Transmission, induction and evolution

Abstract

Many human behaviours are thought to depend upon cognitive capacities enriched with innate domain-specific knowledge. Underpinning this view is the hypothesis that evolution can shape cognition to include strong innate inductive biases. In this thesis, I re-examine that hypothesis with respect to a broad class of behaviours: those that we learn from other individuals. Taking human language as a test case, I present an analysis of the co-evolutionary process that underpins the formation of innate constraints on cognition for behaviours that are culturally transmitted through inductive inference. I derive a series of mathematical models of this process, built around Bayesian models of cognition and cultural transmission, and ask how they can inform our expectations about cognition in a cultural species. I argue that the traditional marriage of nativism and evolutionary reasoning is undermined by this process, as is the suggestion that cognitive adaptation to cultural behaviours is outright implausible. I explore the co-evolutionary dynamics induced by cultural transmission, and conclude that they can radically manipulate the evolution of cognition: culture can intervene in the formation of hard-wired knowledge, but nevertheless facilitate rapid cognitive adaptation. The analyses I report make strong, testable predictions about the nature of inductive biases for cultural behaviours, and offer solutions to a number of long-standing conundrums in the evolution of language.