Sociobiology of aidgiving

1
Sociobiology of aid-giving

• MSc ACSB module 2006/07
• SB Session 4

2
The tragedy of the commons

• Common grazing land is usually over-exploited and
  unproductive
• It would benefit everyone using it if commoners
  kept a smaller stock of animals on the common
• But if any individual showed restraint singly, he
  would suffer the all the cost, while others would
  reap most of the benefits
• So everyone continues to over-exploit the land

3
Wynne-Edwards group-selection

• VC Wynne Edwards had the idea that natural animal
  societies evolved by group selection to avoid
  over-exploiting their food resources
• Most biologists disagreed, e.g. G. Williams, J.
  Maynard Smith, and D. Lack (in Population studies
  of birds)
• Langur infanticide shows that Nice guys finish
  last in the competition to pass genes down to
  the following generation

4
Kin-selection

• Beldings ground squirrel gives warning when
  predators (e.g. badgers) spotted
• Risk to signaller it stands conspicuously rather
  than escaping down the burrow sought out and
  killed by predator because of its alarm call
• Benefit to kin warnings much more common if
  close relatives stand to benefit
• May reflect early learning of sibs rather than
  identification by gene-based cues but would this
  matter? Hare Murie (1996) Behavioral Ecology,
  7, 76-81
• Payoff is indirect, via survival of genes in its
  close relatives Inclusive Fitness (WD Hamilton)

5
Kin-selection (2)

• In Black-tailed ground squirrel, females
  alarm-call throughout life, males only while
  immature
• Males stay in natal group until they mature, then
  leave to breed elsewhere move on to new groups
  each year, and so no relatives will be near them
• Females stay in natal group for whole life so
  always relatives close by to benefit from alarm
  call

6
Kin selection in lions

• Lion prides males (half) brothers (r0.22),
  females resident but not close relatives (r0.15)
• Males fight to death with intruders, but dont
  fight over access to females.
• Chance that a particular mating yields a
  surviving cub is 0.00125, so not worth getting
  killed over
• Females suckle one anothers cubs
• Male more generous than female when cubs try to
  get access to kill (cub-male r.31, -female r.18)

7
Lion prides (2)

• Cooperative hunting hunting group size not
  tailored to prey size and likely success, though
  larger groups allow lions to take larger prey
• Grouping is a response to the danger of being
  killed by other lions even solitary lions must
  jopin forces for safety. Not purely a vehicle
  for gains via inclusive fitness
• Defensive response to taped roars some lions in
  vanguard, some lag is this a PD game?

8
Vampire bats Reciprocal altruism

• Vampire bats roost in groups in which neighbours
  roost together frequently (but not all relatives)
• Significant chance that each bat will be
  unsuccessful in feeding on a particular night
  and if go without for several nights they will
  starve

9
Vampires (2)

• Bats that fail to find food beg from neighbours
  that have been successful
• Donor bats regurgitate some of their blood meal
  to the bat that is begging enough to carry it
  over to the next nights hunting
• Regular neighbours know they will meet again (so
  the favour can be reciprocated) and that X
  reliably repays debts if X was ungenerous, they
  would not offer it food when needed

10
Baboon mating

• In multi-male troop, high-ranking male may escort
  a receptive female
• Other male which wishes to mate cant do this and
  fight off her escort simultaneously
• Solicits help from a friend latter attacks and
  draws off the escort, while its pal mates
• Later, favour will be reciprocated. Males that
  are most ready to help when asked find it easiest
  to recruit aid when they want it

11
Modelling Reciprocal Altruism

• Initially, this kind of reciprocal generosity was
  modelled using the PD game
• More sophisticated recent modelling of
  co-operation Nowak (2006), Science, 314,
  1560-1563
• For single encounters in PD, all-D is
  theoretically the best strategy
• For repeated-encounter PD (IPD) where cant
  predict how many repetitions
• TFT is a successful strategy cooperative
  punitive and forgiving

12
Modelling IPD games

• TFT can be invaded by all-C, then others not ESS
• Nowak Sigmund (1993) Generous TFT, Pavlov, and
  near-Pavlov strategies
• Pavlov(s) copes with noise (occasional errors)
  win-stay, lose-shift strategies
• Previous outcome R S T P
• Prob. (co-operates) 1/1 0/0 0/1 1/0
  Pavlov/ TFT
• Prob. (co-operates) .999 .001 .001 .999
  Near-Pavlov
• Stay Shift
  Stay Shift
• C-C C-D
  D-D D-C

13
Error and discounting in IPD

• Stephens et al., 1995, Journal of theoretical
  biology, 176, 457-469
• Introduce error. Focus on stability of strategy
  X against All-D in the PD space
• TFT zones where stable, sometimes stable, never
  stable
• Pavlov stable in restricted part of space
• Need alternative models to PD for real animals

14
Do PD payoffs reflect co-operative encounters in
life

• Clements and Stevens Jays in skinner boxes
  arranged so that keys offered Co-operate and
  Defect choices on each trial
• Set payoffs to IPD Mutualism IPD
• When Mutualism payoff matrix, showed C
• When PD payoffs, always ended playing all-D
• Suggests that for real animals, PD payoffs may
  not be sufficient to yield cooperative behaviour

15
Cooperative stickleback, guppies

• Predator inspection risky for lead fish and
  safer for those that lag behind
• Present a predator (behind glass) and fix a
  mirror to provide a companion that either swims
  alongside or lags (swims at an angle)
• Fish swim closer when companion is supportive
  than when it defects

16
Fish tit-for-tat

• Dugatkin long debate about whether such
  experiments are relevant to PD and TFT (LA
  Dugatkin, Cooperation among animals, pp. 59-70)
• Showed that payoffs do probably fit PD
• Showed that changes 1st to 2nd half of trial fit
  predictions for TFT in IPD bold/ cautious fish
  are consistent in co-op trials, bold fish
  inconsistent when image defects

17
Human co-operation

• In PD-game contexts, humans play all-D
• Fehr Fishbacher (2003) review other games in
  which strong reciprocity can arise
• Factors such as reputation are important in
  determining whether fair-dealing is stable
• Social norms exist which prescribe fair
  allocation and punishment of transgressors

18
Punishment game

• Ultimatum game in which A decides how much of his
  100 to allocate to B (who has zero initially)
  referee C (50) informed and can keep his own 50
  or spend x to punish A (A loses 3x for
  referees spend of x)
• If A shares 50 with B, no punishment
• If A shares nothing, C spends 14, and A is made
  to suffer 42 loss as punishment

19
Norms and society

• Do social expectations of fairness arise through
  evolutionary pressures? What of reputation?
• Is an altruistic, fair dealing dominant safer to
  be with than a selfish despot?
• NH primates may benefit from using alliances to
  limit the power of dominant group members
• Dunbar says human groups gt150 dont work - social
  disapproval in group cant keep control of
  selfish/ antisocial individual behaviour

20
Human cheating (1)

• Cosmides Humans have an evolved cheat-detection
  module (cf. Wason selection task)
• Yamagishi et al. (2003) SS shown pictures of
  PD-game participants, separating high-D and
  high-C players (intermediates omitted)
• SS later remembered D faces better than C when
  asked to discriminate previously-viewed and novel
  faces

21
Human cheating (2)

• Best remembered targets in post-PD survey
• Didnt trust partner thought partner felt same
• Had less desire to play same partner again
• Saw mutual cooperation as less important
• Saw joint gains in PD as less important
• Considered defectors smarter than co-operators
• Saw real life as NOT involving offered help
• Less willing to help if no pressure/witnesses

22
Human cheating (3)

• Difference in false alarm rates for D and C faces
  that SS had never seen before
• SS mistakenly think they recognise unfamiliar Ds
  (esp. female) more often than for Cs
• So, alternative slant on Cosmides Toobys cheat
  detection module are there differences between
  typically-D and typically-C faces that we are
  tuned to pick up and use in our social
  interactions?

23
References

• Yamagishi et al. (2003) Evol. Hum. Behav., 24,
  290-301
• Krebs Davies (1993) An introduction to
  behavioural ecology (3rd Edn) Chapter 11
• Dugatkin (1997) Cooperation among animals. Esp.
  Ch. 2, section 3.9
• Fehr Fischbacher (2003) Nature, 425, 785-791