GDnD Wiki

GDnD Wiki

This site is for Ulster University students and staff. Enter the class access password to continue.

Flow

18 min read

Summary

Flow is a psychological state of complete, energised engagement first described by Mihaly Csikszentmihalyi (1990, Flow: The Psychology of Optimal Experience). In this state, self-consciousness disappears, time distorts, and intrinsic motivation is at its peak. Jesse Schell identifies flow as one of the most important psychological concepts for game designers, because games are among the most reliable flow-generating experiences humans have devised.

(Schell, The Art of Game Design, see source-art-of-game-design)

Key ideas

The flow channel

Flow occurs in a narrow band between two failure states:

  • Too easy → boredom. When challenge falls below the player’s skill level, the game becomes trivial. Engagement drops.
  • Too hard → anxiety. When challenge exceeds the player’s skill level, they feel helpless or frustrated. Engagement collapses.

The flow channel is the zone between these extremes where challenge is appropriately matched to skill. The channel is narrow — small deviations in either direction break the state.

Challenge
 (high) ↑   ANXIETY          ╱      ╱
        │  (too hard)      ╱      ╱
        │                ╱      ╱
        │              ╱      ╱
        │            ╱ FLOW ╱
        │          ╱      ╱
        │        ╱      ╱   BOREDOM
        │      ╱      ╱  (too easy)
        │    ╱      ╱
  (low) └──────────────────────────────────→ Skill
        (low)                        (high)

Challenge is on the vertical axis and skill on the horizontal axis, as in Csikszentmihalyi’s tennis example reproduced by Schell (The Art of Game Design, Ch. 9). Anxiety lies above the channel, where challenge exceeds skill; boredom lies below it, where skill exceeds challenge.

Flow is a moving target

The flow channel is not a fixed location — it shifts as the player’s skill increases. A mechanic that was appropriately challenging in hour one becomes trivially easy in hour ten. Maintaining flow requires challenge that rises as skill grows, usually in waves with recovery beats rather than a steady climb.

This is why difficulty scaling, new mechanics, and escalating enemy behaviour are not optional polish — they are the mechanism by which a game stays in the flow channel over time.

Schell points out that some ways of moving up the channel are better than others. Moving straight up it is better than ending in anxiety or boredom, but a path that zig-zags within the channel “will probably feel much more interesting to a player”: challenge rises, the player is rewarded, often with more power, which makes the next stretch easier for a while, and then the challenge ramps up again. His example is a gun that kills enemies in two shots instead of three, until new enemies arrive that take three or four. He calls this cycle “tense and release” and argues that “too much tension, and we wear out. Too much relaxation, and we grow bored” (Schell, The Art of Game Design, Ch. 9, see source-art-of-game-design). The same rhythm appears at the scale of a whole game on interest-curves, and Bond’s warning about flow exhaustion below makes the same point.

Autotelic experience

Csikszentmihalyi calls flow “autotelic” — intrinsically motivated, pursued for its own sake rather than for external reward. A player in flow is not playing for a prize; they are playing because the activity itself is completely absorbing. This is the psychological state most associated with deep player investment and long play sessions.

Conditions for flow (Csikszentmihalyi’s list)

  1. Goals are clear
  2. Progress is measurable
  3. Feedback is immediate and unambiguous
  4. Control feels possible
  5. Distraction is minimised
  6. Self-consciousness is reduced
  7. Time perception distorts

Games can be designed to hit all seven conditions deliberately.

In practice

Narrative choice or a difficulty curve?

A common question is whether a core loop will hold players in flow better with more narrative choice or with a difficulty curve matched to their skill. They are different levers:

  • Challenge–skill balance is the flow lever. Boredom and anxiety are positions off the challenge–skill diagonal, and only a change in challenge (or in the player’s skill) moves the player back into the channel. If players are bored or frustrated in the core loop, tune the difficulty curve first.
  • Narrative choice serves autonomy and the sense of control. Choice supports a different condition: Chen lists “a sense of personal control over the game activity” as one of three core elements for flow, alongside matched challenge, and argues that meaningful choices give players that control (Chen, Flow in Games, MFA thesis, University of Southern California, 2006). In SDT terms this is the autonomy need (self-determination-theory, player-agency). Story choices can make a loop feel more owned, but they will not fix a challenge level that is wrong.
  • Shape the curve in waves. Raise difficulty in rising waves with recovery beats (Schell’s “tense and release” above, Bond’s “powerful/awesome” border below, and interest-curves) rather than a steady climb.
  • Combine them where you can. Chen’s own answer is to embed choices that adjust challenge into play, so the player’s decisions both express control and tune difficulty (see flOw below). A branching route with an easier and a harder path is a narrative or level choice that is also a difficulty choice.

As a practical verdict (design judgement rather than a research finding): for flow in a core loop, a skill-matched difficulty curve that rises in waves is the essential lever; narrative choice is a valuable addition for motivation and ownership, not a substitute.

Difficulty scaling approaches

  • Explicit difficulty settings — player-selected (Beginner/Normal/Hard). Simple, but removes designer control over pacing.
  • Dynamic difficulty adjustment (DDA) — system monitors performance and scales challenge accordingly. Can feel manipulative if detected. (course lecture series, see source-player-psychology-lectures)
  • Rubber-band mechanics — opponents or world respond to player success/failure (common in racing games). Maintains competition; risks feeling unfair.
  • Layered complexity — introduce new mechanics over time to sustain challenge as earlier mechanics are mastered. Schell’s preferred approach.
  • Player-driven (active) adjustment — choices embedded in play let players raise or lower their own challenge, such as an optional harder route or, in flOw, choosing when to descend to a harder layer. Jenova Chen proposed this as an alternative to hidden, system-driven DDA (see “Jenova Chen’s flOw” below).

In Unity: Practical levers for flow maintenance include enemy health/damage scaling, spawn rate adjustment, projectile speed, time limits, and gating new abilities behind progression milestones.

Lens #18 (The Lens of Flow): Schell’s lens asks, among other questions:

  • Does my game have clear goals? If not, how can I fix that?
  • Are the goals of the player the same goals I intended?
  • Are there parts of the game that distract players to the point they forget their goal?
  • Does my game provide a steady stream of not-too-easy, not-too-hard challenges, taking into account the fact that the player’s skills may be gradually improving?
  • Are the player’s skills improving at the rate I had hoped?

(Schell, The Art of Game Design, Ch. 9, see source-art-of-game-design)

Flow exhaustion and the “powerful/awesome” border (Bond)

Bond (Ch. 8) adds a critical refinement not present in Schell’s account:

Flow beyond 15–20 minutes is exhausting. Sustained peak engagement is cognitively demanding. Players need breaks from the flow channel to recover — periods of lower challenge where they can reflect on what they have accomplished and how much their skill has grown.

Bond identifies a “powerful/awesome” border between the flow channel and boredom: a zone where the challenge is below the player’s current skill level, but not so far below that it feels trivial. In this zone, the player experiences the feeling of being competent and powerful — a pleasurable contrast to the intensity of flow.

God of War (2005) is cited as a design example of this well. The game alternates intense combat sequences (flow) with traversal and environmental sections (powerful/awesome border), giving the player regular “power trips” that let them appreciate their combat mastery before returning to challenge.

Design implication: Do not design for continuous flow. Design a rhythm of flow and recovery. The recovery phases are not wasted time — they are when the player integrates their skill growth and builds investment in their character or progress.

(Bond, Introduction to Game Design, Prototyping, and Development, see source-introduction-game-design-prototyping)

Koster’s parallel: boredom as mastery (Theory of Fun, 2005)

Koster provides a cognitive underpinning for why the flow channel works. His framework is independently arrived at but structurally parallel:

  • Fun = learning new patterns. The brain rewards pattern recognition and internalisation with endorphins.
  • Boredom = mastery. When a game has been fully “chunked” (all patterns internalised), it no longer teaches, and boredom follows.
  • The six boredom failure modes all map onto positions outside the flow channel: too easy (challenge far below skill), too hard (challenge far above skill), wrong depth (player not interested in the pattern type), wrong pacing.

Where Csikszentmihalyi describes the experiential state (engagement vs. anxiety vs. boredom), Koster describes the cognitive mechanism (pattern learning vs. pattern exhaustion). Both converge on the same design implication: the game’s pattern complexity must track the player’s growing competence.

“Boredom is always the signal to let you know you have failed.” — Koster (A Theory of Fun, Ch. 3, see source-theory-of-fun)

For Koster’s full taxonomy of boredom failure modes, see fun-as-learning.

(Koster, A Theory of Fun, see source-theory-of-fun)

Evidence

“The flow channel is a delicate thing — if the challenge is too low, players become bored; if it is too high, they become anxious. The designer must carefully keep players in the channel through the whole play experience.” — paraphrase of Schell (Lens #18, source-art-of-game-design)

Schell notes that game designers had been creating flow-inducing experiences for decades before Csikszentmihalyi named the concept — the theory is a vocabulary for what skilled designers already intuited.

Implications

  • Tutorials matter: The hardest moment for flow is the beginning. Players have near-zero skill; the challenge must be near-zero too. A bad tutorial breaks flow before the game starts.
  • Failure should teach, not punish. Repeated failure means the challenge exceeds skill — but if the player can see why they failed and how to improve, the frustration converts to engagement. Failure that teaches is in the flow channel; failure that baffles is not.
  • Mastery needs to be honoured. When a player fully masters a mechanic, that mechanic must be retired or escalated. Continuing to challenge players with something they have mastered produces the boredom end of the spectrum.

Sellers’ additions: Yerkes-Dodson and structural coupling

Sellers (Advanced Game Design, 2018) provides two contributions that extend the flow framework with physiological grounding:

Yerkes-Dodson arousal curve

The Yerkes-Dodson law (1908) describes the relationship between arousal and performance as an inverted-U: performance peaks at moderate arousal, drops at both extremes. Sellers argues this is the neurophysiological substrate of the flow channel — the two frameworks are describing the same phenomenon at different levels of abstraction.

Performance
    ↑
    │       ╭─────────╮
    │      ╱           ╲
    │     ╱    FLOW      ╲
    │    ╱                ╲
    └──────────────────────→ Arousal
       (boredom)          (panic)

What this adds to Csikszentmihalyi’s model: The flow channel is not arbitrary — it corresponds to an optimal neurochemical state (dopamine and norepinephrine in the productive range, below the cortisol-dominated stress response). Designing for flow is designing for optimal neurochemical arousal. See neurochemical-engagement for the full neurochemical model.

Design implication: Flow is not just “right difficulty” — it is “right arousal level.” Monotony breaks flow from below (insufficient dopamine stimulus); overwhelming challenge breaks flow from above (cortisol-dominated stress response). Novelty, not just difficulty, is a lever for arousal.

Structural coupling

Sellers draws on Maturana and Varela’s (1987) concept of structural coupling — two systems that interact closely over time, each adapting to the other’s structure — to explain why well-designed games sustain flow:

“A game and player also form a structurally coupled relationship. If the game is systemically designed, it will have defined a sufficiently broad and diverse state-space that it can adapt to the player as the player adapts to it.” — Sellers, Ch. 2

Flow requires that the game continually provide an appropriate challenge as the player’s skill grows. In a structurally coupled system, this happens systemically rather than through explicit difficulty adjustment: the game’s state-space is broad enough that as the player masters lower-level patterns, higher-level patterns and challenges emerge naturally from the system.

This is the systemic design argument for second-order-design: a game with a rich enough state-space adapts to the player without the designer having to author each difficulty curve explicitly. Systems depth is a flow-sustaining mechanism.

(Sellers, Advanced Game Design, see source-advanced-game-design)

Hiwiller: the Fundamental Game Design Directive and casual/hardcore positioning

Hiwiller frames flow not merely as a psychological state to aim for, but as the design directive itself:

“Keep players making interesting decisions.” — Hiwiller, Ch. 9

More specifically, his “Fundamental Game Design Directive” is to keep players in a state where their skill level is a close match for the level of challenge the game presents — i.e. in the flow channel. Every other design technique — difficulty scaling, tutorial design, dynamic difficulty adjustment, new mechanic introduction — is in service of this directive.

Casual vs hardcore positioning

Hiwiller adds a practical calibration point not explicit in Schell or Bond. Different player audiences sit in different positions on the challenge/skill grid:

  • Casual players tend to require lower challenge levels relative to their skill — they become frustrated more easily. Casual games err toward insufficient challenge for their skill level.
  • Hardcore players demand higher challenge levels relative to their skill — they become bored more easily. Hardcore games err toward excessive challenge.

A game designed for one audience can actively drive away the other. This is not a failure of design but a deliberate positioning decision — understanding your target audience’s position on this axis is prerequisite to designing a flow experience for them.

Jenova Chen’s flOw and player-driven difficulty adjustment

Game designer Jenova Chen’s MFA thesis (Flow in Games, University of Southern California, 2006) applied Csikszentmihalyi’s framework directly to game design, proposing that the conditions for flow and the conditions players describe as “fun” are the same. Hiwiller’s reference for Chen’s work is the later article “Flow in Games (and Everything Else)” (Communications of the ACM, 2007); the account below follows the thesis itself.

Hiwiller describes flOw’s main feature as a dynamic difficulty adjustment system that “detects how the player’s abilities are increasing and adjusts the difficulty to compensate” (Hiwiller, Players Making Decisions, Ch. 9). Chen’s thesis describes it differently, and the difference is the thesis’s main argument:

  • Passive, system-oriented DDA monitors player data, infers their state and changes the game for them. Chen argues it struggles because games have no direct data about how a player feels, “performance does not mirror Flow”, assumptions about players fail for a mass audience, and the designer’s adjustment rules are rigid.
  • Active, player-oriented DDA gives players choices that let them steer their own experience: “If they start feeling bored, they can choose to play harder, vice versa.” Chen adds that the choices must be embedded in the core gameplay, because frequent explicit choices interrupt flow (his first prototype, Traffic Light, asked between rounds and found exactly that).
  • In flOw, the game is split into 20 levels. By choosing which food to eat, players descend to a harder level or return to an easier one at any time, and can avoid a challenge and come back later. Dying has a minimal penalty: the player is pushed back to the previous, easier level. Players balance their own experience through swimming and eating, the only two actions in the game.

(Chen, Flow in Games, MFA thesis, 2006)

So flOw is best described as player-driven difficulty adjustment with a light automatic element (the step back on death), not as a hidden system that silently re-tunes challenge. Hiwiller’s wider point, that designers should keep challenge matched to skill, stands either way.

Games have structurally incorporated this principle for decades through explicit difficulty curves — Super Mario Bros.’ World 1-1 provides near-zero challenge for near-zero skill (teaching basic controls through safe experimentation), escalating to complex multi-hazard environments in later worlds that require mastered skills to navigate.

Modern AAA studios apply this with data: Microsoft’s research team produced visualisation tools for Bungie’s Halo 3, mapping where players spent more time or used more ammunition — revealing bottlenecks where challenge was too high, and voids where challenge was insufficient.

(Hiwiller, Players Making Decisions, Ch. 9, see source-players-making-decisions)

Open questions

  • Dynamic difficulty adjustment systems have been criticised for feeling artificial or patronising when detected. Is transparent DDA (where the player knows the system is adjusting) better than opaque DDA? Chen’s thesis sidesteps the question by handing the adjustment to the player through embedded choices; how far that scales beyond a game as minimal as flOw is open.
  • Flow is an individual psychological state — a challenge level that keeps one player in flow will bore or frustrate another. How do designers balance flow across a diverse player population?
  • Some games deliberately break flow for artistic or subversive effect (e.g. Undertale, Doki Doki Literature Club). Is this a violation of flow principles, or a deliberate use of flow violation as a design tool?
  • The Yerkes-Dodson curve predicts a single optimal arousal peak. But players report enjoying both very low-arousal (relaxing games, walking simulators) and very high-arousal (horror, bullet hell) experiences. Does the model need to accommodate this range?

Practice

Chart one play session against the flow channel. Record 15 minutes of yourself playing a game with rising challenge (a platformer, an action game or a puzzle game works well). Stopping to rate yourself during play is itself a distraction, so rate the recording afterwards instead. Every 90 seconds of footage, note:

  • what you were doing at that moment
  • challenge, from 1 (trivial) to 5 (overwhelming)
  • your skill at that task at that moment, from 1 to 5
  • a label: boredom, flow, anxiety, or Bond’s “powerful/awesome” border

Plot the points on a sketch with skill on the horizontal axis and challenge on the vertical axis, draw the diagonal flow channel, and join the points in time order.

Success test:

  • at least ten points are plotted and joined in time order
  • every point outside the channel has a named cause (for example a new mechanic introduced without teaching, or a challenge you had already mastered)
  • at least two of Csikszentmihalyi’s seven conditions for flow are linked to specific moments in the session
  • you note whether the game gave you a recovery phase, and where it was

Extension:

  • pick the worst moment on your chart and propose one fix using a difficulty-scaling approach from this page (explicit settings, dynamic difficulty adjustment, rubber-banding, layered complexity or player-driven adjustment). Say whether your fix suits a casual or a hardcore audience, and why.

Self-test

  1. What happens to the player when challenge falls below their skill, and when it rises above it?
  2. Why is the flow channel described as a moving target?
  3. What does “autotelic” mean?
  4. According to Bond, why should designers not aim for continuous flow?
  5. Name the five difficulty-scaling approaches listed on this page.
  6. How does Hiwiller position casual and hardcore players on the challenge/skill grid?
  7. A designer wants to keep players in flow in a core loop. Should they add more narrative choice or tune a difficulty curve to player skill?

Answers

  1. Challenge below skill produces boredom; challenge above skill produces anxiety or frustration.
  2. Player skill increases with play, so a challenge that was right in hour one becomes trivial later. Challenge must rise as skill grows, usually in waves with recovery beats rather than a steady climb.
  3. Intrinsically motivated: pursued for its own sake rather than for an external reward.
  4. Flow beyond 15–20 minutes is exhausting. Players need recovery phases, such as the “powerful/awesome” border, where they can feel competent and integrate their skill growth.
  5. Explicit difficulty settings, dynamic difficulty adjustment, rubber-band mechanics, layered complexity and player-driven (active) adjustment.
  6. Casual players tend to need lower challenge relative to their skill because they become frustrated more easily. Hardcore players demand higher challenge relative to their skill because they become bored more easily.
  7. Tune the difficulty curve first: only challenge matched to skill moves players back into the flow channel, and it should rise in waves with recovery beats. Narrative choice serves autonomy and the sense of control, so it adds ownership but does not fix boredom or anxiety. Choices that also change difficulty (as in flOw) do both.

Graph View

Backlinks