ArticlePDF Available

Predicates of Personal Taste, Semantic Incompleteness, and Necessitarianism



According to indexical contextualism, the perspectival element of taste predicates and epistemic modals is part of the content expressed. According to nonindexicalism, the perspectival element (a standard of taste, an epistemic situation) must be conceived as a parameter in the circumstance of evaluation, which engenders "thin" or perspective-neutral semantic contents. Echoing Evans (1985), thin contents have frequently been criticized. It is doubtful whether such coarse-grained quasi-propositions can do any meaningful work as objects of propositional attitudes. In this paper, I assess recent responses by Recanati, Kölbel, Lasersohn and MacFarlane to the "incompleteness worry". None of them manages to convince. Particular attention is devoted to MacFarlane's (2014) argument, which states that if perspectives must be part of the content, so must worlds, which would make intuitively contingent propositions necessary. I demonstrate that this attempt to defend thin content views such as nonindexical contextualism and relativism conflates two distinct notions of necessity, and that radical indexicalist accounts of semantics, such as Schaffer's necessitarianism, are in fact quite plausible.
Predicates of Personal Taste, Semantic Incompleteness, and Necessitarianism
Markus Kneer
University of Zurich
Penultimate version. Please consult & cite the final version:
Kneer, M. Predicates of personal taste, semantic incompleteness, and
necessitarianism. Linguistics and Philosophy (2020).
According to indexical contextualism, the perspectival element of taste
predicates and epistemic modals is part of the content expressed. According
to nonindexicalism, the perspectival element (a standard of taste, an epistemic
situation) must be conceived as a parameter in the circumstance of evaluation,
which engenders “thin” or perspective- neutral semantic contents. Echoing
Evans (1985), thin contents have frequently been criticized. It is doubtful
whether such coarse-grained quasi-propositions can do any meaningful work
as objects of propositional attitudes.
In this paper, I assess recent responses by Recanati, Kölbel, Lasersohn
and MacFarlane to the “incompleteness worry”. None of them manages to
convince. Particular attention is devoted to MacFarlane’s (2014) argument,
which states that if perspectives must be part of the content, so must worlds,
which would make intuitively contingent propositions necessary. I demonstrate
that this attempt to defend thin content views such as nonindexical
contextualism and relativism conflates two distinct notions of necessity, and
that radical indexicalist accounts of semantics, such as Schaffer’s
necessitarianism, are in fact quite plausible.
1. Introduction
The mainstream view of propositions is that they are the semantic values of
declarative sentences, the objects of propositional attitudes and illocutionary
For helpful feedback I would like to thank Jonathan Schaffer, François Recanati, Jack Woods,
John MacFarlane, John Perry, John Mackay, Dan Zeman, Julia Zakkou, Adam Marushak, Isidora
Stojanovic and several anonymous referees of this journal.
acts, and the bearers of truth values.
In order to fulfil these roles, propositions
must be minimally specific. That is to say, there are certain types of information
that propositions must contain in order to serve as the content of belief or
assertion, or to be evaluated with respect to truth and falsity. Debates
regarding how much information is required arise across different domains.
The locus classicus is time: Following Frege (1979), eternalists contend that all
propositions must contain temporal information and must thus be time-
specific. Temporalists, by contrast, hold that at least some propositions are
The last few decades have witnessed a broad variety of related
disputes: whether meteorological propositions must be location-specific,
whether propositions regarding epistemic modality require an epistemic
or whether the content expressed by claims of personal taste
must contain a standard of taste.
Advocates of a thin content view (concerning
a particular domain) think that the relevant type of proposition can be neutral
with respect to a particular feature F. Advocates of a rich content view (with
respect to a particular domain) contend that the relevant type of proposition
must always be F-specific.
Our point of departure is the question whether the contents of claims of
personal taste are neutral with respect to standards of tastes (or “judges” or
“perspectives”). Nonindexical contextualists answer this question in the
affirmative: In subjective discourse, perspective-neutral contents, they hold,
are required to account for the phenomenon of faultless disagreement, which,
according to nonindexicalists, requires a cogent explanation (section 2). There
are several ways to understand the argument from faultless disagreement
(section 3). On a rich view of content, what is said includes both the lekton
(Kaplanian content) and the circumstance of evaluation (what Recanati calls
“Austinian propositions”). However, invoking rich contents, it is argued, makes
it impossible to account for faultless disagreement. Alternatively, one might
Stalnaker (1970) is amongst the first to state this view as the orthodoxy; for a detailed defense
against recent alternatives, see Cappelen & Hawthorne (2009).
Eternalists include Stalnaker (1970), Wettstein (1979), Richard (1981, 1982), Salmon (1986),
Stanley (1997a, 1997b), Fitch (1998). Temporalists include Kaplan (1989), Aronszajn (1996),
Ludlow (2001), Recanati (2004), Brogaard (2012).
See inter alia Perry (1986), Carston (1988), Crimmins (1992), Taylor (2001), Recanati (2002, 2007),
Borg (2005), Cappelen and Lepore (2007), Sennet (2011).
See for instance Egan et al. (2005), Kölbel (2009), Von Fintel and Gillies (2008, 2011), Schaffer
(2011), MacFarlane (2011, 2014), Dowell (2011), Yanovich (2013), Marushak (2018), Marushak &
Shaw (in prep.), Roberts (in prep.). For empirical work on the question, see Knobe & Yalcin
(2014), Kneer (2015, in prep. b), Khoo (2015), Beddor & Egan (2018).
See inter alia Wright (2001), Kölbel (2004a, 2004b, 2009), Lasersohn (2005, 2008, 2011, 2016),
Stojanovic (2007, 2012), Recanati (2007), Glanzberg (2007), López de Sa (2007), Cappelen and
Hawthorne (2009), Saebo (2009) MacFarlane (2009, 2014), Egan (2010), Schaffer (2011), Collins
(2013), Ferrari & Zeman (2014), Kompa (2015), Kneer (2015), Dinges (2017), Zakkou (2017,2019).
suggest that the contents of propositional attitudes are exhausted by taste-
neutral contents, and that disagreement must be conceived in terms of
incompatible lekta. Views of this sort run into the incompleteness worrythe
concern that contents thus conceived are too coarse-grained to fulfil their roles
as the objects of propositional attitudes and illocutionary acts (section 4).
MacFarlane (2014), however, has suggested that the worry overgenerates: If
standards of taste must feature in the content, and if we assume that they do
not differ importantly from worlds, the latter, presumably, must also be part of
the content. Problematically, this would render a wide range of propositions
necessary which we standardly consider contingent. A thin content view might
thus be preferable.
After a detailed breakdown of MacFarlane’s argument from modal anxiety
(section 5), I propose several reasons why its conclusion should be resisted:
First, the argument conflates deep necessity (a modal property of sentences
and contents) and superficial necessity (a property of modal sentences and
contents, section 6). In virtue of specifying a world in the content, propositions
turn out superficially necessary. But this is unproblematic, as they remain
contingent in the intuitive, that is, the deep, sense: World-specific propositions
do not purport to represent a particular aspect of all possible worlds as
uniformly true or false across all worlds. They merely turn out true (or false) as
assessed from all possible worlds, since they concern but a single,
determinate, world (section 6.3). The second argument draws on recent work
by Schaffer (2012, 2018). Schaffer argues that the entire case for eternalism
(the view that all propositions must be time-specific) can be mimicked by
necessitarians (advocates of the view that all propositions must be world-
specific). Given such parallelism, and given that eternalism constitutes the
orthodoxy as regards temporal features of propositions, it is not clear why
necessitarianism must be avoided at all costs. But if it isn’t, then MacFarlane’s
argument from modal anxiety is either incomplete – it must establish wherein
the problem with necessitarianism actually lies – or else toothless (section 7).
Third, independent considerations in favour of taste-neutral contents
proposed by Kölbel (2009) and Lasersohn (2008) fail to convince; more general
arguments such as Recanati’s Argument from Innocence (2007) do not carry
over to the domain of personal taste (section 8).
If the considerations proposed in the paper are on the right track, we part with
a number of interdependent lessons: Bringing to bear Evans’ distinction
between superficial and deep necessity on the topic further strengthens
Schaffer’s case in favour of necessitarianism, and it shows why MacFarlane’s
response to the incompleteness worry might not convince. But if the
incompleteness worry remains intact, then the thin content view with respect
to taste claims is implausible. Since a commitment to thin contents constitutes
the only way to coherently formulate an argument from substantive faultless
disagreement, the threat the latter poses to indexicalist semantics with respect
to predicates of personal taste is limited at best.
2. Subjective Discourse
2.1 Perspectival Claims
A central debate in philosophy of language and linguistics concerns
perspectival expressions and claims, i.e. expressions and claims whose
extension depends on a contextually salient perspective. Examples include
predicates of personal taste (“delicious”, fun”), aesthetic predicates
(“beautiful”) and epistemic modals (“might, must, “possibly”).
orthodox approach to perspectival claims in truth-conditional semantics is
indexical contextualism. According to this view, the relevant perspective is
determined by the context of utterance, and it manifests itself at the level of
the content expressed by the utterance. On this approach, a claim of personal
taste such as (1) is standardly
taken to mean (2):
(1) Spinach is delicious.
(2) Spinach is delicious for me.
This intuitively plausible picture has recently been challenged (successfully or
not) on the basis of two widely-discussed arguments: The argument from
faultless disagreement,
and the argument from required retraction.
former has been used to motivate a position called nonindexical
contextualism, the latter is intended to motivate truth relativism. In this article,
we will principally focus on matters related to the argument from faultless
I will use “perspectives” as a general way to refer to perspectival features such as epistemic
perspectives, standards of taste etc.
In certain contexts the standard of taste invoked might be that of a person distinct from the
speaker (which Lasersohn (2005, p.671) calls an exocentric reading), that of a particular group,
or people in general (i.e. a generic reading). To facilitate discussion, we will focus principally
on what Lasersohn calls the autocentric readings of perspectival claims readings that invoke
the speaker’s own standard of taste which stand at the centre of the debate.
As regards the argument from faultless disagreement, cf. inter alia Kölbel (2004a, 2004b, 2009)
and Lasersohn (2005, 2009). For responses sympathetic to contextualism, cf. Glanzberg (2007),
Stojanovic (2007), Schaffer (2011), Sundell (2011) and Cappelen and Hawthorne (2009)
Cf. in particular MacFarlane (2007, 2014). For discussion focusing also on epistemic modals,
see Egan et al. (2005), Egan (2007), von Fintel & Gillies (2008), Schaffer (2011), Dowell (2011),
Yanovich (2013), Kneer (2015), and Lasersohn (2018).
2.2 The Argument from Faultless Disagreement
Consider the following exchange, in which Mary and Frank are having a dispute
about the culinary merits of spinach.
(3) Mary: Spinach is delicious.
(4) Frank: No, spinach is not delicious.
Mary and Frank seem to disagree. In the literature, it is commonplace to
assume that two individuals disagree with respect to a particular issue if the
contents of their beliefs p and q (expressed by their utterances or not) are
doxastically noncotenable,
and this appears to be the case. Curiously,
however, neither of the speakers seems to be at fault, in so far as neither needs
to revise their beliefs or retract their assertion. If the possibility of such faultless
disagreement is an important characteristic of disputes about taste,
indexicalist contextualism comes under pressure (or so the argument goes).
The indexicalist can account for faultlessness, since both Mary and Frank
express a speaker-relative content. Disagreement, however, is lost, as
becomes apparent once the perspectives that tacitly feature in the asserted
contents are made explicit.
The possibility of faultless disagreement in subjective discourse motivates
nonindexical contextualism. On this view, the content expressed by utterances
invoking predicates of personal taste does not contain a tacit standard of taste.
The perspectival element, which is also drawn from the context of utterance,
instead features as a parameter in the circumstance of evaluation. Since the
perspective-neutral content of an utterance of “Spinach is delicious” and the
content of “Spinach is not delicious” stand in direct contradiction,
disagreement is accounted for. Faultlessness is explained by the different
truth-values of the two claims. “Spinach is delicious” is true with respect to
Mary’s perspective, yet false with respect to Frank’s, and vice versa for its
negation. Hence, neither Frank nor Mary need to revise their beliefs or retract
their assertions.
2.3 The Unconvinced and the Nonplussed
Plausible as the nonindexicalist solution might appear at first blush, many
scholars remain unconvinced. Moltmann (2010), for instance, writes:
This has become the standard way to interpret definitions of disagreement by
nonindexcialists such as Kölbel (2004, p. 53-4) and Lasersohn (2005, p. 647). For discussion cf.
e.g. Stojanovic (2007, 2017) and MacFarlane (2014, Ch. 6.3 and 6.7).
The most important problem for the [nonindexicalist contextualist]
account is that it does not really explain faultless disagreement.
Competent speakers […] will know that they mean the utterance of such
a sentence to be true relative to their own context. The problem then
is, why on the [nonindexicalist] account can there be disagreement
among two speakers when the speakers know that the content of their
utterance can be both true, though relative to different contexts? If the
truth conditions of the sentence are clearly different relative to the two
speakers, then this should correspond to a difference in subject matter,
rather than to a single content about which there could be
disagreement.” (p. 194)
Others, such as Stojanovic, question whether faultless disagreement amounts
to a genuine puzzle of any consequence for the linguistic analysis of subjective
[On the assumption of semantic competence,
both speakers] know
that one and the same content may take different truth values when
evaluated at different judges [i.e. standards of taste]. They also know
that the one’s assertion and the other’s denial of the same content are
inconsistent only when evaluated with respect to the same judge.
Hence if each party intends the asserted content to be evaluated at
himself or herself, and if this is mutually clear between them, then they
will realize that there is no clash in truth value between their claims
(when evaluated as they intend them to be), and that their
“disagreement” is thus nothing more than a divergence in preferences.
(2007, p. 697)
The diverging reactions of those who feel the pull of the argument from
faultless disagreement and those who do not, I would like to suggest, is driven
by an undertheorized difference in conceptions of content. To explore the
matter, it is helpful to follow Recanati (2007) in differentiating the truth-
conditionally complete content of an utterance as distributed over two
aspects: the lekton (or, roughly, Kaplanian content) and the circumstance of
evaluation. The approach is neatly captured by two principles:
Both Stojanovic and Moltmann explicitly invoke semantic competence with respect to the
predicates at stake. Stojanovic defines it thus: “Speakers of English are semantically competent
with predicates of taste: they master their meaning and truth conditions.” (2007, p. 696).
: To get a truth-value, we need a circumstance of evaluation as
well as a content to evaluate. (As Austin puts it, “It takes two to make a
: The determinants of truth-value distribute over the two
basic components truth-evaluation involves: content and circumstance.
That is, a determinant of truth-value, e.g. a time, is either given as an
ingredient of content or as an aspect of the circumstance of evaluation.
(2007, pp. 33-34)
Recanati’s framework provides us with two notions of content: The truth-
conditionally complete Austinian proposition, distributed over lekton and
circumstance, as well as the explicit content or lekton itself. Once we have
multiple conceptions of content, our above invoked definition of disagreement
turns out ambiguous the doxastic noncotenability of contents could be
interpreted either as a noncotenability of Austinian propositions or of lekta:
Two individuals A and B disagree if the truth of some
Austinian Proposition believed or uttered by A precludes the truth of
some Austinian Proposition believed or uttered by B.
Two individuals A and B disagree if a lekton believed or
uttered by A and a lekton believed or uttered by B cannot both be true
with respect to any single circumstance of evaluation.
What notion of agreement is at play in the above quoted passages? Both
Moltmann and Stojanovic hold that competent speakers, in using taste-
predicates autocentrically, will evaluate them with respect to their own
perspectives, determined by their respective contexts of utterance. The
speakers, that is to say, keep track not only of those aspects of content
represented in the lekton, but also of those that – like the standard of taste on
a nonindexicalist distribution are part of the circumstances of evaluation.
Genuine disagreement, this position assumes, must arise with respect to
incompatible Austinian propositions, not with respect to lekta. And both
Moltmann and Stojanovic are correct that faultless disagreement cannot arise
on such premises.
Take our sample dialogue consisting of (3) and (4), and let’s suppose that both
interlocutors intend “delicious” to be interpreted autocentrically. If the
circumstances of evaluation include a world w, a time t, and a perspective p,
Austin (1971).
then, on a nonindexicalist view, the Austinian propositions (in square brackets)
uttered by Mary and Frank are:
(5) Mary: [Spinach is delicious. (w, t, Mary)]
(6) Frank: [Spinach is not delicious. (w, t, Frank)]
Semantically competent speakers, who correctly grasp the Austinian
propositions at stake, will notice that there is no substantive disagreement (or,
as Moltmann puts it, that there is a “difference in subject matter” across the
propositions uttered). If, by contrast, Frank evaluates the content explicitly
expressed by Mary’s utterance with respect to his perspective, he fails to grasp
the Austinian proposition she in fact expressed and falls prey to a
misunderstanding. In such a case, there might seem to be disagreement,
though even if there were, it would certainly not be faultless.
The fact that a moderate version of nonindexical contextualism, according to
which disagreement must be spelled out in terms of Austinian propositions,
cannot accommodate faultless disagreement should not be surprising, as the
position is truth-conditionally equivalent to indexical contextualism.
impressed by the phenomenon of faultless disagreement, this suggests, are
committed to a notion of disagreement in terms of lekta – and lekta only, that
is, disagreement in terms of lekta independently of the circumstance of
evaluation. More generally, they are committed to the view that the objects of
belief, assertion and disagreement are thin contents contents that can be
neutral with respect to standards of taste, and perhaps also with respect to
e.g. time and certain other parameters.!
The resulting view of content is radical in nature.
Evans famously scoffed at
the idea that a time-neutral sentence such as “Socrates is sitting” can express
Stojanovic (2007) provides a logical proof, which demonstrates that indexicalism and
moderate nonindexicalism are truth-conditionally equivalent. But by aid of Recanati’s
framework, the point can be made at an intuitive level: It doesn’t matter whether the standard
of taste features in the lekton or the circumstance of evaluation as long as the resulting
Austinian proposition, distributed over both aspects of content, remains the same.
Recanati, a moderate nonindexicalist, is explicit that such a view “by itself, does not
give a solution to the problem to the problem of faultless disagreement, contrary to what
Kölbel and Lasersohn believe.” (2007, p.91). On such an account, “alleged” faultless
disagreement might “arguably” arise (2007, p.94), because both interlocutors invoke a generic
taste parameter referring to the community’s standards, and they might disagree about what
those standards should be. This would of course reduce instances of faultless disagreement to
very few. Furthermore, it is not evident, as Recanati points out, whether the proposal
generalizes to other domains such as epistemic modals.
I follow Recanati (2007, Ch.2) in distinguishing “moderate” nonindexical contextualism from
“radical” nonindexical contextualism. Recanati himself calls nonindexical contextualism
a complete meaning” and considered it “such a strange position that it is
difficult to believe that anyone has ever held it” (1985).
To quickly illustrate
the ramifications of the thin content approach (we will return to it at length in
section 4), let us stick with time: Mary, at 6 am, says “Su is in bed [at home]
and Frank, at noon, says “Su is not in bed”. Though the lekta, by themselves,
appear (at least in some sense) contradictory, it is hard to fathom in what ways
this exchange manifests disagreement in any ordinary sense of the term, and
hence why one would want to endorse a thin content view.
2.4 Thin Content and Truth Relativism
Scholars who do not feel the force of the puzzle of faultless disagreement
sometimes extend their critique from nonindexical contextualist positions such
as Kölbel’s and Lasersohn’s to assessment-relative views like MacFarlane’s.
In an article taking stock of the controversy relating to faultless disagreement,
Stojanovic (2017), for instance, writes:
While there are genuine formal differences between the simpler
[nonindexicalist] framework and MacFarlane's, that does not make the
latter better suited to account for the puzzle [of faultless disagreement].
For assume Kathy and Rob to be competent speakers; they must be
aware, then, that their claims can only be evaluated for truth with
respect to a context of assessment. If Kathy intends her claim to be
evaluated with respect to her own context of assessment, and Rob
intends his denial of Kathy's claim to be evaluated with respect to his
own context, and if this is mutually clear between them, then we have
hardly made any progress towards an explanation of their presumed
disagreement. (2017, pp. 10-11)
“relativism”. I reserve the latter term for MacFarlane’s position, as has become commonplace
in the literature.
Does anyone actually propose such a picture as regards taste-neutral sentences? I think there
is clear evidence that a view like this is advocated by authors such as Kölbel (2004b), Richard
(2004, 2008, 2011) and Lasersohn (2005, 2009) to name but a few.
Perhaps nonindexicalists could argue that time-neutral contents behave in important
respects differently from taste-neutral contents respects that explain why it might strike us as
intuitively implausible in the former, but not in the latter case, to attribute disagreement.
However, in justifying parameter proliferation, much of the efforts of nonindexicalists have
focused on emphasizig the similiarty between the “new parameters” (standard of taste,
epistemic perspective etc.) and more traditional ones such as worlds and times (see e.g. Kölbel,
Note that MacFarlane himself is sceptical of the notion of faultless disagreement (2014, pp.
This might be a little too quick. Just as nonindexical contextualists, truth
relativists such as MacFarlane argue that the extension of taste claims
depends, inter alia, on a taste parameter to which the taste predicate must be
relativized. However, the taste parameter is not determined by the context of
utterance, but by the context of assessment. There are infinitely many contexts
of assessment, so the interpretation of a taste claim is not tied to a single,
determinate context of utterance as on the contextualist view. Let’s assume,
with MacFarlane, that the expression “delicious” is indeed assessment-
sensitive. Then semantic competence with the predicate would entail that
speakers are aware that the extension of a tokened taste claim can change
from context of assessment to context of assessment. If Frank, in our sample
dialogue, evaluates Mary’s utterance with respect to a circumstance
determined by his context of assessment, he would thus be beyond reproach.
Once we take assessment-sensitivity seriously, it is of no help to insist, as
Stojanovic does, that Mary must intend her claim to be evaluated with respect
to her context (that is, her context of utterance), since this presupposes a
single, privileged context – precisely the feature the relativist does away with.
Here’s an alternative way to make the point: In order to introduce his view,
MacFarlane (2014, pp. 62-64) discusses the example of a fictitious expression
noy”, which functions similarly to the ordinary language indexical now. The
difference between the two is that the extension of claims invoking now
depends on a time fixed by the context of utterance, whereas the extension of
noyclaims depends on a time initialized by the context of assessment. If
Mary says “It’s raining noy” at t1 (when it is raining) and Frank assesses her
utterance as false at t2, when it is no longer raining, it is not appropriate for
Mary to complain that she intended her claim to be relative to the context of
utterance. What a complaint of this sort reveals is simply that Mary does not
have the stipulated semantic competence, since she fails to grasp that the
expression “noy” is assessment-sensitive.
3. Accounting for Disagreement
Moderate nonindexical contextualism conceives of the objects of
disagreement as truth-conditionally complete contents, distributed over lekton
and circumstance. Hence, two individuals A and B disagree iff the truth of some
Austinian Proposition believed or asserted by A precludes the truth of some
Austinian Proposition believed or asserted by B. Naturally, if the proposition
that spinach is delicious is true relative to Mary’s standard of taste, but false
with respect to Frank’s, the interlocutors talk past each other: The value of the
perspective parameter is a constituent of the Austinian proposition, hence two
different propositions are at stake, or as John Perry (1986) would have it, the
two propositions concern different standards of taste.
Given contextualist premises (i.e. utterance-sensitivity), the sceptical reactions
vis-à-vis the argument from faultless disagreement discussed above are on the
right track. Paired with a substantive notion of content, faultless disagreement
can simply not arise – no matter whether the distribution accords with
nonindexicalist or indexicalist proposals. This leaves three options to get the
puzzle off the ground:
(1) Truth relativism gives up utterance-sensitivity in favour of
assessment-sensitivity, and can, at least in principle, retain a reasonably
rich view of content.
(2) Moderate nonindexical contextualists acknowledge that they do not
have the resources to account for faultless disagreement in terms of
DisagreementAP as demonstrated above. However, they might argue, in
contrast to indexicalists, they can account for the appearance of
faultless disagreement, and that this is, at root, the phenomenon to be
(3) Radical nonindexical contextualism refuses to retreat to appearances
yet holds fast to utterance-sensitivity. In order to make room for
disagreement, one must thus assume that two individuals can disagree
in terms of taste-neutral lekta, invoking the definition of DisagreementL
from above.
In this paper, the focus lies on the third option, and the more general
plausibility of thin content views, which have received comparatively little
attention. But before we get started, I’ll briefly state why I consider it apt to set
the first two options aside.
3.1 Truth Relativism
Relativism about truth in English, as regards predicates of personal taste, holds
that the ordinary English meaning of expressions such as tasty, “delicious”
A proposition, Perry suggests, is about some feature F, if F is one of its propositional
constituents (articulated or not). For instance, according to eternalists like Frege and Evans,
propositions or “thoughts” always include a temporal specification, even if only tacitly so, and
are thus always about particular times. Alternatively, a proposition can be said to concern a
feature F, if its truth value depends on how things stand as regards F. That’s how a temporalist
understands tensed propositions: Their content is standardly time-neutral, but they concern a
particular time. If Mary utters the time-neutral sentence Socrates is sitting” at midday, it
concerns that specific time since its truth must be evaluated with respect to the world and the
time determined by the context of utterance.
or fun are in fact assessment-sensitive. This is an empirical claim
(MacFarlane, 2014, p. 65). If this empirical claim were true, the relativist would,
as explained in the previous section, have the conceptual resources to account
for faultless disagreement. In this paper, I will not engage in detail with this
view for two reasons. First, I doubt its empirical adequacy. Findings from
experimental linguistics suggest that ordinary language speakers evaluate the
extension of taste claims (Kneer, 2015 Ch.7, in prep. a) and epistemic modals
(Knobe and Yalcin 2014, Marques ms, Kneer, 2015 Ch. 6, in prep. b) with
respect to the context of utterance, not the context of assessment.
whereas faultless disagreement constitutes the central motivation for
nonindexical contextualism, MacFarlane is explicit that the phenomenon “is
not needed for motivating or explaining truth relativism” (2014: 136, cf. also
Chapter 6 more generally).
3.2 The Appearance of Faultless Disagreement
Moderate nonindexcialists potentially have an advantage over indexicalism,
arising from their favoured distribution whereby standards of taste are
anchored in the circumstance. Even though they might be incapable of
accounting for faultless disagreement if disagreement is understood in terms
of Austinian propositions, their options are not exhausted by the radical
alternative notion of lekta disagreement. In contrast to the indexicalist,
nonindexicalists have a story to tell as regards the appearance of faultless
disagreement: there is genuine faultlessness as witnessed at the level of
Austinian propositions, while apparent denial (“No! Liquorice is not tasty.”) or
rejection (“That’s wrong! Liquorice is not tasty.”), and hence at least apparent
disagreement, is accounted for at the level of lekta. The fact that the lekta are
conceived as perspective-neutral ensures that they, at least as long as the
circumstances of evaluation are disregarded, contradict one another. The
indexicalist, by contrast, cannot capitalize on this neat division of labour. Once
standards of taste are part of the content, and relativized to distinct speakers,
it is mysterious how, and why, the appearance of disagreement might arise in
the first place.
In a recent paper, Stojanovic (2017) surveys strategies to account for
appearances of faultless disagreement that do not require a nonindexicalist
framework. Plenty of options are available, drawing on contextual
Whether or not it is empirically adequate, the assessment-sensitive framework remains, of
course, philosophically coherent. But if the meaning of perspectival expressions is in fact not
assessment-sensitive, as the data suggests, then its interest is limited, as it will be devoid of
underdetermination, metasemantic and metalinguistic observations,
presuppositions and disagreement in attitude.
Stojanovic (2007) herself argues that, often times, contextual
underdetermination obscures whether the speaker intends their taste claim to
be understood autocentrically or generically.
In the former case it would
constitute a subjective claim (thus warranting faultlessness), in the latter it
would constitute a garden-variety factual claim about the community’s
standard of taste, warranting genuine disagreement, though not faultlessness.
Appearances of faultless disagreement can arise when it is opaque what, in
fact, the propositions uttered are. Such appearances vanish once the
propositions are further elucidated (“Well, liquorice is tasty for me.”). Another
strategy draws on the observation that predicates of personal taste are
gradable adjectives (Glanzberg, 2007), and that the interpretation of taste
claims thus involves tacit scales and thresholds.
Even if both speakers
envision a joint standard of taste, they might operate with divergent scales and
thresholds. But if they do (thus warranting faultlessness), the appearance of
disagreement might arise nonetheless: subjective discourse frequently triggers
a “presupposition of commonality”, i.e. a presupposition held by all parties
involved that they share a common perspective as to what is tasty, funny or
beautiful (López de Sa, 2008, 2015, cf. also Marques & Garcia-Carpintero,
Sundell (2013) argues that disputes about personal taste, just like disputes
about morality (Plunkett & Sundell, 2013) or aesthetics (Sundell 2017), are
frequently characterised not by disagreement in descriptive content, but by
metalinguistic disagreement. As with non-evaluative gradable adjectives (see
Barker, 2002), situations can arise where the interlocutors disagree less about
whether an object instantiates a particular property, but what the context-
dependent standards for such property ascriptions are or should be. The
appearance of disagreement with respect to property instantiation feeds on
genuine metalinguistc disagreement regarding the scales and thresholds that
govern (or should govern) the application of predicates like “delicious”,
“beautiful” or “wrong” at the determinate context (for detailed discussion, see
Kneer, 2015 Ch.4).
See also Moltmann (2010) and Pearson (2012).
“Tasty” and “fun”, as Glanzberg observes, are more complex than off-the-shelf gradable
adjectives such as “rich” or “tall”, in so far as they can draw on more than a single scale.
However, he argues, “this is not a feature specific to adjectives of personal taste. Many
gradable adjectives can be associated with multiple scales. For instance, someone can be smart
as in ‘book smart’ or ‘street smart’, a large city can be large in population, geography, etc.”
(2007: 10).
Finally, certain philosophers trace the appearance of faultless disagreement
not to disagreement in acceptances, but in attitude (Huvenes, 2012; Marques,
2014, 2015). The fixation on the semantics of taste claims, on this approach, is
somewhat beyond the point. If I say “I like spinach”, and you respond with “I
dislike it”, we have a clash in attitudes, a clash which might frustrate joint dinner
plans, and which might thus constitute disagreement in some sense. Overall,
it is evident that indexicalists have a broad set of option to explain why the
appearance of faultless disagreement in subjective discourse might arise.
3.3 Taking Stock
In section 2, we saw that moderate nonindexical contextualism does not have
the resources to account for faultless disagreement in a substantive sense.
Tellingly, major advocates of moderate nonindexical contextualism, including
Kölbel (2009), have retreated to talk about the mere appearance of faultless
disagreement, and argued that indexical contextualists cannot make sense of
As discussed, however, the options of indexicalists to account for the
latter are plentiful. All this said, a question arises as to how much mileage one
can get out of debates about appearances.
It would be somewhat excessive
to propose far-reaching revisions to truth-conditional semantics on the
grounds that claims of personal taste, under certain circumstances, generate
an impression (and an impression only) of faultless disagreement.
Nonindexicalists might consequently be well advised to opt for a bolder move:
they could cast disagreement in terms of lekta and thus hold fast to a notion
of faultless disagreement that goes beyond appearances only.
According to radical nonindexical contextualism, thin contents or lekta can
constitute the objects of propositional attitudes or illocutionary acts. The view
is radical because disagreement can arise between two speakers asserting
contradictory taste-neutral lekta, in complete disregard of the circumstances
of evaluation determined by the relevant contexts of tokening. In Perry’s terms:
The taste-neutral contents p and q asserted or believed by two individuals can
constitute disagreement even though they concern distinct perspectives. This
Examining situations of apparent disagreement over taste (such as (3) and (4) above), Kölbel
writes that linguistic evidence of this sort “is not meant to consist in the purported fact that
these cases do indeed involve both faultlessness and disagreement in some pre-theoretical
sense. Rather the evidence at best consists in the fact that there appears to be faultless
disagreement.” (2009, p.389).
Beillard (2010), who devotes an entire article to this phenomenon contends that “the
appearance [of faultless disagreement] is possible only under conditions that disqualify it as
evidence: gross ignorance or irrationality, or else a prior commitment to an especially crude
and implausible form of relativism.” (2010, p. 603).
approach, which is not restricted to standards of taste but can extend to other
features that are candidates for parameters in the circumstance of evaluation,
is largely undertheorized. In the following, we will examine an argument
against it the incompleteness worry and a recent response to the latter
proposed by MacFarlane (2014).
4. The Incompleteness Worry and MacFarlane’s Response
4.1 The Incompleteness Worry
Given that a moderate version of nonindexical contextualism can at best
account for faultless misunderstanding, and given that an argument from the
appearance of faultless disagreement has little bite, the radical strategy might
hold most promise for advocates of nonindexicalist contextualism. However,
the thin content picture this strategy invokes is contentious. The worry is this:
If disagreement consists in incompatible lekta, and if we assume the objects of
assertion, belief and disagreement to be the same kinds of entity, then doubts
arise whether propositions or contents so conceived might not rather be too
limited to fulfil their role in propositional attitudes and illocutionary acts.
Differently put, given the outsourcing of relevant aspects of the full truth-
conditional meaning from the content into the circumstance, one might
wonder whether the impoverished explicit content, the lekton by itself, is still
sufficiently fine-grained to explain our attitudes and the actions they drive. If,
for instance, all we know is that Sam thinks it’s raining yet are in the dark as
regards the location at stake, it is reasonable to think that we are in no position
to know what he believes, to predict how he will act, or to explain why he acts
as he does. Similarly, if we don’t know with regards to which standard of taste
to evaluate Mary’s assertion that spinach is delicious, we cannot say what she
believes vis-à-vis the culinary features of spinach, or with regards to whose
tastes we must understand and evaluate her utterance. What to prepare for
dinner? Cappelen & Hawthorne, for instance, write:
There is something of a strain in accepting that each such thin semantic
value cuts the space of possibility into the worlds where it is true and
the worlds where it is not, grounded in felt uneasiness at answering very
simple questions about what it would take for a thin semantic value to
be true. (For example, would Jill is ready be true at a world where she
was ready to play golf, but not ready to get married? [...]) It is immensely
tempting to deny that these kinds of objects reach the level of
propositionality. (2009)
MacFarlane summarizes the worry thus:
One might try to cash out an “incompleteness” worry in the following
way. Propositions are supposed to be the contents of beliefs and other
propositional attitudes. But if we specify the content of someone’s
belief in a way that does not settle what is relevant to the accuracy of
the belief, we have not given its complete content. […] A location-
neutral, time-neutral, or taste-neutral content would only incompletely
determine the conditions for an attitude to be accurate, and so could
not be the complete content of the attitude. (2014, p. 86)
4.2 The Argument from Modal Anxiety
According to MacFarlane, the worry overgenerates. Since his, to my
knowledge, is the only serious attempt to address the problem incompleteness
poses for radical nonindexical contextualism directly,
I will recite his
argument in full, which picks up from the passage just quoted:
[The above] line of thought proves too much. For surely the accuracy of
any contingent belief depends on features of the world in which the
believer is situated – the world of the context of use. Even if we specify
the content of Sam’s belief in a way that builds in time and place – that
it is 0° C at the base of the Eiffel Tower at noon local time on February
22, 2005 it is still not determined whether the accuracy of his belief
depends on the temperature in Paris in world w1 or on the temperature
in Paris in world w2. To know that, we would have to know not just what
Sam believes the content of his belief but in what context, and in
particular in what world, the belief occurs.
One might respond to these considerations by bringing the world of
the context of use into the content of Sam’s thought, so that what he
thinks is that it is 0 C at the base of the Eiffel Tower at noon local time
on February 22, 2005, in this world (Schaffer 2012). […]
[However], bringing the world of the context into the content of Sam’s
thought would make this content a necessary truth about this possible
Another debate fuelled by concerns of incompleteness is the one surrounding “unarticulated
constituents” e.g. in weather reports. The debate differs from the PPT debate in many regards
and I will set it aside in this paper.
At the risk of repetition: Although MacFarlane is one of the few authors who engages with
the incompleteness worry, I do not think that the latter constitutes a challenge for his view, that
is, relativism (see section 2.4).
world, rather than a contingent truth about the weather in Paris. We
should not say, then, that Sam’s thought is about the world of the
context of use. It is not about any particular world. (2014, pp. 86-87)
Assume with MacFarlane that relativity as regards the world parameter is not
special in any way, i.e. it has exactly the same general features as more exotic
parameters such as perspectives, locations or standards of precision.
should pay no heed to the incompleteness worry, the suggestion is, for doing
so gives rise to considerable complications pertaining to modal logic: As soon
as we build the world into the content – for instance by aid of a hidden actuality
operator or a tacit demonstrative reference to the actual world – a true
contingent claim becomes a necessary truth. We will call this argument the
argument from modal anxiety.
Note the limited scope of the argument: It does nothing to explain how thin
propositions could fulfil their role as objects of belief and assertion. As such it
cannot dispel the incompleteness worry. Instead, the argument presents a
dilemma for those impressed by faultless disagreement: Either bite the bullet
as regards incomplete propositions or run into trouble with regards to modal
logic. But it is not obvious that incomplete propositions are the lesser evil.
Facing such a trade-off, we might much rather want to sacrifice the resources
to account for faultless disagreement instead. In contrast to modal
complications and semantic incompleteness, the latter is a comparatively
unimportant phenomenon, if it rises above appearances at all. Hence, what is
presented as an argument against perspective-specific contents, is perhaps
best understood as an argument in favour of moderate nonindexical
contextualism. On this view, perspectives, times, worlds etc. are all safely
outsourced into the circumstance: There, they cannot wreak modal havoc, yet
the objects of belief and assertion if conceived as Austinian propositions
are complete because they consist of lekton and circumstance jointly.
Naturally, as argued by Stojanovic and Moltman, this view presumably cannot
explain substantive faultless disagreement. While the force of the argument’s
conclusion is thus limited, I also have doubts about whether its premises are
sound. To these doubts we turn next.
As such we explicitly refrain from attempting to block the argument in ways familiar from
Evans (1985). Evans argues that the world parameter is special because there is a unique default
value the actual world, whereas there is no such default value for time and other parameters.
In a similar vein, Cappelen and Hawthorne (2009) emphasize that the actual world is “the only
reality there is” (2009, p. 78) and propose a picture according to which world information is
specified neither in the propositional content nor the circumstance of evaluation.
Recanati (2007), who calls this position Strong Moderate Relativism, defends it convincingly
against incompleteness and related worries.
5. The Argument Reconstructed
Let’s look at the argument step by step.
(P1) If time- or location-neutral propositions expressed by utterances such as
“It’s raining” are semantically incomplete, so are world-neutral propositions
like “Paris is the capital of France in 2014”. Differently put: The incompleteness
worry concerns all parameters alike.
(P2) If the objects of assertion and belief must be complete propositions, they
must be world-specific propositions, or propositions about worlds in Perry’s
(1986) sense. Sentences expressing a complete proposition must make
mention of a particular world either explicitly or implicitly. When no world is
explicitly stipulated, a hidden world argument draws a salient value from the
context of utterance. Standardly, the world provided by the context is the
actual world, i.e. the world at which the sentence is uttered. For instance
(ignoring time), “Paris is the capital of France” expresses the proposition “Paris
is the capital of France [in this world]” or “[Actually], Paris is the capital of
France, where the modal operator “actually” (in the following: A) sets the
parameter for the world of evaluation to the world of utterance.
Hence: Any
tokened sentence S apparently expressing a world-neutral proposition P, in
fact standardly expresses a modally complete proposition about the actual
world, AP.
(P3) A sentence which tacitly features the “actually” operator (or an instance of
“in this world”) expresses an actualized proposition. An actualized proposition,
if true, is true necessarily. No matter at which world it is evaluated, it must
always be assessed with regards to the world actual at the context of utterance.
Let N stand for the modal operator “necessarily”, such that: AP
(P4) Given (P2) and (P3): An assertion (or other tokening) of a sentence that
expresses the proposition P in fact always expresses AP, which is equivalent to
NAP. Hence, for any tokened sentence expressing P, P
(C) Since all asserted propositions must be world-specific on pain of
incompleteness, those sentences which do not explicitly specify a world in the
An intuitive grasp of the “actually” operator suffices for our purposes. For discussion of the
operator’s behaviour in propositional modal logic, cf. Crossley and Humberstone (1977),
Gregory (2001) and Blackburn and Marx (2002). Gregory (2001, p. 61ff) is particularly pertinent
for our premises P3 and P4. For “actually” in first-order modal logic based on S5 cf. Hodes
(1984), for a more general first-order modal logic treatment see Stephanou (2005).
lekton must be conceived as carrying an implicit actuality operator (P1). Given
(P4), all sentences, once tokened, express actualized propositions which are
true necessarily if true at all, i.e. P
NAP. As regards Sam’s thought about the
weather in Paris, MacFarlane concludes, the procedure of “bringing the world
of the context into the content of Sam’s thought would make this content a
necessary truth about this possible world, rather than a contingent truth about
the weather in Paris” (quoted above). This, the modal moral is supposed to be,
is deeply counterintuitive.
6. Modal Anxiety and Two Types of Necessity
6.1 Modal Anxiety
What drives MacFarlane’s argument from modal anxiety is, I suspect, an
intuition characteristic of early reactions to Kripke’s (1972) contingent a priori
and necessary a posteriori. Statements of this sort arise as a consequence of
rigid designation, a feature in virtue of which certain expressions such as
proper names or natural kinds designate the same individuals in all possible
worlds. As will be shown below, the privileged role the actual world plays in
determining the extension of such expressions and the prima facie paradoxical
statements it engenders, is exactly what is at work as regards the alleged
necessity of contingent propositions when actualized. The point, however, can
be made at an intuitive level. Suppose the content P of Sam’s thought carries
an implicit reference to the actual world, such that P is “It is C at the base
of the Eiffel Tower at noon local time on February 22, 2005 [at the actual
world].” As MacFarlane highlights, P, if true, is true necessarily. But the
necessity at stake need not clash with our intuition that the content of Sam’s
thought is as contingent as they come. Though actualized propositions are
always necessary (that is, necessarily true or necessarily false), whether they are
true or false in the first place depends on contingent features of the one
particular world at which they are tokened. Differently put: Even if the content
of Sam’s thought happens to be true, and is thus true necessarily, it still holds
good that if it had not been the case that it was 0° C at the base of the Eiffel
Tower on February 22, 2005, the content of Sam’s thought would have been
6.2 Two Types of Necessity and Contingency
The above considerations suggest that there are two different kinds of
necessity (and, correspondingly, two types of contingency) – a proposal which
is hardly new. Following Evans (1979), whose distinction is more fully
elaborated by Davies and Humberstone (1980), we’ll label them “superficial
necessity” and “deep necessity”:
: A sentence or content p is superficially necessary iff p is true
in all possible worlds.
: A sentence or content p is deeply necessary iff p is (actually)
true no matter which possible world is actual.
The actualized and hence world-specific proposition entertained by Sam is
superficially necessary yet deeply contingent. Though epistemic matters are of
no particular concern as regards MacFarlane’s example, it is helpful to discuss
his response to the incompleteness worry in the context of the contingent a
priori, and, in particular, the necessary a posteriori. By aid of the distinction
between the two types of necessity, we can dispel modal anxiety in similar
ways as Evans and his followers countered the widespread contention that
Kripke cases “constitute an intolerable paradox” (Evans, 1979, p. 161).
The content of Sam’s thought is necessarily true in the superficial sense if true,
yet the assessment whether it is true in the first place is a matter of empirical
inquiry. Sam’s thought can thus be seen as an instance of the necessary a
the perplexing epistemic status of which is frequently considered
a direct consequence of the deep contingency which Sam’s thought intuitively
manifests. The case bears considerable likeness to classic examples of the
necessary a posteriori, for instance “scientific identities” like Water = H2O”.
The similarity is even more obvious as regards (deeply) contingent
propositions invoking an “actually” operator, i.e. propositions that constitute
a “fund of simple examples of the necessary a posteriori” as Davies and
Humberstone (1980, p. 10) point out. Given that grass is green in the actual
world, “Grass is actually green” is true in all possible worlds and hence
superficially necessary. Still, it could have been the case that grass was orange,
hence “Grass is actually greenis not true no matter which world is considered
actual – it is deeply contingent.
The same holds for scientific identity statements. If Water = H2O” is in fact
true, it is true in all possible worlds, since expressions denoting natural kinds
designate rigidly. However, the chemical composition of water can only be
determined by means of empirical enquiry, it is known a posteriori. And it is a
The formulations are borrowed, with slight modification, from Hanson (2006, p. 448).
Evans is principally concerned with the contingent a priori, but the strategy carries over to
the necessary a posteriori (cf. Davies & Humberstone, 1980), which is our primary focus.
Not an unusual move, see Davies and Humberstone (1980) as well as (Davies 2004).
posteriori in virtue of its deep contingency. If it had been the case that water
was XYZ, Water = H2O” would have been false. In fact, the expression “water
can be understood as involving a tacit reference to the actual world. Putnam,
in certain moods, describes it as involving such an indexical element, and
Davies and Humberstone suggest to conceive of water” as a descriptive
name (a name whose reference is fixed by description) featuring an implicit
“actually” operator. On this proposal, wateris short for “the actual watery
stuff hereaboutsand we have effectively the same sort of case as the one
MacFarlane is worried about.
What exactly are deep necessity and contingency, and how do they differ from
superficial necessity and contingency? A sentence S, for Evans,
superficial contingency, iff there is a world in which S is false, that is, if neither
Snor ¬Sare true (where the box symbolizes the necessity operator).
Contingency in this sense is a property of a sentence which depends upon
how it embeds inside the scope of modal operators” (Evans 1979, p. 179). By
contrast, deep contingency is introduced not with respect to a sentence’s
behaviour when embedded under standard modal operators, but with regards
to what makes it true”: “If a deeply contingent statement is true, there will
exist some state of affairs of which we can say both that had it not existed the
statement would not have been true, and that it might not have existed” (Evans
1979, p. 185). Conversely, a statement is deeply necessary if it is true
independently of which world turns out actual and hence cannot be falsified
by contingent features of reality.
What is captured by superficial necessity is a property of modal sentences
sentences, that is, which invoke, tacitly or explicitly, some reference to some
particular world. A sentence S and its actualized version AS can come apart in
terms of superficial necessity, because necessity in this sense is responsive to
the modal features of the sentence, in this case the “actually” operator. Deep
necessity, on the other hand, captures not a property of modal sentences, but
a modal property of sentences.
Necessity or contingency regarding a
sentence S and its actualized version AS do not come apart, since necessity in
this sense is unresponsive to the modal element in “AS. In the deep sense, S
and AS are both contingent if dependent on which world happens to turn out
actual, or else both necessary in so far as they hold no matter which world
happens to turn out actual.
Evans’ analysis proceeds in terms of sentences, rather than propositions or contents. We will
follow Kment (2017, p. 2) in making the common assumption that [a] sentence is necessary
(possible, contingent) just in case it expresses a necessary (possible, contingent) proposition.”
Cf. Davies and Humberstone (1980), as well as Davies (2004).
Tacitly world-specific contingent propositions such as “Grass is green [in this
world]” or “Grass is [actually] green”, I suggested, raise as much of a paradox
as necessary a posteriori statements do: None whatsoever. They are necessary
in a superficial sense, that is, true at all possible worlds only in virtue of the
modal element in the content. They are not necessary in a deep sense: had
another world turned out actual, “Grass is [actually] green” would have been
false. The truth of such world-specific propositions is thus just as dependent
on features of contingent reality as the truth of their world-neutral equivalents.
Consequently, the argument from modal anxiety is not a convincing response
to the incompleteness worry. Pace MacFarlane, and in line with Davies,
superficial necessity must be understood as a largely innocent feature of modal
sentences. What matters is that actualized propositions, albeit superficially
necessary, remain deeply contingent.
6.3 Worlds and Times
According to the argument from modal anxiety, all propositions, in virtue of
their being world-specific, turn out necessary. This, MacFarlane argues, is
unwelcome because they are intuitively contingent. However, it does seem
reasonable to assume that necessity and contingency in the intuitive sense
regard modal properties of sentences or contents, not properties of modal (i.e.
world-specific) sentences or contents. Therefore, I have argued, world-specific
propositions remain contingent, in the intuitive – that is, the deep sense, and
there is nothing to worry about. The argument can be further illustrated by aid
of a parallel with time. In his defence of thin contents, MacFarlane appeals to
what Schaffer calls “the most straightforward argument for Contingentism”,
i.e. the feeling “that certain claims are just evidently contingent” (p.143).
Eternalists are familiar with a parallel objection. Claims such as (7) below, the
argument goes, just evidently, or intuitively, capture a transient truth. But on
an eternalist account, they must be classified as true eternally. Differently put,
on an eternalist view or so the argument goes one loses the intuitive
distinction between propositions that are true with respect to all times and
those that are not. Consider:
(7) Merkel is the chancellor of Germany.
Now, eternalists standardly conceive of temporal operators as object-level
On this view, “eternally” constitutes a universal quantifier over
This section draws heavily on Schaffer (2012, section 3.1). See also Schaffer (2018).
Richard (1981) and Salmon (2003) offer an alternative approach: Rather than invoking object-
language quantification they use intensional operators which manipulate semantic values that
do not amount to full-fledged propositions. For discussion cf. Schaffer (2012, p. 131, N. 19).
times, which binds a free time variable t, such that (8), an eternalized version
of (7), must be interpreted as (9):
(8) Eternally, Merkel is the chancellor of Germany.
(9) (t) Merkel is the chancellor of Germany at t.
Given that it is not the case that Merkel is the chancellor of Germany at every
time t, (8) turns out false on the eternalist view, just as it should. The situation
is the same for worlds. Take (10) which, on a quantificational treatment is
understood as (11):
(10) Necessarily, Merkel is the chancellor of Germany.
(11) (w) Merkel is the chancellor of Germany at w.
Merkel is quite clearly not the chancellor of Germany at every possible world
w. Hence, (10) is false, just as it should be. Differently put, the intuitive, that is,
the deep contingency and by extension, the deep transience of
propositions expressed by sentences like “Merkel is the chancellor of
Germany” or “It is 0° C at the base of the Eiffel Tower at noon local time on
February 22, 2005” is preserved.
A potential response on MacFarlane’s behalf could go thus: Conceiving of
propositions as world-specific might not make them deeply necessary. Still, a
semantic view according to which all propositions come out as necessary in
the superficial sense – whether this is the intuitive sense or not is implausible,
too. For a response along these lines to convince, however, it would be helpful
to have an argument why superficial necessity is worrisome. MacFarlane
himself does not provide such an argument. In the following sections I will
examine a few considerations why it might be hard to come by.
7. Parallels between Eternalism and Necessitarianism
7.1 Times and Worlds
As Schaffer (2012) observes, positions analogous to eternalism and
temporalism can be construed for worlds, which he labels necessitarianism and
: For every proposition p, and every bit of time information it
needed for truth evaluation, it is specified in p.
For some proposition p, and some bit of world
information it needed for truth evaluation, it is unspecified in p
(equivalently: p is neutral with respect to it). (2012, p.126)
: For every proposition p, and every bit of world
information iw needed for truth evaluation, iw is specified in p.
For some proposition p, and some bit of world
information iw needed for truth evaluation, iw is unspecified in p
(equivalently: p is neutral with respect to iw). (2012: p. 128)
Schaffer proceeds to argue in favour of Parallelism, that is, the thesis that the
whole case for eternalism can be mimicked by necessitarians: (i) Analogies
between pronouns and tense originally discussed by Partee (1973) carry over
to pronouns and worlds (Stone 1997, Schaffer 2012, section 2.1); (ii)
Complications regarding multiple time-indexing (Kamp 1971, Vlach 1973, van
Benthem 1977, Cresswell 1990) similarly arise with respect to the world
parameter (Cresswell 1990, Schaffer 2012, section 2.2). Finally, (iii) Richard’s
(1981) well-known anti-temporalist argument from belief retention can be
adapted into an anti-contingentist version (Schaffer, 2012, section 2.3).
Assume that Schaffer is right, and that the major arguments proposed in favour
of eternalism carry over neatly to the modal domain. What does this mean for
MacFarlane’s argument, according to which a rejection of thin contents
engenders a view of propositions that are uniformly superficially necessary?
The argument, it seems, loses the little bite it had left. Eternalism, a position
that dates back to Frege and is held inter alia by Stalnaker (1970), Richard
(1981,1982), Salmon (1986), Stanley (1997a, 1997b) and King (2003), is a
respectable view of propositional content, if not the current orthodoxy, as
temporalists like Brogaard (2012: 5) acknowledge. But if eternalism is the
orthodoxy, and if the case for necessitarianism runs parallel, the question arises
why a commitment to necessitarianism should constitute a problem. Quite to
the contrary – it seems much more plausible a position than contingentism. In
short, the alleged threat of necessitarianism does not constitute an effective
response to the incompleteness worry or a convincing argument in favour of a
position as revisionary as radical nonindexical contextualism.
Note that eternalism and necessitarianism are demanding views: For them to be correct, all
propositions must be time-specific or world-specific. Temporalism and contingentism, by
contrast, are comparatively undemanding: For them to be correct, it suffices for there to be a
single proposition that is time-neutral or world-neutral respectively. This point will be of
importance in the next section.
7.2 The Inevitability of Superficially Necessary Propositions
Those inconvenienced by modal anxiety will presumably have to put up with
the phenomenon no matter what. Take actualized sentences such as (12), or
claims that explicitly invoke a determinate world such as the (slightly adapted)
utterance of Roberto Benigni’s character in Jim Jarmusch’s film Down by Law
(12) Merkel is actually the chancellor of Germany.
(13) This is a sad and beautiful world.
Suppose (12) and (13) were tokened in world w1 such that the indexical
expression “actually” or the complex demonstrative “this world” draw w1 from
the context of utterance and hence set the world parameter in the
circumstance to that very value. Said value remains fixed across possible
worlds: On standard assumptions, indexicals and demonstratives are rigid
designators, they denote the same entity across all possible worlds. Hence,
even if we were to evaluate the propositions expressed by (12) or (13) in worlds
that differ in the relevant respects from w1, their truth must still be assessed
with respect to w1. If (12) and (13) are true in the world of tokening, w1, they
are true necessarily in the superficial sense.
Now imagine a hardnosed contingentist: Though she knows that (12) was
tokened in w1 (and is thus in principle w1-specific), she evaluates the
proposition expressed with respect to her actual world w2. In such a case she
would not only change the Austinian Proposition at issue, but the lekton itself.
Differently put, to force superficial contingency back into the picture, one
would have to manipulate the content (in both of Recanati’s senses of content)
from world to world, which seems unacceptable. What this suggests is that the
radical nonindexicalist, too, has to put up with at least some superficially
necessary contents – namely those that make explicit mention of a particular
world in the lekton.
But if superficial necessity is already part of the object
language, it is not evident why it is deemed to troublesome in the first place,
in particular given that superficial necessity does not entail deep necessity.
8. Further Arguments in Favour of Thin Contents
The previous section concludes my response to MacFarlane’s argument from
modal anxiety which, to my knowledge, is the only one that directly addresses
the incompleteness worry directly. There are, however, other arguments in
One possibility is to conceive of “actually” as assessment-sensitive (a view nobody, to my
knowledge, holds). Just like the assessment-sensitivity of predicates of personal taste and
epistemic modals, this would constitute an empirical hypothesis about ordinary English.
general support of thin content views, i.e. views according to which it is
preferable to account for a particular type of feature (worlds, times, tastes, etc.)
by aid of a parameter in the circumstance rather than by postulating tacit
indexical elements in the lekton. I will briefly examine some of these arguments
with regards to personal taste, our original point of departure.
8.1 The Operator Argument
Kaplan famously argued from the existence of operators such as “necessarily”
or “always” to the plausibility of a thin content view (and thus a default
distribution according to which worlds and times are parameters in the
circumstance of evaluation). Sentential operators shift the world or time with
respect to which a content must be evaluated. But if contents were always
world- or time-specific, he suggests, such operators would have no linguistic
function (Kaplan, 1989: 503).
Kölbel (2009) tailors the argument to the domain of personal taste. In ordinary
English, he contents, “for t, p” (where t designates a person and p stands for
a claim of personal taste) shifts the standard of taste with respect to which the
content is to be evaluated.
The FOR operator is supposed to work in similar
ways as modal operators. For instance, it renders otherwise contradictory
utterances such as (14) and (15) felicitous, and it interacts in analogous ways
with quantifiers:
(14) In possible world W, whales are extinct, but whales are not extinct.
(15) For Anna, whale meat is tasty, but whale meat is not tasty. (2009, p. 384)
(16) For some people, Picasso is better than Matisse.
(17) In some possible worlds, the British Empire outlasts the Soviet Union.
(2009: 385).
Persuasive as far as it goes, Kölbel’s proposal runs into trouble with utterances
containing multiple predicates of personal taste, as argued by Cappelen and
Hawthorne (2009, p.75), Kneer (2015, Ch.8) and Kneer, Vicente & Zeman
(2017). Consider:
More formally, the suggestion is the following:
(S1) For all sentences φ and all singular terms α, FOR α, φ is a sentence.
(S2) For all φ, α, w, s and a: if φ is a sentence and α is a personal name referring to a,
w is a possible world, and s is a [perspective]: FOR α, φ is true in a circumstance <w,
s> iff φ is true in <w, s(a)> (where s(a) is a’s [perspective]) (2009, p. 384)
(18) Maria ate something that was tasty for Anna in a dignified way.
(19) Frank showed John how to cook something tasty for his wife in a fun way.
(20) For Jane, even the most tasty steak is disgusting.
As the examples demonstrate, “dignified”, “fun” and “tasty” leap out of the
scope of the FOR operator. In (18), “dignified” (although not a PPT, it works
similarly enough) is relativized to Maria. In (19) “fun” is relativized not to John’s
wife but to John, Frank, the speaker, or several of them. And in (20), “tasty” is
relativized to the speaker (or perhaps people in general). The FOR operator is
thus an unlikely candidate for a sentential operator and must be conceived as
a predicate operator (Kölbel, 2011: 144 acknowledges this, Lasersohn, 2008
as well as MacFarlane, 2012 also conceive of it as a predicate operator). But if
this is so, the argument has no bite. Kaplan’s point is that the existence of
sentential operators justifies the postulation of parameters in the circumstance,
because parameters, like sentential operators and unlike predicate operators,
shift sentential contents. It is, however, not evident how predicate operators
could justify parameter proliferation in the circumstance of evaluation.
8.2 The Argument from Innocence
In his argument from modal and temporal innocence, Recanati (2007), in
contrast to Kaplan, does not infer the existence of modally and temporally
neutral (or “innocent”) lekta from the existence of operators, but takes them
to be explanatorily prior.
He imagines a “modally innocent” linguistic
community whose language does not have modal operators. Suppose its
members now become modally sophisticated and start to engage in “modal
Such modal talk can be formally represented in two ways, as we have
seen tensed talk can: by using sentence operators, or by explicitly
quantifying world variables in the object-language. If we use the modal
framework and introduce modal operators such as “actually” or
“possibly”, nothing will be changed for the fragment of the language
that does not involve those operators. The sentence “Rain is wet” will
still be a simple, modally innocent sentence. The language will simply
have been enriched by the introduction of new resources enabling us
to construct more complex sentences. But if we use the standard
extensional framework [of first-order logic] and represent modal
Thanks to Jonathan Schaffer (p.c.) for this example.
Travis (2006) also argues against rich Fregean contents, but his complex arguments would
take us too far afield. I hope to respond to them at another occasion.
sentences (“It might be that ...”, “Actually ...”) by means of explicit
quantification over possible worlds […] then, unless special precaution
is taken to avoid that consequence, a change of language takes place,
not merely an enrichment. In the new language, all sentences (including
simple sentences) now contain a hidden argument-place for a world.
Modal innocence is lost. (2007:67-68, italics in the original)
In a nutshell, Recanati argues from the theoretical possibility of a modally (and
temporally) innocent language to the preservation of modally (and temporally)
innocent sentences in a modally sophisticated language. The argument relies
on an apparently fundamental distinction between an enrichment and a
change of language. Problematically, what demarcates the difference is left
undefined. In my view, a modally innocent system of language and thought,
deprived of the resources to weigh and decide amongst alternative courses of
action (which requires the notion of alternative worlds), and thus devoid of
concepts such as choice, expectation, decision, agency, responsibility and
many others essential to our ways of thinking and acting would be a very
different system indeed. In fact, it is hard to fathom how the manipulation of
another single feature beyond modality could have similarly far-reaching
consequences for a conceptual system. The introduction of modal elements
into a language previously innocent in this regard must be regarded as a
radical change (as research on the false-belief task
in developmental
psychology also makes abundantly clear), not just an enrichment, and the
argument, I believe, does not succeed.
I have argued that the argument from modal innocence is unsuccessful. But
even if the argument were convincing, it is clear that it doesn’t carry over to
the taste debate. Quite to the contrary, it would instead cast doubt on
MacFarlane’s attempt to take the sting out of the argument from
incompleteness, and Kölbel’s strategy to extend the operator argument to
parameters beyond worlds and times. Here’s why: Suppose Recanati were
right, and we must preserve the possibility to make sense of world-neutral
contents, for instance in order to account for a modally innocent community’s
grasp of “rain is wet”. A similar move is not available in the PPT debate, as
Recanati himself is well aware: The properties designated by PPTs are
inherently response-dependent, whereas modal properties are not. Hence
imagining a linguistic community in which “delicious” does not invoke an
experiencer or judge (and is thus not response-dependent) can teach us but
See Wimmer & Perner (1983).
Note that Recanati himself points out that the option of “going extensional is not off the
table even if the argument did work, as long as special precautions are taken (2007:72; said
precautions draw on variadic functions, cf. Recanati, 2002).
little about our understanding of predicates of personal taste. In such a
community, “delicious” and the like would function like ordinary monadic
predicates, the conundrum regarding apparent faultless disagreement cannot
arise, and the nonindexicalist would be deprived of her paradigm argument.
Nonindexicalists convinced by Recanati’s reasoning would thus be well-
advised to defend a fundamental difference between modal and temporal
features on the one hand, and personal taste on the other. But once they do,
then both Kölbel’s attempt to extend the operator argument to matters of
taste, as well as MacFarlane’s response to the argument from incompleteness
are off the table. On such a view it makes no sense to argue that the
incompleteness worry overgenerates with respect to worlds, precisely because
taste and modality are not relevantly alike.
8.3 The Binding Argument
Just as in the literature on unarticulated constituents, some arguments in the
PPT debate focus on binding. Lasersohn (2008), for instance, examines (21), a
standard interpretation of which would be (22):
(21) Every man rode some ride that is fun.
(22) Every man x rode some ride that is fun for x (not necessarily for the other
men, or the speaker, or the listeners).
Whereas indexicalists would account for (22) by postulating a bindable
perspective variable in the syntactic form of “fun”, Lasersohn has the quantifier
interact directly with the perspective parameter in the circumstance or index.
Binding thus conceived takes place not at the level of the object-language, but
at the level of the meta-language.
Lasersohn acknowledges that index
binding has less expressive power than variable binding, as becomes evident
with respect to claims that contain several predicates of personal taste.
Whereas variable binding allows each PPT to be relativized selectively to a
particular perspective, on Lasersohn’s approach the quantifier phrase binds
into a single (type of) perspective. Consider the following utterance (Lasersohn,
2008, p. 325) and its four candidate readings:
Following Lasersohn, let φ be a sentence, pro a covert pronoun much like the overt pronoun
pro, M a model, c a context, w a world, p a perspective, P a non-empty set of perspectives, g
an assignment and g[x/n] a sequence in which x is the nth element and which agrees with g in
all other positions. Then object-language binding is defined by (i), and meta-language binding
by (ii):
(i) [[λnφ]]M, c, w, p, g = {x P | [[φ]]M, c , w, p , g[x/n] = 1} (2008, p. 313)
(ii) (a) If α is a sentence containing at least one occurrence of pron, then µn α is a sentence abstract.
(b) [[µnφ]]M, c, w, p, g = {x P | [[φ]]M, c, w, x, g[x/n] = 1} ((2008, p. 324)
(23) Every man gave some woman a fun ride and a tasty dish.
(23a) Every man gave some woman a ride and a dish, which were tasty and fun
according to the speaker.
(23b) Every man gave some woman a ride and a dish, which was tasty and fun
according to each man.
(23c) Each woman received a ride and a dish, which were tasty and fun
according to her standards.
(23d) Every man gave some woman a ride fun by his standards, as well as a
dish tasty according to her standards.
Index binding can account nicely for readings 23a, 23b, and 23c, according to
which the rides and dishes are relativized to the same perspectives. According
to Lasersohn, the perspectival uniformity captured by index binding brings out
a central nonindexicalist commitment:
[In index binding] we are employing a single, systematic parameter,
relative to which all denotations are assigned; and if an operator
manipulates this parameter, it will do so for all expressions in its scope.
The intuition behind this pattern can perhaps be expressed this way: In
a relativist theory, in order to assess a sentence for truth or falsity, one
must adopt a stance that is, truth assessment is always done from a
particular perspective. Operators in the sentence may shift the
perspective from which truth assessment is to be done, or quantify over
such perspectives; and when they do so, the relevant perspective must
be adopted for the entire scope of the operator. Because such
operators shift the perspective from which truth is assessed, rather than
shifting the denotation of some particular expression like a pronoun,
they cannot selectively shift only certain items in their scope. (2008, p.
Interpretation 23d, however, requires that “fun” and “tasty” are relativized to
different perspectives: Each woman receives a dish tasty by her standards and
a ride fun by the standards of some man. Whereas variable binding can, index
binding cannot account for this reading. By Lasersohn’s lights this is
unproblematic, since “the sentence [i.e. (23)] cannot mean that each man gave
some woman a ride that was fun for him, and a dish that was tasty for her”
(2008, p. 325), which, consequently, “show[s] that [predicates of personal taste]
cannot have arguments freely chosen from a set of pronouns similar to pro1,
pro2, pro3,…” (2008, p. 326).
Lasersohn’s intuitions as to the nonavailability of reading (26d) seem to be on
the right track. However, this one example by no means shows that the limited
expressive power of index binding suffices in general or that predicates of
personal taste cannot have perspective arguments. Kneer, Vicente and Zeman
(2017) demonstrate that certain claims that contain multiple PPTs do, in
suitable contexts, require relativization to multiple types of perspectives.
(24) We took the kids to a resort in Italy this summer. The wine was delicious
and the water slide was fun.
An interpretation according to which the wine was delicious for the parents
and the water slide for the kids (or the kids and the parents) is not only
available, it is the default reading (for empirical evidence with native English
speakers, see Kneer (2015, Ch.8) and Kneer (in prep. c). The same holds for
quantified versions of such multiperspectival claims:
(25) On Halloween, every child would play a silly trick on some adult or else
get a delicious treat.
(26) Every steak-lover took some vegetarian friend to the Sunday barbecue for
some tasty ribeyes and delicious corncobs.
(27) On father’s day, the fair comes to town. Every dad goes to the fairground
with some kid to taste the delicious local brews and try out the fun new rides.
In contrast to what Lasersohn alleges, then, it is not the case that claims with
multiple PPTs cannot have multiperspectival readings. The resources of index
binding are insufficient to account for such readings, and the full expressive
power of variable binding is required if we are to account for perspectival
plurality. As long as nonindexicalism or relativism cannot account for the
multiperspectival interpretations of (25)-(27), an indexicalist framework might
be preferable.
9. Conclusion
Let’s take stock: Non-indexical contextualism is in no better a position to
account for disagreement than indexical contextualism, as long as the truth-
conditionally complete proposition is understood as distributed over content
and circumstance. To account for genuine disagreement (rather than the
appearance thereof), radical measures are required: The object of
disagreement, assertion and belief must be conceived as the perspective-
neutral content or lekton itself. Such Stoic propositions or thin contents,
however, have a whiff of semantic incompleteness about them, and it is
doubtful at best whether they can serve to individuate, understand and explain
our propositional attitudes and the actions they drive.
However, MacFarlane suggests, the incompleteness worry overgenerates.
Building perspective tacitly into the content of taste-claims carries a fully
specific proposition on its heels: The time and world features are in no relevant
way distinct from perspective, he assumes, and thus need to be included in
the propositional content, too. As regards worlds, this seems to raise a
problem. Once a contingent proposition is made modally specific, for instance
by including an implicit “actually” operator, it is true in all possible worlds. But
a semantics that confers necessary modal status onto all contingent
propositions is troublesome. Hence, semantic completeness must apparently
be resisted on pains of modal anxiety.
MacFarlane’s argument, I have attempted to show, falls prey to several
objections. Even if modal anxiety were a serious problem, the conclusion
supports moderate nonindexical contextualism rather than radical
nonindexicalism, since the inability to account for faultless disagreement
seems a lesser problem than semantic incompleteness. More importantly,
modal anxiety is in fact unwarranted: As has been demonstrated at length,
propositions that turn out necessary in virtue of stipulating a world in the
content remain contingent in the intuitive, that is, the deep sense, which
concerns the modal properties of sentences or contents. Such propositions are
indeed superficially necessary, but necessity thus conceived is only an artefact
of the formal apparatus paired with rigid designation. Superficial
necessitarianism itself is not an unreasonable position: Given the parallels
regarding time- and world-specificity, necessitarianism should be just as much
the default position as eternalism, on which the literature has, by and large,
A number of alternative arguments in favour of thin content views were
considered. Recanati’s argument from modal innocence, I have argued, is not
convincing. But even if it were, its conclusions would not carry over to the
domain of taste, because taste properties, in contrast to modal properties, are
inherently response-dependent. Kölbel’s attempt to mimic Kaplan’s operator
argument in the domain of personal taste is unsuccessful, because “for” does
not behave like a sentential operator. Finally, the expressive power of index
binding, an ingenious move proposed by Lasersohn, is too limited to account
for perspectival plurality. It is not evident how multiperspectival claims can be
formalized if we do not recur to variable binding.
If the above considerations are on the right track, there is little that speaks in
favour of perspective-neutral claims of personal taste. Furthermore, it is
doubtful whether a coherent challenge from faultless disagreement can be
formulated on contextualist premises, at least if a substantial notion of
disagreement, rather than the mere appearance thereof, is at play. The
nonindexicalist cannot have her cake (faultless disagreement) and eat it (invoke
a plausible notion of content). But if the challenge is misconceived,
indexicalism with respect to standards of taste remains a viable option.
Furthermore, even if treating taste variables as tacit constituents of the content
would indeed engender a similar commitment with respect to worlds, there is
little reason to discard the position, since superficial necessitarianism is
modally unproblematic.
Aronszajn, M. (1996). A defense of temporalism. Philosophical Studies, 81(1), 71-95.
Austin, J. L. (1971). Philosophical Papers (Vol. 20, Vol. 4): Oxford University Press.
Blackburn, P., & Marx, M. (2002). Remarks on Gregory's “Actually” Operator. Journal
of Philosophical Logic, 31(3), 281-288.
Barker, C. (2002). The dynamics of vagueness. Linguistics and Philosophy, 25, 136.
Beddor, B. & Egan, A. (2018). Might do better: Flexible relativism and the
QUD. Semantics and Pragmatics 11(7).
Beillard, J. (2010). The appearance of faultless disagreement. Dialogue: Canadian
Philosophical Review, 49(4), 603-616.
Borg, E. (2005). Saying what you mean: Unarticulated constituents and
communication. In R. Elugardo, & R. Stainton (Eds.), Ellipsis and nonsentential
speech (pp. 237-262): Springer.
Brogaard, B. (2012). Transient truths: an essay in the metaphysics of propositions: OUP
Cappelen, H., & Hawthorne, J. (2009). Relativism and monadic truth. Oxford: Oxford
University Press.
Cappelen, H., & Lepore, E. (2007). The myth of unarticulated constituents.In O’Rourke
and Washington (eds.), Situating Semantics: Essays on the philosophy of John
Perry, 199-214.
Cresswell, M. J. (1990). Entities and indices. Dordrecht: Kluwer.
Collins, J. (2013). The syntax of personal taste. Philosophical perspectives, 27, 51-103.
Crimmins, M. (1992). Talk about beliefs. Cambridge, MA: MIT Press.
Crossley, J. N., & Humberstone, L. (1977). The logic of “actually”. Reports on
Mathematical Logic, 8(1977), 11-29.
Davies, M. (2004). Reference, contingency, and the two-dimensional framework.
Philosophical Studies, 118(1-2), 83-131.
Davies, M., & Humberstone, L. (1980). Two notions of necessity. Philosophical Studies,
38(1), 1-30.
Dinges, A. (2017). Relativism, Disagreement and Testimony. Pacific Philosophical
Quarterly, 98, 497-519.
Dowell, J. L. J. (2011). A Flexible Contextualist Account of Epistemic Modals.
Philosophers' Imprint, 11(14), 1-25.
Egan, A. (2007). Epistemic modals, relativism and assertion. Philosophical Studies,
133(1), 1-22.
Egan, A. (2010). De Gustibus Non Disputandum Est (At Least, Not Always). In R.
Feldman & T. Warfield (Eds.), Disagreement. Oxford: Oxford University Press
Egan, A., Hawthorne, J., & Weatherson, B. (2005). Epistemic modals in context. In G.
P. Preyer, G. (Ed.), Contextualism in philosophy (pp. 131-168). Oxford: Oxford
University Press.
Evans, G. (1979). Reference and Contingency. The Monist, 62(2), 161-189.
Evans, G. (1985). Does tense logic rest on a mistake? In Collected Papers (pp. 343-
363): Oxford: Clarendon Press.
Ferrari, F. & Zeman, D. (2014). Radical relativism, retraction, and "being at fault". In S.
Caputo, M. Dell'Utri, M. & F. Bacchini (eds.), New Frontiers in Truth. Cambridge
Scholar. pp. 80-102 (2014)
Von Fintel, K., & Gillies, A. S. (2008). CIA leaks. Philosophical review, 117(1), 77-98.
Von Fintel, K., & Gillies, A. S. (2011). Might made right. Epistemic modality, 108-130.
Fitch, G. (1998). Temporalism revisited. Philosophical Studies, 92(3), 251-256.
Frege, G. (1979). Logic. In H. Hermes, H. Kambartel, & F. Kaulbach (Eds.), Posthumous
Writings. Chicago: University of Chicago Press.
Glanzberg, M. (2007). Context, content, and relativism. Philosophical Studies, 136(1),
Gregory, D. (2001). Completeness and decidability results for some propositional
modal logics containing “actually” operators. Journal of Philosophical Logic,
30(1), 57-78.
Hanson, W. H. (2006). Actuality, necessity, and logical truth. Philosophical Studies,
130(3), 437-459.
Hodes, H. T. (1984). Axioms for actuality. Journal of Philosophical Logic, 13(1), 27-34.
Huvenes, T. (2012). Varieties of disagreement and predicates of taste. Australasian
Journal of Philosophy 90 (1):167-181.
Kamp, H. (1971). Formal properties of 'now'. Theoria, 37, 227-274.
Kaplan, D. (1989). Demonstratives. In J. Almog, J. Perry, & H. Wettstein (Eds.), Themes
From Kaplan (pp. 481-563): Oxford University Press.
Khoo, J.82015) Modal disagreements. Inquiry 5(1), 1-24.
King, J. and Stanley, J. (2005). Semantics, pragmatics, and the role of semantic content.
In Z. Szabó (ed.), Semantics vs. Pragmatics. Oxford: Oxford University Press.
Kment, B. (2017). Varieties of modality. In E. N. Zalta (Ed.), The Stanford Encyclopedia
of Philosophy. Stanford: Stanford University.
Kneer, M. (2015). Perspective in Language. PhD Dissertation, Ecole Normale
Superieure / EHESS, Paris.
Kneer, M., Vicente, A., & Zeman, D. (2017). Relativism about predicates of personal
taste and perspectival plurality. Linguistics and Philosophy, 40(1), 37-60.
Kneer, M. (in prep. a) Predicates of personal taste: Empirical data.
Kneer, M. (in prep. b). Epistemic modals are not assessment-sensitive.
Kneer. M. (in prep. c). Perspectival Plurality: Evidence from native English speakers.
Knobe, J., & Yalcin, S. (2014). Epistemic modals and context: Experimental data.
Semantics and Pragmatics, 7(10), 1-21.
Kölbel, M. Faultless disagreement. In Proceedings of the Aristotelian Society
(Hardback), 2004a (Vol. 104, pp. 53-73): Wiley Online Library
Kölbel, M. (2004b). Indexical relativism versus genuine relativism. International
Journal of Philosophical Studies, 12(3), 297-313.
Kölbel, M. (2009). The evidence for relativism. Synthese, 166(2), 375-395.
Kompa, N. (2015). Contextualism and disagreement. Erkenntnis, 80(1), 137-152.
Kripke, S. A. (1972). Naming and necessity. Springer.
Lasersohn, P. (2005). Context Dependence, Disagreement, and Predicates of Personal
Taste. Linguistics and Philosophy, 28(6), 643-686.
Lasersohn, P. (2008). Quantification and perspective in relativist semantics.
Philosophical Perspectives, 22(1), 305-337, doi:10.1111/j.1520-
Lasersohn, P. (2009). Relative truth, speaker commitment, and control of implicit
arguments. Synthese, 166(2), 359-374, doi:10.1007/s11229-007-9280-8.
Lasersohn, P. (2011). Context, relevant parts and (lack of) disagreement over taste.
Philosophical Studies, 156(3), 433-439.
Lasersohn, P. (2016). Subjectivity and Perspective in Truth-Theoretic Semantics.
Oxford: Oxford University Press
López de Sa, D. (2007). The many relativisms and the question of disagreement.
International Journal of Philosophical Studies, 15(2), 269 279.
López de Sa, D. (2008). Presuppositions of commonality: An indexical relativist account
of disagreement. In M. Garcia-Carpintero and Kölbel, M. (eds.), Relative truth
(pp. 297-310). Oxford: Oxford University Press.
López de Sa, D. (2015). Expressing disagreement: A presuppositional indexical
contextualist relativist account. Erkenntnis, 80(1), 153-165.
Ludlow, P. (2001). Metaphysical austerity and the problems of temporal and modal
anaphora. Noûs, 35(s15), 211-227.
Marushak, A. (2018). Reasons and Modals. PhD Dissertation, University of Pittsburgh.
Marushak, A. & Shaw, J. Epistemics and Emotives. In prep.
MacFarlane, J. (2007). Relativism and disagreement. Philosophical Studies, 132(1), 17-
MacFarlane, J. (2009). Nonindexical contextualism. Synthese, 166(2), 231-250.
MacFarlane, J. (2011). Epistemic Modals are Assessment-Sensitive. Epistemic
modality, 144.
MacFarlane, J. (2014). Assessment sensitivity: Relative truth and its applications.
Oxford: Oxford University Press.
Marques, T. ( ms). Falsity and Retraction: New Experimental Data on Epistemic
Marques, T. (2014). Doxastic disagreement. Erkenntnis 79: 121-142.
Marques, T. (2015). Disagreeing in context. Frontiers in Psychology, 6, 257.
Marques, T., & García-Carpintero, M. (2014). Disagreement about taste: commonality
presuppositions and coordination. Australasian Journal of Philosophy, 92(4),
Moltmann, F. (2010). Relative truth and the first person. Philosophical Studies, 150(2),
Partee, B. H. (1973). Some structural analogies between tenses and pronouns in
English. The Journal of Philosophy, 70(18), 601-609.
Pearson, H. (2012). A judge-free semantics for predicates of personal taste. Journal of
Semantics, 30(1), 103-154.
Perry, J. (1986). Thought without representation. Proceedings of the Aristotelian
Society, Supplementary Volumes, 137-166.
Plunkett, D., & Sundell, T. (2013). Disagreement and the semantics of normative and
evaluative terms. Philosopher's Imprint, 13.
Recanati, F. (2002). Unarticulated constituents. Linguistics and Philosophy, 25(3), 299-
Recanati, F. (2004). Literal meaning. Cambridge: Cambridge University Press.
Recanati, F. (2007). It is raining (somewhere). Linguistics and Philosophy, 30(1), 123-
146, doi:10.1007/s10988-006-9007-1.
Recanati, F. (2007) Perspectival Thought: A Plea for (Moderate) Relativism. Oxford:
Clarendon Press.
Roberts, C. (in prep.) The Character of Epistemic Modality: Evidential indexicals
Richard, M. (1981). Temporalism and eternalism. Philosophical Studies, 39(1), 1-13.
Richard, M. (1982). Tense, propositions, and meanings. Philosophical Studies, 41(3),
Richard, M. (2004). Contextualism and relativism. Philosophical Studies, 119(1), 215-
Richard, M. (2008). When truth gives out. Oxford: Oxford University Press.
Richard, M. (2011). Relativistic content and disagreement. Philosophical Studies,
156(3), 421-431, doi:10.1007/s11098-010-9687-9.
Saebo, J.S. (2009). Judgment Ascriptions. Linguistics and Philosohpy 32, 327-352.
Salmon, N. (1986). Frege's puzzle. Cambridge, MA: MIT Press.
Salmon, N. (2003). Tense and Intension. In Q. Smith, & A. Jokic (Eds.), Time, tense, and
reference (pp. 107-154). Cambridge, MA: MIT Press.
Schaffer, J. (2011). Perspective in taste predicates and epistemic modals. In A. Egan,
& B. Weatherson (Eds.), Epistemic modality: Oxford University Press.
Schaffer, J. (2012). Necessitarian propositions. Synthese, 189(1), 119-162.
Schaffer, J. (2018). Confessions of a schmentencite: towards an explicit
semantics. Inquiry, 1-31.
Sennet, A. (2011). Unarticulated constituents and propositional structure. Mind &
Language, 26(4), 412-435.
Stalnaker, R. (1970). Pragmatics, reprinted in. In Context and Content: Essays on
Intentionality in Speech and Thought (pp. 31-46). Oxford: Oxford University
Stanley, J. (1997a). Names and rigid designation. In B. Hale, & C. Wright (Eds.), A
Companion to the Philosophy of Language (pp. 555-585). Oxford: Blackwell
Stanley, J. (1997b). Rigidity and content. In R. Heck (Ed.), Logic, Language and Reality:
Essays in Honor of Michael Dummett (pp. 131-156). Oxford: Oxford University
Stephanou, Y. (2005). First-Order Modal Logic with an 'Actually' Operator. Notre Dame
Journal of Formal Logic, 46(4), 381-405.
Stojanovic, I. (2007). Talking about taste: disagreement, implicit arguments, and
relative truth. Linguistics and Philosophy, 30(6), 691-706.
Stojanovic, I. (2012). On Value-Attributions: Semantics and Beyond. The Southern
Journal of Philosophy, 50(4), 621-638, doi:10.1111/j.2041-6962.2012.00137.x.
Stojanovic, I. (2017). Context and disagreement. Cadernos de Estudos Linguísticos,
59(1), 7-22.
Stone, M. (1997). The anaphoric parallel between modality and tense. Philadelphia:
Department of Computer and Information Science Technical Reports,
University of Pennsylvania.
Sundell, T. (2011). Disagreements about taste. Philosophical Studies, 155(2), 267-288.
Sundell, T. (2017). Aesthetic Negotiations. In James Young (ed.), The Semantics
of Aesthetic Judgment. Oxford: Oxford University Press.
Taylor, K. A. (2001). Sex, breakfast, and descriptus interruptus. Synthese, 128(1-2), 45-
Travis, C. (2006). Thought's footing: A theme in Wittgenstein's philosophical
investigations. Oxford University Press.
van Benthem, J. (1977). Tense logic and standard logic. Logique et analyse, 20(80),
Vlach, F. (1973). 'Now' and 'then': A formal study in the logic of tense anaphora. (Ph.D.
Dissertation). University of California Los Angeles,
Von Fintel, K., & Gillies, A. S. (2008). CIA leaks. Philosophical Review, 117(1), 77-98.
von Fintel, K., & Gillies, A. S. (2009). 'Might' Made Right. In A. Egan, & B. Weatherson
(Eds.), Epistemic modality: Oxford University Press.
Wettstein, H. K. (1979). Indexical reference and propositional content. Philosophical
Studies, 36(1), 91-100.
Wimmer, H. & Perner, J. (1983). Beliefs about beliefs: Representation and constraining
function of wrong beliefs in young children's understanding of
deception. Cognition. 13 (1): 103128.
Wright, C. (2001). On being in a quandary. Relativism vagueness logical revisionism.
Mind, 110(437), 45-98.
Yanovich, I. (2013). Standard contextualism strikes back. Journal of semantics, ffs022.
Zakkou, J. (2019). Denial and retraction: a challenge for theories of taste
predicates. Synthese, 196(4), 1555-1573.
Zakkou, J. (2019). Embedded taste predicates. Inquiry, 62(6), 718-739.
... The linguistic intuitions inspiring such accounts, however, are not uncontroversial (cf. Hawthorne 2007, Von Fintel & Gillies 2007Yalcin 2011;Braun 2012;Kneer 2015Kneer , 2020. Still, one can safely accept the assessment-sensitivity of future contingents-where intuitions are considerably more uniform-without buying into a rampant relativ- ...
Full-text available
According to Anscombe, acting intentionally entails knowledge in ac- tion. This thesis has been near-universally rejected due to a well-known counter- example by Davidson: a man intending to make ten legible carbon copies might not believe with confidence, and hence not know, that he will succeed. If he does, however, his action surely counts as intentional. Damaging as it seems, an even more powerful objection can be levelled against Anscombe: while act- ing, there is as yet no fact of the matter as to whether the agent will succeed. Since his belief that he will is not yet true while his action is in progress, he can- not possibly know that he is indeed bringing about the intended goal. Knowl- edge in action is not only unnecessary for intentional action, it seems, but–at least as regards success-bound types of action–impossible to attain in the first place. In this paper I argue that traditional strategies to counter these objections are unsatisfactory and propose a new account of knowledge in action which has two core features: (i) It invokes an externalist conception of justification which not only meets Davidson’s challenge, but also casts doubts on the tacit internalist premise on which his example relies. (ii) Drawing on recent work about by John MacFarlane, the proposed account conceives of claims to in action as assessment-sensitive so as to overcome the factivity objection. From a retrospective point of evaluation, previous claims about future events and actions can not only be deemed as having been true, but also as having been known.
Full-text available
Contextualism is the view that the extension of perspectival claims (involving e.g. predicates of personal taste or epistemic modals) depends on the context of utterance. Relativism is the view that the extension of perspectival claims depends on the context of assessment. Both views make concrete, empirically testable predictions about how such claims are used by ordinary English language speakers. This chapter surveys some of the recent empirical literature on the topic and presents four new experiments (total N=724). Consistent with contextualism and inconsistent with relativism, the results suggest that the extension of perspectival claims depends on the context of utterance, not the context of assessment.
Full-text available
Contextualism is the view that the extension of perspectival claims (involving e.g. predicates of personal taste or epistemic modals) depends on the context of utterance. Relativism is the view that the extension of perspectival claims depends on the context of assessment. Both views make concrete, empirically testable predictions about how such claims are used by ordinary English language speakers. This chapter surveys some of the recent empirical literature on the topic and presents four new experiments (total N=724). Consistent with contextualism and inconsistent with relativism, the results suggest that the extension of perspectival claims depends on the context of utterance, not the context of assessment.
Full-text available
According to contextualism, the extension of claims of personal taste is dependent on the context of utterance. According to truth relativism, their extension depends on the context of assessment. On this view, when the taste preferences of a speaker change, so does the truth value of a previously uttered taste claim, and the speaker might be required to retract it. Both views make strong empirical assumptions, which are here put to the test in three experiments with over 740 participants. It turns out that the linguistic behaviour of ordinary English speakers is consistent with contextualist predictions and inconsistent with the predictions of the most widely discussed form of truth relativism advocated by John MacFarlane.
Full-text available
Perspectival plurality is the phenomenon according to which certain claims containing multiple predicates of taste can be sensitive to various contextually salient perspectives. For instance, if a father reports on a family holiday in Italy by saying 'The wine was delicious and the water slide a lot of fun', the predicate 'delicious'-in suitable contexts-must be relativized to the father and 'fun' to the kids. The paper argues that perspectival plurality raises severe problems for nonindexicalist semantics of perspectival expressions. Plurality blocks any attempt to justify parameter proliferation by aid of Kaplanian operator arguments, and it frustrates reasonable nonindexicalist strategies to account for syntactic binding. Both arguments must be taken serious: As the second, empirical part of the paper demonstrates with experiments targeting both predicates of personal taste and epistemic modals, perspectival plurality is a genuine feature of ordinary linguistic discourse.
Full-text available
This paper is an occasion to take stock of the place that the argument from faultless disagreement has occupied in philosophy of language in the last fifteen years, in particular in the debate between contextualism and relativism. The survey offered in the first part of the paper appears to show that the phenomenon of disagreement fails to provide any strong motivation for adopting any novel semantic framework, contrary to relativists’ claims. But, on a more positive note, the interest in disagreement has allowed us to better understand how language works. The second part of the paper explains how the fine-grained structure of meaning – at the level of words, sentences and even entire discourse – reveals a variety of sources that may lead discourse participants to disagree.
Radical relativism was born with a promise: to account for certain phenomena that opposite views are unable to explain. One example is the phenomenon of “faultless disagreement”, according to which two people, while disagreeing, are not at fault in any substantive way. The phenomena of retraction and assessments of truth in cases of eavesdropping are others. All these phenomena have been claimed to pose serious problems for rival views and be best accounted for within a radical relativistic framework. While “faultless disagreement” and the notion of disagreement in general has benefited from extensive discussion in current debates over semantic content, retraction has not been in the spotlight that much. In particular, very few things have been said about what retraction exactly amounts to and how to conceive of its normative profile. This will be the focus of our paper. We will begin by giving an intuitive characterization of retraction by means of some examples (section 2). After presenting the basics of the radical relativist view in section 3, we move to investigating retraction, offering what we take to be some key elements for a substantial analysis of the phenomenon (section 4). Such analysis, we claim, has the virtue of making clear what the normative peculiarity of the notion of retraction is—namely, its retroactive efficacy. In section 5, we inquire into the sense of “fault” in which retractors are said to deem their former selves as not being at fault when making the retracted assertion (MacFarlane 2014). In this connection, we highlight an asymmetry between retractions involving predicates of personal taste and moral terms (section 6). After noting that the epistemic notion of “fault” used by MacFarlane’s cannot explain the asymmetry, in the following section (7) we offer our own explanation, by appealing to a less discussed dimension of assertion evaluation which we call “circumstance-accuracy”. In section 8 we provide support for such an explanation by taking a cue from the legal domain and show how an important distinction found there can be applied to the case of retracting assertions as well. We flag some issues that our paper opens up for further research in section 9.
Wide-ranging semantic flexibility is often considered a magic cure for contextualism to account for all kinds of troubling data. In particular, it seems to offer a way to account for our intuitions regarding embedded perspectival sentences. As has been pointed out by Lasersohn [2009. “Relative Truth, Speaker Commitment, and Control of Implicit Arguments.” Synthese 166 (2): 359â 374], however, the semantic flexibility does not present a remedy for all kinds of embeddings. In particular, it seems ineffective when it comes to embeddings under operators with truth evaluative adverbs such as ‘correctly believes that’ and ‘incorrectly believes that’ and under factive verbs. This paper takes a closer look at the problematic embedding data with respect to predicates of personal taste. It argues that there is indeed no semantic solution for contextualism but a pragmatic way out.
Natural language semantics is heir to two formalisms. There is the extensional machinery of explicit variables traditionally used to model reference to individuals, and the intensional machinery of implicit index parameters traditionally used to model reference to worlds and times. I propose instead a simple and unified extensional formalism – explicit semantics – on which all sentences include explicit individual, world and time variables. No implicit index parameters are needed.