Abstract
In this paper I distinguish two broad construals of the method of conceptual analysis. I call the first construal "the philosophical conceptual analysis" in order to highlight the attitude that comes with it, i.e. one usually starts with some philosophical question that has a long tradition and one then proceeds to view science through this philosophical prism with its historical baggage.
It has the following characteristics: (i) analysis of meta-scientific concepts such as explanation, representation, etc. can be pursued independently of the content of sciences; (ii) formalization is an adequate tool in all contexts; (iii) analyzing concepts consists in searching for necessary and sufficient conditions. I criticize all three characteristics.
First of all, by ignoring the actual scientific practice one engages in a rational reconstruction of science that is filled with philosophical constructs rather than with actual insights into the workings of science. Second, formalization is indeed a powerful tool but it also has its limits.
Inspired by the power of formal logic, many philosophers have used it rather bluntly to the point of absurdity. Third, to understand a concept one needs not to specify the necessary and sufficient conditions.
I call the second construal of the conceptual analysis "the empirical conceptual analysis" to highlight the fact that it builds on two sorts of empirical data. (i) "secondary data" by which I mean using actual scientific findings as a starting point for philosophizing; e.g. cognitive psychology and historical case studies for tracking the development of scientific concepts. (ii) primary data, e.g. qualitative data in empirical philosophy of science. These methods provide us with a solid ground for constructing more general philosophical theories of science that are potentially of interest to scientists as well as to broader public.