Can we be justified in believing things that are false?

Let me shift away from morality for a bit and start to look at epistemology. I’m going to start by looking at this essay by Jonathan MS Pearce over at Tippling Philosopher, where he asks this question: Can there be very strong reasons for believing something although it is false? However, much of the essay is built around developing a specific idea of what is means to know something, which might be rather long and tedious for someone who just finds that question interesting. So, to start, I’m going to try to boil down his discussion here to some critical points using the standard simple philosophical definition of knowledge.

In philosophy, knowledge is generally defined as “justified true belief”. This means that you know something if you believe it to be true, you are justified in believing it to be true, and it is in fact true. So, given that, the question translates to “Is it possible for us to be justified in believing something that, nevertheless, is actually false?”. Now, of course, by that definition I certainly couldn’t be held as knowing that the belief is true; the belief is false so I can’t know it. But if I believe that I am justified in believing that it is true, then I’d certainly think I knew it, which can then be problematic. Surely, we can argue, the reason knowledge is justified true belief is because we want that justification to provide us sufficient reason to think that the belief is true. If we can be justified in believing that something is true even when it’s false it doesn’t look like that justification is doing its job.

Pearce’s big point here, it seems to me, is that the only way we can claim that if we are justified in believing that something is true then it must in fact be true is if our justification provides certainty. It would have to be the case that our justification means that the proposition cannot possibly be false. However, we don’t have access to certainty. Our senses are not, in fact, 100% reliable. We also have potential cognitive lapses and biases that might cause us to accept propositions that are not true; we might make an error in reasoning, for example. Heck, we also might not have all of the information and evidence we’d need in order to come to a proper conclusion. Thus, it seems that if we insist — as Descartes did — that we can only know something if it follows logically from propositions that we have established with certainty then there is vanishingly little that we could ever know. Thus, that definition of justification produces a definition of knowledge that is unworkable. And while it being unworkable doesn’t mean it’s wrong, we probably want some kind of workable definition of knowledge that allows us to, well, actually know things.

So we need to weaken the definition of “justified” so that it doesn’t require certainty. But, of course, as soon as we do that then it becomes possible for us to be justified in believing something that turns out to be false. This leaves us with the problem that we don’t want to make the definition of justified too strong so that we can’t actually know anything by it, but can’t make it too weak either because it would mean that we might accept a lot of false beliefs. There’s a ton of issues around this, but at base what we need out of our justification method(s) is reliability: we’re justified in accepting a method if it produces true beliefs more often than not. How more often than not — and how we determine that — varies by philosophical theory.

Anyway, Pearce sums up his entire essay at the end thusly:

Therefore, let me finally recap the method for reliably finding truth:

1) The thinking entity exists (cogito ergo sum)

2) Any other existence must assume a grounding in faith in an external reality and truth claims assume a similar correlation

3) All phenomena assumed to relate to the outside world count as evidence

4) A fact is a piece of knowledge which is a truth claim supported coherently by overwhelming evidence

5) A fact must be falsifiable

6) A truth claim cannot be supported by very strong reasons if it is unfalsifiable

7) An unfalsifiable truth claim cannot be a fact

8) In one sense, you can have very strong reasons to believe something if you have incomplete evidence. Therefore, very strong reasons fits within the context of the best available evidence.

There are a lot of things to talk about here, but I’m not going to focus on most of them. Let me briefly talk about 2 – 4), and then I’ll move on to the more interesting one — to me — which is 5 – 7). The issue here is that Pearce seems to be limiting “truth” to empirical world claims. But surely we can have truths about things that can’t really be called “external”, can’t we? Are we to infer, for example, that pain claims are really about the external world and not about the internal world of experience? Can there be true statements about my own internal thought processes and imaginations? Heck, by this can we ever claim that there are true statements about fictional realities? What does it mean, then, for something to relate to an external reality, as opposed to an internal one? Pearce admits that there are issues to work through here, but I see these as pretty fundamental … especially if we wanted to adopt this method for things other than direct empirical or scientific claims.

But 5 – 7) is the most interesting one, as it tries to pull out “unfalsifiable” claims and insist that any unfalsifiable claim cannot be a “fact”, but in light of 2 – 4) and 6) it would also imply that we can’t “know” the truth of any unfalsifiable proposition. Now, it becomes clear why Pearce does this — he’s trying to go after God claims — but he is decidedly vague about what “unfalsifiable” is actually supposed to mean. His underlying premise is that unfalsifiable claims aren’t testable, which is a condition for having strong reasons for — read: being justified in accepting — a proposition. But what does he mean by “testable”? He uses, naturally, a religious example to try to highlight that:

If we take, for example, the truth claim that “God answers prayer” we might be able to more clearly understand this point. This is where a claim is unfalsifiable. The claim involves both the empirical (testable) and the untestable. We can test, and have tested, prayer and its effects, using double blind experiments and so on. However, when the experiments (and they have[11]) have shown no positive results for intercessory prayer (when God intervenes for someone on the behest of someone else) defenders of the truth claim move the untestable, unfalsifiable agent (God) outside of the normal causal bounds. For example, a defender might claim “But God may decide not to answer those prayers because he doesn’t like being tested” or use some such device. This means that the agent, or the claim, is effectively unfalsifiable. We have seemingly empirically falsified the belief that God answers prayer using an evidential procedure. However, since God is not consistent, deterministic and testable in such a manner, one cannot reliably say whether God does or doesn’t answer prayer. Thus such a belief is not, in my opinion, supported by very strong reasons, but by unevidenced faith. If excellent and reliable evidence is forthcoming for such a claim, then belief in the claim’s truth value would be more justified. It would rely more on evidence, and less on faith.

In this way a defender of an unfalsifiable claim can move the causal factors of the claim around like a pea in the shell game, defrauding the tester of any ability to test the agent (in this case, God).

He deliberately conflates “empirical” with “testable”, but as it turns out the example isn’t really a case of something being “untestable” and thereby “unfalsifiable”. While the defender here might be accused of moving the goalposts — or, rather, of altering the theory to fit the new evidence instead of abandoning it — they haven’t necessarily moved it into the “unfalsifiable” realm. For “unfalsifiable” to have meaning, it has to be the case that the claim is in some sense unfalsifiable a priori. It can’t be the case that, for example, we could falsify it but that doing so would currently be impractical, or else pretty much all of cosmology is unfalsifiable until we develop or developed the tools to indeed observe those conditions, at which point it suddenly and miraculously became “falsifiable”. If we are to claim — as Pearce does — that unfalsifiable claims cannot be justified, it cannot be the case that those claims are only considered unfalsifiable for precisely as long as we can’t justify them. It has to be the case that they can’t be tested in principle.

And, as it turns out, we can test the case that Pearce uses here: the push to saying that God could decide not to respond to prayers if He had a reason not to, or a reason to not want to be tested. Sure, we can’t test it using the methods of simple science, because it is, in fact, a claim about an intentional agent, and such claims can’t be tested blindly. As we’ve seen in psychology, you can’t simply take intentional agents, put them in a lab, and then get them to act the way they would outside of the lab. If they know or suspect what you are testing, they might consciously or unconsciously act in abnormal ways, for a variety of reasons. So in testing intentions, you need to do less empirical analysis and more intentional analysis: in short, you need to ask questions like “Why would God not want to pass the test?”. Which is how we deal with all intentional agents; if someone wants to justify the statement “X took out some money from the bank because they wanted to buy a video game”, the first place to start is by assessing whether them doing that is consistent with their intentional character. And if we had an empirical result that contradicted that, like us finding out that they didn’t buy a video game, we’d still want to assess our evidence against what we think is true about that person and their intentions, and so might conclude, for example, that they haven’t made it to the store yet.

So it is perhaps true that the move made isn’t testable strictly scientifically, but it is testable using the intentional stance and intentional methods. Thus, it is not clear what Pearce means by “unfalsifiable”. For example, is the statement “2+2=4” unfalsifiable? We know it by definition, which many people take to be the hallmark of an unfalsifiable claim that nevertheless not only is but must be true. However, we could demonstrate that “2+2=5” is in fact false, suggesting that “2+2=4” is falsifiable. But, then again, we wouldn’t demonstrate that empirically either, which seems to clash with Pearce’s idea of “testable”.

In short, I don’t think that adding the “unfalsifiable” part does anything for us here. There are just too many issues and complexities around that term for it to have any meaning. Pearce seems to add it only to try to attack faith-based claims, but all it ends up doing is muddling his definition and confusing the discussions. However, as we’ve seen, the answer to the question he is trying to answer in the essay is “Yes, unless you require certainty for every proposition you claim to know”, and we’ve seen the problems with that.

Tags: epistemology