Chapter 12: Game Theory II: Nonzero-sum and Cooperative Games Flashcards
Nonzero-sum Games
Games where the fixed amount of utility one player wins is not the same as the amount of utility the opponent loses.
Example of a Nonzero-sum Game
Prisoner’s Dilemma.
Why is Prisoner’s Dilemma a Nonzero-sum Game?
- The outcome of one player’s strategy is not the opposite for the other player.
- In a zero-sum game the outcome of one player should be the opposite (negative value) for the other player such that the equilibrium utility = 0.
- In the Prisoner’s Dilemma, if one prisoner receives 1 yr, the other player doesn’t receive -1 yr., they receive 20 years.
- Perhaps the most important point as to why the Prisoner’s Dilemma is a nonzero-sum game, is because each outcome for the players are going to be NEGATIVE.
- The outcomes of Row and Col. will never be zero because they are all negative (-10, -10 / -2, -2 / -1, -20 / -20, -1)
Find the Equilibrium in Prisoner’s Dilemma (SEE NOTES)
Explain the Stag Hunt Game:
- There are 2 pre-historic hunters, Row and Col.
- Each have the option of hunting stag or hare.
- There are more bountiful gains that come with hunting stag. However, hunting stag requires that both hunters hunt the stag.
- If both hunters agree to hunt the stag, they’ll each receive 25 kilos of meat.
- Stag hunting CANNOT be done alone. Hunting stag alone will result in 0 kilos of meat.
- The hunters may also choose to hunt hares individually. This is not a task that needs to be done collectively.
- If the hunters agree to hunt hares by themselves, they’ll each receive 5 kilos of meat.
Find the Equilibrium Strategy in the Stag Hunt, then flip card over to see if your answer was correct (SEE NOTES)
There are actually TWO equilibrium strategies: (R1, C1) stag-stag, and (R2, C2) hare-hare.
In both (____,____) and (____,____), the strategies by both _____________ are ____________ against the strategies of the other __________, where ____________ means means _____-_________-______-________-_____.
(R1, C1); (R2, C2); players; optimal; player; optimal; at-least-as-good-as
Is the Stag Hunt a Nonzero-sum game? Answer yes or no, then explain why.
Yes, the Stag Hunt is a Nonzero-sum Game.
- The values of Row’s strategies are not the opposite (negative) values for Col.’s values.
(5,0)
(0,5)
(5,5)
(25,25)
(R1, C1) and (R2, C2) are ____________ strategy _________________. Calculate and find the mixed strategy equilibrium, where the probability assigned to Row and Col.’s strategies is 1/5.
pure; equilibria
To find the equilibrium between the players mixed strategies, we first have to find the probability assignment that would leave Row and Col. indifferent between the expected utility of his mixed strategies, we have to first find the expected utility of his mixed strategies.
- [C1 1/5, C2 1/5]
- p = C1 & p = C2
- 1-p (unity compliment)
- EU(C1) = (25p) +(0 * (1-p))
= (25p)
- EU(C2) = (5p) + (5*(1-p))
= (5p) + (5 - 5p)
= 5 -5p + 5p
= 5
- EU(C1) = EU(C2)
- 25p = 5
- p = 5/25 ~ 1/5
- Since the game is symmetric, such that row and col. face the same strategies and outcomes, 1/5 is also the probability of Row’s mixed strategy equilibrium (playing R1 1/5, and R2 1/5).
What would a Risk Averse decision maker (hunter) choose during the Stag Hunt.
The risk averse hunter would choose only to hunt for Hare. By hunting for hare, the decision maker knows he will guaranteed leave with something.
- Hunting stag poses a moderate risk. Let’s say one hunter is adamant (strategies remain fixed) on hunting stag. If you decide to hunt for hare instead of the stag (thereby defecting since it’s a noncooperative game), you will get 5 kilos, and now the other hunter has run the risk of receiving nothing (0 kilos).
Pareto Efficiency
A state is Pareto efficient iff. the utility level of one player can’t be increased unless the utility level of the other player is decreased.
Explain why (R2, C2) the Hare-Hare pure strategy in the Stag Hunt is not Pareto Efficient.
(R2,C2) is NOT pareto efficient because the players could unilaterally switch to different strategies, specifically Stag-Stag (R1, C1), where the utility level of BOTH players are increased (one of the players is increased without the other player’s utility level being decreased).
Explain why (R1, C1) the Stag-Stag pure strategy in the Stag Hunt is Pareto Efficient.
(R1,C1) is Pareto Efficient because the players can switch to different strategies wherein one player would have an increased level of utility (5), and the other player would face a decreased level of utility (0).
Explain Battle of the Sexes Game (SEE NOTES)
- Row (Wife) and Col. (Husband) are newly weds.
- Now that they’re married, they want to maximize on spending time together. They’re deciding between attending the Opera, or going to the Football Game.
- Row would rather attend the Opera than go to the Football Game.
- Col. would rather go to the Football Game than attend the Opera, BUT is willing to go to the Opera in case Row tags along.
- In this game, Row and Col. spending time together is more desirable/ of more utility than not spending time together.
Based on Decision Matrices, find the equilibrium for the Battle of the Sexes (SEE NOTES).
There are actually 2 Equilibria!
Now that we know there are 2 pure strategy equilibria for Battle of the Sexes, calculate and find the Mixed Strategy Equilibrium where R1 2/3, C1 1/3.
Why don’t we assume the from the start that Row and Col, are going to reason the same during Battle of the Sexes? Explain why.
Okay, well, let’s reason as Row for a minute.
- if we look at the matrix, I would rather choose R1 because getting utility 2 is more desirable than getting utility 1 (by unilaterally switching strategies ending up w/ outcome of C2 –> going to football game).
- Now let’s reason that Col, my husband, reasons the same. If this were the case, he would opt to choose C2 (Football) because 2 utils is better than 1 (which he would’ve received if he had chosen C1 –> the opera).
- If we converge on this outcome in the matrix, where Row chooses R1 (going to the Opera) and Col chooses C2 (the football game), this outcome has an expected payoff of (0,0), meaning they wouldn’t be spending any time together which is overall undesirable.