On 'Moral Facts Naturally Exist (And Science Could Find Them)' (Pt. 2)

In my last post, I sketched Richard Carrier's moral philosophy. Today I will explain my primary reservation, but first I want to point out several areas of agreement. Like Carrier, I...

• believe morality is concerned with hypothetical imperatives.
• accept a Humean theory of reasons, i.e. what a person has reason to do is dependent on that person's psychology.
• accept the theory of action that a rational person will always try to fulfill her highest-priority desires, according to the information she has.
• agree that science — broadly construed — is vital in finding out the rightness or wrongness of an action.

This definitely puts us in the same neighborhood of metaethics-ville. Carrier characterizes moral facts as objective and his view as realist (by denying the contraries); I quibble with that, but only because I would label the same things differently. Not a big deal.

'Ought' and Internalism

When Carrier defends his view that moral imperatives are a class of hypothetical imperatives, he admits this is an unpopular view among philosophers. "But," he says, "none have ever presented any other identifiable logical relation that can ever be meant by 'ought' (or any other term or phrase semantically equivalent to it) that produces any actual claim to our obedience."¹

This close association of morality, the meaning of 'ought,' and motivational internalism rests at the very beginning of Carrier's chain of deductive logic in the appendix following the chapter. Here are the first three lines, with variables expanded:

1.1 If there is <a moral system>, then <a moral system> is <a system of imperatives that supersede all other imperatives>.
1.2 If <a moral system> is <a system of imperatives that supersede all other imperatives>, then <a moral system> is <what we ought to obey over all other imperative systems (whether they are labeled moral or not)>.
1.3 <what we ought to obey over all other imperative systems (whether they are labeled moral or not)> is <that which we have a sufficiently motivating reason to obey over all other imperative systems>.²

Throughout the chapter, Carrier uses the phrases "what we in actual fact ought to do" and "what we as a matter of actual fact ought most to do" as synonyms, and contrasts this with "other things that carry no sufficient motivating reason for us to do them instead".³ You may recognize this as a strong form of motivational internalism, i.e. recognized moral facts necessarily provide some motivation or — in strong form — overriding motivation.

I think Carrier has a good point that if we start by insisting on internalism, then it's hard to see how moral facts could originate from anywhere but a person's own desires; and if we insist on strong internalism, how they could originate from anywhere but what a person desires most. Or consider a reasons-based version of internalism: a person always has some reason or overriding reason to act morally. If having a reason requires having some appropriate desire — which I affirm — then we're back to the same spot.

'Ought' Externalism

Contrary to Carrier, I hold that sentences like "Michael ought to contribute to UNICEF" or "Josephine ought not fire her pistol into the air when she celebrates" can represent true propositions even if Michael and Josephine happen to lack appropriate desires.

This means I deny (1.3). I'll make this denial punchier: it can be true that we have no reason to do what we ought to do.

How can I get away with saying this? Because I believe the word 'ought' requires an end (or goal) to complete its meaning and make it eligible for being true or false. At the same time, it doesn't require that anyone's desires be a certain way. The logical relation signified by 'ought' works something like this:

Michael ought_{[some end]} to contribute to UNICEF.

or more specifically:

In order that [some end], it ought to be the case that Michael contributes to UNICEF.

The claim being made is that — among the relevant actions open to Michael — the one most likely to precede [some end] is that he contributes to UNICEF. (The 'ought' in the more specific parsing is a non-normative probability 'ought,' like "It ought to rain before midnight." I'm following Stephen Finlay's reductive analysis of normative 'oughts' into non-normative 'oughts' plus ends, which is motivated by making sense of normative language in general.)⁴

Really, though, I just want to drive home the point that 'ought' claims have a gap if you listen for it.

We're normally very adept at filling the gap from context and so we don't notice there ever was a gap. For example, "You ought to eat two cups of green vegetables per week" in typical contexts would suggest a health-related end. In a conversation about minimizing risk for liver cancer, we would fill in the more specific end of minimizing risk of liver cancer. At that point, we have a quite specific claim which is open to empirical investigation.

You ought_{that you minimize your risk of liver cancer} to eat two cups of green vegetables per day.

or

In order that [you minimize your risk of liver cancer], it ought to be the case that you eat two cups of green vegetables per day.

Notice something else: the truth or falsity of this 'ought' claim does not depend on having actual or ideal desires about minimizing the risk of liver cancer.

Laying Claim to Our Obedience

What I'm saying is that true 'ought' statements don't necessarily lay claim to a person's obedience. Some do, because they connect with a person's desires, and this makes them the only imperatives important to that person, in a relevant sense.

I understand the phrases "there is a reason" and "Josephine has a reason" to reflect this distinction. There may be a reason for Josephine to not fire her pistol in the air when she celebrates (it might cause far more suffering than the joy she gains), but if she lacks certain desires she might not have a reason to refrain from pulling the trigger.

Carrier could grant all of the above, adjust his argument a bit, and still identify moral imperatives as imperatives which are both (1) true and (2) matter to a person by virtue of that person's desires. What I'm challenging in this post is the assertion that what a person "in actual fact ought to do" necessarily corresponds with what that person has motivating reason to do.

In other words, Carrier can't simply rule out other (i.e. externalist) uses of 'ought' as invalid. He needs to show that his moral theory is a better solution to metaethics in some way other than winning by default.

...

I may eventually follow up this post with my take on other parts of his overall moral theory, but this will do for now.

ADDED:  A followup on the same topic is here.


1. Carrier, R. (2011). Moral facts naturally exist (and science could find them). In Loftus, J.W. (Ed.), The end of christianity (pp. 333-358). Amherst, New York: Prometheus Books. p. 342
2. Ibid. p. 359
3. Ibid. p. 348
4. Finlay, S. (2009). Oughts and ends. In Philosophical studies, 143(3). pp 315-340. See my post on it.

On 'Moral Facts Naturally Exist (And Science Could Find Them)' (Pt. 1)

This chapter by Richard Carrier comes at the end of The End of Christianity. So far, it's the only chapter I have read because I'm much more interested in popular essays on metaethics than the (anti) religious theme of this anthology.

Can science discover moral facts?

Carrier is following Sam Harris' lead by putting this provocative claim in the title, and — like Harris — doing some philosophy first to analytically reduce moral facts to scientifically-accessible components. So the really controversial steps are philosophical rather than scientific.

I'm familiar with moral philosophy. Give me the short version!

Moral facts are whichever hypothetical imperatives correspond to an individual's deepest desires. Human-universal moral facts exist because all humans share a set of deepest desires. Since science can investigate both hypothetical imperatives and desires, science can discover human-universal moral facts.

What makes 'ought' claims true?

Let's start by looking at conditional 'ought' claims (aka hypothetical imperatives).

If you want to wake up in time for work, you ought to set an alarm clock.
If you want your car to stay put, you ought not park in a tow-away zone.
If you want to become a doctor, you ought to study diligently.

So if I really do want to become a doctor and there really is this connection between studying diligently and becoming a doctor, then I really ought to study diligently. Carrier points out that both of these prerequisites are open to scientific investigation. "And wherever both are an empirically demonstrated fact, the imperative they entail is an empirically demonstrated fact."¹ This means that science can discover 'ought' facts, not just 'is' facts.

What makes an 'ought' claim a moral 'ought' claim?

The majority view has been that moral 'oughts' are different from the above kind of 'oughts' because they're not conditional on what a person wants. Carrier disagrees on the grounds that any system of imperatives which doesn't line up with what a person most wants can't count as morality, because that person will "have a better reason to do something else instead."²

Instead of viewing morality as something that stands in opposition to our desires, morality has to do with what fulfills our deepest desires. It's just that, sometimes, we're mistaken about what promotes our own deepest desires. "What we really want most, and what will really obtain that, are matters of fact that cannot truly be answered from the armchair. Empirical methods must be deployed to ascertain and verify them. Only science has the best tools to do this."³

Doesn't this make morality an individual thing?

Even if morality is grounded on the individual level, there may still be universal moral facts if some moral facts apply to every individual. (Or at least human-universal moral facts if all humans share some moral facts.) Carrier argues that it's likely all humans have the same set of deepest desires.

"Only if what an individual wants most (when rational and sufficiently informed) is not the same as for everyone else will this not be the case. Then, a different set of moral facts will be true for them (yet even then true moral facts still exist, they are just again relative to different groups or individuals.) But that outcome is very improbable for members of the same species."⁴

Carrier's justification for this statement is hard to follow, but it goes something like this:

Humans share many biological facts, and these facts generate a hierarchy of high-order desires that we're stuck with, i.e. we can't just change them without altering our natural humanity. "For example: we all need to eat, breathe, move, think, and cooperate and socialize with a community[....]"⁵ The way these fundamental desires play out for individuals may differ, but we share our most basic biological needs.

Biological differences among humans aren't sufficient to change these high-order desires, at least not without extreme genetic mutation.

Environmental differences only make a difference in how our fundamental human desires play out. Same algorithm; different results. If I had lived life in your shoes, I would want most the things you want most.

So, ultimately, only our shared human biology determines our high-order/foundational/deepest desires, which in turn determine what is morally right for all of us in general terms, and what is right for each of us when applied to our individual situations. To use one of Carrier's examples, we all have fundamental desires to eat and to avoid pointless harm, which might make it morally right for me to eat strawberries but morally right for you to avoid eating strawberries because you're allergic to them; we're both following the same basic imperatives of eating and avoiding pointless harm, which means there is no real difference in moral facts here.

...

To review, Carrier believes moral facts are open to scientific inquiry because his metaethics reduce moral facts to facts about the effectiveness of means to ends (hypothetical imperatives) and psychology (what a person most fundamentally desires). This would be enough to explain how moral facts "naturally exist" and how "science could find them," but he goes one step farther and argues that humanity shares one set of moral facts.

Carrier's chapter is followed by an appendix containing, as he puts it: "formal deductive proofs of every one of these conclusions, fully verifying that they are necessarily true."⁶ I appreciate his boldness! Still, I disagree with his moral philosophy at several points, as I will explain in the next post [which is here, but only covers one point of disagreement].


1. Carrier, R. (2011). Moral facts naturally exist (and science could find them). In Loftus, J.W. (Ed.), The end of christianity (pp. 333-358). Amherst, New York: Prometheus Books. p. 335
2. ibid. p. 343 
3. ibid. p. 342 
4. ibid. p. 351 
5. ibid. p. 352 
6. ibid. p. 334

Words and Things without Ideas

How do words and phrases like "Mars," "the first President of the United States," and "Dante Shepherd" mean what they do? A popular first answer is that they mean what they do because they each stand for a particular thing. The notion these kinds of words are just labels for things is called the Theory of Reference for definite descriptions.

Following William Lycan's Philosophy of Language: a Contemporary Introduction, I will describe four puzzles which make the Theory of Reference appear untenable.

Puzzle #1 — Apparent Reference to Nonexistents

Zaphod Beeblebrox is two-headed.

How can "Zaphod Beeblebrox" have meaning by standing in for something, if that 'something' doesn't exist? Lycan puts this intuition more explicitly (paraphrasing a bit):

I. The sentence above is meaningful.
II. The sentence above is in subject-predicate form.
III. A meaningful subject-predicate sentences is meaningful only because it refers to something then ascribes a quality to it.
IV. "Zaphod Beeblebrox" doesn't refer to anything that exists.
V. Given II through IV, the sentence above isn't meaningful...or it refers to something that doesn't exist.
VI. If something doesn't exist, it's impossible to refer to it.

One of the points from I to VI must be false. (Which would you challenge?)

Puzzle #2 — Negative Existentials

The Lemurian civilization never existed.

Assuming this is true, how could it be a true, meaningful sentence under the Theory of Reference? The sentence explicitly denies the existence of the very thing it is supposedly pointing at and talking about, so to speak.

Puzzle #3 — Identity

Howard O'Brien is Anne Rice.

If the Theory of Reference were correct, this should be a very boring, trivial, non-informative kind of statement along the lines of the equation x = x. There sure seems to be more information here than that! Probably even a commentary on her parents' psychology.

Puzzle #4 — Substitutivity 

Debra Morgan knows that the Bay Harbor Butcher dismembers bodies.

Since Dexter Morgan is the Bay Harbor Butcher, it should be possible to substitute referentially equivalent words without changing the meaning.

Debra Morgan knows that Dexter Morgan dismembers bodies.

But she doesn't know this, at least not in the episodes I've seen. The two sentences obviously have very different meanings if one is true and the other false! (Leaving aside the question of whether facts about fictional characters are genuinely true, since real life cases are possible.)

Now What?

The Theory of Reference is nice and simple, but doesn't work out. There are other theories which do a better job of capturing what we mean by 'meaning.' I don't think any of them has universal support among philosophers (what does?). And of course there's always the possibility our notion of 'meaning' isn't neat, so we can't ever capture it neatly.

On 'Slaves of the Passions'

Mark Schroeder's book, Slaves of the Passions, defends the idea that all reasons are dependent on the psychological features of those to whom they are reasons.

To use his central example, picture Ronnie (who likes to dance) and Bradley (who doesn't even like to be around dancing). Both are invited to a party where dancing will definitely be going on. If you agree the fact that there will be dancing at the party is a reason for Ronnie to attend and a reason for Bradley to avoid attending, then you're on board with the intuition that at least some reasons are explained by psychological differences.

The controversial move is to try explaining all reasons in the same way. In its most general form, Schroeder characterizes this Humean Theory of Reasons¹ as a "parity thesis," which only insists that whatever the difference is between Ronnie and Bradley that constitutes a difference in reasons...that's what constitutes a difference in all reasons. Specific versions of the Humean Theory of Reasons (HTR) may advance different theories about what the difference actually is.

Why make a big deal out of distinguishing general from substantive theories?

1. Past critics of the HTR have tended to challenge some substantive theory while believing they were challenging the general theory.
2. Schroeder offers his own substantive theory, Hypotheticalism, which supposedly dodges historical criticisms. But of course he doesn't want anyone to think discrediting Hypotheticalism would automatically discredit the HTR.

Well, Not THOSE Reasons

The HTR is concerned with normative reasons rather than explanatory reasons.

Explanatory reasons — The reason my car won't start is that it has a dead battery.
Normative reasons — Graduate school requirements are a reason to maintain a high undergrad GPA.

Schroeder makes a further distinction within normative reasons, producing what he (somewhat reluctantly) labels objective normative reasons and subjective normative reasons. This distinction is based on whether the reason applies whatever a person's beliefs may be (objective) or applies because of a person's beliefs regardless of what is actually true (subjective).

Suppose I take a pill I believe is aspirin, but is actually poison. I probably don't have an objective normative reason to take the pill, but if I have a headache it seems right to say I had a reason of some kind to do so. Subjective reasons, Schroeder suggests, are normative reasons which would be objective normative reasons if the person's relevant beliefs were true.

So the two basic kinds of reasons turn out to be explanatory reasons and objective normative reasons, with the HTR aiming at the latter category.² It would be nice to unify these two senses, as Stephen Finlay does for normative and non-normative senses of 'ought,'³ but Schroeder points out difficulties rather than a solution for doing this.⁴

Winning the Intuition War

Or rather, not losing the intuition war. Slaves of the Passions is primarily about rescuing the HTR from disrepute by showing that supposedly knock-down arguments are weak or off-target. I may examine a few of these arguments in the future.

Today I want to take a step back and ask, "What makes a theory of reasons a good theory?" Avoiding incoherence is a start, but philosophers are exceptionally skilled at figuring out how things aren't necessarily incoherent. It's a low bar.

I'm tempted to say that a good theory of reasons provides the simplest possible explanation of our talk about 'reasons.' This is the sort of thing Paul Ziff did in his book Semantic Analysis when he analyzed 'good' as meaning "answering to certain interests."⁵ Schroeder's chapter on normative reduction makes it clear he's trying to say something about the metaphysics of reasons, not just something about English and other languages which might have similar semantic patterns. And a vital way to test the metaphysical correctness of theories about reasons is to check the intuitive data.

For to the extent that a Humean is willing to admit to accepting results that are intuitively false, other philosophers are going to legitimately infer that he has simply changed the subject, and is talking about something else entirely.⁶

This plays out in a pair of chapters alleging that the HTR wrongly calls things reasons which are not reasons...and wrongly denies that things are reasons which are reasons. For example, Schroeder's substantive theory generates the seemingly ridiculous result that I have a reason to eat my car. This would seem to put him in a bind. How can he make make intuitive acceptability a necessary criterion for theories of reasons, yet support a theory with such an unintuitive result?

What I will now do is explain why our negative existential intuitions about reasons are prone to be misleading in this way. The explanation comes in two steps, each of which yields a testable empirical prediction. So I'll then proceed to test these predictions.⁷

I was excited; empirical predictions! No need to rely on intuitions about a metaphysical issue. ...except the empirical tests turned out to be tests of intuitions.

Prediction one: When Schroeder explains that the reason for me to eat my car is that it "contains the recommended daily allowance of iron," it becomes somewhat less unintuitive.
Prediction two: When Schroeder explains that this reason is a terribly weak reason, the unintuitive level drops again.

The lesson here is that initially unintuitive results can be ok if reflecting on them can result in their looking not-so-bad after all. This is probably a good approach to take with people whose big hangup with the HTR is based on on negative intuitions. I don't think it appeals as much to people like me who have an intuitive affinity for the HTR but don't appreciate intuitive 'tests' of metaphysics. Then again, I'm not the sort of reader Schroeder needs to worry about.

Hypotheticalism

I may as well sketch the substantive version of the HTR featured in Slaves of the Passions, to keep anyone curious about that from feeling completely teased by the tangent I took in this post.

Revisiting the Ronnie and Bradley scenario, the fact that there will be dancing at the party is a reason for Ronnie to go to the party because (1) Ronnie desires that he dance and (2) the truth of the proposition there will be dancing at the party helps explain why Ronnie going to the party promotes Ronnie dancing. And if (1) or (2) didn't hold, then the fact that there will be dancing at the party would not be a reason for Ronnie to go. (As is the case for Bradley.)

Reasons depend on a person's desires and external facts about the world. It's important to notice that the only reason identified above is the fact of dancing at the party. This sets Hypotheticalism against other versions of the HTR which count desires as reasons for action (except in special cases where desires fill the role of the fact of dancing in the example above).

Although Hypotheticalism is intentionally constructed as an 'existence proof' of an intuitively viable form of the HTR, Schroeder usually goes back to the dancing scenario and argues that Hypotheticalism turns out to provide a more intuitive explanation of the Ronnie/Bradley difference than other versions of the HTR which are vulnerable to historical criticisms.

I'm definitely on board with the general HTR, and think the basic Hypotheticalist template is about right. Except by 'right' I mean as a concise explanation of how 'reasons' language is actually used, or as a minimally cleaned up version to achieve coherence. Schroeder's applications of Hypotheticalism to moral theory strike me as completely off the wall, but I realize he's trying to show that the range of Hypotheticalism is broad enough to allow for surprisingly Kantian strategies. So it's probably just my usual boggling at Kantians at work there.

1. In Book 2, Section 3, Part 3 of A Treatise of Human Nature, David Hume wrote: "Reason is, and ought only to be the slave of the passions, and can never pretend to any other office than to serve and obey them." This passage inspired so-called Humean theories of reason and, obviously, this book's title.
2. I skipped over the category of 'motivating' reasons which Schroeder takes to be a combination of explanatory and subjective normative reasons. See p. 12.
3. See http://wordsideasandthings.blogspot.com/2010/12/on-oughts-and-ends.html
4. Schroeder, M. (2007). Slaves of the passions. New York, New York: Oxford University Press. p. 35-37.
5. Ziff, P. (1967). Semantic analysis. Ithica, New York: Cornell University Press.
6. Schroeder. p. 86
7. Ibid. p. 94