Executive summary

Gaming is becoming a behavioural laboratory.

For years, the central question was whether skills learned in games transfer into life outside them. That question still matters, but it misses a larger possibility: games can make decisions visible.

People play under rules, incentives, uncertainty, time pressure and the influence of other people. A sufficiently rich game can record not only the outcome, but the sequence of choices that produced it: how someone allocates attention, manages risk, responds to change and adapts after a setback.

Games can create behavioural environments in which human decisions become observable, repeatable and measurable.
ApplicationQuestion
TrainingCan a game improve performance in another environment?
SelectionCan gameplay help identify people suited to a real-world role?
MeasurementCan play reveal behavioural patterns conventional assessments miss?
InterventionCan rules, incentives or consequences change behaviour?
ModellingCan behavioural data help anticipate decisions in unfamiliar situations?

Train people → observe people → measure people → model people → simulate people.

The final step is still largely experimental. It asks whether models built from observed behaviour can help explore how people might respond inside a simulated real-world system.

Transfer

The limits of the traditional transfer argument.

“Games teach skills, therefore gamers have useful skills” is too broad. Different games call on different behaviours: rapid visual tracking, planning across a long horizon, coordination, communication or procedural learning.

The useful unit is not a general category of “gamer”. It is a behaviour performed under a particular set of conditions—and evidence that the behaviour carries over when those conditions change.

Behavioural environments

Why games are unusually useful places to observe behaviour.

Real decisions involve multiple forces at once. Games can preserve some of that complexity while keeping the environment observable, repeatable and adjustable. Researchers can set objectives, rules, rewards, information, time limits and failure states, then record what happens.

State → decision → action → consequence → adaptation → next decision.

Case studies

Twelve examples where the boundary starts to move.

These examples span training, selection, measurement and collective problem-solving. Each shows a different way games can make behaviour useful to study; none establishes that gameplay alone predicts how any one person will act in every real-world setting.

03 / FAA

The FAA looks to gamers

In 2026, the Federal Aviation Administration ran a recruitment effort aimed at gamers for air-traffic-control roles. The open question is whether gameplay can reveal behaviours that overlap with the demands of the job.

04 / DRONE OPERATORS

Screen-based control

Studies of unmanned-aircraft tasks found gamers particularly capable at some visual acquisition and tracking tasks, while trained pilots performed better on some complex tasks. The comparison depends on the task being measured.

05 / SPACE FORTRESS

Skills that are not about the game

Space Fortress was designed to study complex skill acquisition. Related aviation research points to possible gains in attention management, situational awareness and handling competing demands.

06 / SURGERY

Learning through controlled failure

A surgical training trial found participants who used a game solved a median 59% of equipment malfunctions, compared with 33% in the control group. The format enabled repeated practice with disruption.

07 / EMERGENCY MEDICINE

Changing decision heuristics

In a trauma-triage trial, physicians using a serious game under-triaged severely injured patients less often than those receiving traditional material; the difference remained at a six-month follow-up.

08 / WORLD OF WARCRAFT

An accidental behavioural study

The Corrupted Blood incident spread beyond its intended game setting. Players fled, helped, attempted quarantine and deliberately spread the infection. It was not a model of a human pandemic, but it revealed emergent group behaviour.

09 / SEA HERO QUEST

Measurement at population scale

Built as a research instrument, Sea Hero Quest collected navigation behaviour from millions of players. Its scale made it possible to examine wayfinding patterns across a broad population.

10 / FOLDIT

Human problem-solving as search

Foldit turned protein-folding into a game. Players produced a structure that helped researchers investigate a viral protease, demonstrating how human spatial reasoning can contribute to scientific work.

11 / EVE ONLINE

Distributed scientific work

Project Discovery placed scientific classification inside a persistent online game. Players contributed millions of classifications, connecting voluntary play with research infrastructure.

12 / MMO LEADERSHIP

Signals with boundaries

A study of more than 18,000 MMO players found some relationships between in-game social behaviour and leadership in voluntary groups. Other behaviours did not generalise to company leadership.

13 / HUMAN–AI

Working with people, not only agents

Overcooked has been used to examine human–AI coordination. Agents trained only with other agents may struggle with people; systems that model human behaviour can coordinate more effectively.

14 / DECISION MODELLING

From observing to anticipating

Researchers have modelled sequences of human choices from behavioural datasets. A harder question remains: whether those patterns can help anticipate decisions in unfamiliar environments.

Military

High-stakes systems make rehearsal valuable.

Military organisations have used immersive simulations and game-based scenarios to rehearse command, communication and interpersonal decisions. These settings are useful examples because some real-world conditions are costly or dangerous to reproduce.

Any claim about a specific deployment needs public evidence. The broader research question is how controlled scenarios can help people practise and how observed responses can improve the next scenario.

Synthesis

What these environments have in common.

PropertyWhy it matters
RulesDefine which actions are possible.
ConsequencesActions change the state of the system.
FeedbackOutcomes provide information.
AdaptationPrevious outcomes shape later decisions.
TelemetryBehaviour can be recorded in sequence.
RepeatabilityComparable situations can be presented again.
ManipulabilityIncentives, information and other agents can be changed.

Six levels

Different claims need different evidence.

LevelQuestionEvidence today
1 / TrainingCan games improve performance?Strong in selected domains.
2 / SelectionCan play help identify role fit?Promising; predictive validity needs stronger evidence.
3 / MeasurementCan games measure behaviour at scale?Some of the clearest large-scale examples.
4 / InterventionCan games change behaviour?Demonstrated in controlled settings.
5 / PredictionCan behaviour predict future decisions?Shown in constrained environments.
6 / SimulationCan models populate realistic simulations?Early, open and in need of validation.

Behaviour is the data

The valuable object may be the behaviour a game records.

Game businesses often focus on players, engagement and retention. A different lens sees a sequence: players make decisions, interact, adapt and form behavioural trajectories. The dataset is not just a record of activity; it may help describe how people respond under particular conditions.

Conventional assessment often sees fragments. Games can reveal trajectories.

Context

Behaviour is shaped by the environment.

Someone may take a risk because the cost of failure is low, cooperate because cooperation is rewarded, or optimise for the game instead of the intended task. A gameplay pattern does not tell us who a person is.

The useful question is which tendencies appear under defined conditions, and whether they remain meaningful when those conditions change. Transfer has to be demonstrated, not assumed.

The CV

From static credentials to observed behaviour.

A work sample shows how someone performs a task. A behavioural simulation can place someone in a changing environment and record how they respond over time. These approaches may complement conventional credentials, but do not replace them without evidence of added value and fairness.

Decision intelligence

The emerging opportunity is to model response.

Imagine a simulation where many synthetic actors have different goals, memories, risk tolerance, trust thresholds and relationships. The environment changes; information fails; incentives shift; people adapt. The result is only useful if the actors behave plausibly.

Real people → controlled environments → behavioural data → models → simulations → better questions.

Validation

What would have to be true?

HurdleQuestion
Behavioural stabilityDoes a pattern persist when it is observed repeatedly?
Cross-context validityDoes behaviour in a game relate to behaviour outside it?
Incremental validityDoes gameplay add information beyond existing assessments?
Model calibrationCan models describe populations, not just anecdotes?
Simulation validityDo synthetic agents reproduce meaningful behavioural statistics?

Thesis

The thesis.

The evidence does not support the simple claim that gaming makes people better at real life. It points to a more useful possibility: games can create controlled environments where decisions are observable, and those observations can support training, measurement, study and—eventually—models of human decision-making.

Some applications are established in selected settings. Others are emerging. Using behavioural models to populate real-world simulations remains an open research problem.

Implications

If the thesis holds, the lens changes.

Game developersTelemetry may describe behaviour, not only product use.

EmployersObserved work may complement static credentials.

ResearchersGames offer controlled settings for population-scale study.

AI teamsUseful systems need better models of how people actually respond.

Decision teamsBehavioural models may one day help explore complex scenarios.

Conclusion

The game becomes a laboratory. The player becomes a source of behavioural data.

The transfer of skills from games to other settings is real but incomplete as a thesis. A larger possibility is to represent parts of reality in a controlled environment, observe how people make decisions there, and test whether those patterns help us understand behaviour beyond the game.

The work depends on asking the right question, measuring the right behaviour and proving where a finding applies. The boundary between game and real-world system may become less important—but only when evidence supports the connection.

Research notes / Source base

Evidence before extrapolation.

The article draws on peer-reviewed work in game-based surgical and trauma training, unmanned-aircraft tasks, MMO leadership and sequential decision modelling; large-scale projects such as Sea Hero Quest, Foldit and EVE Online Project Discovery; and documented institutional training examples.

Examples that could not be publicly verified are excluded. Evidence of a behaviour in a game is a starting point for a question—not proof that the same behaviour transfers to every real-world context.

← Back to Research