Boyd & Richerson (1995) Culture increases human adaptability

Boyd, R., & Richerson, P. J. (1995). Why does culture increase human adaptability? Ethology and Sociobiology, 16, 125-143. 


Two strategies: Imitators and learners

Two extensions to Rogers’ model

1. Spatially varying environment, more than two behaviors, learning errors. 
• “Consider a model in which organisms live in an environment that consists of a large number of discrete islands, each with a different environment in which a different behavior is favored by natural selection. The populations on different islands are linked by migration of individuals from each island to all other islands. Thus, the rate of migration measures the predictability of the environment. If migration rates are high, individuals’ environments are unlikely to be similar to their parents’. Learners engage in costly learning trials that usually allow them to acquire the locally optimal behavior, but sometimes leads to errors.” 
• This model yields the same qualitative result as Rogers’ model: imitation evolves, but does not benefit the population in the long run. 
2. Imitators can detect learners.
• Learners always acquire the currently favored behavior but at some cost.
• After learners learn, each imitator surveys the behavior of n individuals in its social group, and query each potential model as to how it acquired its behavior. If there are any learners in the group, imitators copy the learner. Otherwise they chose a random individual to imitate. 
• Here as well, the qualitative result is the same. 

How culture can increase average fitness

Social learning would improve the average fitness of a population if it increased the fitness of learners as well as imitators. They suggest at least two ways that imitation can benefit learners.

1. Imitation allows selective learning. 
• Again, two environmental states each with an optimal behavior, and some probability of switching. 
• All individuals attempt to learn by sampling both behaviors. 
• Individuals adopt a particular behavior only if it appears sufficiently better than its alternative (to avoid errors). If they remain indecisive, they imitate a randomly chosen individual. 
• In the model, the difference between the observed payoff of the currently favored behavior and the alternative is an independent, normally distributed random variable with mean m  > 0 and SD = 1. Individuals use the following rule: 
• There is a threshold d that determines how selectively individuals learn. 
• Individuals imitate if the absolute observed difference in payoffs between behaviors is less than d. Otherwise they adopt the better observed behavior. 
• The value of d is a genetically heritable trait, with mutation. 
• Mathematical analysis shows that there is an ESS value of d, d*, at which the fitness of the population is greater than the fitness of a population with no imitation, as long as the probability of the environment switching each generation is less than 0.5. 
2. Imitation allows cumulative improvement
• Environment varies, each with a different optimal behavior.
• Organisms who cannot imitate start with random guess. They can then learn and improve their behavior. However, when they die, the improvements are lost, and their offspring must start anew. In contrast, an imitator can acquire its parents’ behavior after their behavior has been improved by learning. 
• Imitators have higher fitness than learners at evolutionary equilibrium if
• the environment does not change too often compared to the rate at which the population of imitators converges toward the optimum
• learners suffer substantially greater learning costs than do imitators.