EVOLUTION AND LEARNING

Learning is a biological adaptation, and like any other adaptation is the outcome of evolution by natural selection. Because it is acted on by natural selection, learning in different species of animals exhibits both descent with modification and specialized adaptations. Many properties of learning, like the formation of associations, are widely shared among animals. The molecular mechanisms of learning are also remarkably similar among animals as different as sea slugs, honeybees, and rats. But, in addition, learning exhibits specialized adaptations, modifications of learning which differ between species. Evolutionary adaptation in learning is usually investigated using comparative methods to examine similarities or differences among animals in how or what they learn. Learning can also have a reciprocal effect on the process of evolution. Animals can learn to exploit new habitats and new resources within their habitat and thus change the selective pressures they are exposed to. Learning can even have an evolutionary impact that extends beyond the animal itself and affects other animals and plants it interacts with.

There are two basic requirements for evolution of a trait like learning by natural selection. First, the trait must be at least partly heritable: Genotypic variation must produce phenotypic variation in the trait. Second, variation in the trait must have an effect on reproductive success. Learning meets both of these requirements.

Genetic Variation in Learning

Some of the clearest evidence for genetic variation in learning comes from studies of learning mutations in the fruit fly Drosophila (Waddell and Quinn, 2001). Fruit flies are good learners, and can readily learn to avoid an odor that has been associated with electric shock. Mutation of single genes in the Drosophila genome can be induced with chemicals, and some of these mutations have dramatic effects on learning. Mutations in genes with whimsical names like rutabaga and dunce make fruit flies unable to learn an association between odor and shock, or cause them to quickly forget what they have learned. These particular genes code for components of the intracellular signaling system that transforms neural activity into more long-lasting changes in neurons that record experience. The rutabaga and dunce genes code for enzymes that increase and decrease, respectively, the intracellular concentration of cAMP (cyclic AMP), an intracellular second messenger that responds to neurotransmitter signals received by a neuron. Other learning mutations have been discovered that affect different aspects of the learning process, like volado, a gene that affects cell adhesion and influences communication between neurons. These mutations can reveal a great deal about the molecular mechanisms of learning and help unravel the neuroanatomy of learning by pinpointing sites in the brain where expression of the gene makes a difference in learning. But learning mutations are also important discoveries for understanding evolution. They show that changes in genes (in some cases substitution of a single nucleotide) can affect the properties of learning. The phenotypic effects of genetic variation of this kind provide the raw material on which natural selection can act.

Learning and Reproductive Success

Learning contributes directly and indirectly to reproductive success in many ways. Bumblebees learn how to obtain nectar from flowers. Colonial swallows learn to recognize their young, and young herring gulls learn to recognize their parents. Most animals must learn what is edible and what is not; others learn migration routes, how to identify predators, how to defend a territory, and how to attract a mate. In all of these cases, any heritable variant in learning that makes the animal slightly more successful at the task will make it more successful at reproducing itself, and hence more likely to pass on the variant in learning to its offspring. As a consequence of the action of natural selection on learning, learning can differ between species and possess specialized adaptive properties that make learning more effective. Learned food aversions illustrate this kind of specialization. Animals sometimes eat food that makes them ill, either because the food is contaminated or because the plant or animal they have eaten produces toxins to discourage predation. Animals can clearly benefit from learning which foods are edible and which are not, but the natural situation presents them with a problem. Toxic food may not have its effect for several hours after it was eaten. Animals usually have great difficulty learning that two events are related if more than a few minutes separates them in time. Experiments in the laboratory show that rats can associate illness with food even if the food was eaten several hours previously. Rats may require only a single experience with the food to form a strong aversion to it. Furthermore, rats associate the taste and odor of food, not its appearance, with illness. Selectivity in what is learned and the ability to associate events separated in time are distinctive features of tasteaversion learning.

Songbirds exhibit specialized learning in the way they acquire their songs. The songs that male Passerine birds use to advertise territory ownership and to attract a mate are learned. The song-learning system possesses a number of unusual features. Most birds learn only the songs of their own species, even if they are experimentally exposed to songs of other species for an equal period. In addition, the young of many birds have a "sensitive period" during which they learn songs most readily. Songs are not learned by singing them. Instead, songs heard during the sensitive period are remembered until they are first sung many months later, when the breeding season begins. Finally, there are specialized nuclei in the avian brain that are responsible for acquisition and production of song. Restrictions on what is learned, a sensitive period, separation in time of learning and performance, and specialized neural structures make the song learning system different from other kinds of learning, but effective for memorizing and performing species typical songs.

Comparative Methods

Comparative methods that have been used to examine adaptation and evolutionary change in animal physiology and anatomy can also be used to examine the evolution of learning. The clearest way to see the effect of evolution on the process of learning is by comparing the ways in which different species of animals learn. Closely related animals that share most of their evolutionary history but differ in some aspects of their current behavior or ecology may differ in how they learn. Comparisons of learning in closely related animals can thus reveal adaptive modifications of learning that are the result of recent selection. Another approach is to compare animals that are not closely related but share some current aspect of behavior or ecology. Similarities in learning between such animals occur not because of a shared evolutionary history but because they have been exposed to similar selective pressures. If such animals show similarities in how they learn or what they can learn it is likely to be the result of convergent evolution. These animals have independently evolved similar learning capacities because they have been exposed to similar selective pressures. Both kinds of comparison have been used to identify the selective pressures that can influence learning and to illuminate how evolutionary change in learning can occur.

Learning in food-storing birds illustrates how these comparative methods can be used. Some species of birds, notably chickadees, nuthatches, and jays, store food. They make hundreds to thousands of food caches and return between several days and many months later to collect and consume the food they have hidden. Caches are widely scattered and contain only a few food items each. Remarkably, these birds remember precisely where they have placed each cache. Some birds in the jay family, the Corvidae, store a great deal of food and some store little or none. Comparisons among these closely related species of birds have shown that reliance on stored food is associated with the level of performance on spatial tasks. Ecological dependence on food storing has selected for enhanced spatial ability. Food storing evolved independently in jays and in the chickadee family, the Paridae. Food-storing chickadees and food-storing jays both possess enhanced spatial abilities as an evolutionary consequence of their shared reliance on stored food.

Learning and Evolution of the Brain

Evolutionary change in learning requires evolutionary change in the neural apparatus of learning. Evidence for evolutionary modification of learning comes not only from observing differences between species in learning itself, but also from examining differences between species in brain areas that are important for learning. The song control nuclei of species of birds with large song repertoires are larger than the nuclei of birds with small repertoires. An increase in the size of neural structures that participate in learning has been found for a number of other kinds of learning.

The avian hippocampus plays an important role in memory, as it does in mammals, and the hippocampus of food-storing birds is over twice the size of the hippocampus of closely related birds that do not store food (Sherry, 1998). Comparative studies of this kind show that adaptive evolutionary change occurs in brain regions involved in learning. A further example illustrates that such adaptive change can occur within a species. Most voles (rodents in the family Cricetidae) are polygynous: One male has several mates. Male home ranges are larger than female home ranges, and the home range of a polygynous male may encompass the home ranges of several females. Some species of vole, however, are monogamous, and male and female home ranges are of equal size in these species. Laboratory experiments have found that males of polygynous species perform better on spatial memory problems than do females, but in monogamous species males and females perform equally well. The sex difference in spatial ability in polygynous species is an adaptation to their breeding system and to the sex difference in home range size. As with food-storing birds, the consequences of natural selection for learning ability can be seen in the brain. Lucia Jacobs and colleagues (1990) found that the hippocampus of male polygynous voles is larger than that of females, whereas in monogamous voles there is no sex difference in the size of the hippocampus.

The Effect of Learning on Evolution

Evolutionary change occurs in learning, but learning can, in turn, affect the course of evolution. Many species of animals do things that are culturally determined. Behaviors that are traditional within a population of animals are learned from other members of the group, either directly or simply by associating with other group members. Migration routes, learned songs, and food preferences can all be transmitted culturally. Naive birds can learn to recognize predators by participating in mobbing attacks on animals that other members of the social group treat as predators. The effect of such culturally transmitted behavior on biological evolution is not fully understood, though it is clear that learned behavior can consistently expose animals to new environments and new sources of natural selection.

Learning can also have evolutionary effects that extend beyond the animal that does the learning. There is awe-inspiring diversity in protective mimicry among insects. The monarch butterfly contains toxins that make predators ill. The viceroy is a palatable butterfly that mimics the monarch butterfly in appearance so closely that birds that have tasted a monarch avoid both monarchs and viceroys. Many similar model and mimic systems are found in insects. Some syrphid flies have evolved to closely resemble bees and wasps in their appearance, posture, and behavior. They possess the distinctive black and yellow banding pattern found on many bees and wasps, at rest they hold their forelegs in front of their head to resemble wasp antennae, and their seasonal period of activity coincides with that of their bee and wasp models. These mimicry systems have evolved because the animals that would normally prey on the mimetic insects—primarily birds—learn to avoid the toxic, bad tasting, or stinging model and because of its similar appearance, also avoid the mimic. Without learning by potential predators, there would be no mimicry.

As they gather nectar, insects such as honeybees, bumblebees, flies, and wasps carry pollen from one flower to another and serve as the sole means of fertilization for many flowering plants. Among some pollinators, such as bumblebees, different individuals from the same colony learn a preference to visit one kind of flower over others, a phenomenon remarked on by Charles Darwin and known as constancy. Pollinators are probably constant because visiting the same kind of flower makes it easier for them to recognize the flower and extract its nectar. This behavior also affects the flowers, and indeed flowers have probably evolved to promote constancy because it increases the likelihood that pollen will be transferred to another flower of the same species. Constancy can also have a further evolutionary effect (Jones, 2001). Constancy by pollinators may promote speciation in flowering plants by increasing assortative mating, the tendency of similar members of a population to mate with each other. If a population of flowering plants exhibits variation in the appearance or structure of its flowers, constancy by pollinators will result in preferential mating between flowers with the same structure and appearance, leading ultimately to the formation of new species.

Conclusion

Because genetic variation can produce variation in the mechanisms of learning, and because learning makes important contributions to the ability of animals to reproduce themselves, learning evolves by natural selection. Many properties of learning are shared among animals by virtue of their common descent. The formation of associations and some molecular mechanisms of learning are remarkably similar across a wide range of animals. Evolutionary change in learning has also produced specialized adaptations of learning. Food-aversion learning, song learning, and cache retrieval in food-storing birds provide examples of such adaptive specialization. The effects of evolutionary change in learning can also be observed in brain areas that play important parts in learning. Not only has evolution affected learning, but learning can affect evolution, both by exposing animals to selective pressures they would not otherwise encounter and by causing evolutionary change in the animals and plants with which they interact.

Bibliography

Jacobs, L. F., Gaulin, S. J. C., Sherry, D. F., and Hoffman, G. E. (1990). Evolution of spatial cognition: Sex-specific patterns of spatial behavior predict hippocampal size. Proceedings of the National Academy of Sciences of the United States of America 87, 6,349-6,352.

Jones, K. N. (2001). Pollinator-induced assortative mating: causes and consequences. In L. Chittka and J. D. Thomson, eds., Cognitive Ecology of Pollination. Cambridge, UK: Cambridge University Press.

Sherry, D. F. (1998). The ecology and neurobiology of spatial memory. In R. Dukas, ed., Cognitive Ecology. Chicago: University of Chicago Press.

Waddell, S., and Quinn, W. G. (2001). Flies, genes, and learning. Annual Review of Neuroscience 24, 1,283-1,309.

David F.Sherry