Cerebrum Article

Prancing Primates, Turtle with Toys

It’s More Than Just (Animal) Play

Human offspring aren’t the only ones who love to play. In the struggle for survival, why did evolution favor play in species from rats to ravens? Based on new research about the connections between brain development and play, behavioral biologist Lee Alan Dugatkin puts some hypotheses to the test. Why do young ravens play with virtually any new kind of object they encounter but adult ravens fear anything new? Why do fierce older chimps let their little brothers win the tussle? The answers, writes Dugatkin, show that play is serious brain business, both in humans and in other species.

Published: July 1, 2002

Happiness is never better exhibited than by young animals, such as puppies, kittens, lambs, and company, when playing together, like our own children. —Charles Darwin, The Descent of Man, 1872

Animals young and old play. Take Pigface at the Washington Zoo. He was in the habit of clawing his own limbs and neck, causing infections and fungal growth. When new objects were put in his otherwise bland environment, Pigface was diverted and began to engage in “object play.” Brown balls, orange balls, hoops: Pigface would approach them, follow them, push them around—just what you do when you play with something new.

Two things make Pigface’s story compelling. First, Pigface is not a dog or a chimp—the kind of animal we might expect to play—but a turtle. Second, once the balls and hoops were introduced and Pigface opted for play, rather than self-destructive behavior, his health improved. The power of play, even in a turtle, it seems, can be profound.

All I must do to remind myself of the importance of play in human life is to walk through my front door. My eight-year-old son will try to convince me that we need to wrestle—just because it’s fun. Children like all kinds of play, and boys, particularly, are fond of rough-and-tumble games. But adults do not abandon the pleasures of play. Sports and leisure activities are multibillion-dollar industries, and many people drudge through their work week with weekend “play time” as the carrot that keeps them going. Lack of play affects our very psyche. Our brains seem to crave it.

Scientists seek the roots of human play and clues to its importance through studying animal behavior and evolutionary biology. We can use what we learn about play in many different species to build generalizations. For example, only by studying play in various primate species do we see that certain types of play are related to the size of the amygdala in the limbic area of the brain.1 First, though, we need a definition of play.

In his 1975 classic, Sociobiology, E.O. Wilson complained that “no behavioral concept has proved more ill-defined, elusive, controversial and even unfashionable than play.” Definitions are hard to formulate when they focus on something as subjective—as much a matter of intention or attitude—as play. Can we remove that subjective element? A widely cited description of play by Marc Bekoff and John Byers is “all motor activity performed postnatally that appears to be purposeless, in which motor patterns from other contexts may often be used in modified forms and altered temporal sequencing. If the activity is directed toward another living being, it is called social play.”

But Bekoff himself recently observed one problem: Behaviors no one would truly consider play, such as repetitive pacing, satisfy the definition. Moreover, it is notoriously difficult to ascertain whether a behavior is purposeless. Where behavioral scientists work hard enough, and come to understand a phenomenon well enough, we almost always discern a function for it. In the case of play, for example, even if there is no immediate benefit, we may discover that the purpose or potential benefit can be seen later. Nor does Bekoff and Byers’ definition claim that play is actually purposeless—only that it appears to be so.

Where does this leave us? It seems that those who study play take a similar tack to Supreme Court Justice Potter Steward, who said in 1964 “I shall not today attempt further to define [pornography]… but I know it when I see it.” In Animal Play: Evolutionary, Comparative and Ecological Perspectives, a 1998 collection of essays on play behavior in animals, many of the authors argued that since experimental work on play lags behind research in other areas of animal behavior, a wait-and-see approach is reasonable. My hope is that the study of other species will in time uncover commonalities in play behavior that can be used to construct a precise statement of its meaning, and so help us better understand this fascinating phenomenon.

Rehearsing Ravens: Object Play

Even absent a watertight definition, we can study the different kinds of behavior throughout an individual’s development that seem to exhibit the basic characteristics of play. Researchers studying play behavior in animals see three broad types: object, locomotor, and social.

Object play centers on interaction with inanimate objects—sticks, rocks, leaves, feathers, fruit, or things provided by humans—that an animal pushes, throws, tears, or manipulates in some way. Such play has been documented in a wide array of animals and is particularly well studied in zoos, where playthings are used to introduce novelty into an animal’s otherwise relatively constant environment (remember Pigface). Object play is often distinguished from object exploration, in which animals appear to be examining what an object does. Animals engaged in object play appear to be addressing the question, “What can I do with this object?”2

In his delightful Mind of the Raven, Bernd Heinrich builds a case that natural selection in ravens has led to a brain capable of some of the most sophisticated behavior this side of humans. That may be why play occupies an inordinate amount of a raven’s time—far more than one might predict from the avian behavioral literature. Young ravens play with virtually any new kind of object that they encounter: leaves, twigs, pebbles, bottle caps, seashells, glass fragments, and inedible berries. Heinrich, who has studied these birds for thousands of hours, describes young ravens’ almost obsessive drive to touch, manipulate, and play with literally anything.

Yet this fondness of play raises the antennae of “adaptationists,” who immediately want to know what function play has in raven life. Do the birds benefit from it?

If not, why hasn’t natural selection weeded play out of their behavioral repertoire?

Well, object play in young ravens seems to help determine what they will fear, or trust, when they mature. Adult ravens continue to manipulate objects, but when they approach an item, such as a new food source, that they have not encountered when younger, they display great trepidation.

Frolicking Fawns: Locomotor Play

Locomotor play—leaps, jumps, twists, shakes, whirls, somersaults, and similar physical actions—has been studied in rodents, primates, and ungulates (hoofed animals). Young animals, in particular, appear to human observers to enjoy their frequent frolicking. Several hypotheses have been advanced to explain the function of locomotor play, both in animals and in humans. Play both provides exercise and trains specific motor skills that will be needed later in life. It also gives animals a better understanding of the lay of the land in which they have to survive.

John Byers and Curt Walker reviewed 19 potential anatomical and physiological benefits of motor play.3 One of these was cerebellar synapse distribution. The cerebellum in the brain provides critical coordination related to the limbs, smooth movement, postural changes, eye-limb coordination, and other movements in mammals. During development, more cerebellar synapses are created than an animal uses in later life, and the number of these synapses appears to be a function of experience. The question, then, is how well play behavior correlates with formation of synapses. In juvenile mice, the match-up is impressive. Mice start playing at about 15 days of age and peak in their locomotor play activities at days 19 to 25, which corresponds nicely to the pattern of synapse formation.

Does play spur cerebellar synaptogenesis or vice versa? The relationship between play and the growth of synapses in the brain is certainly ripe for exploration. What is more, another major developmental change—the differentiation of muscle fibers into “fast” or “slow” fibers—also correlates nicely with the development of motor play.

Wrestling Rhesus Monkeys: Social Play

Social play, our last category, is perhaps the best-studied type. Three possible functions have received the most attention: honing physical skills, forging long-lasting social bonds, and fostering cognitive skills.

Scientists studying primates and carnivores have speculated in particular about the cognitive benefits of social play. Does social play in immature chimpanzees provide young males with the cognitive skills for coalition formation important in adult life? Another possible cognitive benefit of social play has to do with self-assessment. Infant antelopes prefer same-aged play partners4 and, while this preference could have many causes, there is evidence suggesting that the infants choose play partners that provide them with a reasonable comparison from which to gauge their own development.

Or consider play fighting in squirrel monkeys, for which three possible cognitive benefits have been suggested.5 Play fighting exacts few real costs and may be ideal for learning behavioral flexibility. Since real fighting in adult life is potentially dangerous, play fighting may also train male squirrel monkeys to gauge the intentions of others. In addition, they must work their way up a dominance hierarchy, losing as well as winning many encounters, so they develop experience in both subordinate and dominant roles—roles that they will experience elsewhere in life.

The description of one kind of social play, “rough and tumble,” comes from Harry Harlow’s phrase for this behavior in rhesus monkeys, and was first applied to humans by Nicholas Blurton Jones in his early work on the behavior of nursery school children. Be it wrestling, jumping, rolling, swinging, or chasing, rough-andtumble play is part of a child’s life. Possible explanations for it are similar to those put forward for primates and carnivores. Children’s rough-and-tumble play might function as general physical training (as with locomotor play) or prepare them for serious adult activities like fighting and hunting, both critical in human evolution. Perhaps rough-and-tumble play enables young children to gauge their own strength and potential power, or promotes coalition building skills—again, an important factor in our own evolutionary history.

Knowing When Play Is Play

Recently, Bekoff put forth a novel hypothesis that ties play and cognition to the origins of morality. He started by asking how animals, especially when young, know that they are engaged in play. More to the point, how do they communicate this to each other? Since many behaviors seen during play are associated, in other venues, with aggression, hunting, or mating, animals must have a way to know that play is not the real thing. How has the brain evolved a mechanism for separating the innocence associated with play from the seriousness of survival?

Bekoff posits, first of all, that animals may distinguish play from related activities because the order and frequency of behavioral components in play are often quite different from those in “real” activity. For instance, behavioral patterns during play are often exaggerated and misplaced. In canid play, exaggerated paw slaps and sexual mounts will often break up a sequence of actions normally associated with hunting. If young animals are able to distinguish these exaggerations and misorderings of behavioral patterns by, for example, observing adults not involved in play, there would be a relatively simple explanation for how they know they are playing.

According to Bekoff, animals also may be able to distinguish play from other activities through “play markers.” In dogs and foxes, for example, biting or shaking is usually performed during dangerous activities such as fighting and hunting. Biting and shaking are also part of young canids’ play, but behavioral markers, such as a bow, are made first to communicate that this is play.

Young animals may also use role reversal or self-handicapping on the part of older playmates to distinguish play from related behaviors. Here, older animals may allow subordinate younger ones to assume the dominant role or may perform some act, such as biting or chasing, with a skill or intensity clearly inferior to their capability. Either gambit affords younger playmates the opportunity to recognize that they are involved in a playful encounter.

Where there are games, though, there may be cheating. If animals know they are engaged in play, why do they play fairly? Why not sneak in a seriously aggressive blow once in awhile, instead of playing fairly in rough-and-tumble play? One likely answer is that cheating at play does not net an animal much real benefit and might also lead to being ostracized. If the benefits of social play are great enough, then ostracism could significantly hurt the cheater.

Can we extrapolate from social play in mammals to morality in humans? If we accept that morality is a set of rules designed to ensure fairness at some level, then Bekoff may not be so far afield when he argues that “during social play, while individuals are having fun in a relatively safe environment, they learn ground rules that are acceptable to others—how hard they can bite, how roughly they can interact— and how to resolve conflicts. There is a premium on playing fairly and trusting others to do so as well. What could be a better atmosphere in which to learn social skills than during social play, where there are few penalties for transgressions? Individuals might also generalize codes of conduct learned in playing with specific individuals to other group members and to other situations.6

The Playing Brain

Most researchers of play agree that it involves some of the most complex behavioral patterns, including motor patterns, that young animals display. An understanding of the neurobiology (and neurochemistry) that underlie play is of help.

Broadly speaking, neuroethologists (those who study the neurological basis for the behavior of animals in their natural environments) use two basic techniques to study play behavior. In the first, they target neurotransmitters in the brain to discover their role in play; in the second, they examine the neural pathways involved with a particular form of play. In examining neurotransmitters that may have a role in play, investigators administer a neuroactive compound that either blocks or enhances a specific, known neurotransmitter. Repeating this process with enough transmitters begins to fill in a broad picture of the neurochemistry of play. For example, three neurotransmitter systems (those for dopamine, norepinephrine and serotonin, and opiates) appear to be involved in rat play fighting. Dopamine inhibitors, for example, typically reduce play.

Things may be a bit more subtle than that, however; some researchers argue that dopamine’s function is to invigorate or “prime” an animal’s preparation for play, not to control the play itself. Rats can be trained to anticipate play, making it possible to examine directly whether anticipation of play correlates with changes in dopamine levels. Stephen Siviy constructed an experimental apparatus for rats that consisted of two chambers connected by a tube. After counting the number of times a rat crossed the tube during five minutes, Siviy placed the rats into one of two treatments: play and no-play. In the first, rats were allowed to play with another rat for five minutes. In the second, rats had five minutes more alone in the apparatus, making social play impossible. Total social play time outside the experimental apparatus was equal.

The rats in Siviy’s “play treatment” crossed back and forth in the tube far more frequently before their partner was placed in the experimental apparatus than did rats in the no-play treatment. Does this mean that rats in the play treatment anticipated the opportunity for play and searched it out by means of their increased number of crossings? In a further step, half the rats in Siviy’s experiment were given a dopamine-inhibitor drug. These rats reduced their tunnel crossing “anticipatory behavior” significantly, but their play behavior, once a partner was present, remained level. This supports the interpretation that dopamine acts to increase the anticipation of play rather than play behavior itself.

What neural pathways underlie play? In the past, one way to find out was to damage various areas of the brain and see the effect on play. While this technique has had some limited success, modern biochemical tools should eventually give us a far better window into the neurobiology of play. Some of Siviy’s earlier work showed that lesions to the parafascicular area (PFA) of the rat’s brain reduced play fighting. He then began to search for a neural pathway in the PFA that would underlie play. If the PFA was critical to play, then cells in this portion of the rat brain should be very active during play. Such activity can be measured by quantifying the amount of a protein product (associated with the c-Fos gene) found in the PFA. Rats that had just been involved in play had much higher neural activity in the PFA than did control rats. Somewhat surprisingly, however, this increased neural activity was not limited to the PFA, but was found in other areas of the brain (the cortex and hypothalamus) as well. So play requires the use of much of a rat’s brain, not just its PFA.

Siviy offers two hypotheses for the functions of play. First, it may be a mechanism for coping with stress, because the same neurotransmitters (dopamine, norepinephrine, and serotonin) that are involved in play behavior are also critical in the animal’s stress responses. Second, play may facilitate learning and creativity. The lack of a precise neural circuit for play is frustrating for some, but since both learning and creativity appear to be driven by general brain activity levels, the nonspecificity of the neural circuitry of play suggests a potential link between play and learning. In fact, the transcription of the c-Fos (and related) genes that were shown to be important for play in Siviy’s studies of rats is also an important component of the molecular machinery associated with learning.

Increased creativity also may be a benefit of play, at least in humans. When asked to describe novel uses of an object, young children who had been allowed to play with new objects were much more creative in their suggestions than a comparison group of children who had been instructed by adults on how to use the new object.

Preparing for the Unexpected

So we come back to our initial question: Why play? Marek Spinka and his colleagues have recently suggested a general theory about how play functions in mammals, including humans.7 They argue that all the functions of play that have been suggested by various researchers are consistent with a more global explanation that play helps animals acquire the physical and psychological skills needed to handle unexpected events in which they experience a loss of control. Specifically, they propose that “play functions to increase the versatility of movements used to recover from sudden shocks such as loss of balance and falling over, and to enhance the ability of animals to cope emotionally with unexpected stressful situations.” For example, the loss of control and balance associated with being chased by predators, or the emotions associated with losing an aggressive interaction, may be dealt with more effectively if play has prepared the animal for such eventualities. Play, in short, has prepared the brain to handle the unexpected.

Spinka puts forward a novel set of predictions, commendably laid out in a way amenable to testing his hypothesis. If it is correct, for example, the amount of play experienced should affect an animal’s ability to handle unexpected events. Some research in both humans and nonhumans supports this correlation. Rats, for example, that have been deprived of social play often react more aversely to unexpected stimuli,8 while, in humans, measures of rough-and-tumble play correlate with scores on social problem-solving tests.

If play does help prepare an animal for future events, it should have measurable effects on an animal’s somatosensory, motor, and emotion brain centers. During play, Spinka argues, the brain must deal with sensory inputs in a different way than in other behaviors, and the uncertainty created by these inputs “must be solved by kinematic improvisation and emotional flexibility.” In support of this prediction, rats that have been deprived of social play have long-term changes in certain opiod receptors in the brain, as well as permanently altered levels of dopamine and other neurotransmitters,9 all of which are important for coordinating response to stress.

According to Spinka’s hypothesis, any differences in play behavior that exist between the sexes should increase as individuals mature. The logic is that if between-sex differences in encountering the unexpected do exist, and if play behavior prepares one for the unexpected, then between-sex differences in play should be magnified over time just because the between-sex differences in encountering the unexpected (fights, predators) are increasing. This prediction gains support from data on sea lions, pigs, chimpanzees, rats, and cattle. Another prediction, supported by evidence in rats, dogs, cats, seals, and many other species, is that self-handicapping, in which dominant animals allow subordinates to defeat them during play fights, should be ubiquitous in species that play because it is an excellent means for preparing for the unexpected.

In addition, Spinka predicts that locomotor play should be most common in species that live in the most variable environments. If locomotor play enables an animal to experience loss of physical control, this would be most beneficial to animals in the environments that change most rapidly. More generally, environmental change should favor most types of play, as play is predicted to enable individuals to handle the unexpected, and the unexpected should be more common in changing environments. This interesting prediction, however, has not yet been tested by researchers.

The Serious Business of Play

Thanks to scientists in evolutionary biology, psychology, neuroethology, and other disciplines, things have improved markedly since 1975, when E.O. Wilson lamented that “no behavioral concept has proved more ill-defined, elusive, controversial and even unfashionable than play.” Research continues apace and, although a precise definition of play eludes us, we see an ever-more-complete picture of a fundamentally important activity with implications for the evolution of behavior and make-up of the animal brain. In the potential adaptiveness of play, and its neurobiological underpinnings, we discern how a fascinating feature of animal life may have come into existence and why it has persisted. I must remind myself of that the next time I tell my son not to play with his food.


  1. Pellis, S, and Iwaniuk, AN. “Brain system size and adult-adult play in primates: a comparative analysis of the roles of the non-visual neocortex and the amygdala.” Behavioural Brain Research, in press.
  2. Power, TG. Play and Exploration in Children and Animals. New York. Lawrence Erlbaum Assoc, 2000.
  3. Byers, J, and Walker, C. “Refining the motor training hypothesis for the evolution of play.” American Nature 1995; 146: 25-40.
  4. Thompson, KV. “Play-partner preferences and the function of social play in infant sable antelope, Hippotragus niger.” Animal Behavior 1996; 52: 1143-1152.
  5. Biben, M. “Squirrel monkey playfighting: making the case for a cognitive function for play.” In Bekoff, M, and Byers, J, eds., Animal Play: Evolutionary, Comparative and Ecological Perspectives. Cambridge. Cambridge University Press 1995: 161-182.
  6. Bekoff, M. “Social play behavior: cooperation, fairness, trust, and the evolution of morality.” Journal of Consciousness Studies 2000; 8: 81-90.
  7. Spinka, M, Newberry, R, and Bekoff, M. “Mammalian play: training for the unexpected.” Quarterly Review of Biology 2001; 76: 141-168.
  8. Potegal, M, and Einon, D. “Aggressive behaviors in adult rats deprived of playfighting experience as juveniles.” Developmental Psychobiology 1989; 22: 159-172.
  9. van den Berg, C, van Ree, J, Spruijt, B, and Kitchen, I. “Effects of juvenile isolation and morphine treatment on social interactions and opiod receptors in adult rats: behavioural and autoradiographic studies.” European Journal of Neuroscience 1999; 11: 3023-3032.