Abstract

If evolutionary history were to be replayed from the beginning, what would be the same, and what would be likely different? Would there be a human-like species, multicellularity, or even DNA? There is a great variety in the answers biologists give to this question, despite having the same access to empirical data and biological theory. For instance, Stephen J. Gould has claimed that evolutionary history is radically contingent, while Conway Morris holds that it converges onto specific biological structures that are favored by natural selection. Others such have proposed that evolutionary history is characterized by an inexorable increase in complexity, while others see it as an evolutionary arms race. In this dissertation I investigate the fundaments underlying claims biologists make about contingency and directionality in evolutionary history as a whole. The topics of convergence and contingency have received attention from philosophers of biology in recent years, but the foundations of interpretations of evolutionary history as a whole remains a relatively neglected field. Hence the primary objective of this dissertation was not to defend this or that account, but rather to show a method by which these fundamental issues can be identified and analyzed constructively. The dissertation is organized into two parts, each dedicated to a single problem. The first part concerns the problem of ‘description dependence’: claims about the contingency of evolutionary outcomes depend on how these outcomes and the evolutionary process itself are described. I set out to map the different ways in which the contingency of outcomes changes as the phenomena are described in more and in less detail, and as broader or narrower subsets of evolutionary history are taken into account. That such an analysis is useful to pursue, I attempt to show by applying it to two of the most prominent interpretations of evolutionary history, those of Gould and Conway Morris. According to how their claims about evolutionary history are analyzed, one can arrive at opposing conclusions about the contingency of evolutionary outcomes. The second part concerns the problem of ‘causal complexity’: evolutionary history is a complex mess of unrelated causal processes, and for every generalization there is an exception. This raises the question whether non-speculative generalizations over evolutionary history as a whole are even possible. Limiting the scope of the investigation to the mechanism of natural selection alone, I consider first whether and how natural selection may be expected to give rise to trends at all. Some philosophers reject that natural selection is a cause at all, and that all evolution is simply an accumulation of births and deaths. If any trend occurs at all, it is due to a confluence of unrelated causal processes that could easily not have occurred. I argue against this by showing that these philosophers overlook the issue of time-scale: causal processes may make a difference for reproductive outcome at a time-scale of a single generation without them making a difference at the time-scale of multiple generations. With this distinction in mind, one can argue that natural selection causes a population to tend towards equilibrium. If a yet longer time-scale is taken – not that of multiple generations in a single environment, but that of many species and genuses across many environments – the challenge of causal complexity becomes much more difficult to overcome. Drawing on the phenomenon of phenotypic plasticity and niche construction, I argue that selection for plasticity ‘feeds’ on this complexity and variability in the environment. The trend in plasticity is unique in this regard, since trends in other types of adaptation are interrupted by the variability of complex environments.