The problem of defining life: a case study using family resemblance

Abstract How to define ‘life’ is an unresolved question in the philosophy of biology, but has become more urgent as researchers around the world attempt to create synthetic cells in the laboratory, develop intelligent and autonomous robots, and search for signatures of life elsewhere in the galaxy. Here, we discuss the pros and cons of some of the current approaches to defining ‘life’, then propose an alternative approach based on family resemblance. Using a statistical modelling framework, we find that although living and non-living entities can be grouped according to overall similarity, it is difficult to find a single set of criteria that can both define known forms of life and be useful in identifying or characterizing novel forms of life. We hope that the family resemblance approach will prove to be a fruitful alternative to traditional approaches to defining ‘life’.