Here, I confront the most popular objection to foundationalism: That foundational beliefs must be “arbitrary”.*
[*Based on: “Arbitrary Foundations?”, Philosophical Forum 34 (2003): 141-52.]
1. Foundationalism
Foundationalists hold two theses: (i) there are certain beliefs with “foundational” justification, i.e., justification that does not depend upon one’s having justification for any other beliefs; (ii) all other justified beliefs are based on those foundations. Thesis (i) is the controversial part.
There are two reasons for being a foundationalist. The first you could call the argument from examples: when I think about how I know, or how I am “justified in believing” that I’m in pain when I am, it really doesn’t seem that the answer is that I produce some kind of argument or process of reasoning to the conclusion that I’m in pain.
The second reason is the regress argument: if you don’t have any foundations, then you have to either reason in a circle, or have an infinite regress; either of those alternatives is bad.
2. The Arbitrariness Objection
The most common objection to foundationalism consists of little more than a single word. Critics pretty much just throw out the word “arbitrary”, and that’s supposed to refute foundationalism. Any putatively foundational belief, i.e., a belief that is not supported by other beliefs, is “arbitrary”, and therefore it is unjustified, and therefore it can’t be used to support the rest of your belief system.
Btw, a popular variant is for religious people to declare that all reasoning must start from somewhere, and therefore all belief systems are necessarily based on “faith”, and therefore it’s just as good to start from faith in God as to start from any other starting point. (These people don’t normally use the word “arbitrary”, but they seem nevertheless to be treating all starting points as equally arbitrary.)
In short:
If B is not supported by other beliefs, then B is arbitrary.
If B is arbitrary, then B is unjustified.
Therefore, if B is not supported by other beliefs, then B is unjustified.
Foundationalism says that a belief can be justified without being supported by other beliefs.
So foundationalism is false.
Though many people seem to be impressed with this “reasoning”, I think it’s basically completely empty. What is supposed to be meant by “arbitrary”?
2.1. “Unjustified”
First interpretation: “arbitrary” = “unjustified”. Then premise (1) amounts to this: Any belief that isn’t supported by other beliefs is unjustified. (Steps 2 and 3 are then redundant.)
Problem: Premise (1) is then just the negation of foundationalism (specifically, thesis (i)). The argument is basically “Foundationalism is false; so foundationalism is false.” Great, but why do you think foundationalism is false?
2.2. “Unsupported by reasons”
Second interpretation: “arbitrary” = “not supported by reasons”. In this case, premise (2) amounts to this: “If a belief is not supported by reasons, then it is unjustified.” Once again, that is just the negation of foundationalism, so the “objection” is that foundationalism is false because foundationalism is false. Again, no actual reason has been given for thinking foundationalism is false.
2.3. “Unjustified because unsupported”
Maybe the actual meaning of “arbitrary” is a combination of the above two suggestions: “arbitrary” = “unjustified because it is unsupported by reasons”. Again, the objection on this reading would just blatantly beg the question.
2.4. “Not sufficiently likely”
Maybe “arbitrary” means “no more likely to be true than false”, or just “not sufficiently likely to be true”. Q: What interpretation of probability are we using here? If you’re using some kind of physical probability, then the objection is just wrong; you can’t infer anything about the physical probability of a belief’s being true from the premise that the belief isn’t supported by other beliefs.
The notion of probability here is going to have to be epistemic or logical probability. But then “no more likely to be true than false” means, basically, “no more justified than its negation” and “not sufficiently likely” means “not sufficiently justified”. This interpretation of the objection then collapses into the question-begging version of 2.1 above.
2.5. “No justification-conferring properties”
Here is a more sophisticated objection: foundationalism conflicts with the principle of the supervenience of justification. That’s the principle that if something is going to be justified (or unjustified), it has to be justified (/unjustified) in virtue of its descriptive properties. So there could not be two beliefs that are the same in all respects except that one of them is justified and the other unjustified; there must always be some other difference that explains why one is justified and the other not. Note: this is extremely widely accepted in epistemology.
Why would foundationalism conflict with this principle? Take two propositions:
A 1+1 = 2.
B Purple unicorns live on Mars.
The foundationalist says that (A) has foundational justification, yet (B) does not; (B) is merely arbitrary. What is the descriptive difference between these two beliefs that explains why A is justified an B isn’t?
Suppose the foundationalist answers that proposition A has feature F, which B lacks, and that is why A but not B is foundationally justified. (Maybe F is the feature of seeming obvious, or being something that we’re directly acquainted with, or being analytic, or whatever.)
Q: Is feature F something that makes a proposition likely to be true, or not?
If F doesn’t make a proposition likely to be true, then it’s just bizarre to hold that the presence of F makes a proposition justified. The point of justification is supposed to be that forming (only) justified beliefs is a means to forming (mostly) true beliefs.
But if F does make A likely to be true, then it seems that A isn’t really foundational, because we have a good reason for believing A: namely, that A has F, which makes it likely true.
Hence, there aren’t really any foundationally justified beliefs.
Reasons for belief vs. reasons why belief is justified
Basically, I think the objection confuses the reasons why a belief is justified with the believer’s reasons for holding the belief.
You should distinguish these things, whether or not you’re a foundationalist. Let’s say S is inferentially justified in believing conclusion C based on evidence E. Then the reason why S’s belief is justified is this:
[S justifiedly believed E, S rationally inferred C from E and believed C as a result of that inference, and S had no reason for doubting the inference.]
But the above is not S’s reason for C. S’s reason for C is E, simply. That’s true by definition, because we stipulated that C was based on E.
Since that distinction applies to inferential beliefs, it should also be applicable to foundational beliefs. For a foundational belief, the believer may have no reason for the belief, i.e., the belief isn’t inferred from any other belief. Yet there will be a reason why the belief is justified – say, that the belief seems correct to the person and he has no reasons for doubting it, or that the believer is directly acquainted with a fact that makes the belief true, etc.
Inferential justification is the special case where the reason why a belief is justified is that the subject inferred the belief from other justified beliefs. Non-inferential justification is the case where a belief is justified for any other reason.
Every foundationalist will accept this distinction, so they won’t have trouble differentiating foundational from arbitrary beliefs.
3. The Arbitrariness Objection Is Arbitrary
Most people who reject foundationalism (who are mostly students who just started thinking about philosophy and were lured into skepticism) are incoherent: Basically, they treat the negation of foundationalism as foundational. They do this by starting from the assumption that any belief not supported by reasons is unjustified.
By their own lights, anti-foundationalists can’t just assume that. (A foundationalist could consistently treat foundationalism itself as foundational. But an anti-foundationalist could not treat anti-foundationalism as foundational.) So they have to give an argument for this. They often trick themselves into thinking that they have an argument by deploying the ambiguous word “arbitrary”, which seemingly bridges the gap between “not supported by reasons” and “unjustified”. But on the most obvious interpretations of the word, the “argument” just blatantly begs the question.
The most sophisticated version of the arbitrariness objection is to claim that foundationalists can’t identify any features that would explain what make foundational beliefs justified. At least this version of the objection doesn’t immediately beg the question. But it just confuses the explanation for why B is justified with the subject’s reason for adopting B.
Conclusion: Foundationalism is fine. The most popular objection has nothing to it.
My interpretation of the arbitrariness was that you could sort of whimsically pick which beliefs you want to be the foundational beliefs - or perhaps less strongly, there could be multiple, inconsistent sets of foundational beliefs.
Suppose a belief B is justified by B' which is in turn justified by B'', and so on. Suppose also that we are not sure if the sequence of justificatory beliefs terminates but we know that even if it did not terminate, at each stage the reasoning justifying the previous belief is valid (for e.g. we are sure that no prior justificatory reasons would be repeated, etc). In that case, would we be justified in believing B?