Thursday, December 15, 2011

Frege's Puzzle and In-Between Cases of Believing

There's a huge literature in philosophy of language on what's called "Frege's puzzle" about belief reports. Almost all the participants in this literature seem to take for granted something that I reject: that sentences ascribing beliefs must be determinately true or false, at least once those sentences are disambiguated or contextualized in the right way.

Frege's puzzle is this. Lois Lane believes, it seems, that Superman is strong. And Clark Kent is, of course, Superman. So it seems to follow that Lois Lane believes that Clark Kent is strong. But Lois would deny that Clark Kent is strong, and it seems wrong to say that she believes it. So what's going on? There are several standard options, but all lead to trouble of one sort or another. (If you don't like Superman, try Twain/Clemens or Unabomber/Kaczynski.)

On a dispositional approach to belief of the sort I favor, to believe some proposition P -- the proposition, say, that that guy (variously known as "Superman" or "Clark Kent") is strong -- is to be disposed to act and react, both outwardly and inwardly, as though P were true. (On my version of dispositionalism, this means being disposed to act and react in ways that ordinary people would regard as characteristic of belief that P.) Lois has some such dispositions: For example, she's disposed to say "Superman is strong". But she notably lacks others: She's not disposed to say "Clark Kent is strong". She's disposed to ask Superman/Clark Kent to lift her up in the air when he's in costume but not when he's in street clothes.

Personality traits also involve clusters of dispositions, so consider them as an analogy. If someone is disposed to be courageous in some circumstances and not courageous in other circumstances, it might be neither quite right to say that she is courageous nor quite right to say that she isn't. "Courageous" is a vague predicate, and we might have an in-between case, in which neither simple ascription nor simple denial is entirely appropriate (though there may also be contexts in which simple ascription or denial works well enough -- e.g., battlefields vs. faculty meetings if she has battlefield courage but not interpersonal courage). Compare also "Amir is tall", said of a man who is 5'11". Lois's belief about Superman/Clark Kent might similarly be an in-between case in the application of a vague predicate.

You'll probably object that Lois simply and fully believes that Superman is strong, and it's not an in-between case at all. I have two replies. First, that way of putting it -- in terms of Superman rather than Clark Kent -- highlights certain aspects of Lois's dispositional profile over others, thus creating a conversational context that tends to favor believes-strong ascription (like a battlefield context might favor ascription of courage to a person who has battlefield courage but not other sorts of courage). Second, consider a version of the case in which the belief ascriber doesn't have the name "Clark Kent" available, but only the name "Superman". The ascriber and his friend are looking through a window at Superman/Clark Kent in street clothes. The ascriber's friend, who doesn't know that Lois is deceived, asks, "Does Lois believe that Superman is strong?" What should the ascriber reply? He should say, "Well, um, it's a complicated case!" I see no point in insisting that underneath that hedge there needs to be a determinate metaphysical or psychological or (disambiguated [update Dec. 16: e.g., "de re / de dicto"]) linguistic fact that yes-she-really-does (or no-she-really-doesn't), any more than there always has to be a determinate fact about whether someone is tall simpliciter or courageous simpliciter.

Now this is a heck of a mess in philosophy of language, and I haven't thought through all the implications. I'm inclined to think that excessive realism about the identity of propositions is part of the problem too. I don't claim that this is a full or non-problematic solution to Frege's puzzle. But it seems to me that this general type of approach should be more visible among the options than it is.

[HT: Lewis Powell on Kripke's Puzzle.]

60 comments:

Roy said...

Lois doesn't know that Clark Kent is Superman. Apparently she's not puzzled, she's just dumb.

Tristan Haze said...

Very interesting! In philosophy of language, it's common to distinguish various sorts of belief reports, or readings for belief reports. One well known example is Quine's distinction between 'believes-notional' and 'believes-relational'. In this post, you seem to be fixing on something like 'believes-relational' - i.e. something like a Russellian or objectual view of belief contents (or belief-content-specifications).

I think belief reports sometimes work that way, but not always. When 'S believes that a is F' works that way, it can be paraphrased as 'a is such that S believes of it that it is F'. This in turn could be expanded, I think, to 'a is such that S believes of it, via some mode of presentation, that it is F'.

(In this sense, Kripke's Pierre believes, of Paris, that it is pretty, and also believes, of Paris, that it is not pretty. But since he does it via separate modes of presentation, he need not be being irrational - only ill-informed. It seems like Lois could be in a similar situation, but with Clark/Superman, 'strong' and 'not strong'.)

So, what I would say is this: in the objectual sense ('believes-relational'), Lois believes that Clark/Superman is strong, and also that he is not strong, but via separate modes of presentation.

In the notional sense - i.e. where something like the sense, not just the reference, of the belief's content is specified - Lois lane believes that Superman is strong (metalinguistically: believes 'Superman is strong' to be true), and does not believe that Superman is not strong.

So, I don't see any special need to invoke vagueness here. I think part of the pull toward this might come from confusing relational/objectual reports with notional ones. Or, not clearly realizing that, in the relational/objectual sense, it is perfectly possible for a rational but ill-informed agent to have contradictory beliefs - beliefs which can't both be true.

One reason to favour my approach, I think, is that it copes better with Kripke's puzzle: consider the fact that, if Pierre had never had the postcard/'Londres' presentation of London, but only the presentation he got from going there, we would not hesitate in saying he believes not to be pretty.

Analogously, if Lois Lane hadn't gone into journalism and had never seen Clark Kent in his Clark Kent guise, or heard the name 'Clark Kent', we wouldn't hesitate to say she believes Superman to be strong. Contra your account as I understand it, her lacking dispositions to assent to 'Clark Kent is strong' seems like no bar to her believing that Superman is strong in that case.

What do you think?

Anonymous said...

In this case the system has not been properly contextualized and desambiguated.

Ken Marable said...

"Almost all the participants in this literature seem to take for granted something that I reject: that sentences ascribing beliefs must be determinately true or false, at least once those sentences are disambiguated or contextualized in the right way."

I'm with you (for perhaps only tangentially similar reasons). The day I was introduced to American Pragmatism was the day many of these strict claims of language and knowledge lost all pull on me.

Eric Schwitzgebel said...

Thanks for the interesting comments, folks!

@ Tristan: I have pretty much the same view about notional/relational and about Kripke's puzzle as I have about Frege's puzzle. I even used Kripke's puzzle as part of the motivation for my dispositional account of belief in Chapter 7 of my disseration:
http://www.faculty.ucr.edu/~eschwitz/SchwitzPapers/DissContents.htm

I don't see the need for clean yes-or-no facts in such matters. I think that insistence is part of what leads Kripke into trouble. Same also in de re / de dicto cases, as nicely argued I think in Stich 1983 and Dennett 1987. There's a mushy phenomenon underneath that our belief-attribution language only approximates to differing degrees. So consider the Superman case I discuss near the end of the post where the name "Clark Kent" isn't available. One can try to generate clean yes-or-no belief-ascription facts by appealing to different senses of belief or different types of attribution, but I wager I can build cases that seem plausibly to be in-between for any such disambiguation move one cares to make. It still could be useful to talk about different types of belief attribution -- why not? -- but if the motive is really just to save a clean always yes-or-no approach to belief or belief ascription, I think that motive is ill-grounded.

The view that people do often possess baldly contradictory beliefs does flow from certain views, but that seems to me to a form of biting the bullet, to be avoided if possible.

On your final point, the relevant dispositions, on my account, always have tacit conditions or ceteris paribus defeaters. Not having the name "Clark Kent" available thus excuses the Clark-Kent-is-strong believer from failing to assent to that sentence despite possessing the belief. So in the case you imagine, Lois would be excused from dispositional manifestation and not count at all as failing to conform to the stereotype for believing that Superman/Clark Kent is strong.

Eric Schwitzgebel said...

@ Anon: Just to be clear, my whole point is explicitly to challenge exactly what you said. See the last phrase of my first paragraph. Right?

Eric Schwitzgebel said...

@ Ken: I think you're right that there is something in the spirit of American pragmatism that is very much at odds with the background assumptions of most of the literature on Frege's puzzle. Is it a rejection of clean lines and of sharp true/false categorizations?

Anonymous said...

I'm partial to your position on the puzzles. But what reasons do you think we have for preferring your solution to that of someone like Nathan Salmon. If I'm remembering right, he says something like that Louis believes that Superman is strong; Clark is strong (under the Superman guise); Superman is not strong (under the Clark guise); and Clark is not strong. On this picture of things, there are no "in-between" beliefs, just a few plain old beliefs. We are able to capture Lois' conflicted dispositional profile but we don't have to deny bivalence. (I'm not too worried about denying bivalence, but I can see some folks getting queasy about it.)

Also, I'm hoping you can say a bit more about this "excessive realism about the identity of propositions."

Tristan Haze said...

Thanks heaps for the response. You write: 'It still could be useful to talk about different types of belief attribution -- why not? -- but if the motive is really just to save a clean always yes-or-no approach to belief or belief ascription, I think that motive is ill-grounded.'

I agree with you there. This paragraph makes me think we probably have different focuses here: you're interested in critiquing yes-or-no-ness as a questionable motivation for busting moves in this area, while I'm interested in the project of clarifying different sorts of belief-report (for its own sake, as it were, or to throw light on related issues to do with de re modality).

I'd better have a look at your dissertation.

Tristan Haze said...

Pardon my commenting again, but I've been thinking about this paragraph in your response:

'The view that people do often possess baldly contradictory beliefs does flow from certain views, but that seems to me to a form of biting the bullet, to be avoided if possible.'

That fascinates me, because I just can't see what's bullet-like about it. For a while now, I've felt sure this must just feel bullety because of equivocation over the relational/notional distinction (or something like it). This is the second time I've been in a discussion where someone insists on the bullet-likeness of this thing. I'm now very keen to work on this point, so if there's anything more you could say about it, I'd be interested.

Thanks again and Merry Christmas!

Roy said...

"Frege's puzzle is this. Lois Lane believes, it seems, that Superman is strong. And Clark Kent is, of course, Superman. So it seems to follow that Lois Lane believes that Clark Kent is strong. But Lois would deny that Clark Kent is strong, and it seems wrong to say that she believes it."
So if the premise is that Lois actually does know that Clark Kent is also Superman, why would Lois want to deny that Clark is strong, unless in fact he's shown her that until he changes his persona, he isn't?
Perhaps it's just me, but I still see this lead-in as a very bad example of a belief paradox.

Edward Ockham said...

Frege's puzzle is usually addressed using belief ascriptions. What about sentences like "there is strong evidence that Clark Kent is Superman"? Clearly there is strong evidence that Clark Kent is Clark Kent". That's guaranteed by logic, practically. But there is clearly not so much evidence that Clark Kent is Superman. Substitution seems to fail.

Or 'discovered that'. There is a flake theory that Edward de Vere was really Shakespeare (I think they've even made a film of it). If we discovered in 2012 that the theory were true, it would not be true to say that "in 2012 it was discovered that Edward de Vere was Shakespeare". For the law of identity was discovered in Aristotle's time.

I'm not sure how your theory would engage with these 'impersonal' verbs that take that-clauses, like "it was discovered that", "there is evidence that", "theory X says that" and so on.

Another puzzle. We say "According to the Oxfordian theory, Edward de Vere was Shakespeare", "according to the Baconian theory, Francis Bacon was Shakespeare" etc. All those statements about these theories are true. But it is hard to explain why they are true.

Edward Ockham said...

Eric >>Almost all the participants in this literature seem to take for granted something that I reject: that sentences ascribing beliefs must be determinately true or false
<<

If this works, it must be also true of the impersonal ascription sentences. Thus, per te it is not determinately true that the Oxfordian theory holds that Edward de Vere was Shakespeare.

But most people accept that this is true. I often warn people against unquestioning acceptance of Wikipedia, but what it says here today seems unquestionably right.

It says "The Oxfordian theory of Shakespearean authorship proposes that Edward de Vere, 17th Earl of Oxford (1550–1604), wrote the plays and poems traditionally attributed to William Shakespeare of Stratford-upon-Avon. "

That is unquestionably true. That is what the Oxfordian theory is. Whether the theory is right or not is irrelevant. The question is whether this characterisation of the theory is correct. It is, and it uses, without mentioning, the proper names 'Edward de Vere' and 'Shakespeare'. Its truth would not survive substitution. The proponents of that theory do not hold that Francis Bacon was Edward de Vere, and it would be true to say that even if Francis Bacon were identical with Shakespeare.

Eric Schwitzgebel said...

Thanks for all the interesting comments, folks!

@ Anon Dec 16: Salmon's view seems to me to have more machinery than is really necessary, with the distinction between "believe" and "BEL", and weird patterns of attribution such that one can believe P and also withhold belief from P simultaneously, and one can believe flatly contradictory things. Why accept all this? I suspect it's partly motivated by a desire for clean lines that aren't to be had anyway. Treating attitudes as like personality traits seems to me to allow us to dodge much of that rigmarole.

Eric Schwitzgebel said...

@ Tristan: There seems to me to be an unnaturalness in saying things like: It is the case that both that Lois believes that Superman is strong and that Lois believes that Superman is not strong. On the surface, such a thing just doesn't seem to make sense. At least, I think if you said that to a non-philosopher, they wouldn't know what to make of it. So if your view commits you to that position, that seems to me a cost of the view. It could be a cost worth paying, if the theory has enough other virtues.

Eric Schwitzgebel said...

@ Roy: The premise is that Lois *doesn't* know that Clark Kent is Superman. I guess we're talking about the main 20th century DC world!

Eric Schwitzgebel said...

Oh, more for Anon 12/16, on the excessive realism about the identity of propositions: I don't have a well developed view on this, but I am suspicious about the ontology of propositions and the clean lines there. For the sake of the post, I am allowing that "Superman is strong" and "Clark Kent is strong" express the same proposition, but I'm not really ready to buy into that whole picture. My guess is that it's an excessively clean ontology for an underlying set of muddy, human facts about language and muddy gradations in the world.

Eric Schwitzgebel said...

@ Edward: You're absolutely right that puzzles about non-substitutability extend well beyond belief contexts. Since my approach to the belief ascription cases invokes facts about belief in particular, it cannot straightforwardly be generalized to other non-substitutability cases.

I can see how you might read my post as committed to substitutability within belief contexts and thus maybe substitutability generally. The latter I certainly don't accept. Even fully truth-preserving substitutability within belief contexts isn't entirely to my liking. I have a broadly pragmatic view of language and a distaste for the insistence on applying sharp boundaries to blurry cases. These preferences seem to me to conflict with the view that coreferential names are always substitutable in a truth preserving way.

The aim of my post, then, wasn't to defend substitutability as a universal principle but rather to suggest that some cases that might seem puzzling would seem considerably less puzzling if we permitted vagueness in belief attribution. (I warned you that this wasn't a full or non-problematic solution to the whole Frege's puzzle literature!)

Anonymous said...

Anon 12/16 here- I see what you mean about Salmon's view. Might one abandon the machinery to which you object and still follow Salmon in simply saying that Lois is of two minds--or is it four minds?--she has two sets of contradictory beliefs. So it's true that (1) she believes that Clark is strong (S-guise) and (2) she believes that Clark is not strong (C-guise). And it's also true that (3) she believes that Superman is strong (S-guise) and (4) she believes that Superman is not strong (C-guise).

I realize that you're going to say that this is an odd answer to the puzzle because Lois would then have flatly contradictory beliefs. But why exactly is this problematic? She has contradictory beliefs but doesn't realize it. And it's perfectly understandable why she doesn't know that Clark and Superman are the same person. One wears a cape, not-so-comfy tights, and is charming while the other wears glasses and is a bit of a nerd. It doesn't seem like Lois is epistemically guilty of anything, but she nevertheless seems to have contradictory beliefs (or so the Salmonesque line goes). Again, this reply to the puzzle has the "advantage" of determinate truth-values of (1)-(4) (i.e., they're all true). Your reply is what exactly? Lois simply in-betweenly believes that Clark/Superman is strong. But can you stop there? What of the truth value of (1)-(4)? Do these lack truth-values since she bears the relation of "in-between belief" and not "belief" to the relevant propositions. If that's your answer, then it may appear that it's FALSE that she believes that Clark is strong (since she does not believe that Clark is strong, she in-betweenly believes it). But maybe you want to deny something about the logic of "S does not believe P"; "It's not the case that S believes P"; etc.(?)

So, just in case I'm not being clear, I'm confused about what you think of the truth-value (or lack thereof) of (1)-(4) and also about what is so problematic about having flatly contradictory beliefs.

Also, thanks for clarifying the bit about propositions. I completely agree about clean ontology. I hope you'll consider a post in the near future on this topic.

Edward Ockham said...

Hi Eric,

I posted a longer comment about this on my blog today here. I take your point that the problem of substitutability within belief contexts is maybe not the same thing as the problem of substitutability generally, but it seems to me that it is.

Here’s one argument that the problem is the same. The description “The theory that Edward de Vere was Shakespeare” has a substitution problem. Even if Edward de Vere were Shakespeare, this is not the same theory as the theory that Shakespeare was Shakespeare. For most people hold the Stratfordian theory that Shakespeare wrote Macbeth, whereas only a small group of conspiracy theorists hold the Oxfordian theory. Yet we can easily turn this into the belief problem by asking whether someone believes the Oxfordian theory or not. For to believe the Oxfordian theory is precisely to believe that Edward de Vere was Shakespeare. The problem is surely the same problem. Therefore (applying substitution!) the same problem requires the same solution.

Tony Lloyd said...

I think the puzzle depends on missing a “step”. If we make that step clear then the puzzle goes and with it any need to avoid nice clear "true" and "false" attributions.

Let’s use punctuation to identify three different types of entities:

*Superman* is a word. It’s two syllables if spoken, seven letters if written. It begins with an “S”, ends with an “N” etc.

/Superman/ is a concept. It’s an entity that can be part of an argument, that one can draw conclusions from, a Humean “idea” that can form relations with other ideas.

[Superman] is a fictional character that works at the Daily Planet, calling himself “Clark Kent” and flies around in tights under the name “Superman”.

The first thing to notice is that /Superman/ is not [Superman] and *Superman* is neither /Superman/ nor [Superman]. Naturally that also holds for various combinations of *Superman*, /Superman/, [Superman], *Clark Kent*, /Clark Kent/ and [Clark Kent].

To deal with the puzzle we now have to have a way of incorporating this notation into the puzzle cases:

Lois Lane believes, it seems, that Superman is strong. And Clark Kent is, of course, Superman. So it seems to follow that Lois Lane believes that Clark Kent is strong. But Lois would deny that Clark Kent is strong, and it seems wrong to say that she believes it.

Lois has a concept /Superman/, as do we. Our /Superman/ is not, however, the same as Lois’. One thing we infer from /Superman/ is “calls himself Clark Kent”. One thing Lois draws from /Superman/ is “does not call himself Clark Kent”. We can call our /Superman/ “/Superman-us/” and Lois’ “/Superman-Lois/”.

Now it seems nonsensical to say that Lois believes her concept of Superman. Rather it is the case that Lois thinks there exists an entity that corresponds to /Superman-Lois/. And, of course, she’s wrong: there exists an entity that corresponds to /Superman-us/ and /Superman-us/ differs from /Superman-Lois/.

Lois believes that there are two entities answering the concepts /Superman-Lois/ and /Clark Kent-Lois/ where:

1. /Superman-Lois/ entails superhuman strength, and
2. /Clark Kent-Lois/ entails only human strength

“Clark Kent is, of course, Superman” could be re-phrased as saying that the entity that corresponds to /Superman/ is the entity that corresponds to /Clark Kent/. That we believe that Clark Kent is Superman can be rephrased as there is one entity that answers both /Superman-us/ and /Clark Kent-us/.

I think there are many ways of plugging /Superman-Lois/, /Clark Kent-us/ etc. into the puzzle without generating a paradox. One of the simplest would be:
/Superman-Lois/ entails superhuman strength, whilst /Clark Kent-Lois/ entails only human strength. No entity corresponds to either concept, rather there exists a single entity which corresponds to both /Superman-us/ and /Clark Kent-us/.


And, of course, both Lois and our positions are nicely true or false.

Anonymous said...

@Tony

Is this meant to be like Frege's own solution, just calling senses 'concepts'? (I think that he allowed for a name to have a different sense for each individual, but I'm not quite sure.)

I suppose the difference is that you interpret 'LL believes that Superman is strong' as '/Superman-LL entails strength/'. I think (again, not quite sure), that Frege would say instead that, in this context, the statement 'Superman is strong' itself has a sense, call it 'S', which is derived in some undetermined way from the senses of 'Superman' and '_ is strong'. Then 'LL believes that Superman is strong' is interpreted as 'LL believes that (S)', where 'LL believes that' is now a perfectly well-behaved predicate.

I could well have tripped over the subtleties, but I prefer (what I think is) Frege's way, because it seems that it's possible for me not to realise that everything that's entailed by my concept of e.g. CK is so entailed. E.g. /CK-me/ might entail working in an office for the Daily Planet, and /Daily_Planet-me/ might entail having offices in and only in Metropolis, but I might still fail to believe that CK works in Metropolis.

Tony Lloyd said...

"Is this meant to be like Frege's own solution, just calling senses 'concepts'?"

Possibly, in a Fregean "not realising everything entailed by a concept" way.

[Goes of to re-read "Sense and Reference"]

Roy said...

"@ Roy: The premise is that Lois *doesn't* know that Clark Kent is Superman. I guess we're talking about the main 20th century DC world!"

Then you've lost any resemblance to a paradox. She doesn't believe Clark is strong because (for one of a number of reasons) she's too dumb to believe he's Superman.
No truth in conflict with another truth there in her mind.
Thus no puzzle.

Edward Ockham said...

A question to Eric and all the commenters here. Are the following two statements true?

(1) Lois is never depicted as seeing Clark Kent fly
(2) Lois is sometimes depicted as seeing Superman fly

Roy said...

Lois sees Clark Kent fly, but apparently doesn't know it's Clark that's flying. Although it seems that everyone else here accepts that somehow she unconsciously believes it

Eric Schwitzgebel said...

A lot to catch up on here! I have to dash, but I'll read through and reply soon.

Richard Marshall said...

Roy

She believes that Superman can fly. She believes Clarke Kent can't fly. So, of that guy, does she believe he can fly?

Roy said...

"She believes that Superman can fly. She believes Clarke Kent can't fly. So, of that guy, does she believe he can fly?"
Like I said, she believes Superman can fly, and apparently she is justified in concluding so.
She doesn't apparently or consciously recognize, and thus believe, that Clark can fly. And to the extent that's she's never seen the guy fly as Clark Kent, that belief is also justified.
Bottom line, so far she can't be seen as knowingly holding diametrically contrary beliefs, the essence of a paradoxical puzzle.

Tony Lloyd said...

"She believes that Superman can fly. She believes Clarke Kent can't fly. So, of that guy, does she believe he can fly?"

What guy is “that guy”? As Superman = Clarke Kent, both Superman and Clarke Kent are “that guy”, so Lois believes both that “that guy” can fly and “that guy” can’t fly.

Is the issue one of the “object” of belief?

My theory is that what is believed is that a certain Fregean sense (thanks Anonymous) is believed (wrongly) to correspond to the world. The sense /Superman/ has a referent [Superman].

Eric’s theory is the more commonsense/ordinary use theory that statements about a person are believed. That the referent has certain qualities: “[Superman] can fly”.

Eric’s theory leads to a contradiction that is resolved by holding the truth value of Lois’ beliefs to be indeterminate. Mine avoids the contradiction by denying that Lois is talking directly about “this guy”.

Anonymous said...

OK, Roy, here's one way of making it look more paradoxical. It seems true that, if Clark is Superman, then Clark and Superman have all the same properties. But it also seems true that Superman has the property of being believed by Lois to be strong, and Clark doesn't.

(Paradox!)

I don't know if that's the most popular way of generating a paradox from Lois' case among philosophers, but I think it's the simplest. The one Eric uses in the post is similar, just using the rule that whenever anything is true of Superman, it follows that it's true of Clark. Eric's set-up makes the problem more clearly to do with how names work in language, whereas mine makes it look like it's about how properties are attributed (of course there's a connection). Either way, it's first of all a puzzle in philosophy of language.

Roy said...

*so Lois believes both that “that guy” can fly and “that guy” can’t fly.*
She believes they are different "that guys." Which is a mistake that's clearly not a paradox. And if she's puzzled, she doesn't know it.

*But it also seems true that Superman has the property of being believed by Lois to be strong, and Clark doesn't.*
So what? Mistakes again do not automatically make paradoxical puzzles.

Anonymous said...

Roy,

If Superman has the property of being believed by Lois to be strong, but Clark doesn't, then Superman and Clark don't have all their properties in common. But everything has all its properties in common with itself, so they can't be the same thing, i.e. Superman is not Clark.

NB: The puzzle isn't that Superman and Clark are believed by Lois to have different properties - everyone's fine with that. It's that if she believes different things about them, then they actually have different properties, because having Lois believe in your strength is itself a property you can have. And it's absurd for them to have different properties, because they're the same guy.


Let's stop talking about properties for a bit. Here's a simple and attractive theory about names: the only role of a name is to refer to whatever it is the name of. In a sentence, the names are to point out what things the sentence is about, and the rest of the sentence says what's true about them.

Take the sentence 'Superman is strong', which contains the name 'Superman'. The name 'Superman' refers to Superman, and the sentence as a whole says of Superman that he is strong. But 'Clark' is another name for Superman, so if we swap 'Clark' for 'Superman', to get 'Clark is strong', that's still a sentence about Superman, and it still says of him that he's strong. I.e. it says exactly the same thing as 'Superman is strong'.

(Maybe you think that 'Clark' refers to something else - Clark. But if Clark is Superman, then whatever refers to Clark refers to Superman, so 'Clark' refers to Superman. And anyway, whatever's true of Clark is true of Superman - they're the same guy - so a sentence that says of Clark that he's strong says the same as one that says of Superman that he's strong, and replacing 'Superman' with 'Clark' doesn't change what the sentence says.)

But this nice theory doesn't work for 'Lois believes that Superman is strong'. If we replace the name 'Superman' with 'Clark' in that sentence, it seems it no longer says the same thing. In particular, it was true when we used the name 'Superman', but is false now we use 'Clark'.

This is the same puzzle as before, except now it's not so much a paradox as a counterexample to a nice theory. Of course, if the theory is nice enough, it's paradoxical that it has counterexamples, but the point is that you have to modify the theory or defuse the puzzle in some other way, which is what plenty of philosophers of language have spent time doing.


I had written an even longer comment, but it's probably better if I stop clogging Eric's blog and just redirect Roy to http://plato.stanford.edu/entries/frege/#FrePuz

Roy said...

If this were or was simply a puzzle about identity statements, there would of course be no paradox involved, but also there would be no need to involve Lois.

Anonymous said...

Well, we don't really need Lois. It's just that involving someone's beliefs makes the problem clearer. We could, as noted above, just as well compare 'There is public evidence that Superman is strong' with 'There is public evidence that Clark is strong'. Or 'Superman is so called because he is male and has superhuman powers' with 'Clark is so called because he is male and has superhuman powers'.

Even 'Superman is Superman' and 'Clark is Superman' seem to say different things, because only the former is a logical truth.

Anyway, at its heart the problem isn't so much about identity as about names and reference (though it's hard to talk about co-referring names without saying something about identity).

The Stanford article should make things clearer.

Anonymous said...

(And I know the 'so called' and 'is Superman' cases are in principle different, but I think they illustrate the connection of the Lois puzzle to more general puzzles about names.)

Roy said...

My initial point still stands, however, that the Superman and Lois scenario as written, was a bad example of Frege's puzzle about belief reports.
And in any case, if you take a way the true/false dichotomies, you effectively dissolve such puzzles into dissonance, dis-understanding, and deception.
And if you want to better understand Frege, read Russell.

Roy said...

"Even 'Superman is Superman' and 'Clark is Superman' seem to say different things, because only the former is a logical truth."
Actually the former is an apparently obvious assertion that doesn't require either truth or logic (Superman may be a clever illusion, etc.)
'Clark is Superman' tells us something, if true, that's logical and that otherwise we didn't necessarily know.

(Sorry, couldn't let that one pass without a comment.)

Anonymous said...

I don't understand what you're aiming at in your last comment. Maybe 'logical truth' needs some fleshing out (especially since I sometimes doubt if there should be logical truths about identity, unless we're using a higher-order logic). But I was just being quick, and the point is that, since we need different methods to come to know whether what 'Superman is Superman' and 'Clark is Superman' say is true, they can't be saying the same thing. (Maybe 'can't' is too strong, but I don't see anyone getting much joy arguing that.)

To be honest, I don't much enjoy Russell, especially not on language, but then I think that philosophy of language is just a necessary, but boring, tool for avoiding screw-ups in more interesting philosophy. Maybe Eric agrees, hence the use of Frege's puzzle to explore juicier problems about belief?

Eric Schwitzgebel said...

Thanks for the interesting discussion, all! There's too much here for me to comment on every point, but a few reactions:

* @ Anon 12/16: Part of the issue here is that I find independently attractive both a dispositional approach to belief that treats belief as a vague property and the rejection of two-valued logic for vague properties. So it seems to me quite natural and attractive to reject the insistence on sharp truth-values in these cases, just as I do in many other types of cases that don't involve referential confusion (e.g., in my 2010 PPQ essay and my 2001 Nous essay, both available on my website). And dispositional approaches to belief of the sort I favor don't allow baldly contradictory beliefs because it's not generally possible to have baldly contradictory dispositional properties. So my approach seems to me both to be folksier and more natural than Salmon's and to have independent theoretical advantages at least within philosophy of mind. On your 1-4: I think the talk of "guises" might be helpful as a technical move as a way of capturing the way the dispositions splinter in certain types of in-between cases, similarly to the way "believes implicitly but not explicitly" can be helpful (or "battlefield courageous but not socially courageous"). "Believes" simpliciter is still vague and of indeterminate truth value. And there will be complicated cases in which even talk of guises tends to resolve everything cleanly, anyway.

Eric Schwitzgebel said...

Whoops! Pulled away again....

Eric Schwitzgebel said...

@ Edward: Thanks for the blog post. See my comments there.

@ Tony: That seems like a helpful way of cutting up the territory. But it does seem a major cost of your theory that Lois can't talk directly about this guy (who is perhaps pointed at through a window). Yes, my approach requires abandoning two-valued logic, but I think that is a feature, not a bug, and entirely justified on independent grounds, especially in the case of belief ascription (as I have argued, e.g., in papers in PPQ in 2010 and Nous in 2002).

Eric Schwitzgebel said...

Roy: I'm not sure to what extent your comments are objections just to the Lois/Superman case. Do you have the same concerns about Twain/Clemens or morning star/evening star?

You also suggest that abandoning two-valued logic results in "dissonance, dis-understanding, and deception". Well, of course I beg to differ! If you are committed to the universal applicability of two-valued logic, then of course you won't like my approach. But general issues about two-valued logic are probably a topic for another day.

Roy said...

"Do you have the same concerns about Twain/Clemens or morning star/evening star?"
No, because (or as long as) there's no unknowing Lois involved.

"You also suggest that abandoning two-valued logic results in "dissonance, dis-understanding, and deception"
In the context of it's use to construct a puzzle/paradox, I still do.

Roy said...
This comment has been removed by the author.
Edward Ockham said...

Eric-

I think you did post something because it was in my email alert but it seems to have disappeared.

Edward

Eric Schwitzgebel said...

Aw, shoot. If I recall, my central thought was that claims about informative identities resolve either into claims about attitudes (subject to my analysis) or claims about the co-occurrence or not of various properties (e.g., the entity satisfying X, Y, and Z is also the entity satisfying A, B, and C).

Charles T. Wolverton said...

In order to avoid confusing concepts of belief described in the vocabulary of propositional attitudes and those described in the vocabulary of behavioral dispositions, I'll call the latter D-beliefs. I think a simplified definition of Eric's concept (using language from his paper) is:

To "D-believe P" is to exhibit an (in some sense) adequate subset of the stereotype of context-dependent dispositions that are (in some sense) characteristic of those who hold P true.

(The predicate "is strong" is vague in a way that while relevant for Eric's purposes, doesn't seem relevant for mine in the following. So, I have substituted "can fly" to avoid confusion.)

Eric posits proposition P = "That guy can fly", where the male person referred to by "that guy" is also referred to in some contexts as "Superman" and in other contexts as "Clark Kent". He assumes that as a D-believer in P, Lois Lane is disposed in some contexts to utter "Superman can fly" (and possibly to exhibit other dispositions in the stereotype that relate to Superman). He further assumes that uttering "Clark Kent can fly" is also in the stereotype (presumably because "Superman" and "Clark Kent" are alternative names for "that guy"), but that Lois never exhibits that disposition. Eric therefore concludes that this poses a problem (due, I take it, to widespread reluctance to accept a type of "vagueness" manifest here in Lois's supposedly inconsistent behavior).

Avoiding the problems attendant to the (traditional) mental state view of belief (and its accompanying intensional vocabulary) motivated my move to a behavioral dispositions view, so it would be quite disappointing to think that such a move fails to achieve that objective. But I don't see that it does. As I recall, few (if any) members of the society of the Superman comics know that both names refer to the same "guy", so presumably few (if any) would be disposed to utter "Clark Kent can fly" in any context. Ie, that utterance isn't among the dispositions that are "characteristic of one who holds P true". Therefore, Lois's behavior is consistent with D-believing P.

I see nothing lost by considering D-beliefs Ps = "Superman can fly" and Pc = "Clark Kent can fly" separately. Then, that Lois - like almost everyone else - is not disposed to utter "Clark Kent can fly" in any context (nor presumably to exhibit many - if any - other dispositions in the corresponding stereotype) just means that she doesn't "D-believe Pc".

I don't quite understand the significance re traditional belief of the example of the "Clark Kent"-deficient ascriber. But if we stick to D-belief, then assuming that the ascriber knows that Lois is disposed to utter "Superman can fly" in appropriate contexts (and to exhibit other dispositions in the relevant stereotype) he can answer the query "Does Lois D-believe Ps?" with a simple "yes".

Am I missing something?

Edward Ockham said...

Found it - it was older than 3 days so in the 'moderation' box. Thanks

Eric Schwitzgebel said...

@ Charles: Thanks for the thoughtful comment. You're right about most of the structure of my view, except I don't think that vagueness and truth indeterminacy is a problem -- except for those people hoping for clean ways of describing messy cases!

One thing that comes out nicely in your comment is that on my view of belief the relevant stereotype that has to be matched by the believer-that-P is the stereotype of the ascriber's community. So if that community is us who are in on the secret, Lois is an in-between believer. But if the community is the ordinary folks of the Superman-world, then the issue is more complicated. I could see it going either way. One could work it out as you do. Another possibility might involve emphasizing certain coarse-grained propositions, with co-referring terms substitutable, and then maybe it's still an in-between case. I haven't worked through the mechanics of belief ascription by communities ignorant of pertinent facts. I suspect it will be a mess in the end! But as long as it's a mess for *all* views and not just mine, I'm fine with that. In fact, I like it, since I favor views that highlight the messiness of the mind.

Anthony said...

You haven't shown "that sentences ascribing beliefs must be determinately true or false, at least once those sentences are disambiguated or contextualized in the right way". You've merely shown that some carefully constructed sentences ascribing beliefs are not disambiguated or contextualized in the right ways.

That said, I agree with you that sentences ascribing beliefs can't be disambiguated or contextualized in a way such that the sentence is determinately true or false, at least not for all readers at all times. But then, I believe that no sentence can be disambiguated or contextualized in a way such that the sentence is determinately true or false, for all readers at all times.

"(If you don't like Superman, try Twain/Clemens or Unabomber/Kaczynski.)"

I do have a problem with trying to have a meaningful discussion about this when dealing with a fictional character which was presented with multiple contradictory backstories.

But I don't know what the analogous puzzle would be for Twain/Clemens or Unabomber/Kaczynski. Especially not without inventing fictional (and possibly contradictory) backstories.

Eric Schwitzgebel said...

@ Anthony: I'm inclined to think that the determinate-truth-value motive is much of what's driving others' views. If you join me in not being so motivated in the case of belief ascription, I welcome that.

I'm not sure why Twain/Clemens is so different. Someone might say "Twain wrote Huck Finn" and say "Clemens did not write Huck Finn"; fill in the details in some plausible way, and we're off to the races.

Charles T. Wolverton said...

Eric -

Thanks for the reply. That I picked up on the "community" aspect of your view is presumably due to my having also been biased by "communitarian" thinkers (Rorty, Sellars, et al). Although I didn't make it explicit, I interpreted "those who hold P true" in a Rortian "what your peers let you get away with saying" way.

Perhaps mistakenly, I see your critics here as trying to "patch up" the intensional vocabulary of propositional attitudes, while you are suggesting a quasi-behaviorist way of accommodating its identity issues by applying something along the lines of "fuzzy truth". I have no position on either approach in the context of philosophy of language. But in the context of philosophy of mind, I'm inclined to devote my attention to developing a minimal facility with a neuroscience-based extensional vocabulary, hopefully thereby avoiding such issues entirely.

As a loosely (at best) related aside, while a big fan of Davidson's "3 Varieties" et al, I find it increasingly difficult to read those of his mind essays couched in the vocabulary of propositional attitudes. That may be one of the reasons I too found his arguments that you attacked in your dissertation unconvincing, notwithstanding that I'm inclined to agree with his conclusions about children, language, and thought (which means nothing, but more significant is that per PMN, Rorty seems to as well).

Eric Schwitzgebel said...

Charles: I think you're seeing the general structure of the debate broadly as I see it, though I'm probably less sanguine about the near-term prospects of neuroscience.

Anthony said...

>> Someone might say "Twain wrote Huck Finn" and say "Clemens did not write Huck Finn"; fill in the details in some plausible way, and we're off to the races.

Well, so long as you put the sentences in quotes, there's no problem. People can *say* anything they want.

And "Twain" is not "Clemens", even if Twain is Clemens.

Take it out of quotes, and I'm lost as to how it is plausible that someone could simultaneously believe that Twain wrote Huck Finn and that Clemens did not. Someone might be disposed to say "Twain wrote Huck Finn", and disposed to say "Clemens wrote Huck Finn". But being disposed to say "Twain wrote Huck Finn" does not mean believing that Twain wrote Huck Finn. Besides, then you're putting it back in quotation marks. And the word "Clemens" is not the word "Twain", even if Clemens is Twain.

Anthony said...

>> I'm lost as to how it is plausible that someone could simultaneously believe that Twain wrote Huck Finn and that Clemens did not.

The answer should be the same, and it should be the same as the answer to the question, while pointing to a picture of Twain/Clemens, "Does Lois believe that that guy wrote Huck Finn?"

As for what the answer should be, I think it depends on the situation, but in addition to "Yes" and "No" I do agree that the best answer might be "It's complicated".

Anonymous said...

Hi, Martin Boehme here,

I think the fundamental underlying issue involves criteria for same and different.
If Clark and Superman are identical
(depending upon one's criteria for same and different)then it could be argued, Lois does believe that Clark is strong in some way. And if Clark and Superman are not identical then perhaps Lois does not so believe.
You could also say these are just different views of the same state of affairs, or contrarily that these are two different situations entirely.
In any case like and unlike, it seems to me, is at the heart of it.

Eric Schwitzgebel said...

Martin, how do feel about the idea of being kind of the same? Does it have to be determinately either the same or different?

John Jones said...

A belief is the decision to act or settle on one of two or more unproved or unprovable propositions. Plugging that definition into Frege's puzzle ought to dissolve it. I don't know why philosophers cash it out in terms of truth and falsity. This always seemed perverse to me. Perhaps it's Plato's fault.

John Jones said...

A belief is the decision to act or settle on one of two or more unproved or unprovable propositions. Plugging that definition into Frege's puzzle ought to dissolve it. I don't know why philosophers cash it out in terms of truth and falsity. This always seemed perverse to me. Perhaps it's Plato's fault.