THE SELFISH GENE by Richard Dawkins – Chapter 5 Animated Summary (Aggression)


The Selfish Gene
Chapter 5 by Richard Dawkins. Aggression: Stability and The Selfish Machine. This
chapter continues the assumption that individual survival machines are
selfishly programmed with behaviours for the welfare of the genes as a whole,
acknowledging that all forms of interaction involved some conflict of
interest. Dawkins delves into the misunderstood
topic aggression exploring how behavioural strategies survive and how
these genes that program their behaviours become more successful in the gene pool.
Survival machines treat other survival machines as part of its environment – it
is an obstacle or something that can be exploited like a rock river or food,
except it can retaliate and hit back. This is because the other survival
machine is also trying to preserve its genes for the future. Natural selection
thereby favours genes that allows the survival machine to utilise the
environment to its benefit. Although some survival machines may not seem to impinge
on each other’s lives, for example moles and blackbirds, we must not treat them
totally as insulated because they could both consume a shared resource, such as
earthworms. Different species may be predators, prey, parasites, or hosts.
Usually survive machines within the same species tend to directly impinge on each
other’s lives as mates, compete for mates, or in the example of blackbirds, compete for
worms and everything else. We would assume that the logical thing for
survival machine to do is to murder its rivals and eat them.
However, murder and cannibalism are not as common as we’d expect.
In his book, “On Aggression”, Konrad Lorenz attributes this to the restrained and
gentlemanly nature of animal fighting – a gloved fist view of animal fights –
compared to the savagery of humans. Dawkins disputes this interpretation and
offers an alternative explanation. He notes that there are cost and merit in
taking out your rival, in addition to the investment in cost and time. If B and C
are both rivals, eliminating B might be advantageous, but C may benefit from this
action. Therefore there is no obvious benefit in indiscriminately removing
rivals. We can liken this to agricultural pest control that tries to exterminate
one pest, only to have a worse problem from another pest who benefits from the
action. Even attempting to kill a rival discriminately might be costly. For
example, in attempting to eliminate a rival elephant seal who is in possession
of a large harem of females, the challenger might end up dead.
Although the rival holds a valuable resource, this may be an indication that he
won it through combat, by beating off other challengers, which means he’s
probably a good fighter! Even if the challenger wins the fight, he may be
badly injured in his attempt. Therefore, the decision to fight or not to fight is
a complex and perhaps unconscious cost-benefit calculation. men Maynard Smith, together with GR Price and GA Parker, uses a branch of mathematics called Game
Theory to analyze these decisions. He introduces the concept of evolutionary
stable strategy. When we talk about a strategy, we are speaking of a
pre-programmed behavioural policy because we are treating animals like survival
machines with a pre-programmed computer controlling the muscles, as discussed in
the last chapter. For example, attack opponent, if he flees, pursue him, if he
retaliates, run away. Note, we are not talking about strategy as a conscious
action worked out by an individual. An evolutionary stable strategy (or ESS)
is defined as a strategy, which if most members of the population adopted it,
cannot be bettered by an alternative strategy. In other words, it is the best
strategy an individual should adopt depending on what the majority of the
population are doing. So if an individual in the population is trying to maximize
its own success, the strategy will be the only one that persists and cannot be
bested by deviant individuals. Individuals with alternative strategies
under ESS selection would penalize deviations. In applying this to
aggression, we can think of two sorts of fighting strategies in a population of a
particular species. Number one, a hawk strategy or number two, a dove strategy.
Dawkins points out that the names have no connection with conventional human
usage, as doves are rather aggressive birds. For the hawks, they fight a hard
and unrestrained and only retreat when seriously injured. Doves threaten
but never hurt anybody. If hawks fight a dove, the dove quickly
runs away and doesn’t get hurt. However, if a hawk fights a hawk, they fight until
one of them is seriously injured or dead. If dove meets dove, nobody gets hurt and they
posture at each other until one of them tires or backs down. We assume that
individuals only find out if a rival is a hawk or dove when they fight, and
cannot tell in advance and have no memory of past fights. If we award points to the
contest, 50 points for winning, zero for losing, minus 100 for serious injury, or minus
10 for wasting time over a long contest, these points can be converted to measure
gene survival. An individual with a high point or higher average payoff can leave
many genes behind in the gene pool. The question here is not whether doves will
beat hawks, we know hawks always win. We want to know
if either hawks or dove is an evolutionary stable strategy. If one of
them is an ESS and the other one is not, the one that is an ESS will evolve. It is
also possible that there are two ESS’s and the population would tend to
stick at one of the two stable states it happened to reach first. However, neither
of these strategies would be evolutionary stable on its own. To find
out why we can suppose the populations compose of doves. Then, the average payoff
per contest is plus 40 and minus 10 depending on who gives up. However, if a
hawk arises in the population, given he is the only hope he will score plus 50
every fight, and the net payoff is only plus 15. as Hawks spread throughout the
population, hawks would encounter other hawks. Each hawk can expect to win half
his fights and the hawk that become seriously injured
would score minus 100, while the winner continues to score plus 50 points. In
comparison, doves that encounter hawks run away and are not injured, with an
average payoff of zero. And in a large population of hawks,
dove genes would spread throughout the population. Over time, there is a stable
ratio of hawks to doves. Under this stable ratio, the average payoff for hawks and
doves are the same, which works out to 5/12 doves and 7/12 hawks, and an average
payoff of 6 and 1/4. Of course, if everybody cooperated and agreed to be
doves, every individual would benefit. For example, if there were 1/4 hawks and 5/6 doves, the average payoff would be 16 and
2/3. Group selection predicts that there will be a tendency towards an all dove
conspiracy. The problem is that this is open to abuse or as Dawkins calls it
“treachery from within” because hawks will do extremely well in this population of
mostly doves and will evolve and populate. Under ESS, although the
average payoff is lower for individuals, it is immune to “treachery from within”.
Humans can use conscious foresight and act in long term interests to obey any
agreement against behaving for short term interests. However, there is a
constant danger and temptation of individuals to exploit short term gains
by breaking the pact. We see this in price-cutting behaviour of petrol
stations. Therefore it is difficult to see how wild
animals, dictated by the genes, could evolve strategies for group benefits. In
a stable ratio of hawk genes to dove genes in a gene pool, the technical term in
genetics for this state is called stable polymorphism. If every individual is
capable behaving like hawk or dove, ESS can still be achieved without
polymorphism, with the same probabilities, namely 7/12 as hawk and 5/12 as dove.
Haw and dove is a very simple model but it is useful for understanding a point.
We can introduce some complex strategies. Maynard Smith and Price introduced a more
complex strategy known as the retaliator. The retaliator is conditional
strategist in that its behaviour depends on the opponent. It acts like a dove in
the beginning by not mounting all-out attacks like a hawk, however, if attacked,
the ‘dove’ will retaliate and behave like a hawk. If the retaliator meets
another retaliator, it plays like a dove. We can introduce two more conditional
strategists, the bully, who behaves like a hawk but runs away when attacked, and
prober-retaliator, who continues to play like a hawk if its
opponent does not fight back. If its opponent fights back, it reverts to
playing dove. if the prober-retaliator is attacked, it retaliates like a retaliator.
If all five strategies are run in a simulation, only the retaliate emerges as
evolutionary stable. A population of doves is not stable because it can be invaded
by hawks and bullies. Hawk is not stable because the population of hawks will be
invaded by doves and bullies. Bully is not stable because it could be invaded
by hawks. With a population of retaliators, no other strategy would
invade it. However doves within a population of retaliators do equally well, and its numbers could increase, but prober- retaliators, hawks, and bullies could take
advantage of the doves. Prober-retaliator is nearly stable but
only the retaliator does better, just slightly. Therefore there will be a
gentle escalation between the two with a small dove minority.
This theoretical conclusion is not far from what happens in wild animals. Of
course, constructing a good model depends on the points allocated for winning,
being injured, expending time, etc. which requires a good understanding of the
behaviours of the animals instead of just an arbitrary assignment. For example, in
elephant seals, the prize of winning is near monopoly over the harem, and
therefore the probability of serious injury will also be high, compared to the
cost of wasting time, which is small. In cold climates for a small bird, the cost
of wasting time and waiting around is more fatal. Or the great tit, a passerine
bird, that must catch a prey regularly to ensure its nestling’s survival. The
conclusion here is that ESS will tend to evolve and ESS is not the optimum that
could be achieved by group cooperation. Maynard Smith has considered another
game called “The War of Attrition” where contestants do not engage in dangerous
combat, thereby avoiding injury, and just postures until one backs down. There is
obviously a cost in time. Therefore both contestants pay the price but only one
of them gets the reward. Contestants might decide to give up at the very
beginning and not waste any time. Those that are willing to go longer
would always win, especially if it was just a few seconds more than the
immediate retreater. Therefore, selection progressively favors expending more time
and so it approaches the maximum equal to the economic cost or worth of the
resources being contested. Individuals must not know how long opponents will
give up. Any physical feature or behaviour that betray giving up will be penalized
by natural selection. The poker face and sudden and unpredictable surrender would
evolve. Lying would not evolve in this circumstance because individuals who
call their bluff could invade this lying strategy.
The previous descriptions are examples of symmetric contests. Maynard Smith
considers asymmetric concepts. First these can be if individual vary in size and fighting ability and each individual is capable of gauging a rival’s
size in comparison to its own. Secondly, if the rewards gained from
winning are different across individuals. For example, older males with not long to
live may have less to lose than younger males. Lastly, asymmetry can arise if a
contestant we call ‘resident’ arrives earlier at the location of the contest
than the other, which we call ‘intruder’. If we suppose all individuals play the rule:
resident wins, intruder runs away, we should see half wins and half losses. A
mutant that plays hawk and always attacks will win when his opponent is
an intruder, and risk grave injury if his opponent is a resident. Those that tried
to reverse the rule: if resident, run away if intruder, attack will also do worse.
We’re assuming here that being a resident or intruder does not designate
any advantages. However, being a resident could have advantages in
knowledge of the terrain or the intruder is more likely exhausted from moving into
the battle area. Therefore resident wins, intruder runs away, is more probable as a
stable state in nature. The reverse strategy of intruder wins, resident
retreats would tend towards destruction – a ‘paradoxical strategy’ because contestants
would try to avoid being the resident and hence would ceaselessly move around,
expending time and energy. Therefore natural selection would favour
individuals who strove to be residents, which we know to be the behavior of
territorial defense. Dawkins points to the demonstration by the ecologist Niko
Tinbergen, who had an aquarium with two male sticklebacks, who build their nests
on opposite ends of the tank. When the two sticklebacks in separate test tubes
were placed close to each other, they tried to fight through the glass.
When he moved the test tube to male A’s nest, A assumed an attacking
posture and B attempted to retreat. This was reversed where the two test tubes
were placed in B’s territory. Both males played: if resident – attack, if intruder –
retreat. Hence territorial defense may simply be
an ESS which arose because of the asymmetry in time of arrival over a
location. Memory of the outcome of past fights may also influence the chance of
winning. For example RD Alexander found crickets that were beaten out by a model
cricket lost in fights against other real
crickets. If crickets fought in closed groups, a kind of dominance hierarchy
develops and they would sort themselves out in a ranked order, but the number of
serious fights decreases. Monkeys that are beaten up by other monkeys display
dovish behaviour towards the monkey that beat them. Similarly, hens that have never
met each other fight when introduced to each other, but the fighting dies down
after each individual learns her place relative to each other. Incidentally, egg
production is higher in groups of hens where there is less fighting. So this
helps us explain why we don’t see lions hunting other lions. A cannibal strategy
is unstable because there is too much danger of retaliation as seen in the
hawk strategy. So why don’t antelopes fight back at lions instead of running?
Prey animals usually run away instead of retaliating probably because there’s a
strong asymmetry in the contest. They have become proficient at running and
chasing, that a mutant antelope that stands and fight would not be very
successful. Dawkins warns against social group
thinking when analyzing animal behaviour because of the hidden and flawed group
selectionist assumptions. ESS can help to understand how collections of
independent selfish entities can come to resemble a single organized whole. In conclusion, Dawkins uses this to
answer the question of ‘What make a good gene?’ The gene pool will become an evolutionary
stable set of genes if it cannot be invaded by any new gene. So genes that
arise either by mutation, reassortment, immigration are quickly penalized by
natural selection. If a new gene succeeds in invading the gene pool, an
evolutionary stable set is established. Hence, progressive evolution is like a
series of discrete steps. (Animated and Narrated by J Tsao)