Summary
Raph Koster (A Theory of Fun for Game Design, 2005) argues that fun in games is not an aesthetic or experiential quality — it is a cognitive one. Fun is the pleasure the brain produces when it successfully recognises, internalises, and masters a new pattern. Games are, at their core, teaching machines optimised for pattern delivery.
The thesis has two immediate consequences for designers:
- Boredom is failure — a boring game has either failed to teach or finished teaching too early
- Mastery ends fun — once a player fully chunks a game’s patterns, fun is exhausted
(Koster, A Theory of Fun, see source-theory-of-fun)
The brain as pattern machine
“The brain is mostly a voracious consumer of patterns, a soft pudgy gray Pac-Man of concepts.” — Koster, Ch. 2
The human brain is primarily a pattern-recognition and pattern-compression system. It does not process raw reality — it filters input through expectations, fills in blanks with assumptions, and compresses learned sequences into automated routines.
Chunking
Chunking is the brain’s process of compressing a mastered skill or pattern into an automatic routine that no longer requires conscious attention. What began as step-by-step deliberate action — like getting dressed in the morning, or shifting gears in a car — becomes a single “chunk” that fires as a unit.
Examples:
- A beginner guitarist consciously places each finger; a skilled player chunks chord shapes as single units
- A chess novice evaluates each piece individually; a master chunks board states as recognisable patterns
- Reading: children decode each letter; adults chunk words, phrases, and sentences
The brain’s drive to chunk is relentless. It processes patterns not because it wants to, but because not chunking is cognitively exhausting. The goal is always to convert the conscious into the automatic.
Design consequence: once a player has chunked a game’s core systems, the game becomes boring — the patterns no longer require processing. The game must continually introduce new patterns, new variation within patterns, or new combinations.
Three levels of processing
Koster identifies three levels at which the brain processes learned material:
| Level | Characteristics |
|---|---|
| Conscious / logical | Slow, explicit, mathematical; IQ-type reasoning; can be accessed directly |
| Intuitive / associative | Fast but opaque; builds approximations and chunks; cannot be accessed directly; often wrong (“common sense”) |
| Reflex / muscle-memory | Instantaneous; autonomic nervous system; trained into the body through repetition |
Grokking
Robert Heinlein’s term (Stranger in a Strange Land): to grok something is to understand it so thoroughly that you have become one with it. In Koster’s framework, grokking occurs when all three levels are engaged — conscious understanding, intuitive pattern-matching, and reflex response are all aligned. Grokking a game means it has been completely chunked; the patterns are fully internalised.
Grokking = fun exhausted. Once a game is grokked, boredom follows.
Fun = learning patterns
“Fun from games arises out of mastery. It arises out of comprehension. It is the act of solving puzzles that makes games fun. In other words, with games, learning is the drug.” — Koster, Ch. 3
Games are concentrated, formally abstracted patterns that the brain can practise, run permutations on, and ultimately master. The chemical reward at the moment of pattern-click — the endorphin release that produces a smile — is the brain’s reward for learning. Because the brain is biologically motivated to learn (survival depends on it), this reward is powerful.
Games as puzzles: Every game is a puzzle. The puzzle may be spatial, mathematical, social, physical, or abstract — but in each case, the player is working to infer, understand, and master the underlying pattern.
“Games are just exceptionally tasty patterns to eat up.” — Koster, Ch. 2
The teaching machine: Games function as highly efficient teaching tools because they:
- Abstract and iconify reality into learnable models
- Allow the player to run permutations on the pattern (unlike books, which present patterns passively)
- Provide immediate feedback on pattern-application attempts
- Exclude distracting detail, leaving only the pattern structure
What games teach
Koster observes that the patterns most games teach are dominated by pre-modern survival skills:
- Territory and spatial reasoning
- Power projection and combat
- Aiming and reaction time
- Odds calculation
- Hierarchical status navigation
- Teamwork for resource acquisition
Most games are teaching us to be better cavemen. The challenge for the medium is to evolve games that teach patterns relevant to modern life — empathy, systemic thinking, civic duty, cooperation across difference.
Boredom = mastery (the failure modes)
“Boredom is always the signal to let you know you have failed.” — Koster
When a game stops teaching, the player experiences boredom. Koster identifies six specific boredom failure modes:
| Mode | Player experience | Design failure |
|---|---|---|
| Too trivial | ”Too easy” — pattern grokked immediately | Challenge does not match player’s existing chunk library |
| Irrelevant depth | ”Yeah, there’s depth, but I don’t care” | Pattern exists but is not motivating to this player |
| Invisible pattern | ”This is too hard” — pattern looks like noise | No foothold to begin pattern recognition |
| Pacing too slow | ”It’s repetitive” — same beat, no new variation | Pattern revealed too slowly; player ahead of content |
| Pacing too fast | ”It got too hard too fast” — pattern lost | New variation introduced faster than it can be processed |
| Full mastery | ”I beat it” | All patterns consumed; nothing left to learn |
The good game
“The definition of a good game is therefore ‘one that teaches everything it has to offer before the player stops playing.‘” — Koster
A good game paces the revelation of its patterns to stay just ahead of the player’s chunk-formation. This is structurally similar to flow’s challenge/skill balance — the difficulty curve should track the player’s growing mastery.
Types of enjoyment
Koster distinguishes fun (pattern learning) from other forms of enjoyment that games can provide. All produce positive feelings, but only fun arises from learning:
| Type | What it is | Relation to “fun” |
|---|---|---|
| Fun (proper) | Mastering a problem; the brain’s learning reward | Definition of fun; what this theory is about |
| Aesthetic appreciation / delight | Recognising a known pattern with a surprising wrinkle; the “sensawunda” feeling | Not fun — reward for having already learned; fleeting; exhausted by repeated exposure |
| Visceral satisfaction | Physical endurance/performance; overcoming physical challenge | Only generates fun when combined with mental challenge |
| Social manoeuvring | Status games, dominance play | Enjoyable but not necessarily fun in this sense |
Delight specifically: Koster describes aesthetic appreciation as finding things beautiful when they meet our expectations and exceed them — a perfectly resolved plot with a few loose threads, a landscape that is orderly with one surprising element. Delight is the epilogue to learning. “It’s like the smile from a beautiful stranger in a stairwell — it’s fleeting.”
Fun is contextual. “We went out there to have fun tonight” is categorically different from “We went out there to win.” Fun requires the frame of practice — when the stakes become real, the same activity stops being fun and becomes something else (anxiety, performance, duty).
Social/emotional vocabulary
Koster draws on psychological research to name specific game-relevant emotions. These are distinct from “fun” but commonly occur in play:
| Term | Origin | Meaning |
|---|---|---|
| Fiero | Italian | Personal triumph over adversity; the fist-pump, the involuntary expression of having beaten something hard |
| Schadenfreude | German | Gloating pleasure at a rival’s failure; status signal (“put-down”) |
| Naches | Yiddish | Pride felt when someone you mentored succeeds; tribal continuance signal |
| Kvell | Yiddish | Pride in bragging about a mentee; also a self-value signal |
These terms also appear in Adams’ Bateman survey of player emotions (see game-definition). Koster’s use of them emphasises that games are fundamentally social and tribal experiences even when played alone.
Games vs. stories
Koster (Ch. 5) argues that games and stories are fundamentally different teaching tools and should not be conflated:
| Games | Stories |
|---|---|
| Experiential teaching | Vicarious teaching |
| Good at objectification | Good at empathy |
| Quantise, reduce, classify | Blur, deepen, nuance |
| External — about actions | Internal — about emotions and thoughts |
| Cannot be “fully mastered” in story sense | Good ones cannot be “beaten” |
Peak emotional moments in games often “cheat”: the most cited example of a game making players cry — Floyd’s sacrifice in Planetfall — happens in a cutscene, outside player control. Koster asks: if the peak emotional moment in games happens outside the game, what does that say?
Games excel at mastery emotions: fiero, triumph, satisfaction. Stories excel at empathy emotions: grief, love, regret. They are different instruments.
Design implication: Grafting a mediocre story onto a game (as most games do) serves as a “side dish for the brain” — it does not transform the game’s emotional register. To achieve story-level emotional depth, the mechanics themselves must carry the meaning.
Art vs. entertainment
“Art and entertainment are not terms of type — they are terms of intensity.” — Koster, Ch. 9
Koster’s definitions:
- Media — delivers information
- Entertainment — delivers comforting, simplistic information
- Art — delivers challenging information that must be worked through to be absorbed; alters how people perceive the world
Most of any medium is entertainment. Most novels are read “just for fun.” Most music is “just entertainment.” The existence of pop music does not prevent music from producing Bach. The same should be true for games.
Mere entertainment becomes art when the communicative element is either novel or exceptionally well done — when the work has the power to alter how people perceive the world.
Art is inexhaustible: the test Koster proposes — can you return to the work again and again and keep learning something new? A good game, in this sense, might not need to be won — it needs to teach without being exhausted.
The trellis metaphor
“A game is like a trellis.” — Koster, Ch. 11
A trellis shapes the plants that grow through it. A game shapes the player’s mind — it teaches patterns that become part of how the player thinks. But a simple trellis produces predictable shapes. Great art-trellises (in music, literature, architecture) shape the plant with intent — toward specific, targeted experiences.
Current games know only “fun” and “boring”: they can teach patterns, but they cannot yet reliably choose which patterns to teach or design toward specific insights about the human condition.
The design imperative (Koster’s challenge to the field):
“For games to reach art, the trellis itself, the mechanics, must be revelatory of the human condition.” — Koster, Ch. 11
Not the story wrapper. Not the art assets. The formal system — the rules, the win condition, the feedback loops — must carry the intended meaning. A game about the loneliness of power cannot be made by writing a sad story about a king; it requires a mechanic where accumulating power mechanically produces isolation.
Example Koster offers: a game where power-to-act derives from controlled subjects, but self-healing derives from friendship — and friends fall away as power grows. Victory is not “be on top” but “ensure tribal survival.” Now being powerful is costly and being lower-status is sometimes better. The mechanic encodes the lesson.
The obstacle is intent: the barrier to art-level games is not technical. It is a state of mind — the willingness to approach game design as an act of authorship with communicative purpose.
Implications for design
- Teach something new constantly. Every session should expand the player’s chunk library — new systems, new spatial configurations, new social dynamics.
- Design for mastery curves, not content volume. Throwing more content at a game is not the same as extending its teachable patterns.
- The fiction is secondary. Players are trained by games to see past dressing to the underlying pattern. Running over pedestrians in Deathrace does not teach running over pedestrians — it teaches spatial pickup-object patterns. The ethical and aesthetic concerns of the fiction matter, but they do not change what the game teaches at its core.
- If you want to cause emotion, build it into the mechanic. Emotional experiences borrowed from cutscenes are not the game’s achievement.
- Boredom is diagnostic, not just failure. Each boredom failure mode points to a specific design problem: monotony = inadequate pattern variation; frustration-boredom = no entry point to the pattern; triumph-boredom = patterns exhausted.
- Design toward insight. Ask: what specific aspect of the human condition could the formal system of this game illuminate? That is the direction toward mature game design.
Open questions
- Koster’s theory makes fun primarily cognitive. Does it account for physical fun (sports endorphins, visceral sensation in action games) adequately, or does his category of “visceral satisfaction requiring mental challenge” paper over something distinct?
- The trellis metaphor requires authorial intent at the mechanical level. Is this achievable in games where player agency is high? If players route around the intended lesson, has the art failed?
- Koster (2005) argues games are still teaching primitive caveman patterns. Twenty years later — has the medium evolved? Papers, Please, Disco Elysium, Undertale, This War of Mine suggest it has in some quarters.
- The art/entertainment intensity distinction sidesteps questions about craft, accessibility, and audience. A deeply challenging work that no one can access might be maximally intense but fail as communication. Is intensity enough?
Sellers’ extension: the neurochemical mechanism
Sellers (Advanced Game Design, 2018) provides a biological substrate for Koster’s thesis. Koster describes the cognitive process (pattern recognition → learning reward → boredom as mastery); Sellers adds the neurochemical mechanism:
- The learning reward Koster describes is primarily dopamine: the brain’s “do more of this” signal for novel situations and successfully reached goals
- Dopamine habituates to expected rewards — consistent with Koster’s observation that mastered patterns stop being fun. Once a pattern is chunked and its outcome is predictable, dopamine release decreases
- The “boredom as mastery” failure mode maps directly to dopamine habituation: the pattern is no longer novel, the reward signal diminishes, and the player seeks something new
This alignment validates both frameworks from different directions: Koster arrives at the cognitive pattern-learning model from first principles of game studies; Sellers arrives at the same prediction from neuroscience. They are complementary, not competing.
See neurochemical-engagement for the full neurochemical model, including serotonin (accomplishment), oxytocin (social bonding), and norepinephrine (alertness) as additional engagement pathways Koster’s model does not explicitly address.
(Sellers, Advanced Game Design, see source-advanced-game-design)
Cognitive science support
Koster’s theory — that fun arises from pattern recognition and learning — receives substantial empirical support from cognitive psychology research on how learning actually works. The key findings, documented in Brown, Roediger & McDaniel’s Make It Stick (2014), validate several of Koster’s claims from a different direction:
- Retrieval practice (recalling from memory rather than rereading) is the most effective learning strategy — consistent with Koster’s claim that games teach through active pattern-application, not passive exposure
- Desirable difficulties (spacing, interleaving, generation) produce stronger learning precisely because they require effort — mirroring Koster’s observation that challenge is what makes games fun, and easy games are boring
- Mental models — experts bundle knowledge into automatic procedures that can be deployed without conscious thought — this is exactly Koster’s chunking process by another name
- The paradox that effective learning feels unproductive parallels Koster’s observation that the most engaging games are not the easiest ones
See evidence-based-learning for the full treatment. The learning science page is intentionally separate from the game design wiki but the connection to Koster’s theory is direct.
(Brown, Roediger & McDaniel, Make It Stick, see source-make-it-stick)
Related
- flow — Challenge/skill balance is the mechanism that keeps boredom at bay; Koster’s framework is a cognitive explanation for why the flow channel produces engagement
- neurochemical-engagement — Sellers’ biological grounding for the pattern-learning reward; dopamine habituation as the mechanism of “boredom as mastery”
- game-definition — Koster’s definitions sit alongside Schell, Adams, Poole, and Bond
- fiero — The specific triumph emotion that pattern-learning at high difficulty produces
- chunking — Glossary definition of the cognitive compression mechanism central to this theory
- grokking — Glossary definition of full internalisation of a learned pattern
- mda-framework — LeBlanc’s 8 aesthetics overlap with Koster’s types of enjoyment; both attempt to taxonomise what games produce
- challenge-types — Adams’ challenge taxonomy describes the different pattern-types games use to teach
- interest-curves — Schell’s interest curves describe pacing of emotional engagement; Koster’s boredom theory describes the cognitive equivalent
- holographic-design — Koster’s trellis metaphor and Schell’s skeleton/skin distinction both address the relationship between formal system and player experience
- game-feel — Good game feel sustains engagement between pattern revelations; micro-level interest
- playtesting — Playtesting tests whether the pattern-revelation pacing is correct; whether the game teaches before the player stops playing
- games-vs-film — Koster’s games-vs-stories table
- evidence-based-learning — Cognitive psychology research validating Koster’s pattern-learning thesis; retrieval practice, desirable difficulties, mental models
- source-theory-of-fun
- source-advanced-game-design
- source-make-it-stick