In this passage, Ginet explains why he thinks deontologism supports a kind of access internalism:
Assuming that S has the concept of justification for being confident that p, S ought always to posses or lack confidence that p according to whether or not he has such justification. This is simply what is meant by having or lacking justification. But if this is what S ought to do in any possible circumstance, then it is what S can do in any possible circumstance. That is, assuming that he has the relevant concepts, S can always tells whether or not he has justification for being confident that p. But this would not be so unless the difference between having such justification and not having it were always directly recognizable to S. And that would not be so if any fact contributing to a set that minimally constitutes S’s having such justification were not either directly recognizable to S or entailed by something directly recognizable to S.
Alston offers us this rendition of Ginet’s argument:
1. S ought to withhold belief that p if S lacks justification for p.
2. What S ought to do S can do.
3. Therefore, S can withhold belief wherever S lacks justification.
4. S has this capacity only if S can tell, with respect to any proposed belief, whether or not S has justification for it.
5. S can always tell this only if justification is always directly recognizable.
6. (Therefore) Justification is always directly recognizable.
He takes issue with (5) and invites us to consider an analogous claim having to do with the justification of action:
Consider the ethical analogy that is inevitably suggested by Ginet’s argument. There is an exactly parallel argument for the thesis that the justification of actions is always directly recognizable. But that is clearly false … Would I be morally justified in resigning my professorship as late as April 12 in order to accept a position elsewhere for the following fall? This depends, inter alia, on how much inconvenience this would cause my present department … There is no guarantee that all these matters are available to me just on simple reflection. Why should we suppose, without being given reasons to do so, that the justification of beliefs is different in this respect?
While I happen to agree with Alston that the justificatory status of an action does not depend upon just what the agent can directly recognize (i.e., determine to be true upon reflection if she has the relevant concepts), I am not sure the example has quite the force Alston takes it to have. If Alston knew upon reflection that the permissibility of resigning depended upon facts that he could not know and could easily obtain and he decided to resign without first checking these facts, it seems plausible that he would act without justification. If, however, he did not know upon reflection that he needed to check these facts before making a decision, it seems relatively plausible that he could justifiably make a decision without first considering these facts. If it turns out that his resignation was a major inconvenience, this might show that the action was regrettable, but it is not clear that it shows that the agent acted without justification.
I think the real problem with the argument comes a step earlier. What justification is offered for (4)? Suppose we do sometimes have the power to withhold judgment as to whether p is true. Does our ability to withhold depend upon whether we have sufficient evidence or justification? It seems not. If not, then even if justifying reasons were invisible, unknowable, inaccessible, etc…, their presence or absence would have no bearing on what we are able to do or whether we are able to withhold. So, naturally we could withhold when those reasons happened to be absent, but this lends no support to the thought that we could identify (reflectively or otherwise) whether the reasons to withhold are present or not.
Ginet either conflates “ought” implies “can” with “ought” implies “can tell that you ought” or he takes OIT to be a consequence of OIC:
OIC: In circumstances under which S ought/ought not Φ, S can Φ and can refrain from Φ-ing.
OIT: In circumstances under which S ought/ought not Φ, S can tell whether or not she ought to Φ.
While OIC might be true, OIT is most likely false. It seems there are three serious objections to OIT. First, in cases of conflicting reasons where there is a case both for Φ-ing and a case against Φ-ing, you might know of the relevant reasons without knowing which reasons are stronger. It seems you should not against the stronger reason.
Second, there is a structural problem with OIT. Suppose you have an obligation to remove the left kidney. If so, OIT tells us that you can tell that you have an obligation to remove the left kidney. If your obligation is, inter alia, the obligation to remove the left kidney only if you can tell that this is your obligation, OIT tells us that you are under this obligation only if you can tell that you can tell that your obligation is to remove this left kidney. So, your obligation to remove the left kidney depends upon whether you can know that you know that you know, etc… Assuming, as is plausible, that you can have obligations to Φ where you are not in a position to know that you are in a position to know that you are in a position to know that you ought to Φ, we have to reject OIT.
Third, suppose both OIT and OIC are true. Because water is H2O, you cannot bring someone a glass of water without bringing them something containing oxygen and hydrogen. If you ought to bring them some water, you ought to bring them something containing oxygen and hydrogen. So, according to OIT, you can know that you ought to bring them something containing oxygen and hydrogen. So, if anyone in medieval times ought to have brought someone water or refrained from bringing them water, they were in a position to know that they would bring them something containing hydrogen and oxygen. So, should we conclude that they had no obligations to bring or refrain from bringing someone water because of their ignorance of chemistry?