methods. Watson showed the rat to the child while making a loud and startling sound immediately behind the child’s head. Not surprisingly, the child quickly learned to fear the rat, and Watson made the bold claim that all human phobias had a similar genesis.
Tolman, in contrast, was reluctant to cast away the idea that human heads contained things called minds, whose composition was much more interesting than the collection of associations envisioned by psychologists like Watson. Not only did Tolman take the then radical view that human beings had minds but he wondered whether his collection of rats might possess such things as well. To test his idea, he devised laboratory tasks to assess how much his rats knew about space.
Rats can be trained to do tasks with great ease using methods much like those you might employ to train a pet to perform a trick. In the beginning, they are given a simple task to perform (leave the start box) and are rewarded with food. As they become more accomplished, they are given greater challenges until, eventually, they can complete the whole task. To begin, Tolman placed his rats in a small, square box and trained them to cross a round chamber to enter a narrow alleyway. After a short run in the alleyway, the rats were required to make three right-angle turns into a final alleyway, at the end of which they would find a tasty food reward. The basic setup is illustrated in Figure 4.
Maps in Mouse Minds
Figure 4: Tolman’s special orientation maze for rats
Once his rats had learned their way to the reward, Tolman made a critical change in the shape of the maze. First, he closed off the alleyway leading out of the central chamber partway along, and next he opened up a large number of alleyways radiating from the chamber in the shape of a starburst, as shown in Figure 5. The location of the reward remained unchanged.
Figure 5: Variation on Tolman’s original maze
What will the rat do? According to a behaviorist, Tolman has broken the chain of simple behaviors that the rat has learned. When it crosses the central chamber, enters the usual alleyway, and finds its way blocked, either it will bash into the closed door in frustration or it will wander aimlessly with no clear plan. But what Tollman observed was something much more interesting. Most of his rats chose the alleyway that led most directly to the reward—which, remember, hadn’t moved. What this clever response suggested was that Tolman’s rats possessed an understanding of the spatial relationship between the start box and the food reward that allowed them to follow a route they had never seen before. Indeed, before Tolman made the changes, the route simply did not exist. It was as if the rats had stored an overhead view of the scene in their minds that they could consult to find the best way to respond to an unexpected turn of events.2 For Tolman, the discovery of this “cognitive map” was a major victory in his attempt to demonstrate that simple associations between stimulus and response did not adequately explain all behavior—even the behavior of a laboratory rat. For our story, the main importance of this finding was its implication that rats must possess something like a map in their minds that they can consult to solve the problem.
How do such cognitive maps differ from the gradient maps that guide the movements of homing pigeons and turtles? The main difference is that gradient maps are based on some kind of relatively straightforward physical quantity that is continuously available to animals and can be directly related to position—geomagnetic force lines, patterns of ocean swell, or even panoramas of odors. Cognitive maps, as suggested by the term itself, are constructed using various pieces, including the identity and appearance of individual landmarks and their observed relationship to one another. To some extent, it is possible to observe the spatial relationships of landmarks by simply looking at a scene, but such looking must take place from a variety of observation posts, and to generate a metric map—one in which the directions and distances between things are measured accurately—one must know how the observation posts themselves are connected. In other words, in order for us to build a map based on a set of observations, we have to know where we were when we made the observations. The metric accuracy of a cognitive map is going to depend on the accuracy of path integration. Though a gradient map may serve nicely to bring a turtle to a rich bed of ocean vegetation, a cognitive map of the kind envisioned by Tolman and countless others is a much more sophisticated and flexible wayfinding device.
Not everyone agrees with Tolman’s conclusions about how his rats solved starburst-maze problems. Many scientists continue to try to understand what kinds of maps can be found inside the minds of rats, but almost everyone agrees that rat minds contain some kind of representation of space. Today, more of the debate focuses on questions about where in the neural machinery of the rat such maps can be found. In other animals, the debate is more sharply focused on the question of what it might mean to say that an animal’s head contains a map. Nowhere has this question been more contended than among those who study the mental world of the honeybee.
BUMBLING BEES
The German scientist Karl von Frisch was awarded the Nobel Prize in 1973 along with Konrad Lorenz and Nikolaas Tinbergen, two others who revolutionized the study of the behavior of animals in their natural habitats. Like Tinbergen, von Frisch attributed much of his interest in animals to a free-wheeling youth spent in the countryside watching things grow and move about. As a scientist, von Frisch conducted clever experiments, much in the manner of Tinbergen’s simple studies with digger wasps, that showed how insects can employ landmarks to find their way home.3
Von Frisch’s most dramatic finding concerned what he called the bees’ waggle dance. This peculiar wiggly buzz of movement had first been noted by Aristotle, who speculated that bees used these movements to call attention to themselves.4 Von Frisch suggested that something much more interesting was happening.
When colonies of bees need food, they send out scouts. When successful scouts return to the hive, bringing nectar with them, other bees leave the hive and head directly for the source of nectar. Von Frisch spent many hours hunched over hives, watching bees return to the nest and then leave again for the fields, and he eventually formed an incredible hypothesis. He believed that the scouts were somehow telling the other bees where the food source was located. Like Aristotle, von Frisch focused on the waggle dance—a stereotyped set of movements in which the scout bee walks in a straight line while buzzing, turns in a loop, walks in a straight line again, and then loops in the opposite direction, making a kind of figure eight. Other bees in the colony follow the scout bee as it carries out this display and, after a few repetitions, observing bees will take flight and head in the direction of the food source. Careful observation of the waggle dance convinced von Frisch that the movements made by the scout bee consisted of a kind of body language that encoded symbolically the location of the food source. He noticed that scout bees that had found food placed at very distant locations carried out longer straight buzzing segments in their runs than those that had found food closer to the
hive. So the length of the straight part of the dance signals the distance to the food.
What about direction? The bees that von Frisch studied varied the direction of their dance depending on the bearing of the food from the location of the hive. Von Frisch suggested that the bees were giving a compass bearing to follower bees that was related to the position of the sun. The honeybees that von Frisch studied danced on vertical honeycombs. Von Frisch suggested that scout bees converted direction information such that straight upward was meant to represent the position of the sun. The angle of the dance relative to true vertical, then, was meant to indicate the angle of the route to the food source relative to the sun. There’s one more wrinkle to iron out before the bee language that von Frisch described could work properly.
As you can imagine, when these findings were being reported, beginning with von Frisch’s studies in the 1920s and continuing to the present day, reactions ranged from skepticism to utter disbelief. Von Frisch was suggesting that an insect with a brain about the size of the head of a pin was capable of acts of communication that could be differentiated from true languages only by the careful hairsplitting arguments of philosophers. Von Frisch himself devoted considerable energy to ruling out simpler explanations for his findings (such as, for instance, that the follower bees were simply following some kind of odor trail to the food source), but only very recently have advances in technology enabled researchers to provide what seems like ironclad evidence for the key role of the waggle dance in bee navigation.
In 1989 a team of researchers at the University of Odense in Denmark built a dancing robotic bee. Although this bee robot did not look very much like the real thing (for instance, it had only one wing and this was constructed from a piece of a razor blade), the team, led by Axel Michelsen, was able to show that not only would follower bees pay attention to the dance of the robot but they would also fly off to look for food at directions and distances described by the robot’s dance.5 As we have seen many times before, the ability to trick an animal into behaving in a certain way by massaging the information to which it is exposed makes for a very convincing demonstration of the importance of this information in the animal’s everyday life.
Though it is now well established that bees use the waggle dance to talk about space, controversy still surrounds the question of what kinds of maps of space bee brains might contain. In a widely discussed study,6 James Gould set out to demonstrate map use in bees using a kind of shortcut experiment. Bees were trained to fly from a hive to a feeder. Once the bees had learned to make the trip from hive to feeder and back again, they were picked up and transported to a new release point. Like Tolman’s rats in the starburst maze, Gould reported that many bees flew directly from the release site to the feeder without first returning to the hive. If this route truly represented a novel shortcut, then it suggests that bees possessed a metric cognitive map.
Many researchers took issue with Gould’s conclusions. Those who tried to repeat his shortcut findings noted that the bees could complete shortcuts only under conditions where it was impossible to rule out a simpler strategy than map use. Gould’s original studies, it was argued, were set up in such a way that certain landmarks were observable both from the hive and from the new release point, so the bees could have been using one of the simple landmark strategies I described in chapter 2.7
Until the late 1990s, the tide was turning against the cognitive-map hypothesis for bees, but more recently, the tide has begun to reverse. One breakthrough has been the discovery that bees carry out different kinds of flight missions depending on their history. When young foraging bees are first learning the ropes, they go on what are referred to as orientation flights, whose purpose appears to be to learn the terrain and landmarks surrounding the hive.
Rodolfo Menzel thought that these “orienteering” bees might learn different things about space than experienced foragers. To test his hypothesis, he conducted an experiment in which he first moved an established hive to a new location in order to encourage foraging bees to carry out orientation flights, and then made life even harder for the bees by moving the feeder around from time to time. The feeder was kept close enough to the hive so that it was always easy to find, but because it was never in the same place on two successive foraging trips, the bees would need to pay close attention to their surroundings if they were to find their way home.
Menzel’s surprising finding was that bees trained in the more uncertain world of the moving feeder used a flexible navigation strategy based on something like a map. It was as if these bees, on release, took stock of their surroundings, computed their position, and set off on an accurate “beeline” for home. Recent studies using specialized radar methods to track the full flight paths of foraging bees have confirmed this idea. Bees that had undergone only orientation flights were capable of making direct flights from novel release sites to both the hive and to the feeder, and some of these flights took the bees through terrain they had never experienced before.8
It is very difficult to conceive how a bee could find its way home from a location that it had never visited before using a completely novel route unless it had a map in its head. Such ability, if it exists, is set apart from all the methods of navigation that we have considered so far. Carrying a map of the terrain, with landmarks and their spatial relationships, within one’s nervous system allows one to solve wayfinding challenges at an entirely new level of difficulty, and with the flexibility to solve much more complex and interesting “you are here” problems.
FOOD-CACHING BIRDS
Animals spend much of their time searching for food, securing access to it, and doing what they can to buffer themselves against the vagaries of a perpetually uncertain supply. For many animals this means working carefully to collect and store food during times of abundance so that it can be consumed at a later time, when conditions change. Every young child has been told stories of the prudent squirrel that prepares for winter by gathering and burying nuts to provide a steady food supply for the scarce months ahead, and scientists have demonstrated that such a strategy can indeed be effective in warding off hunger in the bleak winter months.
Some animals store food in “larders” designed specifically to afford defense against theft, whereas others, such as many species of food-storing birds, employ a system of “scatter hoarding” in which they distribute morsels of food across large reaches of the environment for later use. Scatter hoarders have been shown to have a prodigious ability to hide and then retrieve food, and so have been subject to intense scrutiny by scientists interested in how animals use space.
Chickadees, tits, and nutcrackers are among the species of birds that store food in cache sites. Some of these birds have been observed to use up to 80,000 different cache locations in a single fall season. Seeds, decapitated insects, or bits of worms are stored in a wide variety of locations, such as under tufts of dirt on the ground, underneath the bark of trees, and inside hollow plant stems. Cache sites can be either right on the ground or high above it, and birds can cache food either very close to where it was found or at distances of up to about 100 meters away. Most cached food is recovered and consumed within a few days of storage, but some morsels are known to have been hidden away for some months before being retrieved.9
How do we know that the birds actually remember all these cache locations? It would be easy to imagine that they might use some kind of simple rule to find cache sites. For example, if a particular bird always stores food in hollow stems, then it might find its own caches by searching randomly among all the hollow stems that it encounters. I use a similar strategy when I discover, and recache, Halloween candy in my house. Knowing that almost all my children are more vertically challenged than I am, I cache tasty treats on high shelves. But given the sad state of my overtaxed memory, I seldom remember which shelf holds my stash. Luckily for me, it doesn’t take too long to search all the high shelves in my house for a sweet payoff.
We know, however, that food-storin
g birds don’t use this kind of rule-based strategy because of some experiments in which field researchers watched carefully where birds stored seeds and then went out and stored extra seeds in nearby locations that looked almost the same as the original cache site. When the birds revisited cache sites, they showed a strong preference for their own sites as opposed to those prepared by the researchers. All evidence suggests that these birds actually remembered the locations of their cache sites in great detail, days or possibly even months after they stored the food.
Such an extraordinary memory for spatial locations presents myriad opportunities for us to understand how warm-blooded animals manage space. As some birds can also be coaxed to cache and recover food in the artificial setting of the laboratory, scientists have been able to learn a great deal about how spatial memories in food-caching birds are put together.
In one experiment, Clark’s nutcrackers were trained to dig in a specific location on a sawdust-covered floor in order to unearth a small stash of pine seeds. The location of the stash was reliably indicated by a set of prominent landmarks, but the setting was cleverly arranged so that the birds would need to be able to consult a mental map relating the landmarks they could see from the target location to a set of screened-off landmarks that they had learned to associate with the target during training but could not see when near the target site. The nutcrackers were able to accomplish this difficult task, suggesting the existence of a cognitive map, but their accuracy was lower than expected, leaving some lingering doubts about whether the birds possessed a cognitive map or whether they were falling back on some kind of simplifying trick to find the target.10 But even if future studies show that the bird map is not quite up to the lofty standards set by Edward Tolman and others for a cognitive map, birds have shown great flexibility in solving some difficult navigational problems in the face of virtually every curve that wily experimenters have been able to throw at them. These animals cling to their place on the planet with remarkable tenacity, relying on one backup system after another in order to avoid becoming lost.