Hilary Putnam’s position on Quine’s naturalism is very strong. He calls it "sheer epistemological eliminationism" (322). He believes that Quine is arguing that
"we should just abandon the notions of justification, good reason, warranted assertion, etc., and reconstrue the notion of “evidence” (so that the “evidence” becomes the sensory stimulations that cause us to have the scientific beliefs we have)".
Presumably this is because Putnam believes that justification, good reason, warranted assertion, etc. are normative notions and, as we're often told, Quine wants to ‘abandon’ the normative. Whether or not Quine really wants to abandon all these notions or not, the very possibility of abandoning them seems incredible (at least in epistemology or science). What would we have left? Well, for Putnam’s Quine, we would have ‘evidence’ – or ‘evidence’ which is reconstrued to be "the sensory stimulations that cause us to have the scientific beliefs we have".
We only have evidence in the form of beliefs, not in the form of sensory stimulations, at least according to Donald Davidson. And if we are dealing with the beliefs of other subjects, we are dealing with the normative – primarily with attributions or assumptions of rationality on behalf of these subjects. So whereas Putnam is saying that Quine wants to abandon the normative in his project of naturalised epistemology, Davidson and Jaegwon Kim are saying that he couldn't do this even if he wanted to. Of course if sensory stimulations were so free of the normative, or even if they do actually exist as Quine sees them, then talk of sensory stimulations providing ‘input’ and the assertions they cause as ‘output’ this would be a purely descriptive - not a normative - epistemology – which is what Putnam claims Quine (really) wants.
Putnam goes further than this. His basic argument is that if Quine abandons the normative, it follows that he must also abandon or give up on truth (or ‘truth’ as Putnam puts it). Rather than saying, as many philosophers do, that notions of justification, rational acceptability, warranted assertibility, right assertibility, etc. are alternatives to truth, we can argue that they all assume or/and rely on the notion of truth. For example, the justification or warranted assertibility of something is relative to the truth of something. That is my first take on Putnam’s position. However, Putnam himself explains the Quinian implicit rejection of truth in this way. He argues that
"the notions the naturalistic metaphysician uses to explain truth and reference, for example the notion of causality (explanation) and the notion of the appropriate type of causal chain depend on notions which presuppose the notion of reasonableness". (323)
Causality and reference don’t seem like normative notions or indeed normative things. However, to put it simply, there are indefinite amounts of causal process out there in the world. More particularly, there are also an indefinite amount of causal processes which could, or may, account for our theories of truth and reference. That’s where the normative angle comes in. The causal theorist of truth or reference has to decide or choose what the ‘appropriate type of causal chain’ is. The world - or its causal processes - won't choose these things for him. Thus he requires the normative notion of reasonableness. That is, what type of causal process would it be reasonable to select in our accounts of truth and reference? What are the important and relevant causal processes to this act of reference or this account of truth? Here too the predicates ‘important’ and ‘relevant’ are normative in nature. To put this in Davidson’ way. Causation ‘is not itself explanatory’. Causation is neither evidence for - nor an explanation of - anything. In Davidson’s terms, beliefs fulfil these roles. And as I've already said, beliefs can't help but have a normative component.
These arguments against Quine’s naturalism can be applied to a more specific approach – or alternative – to the normative notion of justification.
Take Alvin Goldman’s reliability theory. As Putnam puts Goldman’s position:
"… instead of saying that a belief is justified if it is arrived at by a reliable method, none might say that the notion of justification should be replaced by the notion of a verdict’s being the product of a reliable method." (322)
Here again we can say that normative factors will be involved in our assessments of what actually is an example of a reliable process or method. Indeed the very choice of reliability as an epistemic device or even a notion will have required normative choices and constraints. More specifically, the ‘replacement’ of justification with reliability must involve, or must have involved, normative decisions and normative judgments as to the value of reliability - or reliable processes - in the epistemological project.
For example, why reliable methods or processes rather than the methods or processes of clairvoyance? Because clairvoyance isn't reliable process or method? Wouldn’t that be a circular justification (whether virtuous or vicious can be decided elsewhere)? Even after reliabilism has been chosen as a mode of epistemology, normative decisions and judgements will still be required and used by the reliabilist.