We returned to the idea of "immediate justification." I mentioned at the end of week 1 that the notion of "basing," or whatever we call the relation that holds between prospective justification and well-founded beliefs, should also be applicable when the prospective justification is immediate. So we shouldn't conflate "basing" a belief on some justification with inferring that belief from some other beliefs. (The latter may provide one example of the former.)
We distinguished between your justification for p playing some role in constituting or providing some justification for q, and its merely being a necessary accompaniment of your having that justification for q. Miriam asked for further motivation for or clarification of the former notion. I came up with the following after class, which might be helpful: consider some claim f that it's impossible for you to have justification for, and some claim p that you do in fact have justification for. Then one might argue that you have justification for p iff you have it for p ∨ f and for f ∨ p. (I will in later weeks raise some troubles for Closure Principles in full generality, but we might identify a suitable background story relative to which you have some of this justification iff you have all of it.) Nonetheless I take it we have a strong intuition that your justification for p has some explanatory precedence over your justification for the other propositions. This is the kind of idea I'm trying to invoke.
I mentioned the difference between Closure Principles, which as we'll see later can be formulated in different ways, and Transmission Principles. A Closure Principle for prospective justification would usually be framed as a thesis about the presence of some justification necessitating the presence of other justification (perhaps when certain other conditions hold, for example, your knowing that the first claim entails the second). A Transmission Principle, on the other hand, essentially concerns whether you'd be in a position to justifiably infer some conclusion, or base some belief, on other justification you possess. I mentioned that on some ways of formulating Closure, for example as a thesis about doxastic justification, it becomes harder to see if these principles will remain distinct. Still, the underlying contrast illustrated here, between the modal dependencies of our epistemic properties, and their epistemic dependencies, is one we need to pay attention to.
Next we took up the notion of a defeater. I explained a way that this term has sometimes been used (for example in Gettier literature), as some fact that you need not be aware of but whose obtaining has a negative impact on your epistemic status. I propose to call such facts "disablers" rather than "defeaters". For us, a "defeater" of some justification to believe p will be some information such that when you learn it (or at least become more confident in it), the original justification no longer supports p as much. One way this can happen is that the defeating information gives you some opposing justification, for not p. These are called opposing (or rebutting) defeaters. Another way it can happen is that the defeating information, without directly speaking for or against p, instead gives you some reason to distrust your original justification. These are called undermining (or undercutting) defeaters.
We discussed difficulties in giving a sharp, substantive, and theoretically neutral definition of this contrast. Instead we'll rely on paradigm examples and our rough instincts. I think of defeating and undermining as a matter of degree, so I count some justification as "undermined" if it's subject to any amount of (undefeated) undermining; this may still leave you with some of the original justification remaining. Also, the defeating justification may be acquired before the evidence that it defeats. Also, undermining and opposing justification may come mixed together, and so too can undermining and supporting justification. (You get undermining of some of your justfification to believe p, while simultaneously acquiring some other justification to believe p.) Since undermining is a matter of degree, there can be such a thing as bolstering justification, which has the opposite polarity of underminers: it makes some other justification better, perhaps by repairing or insuring it against some other actual or possible underminers.
I mentioned that in some ways redundancy of evidence (as in reading the same story in two newspapers published by the same press outlet, perhaps even two copies of the same newspaper) is structurally similar to undermining justification. I don't know how to define the difference between these, but am not much bothered by that. (I agree it sound strange to talk of the first newspaper as "undermining" the justification you get from the second.) Another issue nearby (that we didn't discuss in class) is that it's hard to define the difference between getting undermining justification and merely narrowing the reference class of your original justification, in ways that affects what that original justification supports. For instance, you learn that N is a bird. That's justification to believe N can fly. Then you learn N is a bird that lives in Antarctica. That's justification to believe N can't fly. Then you learn N is a bird that lives inside the warm British compound in Antarctica. Now you're again justified in believing N can fly. I don't regard these as examples of undermining. But it's hard to say what the difference is. This bothers me more than the issues with redundancy.
I find it natural to think that higher-order information --- evidence that you're not justified in believing p, or that some particular evidence doesn't justify you in believing that p --- has some effect on your first-order epistemic status wrt p. (Plausibly, positive higher-order information also has an effect.) If we agree it does have a negative effect, I find it plausible to think of this as (to some degree) undermining your first-order justification for p. But I don't know how to argue for this, and I'm aware that others think of these effects as involving different mechanisms, and still others think the higher-order information just doesn't affect your first-order justification wrt p. Sorting out these debates is one of the central aims of our seminar.
I introduced a contrast between liberal and conservative stances about some potential underminer u of some prima facie justification you have to believe p. You can be liberal about some choices for u, p, and what the justification for p is, and conservative about others. Also, you can take stances that are neither liberal nor conservative. The liberal stance says that --- so far as these factors go --- nothing about the truth or falsity of u, nor about independent justification you have to believe not-u, has to be part of your prima facie justification to believe p. (We agree u is a potential underminer, so if you do get some justification to believe u is true, that will to some degree undermine your prima facie justification to believe p.) The conservative on the other hand says that if your justification to believe p is really vulnerable to being undermined by u, then for that justification to really exist in the first place it has to "include" some reason to believe not-u. (The notion of "including" here can be spelled out in different ways, but this is meant to be more than just a modal entailment.)
I've called an underminer "quotidian" when it was a u of the sort the conservative is thinking of. (It'd be nicer if I had used a word that's more cognate with "conservative.") I introduced the term "credulism" for the view that some underminers aren't quotidian. That is, some justification doesn't include any reason to believe not-u but is nonetheless vulnerable to being undermined by evidence for u. A plausible example: my justification for some mathematical calculation might be wholly mathematical, but is vulnerable to being undermined by evidence that I've taken a bad-at-math drug. A more controversial example: my perceptual justification for believing I have hands doesn't include any reason to believe I'm not a brain in a vat, or that my senses aren't malfunctioning, but is vulnerable to being undermined by evidence that I am in those bad circumstances.
"Dogmatists" hold that there can be underminable justification that is nonetheless immediate. This is one form of credulism. The way I use the term "dogmatism", it doesn't commit you to any claims about internalism, or "phenomenal conservatism", or Moorean arguments, though I myself have sympathy for some such claims. Note that some justification's being immediate is compatible with it having some "enabling" conditions, which may be a scarce resouce. The immediacy just concerns the fact that other justification doesn't play certain roles. "Credulism" is a more general position than dogmatism: you might think that even if your mathematical justification isn't immediate, nonetheless it still doesn't include any justification about what drugs you've taken. Yet you think it can be undermined by news about the drugs.
At some point in our discussion, the good points were made that: (a) sometimes thinking of new possibilities can have a negative impact on how much your original evidence justifies belief in p, and also that (b) sometimes entertaining certain doubts, even if not in response to evidence, can have a negative impact. (I don't mean to assume that if there are negative impacts in both such cases, it involves the same mechanism.) I am inclined to think there are effects of these sorts, which we will be trying to explore. I am not inclined to regard these as cases of undermining. However, we shouldn't prejudge that question in the way we initialize our usage of the term "undermining." Let's just say that what we're doing is fixing some agreed core of that notion, and it remains open to further inquiry whether other epistemic effects are sufficiently similar that they should also be counted as cases of undermining.
In trying to get a good handle on the epistemic effects of higher-order evidence, we have to also wrestle with the question what are the epistemic effects of mere higher-order beliefs. That raises the question what are the epistemic effects of mere beliefs more generally. For example, do unjustified beliefs justify us in believing their obvious inferential consequences. I described one extreme view which said these beliefs have no epistemic effect at all, only what justification you have matters. An opposed extreme view says that justification is just a matter of the downward dynamics from the beliefs you start with, so mere beliefs aren't in themselves anyhow handicapped as justifiers. I invited your judgments that both of these extreme views are unsatisfactory. We will spend more time on this in the coming week.
At the end of class, we introduced the difference between "narrow scope" and "wide scope" intepretations of normative principles, like the principle that you ought to believe the obvious consequences of what you believe, or the principle that you ought to take the known necessary means to what you intend. This correlates with whether one is willing to validate the inferential pattern of "factual detachment" or the pattern of "deontic detachment". The narrow scopers interpret the claims as: If q, then Op, and are happy with the analogue of factual detachment (given the further premise q, it does follow that Op). The wide scopers interpret the claims as: O(If q then p), and reject factual detachment (given q, it needn't follow that Op) but tend to be more sympathetic to deontic detachment (given Oq, perhaps it does follow that Op). We will explore all this more in the coming week.