I have just read that Quine relies on his confirmation theory to establish if a scientific theory is “valid” or not. But I am not sure to have understood what is Quine’s confirmation theory and why Quine adopts it.
Quine thinks that humans as a group can "confirm" one another's findings. His view resembles, in some sense, the old idea of summa ratio. That the truth builds up over the generations. On the other hand, it is based on his common sense type attitude to science and not an apodictic attitude of certainty with respect to either intellect or the senses. Whereas he moves between logic and experience without favoring the one or the other. And, likewise, with his "theory" he does not go too far towards the empiricists nor towards the rationalists. That is, towards the sheer empiricism of the principle of induction, or the wild certainty of an a priori logical certainty.
A useful account - no more than that - is to be found in Paul T. Sagal, 'Paradox, Confirmation and Inquiry'. I'm going to draw on three extracts from this article to throw light on Quine's theory of confirmation.
Confirmation - the easy view
A theory is confirmed if the positive evidence for it is sufficient. One theory is more highly confirmed than another if the quality and/or quantity of its positive instances is superior to that of the other. This all goes smoothly. Of course we still need to explain notions like quality of positive instance itself, but we expect to have little difficulty. (Paul T. Sagal, 'Paradox, Confirmation and Inquiry', Philosophy, Vol. 51, No. 198 (Oct., 1976), pp. 467-470: 467.)
In fact we have difficulties aplenty from Hempel's ravens paradox to Goodman's paradox of 'grue' emeralds. Such paradoxes are the background to Quine's account of confirmation.
Quine and confirmation - natural kinds
[Quine] ... attempts to explicate positive evidence in terms of natural kinds, similarity, and innate quality space. The idea is roughly that positive evidence involves entities belonging to some natural kind. The members of a natural kind share among themselves the relevant similarity relation. What is recognized as natural kind is a function of the innate quality space of the organism. The hard data for the last statement depend in turn upon the response patterns to stimulation of such organisms. These patterns tell a tale; and the term innate quality space is the title Quine gives to the tale. Green emeralds form a natural kind. An individual green emerald is positive evidence for the 'theoretical' claim All emeralds are green. Grue emeralds do not form a natural kind. Evolution makes it reasonable to identify the natural kinds we identify with the natural kinds in rerum natura. In short, Quine suggests that confirmation theory must go beyond logic and class (set) theory. It requires a theory of kinds. (Kinds are neither sets nor (intentional) properties.) (Sagal: 468-9.)
(Goodman's grue = X is grue if and only if X is examined before time t1 and is green OR X is not examined before t1 and is blue.)
Natural kinds a halfway house
Quine concedes that the notions of similarity and (natural) kind form something of a halfway house, or perhaps, a detour. In the more developed fields of knowledge, e.g. chemistry, physics, what counts as a kind depends not upon our aboriginal space, but rather upon chemical and physical theory. We are then left in such cases with the problem of explaining theoretical knowledge. Reflections on induction give rise then to reflections on similarity and kind which in turn lead us to the general question of scientific knowledge. But this is where we began, isn't it? For this is the overall problematic situation or context outside of which the confirmation paradoxes do not amount to very much. Kind is introduced to clarify the paradox situation. Kind is tied originally to our primitive dispositions to respond to and distinguish among various stimuli, and to a world in harmony with these dispositions - such harmony being explicable in terms of Darwinian natural selection. As we as individuals and as a race develop, this mechanism retreats in significance. As far as theoretical science is concerned, it is almost completely in the background. What then becomes of similarity and kind? It is at this level that Quine brings Goodman's notion of entrenchment into the picture. An entrenched predicate is roughly speaking the linguistic counterpart of a natural kind. Quine puts the matter as follows: 'A theoretical kind need not be a modification of an intuitive one. It may issue full-blown, without antecedents; for instance the kind which comprises positively charged particles. We revise our standards of similarity or of natural kinds on the strength, as Goodman remarks, of second-order inductions. New groupings, hypothetically adopted at the suggestion of a growing theory, prove favourable to inductions and so become "entrenched". We newly establish the projectibility of some predicate, to our satisfaction, by successfully trying to project it. In induction nothing succeeds like success.'
Entrenchment at best explains why we hypothesize what we hypothesize. In Peircean terms, it is part of a theory of abduction. Our theories must be responsive to problems. We must make use of the most effective instruments in our arsenal. Our stocks of hypotheses and theories are such instruments. They provide the instrumental context of inquiry. Like the very canons of logic, they can be brought to bear in every problematic situation. The entrenched predicates provide raw material out of which our tools are fashioned. We would not know how to wield a strange tool- one made out of unusual (e.g. grue) raw material. (Sagal: 469-70.)
Nelson Goodman, Fact, Fiction, and Forecast, 4th ed., Cambridge, Mass.: Harvard University Press.
W.V.O. Quine, 'Natural Kinds', Ontological Relativity and Other Essays, New York: Columbia University Press, 1969.
Paul T. Sagal, 'Paradox, Confirmation and Inquiry', Philosophy, Vol. 51, No. 198 (Oct., 1976), pp. 467-470.