SF EV L20 reciprocal altruism

10/12/11
Senior Freshman Evolution
Lecture 20: Mutualism and
Reciprocal Altruism
Last lecture we suggested 5 reasons for
apparent altruism, and covered 3 of them:
1) Humans are weird
√ (but more later)
2) Reciprocal altruism
Co-operation where benefit
isn’t assured
3) Kin selection
√
√
4) Best of a bad job
5) Mutualism
Mutualism = co-operation to obtain
an end which one animal could not
achieve alone
Payback is immediate, so the benefits
are obvious, and so mutualism is not
always classed as altruism.
e.g. mobbing of predators by small birds.
They cannot attack alone, but together they can drive the
predator off.
Wildebeest travel in herds to dilute
the predator danger, and because the
ones in the middle benefit…. So
everyone’s trying to be in the middle.
Result – dense herd.
“Selfish Herd” concept.
Co-operation where
everyone always benefits
More co-operative mutualisms
are found in some herds.
e.g. musk oxen make in
impenetrable circle around the
young so wolves can’t attack.
Migrating geese fly in a V, saving
effort by slip-streaming the one in
front.
The front goose is swapped
regularly, so the hardest work is
shared by everyone.
Together they progress faster than
they would alone.
1
10/12/11
In winter mixed flocks of birds feed
together, for selfish herd reasons.
Colonial nesting is a mutualism
despite higher disease risk
Give alarm calls warning of predators,
even though not related to each other
(they are different species).
a) Easier defense from predators
b) Those in the middle safe as
predator bound to have eaten
someone else before getting
to you!
Costly as it draws predator’s attention to
caller but the confusion of lots of flying birds
makes up for the added danger of calling
Saddleback and bicolour tamarins also forage together,
saddlebacks high in the trees, bicolours lower down.
Saddlebacks call to warn of eagles,
bicolours warn of ground predators,
and both species benefit from better
protection from all predators.
Co-operative hunting e.g. lions
Can kill larger prey, but tradeoff as
must share it among group.
May actually get more food if kill
smaller gazelle alone.
Co-operative defence of prey swings
balance toward mutualism.
Male lions show co-operative breeding, as they form coalitions to take over a
pride together.
Can’t do it alone as females help defend
their current male, to avoid infanticide of
cubs by incoming males.
Alpha male takes most of the matings,
but worth it for the beta males as
a)  get some matings,
Synchronous breeding (all
hatching together)
e.g.turtles
Mutualism as gluts the
predators, so your
offspring survive better
than if they hatched alone.
Co-operative courtship between
unrelated males (very rare).
E.g. Long Tailed Manikin
The pair of males sing a duet to attract the
female, then dance by cart-wheeling round
each other.
If female is impressed, the alpha male gets to
mate, but beta male never does…. So why
does he dance for the alpha male?
Dance sites are limited, and a male who serves as a beta
male will inherit it when the alpha male dies.
So their mutualism is to get
matings, by owning a dance site.
b)  inclusive fitness benefits as all the
males are usually brothers.
2
10/12/11
Co-operative breeding
Heinz-Ulrich Reyer (1984) studied pied kingfishers to find out
why unrelated males helped.
Usually when young of the previous year stay behind
and help raise siblings. e.g. moorhen
He found primary helpers worked harder
than secondary helpers, but why did
secondary helpers help at all?
Really best of a bad job, as males who can’t get a
territory do this, increasing their inclusive fitness.
Reyer looked at costs and benefits across
2 years to find out.
But if other males help parents raise offspring
they aren’t related to, this explanation doesn’t
hold.
This happens in the Pied Kingfisher.
Young males without territories play one of
three strategies:
Primary helper helps his own parents
Secondary helper helps a pair to which he is
not related.
Don’t panic – it’s easy! ……
Delayer sits and waits for next year.
Fitness from breeding
tactic in the first year
Chance of surviving to next year, after
knackering year helping (or not)
Fitness
payoff from
breeding
tactic
Reciprocal Altruism – where the payback from your altruistic
action is not assured, but depends on the altruism of another.
e.g. If I buy you a beer in the pub, you may, or may not
buy me one next time. (Why would you?)
For reciprocal altruism to be possible, the animal needs
three conditions:
1)  Good chance of re-encounter, so relatively small
groups
Chance of getting a mate in the second year,
given your good standing (or not) by helping
Fitness from 2nd year
2) Individual recognition, so they know who they met last
time
So best breeding strategy is primary helper
3) Memory of the previous encounter, so they know
whether they owe a favour. i.e. quite intelligent.
2nd best is secondary helper - effort in helping small enough not to reduce
winter survival, and good chance of inheriting the female and the territory
you helped on.
Evolutionary game theorist John Maynard Smith applied models designed
for gaming to whether reciprocal altruism could exist in such animals.
Delayer by far the worst as very little chance of a territory next year either.
JMS suggested that altruism wasn’t an ESS (see last lecture), and
supported this view with his Prisoner’s Dilemma Game.
3
10/12/11
Played once, this is easy. You
always defect. You may score
6 and the worst you can get is
3. If you co-operate the best
you can get is 4.
Prisoner’s Dilemma
2 prisoners do a job and get caught. Prisoner 1 can:
Co-operate = admit some guilt but not rat on Prisoner 2
Defect = deny involvement and say it prisoner 2 did it.
Prisoner
1
Defect
Co-operate
4
2
Defect
6
3
So JMS concluded that altruism shouldn’t evolve; it wasn’t an ESS.
However, if you allow repeated encounters and memory of the previous
deal, i.e. if you allow reciprocal altruism, what happens?
Prisoner 2 has the same options
Payoffs: If both co-operate, both get a small sentence
If P1 defects and P2 co-operates, P2 gets full blame and P1 gets off lightly
If both defect (and blame each other) both get a slightly reduced sentence
Prisoner 2
Co-operate
Prisoner
1
Defect
Co-operate
4
2
Defect
6
3
Result:
Amazingly a single strategy won: Tit for Tat (or “an eye for an eye”)
= co-operate until the other guy defects, then defect on him, just once, then
return to co-operating.
This was a very unexpected result, and a surprisingly forgiving rule to win.
Axlerod asked for new strategies which could invade a population of tit for
tat.
Found that Tit for 2 Tats can = ignore the first offence and only retaliate
after the 2nd (= turn the other cheek?)
BUT Tit for 2 Tats cannot invade a mixed population, only an already
altruistic one.
Axlerod tried Tit for 3 Tats too, but this is too generous, and allows a subpopulation of pure defectors to persist.
So much for modelers – do real animals do it?
Robert Axlerod 1984 wrote a computer programme playing
prisoner’s dilemma repeatedly with many “animals”
meeting and “remembering” their last encounters with each
one.
“Animals” = strategies sent in by delegates of a conference
e.g. always defect, co-operate unless he defected last
time, co-operate 3 times then defect etc.
Payoff for prisoner 1 for each response of prisoner 2:
NB: JMS framed this, not in
years of sentence, but in
payoff, i.e. big scores are
better
Prisoner 2
Co-operate
After each round, killed off those with a low score and
“bred” from the others. Allowed to run until one strategy
won, i.e. to find the ESS (if it ever does).
Vampire bats show reciprocal altruism
If one fails to get food, another will
feed it some blood, as bats must eat
every day.
If the recipient fails to pay back the
favour when the donor needs it, then
the defector isn’t fed next time.
Amazingly, in the WW1 trenches, well before any of this was known, Tit for
Tat evolved.
After a few weeks the death rate in opposing trenches
fell. The soldiers on both sides were shooting to miss.
Without communicating, a system appeared such that if
the Allies killed a German, the Germans killed one of
the Allies, then everyone shot to miss again.
Generals cured the “problem” by moving troops around
more, making the payback less likely so reciprocal
altruism couldn’t evolve.
4
10/12/11
Cheats, policing, and part of why
humans are weird.
The defense against defectors (=cheats) in
any altruistic system is to punish them, but
giving out the punishment is usually costly.
So who chooses to be the policeman?
Fehr and Gachter 2002 let students earn money depending
on how well they did in a game.
Three worked together at any one time, paying stakes into a
community project. The more they paid, the better the project
did and the better the final payout to the whole group.
But you could cheat by putting less in than your fellows.
Group 1 could punish skinflints, but it cost them money to do
so. Group 2 couldn’t punish.
Results
84% of group 1 were prepared to pay €1 to punish the
freeloaders by €3.
10% did it every time they could.
There were strong emotional responses too – furious
outrage with the cheats.
In group 1, the more they could punish, the more they
paid into the communal project.
Group 2, as the rounds of the game went by, invested less and less in the
communal project, and co-operation broke down.
Conclusions: 1) Punishment maintains and feeds social co-operation
2) There is a strong impetus in humans to punish cheats, even at a cost to
themselves (= Altruistic punishment)
3) Outrage and shame may be important drivers of our social evolution.
“The threat of such punishment may have been crucial to the evolution of
human civilization” Fehr and Gachter 2002
Required reading, at least 2 of:
Freeman and Herron “Evolutionary analysis” 3rd Edition chapter 11
Skelton “Evolution: a biological and palaeontological approach” pp.
339-357
Barnard “Animal Behaviour” 2004 Pearson Press Chapter 9.3
Extra if you’re interested:
The rest of chapter 8 in Skelton
New Scientist 8th July 2000 pp 30 – 35 “All for One” – Group selection and
how it might in fact exist, with dissent from Dawkins.
New Scientist 10th May 2003 pp 32-37 “To trust is human” – Why humans
trust one another to a biologically crazy degree (and how having sex can
help the economy!).
Look up “altruistic policing” in google to find lots of fun studies.
5