Why Do Animals Play? Costly, Risky and Almost Everywhere

The difficulty of studying play begins with saying what it is, since it looks like fighting, hunting or mating done badly. The working definition, set out by Gordon Burghardt in 2005 and now standard, requires five criteria to be met together:

• •The behaviour is not fully functional in the form in which it appears, so a play fight does not settle a dispute and play hunting does not produce food
• •It is voluntary, spontaneous, pleasurable or in some sense rewarding, and animals will work for the opportunity
• •It differs from the serious version in being exaggerated, incomplete, repeated or reordered
• •It is repeated in a similar but not rigid form during at least part of the animal's life
• •It occurs when the animal is adequately fed, healthy and free from acute stress, which is why play disappears first when conditions deteriorate and is used as a welfare indicator

Behaviour meeting those criteria sorts into three categories that appear to serve different purposes. Locomotor play is solitary movement for its own sake: leaping, twisting, running, and the pronking of young antelope. Object play is manipulation of things, most developed in carnivores and primates and in birds of the crow and parrot families, and it shades into tool use. Social play is the most complex, involving another individual and requiring signals to establish that what follows is not an attack. Dogs use a play bow, chimpanzees a specific open-mouthed face, rats a particular nape contact, and in every case the signal is given before and repeated during the episode. Social play also requires self-handicapping, in which a larger or older animal restrains its strength and sometimes deliberately loses, and role reversal, in which the chaser becomes the chased, without which the smaller partner stops participating.

Several functions have been proposed and the evidence supports different ones in different species. The training hypothesis holds that play rehearses adult skills, which is intuitive and surprisingly hard to demonstrate, since studies that have looked for a link between play fighting and later fighting success have often failed to find one. A stronger version, training for the unexpected, argues that the point is not the specific skill but the experience of losing balance and recovering, of being in a bad position and getting out, which builds the capacity to cope with surprise; this fits the observation that play deliberately creates awkward situations that a competent animal would avoid. Brain development is a third: play is concentrated in the period when the cerebellum and cortex are maturing, and studies in rats show play experience affecting the development of the prefrontal cortex. Social competence is a fourth and the best supported for social species, since rats deprived of play with peers but otherwise raised normally grow into adults that misread social signals and respond inappropriately to both threat and friendliness.

Play is universal among mammals and common in birds, and both groups share a combination of features that seem to make it possible: parental care, a protected juvenile period, a high metabolic rate and a large brain. Beyond them it becomes patchy and interesting. Reptiles play rarely and do so when warm and well fed, with a Nile softshell turtle in a zoo documented playing with a basketball and Komodo dragons engaging in object play with their keepers. Fish have been recorded repeatedly pushing objects, including a cichlid that flipped a thermometer over hundreds of times. Octopuses, which are molluscs and share no recent ancestor with any of these groups, pass the criteria comfortably. Invertebrate play was long dismissed, and a 2022 study reported bumblebees rolling small wooden balls with no reward attached, doing it more when young, and choosing to enter a chamber where the balls were. The distribution suggests play arises wherever an animal has surplus energy, a flexible nervous system and the safety to waste both.

Play declines sharply with age in most species and does not disappear. Adult social play in primates and canids appears to maintain relationships, defuse tension and test trust, since play requires putting oneself in a vulnerable position with another individual and is therefore a reliable signal of intention. Domestication extends it: dogs play throughout life far more than wolves do, and the general retention of juvenile traits into adulthood, called neoteny, is a consistent side effect of domestication across species. Humans are the extreme case, playing lifelong and building elaborate cultural structures on it, and the argument that human culture itself grows out of play was made by Johan Huizinga in 1938. The practical application is in welfare, since the criteria make play a measurable indicator: an animal that plays is not hungry, frightened or ill, which makes its absence one of the earliest and most reliable signs that something is wrong in a zoo, a farm or a laboratory.

Play is defined by five criteria together: it is not fully functional, it is voluntary and rewarding, it differs from the serious version, it is repeated, and it occurs only when the animal is healthy and unstressed. It divides into locomotor, object and social forms, the last requiring signals, self-handicapping and role reversal. The best-supported functions are building social competence and the ability to recover from unexpected situations rather than rehearsing specific skills, and it appears in mammals, birds, some reptiles, fish, octopuses and bumblebees.