Normative principles and the nature of mind-dependence

Abstract

One of the most fundamental debates in metaethics is whether (subsets of) the normative facts are mind-dependent. Yet some philosophers are skeptical that mind-dependence is a category that's significant in the way metaethicists have assumed it is. In this paper, I consider a puzzle that showcases this skepticism, explaining how it undermines the most natural reading of the mind-dependence claim. I then go on to show that no modification of this reading within a certain class can hope to solve the problem. I conclude by suggesting a new way that mind-dependence should be understood: mind-dependence is ultimately a matter of how normative principles are grounded. I develop this view briefly before concluding.

Introduction

One of the most fundamental debates in metaethics is whether (subsets of) the normative facts are mind-dependent. Yet some philosophers are skeptical that mind-dependence is a category that’s significant in the way metaethicists have assumed it is. In what follows, I’ll consider a puzzle that showcases this skepticism, explaining how it undermines the most natural reading of the mind-dependence claim. I’ll then go on to show that no modification of this reading within a certain class can hope to solve the problem. I conclude by suggesting a new way that mind-dependence should be understood: that mind-dependence is ultimately a matter of how normative principles are grounded.

1 A puzzle

Socrates famously considered whether things are pious because they are loved by the gods.¹ Hamlet famously (and ironically) claimed that things are made good or bad by thinking. In a more technical vein, Sharon Street argues that the fact that A has a reason to Φ depends on A’s evaluative judgments,² and Mark Schroeder thinks that all of an agent’s reasons are explained by that agent’s desires.³ These views all seem to share a common feature: in some as-yet-unspecified sense, some set of normative facts depends on some agent’s or agents’ mental states or events (“mental states” for short).

Yet there are those who disagree, arguing that, in some important sense, the normative does not so depend on the mental—or at least not always. Russ Shafer-Landau summarizes this as the view that “moral principles and facts are objective in a quite strong sense: they are true and exist independently of what any human being, no matter his or her perspective, thinks of them.”⁴ Similarly, Sarah McGrath formulates this stance as follows: “Which moral judgments are true does not depend on what we (either individually or collectively) accept.”⁵ While these formulations and others vary as to whether “thinking” or “acceptance” or some other mental state is relevant—David Enoch thinks it is “attitudes” in general⁶—they form a pattern of denying the dependence of the normative on the mental. (We’ll soon see why some are motivated to be more specific about the kind of mental state involved in this peculiar form of dependence.)

This debate is often described as the debate about the mind-dependence of (sub-domains of) normative facts: the former camp endorse such mind-dependence, while the latter deny it. And furthermore, normative realism is often taken to turn at least in part on the issue of mind-dependence: realists are often taken to be committed to mind-independence (though this has not gone without challenge).⁷

Some philosophers have worried about the mind-dependence/independence distinction.⁸ I won’t rehearse their worries here; suffice it to say that no one has developed a view of the distinction that avoids the problems. Yet a very natural thought is The Straightforward View:

The Straightforward View::

Normative facts are mind-dependent just in case they depend on mental facts.⁹

However, this obviously won’t do. First, we don’t want a view to be mind-dependent just in virtue of depending in any sense on mental facts. Perhaps all facts about artifacts depend causally on intelligent agents’ intentions, etc.—facts about artifacts aren’t therefore mind-dependent in the relevant sense.¹⁰ Second, consider a view on which what’s wrong depends on what agent A desires, but where A’s desires depend on some fundamental moral facts, such as facts about what the good desires are. Such a view, while allowing that wrongness depends on mental states, seems out of place among views like Street’s and Schroeder’s, which are views about the ultimate or fundamental explanations of normative facts.

So we should qualify the Straightforward View:

Qualified Straightforward::

Normative facts are mind-dependent just in case they asymmetrically and non-causally depend on mental facts.

A normative fact asymmetrically depends on mental facts just when it depend on mental facts and those mental facts don’t depend on any normative facts. I’ll refer to this non-causal dependence as “grounding,” for ease of use. However, I don’t want to commit to this grounding being of the metaphysical sort of so much recent discussion, since some have argued that there are multiple fundamentally distinct kinds of grounding, and nothing I say here relies on any particular kind.¹¹ However, later in the paper, I will consider a proposal that does make ineliminable use of different kinds of grounding, at which point I will distinguish the two. For now, let “grounding” be univocal and general dependence, which I treat also as a kind of explanation (though nothing I say here depends on it).¹² Thus I use “X grounds Y,” “Y obtains in virtue of X,” and “X explains Y” interchangeably in this paper.

The puzzle here is that many first-order moral theories seem to posit just such a dependence on mental facts: utilitarians claim that rightness/wrongness facts depend on facts about happiness. Or consider a view on which what determine rightness/wrongness are facts about well-being, where this is understood purely psychologically. Or a view on which one’s intentions determine the rightness/wrongness of one’s actions. The “dependence” or “determination” in these cases is certainly not causal. And in each case, the dependence is asymmetrical in my sense: facts about (e.g.) happiness don’t depend on moral facts (on the utilitarian’s understanding of happiness, at least).

Qualified Straightforward thus seems to entail that these are all mind-dependent theories about their respective normative sub-domains. But utilitarians shouldn’t, just in virtue of their utilitarianism, be cast from the ranks of the moral realists (and mutatis mutandis for the other first-order theories). The first-order theories above do not entail any view concerning mind-dependence.

But before considering solutions to this puzzle, consider one constraint they must meet: any account of mind-dependence needs to be able to accommodate certain kinds of mundane dependencies in different normative sub-domains. They are mundane in the sense that all (non-error-theoretic) views should be able to accommodate claims like them, regardless of whether such views endorse or deny mind-dependence. Here are a few examples: this painting’s symmetry makes it more beautiful; that the evidence is in favor of P grounds that we ought to believe P; this act of infidelity was wrong because it hurt someone so deeply; Euthyphro’s action is impious because he is prosecuting his father. Now, the point is not to give examples the reader finds plausible—how could she, with so little information about the context?—but to give examples of the kind of claims I’m interested in. If the reader is not an error theorist, she will find some such claims undeniable. Call this constraint the Mundane Dependence Constraint.

There are two things to note about the Mundane Dependence Constraint. First, in only some of these examples does the normative depend on the mental—this will be important later. Second, at least in some cases, it seems hard to get away from the idea that such claims are not claims about full explanation (or dependence, or whatever). E.g., we could have the full grounds of why we ought to believe P in the fact that the evidence is in favor of P.

In what follows, I’ll have various kinds of objections to make to the views I reject, but one feature many of them share is that they fail the Mundane Dependence Constraint. The only ways to avoid this problem are to be a certain kind of pluralist about kinds of dependence—so that normative facts can depend fully on the mental in one sense but the mundane dependencies can obtain in another—or to endorse my view. It shouldn’t surprise anyone that I’ll argue that we should take the latter path.

2 Modifying the grounds

In this section I want to explore one natural type of attempted solution to the above puzzle. I call members of this class “Contents of Grounds (CoG)” solutions, because they attempt to avoid the puzzle above by denying that just any mental facts amongst a normative fact’s grounds to any degree make that fact mind-dependent. Instead, they claim that we need to get more precise about the nature of the mental states in the grounds of a fact before we can tell if that fact is mind-dependent. There are other kinds of solution—e.g., those that draw the distinction by the kind of dependence involved—that I’ll explore in the next section. But here I’ll just argue that CoG solutions fail, so some other way of drawing the distinction is necessary, if it’s to be considered a legitimate distinction.

2.1 Kinds matter

Perhaps the problem with the Straightforward View is that it is not specific enough about the kinds of mental states that matter for mind-dependence. Consider:

Kinds Matter::

Normative facts are mind-dependent just when they are asymmetrically grounded in some particular kind K of mental state.

That view is really a schema, since it doesn’t specify the dependence-inducing kinds of mental states. Perhaps a view is mind-dependent, for example, just when the normative facts depend on desires. Utilitarians then would get to avoid mind-dependence (assuming happiness can be understood independently of desire).

Some explicitly consider Kinds Matter, as Rosati does: “Does it make a difference to the plausibility of mind-independence characterizations of moral realism whether claims of mind-independence concern, say, independence from emotions, desires, or responses as opposed to judgments?”¹³ Others are slightly less explicit. Sharon Street, for example, says K is our values, or evaluative judgments, while Karl Schafer says it is “judgments or values of individuals or communities.”¹⁴ Shafer-Landau thinks mind-dependence (“stance-dependence” in his words) obtains when a moral standard is made true by being ratified from within some perspective.¹⁵ These authors, I propose, are aware of problems like I’ve presented above and are trying to preempt them via Kinds Matter. (This is explicit in Shafer-Landau: his formulation follows consideration of a problem like mine, above.)

Of course, there’s one problem right off the bat: what we wanted was an account of mind-dependence on which it was left open whether utilitarians were mind-dependence theorists, not one which entailed that utilitarians are not mind-dependence theorists. In fact, any Kinds Matter view will commit the utilitarian one way or the other on the mind-dependence issue. That is, for any instantiation of K, K either will or will not include the kind of happiness facts the utilitarian posits as the unique non-normative grounds for moral facts.¹⁶ But then, once we get our instantiation of Kinds Matter, utilitarians will be determinately committed to one answer over the other on the question of mind-dependence. And as I’ve said, that’s not what we wanted: utilitarians are not mind-dependence or -independence theorists just in virtue of their utilitarianism.

But let’s assume advocates of Kinds Matter can get past this, and past the further difficulty of giving a specification of K that is not counterexample-able (for the above view, counterexamples could consist in, e.g., desires grounding normative facts without being mind-dependent, or mental states other than desires giving rise to mind-dependence). Even then, they’ll have a deeper problem. To justifiably identify one kind of mental state as the kind that uniquely engenders mind-dependence requires an explanation of why that kind is special in this way. To see why, imagine someone, by clever philosophical concoction, gerrymandering a kind of mental state that avoids all counterexamples. For example, take all the particular mental states S₁-S_N such that, when a normative fact depends on some S_X, this fact seems mind-dependent. Now define a kind of mental state K_S such that one has a K_S mental state if one has some S_X. We can then say that any normative fact depends on a K_S state, it is mind-dependent. Such a view would (by design) avoid counterexamples like those above, but is not philosophically satisfying: we don’t (yet) know what is it about K_S states that makes them special, such that when a normative fact depends on a K_S state, it is mind-dependent. Call this “the significance question”. Without an answer to the significance question, a Kinds Matter view will be dissatisfying even despite achieving extensional adequacy.

The problem generalizes beyond gerrymandered kinds. For any given kind of mental state, no explanation of why that kind is dependence-inducing seems forthcoming. Why, for example, should we get a controversial metaethical view when normative facts depend on beliefs (“It’s wrong because I think it’s wrong”) but not when they depend on desires (“It’s wrong because I want it to be wrong”)? The same question could be asked about any purported dependence-inducing mental state. And no answer seems forthcoming, for beliefs or any other kind of mental state. Defining mind-dependence in terms of the kinds of mental states in the dependence base just doesn’t seem to capture the importance of the mind-dependence versus mind-independence distinction.

2.2 Agents matter

Perhaps it’s not the kind of mental state that matters, when distinguishing mind-dependence from mind-independence, but which agent has the mental state. Of course the morality of my action depends on the mental states it engenders in you: whether we’re realists or not, its wrongness depends on its causing you pain. But once we say that my action of causing you pain is wrong because it engenders some mental state in me, that’s when wrongness becomes mind-dependent (at least in this instance). So consider:

Agents Matter::

Normative facts are mind-dependent just when they are asymmetrically grounded in certain agents’ mental states, but not others.

Again, the view is a schema, and the relevant agents could be defined in a number of ways.

However, according to the extant (and most natural) motivations for Agents Matter, normative facts are mind-dependent when they are grounded in the agent’s own mental states. E.g.: Φing is wrong for A because of A’s mental states. This is the most natural way of developing Agents Matter because it is the only way that has a clear answer to the significance question: of course the normative facts for A should depend on A’s mental states. This is, I propose, why so many reasons internalists have relied on this very motivation.¹⁷ It could similarly explain the appeal of views like Schroeder’s Humean Theory of Reasons, according to which all an agent’s reasons are explained by that agent’s mental states.¹⁸ Just as, for these theories, the agent’s own mental states seem particularly apt for grounding reasons, so too the mind-dependence theorist could think that mental states of others are too “alienated” (to use Railton’s phrase¹⁹) to engender mind-dependence.

Unfortunately, this answer to the significance question seems to come at the cost of extensional adequacy: there are cases where a normative fact (for A) depends on A’s mental states, but where this normative fact does not seem mind-dependent. Duties to self will often meet this description: I have a duty not to shoot myself in the foot, which depends on the pain it would cause me. Or consider the utilitarian’s verdicts for an agent in a world where she alone is capable of happiness: all moral verdicts will depend on her mental states. Finally, consider the simple view on which moral verdicts for A depend only on A’s intentions. None of these views seem inherently committed to the mind-dependence of the relevant phenomena, yet all claim that those phenomena depend (asymmetrically and non-causally) on the agent’s own mental states.

Similarly, it seems like normative facts can be mind-dependent because they depend on mental states of agents other than A: if it is wrong for A to Φ because A’s culture judges that Φing is wrong, wrongness certainly is mind-dependent. So we should say of Agents Matter that it can only answer the significance question if it fails at extensional adequacy on multiple fronts.

2.3 Completeness matters

Finally, it might seem like what really distinguishes mind-dependence from mind-independence is not the kinds of mental states the normative facts depend, or whose mental states they are, but rather how complete of an explanation the mental states can give of the moral facts. That is:

Completeness Matters::

Normative facts are mind-dependent just when they are asymmetrically and fully, not merely partially, grounded in mental states.²⁰

In other words, a normative fact F is mind-dependent just when:

Here, the solid line represents a full grounding relation, and “M” represents some fact(s) about mental states. Similarly, F is mind-independent just when:

Here, “P” is a non-mental state fact, and the dotted lines represent (mere) partial grounding, where for A to (merely) partially ground B is just for A and some other non-A facts to fully ground B.

Completeness Matters has some initial plausibility. It would make sense of the intuitive importance of the debate if what was at stake was whether something other than mental states had to be brought in to explain the normative facts. And, while I’m here concerned with the distinct debate over mind-dependence versus -independence, this way of dividing things up also tracks well with how some have defined non-naturalism: on such views, non-naturalists’ core claim is that the normative is not fully grounded in the non-normative.²¹ (Note, however, that there are dissidents to this strategy: Stephanie Leary presents an argument against it, as well as an alternative way of understanding the naturalism vs. non-naturalism debate.)²²

Completeness Matters, too, has problems. Let’s look at the space it carves out for mind-independence: normative facts that are not fully grounded in the mental, but in something non-mental in addition (“P”). And let’s assume that Completeness Matters only passes muster if mind-independence comes out legitimately possible—otherwise the view doesn’t do its primary job, which was making sense of a distinction. Below I’ll argue that Completeness Matters fails, because while P must be a normative principle, Completeness Matters cannot allow the conceptual space for mind-independence without endorsing a false view of normative principles.

It is most natural to think of whatever non-mental stuff goes in the grounds as a principle—hence “P”—which connects the mental facts to the normative facts—e.g., “If Φing results in desire-frustration, then it is bad.” After all, if P is not a principle, it needs to be either another particular normative fact or some kind of non-normative fact. If it is another particular normative fact—call it F*—then we’ve just moved the bump in the carpet: we don’t know whether F is fully grounded in mental facts until we know how F* is grounded.²³ But since (plausibly) no particular normative fact is ungrounded,²⁴ this view entails that any mind-independent particular normative fact will have an infinitely long grounding chain. Each mind-independent particular normative fact is partially grounded in a further (mind-independent) particular normative fact.

Of course, if we turn to something non-normative to stand in for P, it can’t be a fact concerning mental states—otherwise we get full grounding of F in the mental, which would make F mind-dependent. But what, then? I can’t see any plausible answer here. This is not because non-mental yet non-normative facts never seem normatively relevant—think of the Mundane Dependence Constraint. That the probability of P is very high is a reason to believe it. That Φing would save a life is a reason to Φ.

But the problem is that these do not seem like the kind of facts that could complement mental state facts in a grounding explanation. That is, while both “I wanted to Φ” and “Φing saved a life” can seem somehow relevant to the fact that Φing is good, they do not seem like they could combine to constitute the full grounds of the moral fact. They seem instead to be either competing explanations within the same type, or explanations of fundamentally different types, answering different sorts of question about the fact that Φing is good.

The natural view then, if we assume Completeness Matters, is that P is a normative principle. In the case at hand, we get something like this:

Put off worries about the proper form of the principle, which I’ll address later. In contrast to the former case, the principle here does seem like the kind of thing that could combine with the mental facts to produce the full grounds of the moral fact. And, unlike particular normative facts, some principles (any that are not derivative of other principles) do not seem in further need of normative justification.

But principles cannot play the role advocates of Completeness Matters want them to. The problem is that we have to find a way of making sense of the nature of principles that is compatible with Completeness Matters. Berker makes the distinction between some claim’s being explanation-involving as opposed to its being explanation-serving. In the former case, the principle’s contains as part of its content that some explanation of normative phenomena obtains; in the latter, the principle itself partially explains a particular normative fact.²⁵ Below, I’ll argue first that principles must be explanation-involving, before arguing that Completeness Matters requires them to also be explanation-serving, and that this causes a serious problem.

Principles seem to be making some sort of explanatory claim: utilitarians are not claiming a mere covariation between happiness facts and rightness/wrongness facts, but rather that the former (somehow) explain the latter.²⁶ Consider how quickly we get into trouble if principles are something like universal generalizations.²⁷ Universal generalizations are plausibly grounded in their instances, which in this case, will be an unquantified conditional, e.g., “If this act of kicking the cat promotes my desires, then this act of kicking the cat is good.” Conditionals are grounded in the truth or falsity of their antecedent and consequent. But where “This act of kicking the cat is good” is that which is being grounded, we know it’s true. That means that the conditional is partially grounded in the truth of the consequent. So we’d end up with this (where “Gx” is the instantiation of some normative property, and “Mx” is the instantiation of some mental property):

And this would violate the intuitive thesis that grounding is irreflexive: Ga cannot partially ground itself.²⁸

This is just to say (though inconclusively, for reasons of space) that principles are explanation-involving. But Completeness Matters is hard to square with this. Consider views on which principles make grounding claims²⁹: maybe something like $\forall$ x((Mx&Gx) → Mx grounds Gx),³⁰ or—if you prefer the Armstrongian view of principles as relations between properties—M-ness grounds G-ness.³¹ Notice that on the first view, we don’t wholly avoid the problem with principles as universal conditionals: the principle will be partially grounded in [Ma grounds Ga]. How is this latter fact grounded—in Ma, Ga, or both?³² The latter two options won’t work, because just like before, irreflexivity would fail. But we also can’t have [Ma grounds Ga] grounded just in Ma, since all views will collapse into mind-dependence views (assuming transitivity). (Recall that the point of bringing in principles was to open space for a mind-independence option, where normative facts are not fully grounded in mental facts.)

But suppose we find some way of avoiding such problems. There still seems to be a problem with Completeness Matters. Roughly: its proponents must choose between all normative facts being mind-dependent, or getting a false view of how normative principles work. Let “normative principles” in this argument be short for “the normative principles that partially ground normative facts, according to Completeness Matters”. The argument goes:

• 1. Assume that principles are amongst the grounds of particular normative facts.

• 2. Normative principles state full grounds for normative phenomena.

• 3. Those grounds are either fully mental or not.

• 4. If so, there are no mind-independent normative facts.

• 5. If not, then (assuming (1)) normative principles are self-referential (i.e., name themselves as partial grounds) and give partially normative grounds for normative phenomena.

• 6. It is not the case that normative principles are self-referential and give partially normative grounds for normative phenomena.

• 7. Therefore, if (1), then there are no mind-independent normative facts.³³

Completeness Matters is committed to (1), assuming, as I’ve argued, that P needs to be a principle. So the argument purports to show that on Completeness Matters, there is no room for mind-independence, just as a matter of broad metaphysical constraints.³⁴

(2) seems true as a constraint on normative principles: they aim to give a full explanation of normative phenomena.³⁵ For example, utilitarians say that failure to maximize happiness fully explains wrongness.³⁶ (3) is not in need of defense. (4) is important: if normative principles give full, mental grounds for normative facts—e.g., “That I desire to Φ fully grounds that it is good to Φ”—then assuming the truth of Completeness Matters, all normative facts are mind-dependent. Yet it was the appeal to principles that was supposed to save mind-independence, on Completeness Matters. But now we find out that those very principles just state the core commitment of mind-dependence—that the relevant normative fact is fully grounded in mental facts. So mind-independence turns out to be incoherent.³⁷

Regarding (5): principles need to describe the full grounds of normative facts (as (2) says). But we’ve seen that if mind-independence is to be possible, those grounds must include normative principles. So normative principles end up being self-referential in this way:

Likewise, since the principle makes itself part of the grounds of the particular normative fact (and so for all normative facts), the business of normative principles is—surprisingly—not to give non-normative grounds for normative facts. Normative principles are normative facts, broadly construed. (I think Leary’s criterion for what counts as a normative fact—any fact involving a normative property—is too broad, but we don’t need such a broad criterion for normative principles to be normative facts.³⁸) So, by necessity, principles always mention normative conditions—themselves—for normative facts.

Finally, (6): normative principles—at least the ones that would help ground particular normative facts³⁹—aim to give purely non-normative grounds for normative phenomena. That is one of the main points of normative theorizing, after all—at least insofar as principles figure into such theorizing: to give non-normative conditions for normative phenomena. When should we believe a proposition? When the probabilities line up in such-and-such a way. When should we perform an action? When it maximizes pleasure. The list could go on, but in each case our goal is to give purely non-normative conditions for the normative phenomenon in question. Independently of this worry, it also seems implausible that normative principles are categorically self-referential. Berker argues that it is not self-reference alone that is worrisome—I agree—but rather that for normative principles to be self-referential would lead to endlessly iterated grounding relations.⁴⁰ I think that while self-reference is not impossible, it is simply implausible that all the normative principles—e.g., the principle of utility—that help ground particular normative facts are covertly self-referential. So we cannot countenance normative principles being categorically self-referential.

Generally speaking: what is causing us trouble is that on Completeness Matters, principles are being made to play two different grounding roles. Completeness Matters is trying to have them both ground and describe the very grounding they’re engaged in—or in Berker’s terms, it’s trying to make principles both explanation-involving and explanation-serving. I’ve given independent reason to think that principles describe grounding relations. And Completeness Matters is committed to them doing grounding work for normative facts. So Completeness Matters won’t work, at least if we want to leave open even the possibility of mind-independence. Of course, there may actually be no mind-independent normative facts. But this shouldn’t be settled at the level of the mind-dependence vs. mind-independence distinction itself.

2.4 COG views and the mundane dependence constraint

Before I leave the topic of COG views, it will be worth cashing in on our investment earlier, and briefly show that they categorically fail the Mundane Dependence Constraint. Showing this exhaustively would be hard, since COG views are diverse and concern diverse normative sub-domains. So let me just take a particular instantiation of a COG view, show how on that view there will be no space for mind-dependence, and then argue more abstractly that this will hold across COG views.

Consider the following Kinds Matter view: that a normative fact is mind-dependent when grounded in desires. Now consider a metaethical view according to which goodness facts are grounded in desire facts. So, on this (simplistic) view, goodness is mind-dependent. But now, consider the following mundane dependence: that my action was good is grounded in the fact that by performing it I kept a promise. This mundane dependence is incompatible with our simplistic metaethical view: if goodness can be grounded in promise-keeping, then it’s not universally grounded in desire!⁴¹

More generally, this mundane dependence is incompatible with any grounding of goodness (purely) in mental states, so on any version of Kinds Matter, all mind-dependent accounts of goodness would be false. Similar mundane dependence claims would be incompatible with mind-dependence accounts in other normative sub-domains (wrongness, beauty, etc.). And so assuming Kinds Matter, mind-dependence in general would be impossible.

More generally still, this has nothing to do with Kinds Matter as opposed to the other two COG views at all! For any given COG view that makes mental states M₁-M_X the relevant states for mind-dependence for instantiations of normative property N, we will be able to easily find mundane dependence claims that do not mention any of M₁−M_X in the grounding base. (This is true even for Completeness Matters, on which M₁−M_X would not exhaust the grounding base, since the mundane dependence claim would also not mention normative principles in their grounding bases.) This entails that mind-dependence would be impossible on COG views, as long as we’re committed to virtually any plausible set of mundane dependence claims. And we do not want our account of mind-dependence to entail that it is impossible.

3 A proposal

I hope to have established that we cannot save mind-(in)dependence by fiddling with the grounds of normative facts. In this section, I’ll look at other avenues, arguing against one possible solution before ultimately endorsing and developing a different answer.

3.1 An inadequate solution

First, consider one answer to the question that I won’t get behind here. On it, the problem is not that we haven’t gotten specific enough about the grounds of mind-dependent vs. independent normative facts; it’s that we haven’t differentiated the kinds of grounding involved. Up till this point I have tried to stay maximally ecumenical about the nature of the non-causal dependence I’m concerned with. But perhaps that’s precisely what’s keeping us from solving the problem! After all, when the utilitarian says that happiness facts make an action right, and the relativist says that desire facts make an action right, it can seem like the “making” can’t be the same kind of thing in both cases.

Some involved in the grounding literature have, for similar reasons, proposed that we distinguish between two fundamentally distinct kinds of grounding: metaphysical grounding and normative grounding.⁴² The former is involved in cases like [The ball is red] grounding [The ball is colored]. The latter is involved in cases like [The act maximized happiness] grounding [The act is morally right]. Perhaps then:

Type of Grounding Matters::

Normative facts are mind-dependent just when they are metaphysically grounded in mental states.

For the pictorially-oriented:

According to Type of Grounding Matters, it is the lefthand side that represents mind-dependence; the righthand side represents a mundane first-order normative claim.

The payoff is that this allows us to get out of the initial problem: that all kinds of first-order normative theories posit dependence on the mental without thereby committing to mind-dependence. With two fundamentally distinct kinds of grounding, we have the logical space to say that the utilitarian is committed only to the normative grounding of the moral in the mental; it is open to her, though importantly not required in virtue of her utilitarianism, to go on to endorse the metaphysical grounding of the normative in the mental. So utilitarians are not committed one way or the other on the mind-dependence question. This is because—on this solution—the kind of explanation utilitarians trade in is different from the kind of explanation metaethicists trade in.

Type of Grounding Matters also has the virtue of compatibility with the Mundane Dependence Constraint. Recall the problem COG views had: because they did not differentiate kinds of grounding, mind-dependent views on which (say) goodness was grounded in (say) desire could not accommodate mundane examples like the goodness of my acting being grounded in the fact that I thereby kept a promise. On Type of Grounding Matters, this mind-dependence theorist can accommodate both kinds of claim: goodness is metaphysically grounded in desire even though it is sometimes normatively grounded in promise-keeping.

There’s a lot of appeal to this solution. But like I said, I won’t pursue it. I won’t give a knock-down objection to it, either.⁴³ I bring it up mainly to show, by one of its shortcomings, the need for my proposed solution. Let’s get a concrete case: suppose Norma wants to assert a connection between being desired and being good. But she doesn’t want to commit to the mind-dependence of goodness. Fine: on Type of Grounding Matters, Norma will say that [Φing is desired] normatively grounds [Φing is good], while denying that the former metaphysically grounds the latter. And let’s assume for simplicity that goodness is always normatively (but never metaphysically) grounded in desire in this way, on Norma’s view. So far so good: Norma can articulate her view while maintaining the mind-independence of goodness.

But it seems to me that, even though we have established that no particular goodness facts are mind-dependent on Norma’s view, it’s still unclear whether a certain class of goodness-related phenomena are mind-dependent. Norma is asserting a connection between certain kinds of non-normative, mental facts (desire facts) and certain kinds of normative facts (goodness facts). So what about this connective fact—this normative principle, which states the connection between the non-normative and the normative? Is it mind-dependent in the way that she doesn’t want particular goodness facts to be?

It seems to me that this is an important question—mind-dependence is at least just as important to establish concerning principles in domain D as it is to establish about particular facts in D. Imagine a utilitarian explaining the principle of utility to you, after which you ask whether the particular moral facts are metaphysically or merely normatively explained by pleasure/pain facts. He indicates the latter. You thereby learn, on Type of Grounding Matters, that the particular moral facts are mind-independent. But then suppose you get curious about the principle of utility itself. That is, you want to know whether this utilitarian thinks that the principle of utility is itself mind-dependent—whether, in other words, our mental states make it the case that happiness-maximization makes something obligatory. This seems like a natural next step in the discussion. To me, it seems like a deepening of the discussion—like we’re really getting to the heart of the issue now. My happiness might give rise to moral goodness, but it would seem like a deeper mind-dependence if happiness gave rise to moral goodness in general because (say) I approved of the connection between the two.

Of course, Type of Grounding Matters can deliver verdicts about normative principles: they are mind-dependent when they are metaphysically grounded in the mental. And while normative principles are not the sort of thing that are apt for normative grounding—it obtains between particular normative facts and their (normatively relevant) non-normative grounds—it could be that principles are ungrounded entirely. This would provide the space within Type of Grounding Matters for a mind-independence option for normative principles.

The problem is that Type of Grounding Matters is not able to account for the influence of the mind-dependence of principles on the mind-dependence of particular normative facts. Suppose utilitarianism is true, and that all particular facts about rightness and wrongness are (merely) normatively grounded in facts about happiness. But the principle of utility itself is only true because I think it’s true. In this case, it seems that we should say that some broader mind-dependence of the moral obtains. After all, if I shouldn’t divorce my wife because it will not maximize happiness, but that connection only obtains because I think it obtains, then the fact that I shouldn’t divorce my wife seems to depend strongly on my mental states.

Yet Type of Grounding Matters says that in this case, the particular facts about what is right and wrong are mind-independent—just as much as if the principle were ungrounded! And it will do so in all such cases: cases in which the particular normative fact is (merely) normatively grounded in the non-normative, but its connective principle is metaphysically grounded in the mental. But in such cases—as I argued by example above—that seems decidedly like the wrong verdict.

3.2 The general solution

So what we need is a view that makes sense of the fundamental importance of the mind-(in)dependence of normative principles, and especially of their influence on the mind-dependence of particular facts in a domain. That is just what I hope to develop in the rest of this paper. I give a view of the mind-(in)dependence of normative principles and propose that the mind-(in)dependence of other, particular normative facts can be derived from the status of the principles that play a role in their explanation.

In this section, I develop the view in the most ecumenical way possible. This is an important task, since I’m proposing that this is the best way to make sense of a diverse extant debate; it’s therefore important not to get committed to too much theoretical baggage in the account. However, in the next section, I’ll deal with a problem with making sense of the explanatory role of normative principles, a problem which threatens to derail my ecumenical solution. If we’re committed to my ecumenical solution, then solving this problem does end up committing us to some surprising theoretical baggage.

Here’s the view I want to defend:

Inheritance::

A normative fact N is mind-dependent just when some principle connecting N to its non-normative grounds is mind-dependent.

A few clarifications: first, for now let the “connecting” talk be a stand-in for whatever role principles play in explanations of particular normative facts: the principle of utility connects [Φing is wrong] in [Φing did not maximize utility]. Second, in normal cases, there will only be one principle connecting N to its non-normative grounds. However, Inheritance is formulated (“some principle,” not “the principle”) to account also for cases of overdetermination, like the following: [This action is wrong] is fully (and separately) grounded both in [This action caused harm H] and [This action broke a promise]. This would mean there are two different principles, each connecting the same normative fact to a separate full ground, which would open up the possibility that one principle is mind-dependent while the other isn’t. Inheritance says that, in such cases, N is mind-independent: I think the interesting question about mind-independence is whether a normative fact has some mind-independent grounding. (One reason: N will have the modal robustness of a normal mind-independent normative fact.)

Of course, Inheritance wouldn’t be satisfying without an account of the mind-dependence of normative principles:

Principle Dependence::

Normative principles are mind-dependent just when they are asymmetrically grounded in mental facts.

So the mind-(in)dependence of particular normative facts is inherited from the mind-(in)dependence of their principles, and the mind-(in)dependence of normative principles is determined by whether they are asymmetrically grounded in the mental. We can call the combination of Principles Matter and Principle Dependence “the principles view”.

This view captures the sense that it is principles that are most important for mind-(in)dependence, and it does so in a way that allows us to derivatively capture the mind-(in)dependence of particular normative facts, as well. Finally, it gets us out of our motivating puzzle: utilitarians are not committed to the mind-dependence or -independence of moral facts just if virtue of their utilitarianism. That all particular moral facts are grounded in mental facts has nothing to do with mind-dependence, on the principles view. If we want to know whether some particular utilitarian is committed to mind-dependence, we need to ask her whether she thinks the principle of utility itself is asymmetrically grounded in mental facts. And this question is simply not decided by the commitments of utilitarianism. Thus we get the desired result that our account of mind-(in)dependence leaves utilitarianism compatible with either the mind-dependence or the mind-independence of moral facts. And similarly for other merely first-order normative views on which the normative phenomena in question are fully and asymmetrically grounded in the mental.

3.3 Can the principles view explain inheritance?

The principles view, I believe, needs to give at least a cursory explanation of Inheritance. Why is it, after all, that particular normative facts are mind-(in)dependent just when a certain principle is? Just for contrast, consider a Humean view of normative principles: they merely describe the patterns of coinstantiation of normative and non-normative phenomena, but play no robust role in explaining the normative.⁴⁴ On such a view, I submit, it would be perplexing why Inheritance obtains. With no explanatory connection between principles and particular facts, there’s reason to doubt that the latter could inherit mind-(in)dependence from the former.

Failing to explain Inheritance doesn’t immediately relegate a view of the role of principles to the trash can; however, it may be a mark against it. Thus, while my goal here is not to argue for any particular understanding of principles, I do think it would bolster the plausibility of the principles view if we could show that there is some view of principles that could potentially explain it. (It would be bad, after all, for there to be no plausible account of principles that could.) So in the rest of this section, I want to do that. To be clear: I do not intend to argue for any particular view. That would require far more space than I have left here. It will be enough to meet my limited goal here to show how at least one view on offer could, in principle, explain Inheritance.

This would seem very easy if principles were part of the grounds of particular normative facts (on the model we considered in Sect. 2.3). It might seem that, on this view, it’s obvious that normative facts will inherit their mind-dependence status from normative principles. That’s because plausibly, grounding is transitive: if A partially grounds B and B partially grounds C, then A partially grounds C (and similarly for full grounding).⁴⁵ So, if [normative fact] was partially grounded in [normative principle], then whenever [normative principle] is fully (and therefore partially) grounded in the mental, [normative fact] will be partially grounded in the mental. So, assuming the principles view, whenever [normative principle] is mind-dependent, so will [normative fact], so long as [normative fact] is partially grounded in [normative principle]. And that would allow the advocate of the principles view to explain Inheritance.

Unfortunately, there are a couple of problems with this strategy. The first is that, as I argued above, it doesn’t preserve the mind-dependence/-independence distinction—so it would be a poor candidate for explaining a feature of a proposed account of that distinction! The second is that, paying close attention to the fullness of the grounding involved, we see that normative facts will be grounded in the mental differently than normative principles. Because normative facts will only be partially grounded in principles, they can only inherit partial grounding in the mental from normative principles.⁴⁶ But on the principles view, a principle is mind-dependent when it is fully (and asymmetrically) grounded in the mental. So it seems that, if the advocate of the principles view takes this route, she’ll have to a disjunctive account of mind-dependence. That is, particular normative facts will be mind-dependent in a different way than normative principles: normative facts are mind-dependent when partially grounded in the mental, whereas principles are mind-dependent when fully grounded in the mental. And that seems to purchase an explanation of Inheritance only at the expense of a unified theory of mind-dependence.

Luckily some philosophers have proposed an alternative: normative principles play a governing role in normative explanations.⁴⁷ Consider causal laws as an analogy: a causal law clearly plays some important explanatory role in causal explanation. But it is neither the role of cause nor effect. For example, it’s not as though when the window’s breaking is caused by the baseball hitting it, the causal laws involved (about fragility, force, etc.) are part of the cause. Rather, they govern the causation. Or as another analogy, consider valid arguments and valid rules of inference. We don’t think that rules of inference like modus ponens enter into the argument as extra premises; rather, they govern the inference from premises with certain forms to conclusions with certain forms.⁴⁸ Ralf Bader summarizes the view like this: “Normative laws are not fundamental in the sense of occupying the basic level of a grounding hierarchy, but rather in the sense of standing outside this hierarchy and inducing its structure.”⁴⁹ Elsewhere he says that principles “modify the grounding relation rather than featuring as relata.”⁵⁰ What’s clear is that we now have a grounding relation on which there are three relata—ground, grounded, governor—instead of just two (ground, grounded). Call this view “the principles-as-governors view.”⁵¹

Recall that the principles-amongst-the-grounds view required principles to be both explanation-involving and explanation-serving. By pulling principles out of the grounds, the principles-as-governors view saves a robust explanatory role for them while solving this problem: they can now describe the explanation (roughly speaking) since they need not enter into the explanans themselves.

Admittedly, the governors view does not offer an explanation as clean as the principles-amongst-the-grounds explanation initially seemed: since principles are no longer amongst the grounds of normative facts, we can’t appeal to a traditional transitivity principle to establish Inheritance. But I don’t think there’s a problem here; in fact, I think it’s natural to endorse Inheritance once you’ve accepted a governing role for principles. To see this, consider two pieces of support, which I’m happy to admit are not wholly conclusive.

First, consider again our causal and logic analogies. Suppose we found out that causal laws were all in our heads: it’s only because we believe in (e.g.) the law of gravitation that it obtains. Now consider the case in which someone’s hitting the baseball causes it to take a certain trajectory. This causation only occurs because it is governed by certain laws, L. Now assume L are mind-dependent—in which case, so is the flight of the ball. That is, because L only obtain because we believe them, so also the ball only flies as it does because of our beliefs. Similarly with logic: if the “law” of modus ponens only obtains because I like it, then I can reasonably conclude Q from P and P $\rightarrow$ Q only because of my likings. So based on our analogies, we’d expect the same to obtain in the normative case.

Second, consider the modal landscape. We need to consider a tightly constricted case, just for illustration: assume mind-dependence about the moral, that the entity whose mental states “construct” the moral is agent A, and the relevant mental state is endorsement. Assume also that A endorses the principle of utility—i.e., that an act is morally wrong if and only if, and because, it fails to maximize utility. Finally, assume that A does not endorse any other moral principle which would overdetermine wrongness in this case. It follows, on our sample view, that the principle of utility is true, and that—to take a concrete case—a particular act of adultery was wrong because it failed to maximize happiness. But it also follows, given our assumptions, that were A not to endorse the principle of utility, not only would it fail to be true, but the act of adultery would fail to be wrong, too.⁵² Assuming principles-as-governors, particular acts of wrongness don’t result if one doesn’t endorse a principle that could govern the grounding of that wrongness in the non-normative. So, the particular normative facts supervene on the endorsement of a such a governing principle. And that makes it unsurprising that Inheritance would be true.

I need not here commit to principles-as-governors, nor do I think we ought to see the mind-(in)dependence debate as one about governors. That would read too much into the debate. My goal was simply to show that there is some way of understanding the role of principles that is both initially plausible and could explain Inheritance. I do not claim that principles-as-governors has no problems: for example, it’s not clear whether we need to give further explanation of the governor role, and if so, whether we can. (On Berker’s presentation of the view—which he rejects—it is actually a part of the principles-as-governors view that the governor role is sui generis.⁵³) But I do not need to answer all the questions about the principles-as-governors view. I take myself to have shown the rough outline of how it could explain Inheritance, which for my purposes is enough.

3.4 An objection to the principles view

Before concluding, I want to briefly consider an objection to the principles view: it seems to categorize apparently mind-dependence theories as mind-independent.⁵⁴ In addressing this objection, I hope to actually highlight one of the boons of the principles view: roughly speaking, it can accommodate platitudinous particular normative claims, in a way some first-glance readings cannot.

Consider the paradigm case of mind-dependence, from the Euthyphro: what's pious is pious because the gods love it. It is natural (so the objection goes) to construe this as the claim that piety facts are grounded in facts about what the gods love—so, mental facts. But now consider a version of this view on which the relevant principle—that X is pious if (and because) X is loved by the gods—is ungrounded. The principles view thus says that piety is mind-independent on this view. But this seems like a paradigm case of mind-dependence! Even granting that it’s mind-independently true that piety is determined by the gods’ lovings, that doesn’t seem to abate the mind-dependence of piety facts.

The problem is that taking the Euthyphro example at face-value does not accommodate the Mundane Dependence Constraint. If piety is always grounded in the gods’ lovings, then there is no way for an act to be impious because it was a prosecution of one’s father. Once we are committed to piety facts being grounded in the gods’ lovings, they can’t be grounded in facts about respecting one’s parents.⁵⁵

So, in sum, we cannot and should not take the Euthyphro case at face-value.⁵⁶ But then, can the principles view offer an alternate reading that is still plausible? Yes: what is actually at stake, on the principles view, is whether the normative principle, “If an act is respectful of one’s parents, then it is pious” is grounded in the gods’ lovings. This allows for the fact that an act is pious to still be grounded in the fact that it is respectful of one’s parents, thus maintaining the relevant mundane particular normative claim.

Now, there is the question of whether this interpretation is faithful to what Socrates actually says. How important that question is, I don’t know—my suspicion is that it is not all that important. The important question is what mind-(in)dependence consists in, and it’s not too wild for belief, in my view, if the correct answer to that question entails that maybe Socrates was speaking a little loosely.

But I think that the principles view actually doesn’t do violence to the Euthyphro claim, for what it’s worth. On the mind-dependence horn of the dilemma, the pious is pious because it is loved by the gods. This could be read in two ways, with equal plausibility as interpretations. First, it could mean that particular facts about piety—e.g., that John’s action of obeying his dad is pious—obtain because of the gods’ love. In other words, mental states ground particular normative facts. But it seems equally plausible to read “the pious” as “the (non-normatively construed) types of act that are pious.” In that case, the claim is that what makes it the case that acts of obedience to one’s parents are pious is the gods’ love. But “acts of obedience to one’s parents” is not a particular act—it is a type of act. And to say that acts of a certain (non-normative) type have a certain normative property is to state a principle. So, on this reading, the claim is that the gods’ lovings ground a normative principle.

I think this is (at least) often the case. When people say “What’s N is N because of M,” or “N-facts obtain in virtue of M-facts,” or even “N-facts are grounded in M-facts,” they’re giving an abbreviated version of the following claim: “T-facts ground N-facts, and this general connection is because of M-facts,” where T is some non-normative category. For example, when Hamlet says “Nothing is good or bad, but thinking makes it so,” we shouldn’t read him as saying that thinking grounds goodness. Instead, he’s claiming that whatever mundane non-normative stuff grounds goodness—e.g., that the Dr. Pepper is sweet grounds that it is good—that principled connection (here, between sweetness and goodness) only obtains because I think it does.

Conclusion

Recent attempts to undermine the distinction between mind-dependence and mind-independence are worrying, and well-motivated. Many merely first-order normative theories seem committed to some kind of dependence of the normative on the mental, which raises the issue of whether we can make sense of the distinction without committing these first-order theories to mind-dependence. I’ve argued here that a whole class of attempted solution to this problem should be tossed out: we can’t solve the problem by simply modifying the contents of the grounds of particular normative facts. Rather, we ought to see the heart of the distinction in how normative principles are grounded. Doing so allows us to vindicate the mind-dependence vs. mind-independence distinction: there really is a meaningful debate as to whether normativity is all “just a matter of opinion.”