I’ve been reading Sam Harris’s Letter to a Christian Nation, which has been in the news quite a bit recently. Early in the book, Harris puts forward an argument to the effect that Christians are inconsistent in claiming, on the one hand, that they have good reasons for holding Christianity while believing, on the other, that Muslims do not have good reasons for believing Islam. Maintaining consistency, concludes Harris, requires Christians to reject Christianity. I thought it was an interesting argument. So, I’ve been thinking about a possible response, and thought I would share what I’ve come up with so far.
As I understand it, Harris’s argument, when boiled down to its essentials, can be put thus:
(1) One ought to hold a belief only if one has reason to hold that belief.
(2) The set of reasons for believing Christianity is identical [in a sense described below] to the set of reasons for believing Islam.
(3) The belief-content entailed by Christianity is inconsistent with content entailed by Islam.
(4) If (1), (2), and (3), then any bias toward Christianity or Islam is unwarranted, in which case one has insufficient reason to believe either Christianity or Islam.
(5) Therefore, one has insufficient reason to believe Christianity or Islam, and thus insufficient reason to believe Christianity.
Statement (2) is the obvious point of contention. (For the sake of argument, I shall assume all the other statements are true.) Harris, addressing the Christian, supports what (2) expresses in the following paragraph:
Every devout Muslim has the same reasons for being a Muslim that you have for being a Christian. And yet you do not find their reasons compelling. The Koran repeatedly declares that it is the perfect word of the creator of the universe. Muslims believe this as fully as you believe the Bible’s account of itself. There is a vast literature describing the life of Muhammad that, from the point of view of Islam, proves that he was the most recent Prophet of God. Muhammad also assures his followers that Jesus was not divine (Koran 5:71-75; 19:30-3
and that anyone who believes otherwise will spend eternity in hell. Muslims are certain that Muhammad’s opinion on this subject, as on all others, is infallible. (Harris 2006, 6)
Now, let us grant that, with the exception of the first statement, all the other statements in the quoted paragraph are true. In fact, let’s go further:
Suppose that for every belief C holds in support of her Christianity, M has a parallel belief supporting her Muslim faith. Two supporting beliefs are, in a rough and ready way, parallel if they are of the same schema where substitutions can occur on the order of ‘Koran’ for ‘Bible’, ‘Allah’ for ‘God [the Father of Jesus Christ]’, and so on. For example, if C has the belief, “Christianity is true because the Bible says it is true,” then M has the belief, “Islam is true because the Koran says it is true.” For every faith-supporting belief C has, M has the parallel supporting belief. (Of course, the beliefs that comprise the two faiths are not the same, otherwise there would be no difference between the faiths other than names. But I am concerned here not with the doctrinal content of the faiths themselves, but only with faith-supporting beliefs.) Let us call this the Parallel Supporting-belief Scenario.
As I see it, Harris’s argument for the truth of (2) requires the truth of the following conditional: If the Parallel Supporting-belief Scenario obtains, then neither C nor M has more reason than the other to believe her respective faith. My question is, does this conditional hold? I think the answer is ‘no’. For, a ‘yes’ would imply that reasons for holding beliefs must be other beliefs, which seems implausible.
To support my claim of implausibility, I draw from two distinct areas in philosophy: meta-ethics and philosophy of language. First, in meta-ethics many hold that the term ‘reason’ is ambiguous. For example, suppose Mort shoots Nate because the latter stole the former’s candy bar. The sheriff asks Mort what his reason was for shooting Nate, and Mort responds, “The reason is, Nate stole my candy bar!” “But that’s no reason to shoot Nate!” cries the sheriff. By the lights of many meta-ethicists, Mort is using ‘reason’ to pick out a motivational reason, while the sheriff is using ‘reason’ to refer to a normative reason. Mort’s action was unjustified because he failed to have a normative reason, although presumably he thought he had one. But what about the other direction: can one have a normative reason for believing that P, fail to believe he or she has that reason, and yet be justified in believing that P?
I think so. One reason(!) is why many think the causal theory of reference is true. Consider the following piece of fiction:
Russell is widely believed to have written Why I am not a Christian. In fact, however, McTaggart wrote Why I am not a Christian, and it turns out Russell actually stole the manuscript from him; then, Russell, a couple of years after McTaggart’s passing, published the book under his own name. Many have noticed the argumentation is not up to Russell’s usual standards, but no one suspects Russell of foul play.
Now, take the following two claims:
(6) Russell wrote Why I am not a Christian.
(7) ‘Russell’ in (6) refers to Russell.
If the fictitious story about Russell were true, then surely (6) would be false and (7) would be true. That is, in uttering (6), we would not accidently refer to McTaggart even if neither I nor any other living person were to know of Russell’s pilfering. What makes (7) true (and thus (6) false) is presumably a causal chain running from Russell to our use of ‘Russell’. Now, long before Donnellan and Kripke introduced causal theories of reference, we had reason for believing that names referred to the individuals who were so baptized. I, as a young lad, had reason to believe that I referred to George Washington when I said, “George Washington spilled his guts to his dad about the cherry tree,” even though the story is false and I said it long before I had heard of the causal theory of reference. It is not belief in a theory of reference that gives reason for believing that we refer to someone correctly, even when we make false claims about him or her; what provides the reason is the causal chain running from the named individual to the utterance of the name, whether or not we know about the causal chain.
Now, take the event, central to the Christian faith, of the Resurrection (my use of a Christian event is unimportant; I could have used an event central to the Muslim faith). Suppose the Resurrection occurred and that the event was baptized, as it were, ‘the Resurrection’ by those whom Mary and Martha saw at the empty tomb. Now, if there is a causal chain running from the baptism to me and the chain results (in some sense) in my saying ‘the Resurrection’, I have reason to believe that I am referring to the Resurrection. This is so, even if I have no belief about theories of reference or about the causal chain linking me to the event.
But if I have reason to believe that I am referring to a certain event, even if I do not know of the reason, then I have reason to believe the event occurred, even if I do not know of that particular reason. To see this, notice, first, that my claim does not fail to take into account the de re/de dicto distinction: the reason for the utterance just is the causal chain linking the utterance and the event, all of which are on the de re side of the distinction. Second, notice that I do not have a similar reason to believe in Santa Claus’s landing on my roof last Christmas. For, there is no causal chain linking Santa’s landing and my reference to it. The sort of reason I am describing here for believing an event occurred will obtain only when the event actually occurs.
So, in short, the soundness of Harris’s argument relies the truth of this conditional: if two people having all the same supporting belief-schema for their respective faiths and the faiths are inconsistent with each other, then neither has more reason than the other to believe her faith. I am arguing, to the contrary, that together the meta-ethical distinction between kinds of reasons and the support for the causal theory of reference which I described indicate that some reasons for believing a proposition are not included among a cognizer’s beliefs. Clearly, the argument I am offering does nothing at all to show that Christianity, Islam, or any other faith is true or even that one in fact has reason to believe any one faith over another. My argument, if it is sound, shows only the falsity of a conditional crucial to Harris’s argument.
- Joseph Long
You suppose that the reasons Harris points to are other beliefs, but why suppose that, especially if you think that reasons supporting other need not be other beliefs.
It seems to me that Harris is not talking about beliefs but about self-autheticated declartions: “The Koran repeatedly declares that it is the perfect word of the creator of the universe.” And presumably the bibe says something similar. So declartive evidence is the same, let’s suppose.
Of course your point comes back: if one set of those declartions are true and the other false then the true ones give us more reason to believe them. I accept that and it is an important point. But such an observation does not help us in deciding to believe precisely because we cannot discern the true declarations from the false. So as a practical matter there seems to be nothing to favour being Christian over Muslim.
Lee,
Thanks for the comment. Your point regarding the practical matter of choosing between faiths is right. Indeed, the purpose of the argument is not to show that any particular faith is true or is more justified than the others. Clearly this argument does not do that.
In order to show that one faith is true, more justified, etc., independent arguments would need to be advanced. But until all the arguments are in, as it were, I don’t think we can rightly suppose that all the declarative evidence is the same, for there might be declarative evidence waiting to be unearthed or amassed and put together so that proper comparisons can be made. And, of course, to concede that all the declarative evidence is the same would be, for the believer of either faith, tantamount to giving up the game. So, I don’t yet want to say that all the declarative evidence is the same. Having said that, however, my limiting the argument to beliefs might not be a completely accurate rendering of Harris, but if neither mere belief nor the totality of declarative evidence is quite right, I’m not sure how else to put it. Perhaps I should say something like “all the declarative evidence two individuals have at hand is the same.”
The purpose of the this argument is really just to show that one often has reasons for believing that P even if the believer herself is unaware of those reasons. The importance of this argument is that, if it is sound, it undermines the oft heard claim that one has no reason to hold to her faith since there are many faiths that are inconsistent with each other and whose members all firmly believe their own faith is the true one. If my argument undermines that claim, then I think we’ve made at least some progress in the debate–even if it is only a little bit.
Joseph,
As I suspected, that was a very interesting post. However, I do have some questions. I did not have time to read the other comment so forgive me if anything I bring up is redundant. You mentioned the causal history of each set of beliefs, that each belief goes back to a specific event. Does each person’s causal history of belief really go back to the event in question? The claim that it does seems too difficult to prove. If the person did not witness the event how can they know it happened? Especially when we are talking about religious miracles for which there are many reasons to edit and change the records of events. If that happened then the belief goes back to an edit made in a book and not to an event. Given that so many changes were made/could have been made, it seems too difficult to pin point when the cause of the belief really happened. Also, if you are making a causal chain that goes on that far back in history then the belief in question is probably strongly related to many more things then just the event in question. Everything will be related to everything else. In such a case all you have is a tautology.
Not only is the chain from belief to event to difficult to prove, but it may not really matter. After all, how do we know anything in history really happened as a book reports? Moreover, I think that a person’s belief formation is likely to take place just in the course of his/her life. Yes, a belief can go back to a certain event but that belief is not sufficient in producing the belief (although it would be necessary). When looking at how/why people come to believe what they do, then you have to look at the influences of their lives. There are many reasons why people belief things, but when looking at religion one of the most important factors is culture. Culture is why (the reason) many people are so committed to one faith or another. Along those same lines, people who were raised by there parents to believe X and in turn have belief X continuously reaffirmed by their society. Naturally he/she will likely carry that believe into adulthood. A reason why people continue to believe one thing over another is that they believe their belief happens to be under attack. There are probably many more reasons where the beliefs are causally connected in the context of the believing agent’s life. Based off of what I read in your post about what Harris said, I think that Muslims and Christians do have a lot of the same reasons for believing what they do, if you look at the reasons in their life for believing. What do you think?
I’ve been thinking about Harris’ argument, and I’m starting to think that the real slippery part is (4) — I don’t see that it really follows from (1), (2) and (3), in that I can hold (1) - (3) to be true and yet reject (4).
It’s possible to interpret the kinds of reasons in (2) to be more than just beliefs (or self-authenticating declarations): perhaps forms of life, or cultural embeddedness, or a psychological need to belong, or a desire to honor the traditions of one’s parents, and so on. These are obviously straight out of William James and The Varieties of Religious Experience. Accepting this requires us to swallow a bit of pragmatism, and to treat religious belief as having a different kind of grounding in reasons than other kinds of belief (scientific, empirical, etc.) but it is, I think, more true to the psychological facts of human beings than Harris’ assumption that the only reason to accept the Koran is Muhammad’s putative infallibility.
If we accept this, then (2) can be true — along with (1) and (3) — yet (4) be false. (4) tries to draw a contradiction from (2) and (3). But (2) is in terms of reasons whereas (3) is in terms of belief-content. There is no obvious contradiction there. It seems possible for a Christian (I’ll follow you and call her C) and a Muslim (named M) to be in exactly parallel situations — such that not only is the is the Parallel Supporting-Belief Scenario true, but the Parallel Supporting-Reason Scenario is also true. So for every belief that C has, her counterpart M has the parallel belief, and for every reason for belief that C has, M has the parallel reason for belief.
(4) seems to imply that if this is so, C and M stand in the exact same relation to Islam and to Christianity, and so to privilege one over the other is a violation to (1) — they profess belief in Christianity (or Islam) but they have no reason to hold that belief. But (2) is true only for specific individuals in a particular context. So while C and M are in parallel situation vis-à-vis their own religions, they are not in parallel situations to both religions simultaneously. Considering their contexts, they have a reason to choose one over the other.
Hi, Amy,
Thank you for writing and for you comments. If I understand you, the argument in your first paragraph can be boiled down to this:
(
If Joe, then S has a reason to believe P only if there is a causal chain running from the correspondent event to P.
(9) If S has a reason to believe P only if there is a causal chain running from the correspondent event to P, then we can discover the reason-providing causal chain running from the correspondent event to P.
(10) Since (10.a) there are reasons to change the historical documents, (10.b) the causal chain must run through those documents, and (10.c) we cannot prove that such changes were not made, we cannot discover the reason-providing causal chain running from the correspondent event to P.
(11) Therefore, not Joe.
Premise (9) is not explicitly stated, but it seems implicit and necessary to your argument (again, if I understand you). It is this which I will deny: I don’t think having a reason to believe P requires our discovering (knowing, etc.) the reason or even being able to discover (know, etc.) the reason for believing P. The causal theory of reference bit in the post is supposed to show this: even when make false claims about George Washington and even though I cannot trace the causal chain back to Washington’s being baptized ‘George Washington’, I have reason to believe I’m referring to George Washington.
I’m a bit confused by the tautology remark. Perhaps you mean this: At best I can show that for every true belief, S has a reason to hold that belief only if there is a causal chain running from the correspondent event to the belief in question; but, you say (if I understand you), that is insufficient to show that for some *particular* belief (a belief in a particular miracle, say) there is *in fact* causal chain running from from that particular event to a corresponding belief, and thus it is insufficient to show that for a particular belief S has a reason to hold that belief. If that’s what you mean, then I certainly agree: I cannot derive the antecedent of a conditional from a conditional alone.
I take this to be the thrust of your second paragraph: a chain leading from only an event to a corresponding belief is insufficient to cause the belief. Indeed, I would agree. I do not at all mean to argue that it is sufficient for producing a belief; I don’t think it is sufficient even for justifying a belief. All I mean to show is that the causal chain between the belief and the correspondent event would provide an additional reason for believing the event occurred. Thus, contrary to Harris (as I understand him), even if C and M have all the same belief-schema supporting their respective faiths, one of the two might nonetheless have more reason to believe her faith than the other has for holding hers. I make no claim regarding who actually has more reason.
Casey,
Thanks for writing. I think you make a good point: If C had never heard of any other faith, then whether or not C has all the same supporting beliefs and all the same reasons as every other faith, then C might indeed have sufficient reason to believe Christianity, even if Christianity is inconsistent with the other faiths. Moreover, as you point out, if whether or not to believe Christianity were a live, forced, and momentous decision (in James’s terminology), then, by James’s lights, it seems C might have even more reason for believing Christianity.
I don’t think this move will work against Harris, however. Says Harris: “Every devout Muslim has the same reasons for being a Muslim that you have for being a Christian. And yet you do not find their reasons compelling.” From this, it seems Harris is assuming that, in terms of the Parallel Supporting-belief Scenario, C not only knows that M’s faith is inconsistent with her own, she also knows that the Parallel Supporting-belief Scenario has obtained, and even more importantly he (Harris) is assuming that reasons themselves are rational, rather than passional. Given these assumptions, C is in the proper context for (4) to hold.
Hey Joe; just a few quick points.
First, if Harris is assuming that rational reasons and passional reasons are mutually exclusive, that’s where he’s making his mistake — and it’s a mistake made by many of the New Atheists (Dawkins, PZ Meyers, etc.). The new trend in psychology research is showing that emotions are crucial to good reasoning, so if Harris’ argument requires that assumption I reject his thesis out-of-hand.
Second, it’s not (4) that I think is contextual, it’s (2). This is important because in this case C and M can both be aware that the Parallel Supporting-Belief scenario holds, yet they aren’t required to accept Harris’ pessimistic conclusion that they are irrational to believe one thing over another. If (2) is the correct rendering of Harris’ premise (and, give the paragraph you’ve quoted, it certainly seems to be) it’s ambiguous between the set of reasons being equivalent from a neutral vantage point and the set of reasons being equivalent for each individual involved — that is, Harris is equivocating between different senses of ’same’ when he says “Every devout Muslim has the same reasons for being a Muslim that you have for being a Christian.” Obviously they aren’t the ’same’ as in identical, while they might be the ’same’ as in equivalent.
Let’s imagine that we can label all the elements in the set of reason-schemas: (R1x, R2x, R3x,…). So for every reason in the set C has (R1c, R2c, R3c,…) and M has (R1m, R2m, R3m,…). So (2) is true in the sense that for every instance of an arbitrary reason-schema Rnx, C has Rnc and M has Rnm, and so the sets of reasons are equivalent — they have the appropriate one-to-one correspondence. But (2) isn’t true in the sense that there is any specific reason that both C and M have in common as elements of the sets of reasons; the intersection of the sets is empty.
I’ll call the set of reasons for C, RC; the set of reasons for M, RM; and the set of reason-schemas, R#. Suppose the set RC is sufficient to believe in Christianity, in the sense of (1), and likewise that RM is sufficient to believe in Islam. From here I think the rest of my argument still works: (1), (2) and (3) can all be true but (4) still be false. (1) and (3) are relatively uncontroversial; we’ll assume they are true. Let’s assume that (2) is true in the substitution-from-schema sense but not the strict-identity sense. When would (4) be true? When would it be unwarranted to have a bias in favor of Christianity over Islam?
(4) would be true if RC was sufficient to believe in Christianity and equally sufficient to believe in Islam. But that doesn’t follow from the identity-through-substitution at all. (4) might be true if R# was sufficient to believe in either Christianity and Islam, but R#, being a set of schema, probably doesn’t give any reason to believe in anything. (4) would also be true if (2) were true in the strict-identity sense — but it’s not, by our original assumptions. This is where I think Harris falls prey to equivocation.
I think, in the big picture, what Harris is getting at is this: C believes in Christianity because she has the reasons in set RC; C also believes that M has reasons RM to believe in Islam. C is rational in believing in Christianity because — via (1) and RC — C has good reasons to so believe. But C must also believe that M is equally rational in her belief since M has exactly identical good reasons to believe in Islam. So if Christianity and Islam are equally rational there is no warranted reason to choose one over the other. Or: Harris assumes that there is a second-order irrationality in C thinking that M is both rational and wrong simultaneously.
But neither of these conclusions are valid. C can think that Islam is both rational (as in properly supported by reasons) yet wrong (as in the supporting reasons are false); therefore C can think that Christianity and Islam are both rational and also have a reason to choose one over the other.
So where Harris was shooting for an argument that proves any religious belief is unwarranted, the most he can prove is that one can’t hold other religions to be irrational when the situation described by (2) holds — at most, one can hold that they are wrong. Ultimately, I think that’s an OK result. ‘Wrong’ and ‘irrational’ aren’t coextensive. The problem — as your argument points out — is that it is difficult to know whether the situation described by (2) does actually hold. But if my reading of (your presentation of) Harris’ argument is right, someone can think that — at least as far as (2) is concerned — the declarative evidence is the same (via substitution from schemas), but still be rational in choosing one religion over another.
Of course, since this allows beliefs to act as reasons for other beliefs, I might be subject to an infinite regress problem. But at that point I’ll just bring in James again.
Not to be simplistic but it seems premises
(2) The set of reasons for believing Christianity is identical [in a sense described below] to the set of reasons for believing Islam.
and
(3) The belief-content entailed by Christianity is inconsistent with content entailed by Islam.
are inconsistent.
The simple reason is that if both the set of Christian beliefs and the set of Muslim beliefs are identical (there are no members of set C which are not also members of set M and the converse is also true) then it seems that it is impossible for these two identical sets to be inconstant. The real-world practical problem of Christians and Muslims killing each other is then just the result of misnomer.
If, instead, by identical we mean that God can be replaced with Allah etc, the two sets are only different via translation and once again cannot be inconstant. This would be like saying that there is a possible world in which the sentances “donde esta el bano?” and “where is the bathroom?” can be logically inconstant.
Perhaps then the point that I am missing is that what we are discussing is that despite the differences between the “belief-content” of these two faiths, the reasons for belief remain identical on a meta-level regardless of the belief-content. My simple and pedantic reply to this is that if the belief-content is sufficiently different enough to be inconstant, and the differences are more than simple translation issues, then it is highly likely that the reasons for belief will not be identical.
In other words, if what Mort believes in is different from what Nate believes in, by virtue of the content itself, it seems highly likely that their reasons for belief also differ—if only because part of Mort’s reason for belief is belief content A and Nate’s reason for belief is belief content B.
My intuition is that Harris’s aim is to reduce the reason for belief of the Christian and Muslim to something similar to “the Bible tells me so” and the “Koran tells me so.” Ignoring the fact that the difference between these two texts in not solely a translation issue, the schema itself as a description of why most people believe is fallacious. Although academics tend to want to reduce the average person’s belief to it, it is likely not the case. A more accurate schema would be something akin to “I believe because of what the Bible says” or “I believe because of what the Koran says.” In which cases these schemes, by virtue of the content, are not identical and choosing to believe one over the other is then not inconsistent.
[...] Long, an FSU Philosophy grad student, critiques Sam Harris’ argument (found in Letter to a Christian Nation) that the symmetry between religions means there is no [...]
very interesting, but I don’t agree with you
Idetrorce