"Would you accept the testimony of a crazy man?" and other thoughts

It has been a busy few days, but at last I have some time to return to the bevy of comments on our ongoing epistemology debates. Given the sake of space limits, I shall have to limit myself here to AnAtheist.Net's counter-claim (in the "nailing jelly to walls" thread) that we only accept the testimony of others when we have evidence for their reliability.

My argument was concerned to defend the claim that a wide variety of our properly basic beliefs are innocent-until-proven guilty. Among the examples I provided are testimony beliefs, memory beliefs, and sense perception beliefs.

The point is simply this: if many of these beliefs are innocent-until-proven-guilty (that is, rationally held and if true then known, without evidence and in the absence of defeaters), then it may be that some religious beliefs are likewise innocent-until-proven-guilty. That is, a Christian could know "God loves me" or "Jesus is God" in a properly basic way, apart from evidence.

(How might they know it? Well for starters, many religious beliefs are held on the basis of testimony. Thus if testimony is generally properly basic absent defeaters, then so may religious beliefs accepted on the basis of testimony be properly basic absent defeaters.)

AnAtheist.Net responds by challenging my three proposed examples of rational (sans defeater) beliefs. When it comes to testimony beliefs I argued: "If you ask for directions on the street, do you first have to establish the reliability of the passerby?" (My view is that we do not first establish grounds for reliability. Instead, we trust the testimony of others.)

AAN disagrees with this analysis. Instead, he argues that we look for evidence of reliability prior to accepting testimony of others: "When asking for directions on the street one tries and locates people who appear to be locals and not obvious out of towners. So we are establishing at least a minimum baseline of reliability."

I think AAN's analysis of testimony is mistaken. Let's say that I am visiting London and find myself lost in Leicester Square, unsure how to get to a musical in the West End. Convinced that I need to ask somebody for directions I bypass the first person I see because he is flapping his arms up and down rapidly and shouting obscenities. Instead I ask the second individual who is quietly reading a newspaper on a park bench.

AAN would argue that I sought evidence both for the first and second persons. I don't think this is a proper analysis of what is going on. Rather, I argue that we accept testimony unless we have a defeater to the testifier's reliability. It is innocent until proven guilty. The first man's erratic behavior defeats his reliability as a witness but I simply have no reason to think the second man unreliable and so I am justified in believing him reliable (even knowing that he could be a compulsive liar ... or something worse).

AAN's analysis is thus not the most plausible way to approach testimony. It also creates problems for it leaves us either having to abstain from accepting testimony which it is surely rational to accept or it forces us to expand the notion of "evidence" so broadly that it becomes basically indistinguishable from my "granting the benefit of the doubt" attitude. 

Here's another example:

You are having problems with your computer. You call the help line and a person named "John" answers. He instructs you to download a program on the internet, assuring you that this will solve your problem.

Is it rational to accept John's testimony? After all, it is possible that he is planning to quit his job and that he intends to take down as many computers as possible by having them download a malicious virus. It is also possible that he is incompetent and has inappropriately advised you on this particular issue. Either way, taking his advice could thus cause your computer irreparable harm.

Keep in mind that you have no direct evidence that John is competent and of good character, for you don't know him from Adam. All you have is an anonymous voice at a call center which could be in Minneapolis or Mumbai or anywhere in between. So where does this leave you?

If AAN is correct then you need some evidence that John is trustworthy; otherwise you should refrain from downloading the program. And yet you are surely rational to take John's advice.

This leaves AAN with one option: argue that we accept John because we have evidence that he is reliable. Since we have no personal knowledge of John, AAN must switch to a broader inferential knowledge: e.g. people hired to provide tech support are generally reliable.

But is AAN really aware of any such evidence for the specific reliability of tech support? For instance, does he know of a study that shows statistically that support people are likely to be of good character and competent (and for this company in particular)?

Clearly no such evidence is required to accept John's testimony.

At this point AAN is forced to redefine "evidence" more broadly yet until it becomes basically indistinguishable from Thomas Reid's famous principle of credulity. But once we have arrived there, we are actually accepting my view that all things being equal it is rational to accept testimony unless we have defeaters to the individual's character or competence.

Thus on this first point - the issue of testimony - AAN's analysis must be rejected.