You are on page 1of 22

November

2018, Draft 1.0


Sentimentalism and the Moral Self
Jesse Prinz


1. Introduction

Sentimentalism is an approach to value theory that attempts to ground evaluative
judgments in emotional responses or other affective states. In the moral domain,
sentimentalism has been associated with theories such as moral sense theory,
emotivism, sensibility theory, and norm expressivism. These theories differ subtly
in their semantic and metaethical commitments. Debates between them rage on.
These theories also share certain assumptions. For example, they tend to
presuppose similar approaches to moral motivation. They are also vulnerable to
some of the same objections. Less obviously, there is a shared tendency among
defenders of these theories to neglect a topic that has recently made its way into
empirically oriented moral psychology: personal identity. New work suggests that
moral values are integral to how we construct ourselves. To date, little has been
said about how this new work bears on sentimentalist theories of moral judgment.
Here I want to bridge that gap by suggesting that the concept of moral identity can
help adjudicate debates between competing versions of sentimentalism, and
perhaps even offer new strategies for deflecting more general objections that have
been levied against the sentimentalist tradition. An identity-theoretic approach to
sentimentalism has implications for moral psychology and moral ontology.
Below, I will begin with a brief review of sentimentalist theories, including
their differences, the challenges the face, and the state of empirical evidence for
them. I will then turn to work on moral identity. After reviewing recent findings, I
will suggest ways in which this work might be used to make progress of debates
between competing sentimentalist theories, and debates between sentimentalists
and theory opponents. Along the way, I will suggest that the two major metaphors
associated with the sentimentalist tradition—projection and perception—be
replaced by a third—performance. Correspondingly, I will suggest that morality has
a performative ontology.

2. Sentimentalism

2.1 The Flight from Reason

The sentimentalist tradition emerged in British moral philosophy in the first half of
the 18th century at a time when the moral rationalism of the Cambridge Platonists
had been dominating discussion. The sentimentalists rejected this tradition, and
advanced the view that moral judgments are based on sentiments or passions.
These terms were used somewhat interchangeably; the term “emotion” was
reserved for strong feelings at that time. Anthony Ashley-Cooper, the 3rd Earl of
Shaftesbury, was among the first to advance a sentimentalist theory, and he was
followed by Francis Hutcheson, David Hume, and Adam Smith in Scotland.
These sentimentalists do not eschew reason entirely. They recognize, for
example, that reason could play a role in correcting sentimental biases, such as out
tendency to care more for the near and dear. But they reject the rationalist
conjecture that we arrived at moral judgments by reason alone. They also believe
that sentiments are essential to moral psychology at least two ways. Sentiments are
partially constitutive of moral judgments, and sentiments are required for moral
motivation. The judgment that some action is right or wrong is constituted by a
sentiment towards that action (or a meta-sentiment, as we will see momentarily),
and that sentiment compels us to act in accordance with the judgment.
The sentiments in question are not always named, though we find reference
to approval, approbation, liking, and their opposites. Hume also refers to the
“peculiar” sentiments, implying that there might be a class of sui generis moral
emotions. The British sentimentalist also make frequent reference to other morally
relevant feelings, such as gratitude, pity, and sympathy. On some versions, these
play a role in the generating moral judgments, though they are not constitutive of
moral judgments. Thus, for Shaftesbury, moral judgments comprise second-order
sentiments toward conduct that is guided by first order sentiments: we like it when
someone acts kindly to another out of pity, and this liking constitutes our approval
of that action. Moral judgments are constituted by the liking of benevolent
sentiments and the actions they promote. For Hume, we sympathetically experience
the psychological effects of actions on others—such as the pleasure induced by acts
of charity—and that sympathetic response prompts our feelings of approbation. On
both these views, sentiments arise in two places: as precursors to moral judgments
and as components of moral judgments. In principle, though, it is the latter that
qualifies the views as sentimentalist.
The British sentimentalists tend to give three kinds of argument for their
position. These are especially vivid in Hutcheson and Hume. The first is an appeal
to psychological observations. They are attempting to describe what it is like to
make a moral judgment, and each finds that emotions are present. It is not always
clear exactly how they arrive at these conclusions. For example, they do not tend to
speak in the first-person, which suggests that they are not trying to develop
explicitly phenomenological evidence for their theories. But, one can assume that
they are drawing on both personal experience and observation of others. Hume, for
example, makes detailed observations about the way in which sympathy can be
affected by such factors as the intentions of a person under consideration, their guilt
of innocence, their social distance, and the conditions of judgment (***).
Rationalists, such as Clarke and Cudworth, tend to develop to develop their theories
with little attention to human behavior. Sentimentalists are coming out of the
empiricist tradition, which assigns a central role to psychology in philosophical
theorizing, and aims at a descriptively adequate, observationally grounded, account
of human nature. Not all empiricists were sentimentalists, of course. Hobbes, for
example, was an egoist, but Shaftesbury and his Scottish followers present
psychological arguments against egoism. They take sentimentalism to accord with
the most careful psychological observations.
The second argument that can be found in these authors emphasizes the role
of sentiments in motivation. Shaftesbury, Hutcheson, and Hume all argue that
reason alone cannot motivate us to act. They subscribe to a faculty psychology
according to which reason is motivationally inert. Only sentiments can motivate.
Hume develops the argument in most detail, though there are clear statements of it
in both Shaftesbury (***) and Hutcheson (***). Hume argues that the will—which is
the faculty that causes action—is influenced by direct passions, which include both
primary emotions, such as fear, and instinctive drives, such as hunger. These
involve pleasure and pain, as causes and effects respectively, and those pleasures
and pains prompt the will to act. Hume also discusses “indirect passions” such as
pride and humility, which principally involve ideas of the self. Curiously Hume
believes that these do not directly act on the will, but they do so indirectly by
intensifying direct passions. In any case, the basic idea is that reason alone cannot
motivate; it is, in this sense, a slave of the passions.
To serve as an argument for sentimentalism, the claim that motivation
requires passions must be supplemented by a second premise: moral judgments
motivate. This premise, sometimes called motivational internalism, is controversial.
In contemporary ethical theory, there are some authors who deny that moral
judgments are inherently motivating. They hypothesize that there could be
“amoralists” who recognize the difference between right and wrong, but feel no
inclination to act accordingly. The argument from motivation implicitly denies that
this is possible. Hume discusses moral apathy (his sensible knave, ***), but he
assumes that those who are indifferent to morality are also oblivious to it. In any
case it is noteworthy that such characters are generally presented as fictional. It is
not clear whether the all British moralists intend their theories to be modally
robust; that is, they may not care whether it is literally impossible to think about
moral rules without motivating feelings. They are most concerned to describe how
moral psychology ordinarily operates. Here, there seems to be firm evidence in
their favor: moral judgments tend to be occasioned by strong feelings, and these, in
turn, influence various forms of behavior: we avoid temptations, treat each other
kindly, punish transgressors, and so on.
The final argument associated with the British sentimentalist tradition
emphasizes the inadequacy of reason in deriving moral norms. Hutcheson and
Hume both develop versions of the argument. Sometimes the point is advanced in
combination with the motivation argument: reason does not motivate. But
Hutcheson and Hume both argue that reason is simply incapable of arriving at moral
judgments. Hume is especially insistent on this point. He argues that we cannot find
anything inherent in bad behavior that might be said to constitute its badness:

Take any action allow’d to be vicious: Wilful murder, for instance. Examine it
in all lights, and see if you can find that matter of fact, or real existence, which
you call vice… The vice entirely escapes you, as long as you consider the
object. You never can find it, till you turn your reflexion into your own breast,
and find a sentiment of disapprobation. (***)

Hume argues that reason can be used only to tell us about the relation between
ideas or to posit causal relations (***). It tells us what follows from what. But
nothing moral follows from non-moral premises (Hume’s law), so reason cannot be
a source of moral truth.
Proponents of reason have rejected Hume’s argument (Joseph Butler, Mary
Wollstonecraft, Immanuel Kant, among others), but none have identified an
uncontroversial method for deriving moral truths from reason alone. Kant’s effort is
the most ambitious and the most influential, but its key premise is manifestly
implausible and far too powerful. Kant says it is irrational to do anything that would
couldn’t will as a universal law—that it we should avoid any action that would be
self undermine were everyone do it. Consider the choice to become a philosopher.
If everyone did that, then no one would be around to cultivate food or other basic
necessities, so philosophizing would be unsustainable. So Kant unwittingly applies
that his own life’s work is irrational and unethical. Kantians would undoubtedly
balk at this offhand rejection. I make the point only to emphasize that Kant has
failed to convince many readers and, for that reason, the sentimentalist challenge
continues to be taken seriously. When combined with the arguments from
psychology and motivation, it also has some force as an argument to the best
explanation: even if reason could be used to derive moral norms, psychological
evidence and moral motivation seem to indicate that ordinarily moralizers rely on
sentiment rather than reason. As a theory of how people actual moralize, even Kant
would concur that the sentimentalists are onto something. In what follows, I will
pursue this line of thought, treating sentimentalism as a descriptive theory, and
setting aside the possibility that an alternative moral psychology might be possible
or preferable.

2.2 Sentimentalisms

Each of the British moralists developed somewhat different versions of
sentimentalism and many others have developed since. Here I want to give some
indication of that variation and also to introduce some of the enduring objections
that these theories face. I will divide them into three classes: moral sense theories,
expressvisms, and sensibility theories.
Shaftesbury passingly uses the phrase “moral sense,” and Hutcheson adopted
this label and developed it more explicitly. I will use the term here to cover the
conjunction of two principles, pertaining to moral ontology and epistemology
respectively. The ontological principle is moral realism, which I will define as the
view that there are moral truths and these truths obtain independently of our
capacity to recognize them. There is some scholarly controversy about whether
Shaftesbury and Hutcheson were moral realists, but I think the preponderance of
evidence favors this interpretation (Rivers, 200; Rauscher, 2003; Irwin, 2015).
Shaftesbury develops a positive view of objective moral facts—to a first
approximation, the good is that which contributes to the well-being of the species.
Hutcheson is more reticent, but he claims that God is good, and that God had
furnished us with a capacity to recognize the good. He also says that God has
arranged things so that goodness also serves self-interest. Both authors then
advance an affect-based epistemology form morality. We already caught a glimpse
of Shaftesbury’s view. Our capacity to recognize the good consists in in the fact that
we respond positively when we see people acting out of benevolent motives. He
argues that non-human animals can behave morally, but we alone have a moral
sense: a capacity to reflect on such actions, and a natural tendency to like them.
Hutcheson says that God has furnished us with a natural tendency to register good
behavior with approval.
The term “moral sense,” which Hutcheson regards as an innate faculty, draws
an explicit analogy between moral judgment and perception. This analogy can be
developed in different ways, but it invites us to think of moral judgment as an
automatic response—a passive and immediate registration of moral facts. For
Hutcheson, the analogy also implies that our moral faculty, like vision or touch, is
innate and not reducible to any other psychological capacities.
Hume rejects the moral sense theory. He denies that there is an innate moral
faculty, and he rejects moral realism. Hume agrees that we are innately prosocial, at
least to some degree. We are, by nature benevolent, but, in the first instance, that
benevolence is restricted to friends and family. We must condition ourselves to
extend these feelings outward. Morality is, to this degree, “artificial” (***). When we
extend kindness to distant others, we are not discovering mind-independent moral
truths. This is a matter of human invention. That does not mean we can develop
morality in any direction we like. Hume is no relativist (1739, 3.II.viii, note 2; 1751:
appendix 4). He thinks that human nature, combined with the shared challenge of
developing rules to peacefully co-exist with others, will converge on specific rules,
including respect for property (***) and female chastity (***). These rules are
universal, but not innate. We condition ourselves to disapprove of violations.
In rejecting the moral sense theory, Hume does not abandon the analogy to
perception entirely; he simply abandons the realist assumption that moral facts are
out there to be perceived. Instead,

[V]ice and virtue… may be compar’d to sounds, colours, heat and cold, which,
according to modern philosophy, are not qualities in objects, but perceptions
in the mind. (***)

Here, Hume is drawing an explicit analogy between morals and what Locke calls
secondary qualities. For Locke, the external world is colorless, but it has the power
to cause color sensations in us. This does not entail that we should be skeptics or
anti-realists about color. On the standard reading of Locke, colors are real, but
mind-dependent. Likewise, Hume says the wrongness of murder, “is a matter of fact
but ’tis the object of feeling” (ibid.). On this interpretation, Hume treats moral
judgments as truth-apt. They are true when and only when they correspond to our
moral sentiments—or, one might say, moral sentiments under the right conditions.
This qualification is meant to capture the possibility that Hume has a nascent ideal
observer theory. This is most explicit in his sentimentalist aesthetics where he
develops a theory of ideal critics (***). In his ethics, we get a hint of it when Hume
talks about adopting “the general point of view” to overcome biases of sympathy
(***).
In recent moral philosophy, the analogy between morals and secondary
qualities is sometimes called the “sensibility theory” (cf. Wiggins, ***; McDowell, ***;
McNaughton, ****; Prinz, 2007). It is a form of subjectivism, since it grants to
existence of moral facts but insists that these facts are dependent on our response.
Some heirs to Hume have rejected this form of subjectivism. Blackburn (1994), for
example, takes aim at the analogy with secondary qualities. Colors, he says, may
depend on us, but they also supervene on physical properties in the external world,
and there is a special faculty for detecting those physical properties. One might
address this concern by downplaying the color analogy and emphasizing, instead, an
analogy between qualities of taste, such delicious or revolting. But this shift in
emphasis would raise a more general concern that, if moral judgments were mere
matters of taste, then moral disagreements would be manifestly spurious. Color,
one might say, is too objective, and taste is too subjective.
Blackburn tries to bypass the secondary quality analogy by arguing that
moral judgments are not in the business of asserting facts at all. They are, instead,
expressions of our attitudes. Blackburn encourages us to move away from the
perception analogy, deployed by both realist sentimentalists and sensibility
theories. He prefers instead the metaphor of projection, which picks up on Hume’s
observation that, “the mind has a great propensity to spread itself on external
objects” (***). The idea here is that moral facts are not real but only “quasi-real.”
We talk as if there were moral facts but really we are projecting feelings onto the
word. To say that stealing is wrong is like saying “Disco sucks.” It looks
superficially like a statement of fact—a statement about something in the world—
when it is actually a manner of expressing a negative attitude.
Blackburn’s account closely follows a tradition that began earlier in the 20th
century, with authors such as Stevenson (1937) and Ayer (1946) who developed
“emotivist” theories of moral judgment. Stevenson doesn’t discuss the secondary
quality theory explicitly but he does argue that standard subjectivist theories face
difficulties when it comes to disagreement. His objection is akin the worry about
treating moral judgments as expressions of taste. He proposes instead that moral
judgments have “emotive meaning”: they express affective responses. The concept
“morally good” expresses approval. “Morally bad” expresses disapproval (or
perhaps indignation or shock; p. 26). “Express” here doesn’t simply mean that such
judgments convey the speakers feelings; they also have what calls a “quasi-
imperative force” (p. 19). Moral judgments encourage listeners to adopt a similar
attitude. They function as recommendations. Stevenson proposes that moral
disagreements are not disagreements over facts, but rather over interests (p. 27)
Saying “X-ing is good” is akin to saying “Let’s do X” and those who do not think X-ing
is good will resist this suggestion.
Expressivist theories are not restricted to emotivism. A more recent version,
called norm expressivism, has been developed by Gibbard (1990). Whereas
Blackburn, Stevenson, and Ayer would say that moral judgments directly express
feelings (and thereby serve as suggestions), Gibbard contends that they express the
acceptance of feeling norms. Saying “X is bad” expresses that one accepts a norm
according to which is would be appropriate to experience guilt about X-ing or anger
toward those who do X. This makes his theory more meta-cognitive than its
expressivist predecessors. Rather than assuming that speakers are having
emotional responses when they make moral judgments, Gibbard can allow that they
are conveying endorsement of norms about emotions.
Some sensibility theorists have also formulated their accounts
metacognitively. For example, McDowell (***) suggests that morally bad behavior is
that which merits disapproval. This metacognitive move has been called “neo-
sentimentalism” in the literature, to contrast it with more traditional first-order
theories (D’Arms and Jacobson, ***). The difference between McDowell and
Gibbard is subtle but important. For McDowell, moral judgments can be true “X-ing
is wrong” says that X-ing merits disapproval. For Gibbard, “X-ing is wrong”
expresses the speakers endorsement of a norm to feel disapproval; so it is not
saying something true or false, but rather conveying the acceptance of that norm
(for Gibbard, acceptance is a primitive propositional attitude). In more recent work,
Gibbard has shifted to a formulation in terms of plans or decisions (Gibbard, 2003),
but he remains committed to the view that moral judgments are not, strictly
speaking, true or false.

2.3 Empirical Adjudication?

Philosophical debates about these various forms of sentimentalism (and between
sentimentalists and their opponents) have not come to any resolution, and some
authors have hoped to escape the impasse by appealing to empirical evidence. This
is in keeping with a broader empirical turn in philosophy. Advocates of empirical
methods claim that philosophy is good for developing coherent theories, but not
always ideal for confirming which theories are right—especially when those
theories carry commitments about psychological states and processes.
Over the last 30 years, considerable evidence has amassed linking moral
judgments to emotions. As a matter of empirical fact, it seems to be the case that
people are ordinarily in emotional states when they make moral judgments. Every
neuroimaging study of moral judgments to date has been consistent with this
assessment, even in the rare cases where their authors have claimed otherwise
(Prinz, ***). There is also overwhelming evidence that experimentally manipulating
people’s emotions has an influence on their moral judgments (Haidt & Wheatley,
***; Schnall et al., ***; Seidel & Prinz, 2013a, 2013b). Those with emotional
impairments show corresponding abnormalities in moral judgment tasks (Blair,
***). Such findings support sentimentalism, at least as a descriptive theory about
how people ordinarily make moral judgments.
Empirical work has also filled in some details about which emotions
contribute to morality. As noted earlier, the British moralists tended to speak
vaguely of (dis)approbation or (dis)approval, or sometimes of “peculiar”
sentiments. Psychological studies have identified more specific emotional
correlates of moral judgments, along with job descriptions for each. When crimes
are committed against persons (e.g., property violations, injustice, and physical
harm), we experience anger, and when actions are deemed to violate natural laws
(e.g., incest, cannibalism, mutilation, and even flagrant hypocrisy), we respond with
disgust (cf. Rozin et al., 1999; Seidel and Prinz, 2013b). Guilt and shame are evoked
when people consider their own transgressions in these two categories. When we
reflect on good behavior, we experience emotions such as elevation (Haidt, 2000),
gratitude (Desteno et al., 2010), and admiration (Algoe & Haidt, 2009). Individuals
deficient in specific emotions often show islands or moral insensitivity, as indicated
by work on deficits in disgust (Schmidt & Bonelli, 2008; Igoumenou et al., 2017) and
guilt (Pletti et al., 2017).
Such finding both support and enrich the case for sentimentalism. But which
form of sentimentalism is most empirically plausible? This question has received
less attention than one might hope, but there is some relevant evidence in the
literature. Realist theories are difficult to refute outright, but they have long been
subjected to various kinds of debunking arguments. One classic example is
Harman’s (***) argument that we can fully account for moral beliefs without
positing objective moral truths. Stated as an empirical claim, the argument begins
with the premise that when we try to account for why people have the moral values
that they do, facts about evolution and cultural inculcation suffice. With many other
kinds of beliefs—beliefs about the weather or current events—the best explanation
implicates mind-independent facts. There is also empirical evidence that many
people do not harbor objectivist theories of morality, and that objectivism is
correlated with religiosity (Goodwin & Darley, ***). This contrasts with other
domains, where presumptions of realism are more deeply entrenched and not
wedded to specific religious doctrines. Moreover, no one has identified what
empirical evidence for moral reality would look like. Some moral realists would say
that moral facts are just those that correspond to the preponderance of moral
beliefs (***). But this too is empirically dubious. There is abundant evidence moral
beliefs are interpersonally and interculturally variable, so there is little reason to
think that moral beliefs converge on any set of objective facts, since there is little
moral convergence.
Expressivist theories are most frequently criticized for their failure to explain
why moral judgments usually come in the form of indicative sentences rather than,
say, expletives or emperative. Both Blackburn and Gibbard have proposes semantic
theories to explain this superficial appearance. To my mind, such technical
acrobatics underestimate the force of the worry. The hard question isn’t whether
we can devise an expressive semantics consistent with the use of indicate sentences;
rather, the question is why we use such sentences at all if moral discourse is
expressive. Stated as an empirical argument, one can simply point out that, across
the world’s languages, there are moral predicates that can be used in ordinary
indicative sentences. Indeed, I know of no language that includes an explicitly moral
imperative form. We can construct imperatives in English like “Though Shall Not
Kill” or “Don’t cheat!” or “Be nice!” – but notice no moral terms are used here. If
moral concepts were imperatives, there should be forms such as “Wrong killing!”
Empirical objections have also been made against metacognitive versions of
sentimentalism—the so-called neo-sentimentalist theories of Gibbard and
McDowell. Nichols (***) argues that they are too cognitively demanding since they
falsely imply that individuals without developed metacognitive abilities can make
moral judgments. The empirical motivation for such theories can also be called into
doubt. Metacognitive theories emphasize the possibility of making moral judgments
without actually having emotions, but empirical evidence provides no indication
that this happens. When people make moral judgments, they do, in fact, have
emotional responses. It is an interesting empirical question whether they also
ordinarily also form metacognive attitudes. I would conjecture that this is not
always the case. We sometime say things like “You should be ashamed of yourself,”
but it seems plausible that one can morally condemn an action without this added
thought.
I left the first-order sensibility theory to the end, since it is the position that I
have defended in the past, and I think it is more empirically plausible than its near
neighbors (Prinz, 2007; Prinz, ****). I continue to think that something along these
lines is right. Here, however, I want to suggest that the view faces serious various
challenges. One of these was passed over without comment earlier when I
mentioned the need for a criterion of truth. Expressivist theories forego true, so
they do not need to specify conditions under which an emotion-grounded moral
judgment qualifies as correct. Sensibility theorists do. Hume leaves this task
undone, though he hints at an ideal observer theory. If we take his aesthetics to be a
guide, then Hume might have recommend that we identify the traits that make ideal
moral critics: disinterest, full information, and so on. This approach can only get us
so far, however, since there is little reason to believe that individuals meeting such
conditions would converge. Empirical work on moral variation across groups
suggests that people who are equally reasonable and informed can have different
moral values. Even moral philosophers, who can be regarded as disinterested
experts, do not agree about basic normative question. This is shown both by the
history of philosophy and by recent experimental philosophy surveys (e.g.,
Chalmers, ****). Any sensibility theory owes us a standard of truth. Call this the
Semantic Challenge.
Other challenges to first-order sensibility theories can be adduced by
reflecting further on disanalogies with secondary qualities. I mentioned Blackburn’s
objections earlier and suggested that moral predicates function more like taste
predicates than color predicates. Calling something good, I suggested, is like calling
it delicious. This preserves a perceptual model of how moral judgment works. With
food, we have a bite, and then react. Some cases of moral judgment are like this;
Harman (***) gives the example of how we might react if we saw someone lighting a
cat on fire. But there are many cases where the perception analogy would be
misleading. As is often pointed out moral judgments sometimes arise after much
reflection. Call this the Deliberation Challenge.
There is also a less widely recognized problem with the perceptual analogy.
There is something very individualist about the idea that morality is like perception.
Each of us uses perception to make our way about the world. We can do so in
situations that are not social, and we rarely have a need to broadcast what we
experience. Even judgments of taste are like this. A person in isolation would form
preferences, and use these as a basis for decision. Morality, in contrast, is more
social: moral judgments arise primarily in social contexts and they are calibrated to
align with others. Empirically, there is much evidence that we learn moral through
socialization and cultural transmission (Newcomb, 1948; Edelstein, 1962; Cavalli-
Sforza & Feldman 1981; Henrich et al., 2005; Graham, et al., 2016). In addition,
there is evidence that people adjust their moral judgments to promote positive
assessments form others (Rom & Conway, 2018; Lee, 2018). These ongoing efforts
at social calibration are easy to neglect if we focus too heavily on the analogy
between morality and secondary qualities, including taste. Call this the Sociality
Challenge.
Let me mention, too, a more general worry that threatens sentimentalist
accounts. Recall that a key argument for sentimentalism appeals to the link
between morality and motivation. I think there are solid empirical reasons to
believe that morality is motivating, but the relationship is complicated. When we
are personally involved in a situation, moral judgments can certainly impact
behavior. For example, while playing economic games, we will punish non-
cooperative players at considerable personal cost (***). But motivation tends to
trail off with distance. We are less reliable at making third-party interventions, and
we rarely go out of our way to help others (***). For example, few people would
cross the street to give some change to a homeless person. Most of us also fail to
engage in much political activism, and, when we do, it tends to be for local issues
that personal affect us, rather than issues effecting distant others (***; there are, of
course, exceptions to this, as I will mention below). Call this the Motivation
Challenge.
Sentimentalists might explain disproportionate motivation on analogy with
Hume’s account of sympathy bias: emotions lose intensity with greater distance.
But this explanation also exposes a weakness of first-order sentimentalist theories:
if moral judgments are constituted by emotions, then one might expect the
judgments to become more emotionally intense when they pertain to matters that
are more morally serious. But, most people are more outraged by minor slights
from co-workers than massive injustice around the globe. We judge that these
slights are less serious, but we experience them more intensely. This is confirmed
by everyday observation as well as empirical research (***). So clearly the degree of
wrongness is not well calibrated to emotional intensity. This is another disanalogy
with perception: felt intensity is usually a reliable guide to stimulus strength. Call
this the Intensity Challenge.
Neo-sentimentalists, who adopt a metacognitive approach, might appear to
have an easy solution to the Intensity Challenge. They would say that minor slights
merit less anger than global injustice. This is a tempting move, but it’s not clear that
our metacognive norms would come out the right way: after all, what’s the use of
getting worked up over remote situations over which one has no control?
Practically speaking, a personal slight merits a stronger reaction that global
injustice. A Humean might intervene at this point and say that we cannot determine
the appropriate emotional intensity until we divorce ourselves from our actual
positions in the world and adopt a general point of view, imagining how we would
respond were we in the position of other affected parties. This sounds promising,
but it may be unworkable in some cases. Suppose we are reflecting on a society in
which women are treated as male property and denied basic liberties. Does the
general point of view require us to adopt the perspective of each woman there to
determine the appropriate intensity of our judgment? And what if women there
largely endorse the value system that places them in an unequal position? Putting
aside the normative question about how to think about such cases, there is are
psychological problems with Hume’s suggestion: we cannot actually adopt all the
required viewpoints, and we often condemn things that would not be condemned by
individuals directly involved.
I want to suggest that these challenges require would benefit from a fresh
look at sentimentalism. We might be well advised to move beyond the dominant
alternatives. I will ultimately propose a variant that abandons the perception
analogy. To get there we need to make a detour through another topic that has
recently drawn attention in the empirical literature.

3. Towards a Selfish Sentimentalism

3.1 Morality and Personal Identity

Sentimentalism got a significant boost with the advent of empirical moral
psychology. Since the 1990s numerous studies have come out investing the
relationship between emotions and moral judgment. There is now a sizable
literature suggesting that emotions are operative when people engage in moral
decision-making and that manipulating emotions can impact the moral judgments
that people draw. This provides support for sentimentalism, but also leaves
questions unanswered, including questions about which version of sentimentalism
is most plausible. Over the last few years, there has been a move in a new direction.
Until recently, most research in moral psychology focused on moral judgments. The
broader psychological role of these judgments had gone larger unexplored. As
research advanced, it became increasingly clear that moral judgments issue from
moral value systems, and moral values systems vary across groups. From here, it
was a small step to realize that morality may play a role in social identity: moral
judgments can reflect membership in a culture or subculture including nationality,
region, religion, and political party affiliation. This insight led, in turn, to the
conjecture that moral values may also play a role in personal identity.
Social identity and personal identity are often studies independently. For
example, sociologists and social psychologists who study group behavior often care
a great deal about social identity but make no assumptions about how group
members construct their identities as individuals. Personal identity has been most
actively studies by philosophers, not social scientists, and the most influential
theories make no reference to group membership. According to one theory
personal identity is a matter of psychological connectedness or continuity, and, in
that tradition, emphasis has been placed on memory links (Locke, ***; Grice, ***;
Parfitt, ***). A related tradition proposes that identity emerges through the
construction of personal narratives (MacIntyre, ***; Ricouer, ***; Schechtman, ****).
Memories and narratives make reference to other people, but they are generally
presumed to be highly idiosyncratic, chronicling each person’s unique path through
the world. Other authors have emphasized agency, suggesting that we are defined
by the choices that we make—thing we do rather than things done to us (Korsgaard,
***). Agency is closely linked to the notion of autonomy, and, thus, this theoretical
orientation tends to think about personal identity as contrasting with aspects of
psychology that strongly reflect social influence.
Against these prevailing theoretical assumptions, there has been a new
interest in the relationship between moral group membership—an aspect of social –
and personal identity. A hypothesis has emerged according to which moral values
an integral to how we construct ourselves as individuals. I say “construct ourselves”
since few in this literature think there are deep metaphysical facts about what
constitutes personal identity. There are many different traits that can be used to
settle what makes me me. Memory, narratives, and agency presumably matter for
questions of identity, at least in some contexts, as might other traits such as
personality and long-term projects. The hypothesis that morality is also part of
identity is intended to supplement these other proposals, but, some of its defenders
go a step further and claim that moral values are even more important for identity
than some of these other constructs.
This has been borne out by a number of studies in experimental philosophy.
The first of these, I carried out with Shaun Nichols (Prinz, & Nichols, ***). We
decided to focus what philosophers call diachronic moral identity: what makes
someone count as the same person as they undergo transformations over time. We
devised vignettes in which a person sustains a head injury that impairs one capacity
but leaves others in tact. In one scenario, the person loses his moral values, in
another he loses his memory, in a third he loses his capacity to construct narratives.
In each case we asked, “is he the same person after the change?” We found that
moral changes had a dramatic impact on identity, and the other changes has only a
minor impact. This was true in cases where a person’s morals change for the better
as well as when the change for the worse. We also included cases where morals
were changed by choice rather than head injury. Here we reasoned that an agency
theory of personal identity would predict a big difference between these cases: if I
choose new values, then that is an expression of agency, and, if agency is crucial for
identity, these changes should be regarded as mine. Against this prediction, we
found that moral changes have an equally strong impact on identity, whether the
changes were chosen or not.
Subsequent work by Nina Strohminger and Nichols (2015) replicated and
extended this work. I have also conducted numerous replications and extensions
with other collaborators, especially Javier Gomez-Lavin and Joerg Fingerhut. The
results are incredibly consistent. People regard continuity of moral values to be
crucial for personal identity. They think a moral changes is a change in self, literally
and not just figuratively. The result holds up if we ask about morality in the
abstract, or if we specify specific values systems such as political party affiliation
and religion. We have replicated these findings in Western Europe and in Taiwan.
We have also presented participants with cases involving moral changes that can
occur in real life, including prison reform, war trauma, and immigration. No matter
how we test it, moral changes are said to dramatically impact identity, and we have
not obtained comparable effects for any other trait.
To date, there has been no systematic attempt to relate work on
sentimentalism to this newer work on the moral self. In the next section I will bring
these two research traditions together.

3.2 From Moral Judgment to the Moral Self

Sentimentalism is, first and foremost, a theory of judgment. As such, there is an
immediate difficulty relating the research on morality and personal identity.
Identity is an enduring trait and judgments are ephemeral. Two steps are needed to
bring these two into contact. We need an account of how judgments issue from
standing dispositions of the mind, and how those dispositions get bound up with
one’s sense of self.
As to the first issue, it is helpful to draw a distinction between moral
judgments and moral values. Moral judgments are brief psychological episodes.
These episodes do not arise ex nihilo. Suppose I you see someone wearing a hat that
says Make America Great again. You realize this is a Donald Trump supporter and
you form a moral judgment about the person. If you are a Trump enthusiast, you
might experience feelings of moral approval, such as admiration, gratitude, or
elevation. This is someone you feel a sense of solidarity with and you appreciate
their decision to advertise the President you admire so deeply. Alternatively, if you
have a negative opinion of Donald Trump, you might experience outrage or even
disgust: how can this person wear a hat that is an emblem of hate? There is nothing
in the stimulus (a person wearing a hat) that determines the judgment. Your
reaction depends on attitudes that are in place before the encounter. It depends on
moral values.
Moral values can be defined as standing dispositions to form moral
judgments. Values might be directed towards kinds of behavior or kinds of people.
They determine whether we experience negative, positive, or neutral emotions.
They are also context-sensitive in various ways. Values pertaining to actions are
rarely simple generics, such as killing is bad. To kick in, we often need to know who
did what to whom, how they did it, and why? Killing in self-defense, for example, or
in a just war, may differ from joy killing, and that may differ from instrumental
killing or accidental killing. We weigh in facts about the both the killer and the
victim, and the circumstances and methods (was it premeditated, violent, and so
on). So values are quite complex mental rules. They produce emotional responses
in a way that reflects the complexity. For example, a political assassination may
excite outrage and a serial killer may instill disgust. Those who have taken the life
of another (e.g., drivers who commit unintentional manslaughter) may experience
intense guilt, or even shame, if the event leads one feel like a bad person. Elsewhere
I use the term “sentiment” to refer to such context-sensitive emotion dispositions
(Prinz, 2007).
So described, values, unlike judgments, satisfy one condition on contributing
to identity. They are long-standing traits. It is tempting, therefore, to simply define
one’s moral identity as the set of values that one has internalized. There is,
however, one complication. Most values get acquired through inculcation, and
among these there may be some that we would rather not have. Many people
harbor values that they would disavow. Consider implicit bias. Traits such as
agism, racism, sexism, homophobia, transphobia, and ablism are deeply entrenched,
and function as enduring dispositions. These usually don’t always qualify as moral
values, but they can inform and infect our moral values. One might be more tolerant
than one would like of sexual harassment, one might have different reactions to
public displays of affection by straight and gay couples, one might be inclined to
blame people for failing to perform in expected ways even if one knows that mental
illness was factor. These are moral reactions, and the issue from long-standing
dispositions. But some people would rather not have these reactions. They are
disavowed.
In a study with Javier Gomez-Lavin, we found that disavowal weakens the
relationship between morality and identity (***). We gave people scenarios about
people who were raised in homophobic societies, and end up with negative feelings
towards homosexuality as a result. One person embraces those attitudes and
another disavows them. In both cases, the values then change, and we ask, “Is this
the same person?” The impact on identity is significantly lower for the person who
disavowed the homophobia.
Such considerations suggest that there is a difference between the moral
values that we associate with personal identity and those that we do not. Merely
having a value does not necessarily make that value qualify as part of the self. So, in
addition to a theory of valuing, we need an account of which values are part of
identity. This is a topic that still needs empirical and philosophical attention, but I
will offer some speculations here.
The concept of disavowal is a helpful starting place. Values that we disavow
are, I propose, not regarded as aspects of identity. This, however, does not settle the
matter. For it leaves us with two other cases: values we avow and those about
which we have neither avowed nor disavowed. Is identity restricted to the former?
And what, in any case, is avowal?
It is safe to assume that moral values are construed as part of identity when
they are explicitly avowed. Avowal, as I a using the term, is matter of endorsing
one’s values. Endorsement, in turn, can be understood on the model of what
Frankfurt (1971) calls second-order desires. In develop a compatibilist account of
free will, Frankfurt distinguishes the desires we desire to have from those we desire
not to have, and those about which we have no second-order desires (what he calls
“wanton” desires). I am not a compatibilist, and won’t weigh in on the free will
debate here, but I think Frankfurt points us towards a useful theory of avowal.
There are sentiments we want to have and other sentiments we don’t want to have.
There are different reasons for wanting certain sentiments, but, in the present
context, what seems to matter most are cases where we morally approve and
disapprove of our sentiments. For this reason, rather than using the term second-
order desires, I prefer to talk of meta-sentiments. We have sentiments towards
which we feel approbation and those towards which we feel disapprobation.
Someone might take moral pride in condemning racism and feel shame about
residual homophobia. The former is treated as part of identity; the latter not.
What about wanton sentiments—those towards which we have no
metasentiments? Here things get a bit tricky. There are several kinds of cases to
consider. First, there are cases that one would endorse were one to think about it.
These can safely be included in our constructions of identity. Then, among those for
which there are no fixed dispositions, I think it may help to distinguish two kinds of
cases. Some wanton values may be peripheral in that they do not interact with
many other aspects of the mind. Certain pet peeves are like this, and so are some
judgmental tendencies. Consider fattism—a negative bias against those deemed
“overweight.” Many people have an implicit tendency to blame those whose bodies
do not conform to prevailing norms. For some people, such tendencies are
recognized as a bias, for others they are endorsed, and for still others, there is no
fact of the matter based on current psychological dispositions whether their fattism
would survive critical scrutiny: there are no meta-sentiments, even dispositonally.
Imagine that such a person somehow gets over this bias. Are they the same person?
My guess (an empirical prediction) is that they are. Fattism may simply be too
disconnected from other aspects of the mind, including other values, to matter very
much for identity.
Now consider a wanton value that is more intimately connected to other
things. One possible example is a moralized form of classicism: a tendency to blame
the poor, and to morally excuse affluent people for living lifestyles that depend on
exploitation. When confronted with this bias, some people might find it difficult to
settle firmly settle on whether to endorse it or reject it. For an affluent person,
exorcizing classicism might have considerable person cost, so there is a strong
motivation to preserve classist values, but these may be accompanied by a
recognition that class divisions are often unjust. Should we say that classicism is
part of the identity of the wanton wealthy who blithely resist metasentiments?
Here, my prediction in that the value really is part of identity. It is part of identity
precisely because it is so deeply connected to other aspects of how such a person
conducts their life. A change of values here would be quite significant.
Such cases point to a limitation in the Frankfurt approach to endorsement.
The classist may not explicit avow her classicism, but she can be said to implicitly
endorse it in the way she conducts her life. We have seen three kinds of cases:
explicit endorsement, dispositional endorsement, and implicit endorsement. I think
all three would be regarded as aspect of identity. Peripheral wanton values and
disavowed values would be excluded. With these distinctions in hand, we can define
the moral self as the set of values that a person endorses, explicitly, implicitly, or
dispositionally.

3.3 Beyond Perception and Projection: Morality as Performance

I have just described how we can bridge research on the emotional basis of moral
judgment with research on moral identity. I know want to consider how this
marriage of ideas may inform a sentimentalist ethical theory. There is one crucial
ingredient missing. So far I have been describing moral values as if we had them
and endorsed them in isolation from others. But, as noted earlier, we normally
acquire values from others, and endorsement also has important social dimensions.
Moral values grant us membership into like-minded groups, and the approval of
those groups plays a role in determining our approval of our own values. There are
occasional moral renegades, who don’t care about social approval, but most of us do,
whether we admit it or not, and there is strong pressure to keep our values in
alignment with the communities to which be long.
There is much research showing that people give preferential treatment to
those who are morally like-minded, and members of moral out-groups are
derogated. Consider, for example, work on polarization in the United States.
Republicans and Democrats have deep animosity towards each other (Pew
Research, ***), and give unfair advantages to members of the own group (***).
Those whose values differ can be socially ostracized. Such partisanship is not
simply a function of having first-order values that happen to align with a party
platform. People identity with parties, and that identification plays a regulatory role
in driving partisan behavior, political participation, and value preference (***).
Such group-driven endorsement is not restricted to political values. It can also be
promoted by religious communities (***), nationalities (***), heritage (***), and
subcultures (***). By joining such groups we reap rewards and avoid sanctions.
Membership usually comes with a requirement that we morally conform.
This social aspect of moral endorsement has implications for a theory of
moral judgment. According to sentimentalist theories, moral judgments are
grounded in emotions. Emotions, in turn, are usually thought to be inner feelings.
When I say “X-ing is wrong,” I am conveying how I personally feel about X-ing. If
such feelings have social origins, however, and a regulated by ongoing social
pressures, it might be more accurate to render what is conveyed in the first-person
plural: this is how we feel about X-ing. Or, even more accurately, when a I say “X-ing
is wrong,” I am conveying my credentials as a member of a moral group. I am
reassuring group members that my values align with theirs, and informing out-
group members that I am a moral opponent.
This communicative aspect of moral judgment is easy to neglect when we
deploy metaphors of perception and projection. Those concepts invite us to imagine
a solitary observer spreading herself onto the world or being impinged upon by
moral stimuli. I think it is more illumining to think of moral judgment as a kind of
performance. Here I am inspired by Judith Butler’s (1990) work on gender, as well
as a long tradition within sociology that likens social behavior to acting (Goffman,
1956). There is also a large literature within emotion theory that characterizes
emotional responses as performative (Hochschild, 1979; Averill, 1981; Fehr &
Russell, 1984; Shaver et al., 1987; Eickers, 2018). Morality is something we
perform. We publically display our moral preferences so that people can see where
we belong in social space.
This performative aspect can be seen in many ways. Consider social media,
which have been increasingly described as echo chambers where people preach
their values to an approving choir. There has also been much recent discussion of
virtue signaling. Both online and offline we work hard to make sure that others
know where we stand on moral issues. Often this is not done to confront opponents,
but to secure respect and fellow-feeling from members of out moral groups.
We also singal moral group membership in our preferences for art, music,
fashion, sports, and pastimes. In recent studies, I have shown that such matters of
taste are highly predictive of moral values (***). These recreational interests may
be, in part, moral performances. Be expressing enthusiasm for a kind of music, for
example, we indicate something about our moral identities. Choice of clothing,
transportation, and diet can all play a role in such signaling.
I don’t want to suggest that projection metaphors (such as “spreading”) and
perception metaphors (“moral seeing”) are uninformative. Projection captures the
idea that morality comes from us, and perception captures the idea that moral
judgments can be palpably felt and automatic. But the metaphor of performance has
advantages. Performances also come from us, and, when highly rehearsed, they are
automatic. Crucially, too, they are social.
Performance also suggests a somewhat different ontology that we are used to
from the sentimentalist tradition. For expressivists, moral facts don’t exist. For
sentimentalist moral realists, facts are out there in the world. For sensibility
theorists, moral facts are generated by or “responses.” On the performative
approach, it would be more accurate to say that moral facts are “enacted.” They
come into being through things we do: our communicated performances. This
analysis also accommodates moral reformers who bring new systems of value into
being through their rhetoric and theories.
Once a system of values has been constructed, one might say that it is there
to be perceived, so the perception metaphor can take hold again. This way of
describing things is misleading, however. Admittedly, we sometimes go looking for
what to believe, and the passivity of perception might apply to that exercise, but
such cases are unusual. More often, moral judgments are expressions of standing
moral dispositions—what I called values. When we give voice to them, we are not
simply picking up on social facts; we are playing in active role in sustaining those
facts.
One might worry that the performative approach does not apply in cases
where we make moral judgments privately, as when we read a newspaper at home
alone. This worry is misplaced, however, since it falsely presupposes a sharp
distinction between public and private. Private reactions are public in that they
both reflect social influences, and dispose us to share our attitudes with others.
They may also function as a kind or moral rehearsal, and they help confirm who we
are to ourselves.
A harder case in the aforementioned moral renegade, whose values depart
from local moral communities. But here, too, judgments can be seen as
performative—perhaps all the more so. Those of us who confirm to local moralities
convey that we are good group members when we voice our values. That is to be
expected; we are saying what others expect us to say. Those who rebel against the
prevailing moral order are saying something surprise, and that captures our
attention. They are not picking up on moral facts; they are trying to entrain our
sentiments, and bring a new form or moral conscious into being. Passive perceptual
methaphors are inapplicable here, but performance is not.
I don’t mean to suggest that we take ourselves to be pretending when we
make moral judgments. Defenders of performative theories of emotion and gender
do not think of performance as a pretense or deception. It is as if we are method
acting, and don’t even realize it (Hochschild, 1979). The crucial thing is that moral
performances are active doings, not passive seeings, and they have communicate
functions. Moral judgments are social actions, and moral facts are collectively made
and perpetually remade, not detected.
I have moved here from moral identity to performativity. These are not
separate theories but part of a whole. Once we see the link between morality and
identity, and the social nature of both, we can see that morality requires social
performances. I now want to consider how this framework may help to update
sentimentalism and address outstanding objections.

3.4 Putting Selfish Sentimentalism to Work

I have been proposing an approach to sentimentalism that can be characterized as
self-involving or “selfish.” It begins with the standard approach: moral judgments
have an emotional basis. It then steps back and grounds these emotions in values.
Among those values are some we endorse. The endorsed values are regarded as a
core aspect of personal identity. This aspect of personal identity is also part of our
social identity. Consequently, when we make moral judgments, we are affirming
that identity to others; moral judgments are performed.
This embellishment of sentimentalism may have some advantages when it
comes to addressing the challenges that I laid out for the sensibility theory in
section 2.3. I will consider each of these challenges in turn.
Let’s begin with the Semantic Challenge. Sentimentalist subjectivists need to
provide a theory of what makes moral judgment true. Ideal observer theories,
developed on the model of Hume’s theory of ideal critics in aesthetics, are unlikely
to succeed. Disinterested, fully informed critics disagree. In earlier work, I defend
the view that each of us is individually the own arbiter of moral truth (Prinz, 2007:
chap. *). I still think that is correct. Pace Hume, there is unlikely to be a single set of
moral truths onto which human beings converge. Individual relativism—the view
that locates truth in each judge—in the most stable relativist position. But in earlier
work I didn’t fully appreciate two things about this position. First, it implicitly
depends on something like the moral self. If moral truth depends on my values, then
we need a theory of what makes values mine. I may occasionally make moral
judgments that are erroneous by my own standards. We all make judgments that
are out of character from time to time. So, a properly develop individual relativism
would include a theory of endorsement, and that is tantamount to an invocation of
the moral self. Second, because our personal identities are social, our personal
values are often deferential: we try to calibrate with others. So individual relativism
generally collapses into a kind of group relativism. I choose what groups to defer to,
but those groups then plan an active role in guiding my values. We listen to group
members and remain receptive to moral correction. Though details need to be
worked out, the basic proposal that emerges can be stated as follows: “X-ing is
wrong/right” when judged by O just in case O endorses a value according to which
X-ing is wrong/right” or O defers to a group whose members endorse such a value.
Next consider the Deliberation Challenge. The perception analogy
encourages one to think that moral judgments always arise passively and
automatically, but this is not the case; we often deliberate. Selfish sentimentalism
makes no such prediction. Moral values can issue automatic moral judgments, but,
as just noted, the arbiters of moral truth are endorsed values and, if we defer, moral
communities. Neither endorsement not community assent are easily to discover by
immediate reflection. In the case of implicit endorsements, for example, discovering
whether a value is mine is a complex and holistic process. As noted, too, moral
judgments issue from values and values are complex mental structures with many
contextually determined parameters. Figuring our whether a value applies may
take cognitive work (we must also often analyze the cases under consideration to
figure out whether they are instances of values we endorse). The performative
theory of judgment adds another level of complexity as well. In addition to
determining who has done what to whom and how, we must think about our
audience. In making moral judgments, we aim to confirm our moral credentials to
those who might be listening, and this can require added reflection.
Performativity also helps with the Sociality Challenge. Selfish sentimentalism
is not individualistic, despite the central role in assigns to personal identity.
Personal identity, we have seen, is social. We aim to calibrate our values of others,
and we make judgments with the aim of explicitly upholding views that others
share.
It is less obvious how the selfish sentimentalist should handle the Motivation
Challenge, but I think there are resources here as well. Once again performativity
plays a role. On perceptual views, moral judgment is regarded as fundamentally
epistemological: judgments tell us what is right and what is wrong. Such judgments
are then thought to guide action: pursue the good and avoid the bad. The
performative story places emphasis elsewhere. Moral judgments are primarily
communicative and constructive. They convey our values to others and thereby
contribute to sustaining the norms we have socially constructed. From this
perspective, it is unsurprising that a moral judgments are often oriented toward
virtue signaling rather than action. They aim to secure group ties. Of course people
who act in ways that grossly contradict their values incur penalties from the moral
groups to which they belong. Compliance with values is important. But often
compliance is more a matter of saying that doing. Announcing contempt for racism
can achieve its perfomative ends even if it isn’t backed up by active efforts to end
structural inequality. The emotions that arise when we make such judgments have
some motivational force, and they might lead us to react behaviorally if we were
directly confronted with a norm violation, but emotions do not always issue in overt
behavior, especially when there are other costs to consider. When we are not
directly effected, emotions may motivate verbal behavior (impassioned
denouncements, for example) much more than other forms of action. The
performative theory helps shed light on this phenomenon.
These last observations may seem cynical, since they imply that a lot of moral
life is just a verbal performance. We are often content to say the right things
without doing much. But selfish sentimentalism also provides some reason for
hope. First of all, moral behavior is a good way to win points with our moral groups.
Being outspoken may be sufficient for group membership, but moral activism can
earn high esteem. So the performative aspect of morality draws attention to a
source of moral motivation that has been woefully neglected in analytic moral
psychology. In addition, the selfish sentimentalist introduces another source of
motivation. Consider someone who profoundly fails to live in accordance with their
values. This is not just unpleasant (evoking guilt and shame); it is also an existential
threat. Those who profoundly violate their own values expose themselves to the
possibility that the values are not really their. Such individuals are not true to
themselves, and the breach between self-conception and behavior can trigger an
identity crisis: Who am I really? Do I really hold these values if I don’t live by them?
Research in military psychiatry has explored the phenomenon of “moral injury”—
trauma brought on by involvement in conduct that one deems to be immoral in
civilian life. This can be extremely disorienting, and an identity-theoretic approach
to morality can help explain why.
Let’s turn finally to the Intensity Challenge. This is the hardest to account for.
Standard first-order sentimentalist theories have an especially had time with it,
since the intensity of first-order emotional responses does not reliably track the
intensity of moral wrongness. Neo-sentimentalism and Hume’s ideal observer
theory looks more promising, but I raised some prima facie worries for these
approaches above. I think selfish sentimentalism has some tools it can bring to bear
on this challenge. First of all, the locus on moral truth is value endorsement. If I
endorse the value that people should treat each other with respect, then, by my own
lights, this values is violated when there is a minor slight against me and when there
are cases of structural injustice affecting millions of people. In the latter case, the
value I endorse has been violated to a far greater degree than the former, so, by my
lights, it is a far greater wrong. Here, intensity is irrelevant . The fact that I react
more strongly to the personal slight does not bear on the content or truth-
conditions of my judgments. It is true that, when introspecting, I may sometimes
use intensity to gauge how I feel about something, but this is not a failsafe
procedure. Intensity is just one indicator of what values I endorse, and it is an
imperfect one, since it intensity can also scale up with proximity and other factors.
In addition, the selfish sentimentalist can appeal to performative norms. In
practice we sometimes get more worked up about small slights that global
problems, but we can be called out for doing so. When publically conveying our
values, it would be uncouth to imply that we are more bothered by minor slights
than major injustice. So the norms governing the public presentation of moral
judgments serves as a kind of corrective for moral intensity. We can bring intensity
into better alignment with wrongness when we bear in mind how we are perceived
by others. Neo-sentimentalists emphasize norms of merit; whereas the
performative approach emphasizes the aspiration to look good in the eyes of one’s
moral groups. If group members care about global injustice more than minor slights
against you, then, when publically performing moral judgments, there will be
pressure to express greater outrage for global problems. This may seem to depend
on prior judgments of merit (we aim to show that we are most outraged at things
that merit greatest outrage), but it can be formulated with reference to merit: we
aim to show that we are most outraged at the things our moral groups encourage us
to be most outraged about.
This last observation highlights a subtle difference between selfish
sentimentalism and neo-sentimentalism, but it also draws attention to a similarity.
Both views move beyond first-order judgments and recognize that we also have
norms about norms. One might worry that this similarity exposes selfish
sentimentalism to an empirical objection that has been raised against neo-
sentimentalism. As we have seen, Nichols (***) accuses neo-sentimentalists of
making morality too cognitively demanding: individuals who have limited social
cognition abilities can nevertheless form moral judgments. Nichols sites young
children and people with autism in this context. Here I would say two things in
reply. First, selfish sentimentalism does not advance a metacognitive view of moral
judgments. Neo-sentimentalists says that moral judgments are meta-cognitive: they
express the conviction that moral emotions are warranted or merited. Selfish
sentimentalism has a first-order view of moral judgments; performative norms play
a regulatory role. Second, these performative norms need not involve explicit
deployment of mental concepts. There may be minimal forms of sensitivity to social
pressures such as the recognition that we will be socially sanctioned or rewarded
for making certain judgments. Individuals who lack mental-state concepts can show
this kind of sensitivity. Young children, for example, acquire their norms through
social pressures, and they are keenly aware of those pressures. High functioning
individuals with autism are highly motivated to align their behavior with others,
even if they don’t spend as much time as neurotypical individuals reflecting on how
they are viewed. It may be that such individuals are also less concerned with moral
group membership. This difference does not imply a lack of moral understanding,
but it may indicate a difference in the degree of performativity. Individual variation
on this dimension does not threaten the thesis that, for the vast majority of mature
neurotypical moralizers, moral judgments are highly performative.
These replies to the various challenges facing sentimentalism are incomplete.
Each deserves much more discussion. My goal here is simply to show that selfish
sentimentalism introduces some resources that may be useful in developing more
adequate replies. Both the personal and the social character of moral judgments
have been curiously underappreciated within the sentimentalist tradition. Both
factors, which are intimately linked on my view, shed light on moral psychology in
ways that help arm the sentimentalist against enduring objections.

4. Conclusions

This discussion began with a review of arguments for sentimentalism and a menu of
sentimentalist theories. Each of these theories has been criticized, and
sentimentalist theories face important challenges. I introduced several of these. I
then developed a new perspective on sentimentalism, inspired by recent research
on the relationship between morality and identity. This new perspective, which I
have been calling selfish sentimentalism, highlights new strategies for meeting
challenges to other sentimentalist theories.
Selfish sentimentalism makes several claims that distinguish it from other
forms. It shares the view that moral judgments have an emotional basis, but it
grounds these in values, and then distinguishes the values we endorse from those
that we disavow. Endorsed values comprise the moral self: a dimension of personal
identity. The moral self is also important to social identity, since most people aim to
align their values with moral groups. All this has implications for how we should
think about moral judgments. Such judgments are not mere expressions of feeling;
they are also expressions of identity. They tell people who we are as persons. In the
same gesture, they also indicate the moral groups we which we want to be allied.
Therefore, they signal solidarity to group members. This communicate function
implies that moral judgments are not simply in the business of passively registering
moral violations, but also play a role in actively indicating lines of moral solidarity.
Moral judgments are performative. In performing morality through judgment, we
also play a role is sustaining the moral truths that have been constructed by
ourgroups. Moral judgments do no simply register moral reality; they enact moral
reality, by creating and sustaining systems of value. Selfish sentimentalism moves
us beyond the perception analogy that informs moral sense theory and the
sensibility theory, while also rejecting the expressivist view that morality is a mere
projection. Morality is something we bring into being through the impassioned
judgments that forge social ties and establish personal identities identities. It
makes just as much sense to say that the self is a projection of morality than to say
that morality is a projection of the self. Both are co-constituted. The emotional
construction of morality is a form of self-construction, and selves are both the
products of social groups and the sustaining elements of such groups. Morality, me,
and we are intimately and metaphysically linked. Sentimentalism must be updated
to reflect these interdependencies, and, the resulting theory may prove to be more
robust than its predecessors.

You might also like