In the first part of Regard for Reason in the Moral Mind, Joshua May (Reference May2018) mounts a bold defense of a form of moral rationalism against sentimentalism. But what exactly is his target, and does he offer a credible alternative?
As I have observed previously (Kauppinen Reference Kauppinen and Zalta2013b), sentimentalism comes in many logically independent forms, in which emotions or more broadly pro- and con-attitudes play different roles. Explanatory sentimentalists hold that sentimental reactions fundamentally explain our moral verdicts; judgment sentimentalists hold that moral judgments consist in sentiments or otherwise make essential reference to sentiment; metaphysical sentimentalists hold that moral properties are grounded in actual or possible sentimental responses; and epistemic sentimentalists hold that we come to know moral truths ultimately by way of sentimental responses. Sentimentalists offer different sorts of a priori arguments for these claims, appealing, for example, to the apparent importance of attitudes that have a world-to-mind fit in explaining the action-guiding character of moral thought. Recently, some sentimentalists, most notably Jesse Prinz (Reference Prinz2007) and Shaun Nichols (Reference Nichols2004), have also offered a posteriori arguments for these views, drawing on scientific findings.
It is the a posteriori arguments that are May's main target, although he merely points to arguments of others when it comes to a priori sentimentalism. This is worth emphasizing for two reasons. First, though May mounts a very promising case against the a posteriori arguments, we may nevertheless have sufficient reason to subscribe to a sentimentalist view on a priori grounds. Second, I think it is fair to say that what defines the various sentimentalist views are the conclusions of the a priori arguments. Only explanatory sentimentalists, for example, are committed to causal claims about the role of emotion in generating moral judgments, and these claims are sometimes significantly weaker than May's targets. Adam Smith, for example, holds that “the greater part of our moral judgments […] is regulated by maxims and ideas derived from an induction of reason,” while arguing that it is “absurd and unintelligible to suppose that the first perceptions of right and wrong can be derived from reason” (Smith Reference Smith and Haakonssen2002, p. 377). On this kind of view, emotions do not play a causal role in every moral judgment, but rather explain why we find certain act-types right or wrong in the first place. According to even more modest social transmission views, emotions play a causal role in explaining why certain patterns of moral judgment prevail and get transmitted (Kauppinen Reference Kauppinen and Maibom2014; Nichols Reference Nichols2004). Assuming that people pick up their moralizing tendencies from others, this view entails that emotions ultimately (but indirectly) explain even the judgments of those who never respond emotionally.
The evidence May adduces in chapter 2 against exaggerated claims about the causal influence of emotion on moral judgment is compatible with a view like Smith's being true. And of course it does not bear on other varieties of sentimentalism, which make no causal claims in the first place. The best kind of evidence against a Smithian sort of explanatory sentimentalism would show that there are individuals who lack the postulated kind of sentiments altogether, but nevertheless make genuine moral judgments. The closest results in this respect come from studies on psychopaths – but alas, it is far more ambiguous, because psychopaths do have emotions (even if abnormal), and there is active debate on whether their moral judgments are genuine (see, e.g., Smith Reference Smith1994). And the social transmission view is of course not committed to the claim that emotions directly explain the judgments of particular individuals, so it is not necessarily threatened even if psychopaths know perfectly well what is right or wrong.
Why hold on to even modest explanatory sentimentalism, however, if the observed effects of emotional manipulation are as weak as May argues? Perhaps the most convincing argument is based on the close parallel between independently evolved emotional tendencies and widely accepted moral principles. There is an extremely plausible adaptive rationale for the tendency of social animals like us to have negative emotional responses to actions like cheating, failing to reciprocate, insulting, and grabbing a share of resources that is disproportionate to one's contribution (e.g., Sober and Wilson Reference Sober and Wilson1998). Other primates have analogous responses, which lends additional credence to the claim that they are independent of moral judgment. Yet there is a striking parallel – even if not an exact correspondence – between these adaptive emotional tendencies and widespread patterns of moral judgment (e.g., Boehm Reference Boehm2012). Some use such facts as a premise in a debunking argument of moral beliefs (Street Reference Street2006), but that is not the sentimentalist claim. The explanatory sentimentalist contention is that the parallel is best explained by the fact that moral judgment is deep down driven by emotion, though competing accounts differ on the details of just how this happens. David Hume (Reference Hume and Sayre-McCord2006, p. 260), for example, emphasizes the need to correct for bias in our untutored responses for morality to perform its social function. (This would explain why there is only a parallel, not an exact correspondence.) For the rationalist, in contrast, the parallel between adaptive emotion and moral judgment is a coincidence: Reason just happens to tell us to disapprove of the very things we in any case tend to feel negatively about, at least when we are ourselves at the receiving end. This comparison does not flatter the rationalist.
So far, I have focused on what sentimentalism is and what it is not. Let us now turn to rationalism, as May understands it. His claim is that “moral judgment is fundamentally an inferential enterprise that is not ultimately dependent on non-rational emotions, sentiments, or passions” (p. 7). May relies here on an extremely broad conception of inference, which includes “unconscious, unreflective, or implicit processes that nonetheless amount to reasoning” (p. 55). But he acknowledges that not every transition among beliefs (or other contentful states) amounts to reasoning (p. 9). Otherwise rationalism would be devoid of distinctive content.
What is reasoning, then? Here we must bear in mind that bad reasoning, too, is a kind of reasoning, so we cannot appeal to what are in fact genuine requirements of rationality (Broome Reference Broome2013). It is common to hold that at least the following elements are necessary: doxastic states whose contents serve as premises, doxastic or conative states whose contents express the conclusion, and some form of endorsement of the move from the premises to the conclusion, such as tacit acceptance of a pertinent rule of inference or taking the conclusion to follow from the premises (Boghossian Reference Boghossian2014).
Although this minimalist account is compatible with non-conscious reasoning, many of the computational mental processes that May argue play a role in moral judgment do not qualify as inference by its lights, because any kind of inference requires both premise-beliefs and somehow basing the conclusion on their content. For example, May holds that categorization of an ordinary object as a piece of furniture involves inference from a belief or belief-like state like “This objects resembles sofas, chairs, and tables” (p. 70). However, this is a non-starter as an account of categorization, as the very same (non-inferential) recognitional capacity that allows us to categorize something as furniture is required to make the judgment that it resembles items in the furniture category. If we can perform the latter without inference (and surely there are some such judgments on anyone's view), there is no reason to think unconscious inference must be involved in the former. Similar considerations hold for high-level perception (e.g., Audi Reference Audi2013), like the perception that someone is on drugs – we can be sensitive to complex information without any kind of inference from premises to a conclusion.
The same goes for moral categorization: there is no evidence for a necessary inferential step. Curiously, May half-acknowledges that the evidence fits the alternative hypothesis that our principles like the Doctrine of Doing and Allowing merely describe the pattern of our moral judgments rather than guide them (p. 70). What makes this only a half-acknowledgment is that he describes this in terms of “reasoning in accordance with” the principle. But that our judgments accord with a principle is no evidence at all that they result from reasoning – indeed, if it is acknowledged that the principle does not guide our reasoning, it would be a miracle of sorts if reasoning guided by some other rule yielded the same output in every case.
Second, even if we were to accept May's broad notion, the evidence he cites only shows that inferences about non-normative facts, such as the extent to which the agent was involved in bringing about the outcome, play a role in moral judging. This is something that sentimentalists accept. Already Hume emphasized that although sentiment renders the final verdict, “in order to pave the way for such a sentiment, and give a proper discernment of its object, it is often necessary, we find, that much reasoning should precede, that nice distinctions be made, just conclusions drawn” (Hume Reference Hume and Sayre-McCord2006, p. 189). So even on the arch-sentimentalist Hume's view, it is not only true that moral sentiments are “sensitive to information” (p. 74), but also that they sometimes require conscious reasoning about non-normative facts. What he and other sentimentalists deny is simply that this suffices to explain or justify moral judgment, because there is a gap between non-normative and normative conclusions. On their view, emotions do not just “facilitate” inference by directing attention, but either fundamentally explain or justify crossing the gap. Unless May shows that the process that takes us from non-moral premises regarding, say, intentions and consequences, to moral verdicts is distinctively rational, his view is importantly incomplete.
Finally, and related to the previous point, any process of inference must begin from premises, which on pain of regress cannot always be justified by further inference. Take the following simple piece of (good) reasoning:
1. Clinton lied.
2. Lying is wrong.
3. So, Clinton did something wrong.
No one denies that it is possible to reason from premises 1 and 2 to conclusion 3, and thereby gain justification to believe 3, if one is justified in believing the premises. But what justifies belief in premise 2? (Let us assume for simplicity that it is true.) On pain of a different regress, the answer cannot be “testimony.” So traditional intuitionists say, roughly, that it is self-evident: Anyone who understands the content thereby has justification to believe in it (Audi Reference Audi2013). Many epistemic sentimentalists say, roughly, that it is a legitimate inductive generalization from the contents of emotional responses, such as resentment, that present particular acts of lying as wrong (e.g., Tappolet Reference Tappolet2016). These are both the right kind of answers in that they do not appeal to further premises. May does not argue against such views. But more importantly, while he discusses evidence that we engage in reasoning from moral principles, I was unable to find any discussion of how we reason to moral principles, although he acknowledges the need in passing (p. 79).
To sum up, May tends to construe sentimentalism extremely thinly, as a claim that moral judgments are explained by or consist in purely non-cognitive feelings, and rationalism extremely broadly, as something like the claim that moral judgments are sensitive to information about their targets. On such construals, it is easy to declare rationalism as the better theory. But as I have tried to sketch here, at least when it comes to philosophy, both of these characterizations are ill-fitting. More work is needed to refute the arguments that sentimentalists actually make, and to develop a credible rationalist alternative.
It is thus fortunate that most philosophical sentimentalists from Hume and Smith onward are no less optimistic than May. They hold that as long as there is “some particle of the dove, kneaded into our frame, along with the elements of the wolf and serpent” (Hume Reference Hume and Sayre-McCord2006, p.259), we will approve of just and benevolent actions, constrain our egoism in virtue of internalizing the reactive attitudes of actual or imagined others, and make moral progress by reasoning about non-moral facts before rendering our judgment and by extending our natural empathy beyond our immediate circle. Doesn't it warm your heart just to think about it?
In the first part of Regard for Reason in the Moral Mind, Joshua May (Reference May2018) mounts a bold defense of a form of moral rationalism against sentimentalism. But what exactly is his target, and does he offer a credible alternative?
As I have observed previously (Kauppinen Reference Kauppinen and Zalta2013b), sentimentalism comes in many logically independent forms, in which emotions or more broadly pro- and con-attitudes play different roles. Explanatory sentimentalists hold that sentimental reactions fundamentally explain our moral verdicts; judgment sentimentalists hold that moral judgments consist in sentiments or otherwise make essential reference to sentiment; metaphysical sentimentalists hold that moral properties are grounded in actual or possible sentimental responses; and epistemic sentimentalists hold that we come to know moral truths ultimately by way of sentimental responses. Sentimentalists offer different sorts of a priori arguments for these claims, appealing, for example, to the apparent importance of attitudes that have a world-to-mind fit in explaining the action-guiding character of moral thought. Recently, some sentimentalists, most notably Jesse Prinz (Reference Prinz2007) and Shaun Nichols (Reference Nichols2004), have also offered a posteriori arguments for these views, drawing on scientific findings.
It is the a posteriori arguments that are May's main target, although he merely points to arguments of others when it comes to a priori sentimentalism. This is worth emphasizing for two reasons. First, though May mounts a very promising case against the a posteriori arguments, we may nevertheless have sufficient reason to subscribe to a sentimentalist view on a priori grounds. Second, I think it is fair to say that what defines the various sentimentalist views are the conclusions of the a priori arguments. Only explanatory sentimentalists, for example, are committed to causal claims about the role of emotion in generating moral judgments, and these claims are sometimes significantly weaker than May's targets. Adam Smith, for example, holds that “the greater part of our moral judgments […] is regulated by maxims and ideas derived from an induction of reason,” while arguing that it is “absurd and unintelligible to suppose that the first perceptions of right and wrong can be derived from reason” (Smith Reference Smith and Haakonssen2002, p. 377). On this kind of view, emotions do not play a causal role in every moral judgment, but rather explain why we find certain act-types right or wrong in the first place. According to even more modest social transmission views, emotions play a causal role in explaining why certain patterns of moral judgment prevail and get transmitted (Kauppinen Reference Kauppinen and Maibom2014; Nichols Reference Nichols2004). Assuming that people pick up their moralizing tendencies from others, this view entails that emotions ultimately (but indirectly) explain even the judgments of those who never respond emotionally.
The evidence May adduces in chapter 2 against exaggerated claims about the causal influence of emotion on moral judgment is compatible with a view like Smith's being true. And of course it does not bear on other varieties of sentimentalism, which make no causal claims in the first place. The best kind of evidence against a Smithian sort of explanatory sentimentalism would show that there are individuals who lack the postulated kind of sentiments altogether, but nevertheless make genuine moral judgments. The closest results in this respect come from studies on psychopaths – but alas, it is far more ambiguous, because psychopaths do have emotions (even if abnormal), and there is active debate on whether their moral judgments are genuine (see, e.g., Smith Reference Smith1994). And the social transmission view is of course not committed to the claim that emotions directly explain the judgments of particular individuals, so it is not necessarily threatened even if psychopaths know perfectly well what is right or wrong.
Why hold on to even modest explanatory sentimentalism, however, if the observed effects of emotional manipulation are as weak as May argues? Perhaps the most convincing argument is based on the close parallel between independently evolved emotional tendencies and widely accepted moral principles. There is an extremely plausible adaptive rationale for the tendency of social animals like us to have negative emotional responses to actions like cheating, failing to reciprocate, insulting, and grabbing a share of resources that is disproportionate to one's contribution (e.g., Sober and Wilson Reference Sober and Wilson1998). Other primates have analogous responses, which lends additional credence to the claim that they are independent of moral judgment. Yet there is a striking parallel – even if not an exact correspondence – between these adaptive emotional tendencies and widespread patterns of moral judgment (e.g., Boehm Reference Boehm2012). Some use such facts as a premise in a debunking argument of moral beliefs (Street Reference Street2006), but that is not the sentimentalist claim. The explanatory sentimentalist contention is that the parallel is best explained by the fact that moral judgment is deep down driven by emotion, though competing accounts differ on the details of just how this happens. David Hume (Reference Hume and Sayre-McCord2006, p. 260), for example, emphasizes the need to correct for bias in our untutored responses for morality to perform its social function. (This would explain why there is only a parallel, not an exact correspondence.) For the rationalist, in contrast, the parallel between adaptive emotion and moral judgment is a coincidence: Reason just happens to tell us to disapprove of the very things we in any case tend to feel negatively about, at least when we are ourselves at the receiving end. This comparison does not flatter the rationalist.
So far, I have focused on what sentimentalism is and what it is not. Let us now turn to rationalism, as May understands it. His claim is that “moral judgment is fundamentally an inferential enterprise that is not ultimately dependent on non-rational emotions, sentiments, or passions” (p. 7). May relies here on an extremely broad conception of inference, which includes “unconscious, unreflective, or implicit processes that nonetheless amount to reasoning” (p. 55). But he acknowledges that not every transition among beliefs (or other contentful states) amounts to reasoning (p. 9). Otherwise rationalism would be devoid of distinctive content.
What is reasoning, then? Here we must bear in mind that bad reasoning, too, is a kind of reasoning, so we cannot appeal to what are in fact genuine requirements of rationality (Broome Reference Broome2013). It is common to hold that at least the following elements are necessary: doxastic states whose contents serve as premises, doxastic or conative states whose contents express the conclusion, and some form of endorsement of the move from the premises to the conclusion, such as tacit acceptance of a pertinent rule of inference or taking the conclusion to follow from the premises (Boghossian Reference Boghossian2014).
Although this minimalist account is compatible with non-conscious reasoning, many of the computational mental processes that May argue play a role in moral judgment do not qualify as inference by its lights, because any kind of inference requires both premise-beliefs and somehow basing the conclusion on their content. For example, May holds that categorization of an ordinary object as a piece of furniture involves inference from a belief or belief-like state like “This objects resembles sofas, chairs, and tables” (p. 70). However, this is a non-starter as an account of categorization, as the very same (non-inferential) recognitional capacity that allows us to categorize something as furniture is required to make the judgment that it resembles items in the furniture category. If we can perform the latter without inference (and surely there are some such judgments on anyone's view), there is no reason to think unconscious inference must be involved in the former. Similar considerations hold for high-level perception (e.g., Audi Reference Audi2013), like the perception that someone is on drugs – we can be sensitive to complex information without any kind of inference from premises to a conclusion.
The same goes for moral categorization: there is no evidence for a necessary inferential step. Curiously, May half-acknowledges that the evidence fits the alternative hypothesis that our principles like the Doctrine of Doing and Allowing merely describe the pattern of our moral judgments rather than guide them (p. 70). What makes this only a half-acknowledgment is that he describes this in terms of “reasoning in accordance with” the principle. But that our judgments accord with a principle is no evidence at all that they result from reasoning – indeed, if it is acknowledged that the principle does not guide our reasoning, it would be a miracle of sorts if reasoning guided by some other rule yielded the same output in every case.
Second, even if we were to accept May's broad notion, the evidence he cites only shows that inferences about non-normative facts, such as the extent to which the agent was involved in bringing about the outcome, play a role in moral judging. This is something that sentimentalists accept. Already Hume emphasized that although sentiment renders the final verdict, “in order to pave the way for such a sentiment, and give a proper discernment of its object, it is often necessary, we find, that much reasoning should precede, that nice distinctions be made, just conclusions drawn” (Hume Reference Hume and Sayre-McCord2006, p. 189). So even on the arch-sentimentalist Hume's view, it is not only true that moral sentiments are “sensitive to information” (p. 74), but also that they sometimes require conscious reasoning about non-normative facts. What he and other sentimentalists deny is simply that this suffices to explain or justify moral judgment, because there is a gap between non-normative and normative conclusions. On their view, emotions do not just “facilitate” inference by directing attention, but either fundamentally explain or justify crossing the gap. Unless May shows that the process that takes us from non-moral premises regarding, say, intentions and consequences, to moral verdicts is distinctively rational, his view is importantly incomplete.
Finally, and related to the previous point, any process of inference must begin from premises, which on pain of regress cannot always be justified by further inference. Take the following simple piece of (good) reasoning:
1. Clinton lied.
2. Lying is wrong.
3. So, Clinton did something wrong.
No one denies that it is possible to reason from premises 1 and 2 to conclusion 3, and thereby gain justification to believe 3, if one is justified in believing the premises. But what justifies belief in premise 2? (Let us assume for simplicity that it is true.) On pain of a different regress, the answer cannot be “testimony.” So traditional intuitionists say, roughly, that it is self-evident: Anyone who understands the content thereby has justification to believe in it (Audi Reference Audi2013). Many epistemic sentimentalists say, roughly, that it is a legitimate inductive generalization from the contents of emotional responses, such as resentment, that present particular acts of lying as wrong (e.g., Tappolet Reference Tappolet2016). These are both the right kind of answers in that they do not appeal to further premises. May does not argue against such views. But more importantly, while he discusses evidence that we engage in reasoning from moral principles, I was unable to find any discussion of how we reason to moral principles, although he acknowledges the need in passing (p. 79).
To sum up, May tends to construe sentimentalism extremely thinly, as a claim that moral judgments are explained by or consist in purely non-cognitive feelings, and rationalism extremely broadly, as something like the claim that moral judgments are sensitive to information about their targets. On such construals, it is easy to declare rationalism as the better theory. But as I have tried to sketch here, at least when it comes to philosophy, both of these characterizations are ill-fitting. More work is needed to refute the arguments that sentimentalists actually make, and to develop a credible rationalist alternative.
It is thus fortunate that most philosophical sentimentalists from Hume and Smith onward are no less optimistic than May. They hold that as long as there is “some particle of the dove, kneaded into our frame, along with the elements of the wolf and serpent” (Hume Reference Hume and Sayre-McCord2006, p.259), we will approve of just and benevolent actions, constrain our egoism in virtue of internalizing the reactive attitudes of actual or imagined others, and make moral progress by reasoning about non-moral facts before rendering our judgment and by extending our natural empathy beyond our immediate circle. Doesn't it warm your heart just to think about it?