Evidence Is Seemings

Here, I explain for the umpteenth time why I think justification comes from seemings.* I was invited to contribute to a debate on whether “evidence is seemings”, defending the affirmative.

[ *Based on: “Evidence Is Seemings” and “Four Challenges for Phenomenal Conservatism,” pp. 127-34 and 142-5 in Contemporary Debates in Epistemology, 3rd ed. (Wiley, 2024). ]

1. A Theory of Justification

I was asked to talk about whether “evidence” consists of seemings. Some epistemologists use “evidence” to mean whatever confers epistemic justification for belief, so let’s just talk about that.

I think that all justification for belief derives from appearances, or seemings (mental states one has whereby something seems to one to be the case). These include sensory experiences, memory experiences, and intuitions, perhaps among other things. It’s rational, I claim, to presume that things are the way they seem, unless and until you acquire reason (which would itself have to come from other seemings) for doubting that. I call this principle Phenomenal Conservatism (PC).

2. Examples

Usually, the best way to assess a theory of X is to think of examples of things that we pretheoretically consider examples of X, and see how well the theory does explaining them.

Most beliefs derive (directly or indirectly) from observation by the 5 senses. Observation is a matter of having sensory appearances (things look a certain way, sound a certain way, etc.) that are caused in the right way by the objects they represent. PC explains why we (justifiedly) believe what we seemingly observe, until we get reasons for doubting it.

Or consider a priori beliefs, such as the belief [If a=b and b=c, then a=c]. When you think about this proposition, it just seems obvious. That is a type of appearance (sc., an “intuition”). PC can explain a priori knowledge as resting on these sorts of intuitions.

Memory beliefs are similar. You don’t start by assuming that memory is untrustworthy until proven trustworthy. You start by assuming that what you seem to remember is what happened, unless and until you have grounds for doubt.

Regarding self-knowledge (knowledge of your own mental states), note that the appearance need not be a separate state. Maybe conscious mental states are identical with their own appearances. E.g., if you’re in pain, the “appearance of being in pain” just is the pain.

In brief, Phenomenal Conservatism does a good job of explaining the beliefs that we normally take to be justified.

3. Alternative Accounts

3.1. Awareness

Some people claim to have justified beliefs based on states of direct awareness of the things their beliefs are about. How does this awareness differ from mere appearance?

The main difference is that it is possible to have a false appearance (it can appear to you that P when P is false), but there cannot be false awareness (you cannot be aware of X unless X exists).

So a test case would be one where a person has a perfectly realistic hallucination, say of a porcupine. In this case, they have the porcupine appearance, but they are not aware of a porcupine, since there is no porcupine to be aware of.

Intuitively, this person does still have justification for thinking there is a porcupine (in the absence of grounds for doubt). This shows that it is appearance, rather than awareness, that grounds justification.

3.2. Other alternatives

Sometimes, people claim to base beliefs on something other than appearances. Most of the time, this will be an appeal to inference. E.g., a detective might claim to base the belief that suspect S is guilty on the fact that S’s DNA was found at the crime scene (rather than on an “appearance” that S is guilty). But then we’d have to ask what the belief that S’s DNA was at the crime scene was based on. And so on—we have to follow back the chain of reasoning to the start. The starting point, I claim, is going to be appearances. E.g., an investigator at the crime scene had a sensory appearance of a hair found at the scene, then other investigators had sensory appearances of the DNA testing apparatus, etc.

Philosophers may propose other bases for justification (other than appearance), e.g., the actual reliability of a given belief-forming method. This implies that if you have normal appearances but your belief-forming method is (unbeknownst to you) unreliable, then your belief is unjustified. E.g., the brain in a vat is being irrational all the time. This seems wrong.

Likewise, the view implies that if you have a reliable method, but it doesn’t seem reliable to you, then you would be justified in forming beliefs by that method. This also seems wrong.

Reliability is just one example. I think similar arguments will go for any alternative to appearances.

4. The Self-Defeat Argument

If you carefully reflect, I claim, you will find that almost all of your beliefs rest on appearances: you believe things because they seem right to you. (The exceptions are things like blind faith, wishful thinking, and similar obviously irrational belief-formation methods.) Briefly, I think this makes it self-defeating to deny PC, because that denial would itself rest on appearances, which you’re denying are a valid source for forming beliefs.

For more about this, see my earlier posts:

Compassionate Pheomenal Conservatism

Alternative Self-Defeat Arguments

Phenomenal Conservatism and Self-Defeat

5. The Internalist Argument

I claim that your appearances are the only things that affect the justification for your beliefs. Suppose I am wrong and that there is some other factor, E, which is independent of appearances, that affects whether beliefs are justified.

Then it should be possible for a person to have two beliefs, say, P and Q, where everything about the beliefs appears relevantly alike for the person (P and Q seem equally correct, equally reliable, equally free from defeaters, etc.), and yet the belief that P has E while the other belief lacks E.

In that case, the externalist would apparently say that the person should believe P but not Q.

Imagine the subject telling us about this: “Well, P and Q each seem equally correct to me, equally reliable, and in general equally good in all epistemically relevant respects. Nevertheless, I just believe P and don’t believe Q, for no apparent reason.”

This, I submit, could not be an accurate expression of a rational mental state. One should conclude, then, that it could not be that P is justified while Q is not; and therefore, it could not be that the justification for beliefs depends on factors extraneous to the appearances.

For more, see my earlier post:

Phenomenal Conservatism and the Internalist Intuition

6. Objections

6.1. Memory

Maria Lasonen-Aarnio wrote the other side of the debate, denying that “evidence is seemings”. First objection: when you believe something simply because you remember it, without remembering how you learned it, this belief can be justified without relying on seemings. It is justified simply because it results from the normal operation of memory.

Reply: This isn’t an example of evidence. Once a belief is acquired, one normally holds on to it, and is justified in holding on to it, as long as one hasn’t forgotten it and hasn’t gotten new evidence against it. But neither of these things—failure to forget, failure to acquire counter-evidence—constitute evidence for the belief.

Evidence is only required to justify acquiring a belief in the first place; no evidence is required to justify retaining a belief. Retention only requires the absence of defeaters plus the normal working of memory.

6.2. Corrupt-source appearances

Suppose that a person adopts an irrational belief that P. Beliefs can affect your appearances (some people call this phenomenon “cognitive penetration”). So suppose that this irrational belief causes the person, later, to experience an appearance that Q. At the later time, however, the person has forgotten that they adopted an irrational belief, or they are unaware that the belief caused the appearance. So the person has no specific reasons for doubting Q. So, according to PC, the person has justification for believing Q.

But some philosophers think, intuitively, that Q should count as unjustified. This is supposed to be a counter-example to PC.

Reply: I think Q is justified. This is just a case in which a person gives himself misleading evidence at a later time.

Take a more normal example of misleading evidence. Suppose you write a false entry in your diary, saying that you pet a black cat today. Ten years later, you’ve completely forgotten this, but you happen to look through the diary, and you see the entry about petting the cat. You then adopt the belief that, on that day ten years ago, you pet a black cat. This is completely justified.

Now add a little to the story: suppose that at the time you wrote the false diary entry, you were motivated by an irrational belief—say, you irrationally believed that you were petting a black cat when you were actually petting a dog. This was just a weird quirk that happened to you on that day, that you later completely forgot about.

This doesn’t matter to what you rationally ought to believe at the later date. At the later time when you read the diary entry, it is still rational for you to think that you pet a cat, because that is what your current evidence supports. It’s irrelevant that this evidence was caused by an earlier unjustified belief.

Similarly, if an irrational belief produces an appearance that Q, with no defeaters, then it’s rational for the subject to believe Q.

6.3. Prior probabilities

Some contingent propositions have a prior probability of zero. E.g., suppose there are two line segments in a continuous space. The initial probability of their having exactly the same length is 0. If a proposition has an initial probability of zero, then it should also have a probability of zero conditional on any evidence. Thus, no evidence supports it. This is due to Bayes’ Theorem:

P(h|e) = P(h) * P(e|h) / P(e).

If P(h) = 0, then P(h|e) must also = 0 (assuming P(e) is nonzero; otherwise, the above equation is invalid).

But one of these propositions could also seem true, e.g., two lines might appear to someone to be equally long. So, maybe, this is a case where you could have an appearance that P and yet have no evidence for P, even in the absence of defeaters?

Replies:

a) Actually, there is a nonzero prior probability that space is discrete, and thus a nonzero probability that two lines are of exactly the same length.

b) Actually, two lines never appear exactly the same length. Our vision has limited discriminatory capacity, so things never appear to have any perfectly precise length. Two lines can only appear to be approximately the same length (to within our discriminatory capacity).

c) If you had a being with infinite discriminatory capacity, that being could have appearances of precise lengths. However, in this case, P(e) (where e is a proposition assigning some perfectly precise apparent length to something) would also be 0. In that case, the argument that P(h|e) = P(h) would fail.

6.4. Access Requirements

What if someone has an appearance that P, but no appearance that Q, but the person doesn’t know (does not “have access to”) the fact that they lack an appearance that Q. Would they then be justified in believing P but not Q? But they would be unable to give any rational explanation for why they believed P but not Q. What should the phenomenal conservative say about this?

Reply: I find this “example” pretty opaque. Lasonen-Aarnio raised it but did not give any details about what might be happening so that one could lack access to the fact that one lacks an appearance that Q.

The best I can come up with is that a person might have some other mental state that isn’t an appearance but that they confuse with an appearance that Q. If that person also accepts PC, then they would think they were justified in believing Q, but they wouldn’t actually have justification because they don’t actually have an appearance.

If this is possible, and if the person could be justified in their confusion, then I think the person would have inferential justification for believing Q (based on inference from the proposition that they have an appearance that Q) but no foundational justification for it.