A yet further difficulty comes from the idea of defining theoretical terms in an observational
language. This implies that if the observations associated with a term change as a
result of investigation, then the meaning of the term changes. But this then makes it
difficult to explain how different theories can disagree––a difficulty that is if anything
increased by adopting Quine’s claim about the relation of theory to experience; and this in
turn makes it hard to explain how preference of one theory over another can be rationally
justified. Paul Feyerabend, indeed, drew from the notion of incommensurability the conclusion
that such preference cannot be justified, and denounced the imperialism of
scientists who claimed that their theories were in any way superior to the views of believers
in witchcraft