Biology Reference
In-Depth Information
from A today but refuse to repay the favour tomorrow. We will explore the conditions
under which reciprocity can be evolutionarily stable by using a simple model.
… and the
problem of
free riding
Repeated interactions in the Prisoner's dilemma
To consider how reciprocity could favour cooperation, it is useful to return to the Prisoner's
dilemma. Is there any way in which individuals can escape this dilemma and come to stable
cooperation? The answer is no if two players only meet once; defect is the only stable
strategy in Table 12.1. Defect is likewise the stable strategy if the total number of interactions
is precisely known in advance because defection will be optimal on the last encounter and,
therefore, also on the next-to-last and so on back to the first encounter. However, if the
series of encounters goes on indefinitely or, more realistically, if there is always a finite
probability,
w
, that the two players will meet again, then some form of cooperation may be
stable. This is because the long-term benefits of cooperation between players could
outweigh the short-term benefit of defecting. Although such behaviour is sometimes
referred to as 'reciprocal altruism', this is misleading, because although helping is costly in
the short term, it will only be favoured when this cost is outweighed by the benefit of being
helped in the future, and so it is mutually beneficial, not altruistic (Box 11.1).
Axelrod (1984) investigated this problem with a famous computer tournament in
which he competed 62 different strategies, submitted by scientists from all over the
world. These strategies involved mixtures of cooperation and defection in various
sequences, and Axelrod's simulations suggested that the best strategy was one called 'tit
for tat' (TFT): cooperate on the first move and thereafter do whatever your opponent did
on the previous move. TFT is a combination of nice (it starts by cooperating), retaliatory
(punishes defection) and forgiving (respond to cooperation of others, even if they had
defected previously).
Axelrod's computer experiment has been hugely influential, leading to the widespread
assumption that reciprocity via TFT is an evolutionarily stable strategy (ESS) that can
resist invasion by any other strategy. However, it has since been realized that the case for
TFT was overstated, that it can be beaten, and that it is hard for any one single strategy to
dominate. For example, other strategies which can perform well include 'suspicious TFT',
which defects in the first interaction and then plays TFT, or 'tat for tit', which starts by not
cooperating, and only switches to cooperation in response to the cooperation of others
(Boyd & Lorberbaum, 1987; Binmore, 1994, 1998). Consequently, while reciprocity is
theoretically possible, we shouldn't necessarily expect simple TFT rules in nature.
Repeated
interactions could
make reciprocal
cooperation
stable
Tit for tat was
the winner of
Axelrod's
computer
simulations …
… but it can be
beaten
Reciprocity in animals
Humans
Reciprocity seems common in the human world. One example comes from the Antwerp
diamond market, where a relatively small group of experts trade diamonds. An
individual will give another a bag of diamonds to take home to study. These bags are
often very valuable indeed, but no receipts are exchanged and no contracts are signed.
If anyone cheats, he is expelled from the community. So, here we have an example of
reciprocal cooperation being sustained in a repeated game by use of a strategy that is
termed GRIM, where an individual starts by cooperating, but then never cooperates
again after a single defection. There is also an extensive literature using 'economic
GRIM reciprocity
in humans
