In a manner similar to my treatment of both Searle and Deacon’s books, I now plan to consider, over the next few posts, Brian Skyrms’ books “Evolution of the Social Contract” and “The Stag Hunt and the Evolution of Social Structure.” These accounts will serve as a prelude to a consideration of Rawls’ “Theory of Social Justice”, Wright’s “Non-Zero” and perhaps a constructivist account of meta-ethics which parallels Rawls’ in its ontology but matches W. D. Ross’s in terms of content.
According to Rawls, the just society is one which is characterized by two features: 1) it is the social structure which would be chosen by an agent placed behind a veil of ignorance and is motivated by a rational self-interest; 2) any inequalities are justified by being beneficial for all involved. Skyrms uses the well-known example of a person cutting a cake into two pieces without knowing which piece he will get for himself. The ‘just’ thing to do, all would agree, is to cut the cake in half.
But why? Why wouldn’t a fully rational person cut the cake into 1/3 and 2/3 hoping that the larger half will end up on his plate? Indeed, how will it be decided who gets what piece? Harsanyi argues that the (hypothetical) decision is made by the flip of a coin, but again, this is no reason to suggest that the 50/50 scenario is any better than the 66/33 version.
“Now, if all I care about is expected amount of cake – if I am neither risk averse nor a risk seeker – I will judge every combination of portions of cake between A and B that uses up all the cake to be optimal: 99% for A and 1% for B is just as good as 50%-50%, as far as I am concerned.” (6)
Skyrms approaches this problem from the angle of evolutionary game theory. Suppose a population is composed of agents who adopt various strategies (asking for 33, 50 or 66 out of 100). When interacting with one another, an agent will get what he asks for unless the sum of what is asked for by both the agent and agent with which he is interacting surpasses 100, at which point both parties get nothing. The number which an agent may or may not “get” corresponds to the number of agents which will adopt the same strategy in the next generation.
Notice, however, that the actual agent in this scenario means nothing. What are important are the strategies which are interacting within the population.
“The identity of the individuals playing is unimportant and is continually shifting. This is the Darwinian veil of ignorance… For this reason, the Darwinian story can be transposed into the context of cultural evolution, in which imitation and learning may play an important role in the dynamics.” (10-11)
In such a scenario, 50-50 is an evolutionarily stable strategy. In a population of 60’s nobody gets anything in interacting with one another. Indeed, even if mutations are allowed, nobody would get anything until a ‘less-than-40’ came along. Then again, in a population of 40’s everybody gets what they ask for every time, but it’s only 40. 50-50 is the only “attracting dynamical equilibrium in the evolutionary replicator dynamics” when considering pure strategies.
When considering a population of various strategies interacting with one another, however, things get a little dirty. Consider a population made up of half 33’s and half 66’s. The 33’s will get what they ask for every generation and the 66’s will get what they ask for half of the time, thus averaging 33 per generation. Of course 33 per generation is not 50, but any variation from these two numbers will be selected against, for 67’s will over shoot every time and 65’s will not do as well as 66’s. While it is certainly more likely that a population which initially starts with random concentrations of 33’s, 50’s and 66’s will end up being taken over by 50’s there is still a significant possibility that it might end up with the 33/66 scenario.
There are two ways which might contribute to the 50/50 scenario: 1) The greater number of possible strategies which a population has or can potentially have the more likely the population is to evolve toward 50/50 or something very near it; 2) The introduction of a positive correlation into the population wherein agents are more likely to interact with others who have similar strategies strongly pulls the population toward 50/50 as well. Of course, with regards to the second path toward social justice one wonders what a negative correlation would do to the population wherein 33’s are encouraged to interact with 66’s rather than 33’s.
If the population continues to experience significant degrees of variation then the population will not remain in a 50/50 or a 33/66 equilibrium forever.
“If there is enough random variation in the evolutionary process, a population caught in a polymorphic pitfall (33/66 equilibrium) will eventually bounce out of it and proceed to the fair division equilibrium. It will also eventually bounce out of the fair division equilibrium as well, but the amount of time spent at fair division will be large relative to the amount of time spent in polymorphic traps, because of the larger basin of attraction of the fair division equilibria.” (21)
Skyrms means this account of the evolution of justice as a supplement to the Rawlsian veil of ignorance, a way of accounting for the near universal preference for 50/50 over 33/66 scenarios. While such an account is interesting, and does suggest that the 50/50 strategy does seem to have an evolutionary advantage over others, one wonders how easily such a strategy would spread in a structured society with hierarchal relationships. Then again, this is only the first of several chapters which we will consider.
