Computational epistemology provided a brand new normative research scheme for the logic of discovery when it was restricted within the traditional descriptive research paradigm. In this normative perspective, a detailed justification for the logic of discovery and many normative discussions concerning the solvability of discovery problems have been put forward. This approach is very enlightening for the further study of the logic of discovery, though the findings of the specific normative research is still quite inadequate for the scientific discovery in practice.