An Adaptive Model for Predicting !Kung Reproductive Performance: A Stochastic Dynamic Programming Approach

A stochastic dynamic programming model is presented that supports and extends work on the reproductive performance of the !Kung Bushmen (Lee 1972), (Blurton Jones et al. 1978), (Blurton Jones 1986) proposing that !Kung women and their reproductive systems may be maximizing reproductive success. The stochastic dynamic programming approach allows the construction of a \whole-life" model where the physical/environmental constraints along with the uncertainty about future events !Kung women face when making reproductive choices can be explicitly built in. The model makes quantitative predictions for the optimal reproductive strategy assuming !Kung women are maximizing expected lifetime reproduction (ELR) given the physical parameters of !Kung life. The model relies on data gathered from the works cited above and some considerations from simple probability theory. The model predictions for optimal birth spacing match the !Kung reproductive data very well and support earlier ndings (Blurton Jones and Sibly 1978), (Blurton Jones 1986). The utility of the dynamic modeling approach is illustrated when the e ects of varying certain model parameters are investigated. By including the e ect of the mother's mortality which was not included in the Blurton Jones and Sibly (1978) analysis, the model allows for further exploration of the application of an adaptive approach to human reproductive performance. By adding some considerations about the risks of childbirth for the mother the model not only predicts optimal birth spacing which is site speci c but also predicts the optimal time for a woman to begin and cease having children. These predictions coincide with menarche and menopause and shed light on their possible adaptive value.