J L Gould, Princeton University, Princeton, NJ, USA © 2008 Elsevier B.V. All rights reserved.
Behavior is a critical component in the ecology of most animal species. The location of a species' nests, the nature of the social system, and the timing of reproduction, for instance, help define the niche of otherwise similar sorts of animals. Bank swallows and cliff swallows, for example, coexist because of different nesting habits - excavating a burrow in a soft bank versus constructing an external cavity on cliffs from mud pellets. Honey bees and bumble bees may forage on the same blossoms, but neither drives the other extinct because their different social systems, communication behavior, learning techniques, and overwintering strategies minimize competiton; as a result, honey bees tend to dominate in the spring, bumble bees in the fall.
Work on understanding the behavior of animals falls into two overlapping categories: mechanisms (ethology) and evolution (behavioral ecology). Mechanistic approaches seek to uncover the neural bases of behavior, either at the level of programming strategies - the instructions - or in terms of sensory organs and the actual processing circuits in the nervous system. The emphasis is on species-specific behavior; evolution and ecology are important to ethologists as they relate to species-specific adaptations. Evolutionary approaches, on the other hand, more often look for generalized, species-independent rules for foraging, social organization, and decision making that link a species' ecology to its behavior. Other than the programming instructions, mechanistic considerations at this more global level are rarely thought to be very important; basically, ecology determines behavior.
Although ethology and behavioral ecology look at behavior from essentially different directions - bottom ! up versus top ! down - both approaches agree on the basic behavioral building blocks and typically take the animal to be a machine, operated by its nervous system and programmed by natural selection to sense and evaluate stimuli, make 'decisions', and effect responses. Beginning about 1980, however, an ever-increasing body of evidence shows that some animals in at least certain situations act as though they possess a degree of intelligence. Though covering only a limited number of species and behavioral contexts, these observations are compelling as far as they go. Consequently they alter the way ethologists and behavioral ecologists think about how animals interact with each other and the environment.
The view of animals as machines - still the most common perspective - was a reaction to the Clever Hans incident early in the twentieth century. A German schoolteacher named von Osten set out to teach his horse to understand spoken language, read simple words, and perform mathematical calculations. The horse would answer questions by pointing with his nose or tapping out the response with a hoof.The questions could be posed by anyone, and Mr. von Osten need not be present. The widely accepted inference was that Hans was highly intelligent. After careful study, however, the psychologist Oskar Pfungst was able to show that most of Hans' answers were cued by the audience: the observers grew tense as the horse neared the correct answer, then relaxed when the correct number of taps had been delivered. (This cannot be the whole story, however: Hans shook his head and refused to answer questions put to him in languages other than German, and began tapping his hoof rapidly when the correct response was a large number. These two well-documented behaviors have never been explained.)
The main response to the Hans debacle was to exclude intelligence as a possible interpretation of animal actions. Work focused on the more automatic aspects of behavior. Ethologists showed that animals live in species-specific worlds, with unique sensory experiences. Honey bee vision, for example, has much lower spatial resolution than that of primates, but much higher temporal resolution. Bees may be blind to red, but they can see in the ultraviolet and detect polarized light. Most dramatically, two closely related species can see the same objects or hear the same sounds and, without the benefit of learning, interpret them differently. A nesting tern, for instance, does not respond to its egg, but rather triangulates its nest based on local landmarks and thus will settle down to incubate on an empty nest, ignoring its egg which has been moved a few centimeters to the side. A gull, on the other hand, fixates on any displaced egg within half a meter or so, reaches out with the neck, and rolls it back into the nest.
Moreover, recognition of the egg in most species is quite schematic and independent of learning. Gulls will roll in cubes, light bulbs, batteries, and grapefruit; if given a choice, however, they opt to first recover the object with the more ovoid shape, the greener color, the most numerous speckles, and the largest size. These innate cues are known as sign stimuli, and are extracted by the nervous system to identify important objects or individuals and trigger appropriate responses. Nothing about sign stimuli requires intelligence on the part of the animal.
Similarly, complex behaviors like the egg-rolling response are innately encoded and called upon when needed. All geese, for instance, recover eggs in exactly the same way (with the underside of the bill, rolling the object between the legs), and once the recovery sequence has begun, the egg can be removed in plain sight and the bird will nevertheless continue with the rolling and settling behavior. Even the apparent variability (geese will sometimes ignore eggs) appears to be an automatic consequence of motivation or drive rather than decision making: the threshold for responding to an egg changes in concert with nestbuilding and incubation behavior, all of which are controlled by hormonal levels.
Learning, a behavior that might seem to require understanding, was seen by both behaviorist psychologists and ethologists as equally mindless. An innately recognized stimulus (sign stimulus to an ethologist, an unconditioned stimulus to a psychologist) releases an innate response; pairing of this stimulus with another cue (the conditioning stimulus) leads the animals to deliver the response to the novel stimulus. In the case of imprinting (as when goslings a few hours old memorize the voice and appearance of their parents), the learning also has a critical period outside of which the animals cannot make the association.
Psychologists showed that the strength of conditioning is mathematically scaled to the observed probability that the novel stimulus predicts the sign stimulus versus the probability that it is a false alarm. In the lab at least, learning of novel behavior (as opposed to associating cues) works in the same probabilistic way. Moreover, learning is often highly biased in apparently sensible species-specific ways. Pigeons, for instance, will learn to associate a visual stimulus with food, but not a sound; on the other hand, they can learn to connect a sound with imminent danger. Given that their natural food
(seed) is mute, ignoring the sounds around them as predictive cues while feeding probably speeds learning and minimizes spurious associations. And in developing novel behavior, pigeons experiment with the beak rather than wings or feet when food is the reward, whereas they try wings and feet first if avoiding danger is the goal. Conditioning, it would seem, is automatic and to some degree species specific, a consequence of circuits that extract cause-and-effect relationships from context-specific cues or actions and rewire the animal in consequence.
The machine model, at least at its most basic level, has outlived its usefulness for most animals. Behavioral ecologists, for instance, found that a species can have quite different social systems at two different locations; wildebeests, for instance, are territorial in a relatively rich habitat, but depend on dominance in poorer ones. Hedge sparrows, again depending on habitat quality, may be monogamous, polygamous, or polyandrous. Diet choice shows equal attention to ecological variables and economic realities, as does the time invested in harvesting (or attempting to attract mates) at a particular location before moving on. While these decisions could be controlled by innate programming, with the animal prewired to evaluate and weigh alternative factors, there seems little value (Clever Hans notwithstanding) to assume so without considering other explanations, particularly as the number of alternative responses multiplies.
Meanwhile, behavioral psychologists have accumulated substantial evidence that rats can build up mental maps of their surroundings, and use these maps to plan routes in the face of novel problems. Map use has generally been thought a high-level cognitive activity. At about the same time, pigeons were shown to form visual concepts such as 'tree', 'fish', and 'chair'. These abstractions of the specific to the general are essential to human thought, and had formerly been considered far beyond the capacity of other animals. The ability to accurately identify and categorize novel objects would make even innate decision making far more potent. Quite old work on chimpanzees (mentioned below), which had shown them capable of planning while solving food-gathering problems, has taken on new meaning.
Ethologists and cognitive psychologists have accumulated the most impressive body of evidence for cognitive potential in animals, including (most strikingly, given our biases) invertebrates. With regard to mental maps for instance, researchers showed that honey bees can plot novel routes to food when displaced from the hive and released, while hunting spiders can formulate indirect routes to prey. Work with birds shows that essentially every species with a need for a map seems to have one, and species like the pinon jay (which cache thousands of seeds that are later recovered in the winter) possess especially wide-ranging and detailed maps. Whether the need for a map leads to the capacity, or the capacity leads to the map-use behavior, is another question. In either case, the ability to devise novel routes and remember numerous landmarks is likely to change the niche (or at least the niche width) of a species.
Some workers argue that route planning may not be a particularly demanding cognitive process. Since there is a structure in the hippocampus of mammals that appears dedicated to this task, perhaps the ability is hard-wired. Instances of planning solutions for novel, map-independent tasks are more convincing. When presented with bananas hung out of reach, the chimpanzees mentioned above would look around at the objects in their enclosure, and with no trial-and-error experimentation begin an attempt to reach the fruit. For instance, the chimp might carry a box and set it down under the bananas, then stack another box on top, then bring a stick to the pile, climb to the top, and knock the food down. This appears to be a case of mental trial and error; there is no preliminary exploration or testing of the alternatives. But essential to this behavior is prior experience with the boxes and sticks - not experience using them in recovering food, though, but familiarity with their potential uses, which is acquired during play. The essential cognitive achievement is an ability to assemble a novel plan based on independently acquired pieces of knowledge, arranged in a logical order.
Another highly playful group of species that show clear evidence of planning are the corvids, which include crows and ravens. Hand-reared ravens were presented with food hung from branches by strings. After a time (and with no trial-and-error experimentation), most of the birds solved the problem by repeatedly pulling up on the string and stepping on the accumulating length as it piled up on the branch. New Caledonian crows, who make and use twig tools in the wild to remove insect larvae from holes, will bend wires to create a novel tool in the lab which can then be used to lift a food bucket out of a clear plastic tube. Here is a wholly new behavior directed at a material (wire) not found in nature used to solve what is, for the species, a vaguely conventional problem. The ability to imagine and then implement a solution to a novel problem can fundamentally change a species' niche. The need for an expended period of play, however, constrains the development of such behavior to species with a life history that permits this element of juvenile experimentation.
Nesting sites and structures are often an important part of a species niche. While some sorts of animals move through a stereotyped sequence of steps in building, a few are capable of making repairs to unlikely damage or of changing the nest basic design to suit the site. There seems to be some basic understanding of the purpose of the structure. The most impressive example of flexible innovation in building is seen in beavers, where no two lodges, dams, or sets of canals are the same. Lodges may be nothing more than burrows set into a bank, or excavated into peninsulas created by the beavers for this purpose, or built within a moat dug by the animals, or constructed on an artificial island set upon stones brought in the beavers, or fabricated within a human building at the edge of the water. Of course, there could be a separate set of programmed behavioral routines for each of these alternatives (and the many others that have been observed), but common sense suggests that the animals simply understand the goal and, using a mixture of innate and novel behavior, set about accomplishing it based on the characteristics of the site and the materials available. Such an ability must greatly broaden a species' niche by widening the range of habitats it can occupy.
In a behavior doubtless derived from building, male bowerbirds create elaborate, highly decorated structures to attract females. These sculptures (there is no other good word) may weigh hundreds of times more than the male. In many bowerbirds, the preferred colors of the decorations are relatively consistent and the design largely standardized. But in the species with the most elaborate bowers, the variation can be tremendous. Some males may paint the inside while others will invest their time in adding more sticks. Some will collect dozens of snail shells while others will focus on green berries. In one region, males may create elaborate maypoles surrounded by a densely woven moss mat and covered with fresh flowers; in another, the males may omit decorating the pole, and instead build a hut over the mat and decorate the periphery. It is hard to believe that the behavior is not driven by an esthetic sense - an appreciation of beauty we assume (in ourselves) to be cognitive. It is also hard to see how to apply conventional cost-benefit and optimization models to esthetics.
Self-directed learning and teaching are some of the most intellectually challenging behaviors animals might benefit from. For many years, there was no good evidence of teaching, though cultural traditions seemed common in at least chimpanzees and bowerbirds. More recently, however, the situation has changed. Border collies, for instance, are famous for their ability to understand (and accurately execute) human verbal and visual commands. Researchers have found individuals of this breed able to learn the spoken names of up to 250 objects, an achievement that in some ways far surpasses that of language-trained chimpanzees. One reason the dogs are able to acquire new names with such speed and accuracy is that they use a strategy similar to humans: novel names can be associated with novel objects through a simple process of elimination. If you know the name of everything else, the new word must refer to the unknown object. This logical inference is basic to human concept formation and thinking. It would be surprising indeed if this talent were restricted to a single breed of one domesticated species. The ability to make logical inferences is rarely a factor explicitly considered by behavioral ecologist, and yet decision making is a key component in ecological models.
A good example of teaching comes from meerkats, a highly social species of African mongoose. Adults teach pups prey-handling skills with a careful regard to the youngsters' existing knowledge - knowledge they gauge on the basis of age. Only dead scorpions, for instance, are initially fed to the young, but as the pups become older, the prey is merely disabled to some degree. The older the youngster, the less injured the scorpions are as the pups become increasingly adept at killing these dangerous food items on their own. The essence of teaching is modifying behavior based on the knowledge of the pupil. The reason the cultural learning in chimpanzees - passing on the practice of hammering nuts open with a stone, for instance - is not considered teaching is that adults appear to make no effort to show youngsters the techniques involved. The cognitive achievement of adult meerkats is perhaps matched by that of young chimps, who learn despite the apparent indifference of their elders. Skeptics, however, can readily imagine noncognitive programming to account even for teaching and copying; the definitive experiment has yet to be performed.
Cognition can also affect the way social systems work. The standard models of behavioral ecology suppose that animals must understand their position in the group hierarchy and/or their degree of relatedness to other individuals. But work from the field and zoos indicates that in some species individuals know far more - that they understand, for instance, which offspring belongs to which female, which pair of adults are likely to form an alliance, as well as how to create temporary partnerships to achieve short-term and long-term goals. Moreover, in at least some species, individuals appear to understand the state of knowledge of others, warning other individuals only if they are ignorant of a threat, and even then only if it is in the animal's best interest to bother with a warning. An animal with the cognitive wherewithal to understand what others know vis-a-vis itself is in a better position to increase its reproductive fitness.
One of the most difficult issues in cognitive ethology is the question of whether any animals are self-aware. The most persuasive evidence to date comes from studies allowing animals access to mirrors or real-time video feeds. Dolphins and higher primates quickly come to treat their mirror images as themselves, using the reflection for self-examination and grooming rather than threatening the image. A real-time video elicits the same response, with the animal engaging in contingency-checking behavior (seeing if moving a limb or head produces the same action on the screen) followed by exploration. In control tests using video tapes of the same animal from an earlier time, the contingency-free replays are ignored. What are the ecological consequences of an individual understanding that it is an independent actor in its group and environment?
The standard models of behavioral ecology assume that individuals make decisions based on circuitry fashioned by natural selection, attuned to measure a set of relevant but predefined variables and calibrate themselves through directed learning. But if some species are instead using logic and insight, devoting energy to strate-gizing and planning, learning independently via inference, short-circuiting their innate circuitry through concept formation, ascribing knowledge and intentions to others, then our models grossly underestimate what they are capable of, and fail to capture what is actually going on. The human analogies so mercilessly weeded out ofthe study of behavior after Clever Hans are now essential tools for formulating hypotheses. The idea ofintelligence, and especially different degrees within a species and different kinds between species, may be critical to understanding competition and evolution.
Was this article helpful?