Kant famously held that reason plays a fundamental role in our grasp of morality. Because he believed that reason was outside the empirical realm (Kant Reference Kant and Gregor1996, p. 99), Kant denied that our moral judgments could be ultimately understood through empirical investigation (though he gave “moral anthropology” an important secondary role [Kant Reference Kant and Gregor1996, p. 372]). Kant even claimed that moral facts could come radically apart from any empirically detectable facts (Kant Reference Kant, Guyer and Wood1998, p. 544). However, the philosophical climate has shifted from Kant's time. Most philosophers today believe that any complete account of human moral judgment must be closely tied to empirical psychology. Some, such as Joshua Greene, think this bodes poorly for deontological rationalist views like Kant's (Greene Reference Greene and Sinnott-Armstrong2008).
Joshua May's (Reference May2018) Regard for Reason in the Moral Mind argues that the best empirical evidence does not threaten either the rationalist claim that moral judgments are based in reason or our acceptance of broadly deontological moral principles. The book is clearly written, philosophically rich, and enjoyable to read. Its optimistic claims and tone was, for this reader at least, very welcome. More often than not, I found May's arguments persuasive. In case after case, May provides a “hold on – let's look at the details” check on empirical results from which many have drawn pessimistic, anti-rationalist conclusions. Whether or not May intends it to be, his general defensive approach is itself in the Kantian tradition, because Kant principally emphasized that the empirical facts left room for strong moral facts.
Although May's primary aim is to defend a broadly rationalist view, he also offers some pieces of the particular moral view he himself is drawn to (on both philosophical and empirical grounds). That view is more modest than many earlier rationalist moral views, but it is far from trivial. May claims that “a creature with unlimited time and resources needn't possess emotion [or, more specifically, affect] to make distinctively moral judgments” (p. 13 – though May later allows that emotional affect might be a requirement for all cognition [p. 80]). In chapter 3, May offers four principles that may underlie moral inference, only one of which is directly consequentialist:
Consequentialist Principle: All else being equal, an action is morally worse if it leads to more harm than other available alternatives. (p. 57)
Intentionality Principle: All else being equal, it's morally worse to cause harm intentionally as opposed to accidentally. (p. 61)
Action Principle: All else being equal, harm caused by action is morally worse than harm consequent upon omission. (p. 62)
Principle of Agential Involvement: All else being equal, it is morally worse for an agent to be more involved in bringing about a harmful outcome. (p. 69)
May does not commit to these principles being foundational, though he notes that he is inclined to think there are at least some foundational moral principles which guide our thought in the roughly the same way that rules of grammar do (see pp. 70, 78–79). Though May generally characterizes reason as the capacity for inference (basing beliefs on other beliefs [p. 11]), he does not claim that these principles are the result of inferences. They would count as rational, presumably, just by being non-affective and by supporting inferences in some way.
My aim here is to assess the form of rationalism that would hold if May's four principles were the rational foundation of moral belief. The contrast I will consider is with a sentimentalist view that agrees with May about the psychology reality of the principles, but takes them to somehow rest on affect. May acknowledges that a sentimentalist could agree about these principles (pp. 55–56, 71), but argues at length in chapter 2 that we lack any strong empirical grounds for accepting sentimentalism. In this discussion, I am going to assume that May is right about the empirical factors he considers, and instead offer abductive considerations that favor sentimentalism about the principles over rationalism. If I am right about those considerations, then May's most obvious line of response will be to revise or expand the principles. My remarks are thus primarily intended to provide May an opportunity to further develop his positive view.
To begin, consider again the most famous rationalist view of moral judgment: Kant's. Kant held that there is a single fundamental moral principle, whose main formulation is: “act only in accordance with that maxim through which you can at the same time will that it become a universal law” (Kant Reference Kant and Gregor1996, p. 73). Though there are hard questions about how exactly Kant's principle gets applied (see, e.g., Herman Reference Herman1993), it is fairly clear that Kant believed that the principle of reason deductively entails some substantive verdicts about what to do. This is what we would expect given Kant's general view of reason, which, in his mature work, he introduced in terms of syllogistic inference (Kant Reference Kant, Guyer and Wood1998, pp. 387–91). Kant was not alone in this – other rationalists like Spinoza also believed that foundational moral principles have some substantive, deductive implications (see, e.g., Spinoza Reference Spinoza and Curley1988, pp. 586–87). No doubt this was tied to their view of reason as the capacity that allows us to make mathematical inferences and grasp necessary truths.
By contrast, May's four principles have no substantive deductive implications by themselves, because they are ceteris paribus principles. They apply only when other things are equal, but give no indication of which other things are relevant. For all the principles say, for example, consequentialist considerations cease to be relevant when trolleys are involved. Hence, if these principles were the true foundations of moral judgment, then any substantive moral judgment based on them (as in the trolley case) would involve an irrational inferential jump. This matters both for May's account of moral judgment and for his defense, in chapter 8, of the idea that moral beliefs can generate the primary motivation for a particular action by generating a desire. It is hard to see how this later idea could work if the beliefs in question had no substantive implications for particular actions.
Contrast this with rules of grammar for English, such as the rule that every declarative sentence requires a verb. At face value, this principle has substantive deductive implications – it dictates that “I cat” is not a declarative sentence. That does not mean that speakers of English have an articulated belief in the rule, of course. But it has some psychological reality for them, and explains how they are able (without any irrational inferential jumps) to recognize some novel sentences as problematic. By contrast, consider the difficulty in learning spelling in English or learning logograms in Chinese. Although certain rules of thumb apply, the rules are not generative, so brute memorization is needed for accurate recognition.
Presumably, though, May would take fundamental moral principles to be more than rules of thumb that require supplementation by memorization. Structurally, then, they seem to be closer to rules about seasoning foods such as “other things being equal, add some salt.” This principle has no direct substantive entailments. It is not, however, merely a rule of thumb that is supplemented by memorization. Instead, it gestures at a pattern of responses we have to food: most humans respond positively to salt, but not always. These responses are not determined by reason in any sense, however, but are instead arational and affective. Hence, I suggest that if the most foundational principles for a domain are merely ceteris paribus principles, we have abductive reason to think that the principles are really just gesturing at some arational pattern of responses.
May's book offers what might be a potential response by analogy: our way of classifying objects as furniture or non-furniture (p. 70). There might be principles of furniture-identification, such as “if the function of an object is for sitting, then it is furniture.” Such principles, though, “merely identify prototypical features that are statistically frequent in the category or exemplars with which I can compare the object in question” (p. 70), and are stronger than mere ceteris paribus principles. Recognizing furniture, however, is a broadly rational accomplishment, in May's broad sense of “rational,” because it involves forming a belief on the basis of other beliefs – albeit in a non-deductive way.
Understanding May's principles in this way, however, suggests that they are not really foundational principles, but instead stem from some more foundational representations of moral prototypes. By contrast, principles of grammar and seasoning do not seem to hinge on prototypes, though we might use prototypes to identify particularly good examples of grammar and seasoning. This may be connected to why it seems harder to explain cross-historical and cross-cultural convergence by appeal to prototypes than to innate rules (this is part of why traditional rationalists modeled their moral principles on logical and mathematical principles). Because May is drawn toward the grammar analogy partly to explain intercultural convergence (p. 78), he therefore has reason to not rely too heavily on the furniture analogy.
I will make one more broadly Kantian point. Arguably, part of the reason that Kant held that the moral law was unconditionally binding (“categorical”) was that its verdicts were clear (at least, when the right questions were posed). However, it seems that one way to not be bound by a principle is to be unable to see what it implies. For example, there are limits to how much we can hold someone accountable for being un-American (even given full knowledge of other relevant facts), because it is often not clear what being “American” requires. Likewise, there are many cases where we would not fault someone for being unsure whether something was furniture, even if she had a full knowledge of the physical properties of the item in question. If we do indeed regard morality (unlike convention) as unconditionally binding in situations where we know the relevant non-moral facts, then that suggests that we take moral requirements to be reliably clear. Assuming we are coherent in seeing morality as binding and clear, it would therefore seem that we must take morality to be guided by something with clearer substantive implications than ceteris paribus principles or exemplars. If absolutist principles like Kant's are off the table, then strong affects would seem like the best candidate, because even in novel cases, we are often very confident (albeit sometimes wrongly) about how other people will respond to a given level of saltiness. Of course, May could say that we are simply wrong to assume that our rational principles have clear implications, and so wrong to see them as unconditionally binding, but attributing such an error to us seems like a cost of a view.
In sum, then, if May's four principles are the foundations of moral judgment, there is reason to think that, like principles of seasoning, they rest on sentiments in some way. The obvious response is for May to deny that, as stated, these principles exhaust the foundations of moral judgment, perhaps leaving the matter up to further empirical investigation. I expect that May is not willing to go as far as Kant and identify a single clear principle with substantive, deductive implications, because he later appeals to the fact that there is “nearly always … wiggle room in the application of moral principles” (p. 167). Even so, there is plenty of middle ground between weak ceteris paribus principles and a Kantian view. I therefore hope that May will develop his view further, perhaps in a way that can explain why we are inclined to think that the right moral answer is often obvious, even in novel situations.
Kant famously held that reason plays a fundamental role in our grasp of morality. Because he believed that reason was outside the empirical realm (Kant Reference Kant and Gregor1996, p. 99), Kant denied that our moral judgments could be ultimately understood through empirical investigation (though he gave “moral anthropology” an important secondary role [Kant Reference Kant and Gregor1996, p. 372]). Kant even claimed that moral facts could come radically apart from any empirically detectable facts (Kant Reference Kant, Guyer and Wood1998, p. 544). However, the philosophical climate has shifted from Kant's time. Most philosophers today believe that any complete account of human moral judgment must be closely tied to empirical psychology. Some, such as Joshua Greene, think this bodes poorly for deontological rationalist views like Kant's (Greene Reference Greene and Sinnott-Armstrong2008).
Joshua May's (Reference May2018) Regard for Reason in the Moral Mind argues that the best empirical evidence does not threaten either the rationalist claim that moral judgments are based in reason or our acceptance of broadly deontological moral principles. The book is clearly written, philosophically rich, and enjoyable to read. Its optimistic claims and tone was, for this reader at least, very welcome. More often than not, I found May's arguments persuasive. In case after case, May provides a “hold on – let's look at the details” check on empirical results from which many have drawn pessimistic, anti-rationalist conclusions. Whether or not May intends it to be, his general defensive approach is itself in the Kantian tradition, because Kant principally emphasized that the empirical facts left room for strong moral facts.
Although May's primary aim is to defend a broadly rationalist view, he also offers some pieces of the particular moral view he himself is drawn to (on both philosophical and empirical grounds). That view is more modest than many earlier rationalist moral views, but it is far from trivial. May claims that “a creature with unlimited time and resources needn't possess emotion [or, more specifically, affect] to make distinctively moral judgments” (p. 13 – though May later allows that emotional affect might be a requirement for all cognition [p. 80]). In chapter 3, May offers four principles that may underlie moral inference, only one of which is directly consequentialist:
Consequentialist Principle: All else being equal, an action is morally worse if it leads to more harm than other available alternatives. (p. 57)
Intentionality Principle: All else being equal, it's morally worse to cause harm intentionally as opposed to accidentally. (p. 61)
Action Principle: All else being equal, harm caused by action is morally worse than harm consequent upon omission. (p. 62)
Principle of Agential Involvement: All else being equal, it is morally worse for an agent to be more involved in bringing about a harmful outcome. (p. 69)
May does not commit to these principles being foundational, though he notes that he is inclined to think there are at least some foundational moral principles which guide our thought in the roughly the same way that rules of grammar do (see pp. 70, 78–79). Though May generally characterizes reason as the capacity for inference (basing beliefs on other beliefs [p. 11]), he does not claim that these principles are the result of inferences. They would count as rational, presumably, just by being non-affective and by supporting inferences in some way.
My aim here is to assess the form of rationalism that would hold if May's four principles were the rational foundation of moral belief. The contrast I will consider is with a sentimentalist view that agrees with May about the psychology reality of the principles, but takes them to somehow rest on affect. May acknowledges that a sentimentalist could agree about these principles (pp. 55–56, 71), but argues at length in chapter 2 that we lack any strong empirical grounds for accepting sentimentalism. In this discussion, I am going to assume that May is right about the empirical factors he considers, and instead offer abductive considerations that favor sentimentalism about the principles over rationalism. If I am right about those considerations, then May's most obvious line of response will be to revise or expand the principles. My remarks are thus primarily intended to provide May an opportunity to further develop his positive view.
To begin, consider again the most famous rationalist view of moral judgment: Kant's. Kant held that there is a single fundamental moral principle, whose main formulation is: “act only in accordance with that maxim through which you can at the same time will that it become a universal law” (Kant Reference Kant and Gregor1996, p. 73). Though there are hard questions about how exactly Kant's principle gets applied (see, e.g., Herman Reference Herman1993), it is fairly clear that Kant believed that the principle of reason deductively entails some substantive verdicts about what to do. This is what we would expect given Kant's general view of reason, which, in his mature work, he introduced in terms of syllogistic inference (Kant Reference Kant, Guyer and Wood1998, pp. 387–91). Kant was not alone in this – other rationalists like Spinoza also believed that foundational moral principles have some substantive, deductive implications (see, e.g., Spinoza Reference Spinoza and Curley1988, pp. 586–87). No doubt this was tied to their view of reason as the capacity that allows us to make mathematical inferences and grasp necessary truths.
By contrast, May's four principles have no substantive deductive implications by themselves, because they are ceteris paribus principles. They apply only when other things are equal, but give no indication of which other things are relevant. For all the principles say, for example, consequentialist considerations cease to be relevant when trolleys are involved. Hence, if these principles were the true foundations of moral judgment, then any substantive moral judgment based on them (as in the trolley case) would involve an irrational inferential jump. This matters both for May's account of moral judgment and for his defense, in chapter 8, of the idea that moral beliefs can generate the primary motivation for a particular action by generating a desire. It is hard to see how this later idea could work if the beliefs in question had no substantive implications for particular actions.
Contrast this with rules of grammar for English, such as the rule that every declarative sentence requires a verb. At face value, this principle has substantive deductive implications – it dictates that “I cat” is not a declarative sentence. That does not mean that speakers of English have an articulated belief in the rule, of course. But it has some psychological reality for them, and explains how they are able (without any irrational inferential jumps) to recognize some novel sentences as problematic. By contrast, consider the difficulty in learning spelling in English or learning logograms in Chinese. Although certain rules of thumb apply, the rules are not generative, so brute memorization is needed for accurate recognition.
Presumably, though, May would take fundamental moral principles to be more than rules of thumb that require supplementation by memorization. Structurally, then, they seem to be closer to rules about seasoning foods such as “other things being equal, add some salt.” This principle has no direct substantive entailments. It is not, however, merely a rule of thumb that is supplemented by memorization. Instead, it gestures at a pattern of responses we have to food: most humans respond positively to salt, but not always. These responses are not determined by reason in any sense, however, but are instead arational and affective. Hence, I suggest that if the most foundational principles for a domain are merely ceteris paribus principles, we have abductive reason to think that the principles are really just gesturing at some arational pattern of responses.
May's book offers what might be a potential response by analogy: our way of classifying objects as furniture or non-furniture (p. 70). There might be principles of furniture-identification, such as “if the function of an object is for sitting, then it is furniture.” Such principles, though, “merely identify prototypical features that are statistically frequent in the category or exemplars with which I can compare the object in question” (p. 70), and are stronger than mere ceteris paribus principles. Recognizing furniture, however, is a broadly rational accomplishment, in May's broad sense of “rational,” because it involves forming a belief on the basis of other beliefs – albeit in a non-deductive way.
Understanding May's principles in this way, however, suggests that they are not really foundational principles, but instead stem from some more foundational representations of moral prototypes. By contrast, principles of grammar and seasoning do not seem to hinge on prototypes, though we might use prototypes to identify particularly good examples of grammar and seasoning. This may be connected to why it seems harder to explain cross-historical and cross-cultural convergence by appeal to prototypes than to innate rules (this is part of why traditional rationalists modeled their moral principles on logical and mathematical principles). Because May is drawn toward the grammar analogy partly to explain intercultural convergence (p. 78), he therefore has reason to not rely too heavily on the furniture analogy.
I will make one more broadly Kantian point. Arguably, part of the reason that Kant held that the moral law was unconditionally binding (“categorical”) was that its verdicts were clear (at least, when the right questions were posed). However, it seems that one way to not be bound by a principle is to be unable to see what it implies. For example, there are limits to how much we can hold someone accountable for being un-American (even given full knowledge of other relevant facts), because it is often not clear what being “American” requires. Likewise, there are many cases where we would not fault someone for being unsure whether something was furniture, even if she had a full knowledge of the physical properties of the item in question. If we do indeed regard morality (unlike convention) as unconditionally binding in situations where we know the relevant non-moral facts, then that suggests that we take moral requirements to be reliably clear. Assuming we are coherent in seeing morality as binding and clear, it would therefore seem that we must take morality to be guided by something with clearer substantive implications than ceteris paribus principles or exemplars. If absolutist principles like Kant's are off the table, then strong affects would seem like the best candidate, because even in novel cases, we are often very confident (albeit sometimes wrongly) about how other people will respond to a given level of saltiness. Of course, May could say that we are simply wrong to assume that our rational principles have clear implications, and so wrong to see them as unconditionally binding, but attributing such an error to us seems like a cost of a view.
In sum, then, if May's four principles are the foundations of moral judgment, there is reason to think that, like principles of seasoning, they rest on sentiments in some way. The obvious response is for May to deny that, as stated, these principles exhaust the foundations of moral judgment, perhaps leaving the matter up to further empirical investigation. I expect that May is not willing to go as far as Kant and identify a single clear principle with substantive, deductive implications, because he later appeals to the fact that there is “nearly always … wiggle room in the application of moral principles” (p. 167). Even so, there is plenty of middle ground between weak ceteris paribus principles and a Kantian view. I therefore hope that May will develop his view further, perhaps in a way that can explain why we are inclined to think that the right moral answer is often obvious, even in novel situations.