Metaethical Internalism: Another Neglected Distinction

Abstract

‘Internalism’ is used in metaethics for a cluster of claims which bear a family resemblance. They tend to link, in some distinctive way—typically modal, mereological, or causal—different parts of the normative realm, or the normative and the psychological. The thesis of this paper is that much metaethical mischief has resulted from philosophers’ neglect of the distinction between two different features of such claims. The first is the modality of the entire claim. The second is the relation between the items specified in the claim. In part one I explain this distinction and the problems neglecting it may cause. In part two I show that it has been neglected, and has caused those problems, at least with respect to one version of internalism. That is judgment internalism, which claims that moral beliefs are necessarily related to pro- or con-attitudes; e.g., that if you believe you ought to x you must have some motivation to x. The considerations standardly adduced in favor of judgment internalism support only a version which lacks the metaethical implications typically attributed to it, at least so far as anyone has shown. Proponents and opponents of judgment internalism fail to realize this because of their neglect of the modality/relation distinction. I illustrate by considering discussions of judgment internalism by Russ Shafer-Landau, Simon Blackburn, James Dreier, David Brink, and others.

‘Internalism’ is used in metaethics for a cluster of claims which bear a family resemblance. They tend to link, in some distinctive way—typically modal, mereological, or causal—different parts of the normative realm, or the normative and the psychological. Such claims play a significant role in metaethics. This makes sense, since any such relation would seem to reveal much about the items so related.

For instance, judgment internalism posits necessary relations between moral beliefs and pro- and con-attitudes; e.g., that, necessarily, if you believe you are morally obligated to A, you have some motivation to A. If judgment internalism is true, the implications for metaethics seem enormous. Historically, the prime beneficiary has been noncognitivism, since it effortlessly explains any such necessities: if moral beliefs are attitudes, of course one cannot have the former without the latter. Cognitivists who accept judgment internalism tend thereby to be led to quite specific and distinctive accounts of moral ontology or psychology. For instance, some explain it by adverting to the special, intrinsically magnetic or repulsive nature of moral properties. Some explain it by supposing that moral beliefs are unlike typical nonmoral beliefs in some central respect: they necessarily generate attitudes, perhaps, or are beliefs-which-are-also-attitudes. Some explain it by supposing that moral properties just are relations to attitudes, or are whichever properties bear certain relations to attitudes, or are not really properties at all but merely projected attitudes. The truth of judgment internalism would not settle all metaethical questions, but, it seems, it would give us an excellent start.¹

Early statements of internalist claims often obscured crucial distinctions, as between facts (about morality or reasons), recognitions of such facts, or mere beliefs about them; or between normative and motivating reasons. Happily, in recent years a number of philosophers have succeeded in bringing these distinctions to our attention, so that they are now impossible to miss.² The thesis of this paper, though, is that a distinction is still neglected–and it is crucial. My aim is to make it impossible to miss. A subsidiary aim is to provide a handy way of distinguishing internalist claims.

In part one (Sects. 1 and 2), I present the distinction and explain in abstract terms how neglecting it is apt to cause metaethical mischief. In a nutshell, doing so makes certain metaethical possibilities invisible, and makes internalist claims seem more revelatory than they really are about the items our interest in which leads us to consider them in the first place. I dub this pattern–of neglect, oversight, and over-estimation of epistemic value–the ‘internalist fallacy’.

In part two (Sects. 3 and 4), I show the fallacy at work in recent work on judgment internalism. Contrary to appearances, judgment internalism does not narrow the range of possible views about moral psychology or ontology, nor does it even favor or oppose any. To be precise, although there are versions of judgment internalism which would have such an impact, none are supported by the considerations standardly brought forth in its favor. Those considerations do support a version of judgment internalism, but it is metaethically innocuous. Philosophers think a metaethically significant version is supported only because they have committed the internalist fallacy.

Part One: A Distinction and Why it Matters

Five Internalist Variables

Internalist claims can be distinguished by the values they give to five variables. The neglected distinction of this paper is between two of them.

Three Internalist Claims

To identify the variables, I will start with three common internalist claims, stating them, then briefly explaining each.³ I use ‘attitude’ as a cover term for any affective or motivational state.

Judgment Internalism	Moral beliefs entail attitudes.
Reasons/Motives Internalism	Reasons entail attitudes.
Morality/Reasons Internalism	Moral facts entail reasons.

Judgment Internalism (aka Motivational Internalism, Appraiser Internalism about Motives), as noted, posits necessities linking moral beliefs and attitudes. Clearly there are attitudes characteristic of moral beliefs; e.g., con-attitudes such as guilt and indignation are characteristic of beliefs about what is wrong. The dispute between judgment internalists and externalists is over whether the attitudes are necessary for the beliefs.

Reasons/Motives Internalism (aka Instrumentalism, Internalism about Reasons, Metaphysical Existence Internalism) says there is a necessary link between the realm of (practical) reasons and that of attitudes. In the simplest version, S’s having a reason to do A entails that S has some pro-attitude served by A. For instance, you have a reason to go to Nepal only if doing so serves a pro-attitude you have. In more sophisticated versions A must bear some counterfactual relation to S’s attitudes, involving full information, rational deliberation, and the like.

Morality/Reasons Internalism (aka Rationalism, Agent Internalism about Reasons) tells us that moral facts entail reasons; e.g., necessarily, if you are morally obligated to A then you have a reason to A. This differs from judgment internalism both in what has the entailment (moral facts vs. moral beliefs) and what is entailed (reasons vs. attitudes). The reasons meant here are normative, and not just moral reasons, which would render the thesis trivial.

Four Variables

Each claim tells us that certain sorts of things, of some normative or evaluative type, entail something. This gives us four possible variables: the general sort of thing, or Genus, the claim is about (e.g., beliefs, facts); the normative or evaluative Type of those things (e.g., moral, reasons); the Relation (so far indicated by ‘entail’ in each case); and the Relata (e.g., attitudes, reasons).

Internalism	Type	Genus	Relation	Relata
Judgment	Moral	Beliefs	Entail	Attitudes
Reasons/Motives	Reasonsa	Facts	Entail	Attitudes
Morality/Reasons	Moral	Facts	Entail	Reasons

1. ^aWe can further distinguish facts about reasons (S has a reason to A) from facts which are reasons (F—some fact—is a reason for S to A)

We can get further internalisms worth considering by filling in different values; e.g., replace ‘Beliefs’ with ‘Facts’ or vice-versa in the above claims, try plugging in entirely new Types, like Aesthetic, Prudential, Legal, etc., or provide details about the Relation or Relata.⁴ But my interest here is not in what is brought out by this schema, but in something hidden by it. In particular, complexities hidden by ‘Entail’ in Relation.

Wide- vs. Narrow-Scope Necessities

We have four variables down. To see the fifth, consider exactly what ‘Entail’ tells us.

From our three internalisms we can discern that it means, at least, accompaniment. That is, moral beliefs (or reasons) are accompanied by attitudes, moral facts by reasons: where there is one, there is the other. (Throughout I take ‘x is accompanied by y’ to leave open the possibility that y is part of x, or even that they are identical.)

Clearly, though, accompaniment does not exhaust the content of ‘Entail’, for it leaves out the modal element. To get internalism we must posit accompaniment, not just actually, but throughout possible worlds. That is, ‘Entail’ indicates that the accompaniment is necessary. But we must distinguish between wide and narrow-scope, or de dicto and de re, necessities. Since this distinction is crucial for seeing our fifth variable, I will say a word about it.

Wide-scope necessities say that a certain proposition is necessarily true; the operator scopes over an entire proposition. Narrow-scope necessities tell us that something(s) necessarily has a certain property; the operator scopes over that property. Which is intended is sometimes indicated by the location of the operator (‘necessarily, p’ vs. ‘x is necessarily F’). Consider, e.g., the following claims.

• (1) Necessarily, parents have children.

• (2) Parents necessarily have children.

(1) puts the necessity out front so it scopes over the proposition parents have children. As such, it is clearly true: no possible world has a parent without a child. But (2) says something different, and false. What it says is that those individuals who (in the actual world) are parents necessarily have children. Not so.Your parents did not have to have you, or any other children. They could have remained childless, as is true of every parent

The Fifth Variable: Wide-Scope Modality

Returning to our internalist claims, we can see that ‘Entail’ really covers two distinct possibilities. It could mean that the Relation is that of necessary accompaniment (e.g., moral beliefs are necessarily accompanied by attitudes). Or it could mean that there is a necessity which scopes over the entire proposition (necessarily, moral beliefs are accompanied by attitudes). In the latter case, we get a necessity, but not in the Relation, which might be simple accompaniment. These same two possibilities hold no matter what gets filled in for Type, Genus, and Relata.

A narrow-scope necessity is a matter of Relation, but none of the four variables so far identified can distinguish claims which incorporate a wide-scope necessity from those which do not. Wide-scope necessity is a matter neither of Genus, Type, Relation, nor Relata. A fifth variable is thus demanded: (wide-scope) Modality. In practice, its value tends to be either ‘Necessity’ or simply unspecified.⁵

The distinction between Modality and Relation is the ‘neglected distinction’ of this paper.

Consequences of Neglect of the Modality/Relation Distinction

If the Modality/Relation distinction is neglected then features of values for one will not be clearly distinguished from features of values for the other. I will focus on one way in particular this could occur–since it is what (I will argue in part two) actually has occurred in the case of judgment internalism. This is the confusion of strength in Modality with strength in Relation.

Weakness and Strength in Modality and Relation

The intuitive idea of ‘strength’ is that of robustness or interestingness: a given value is strong if part of what makes the claim interesting is that it incorporates that value. More strictly ‘strength’ can be understood as relative power of entailment. With respect to some variable, a value V ₁ is stronger than V ₂ iff an internalist claim which incorporates V ₁ entails a claim which incorporates V ₂ but not vice-versa. For instance, in Modality, a necessity is stronger than nothing, since necessarily p entails p but not vice-versa. Likewise, in Relation, necessary accompaniment is stronger than accompaniment.

Given the paucity of options, strength in Modality is just necessity, weakness just nothing. In Relation there are more options, and which are in play will vary according to other details of the internalist claim. For instance, regarding judgment internalism (relating moral beliefs and attitudes), possibilities include constitution (total or partial), necessary accompaniment, and various causal relations (e.g., that moral beliefs alone cause attitudes).⁶

Prima facie, there is nothing incoherent about a strong Modality/weak Relation internalism; i.e., one which assigns necessity to Modality but a relatively weak value, such as accompaniment, to Relation. As (1) and (2) show, strength in Modality need not carry over into strength in Relation. So strong Modality/weak Relation internalism is a possible view. But if the Modality/Relation distinction is neglected, it will be overlooked. This is the first bad effect of neglecting the Modality/Relation distinction. There is a second.

The Epistemic Value of Strength in Relation vs. Modality

Let us call the ‘Subject Matter’ of an internalist claim the things of its Type and Genus (e.g., moral beliefs). Our interest in such things is what leads us to consider any given internalist claim in the first place. But there is a huge difference between the light cast on a Subject Matter by strength in Modality and strength in Relation. In general, the latter is much more revealing.

Relations potentially reveal much about what bear them, since some relations can only be had by certain sorts of things. For instance, suppose, as many do, that there are no necessary relations between distinct things. It follows that if x and y bear a necessary accompaniment relation, x and y are not distinct. A mere accompaniment relation tells us no such thing, since, obviously, distinct things can bear that relation. In general, stronger relations hold out more hope of revealing the nature of the Subject Matter than weaker ones.

That is not true of Modality. Surprising as it may sound, a claim with a wide-scope necessity reveals no more about its Subject Matter than the same claim without one. What, e.g., does (1)—that, necessarily, parents have children—tell us about parents? Well, that they can and do have children, since what is necessary is possible and actual. But these two things are also revealed by the fact that parents have children (sans modality). (1) tells us that parents are such that, necessarily, if they are parents, they have children. But this reveals nothing distinctive about parents, since it is equally true of non-parents. Adding the necessity reveals nothing about the subject of the claim not also revealed by the same claim minus the necessity.

Necessity is so unrevealing because it may be due to the concept used to pick out that subject rather than the nature of the subject itself. Parent is a relational concept: to pick something out as a parent is to pick it out in virtue of its relation to a child. The necessity thus holds, but tells us little about the things to which the concept applies except that they do, in fact, bear that relation.

The lesson for internalism is this. If we are interested in the nature of the Subject Matter, we must look to Relation not Modality. If we neglect the Modality/Relation distinction, we are apt to conflate strength in Modality with strength in Relation, and claims with only the former will seem more revealing about their Subject Matter than they really are. This overestimation of internalism’s evidential import vis-à-vis its Subject Matter is the second bad effect of neglect of the distinction.⁷

This pattern—of neglect of the Modality/Relation distinction, consequent oversight of the possibility of strong Modality/weak Relation internalism, and consequent overestimation of the epistemic value of a merely strong Modality internalism—is what I dub the ‘internalist fallacy’. In part two I show it at work in the case of judgment internalism.

Part Two: The Case of Judgment Internalism

Strong Modality/Weak Relation Judgment Internalism: A Sketch

To make the charge of the internalist fallacy with respect to judgment internalism, I do two things. First, in this section, I sketch a version of strong Modality/weak Relation judgment internalism (henceforth ‘SM internalism’) and argue that it is not so obviously false as to be simply ignorable. Secondly, in the next section, I will show that much thinking about judgment internalism involves, precisely, ignoring it–strength in Modality is taken as tantamount to strength in Relation.⁸

The Concept Moral Belief as Relational-to-Attitudes

The core idea of SM internalism is that the concept moral belief is relational-to-attitudes, as the concept parent is relational-to-children. That is, moral belief picks out a state only if that state bears certain relations to attitudes. This underwrites a strong Modality: if concept C applies only to things which bear R to an F, it follows that, necessarily, Cs bear R to Fs. That is why (1) is true. But it is consistent with weak relations between moral beliefs themselves and attitudes. Here it is important that relational concepts do not apply to everything required for their application: parent applies to Sue only if she has kids, but applies only to Sue, not to her and her kids. Likewise, even if moral belief is relational-to-attitudes, the items to which it applies–i.e., moral beliefs–need not be even partly constituted by attitudes.

There are nice nonmoral models for the claim that moral belief is relational-to-attitudes. Consider, for instance, the concept inspiring belief. Inspiring beliefs are beliefs which inspire. If Joe’s belief that God loves him gets him out of bed each morning with a spring in his step, it is an inspiring belief. Thus, necessarily, inspiring beliefs are accompanied by an attitude—viz. inspiration: a belief which always leaves you cold is not inspiring. So there is a true claim with a wide-scope necessity relating inspiring beliefs and attitudes. But the relations between inspiring beliefs and inspiration are not very strong: inspiring beliefs are neither constituted by inspiration nor necessarily accompanied by it (any inspiring belief could have failed to inspire). The strongest relation is merely causal. Inspiring belief, then, underwrites a strong Modality but only weak Relations, in claims relating inspiring beliefs and attitudes. Many other examples could be given: replace inspiring with fanatical, reviled, calming, demoralizing, self-serving, etc.

Another nice nonmoral model is that of the concept tourist map, which unlike world map applies to maps in virtue of their role, not their content. Reflection on this case helps explain what must be going on if moral belief is indeed relational-to-attitudes. Concepts of the form T map indicate a relation to the realm T denotes. Typically this is the relation of being about that realm. But it need not be. When maps themselves play a salient role in the realm, that role may be indicated. That is why tourist maps are for tourists, not of them. If moral belief is relational-to-attitudes, it must be because the role moral beliefs play in the moral realm is such a salient feature of them that it has shaped the very way we conceptualize them. Moral beliefs are those with a distinctive role in moral practices, not beliefs about moral practices.

Three Differences Between the Moral and Nonmoral Cases

There are, of course, differences between the moral and nonmoral cases. Let us note three.

Moral Belief Type Concepts as Relational-To-Attitudes

If moral belief is relational-to-attitudes then a further kind of concept is relational-to-attitudes as well–a kind for which there is no ‘inspiring’ analogue. Moral belief is not our only concept of moral beliefs. We have a slew of more specific ones as well, such as belief that racism is wrong, belief that kindness is good–indeed, for any moral sentence m, belief that m counts. Correspondingly, there are two sorts of internalist claims, those for which the Type is moral and those for which it is That m, for some moral sentence m. The SM internalist versions of these are captured below (note that ‘Specific’ names a type of claim–to get tokens, fill in moral sentences for m).

Internalism	Modality	Type	Genus	Relation	Relata
General	Necessity	Moral	Beliefs	Accomp.a	Attitudes
Specific	Necessity	That m	Beliefs	Accomp.	Attitudes

1. ^a For ease I have indicated the Relation as accompaniment, though there are other possibilities (see note 24)

Now, if we endorse General SM internalism, we should endorse the Specific ones as well. That is because, necessarily, for any moral sentence m, beliefs that m are moral beliefs. If this were not so, it would be possible for there to be, for instance, a belief that racism is morally wrong which is not a moral belief–but that is incoherent. So if it is necessary that moral beliefs are accompanied by attitudes, it is necessary that beliefs that m are as well.

The nonmoral models involve claims analogous to the General internalist claim, but not the Specific ones. There is no sentence s such that, necessarily, a belief that s is inspiring. There is thus a wider variety of moral internalist truths, if there are any, than nonmoral ones.⁹

Notice that all of the Specific internalist claims are versions of SM internalism. The only difference between them and the General SM internalist claim is what goes in Type. The core idea remains the same: just as the concept moral belief is relational-to-attitudes, so are the concepts belief that racism is wrong, belief that kindness is good, and so on.

Relational to Which Attitudes?

Naturally the particular attitudes and relations implicated by moral belief will differ from those implicated by inspiring belief and the others. What is crucial in the moral case is not inspiration but, plausibly, the guidance of action. Moral beliefs play a certain action-guiding role, and are instilled to play that role. Types of moral beliefs, right-beliefs, wrong-beliefs, etc., play specific types of action-guiding roles, involving particular pro- and con-attitudes. All this, SM internalism says, is necessarily true.

Space limits forbid a detailed discussion of these relations. But note how weak they might be. For we could say: a belief is action-guiding just in case it tends to lead to action. This leaves it open that when it does lead to action, it does so in the typical (Humean) way: by linking with a distinct desire. There is no implication that the belief is a desire, or is necessarily accompanied by a desire, or alone causes a desire, or is about intrinsically magnetic or repulsive properties. All SM internalism demands is that moral beliefs are in fact accompanied by the relevant desires.

In fact, SM internalism could be even weaker: it need not even demand that moral beliefs be accompanied by attitudes in their every owner. One could hold that, necessarily, moral beliefs find their home in social practices involving the guidance of action, but that not every moral believer need participate in these practices. Likewise, necessarily, money finds its home in practices of exchanging goods and services; this follows from the concept money. But not every bit of money need be used in this way, as when a proud retailer frames her first bill. SM internalism, then, can even allow amoralists (moral believers without the typically accompanying attitudes), though not without some moralists.¹⁰

SM Internalism and Moral Content

Another difference, at least potentially, between the concept inspiring belief and moral belief concepts is that the former implicates nothing about the content (truth-conditions) of the states to which it applies. For any p, it is conceptually possible to have an inspiring belief that p. But that is not a necessary feature of relational-to-attitudes concepts. It is illuminating to compare the concept money with concepts of specific kinds of money like coin and bill. They too are relational concepts: they apply only to items found in the relevant social-practical context. But they implicate more than a role. To be a coin or a bill is not just to play the relevant sort of money role, but to have a certain material constitution or feature (metal vs. paper, or perhaps rigidity vs. papery-ness). Similarly, moral belief concepts might implicate both a content and relations to attitudes. For instance, belief that A is right might apply to a belief B only if (a) B’s content is that A is F and (b) B plays the relevant action-guiding role. And F may be whatever you like. Regarding moral content, SM internalism does not narrow the range of independently available options.

Three Challenges

Let us consider three worries about SM internalism. Though each needs more thorough treatment, I will say why I think none is obviously decisive.

Nonmoral Belief Concepts are Not Relational-To-Attitudes

Concepts of the form belief that s are not in general relational-to-attitudes. Rather, they indicate a kind of state (belief) and its content (that s). There seems a powerful inductive case against the SM internalist account of moral belief concepts.

In reply, first, the inductive case may not be very strong. Often, albeit controversially, belief that s concepts are taken to do more than just indicate kinds of states and contents. Many think it possible, e.g., to believe that x is Clark Kent (or water) without believing that x is Superman (or H₂O) despite their being beliefs with the same content. If more than content is indicated here, why not relations to attitudes elsewhere? There may even be nonmoral cases in which attitudes are implicated; e.g., concepts of racist beliefs such as belief that x is a wop. Plausibly, to believe that, you must have a con-attitude to Italians. At very least, racist beliefs seem to require that there have been some racist attitudes. It is hard to imagine how there could have been wop-beliefs if no one had ever had even a hint of con-attitudes to Italians.

Secondly, there is obviously something distinctive about morality—as evidenced, inter alia, by the existence of the field of metaethics itself—even if what is distinctive is disputed. So even if nonmoral belief that s concepts uniformly fail to implicate attitudes, that is hardly decisive. Similarly, although anti-realist expressivists face challenges due to their suggestion that moral sentences have a distinctive semantics for items of their syntax, those challenges arise from the details of their views and not from the mere fact of distinctiveness.

Are Moral Beliefs Only Contingently Moral?

SM internalism entails that moral beliefs are only contingently moral beliefs. That follows from the strong Modality and weak Relation. If it is necessary that moral beliefs be accompanied by attitudes, but moral beliefs are not themselves necessarily accompanied by attitudes, then moral beliefs are not necessarily moral beliefs. But intuitively moral beliefs are necessarily moral beliefs.¹¹

Intuitively that is so, but we should resist such intuitions. It can be hard to disentangle things from our concepts of them and this can throw off our intuitions. For instance, intuitively, cars are necessarily cars. How could that very thing sitting in your driveway have existed but not have been a car? But plausibly it could have, since car is a functional concept which applies only to items linked in the relevant way to the relevant intentions, and the object in your driveway could have existed even in the absence of such intentions. For instance, it could have come together coincidentally, in a world without sentience. Disentangling things from our concepts of them is especially difficult when the things in question are obscure to us. This is why the intuition that mental states are necessarily mental states is not taken as decisive against views which conflict with it; e.g., which identify mental states with the brain states which, contingently, play a certain role. Likewise, the intuition that moral beliefs are necessarily so is hardly decisive against SM internalism.

Too Shallow an Account of Moral Belief-Attitude Relations?

SM internalism looks to moral belief concepts rather than moral beliefs or moral properties for the source of morality’s distinctively intimate involvement with attitudes. But this seems rather shallow—a mere semantic truth when something psychological or metaphysical is called for.

An SM internalist need not claim that her account of moral belief concepts captures everything–or even nearly the most interesting things—about the relations between moral judgments and moral attitudes. Indeed, if the lesson drawn from tourist map is correct, our concepts of moral beliefs are as SM depicts precisely because they bear highly salient relations to moral attitudes independently of our concepts of them. Those rich relations are neither explained nor much described by SM internalism.

It is true that SM internalists are not free to posit strong Relations between moral beliefs and moral attitudes. But neither can full-fledged externalists, and no one thinks externalism can simply be ignored. If there are stronger relations than SM internalism can allow, they must be shown. Sometimes an appeal to moral phenomenology is thought to suffice for establishing such a relation (e.g., McNaughton 1988). But such appeals seem weak. If the stronger relation is modal (moral beliefs or properties are necessarily linked to attitudes) then we may object that phenomenology reveals what is, not what must be. If the idea is that moral beliefs are attitudes, then we may wonder whether phenomenology is adequate to distinguish that from the nearby possibility that moral beliefs are merely accompanied by attitudes. Phenomenology is not what reveals that inspiring beliefs are not themselves inspiration. If the stronger relation claimed is that moral beliefs alone cause attitudes or actions we should worry about the presence of ‘calm passions’—attitudes known by their effects rather than their phenomenology. SM internalists, like externalists, can hold that moral properties as a matter of fact have a distinctively strong tendency to generate attitudes in human beings aware of them, and account for the phenomenology that way.¹²

The Internalist Fallacy in Recent Discussions of Judgment Internalism

Judgment internalism has played a prominent role in recent metaethics. I argue that much of its felt importance is due to the internalist fallacy.

The Traditional Inference to Noncognitivism

A main source of interest in internalism¹³ is its supposed tendency to favor noncognitivism. Here I will quote two recent (and I think fair) explications of how the inference is apt to go. In both cases the argument is that internalism ascribes to moral judgments a property incompatible with their being or expressing beliefs. The arguments are generally taken to be valid, including by these authors; objections focus on the truth of the premises. But the appearance of validity is due to the internalist fallacy. In both cases SM internalism could underwrite the internalist premise without favoring the noncognitivist conclusion. But in neither case is SM internalism even considered, much less ruled out.

Shafer-Landau (2003, pp. 120–121) describes a “classic antirealist argument”, sufficiently widespread to be called “the Non-cognitivist Argument”

1. Necessarily, if one sincerely judges an action right, then one is motivated to some extent to act in accordance with that judgment. (Motivational Judgment Internalism)

2. When taken by themselves, beliefs neither motivate nor generate any motivationally efficacious states. (Motivational Humeanism)

3. Therefore, moral judgments are not beliefs. (Moral Non-cognitivism)

2 rules out beliefs bearing certain relations to motivation, viz. constituting or alone causing it. But the Relation of 1 is neither of those: it is, rather, mere accompaniment. The strength of 1 lies in its Modality. SM internalism could thus underwrite 1. And, as have seen, SM internalism does not entail that moral judgments, taken by themselves, either motivate or generate any motivationally efficacious states. So the inference is invalid: we need some reason to think that 1 supports ascribing to moral judgments the relations 2 says no beliefs have. The appearance of validity must be due to 1’s being thought tantamount to ascribing to moral judgments such relations. That is, it must be due to the internalist fallacy.

Miller (2003, pp. 120–121) in his review of “arguments which the non-cognitivist uses against cognitivism”, describes “the argument from moral psychology”

[A] Being motivated to do something or to pursue a course of action is always a matter of having a belief and a desire … But [B] it is an internal and necessary fact about an agent that, if she sincerely judges that X is good, she is motivated to pursue the course of action X. So [C] if a moral judgment expressed a belief, it would have to be a belief which sustained an internal and necessary connection to a desire: it would have to be a necessary truth that an agent who possessed the belief would inter alia possess the desire. But [D] no belief is necessarily connected to a desire because, as Hume claimed, ‘beliefs and desires are distinct existences’, and it is impossible to have a necessary connection between distinct existences. So [E] it cannot be the case that moral judgments express beliefs.¹⁴

In a nutshell: Given the Humean claim about motivation [A], internalism [B] entails [C], that if moral judgments express beliefs, those beliefs are necessarily accompanied by desires. But [D] no beliefs are necessarily accompanied by desires, so noncognitivism [E]. Notice that the strong relation Miller takes internalism to posit, necessary accompaniment, differs from Shafer-Landau’s, constitution or alone causation. There is more than one route from internalism to noncognitivism.¹⁵

Let us focus on the inference from [B] to [C], taking [A] for granted. Simplifying harmlessly,¹⁶ it goes like this:

• [B*] Necessarily, if an agent judges that X is good, she has a desire.

So,

• [C*] If judgments that X is good express beliefs, those beliefs are necessarily accompanied by desires.

[B*] has a strong Modality (necessity) but a weak Relation (accompaniment). If it is true, that could be because moral judgments express desires, or express beliefs which are necessarily accompanied by desires. But it could also be because the concept judgment that X is good is relational-to-attitudes: a judgment counts as a judgment that X is good only when the belief it expresses is accompanied by the relevant attitudes. If so, then the judgments would express beliefs, but not beliefs which are necessarily accompanied by desires. Since SM internalism has not been ruled out, the argument is invalid. We need a reason for thinking the strong Modality of [B*] translates into the strong Relation of [C*]. But that need is not noticed, much less satisfied.¹⁷ This bespeaks the internalist fallacy.

What Amoralists Could Show?

Perhaps the accusation of ‘fallacy’ is too hasty. Perhaps, charitably, we should read claims such as Shafer-Landau’s 1 and Miller’s [B] as infelicitous attempts to express a strong Relation internalism. If so, the charge of fallacy falls away.

Of course, it is not so simple. To see if this move succeeds we must look at why we suspect internalism true in the first place. The evidence for it will support either a strong Relation, or merely SM internalism. If the latter, then replacing our SM internalism with a strong Relation internalism avoids fallacy at the cost of support. And with respect to at least one main source of support for internalism, that is precisely the price that would be paid. The source I mean is our tendency to withhold moral belief ascriptions when the relevant attitudes are absent. That tendency does support internalism, but only SM internalism. It gives no support to a strong Relation internalism, at least so far as anyone has shown. To see why, let us consider two recent defenses of internalism, by Blackburn and Dreier.

Blackburn’s most extensive discussion of internalism is in Ruling Passions. There he considers a variety of putative counterexamples; e.g., Milton’s Satan, who appears to commit himself to evil. Blackburn’s main reply is that we run the risk of overhastily rejecting internalism in light of such cases, due to an oversimple view of the psychological role attitudes might play. If we think they must straightforwardly translate into action, or even into a push to action in a predictable direction, then we ignore psychological complexities. For even love bears no such simple relation to action and impulse. One may even behave in the way opposite to what we would expect on a simple model of love. Blackburn cites Othello, who “still loves Desdemona as he smothers her” (1998, p. 63). Likewise, the attitudes entailed by moral beliefs may manifest in a variety of ways, and we should not suppose they are absent when their typical manifestations are. Thus, a Satan-like figure could appear to be a counterexample to internalism, while still actually having the relevant attitudes.

However, Blackburn suggests, there is a limit:

these cases [of dislocation between ethics and action] only exist against either a psychological background of motivation by what is perceived of as duty, or a social background of insistence upon duty as a practical constraint. That is, if we try to stretch the case away from the Satan/Othello model by sketching an agent who has absolutely no conflict, but views duties with the utmost unconcern, we will find that he exists only against a backdrop in which talk of duty does express concern.

If there is nothing but settled, cold unconcern from an agent for what he verbally acknowledges as his duty, then of course we do begin to talk of mere lip-service. The agent is using evaluative vocabulary in a parasitic way, as mere labeling for what other people regard as good.

“So”, he concludes, “externalists can have individual cases, but internalism wins the war. Ethics remains essentially practical, a matter of attitude, disposition, and emotion” (Blackburn 1998, pp. 64–65).

Dreier (1990) calls upon similar cases, with a twist. He describes a community of folks who, though otherwise like us, are ‘Sadists’: they have inverted attitudes. They have a predicate whose application tracks our ‘right’ (applied to helping, honesty, fidelity, etc.) which is not just unaccompanied by pro-attitudes, it is accompanied by repulsion. (And similar correspondents to all our moral terms.) Dreier holds that, intuitively, such folks are not thereby expressing beliefs that helping is right. Hence, right-beliefs must bear the relevant relation to pro-attitudes.

Both Blackburn and Dreier cite our tendency to withhold moral belief ascriptions in the absence of the appropriate attitudes.¹⁸ It is indeed a powerful tendency, especially when it comes to entire communities of amoralists. I think it accounts for much of our attraction to internalism. And SM internalism explains it. If the concept moral belief applies only to states which play the relevant role, then amoralists lack moral beliefs, just as they lack inspiring beliefs if their beliefs do not inspire.¹⁹

Since SM internalism explains our intuitions about amoralists, it would be a mistake to infer from those intuitions something which SM internalism fails to support—at least not without even considering SM internalism. But both Blackburn and Dreier appear to do so.

Blackburn’s conclusion—“ethics [is] essentially practical”—sounds like the assertion of a strong Relation. And he seems to be reading it that way. Consider his admonishment of Jackson and Pettit for taking as platitudinous in their account of moral concepts like fairness “Commonplaces about motivation (anyone who believes one option only to be fair will prefer it, other things being equal)”.

[T]he platitude that moral judgments motivate is a dangerous element in a theory that aims at showing that their content is purely descriptive.… [I]t is not an accident that ethical judgments motivate: ethics, as I have already explained, is essentially practical. But then the problem is how there can be a purely descriptive content that is at the same time essentially practical – one that cannot, as we might say, be apprehended in general just with a shrug… If they ‘just happened’ to motivate us, in the same way that discovery that the bedroom is yellow might just happen to motivate someone to change it, this would be fine, but that is not the way it is. (1998, pp. 113–114)

Here Blackburn offers cognitivists two options. First, externalism: it is an accident that moral beliefs motivate—they just happen to do so, like color beliefs. Second, a mysterious internalism: the content of moral beliefs is essentially practical—cannot be apprehended with a shrug. The essential practicality of ethics he takes to rule out the former, leaving only the latter.

But there is a third option: SM internalism. On that view, it is not an accident that moral beliefs motivate; it is, rather, necessary. But that necessity is not explained by anything’s mysteriously being impossible to apprehend with a shrug. The content of moral beliefs, on this view, could be apprehended with a shrug—though such apprehensions would not be moral beliefs. Blackburn, guided by the thought that ethics is “essentially” practical, has taken internalism to involve a strong Relation (necessary accompaniment). But the impossibility of amoralists as yet reveals no such Relation.²⁰

Dreier also draws conclusions unwarranted by SM internalism. His ultimate view, Speaker Relativism, is that moral terms are indexicals—terms with variable content, like ‘I’, ‘here’, and ‘now’. He offers us a nice analogy with demonstrative indexicals like ‘this’ and ‘that’. Just as the contents of those terms is pinned down by demonstrations, so the content of moral terms is fixed by attitudes.²¹ Attitudes are intentional states. They can be directed at properties: e.g., Mary approves of maximizing happiness, Ruttiger of treating people as ends, Sally of being honest, faithful, and benevolent, and so on. On Dreier’s view, the content of ‘right’ in their respective mouths is fixed by those attitudes: in Mary’s mouth, the content of ‘right’ is the maximization of happiness, in Ruttiger’s, the treatment of others as ends, in Sally’s, being honest, faithful, and benevolent, etc. This gives us internalism: just as no belief is expressed by ‘that is …’ without something to pin down the content of ‘that’, so no belief is expressed by ‘A is right’ without an attitude to pin down the content of ‘right’.²²

Our tendency to withhold moral belief ascriptions can be explained in this way, but it need not be. SM internalism explains it, and (as noted in 3.2.3) SM internalism is compatible with any independently possible view of the contents of moral beliefs. We could add to SM internalism the view that moral content is fixed by attitudes; we could also add that moral content is invariant. Since SM internalism explains our response to amoralists, that response does not yet favor Speaker Relativism. We need a further argument.²³

Dreier’s example reminds us that noncognitivism is not the only view internalism has been taken to support. The others fall into two main (non-exclusive) categories: those which explain internalism by reference to the content of moral beliefs and those which explain it by reference to their distinctive psychology. The former camp includes views which make moral content sensitive to responses (e.g., Johnston 1989; Wiggins 1997), and the view that moral properties are (or would be were they instantiated) intrinsically motivating (e.g., Mackie 1977; Platts 1979). The latter camp includes the view that moral beliefs are besires (both beliefs and attitudes), beliefs which alone cause attitudes, beliefs which can yield action without aid of attitudes, or beliefs necessarily accompanied by attitudes (e.g., McDowell 1978; McNaughton 1988). In each case, the view goes beyond what is warranted by SM internalism, which is simply neutral on both moral content and the psychology of moral beliefs (except for demanding that they do in fact bear the relevant weak relations). Since SM internalism fully accounts for our tendency to withhold moral beliefs from (individual or communal) amoralists, such views are not supported by that tendency.²⁴

The Metaethical Beneficiaries of Exposing the Internalist Fallacy

I have noted views which appear supported by internalism, but—so far as anyone has shown—only due to the internalist fallacy. Other views appear threatened by internalism, but only due to the fallacy. They are the metaethical beneficiaries of a clear view of the Modality/Relation distinction.

Central among those beneficiaries are views which combine full-fledged moral objectivism with an austere naturalism (e.g., Boyd 1988; Brink 1989; Railton 1997; Sturgeon 2006).²⁵ Given the neglect of the Modality/Relation distinction, it makes sense that such views have seemed threatened by internalism. For they would be threatened by a strong Relation internalism. As we have seen, the ways of accounting for such an internalism all involve either rejecting full moral objectivism, or retaining objectivism but only by positing something which would make an austere naturalist uncomfortable. The former include adopting noncognitivism, taking moral ‘properties’ to be projected attitudes (i.e., an error theory), or some form of relativization of moral properties to attitudes such as with Dreier. The latter include taking moral properties to be intrinsically motivating, or taking moral beliefs to be unlike typical nonmoral beliefs in some deep way: necessarily accompanied by attitudes, able to yield attitudes or action without aid of a distinct attitude, constituted by attitudes, or some such. If we reject all these moves, we are left with the view that moral properties are inert, objective, natural properties. And that moral beliefs are in relevant respects like any other beliefs about inert, objective, natural properties, such as beliefs about the masses of objects. Accounting for a strong Relation internalism with such materials is simply impossible.

Objective naturalists typically reply to internalist challenges by citing the possibility of amoralists. This reply is weak, since the internalist challenge can be re-raised at the communal level (3.2.2), where internalist intuitions are far more robust.²⁶ Fortunately for them, it is also not necessary. They can give a vindicating explanation of those intuitions, by adopting SM internalism.

Note that objective naturalists can adopt SM internalism without the slightest change in their account of moral beliefs and properties. They will simply be supplementing it with a distinctive account of how we conceptualize those items. SM internalism is compatible with any independently possible account of moral content (3.2.3). This includes those which depict moral properties as inert, objective, natural properties. And SM internalism makes no demands on our account of moral beliefs beyond those made by a nonmodal accompaniment claim (2.2). Such a claim is compatible with whatever account you like of the beliefs thus accompanied. Indeed, objective naturalists already accept such an accompaniment claim, as it is empirically evident. SM internalism may be a mistaken account of moral belief concepts. But there is no chance of it carrying problematic implications regarding anything beyond those concepts.

Modality/Relation in Influential Discussions of Internalism

One reason the Modality/Relation distinction is neglected is that it has been unnoted, and even somewhat blurred, in the work of those who have done most to bring internalist subtleties to our attention. I will show how easy it is for this to occur, by considering two such philosophers, Brink and Darwall.

In Moral Realism and the Foundations of Ethics, Brink nicely lays out many internalist distinctions, but not Modality/Relation. His “first approximation” of “Appraiser Internalism about Motives” (judgment internalism) is this: “the claim that it is a part of the concept of a moral consideration that such considerations motivate the agent to perform the moral action …” (1989, p. 39). This sounds like the core idea of SM internalism, that moral belief concepts are relational-to-attitudes. There is wiggle room in what Relation is suggested by “moral considerations motivate” (or, later, “provide… the appraiser with motivation” (1989, p. 40)). Perhaps it means that they are themselves motivational states, or alone cause such states. But it may mean something weaker, such as that they are action-guiding in the sense of 3.2.2.

The discussion which follows, though, suggests that Brink has a stronger Relation in mind. Consider his characterization of externalism: “Externalism is the denial of internalism; externalism claims that the motivational force … of moral considerations depend[s] on factors external to moral considerations themselves” (1989, p. 42). Here the idea is that internalism and externalism differ over whether motivation (or “motivational force”) is internal or external to moral beliefs–not the concept moral belief but moral beliefs “themselves”. That is a disagreement about a strong Relation (probably constitution, but perhaps alone causation).

A strong Relation is also suggested by Brink’s later characterization: “Internalism [holds] that moral considerations necessarily motivate[, and that] it is the concept of morality that shows that moral considerations necessarily motivate …” (1989, p. 42). Although not mandatory, it is natural to read “moral considerations necessarily motivate” as indicating necessity in Relation. It sounds like something those states necessarily do. One might easily take internalism as the claim that the concept of a moral consideration is that of a necessarily motivating state.

Finally, Brink’s discussion of “the antirealist argument from internalism about motives” suggests that internalism involves a strong Relation.

Internalism is a premise in many arguments for noncognitivism. Some claim that no set of facts or cognitive states (e.g., beliefs) could necessitate any affective or motivational attitude…. It seems possible to be indifferent to any set of facts, but this cannot be true of moral facts if internalism is true.…

I don’t think we should accept this antirealist argument. One realist reply would be to question the incompatibility of realism and internalism. A realist might argue that moral belief can itself be motivational …. However, this is not my preferred strategy; I think we should reject the internalist premise…. (1989, pp. 43–44)

Brink notes only two possible realist rejoinders: argue that “moral belief can itself be motivational” or deny internalism. This suggests that internalism demands that moral beliefs are themselves motivational. That is, it suggests that internalism has a strong Relation.²⁷

Darwall has also contributed much to our understanding of the variety of possible internalisms. In his most extensive discussion, he characterizes judgment internalism in the following way.

Judgment internalism holds that if S judges (or believes, or sincerely asserts) that she ought to do A (or that she has reason to do A), then, necessarily, she has some motivation to do A. This is a view about the nature of normative thought and language. It says that nothing counts as a genuinely normative thought or as a sincere normative assertion unless it is related in this way to motivation. (1997, p. 308)

The definition offered is a version of SM internalism: the “necessarily” is, on reflection, most naturally read as wide-scope, the Relation as accompaniment. But confusion is possible. Consider a reader who asks herself, after the last sentence, ‘related in what way?’. Though the correct answer given the definition is simply ‘accompanied by’, it would be natural to think the answer is ‘necessarily’. The gloss “a view about the nature of normative thought and language” also suggests a stronger relation, since as we saw (in 2.2), SM internalism as yet offers us no such thing. Darwall’s words do not especially lend themselves to conflation of Modality/Relation, but philosophers taking their cue from them will not easily see the distinction or that it needs marking.²⁸

Conclusion

Neglect of the Modality/Relation distinction in discussions of internalism is widespread. Perhaps this is no surprise, given the plethora of claims which get the title ‘internalism’. There are only so many details we can attend to. But concern for metaethical upshot should move us to attend to this distinction. I have argued that there is one sort of internalism—judgment internalism—about which it is clear that such attention has important results. It reveals a neglected possibility, SM internalism. Since SM internalism absorbs the evidence from amoralist thought-experiments, those thought-experiments leave strong Relation internalisms hanging. And since such a Relation is needed for the metaethical upshot often ascribed to judgment internalism, we have to re-think whether it really has that upshot.