Rationality and common knowledge

In this section:
1. We study the implications of imposing the assumptions of rationality as well as common
knowledge of rationality.
2. We derive and explore some solution concepts that result from these two assumptions

Some notation:
Denote the payoff of a player i from a profile of strategies s = (s1, s2, . . . , si -1, si , si +1, . . . ,
sn) as vi(s).

The actions chosen by the players who are not player i are denoted by the profile

Without loss of generality, define

as the set of all the strategy sets of all players who are not player i.

We then define:

as a particular possible profile of strategies for all players who are not i.
Hence we can rewrite the payoff of player i from strategy s as:
vi(si, s−i), where s = (si, s−i).

Section 1 : Dominance in Pure Strategies

Dominated strategies

Consider the Prisoner’s dilemma example

Both players have a strategy (an action) that is best regardless of what their opponent
chose.
→ Rationality implies that each player will choose whatever they deem to be best for
them.
→ If it turns out that a player’s best strategy does not depend on the strategies of his
opponents then we should be all the more confident that he will choose it.
It is not too often that we will find ourselves in situations in which we have a best action
that does not depend on the actions of our opponents. We begin, therefore, with a less
demanding concept that follows from rationality.

In particular consider the strategy mum in the Prisoner’s Dilemma: As we argued earlier,
playing M is worse than playing F for each player regardless of what the player’s opponent
does. We say that such a strategy is dominated.
Formally we have:

Now that we have a precise definition for a dominated strategy, it is straightforward to

draw an obvious conclusion:

Claim 4.1 A rational player will never play a strictly dominated strategy.

When we apply the notion of a dominated strategy to the Prisoner’s Dilemma we argue
that each of the two players has one dominated strategy that he should never use, and
hence each player is left with one strategy that is not dominated.
→ Therefore, for the Prisoner’s Dilemma, rationality alone is enough to offer a
prediction about which outcome will prevail: (F, F) is this outcome.

Many games, however, will not be as special as the Prisoner’s Dilemma, and rationality
alone will not suggest a clear-cut, unique prediction.
As an example, consider the following advertising game. Two competing brands can
choose one of three marketing campaigns—low (L), medium (M), and high (H )—with
payoffs given by the following matrix:
It is easy to see that each player has one dominated strategy, which is L. However, neither
M nor H is dominated. Hence rationality alone does not offer a unique prediction.

Dominant strategy equilibrium

Rationality thus implies players will always choose dominant strategies → sensible
predictor of behavior → first interesting solution concept

Definition 4.3 formalizes (F, F) “both players fink” as the dominant strategy equilibrium.

Properties of the SDSE

1. Uniqueness

2. Existence

→ Proposition 4.1 implies that when it exists, the strict-dominance solution concept
guarantees uniqueness.
NOTE: The SDSE may not always exist

Consider the Battle of the Sexes game

→ No strictly dominant strategies here! Thus no equilibrium if we use the solution

concept of strict dominance.

Pareto optimality

Going back to the Prisonner’ dilemma, the unique SDSE is not pareto optimal

“Institutional design,” which changes the game that players play, can be very useful in
affecting the well-being of players.

Assuming a punishing Mafia, the game becomes:

And for z > 1, the SDSE becomes (M, M)

Section 2 : Iterated Elimination of Strictly Dominated
Strategies

We saw, that the strong solution based on strict dominance will often fail to exist.

Consider a less extreme concept that is based on the idea of :

1. dominance, and
2. common knowledge of rationality, but
3. does not require the existence of a strictly dominant strategy.

The concept we introduce here is called Iterated Elimination of Strictly Dominated

Strategies (IESDS),
→ builds on the premise that rational players will not play strictly dominated
strategies.

→ that both the payoff’s of the game and the rationality of the players is common
knowledge,

i,e, If all the players know that each player will never play a strictly dominated strategy,
fining a new
This may reduce the effective strategy space for each of the players, thus de
“smaller” game. Taking this logic further, because it is common knowledge that all players
are rational, then everyone knows that everyone knows that the game is effectively a
smaller game.
This process can be repeated for the reduced game.

We can think of this process as an algorithm that works as follows:

Step 1: Consider the strategy sets of each player.

Step 2: For each player, are there strategies that are strictly dominated?

Step 3: If the answer in 2 is ‘yes’, eliminate these strategies and define a new game, go back
to step 1. If the answer is ‘no’ for all players then go to step 4

Step 4: The remaining strategies are reasonable predictions for behavior.

A Simple Example

Consider the following two-player discrete game:

A quick observation reveals that there is no strictly dominant strategy, neither for player 1 nor
for player 2
However, for player 2: C is strictly dominated by R. Thus, since this is common knowledge,
both players know that effectively we can eliminate C, which results in the following reduced
game:

In this reduced game both M and D are strictly dominated by U for player 1, which could only
be concluded after the first stage of reasoning we did above. Eliminating these two strategies
yields the following trivial game
L R
U 4,3 6,2

in which player 2 has a strictly dominant strategy, playing L. Thus, for this example IESDS
yielded a unique prediction: the strategy profile we expect these players to play is (U, L),
yielding the payoffs of (4,3).

Note : As the example demonstrates, this process of iterated elimination of strictly

dominated strategies (IESDS) builds on the assumption of common knowledge of
rationality.
→ The first step of iterated elimination is a consequence of player 2’s rationality;
→ the second stage follows because players know that players are rational;
→ the third stage follows because players know that players know that they are
rational, and this ends in a unique prediction.
 A More Complicated Example: Cournot Revisited

Cournot Duopoly
linear costs:

Assume that marginal costs (and total costs) of production are zero, i.e.,

c(qi) = 0 for i ∈ {1, 2}

Now consider the profit function of firm 1.

First order condition is :

(1)
From (1), Since q2 ≥ 0, firm 1 will never choose to produce more than 50. implying that
the effective strategy space is q1 ∈ [0, 50]. A similar argument works for firm 2 → first
step of iterated elimination

Second stage of elimination: Since q2 < 50, equation (1) implies that firm 1 will choose a
quantity no less than 25, → qi ∈ [25, 50] for i ∈ {1, 2}.

You should be able to convince yourself that the next stage will reduce production to qi ∈
[25, 37,5]

Interestingly, the set of strategies that survives this process converges to a single
quantity choice of qi = 33,33

PROPERTIES :

- We can apply IESDS to any game. It does not require the existence of a strictly
dominant strategy, nor does it require the existence of strictly dominated strategies.

- When strictly dominated strategies exist, the process of IESDS is able to say
something about how common knowledge of rationality restricts behavior.
 - Unlike strict dominance, an IESDS solution always exists BUT IS NOT UNIQUE. In
the two examples above, IESDS lead to the survival of a unique strategy.

However, consider the battle of the sexes:

IESDS is not useful here: neither O nor F are strictly dominated strategies for the players. As
we can see, this solution concept can be applied (it exists) to any game, but it will often fail to
provide a unique solution. For the Battle of the Sexes game, IESDS can only conclude that
“anything can happen.”

NOTE: Relationship between the IESDS solution concept and the strict-dominance solution
concept:

Resume: Assumption of common knowledge of rationality → IESDS may offer a fine

prediction when strict dominance does not apply. This is exactly what the extra assumption of
common knowledge of rationality delivers: a more widely applicable solution concept.

Section 3 :Beliefs, Best Response, and Rationalizability

Consider the Battle of the Sexes

BEST RESPONSES

If a strategy si is not strictly dominated for player i then it must be that there are
combinations of strategies of player i’s opponents for which the strategy si is player i’s best
choice → rationalize the choice of player i.
In order for a player to be optimizing in a game, he has to choose a best strategy as a response
to the strategies of his opponents. We therefore introduce the following formal definition:

In fact rationality implies that given any belief a player has about his opponents’ behavior,
he must choose an action that is best for him given his beliefs. That is,

NOTE : Concept of playing a best response and the concept of dominated strategies.
→ First, if a strategy si is strictly dominated, it means that some other strategy si is
always better. This leads us to the observation that the strategy si could not be a best
response to anything:

Also, if a strategy sD is a strictly dominant strategy then it must be a best response to

anything i’s opponents can do.

(see proof in book)

Beliefs and best response correspondances

We now need to give players the ability to form conjectures about what others are doing.
→ Introducing the concept of beliefs, and actions that best respond to beliefs, is
central to the analysis of strategic behavior

Player i’s best-response function maps beliefs into a choice of action, and this choice of
action must be a best response to the beliefs.

Note: there are games in which for some beliefs a player will have more than one best-
response strategy (see book p 72)

That is, given a belief player i has about his opponents, s−i , the set of all his possible
strategies that are a best response to s−i is denoted by BRi(s−i). If he has a unique best
response to s−i then BRi(s−i) will contain only one strategy from Si.

Rationalizability

Equipped with the idea of beliefs and a player’s best responses to his beliefs, the next
natural step is to allow the players to reason about which beliefs to have about their
opponents.
→ This reasoning must take into account the rationality of all players, common
knowledge of rationality, and the fact that all the players are trying to guess the
behavior of their opponents.
→ In what follows we introduce another way of reasoning that rules out irrational
behavior with an iterated process that is, in many ways, the mirror image of IESDS.
→ This new concept asks “What might a rational player do?” A rational player will
select only strategies that are a best response to some profile of his opponents. Thus
we have
→ Next step is to use common knowledge of rationality to build an iterative process that
takes this reasoning to the limit.

→ After employing this reasoning one time, we can eliminate all the strategies that are never
a best response, resulting in a possibly “smaller” reduced game that includes only strategies
that can be a best response in the original game.
→ The solution concept of rationalizability is defined precisely by iterating this thought
process. The set of strategy profiles that survive this process is called the set of
rationalizable strategies.

A Discrete Example: the “p-beauty contest”

• Each players i ∈N = {1, 2, ..., n} choose an integer si ∈ Si = {0, 1, ..., 20}

• The set of winners W ⊂ N (may be one or more) are those players that are closest to
3/4 (here p = 3/4) of the average, and the rest are all losers. That is, the winners are

• Each player pays 1 to play the game, and winners split the pot equally among
themselves. This implies that if there are k ≥ 1 winners, each gets a payoff of
(N-k)/k (their share of the pot net of their own contribution) while losers get -1
(they loose their contribution)

Rationalization: look for strategies that are never a best response

1st Round
- If you believe that the average is below 20, then 20 cannot be a best response —
there will be a lower number that is.
- If you believe the average is 20,
→ playing 20 → split the pot with all the other players.
→ If you choose 19, then you will win the whole pot for sure, because for any n,
19 will be closer to (3/4)(1/n)[(n − 1)20 + 19] (which lies between 145/8 (if n = 2) and
15 (for n → ∞))

Note: 19 is not the unique best response to the belief that all others are playing 20
(left as an exercise)

Thus only the numbersi =

S1{ }
0, 1,...,19 survive the first round of rationalizable
behavior.
Similarly after each additional round we will “lose” the highest number

19th round: we are left withi = {

S19 }0,1 → after 19 rounds of dropping strategies that
cannot be a best response → two strategies that survive: 1 and 0.
→ If n > 2, we cannot reduce this set further: if, for example, player i believes
that all the other players are choosing 1 then choosing 1 is a best response for
him. (This is left as an exercise.)
→ Similarly regardless of n, if he believes that everyone is choosing 0 then
choosing 0 is his best response.
→ Thus we are able to predict using rationalizability that players will not
choose a number greater than 1, and if n = 2, then we will predict that both
will choose 0.

Remark : For two-player games IESDS and rationalizability result in the same
outcomes. We discuss this issue briefly in Section 6.3, after we introduce the
concept of mixed strategies.

Summary
- Rational players will never play a dominated strategy and will always play a
dominant strategy when it exists.
- When players share common knowledge of rationality, the only strategies
that are sensible are those that survive IESDS.
- Rational players will always play a best response to their beliefs. Hence any
strategy for which there are no beliefs that justify its choice will never be
chosen.
- Outcomes that survive IESDS, rationalizability, or strict dominance need not
be Pareto optimal, implying that players may not be able to achieve desirable
outcomes if they are left to their own devices.