In Philosophical Papers 29 (2000), 121-53
Nondescriptivist
Cognitivism:
Framework for a New
Metaethic
Terry Horgan and Mark
Timmons
We propose to break
some new ground in metaethics by sketching a view about moral judgments and statements
that departs from traditional ways of thinking about them. As the title
suggests, our view combines a nondescriptivist account of moral judgments and
statements—they are not in the business of describing moral facts—with the
cognitivist idea that moral judgments are genuine beliefs and moral statements
are genuine assertions. We claim that in addition to descriptive beliefs, there
are (moral) evaluative beliefs which are neither reducible to, nor a species
of, beliefs of the former type. We think that our kind of metaethical view has
obvious advantages over the standard menu of options (versions of realism,
rationalism, relativism, error theory, and forms of standard
nondescriptivism)—advantages that will become apparent as we proceed.
Our plan is to begin
(section I) by questioning a deeply
embedded assumption of traditional metaethical thinking which we think has
unfortunately and unnecessarily blocked from view the metaethical theory we
favor and which, when rejected, opens up some new metaethical territory worth
exploring. We then proceed in sections II—VI to outline our positive view by
developing a new framework for understanding belief and assertion within which
nondescriptivist cognitivism emerges as a consistent and plausible metaethical
contender. In section VII we consider various challenges to our brand of
cognitivism, explaining how our view can answer such challenges and also
indicating some of the main tasks that lie ahead for any attempt to develop the
view further.
Our central focus will
be on moral judgments, with much of what we say applying mutatis muntandis to
moral statements. Sometimes, but not always, we will explicitly extend points
made about judgments to the case of statements too.
I. The Semantic Assumption
In order to focus on the semantic assumption that we think ought to be
rejected, we distinguish three notions of semantic content.
First, let judgments whose overall content is expressible by
declarative sentences be called declarative
judgments, and let the overall content of such a judgment be called its declarative content. Declarative content, then, is possessed simply as a result of
grammatical form. Typical moral judgments are expressible by declarative
sentences (e.g., ‘Apartheid is wrong’; ‘Himmler was an evil man’), and so in
metaethics all competing views—descriptivist and nondescriptivist alike—must
grant that moral judgments have declarative content.
Even if all metaethical views recognize that moral judgments have
declarative content, they disagree over whether such judgments also have cognitive content. Cognitive content is
belief-eligible and assertible content, and so to say that a judgment has such
content is to say that the judgment is a genuine belief. Correspondingly, to
say that a statement has cognitive content is to say that it is a genuine
assertion. Although talk of cognitive content
might be a relatively recent bit of philosophical nomenclature, talk of
cognitive meaning has a history of
use in metaethics, though the two expressions may be used to signify the same
thing. Of course, there have been sharp divisions within metaethics over the
question of whether or not moral judgments and statements have cognitive
content and if so, whether such content is semantically primary. Cognitivists
in metaethics affirm that typical moral judgments have cognitive content, while
their noncognitivist opponents deny that the declarative content of a moral
judgment is cognitive (or primarily cognitive).[1]
But notice that what has been taken for granted in analytic philosophy
generally, and metaethics in particular, is the idea that for content to be
genuinely cognitive it must be in the business of purporting to represent how
the world is. And this brings us to a third notion of content—descriptive content. Descriptive content
is content that purports to represent the world as being a certain way, and is
characteristic of ordinary nonmoral beliefs about the world. The judgment that
Now, according to our view, moral judgments are genuine beliefs and
moral statements are genuine assertions. Consequently, moral judgments and
statements have declarative content that is genuinely cognitive—that is, they
have belief-eligible, assertible content. Cognitivism in ethics is the view
that moral judgments are genuinely cognitive in their content, and so we are
ethical cognitivists. Cognitive content has been assumed, by all parties in
these discussions, to be the same thing as descriptive content. Thus,
‘descriptivism’ and ‘cognitivism’ have been seen as alternative labels for the
same kind of metaethical position. In opposition to the tradition, we maintain
that the declarative content of moral beliefs and assertions is not a species
of, nor is it reducible to, descriptive content—content that represents the
world as being a certain way. We therefore reject metaethical descriptivism; on
our view, moral beliefs (and the sentences expressing them) are not descriptive.
This combination of cognitivism and nondescriptivism flies in the face
of a deeply embedded assumption that we call the semantic assumption:
SA All genuinely cognitive content is descriptive content—i.e.,
way-the-world-might-be content. Thus, mental states like beliefs and linguistic
items like sentences that have cognitive content are in the business of
representing some (putative) state of affairs or stating some (putative) fact.
This assumption, we claim, is a largely unquestioned dogma of both descriptivist
and nondescriptivist views in metaethics, and (we think) is the main culprit
that stands in the way of developing a fully adequate metaethical account of
moral thought and discourse.[3] Let us briefly review how
it figures in traditional metaethical thought.
Suppose one accepts
what we call the thesis of semantic unity:
SU Sentences with the grammatical and logical trappings of
assertion have genuine cognitive content. Similarly, judgments whose content is
expressible by such sentences have genuine cognitive content.
So now consider a typical (if somewhat simplified) line of thought
behind versions of metaethical descriptivism. The descriptivist begins with the
following observation about moral thought and discourse:
M Moral thought and discourse manifest the relevant
grammatical and logical features that are characteristic of genuine belief and
assertion.
Now this claim, together with the idea that judgments having such
grammatical and logical trappings really are genuinely cognitive (SU) and that all
genuinely cognitive content is descriptive (SA) entail the main descriptivist
claim:
D Moral thought and discourse have descriptive content, i.e.,
declarative moral content is descriptive.
By contrast, the
traditional nondescriptivist rejects metaethical descriptivism, recognizes that
moral discourse has all the grammatical and logical trappings of genuine
cognitive content, but then, given the semantic assumption, is forced to reject
the thesis of semantic unity. That is, the traditional nondescriptivist reasons
as follows: not-D; M; SA; therefore not-SU. And so the nondescriptivist,
rejecting the thesis of semantic unity, must distinguish, for moral discourse,
between surface features of moral thought and discourse and the supposedly deep
features that reveal its true semantical workings. Hence, the project of the
traditional nondescriptivist was to characterize the deep semantic workings of
moral thought and discourse—often through reductive meaning analyses that
essentially equated declarative moral content with some kind of non-cognitive
content expressible in non-declarative language. Eschewing descriptive
declarative content for moral thought and discourse, the traditionalist
embraced some form of noncognitivism (e.g., emotivism).[4]
Our proposal is to
break away from all this by rejecting the semantic assumption that weds genuine
cognitive content to descriptive content. The line of thought we employ, then,
could be expressed this way: we do recognize and take seriously the fact that
moral thought and discourse display the grammatical and logical trappings of
cognitive content, and along with the traditional descriptivists, we agree that
such trappings are indicative of genuine, deep, cognitive content for moral
thought and discourse; but since we reject descriptivism in ethics, we must
hold (and think there is good reason to hold) that some forms of genuinely
cognitive thought and discourse are not descriptive.[5] Our project involves
staking out a metaethical position according to which this claim, in connection
with moral thought and discourse, is both consistent and plausible.
Here, then, is an
initial statement of our nondescriptivist cognitivism (henceforth, NDC):
(1) Declarative judgments with moral content are genuine beliefs,
having genuinely belief-eligible, cognitive content. Thus, declarative
statements with moral content are genuine assertions—their declarative content
is cognitive.
(2) However, the cognitive content of such judgments and statements is
not descriptive (way-the-the-world-might-be) content.[6]
It will perhaps help
if we locate our metaethical position vis-à-vis standard views using a visual
aid:
METAETHICAL
VIEWS ABOUT
MORAL JUDGMENTS
DESCRIPTIVIST NONDESCRIPTIVIST NONCOGNITIVISM
COGNITIVISM COGNITIVISM
Descriptive Content Nondescriptive
Content
Cognitive Content Noncognitive Content
Declarative Content
Notice that on our
diagnosis of what is wrong with traditional metaethics, two levels of
content—descriptive/nondescriptive and cognitive/noncognitive—are simply
conflated owing to the semantic assumption. Rejecting the assumption and
distinguishing these types of content opens up fertile metaethical territory
that we plan to explore and cultivate.[7]
II. A Fresh Start
In developing a metaethical theory, one would like to accommodate what
seem to be deeply embedded features of moral thought and discourse as plausibly
and coherently as possible. One thing that seems clear is that moral judgments
and moral statements exhibit many of the characteristics distinctive of genuine
belief. First, we have already mentioned that moral judgments have the logico-grammatical trappings of genuine
beliefs: the content of a moral belief is declarative, and can embed as a constituent
of a judgment that has logically complex declarative content (e.g., the
judgment that either Jeeves has already mailed Uncle Willoughby’s parcel or
Bertie ought to mail it).[8]
As such, moral judgments can figure in logical inferences. They can combine
with other beliefs to yield new beliefs that are content-appropriate given
prior beliefs. Second, moral judgments also exhibit phenomenological features characteristic of beliefs. They are
experienced as psychologically involuntary, and as grounded in reasons: given
one’s evidence, one cannot help but make certain moral judgments. And because
of their reason-based involuntariness, moral judgments exert a felt rational
authority upon us.[9] The belief-like nature of
typical moral judgments is widely enough recognized and uncontroversial enough
that we need not digress here in order to elaborate the case for this claim.
Moral judgments also
seem to play a distinctive action-guiding role in a person’s overall
psychological economy that makes them in some ways unlike ordinary nonmoral
beliefs. Typically, anyway, moral judgments directly dispose us toward
appropriate action, independently of our pre-existing desires—whereas ordinary
nonmoral beliefs only become action-oriented in combination with such prior desires.
(Thus, the reason-based authority of moral beliefs typically gives them
motivational force, over and above the motivational force of our pre-existing
desires and often capable of “trumping” them.[10])
Associated with this action-guiding role are certain distinctive
phenomenological features too—notably, a felt demandingness, a phenomenological
“to-be-done-ness.” The action-oriented nature of typical moral judgments, with
its accompanying typical phenomenology, has led many moral philosophers to
embrace some form or other of ethical internalism. Despite difficulties in
formulating a plausible form of internalism, we think the insight behind such
philosophical views is correct—distinctive of moral judgments is their
action-guiding role.[11]
The problem is to plausibly
combine these two dominant features of moral judgments—their being a kind of
belief and yet mainly in the business of action-guidance—into a plausible
metaethical view. Many moral philosophers see a tension here, some opting for
nondescriptivist views that would deny that moral judgments have overall
cognitive content, others denying internalism. Of course, there are those who
attempt to defend cognitivism and internalism, but not too successfully we
think.
We are
nondescriptivists, and we aim to develop a strain of this general kind of view
that fairly accommodates both features just mentioned. Doing so requires that
we face three serious tasks:
First Task: Articulate a conception of belief that does not require the
overall declarative content of beliefs to be descriptive content.
Second Task: Make a case
for the independent plausibility of this conception of belief.
Third Task: Argue that nondescriptivist cognitivism, formulated in a way
that draws upon the proposed conception of belief, has significant comparative
advantages over descriptivist forms of cognitivism.
The first task is the
most basic, because it is not antecedently clear how the semantic assumption,
which effectively equates cognitive content with descriptive content, could
possibly be mistaken. This task is also the most important, because it is what
will open up the new metaethical territory we seek to occupy. We propose to
address this challenge by developing a generic framework for belief that does
not presuppose that all cognitive content is descriptive content, and therefore
is consistent with the claim that some beliefs have overall cognitive content
that is not descriptive. (The framework is also consistent with the denial of
this claim.) This is the business of section III.[12]
Of course it is not
enough just to propose a conception of belief that is consistent with
the claim that some beliefs have overall content that is not descriptive. For,
the proposal might complicate the notions of belief, assertion, and cognitive
content in ad hoc, implausible ways, and/or it might seem theoretically
unmotivated (and hence, question-begging) from the perspective of advocates of
the semantic assumption. The second task, then, is to show that the framework
is theoretically plausible independently of the fact that it is
consistent with the possibility that some beliefs have overall content that is
not descriptive. We take up this project in section IV, where we argue that the
proposed framework for belief is attractive even for those who accept the
semantic assumption, because it provides a way for descriptivist versions of
cognitivism to accommodate the internalistic, action-guiding, aspect of moral
judgments.
Insofar as the
framework turns out to be independently plausible, however, the third task then
arises: arguing that nondescriptivist cognitivism, as situated within the
framework, is more plausible than descriptivism—and in particular, is more
plausible than the kind of descriptivist cognitivism that is situatable within
the same framework (thereby successfully combining descriptivism with
internalism). Addressing this issue is the business of section V.
The discussion in
sections III-V thus will constitute an articulation of both the metaethical
position we advocate and the reasons for embracing it. In section VI we make
some observations about the philosophical methodology employed in the preceding
sections, in order to underscore how our approach departs from standard
metaethical debates not only in substance but also metaphilosophically.
III. A Framework for Belief and Assertion
We will describe a
generic approach to belief and assertion that provides the backbone of our
brand of nondescriptivist cognitivism. We begin with a characterization of the
base case for understanding beliefs and assertions—that is, beliefs and
assertions whose declarative content lacks truth-functional or quantificational
complexity, and also lacks any embedded deontic operators—and then turn to
cases that have that kind of logical complexity.
1. The base case
Speaking most generally, a base-case belief is a kind of psychological
commitment state, of which there are two main species: is-commitments and ought-commitments.
Beliefs of both sort have what we call core
descriptive content—a way-the-world-might-be content. So, for instance, the
belief that Bertie will mail the parcel, and the belief that Bertie ought to
mail the parcel, share the same core descriptive content, expressible by the
non-evaluative that-clause, that Bertie
mail the parcel. A parallel point applies to assertions, about which we say
more below.
An ordinary descriptive belief (purporting to represent how the world
is) is an is-commitment with respect to a core descriptive content, and so the
belief’s declarative content coincides with its core descriptive content. For
descriptive base-case beliefs and assertions, then, their overall declarative
content is descriptive.
By contrast, an
evaluative belief is an ought-commitment with respect to a core descriptive
content. Evaluative beliefs differ essentially from descriptive beliefs in the
following respect: the core descriptive content of an evaluative belief does
not coincide with its overall declarative content. For instance, the belief
that Bertie ought to mail the parcel is an ought-commitment with respect to the
core content, that Bertie mail the parcel;
however, its overall declarative content is that
it ought to be that Bertie mail the parcel, and so its overall declarative
content does not coincide with its core descriptive content. Thus, whereas
descriptive beliefs involve an is-commitment (a how-it is-with-the-world
commitment) with regard to a core descriptive content, moral beliefs involve a
different type of commitment: a how-it-ought-to-be-with-the-world commitment
with regard to a core descriptive content.[13]
Some observations are
in order. First, we previously distinguished three species of content:
declarative, cognitive, and descriptive, where we were focusing on an item’s overall content. Moral judgments
certainly have overall declarative content because their overall content is
expressible by declarative sentences. Furthermore, within the framework we are
proposing, their overall content is also cognitive
content since they count as genuine beliefs. The framework is officially
neutral, however, about whether or not their overall content is descriptive content. According to the
metaethical position we will be advocating, moral beliefs do not have overall
descriptive content, but the framework could be adopted by someone who thinks
their overall content is descriptive. (More on this below.)
Second, even if one
denies that the overall content of moral judgments is descriptive content,
there is still a kind of descriptive content that is possessed both by ordinary
descriptive beliefs and by moral beliefs (as illustrated above in the pair of
statements about Bertie). We introduced our notion of core descriptive content
to refer to such content. Once one construes a base-case moral judgment as an
ought-commitment with respect to a core descriptive content, conceptual space
thereby opens up for the claim that the judgment’s overall declarative content
is cognitive content on the one hand (so that the state is a genuine belief), but is nondescriptive on the
other hand. Even though the state is a genuine belief, by virtue of being an
ought-commitment with respect to a core descriptive content, it doesn’t follow
that its overall declarative content is descriptive content.
Third, on standard
accounts of these matters, a belief involves a relation between a believer
(speaker) and a proposition (or sentence, or whatever) such that what is
believed is something having overall descriptive content. This conception of
belief presupposes the semantic assumption and makes the very idea of nondescriptivist
cognitivism incoherent. By contrast, our framework opens up the possibility
that certain genuine beliefs have overall declarative content that is not
descriptive. Thus, the framework calls into question the common assumption that
a belief is always a relation between a believer (speaker) on the one hand, and
on the other hand a proposition
constituting the belief’s overall declarative content.
Fourth, in maintaining
that there are two distinct base-case species of belief—is-commitments and
ought-commitments—we are maintaining that states of both types exhibit certain
generic kinds of functional and phenomenological features that qualify them as
genuine beliefs. However, in maintaining that ought-commitments are a distinct
kind of commitment, to be distinguished from is-commitments, we are also
maintaining that ought-commitments exhibit certain functional and
phenomenological features that are distinctive of this sort of judgment. We
have noted the action-guiding character of typical moral judgments, and here it
is worth mentioning that understanding base-case moral beliefs as essentially
ought-commitments with regard to a core descriptive content, helps accommodate
the widely shared internalist intuition that there is some intimate relation
between having a moral belief and action. That is, the very idea of an
ought-commitment suggests a kind of commitment oriented toward appropriate
action vis-à-vis the specific core descriptive content of the belief. The way
to understand this manner of action-orientation is by way of examining the role
of such beliefs in the overall cognitive economy of agents.
Just as
beliefs are psychological commitment states with a certain distinctive role in
psychological economy, assertions are speech acts that play a certain
distinctive sociolinguistic role—a role in interpersonal dynamics. An assertion
is a stance-taking speech act, an act
through which (i) one expresses an is-commitment or an ought-commitment with
respect to a core descriptive content, and thereby (ii) one positions oneself,
within the context of sociolinguistic dynamics, vis-à-vis that core content. A
stance is an orientation thereby occupied, within an interpersonal situation.
An ought-stance, in particular, is a distinctively action-guiding orientation. For instance, to take an ought-stance
with respect to the core descriptive content, Bertie’s mailing the parcel, is to engage in an action-guiding
speech act whose role within interpersonal dynamics is importantly similar to
the role of the corresponding psychological ought-commitment (the moral belief)
within intrapersonal cognitive economy. This sociolinguistic role involves reasons for action, and a preparedness
to provide them. By asserting that Bertie ought to mail the parcel, one
normally signals one’s willingness to defend one’s ought-commitment on this
matter over and against opposing ought-commitments, including a willingness to
give reasons for such a commitment. Normally it is understood that the reasons
one is prepared to give are of a certain distinctive kind that, e.g., appeal to
impartial considerations bearing on the issue. In general, one enters the space
of interpersonal moral discourse and reasoning bound by the sorts of
conventions (often unstated and partly inchoate) that govern interpersonal
deliberation and discussion about moral issues.[14]
2. The framework continued: logically complex cases
We now generalize our framework, by extending it to
beliefs and assertions whose overall declarative content has truth-functional
and/or quantificational logical complexity, and/or embedded deontic operators.[15]
To begin with, let us restrict the notion of “core descriptive content” to atomic descriptive content—the kind of
content expressible by atomic sentences. Given this stipulation, here is the key
idea for generalizing our approach: whereas a base-case belief is a logically simple commitment-state with
respect to a single core descriptive
content, a non-base-case belief is a logically
complex commitment-state with respect to several core descriptive contents. Whereas base-case beliefs
comprise two logically simple commitment-types (viz., is-commitment and
ought-commitment), non-base-case beliefs comprise a whole recursive hierarchy
of logically complex commitment-types, corresponding to the various logical
forms that can be exhibited by logically complex declarative sentences. The
essential feature of any given logically complex commitment-type is its
distinctive constitutive inferential role
in an agent’s cognitive economy (insofar as the agent is rational), a role
involving the relevant core descriptive contents.
First let us consider
cases of moral belief exhibiting truth-functional complexity, i.e., complexity
involving connectives but not quantifiers. On our view, such a belief is to be
understood as a logically complex commitment state with respect to a sequence of core descriptive contents.
So, for example, consider the belief that either
Jeeves mailed the parcel or Bertie ought to mail the parcel. This belief is
a logically complex commitment-state of the logical type [f v (Ought)y], with respect to the
sequence of core descriptive contents <Jeeves mailed the parcel, Bertie
mails the parcel>. The key to understanding this belief, and others of the
same logical type, involves understanding their constitutive inferential role
in the psychological economy of the agent. In particular, their role is to
combine in a distinctive way with other beliefs (other commitment-states) to
inferentially yield further beliefs (further commitment-states). One way to put
the main idea about such logically complex commitment states is that the simple
constituents of complex commitment states are logically “in the offing” in the
sense that the complex commitment state involved in the disjunctive belief,
when combined with an appropriate additional belief, rationally-inferentially
yields an ought-commitment with declarative moral content. In the example at
hand, the embedded moral constituent, Bertie
ought to mail the parcel, is in the offing in the sense that the complex
commitment-state in question, together with the belief that Jeeves did not mail
the parcel, inferentially yields (at least for the minimally rational agent) an
ought-commitment with respect to Bertie’s mailing the parcel.
Now consider cases of belief with quantificational complexity, i.e.,
complexity involving quantifiers (and perhaps connectives too). Such a belief
is a logically complex commitment-state vis-à-vis a set of sequences of core descriptive contents. So, for instance,
the belief that anyone who pinched Uncle
Willoughby’s parcel ought to mail it, is a logically complex commitment
state of the logical type (a)[Fa É (Ought)Ya], with respect to a set of sequences of core
descriptive contents {<Bertie pinched the parcel, Bertie mails the parcel>,
<Aunt Agatha pinched the parcel, Aunt Agatha mails the parcel>, . . .},
etc. Again, the essential feature of this type of commitment state is its
constitutive inferential role in the psychological economy of the agent. For
someone whose belief has the universally quantified declarative content in
question, other beliefs with declarative moral content are “in the offing” in
the sense that the complex commitment involved in the universally quantified
belief, when combined with an appropriate additional belief (e.g., the belief
that Bertie pinched the parcel) rationally-inferentially yields an
ought-commitment with declarative moral content (e.g., the belief that Bertie
ought to mail the parcel).
The aspect of logical complexity arising from embedded ‘Ought’
operators gets accommodated too, within this framework. Each belief-type
involving embedded deontic operators will have its distinctive, constitutive,
inferential role in the psychological economy of the rational agent. It is the
business of deontic logic to systematize these logical roles.
As we said, on this approach there is a whole recursive hierarchy of
commitment-types of increasing logical complexity, corresponding to the
hierarchy of increasingly complex logical forms exhibited by declarative sentences
that can express the overall declarative content of a belief. Each such
commitment is directed toward a core descriptive content, or a sequence of core
descriptive contents, or a set of sequences of core descriptive contents. And
each such commitment has a constitutive inferential role in psychological
economy—a role involving the core descriptive content(s) toward which the
commitment is directed.[16]
These
observations about beliefs with logically complex declarative content can be
extended, mutatis mutandis, to assertions. Whereas a base-case assertion is a
speech act of taking a logically simple
stance with respect to a single core
descriptive content, a non-base-case assertion is a speech act of taking a logically complex stance with respect to
several core descriptive contents. A
logically complex stance plays a constitutive inferential role in the dynamics
of sociolinguistic intercourse that is analogous to the constitutive
intra-psychological inferential role of logically complex beliefs. The constitutive
inferential role is this: to combine with other sociolinguistic stances, taken
by making additional assertions, to generate—often automatically and
implicitly—certain further stances that are logically implied by one’s overt
stance-taking speech acts. Implicit is-stances and ought-stances are thus “in
the offing” when one makes a logically complex assertion: such an assertion, in
combination with appropriate additional ones, will logically generate implicit
is-stances or ought-stances with respect to certain core descriptive contents.[17]
Suppose, for
example, that one asserts, either Jeeves
mailed the parcel or Bertie ought to mail the parcel, and one also asserts Jeeves did not mail the parcel. The
former assertion is a logically complex stance-taking speech act, of the
logical type [f v (Ought)y], with respect to the sequence of core
descriptive contents, <that Jeeves
mailed the parcel, that Bertie mails that parcel>. The latter assertion is a
speech act of logical type ~f, with respect to the core descriptive
content, that Jeeves mailed the parcel. In performing these two speech acts
together, one thereby comes to occupy, as a matter of the logic of speech acts,
an ought-stance with respect to the core descriptive content, that Bertie mails
the parcel.[18]
3. NDC as a consistent metaethical position
Our
main task has been to provide a framework for belief and assertion that renders
the basic tenets of NDC consistent. According to NDC, judgments and statements
with moral content are genuine beliefs and assertions, having cognitive
content, and yet the overall declarative content of such an item is not
descriptive. If one accepts the semantic assumption, then such a view is
outright inconsistent (since according to that assumption cognitive content
just is descriptive content). According to our framework this assumption is not
taken for granted; it is quite consistent with our framework to hold that some
beliefs and assertions lack overall descriptive content. Consider, once again,
base-case moral beliefs and logically complex moral beliefs.
As we have already noted in passing, nothing in the notion of a
base-case belief or assertion, construed as an ought-commitment with respect to
a core descriptive content, forces on us the claim that the overall declarative
content of a such a belief or assertion is descriptive content. And the point
generalizes: in light of the previous section, nothing in the notion of a
morality-involving logically complex belief (or assertion)—understood as a
logically complex commitment with respect to a multiplicity of core descriptive
contents (where what is essential about the belief or assertion is its
constitutive inferential role)—forces on us the claim that the overall
declarative content is descriptive. Thus, the position we call nondescriptivist
cognitivism is rendered consistent by our proposed framework.
On the other hand, the framework certainly does not entail nondescriptivist cognitivism.
Rather, it is neutral with respect to competing metaethical positions that
recognize that moral thought and discourse involves genuine beliefs and
assertions—that is, competing versions of cognitivism. In particular, our
framework is consistent with descriptivist metaethical views. The
descriptivist, that is, could grant what we have said about ought-commitments
and is-commitments being distinct commitment types, and about beliefs and
assertions with complex overall declarative content being logically complex
commitments vis-a-vis core descriptive contents, without having to deny that morality-involving
beliefs and assertions have overall descriptive content (the fundamental claim
of the descriptivist). Our framework, recall, leaves open whether or not the
overall declarative content of a moral belief is descriptive.
A critic might be inclined to say that we are trading in the
implausibility of metaethical descriptivism, with its burdensome metaphysical
commitments, for a complicated and ad hoc
framework for belief and assertion, and thus that there is a more or less
straight trade off—metaphysical extravagance for semantic complexity. Not so,
however, as we will now explain.
Not only is the framework consistent with descriptivism (as already
explained), but there is good reason for the descriptivist to embrace our
framework: viz., doing so allows the descriptivist to accommodate strongly held
and deeply shared internalist intuitions about moral thought and discourse.
Adopting the framework, descriptivists would maintain that the belief that,
e.g., Bertie ought to mail the parcel is both
an is-commitment with respect to the overall declarative content (which they
understand to be descriptive), that it
ought to be that Bertie mails the parcel, and an ought-commitment with
respect to the core descriptive content, that
Bertie mails the parcel. Given the specific action-oriented functional role
and phenomenology distinctive of ought-commitments, descriptivists could
thereby neatly combine their view with internalism. (The point generalizes to
encompass morality-involving logically complex commitments as well, since
action orientation is inferentially ‘in the offing’ for these too.) So
descriptivists have no reason to suppose that our framework begs any important
metaethical questions against them, and they have good reason to positively
embrace it.
An adequate metaethical position should be faithful to the phenomena it
seeks to understand. If the phenomena are sufficiently complex, then a
corresponding degree of complexity in one’s metaethical position is theoretically
appropriate—not ad hoc. Moral
judgment and moral discourse have internalist aspects—a form of complexity in
the phenomena whose theoretical illumination evidently requires the kind of
complexity exhibited by our proposed framework. So even descriptivists have
ample reason to embrace the framework.[19]
V. Nondescriptivist Cognitivism versus
Descriptivist Cognitivism
Although we will not
attempt to explain why we think that all of the various traditional metaethical
views are unsatisfying, we do want to say something about the plausibility of
our view vis-à-vis descriptivist versions of cognitivism. Doing so is
especially important because, as just explained, there is a version of
internalist descriptivism that draws upon our own proposed generic framework
for belief as a way of combining the idea that moral judgments are genuine
beliefs (and moral statements are genuine assertions) with the idea that they
are action-guiding. Why prefer our nondescriptivist cognitivism to
descriptivism? In particular, why prefer our view to the kind of descriptivist
cognitivism that accommodates the internalistic aspects of moral judgment and
moral discourse?
We will briefly mention three philosophical reasons for doubting that
the declarative content of moral beliefs is descriptive. First is what
Second, in arguing
that moral judgments are a species of belief, part of our plan was to show that
construing them as beliefs does not commit one to the further theoretical claim
that they possess descriptive cognitive content. The point here is that
attributing to such beliefs this sort of content is gratuitous for purposes of
understanding them as beliefs and understanding their distinctive
action-guiding role in our lives. In light of their psychological role and
associated phenomenology, there simply is no apparent need to burden them with
a kind of theoretical commitment which, given the location problem, cannot be
discharged.
Third, the case against descriptivism receives additional support from
considerations of convervatism with respect to the nature and evolution of
human concepts. Applied to moral notions the argument would go like this. Moral
discourse, and moral concepts employed in such discourse, play an indispensable
role in human life that would survive rejection of the idea that there are
objective moral facts that moral claims purport to describe. Indeed, after Mackie
argued that all affirmative moral sentences are false because they involve (so
he thought) metaphysical commitments to ontologically ‘queer’ properties, he
did not advocate eliminating the use of moral concepts and moral discourse;
rather, he went on to propose a normative ethical system based on a certain
conception of human flourishing. Now if we assume that human concepts tend to
evolve in a broadly pragmatic way and are thus not likely to have application
conditions that are more demanding than is required for the purposes they
serve, then the fact that moral discourse would survive the rejection of
objective moral facts and properties strongly suggests that such discourse does
not have any such metaphysical commitments.
VI. Semantic Illumination by Triangulation
Our main task is completed: we have sketched the rudiments of a new
kind of metaethical theory, involving a generic conception of belief and
assertion that renders the view a consistent position, and we have indicated
briefly what virtues our view has vis-à-vis the more standard metaethical
options. Obviously, filling out the theory and defending it against all
relevant challenges would require a book or at least a series of articles.
However, in the space remaining we will address, if only in a preliminary way,
certain questions and matters of detail that have very likely occurred to the
attentive reader. In this section we will make some remarks about philosophical
methodology in relation to filling out our positive metaethical story about the
semantics of moral thought and discourse. Then, in the following section, we
will take up more specific questions concerning truth ascription, logical
embedding, moral progress, and moral seriousness.
According to our
nondescriptivist cognitivism, the contents of moral beliefs and assertions are
sui generis in the sense that they cannot be reduced to or analyzed as
equivalent to other types of declarative or nondeclarative contents (or even a
combination of the two). In this respect, our view is unlike older
nondescriptivist views according to which, for instance, moral beliefs and
assertions are primarily commands and so have prescriptive content as primary
in addition to any descriptive content they may also possess. In rejecting all
reductive semantic projects in relation to understanding moral thought and
discourse, the appropriate response to questions like ‘What is the content or
meaning of moral judgment, M?’ is simply to repeat the content of the judgment
in question. Thus: ‘What is the content of Genocide
is wrong?’ Answer: genocide is wrong. However, offering only such a
disquotational response to these kinds of questions about content does not mean
that our view is deeply mysterious or that we are obscurantists about matters
of moral semantics. Quite the contrary. We maintain that one gains sufficient
semantic illumination of the nature of nondescriptive cognitive content
precisely by coming to understand the psychological states and speech acts that
have it, as states and speech acts
involving a certain distinctive kind of commitment (or stance taking) with
respect to certain core descriptive contents. Such understanding involves
coming to appreciate in enough detail the psychological role and associated
phenomenology definitive of the relevant psychological states, and,
correspondingly, by coming to appreciate in enough detail the sort of
sociolinguistic role of the relevant speech acts. In short, illuminating the
characteristic roles of moral thought and discourse helps one understand the sui
generis kind of cognitive content moral beliefs and utterances possess. We call
this kind of methodology for illuminating content, triangulation, which we have employed in sketching our semantic
story about both base case and logically complex moral beliefs and assertions.
Thus our break with metaethical tradition involves not only our proposed
metaethical theory but our methodology as well.
VII. Work to be Done
We
turn finally to various challenges that may have occurred to our readers, in
order to indicate at least roughly how we propose to deal with them.
Specifically, we take up issues of truth ascription, logical embedding, moral
progress, and moral seriousness.
1. Truth ascription
According to NDC,
moral judgments are genuine beliefs, and moral utterances are genuine
assertions. But the concepts of belief and assertion are linked by platitudes
to the concept of truth: a belief is a psychological state that aims at truth;
to assert is to set forth as true. How does our view deal with matters of truth?
After all, being nondescriptivists, we claim that moral beliefs and associated
speech acts lack overall descriptive content; they are not in the business of
representing or purporting to describe the world.
On our view, the
proper way to gain illumination about matters of truth in relation to moral
thought and discourse is to focus on truth ascriptions to moral statements as
metalingustic speech acts, and ask about the nature of these speech acts. When
one thinks or remarks, ‘The claim that apartheid ought to be stopped is true’,
what is one doing? The appropriate answer involves noting that such a truth
ascription constitutes a morally engaged
semantic appraisal: one that is infused with one’s own moral commitment. The
main idea can perhaps be conveyed by saying that truth ascriptions to moral
statements involve a kind of appraisal in which semantic and moral are
“fused”—which is to be expected, since ordinary uses of the truth predicate
operate in accordance with schema T.[20]
In recent years, so-called minimalist
treatments of truth have been developed and defended—views that attempt to make
sense of truth ascription without robust metaphysical commitments. Our view is
in the minimalist spirit though we would insist on two things. First, to
understand truth minimalistically in one discourse does not commit one to
minimalism in relation to every mode of discourse.[21]
Second, there is an interesting story to be told about moral truth ascription;
our view is not a simple redundancy view.
2. Embedding
A
certain problem involving embedded contexts has been frequently pressed against
various forms of nondescriptivism. One common way of raising the embedding
challenge is to point out that inferences like the following seem to be valid:
(1) One ought not to kill; (2) If one ought not to kill, then one ought not pay
someone to kill; thus, (3) One ought not pay someone to kill. The problem for,
say, an emotivist is that according to emotivism, the meaning of premise (1) is
to be understood in terms of its noncognitive emotive role in thought and
assertion, viz., to express one’s emotion and influence the attitudes of
others. However, in premise (2), where (1) occurs as the antecedent of the
conditional, (1) is not expressed with its typical emotive role; one who affirms
premise (2) is not thereby committed to affirming its antecedent. But then it
appears that one has to say that ‘one ought not to kill’ differs in meaning in
its two occurrences in the argument which implies that, despite appearances,
the argument is not valid; it commits the fallacy of equivocation. The critic
pressing this objection presumably thinks that only if moral statements have
descriptive content, and so can be understood in terms of some set of
descriptive truth conditions—something that a statement carries from unembedded
to embedded contexts—can we make sense of moral modus ponens and other such
valid inferences.[22]
Our reply to this
challenge is implicit in our above discussion of logical complexity. In
developing our framework in connection with logically complex moral beliefs and
assertions, we noted that the declarative content of such beliefs and
statements can be triangulated in terms of their constitutive inferential role
in modus ponens and other argument forms. Thus, the conditional statement, ‘If
one ought not to kill, then one ought not pay someone to kill’, is to be
understood primarily in terms of its role in mediating inference from an
affirmation of its antecedent to an affirmation of its consequent, as in the
little argument featured above. So on our view, to get a handle on embedded
moral claims involves understanding the role of the kinds of logically complex
statements that embed them. What one can say about the contents of embedded and
unembedded occurrences of some one moral claim is that (1) they share the same
core descriptive content, (2) in an embedded context an ought commitment with
respect to that core content is “suspended,” but nevertheless (3) the overall
claim containing the embedded context expresses a logically complex commitment
state whose constitutive role in inference is such that an ought commitment
with respect to the relevant core descriptive content is “in the offing.” To
make these observations, we think, is to make sense of valid inference
involving embedded moral constituents.
Often when
the embedding issue is raised, those posing the challenge assume that one must
first give an account of the meaning of moral statements, and then show that
their meaning (according to the given account) remains constant when the
statements are embedded. But, given our proposed framework for belief and
assertion, this methodological assumption gets called into question. On our
approach, what it is for a statement
S with nondescriptive cognitive content to have constant meaning, whether
unembedded or in various embedded contexts, just
is for the states and speech acts whose overall declarative content
includes S (i.e., whose overall declarative content is expressible by a
statement with S as constituent) to figure in certain specific constitutive
inferential connections involving S’s core descriptive content. This is a
dialectical reversal, turning the standard embedding problem on its head.
(Remember: on our approach, one explains nondescriptive cognitive content by
explaining the psychological states and speech acts that have it, as certain distinctive kinds of
psychological or sociolinguistic commitments with respect to certain core
descriptive contents. Such commitments bear constitutive inferential
connections to one another.)[23]
3. Moral progress and taking morality seriously
For a
descriptivist-realist, intellectual moral progress is a matter of one’s moral
beliefs coming to better approximate the moral facts. But if moral belief and
assertion are not primarily in the business of describing or representing
in-the-world moral facts, then how can we make sense of genuine moral progress?
Put another way, how can our view distinguish between mere change in moral
belief and genuine progress? And, relatedly, if there is no metaphysical anchor
for moral thought and discourse, then why take it seriously, why not construe
moral discussion and disputes as being more like disputes about matters of
taste?
These challenges focus
on our irrealist moral metaphysics, and we consider them to be some of the most
difficult for any moral irrealist. Here, then, is an indication of how we would
respond to these challenges, though they certainly deserve a more thorough
reply than we can offer here.
Of course, on our
view, moral progress of the sort in question is not to be understood as a
matter of bringing one’s beliefs into closer proximity to a realm of moral
facts. We propose that, instead, one think of moral progress as something to be
judged from within a committed moral outlook: when one makes judgments about
moral improvement, one does so from an engaged moral perspective. In judging,
for example, that moral progress was made in the
[T]he minimalist will have to admit that such ideas of progress, or
deterioration, are ones for which we can have use only from within a committed moral point of view;
and that the refinement of which our moral sensibilities are capable can only
be a matter of approaching a certain equilibrium as appraised by the exercise
of those very sensibilities. (Wright, 1992: 168-9).
Again, we think the
challenge to make sense of moral seriousness does not require some metaphysical
backing for moral thought and discourse. Rather, on our view, the challenge
regarding moral seriousness is plausibly understood as a moral challenge: why ought we take our moral views seriously? And
the appropriate response to such a challenge is to give moral reasons—reasons
that, for instance, will likely appeal to the important role of morality in
people’s lives. Like our reply to the moral progress challenge, our reply here
is to view the challenge as one to be appropriately dealt with from within a
committed moral outlook.[24]
VIII. Conclusion
We think it is time
for a change in metaethics, and only by challenging certain pervasive
philosophical assumptions is one likely to make progress. Our proposal is to
rethink fundamental assumptions about the nature of belief and assertion;
specifically, we challenge the idea that all belief-eligible and assertible
contents are descriptive—what we call the semantic assumption. We have set
forth a framework for belief and assertion that does not presuppose the
semantic assumption, thus allowing for the possibility
of beliefs and assertions that are not descriptive. Nondescriptivist
cognitivism embraces the framework, and also maintains that the overall
declarative content of moral beliefs and assertions is in fact not descriptive. The virtues of this metaethical position
are great. It surely deserves to be taken seriously as a theoretical option in
metaethics. Indeed, we submit that it ought to be the default view.[25]
References
Dreier, James. 1996.
“Expressivist Embedding and Minimalist Truth.” Philosophical Studies 83: 29-51.
Graham, G. and T. Horgan.
1994. “Southern Fundamentalism and the End of Philosophy.” Philosophical Issues 5: 219-47.
Hare, R. M. 1952. The Language of Morals. London: Oxford
University Press.
Hare, R. M. 1970. “Meaning
and Speech Acts.” Philosophical Review
79: 3-24.
Horgan, T. 1994.
“Transvaluationism: A Dionysian Approach to Vagueness.” Southern Journal of Philosophy 23: 97-126. Supplementary volume on
Vagueness.
Horgan, T. 1995. “Critical
Study of Crispin Wright’s Truth and
Objectivity. Nous 29: 127-38.
Horgan, T. 1996. “The Perils
of Epistemic Reductionism.” Philosophy
and Phenomenological Research 66: 891-97.
Horgan, T. and M. Timmons.
1991. “New Wave Moral Realism Meets Moral Twin Earth.” Journal of Philosophical Research XVI: 447-65. Reprinted in Rationality, Morality, and Self-Interest,
edited by John Heil, Rowman & Littlefield (1993): 115-33.
Horgan, T. and M. Timmons.
1992a. “Troubles on Moral Twin Earth: Moral Queerness Revived.” Synthese 92: 221-260.
Horgan, T. and M. Timmons. 1992b. “Troubles for New Wave Moral
Semantics: The Open Question Argument Revived.” Philosophical Papers 21: 153-175.
Horgan, T. and M. Timmons. 1993. “Metaphysical Naturalism, Semantic
Normativity, and Meta-Semantic Irrealism.” Philosophical
Perspectives 4: 180-204.
Horgan, T. and M. Timmons. 1996a. “From Moral Realism to Moral
Relativism in One Easy Step.” Critica
38: 3-39.
Horgan, T. and M. Timmons. 1996b. “Troubles for Michael Smith’s
Metaethical Rationalism.” Philosophical
Papers 25: 203-231.
Horwich, Paul. 1990. Truth.
Oxford: Basil Blackwell.
Jackson, Frank. 1998. From
Metaphysics to Morals. Oxford and New York: Oxford University Press.
Leblanc, Hugues and Wisdom, William. 1993. Deductive Logic, Third Edition. Englewood Cliffs, NJ: Prentice
Hall.
Mackie, J. L. 1977. Ethics:
Inventing Right and Wrong. Harmondsworth: Penguin.
Mandelbaum, Maurice. 1955. The
Phenomenology of Moral Experience. Glencoe, IL: The Free Press.
Schiffer, Stephen. 1990. “Meaning and Value.” Journal of Philosophy 87: 602-17.
Sinnott-Armstrong, Walter. Forthcoming. “Expressivism and Embedding.” Philosophy and Phenomenological Review.
Smith, Michael. 1993. “Objectivity and Moral Realism: On the
Significance of the Phenomenology of Moral Experience.” In J. Haldane and C.
Wright, eds., Reality, Representation and
Projection. Oxford: Oxford University Press.
Stevenon, C. L. 1944. Ethics and
Language. New Haven: Yale University Press.
Stoljar, Daniel. 1993. “Emotivism and Truth Conditions.” Philosophical Studies 70: 81-101.
Timmons, M. 1999. Morality Without Foundations: A Defense of Ethical Contextualism.
Wodehouse, P. G. 1967. The World of Jeeves.
Wright, Crispin. 1992. Truth and Objectivity.
[1] Both sides agree that something more is required for being a genuine belief and a genuine assertion than having declarative content. They agree, for instance, that if the declarative content of moral judgments and statements is reducible to (i.e., synonymous with, or paraphrasable by, or theoretically modelable by) the kind of content that is linguistically expressable by certain nondeclarative sentences—e.g., imperatives—then the judgments and statements in question are not full-fledged beliefs and assertions. So, even though it is grammatically permissible to append ‘believes that’ (and likewise ‘asserts that’) to any sentence with declarative content, the shared assumption is that being a full-fledged belief or assertion requires more—viz., declarative content that is not reducible to nondeclarative content.
[2] In this paper we restrict talk of descriptive content to indicate content that represents the world as being a certain way. One might use ‘descriptive content’ in a broad way that would apply to any meaningful declarative sentence, but that is not how we are using the expression. To say that the declarative content of a mental state, judgment, or sentence is descriptive, then, is to say that it purports to describe or pick out some kind of fact in the world. In metaethics, such facts might be understood to have a strong mind-independent status, as the moral realist claims, or they might be tethered to the beliefs or attitudes (actual or ideal) of individuals or groups, as relativists and rationalists would have it. In short, to have descriptive cognitive content is to purport to be descriptive of some sort of fact more robust in nature than is consistent with a minimalist understanding of fact talk.
[3] It is also worth noting that the semantic assumption is presupposed when philosophers employ the ‘direction of fit’ metaphor in attempting to distinguish beliefs from desires. Beliefs, it is said, aim at the truth and can be appropriately characterized (at least in contrast to desires) as psychological states that are supposed to fit the world; beliefs that fail to do so are mistaken. Desires, it is said, have an opposite direction of fit: they aim at satisfaction, which obtains when the world fits them.
[4] Error theories too embrace the semantic assumption, although with a theoretical twist. An error theorist assumes that non-defective cognitive content is descriptive, way-the-world-might-be, content. Given this assumption, moral-evaluative content is then construed as defective cognitive content: on the one hand it is belief-eligible and assertoric, because it purports to constitute or specify a genuine way the world might be; but on the other hand it is defective, because it does not in fact do so. This characterization holds for the classic version of error theory in Mackie (1977), and also for the more recent version in Schiffer (1990).
[5] It is crucial to understand, however, that we retain the traditional assumption that genuine cognitive content is not reducible to content expressible by nondeclarative sentences; i.e., we assume that if moral declarative content were reducible to nondeclarative content, then it would not be cognitive content, and moral judgments and statements would not be full-fledged beliefs and assertions (cf note 1).
[6] Are we, then, so-called minimalists about belief and assertion? That depends on how one uses the term ‘minimalism’. Let type-1 minimalism be the claim that moral declarative content counts as cognitive content even if it is reducible to nondeclarative content; and let type-2 minimalism be the claim that moral declarative content counts as cognitive content even if it is not descriptive content. We espouse minimalism of type 2, but not of type 1. (The two types of minimalism will be regarded as equivalent by someone who accepts the following modified semantic assumption: all declarative content either (i) is descriptive content, or (ii) is reducible to to nondeclarative content. We deny the modified semantic assumption, of course, in addition to denying the semantic assumption itself.)
[7] Someone who is a type-1 minimalist about belief and assertion (cf. note 6) will also reject the semantic assumption, but on different grounds than we do—viz., on the basis of the claim that declarative content automatically counts as cognitive content even if it is reducible to nondeclarative content. (A type-1 minimalist will consider the term ‘noncognitivism’ an inappropriate label for metaethical positions affirming the reducibility of declarative to nondeclarative content.) But insofar as the type-1 minimalist embraces the modified semantic assumption (cf. note 6), the menu of metaethical options will remain largely as it was before, except that the categories of ‘belief’ and ‘assertion’ will now be applied to the kinds of psychological states and speech acts described by traditional versions of nondescriptivism like emotivism and prescriptivism. Since we ourselves deny the modified semantic assumption, however, our position opens up fertile new metaethical territory even from the perspective of type-1 minimalism. For, it remains an important theoretical novelty to claim, as we do, that moral content is a kind of cognitive content that is neither descriptive nor reducible to nondeclarative content.
[8]
The parcel contains Uncle Willoughby’s book manuscript that he left on the hall
table, to be mailed to the publisher. Bertie (
[9] See for example, Mandelbaum (1955) and Smith (1993) for characterizations of these features and also those we mention in the next paragraph.
[10] To say that moral judgments directly dispose us toward action independently of pre-existing desires, and that they have motivational force independently of such desires, leaves it open whether (i) these judgments play this causal role all by themselves, or instead (ii) they generate new desires which then play that role.
[11] A problem with standard versions of internalism is that they make the connection between moral judgment and appropriate motivation exceptionless. Although we ourselves maintain that part of the concept of a moral judgment is that such judgments typically are motivational, we also hold that the connection to motivtion is “soft”: it allows the possibility of abnormal cases in which (for some explicable reason) the typical motivating aspect is deadened or absent (cf. Timmons 1999, pp. 140-42). Moreover, it should be acknowledged that the action-guiding role of moral judgments is sometimes somewhat indirect, for instance when one condemns persons long dead for actions they performed long ago. Still, normally the action-guiding aspect of moral judgment is operative either directly or at least indirectly, with respect to potential behavior in situations either actual or counterfactual.
[12] Unless otherwise indicated, when we speak of the content of a judgment (or assertion) we mean its overall declarative content. Nondescriptivist cognitivism claims that this kind of content is cognitive, while also claiming that it is not descriptive. In the course of the discussion below we will describe an additional, “inner,” kind of content involved in moral judgments which is descriptive but is distinct from their overall declarative content.
[13] For simplicity’s sake, we focus exclusively on moral beliefs expressible linguistically by the deontic operator ‘it ought to be that’, thus ignoring those kinds of moral beliefs expressible linguistically by operators like ‘it is permissible that’ and ‘it is good that’. We leave open how exactly to understand these latter beliefs as types of evaluative commitment state, although we expect that our general approach to understanding ought-commitment states can be appropriately adapted to the understanding of evaluative commitment states of these other sorts.
[14] What we are calling a “stance” is a sociolinguistic orientation whose role in social dynamics is largely parallel to the role of a commitment state within a person’s own psychological economy. (Indeed, a stance is an interpersonal kind of commitment-state, as distinct from the psychological kind.) The notion of an assertion as a stance-taking speech act certainly deserves further elaboration—as does the notion of a stance itself, and the distinction between is-stances and ought-stances. In our view, one can make a good start on these matters by considering the treatment of the speech-act dimension of moral language in Hare (1952, 1970). Much of what Hare says about the moral statements as speech acts is both plausible and consistent with our own proposed framework for belief and assertion.
[15] Note that base-case moral commitment states are not logically complex in this sense, even though they do have deontic logical complexity, formalizable in terms of a single deontic operator appended to an atomic sentence. Also, we should now make a clarificatory comment about our official formulation of nondescriptivist cognitivism in section I above. Thesis (1) of NDC mentions declarative judgments and statements “with moral content”; these include not only base-case moral judgments and statements, but also logically complex ones with constituent moral content.
[16] What we have said here represents only a sketch of an account of logically complex commitment states; various questions are left open, for more extensive treatment elsewhere. For instance, we are inclined to add two further claims. First, a logically complex belief or assertion whose overall declarative content is descriptive will count both as a logically complex commitment with respect to a multiplicity of core descriptive contents, and as an is-commitment with respect to its overall declarative content. Second, although a logically complex declarative content can itself be the object of an is-commitment (or an ought-commitment), this is so only if this declarative content is descriptive.
We also suspect that a more
extensive treatment should distinguish between logical commitments and psychological
commitments, and should allow for the possibility that a logical commitment can
exist even in the absence of corresponding psychological commitment. (Such a
commitment would not be a belief,
since beliefs are psychological states.) If an agent fails to make a rationally
dictated inference, for instance, then that agent still has a logical
commitment whose declarative content is the conclusion of the inference, even
though the agent lacks the appropriate belief.
[17] In calling a stance implicit we mean that it is operative within sociolinguistic dynamics without being explicitly expressed by means of a new assertion. That is, it is sociolinguistically implicit. On the other hand, one or more of the parties in the relevant discourse-community might fail to realize that certain stances expressed overtly by assertions logically generate a specific further stance. Perhaps one should say that relative to those parties, the logically generated stance is merely logical, rather than being sociolinguistically implicit. Cf. the analogous point about merely logical commitments vs. psychological commitments, in the preceding note.
[18] Our approach to logically complex beliefs and assertions with moral content can be suitably generalized to account for noncommittal psychological states (and associated utterances), such as wondering whether Bertie ought to mail the parcel. Such a state involves hypothetically “trying on” an ought-commitment. To understand such states involves understanding their various roles in the overall psychological economy of typical agents, especially their role in moral reasoning. Often, when wondering about some moral claim, an agent thinks through the issue by combining the hypothetical moral commitment in question with beliefs (both moral and nonmoral) in a process of moral reasoning in which she or he is tracing out the implications of adopting the hypothetical commitment. The generic notion of “trying on” an ought-commitment applies mutatis mutandis to a full range of noncommittal psychological states, including, for example, hoping-that and fearing-that states with moral content.
[19] Moreover, our point here about being faithful to the complexity of the phenomena applies mutatis mutandis to noncommittal psychological states (and associated utterances) with moral content, states of the sort discussed in the preceding note. When one hypothetically “tries on” an ought-commitment in one’s state of wondering (or hoping, or fearing, etc.), this includes trying on the internalist, action-oriented aspect of the ought-commitment.
[20] According to what we call contextual semantics, many terms—including the truth predicate—are subject to contextually variable semantic standards. In the case of moral thought and discourse, which is nondescriptive in overall declarative content, typically the contextually operative semantic standards governing the truth predicate dictate a morally engaged use—the use we have just explained. But in some contexts the semantic standards dictate a morally detached use of the truth predicate, under which ‘true’ signals language-world correspondence; on this usage, only statements whose overall declarative content is descriptive are either true or false. For further discussion of contextual semantics in general and of the truth predicate in particular see Horgan (1994, 1995, 1996), Horgan and Timmons (1993), and Timmons (1999), ch. 4.
[21] Thus, we are inclined to advocate a kind of pluralism about truth according to which there is a univocal notion of truth even though truth ascription may involve more or less robust metaphysical commitments in relation to different areas of thought and discourse. Pluralism about truth is also featured in Wright (1992).
[22] There is some controversy about how forceful this kind of objection really is; some claim that it can easily be skirted by the emotivist and by other brands of noncognitivism. See Horwich (1990) and Stoljar (1993). But see Dreier (1996) and Sinnott-Armstrong (forthcoming) who press the difficulty of the problem.
[23] Our approach does assume, of course, that there is an intelligible notion of logical consequence that applies to beliefs and assertions whether or not their overall declarative content is descriptive. But it is surely plausible that this is so—i.e., that logic governs psychological commitments and sociolinguistic stances, even those with non-descriptive declarative content. Indeed, in light of our remarks about truth ascription in section VII.1 above, consider truth-theoretic accounts (in contrast to model-theoretic accounts) of notions like logical truth and logical consequence—for instance, that of LeBlanc and Wisdom (1993), with its substitutional treatment of the quantifiers. If truth-value assignments are extended to encompass base-case ought-statements in addition to atomic statements, then the resulting truth-theoretic account of logical truth and of the logical-consequence relation can be interpreted (i) as applicable to morally engaged uses of the truth-predicate, hence (ii) as also applicable to morally engaged thought and discourse (which conforms to schema T), hence (iii) as applicable to beliefs and assertions even when their overall declarative content is not descriptive. Moreover, presumably the possible-world semantics of deontic logic could be smoothly incorporated into such a truth-theoretic approach, with possible worlds construed as specifiable by Carnap-style “state descriptions”: maximal consistent sets of atomic statements and negations of atomic statements. Elaborating these brief remarks is a task for another occasion.
[24] For some elaboration of the various challenges and replies featured in this section, see Timmons (1999), ch. 4.
[25] We
respectfully dedicate this paper to R. M. Hare, whose pioneering work in
metaethics has inspired us in many ways. A predecessor of the paper was
presented at a conference in 1994 at the