The Explanatory Autonomy of the Biological Sciences

Abstract

This thesis aims to argue for the explanatory autonomy of the biological sciences. According to many philosophers and scientists, the biological sciences do not have their own explanatory autonomy because they either can be reduced to other “hard” sciences such as physics and chemistry, or cannot really explain phenomena since they do not have laws of nature. To maintain the explanatory autonomy of the biological sciences, I first argue against one influential form of reductionism, i.e., explanatory reductionism, by showing that explanation in the biological sciences can be achieved without reduction. Then, I demonstrate that the biological sciences do not have laws of nature. Instead, I suggest that it is usually biological models that do the explanatory work. Given the fact that biological models usually do the explanatory work, we still need a story of how biological models can be explanatory. However, to understand how a biological model can explain phenomena in the world, we first need to understand what the model-world relationship is. To this end, a holistic view of the model-world relationship is developed by looking closely at scientific practice. Based on this, a holistic account of model explanation is proposed. The basic idea behind the holistic account of model explanation is that, for a model to be explanatory, it must answer two kinds of questions: counterfactual dependence questions (corresponding to Woodward’s what-if-things-had-been-different questions, “w-questions” for short) that concern the model itself, and hypothetical questions that concern the relationship between the model and its target system. Furthermore, I suggest that the reason a biological model can answer these two kinds of questions is due to the fact that: (a) a model is a structure, that is, a set of dependence relationships that can be employed to answer w-questions, and that (b) the holistic fit between the model and its target warrants the hypothetical inference from the model to its target and thus helps to answer the second kind of question.