In this paper, we first set out three requirements that each e-theory – a theory whose task is to explain data – must fulfill in order to be one such good theory: i) an ontological requirement, i.e. adequate simplicity, ii) a methodological requirement, i.e. plurality of research procedures, iii) an epistemological requirement, i.e. compatibility with the best available epistemical procedures. Moreover, we will claim that from the metaphilosophical point of view, unlike scientific naturalism on the one hand and supernaturalism on the other, liberal naturalism is the only philosophical approach capable of fulfilling all such requirements.