You are on page 1of 6

Shen Dai, Nguyen Son Tung, Li Shilei, Xu Bingzhou

Game theory
Report 3

A closer look at game theory


in animal kingdom
by studying evolutionary game theory

I. Background Information
Evolutionary game theory (EGT) is the application of game theory to interaction
dependent strategy evolution in populations. EGT is useful in a biological context by
defining a framework of strategies in which adaptive features can be modeled. It
originated in 1973 with the formalization of evolutionarily stable strategies as an
application of the mathematical theory of games to biological contexts; arising from
the realization that frequency dependent fitness introduces a strategic aspect to
evolution. EGT differs from classical game theory by focusing on the dynamics of
strategy change more than the properties of strategy equilibria. Despite its name,
evolutionary game theory has become an increasing interest to economists,
sociologists, anthropologists, and philosophers.

One of the underlying principles of evolutionary game theory is that each player
makes individual decisions, and the resulting payoff to each player relies upon the
group decisions made by everyone. As a result, an important question in EGT is
predicting the behavior of other players engaged in any given game.

For EGT to be applicable to organisms, they cannot be following a set of random


rules, but rather a specific strategy that responds to specific pressures. The value of
any particular strategy is always in relation to an organism’s environment. When
applied to an evolutionary context, a payoff for an outcome of a game is analogous to
the fitness of an organism.

EGT assumes large populations who interact in randomly matched pairs in repeated
games. The game is symmetric in two senses. The first is that players have the same
set of strategies to choose from and the payoff for a strategy is the same for any player
or organism, irrespective of the features of the other player or organism that chooses
an alternative strategy.

An important feature of the set of models under the umbrella of evolutionary game
theory is repetition. If the games were not repeated, these EGT models would not be
Shen Dai, Nguyen Son Tung, Li Shilei, Xu Bingzhou

able to provide any insight into adaptive behaviors and strategies due to the dynamic
nature of the mechanisms of evolution. Further, this biological application is
meaningful because it provides an understanding of the adjustments that occur
between equilibriums.

The successful application of game theory to evolution has brought further insights to
human behavior. Whereas game theory traditionally assumes rational actors, in the
real world this does not always describe human behavior. EGT has predicted
behaviors in animals where strong assumptions of rationality cannot be made.

II. Our Game


Consider the story of the desert spider Agelenopsis aperta, found in New Mexico.
Female lays its eggs within a web, but sometimes the webs are scarce because they
are difficult to build. Henceforth female spiders always fight--or almost fight--over an
existing web. The conflict is settled when one spider retreats leaving the other in sole
possession of the web. The following table gives the payoff of the spiders adopting
different strategies.

Spider1/Spider2 Concede Fight


Concede 5, 5 0,10
Fight 10, 0 x, x

In this case, the payoff of a strategy represents the reproductive fitness, which literally
means the expected number of offspring of a spider adopting that strategy. From the
table, x is the payoff of each of the two spiders that choose to fight each other and is
less than 5 as damage to the web, injuries, and sometimes death may result during
their fight. With different values of x, it will give different proportions of ‘Hawks’
(spiders always fight against their opponents) and ‘Dove’ (spiders always retreat from
a fight) when the population of spiders reaches evolutionary equilibrium, a point
where the proportions of the two strategies become immune to the invasion of
outsiders, regardless of which strategies they adopt. For instance, will the proportions
of ‘Hawks’ and ‘Dove’ in a population change when it is joined by a group of
"Hawks" from other place, given the payoff of the two strategies? Our focus in this
project is to find out a simple and efficient way to calculate the equilibrium
proportions using Nash Equilibrium. However, in this study we exclude the effect of
mutations which may in reality also play a significant role in affecting the spiders’
behaviors and assume that each individual spider can produce accurate copies of
itself.
Shen Dai, Nguyen Son Tung, Li Shilei, Xu Bingzhou

We will first solve the problem using evolutionary equilibrium, and then solve it again
on a specific range of x values using Nash Equilibrium.

In this game we introduce the strategy set S={F, C} where:

• F is to fight

• C is to concede

Spider1/Spider2 Concede Fight

Concede 5,5 0,10

Fight 10,0 (10-C)/2, (10-C)/2

Where C is the cost of injury to fight (note that x = (G - C)/2) = (10 - C)/2)

Denote: Wij is the payoff of one spider playing i when the other plays j

Fi is the fitness of strategy i

If we have 2 strategies F and C then:

Where p is the proportion of spiders that play C in the population

Next, we will introduce the concept of Evolutionary Stability:

A strategy is evolutionary stable no other strategy can invade it under the influence of
natural selection. We say a strategy i can invade a population of j if Fi > Fj.
Strategy i is an evolutionary stable strategy if, for all strategies j ≠ i:
Wii ≥ Wji
and if Wii = Wji:
Wij > Wjj

Let's get back to our game. Consider the case when C ≤ 10:

Spider1/Spider2 Concede Fight

Concede 5,5 0,10


Shen Dai, Nguyen Son Tung, Li Shilei, Xu Bingzhou

Fight 10,0 (10-C)/2, (10-C)/2

(10 - C)/2 ≥ 0

In this case WFF≥ WCF and WFC≥ WCC, hence F is an evolutionary stable strategy. This
suggests that a population consisting entirely of ‘Hawks’ is not susceptible to the
invasion of ‘Doves’ as all the ‘Doves’ will gradually convert to ‘Hawks’; whereas a
population of ’Doves’ is very susceptible to the invasion of ‘Hawks’ and the entirely
population will convert to ‘Hawks’ after sufficient time.

On the other hand, when C > 10 => (10 - C)/2 < 0

In this case WFF< WCF, hence F is not an evolutionary stable strategy. And since WCC<
WFC, C is not an evolutionary stable strategy either. Hence neither strategy is
evolutionarily stable. Each type of spider will coexist with the other to make up a
certain proportion of the whole population. We need to find the specific proportion of
spiders which play strategy C and the respective proportion of spiders which play
strategy F.

Original approach

One method which has been reported in literature is to calculate the equilibrium
proportions of the two strategies is by using replicator equations.

Replicator equation is a differential equation which describes how the average


frequency of each strategy in a population changes over time. If we express
evolutionary success as the difference between the fitness of a replicator (spider in
this case) and the average fitness in the population, we obtain the following
differential equation:

Where is the relative abundance of the strategy i in a population with n types of


strategies in total, is the rate of change of the frequency of strategy i, is the
average fitness of a replicator of strategy i which depends on the population make-up
x, and is the average fitness in the population.

From the definition, it is not difficult to derive the following relationships:

, , and
Shen Dai, Nguyen Son Tung, Li Shilei, Xu Bingzhou

It is not difficult to see that within the context of our game, n=2 and i={F, C}. By
solving the replicator equation we can find the composition of the spider population at
any instant. For instance, consider when C=15 and G-C<0, we can plot the relative
frequency of the strategies over time as shown in the following diagrams.

It can be seen that in both cases it reaches the same equilibrium regardless of the initial
proportions of the two strategies (for the first case, pi=9/10, second case pi=1/10, and for both
cases pf=1/3). Hence the equilibrium relative proportions of the two strategies are
evolutionary stable.
However, solving differential equations like this is very tedious and impractical, especially
when only the final evolutionary equilibrium is concerned.

Alternative approach

Here we introduce another method finding the final equilibrium proportions by using Nash
Equilibrium for the case in which C>10
Set FF equal to FC, where

and p is the proportion of spiders that play C in the population

Hence we have (1-p)∙(10-C)/2 + 10p = (1-p)∙0 + 5p

Solve the equation, we have p=1/3 when C = 15, which is the same as what we
obtained previously.

If we examine the results closely, it does make sense. Note that


Shen Dai, Nguyen Son Tung, Li Shilei, Xu Bingzhou

FF – FC = 5p –(C-10)/2∙(1-p)

hence when p increases, which means the proportion of ‘Dove’ increases, ‘Hawks’
will have a relatively higher payoff, and subsequently, the proportion of ‘Hawk’ will
also increase and the two strategies will reach equilibrium again. The converse is also
true.

Hence the result obtained using this method is valid.

III. Conclusion
In conclusion, we have applied to concept of Nash Equilibrium in Game theory to
solve for the evolutionary stable strategy of desert spiders Agelenopsis aperta, in
disputes over possession of webs.

In addition, we have found an easy and fast method to calculate the evolutionary
equilibrium of proportions of the ‘Hawk’ and ‘Dove’ strategies in the case where x<0,
which proves to be much more efficient than the traditional method using replicator
equations. This method applies mix-strategy Nash Equilibrium, which we have learnt
and have been tested many times.

You might also like