Prisoner's Dilemma
Rewards cooperation but creates a temptation to defect. Useful for reciprocity and learning.
Results
Rounds0
Top agent—
Avg payoff0
Cooperation / match rate0%
How this teaches game theory
This final laboratory compares strategy classes rather than only one game. Students can observe how success depends on the environment, opponent, feedback, noise, and adaptation rule. There is no universally best strategy across all strategic settings.
Complete scenario, mode, level, and agent guide
Matching Pennies
A zero-sum game with no pure-strategy equilibrium; unpredictability matters.
Coordination Game
Agents benefit from matching actions, highlighting convention formation.
Rock–Paper–Scissors
Cyclic dominance rewards mixed or adaptive behavior rather than a single fixed action.
Modes
Agent vs Agent
Two selected algorithms play many rounds against each other.
Human vs Agent
The human is represented by a reactive strategy input model while the selected opponent adapts.
Full Tournament
Every available agent plays every other agent and a ranking is produced.
Levels
Level 1 — Fixed Strategies
Random, always-first, always-second, and simple best-response agents.
Level 2 — Learning Agents
Adds fictitious-play and reinforcement-style agents that update from observed history.
Level 3 — Noisy Environment
Actions occasionally flip, testing robustness and forgiveness under imperfect execution.
Interpretation: tournament success is contextual. A strategy can dominate one opponent and perform poorly against another, so evaluation should include payoff, stability, exploitability, and robustness.