I have written a revised version of this post. Please read it instead.

---

The first formulation of Kant’s categorical imperative is “act only on that maxim by which you can at the same time will that it should become a universal law.”^[i] Many of Kant’s critics charge that the first formulation of the CI is so vague and general that specific recommendations about how to act cannot be derived from it. In this post, I will attempt to reply to this criticism. I will do so by building on the ideas that I introduced in the series called The Self-Confidence Principle.

Choice Guiding Concepts

In “A Better Moral Theory,” Sayre-McCord draws our attention to the following scenario:

“Imagine…that on our understanding of what morality’s standards actually are, a different set of standards would be (morally) better. Would this give us any reason to think that our original theory of what the standards of morality are is mistaken? I think it would; we would be finding ourselves with reason to think we had originally gotten the standards wrong…”

“A Better Moral Theory” page 3 version of May 2018

Later on, he explains why he thinks that, in this scenario, we have found ourselves with reason to think that the original standards are wrong. His explanation turns on a feature that, as he points out, all normative concepts share:

“…normative concepts (whether moral or not) are concepts that make a claim on our allegiance, purporting to be such that their satisfaction ipso facto means – and here I purposely use a generic characterization that reaches beyond morality – someone or other has reason to act or react in certain ways to what satisfies them…”  

“A Better Moral Theory” page 16 version of May 2018

Because normative concepts are such that their satisfaction establishes reasons, I will say that they are choice guiding. Sayre-McCord suggests that, if we discover that “on our understanding of what morality’s standards actually are, a different set of standards would be (morally) better,” then we have discovered a reason to think that our current conception of morality is not choice guiding.^[ii] Because the correct conception of morality (whatever it is) is choice guiding, we have discovered a reason to think that our current conception of morality is wrong.

Crucial to Sayre-McCord’s line of reasoning is the claim that if we discover that “on our understanding of what morality’s standards actually are, a different set of standards would be (morally) better,” then we have discovered a reason to think that our current conception of morality is not choice guiding. To conclude this section, I will make a very tentative attempt to say why this crucial claim is true.

Recall from The Self-Confidence Principle 2 that a moral theory is self-confident if and only if it follows the GVp↔Vp rule. I think it is plausible that, if a moral theory lacks self-confidence, then it is not choice guiding. I think this for the following reason:

Suppose that you accept moral theory T, and you use it to guide your choices. Further suppose that T fails to follow the GVp↔Vp rule. If this is the case, then by T’s own lights, T has a problem. Because you use T to guide your choices, you have reason to choose to believe that T has a problem. If you believe that T has a problem, then, it seems, you shouldn’t use T to guide your choices. To put the point another way, if you believe that T has a problem, then, it seems, you shouldn’t believe that the satisfaction of the conception of morality that T expresses is on its own enough to establish that you have a reason to do anything. 

I strongly suspect that something like this line of argument is correct, but I struggle to phrase it in a way that is completely convincing. For that matter, I struggle to phrase it in a way that makes it easy to understand. I still have lots of questions about it, and I will turn to these questions in the last section of this post.^[iii] For now, I will return to Kant.

Law-Like Principles

Again, the first formulation of the CI is “act only on that maxim by which you can at the same time will that it should become a universal law.” It is obvious that you cannot will that a principle should become a universal law unless that principle is law-like. Laws tell you what to do, so a principle is not law-like unless it is choice guiding. In the previous section, I reached the tentative conclusion that if a moral theory is choice guiding, then it is self-confident. Putting these claims together, if a moral theory’s principles are law-like, then that theory is self-confident.

This gives us some information about what we need to do in order to satisfy the first formulation of the CI. One thing that we need to do is develop a self-confident moral theory. Unfortunately, self-confidence is a very high bar. A self-confident moral theory delivers all and only those verdicts that, by the theory’s own lights, are good verdicts. A self-confident theory is, by its own lights, completely free of problems and counterexamples. This is such a stringent requirement that, practically speaking, we cannot hope to construct a self-confident theory.

It follows that we cannot hope to satisfy the first formulation of the CI. This means that we are required to either (a) abandon the CI or (b) keep the CI but admit that we cannot hope to construct a correct moral theory. In the next section, I will argue that we should pick option (b) instead of option (a) because, I will maintain, option (b) really isn’t so bad. 

Self-Confidence as an Ideal

Although it is practically impossible to construct a self-confident moral theory, it is possible to pursue self-confidence as an ideal. An ideal is a model of perfection. Even though, practically speaking, perfection is usually out of our reach, we can still use ideals to guide our choices. A better choice is one that more closely approximates the ideal.

Even though we cannot realistically hope to construct a moral theory whose principles satisfy the first formulation of the CI, the CI still gives us an ideal to work towards. A better moral theory, one that more closely approximates the ideal, is one whose principles come closer to being such that we can will that they should become universal laws. A better theory, then, is one that comes closer to being self-confident.

Because the first formulation of the CI gives us guidance that we can follow when constructing a moral theory, option (b) does not look so scary. Although a prefect moral theory is out of our reach, we still have useful guidance that will help us make improvements to whatever moral theory we happen to start out with.

We are now in a position to return to the question with which I began this post. Is the first formulation of the CI so vague and general that specific recommendations about how to act cannot be derived from it? It seems to me that the ideal of self-confidence can help us decide between competing moral theories. Recommendations about how to act follow from such decisions.

Questions

Everything that I have said in this post is tentative and incomplete. I still have lots of questions. Here are some of them:

1) I said “Even though we cannot realistically hope to construct a moral theory whose principles satisfy the first formulation of the CI, the CI still gives us an ideal to work towards. A better moral theory, one that more closely approximates the ideal, is one whose principles come closer to being such that we can will that they should become universal laws.”

In this passage, I talk about the CI like it is not part of a moral theory. Rather, I talk about it like it is a principle that is used to regulate the selection of moral theories. Is it okay to understand Kant, or Kantianism, in this way?

I guess what I mean is that the CI is a general principle that cannot be applied to real world situations unless we elaborate on it by developing more specific principles. A ‘moral theory,’ as I am using the term, is a set of more specific principles. That set will either be self-confident, or it will not. If it is not, I suspect, it does not satisfy the CI.  To put the point another way, I guess what I am thinking is that you can’t will that a maxim should become a universal law unless that maxim is part of a set of more specific principles that is self-confident. Is this correct? Or am I totally off track here?

Does the way that I use the term ‘moral theory’ here conflict with the way that I use it in The Self-Confidence Principle 1-3?

2) Kant clearly thinks that some rules, such as ‘don’t make lying promises,’ satisfy the first formulation of the CI. Where do I get off, then, saying that it is practically impossible to develop a moral theory that satisfies the first formulation of the CI? I guess what I mean is that it is practically impossible to develop a whole set of principles each of which satisfies the first formulation of the CI. Is this true?

3) Kant and Hume hold opposing views about the relationship between reason and morality. One way to understand why they disagree is to look at what is sometimes called Hume’s Motivation Argument. I will outline the Motivation Argument without trying to render it deductive.^[iv]

Reason, as Hume understands it, is the capacity to determine truth and falsity.^[v] If this is all that reason can do, then, Hume observes, “reason can never immediately prevent or produce any action by contradicting or approving of it.”^[vi] Reason can contradict or approve of beliefs as true or false, but it cannot contradict or approve of actions because actions are not the sort of thing that can be true or false. In contrast, morality can, at least sometimes, prevent or produce actions by contradicting or approving of them. This happens, for example, when someone is motivated to give to charity by the recognition that it is good to help others. Because morality can do something that reason cannot do, Hume concludes that morality cannot be founded on reason alone.^[vii]

Kant, on the other hand, thinks that reason can prevent or produce actions by contradicting or approving of them. Kant thinks this because he understands reason differently from Hume. For Kant, reason is the capacity to make non-arbitrary choices about what to believe and how to act. Rational agents have the capacity to choose actions and beliefs on the basis of reasons, as opposed to acting and believing at random or because forced to.^[viii] Non-arbitrary choices are choices that are based on principle, and principled choices are choices that are based on maxims that could be universal laws. For this reason, Kant thinks that the fundamental principle of morality is “act only on that maxim by which you can at the same time will that it should become a universal law.” Because this principle is a requirement of reason, it is founded on reason alone.

I have been going back and forth about whether the issues that I have addressed in this post are relevant for this debate. Kant believes that the fundamental principle of morality can be known a priori. Does Hume believe, in contrast, that the only way to know anything at all about what morality requires of us is to investigate the matter empirically? If this is Hume’s position, then he needs to deny that the fundamental principle of morality can be known a priori. Or, instead, is Hume’s position that the only way to know anything substantial about what morality requires of us is to investigate the matter empirically?^[ix] If this is his position, then he can admit that that CI can be known a priori (in the same way that he can admit that ‘unmarried man’ can be derived from the concept of bachelor a priori), but he must deny that anything substantial can be derived from the CI. If this Hume’s view, then he must contend with the argument that I have offered in this post.

Famously, Hume holds that “’Tis not contrary to reason to prefer the destruction of the whole world to the scratching of my finger.”^[x] Perhaps, then, he could take the position that (1) the CI can be known a priori and (2) specific choice guiding recommendations can be derived a priori from the CI but (3) reason alone does not require us to follow these recommendations? To put the point another way, Hume could say that, if we care about morality, then reason requires us to follow certain recommendations, but reason does not require us to care about morality.

The disagreement between Kant and Hume rests in part of their different understandings of rationality. Are they just using the term ‘reason’ differently? Is it possible to construe their disagreement such that it is more than just a disagreement over the definition of the word ‘reason’?

Does Hume deny that we have a capacity that Kant thinks that we do have? If so, how might this disagreement be resolved? Does Kant think that, in order to have any understanding of the world at all, we must have the capacity that he describes, forcing Hume either to be an error theorist or to accept Kantianism?

4) On page 1 of “A Better Moral Theory,” Sayre-McCord writes that “there being a difference between one’s theory of the best normative X (the best morality, the best standards of inference, the best rules of justification…) and one’s (so far) best theory of X, necessarily provides a reason (though perhaps not a decisive reason) to think one’s (so far) best theory is wrong” (emphasis added).

In order to establish that it is possible to know P→Q on the basis of reason alone: (1) Do you need to establish ‘it is necessary that (P→Q)’? (2) Is it sufficient to establish ‘it is necessary that (P→Q)’? (3) Do you need to establish ‘P→(it is necessary that Q)’? (4) Is it sufficient to establish ‘P→(it is necessary that Q)’? (5-6) What about ‘(it is necessary that P)→Q’?

Or maybe none of (1)-(6) is correct?

Is (1) equivalent to (3) equivalent to (5)? Same question for (2), (4), and (6)?

5) In The Self-Confidence Principle 1-3, I focused on morality, but Sayre-McCord thinks that the conclusion that he reaches in “A Better Moral Theory” applies to other normative domains as well. What changes do I need to make in order to broaden the scope of my discussion in The Self-Confidence Principle 1-3?

6) Say we do discover “that on our understanding of what morality’s standards actually are, a different set of standards would be (morally) better.” Because of this discovery, we conclude that there is a problem with our current understanding of morality’s standards, and we decide to revise that understanding. How should we revise it? Our current understanding of morality’s standards is suspect, so we cannot rely on it to guide our revision. But what else is there?

The only answer I can think of is: There is some higher level principle that is used to regulate our understanding of morality’s standards. If there is such a principle, we can appeal to it in order to revise that understanding. The ideal of self-confidence provides such a principle.

7) How do we determine which moral theories are closer to the ideal of self-confidence and which are further away? Do we have to make moral judgments in order to make this determination? If so, we’ll have to rely on the moral theory that we currently accept. Is that a problem?

8) Onora O’Neill writes that, according to Kant, reason requires “that no principle incapable of being a law be relied upon as a fundamental principle for governing thought and action.”^[xi] Should this instead be ‘the fundamental principle for governing thought and action is that no principle incapable of being a law be relied upon’? The first version only places a restriction on our fundamental principles, not all our principles (although other principles would be restricted indirectly by the fundamental principles). The second restricts all our principles.

9) In Moral Incompleteness, I argued that every sound moral theory is incomplete. Sayre-McCord suggests that “our concept of morality is a concept of a system of rules, principles, and values than which none morally better can be conceived.”^[xii] If this is the case, then even the best conceivable moral theory is incomplete.

10) I said “It seems to me that the ideal of self-confidence can help us decide between competing moral theories.” An improved version of this post would give examples to illustrate how this works. However, writing is time consuming, and I do not think that it is worth my time to revise this post before I get feedback from other people. 

11) To pursue self-confidence as an ideal is to (a) hold that a better moral theory is the one that comes closer to self-confidence, everything else equal and (b) admit that there is little hope of constructing a theory that is self-confident. It seems to me that anyone who accepts the WR version of the self-confidence principle is committed to (a). Is this correct?

---

[i] Groundwork ed. Hill and Zweig page 222

[ii] As he puts it, “when we discover a theory of a better morality,” we discover “evidence that satisfying the criteria we are relying on does not ipso facto establish reasons” (“A Better Moral Theory” page 16 version of May 2018).

[iii] See especially questions 5 and 6.

[iv] In “Hume on Practical Morality and Inert Reason,” Sayre-McCord’s aim is to render it deductive.

[v] In “Hume on Practical Morality and Inert Reason,” Sayre-McCord argues that, for Hume, rationality also includes the disposition to form beliefs that line up with the recommendations that reason makes.

[vi] T 3.1.1.10

In “Hume on Practical Morality and Inert Reason,” Sayre-McCord emphasizes the importance of the qualifier “by contradicting or approving of it.”

[vii] “Reason is wholly inactive, and can never be the source of so active a principle as conscience, or a sense of morals” (T 3.1.1.10)

[viii] “Kant begins by characterizing a will as ‘a kind of causality that living beings have so far as they are rational.’ Unlike computers and robots, human beings have wills because they are living beings with the capacity to make things happen, intentionally and for reasons. We can bring about changes in the world in accord with our purposes and principles.” (“Editor’s Introduction” to Groundwork ed. Hill and Zweig page 94).

[ix] This would allow him to admit that statements such as ‘morality requires you to do your duty’ can be known on the basis of reason alone. On the other hand, perhaps Hume thinks that it is not possible to rule out the error theory without empirical investigation, and perhaps the claim ‘morality requires you to do your duty’ cannot be established without ruling out the error theory.

Also, by ‘something substantial’ I mean ‘specific recommendations about how to act or react.’ Is this a precise enough definition?

[x] T 2.3.3.6

[xi] “Vindicating Reason” page 296 in The Cambridge Companion to Kant 

[xii] “A Better Moral Theory” page 17 version of May 2018

Previous posts in this series:
The Self-Confidence Principle 1
The Self-Confidence Principle 2

In my last post, I introduced the self-confidence principle. I will begin this post by clarifying it. As it stands, the self-confidence principle runs:

If a moral theory is not self-confident, then there is reason to think that it is incorrect.

The self-confidence principle identifies a consideration that we should take into account when we assess a moral theory. However, it does not tell us whether that consideration is decisive. If it is decisive, then a moral theory needs to be self-confident in order to be correct. If it is not decisive, then it is possible for a moral theory to be correct even though it lacks self-confidence. To keep things clear, it is best to distinguish between two different versions of the self-confidence principle:

The Strong Self-Confidence Principle: If a moral theory is not self-confident, then there is decisive reason to think that it is incorrect.

The Weak Self-Confidence Principle: If a moral theory is not self-confident, then there is defeasible reason to think that it is incorrect.

In “A Better Moral Theory,” Sayre-McCord sometimes writes as though a moral theory needs to be self-confident in order to be correct. For example, he writes that:

…an acceptable normative theory in the relevant domain must not be one that fails to meet the justificatory standards it sets. Again, satisfying this requirement does not establish a normative theory as right, but failing to meet it rules a theory out as mistaken.

“A Better Moral Theory” page 15 version of May 2018

In other places, he seems to say that a moral theory does not need to be self-confident in order to be correct. For example, he writes that:

There being a difference between one’s theory of the best normative X (the best morality, the best standards of inference, the best rules of justification…) and one’s (so far) best theory of X, necessarily provides a reason (though perhaps not a decisive reason) to think one’s (so far) best theory is wrong.

“A Better Moral Theory” page 1 draft of May 2018

This seeming inconsistency can be resolved by distinguishing between (1) the correct moral theory and (2) the correct moral theory to hold given the available evidence. I will refer to this distinction by speaking of ‘the perfect moral theory’ and ‘the best available moral theory.’

The distinction between the perfect moral theory and the best available moral theory cuts across the distinction between the strong self-confidence principle and the weak self-confidence principle, creating four versions of the self-confidence principle:

The Strong Idealistic Self-Confidence Principle: If a moral theory is not self-confident, then there is decisive reason to think that it is not perfect.

The Weak Idealistic Self-Confidence Principle: If a moral theory is not self-confident, then there is defeasible reason to think that it is not perfect.

The Strong Realistic Self-Confidence Principle: If a moral theory is not self-confident, then there is decisive reason to think that it is not the best available moral theory.

The Weak Realistic Self-Confidence Principle: If a moral theory is not self-confident, then there is defeasible reason to think that it is not the best available moral theory.

It seems to me that the SI self-confidence principle and the WR self-confidence principle are correct, but the WI self-confidence principle and the SR self-confidence principle are incorrect. If a theory lacks self-confidence, then it is certain that it is not perfect. However, it may still be the best available theory, so long as it does a better job of accommodating the evidence than any of the available alternatives.

Now that all four versions of the self-confidence principle are in play, it is possible to resolve the apparent inconsistency in “A Better Moral Theory.” On the one hand, when Sayre-McCord suggests that a moral theory needs to be self-confident in order to be correct, he can be understood as appealing to the SI self-confidence principle. On the other hand, when he suggests that a moral theory does not need to be self-confident in order to be correct, he can be understood as appealing to the WR self-confidence principle.

The second change that I think Sayre-McCord should consider making to “A Better Moral Theory” is to clearly distinguish appeals to the SI self-confidence principle from appeals to the WR self-confidence principle.

Conclusion

I began this series of posts by arguing that Enoch’s GGp→Gp formula can be defended by clarifying the operator ‘G’. Once that clarification is made, the obvious next step is to state the self-confidence principle. Once the self-confidence principle is on the table, the obvious next step is to clarify it by distinguishing between four different versions of it. After that, it becomes clear that a seeming inconsistency in “A Better Moral Theory” can be resolved by appealing to the difference between the SI self-confidence principle and the WR self-confidence principle.  

Previous post in this series:
The Self-Confidence Principle 1

The purpose of this series of posts is to suggest two changes to Geoffrey Sayre-McCord’s draft paper “On a Theory of a Better Moral Theory and a Better Theory of Morality.” In my last post, I made the first suggestion. Before making the second, I need to introduce a rule that I will call the Self-Confidence Principle. I will do so by breaking the self-confidence principle down into its component parts. To begin with, consider:

The GGp→Gp Principle: If a moral theory does not follow the GGp→Gp rule, then there is reason to think that it is incorrect.

Moral theory T fails to follow the GGp→Gp rule if and only if (1) according to T, it is good for T to deliver the verdict that it is good that p and (2) T does not deliver the verdict that it is good that p.^[1] If both (1) and (2) are the case, then theory T fails to deliver a verdict that, by its own lights, is a good verdict. By its own lights, then, T has a problem. According to the GGp→Gp principle, this problem constitutes a reason to think that T is incorrect.

The thought behind the GGp→Gp principle is not limited to claims about goodness.^[2] For example, consider statements such as ‘If moral theory T delivers the verdict that it is good that T delivers the verdict that it is bad that p, then T delivers the verdict that it is bad that p’ and ‘If moral theory T delivers the verdict that it is good that T delivers the verdict that it is obligatory that p, then T delivers the verdict that it is obligatory that p.’ These statements can be abbreviated as GBp→Bp and GOp→Op. In order to capture the thought behind these statements, the GGp→Gp principle needs to be generalized. Towards that end, define:

 Vp: Moral theory T delivers verdict V about p, where that verdict is construed as a judgment about the way the world is, not as a hope about the way it might be.

Using the operator ‘V,’ the GGp→Gp principle can be generalized into:

The GVp→Vp Principle: If a moral theory does not follow the GVp→Vp rule, then there is reason to think that it is incorrect.^[3]

To illustrate the GVp→Vp principle, consider an example. Suppose that, according to utilitarianism, it is good for utilitarianism to deliver the verdict that stealing is always wrong. Further suppose that utilitarianism does not deliver the verdict that stealing is always wrong. If this is the case, the utilitarianism fails to deliver a verdict that, by its own lights, is a good verdict. By its own lights, then, utilitarianism has a problem. According to the GVp→Vp principle, this problem constitutes a reason to think that utilitarianism is incorrect.

The GVp→Vp principle is the first of two components of the self-confidence principle. The second is the Vp→GVp principle:

The Vp→GVp Principle: If a moral theory does not follow the Vp→GVp rule, then there is reason to think that it is incorrect.

Moral theory T fails to follow the Vp→GVp rule if and only if (1) T delivers verdict Vp and (2) it is not the case that, according to T, it is good for T to deliver verdict Vp. If both (1) and (2) are the case, then theory T delivers a verdict that, by its own lights, is not a good verdict. By its own lights, then, T has a problem. According to the Vp→GVp principle, this problem constitutes a reason to think that T is incorrect.

For example, suppose that utilitarianism delivers the verdict that it is sometimes permissible to steal. Further suppose that, according to utilitarianism, it is not the case that it is good for utilitarianism to deliver that verdict that it is sometimes permissible to steal. If this is the case, then utilitarianism delivers a verdict that, by its own lights, is not a good verdict. By its own lights, then, utilitarianism has a problem. According to the Vp→GVp principle, this problem constitutes a reason to think that utilitarianism is incorrect.

Say that a moral theory is self-confident if and only if it follows both the GVp→Vp rule and the Vp→GVp rule, or, we might say, follows the Vp↔GVp rule. Using this terminology, the self-confidence principle can be stated as follows:

The Self-Confidence Principle: If a moral theory is not self-confident, then there is reason to think that it is incorrect.

A self-confident theory delivers all and only those verdicts that, by its own lights, are good verdicts. If a theory is not self-confident, then, by its own lights, it has a problem. According to the self-confidence principle, this problem constitutes a reason to think that it is incorrect.

In my next post, I will finally get around to suggesting the second change that I think Sayre-McCord should consider making to “A Better Moral Theory.”

---

[1] It is worth noting that the counterexamples to the GGp→Gp formula that van Someren Greve and Sayre-McCord offer are not counterexamples to the GGp→Gp principle. The examples purport to show that a moral theory could deliver verdicts that follow the GGp→Gp rule yet still be inadequate. A counterexample to the GGp→Gp principle must show that, even though a given moral theory does not follow the GGp→Gp rule, there is no reason to think that it is incorrect.

[2] Sayre-McCord makes a similar point on page 12 of “A Better Moral Theory” version of May 2018.

[3] At first glance, the GVp→Vp principle resembles a principle of completeness. This resemblance, however, is misleading. Say that a theory of domain D is complete if and only if all true statements about the objects in D are provable within the theory. In this definition of completeness, two standards of correctness are in play, truth and provability. Stated imprecisely, the GVp→Vp principle tells us that, in the domain of morality, if a verdict is good, then it can be derived from moral theory T. When the principle is stated in this imprecise way, it appears that there are two standards of correctness in play, goodness and derivability from the theory. Stated precisely, the GVp→Vp principle tell us that, in the domain of morality, if a verdict is good according to theory T, then theory that verdict can be derived from T. When the principle is stated this way, it is clear that only one standard of correctness is in play, namely, derivability from the theory. No outside standard of goodness is in play. This point makes it clear that the GVp→Vp principle is a principle of self-assessment.

The purpose of this series of posts is to suggest changes to Geoffrey Sayre-McCord’s draft paper “On a Theory of a Better Moral Theory and a Better Theory of Morality.”^[1] Some of the terminology that Sayre-McCord uses in “A Better Moral Theory” is drawn from David Enoch’s paper “Wouldn’t It Be Nice If p, Therefore, p (for a moral p).” For this reason, I will begin by discussing Enoch’s paper.

Enoch on the GGp→Gp Formula

In “Wouldn’t It Be Nice,” Enoch focuses on arguments that run ‘It would be good for q to be the case, therefore q is the case.’ As he points out, these arguments are rarely convincing. Most of the time, they are “ridiculous, paradigmatic cases of wishful thinking.”^[2] However, he thinks that they are not always hopeless. There is hope, he suggests, when q stands for ‘it would be good if p.’ In this case, the argument form under consideration becomes ‘It would be good for it to be good that p, therefore it is good that p.’ Enoch abbreviates this argument form as GGp→Gp.

To illustrate the GGp→Gp argument form, Enoch gives the following example:

Suppose that a world in which we have an utterly non-consequentialist moral status is a better world than one in which we don’t have such a status. Does this give any reason to believe that we have such moral status? 

“Wouldn’t It be Nice” 223

If it is true that GGp→Gp, then the answer to this question is ‘yes.’

First Objection to Enoch

In the remainder of this post, I will discuss two objections to Enoch. In my opinion, both objections can be overcome by clarifying the GGp→Gp argument form. The first objection is raised by Rob van Someren Greve in a paper called “Wishful Thinking in Moral Theorizing: Comment on Enoch.” The second is raised by Sayre-McCord in “A Better Moral Theory.” Both objections take the form of counterexamples.

To paraphrase van Someren Greve’s counterexample, suppose that p stands for ‘the entire actual world.’ Then GGp→Gp becomes ‘It would be good for the entire actual world to be good, therefore the entire actual world is good’.^[3] Certainly this line of reasoning is ridiculous wishful thinking.

Sayre-McCord says of van Someren Greve’s counterexample that “this version of a counter example doesn’t really work, though, since ‘good’ is not being used consistently as an operator.”^[4] However, Sayre-McCord does not say what the inconsistency is. Once the inconsistency is identified, I will argue, it becomes clear that Sayre-McCord’s counterexample is inconsistent in the same way. 

Reply to the First Objection

As I have said, I think that van Someren Greve’s counterexample can be dealt with by clarifying the GGp→Gp argument form. Two clarifications are in order.

First, van Someren Greve’s counterexample does not clearly distinguish between (a) the way the world actually is and (b) hopes about the way it might be. To keep things clear, I will use ‘it would be good if p’ to express hopes about the way the world might be and ‘p is good’ or ‘it is good that p’ to express judgments about the way it actually is.^[5]

Second, for the purposes of the GGp→Gp formula, ‘it is good that p’ should be understood to mean ‘moral theory T delivers the verdict that it is good that p.’^[6] When both clarifications are combined, the understanding of Gp that results is as follows:

Gp: Moral theory T delivers the verdict that it is good that p, where that verdict is construed as a judgment about the way the world is, not as a hope about the way it might be.

When Gp is understood in this way, the GGp→Gp formula becomes ‘If moral theory T delivers the verdict that it is good that moral theory T delivers the verdict that it is good that p, then moral theory T delivers the verdict that it is good that p.’

van Someren Greve’s example is not a counterexample to this version of the GGp→Gp formula. It is clear that ‘the actual world’ may be substituted for p without generating an unacceptable result. The statement ‘If it would be good for the entire actual world to be good, then the entire actual world is good’ generates an unacceptable result because its antecedent is obviously true. It is not obvious, however, that the antecedent of ‘If moral theory T delivers the verdict that it is good that moral theory T delivers the verdict that the entire actual world is good, then moral theory T delivers the verdict that the entire actual world is good’ is true. Whether or not it is true depends on the theory in question. Even if it is true, ‘moral theory T delivers the verdict that the entire actual world is good’ may be true as well.

Second Objection to Enoch

Sayre-McCord’s doubts about the GGp→Gp argument form rest on the following counterexample. Suppose that p stands for ‘you express sincere appreciation.’ Then GGp→Gp becomes ‘If it would be good for it to be good for you to express sincere appreciation, then it is good for you to express sincere appreciation.’ For the sake of argument, assume that it is good for you to express sincere appreciation when someone has done something good for you. If we make this assumption, then the fact that it is good for you to express sincere appreciation reflects the fact that something good has happened, namely, that someone has done something good for you. For this reason, it is true that it would be good for it to be good for you to express sincere appreciation. In other words, it is the case that GGp. However, it does not follow that Gp, for it may be the case that, in fact, no one has done anything good for you.

Reply to the Second Objection

Sayre-McCord’s counterexample can be dealt with in the same way as van Someren Greve’s. Say that GGp→Gp stands for ‘If moral theory T delivers the verdict that it is good that moral theory T delivers the verdict that it is good that p, then moral theory T delivers the verdict that it is good that p.’ When GGp→Gp is understood in this way, ‘you express sincere appreciation’ may be substituted for p without generating an unacceptable result. The statement ‘If it would be good for it to be good for you to express sincere appreciation, then it is good for you to express sincere appreciation’ generates an unacceptable result because its antecedent is true (it is true, as Sayre-McCord says, because the fact that it is good for you to express sincere appreciation reflects the fact that something good has happened, namely, that someone has done something good for you). It is not obvious, however, that the antecedent of ‘If moral theory T delivers the verdict that it is good that moral theory T delivers the verdict that it is good for you to express sincere appreciation, then moral theory T delivers the verdict that it is good for you to express sincere appreciation’ is true. Whether or not it is true depends on the theory in question. Even if it is true, ‘moral theory T delivers the verdict that it is good for you to express sincere appreciation’ may be true as well.^[7]

Conclusion

In “A Better Moral Theory,” Sayre-McCord rejects the GGp→Gp formula because he thinks that it is subject to counterexample. In my opinion, once the GGp→Gp formula is properly clarified, neither van Someren Greve’s example nor Sayre-McCord’s is a counterexample to it. I suggest modifying “A Better Moral Theory” to take this into account. 

In my next post, I will generalize and elaborate on the GGp→Gp formula. In the post after that, I will suggest another change to “A Better Moral Theory.”

---

[1] Sayre-McCord, G. (2018, May) On a Theory of a Better Moral Theory and a Better Theory of Morality. Retrieved from http://philosophy.unc.edu/people/geoffrey-sayre-mccord

[2] Enoch, D. (2009). Wouldn’t It Be Nice If p, Therefore, p (for a Moral p). Utilitas, 21(2), 1. Retrieved from http://ssrn.com/abstract=2607074

[3] van Someren Greve, R. Wishful Thinking in Moral Theorizing: Comment on Enoch. Utilitas, 23(4).

van Someren Greve phrases this counterexample as “for every state of affairs that the actual world contains, it is true that (it would be good if that state of affairs were good).”

[4] “A Better Moral Theory” page 11 version of May 2018

[5] It is difficult to state van Someren Greve’s and Sayre-McCord’s examples in a way that both respects this distinction and is grammatical. Both examples center on the phrase ‘It would be good if p were good,’ which is supposed to correspond to GGp. This phrase should not be restated as ‘It would be good if it would be good if p’ because that is difficult even to understand. It should not be restated as ‘It would be good if it is good that p’ because that is not grammatical. It should not it be restated as ‘It is good that it is good that p’ because that does not capture the intention of either author. Finally, ‘It is good that it would be good if p’ is so odd that I am not sure whether it is ungrammatical or grammatical but difficult to understand. These difficulties suggest that the operator ‘good’ requires further clarification if it is to be used in the context of the GGp→Gp formula. I provide this clarification presently.

[6] Sayre-McCord uses expressions such as ‘by the lights of the original theory’ in a number of places in “A Better Morality.” For example, he says “still, imagine that we (tentatively, but with some confidence) arrive at a theory of what the best standards of morality would be (by the lights of our original theory)” (“A Better Moral Theory” page 3 version of May 2018). Expressions such as “by the lights of our original theory” suggest understanding Gp to mean not ‘p is good’ but ‘theory T delivers the verdict that p is good.’ However, Sayre-McCord does not clarify the GGpGp formula by explicitly stating that Gp ought to be understood in this way. As a result, he rejects the GGpGp formula.

[7] I would like to express sincere appreciation to anyone who actually read this paragraph.