December 02, 2005

Norms on Acts and Actors

Jon Matheson just was asking me about Jennifer Lackey's Norms of Assertion, which contains some criticisms of my Must We Know What We Say? The paper raises a lot of interesting points--for reasons that I might explain later,* I think that we're actually fairly close to agreement about some of the points she's criticizing me on.

The real substantive disagreement here, I think, comes down to the question of primary and secondary propriety. Keith DeRose has argued (and I agree) that, if there is a norm of assertion, the asserter can be subject to or exempt from criticism according to whether she had reason to believe that the assertion conformed to the norm. If the agent has reason to believe her asssertion confirms to the norm, it's secondarily proper even if it doesn't conform to the main norm of assertion.

Lackey argues that this doesn't make sense; you can't have a primary/secondary propriety distinction for actions. If an action violates the norm, it just violates the norm, and you can be criticized for it. She gives the example of a basketball player, Mabel, who has lost a contact lens. Mabel takes a foul shot from in front of the foul line, but she reasonably believes that she is behind it. Still, her shot is in no way proper--it is merely that Mabel has a good excuse for taking an improper foul shot. Similarly, if she takes the shot from behind the line while thinking she's in front of it, there's nothing wrong with the shot. So the primary norm, that foul shots should be from behind the foul line, is the only norm on the shots.

I've just been talking about this with Allan Hazlett, and I think that the way to make this distinction out is to distinguish norms applying to the acts and norms applying to the actors. Primary propriety is a norm on the act; secondary propriety is a norm on the actor, as to whether she did a reasonable job of trying to live up to the norm on the act.

Below the jump a couple of sports examples, which seemed amusing at the time. In fact those examples are this post's reason for existence. I'll perhaps try to post something a bit more worked out about primary and secondary propriety here or at CD. You may also want to read this thread from Certain Doubts (I especially like Keith's comment about how we have intuitions reflecting both kinds of propriety) and my earlier response.

The first example I came up with was the norm of kicking a field goal in American Football. Not the rules for field goals (how many players must be on the line of scrimmage, whatever) but what makes a field goal Good or No Good--which is, clearly, whether the ball goes through the goalposts.

But whether a field goal is good qua field goal does not necessarily correlate with whether the field goal kicker has done anything right or wrong. In the Bears-Niners game** the kicker lined up for a field goal that was kicked perfectly straight, and seemed as though it would travel through the uprights, but that was seized by a gust of wind and blown sideways out of bounds. The kicker is exempt from criticism--he did everything he could. But the field goal was still no good.

To echo what Keith said in his CD comment, I think intuitions about whether assertions are improper sometimes rely on primary propriety of the assertion and sometimes on secondary propriety of the asserter. So we may say, "Well, this assertion is false, even though the asserter can be excused." Or, "The asserter is blameworthy even though she made a true assertion." If we feel that an assertion is basically OK, that's not necessarily a sign that it is in fact in conformity with the primary norm of assertion. Our feeling may be that the asserter is exempt from criticism, even though the assertion itself is improper; just as we wouldn't criticize the kicker even though the field goal isn't a good field goal.

Here Allan objected that there can be things that meet the criteria of proper assertions even though they don't involve any agency at all. Suppose that I receive an electric shock which causes my throat to contract involuntarily, producing sounds indistinguishable from some true utterance. Would I be forced to say that that is a proper assertion, with only secondary impropriety? We realized pretty quickly that in this case I can say that it's not an assertion at all, because of the lack of agency. So the norm simply doesn't apply.

The football analogy is this: Suppose that the wind was so strong that it blew the ball from the holder's grip and through the uprights, without the kicker's foot touching it. It would not be a field goal at all, good or no good. Since there is no action by the kicker, there is nothing to apply the norm to. [Agency as such isn't necessary in football--if the kicker is blown into the ball it is a field goal I think. But that's just an analogy.]

The punchline of the shaggy-dog story is our attempt to figure out what this would be. We decided it would be a fumble through the end zone, and touchback. I assure you that it did seem very funny at the time.

*I reserve the right to explain it only in the privacy of my own home, not on the blog.

**Speaking of Chicago, Sufjan Stevens' Illinois is amazing. I put it on last night and have been listening to it obsessively. Hear a bit or two.

Posted by Matt Weiner at December 2, 2005 11:42 PM

Something (typepad?) isn't letting me comment on the proper post, but Apostropher has a great example of things that are little-known because of their complete lack of truth.

Posted by: washerdreyer at December 3, 2005 03:23 PM

You wanted this post I assume--it's closed for anti-spam reasons. (Wish I didn't have to do that, but I'm not willing to actively monitor comments on old posts, so I just close them.)

Apostropher's link did remind me of my old post--unfortunately Hindrocket also used "little-known fact," which was what got me pwned the first time around.

Posted by: Matt Weiner at December 3, 2005 03:43 PM


I'm quite sympathetic to the idea that there are two types of evaluations--those that apply to agents and those that apply to the products of agency. It seems you create unnecessary headaches when you suggest there are really two types of norms or standards of correctness because this suggests a picture Dancy gives a good mocking to according to which one set of norms clamor for the agent's attention (Do that thing!) and another clamor for the act's attention (Hey, get yourself done!).

The two-norm view really gets you into trouble when you get situations like this: you are rationally compelled to believe that the impermissible is mandatory. According to level 1 norms, X-ing must be avoided. According to level 2 norms, X-ing must be done. So, what is to be done? None of the obvious theoretical options look all that attractive. You can try to 'shrink' the norms and internalize them to try to banish all such conflicts; you can conceive of the fight as the fight between pro tanto reasons; you can hope that such situations will cause aneurisms so severe that an agent's head will explode. They're all bad.

Instead of two norms, it's best (I think) to think of evaluations that apply in the first instance of agent and those that apply in the first instance to acts. Only the second sort of judgment appeals to norm satisfaction and permissibility; the correctness of the first sort of judgment is never explained in terms of permissibility or the satisfaction of norms.

The picture that I think is right (and which I suspect is yours and strongly suspect cannot be sensibly denied) is this one:

In evaluating acts, the question is just 'Did the act satisfy the relevant norms?' In evaluating agents, given that there is a rational explanation available for the act, can the agent be credited or faulted?' The point of a justification is to show that the act did in fact conform to the demands of overall reasons. The point of an excuse is to show how there could be a rational explanation of the act and yet the agent cannot be credited or faulted for acting as she did (this is why the easiest excuses involve showing that the agent was failed). Allan's case is strictly speaking not one in which there is a justified act or an excusable failure (depending upon whether the 'assertion' expressed a truth or falsehood) but rather one in which there is an exemption available--an explanation for why the appearance that there is a rational explanation for the act is missing.

Posted by: Clayton at December 3, 2005 10:24 PM

I think this issue here discussed is why Williamson gave the Snow Case at
Williamson 2000, 257. 

In the Snow Case, a film crew has put fake snow around my house in the winter
and I thus falsely but reasonably believe it's snowed.  I can thus
reasonably believe I know and so the assertion is reasonable. 
Reasonableness is a category of evaluation that applies to both beliefs and
acts, including speech acts (I don't say it is univocal for beliefs and acts,
neither does Williamson).  In the Snow Case the belief and the assertion
are reasonable but the assertion is speech-act-wise improper according to the

Contra Clayton, I don't see any real harm in talking about norms in the way
he reprobates, the problems he raises don't seem compelling to me, but I'd have
to see the view worked out to say more.  

Posted by: Trent Dougherty at December 5, 2005 07:32 AM

I've been thinking about Lackey's examples some more and I think they fail for another reason. I think that the sports anaologies aren't very good, despite the fact I enjoy you bringing up the field goals in the Bears game. The problem is that in sports we really don't care about assessing the agents apart from the success of their actions (save little girls soccer). To be a good football player is just to have good stats, not to try really hard all the time and often fail. In the language game, we really do care about the agents and how they are acting, I think perhaps even more than we even care about the success of their actions. This is an important difference that prevents Lackey's examples from being counterexamples.

Posted by: Jon Matheson at December 5, 2005 03:49 PM

Thanks for the comments!

Clayton, I'm not sure to what extent this is a verbal dispute about the use of the word 'norm'. On the kind of level 1/level 2 distinction we have here, the level 2 norm is always something along the lines of "do your best to make sure your act satisfies the level 1 norm." So there's one way in which it'll never clash with the level 1 norm. (Other norms on actors may tell you to strive for an act that in fact clashes with the level 1 norm, but that'll be just like an ordinary clash of norms. I guess what I'm saying is, what's wrong with a fight between pro tanto reasons?)

Now, sometimes an actor who satisfies the level 2 norm may in fact produce an act that does not satisfy with the level 1 norm, as when (on the truth account of assertion) someone makes a justified false assertion. But then it seems reasonable to say that the asserter acted properly, but that the assertion didn't live up to the norm of assertion.

If you want to say norms always clamor for the attention of something or other--they're always about permissibility--then there's a problem, but I don't see why the word 'norm' should be restricted that way. We can say "X isn't a good F" without thinking that someone ought to have made X a good F. [I agree about Allan's example--though I'd say that there's no act there at all, not an act without rational explanation.]

Thought you'd appreciate the Bears example, Jon. I don't think it'ss absolutely accurate that in sports we care only about good stats; scouts, for instance, may care about whether players are doing what ought to succeed as well as what will succeed (though in the long run success is by far the best measure of 'ought to succeed'). But I do think Lackey's sports examples provide the very clearest case of norms on acts, because the most important and fundamental thing is definitely whether the play succeeds according to the rules. That's what's reflected in the score.

The problem arises when we try to transfer lessons from these examples to endeavors, like assessing assertions, where we don't explicitly keep score. In that case, we sometimes feel "something's wrong." Williamson's argument for the knowledge account is largely based on our feeling that something is wrong with the lottery assertion, for instance. But--and this is the main philosophical point of the post (and I think what Keith was saying on CD)--those intuitions can be of the sort "That act was wrong" or "The agent was doing something wrong," and we shouldn't assume we have an easy way to pick those apart.

That's why (and this brings it around to Trent's comment) Williamson brings up the Snow example, to say that when we feel that the speaker is doing something right, it's because he is reasonable to think he's matching the norm. And it's also why (as Keith DeRose says in "Assertion Knowledge and Context") when you're looking at cases to test the Knowledge Account, it's best to look at cases in which the speaker is well aware of the evidential status of her assertion. So in the lottery example, and in my Aubrey/Holmes examples, the speakers not only don't know what they say, they know that they don't know it.

Posted by: Matt Weiner at December 5, 2005 07:55 PM


Chances are it could remain a mere verbal dispute, but as the sort of view that I suspect that you are going for is rejected by some because of its alleged commitment to the claims some people (e.g., Dancy) take to be absurdities, it's useful to see if there are ways of stating the view that aren't committed to any such things. Anyway, I apologize for length, but here goes...

I posted a bit on this at my place and wrote oodles about it in my snappily titled dissertation, but suppose you accepted the following norm for action:

U: One must X iff X produces at least as good an outcome as the alternatives to X-ing.
U2: One must X iff X is such that one is rationally compelled to believe that X-ing is at least as good an outcome as the alternatives to X-ing.

It seems given imperfect information, if you thought that U and U2 were both norms (where U2 stems from U because that's what norms do), you'd have to accept that there are mandatory impermissibles. There are three standard ways of avoiding this:
(1) deny that we could have imperfect information because there is an epistemic filter or because we should aim at maximized expected utility;
(2) deny there are positive duties;
(3) deny that there is a real contradiction and insist that norms confer pro tanto reasons.

In the case of assertion and belief, I suppose (2) might be more attractive than it is for action but because the problem arises in the case of ethical action and seems to be a general problem about norms, we should look elsewhere for the solution.

We don't want to follow (1) because it's clear that there are norms that tell us to respond to conditions that are external. If I assert p when p is false, my assertion is defective and there is a perfectly good sense in which I shouldn't have said it (even if there is a perfectly good sense in which there was something good about asserting it).

As for (3), we get into difficult issues about what pro tanto reasons are. It is hard to know what is essential to pro tanto reasons, but typically when pro tanto reasons conflict, one is capable of coming to an all things considered judgment that sides with one reason against the other without thereby thinking of the other reason as spurious. The agent can reach no decision of this type when U and U2 conflict. Like I said, I'm not certain that this is essential to the pro-tanto, but there is a second piece of evidence I think is much more striking.

Suppose that I know there is a p.t. reason for you to X and another for you to Y and know that you can't both X and Y and you've asked for my advice. As an advisor, I'd feel the pull of both reasons and have to negotiate their demands in determining what the correct advice to offer is. There is no such conflict when I (as an advisor) see that you satisfy only one of the primary and secondary norms. From the perspective of the advisor, the advisee is quite simply not under the requirement of any secondary norm at all. Even if I know you cannot rationally accept my advice on the matter because I cannot tip the balance of evidence in favor of thinking that X will maximize utility, if I know that X will in fact maximize utility, I know that the correct advice to give is this: you should X. I also know that there is no reason for you to Y. While the former judgment is one I can make when I see (for example) that the demands of justice outweigh those of friendship, I cannot make the latter type of judgment when I see that your evidence makes it look as if Y maximized utility when X is really what does so.

This seems to be some evidence against thinking of secondary norms as pro tanto reasons. I'd also add that when a pro tanto reason is defeated, it still places a demand on somebody (e.g., to apologize, to compensate, generally to do the second best thing). If the duty to render aid defeats the duty I had to meet you for a drink, the demand placed on that defeated reason doesn't go away during or after the period of time in which I can comply and meet you for a drink. That's why afterwords, I should apologize, offer to meet you another time, etc. Again, the secondary norms seem to require no such thing. They lack all the marks of resultancy and, if Ross is right, what is crucial to the pro tanto is that it has a real moral force that can be understood by way of his analogy of winds and vector forces.

I think it's much better to think of these secondary norms, reasons, and duties as decoys. Their presence is crucial to explaining why someone should be excused for a failure to hit a real duty or conform to a real reason or norm but they don't show that there is some positive feature about the act itself that spoke in favor of choosing it over the available alternatives.

At any rate, if there are theoretical problems that arise when we speak of secondary norms as if they were norms and there is no trouble accounting for the intuition that leads people in this direction simply by helping ourselves to the justified/excused distinction, I think talk of secondary norms and propriety is best avoided. There are other theoretical problems that arise in connection with basing, but as there are likely problems with what I've said thus far, it might be better not to list the 239834 problems I think arise for the view since really they can all be handled rather straightforwardly without really altering the substance of the position.

Quick question: do you defend the T account of assertion? I can't open pdf files on my parents' computer--they keep freezing up. I'm convinced that Williamson's brusque dismissal of the T account actually is due to taking the primary/secondary norm/propriety distinction in the wrong direction so sometimes the dispute isn't just verbal, sometimes it leads you to write one of the best books in epistemology I've ever read.

Posted by: Clayton at December 5, 2005 10:26 PM

Oh, and that Sufjan is really good but when I told a friend of mine that the 'really beautiful' song we were listening to when driving to dinner was about John Wayne Gacey, I think I might have spoiled the album for him.

Posted by: Clayton at December 5, 2005 10:28 PM

I don't see the problem with pro tanto reasons.  I think it's a pretty
basic notion and I'm not much more worried about a lack of analysis than in the
case of causation.  Sure, I'd like to have an account, but both concepts
are intuitive enough.  Can you maybe set that argument out with premises
and all?  I'm not sure how considerations regarding some hypothetical
advisor are relevant. 

It's not clear to me that secondary norms don't have the sorts of results you
say they don't.  In fact, Williamson argues for the Knowledge Account
specifically by noting that people make excuses when it is pointed out that what
they asserted wasn't known (see
Williamson, “Knowing and Asserting,” The Philosophical Review 105.4
(October 1996), pp. 489-523).  This fits nicely with secondary norms having
such requirements. 

I'm just not sure why you want to
limit the use of "norm" in the way you apparently do.  Norms are not
necessarily imperatives, the normal usage is a pattern or standard.  So we
can evaluate acts or we can evaluate agents, we've got one standard for one and
another for the other.  So we've got two kinds of norms, as long as we know
what we're talking about what's the big deal?

Posted by: Trent Dougherty at December 5, 2005 11:31 PM


I suspect that there are some pretty deep disagreements between us.

You wrote "the normal usage is a pattern or standard". I don't get that at all. Standards are normative; patterns are not. It seems pretty clear that what W is up to in specifying the norms of assertion is specifying the normative requirements on speakers rather than telling us what is typical or standard for an assertion.

As for pro tanto reasons, I don't think it is just some intuitive notion like the notion of a cause. It is a rather technical notion that finds its origins in Ross' The Right and the Good. The basic idea is that when one has a p.t. reason to X that isn't opposed or excluded, one should X. I'm not sure whether you have this in mind or not, but it's what I have in mind when I use the term.

So, here's an argument against thinking that secondary norms provide normative reasons for belief/action/assertion. It's partially inspired by Judith Thomson but I've added a slight twist. Here it is:

(1) An advisor may be warranted in asserting 'You should X' even if the advisor knows the advisee would violate the secondary norm so long as the advisor knows that the advisee would satisfy the primary norm.
(2) The advisor knows 'You should X' is true even if the advisee violates the secondary norm. [K account of assertion]
(3) The advisee should X even if the advisee violates the secondary norm. [Factivity of knows]
(4) Either the secondary norm provided a defeated reason to refrain from X-ing or provided no reason at all. [General claim about reasons and oughts].
(5) If the secondary norm provided a defeated reason, the advisor would not be warranted in asserting 'There is not really any good reason for you not to X'.
(6) But the advisor is warranted in asserting 'There is not really any good reason for you not to X'.
(C) Secondary reasons provide no reasons at all.

An example to make this concrete. I've promised to keep an appointment, forgotten, and acquired massive amounts of evidence to suggest that my rival will try to convince me that I have an appointment to keep because he likes me to waste my lunch hour. You know all this and know that I'm supposed to meet you for lunch. You advise me, calling me from the restaurant. You know that your testimony won't tip the balance of evidence making it more likely than not that given my evidence what you say is true, but nevertheless, you call and tell me that I should meet you for lunch because I promised. If you think what you say is true, we're off to the races.

Even if you want to maintain that there really are pro tanto reasons associated with secondary norms, let me ask you this: will they ever win the day? Will there ever be a case in which there is a conflict between a primary and secondary norm where the primary norm would assign some deontic status to an act, the secondary norm would assign a different status, and the secondary norm assigns the right all things considered status? If not, why call 'em norms? Why not norm shadows or cliff clavens?

Sorry Matt for taking up so much room.

Posted by: Clayton at December 6, 2005 01:31 AM

Oh, and that Sufjan is really good but when I told a friend of mine that the 'really beautiful' song we were listening to when driving to dinner was about John Wayne Gacey, I think I might have spoiled the album for him.

Had a brief exchange with someone about this at Unfogged; I'm thinking of posting about it. If I do, it'll allude to the Updike's entry in the bad sex writing contest. To say any more would spoil it.

Thoughts about philosophy later, I hope.

Posted by: Matt Weiner at December 6, 2005 07:46 AM

New Page 1

That patterns can be standards I take to be uncontroversial.  It's one
way to think of syntactic inference rules for example.  It is also
normative for how to set up a chess board, etc.  You must have been taking
me to mean more than I meant to mean (or less).  In your second sentence
you supply "typical"; that's not mine.  You must have thought that by
"pattern" I meant something like a normal curve (not that that can't be
normative mind you).  That I combined it with "standard" I thought made
that clear that I wasn't speaking of "normalcy" in that sense.  Rather, I
meant a pattern one had some obligation to follow.

I'm quite familiar with Ross and PTR's, and so is almost every ordinary
person, though not under that description.  Prior to Ross people knew that
if they promised to do something, but following through on that now would get
someone killed, then they had to break the promise, even though ordinarily
they'd have to keep the promise.  They might not have had a fancy Latin
name for it, but it was part of their conceptual machinery.  You might even
see Sophocles Antigone as expressing just how seriously ancient folk took
conflicts in duties.   

Thanks for the argument though, that's the kind of stuff I like. 
Unfortunately, as you suggested, I think we are at the antipodes w.r.t. moral
theory.  Insofar as I understand the argument, I think a fallacy of
equivocation is committed in (4).  As far as I can tell the sense of
"reason" there invokes a different oughtness than the "should" of (1). 
Primary and secondary norms are of a different kind: the objective sense of "S
should A" is something like "all things considered it would be best if S A'd"
like when you are watching someone play chess (sorry, couldn't think of a sports
analogy) and say "He should have taken that Pawn en passe" but the
reasons I have to do an action have little or nothing to do with that, those
things are internal to the subject.  The standard for how the agent should
be evaluated along the dimension of how he used his agency is how things looked
to him at the time.  Even if I think it would have been better for S if she
had A'd, from S's perspective, say S didn't even know about the en passe
move, then it's not the case that she should have moved that way.  Indeed
she an obligation *not* to do that.  So, to turn to your example, I would
not make the call from the restaurant in the first place if I was certain I
couldn't provide you with a reason to come meet me, because if I did I couldn't
say "You should come meet me" because I wouldn't want to say something I knew
was false or irrelevant (false if subjective sense, irrelevant if objective

For the record, I think what Williamson and Co. call "secondary norms" are
really the primary norms: I think they win the day every time.  I'm
not even comfortable with calling the other things "norms," but, as you know, I
don't like to legislate limits in linguistics.  In particular, I'm inclined
to think there is no such thing as the sort of "norm" Williamson and Co. call
"constitutive" and if there are such things I don't think there are any
interesting ones involving knowledge (probably none at all), certainly not a
norm of assertion (but then again I don't find knowledge interesting in the
first place). 

Posted by: Trent Dougherty at December 6, 2005 07:47 AM

You wrote:
I'm quite familiar with Ross and PTR's, and so is almost every ordinary
person, though not under that description. Prior to Ross people knew that
if they promised to do something, but following through on that now would get
someone killed, then they had to break the promise, even though ordinarily
they'd have to keep the promise. They might not have had a fancy Latin
name for it, but it was part of their conceptual machinery. You might even
see Sophocles Antigone as expressing just how seriously ancient folk took
conflicts in duties.

I didn't mean to imply that you didn't know the Ross, but there seem to be at present two very different uses of pro tanto. On one usage, (the pleonastic usage), there is a pro tanto or prima facie reason to X whenever the agent is rational in X-ing. On the other usage (the Rossian usage) a pro tanto reason is a right making feature. It is controversial whether the two come to the same thing and for the most part, epistemologists seem inclined to think of pro tanto reasons pleonastically. I think this is a serious, serious mistake, but at least if it is explicit, there is something to talk about.

At any rate, I agree that the ancient Greeks seemed to have a decent understanding of the phenomenon but things got fouled up by Kant and the utilitarians who both seemed to think there couldn't be real conflicts of duties [The best thing I've read on this, by the way, is Gardner's 'In Defense of Defenses' (warning: pdf)]. Because people have fouled it up, I can't just assume we're all on the same page. The Antigone story, I take it, is a serious threat to ought implies can whereas it wasn't really until Ross that we recovered what we needed to do justice to this.

Alright, so I think the two ways in which prima facie and pro tanto come apart are precisely in cases where excusable mistakes are made. That they are excusable implies that the mistake can be given a rational explanation but that the agent's failure to do what she should doesn't prevent us from seeing her in a positive light on that occassion. There is a temptation in such cases to say that she was prima facie justified, but if you don't carefully distinguish the two uses, you are thereby committed to saying that there really was some good reason for her to do what she did. But then, you aren't offering excuses any more, you're offering a justification (one that the agent disowns, by the way, once the facts come to light). This comes out in the argument. When we compare forgotten promises to overridden promises, that's where we see the important contrast. When there is an overridden promise, we can say two things:

(i) The agent should break the promise.
(ii) The agent has a good reason to keep the promise.

When there is a forgotten promise, we cannot say (i) but can say (ii). If you think that in the case of the forgotten promise, the agent should break the promise, you have to say (ii) is true [Because whenever you ought to X there is a reason to X]. However, I take it we all think that the advisor cannot truly assert (ii) in the case of the forgotten promise.

You seem to be saying two things in response to the restaurant example: there is an equivocation; you aren't warranted in asserting that there is an obligation to meet at the restaurant.

I can argue against the equivocation claim but I'll have to wait for an outside judge [Though I will note that precisely the people who this argument is directed against share the intuition that it is true to say 'You should keep the appointment' and come up with all manner of ad hocery and semantic dodginess to account for it--see Jackson's decision theoretic consequentialism paper and Dancy's Practical Reality for example].

So, you'll not be impressed with THIS argument against your claim regarding secondary norms, but here goes. Cooper is a BIV. You can communicate with him via a computer that creates an image of you that is a seamless part of his virtual reality. He asks 'Should I believe I have hands?'. I take it you think you'd not be dishonest if you said 'Yes, you should believe that' but you would be lying if you said 'No, you should not believe that'.

Posted by: Clayton at December 6, 2005 09:45 AM

re: Promises
I'm not sure I understand "breaking" a forgotten promise, so I don't understand the comparison.

re: BIV
Right, that's just what I'd say.

Posted by: Trent Dougherty at December 6, 2005 10:47 AM

I would have thought that a promise was broken if not fulfilled and while the person to whom the promise was made could dissolve the promise, the promise isn't dissolved just because someone's forgotten. At any rate, it doesn't matter whether it is promises, all we need is some right-making feature which can obtain, intuitively shift the balance at the level of overall reason while the subject is mistaken about it. As the semester comes to an end, though, and I break all manner of promises to students I've made over the past few months because I've forgotten making them, maybe I can refer some to you to correct them when they accuse me of breaking promises I made to them. I'm not sure how this defense will play out:

Of course he promised you you'd get extra credit in the past, but just because he didn't give you the extra credit doesn't mean he broke his promise--he forgot. In fact, because he forgot and wouldn't believe either of us if we reminded him, there is no reason at all for him to give you extra credit and students who try to get extra credit when there is no reason for them to get it are in the wrong.

I thought that's what you'd say about BIV. But then, what role do you think advisors play? Coop asks whether he should believe he has hands. You tell him he should. If Coop later discovers that he had been deceived and recalls the incident where you told him that he should believe he had hands, had a loving family, etc... when you knew he didn't, can't he accuse you of being dishonest? If the advice you offer was a mistake (he'll later say it was), then the advice wasn't correct. If it's not correct, he should have done something else. If he should have done something else, you should have told him to do something else. If your theory assigns T to the assertion 'You should believe you have hands', your theory is mistaken and needs to be fixed.

Posted by: Clayton at December 6, 2005 11:30 AM

Hmmm.  To "break" a promise just sounds like an intentional act to me,
but I think I can still address the case.

One wrinkle in the extra credit case is that my memory is pretty bad and
being unfair has pretty high disutility for me, so the students wouldn't have to
convince me that it was probable that I promised, I'd give the extra credit if I
thought there was much chance at all that I'd promised.  I wouldn't know
how to advise your students without knowing if you're like me in that respect. 
I guess I'd assume you were and tell them to ask for the extra credit. 
However, in the case where you are certain you didn't promise, then you have no
(intrinsically motivating) reason to do what you promised (and I think it would
be wrong of you to do so, morally and rationally). Again, I would say "Ouch, he
should have given them the extra credit" in the same since as "Ouch, she should
have taken that pawn" where this means "It would have been better for you both
if you'd seen things from God's perspective and acted accordingly."  I just
don't think that's a standard according to which we should judge agent's use of
agency which is what I think ethics should be about. 

I have no general theory of the role of advisors, nor do I suppose that there
is any univocal role across contexts, but my gambit would be to try to convince
them of what I think is true in cases where we disagree and I think there is a
reasonable chance of success.  This would include truths about what people
ought to do in various circumstances.  Ultimately, I'd tell them to follow
their conscience. 

Now the following is just too good to be true given the talk of patterns
above.  I got this in an email from Henry Kyburg in response to a question
about the study guide for a nonmonotonic logic final a few minutes ago:

"I never mentioned (what, of course, I thought would be obvious) what I
would ask about the patterns.  The patterns are rules of
inference. You got stuff that looks like what is above the line, you
may infer what is below the line.  An interpretation of the
makes a rule of inference o.k. If, whenever what's above
the line is true, according to an interpretation, so is what's below the line,
according to the same interpretation.  So to show the rule valid for an
interpretation, that's what you show. "

I should specify that this is not a "in your face" quotation, it's a "what a
coincidence" quotation.  Emotions are hard to read on the page and I find I
get misinterpreted with such frequency that it's better to say the obvious than
to risk misunderstanding and offense.  What's kind of funny is that when I
wrote that rules of inference were normative syntactic patterns I was like "Man,
I hope that's not a misleading statement" (I'm more of a semantics guy) then I
get this message which is just as blatant as could be.  (It's also
interesting to note that some theories of ancient geometry say that the figures
were what were really considered the proofs and what we call the proofs were
originally just annotations.)

Posted by: Trent Dougherty at December 6, 2005 04:41 PM


It's too bad there wasn't a character on Seinfeld named Kyburg--that would have sounded cool coming from Kramer.

I have no theoretical account of the role of advisors, but when I think about intuitions about correct advice, I'm pretty convinced (thanks to Judith Thomson) that the advisor is not concerned with the advisee's subjective motivational states in determining what advice to give apart from those aspects of their motivational states that specify their ends. Intuition tells me that if I know that p and know that you're in no position to know p (and won't be after I tell you), this is consistent with it's being the case that the correct advice to give is for you to believe p (or at least, to refrain from believing ~p). My favorite example: a secret cabal involving the queen, myself, and your former mailman has conspired to poison your soup. Your sometimes insane friend learns of this and says 'You should not eat the soup'. What he says is true and you'll learn this in heaven after our plan has worked (mwah hah hah hah!). But, if you don't have the intuition, that's alright.

For the record, I didn't want to pin on you the view that you thought norms were things like generalizations about what's normal--I just needed to draw you out to be certain.



Posted by: Clayton at December 6, 2005 05:38 PM