embodying a
strong respect for both reciprocity and reputation—even though
no
concern for such
emergent phenomena need be built into the utility functions as-
signed to the
players. Experimental findings that reciprocity and reputation
matter
in some social
interactions therefore come as no surprise to orthodox game
theorists.
The biologist
Robert Trivers [30] says that strategies which support full
coopera-
tion in repeated
games exhibit reciprocal altruism. He cares about this
phenomenon
because Nash
equilibria do not only describe the outcome of rational play; under
appropriate
conditions, they also describe the end-product of evolutionary
processes.
It is therefore
of some importance that the only one of Axelrod’s conclusions that
seems to be
genuinely robust is his claim that we should normally expect the
kind
of evolutionary
computer simulations that he pioneered to lead to efficient (fully
cooperative)
Nash equilibria (Binmore
[4, p.313]).
A social norm
can be seen as a device for solving the equilibrium selection
problem that the
folk theorem says is built into a society’s indefinitely repeated
“game of life”.
We then obtain a putative explanation for the cultural evolution of
different social
norms in different societies. The folk theorem therefore provides a
theoretical
backdrop for the ideas on cultural evolution pioneered by Boyd
(author
number two) and
Richerson [10, 11].
The important
point is that only the Nash equilibria of a society’s game of life
can be
evolutionarily stable, but nothing says that evolution must select the
same
equilibrium in
different societies. (See Samuelson [26] and Young [31].) Just as
the
French use the
Nash equilibrium for the Driving Game in which everyone drives on
the right and
the English use the Nash equilibrium in which everyone drives on
the
left, so we must
expect different social norms that select different equilibria to
have
evolved in
different societies.
The social norms
that interest me most are those that we normally describe in
terms of
fairness or justice (Binmore [3, 4]). My recent
Natural Justice [5] offers
an algebra-free
version of my theory of fairness that draws on both psycholological
and
anthropological thinking. In passing, I try to explain that the reason
neo-
conservative
economists see no role for fairness in their models of the world is
that
their absurdly
over-simplified models only have one equilibrium, and so there is
no equilibrium
selection problem for fairness to solve. I do not like the policies
advocated by
such economists any more than the authors of the paper we are
discussing, but
the answer is not to throw the baby out with the bathwater by
seeking to
discredit the basic methodology of economics using whatever
rhetoric
seems currently
persuasive, but to direct the attention of neo-conservatives toward
economic models
that are genuinely descriptive.
Game theorists
think the folk theorem is particularly relevant to the social
norms (or social
contracts) of small-scale societies, because—unlike our own large
societies—the
no-secrets proviso of the folk theorem has a good chance of being
reasonably
descriptive. It is no problem that kinship is likely to be an
important
explanatory
variable in such societies, because we can simply write this fact into
a
player’s utility
function using an appropriate version of Hamilton’s rule before ap-
pealing to the
folk theorem. (Remember that the idea that our theories necessarily
6