Normal-Form Games
762 words
4 min read
Visual companion
Python
Type and operator map
Python Week 1: the first filter for runtime behavior
View
Revision summary
What this note is really saying
Short form
# Normal-Form Games ## 🎯 Learning Objectives - Represent any static game of complete information using a normal-form payoff matrix - Identify players, strategies, and payoffs from a game description - Apply common knowledge of rationality assumptions - Analyze classic games: Prisoner's Dilemma, Battle of the Sexes,...

Normal-Form Games
🎯 Learning Objectives
- Represent any static game of complete information using a normal-form payoff matrix
- Identify players, strategies, and payoffs from a game description
- Apply common knowledge of rationality assumptions
- Analyze classic games: Prisoner's Dilemma, Battle of the Sexes, Coordination, Chicken
- Distinguish between strict and weak dominance
📋 Prerequisites
- Basic algebra and set notation from high school mathematics
- From BSCS3003-AI Search: Minimax algorithm for zero-sum games provides intuition for strategic thinking
1.1 What Is a Game?
A game involves:
- Two or more players (decision-makers)
- Each player chooses from a set of strategies Si
- The payoff for each player depends on the combination of strategies chosen
🔑 Key Insight: Your payoff depends not only on your own choice, but on others' choices too.
Formal Definition
A normal-form game is ⟨N,{Si}i∈N,{ui}i∈N⟩ where:
- N={1,2,…,n} — set of players
- Si — strategy set for player i
- ui:S1×⋯×Sn→R — payoff function For two-player games, we use a payoff matrix:
The first entry is Player 1's (row) payoff; the second is Player 2's (column) payoff.
1.2 Rationality Assumptions
| Assumption | Meaning |
|---|---|
| Rationality | Players maximise their own payoff given beliefs |
| Common Knowledge of Rationality (CKR) | Everyone knows everyone is rational, knows they know, ad infinitum |
| Complete Information | All players know the game structure (players, strategies, payoffs) |
| Simultaneous Moves | No one observes another's choice before deciding |
1.3 Classic Games
Prisoner's Dilemma
SilentConfessSilent(−1,−1)(0,−10)Confess(−10,0)(−5,−5)- Confess strictly dominates Silent for both
- Outcome (Confess,Confess) is worse for both than (Silent,Silent)
- Paradox: Individual rationality → collective irrationality
Battle of the Sexes
BalletFootballBallet(2,1)(0,0)Football(0,0)(1,2)- Two Nash equilibria: (Ballet, Ballet) and (Football, Football)
- Coordination problem with distributional conflict
Chicken (Hawk-Dove)
SwerveStaySwerve(0,0)(1,−1)Stay(−1,1)(−5,−5)- Two asymmetric equilibria: (Swerve, Stay) and (Stay, Swerve)
- Models brinkmanship
1.4 Dominance
Strict Dominance
si strictly dominates si′ if:
Weak Dominance
si weakly dominates si′ if:
And strictly for at least one s−i.
Iterated Elimination of Strictly Dominated Strategies (IESDS)
(Diagram)
1.5 Best Response
si is a best response to s−i if:
The best response correspondence BRi(s−i) collects all best responses.
📊 Formula Summary
| Concept | Definition |
|---|---|
| Normal-form game | ⟨N,{Si},{ui}⟩ |
| Strict dominance | ui(si,s−i)>ui(si′,s−i) ∀s−i |
| Weak dominance | ui(si,s−i)≥ui(si′,s−i) ∀s−i , strict for some |
| Best response | BRi(s−i)=argmaxsiui(si,s−i) |
✅ Practice Questions
Q1: Two firms compete on price (High/Low). Profits:
Does either firm have a strictly dominant strategy?
XYZA(5,3)(3,1)(2,0)B(4,2)(4,4)(1,5)C(3,1)(5,2)(4,3)SolutionFor Firm 1: If P2 plays H, u1(L)=6>u1(H)=4. If P2 plays L, u1(L)=2>u1(H)=1. L strictly dominates H. By symmetry, L also dominates for Firm 2. This is a Prisoner's Dilemma. Q2: Find weakly dominated strategies:
UDL(10,2)(5,6)R(3,4)(8,1)SolutionZ is strictly dominated by Y (3>2,4>1,5>4). C is strictly dominated by B (2>1,4>2,5>3). After eliminating Z and C, no further dominated strategies remain. Q3: Find all best responses:
BR1(L)={U}, BR1(R)={D}, BR2(U)={R}, BR2(D)={L}. No pure-strategy Nash equilibrium exists.