Here Dennett seems to assume that if you can't show by reason that a given proposed source of truth is in fact reliable, then it is improper to accept the deliverances of that source. This assumption goes back to the Lockean, Enlightenment claim that, while there could indeed be such a thing as divine revelation, it would be irrational to accept any belief as divinely revealed unless we could give a good argument from reason that it was. But again, why think a thing like that? Take other sources of knowledge: rational intuition, memory, and perception, for example. Can we show by the first two that the third is in fact reliable--that is, without relying in anyway on the deliverances of the third? No, we can't; nor can we show by the first and third that memory is reliable, nor (of course) by perception and memory that rational intuition is. Nor can we give a decent, non-question-begging rational argument that reason itself is indeed reliable. Does it follow that there is something irrational in trusting these alleged sources, in accepting their deliverances? Certainly not. So why insist that it is irrational to accept, say, the Internal Testimony of the Holy Spirit unless we can give a rationally conclusive argument for the conclusion that there is indeed such a thing, and that what it delivers is the truth? Why treat these alleged sources differently? Is there anything but arbitrariness in insisting that any alleged source of truth must justify itself at the bar of rational intuition, perception and memory? Perhaps God has given us several different sources of knowledge about the world, and none of them can be shown to be reliable using only the resources of the others.My first observation was that Plantinga sees that reason relies on certain unprovable axioms, without all of which, no reasoned conclusion can be reached. This is a good start. Plantinga claims the axioms are logic, memory, and perception, but I suspect these equate to my axioms of logic, regularity, and the axiomatic nature of experience (that experiences need to be explained).
Plantinga then argues that, since we are comfortable accepting these axioms without proof, why not accept additional axioms (e.g., that there are non-rational sources of knowledge)? Maybe "Internal Testimony" (whatever that is supposed to mean), is just an extra-rational assumption, rather than an irrational one.
Alternate Sources of Knowledge
Assuming that knowledge is defined as justified true belief, what does rationality say about sources of knowledge?
If I have some source of knowledge, S, then I am saying that there is some associated test, TS, I can apply to a proposition, P, to test its truth:
S: Truth(P) = TS(P)
I am also saying that there is some (potentially different) form of justification for belief in P:
S: Justification(P) = JS(P)
In the case of science, justification and test are one. The truth of a scientific belief is fixed by the test of its truth.
If I have multiple sources of knowledge, then I may have multiple definitions of the truth:
S1: Truth(P) = TS1(P)
S1: Justification(P) = JS1(P)
S2: Truth(P) = TS2(P)
S2: Justification(P) = JS2(P)
At this point, I'm going to assume that S1 is science and S2 is supernaturalism. This means I can write:
S1: Truth(P) = Justification(P) = TS1(P)
S2: Truth(P) = TS2(P)
S2: Justification(P) = JS2(P)
There's no guarantee that a truth from one source of knowledge is a truth in the other. An intuitive personal truth (one tested by asking a person for his opinion) may not be a scientific truth. Thus, in general, the "truth" of a proposition has no fixed meaning when there is no preferred source of knowledge.
There are three ways to avoid the problem of propositions having no truth values:
1) Assume that there is only one source of knowledge. In this case, most of us would likely choose science.
2) Assume that there are multiple sources of knowledge, but that they should all agree on the truth of any applicable proposition. This means that:
TS1(P) = TS2(P)
3) Assume that no single proposition can be evaluated by every source. That no truth value revealed by S1 can be revealed by S2, and vice versa. This might be akin to Stephen Jay Gould's Non-overlapping Magisteria.
I'm pretty sure theists would reject (1). I won't entertain the claim that science is not the preferred source of knowledge where available because no one reading this blog could consistently make that claim.
This means that either (2) or (3) is the case.
Case (2) is ruled out because supernatural knowledge sources are broadly inconsistent with scientific ones where they overlap. Psychics and miracles are routinely shown to be fraudulent, and supernatural sensation isn't any better than guessing. Thus, we have ample evidence that supernatural knowledge sources fail to give the same truth values as scientific methods, as would be expected in case (2).
Case (3) is problematic for the theist because it means that any question that can potentially be settled by predictive means cannot be answered by a supernatural source. Indeed, it means that any phenomenon that one claims to know through supernatural methods must be unknowable through natural methods. I think this is one reason why theists assert that the mind cannot be purely physical, for otherwise, there would be no meaningful truth claims about supernatural souls.
As is well known, history's trash heap is littered with supernatural beliefs that were displaced by scientific truths, and the boundary of the supernatural domain has been in monotonic retreat for centuries. Still, the inductive inference that all supernatural claims are rot isn't deductive proof that they are rot. So, let's suppose that there is a domain, non-overlapping with science, in which supernatural sensation does reveal the truth (at least, most of the time). It is implicit in a knowledge claim (supernatural or otherwise) that the knowledge will be true as well as justified. Can we not then ask science to assess the efficacy of the supernatural knowledge source?
Symbolically, what we're evaluating is this:
S1: Truth(S2 is effective) = TS1(S2 is effective)
If it were possible for S1 to find S2 to be ineffective, then the space of truths of S2 would overlap (and potentially conflict) with those of S1 because S2 also implicitly asserts that it is effective. This would violate the premise that S1 and S2 don't overlap.
Therefore, S2 cannot give an answer that S1 might later determine to have been wrong. Thus, a fortune-teller cannot tell me that I'll be a millionaire by age 30 because I could use scientific means to know that he was wrong, and the fortune-teller implicitly claims he is right (i.e., he claims that his knowledge is not just supernaturally justified, but true). Unfortunately, this principle also negates all knowledge about future experience derived by S2 because such knowledge could be falsified by S1.
The practical upshot of all this is that S2 is unable to tell me anything about future experience. So why should I care what S2 has to say?
I might care about what S2 says if the execution of the method of S2 is a source of amusement. Is this why TV commercials for psychic hotlines display a "for entertainment only" disclaimer?
Science and Rationality
The beautiful thing about science (apart from the fact that it works) is that it is derived from rationality itself. In assuming that science is a source of knowledge, what am I assuming? I am assuming that I can make inductive inferences from past experience to predict future experience. Can I drop this assumption without destroying rationality itself?
If I assume that past experience is no guide to future experience, why should I assume that a theorem that I have just deductively proven true won't be false before I perform the next step in a proof? I cannot. I must assume that past experience, whether mental or physical, is a guide to future experience. Once I make this assumption for purposes of rationality, science automatically follows.
I have shown that supernatural sources of knowledge are either totally unreliable, or can only tell me about things that are irrelevant to experience. I have shown that science as a source of knowledge follows from the assumed axioms of rationality.
If science is accepted as the primary source of knowledge in any domain, it is the only relevant source knowledge about experience.
Nov 14 2006 Clarification: Science here refers to methods of deductive and inductive inference. It could be mathematics as easily as it is physics or linguistics.