The evolutionary accessibility of cultural fitness landscapes

Many human cultural and technological traits have a complex and peculiar structure. They are often characterized by multiple closely interacting parts that can form modules, and change as a result of innovation, collaboration, and social learning. A full understanding of the ways in which such cultural traits evolve requires a model that accounts for the close epistatic interaction of the constituent parts, and for the uniquely cultural processes that act on them. Here, we generalize the concept of 'adaptive landscapes', where populations search a landscape for optimal combinations of components. This generalization accounts for epistatic interactions, and includes processes such as modularization, communication, and collaboration, that are not features of genetic systems but may be quite common features of cultural systems. We show that modularization in combination with collaborative communication of search results is more successful, but sometimes more costly, than searching a full adaptive landscape. However, in many of the circumstances under which technology evolves, a full search of many thousands of possible modules may not be feasible since time, financial, energetic, and environmental constraints are likely to be strong. Under these conditions, our models suggest that there is a critical value for the number of modules to search below which the global optimum is more likely to be reached by searching the full landscape. The value of this threshold depends on the complexity of the technology. Overall, we suggest that analysis of cultural fitness landscapes is likely to be a fruitful avenue for research into technological change especially when combined with elements of human learning and interaction known to be important features of cultural evolution.