Solutions to Chapter 12 Exercises

SOLVED EXERCISES

S1. (a) The payoff table for the two types of travelers follows:

High Low

High 100, 100 30, 70

Low 70, 30 50, 50

(b) The graph is shown below:

(c) There are three possible equilibria: a stable monomorphic equilibrium of all Low types (h
= 0), a stable monomorphic equilibrium of all High types (h = 1), and an unstable polymorphic
equilibrium where two-fifths of the population are High types (h = 0.4).

S2. Throughout the answers for this exercise, we let x represent the population proportion of the
invading type.

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company

 (a) In a population primarily consisting of A types with only a small proportion (x) of
invading T types, the A-type fitness is F(A) = 864(1 – x) + 936x = 864 + 72x and the T-type fitness is
F(T) = 792(1 – x) + 972x = 792 + 180x. F(A) > F(T) as long as 864 + 72x > 792 + 180x, or 72 > 108x, or
x < 72/108 = 2/3. An all-A population can’t be invaded by T types unless the T types are more than two-
thirds of the population, so a small number of mutant Ts cannot successfully invade. Similarly for a small
proportion (x) of invading Ns: F(A) = 864(1 – x) + 1,080x = 864 + 216x and F(N) = 648(1 – x) + 972x =
648 + 324x. F(A) > F(N) as long as 864 + 216x > 648 + 324x, or 216 > 108x, or x < 216/108 = 2. This is
always true because x must be between 0 and 1. Therefore, the A types are always fitter than the N types,
so an all-A population can’t be invaded by Ns.

(b) In a primarily N population, mutant Ts have fitness F(T) = 972(1 – x) + 972x = 972 and
Ns have fitness F(N) = 972(1 – x) + 972x = 972. The fitnesses are equal, so Ts and Ns do equally well in
the population and Ns cannot prevent Ts from invading. A population of Ns invaded by Ts thus exhibits
neutral stability, where both the primary and secondary criteria for an ESS give ties. Since neither type is
more fit than the other, their proportions in the population will persist, only slightly adjusting as
mutations occur.

Against a group of mutant As, the N types have fitness F(N) = 972(1 – x) + 648x = 972 – 324x
and the A types have fitness F(A) = 1,080(1 – x) + 864x = 1,080 – 216x. F(N) > F(A) when 972 – 324x >
1,080 – 216x, or 108 + 108x < 0. This condition never holds, so As can invade an all-N population. An
all-N population is unstable when an A mutation is possible.

(c) In a primarily T population, mutant As have fitness F(A) = 936(1 – x) + 864x = 936 –
72x, and the T-type fitness is F(T) = 972(1 – x) + 792x = 972 – 180x. F(T) > F(A) when 972 – 180x > 936
– 72x, or 36 > 108x, or x < 36/108 = 1/3. An all-T population can’t be invaded by A types unless the A
types are more than one-third of the population, so a small number of mutant As cannot successfully
invade. When the mutants are type N, F(T) = 972(1 – x) + 972x = 972, and the mutant Ns also have
fitness F(N) = 972(1 – x) + 972x = 972. Again, the fitnesses are equal, so Ts and Ns do equally well in the
population, and Ts cannot prevent Ns from invading.

S3. (a) The payoff table follows:

Column

A T

A 20, 20 11, 35
Row
T 35, 11 6, 6
Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company
 (b) Let x be the population proportion of T players. Then the As’ expected years in jail are
20(1 – x) + 11x = 20 – 9x, and the Ts’ expected years in jail are 35(1 – x) + 6x = 35 – 29x. Then the T
type is fitter than the A type if the former spends fewer years in jail: remember that payoffs are years in
jail here, so smaller numbers are better. Ts are fitter than As when 35 – 29x < 20 – 9x, or 15 < 20x, or x >
0.75. If more than 75% of the population is already T, then T players are fitter, and their proportions will
grow; we have one stable ESS when the population is all T. Similarly, if the population starts with fewer
than 75% T types, then the A types are fitter, and their proportion will grow; we have another stable ESS
when the population is all S. Finally, there is another equilibrium in which exactly 75% of the population
is type T and 25% of the population is type A; that equilibrium is polymorphic but unstable.

(c) See the payoff table below. The strategy N does not do well in this game. In a mixed
population that includes all three types, let x and y be the population proportions of T and N, respectively.
Then the As’ expected years in jail equal 20(1 – x – y) + 11x + 2y, the Ts’ expected years in jail are 35(1 –
x – y) + 6x + 6y, and the Ns’ expected years in jail are 50(1 – x – y) + 6x + 6y. T is strictly fitter than N in
this population (remember that smaller numbers are better) and will eventually dominate N. Using the
same equations, we see that if the population consists initially of only N and A, a mutant T can invade
successfully and then eventually dominate N. Even if the population is initially all N, a mutant A will be
able to invade. In that case, for a small proportion a of A types, the Ns get 50a + 6(1 – a) years in jail,
whereas the As get 20a + 2(1 – a); the As are fitter for all values of a, so they can successfully invade.
This analysis implies that N cannot be an ESS in this game. The ESS here are the same as those in part
(b):

Column

A T N

A 20, 20 11, 35 2, 50

Row T 35, 11 6, 6 6, 6

N 50, 2 6, 6 6, 6

S4. (a) The payoff table is shown below:

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company

 Male

Theater Movie

(prop. y) (prop. 1 – y)

Theater
1, 1 0, 0
(prop. x)
Female
Movie
0, 0 2, 2
(prop. 1 – x)

Fitness of female T individual = 1y + 0(1 – y) = y.

Fitness of female M individual = 0y + 2(1 – y) = 2(1 – y).

So for females, the theater type is fitter than the movie type (and the population proportion x of
theater types increases) if y > 2(1 – y) or if y > 2/3. Similarly, for males, the theater type is fitter than the
movie type (and the population proportion y of theater types increases) if x > 2(1 – x), or if x > 2/3.

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company

The diagram below shows the dynamics of the game:

(b) There are two ESS: (0, 0) and (1, 1).

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company

S5. (a) The most obvious mixture to play is the mixed-strategy equilibrium of the two-player
version of the game: report $100 (that is, act like a High type) with probability 0.4.

(b) To calculate expected payoffs for pairs that include a Mixer type, remember that the
Mixer reports $100 with probability 0.4 and $50 with probability 0.6. So when a Mixer plays against a
High type, there is a 40% chance that both report $100 and a 60% chance that there is one report of $100
and one report of $50. The expected payoff of the pairing (Mixer, High) is thus 0.4 * (100, 100) + 0.6 *
(70, 30) = (82, 58). Similarly, the expected payoff of the pairing (Mixer, Low) is 0.4 * (30, 70) + 0.6 *
(50, 50) = (42, 58).

The expected payoff when two Mixer types meet is a little more complicated, since there are four
cases to consider. With probability 0.4*0.4, both Mixers report $100. With probability 0.4*0.6, the first
reports $100 and the second $50. With the same probability, the first reports $50 and the second $100.
Finally, with probability 0.6 * 0.6, they both report $50. The expected payoff for the pairing (Mixer,
Mixer) is thus 0.4 * 0.4 * (100, 100) + 0.4 * 0.6 * (30, 70) + 0.6 * 0.4 * (70, 30) + 0.6 * 0.6 * (50, 50) =
(58, 58).

The three-by-three table of expected payoffs with the Mixer type is then:

High Mixer Low

High 100, 100 58, 82 30, 70

Mixer 82, 58 58, 58 42, 58

Low 70, 30 58, 42 50, 50

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company

 (c) Consider the case when a single High type attempts to invade a population entirely
composed of Mixer types. In this situation we need only the upper-left corner of the payoff table in part
(b):

High Mixer

High 100, 100 58, 82

Mixer 82, 58 58, 58

In terms of m—the proportion of Mixer types in the population—the fitness of the High type is
100(1 – m) + 58m, whereas the fitness of the Mixer type is 82(1 – m) + 58m. When m = 1, it is true that
the fitnesses are equal: 100(0) + 58(1) = 82(0) + 58(1). However, whenever m < 1, the High type has a
higher fitness, because 100(1 – m) + 58m > 82(1 – m) + 58m  18(1 – m) > 0 whenever m < 1. A single
High type will be more fit than the Mixer types in the rest of the population, so the High type will
successfully invade.

Now consider the case when a single Low type attempts to invade a population entirely composed
of Mixer types. Now we need only the lower-right corner of the payoff table in part (b):

Mixer Low

Mixer 58, 58 42, 58

Low 58, 42 50, 50

In terms of m—the proportion of Mixer types in the population—the fitness of the Low type is
58m + 50(1 – m), whereas the fitness of the Mixer type is 58m + 42(1 – m). When m = 1, the fitnesses of
the two types are equal: 58(1) + 50(0) = 58(1) + 42(0). However, whenever m < 1, the Low type has a
higher fitness, because 58m + 50(1 – m) > 58m + 42(1 – m)  8(1 – m) > 0 whenever m < 1. A single
Low type will be more fit than the Mixer types in the rest of the population, so the Low type will
successfully invade.

Since either a High type or a Low type could successfully invade a population of Mixer types, the
Mixer phenotype is not an ESS of this game.

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company

S6. (a) The fitness of the solar type, FS, is 2s + 3(1 – s) = 3 – s, and the fitness of the fossil fuel
type, FFF, is 4s + 2(1 – s) = 2 + 2s. Graphing the fitness curves of the solar type and the fossil fuel type
with respect to s, we have

There are three possible equilibria: an unstable monomorphic equilibrium where everyone uses
fossil fuels (s = 0), an unstable monomorphic equilibrium where everyone uses solar (s = 1), and a stable
polymorphic equilibrium where one-third of the population uses solar (s = 1/3). The monomorphic
equilibria exist as possibilities because in the absence of any alternative types the lone existing type will
simply reproduce in perpetuity.

(b) With the change in the (Solar, Solar) payoff, FS = ys + 3(1 – s) = 3 + (y – 3)s. There is no
change in FFF. At a polymorphic equilibrium, the fitness of the solar type is equal to the fitness of the
fossil fuel type, so 3 + (y – 3)s = 2 + 2s. For the polymorphic equilibrium to occur at s = 0.75, it must be
true that 3 + (y – 3)(0.75) = 2 + 2(0.75)  y = 11/3 ≈ 3.667.

S7. (a) When t = 0.5, the fitness of tortoises is c * 0.5 + (–1) * 0.5 = 0.5c – 0.5, and the fitness of
hares is 1 * 0.5 + 0 * 0.5 = 0.5. The fitness of tortoises will be greater than the fitness of hares when 0.5c
– 0.5 > 0.5  c > 2. That is, when t = 0.5, tortoises will be more fit than hares if c > 2.

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company

 (b) When t = 0.1, the fitness of tortoises is c * 0.1 + (–1) * 0.9 = 0.1c – 0.9, and the fitness of
hares is 1 * 0.1 + 0 * 0.9 = 0.1. Tortoises are more fit when 0.1c – 0.9 > 0.1  c > 10. When t = 0.1,
tortoises will be more fit than hares if c > 10.

(c) When there is a population of pure tortoises (that is, when t = 1), the fitness of tortoises is
c * 1 + (–1) * 0 = c, and the fitness of hares is 1 * 1 + 0 * 0 = 1. If c = 1, tortoises and hares are equally fit
when t = 1, so an invading hare will at least hold its own against the tortoises. But really, after a hare has
invaded, it isn’t quite a population of pure tortoises; that is, t = 1 – ε, where ε is some very small but
positive number. With this in mind and given that c = 1, we find that the fitness of tortoises is 1 * (1 – ε)
+ (–1) * ε = 1 – 2ε, and the fitness of hares is 1 * (1 – ε) + 0 * ε = 1 – ε. Since 1 – ε > 1 – 2ε, hares are
strictly more fit than tortoises once a single hare has invaded. If c = 1, a single hare will indeed
successfully invade a population of pure tortoises.

(d) In terms of t and c, the fitness of tortoises is ct + (–1)(1 – t) = t + ct – 1. In terms of t, the

fitness of hares is 1t + 0(1 – t) = t. Tortoises are more fit than hares when t + ct – 1 > t. Solving for c, we
have c > 1/t. As we saw in parts (a), (b), and (c), for tortoises to be more fit the value of c needs to be
larger as the proportion of tortoises, t, declines.

Note that as t approaches zero, c would need to be infinitely great for a tortoise to be more fit.
Regrettably, a single tortoise in a population of pure hares would never have a chance to have a pleasant
conversation with another tortoise. Thus—no matter the value of c—a single tortoise could never
successfully invade a population of pure hares.

(e) In a polymorphic equilibrium, the fitness of tortoises must equal the fitness of hares. That
is, t + ct – 1 = t  t = 1/c, or c = 1/t. Since t is a proportion, it can only possibly take on values between 0
and 1 (inclusive). When t = 1, c = 1, and when t approaches 0, c approaches infinity. Polymorphic
equilibria are thus possible when 1 ≤ c < ∞. (Note, however, that any polymorphic equilibrium of this
evolutionary game will be unstable.)

S8. (a) The fitness of X, FX, is 2x + 5(1 – x) = 5 – 3x.

The fitness of Y, FY, is 3x + 1(1 – x) = 1 + 2x.

(b) Following the expressions given for FX and FY in part (a) and given that x0 = 0.2, FX0 = 5 –
3(0.2) = 4.4, and FY0 = 1 + 2(0.2) = 1.4.

(c) Following the model, we know that x1 = x0 * FX0/[x0 * FX0 + (1 – x0) * FY0], so x1 = (0.2 *
4.4)/[0.2 * 4.4 + 0.8 * 1.4] = 0.44.

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company

 (d) Given that x1 = 0.44, FX1 = 5 – 3(0.44) = 3.68, and FY1 = 1 + 2(0.44) = 1.88.

(e) We know that x2 = x1 * FX1/[x1 * FX1 + (1 – x1) * FY1], so x2 = (0.2 * 3.68)/[0.2 * 3.68 + 0.8
* 1.88] ≈ 0.60599.

(f) Given that x2 ≈ 0.60599, FX2 ≈ 5 – 3(0.60599) = 3.18203, and FY2 ≈ 1 + 2(0.60599) =
2.21198.

S9. (a) The fitness of the purple type, Fp, is 3g + 2(1 – g) = 2 + g.

(b) The fitness of the green type in terms of a, Fg(a), is a * g + 4(1 – g) = 4 + (a – 4)g.

(c) For there to be a polymorphic equilibrium, it must be the case that the fitness of purple
types is equal to the fitness of green types for some proportion of greens, g*, less than 1. For this
polymorphic equilibrium to be stable, green types must be more fit for values of g less than g*, and purple
types must be more fit for values of g greater than g*. That is, for the polymorphic equilibrium to be
stable the game must be similar to a game of Chicken rather than an assurance-type game (see Figures
13.7 and 13.9).

Graphing the fitness curve of the purple type and multiple fitness curves for green types of
varying levels of a helps us to visualize the mechanics of the game, as seen in the figure below:

The fitness of the purple type, Fp, is equal to the fitness of a given green type, Fg, whenever the
two fitness curves intersect; their point of intersection occurs at some g* < 1 whenever a < 3. That is,
there will be a polymorphic equilibrium whenever a < 3. (When a  3 there will be a monomorphic
equilibrium of all green types.) As for stability of the polymorphic equilibrium, we see that green types

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company

will always be more fit than the purple type for values of g < g* no matter the value of a. Furthermore,
when g > g* the purple type is more fit than the green type, so whenever a polymorphic equilibrium
occurs in this game, it will be a stable one.

(d) The equilibrium proportion of green types, g*, in the stable polymorphic equilibrium is
the one where the fitness of the purple type is equal to the fitness of the green type:

Note that the expression for g* makes sense only as a proportion (which by definition must be
bounded between 0 and 1) when a ≤ 3. Reassuringly, this agrees nicely with the answer in part (c).

S10. Suppose the strategies are S1, S2, . . . , Sn, and the payoff of a player genetically programmed to
play i when matched against one genetically programmed to play j is Aij. Suppose the proportions of the
types are p1, p2, . . . , pn. Then the fitness of type i is

Fi = Ai1p1 + Ai2p2 + · · · + Ainpn.

If strategy 1 is strictly dominated by strategy 2, then A1j < A2j for all j. The proportions sum to 1,
so at least one of them must be positive. Therefore F1 < F2. Then even a small initial proportion of two
players will grow at the expense of that of one player and eventually drive the latter proportion to zero.
Next, suppose strategy 1 is weakly dominated by strategy 2. The two can have equal fitnesses if the
population consists of only those types against which the two have equal payoffs, but not if the population
has positive proportions of all types. For example, suppose A1j = A2j for j = 1, 2, 3 and A1j < A2j for j = 4,
5, . . . , n. Then type 1 can survive in a population that consists only of types 1, 2, and 3, but not in a
population that has positive proportions of types 4, 5, . . . , n.

Games of Strategy, Fourth Edition Copyright © 2015 W. W. Norton & Company