There were lots of useful comments on the “pragmatics, belief and knowledge thread”:http://tar.weatherson.org/archives/004282.html and I wanted to follow up with a list of cases designed to spell out the consequences of the position and some choice-points in the development of the theory. (Although they may not look related, the cases below arose out of some considerations “Matt Weiner”:http://mattweiner.net/blog/archives/000508.html brought up in a response to my post.)
These cases are reverses of the train cases that Fantl and McGrath use. All of the characters below are trying to get from Providence to Boston. All of the characters want to get to Boston sooner rather than later, though this isn’t the most important thing in the world.
The background facts are that there are trains and buses that run from Providence to Boston. There are both buses and trains at 8:00 and at 8:30. The existence of the 8:30 trains and buses are common knowledge to all participants. It is also common knowledge that bus and train timetables change every few months, but the changes are usually fairly minor. What they know about 8:00 is less certain.
The cases are all set a few minutes before 8. The characters can only check whether the 8:00 train or bus is leaving by going to the train station or bus station respectively, and they don’t have time to go to both.
A caught the 8:00 train three months ago, and the 8:00 bus yesterday. She doesn’t have any more information about whether those trains and buses are still running. So she has reasonably good evidence that there is an 8:00 train, but considerably better evidence that there is an 8:00 bus. A is indifferent between catching trains and buses in general.
In this situation it would be a practical mistake for A to go to the train station, since she has better evidence that there is an 8:00 bus than that there is an 8:30 train. And my analysis of this (following broadly the Hawthorne-Stanley line) is that she shouldn’t go to the train station because she doesn’t _know_ that there is an 8:00 train, but she does know that there’s an 8:00 bus.
B has just about the same background as A. She last caught the 8:00 train three months ago. She saw yesterday when she was passing the bus station that there was an 8:00 bus. She didn’t catch it because she always gets severely motion sick on buses. She would rather catch the 8:30 train rather than the 8:00 bus – arriving 30 minutes earlier doesn’t make up for the sickness. She is making the right practical decision in going to the train station rather than the bus station. And I think that she even knows that there is an 8:00 train. The possibility that the train timetable is too remote given her interests.
So for A the existence of the 8:00 bus is a practical defeater of any possible knowledge she might have about the 8:00 train. It isn’t a defeater for B because given her disposition to be ill on buses, the bus isn’t a practical option. This might seem like an odd kind of defeater, but if you take the practical interests considerations seriously there will I think have to be cases like this. And I don’t think that this pair of judgments is too outlandish.
But once we have the idea of defeaters in mind, we can see some options for how to develop the theory. Some cases help spell this out.
C has the same evidence as A in the sense that she caught the same trains and buses. But C refuses to use induction to gain evidence about how buses work. No matter how often she catches the 8:00 bus, she doesn’t come to believe that there regularly is an 8:00 bus. And she never plans her actions around this. So she, like B, goes to the train station rather than the bus station.
This seems to me like a practical mistake, and seems to indicate a lack of knowledge. This is despite the fact that going to the train station maximises expected utility according to her credences. And (this is what makes the case tricky for my view) she has a perfectly rational belief about trains. So it isn’t the case that S knows that p just in case S has a high enough degree of belief in p to take it for granted in practical deliberation, and her degree of belief in p is justified and properly grounded in facts about p. Rather, we’ll have to say that she _doesn’t_ know that p if there is some other proposition q such that given her evidence she _should_ believe q to a degree such that if she believed q to that degree, she should not take p for granted in practical deliberation.
D is like B in that she caught the 8:00 train three months ago, but has never caught the bus. She has in fact systematically ignored chances to gain evidence about buses in the time she’s been in Providence. So while going to the train station is the right thing to do given her credences, and even given the rational credences she should have given her evidence, we might think the buses are still a defeater. We’ll come back to D.
E is also like B in that she caught the 8:00 train three months ago, but has never caught the bus. In her case this is because of an irrational and unjustifiable dislike of buses. She doesn’t like buses because sometimes on buses you end up sitting next to a _philosopher_. Given her utilities, it is utility-maximising to go to the train station not the bus station. But it seems like this would be a strange way to come to have knowledge. It’s one thing to say the defeater goes away because of B’s illness, but another to say it goes away because of E’s prejudices.
C, D and E all offer us choice points in the theory. As we’ll see the choices are not quite independent, so there are 6 choices available. We’ll set out the choices by writing out the position with the options written in.
bq. S knows that p iff her degree of belief in p is justified and properly grounded in the facts, and she can take p for granted in practical deliberation given her (actual/ideal) degrees of belief given her (actual/ideal) evidence and (actual/ideal) values.
By ‘ideal’ degrees of belief I mean something like the Keynesian probabilities given the evidence.
By ‘ideal’ evidence I mean something like the evidence she would have were she to be paying proper attention to evidence easily available in her enivronment.
I mean to stay as neutral as possible on ‘ideal’ values – whether it means something like what values her ideal self would advise her to have, or whether it means something like Moorean/Platonic perfect values.
There is little point in taking ideal evidence but actual degrees of belief, since what the evidence is only matters if we are considering ideal degrees of belief. So there are 6 possible choices here.
I’m inclined to take the option that chooses ‘ideal’ in each of these cases, but I’m not wedded to that view. This would in effect give us a lot of defeaters, and say that none of C, D and E know that there is an 8:00 train, like A and unlike B. (I would say that they all believe that there is an 8:00 train, which makes them like B and unlike A.)
I also want to make a terminological suggestion, one that isn’t too tightly tied to my way of doing things. If we use actual values, I’d call it an interest-based theory, if we use ideal values I’d call it a value-based theory. This might generate an interesting (i.e. worth investigating) difference between the Fantl-McGrath-Hawthorne-Stanley theories, which sound like interest-based theories, and the feminist theories (such as Elizabeth Anderson’s) that sound like value-based theories. Though since FMHS don’t discuss (to the best of my knowledge) cases where the subjects have unjustified or improper values, I am speculating a little here. (Note that this distinction is not the distinction between actual interest and perceived interest, one on which Jason Stanley does take a position, and correctly says that it’s actual interest that matter. But that’s consistent with the interests being things like prejudice that we might not think could generate knowledge.)