epistemologists seem to be very interested in practical reasons and practical
rationality. One good example of this is an interesting new paper entitled
‘Knowledge and Action’ by John Hawthorne and Jason Stanley (here) forthcoming
in the Journal of Philosophy.
Usually, when I read epistemologists writing about reasons I feel like entering
a strange new world where things look slightly peculiar.
paper, Hawthorne and Stanley argue that there are interesting relations between
knowledge (as opposed to mere beliefs), rationality and reasons. They claim
that the theories of rational action they are familiar with leave this connection
unexplained. This claim is difficult to assess. The only theory of rational
action they mention is that rational action is maximizing expected utility
(well, they do talk about a proposal by Fantl and McGrath about the connection
between knowledge and rationality). If that is the only theory they know of,
then their claim might well be true.
course, there is a huge, rich recent literature about rational action,
practical rationality, and reasons that others are familiar with – the work of
Williams, Smith, Wallace, Dancy, Scanlon, Parfit, Broome, Korsgaard, Velleman
and others. Hawthorne and Stanley bring up none of this literature and the views
about beliefs, reasons and rationality developed in this tradition. Couldn’t any of this work shed any light on how to account for the
connection between knowledge and rationality? Have all these people done really
so badly that their work is not even worth mentioning in the context?
Hawthorne and Stanley really believe this, then I would like to see some
arguments and investigation. Otherwise, the bold claim above just reveals
ignorance from their part. Sorry – this is just a personal methodological rant.
I do have though also more substantial questions upon (even if I probably
haven’t read the paper well enough yet).
and Stanley begin from what they think is the
view about rational action: Action is rational if it maximizes expected
utility. The latter is a function of subjective degrees of belief of the
consequences of the actions and their utility. On this view, knowledge about
the relevant consequences is not sufficient (one may know that p without high
enough degree of belief for being warranted to act on the belief that p) or
necessary (high degree of belief can rationalise actions without amounting to
knowledge) for rationality.
and Stanley resist this picture by referring to folk-appraisals of actions – by
what we would intuitively say to criticise people if they act without knowledge
of the relevant facts. So, they talk about Hannah and Sarah who go down a wrong
street because Hannah believes that the restaurant they are looking for is
there. Hawthorne and Stanley say that it would be natural for Sarah to say to
Hannah that ‘you shouldn’t have gone there since you didn’t know the restaurant
was there’. Likewise, we blame a doctor for negligence if she merely
justifiably believes that the needle is safe without knowing it. They really
give a lot of this sort of examples.
problem is that these appraisals do not seem to be appraisals of rationality per se. Instead, what is
assessed is what the agents ought to have done, what they had reason to do, or
whether they are to be blamed for negligence. Someone holding the maximizing
expected utility view about rationality could well accept all these appraisals
without giving up the idea that the described agents acted rationally. All this
would show is that oughts, reasons, and blame come apart from rationality
(which so many people think is true anyway). One can rationally do what one
ought not to do, what one has reason not to do, and what one is blameworthy
for. Furthermore, Hawthorne and Stanley do not give a view about rationality to
fix the problem but rather a view of reasons. Given that the connection between
reasons and rationality is controversial to say the least – we do not get much
further on the rationality side of things in the end. H & S actually
acknowledge this towards the end as they seem to fall back on the expected
utility view about rationality.
positive view which is supposed to fit the appraisals better is offered in few
principles. The first one of them is:
Action-Knowledge Principle: Treat the proposition that p as a reason for acting
only if you know that p.
principle is supposed to capture what is implicit in our assessments, i.e.,
that if one acts on a belief that does not amount to knowledge, one violates a
norm and is therefore criticisable. I would have liked to have heard much more
about which norm one is violating and for what precisely one is criticisable
case, the claim here is that knowing that p is necessary but not sufficient for
treating a proposition as a reason. It’s not sufficient because there is
irrelevant knowledge for actions. To get round this, they first define
p-dependency; choice between x1, …, xn is p dependent iff the most preferable
of x1, …, xn conditional on the proposition that p s not the same as the most
preferable of x1, …, xn conditional on the proposition that not-p. With the
help of this definition we get the principle which is gives both sufficient and
Reason-Principle: Where one’s choice is p-dependent, it is appropriate to treat
the proposition that p as a reason for acting iff you know that p.
If I get
this view right, there are propositions that are good reasons. They are of the ‘that
p’ sort. These propositions can become the reasons for which we act only if
they are known. This implies two things. First, if knowledge is a ‘factual’
term, then only true propositions can be reasons. And, second, one cannot act
for a good reason that p unless one knows that p (the proposition that p can be
a proposition that it is likely that q). Thus, being known ‘converts’ certain
propositions, i.e., reasons, into ‘personal’ reasons.
I have two
worries about this account. The first are the all too familiar Dancyan worries
about a category mistake – that the view just gets reasons into a wrong
metaphysical class. I don’t know what H&S think propositions are, but they
must be some sort of abstract, linguistic entities – sets of possible worlds,
functions from worlds to truth-values, structured thoughts, and so on. But,
these are not considerations that count in favour of actions like reasons do.
It’s the snake, its poison, and other such wordly things out there that give me
reasons to be scared and run away rather than the abstract objects that are in
some semantic relation to the snake. Why would they give me reason to run? So,
any view that requires treating propositions as reasons seems to require
treating as reasons things that are not reasons at all. And, that the
propositions are known doesn’t make these propositions that are not reasons any
more reasons. Of course, knowing that p implies that one should regard p, the
thing in the world, as a reason in the case that p really is a reason since
knowing that p guarantees that p is the case.
I do also
have a more substantial worry. In the place I live, the fire-alarm goes off every
now and then. In each case, I know that the fire-alarm goes off, I believe that
the house is on fire, and I run out. So far, it’s been either a false alarm or
a test. My robust intuition is still that I run out for a good reason. My worry
is that the H&S account either gives me no reason at all to run out or it
gives the wrong reason.
potential reason to run out would be the fact or the state of affairs that the
house is on fire. Dancy’s account gets into problems here. He wants to say that
this fact, even when it is a ‘non-existent’ fact as believed by me, is the good
reason for me to run out. But, most people would want to say that the good
reason I run out for is my justified belief
that the house is on fire. The H&S account does not allow me to have this
reason (or the believed proposition that the house is on fire as a reason). My
belief is after all false, so it cannot count as knowledge. Thus, the necessary
condition for my reason is not satisfied.
I know that the alarm has gone off. But, that the alarm has gone off seems like
a wrong reason to run out of the house. I could for instance set the alarm off
myself and yet I wouldn’t have any reason to run out. The alarm itself doesn’t
count in favour off anything. It is just supposed to indicate the real reason.
What other options H&S would have? Well, they could say that, given the
alarm, I know the proposition that
the house could be on fire. However, if this proposition was a good reason,
then I would have a reason to run out now because I do believe even now that
this house could be on fire. Maybe they could say that, as I believe that the
house is on fire in the alarm case, I also believe that the house is probably
on fire. That’s true, but does that belief count as knowledge? It could well be
false given that there’s never been a fire when the alarm has gone off. So,
it’s hard to tell what reason they would give for me to run out.
All in all,
I’m not convinced yet. And, I wished they would have engaged a bit with the great
literature on reasons and practical rationality we have and the more plausible views about reasons and rationality therein.