Saturday, July 23, 2011

The Outsider Test for Faith

I've entered the fray (jparris8) over at Debunking Christianity, and as those familiar with John Loftus and the blog know, there is no avoiding discussion of `The Outsider Test for Faith' (OTF). I had heard of this before while watching the threads at Victor Reppert's blog, and as I will inevitably have to discuss it again and again, I will save myself and others misery by collecting my thoughts on the OTF in this post.

I will begin by discussing what Loftus describes as the most comprehensive explanation of the OTF on the web.
The question I’ll be addressing today is whether we should adopt a believing or a skeptical predisposition prior to examining the evidence for a religious set of beliefs. I’ll argue that a skeptical predisposition is the preferred one to adopt.
He did say `prior', so I have to say something Bayesian: one can leave aside entirely the question of prior absolute probabilities and focus on a cumulative Bayes factor, i.e. the number by which your old odds ratio on Christianity is multiplied to yield your new odds ratio on Christianity. Granted, this is a very big task, but once done, you can say what initial commitment to Christianity would be required to have confirmed it as `more likely true than not' or to whatever value you feel significant. That analysis would necessarily include evidence concerning contradictory religious beliefs.

But I have a nagging feeling that Loftus wants to do more with the OTF than the above generalization suggests, just as he obviously wants it to be more than a handy thought experiment.
There is overwhelming, undeniable and non-controversial evidence for the test itself that can be found in the sociological, anthropological, and psychological data. I’ll start with some of this data that forms the basis for the test.
Here I have an issue: he is claiming `undeniable' evidence for a norm. As good is/ought distinguishing Humeans, we should be looking out for the goals or desiderata - presumably themselves uncontroversial and undeniable to theists - by which this evidence translates into support for the OTF.

What might this be? In comments and posts, Loftus is ever touting this:
The basis for the outsider test has been stated adequately by liberal Christian philosopher John Hick: “It is evident that in some ninety-nine percent of the cases the religion which an individual professes and to which he or she adheres depends upon the accidents of birth.” That is to say, if we were born in Saudi Arabia, we would be Sunni Muslims right now. If we were born in Iran, we’d be Shi’a Muslims. If we were born in India, we’d be a Hindus. If we were born in Japan, we’d be Shintoists. If we were born in Mongolia, we’d be Buddhists. If we were born in the first century BCE in Israel, we’d adhere to the Jewish faith at that time, and if we were born in Europe in 1000 CE, we’d be Roman Catholics. For the first nine hundred years we would’ve believed in the ransom theory of Jesus’ atonement. As Christians during the later Middle Ages, we wouldn’t have seen anything wrong with killing witches, torturing heretics, and conquering Jerusalem from the “infidels” in the Crusades. These things are as close to being undeniable facts as we can get in the sociological world.
Ok, this is all fine and well (call it and other details the fact of context-dependence): but why does this imply the OTF? I suppose it will have to wait:
Michael Shermer, a former Christian turned atheist, has done an extensive study of why people believe in God and in “weird things.” He argues: “Most of us most of the time come to our beliefs for a variety of reasons having little to do with empirical evidence and logical reasoning. Rather, such variables as genetic predispositions, parental predilections, sibling influences, peer pressures, educational experiences, and life impressions all shape the personality preferences and emotional inclinations that, in conjunction with numerous social and cultural influences, lead us to make certain belief choices. Rarely do any of us sit down before a table of facts, weigh them pro and con, and choose the most logical and rational belief, regardless of what we previously believed. Instead, the facts of the world come to us through the colored filters of the theories, hypotheses, hunches, biases, and prejudices we have accumulated through our lifetime. We then sort through the body of data and select those most confirming what we already believe, and ignore or rationalize away those that are disconfirming. All of us do this, of course, but smart people are better at it.”
Ok, so there are reductive accounts of religious tendencies. Ok, why does this imply the OTF? Damn, I still have to wait:
This whole inside/outside perspective is quite a dilemma and prompts me to propose and argue on behalf of the OTF, the result of which makes the presumption of skepticism the preferred stance when approaching any religious faith, especially one’s own. The outsider test is simply a challenge to test one’s own religious faith with the presumption of skepticism, as an outsider. It calls upon believers to "Test or examine your religious beliefs as if you were outsiders with the same presumption of skepticism you use to test or examine other religious beliefs." Its presumption is that when examining any set of religious beliefs skepticism is warranted, since the odds are good that the particular set of religious beliefs you have adopted is wrong.
Ok, now tell me: why do the apparent cultural dependency of religion and the at least partial success of reductive accounts in explaining common tendencies in religion suggest the OTF?
The OTF is no different than the prince in the Cinderella story who must question forty-five thousand girls to see which one lost the glass slipper at the ball last night. They all claim to have done so. Therefore, skepticism is definitely warranted. This is especially the case when an empirical foot match cannot be had.
Ok, warranted. Not a big deal, what's important is the demand that theists apply the OTF.
The amount of skepticism warranted depends on the number of rational people who disagree, whether the people who disagree are separated into distinct geographical locations, the nature of those beliefs, how they originated, how they were personally adopted in the first place, and the kinds of evidence that can possibly be used to decide between them. My claim is that when it comes to religious beliefs a high degree of skepticism is warranted because of these factors.
These can be factors, but they are not decisive. Anyways, TELL ME ALREADY.
If she refuses to [apply the OTF] then she must justify having such a double standard. Why does she test other religious beliefs differently than her own? For someone to object that what I’m asking is unfair, she has the burden of proof to show why her inconsistent approach to religious faith is justified in the first place.
Pardon me, but assessing `other beliefs' as `outsiders' and our own as `insiders' applies to everyone. About anything. Forget religion. That's not `double standards', that's the necessary product of having beliefs. A `double standard' with respect to the evidence and argument, if it should emerge, should emerge in the normal way... in the course of normal argument. The OTF adds nothing to this.

I suspect I'm not going to get anywhere on the is/ought gap today. And I see I'm not going to get anywhere on the `X is warranted/X is binding on all reasonable people' gap, either. Oh well, I can do other things:
Nonetheless, if after having investigated your religious faith with the presumption of skepticism it passes intellectual muster, then you can have your religious faith. It’s that simple. If not, abandon it like I did. I suspect that if someone is willing to take the challenge of the outsider test, then her religious faith will be found defective and she will abandon it along with all other religious faiths, like it has me.
See, I'm not sure that a positive or negative result for the OTF translates into any probability threshold, and by that virtue it need not translate into any final stance on one's beliefs. This ignores completely the problem of the priors. It ignores completely the incompleteness of evidential calibration that we have. It assumes from the outset that believers should adopt for-the-sake-of-examination probabilities about skeptical alternatives - yes, positive claims - and apply them to religious claims, alternatives which are presumably non-binding to adopt in the first place. Else, why insist on the OTF? The OTF would merely be restating an undeniable state of the evidence, not a norm mysteriously implied by the existence of other religions. So then what happens when believers adopt their former commitments, something which is apparently not intrinsically unreasonable to do?

A believer who failed the outsider test would conclude, and need only conclude, that skeptics of her faith can be reasonable. It would not entail that her beliefs are unwarranted and incapable of reasonable commitment. At all. And I'm not sure what the point of the OTF is if not that.

I now move on to some other objections. As far as I can tell, none of his responses to potential objections substantively address of mine, but he does say interesting things in his response to one objection:
After all, someone can be right if for no other reason than that she just got lucky to be born when and where she did. But how do you rationally justify such luck? This is why I’ve developed the challenge of the outsider test in the first place, to test religious faiths against such luck.
Here, Loftus is definitely saying something about prior distributions. Namely, that believers must equivocate across contradictory religions in assigning prior probabilities, or that they should match the likelihood of the truth of their religion to the `chance' of their believing it.

First big problem: the demand is impossible if you believe that the number of possible religions and/or gods is uncountably infinite. No non-zero probability of each particular god/religion is possible if that number is (countably) infinite.

Second big problem: the initial facts Loftus describes may not be neutral with respect to some initial distribution. The fact that a religion has died out can affect its probability. The fact that Christianity is `a religion amongst others' does not entail that there are no important qualitative differences between Christianity and other religions which could lead someone to reasonably assign it a quite significant prior: we lack coherent descriptions of large numbers of extinct religions or sufficient detail to distinguish them from others apart from relabeling of terms. Of extant religions, many lack coherent, unified systematic treatments like that presented by Christian philosophers who have spent centuries dealing with rational argument.

Third big problem: if the initial confidence in a religion is to be determined by the fact that there are many other religions, and we should equivocate across them, we make very big commitments to features shared by most or all of the important contenders. In particular, the OTF would have us practically assume supernaturalism and some supernatural cosmogony, though any particular version may be initially implausible. I don't think Loftus should be so eager to demand the starting point he does. I think he's sneaking in the assumption that materialistic alternatives dominate this space. If that's the case, we are more than past the importance of the number of alternatives. Unless he's asking us to assume supernaturalism, this sure does look, smell, and taste like an unjust demand that believers assume naturalism.

I could say more. Lots more. (Must personal, subjective background experience count for nothing?) But I really don't see the point, since Loftus never presents a sound, valid argument. Since the conclusion of his argument is that supernaturalism should be part of the outside position, I suppose I can be happy about that.

There is a way of preserving the legitimate core of the OTF without lapsing into absurdities and unreasonable dictations. It's called normal argument. Inconsistent treatments of evidence by believers, where important, will show up there. I do not see how the OTF does anything more than to prolong outstanding confusions. I would even say that it does more harm than good to the rational faculties of those who are introduced to it.

Bonus fun-time exercise: What would Calvin say about all this?

Edit 7/25/11: I cleaned up some of the sentences, added a few phrases, and added the last paragraph. I also want to add another observation:

If we should set our prior odds on Christianity to match those of our being Christians - as calibrated by sociological data - we get a very bad result for skeptics of Christianity. What, for example, should be our prior odds on the Resurrection before arguing over the evidence? Normally, and I think correctly, we calibrate this with respect to the well-documented tendency of dead people to remain dead. But what happens if we set our prior odds to the frequency of belief in the Resurrection? Since Christians argue that the Resurrection was an exceptional event, the fact that we normally observe dead people remaining dead would do nothing to reduce these odds before engaging the texts. If we take as a low estimate that Resurrection-believers account for 10% of the population, then our prior odds on the Resurrection will be absurdly high. I think that Christians can very easily overcome such a prior improbability with even a skeptical account of the New Testament. Yes, I think that Christians can easily pass the OTF as presented, at least with respect to the Resurrection.

7 comments:

  1. "A believer who failed the outsider test would conclude, and need only conclude, that skeptics of her faith can be reasonable. It would not entail that her beliefs are unwarranted and incapable of reasonable commitment. At all. And I'm not sure what the point of the OTF is if not that."

    Really? Doesn't it seem obvious that, what with all the competing religious claims, a naturalistic explanation makes more sense than a theistic one? A crucial attribute of the OTF, it seems to me, isn't just that it gets the theist to question her faith in particular, but to question the basis for theistic belief in general as well.

    Think of one million Mars rovers. Each NASA analyst is allowed to independently test the signal sent by the other Mars rovers, but not her own. All of the other Mars Rovers are discovered to have a faulty water detector that signals that water is present on Mars. The NASA analyst sees that all other reports on all other Rovers are faulty. So why should she trust that her Rover has detected water?

    ReplyDelete
  2. My first non-spam comment! *Happiness*

    "Doesn't it seem obvious that, what with all the competing religious claims, a naturalistic explanation makes more sense than a theistic one?"

    Not really, as I will soon explain.

    "A crucial attribute of the OTF, it seems to me, isn't just that it gets the theist to question her faith in particular, but to question the basis for theistic belief in general as well."

    That would be an excellent line of argument; were the OTF presented as a normal argument instead of a demand about prior odds or whatever else the OTF is supposed to be, this would have been a completely different post. More on this in a moment, but first:

    As I stated in the final paragraph, the usually minor differences between the various sects of Christianity do not delete from memory where they agree or approximately agree. If Joe and Tom disagree about the Trinity, they may still agree about the Resurrection. If Joe and Tom were our entire population, Loftus seems to be suggesting something like the following: ascribe probability ~0.5 to each stance on the Trinity and probability ~1 to the Resurrection. If he couches his arguments in terms of `chance of believing X', that is the sort of conclusion we find.

    Now let's look at a normal argument and see what it looks like. Your Mars Rover example is perfect: what we are doing there is moving from some prior confidence in the reliability of the water detection systems installed in the rovers to some posterior confidence - which is much lower - based on the faulty findings. We notice the correlation and consequently infer that the failures are more likely systematic to the design of the detector, as opposed to having a million coincidental failures. The analyst might or might not end up trusting her Rover; whether or not she does will depend on other data affecting her prior confidence.

    For example: she could have designed her own water detection system while the other systems used parts from a common source. She may infer that the problems with the other rovers trace back to that source, while her source is more reliable. She might plausible grounds for suspecting in advance that the other rovers would use flawed systems, hence why she built her own.

    It's also quite possible that she does not trust her rover, even if she built her own system. She may have grounds for thinking that interfering elements in the atmosphere and soil are responsible.

    In other words, facts specific to the analyst - other than that she is an analyst - must be addressed before locking in priors. And to do so in a convincing manner, we engage in the rather broad activity of normal argument.

    Let me know if that helps.

    ReplyDelete
  3. Hi Jesse
    I will comment on your OTF take here on your site rather than Reppert's. Mainly because I need some time to think about your commentary.

    I will also include your last comment to me in the mix.

    Cheers

    ReplyDelete
  4. I also note Tony Hoffman commenting here. He has a good head for reasoning.

    Cheers

    ReplyDelete
  5. JP: " That would be an excellent line of argument; were the OTF presented as a normal argument instead of a demand about prior odds or whatever else the OTF is supposed to be, this would have been a completely different post."

    Since you seem to agree that the OTF is somewhat amorphous (is it an argument? a challenge? a process? I suppose parts of it could be used to form all three.) I am not sure that you're not being charitable in your criticism of the OTF. Sure, the OTF fails as a deductive logical argument that, say, proves that one's theistic belief is false, but it seems uncharitable to insist that the OTF must meet that standard to be coherent. It seems that if it works, by and large, it would be coherent.

    JP: " As I stated in the final paragraph, the usually minor differences between the various sects of Christianity do not delete from memory where they agree or approximately agree. If Joe and Tom disagree about the Trinity, they may still agree about the Resurrection. If Joe and Tom were our entire population, Loftus seems to be suggesting something like the following: ascribe probability ~0.5 to each stance on the Trinity and probability ~1 to the Resurrection. If he couches his arguments in terms of `chance of believing X', that is the sort of conclusion we find."

    I'm not sure why this is an issue. Yes, current sects of Christianity largely agree about swathes of Christian dogma. But the OTF applies as well to previous generations and all other religions (known and unknown), and it seems fairly easy, even using a conservative heuristic, to come up with a multitude of divergent theistic beliefs. I'd suggest that divergence occurs along a fault line where adherents feel compelled to proselytize -- this would include everything from the most extreme (violent) methods, to the most gently persuasive -- say, door-to-door proselytizing or asking a spouse to convert, etc. Using this standard, we are still looking at a highly segmented population.

    JP: "For example: she could have designed her own water detection system while the other systems used parts from a common source. She may infer that the problems with the other rovers trace back to that source, while her source is more reliable. She might [have] plausible grounds for suspecting in advance that the other rovers would use flawed systems, hence why she built her own."

    JP: " In other words, facts specific to the analyst - other than that she is an analyst - must be addressed before locking in priors. And to do so in a convincing manner, we engage in the rather broad activity of normal argument."

    Well, I meant for my analogy to be reasonably tight, in that I wasn't granting my analyst any different equipment on her Mars Rover than was on the other Mars Rovers. But I agree that she is free to make an argument for a process or technique that demonstrates that her Rover has found water, knowing that the overwhelming evidence is that the water detectors on the Mars Rovers are faulty.

    ReplyDelete
  6. "Since you seem to agree that the OTF is somewhat amorphous (is it an argument? a challenge? a process? I suppose parts of it could be used to form all three.) I am not sure that you're not being charitable in your criticism of the OTF. Sure, the OTF fails as a deductive logical argument that, say, proves that one's theistic belief is false, but it seems uncharitable to insist that the OTF must meet that standard to be coherent. It seems that if it works, by and large, it would be coherent."

    I'm not sure if you've seen it already, but I have a new post which spells the problems out in more detail. If it isn't deductive or probabilistic, and believers may rationally and reasonably reject it while accounting for its motivating observations, I think that the failure is complete.

    "I'm not sure why this is an issue."

    Ask yourself whether there is a better and more convincing way to calculate the prior odds on e.g. Resurrection that happens to be more favorable to skeptics.

    "Well, I meant for my analogy to be reasonably tight, in that I wasn't granting my analyst any different equipment on her Mars Rover than was on the other Mars Rovers. But I agree that she is free to make an argument for a process or technique that demonstrates that her Rover has found water, knowing that the overwhelming evidence is that the water detectors on the Mars Rovers are faulty."

    I know you weren't granting that. I should have been clearer: the point was to illustrate that specifics about X which do not follow from large numbers of beliefs/claims contradicting X must be considered before making a judgment. The observation of many bad water detections might or might not be important. In order to decide whether or not it is important, and if so how important, we have to do something other than insist on prior distribution, as I explained my previously referenced post.

    ReplyDelete