I’m attracted to a traditional account of the semantics of belief reports which holds that the semantic content of a belief report of the form ‘S believes that P’ is a two-place relation that holds between a person and a proposition.
I’m curious about other common phrases in English that (a) use the word ‘believe’, (b) appear to be something like a belief report, and (c) do not appear to fit the above schema.
These reports use ‘believes in’ and appear to express relations that obtain (if at all) between a concrete particular thing and another concrete particular thing.
Here are some examples.
(A) Virginia believes in Santa Claus
(B) Bob believes in God
(C) Dave believes in Jesus
(D) Andy believes in Tim
I we only had (A) and (B) to worry about, then we could reduce ‘believes in’ to ‘believes that’ as follows:
(BIBT) S believes in R =df.) S believes that R exists
But (C) and (D) pose trouble, right? Dave can believe that Jesus exists without believing in Jesus (in the sense that I think Christian theists talk about when they say they believe in Jesus)
An easier example would be when persons talk about believing in other ordinary people as in (D).
I’m wondering if there is a uniform treatment of ‘believes in’ locutions that captures (A)-(D) and is cashed out in terms of ‘believes that’.
Bleg: Anyone see discussions of this in the literature on the semantics of belief reports? I’d love a few references.
(Posted from my Android Phone)
UPDATE: [November 11, 2009]
Robbie Williams brought a recent paper by Zoltan Szabo to my attention on this topic called “Believing in Things”. Here’s the JSTOR link. A cursory glance at the paper suggests that Szabo argues that we should not analyze “believes-in” in terms of “believes-that” – should be an interesting read. I hope to get to it soon.
If someone asks a non-theist, “Do you believe in Jesus?”, the non-theist, it seems, will say something like this:
A: Well, I believe that Jesus exists(/existed), but I don’t believe that he was the savior/son of God/etc.
I’m not sure if similar things come up for other examples of ‘believes in …’; I can’t think of any right now.
This suggests that there may be something peculiar about the question, not because of ‘believes in’, but because of ‘Jesus’. ‘Believing in’ Jesus has just come to mean something like ‘believing that he exists, and that he did x, y, and z’. It would be interesting to see if this explanation is plausible for other potentially problematic uses of ‘believes in’.
As far as believing in other concrete particulars, I think cashing them out in terms of ‘believes that’ is correct.
(1) I believe in Emmy.
I suppose (1) is actually ambiguous. It could be analogous to (2):
(2) I believe in electrons.
Or it could mean something like ‘I believe that Emmy has the skills to get it done’. In either case, I think cashing things out in terms of ‘believes that’ works fine.
So to sum up: I think that ‘believes in’ can plausibly be cashed out in terms of ‘believes that’, and if a purely existential proposition doesn’t seem to capture everything we mean when we say ‘I believe in X’ (as in the Jesus example), I think there is just another proposition, like the proposition that X did such and such, which we also have to believe to make a ‘believes in’ report sound true. I don’t know if this last bit is pragmatic, or if it would enter into the semantics somehow, though.
I’m not exactly sure what to think about the semantics of all this, and I don’t know of any literature that would help, but here’s another case that seems problematic. The first result of my Google search of “believes in” was titled “Hans Zimmer believes in the legitimacy of games [as an art form]….” Here it’s not quite like (C) & (D) since it doesn’t involve being confident or trusting in someone’s abilities, etc. But it also doesn’t follow (A) & (B) in that (BIBT) doesn’t really apply, since it’s not about existence. It seems “Hans Zimmer believes in the legitimacy of games [as an art form]” means that Hans believes that games are a legitimate form of art. It might be translatable into “believes that” talk, but I’m not sure a single principle could generate the correct translation it and the other cases.
The Zimmer example might fit Justin’s more general model…e.g…
(H) Hans Zimmer believes in the legitimacy of games as an art form.
Might be
(H*) Hans Zimmer believes that games are a (legitimate) form of art.
The most natural way to capture Justin’s suggestion would be…
(BIBT*) S believes in R =df.) There is some proposition X (that has R as a constituent) and S believes that X.
Although, negative existentials fit this. I would qualify as believing in Obama, if I believed the proposition that Obama did not exist.
So, how about.
(BIBT**) S believes in R =df.) There is some proposition X that ascribes an existence entailing property to R and S believes that X.
Although, I’m worried about this because non-theists would qualify as believing in Jesus on this account.
What I’m worried about is that we’ll need to stick in some weird context sensitive features on a case-by-case basis.
But if we could get some general features to analyze ‘believe in’ that worked in all cases – that would be awesome.
Could beliefs in be justified, or become knowledge? We can say things like “Your belief in Jesus is justified”, but not, “You know in Jesus”. For knowledge, we definitely need a proposition there. But I think if your belief in Jesus is justified and true (and there’s no funny business), then that belief can turn into knowledge. So there has to be a proposition that you believe when you believe in Jesus.
So I wouldn’t be too surprised if the right semantics for ‘believes in’ had some sort of context-sensitivity built in, because we’re going to need to complete the proposition that you believe somehow (since ‘Jesus’ does not express a proposition by itself), and context seems like the right way to do this. And sure, what the context adds will be different from case-to-case, since the context will be different from case-to-case, but maybe we can come up with an underlying sort of function from context to what it contributes.
That’s a good way to motivate the claim that there is likely some reduction…”believe in” statements seem to be subject to the same sorts of normative constraints that “believe that” statements are.
I won’t be too worried about context filling in that proposition if we can come up with a nice, informative way to state what the function is.
Now the fun part is to try and figure out what that function is.
Interesting post and comments. But why not let apparent similarities stand between ‘believe-in’ and ‘believe that’ statements at a level of the semantics and then allow a reduced propositional correspondence? So that a proposition (or statement-like or clausal correspondence) just had whatever ‘architecture’ suited any type of quite minimal assertion.
Hi Matthew,
I’m not sure I follow. Is this a suggestion that we need not analyze believe-in in terms of believe-that? And instead we could, as you say, “let apparent similarities…” If the suggestion is to abandon analyzing one in terms of the other – here’s a reason not to abandon. We would simply have a cleaner picture of the mental if the properties/relations involved in the “believes-in” relation were reducible to the properties/relations involved in the “believes-that” relation.
I’m also not sure what the allowance of “reduced propositional correspondence” does. Can you explain that more with an example (e.g. by comparing a believe-in sentence and a believe-that sentence and indicate what this propositional correspondence is) – I just need to see an example of a particular instance of correspondence to get a sense of what this relation is that you’re talking about.
Following up on Justin’s most recent comment: I thought I’d get some more examples of believes-in sentences.
(A) Clark believes in Truth, Justice, and the American Way
(B) George believes in the Constitution (of the United States)
(C) Thomas believes in what the Constitution stands for.
I’ve heard instances of these kinds of uses of “believes in” before. Maybe they’ll help to articulate whatever that function is that Justin mentioned.
There’s evidence for the ambiguity hypothesis from VP ellipsis. “I believe in properties and Jesus” is only comfortably heard as an existence claim. Reading it in the “I believe that properties exist and I have faith in Jesus” (or whatever) way strikes me as zeugmatic and punny like “I run marathons and the largest banking system in Connecticut” is. It’s okay to posit ambiguity if there is evidence for it.
Chris,
That seems right to me. So suppose we disambiguate between two different senses of “believes in”
S believes-in1 R = df.) S believes that R exists
S believes in2 R =df.) [whatever the heck it is we mean in the other sense of “believes in”]
I’m curious to know if we can analyze that latter sense of “believes-in” in terms of “believes-that”. The ambiguity would be fleshed out in terms of the types of propositions ranged over. The former would be straight forwardly existential propositions. I suspect that the latter sense can be fleshed out in terms of believes-that, but I’m at a loss as to where to begin.
Answering your second question to me. Putting something in terms of a statement like: ‘I believe that there is a Santa’, rather than just believing-in, seems to me to be obviously offering oneself as in some relation to a sentence-like entity, at least the one following the that-clause, so rather like asserting, whereas believing-in might just be some kind of standing state or whatever (even one I might not affirm). I think your move to treat both things/events (say) on par (so that there would be some equivalence in terms of some relation … a relation to a believed-in sentence) overrides this difference, the difference between being in some relation to some sentence like entity and being in a state of some sort. Of course describing or assigning some content in the case of a belief-in might end up being just like having a belief-that, and then … and so-on.
I think the point you make in the first part of the response to me is important and I’m going to digest it, although actually I think you have me the wrong way around. I think there are specific contingencies associated with asserting, and in terms of knowledge claims made, which need not export. Going back to the Santa example then its statement would incur some kind of propositional obligations, which it has been possible to think of in terms of correspondence, as you know, and if the ‘believe-in’ ‘relation’ whatever it was inherited a comparable truth-conditionality then I think you might be inflating that relation. (I think the ‘relation’ talk comes from squeezing-out propositions or statements down the line.)
Here is a method for generating the relevant proposition:
1) Start with the [S believes in Y] construction.
2) Determine an equivalent “believes in” claim of the form “S believes in the X of Y”
3) Convert to “S believes that Y Xs”
For instance:
Dave believes in Jesus -> Dave believes in the divinity of Jesus -> Dave believes that Jesus is divine.
Andy believes in Tim -> Andy believes in [the ability to succeed of] Tim or Andy believes in [the honesty and virtue] of Tim -> Andy believes that Tim is able to succeed/Andy believes that Tim is honest and virtuous.
Lewis,
I like that. What do you say about things like
(A) Clark believes in Truth, Justice, and the American Way
(B) George believes in the Constitution (of the United States)
What are we saying there? Perhaps.
(A*) Clark believes in the pursuit-worthiness of Truth, Justice, and the American Way.
(A**) Clark believes that Truth, Justice, and the American way are worth of pursuit.
I wonder if we can extract something more informative by trying to find some feature that the Xs have in common.
If the Jesus one isn’t divinity but something like follow-worthiness, then maybe we have something that’s common across all of these.
Help me out here. So, focusing on The Content of a belief report, whatever falls between quotes—as in the examples A,B,C,D in Andrew’s original post—is just going to allow one to generate relevant propositions/contents as in the last two posts. But doesn’t having those contents just miss something about the point in having belief reports in the first place? That they are in some sense preservative? Or at least it doesn’t seem to respect an order or embedding in contents and some hierarchy of commitments determined by contents, given in a principled report. Isn’t there a separate problem in attempting to identify the contents of a report with some conditions pertaining to the generation of specific contents; those reported on? So that there might be something like literality in their restatement. (Andrew, I was resisting separating them out (so you might check out what I meant by ‘correspondence’) because I believe that what this could be about is some conventional ordering in belief reports which just does tend to treat believing-in (as reported-on) as equivalent to believing that—as a function of reporting-on. And so the minimalism that interests me is the one associated with the more minimal commitments traditionally associated with reporting.)
I’m not sure how to extend Lewis’s suggestion into an analysis. Here is my first attempt:
S believes in Y iff there is an X such that S believes in the X of Y
S Believes in the X of Y iff X believes that Y is X
But, this is too permissive. After all, I believe that Jesus was bearded. I guess I believe in the beardedness of Jesus (although property anti-realists might not). But I don’t believe in Jesus. There must be a restriction on the appropriate kinds of X’s. But, the restriction can’t be too strict. For example, it can’t be that that X must be a good thing. After all, the villain of a movie might believe in mankind in virtue of believing in the selfishness of mankind. But, clearly selfishness isn’t a good thing.
I guess we could go with a contextually salient X, but I would prefer to avoid doing that.
Joshua,
My inclination would be to suppose that there are only a handful of appropriate values for X, and that (broadly) contextual parameters determine which value is supplied.
For instance, if we just had cases like, “believes in ghosts” and “believes in Tim” to account for, we would have “existence” and “trust-worthiness” as possible values for X. Since it is routinely questioned whether ghosts exist, it is much easier to get the “believes in the existence of ghosts” reading. Suppose, however, that we all believed in ghosts, and were hoping ghosts (as a species(?)) would come through for us on something important. In that context, if I said, “I believe in ghosts, they always come through in a pinch” it would be easier to get the “have faith in” reading.
It seems like Andy’s cases break down into two group: straightforwardly existential and “have faith/confidence in” readings. If we could figure out the right X that works for all cases of “have faith in”, I’d think we were in good shape.
I’ll also note that the “believe in Jesus” case is not as clear-cut for me as the others, in the sense that I’m willing to concede that the non-christian can say, “I believe in Jesus, I just don’t think he was more than a human” or something like that.
Interesting update concerning the literature on this…Robbie Williams directed me to a paper by Zoltan Szabo on this in PPR 2003 called “Believing in Things”
http://www.jstor.org/stable/20140563