Political Theory and the Sciences: a Surprising Parallel Between Definitions, Principles and Laws

Debates about the so-called ‘demarcation problem’ – how to differentiate science and non-science – are mostly unhelpful. One largely unnoticed problem with these debates is the very assumption that something either is or is not a science. This paper argues, by contrast, that political theory, while not being a full science, has important scientific *elements*. Recognising this has significant implications for the methodology of political theory. To put it another way, political theory can (in some respects) be scientific without being a science – and grasping this helps us do political theory better. That view has, of course, already been defended in certain ways by scholars commending the use of social science *methods* within political theory (e.g. Miller 1999, Swift 1999, Floyd 2017, Dowding 2020, Perez 2023). But to the extent that we distinguish methods from *methodology* (the latter defined as the logic of inference – how we reach, test and justify our conclusions), then political theorists can indeed use aspects of scientific methodology even when we not using scientific methods (e.g. Blau 2016; Wilson 2016). This argument rests on a surprising parallel between scientific generalisation, on the one hand, and conceptual definition and normative principle on the other. A conceptual definition is like a scientific generalisation. E.g. ‘corruption is the misuse of public office for private gain’ is in key respects equivalent to ‘water boils at 100 degrees Celsius’. Certain variables are included in the definition of corruption (misuse, public, office, private, gain), and certain variables are implicitly excluded (e.g. sex, height, skin colour). Likewise, a normative principle such as ‘do what you like provided you do not harm others’ also includes certain variables and excludes others. The same applies to generalisations: ‘water boils at 100 degrees Celsius’ says nothing about atmospheric pressure or the purity of water. Indeed, testing water’s boiling point at different atmospheric pressures and with impure water led scientists to modify the initial generalisation. Methodologically, this parallel, and the last point about modifying scientific laws/generalisations by further testing, opens the door to an important methodological insight: we can use thought experiments to probe conceptual definitions and normative principles, just as we can use actual experiments to probe scientific laws/generalisations. It is, indeed, *sensible* to test a definition by throwing actual or hypothetical examples at it. And only a very brave policy-maker would not try to test a normative principle with actual or hypothetical examples. Such testing will often help us refine the definition or principle (think of chapters 4 and 5 of ‘On Liberty’, where Mill ends up adding extra variables to the harm principle). So, recognising the ontological parallel between definitions, principles and empirical generalisations leads to important epistemological and methodological contributions.