Overturning a paradigm is one thing. An 'a priori' assumption, however is not a paradigm and is not reached inductively. It is assumed 'a priori' and overturning an 'a priori' assumption is impossible. That is the problem. Any 'new stuff' is still interpreted in the context of the 'a priori' assumtion, which never changes.
Therefore, while the paradigm may change, the 'a priori' assumption of naturalism does not. Ultimately, a naturalistic model (any model as long as it remains naturalistic) is assured.
Understand?
So give me an example of a problem in science where the assumption of naturalism is not productive.