Theoretical anthropology tries to develop models of human biology and culture. In this thesis, we investigate how different computational models from theoretical biology can be applied to evolutionary anthropology. We study two different types of models, applying them to two different sub-fields of evolutionary anthropology, and highlighting alternative choices in their construction. On the one hand, we observe that the evolutionary simulations are composed of three main components: an updating rule, a game and a population structure. We find that the updating rule can alter the qualitative and quantitative evolutionary outcome of a model. A dominant language is more resilient to learning errors and more frequent when selection primarily weeds out maladapted individuals, instead of promoting well-adapted ones. We study the evolution of cooperation and institutional punishment. Group selection can support cooperation, even when implemented through the selection of individual agents migrating between communities at different rates. Institutional punishment on the other hand is highly complex and cannot arise from simpler strategies in either wellmixed or community-structured populations. On the other hand, Bayesian inference models used for linguistic phylogenies can incorporate highly correlated typological information, without a priori knowledge about the underlying linguistic universals. While close in subject, models in theoretical biology and profound anthropological expertise express all but disjoint theories in terms of scope and complexity. This thesis acknowledges this challenge and contributes to bridging the gap.