epistemic game theory lecture 1 - esslli'12, opole - eric pacuit
TRANSCRIPT
![Page 1: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/1.jpg)
Epistemic Game TheoryLecture 1
ESSLLI’12, Opole
Eric Pacuit Olivier Roy
TiLPS, Tilburg University MCMP, LMU Munichai.stanford.edu/~epacuit
http://olivier.amonbofis.net
August 6, 2012
Eric Pacuit and Olivier Roy 1
![Page 2: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/2.jpg)
The Guessing Game
Eric Pacuit and Olivier Roy 2
![Page 3: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/3.jpg)
Plan for the week
1. Monday Basic Concepts.
2. Tuesday Epistemics.
3. Wednesday Fundamentals of Epistemic Game Theory.
4. Thursday Puzzles and Paradoxes.
5. Friday Extensions and New Directions.
Eric Pacuit and Olivier Roy 3
![Page 4: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/4.jpg)
Plan for the week
1. Monday Basic Concepts.
• Basics of Game Theory.• The Epistemic View on Games.• Basics of Decision Theory
2. Tuesday Epistemics.
3. Wednesday Fundamentals of Epistemic Game Theory.
4. Thursday Puzzles and Paradoxes.
5. Friday Extensions and New Directions.
Eric Pacuit and Olivier Roy 3
![Page 5: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/5.jpg)
Plan for the week
1. Monday Basic Concepts.
2. Tuesday Epistemics.
• Logical/qualitative models of beliefs, knowledge andhigher-order attitudes.
• Probabilistic/quantitative models of beliefs, knowledge andhigher-order attitudes.
3. Wednesday Fundamentals of Epistemic Game Theory.
4. Thursday Puzzles and Paradoxes.
5. Friday Extensions and New Directions.
Eric Pacuit and Olivier Roy 3
![Page 6: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/6.jpg)
Plan for the week
1. Monday Basic Concepts.
2. Tuesday Epistemics.
3. Wednesday Fundamentals of Epistemic Game Theory.
• Common knowledge of Rationality and iterated strictdominance in the matrix.
• Common knowledge of Rationality and backward induction(strict dominance in the tree).
4. Thursday Puzzles and Paradoxes.
5. Friday Extensions and New Directions.
Eric Pacuit and Olivier Roy 3
![Page 7: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/7.jpg)
Plan for the week
1. Monday Basic Concepts.
2. Tuesday Epistemics.
3. Wednesday Fundamentals of Epistemic Game Theory.
4. Thursday Puzzles and Paradoxes.
• Weak dominance and admissibility in the matrix.• Russell-style paradoxes in models of higher-order beliefs. (The
Brandenburger-Kiesler paradox).
5. Friday Extensions and New Directions.
Eric Pacuit and Olivier Roy 3
![Page 8: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/8.jpg)
Plan for the week
1. Monday Basic Concepts.
2. Tuesday Epistemics.
3. Wednesday Fundamentals of Epistemic Game Theory.
4. Thursday Puzzles and Paradoxes.
5. Friday Extensions and New Directions.
• Nash Equilibrium and mixted strategies.• Forward Induction.• Are the models normative or descriptive?• Theory of play.
Eric Pacuit and Olivier Roy 3
![Page 9: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/9.jpg)
Practicalities
I Course Website:
• ai.stanford.edu/~epacuit/esslli2012/epgmth.html
I There you’ll find handouts, reading material and additionalreferences.
I In case of problem:
• Olivier Roy: [email protected]
• Eric Pacuit: [email protected]
Eric Pacuit and Olivier Roy 4
![Page 10: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/10.jpg)
Basics of Game Theory
Eric Pacuit and Olivier Roy 5
![Page 11: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/11.jpg)
Basics of Game Theory
Key Concepts
I Games in Strategic (matrix) and Extensive (tree) form.
I Strategies (pure and mixed).
I Solution Concepts: Iterated Strict Dominance, Iterated WeakDominance, Nash Equilibrium,
Eric Pacuit and Olivier Roy 6
![Page 12: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/12.jpg)
Basics of Game Theory
The Matrix: games in strategic forms.
Alexei
StrangeloveDisarm Arm
Disarm 3, 3 0, 4
Arm 4, 0 1, 1
Players, Actions or Strategies, Strategy profiles, Payoffs on profiles.
Eric Pacuit and Olivier Roy 7
![Page 13: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/13.jpg)
Basics of Game Theory
The Matrix: games in strategic forms.
Alexei
Strangelove
Disarm Arm
Disarm 3, 3 0, 4
Arm 4, 0 1, 1
Players,
Actions or Strategies, Strategy profiles, Payoffs on profiles.
Eric Pacuit and Olivier Roy 7
![Page 14: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/14.jpg)
Basics of Game Theory
The Matrix: games in strategic forms.
Alexei
StrangeloveDisarm Arm
Disarm
3, 3 0, 4
Arm
4, 0 1, 1
Players, Actions or Strategies, Strategy profiles,
Payoffs on profiles.
Eric Pacuit and Olivier Roy 7
![Page 15: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/15.jpg)
Basics of Game Theory
The Matrix: games in strategic forms.
Alexei
StrangeloveDisarm Arm
Disarm 3, 3
0, 4
Arm
4, 0
1, 1
Players, Actions or Strategies, Strategy profiles, Payoffs on profiles.
Eric Pacuit and Olivier Roy 7
![Page 16: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/16.jpg)
Basics of Game Theory
The Matrix: games in strategic forms.
Alexei
StrangeloveDisarm Arm
Disarm 3, 3 0, 4
Arm 4, 0 1, 1
Players, Actions or Strategies, Strategy profiles, Payoffs on profiles.
Eric Pacuit and Olivier Roy 7
![Page 17: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/17.jpg)
Basics of Game Theory
A three players game
Fidel - DAlexei
StrglvD A
D 3, 3, 3 1, 4, 5
A 4, 1, 1 2, 2, 2
Fidel - AAlexei
StrglvD A
D 3, 3, 2 1, 4, 4
A 4, 1, 0 2, 2, 2
Eric Pacuit and Olivier Roy 8
![Page 18: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/18.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions,
Players, Payoffs on leaves, Strategies
Eric Pacuit and Olivier Roy 9
![Page 19: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/19.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions, Players,
Payoffs on leaves, Strategies
Eric Pacuit and Olivier Roy 9
![Page 20: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/20.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions, Players, Payoffs on leaves,
Strategies
Eric Pacuit and Olivier Roy 9
![Page 21: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/21.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions, Players, Payoffs on leaves, Strategies
Eric Pacuit and Olivier Roy 9
![Page 22: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/22.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions, Players, Payoffs on leaves, Strategies
Eric Pacuit and Olivier Roy 9
![Page 23: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/23.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions, Players, Payoffs on leaves, Strategies
Eric Pacuit and Olivier Roy 9
![Page 24: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/24.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions, Players, Payoffs on leaves, Strategies
Eric Pacuit and Olivier Roy 9
![Page 25: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/25.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions, Players, Payoffs on leaves, Strategies
Eric Pacuit and Olivier Roy 9
![Page 26: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/26.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions, Players, Payoffs on leaves, Strategies
Eric Pacuit and Olivier Roy 9
![Page 27: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/27.jpg)
Basics of Game Theory
The Tree: games in extensive forms.
S
A A
3, 3 1, 4 4, 1 2, 2
D A
D A D A
Actions, Players, Payoffs on leaves, Strategies
Eric Pacuit and Olivier Roy 9
![Page 28: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/28.jpg)
Basics of Game Theory
Extensive and strategic form games are related
A
SD A
D 3, 3 1, 4
A 4,1 2, 2
S
A A
3,3 1,4 4,1 2,2
D A
D A D A
Eric Pacuit and Olivier Roy 10
![Page 29: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/29.jpg)
Basics of Game Theory
Extensive and strategic form games are related
A
SD A
D 3, 3 1, 4
A 4,1 2, 2
S
A A
3,3 1,4 4,1 2,2
D A
D A D A
Eric Pacuit and Olivier Roy 10
![Page 30: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/30.jpg)
Basics of Game Theory
Some types of non-cooperative games of interest
I 2 players games.
I 2 players, zero-sum: if one player “wins” x then the other“looses” −x .
I 2 players, win-loose games.
I Perfect/imperfect information.
Eric Pacuit and Olivier Roy 11
![Page 31: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/31.jpg)
Basics of Game Theory
Pure and mixed strategies.
Alexei
StrangeloveHead Tail
Head 1, -1 -1, 1
Tail -1, 1 1, -1
I Strangelove has two pure strategies: Head and Tail.I A mixed strategy is a probability distribution over the set of
pure strategies. For instance:
• (1/2 Head, 1/2 Tail)• (1/3 Head, 2/3 Tail)• ...
I Additional subtleties in extensive games. (mixing at a node vsmixing whole strategies).
Eric Pacuit and Olivier Roy 12
![Page 32: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/32.jpg)
Basics of Game Theory
Pure and mixed strategies.
Alexei
StrangeloveHead Tail
Head 1, -1 -1, 1
Tail -1, 1 1, -1
I Strangelove has two pure strategies: Head and Tail.I A mixed strategy is a probability distribution over the set of
pure strategies. For instance:
• (1/2 Head, 1/2 Tail)• (1/3 Head, 2/3 Tail)• ...
I Additional subtleties in extensive games. (mixing at a node vsmixing whole strategies).
Eric Pacuit and Olivier Roy 12
![Page 33: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/33.jpg)
Basics of Game Theory
Pure and mixed strategies.
Alexei
StrangeloveHead Tail
Head 1, -1 -1, 1
Tail -1, 1 1, -1
I Strangelove has two pure strategies: Head and Tail.
I A mixed strategy is a probability distribution over the set ofpure strategies. For instance:
• (1/2 Head, 1/2 Tail)• (1/3 Head, 2/3 Tail)• ...
I Additional subtleties in extensive games. (mixing at a node vsmixing whole strategies).
Eric Pacuit and Olivier Roy 12
![Page 34: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/34.jpg)
Basics of Game Theory
Pure and mixed strategies.
Alexei
StrangeloveHead Tail
Head 1, -1 -1, 1
Tail -1, 1 1, -1
I Strangelove has two pure strategies: Head and Tail.I A mixed strategy is a probability distribution over the set of
pure strategies. For instance:
• (1/2 Head, 1/2 Tail)• (1/3 Head, 2/3 Tail)• ...
I Additional subtleties in extensive games. (mixing at a node vsmixing whole strategies).
Eric Pacuit and Olivier Roy 12
![Page 35: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/35.jpg)
Basics of Game Theory
Pure and mixed strategies.
Alexei
StrangeloveHead Tail
Head 1, -1 -1, 1
Tail -1, 1 1, -1
I Strangelove has two pure strategies: Head and Tail.I A mixed strategy is a probability distribution over the set of
pure strategies. For instance:
• (1/2 Head, 1/2 Tail)• (1/3 Head, 2/3 Tail)• ...
I Additional subtleties in extensive games. (mixing at a node vsmixing whole strategies).
Eric Pacuit and Olivier Roy 12
![Page 36: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/36.jpg)
Basics of Game Theory
Interpretation of mixed strategies
1. Real randomizations:
• Side of goal in penalty kicks.• Serving side in tennis.• Luggage check at the airport.
2. Epistemic interpretation:
• Mixed strategies as beliefs of the other player(s) about whatyou do.
Eric Pacuit and Olivier Roy 13
![Page 37: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/37.jpg)
Basics of Game Theory
Interpretation of mixed strategies
1. Real randomizations:
• Side of goal in penalty kicks.• Serving side in tennis.• Luggage check at the airport.
2. Epistemic interpretation:
• Mixed strategies as beliefs of the other player(s) about whatyou do.
Eric Pacuit and Olivier Roy 13
![Page 38: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/38.jpg)
Basics of Game Theory
Interpretation of mixed strategies
1. Real randomizations:
• Side of goal in penalty kicks.• Serving side in tennis.• Luggage check at the airport.
2. Epistemic interpretation:
• Mixed strategies as beliefs of the other player(s) about whatyou do.
Eric Pacuit and Olivier Roy 13
![Page 39: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/39.jpg)
Basics of Game Theory
Solution Concepts
I Set of profiles or outcome of the game that are intuitivelyviewed as “rational”.
I Three well-known solution concepts in the matrix:
• Nash Equilibrium.• Iterated elimitation of:
I Strictly dominated strategies.I Weakly dominated strategies.
I In the tree we will focus on one:
• Backward induction.
Eric Pacuit and Olivier Roy 14
![Page 40: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/40.jpg)
Basics of Game Theory
Solution Concepts
I Set of profiles or outcome of the game that are intuitivelyviewed as “rational”.
I Three well-known solution concepts in the matrix:
• Nash Equilibrium.• Iterated elimitation of:
I Strictly dominated strategies.I Weakly dominated strategies.
I In the tree we will focus on one:
• Backward induction.
Eric Pacuit and Olivier Roy 14
![Page 41: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/41.jpg)
Basics of Game Theory
Solution Concepts
I Set of profiles or outcome of the game that are intuitivelyviewed as “rational”.
I Three well-known solution concepts in the matrix:
• Nash Equilibrium.• Iterated elimitation of:
I Strictly dominated strategies.I Weakly dominated strategies.
I In the tree we will focus on one:
• Backward induction.
Eric Pacuit and Olivier Roy 14
![Page 42: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/42.jpg)
Basics of Game Theory
Solution Concepts
I Set of profiles or outcome of the game that are intuitivelyviewed as “rational”.
I Three well-known solution concepts in the matrix:
• Nash Equilibrium.• Iterated elimitation of:
I Strictly dominated strategies.I Weakly dominated strategies.
I In the tree we will focus on one:
• Backward induction.
Eric Pacuit and Olivier Roy 14
![Page 43: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/43.jpg)
Basics of Game Theory
Nash Equilibrium
A B
a 1, 1 0, 0
b 0, 0 1, 1
I The profile aA is a Nash equilibrium of that game.
DefinitionA strategy profile σ is a Nash equilibrium iff for all i and all s ′i 6= σi :
ui (σ) ≥ ui (si , σ−i )
Eric Pacuit and Olivier Roy 15
![Page 44: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/44.jpg)
Basics of Game Theory
Nash Equilibrium
A B
a 1, 1 0, 0
b 0, 0 1, 1
I The profile aA is a Nash equilibrium of that game.
DefinitionA strategy profile σ is a Nash equilibrium iff for all i and all s ′i 6= σi :
ui (σ) ≥ ui (si , σ−i )
Eric Pacuit and Olivier Roy 15
![Page 45: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/45.jpg)
Basics of Game Theory
Some Facts about Nash Equilibrium
I Nash equilibria in Pure Strategies do not always exist.I Every game in strategic form has a Nash equilibrium in mixed
strategies.
• The proof of this make use of Kakutani’s Fixed point thm.
I Some games have multiple Nash equilibria.
Eric Pacuit and Olivier Roy 16
![Page 46: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/46.jpg)
Basics of Game Theory
von Neumann’s minimax theoremFor every two-player zero-sum game with finite strategy sets S1 andS2, there is a number v , called the value of the game such that:
v = maxp∈∆(S1)
minq∈∆(S2)
u1(s1, s2)
= minq∈∆(S2)
maxp∈∆(S1)
u1(s1, s2)
Furthermore, a mixed strategy profile (s1, s2) is a Nash equilibriumif and only if
s1 ∈ argmaxp∈∆(S1) minq∈∆(S2)
u1(p, q)
s2 ∈ argmaxq∈∆(S2) minp∈∆(S1)
u1(p, q)
Finally, for all mixed Nash equilibria (p, q), u1(p, q) = v
Eric Pacuit and Olivier Roy 17
![Page 47: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/47.jpg)
Basics of Game Theory
Strictly Dominated Strategies
A
SD A
D 3, 3 1, 4
A 4,1 2, 2
Eric Pacuit and Olivier Roy 18
![Page 48: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/48.jpg)
Basics of Game Theory
Strictly Dominated Strategies
A
SD A
D 3, 3 1, 4
A 4,1 2, 2
Eric Pacuit and Olivier Roy 18
![Page 49: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/49.jpg)
Basics of Game Theory
Strictly Dominated Strategies
A
B
In general, the idea applies to both mixed and pure strategies.
Eric Pacuit and Olivier Roy 19
![Page 50: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/50.jpg)
Basics of Game Theory
Strictly Dominated Strategies
A
B
In general, the idea applies to both mixed and pure strategies.
Eric Pacuit and Olivier Roy 19
![Page 51: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/51.jpg)
Basics of Game Theory
Strictly Dominated Strategies
A
B
> > > > >
In general, the idea applies to both mixed and pure strategies.
Eric Pacuit and Olivier Roy 19
![Page 52: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/52.jpg)
Basics of Game Theory
Strictly Dominated Strategies
A
B
> > > > >
In general, the idea applies to both mixed and pure strategies.
Eric Pacuit and Olivier Roy 19
![Page 53: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/53.jpg)
Basics of Game Theory
Iterated Elimination of Strictly Dominated Strategies
Bob
AnnU L R
U 1,2 0,1 U
D 0,1 1,0 U
Eric Pacuit and Olivier Roy 20
![Page 54: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/54.jpg)
Basics of Game Theory
Iterated Elimination of Strictly Dominated Strategies
Bob
AnnU L R
U 1,2 0,1 U
D 0,1 1,0 U
Eric Pacuit and Olivier Roy 20
![Page 55: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/55.jpg)
Basics of Game Theory
Iterated Elimination of Strictly Dominated Strategies
Bob
AnnU L R
U 1,2 0,1 U
D 0,1 1,0 U
Eric Pacuit and Olivier Roy 20
![Page 56: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/56.jpg)
Basics of Game Theory
Iterated Elimination of Strictly Dominated Strategies
Bob
AnnU L R
U 1,2 0,1 U
D 0,1 1,0 U
Eric Pacuit and Olivier Roy 20
![Page 57: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/57.jpg)
Basics of Game Theory
Facts about IESDS
I The algorithm always terminates on finite games. Intuition:this is a decreasing (in fact, monotonic) function onsub-games. It thus has a fixed-point by the Knaster-Tarskithm.
I The algorithm is order independent: One can eliminate SDSone player at the time, in difference order, or allsimultaneously. The fixed-point of the elimination procedurewill always be the same.
I All Nash equilibria survive IESDS. But not all profile thatsurvive IESDS are Nash equilibria.
Eric Pacuit and Olivier Roy 21
![Page 58: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/58.jpg)
Basics of Game Theory
Weak Dominance
A
B
I All strictly dominated strategies are weakly dominated.
Eric Pacuit and Olivier Roy 22
![Page 59: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/59.jpg)
Basics of Game Theory
Weak Dominance
A
B
I All strictly dominated strategies are weakly dominated.
Eric Pacuit and Olivier Roy 22
![Page 60: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/60.jpg)
Basics of Game Theory
Weak Dominance
A
B
> = > = =
I All strictly dominated strategies are weakly dominated.
Eric Pacuit and Olivier Roy 22
![Page 61: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/61.jpg)
Basics of Game Theory
Weak Dominance
A
B
> = > = =
I All strictly dominated strategies are weakly dominated.
Eric Pacuit and Olivier Roy 22
![Page 62: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/62.jpg)
Basics of Game Theory
Iterated Elimination of Weakly Dominated Strategies
Bob
AnnU L R
U 1,2 0,1 U
D 0,1 1,1 U
Eric Pacuit and Olivier Roy 23
![Page 63: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/63.jpg)
Basics of Game Theory
Iterated Elimination of Weakly Dominated Strategies
Bob
AnnU L R
U 1,2 0,1 U
D 0,1 1,1 U
Eric Pacuit and Olivier Roy 23
![Page 64: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/64.jpg)
Basics of Game Theory
Iterated Elimination of Weakly Dominated Strategies
Bob
AnnU L R
U 1,2 1,1 U
D 0,1 1,1 U
Eric Pacuit and Olivier Roy 23
![Page 65: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/65.jpg)
Basics of Game Theory
Iterated Elimination of Weakly Dominated Strategies
Bob
AnnU L R
U 1,2 0,1 U
D 0,1 1,1 U
Eric Pacuit and Olivier Roy 23
![Page 66: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/66.jpg)
Basics of Game Theory
Facts about IEWDS
I The algorithm always terminates on finite games.
I The algorithm is order dependent!: Eliminating simultaneouslyall WDS at each round need not to lead to the same result aseliminating only some of them.
I Not all Nash equilibria survive IESDS.
Eric Pacuit and Olivier Roy 24
![Page 67: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/67.jpg)
The Epistemic View on Games
Hey, no, equilibrium is not the way to look at games.Now, Nash equilibrium is king in game theory. Absolutelyking. We say: No, Nash equilibrium is an interestingconcept, and its an important concept, but its not themost basic concept. The most basic concept should be:to maximise your utility given your information. Its in agame just like in any other situation. Maximise yourutility given your information!
Robert Aumann, 5 Questions on Epistemic Logic, 2010
Two views on games:
I Based on solution Concepts.
I Classical, decision-theoretic.
Eric Pacuit and Olivier Roy 25
![Page 68: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/68.jpg)
The Epistemic View on Games
Hey, no, equilibrium is not the way to look at games.Now, Nash equilibrium is king in game theory. Absolutelyking. We say: No, Nash equilibrium is an interestingconcept, and its an important concept, but its not themost basic concept. The most basic concept should be:to maximise your utility given your information. Its in agame just like in any other situation. Maximise yourutility given your information!
Robert Aumann, 5 Questions on Epistemic Logic, 2010
Two views on games:
I Based on solution Concepts.
I Classical, decision-theoretic.
Eric Pacuit and Olivier Roy 25
![Page 69: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/69.jpg)
The Epistemic View on Games
Hey, no, equilibrium is not the way to look at games.Now, Nash equilibrium is king in game theory. Absolutelyking. We say: No, Nash equilibrium is an interestingconcept, and its an important concept, but its not themost basic concept. The most basic concept should be:to maximise your utility given your information. Its in agame just like in any other situation. Maximise yourutility given your information!
Robert Aumann, 5 Questions on Epistemic Logic, 2010
Two views on games:
I Based on solution Concepts.
I Classical, decision-theoretic.
Eric Pacuit and Olivier Roy 25
![Page 70: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/70.jpg)
The Epistemic View on Games
Hey, no, equilibrium is not the way to look at games.Now, Nash equilibrium is king in game theory. Absolutelyking. We say: No, Nash equilibrium is an interestingconcept, and its an important concept, but its not themost basic concept. The most basic concept should be:to maximise your utility given your information. Its in agame just like in any other situation. Maximise yourutility given your information!
Robert Aumann, 5 Questions on Epistemic Logic, 2010
Two views on games:
I Based on solution Concepts.
I Classical, decision-theoretic.
Eric Pacuit and Olivier Roy 25
![Page 71: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/71.jpg)
The Epistemic View on Games
Component of a Game
A game in strategic form:
Ann/ Bob L R
T 1, 1 1, 0
B 0, 0 0, 1
A coordination game:
Ann/ Bob L R
T 1, 1 0, 0
B 0, 0 1, 1
G = 〈Ag, {(Si , πi )i∈Ag}〉I Ag is a finite set of
agents.
I Si is a finite set ofstrategies, one for eachagent i ∈ Ag.
I ui : Πi∈AgSi −→ R is apayoff function defined onthe set of outcomes of thegame.
Solutions/recommendations: Nash Equilibrium, Elimination ofstrictly dominated strategies, of weakly dominated strategies...
Eric Pacuit and Olivier Roy 26
![Page 72: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/72.jpg)
The Epistemic View on Games
A Decision Problem: Leonard’s Omelette
Egg Good Egg Rotten
Break with other eggs 4 0
Separate bowl 2 1
I Agent, actions, states, payoffs, beliefs.
I Ex.: Leonard’s beliefs: pL(EG ) = 1/2, pL(ER) = 1/2.
I Solution/recommendations: choice rules. Maximization ofExpected Utility, Dominance, Minmax...
Eric Pacuit and Olivier Roy 27
![Page 73: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/73.jpg)
The Epistemic View on Games
A Decision Problem: Leonard’s Omelette
Egg Good Egg Rotten
Break with other eggs 4 0
Separate bowl 2 1
I Agent, actions, states, payoffs, beliefs.
I Ex.: Leonard’s beliefs: pL(EG ) = 1/2, pL(ER) = 1/2.
I Solution/recommendations: choice rules. Maximization ofExpected Utility, Dominance, Minmax...
Eric Pacuit and Olivier Roy 27
![Page 74: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/74.jpg)
The Epistemic View on Games
A Decision Problem: Leonard’s Omelette
Egg Good Egg Rotten
Break with other eggs 4 0
Separate bowl 2 1
I Agent, actions, states, payoffs, beliefs.
I Ex.: Leonard’s beliefs: pL(EG ) = 1/2, pL(ER) = 1/2.
I Solution/recommendations: choice rules. Maximization ofExpected Utility, Dominance, Minmax...
Eric Pacuit and Olivier Roy 27
![Page 75: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/75.jpg)
The Epistemic View on Games
A Decision Problem: Leonard’s Omelette
Egg Good Egg Rotten
Break with other eggs 4 0
Separate bowl 2 1
I Agent, actions, states, payoffs, beliefs.
I Ex.: Leonard’s beliefs: pL(EG ) = 1/2, pL(ER) = 1/2.
I Solution/recommendations: choice rules. Maximization ofExpected Utility, Dominance, Minmax...
Eric Pacuit and Olivier Roy 27
![Page 76: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/76.jpg)
The Epistemic View on Games
The Epistemic or Bayesian View on Games
I Traditional game theory:Actions, outcomes, preferences, solution concepts.
I Decision theory:Actions, outcomes, preferences beliefs, choice rules.
I Epistemic game theory:
Eric Pacuit and Olivier Roy 28
![Page 77: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/77.jpg)
The Epistemic View on Games
The Epistemic or Bayesian View on Games
I Traditional game theory:Actions, outcomes, preferences, solution concepts.
I Decision theory:Actions, outcomes, preferences beliefs, choice rules.
I Epistemic game theory:Actions, outcomes, preferences, beliefs, choice rules.
Eric Pacuit and Olivier Roy 28
![Page 78: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/78.jpg)
The Epistemic View on Games
The Epistemic or Bayesian View on Games
I Traditional game theory:Actions, outcomes, preferences, solution concepts.
I Decision theory:Actions, outcomes, preferences beliefs, choice rules.
I Epistemic game theory::= (interactive) decision problem and choice rule +higher-order information.
Eric Pacuit and Olivier Roy 28
![Page 79: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/79.jpg)
Basics of Decision Theory
Eric Pacuit and Olivier Roy 29
![Page 80: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/80.jpg)
Basics of Decision Theory
A Decision Problem: Leonard’s Omelette
ui P ¬P
A 4 0
B 2 1
pi P ¬P
A 1/8 3/8
B 1/8 3/8
I Actions, states, payoffs, beliefs.
I Solution/recommendations: choice rules.
• Which choice rule is normatively or descriptively appropriatedepends on what kind of information are at the agent’sdisposal, and what kind of attitude she has.
Eric Pacuit and Olivier Roy 30
![Page 81: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/81.jpg)
Basics of Decision Theory
A Decision Problem: Leonard’s Omelette
ui P ¬P
A 4 0
B 2 1
pi P ¬P
A 1/8 3/8
B 1/8 3/8
I Actions, states, payoffs, beliefs.
I Solution/recommendations: choice rules.
• Which choice rule is normatively or descriptively appropriatedepends on what kind of information are at the agent’sdisposal, and what kind of attitude she has.
Eric Pacuit and Olivier Roy 30
![Page 82: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/82.jpg)
Basics of Decision Theory
A Decision Problem: Leonard’s Omelette
ui P ¬P
A 4 0
B 2 1
pi P ¬P
A 1/8 3/8
B 1/8 3/8
I Actions, states, payoffs, beliefs.
I Solution/recommendations: choice rules.
• Which choice rule is normatively or descriptively appropriatedepends on what kind of information are at the agent’sdisposal, and what kind of attitude she has.
Eric Pacuit and Olivier Roy 30
![Page 83: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/83.jpg)
Basics of Decision Theory
A Decision Problem: Leonard’s Omelette
ui P ¬P
A 4 0
B 2 1
pi P ¬P
A 1/8 3/8
B 1/8 3/8
I Actions, states, payoffs, beliefs.
I Solution/recommendations: choice rules.
• Which choice rule is normatively or descriptively appropriatedepends on what kind of information are at the agent’sdisposal, and what kind of attitude she has.
Eric Pacuit and Olivier Roy 30
![Page 84: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/84.jpg)
Basics of Decision Theory
A Decision Problem: Leonard’s Omelette
ui P ¬P
A 4 0
B 2 1
pi P ¬P
A 1/8 3/8
B 1/8 3/8
I Actions, states, payoffs, beliefs.
I Solution/recommendations: choice rules.
• Which choice rule is normatively or descriptively appropriatedepends on what kind of information are at the agent’sdisposal, and what kind of attitude she has.
Eric Pacuit and Olivier Roy 30
![Page 85: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/85.jpg)
Basics of Decision Theory
A Decision Problem: Leonard’s Omelette
ui P ¬P
A 4 0
B 2 1
pi P ¬P
A 1/8 3/8
B 1/8 3/8
I Actions, states, payoffs, beliefs.I Solution/recommendations: choice rules.
• Which choice rule is normatively or descriptively appropriatedepends on what kind of information are at the agent’sdisposal, and what kind of attitude she has.
Eric Pacuit and Olivier Roy 30
![Page 86: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/86.jpg)
Basics of Decision Theory
Decision Under Risk
When the agent has probabilistic beliefs, or that her beliefs can berepresented probabilistically.
ui P ¬P
A 4 0
B 2 1
pi P ¬P
A 1/8 3/8
B 1/8 3/8
Expected Utility: Given an agent’s beliefs and desires, theexpected utility of an action leading to a set of outcomes Out is:
∑o∈Out
[ subjective prob. of o]× [utility of o]
Eric Pacuit and Olivier Roy 31
![Page 87: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/87.jpg)
Basics of Decision Theory
Why don’t we just give our best guess of wet or dry?Often people want to make a decision, such as whetherto put out their washing to dry, and would like us to givea simple yes or no. However, this is often a simplificationof the complexities of the forecast and may not beaccurate.
By giving PoP we give a more honest opinionof the risk and allow you to make a decision dependingon how much it matters to you. For example, if you arejust hanging out your sheets that you need next week youmight take the risk at 40% probability of precipitation,whereas if you are drying your best shirt that you needfor an important dinner this evening then you might nothang it out at more than 10% probability. PoP allowsyou to make the decisions that matter to you.
http: // www. metoffice. gov. uk/ news/ in-depth/
science-behind-probability-of-precipitation
Eric Pacuit and Olivier Roy 32
![Page 88: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/88.jpg)
Basics of Decision Theory
Why don’t we just give our best guess of wet or dry?Often people want to make a decision, such as whetherto put out their washing to dry, and would like us to givea simple yes or no. However, this is often a simplificationof the complexities of the forecast and may not beaccurate. By giving PoP we give a more honest opinionof the risk and allow you to make a decision dependingon how much it matters to you.
For example, if you arejust hanging out your sheets that you need next week youmight take the risk at 40% probability of precipitation,whereas if you are drying your best shirt that you needfor an important dinner this evening then you might nothang it out at more than 10% probability. PoP allowsyou to make the decisions that matter to you.
http: // www. metoffice. gov. uk/ news/ in-depth/
science-behind-probability-of-precipitation
Eric Pacuit and Olivier Roy 32
![Page 89: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/89.jpg)
Basics of Decision Theory
Why don’t we just give our best guess of wet or dry?Often people want to make a decision, such as whetherto put out their washing to dry, and would like us to givea simple yes or no. However, this is often a simplificationof the complexities of the forecast and may not beaccurate. By giving PoP we give a more honest opinionof the risk and allow you to make a decision dependingon how much it matters to you. For example, if you arejust hanging out your sheets that you need next week youmight take the risk at 40% probability of precipitation,whereas if you are drying your best shirt that you needfor an important dinner this evening then you might nothang it out at more than 10% probability.
PoP allowsyou to make the decisions that matter to you.
http: // www. metoffice. gov. uk/ news/ in-depth/
science-behind-probability-of-precipitation
Eric Pacuit and Olivier Roy 32
![Page 90: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/90.jpg)
Basics of Decision Theory
Why don’t we just give our best guess of wet or dry?Often people want to make a decision, such as whetherto put out their washing to dry, and would like us to givea simple yes or no. However, this is often a simplificationof the complexities of the forecast and may not beaccurate. By giving PoP we give a more honest opinionof the risk and allow you to make a decision dependingon how much it matters to you. For example, if you arejust hanging out your sheets that you need next week youmight take the risk at 40% probability of precipitation,whereas if you are drying your best shirt that you needfor an important dinner this evening then you might nothang it out at more than 10% probability. PoP allowsyou to make the decisions that matter to you.
http: // www. metoffice. gov. uk/ news/ in-depth/
science-behind-probability-of-precipitation
Eric Pacuit and Olivier Roy 32
![Page 91: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/91.jpg)
Basics of Decision Theory
Maximization of Expected Utility
Let DP = 〈S ,O, u, p〉 be a decision problem. S is a finite set ofstates and O a set of outcomes. An action a : S −→ O is afunction from states to outcomes, ui a real-valued utility functionon O, and pi a probability measure over S . The expected utilityof a ∈ A with respect to pi is defined as follows:
EUp(a) := Σs∈Sp(s)u(a(s))
An action a ∈ A maximizes expected utility with respect to pi
provided for all a′ ∈ A, EUp(a) ≥ EUp(a′). In such a case, we alsosay a is a best response to p in game DP.
Eric Pacuit and Olivier Roy 33
![Page 92: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/92.jpg)
Basics of Decision Theory
Decision under Ignorance
What to do when the agent cannot assign probabilities states? Orwhen we can’t represent his beliefs probabilistically? Manyalternatives proposed:
I Dominance Reasoning
I Admissibility
I Minimax
I ...
Eric Pacuit and Olivier Roy 34
![Page 93: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/93.jpg)
Basics of Decision Theory
Dominance Reasoning
A
B
> > > > >
Eric Pacuit and Olivier Roy 35
![Page 94: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/94.jpg)
Basics of Decision Theory
Some facts about strict dominance
I Strict dominance is downward monotonic: If ai is strictlydominated with respect to X ⊆ S and X ′ ⊆ X , then ai isstrictly dominated with respect to X ′.
• Intuition: the condition of being strictly dominated can bewritten down in a first-order formula of the form ∀xϕ(x),where ϕ(x) is quantifier-free. Such formulas are downwardmonotonic: If M, s |= ∀xϕ(x) and M′ ⊆M thenM′, s |= ∀xϕ(x)
Eric Pacuit and Olivier Roy 36
![Page 95: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/95.jpg)
Basics of Decision Theory
Some facts about strict dominance
I Strict dominance is downward monotonic: If ai is strictlydominated with respect to X ⊆ S and X ′ ⊆ X , then ai isstrictly dominated with respect to X ′.
• Intuition: the condition of being strictly dominated can bewritten down in a first-order formula of the form ∀xϕ(x),where ϕ(x) is quantifier-free. Such formulas are downwardmonotonic: If M, s |= ∀xϕ(x) and M′ ⊆M thenM′, s |= ∀xϕ(x)
Eric Pacuit and Olivier Roy 36
![Page 96: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/96.jpg)
Basics of Decision Theory
Some facts about strict dominance
I Relation with MEU:Suppose that G = 〈N, {Si}i∈N , {ui}i∈N〉 is a strategic game.A strategy si ∈ Si is strictly dominated (possibly by a mixedstrategy) with respect to X ⊆ S−i iff there is no probabilitymeasure p ∈ ∆(X ) such that si is a best response withrespect to p.
Eric Pacuit and Olivier Roy 37
![Page 97: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/97.jpg)
Basics of Decision Theory
Some facts about admissibility
I Admissibility is NOT downward monotonic: If ai is notadmissible with respect to X ⊆ S and X ′ ⊆ X , it can be thatai is admissible with respect to X ′.
• Intuition: the condition of being inadmissible can be writtendown in a first-order formula of the form ∀xϕ(x) ∧ ∃xψ(x),where ϕ(x) and ψ(x) are quantifier-free. The existentialquantifier breaks the downward monotonicity.
Eric Pacuit and Olivier Roy 38
![Page 98: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/98.jpg)
Basics of Decision Theory
Some facts about admissibility
I Admissibility is NOT downward monotonic: If ai is notadmissible with respect to X ⊆ S and X ′ ⊆ X , it can be thatai is admissible with respect to X ′.
• Intuition: the condition of being inadmissible can be writtendown in a first-order formula of the form ∀xϕ(x) ∧ ∃xψ(x),where ϕ(x) and ψ(x) are quantifier-free. The existentialquantifier breaks the downward monotonicity.
Eric Pacuit and Olivier Roy 38
![Page 99: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/99.jpg)
Basics of Decision Theory
Some facts about admissibility
I Relation with MEU:Suppose that G = 〈N, {Si}i∈N , {ui}i∈N〉 is a strategic game.A strategy si ∈ Si is weakly dominated (possibly by a mixedstrategy) with respect to X ⊆ S−i iff there is no full supportprobability measure p ∈ ∆>0(X ) such that si is a bestresponse with respect to p.
Eric Pacuit and Olivier Roy 39
![Page 100: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/100.jpg)
Road Map again
1. Today Basic Concepts.
• Basics of Game Theory.• The Epistemic View on Games.• Basics of Decision Theory
2. Tomorrow Epistemics.
• Logical/qualitative models of beliefs, knowledge andhigher-order attitudes.
• Probabilistic/quantitative models of beliefs, knowledge andhigher-order attitudes.
Eric Pacuit and Olivier Roy 40
![Page 101: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/101.jpg)
Road Map again
1. Today Basic Concepts.
• Basics of Game Theory.• The Epistemic View on Games.• Basics of Decision Theory
2. Tomorrow Epistemics.
• Logical/qualitative models of beliefs, knowledge andhigher-order attitudes.
• Probabilistic/quantitative models of beliefs, knowledge andhigher-order attitudes.
Eric Pacuit and Olivier Roy 40
![Page 102: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/102.jpg)
Formal Definitions
Strategic Games
DefinitionA game in strategic form G is a tuple 〈A,Si , ui 〉 such that :
I A is a finite set of agents.
I Si is a finite set of actions or strategies for i . A strategyprofile σ ∈ Πi∈ASi is a vector of strategies, one for each agentin I . The strategy si which i plays in the profile σ is noted σi .
I ui : Πi∈ASi −→ R is an utility function that assigns to everystrategy profile σ ∈ Πi∈ASi the utility valuation of that profilefor agent i .
Eric Pacuit and Olivier Roy 41
![Page 103: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/103.jpg)
Formal Definitions
Extensive form games
DefinitionA game in extensive form T is a tuple 〈I ,T , τ, {ui}i∈I 〉 such that:
I T is finite set of finite sequences of actions, called histories,such that:
• The empty sequence ∅, the root of the tree, is in T .• T is prefix-closed: if (a1, . . . , an, an+1) ∈ T then
(a1, . . . , an) ∈ T .
I A history h is terminal in T whenever it is the sub-sequence ofno other history h′ ∈ T . Z denotes the set of terminalhistories in T .
I τ : (T − Z ) −→ I is a turn function which assigns to everynon-terminal history h the player whose turn it is to play at h.
I ui : Z −→ R is a payoff function for player i which assigns i ’spayoff at each terminal history.
Eric Pacuit and Olivier Roy 42
![Page 104: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/104.jpg)
Formal Definitions
Strategies
Definition
I A strategy si for agent i is a function that gives, for everyhistory h such that i = τ(h), an action a ∈ A(h). Si is the setof strategies for agent i .
I A strategy profile σ ∈ Πi∈ISi is a combination of strategies,one for each agent, and σ(h) is a shorthand for the action asuch that a = σi (h) for the agent i whose turn it is at h.
I A history h′ is reachable or not excluded by the profile σ fromh if h′ = (h, σ(h), σ(h, σ(h)), ...) for some finite number ofapplication of σ.
I We denote uhi (σ) the value of utili at the unique terminal
history reachable from h by the profile σ.
Eric Pacuit and Olivier Roy 43
![Page 105: Epistemic Game Theory Lecture 1 - ESSLLI'12, Opole - Eric Pacuit](https://reader031.vdocuments.mx/reader031/viewer/2022021007/6203913eda24ad121e4aea71/html5/thumbnails/105.jpg)
Formal Definitions
Nash Equilibrium - General Definition
DefinitionA profile of mixed strategy σ is a Nash equilibrium iff for all i andall mixed strategy σ′i 6= σi :
EUi (σi , σ−i ) ≥ EUi (σ′i , σ−i )
Where EUi , the expected utility of the strategy σi against σ−i iscalculated as follows (σ = (σi , σ−i )):
EUi (σ) = Σs∈ΠjSj
((Πj∈Agσj(sj))ui (s)
)
Eric Pacuit and Olivier Roy 44