Intuitions as truth-vectors vs. intuitions as empirical data

Or, why weird views can seem correct to me

Here’s a question: suppose there are two philosophical views, A and B. Both of them equally well predict everything we see, including our intuitions about various philosophical matters. They also both have the same theoretical virtues—simplicity, unity, explanatory power, etc. To a detached observer with no prior commitments, they would seem equally likely. However, there is one difference between them: given our intuitions, A seems more true than B. A vindicates many of our intuitions, and considers as true many things we pre-theoretically belief. B, in contrast, has to be revisionary, considering our intuitions biased or otherwise in error. However, keep in mind that we would be just as likely to have the intuitions we do conditional on B being true. The question: given this situation, what, if anything, do you decide to believe? In particular, do you suspend judgment, or do you believe A?

I’ve recently discovered that I’m weird in saying we should suspend judgment in a case such as the above, given the stipulations. I’ve talked about what I think the proper role of intuition in philosophy is, and I’ll quote myself here:

I am not “against intuition” in general. I am against intuition as the ultimate justification for moral principles. I believe the philosopher should use intuitions in exactly the same ways the mathematician does. For the mathematician, intuition is indispensable: it is needed to know what lines of inquiry are worth putting effort into, making guesses as to what is true so that you know what to try to prove, and looking for flaws in others’ proofs. But mathematicians do not publish papers where their ultimate justification is that their claim achieves the best balance of intuitions; insofar as they might explicitly rely on such, they recognize their understanding as defective, and continue looking for an actual proof. Likewise, in ethical philosophy, we need intuitions for the same reasons: if the Categorical Imperative turns out to imply it’s immoral to wear a green shirt on a Tuesday, I need to go back and check my work; if I find myself having to choose whether to kill one man to harvest his organs and save five, I will refrain from doing so because it is intuitively wrong, as I do not yet have an Ethical Theory of Everything. But as long as I am relying on intuition, my judgments are merely provisional, and I need to continue the search for actual justifications for them.

In sum, I think intuitions are empirical evidence, nothing more. I think, at least when it comes to theoretical and empirical knowledge, we have good evidence that our intuitions do better than chance at tracking truth, so they do count as evidence for the intuitively true proposition.¹ But the way empirical evidence works is that if two hypotheses are both equally good at predicting the evidence, and if your priors in the two hypotheses are similar, then the two hypotheses get similar credences.

But does the scenario I opened with ever actually happen? Maybe it’s a bit extreme, to assume two views have equal priors and that they both equally well predict our intuitions. Maybe the intuitively-correct views have more going for them than that they predict our intuitions while also vindicating them.

As a matter of fact, I think the scenario I opened up with is very unlikely. But generally things fail in the other direction. Non-intuitive views have quite a bit going for them by way of simplicity, naturalness, arguments, etc., while also being compatible with perfectly likely and independent stories about why our intuitions are the way they are—as a matter of fact, that’s why anyone believes them in the first place. On the other hand, the intuitive views are propped up, mainly, by the fact that they vindicate intuitions. Were we banned from appealing to intuitions except as empirical data, I think proponents of the latter views would be left utterly unarmed.

Consider, for example, the issue of robust moral realism—which holds that some actions are objectively obligatory, and that such involves the actions having metaphysically irreducible, intrinsically normative properties that are as part of the fabric of reality as mass and charge—and moral skepticism. Moral skeptics are often motivated by the idea that intrinsically normative properties would be strange and unlike anything we know to exist (see Mackie’s “Argument from Queerness”²), and that the reality we are acquainted with can be perfectly well accounted for without them. Robust realists, on the other hand, argue that their view is the only or best way to vindicate our intuitions, and they address skeptical arguments piecemeal, often with the charge that accepting them would lead to skepticism in other areas.

What evidence do our intuitions give? Conditional on robust realism being true, we can grant that the probability that our intuitions pointing to the position would be fair. But also conditional on moral skepticism, the way things seem to us make perfect sense—when we consider a horrendous action, we very powerfully feel as though it is simply to be ruled out, and it is not hard to give an evolutionary explanation as to why such a feeling would arise. I think a neutral outsider would judge it to be fairly likely that things appear to us the way they do conditional both on robust realism and moral skepticism being true. And what of the prior probabilities, that is, the likelihood of either view without the evidence of intuition? I’d wager that, for aliens without these intuitions, the idea of robust moral realism would not occur to them in a million years, and it would seem very implausible if suggested to them. If anything, the prior likelihood of moral skepticism is much higher. And how are we any different from them, except in that we have the evidence of intuition?³ Why, then, should we believe anything different from what they believe if they updated on the empirical evidence provided by intuition?

What about skepticism in general, the idea that most or all of the knowledge we take ourselves to have is not in fact knowledge? Again, there are powerful arguments for skepticism, which is why people are inclined to accept it in the first place. And we can very easily explain why skepticism seems strange conditional on skepticism being true; evolution optimizes us to live in the world we actually live in, not the world we can justifiedly believe ourselves to live in, so it is natural that we would see a tiger and take ourselves to know there’s a tiger. Requiring that we rule out being a brain in a vat isn’t useful for beings optimized to live in a world where we are not brains in vats. So, our intuitions about the matter don’t seem to be much evidence against skepticism.⁴

And because of our inner desire to vindicate intuitions, positions have been dreamt up that should have never even seen the light of day.⁵ Consider, for example, the “modal realism” of David Lewis. On this view, all possible worlds are as real as the actual world—the only thing special about the actual world is that we’re in it. It’s possible that the world could have just consisted of a single cat eternally eating some grass—so, for Lewis, there is a real possible world out there where such is the case. Why did he believe this? Was he a Leibnizian who thought it would be arbitrary for one possibility to be actualized instead of another? No, the reality is far less respectable. He believed it because it is a useful assumption which helps our other philosophical theories work out. We need to talk about what’s possible to do nearly any metaphysics; the Lover of Wisdom finds Problems with all other accounts of possible states of affairs; therefore, we are left with modal realism.

I think it should be clear to those unfamiliar with the Love of Wisdom that no considerations of that sort can be used to justify a claim of that sort. “Well, it seems this is what has to be posited for our Best Theories to work out, and I’m justified in believing in whatever is quantified over in our Best Theories”—is this reasoning sensitive to the concrete existence of possible worlds? Wouldn’t we be getting along just the same, doing everything we want to do without concrete possible worlds? Frankly, I think it is a scandal to the field that it developed in a manner so as to cause these thoughts to exist and not be immediately dismissed.

Why did this happen? Well, most philosophers do not treat intuitions as empirical evidence which should be dispensed with once we actually figure things out. They think that if two views differ equally in every respect—including how well they predict the fact of our having certain intuitions—then the one that better vindicates our intuitions is to be preferred. Intuitions, for them, are little vectors pointing at truths, though imperfect ones, so we need to add all of them up and see what they point to on net, and then we have a strong reason to believe whatever they point to. On this view, of course they can be used as a tiebreaker.

Intuitions do not count as empirical evidence, the lovers of intuition often say—rather, they are or can be the operation of a priori reasoning. But come on, what actually happens when you see a proposition as self-evident, or as intuitively true? You think about a claim, and you have a reaction. You’re not getting any information other than that you associate a certain feeling with a certain proposition.

What if half of people found moral realism self-evident, and the other half didn’t, with the realists being blonde and the skeptics being brunette? Would you have a different belief conditional on finding out you’re a blonde than you would conditional on finding moral realism intuitive? No, you should believe depending on the probability of the claim in question given your evidence, whether it be evidence about your hair color or what feelings you associate with a proposition. But if it is indeed true in the actual world that the probability of moral realism conditional on the empirical fact of your having a certain intuition is, say, less than 0.5 (which should be more or less true, if we can indeed account for our having certain intuitions conditional on moral skepticism being true), then why should the actual existence of people for whom the evidence is misleading matter? Why is a strategy for belief that foreseeably leads half of the actual people to failure an intolerable one, but a strategy that foreseeably leads half of the possible people to failure a fine one? It seems, then, the feelings you associate with a proposition should relate to your belief in the same kind of way any empirical evidence should.

[EDIT—A lot of people are bringing up an important point, so I’ll address it here. The point is that in order to use intuitions as empirical evidence, you need a prior, which seems like it must be formed through the use of intuition. Indeed, in the opening and later I appeal to simplicity, which is a relevant consideration for many people because they think simple things are more likely to be true because that’s intuitively so. I will not hinge this post on the a priori status of Occam’s razor or on its restriction to the world of experience, and indeed I think this worry is easier to get around. Put simply, I can grant that intuitions are necessary for us to figure out the correct priors in the first place, but that is irrelevant to what I say here. My argument is successful if what I say goes through given the actual priors I and the reader have, independently of where the priors comes from.

I am meeting the reader where they are at, not questioning how they got to where they are. From our savior Bayes, we know that the odds of, say, robust moral realism conditional on our intuitions is equal to the likelihood ratio (how relatively well the truth of robust realism predicts our having certain intuitions) multiplied by the prior odds of robust realism. Prior, that is, to our intuitions about robust realism. It hardly seems that these prior odds are in favor of robust realism. Even if I have to resort to using intuitions somewhere else, say when it comes to Occam’s razor, it would remain that this likelihood ratio is 1 and these prior odds are not high, so belief in robust realism is not justified.

It should be instructive that the scientist does ultimately rely on intuitive justification—I imagine if you raised the possibility that his results are simulations in the Matrix, he would just say “Okay, I don’t know, I just assume otherwise”—and yet he does not spend any time thinking about intuitions as such. There’s just the usual matters of unity, simplicity, etc., and then how likely the data is given a theory. Even granting that intuitions are indispensible somewhere would not require one to think they should be appealed to in any way remotely resembling what analytic philosophers do.]

---

Well, there is one way it can make sense to give accounts of knowledge, morality, etc. using “the sum of our intuitions is basically correct” as a premise. That would be if the relevant facts about the subject matter were such that the truths about the matter and our intuitions about them were no coincidence. Consider, for example, someone who thinks that diamond is stacked layers of hexagonal lattices of carbon atoms, while graphite is carbon atoms arranged in a cubic lattice. No, they are not misunderstanding the conventions around the words—they know full well that people think diamond is the cubic lattice and that graphite is the hexagonal lattice, but they think things are in fact the other way around. Nor does our individual think they have any shocking empirical discoveries, such as that scientists screwed up when investigating the molecular structure of paradigmatic diamonds. This person just thinks graphite is itself the thing people put on engagement rings, and diamond is the stuff pencils write with, knowing full well that the stuff on engagement rings have a cubic structure and the pencil lead has a hexagonal one.

A normal person considering this scenario will think the individual is confused about the words they use. “Diamond” just is the word we use to refer to the hard, shiny rock people often put on engagement rings, and “graphite” just is the word we use to refer to the stuff pencils usually write with, and they turn out to have the molecular structures people take them to have. Now, if they take the claim at face value, Lovers of Wisdom will think it is a level confusion to suppose that diamonds are such-and-such rocks with such-and-such structure because we use the word “diamond” in a certain way; note that the former claim is a claim about diamonds, not the word “diamond.”⁶ But I trust that the reader can understand me when I refer to the sense in which the guy from our example is merely confused about words, and that the facts about our conventions surrounding the relevant words are enough to rule his position out.⁷

It is, then, no surprise that we surmise the nature of diamonds in part by looking at what people intuitively refer to as diamonds; diamonds are the shiny hard rocks people put on rings, and we should investigate those paradigmatic rocks if we want to learn more about the diamond’s nature. If people disagreed about the molecular structure of diamonds, not due to empirical disagreement but rather due to disagreement as to which rocks count as a diamond, then weighing up our intuitions would make sense, since it’s true a priori that diamonds are the things that are sufficiently similar to paradigm examples of diamonds.

So, likewise, we may think that the concepts of morality and knowledge have to correspond to our intuitions as a whole, because their natures are determined conventionally, in the same sense the nature of a diamond is in part determined conventionally. Maybe knowledge just is the thing English speakers express with the word “knows,” and morality just is the thing English speakers express with the word “ought,” and so it’s no wonder that there is a preestablished harmony between their natures and our intuitions. For the philosopher who is skeptical of the possibility of completely a priori normativity (whether about belief or action) or about the existence of any a priori concepts, this seems like an entirely reasonable view.

But this is emphatically not what the average Standard Realist about morality or knowledge wants to conceive of themselves as doing.⁸ They think these concepts are real and significant independently of the fact that the words for them play an important role in our language. The choice to worry about the good instead of some other possible concept that could apply to action is not supposed to be determined by accident; the good intrinsically matters to any rational being, or at least it should, in a sense deeper than merely saying “the good is good.”

Look. The significance of your conclusions cannot go any further than the scope of your justification for them. If the reason our diamond-truther is wrong is because (in a sense) he is using words in an unconventional way, then what we say to him has no more significance than does the claim “The things that people paradigmatically refer to as ‘diamonds’ have a cubic structure,” as semantically different as that claim may be from “Diamonds have a cuboid structure.” You can’t get things for free by going down a level. Likewise, if paradigm cases of immoral action are by themselves to in-principle constraints on what can be believed to be immoral, then the Lover of Wisdom’s conclusions cannot have any more significance than does the claim “The actions paradigmatically referred to as ‘immoral’ have such-and-such properties.” The view on which intuitions intrinsically, and not as a matter of empirically verifiable correlation, point towards truth is necessarily an antirealist one.

---

I don’t think it can be overemphasized how set apart I am from most contemporary philosophy by viewing the evidence from intuition as merely empirical, and by viewing as indifferent theories that predict our intuitions (whether they vindicate them or not) equally well, all else equal. This problem is everywhere. To name a few more places: the idea of “thick moral concepts” as being a problem for theories that cleanly separate the descriptive and the normative; the demandingness objection to consequentialism; dismissal of skepticism about weakness of will; opposition to nominalism; and so on.

Really, so much of the contemporary discourse is shaped by the fact that people are allowed to give easy answers to hard questions, expecting that the mere fact that they say things are as they seem outweighs the a priori implausibility of their views. To a large extent, when it comes to foundational matters regarding normativity and metaphysics, your job is to either formulate skeptical arguments against views that hardly had any actual support in the first place but which are given an arbitrary advantage, or else address such arguments.

1. I also think, as a matter of fact, that intuitions do better than chance when it comes to moral claims, though most philosophers are not in a position to justifiedly believe this. The reason they are better than chance is because the processes that formed our intuitions weakly track universalizability.

2. See his Ethics: Inventing Right and Wrong.

3. Sure, they do have the evidence of lacking an intuition. If you think this is a crux, think carefully about whether things would appear any different before reflecting in a manner that makes your intuitions manifest.

4. “But,” you say, “you are appealing to knowledge in order to provide this explanation of our intuitions.” So? Non-skeptics believe evolution is true, and so if their view of the world is one on which their intuitions do not provide evidence against skepticism, then the skeptics win.

5. Okay, we can argue about whether intuitions are the culprit here, but it’s the same problem.

6. This is another example of how Lovers of Wisdom prevent objections to their views by making them inarticulable.

7. Properly, what I should say is that “diamond” refers to the concept dthat[λx.x is the same kind of thing as what English speakers paradigmatically refer to as diamonds]. This explains why there’s a preestablished harmony between intuition and facts about diamonds without making any apparent level confusions.

8. Usually, at least. But Scanlon, in Being Realistic about Reasons, seems to be quite comfortable portraying himself as merely talking about accidental conventions (or “frameworks”).