Three ideas at the core of cultural anthropology that require explanation:
1. Individuals living in groups tend to behave in similar ways, and hold similar moral values, beliefs.
2. Such beliefs and values are transmitted within groups by various forms of social learning.
3. These differences create boundaries between groups that persist over time.
A complete theory should address these questions at 2 levels:
1. At a cultural evolutionary level, how are beliefs and values transmitted among individuals, and why does this process generate and maintain between-group differences?
2. At a genetic evolutionary level, under what conditions does natural selection favor psychological mechanisms posited to explain the cultural evolution of groups?
Previously, Boyd and Richerson (1985) have shown that:
(1) a particular form of social learning, termed conformist transmission, can maintain similarities within cultural groups and differences between cultural groups; and
(2) natural selection favors genes leading to conformist transmission in a simple model of evolution in a spatially varying environment.
This paper shows that selection favors psychological mechanisms that give rise to a conformist transmission under a wider range of conditions, including both spatially and temporally varying environments.
"Conformist transmission implies that individuals possess a propensity to preferentially adopt the cultural traits that are most frequent in the population. Under conformist transmission, the frequency of a trait among the individuals within the population provides information about the trait’s adaptiveness. This psychological bias makes individuals more likely to adopt the more common traits than they would under unbiased cultural transmission."(p. 219)
Unbiased transmission is insufficient to maintain cultural variation/boundaries if there is any cultural mixing.
Boyd & Richerson (1985) used a spatially-varying environment to show that conformist transmission could be adaptive. However, if the environment varies temporally, then old behaviors might suddenly be maladaptive, and conformist transmission could be deleterious. This paper explores this issue.
The Model
2 types of environmental cues:
Nonsocial cues predict the current environment, but sometimes lead to errors.
Social cues: observe the behavior of other individuals in the population.
A population of individuals is divided among n subpopulations.
The model is divided into 4 stages:
1. cultural transmission (including conformist transmission)
2. individual learning. The environment varies in how accurately individuals can ascertain the correct behavior, by parameter \rho, which is the probability that the individual correctly ascertains the state of the environment (\rho varies from 0.5 to 1). Individuals select a new behavior only if the probability that it is correct is sufficiently better than its alternative, meaning that individuals have acquired sufficiently ac- curate environmental information through trial-and-error learning, for example, to allow them to select the adaptive behavior with an acceptable degree of certainty. When environmental cues are not of sufficiently high quality, individuals imitate. A tradeoff: "selection can increase the accuracy of individual learning, but only by also increasing the probability that environmental cues will be indecisive and thereby causing individuals to increase their reliance on imitation." (p. 222)
Individuals have two traits: L: the reliance on social vs. individual learning, and \Delta: the strength of conformist bias (vs. unbiased transmission) in social learning. At the start of simulations, 99.99% of individuals have genes for pure individual learning and unbiased transmission (L = 0, \Delta = 0)
3. migration. A proportion m of each subpopulation moves to a new subpopulation. Immigrants with a heavy reliance on social learning (high L) and a strong conformist effect (\Delta ~ 1) can more rapidly learn the common behavior in their new environment through imitation. However, increasing the number of immigrants affects the extent to which the current subpopulation can be used as an accurate indicator of the recently adaptive behavior.
4. natural selection. Individuals with the nonadaptive behavior are less likely to survive.
Temporal variation
During each generation, a subpopulation may experience a sudden shift in the environmental state, say from State 1 to State 2, such that the previously adaptive behavior is no longer adaptive. The probability that the environment does not shift is given by \epsilon.
Initial conditions:
We set the initial conditions such that 99.99% of the individuals in each sub- population possessed the allele for all individual learning (vs. social learning) and no conformist effect. The remaining 0.01% of the population was distributed equally over the other 399 genotypes representing all possible combinations of both social learning reliance (L) and conformist transmission (\Delta). Similarly for the cultural traits, 80.0005% of the individuals in each subpopulation started with behavior 1 (and the environments for each subpopulation always began in environmental state 1), whereas 19.995% of the population started with behavior 2.
All simulations were run to 300,000 generations to assure that near-equilibrium was reached.
Results
Summary
Evolution of social learning: Reliance on social learning (L) depends strongly on the accuracy of environmental information and the frequency of environmental fluctuations. The evolution of high values of L requires low-accuracy environmental information and infrequent environmental shifts.
Evolution of conformist transmission: \Delta evolves to near maximum strength under a wide range of parameters, provided there is a small amount of migration. Also, the presence of conformist transmission effectively increases the accuracy of social learning in most situations and consequently facilitates a greater reliance on social learning.
Main results:
Selection Favors Conformist Transmission as Long as Environments do not Change too Rapidly (Fig 1)
AND
Greater Rates of Environmental Change Favor Less Reliance on Social Learning (Fig 1)
Selection Favors Strong Conformist Transmission Unless Individual Learning is Very Error Prone or Highly Accurate (Fig 3)
Conformist Transmission is Favored for a Wide Range of Migration Rates (Figs 4 & 5)
Varying the Number of Subpopulations Has Little Effect on the Model’s Behavior (very small effects of varying n between 2 and 5)
No comments:
Post a Comment