Strategic games 1

Atiye Sarabi-Jamab
Presented
email:by: Atiye Sarabi-Jamab, Ph.D.
[Link]@[Link]
Institute for Research in Fundamental Sciences (IPM),
School of Cognitive Sciences,
e-mails: [Link]@[Link], [Link]@[Link] 1
Nash equilibrium

2
 How does a rational individual choose?
➢ The theory of rational choice is that a decision-maker chooses the best action
according to his/her preferences, among all the actions available to his/her.

➢ No qualitative restriction is placed on the decision-maker’s preferences; her

“rationality” lies in the consistency of her decisions when faced with different
sets of available actions, not in the nature of her likes and dislikes

‫نظریه انتخاب عقالئی‬

‫ حداقل‬،‫ عملی است که با توجه به ترجیحات او‬،‫عمل انتخاب شده توسط یک بازیکن‬
.‫به خوبی هر عمل دیگری است که در دسترس او قرار دارد‬

3
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

how would empirical testing of these kinds of axiomatic models work?

4
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

how would empirical testing of these kinds of axiomatic models work?

5
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

how would empirical testing of these kinds of axiomatic models work?

‫رابطه ترجیح‬
‫𝑥 مطلوب تر مساوی 𝑦 است‬

6
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

how would empirical testing of these kinds of axiomatic models work?

‫آیا همیشه تابع مطلوبیت وجود دارد؟‬

7
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

how would empirical testing of these kinds of axiomatic models work?

‫آیا تابع مطلوبیت یکتا است؟‬

8
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

how would empirical testing of these kinds of axiomatic models work?

‫مطلوبیت در موقعیت های تصادفی چطور محاسبه می شود؟‬

9
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

how would empirical testing of these kinds of axiomatic models work?

‫مطلوبیت در موقعیت های تصادفی چطور محاسبه می شود؟‬

10
‫مطلوبیت در موقعیت های تصادفی چطور محاسبه می شود؟‬

‫‪11‬‬
 ‫مطلوبیت در موقعیت های تصادفی چطور محاسبه می شود؟‬

‫‪𝟏ൗ‬‬ ‫‪𝟏ൗ‬‬
‫𝟐‬ ‫𝟎‬ ‫𝟐‬

‫)𝟐‪(𝟏Τ𝟐 , 𝟎, 𝟏Τ‬‬
‫‪𝟏ൗ‬‬ ‫‪𝟏ൗ‬‬ ‫‪𝟏ൗ‬‬
‫𝟔‬ ‫𝟐‬ ‫𝟑‬

‫)𝟑‪(𝟏Τ𝟔 , 𝟏Τ𝟐 , 𝟏Τ‬‬

‫‪12‬‬
 ‫مطلوبیت در موقعیت های تصادفی چطور محاسبه می شود؟‬

‫‪𝟏ൗ‬‬ ‫‪𝟏ൗ‬‬
‫𝟐‬ ‫𝟎‬ ‫𝟐‬

‫)𝟐‪(𝟏Τ𝟐 , 𝟎, 𝟏Τ‬‬
‫‪𝟏ൗ‬‬ ‫‪𝟏ൗ‬‬ ‫‪𝟏ൗ‬‬
‫𝟔‬ ‫𝟐‬ ‫𝟑‬

‫)𝟑‪(𝟏Τ𝟔 , 𝟏Τ𝟐 , 𝟏Τ‬‬

‫‪13‬‬
 ‫مطلوبیت در موقعیت های تصادفی چطور محاسبه می شود؟‬

‫‪𝟏ൗ‬‬ ‫‪𝟏ൗ‬‬
‫𝟐‬ ‫𝟎‬ ‫𝟐‬

‫)𝟐‪(𝟏Τ𝟐 , 𝟎, 𝟏Τ‬‬
‫‪𝟏ൗ‬‬ ‫‪𝟏ൗ‬‬ ‫‪𝟏ൗ‬‬
‫𝟔‬ ‫𝟐‬ ‫𝟑‬

‫)𝟑‪(𝟏Τ𝟔 , 𝟏Τ𝟐 , 𝟏Τ‬‬

‫‪14‬‬
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

Von Neumann/ Morgenstern: Expected Utility Theory (1944)

• Axiomatization of utility based on revealed preferences

• Individuals have preferences over probability distributions
over outcomes (can be anything, not just money)
• Probabilities are objective, Nature-given
• Representation theorem: Preference relation obeys
rationality axioms which lead to subjective utility function

15
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

Von Neumann/ Morgenstern: Expected Utility Theory (1944)

16
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

Von Neumann/ Morgenstern: Expected Utility Theory (1944)

17
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

18
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

Von Neumann/ Morgenstern: Expected Utility Theory (1944)

≽ ≽

? ≽ +

19
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
REVEALED PREFERENCE

Von Neumann/ Morgenstern: Expected Utility Theory (1944)

≽ 𝟏ൗ
𝟐 + 𝟏ൗ
𝟐

20
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern: Expected Utility Theory (1944)

21
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern: Expected Utility Theory (1944)

(𝟏, 𝟎, 𝟎)

(𝟎, 𝟏, 𝟎) (𝟎, 𝟎, 𝟏)
22
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern: Expected Utility Theory (1944)

(𝟏, 𝟎, 𝟎)
≽ ≽
U(Pear) = 1 U(Apple) = 0

(𝟎, 𝟏, 𝟎) (𝟎, 𝟎, 𝟏)
23
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern: Expected Utility Theory (1944)

(𝟏, 𝟎, 𝟎)
≽ ≽
U(Pear) = 1 U(Apple) = 0

ℒ ≈𝑝 + (1 − 𝑝)
𝑢(ℒ) ≈ 𝑝𝑢 + (1 − 𝑝)𝑢

(𝟎, 𝟏, 𝟎) (𝟎, 𝟎, 𝟏)
24
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

25
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

?
𝑢 𝑝1 𝐴1 + ⋯ + 𝑝𝑛 𝐴𝑛 = 𝑝1 𝑢 𝐴1 + ⋯ + 𝑝𝑛 𝑢 𝐴𝑛

26
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)
?
𝑢 𝑝1 𝐴1 + ⋯ + 𝑝𝑛 𝐴𝑛 = 𝑝1 𝑢 𝐴1 + ⋯ + 𝑝𝑛 𝑢 𝐴𝑛
0𝐴 ≼ 𝐴 ≼ ⋯ ≼ 𝐴 1
1 2 𝑛
𝑢 𝐴𝑖 = 𝑢𝑖

27
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)
?
𝑢 𝑝1 𝐴1 + ⋯ + 𝑝𝑛 𝐴𝑛 = 𝑝1 𝑢 𝐴1 + ⋯ + 𝑝𝑛 𝑢 𝐴𝑛
0𝐴 ≼ 𝐴 ≼ ⋯ ≼ 𝐴 1
1 2 𝑛
𝑢 𝐴𝑖 = 𝑢𝑖 𝑢𝑖 𝐴𝑛 + (1 − 𝑢𝑖 )𝐴1 = 𝑢𝑖

28
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)
?
𝑢 𝑝1 𝐴1 + ⋯ + 𝑝𝑛 𝐴𝑛 = 𝑝1 𝑢 𝐴1 + ⋯ + 𝑝𝑛 𝑢 𝐴𝑛
0𝐴 ≼ 𝐴 ≼ ⋯ ≼ 𝐴 1
1 2 𝑛
𝑢 𝐴𝑖 = 𝑢𝑖 𝑢𝑖 𝐴𝑛 + (1 − 𝑢𝑖 )𝐴1 = 𝑢𝑖

𝑝1 𝑢 𝐴1 + ⋯ + 𝑝𝑖 𝑢 𝐴𝑖 + ⋯ + 𝑝𝑛 𝑢 𝐴𝑛

29
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)
?
𝑢 𝑝1 𝐴1 + ⋯ + 𝑝𝑛 𝐴𝑛 = 𝑝1 𝑢 𝐴1 + ⋯ + 𝑝𝑛 𝑢 𝐴𝑛
0𝐴 ≼ 𝐴 ≼ ⋯ ≼ 𝐴 1
1 2 𝑛
𝑢 𝐴𝑖 = 𝑢𝑖 𝑢𝑖 𝐴𝑛 + (1 − 𝑢𝑖 )𝐴1 = 𝑢𝑖

𝑝1 𝑢 𝐴1 + ⋯ + 𝑝𝑖 𝑢 𝐴𝑖 + ⋯ + 𝑝𝑛 𝑢 𝐴𝑛
= ෍ 𝑝𝑖 𝑢𝑖 𝐴𝑛 + 1 − 𝑢𝑖 𝐴1
𝑖
30
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)
?
𝑢 𝑝1 𝐴1 + ⋯ + 𝑝𝑛 𝐴𝑛 = 𝑝1 𝑢 𝐴1 + ⋯ + 𝑝𝑛 𝑢 𝐴𝑛
0𝐴 ≼ 𝐴 ≼ ⋯ ≼ 𝐴 1
1 2 𝑛
𝑢 𝐴𝑖 = 𝑢𝑖 𝑢𝑖 𝐴𝑛 + 1 − 𝑢𝑖 𝐴1 ≈ 𝑢𝑖

𝑝1 𝑢 𝐴1 + ⋯ + 𝑝𝑖 𝑢 𝐴𝑖 + ⋯ + 𝑝𝑛 𝑢 𝐴𝑛

= ෍ 𝑝𝑖 𝑢𝑖 𝐴𝑛 + 1 − 𝑢𝑖 𝐴1 = ෍ 𝑝𝑖 𝑢𝑖 𝐴𝑛 + (1 − ෍ 𝑝𝑖 𝑢𝑖 )𝐴1
𝑖 𝑖 𝑖
𝑢 31
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

(𝟏, 𝟎, 𝟎)

ℒ ≈ 𝑝1 + 𝑝2 + 𝑝3 ( )
𝑢(ℒ) ≈ 𝑝1 𝑢 + 𝑝2 𝑢 + 𝑝3 𝑢

(𝟎, 𝟏, 𝟎) (𝟎, 𝟎, 𝟏) 32
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

‫طبق اصل‬
‫استقالل‬
ℒ ≽ ℒሖ ⇔ 𝛼ℒ + 1 − 𝛼 ℒ " ≽ 𝛼 ℒሖ + (1 − 𝛼)ℒ "

33
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

‫طبق اصل‬
‫استقالل‬
ℒ ≽ ℒሖ ⇔ 𝛼ℒ + 1 − 𝛼 ℒ " ≽ 𝛼 ℒሖ + (1 − 𝛼)ℒ "

ℒ ≈ ℒሖ ⇔ 𝛼ℒ + 1 − 𝛼 ℒ " ≈ 𝛼 ℒሖ + (1 − 𝛼)ℒ "

34
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

‫طبق اصل‬
‫استقالل‬
ℒ ≽ ℒሖ ⇔ 𝛼ℒ + 1 − 𝛼 ℒ " ≽ 𝛼 ℒሖ + (1 − 𝛼)ℒ "

ℒ ≈ ℒሖ ⇔ 𝛼ℒ + 1 − 𝛼 ℒ " ≈ 𝛼 ℒሖ + (1 − 𝛼)ℒ "

ℒ ≈ ℒሖ ⇔ 𝛼ℒ + 1 − 𝛼 ℒ ≈ 𝛼 ℒሖ + (1 − 𝛼)ℒ
ℒ
35
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

ℒ ≈ ℒሖ ⇔ 𝛼ℒ + 1 − 𝛼 ℒ ≈ 𝛼 ℒሖ + (1 − 𝛼)ℒ
ℒ

1 36
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

1 37
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

1M

0.5

1M
? 1M

0.5

38
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

1M

0.1

200 K
? 1M

0.9

39
QUANTITATIVE TESTS OF QUALITATIVE THEORIES:
Von Neumann/ Morgenstern:
Expected Utility Theory (1944)

1M

0.000001

2
? 1M

0.9

40
MODELS OF RISKY CHOICE: Descriptive Modifications
by the early 1970s there were a significant number of empirical observations
that could not be accounted for by expected utility (Kahneman and Tversky,
1979).

Subjective Probability
41
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

42
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L 1 0

R 0 2

43
Strictly Competitive games
Definition: A two-Player strategic game is a strictly competitive game if
for any two strategy profile 𝑎, 𝑏 ∈ 𝐴 we have:
𝑢1 𝑎 ≥ 𝑢1 𝑏 ⟺ 𝑢2 (𝑎) ≤ 𝑢2 𝑏
In particular if ∀𝑎 ∈ 𝐴 ∴ 𝑢1 𝑎 + 𝑢2 𝑎 = 0 then the game is a zero-
sum game

44
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L (1,-1) (0,0)

R (0,0) (2,-2)

45
 Nash equilibrium
▪ A Nash equilibrium is an action profile 𝑎∗ with the property that no player i can do better
by choosing an action different from 𝑎𝑖∗ , given that every other player j adheres to 𝑎𝑗∗

In the idealized setting in which the players in any given play of the game are drawn randomly
from a collection of populations, a Nash equilibrium corresponds to a steady state. Expressed
differently, a Nash equilibrium embodies a stable “social norm”: if everyone else adheres to
it, no individual wishes to deviate from it.

The second component of the theory of Nash equilibrium—that the players’ beliefs about
each other’s actions are correct—implies, in particular, that two players’ beliefs about a third
player’s action are the same. For this reason, the condition is sometimes said to be that the
players’ “expectations are coordinated”.

46
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L (1,-1) (0,0)

R (0,0) (2,-2)

47
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L (1,-1) (0,0)

R (0,0) (2,-2)

48
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L (1,-1) (0,0)

R (0,0) (2,-2)

49
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L (1,-1) (0,0)

R (0,0) (2,-2)

50
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L (1) (0)

R (0) (2)

51
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L (1) (0)

R (0) (2)

max min 𝑎𝑖𝑗

𝑖 𝑗

52
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L (1) (0)

R (0) (2)

max min 𝑎𝑖𝑗

𝑖 𝑗
➢ A player’s max-minimizing mixed strategy guarantees her max-minimized payoff.
Furthermore, no strategy guarantees her a higher payoff 53
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

Hider
L1 R2

Chooser
L (1) (0)

R (0) (2)

min max 𝑎𝑖𝑗

𝑗 𝑖

54
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

max min 𝑎𝑖𝑗 ≤ min max 𝑎𝑖𝑗

𝑖 𝑗 𝑗 𝑖

55
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

max min 𝑎𝑖𝑗 ≤ min max 𝑎𝑖𝑗

𝑖 𝑗 𝑗 𝑖

𝑖 ∗ 𝑗∗

56
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

max min 𝑎𝑖𝑗 ≤ min max 𝑎𝑖𝑗

𝑖 𝑗 𝑗 𝑖

𝑖 ∗ 𝑗∗
min 𝑎𝑖 ∗𝑗 max 𝑎𝑖𝑗 ∗
𝑗 𝑖

57
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

max min 𝑎𝑖𝑗 ≤ min max 𝑎𝑖𝑗

𝑖 𝑗 𝑗 𝑖

𝑖 ∗ 𝑗∗
min 𝑎𝑖 ∗𝑗 ≤ 𝑎𝑖 ∗ 𝑗 ∗ ≤ max 𝑎𝑖𝑗 ∗
𝑗 𝑖

58
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

0
max min 𝑎𝑖𝑗
𝑖 𝑗
≤ min max 𝑎𝑖𝑗
𝑗 𝑖
1 Hider
L1 R2

Chooser
∗ L (1) (0)
𝑖 ∗ 𝑗
min 𝑎𝑖 ∗𝑗 ≤ 𝑎𝑖 ∗ 𝑗 ∗ ≤ max 𝑎𝑖𝑗 ∗ R (0) (2)
𝑗 𝑖

59
Max-minimization and NE equilibrium

min 𝑎𝑖 ∗𝑗 = 𝑎𝑖 ∗ 𝑗 ∗ = max 𝑎𝑖𝑗 ∗

𝑗 𝑖

60
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

𝑝𝐻
1 − 𝑝𝐻

61
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

max(𝑝𝐻 , 2(1 − 𝑝𝐻 ))

𝑝𝐻
1 − 𝑝𝐻
𝑝𝐻

62
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

max(𝑝𝐻 , 2(1 − 𝑝𝐻 ))

𝑝𝐻
1 − 𝑝𝐻
𝑝𝐻
2ൗ
3

63
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

min(𝑝𝐶ℎ , 2(1 − 𝑝𝐶ℎ ))

𝑝𝐶ℎ
1 − 𝑝𝐶ℎ 𝑝𝐶ℎ
64
 Let’s back to Strategic Games with Perfect information
▪ Zero-sum game: guess which hand

min(𝑝𝐶ℎ , 2(1 − 𝑝𝐶ℎ ))

𝑝𝐶ℎ
1 − 𝑝𝐶ℎ 𝑝𝐶ℎ
65