Play is universal

Chapter 1The Universality of Play

On the cross-species, cross-cultural persistence of play - the archaeological record of games - and why the existence of games is a phenomenon that requires explanation rather than a trivial consequence of culture.

1.1 The problem: games exist everywhere

There is a simple observation that, once stated plainly, becomes difficult to ignore: every known human culture plays games.

This is not a statement about modern entertainment industries or digital media. It is a statement about human behaviour across time and geography. Archaeological evidence reveals structured games in ancient Mesopotamia (the Royal Game of Ur), Egypt (Senet), India (Chaturanga), China (Go), Mesoamerica (ballgames), and medieval Europe (Chess). These systems differ in surface features - materials, rules, symbolic meaning - but share a deeper structural property: they are rule-bound environments in which agents act under uncertainty toward defined outcomes.

More strikingly, these systems arise independently. Chess is not a derivative of Go; the Royal Game of Ur is not a precursor to Mesoamerican ballgames. The recurrence of games across cultures is not explained by diffusion alone. It reflects repeated invention.

The implication is immediate. If a behaviour appears:

• across geographically isolated populations
• across historical epochs
• in structurally similar forms

then it is unlikely to be arbitrary. It is more plausibly the expression of an underlying constraint in human cognition.

Games, in this sense, are not merely cultural artifacts. They are solutions to a problem we have not yet fully specified.

1.2 Play is older than culture

The anthropological universality of games becomes more compelling when placed in a broader biological context: play predates humanity.

All four great apes - chimpanzees, bonobos, gorillas, and orangutans - engage in structured play behaviours. These include:

• Rough-and-tumble play, involving mock aggression without escalation
• Role reversal, where dominant individuals temporarily adopt subordinate positions
• Self-handicapping, where stronger individuals deliberately limit their advantage
• Teasing and provocation, including behaviours that appear to anticipate reactions

A landmark 2024 study by Laumer, Winkler, Rossano, and Cartmill (Proceedings of the Royal Society B) documented spontaneous playful teasing in all four great ape species, involving intentionally provocative, one-sided behaviours that include elements of surprise. This suggests that the cognitive prerequisites for humour - expectation, violation, and resolution - were present at least 13 million years ago in the hominoid lineage. Laughter-like vocalisations show homologous acoustic structures across great apes, reinforcing the evolutionary continuity of play-related affect.

These behaviours are not random. They are rule-governed in an implicit sense. Play fighting, for example, follows constraints:

• Biting is inhibited
• Signals (play faces, vocalisations) indicate non-serious intent
• Violations of these constraints terminate the interaction

In other words, even in non-human animals, play exhibits the core features of games:

• Bounded interaction spaces
• Implicit rules
• Uncertain outcomes
• Repeated engagement for its own sake

Jane Goodall's foundational observations at Gombe, published in The Chimpanzees of Gombe (1986), documented play across all age classes of wild chimpanzees. A subsequent analysis of 33 years of Gombe data by Lonsdorf and colleagues (2017) demonstrated that infant chimpanzees who spent more time in social play achieved motor and social milestones - independent travel, first grooming of non-maternal kin, first mating attempts - at significantly earlier ages. This was the first study in great apes linking play directly to developmental outcomes.

The conclusion is difficult to avoid. What we call "games" in human culture are not inventions ex nihilo. They are formal elaborations of a behavioural system that predates culture itself.

1.3 From play to games: formalisation and abstraction

If play is the biological substrate, games are its cultural crystallisation.

The transition from play to games involves three transformations.

First, constraint becomes explicit. In animal play, rules are implicit and enforced through social signalling. In human games, rules become externalised and formalised. "Do not bite" becomes "no fouls." "Take turns" becomes turn-based systems. Spatial boundaries become boards, fields, or arenas. This externalisation allows rules to be transmitted across individuals, stabilised across generations, and modified deliberately.

Second, uncertainty becomes structured. Play contains uncertainty, but games engineer it. Dice introduce stochastic outcomes. Hidden information creates informational asymmetry. Skill-based systems create performance uncertainty. The result is not mere unpredictability, but controlled unpredictability; uncertainty that is neither trivial nor overwhelming.

Third, interaction becomes symbolic. Games detach from immediate physical reality and operate in abstract spaces. Stones on a board represent territory (Go). Pieces represent hierarchical power (Chess). Cards represent probabilistic distributions (poker). This abstraction allows games to scale in complexity far beyond what is possible in raw play behaviour.

The archaeological record documents this transition with striking clarity. Senet, the most iconic ancient Egyptian game, appears in fragmentary form in First Dynasty burials at Abu Rawash around 3100 BCE. Played on a grid of 30 squares with casting sticks determining movement, it began as secular entertainment but by the New Kingdom (c. 1550-1077 BCE) had acquired profound religious significance, appearing in Chapter 17 of the Book of the Dead as a representation of the soul's journey through the netherworld. Peter Piccione's 1990 University of Chicago dissertation traced this evolution in detail. Four senet boards were found in Tutankhamun's tomb; the game was played across all social classes, from elaborate inlaid boards to grids scratched into stone floors.

The Royal Game of Ur, discovered by Leonard Woolley during excavation of the Royal Cemetery at Ur between 1922 and 1934, dates to approximately 2600-2400 BCE. Irving Finkel of the British Museum reconstructed its basic rules from a cuneiform tablet written by the scribe Itti-Marduk-balāṭu in 177-176 BCE; a race game combining strategy and luck, broadly similar to backgammon. The earliest possible game board, found at Ain Ghazal in modern

Jordan, dates to approximately 5870 BCE.

1.4 Independent invention and convergent structure

One of the most important features of games is not their diversity, but their convergence.

Across cultures, we repeatedly observe similar structural solutions:

• Grid-based territorial games (Go, various indigenous strategy games)
• Race games involving probabilistic movement (Senet, Royal Game of Ur, Pachisi)
• Abstract strategy games with hierarchical pieces (Chess and its relatives)
• Physical competition games with bounded arenas (ball sports across continents)

These are not arbitrary categories. They represent distinct ways of structuring uncertainty and action: spatial control, temporal progression, hierarchical interaction, and motor skill execution. The recurrence of these forms suggests that there are only a limited number of ways to construct engaging systems under the constraints of human cognition. Cultures discover these solutions independently because they are locally optimal in a shared design space.

Roberts, Arth, and Bush's foundational cross-cultural study "Games in Culture" (1959, American Anthropologist) established the standard tripartite classification: games of physical skill, games of strategy, and games of chance. Analysing 50 societies, they found that games of strategy correlated with social stratification, while games of chance correlated with beliefs about supernatural controllability. Games, they argued, function as "expressive models" of real-world activities: physical skill models hunting and combat, strategy models political competition, and chance models the unpredictability of nature.

Certain children's games appear to be cross-cultural universals: hide-and-seek, tag and chase games, wrestling, and catch appear across societies with no historical connection, from Aboriginal Australia to the Arctic to sub-Saharan Africa to South America. Evolutionary psychologists interpret this convergence as evidence of deep biological function; these games simulate predator avoidance, terrain navigation, and emotional regulation under stress.

1.5 The invariants of games

Despite surface variation, games share a set of invariant properties:

1. Rules. Games define constraints on action. These constraints create the possibility space within which behaviour occurs.
2. Agency. Players can act within the system. Outcomes are not purely observational.
3. Uncertainty. The outcome is not fully determined in advance. This may arise from randomness, hidden information, or skill.
4. Feedback. Actions produce discernible consequences. The system responds to the player.
5. Autotelic engagement. The activity is performed for its own sake, not merely for external reward.

These properties are not definitional conveniences. They are necessary conditions. Remove any one of them and the system ceases to function as a game:

• Remove uncertainty and the system becomes trivial
• Remove agency and it becomes spectacle
• Remove feedback and it becomes opaque
• Remove rules and it becomes unstructured play

The persistence of these invariants across cultures reinforces the central claim: games are not arbitrary constructions. They are constrained systems shaped by the architecture of the human mind.

1.6 The inadequacy of cultural explanations

A common response to the universality of games is to treat them as cultural byproducts: humans enjoy competition, humans enjoy storytelling, humans enjoy social interaction. Games, on this view, are composites of these tendencies.

This explanation is insufficient for two reasons.

First, it is descriptive rather than explanatory. It lists associated features without identifying the mechanism that makes games compelling.

Second, it fails to explain why games take the specific forms they do. Competition, for example, does not require rule-bound systems with structured uncertainty. Social interaction does not require abstract symbolic spaces. The particular architecture of games - rules, uncertainty, feedback - remains unexplained.

To explain games, we must move beyond surface motivations and ask a deeper question:

What kind of system does the brain find intrinsically engaging, and why?

1.7 Games as a problem in cognitive science

The existence of games poses a problem that sits at the intersection of anthropology, neuroscience, and computation.

We have established three facts:

1. Play is evolutionarily ancient, present across mammalian species and especially developed in primates.
2. Games are culturally universal, appearing independently across human societies.
3. Games share invariant structural properties, suggesting common underlying constraints.

These facts demand a unifying explanation.

The hypothesis that will guide the remainder of this book is the following:

Games are structured environments that exploit fundamental properties of the brain's learning and uncertainty-processing systems.

This is not yet a theory. It is a direction. To make it precise, we must understand how the brain processes uncertainty, how it updates its internal models of the world, and why certain patterns of interaction feel intrinsically rewarding. Only then can we explain not just why games exist, but why some games succeed where others fail.

1.8 Transition

Anthropology establishes that games are universal. Evolutionary biology suggests that play is functional. But neither tells us how games work at the level of mechanism.

To answer that, we must descend a level; from behaviour to biology.

The next chapter examines the evolutionary function of play and the neural systems that support it, beginning with a claim that will become central:

Play is not a luxury. It is a biological system for training the organism to act under uncertainty.

Chapter 2The Evolutionary Function of Play

On play as a biological system for training organisms to act under uncertainty - the neural circuits that generate it - and why its persistence across species implies function rather than frivolity.

2.1 Play is not optional

If a behaviour is metabolically expensive, developmentally prolonged, and evolutionarily conserved, it is unlikely to be incidental.

Play meets all three criteria.

• It consumes time and energy without immediate survival payoff
• It appears most intensely during development, when resources are most constrained
• It is present across mammalian species, with especially elaborate forms in primates

From an evolutionary perspective, this is paradoxical. Natural selection is not in the business of maintaining costly behaviours without function. The existence of play therefore demands explanation.

The most parsimonious hypothesis is not that play is leisure, but that it is work of a different kind:

Play is a training system.

The question is: training for what?

2.2 The structure of play behaviour

Across species, play exhibits a remarkably consistent structure.

Consider rough-and-tumble play in juvenile mammals. Individuals engage in mock aggression: they pursue, grapple, and attempt to dominate. Yet critical constraints are enforced: bites are inhibited, vulnerable positions are tolerated, and encounters terminate before escalation. The behaviour contains real motor patterns - chasing, striking, balancing - but stripped of lethal consequences.

Similarly, in object play, animals manipulate objects in ways that resemble foraging or tool use. Actions are repeated, varied, and recombined. Outcomes are explored rather than optimised. And in social play, individuals test boundaries: they provoke, retreat, re-engage. They learn the contingencies of interaction.

Across all forms, the same pattern appears:

• Action under uncertainty
• Immediate feedback
• Low cost of failure
• Repetition with variation

This is not random behaviour. It is a structured exploration of possibility space.

2.3 The PLAY system

Jaak Panksepp's work on affective neuroscience provides the clearest biological account of play.

He identified a primary emotional system - alongside SEEKING, FEAR, RAGE, CARE, LUST, and PANIC/GRIEF - dedicated specifically to play: the PLAY system. His landmark book Affective Neuroscience (Oxford University Press, 1998) and its successor The Archaeology of Mind (2012, with Lucy Biven) mapped these systems using electrical brain stimulation, pharmacological challenges, and lesion studies. Critically, all seven systems arise from subcortical brain regions homologous across mammals. They do not require the neocortex: decorticated rats still play, and hydranencephalic children - born without a cerebral cortex - still laugh and smile (Merker, 2007).

Key properties of the PLAY system:

• It is subcortical, originating in ancient brain structures
• It is intrinsically motivated; play is pursued for its own sake
• It is socially modulated, especially in mammals
• It exhibits homeostatic rebound: deprivation increases subsequent play intensity

Panksepp's most celebrated experiments involved tickling rats. In publications from 2000 to 2003 (Behavioural Brain Research, Physiology & Behavior), he demonstrated that rats emit 50-kHz ultrasonic vocalizations during rough-and-tumble play and when tickled by human hands at the nape of the neck; the same area targeted during conspecific play. These frequency-modulated calls, proposed as a homologue of human laughter, are positively correlated with the rewarding value of tickle stimulation, can induce approach behaviour in other rats when played back (Wöhr & Schwarting, 2007), and are distinct from the 22-kHz aversive calls emitted during distress. Rats selectively bred for high rates of 50-kHz vocalizations showed increased playfulness (Burgdorf et al., 2005, Behavior Genetics).

That play is a primary drive - not learned, not secondary - is supported by several converging lines of evidence. Play is tightly homeostatic: its amount can be titrated by varying social isolation duration, with longer deprivation producing rebound increases. Rats develop conditioned place preferences for play-associated environments and perform operant responses to obtain play opportunities. Neonatal decortication does not abolish play (Panksepp et al., 1994), confirming its subcortical origin. Conversely, frontal lobe lesions increase playfulness (Panksepp et al., 2003), suggesting the maturing cortex normally inhibits subcortical play urges.

The neurochemical basis involves four interacting systems:

• Dopamine drives approach and engagement (motivational "wanting")
• Opioids mediate hedonic pleasure (hedonic "liking")
• Endocannabinoids enhance positive affect and flexibility
• Oxytocin supports social bonding

The neural architecture involves a distributed circuit centring on subcortical structures. Siviy and Panksepp (2011, Neuroscience & Biobehavioral Reviews) proposed a primary-process executive circuit including the parafascicular area of the thalamus (PFA), the dorsal and ventral striatum, and the frontal cortex, with contributions from the amygdala, periaqueductal grey, and ascending dopamine systems. The PFA appears to be a critical integration node: electrolytic lesions of the PFA reduced play pinning by 73% while having minimal effects on other sensory-motor processes (Siviy & Panksepp, 1985, Behavioral Neuroscience).

The PLAY system is not a cultural overlay. It is a biological drive, as fundamental as hunger or fear.

2.4 Neuroplasticity and skill acquisition

If play is a training system, it must produce measurable changes in the brain.

It does.

Experimental studies of play deprivation provide a critical insight. Animals deprived of play during development show:

• Reduced synaptic density in prefrontal cortex
• Impaired cognitive flexibility
• Deficits in social competence
• Increased anxiety-like behaviour

Bijlsma et al. (2022) demonstrated that play-deprived rats showed reduced inhibitory synapses in prefrontal cortex and impaired cognitive flexibility. The prefrontal cortex is not required to generate play but is profoundly shaped by it. Gordon et al. (2003) showed BDNF elevation after play; BDNF is the primary growth factor supporting synaptic plasticity throughout the brain.

Conversely, play enhances:

• Brain-derived neurotrophic factor (BDNF) levels
• Synaptic plasticity
• Motor coordination
• Social prediction ability

These effects are not subtle. They indicate that play is directly involved in shaping the neural architecture required for adaptive behaviour. Stuart Brown, in his clinical work, found that 26 young male murderers had no memories of normal play; a striking if preliminary correlation between play deprivation and catastrophic failures of social calibration.

The implication is clear:

Play is not rehearsal of specific behaviours; it is training the capacity to learn, adapt, and respond to novelty.

2.5 Safe failure and error tolerance

A defining feature of play is that it allows failure without catastrophic consequence.

This is not incidental. It is the central design feature.

In non-play contexts, failure may result in injury or death; errors are costly and must be minimised. In play, failure is expected, errors are tolerated, and exploration is encouraged. This creates a unique learning environment where the organism can test hypotheses, explore edge cases, and experience near-failure states repeatedly.

From a learning perspective, this is ideal. Systems that learn through error correction require error signals, opportunity for adjustment, and repeated exposure. Play provides all three. The cost structure of play is what makes it such an effective training system: prediction errors are frequent and informative, but the consequences of getting things wrong are minimal. This maps directly onto how games function. Death in Dark Souls costs a few minutes of progress and some virtual currency. Death in real combat costs everything. The prediction error is structurally similar; the cost is incomparably different.

2.6 Uncertainty as the core variable

The common thread across all forms of play is uncertainty.

• In physical play: uncertainty of movement outcomes
• In social play: uncertainty of others' responses
• In object play: uncertainty of environmental interaction

Play is not simply activity. It is activity structured around unknowns.

Crucially, the uncertainty is:

• Non-trivial: outcomes are not predetermined
• Reducible: patterns can be learned over time

This places play in a specific regime; not pure randomness (which cannot be learned), not full determinism (which requires no learning), but environments where prediction is possible but not yet achieved. This is the regime in which learning systems operate most effectively.

2.7 Social calibration and boundary testing

In social species, play serves an additional function: calibration of interaction.

Through play, individuals learn how far they can push others, how others respond to provocation, how to signal intent, and how to repair violations. The mechanisms are subtle: a play face signals non-serious intent - overly aggressive behaviour leads to disengagement - successful interaction requires continuous adjustment. This is effectively real-time modelling of other agents. The organism is learning prediction of others' actions, adjustment of its own behaviour, and maintenance of cooperative equilibrium; precisely the skills required for navigating complex social environments.

2.8 The paradox of inefficiency

From a narrow perspective, play appears inefficient. Time spent playing is time not spent foraging. Energy is expended without immediate reward. Risks are taken without necessity.

Yet evolution has preserved it.

The resolution of this paradox lies in timescale. Play is inefficient in the short term but efficient in the long term. By investing in flexibility, adaptability, and learning capacity, the organism gains increased survival under novel conditions, improved social integration, and enhanced problem-solving ability. Play is therefore not a waste of resources. It is an investment in general competence.

Stephen Jay Gould described humans as a neotenous species; retaining juvenile characteristics including curiosity, behavioural flexibility, and attraction to novelty into adulthood. Adult play persists in every human culture through sports, board games, music, dance, humour, and creative expression. Johan Huizinga's Homo Ludens (1938) argued that culture arises in the form of play and that civilisations lose their vitality as they lose their playfulness. Six evolutionary theories have been proposed for play's function:

Groos's practice theory, Spencer's surplus energy theory, Burghardt's developmental model, Spinka et al.'s training for the unexpected, Gray's self-education hypothesis, and Bateson's creativity engine. All converge on the same core insight: play builds the organism's capacity to handle novelty.

2.9 From play to games (revisited)

We can now reinterpret games in light of play's function.

If play is a biological system for training under uncertainty, then games are artificial environments that replicate and amplify this training process. Games take the core features of play - uncertainty, feedback, safe failure, repetition - and make them more structured, more controllable, and more scalable. They are, in effect, engineered play systems.

This reframes the design problem entirely. The question is no longer "How do we make games entertaining?" It becomes:

How do we construct environments that optimally engage the brain's learning systems?

2.10 Transition

We have established that play is functional: it is biologically grounded, it shapes neural development, and it trains organisms to act under uncertainty. We have also seen that games are formalised extensions of play.

The next step is to move from function to mechanism. If play trains the organism through interaction with uncertainty, then: what neural signals track that uncertainty? What makes certain interactions rewarding? Why do some patterns sustain engagement while others do not?

To answer these questions, we must examine the brain's reward and learning systems directly.

The next chapter begins with a claim that will anchor the rest of this book:

The brain does not reward outcomes. It rewards the reduction of uncertainty.