Brains combine critical, immediate payoffs with exceptional costs.

Workers and soldiers of the army ant Eciton burchellii cupiens, Tiputini Biodiveristy Station, Ecuador. Note the large, whitish head capsules and long, curved mandibles (mouthparts) of the soldiers.

Sean O'Donnell

Brains are excellent organs for exploring the interplay between evolutionary opportunities and limits. The nervous system, including such sensory structures as eyes and noses (or antennae), is the principal interface of an animal with its environment; the cognitive aspects of the nervous system process information and generate behavior. Fast, accurate, and, in some cases, flexible nervous system performance can be expected to yield substantial evolutionary payoffs. But nervous systems are expensive. In fact, nervous tissue is among the most metabolically costly tissue to use and maintain, and the developmental costs of producing and growing nervous systems can be substantial. For an animal of a given body size, or one with a limited total energy budget, investment in brain tissue will come at the disproportionate expense of meeting other bodily needs. Given this balance of great benefits tempered by high costs, we can expect brain investment to be generally favored but tightly constrained. Brain investment should closely match the cognitive demands a species faces.

A fascinating illustration of matching cognitive demand with brain investment comes from seasonal brain dynamics in songbirds. In the temperate zone, songbird brain structure changes over the year. These changes illustrate both the notion that a brain structure’s size matches its cognitive power, and the assumption that brain tissue is expensive and tightly restricted. Mating behavior for most temperate songbirds is strongly seasonal. Songbirds experience a surge of circulating sex hormones in late winter and early spring, driving dramatic increases in the size of the testes in the males. After mating season, the testes regress to smaller size. The male birds’ brains, or more accurately particular regions of their brains, undergo similar annual cycles of developmental increase and regression. Songbird brains include several anatomically distinct regions that are important to the ability to sing, mainly in males. Neurobiologist Eliot Brenowitz and colleagues at the University of Washington have shown that these brain regions—the song nuclei—increased in size prior to the onset of breeding and song production and regressed substantially afterwards as the need for song waned. This cycle of brain region growth and reduction is repeated annually throughout the bird’s life. Production and recruitment of new nerve cells (neurons) is part of the process of song nucleus size increase.

Male and female wild form guppies (Poecilia reticulata)

The field of neuroecology explores how brains—their size, structure, and information processing abilities—are shaped by species’ differences in environmental and cognitive challenges. One of the main underlying assumptions of neuroecology is that more brain investment—larger brains, or more brain tissue—will correspond to greater cognitive capacity. Recent experimental studies by biologist Alexander Kotrschal at Stockholm University directly tested and supported this assumption. Guppies (Poecilia reticulata), the popular aquarium fish, were artificially selected in the lab for increased relative brain size. Brain size evolved quickly, and the resulting larger-brained guppies showed improvements in cognitive ability (learning tasks). The expense of the larger brains drove reductions in some other body tissues and even led to lowered offspring production.

An illustration of Haller’s rule, using brain volume data for eight species of Eciton army ants. The ratio of total brain volume to body size (head capsule volume) is plotted against head capsule volume for workers (gray symbols) and soldiers (white symbols). The symbol shapes used to plot the data points for each species are indicated in the legend.

Sean O'Donnell
Relatively large brain size or brain region size is often tied to cognitive performance, but this does not mean that small brains are universally disadvantaged. Body size is an important factor in understanding the evolution of brain size. In general, larger-bodied species have larger brains, indicating that bigger bodies require bigger brains to function properly. But within animal lineages, the evolutionary increase in brain size with body is strongly non-linear. The relative amount of increase in brain size with body size is minimal for the biggest species; in contrast, as the smallest body sizes are approached, relative brain size skyrockets. It is almost as if minimum brain size hits a wall—the body can go on shrinking, but the brain cannot. This common, if not nearly universal, pattern of brain-body size relationships is known as Haller’s rule. Exactly why Haller’s rule holds is not completely clear, but one idea is that each animal lineage has fundamental cognitive requirements that cannot be sacrificed. These lineage-specific cognitive needs impose a strict minimum on the brain size that is possible within that lineage: for example, if you are going to be a successful monkey, you need to have a brain that supports some basic monkey behavioral abilities. If you are a small monkey, less monkey brain simply won’t cut it below some required minimum, even if a species of similar size in another mammal lineage could function with less brain investment. Both body-relative brain size, and evolutionary lineage and relatedness, need to be considered in studies of brain evolution. In any case, there is little evidence for a general cognitive decline in smaller animals even though they have smaller brains. Whether and how brain function might meet limits at the very smallest animal body sizes, such as in tiny in-sects and spiders, is an open question of considerable interest to researchers. 

Myriad examples from across the animal kingdom support the general predictions of neuroecology: that fundamental differences in species’ ecology affect the cognitive challenges they face, and that a species’ ecology is reflected in the structure of its brains. For example, the typical sensory environment a species occupies is often strongly reflected in its brain structure. A well-studied test case explores the effects of variation in the light levels a species normally experiences during activity. Within several animal lineages that typically occupy high-light environments, such as above-ground activity during daylight hours, some species have evolved to make use of darker environments. Decreases in light levels accompany species shifts to nocturnal activity, and darkness also accompanies evolutionary moves to underground and cave life. There are consistent changes in brain structure that follow the evolution of these dimly-lit lifestyles—for example, decreases in the relative size of brain regions that process visual information. This type of evolutionary brain change has occurred independently in nocturnal mammals, and in subterranean mammals and ants. In one lineage of Mexican cave fish, surface stream-dwelling ancestors have given rise to several independent lines of cave-dwelling fish, with remarkably similar changes in brain structure accompanying each invasion of the stygian realm.

Same-scale 3-D reconstructions of head capsules and brains of an army ant (Eciton burchellii parvispinum) worker (left) and soldier (right) from the same colony. Note the much larger brain size of the smaller worker realtive to its head size. The scale bar represents 0.5 mm. Brain structures are the colored bodies in the center of each head capsule. The mushroom bodies are dark blue, and the antennal lobes are light blue.

Sean O'Donnell
Work in my lab showed evidence for a reversal of this trend among species of New World army ants. Army ants evolved from surface-dwelling ants, but the most basal army ants and their immediate ancestors are largely or completely subterranean. The most recently evolved army ant genus, Eciton, is also the most above-ground active: Eciton returned to the surface in a lineage with a long history of below-ground activity. Estimated dates of divergence suggest the ancestral Eciton came back above ground about 18 million years ago, after 60 million, or so, years in the dark. Eciton workers have much larger visual processing brain regions called optic lobes than all the other New World army ant genera; one Eciton species has optic lobes that average six times the relative optic lobe volume of the most subterranean army ant species.

Beyond the impact of the sensory world, the evolution of more complex, integrative cognitive abilities can leave its mark on brain structure as well. An example from vertebrates is the strong association of increases in the relative size of the hippocampus—a brain region critical for spatial learning and memory—with the evolution of behavior that requires persistent knowledge of special locations. In birds that store, and later recover, non-perishable food items, such as seeds or nuts, spatial navigation and place memory are critical to successful retrieval of the stored food items. The relative volume of the hippocampus covaries with reliance on caching among bird species, and even among different populations within bird species.

Another potentially relevant feature of an animal’s ecology is its social environment—the array of interactions and relationships an animal has with other members of its own species. Most animals are minimally social; aside from the necessary but brief interludes when males and females get together for courtship and mating, most animal species are solitary. Yet in many animal lineages, famously including our own, varying degrees of duration, intensity, and complexity of social interactions have arisen. Species-typical social interactions can include combinations of extended parental care, male-female partnerships, and integrated groups that in-clude mainly relatives and even non-relatives. Group living can be temporary or life-long. Research on a wide array of species suggests variation in sociality plays a strong role in driving brain evolution.

Group infighting and nasty politics are pervasive, omnipresent, and, sometimes, deadly. Even positive, affiliative social interactions—friendships and consort-ships—often take the form of alliances to ward off aggressive competitors. Sometimes, backstabbing occurs.

One idea called the “social brain hypothesis” suggests that animals’ social environments can impose novel cognitive challenges and select for changes in brain capacity. In diverse animal taxa, from mammals to birds to fish, measures of species-typical social behavior correlate positively with either total brain size, or with the relative size of targeted brain regions. Among vertebrates, social brains are bigger brains. But why? What is it about sociality that demands greater cognitive capacity and avors greater brain investment? There may be a dark side to social brain effects.

Among vertebrates, evolutionary increases in social group size or social complexity often require animals to interact with, and navigate relationships with, ever larger numbers of individuals. As group size goes up, an animal’s social partners are more likely to include distantly-related, or even unrelated, individuals. Consequently, social partners often do not overlap in their evolutionary agendas. Even a quick glance at the literature on primate sociality—white-faced capuchin monkeys, baboons, and chimpanzees—suggest group infighting and nasty politics are pervasive, omnipresent, and, sometimes, deadly. Even positive, affiliative social interactions—friendships and consort-ships—often take the form of alliances to ward off aggressive competitors. Sometimes, backstabbing occurs. Many social animals may be selected to exhibit what is referred to as Machiavellian intelligence regarding group dynamics: the ability to interpret, anticipate, and counter enemy advances, and the ability to outwit rivals for personal gain. If social brains are big brains, they may result from social conflict. 

Social insects, such as paper wasps, ants, and honey bees, provide some of the premier examples of group-living animals. Do social insects exhibit social brain effects? My lab used a family of wasps, the Vespidae, to explore whether social brain predictions held for insects. Indeed, they did, but the brain/behavior patterns were strikingly different from those seen among so many kinds of vertebrates.

We compared the relative size of a key brain region, called the mushroom bodies, among vespid wasp species ranging from solitary nesters (potter wasps, Eumeninae), to those with small, simple societies (Polistes, with their small nests occupied by a few dozen adults), to some of the largest and most complex societies known (swarm-founding species of Agelaia and Polybia, with colony populations in the thousands or tens of thousands). The mushroom bodies (named for their stalked and capped shape) are an insect forebrain region involved in learning, memory, and sensory integration. We found the mushroom bodies of solitary species were the largest relative to brain size, exactly the opposite pattern of brain investment seen among most vertebrates. There was a steep drop-off in relative mushroom body size going from the solitary to the social species. Curiously, further changes in wasp social structure, such as colony size and the degree of queen/worker differentiation, were not associated with further changes in mushroom body investment. The jump from solitary to social was the main factor that predicted decreased brain investment, suggesting the origin of sociality was an important cognitive transition.

Why are social brain effects in insects so different from vertebrates? I believe the key lies in the ways social insect colonies form and grow. Insect colonies, even the largest and most complex, are almost always families. Insect colonies comprise a parent (in the wasp and ant Order, the Hymenoptera, a queen or several related queens) and her offspring. Although there is often conflict within insect colonies, frequently over who gets to reproduce, reproduction is generally highly restricted to one or a very few group members. Rather than accumulating an increasing array of potential rivals, insect colonies grow by adding more close relatives. Although not perfectly aligned, evolutionary interests within insect colonies probably overlap more than they do within most vertebrate societies. If natural selection operates on the colony, such as when some colonies have features that allow them to out-compete or out-produce other colonies, then colony social integration and coordination can be favored. Colony integration is achieved by information sharing and communication among colony mates. To the extent the members of a colo-ny can rely on shared information to guide their decisions and activities, their personal cognitive demands decrease. An individual can rely on her nest mates to get things right. When cognition is distributed across the society, individual brain investment can decrease. This is only possible to the extent that group mates are collaborators with aligned interests and goals, rather than rivals.

Insect sociality appears to have permitted species-wide brain reductions via distributed cognition. We can take this idea further and ask whether the allocation of brain tissue within societies shows similar evidence of colony-level adaptation. Maybe not all social brains are created equal. If some members within a society reliably face weaker cognitive demands, those individuals can afford to have even further reductions in brain investment. This could benefit the colony by saving investment in expen-sive brain tissue; the costs of impaired cognition would be minimized because cognition is distributed among the colony members.

A compelling test case is offered by the two-species integrated societies of slave-making ants. Slave-making ants in the genus Polyergus are highly specialized social parasites. The slave-maker workers are fighting machines; their mouthparts are elongated, piercing tines that are excellent for dismembering other ants in fights, but nearly useless for performing most day-to-day ant work. Polyergus workers cannot collect food, excavate nests, care for offspring, or even feed themselves, but they can and do fight and kill other ants. Polyergus colonies can only succeed by raiding and stealing immature ants (pupae) from host ant colonies in the related genus Formica. The stolen Formica workers that emerge in a Polyergus nest work for the slave-makers and carry out all normal ant nest activities. I predicted the reduced task repertoire and specialized behavior of Polyergus workers would permit a reduction of brain investment in the slave-makers. Collaborating with Joseph Sapp, at the University of California-Santa Cruz, my lab compared the relative mush-room body investment of Polyergus workers with their enslaved Formica workers and with free-living Formica from the colonies they had raided. The free-living and enslaved Formica both perform normal ant workloads, and their brains did not differ. Yet as predicted, relative mushroom body size was reduced, on average about 15 percent lower, in the slave-making Polyergus.

A Polyergus breviceps worker returning from a raid of a Formica fusca group nest mound.


My newest study on brain investment shows that similar trends can hold within colonies of a single species. In the army ant genus Eciton, the soldiers are a unique class of workers that are among the most behaviorally and morphologically specialized among social insects. Eciton soldiers specialize in colony defense, particularly against vertebrates. I can personally attest that their bites and stings are painful, and their mandibles are exceptionally difficult to remove from skin; Eciton soldiers are suicidal in defense of their colony. Much like Polyergus workers, Eciton army ant soldiers have long, ice tong-like mandibles and they are incapable of caring for offspring, hunting or carrying prey, and (again) they cannot feed themselves. As in Polyergus workers, Eciton soldiers have reduced investment in the mushroom bodies, and in a major chemical-sensing brain region (the antennal lobes). The reduced job repertoire of soldiers appears to have permitted reduced brain tissue investment in that class of workers. Biologist James F. A. Traniello and colleagues at Boston University found a similar pattern when comparing large– and small-bodied workers of distantly-related ants in the genus Pheidole. Behavioral challenges match individual worker brain investment in a wide array of social insects.

Are there lessons here for human brain evolution? Perhaps. Although human brains are exceptionally large, there is some evidence that body-size corrected brain volume has declined during recent human evolution. Over the past 20,000 years, human brains have lost about the equivalent of a tennis ball’s worth of volume on average. Although the interpretation of these findings is controversial, I believe it is plausible that the advent of increased communication and information transmission via abstract symbols, in other words human language, led to a form of cognitive offloading. The demand for individual cognitive capacity may have relaxed somewhat. If so, individual investment in brain tissue would be expected to decrease. Does this mean we have lost cognitive ability—have we “dumbed ourselves down”? In some ways, at the individu-al level, possibly so. But I suspect not overall. If anything, it seems clear that the information available to favor our survival and reproduction has increased. Humans, like no other species, are relying less on our physical organic brains and more on external, collectively shared cognition. We are engaged in an unprecedented ongoing experiment in cognition that may be shaping our brain evolution.