Professional Documents
Culture Documents
Alex Silk
a.silk,bham.ac.uk
Draf of May io1
Abstract
It is a commonplace that some of our desires are stronger than others;
that certain values, norms, and rules are more important than others; and
that states of aairs can be more likely or typical than others. Some authors
have claimed that standard analyses of modals cannot capture how the truth
conditions of modal sentences can be sensitive to such dierences in strength
and priority. I develop an interpretation of a standard premise semantics for
modals that captures various ways in which weights and priorities can fgure
in the interpretation of modal sentences. I then extend the proposed analyses
in considering certain inference patterns involving modals and comparatives,
and modals and conditionals. Te resulting framework provides an attractive
account of the relation between the conventional meanings of modals and the
logic of weights and priorities.
Tanks to Fabrizio Cariani, Kai von Fintel, Irene Heim, Ezra Keshet, Dan Lassiter, Bob Stalnaker,
Eric Swanson, and participants at SALT ii for helpful discussion, and to two anonymous reviewers
for DEONio1 for their valuable comments. Apreliminary version of the ideas in this paper can be
found in Anca Chereches (Ed.), Proceedings of Semantics and Linguistic Teory (SALT) ::, o, .
Introduction
It is a commonplace that some of our desires are stronger than others; that certain
values, norms, and rules are more important than others; and that states of aairs
can be more likely or typical than others. Tese dierences in strength and prior-
ity can aect which modal claims are true. Suppose you promised Alice that you
would meet her for lunch, but you also promised your ailing mother that you would
drive her to the hospital for a critical operation. You realize that you cant keep both
promises, but it is much more important that you keep your promise to your mother.
Suppose also that there are no other normatively relevant factors. Intuitively, (1) is
true and (i) is false. (Here and throughout, assume that the modals are given a
uniform type of normative reading.)
(1) You must keep your promise to your mother.
(i) You must keep your promise to Alice.
Call this case wiicu1iu vvomisis.
Cases like wiicu1iuvvomisis raise a prima facie challenge for the classic premise
semantic framework for modals. Modals are treated as receiving their reading or in-
terpretation from a contextually supplied set of premises.
1
Since modals can them-
selves occur in intensional contexts, it is standard to index premise sets to a world of
evaluation(written: P
w
). To a frst approximation, givena consistent set of premises
P
w
, Must says that the prejacent follows from P
w
. Making room for non-trivial
interpretations of modal claims given inconsistent premise sets, Must says that
follows from every maximally consistent subset of P
w
. Slightly less roughly, given
a premise set F
w
(a modal base) that describes some set of relevant background
facts in w, and given a further premise set G
w
(an ordering source) that represents
the content of some ideal (morality, your goals, etc.) in w, Must says that fol-
lows from every maximally consistent subset of F
w
G
w
that includes F
w
or, in
terms of the simplifying notation in Defnition 1, that follows from every set in
m.x(F
w
, G
w
).
i
1
Equivalently, a preorder (Liwis 181). See especially Kv.1ziv 1,,, 181a, 11; also
v. Fv..ssi 1,, Liwis 1,, Vii1m. 1,o.
i
Kv.1ziv 181a, 11; cf. Gonii io1: i8oi81. Tis premise semantic implementation is
equivalent to the perhaps more familiar implementation in Kv.1ziv 181a, 11 which uses the
ordering source to induce a preorder on the worlds compatible with the modal base (Liwis 181).
For simplicity I make the limit assumption (Liwis 1,: 1io) and assume that ordering the con-
sistent subsets of F
w
G
w
(that include F
w
as a subset) by set inclusion yields a set of subsets that
are -maximal. For semantics without the limit assumption, see Liwis 1,, Kv.1ziv 181a, 11,
1
Denition . m.x(F
w
, G
w
) = {P cons(P) P
[P P
F
w
G
w
cons(P
)]
F
w
P F
w
G
w
}, where, for a set of propositions S, cons(S) i S
Denition . Must is true at w i P m.x(F
w
, G
w
) P
What premise sets are called for in a case like wiicu1iu vvomisis: As David Lewis
counsels us, We must be selective inthe choice of premisesBy judicious selection,
we can accomplish the same sort of discrimination as would result from unequal
treatment of premises (181: iioii1; cf. Kv.1ziv 181b: i1o). Te question is
whether we can do so in a way that captures the full range of data and refects our
intuitive views about the norms that are relevant in the context.
Te musts in (1)(i) are interpreted with respect to the relevant norms. Te
normative force of your promise to Alice should be represented, even if this promise
is ultimately outweighed by the promise to your mother. Intuitively, the norms that
fgure in the interpretation of must in (1)(i) are the same as the norms that fgure
in the interpretation of have to in ().
() If you didnt visit your mother, you would have to meet Alice for lunch.
So, we might treat wiicu1iu vvomisis as calling for premise sets like the ones in
() that describe the relevant circumstances and the contents of the relevant prac-
tical norms, where l is the proposition that you meet Alice for lunch, and h is the
proposition that you take your mother to the hospital.
() F
w
= {(h l)}
G
w
= {h, l}
P = F
w
G
w
= {(h l), h, l}
But this wont do. P is inconsistent. Holding fxed the relevant circumstance that you
cant keep both promises, P has maximally consistent subsets P
1
= {(h l), h} and
P
i
= {(h l), l}. Since neither h nor l follows from both P
1
and P
i
, this incorrectly
predicts that each of (1) and (i) is false (though it correctly predicts that You must
keep your promise to Alice or your mother is true). Te problemis that there seems
Sw.so io11. I assume that the prejacents of modals and the elements of premise sets are proposi-
tions, conceived as sets of possible worlds. For expository purposes I focus only on strong necessity
modals, like must and have to; weak necessity modals, like ought and should, raise complications
orthogonal to our discussion here (see Siix io1 for my preferred account).
=
V/. I will ofen use unitalicized capital letters or w in referring to worlds, italicized lowercase
(English or Greek) letters in referring to propositions, and italicized capital letters in referring to sets
of propositions.
i
to be no room for representing how your promise to your mother is stronger than
your promise to Alice, or how the premise h has priority to the premise l. Tere
seems to be no mechanism for breaking the tie between the maximally consistent
subsets of P.
It is rare to fnd explicit articulations of the premises that fgure in the inter-
pretations of modals. Not unsurprisingly, problems concerning weights and prior-
ities among premises have received little attention in the formal semantics litera-
ture. Daniel Lassiter (io11) is one of the few to address these issues; his assessment
is not optimistic: Te problem is fundamentally that the theory makes no room
for one [premises] being stronger than another; instead any confict of [premises]
leads to incomparability; the theory doesnt leave any roomfor [one premises] be-
ing stronger or weaker than another [Premises] are all-or-nothing (io11: o1oi,
oo; see also 1,1, 1,).
In context Lassiter is considering the implications of this alleged failure to represent weights for
Kratzers account of (epistemic and deontic) comparatives.
,
For instance, abstracting away from details of implementation, we might introduce a ranking
Gw
on the propositions in a given premise set G
w
. We could then introduce a function G
w
that
takes F
w
and
Gw
as argument and returns a preorder on the elements of m.x(F
w
, G
w
). Making an
analogue of the limit assumption, we could then say that Must is true i follows from every set
in m.x(F
w
, G
w
) that ranks among the highest in the preorder G
w
(F
w
,
Gw
). In wiicu1iu vvomisis,
G
w
and
Gw
would privilege the maximally consistent subset P
1
over P
i
; it would only be P
1
from
which the modals prejacent would be said to follow. Tis would correctly predict that (1) is true but
(i) is false. Tis is essentially the route taken by Goble (io1: i8,i), following work by H.si
(iooo) and Hov1v (ioo,, io1i); cf., e.g., Asuiv & Boiv.c 1o, Bvow 1o, Biiziv & Loiwiv
1,, v. uiv Tovvi & T. 18, H.sso ioo1. See K.1z i1 .i. io1i for a related account that
represents cascades of priorities via an operation of ordered merging of ordering sources that mirrors
the lexicographical product for posets. Tis account canbe seenas developing Kratzers (181b, 18,
iooi) strategy of lumping propositions into additional premises to capture the relative importance
of facts in the interpretation of counterfactuals.
semantics that captures various ways in which weights and priorities can fgure in
the interpretation of modal sentences (i). I then extend the proposed analysis
to the case of graded modal expressions in considering certain inference patterns
involving modals and comparatives (). Te resulting framework provides an at-
tractive account of the relation between the conventional meanings of modals and
the logic of weights and priorities (,).
Weights, priorities, and applicability conditions
We need a more nuanced enough understanding of the considerations that fgure
in the interpretation of modals. Norms like values, goals, desires, etc. typically
dont come in the form of blunt categorical imperatives or commands. Tey arent
usually of the form ^o matter what, !. Rather they ofen come with what I will call
applicability conditions (ACs), or conditions under which they apply. If I want to go
for a run, my desire neednt be that I go for a run, come what may. More plausibly, it
is that I go for a run given that its nice outside, among other things. Our preferences
are ofen conditional, preferences for certain circumstances. Similarly with moral
norms. A norm against lying neednt take the form of a categorical prohibition. It
might be something to the eect that you dont lie unless not lying would lead to an
innocent persons death, or would put your family in grave danger, or . Norms
can thus be understood on the model of conditional imperatives, imperatives that
enjoin an action or state of aairs given that certain circumstances obtain. Tis
captures the intuitive idea that depending on the circumstances i.e., depending
on which applicability conditions are satisfed only certain norms may apply, or
be in force. Fixing terminology, call the content of a conditional norm, goal, etc. a
consideration. Given a consideration If C, , let C be the considerations applicability
condition, and be the considerations premise, or what the consideration enjoins
given C. (Categorical considerations can be treated as conditional on the tautology.)
Tere are a number of ways applicability conditions might be integrated into the
semantics. One option, suggested by the few explicit remarks that there are on the
contents of ordering sources in concrete examples, would be to build them into the
propositions that are the elements of ordering sources. Te elements of ordering
sources would be identifed with considerations, construed as material conditionals
C .
o
Simplifying quite a bit, the relevant deontic premise set in wiicu1iu co.is
o
For example: one should take the propositions that make up the relevant deontic background
context l(w) to consist just of such conditional propositions, like that If someone owns a car and is
not handicapped, he must pay taxes (Fv.x 1o: ,; see also ,1,i, 18o181, Pov1iv ioo: 8,
= {(b p), h b, h p}
Intuitively, in light of his parents commands, Timmy must work on his home-
work frst. However, since h is compatible with P
w
), where G
w
is the set of premises
such that there is a pair C, in G
w
such that F
w
C. Since my aim here is to capture the role of
weights and priorities within the classic semantic framework, I wont pursue this approach. We can
capture the intuitions driving it utilizing resources already present in the classic semantics.
o
Timmys parents command that determines the intended reading of must in ()
can be treated as a function that assigns to every possible world the set of proposi-
tions describing the house rules in that world.
() Inlight of what Little Timmys parents command, he must do his homework
frst.
Similar remarks hold for the meanings of phrases like in view of the relevant cir-
cumstances, according to U.S. law, and so on. It is these functions that context
supplies for the interpretation of modals. Call these functions unsaturated premise
sets (written S). Call the value of an unsaturated premise set given a world of eval-
uation a saturated premise set, or simply a premise set (written S
w
).
We can capture the role of applicability conditions in terms of variability in the
values of unsaturated premise sets at dierent worlds. Suppose we have a consid-
eration If C, which enjoins given that conditions C obtain. We can represent
the content of this consideration with an unsaturated premise set S that assigns to
every relevant C-world a premise set that includes . For example, the import of
your desire to go for a run, mentioned above, would be refected in Ss assigning a
premise set that includes the proposition that you go for a run to worlds in which the
weather is nice (among other things). Te premises in a saturated premise set thus
refect what follows from a body of considerations what is enjoined by a body of
conditional norms, what is preferred in light of a body of conditional preferences,
what is expected in light of a body of evidential relations, etc. given the circum-
stances that obtain in the evaluation world.
G
AM
= {h}
G
AM
=
Te normative import of your promise to Alice is refected in Gs assigning a premise
set that includes l to some world in which you make this promise, namely AM. Te
normative import of your promise to your mother is refected in Gs assigning a
premise set that includes h to some world in which you make this promise, e.g.
AM (or AM). And the priority of keeping your promise to your mother over your
promise to Alice is refected in Gs assigning a premise set that includes h to some
world in which you make both promises, namely AM. (Had your promises been of
equal importance, this would be refected in a premise set that contains the disjunc-
tion h l.) Tis unsaturated premise set G correctly predicts that (1) is true and
(i) is false in the given context, i.e. at AM: h, but not l, follows from every set in
m.x(F
AM
, G
AM
).
(1) m.x(F
AM
, G
AM
) = {{(h l), h}}
Prima facie conficts among norms neednt lead to incomparabilities.
One might worry that this treatment of wiicu1iu vvomisis makes incorrect
predictions concerning contrary-to-duty imperatives. Consider (1,)(1o).
1i
(1,) You have to take your mother to the hospital.
(1o) If you dont take your mother to the hospital, you have to meet Alice for
lunch.
Assuming a standard Kratzerian (11) treatment of conditionals (Defnition ),
the antecedent of (1o) adds the proposition h to F
AM
, and the modal have to is
interpreted with respect to this updated modal base F
+
AM
= F
AM
{h}. Given the
proposed premise sets for F
AM
and G
AM
, this seems to incorrectly predict that (1o)
is false, as refected in (1,)(18):
Denition . If , is true at w i P m.x(F
+
w
, G
w
) P , where F
+
w
=
F
w
{}
(1,) F
+
AM
= {(h l), h}
G
AM
= {h}
1i
I use have to instead of must to avoid potential complications fromthe entailingness of must.
(I.e., many speakers fnd Must , but and Must , but might to be marked, even for deontic
readings of must. For such speakers, accepting Must has the potential to violate the presupposi-
tion of conditionals If that is a live possibility.)
1o
m.x(F
+
AM
, G
AM
) = {F
+
AM
}
(18) (1o) is true at AM i
P m.x(F
+
AM
, G
AM
) P l i
{(h l), h} l
Tere are subtle issues concerning time, andwhat informationis takenfor granted,
which may complicate the interpretation of pairs of claims like (1,)(1o). Intu-
itively, in interpreting the modal in (1,) one assumes that the acts of keeping your
promise to your mother and keeping your promise to Alice are both available to you.
In interpreting the modal in the consequent of the conditional in (1o), by contrast,
one assumes that the act of keeping your promise to your mother is no longer an
option. One way of capturing this is to refne our indices of evaluation. Te norms
encoded in G must be conditional not only what promises you have made but also
on what acts are available to you. Let AM1 be a circumstance in which the option
of keeping your promise to your mother and the option of keeping your promise
to Alice are both available, and let AMi be a circumstance in which only the op-
tion of keeping your promise to Alice is available.
1
A more fne-grained deontic
unsaturated premise set G
AM1
= {h}
G
AMi
= {l}
Assuming that (1,) is evaluated at a circumstance like AM1 at which keeping your
promise to your mother is still available, we continue to predict that (1,) is true.
Adopting a double modal analysis of deontic conditionals delivers the correct inter-
pretation for (1o).
1
Roughly, on such an analysis, in interpreting a deontic condi-
tional one checks whether the modalized consequent clause is verifed at all relevant
(circumstantially accessible, epistemically accessible, closest) worlds in which the
antecedent holds. Tis delivers the following simplifed truth conditions for (1o):
(io) (1o) is true at wifor all relevant h-worlds w
P m.x(F
w
, G
w
) P l
1
As before, AM1 and AMi can be treated as representatives of suitable equivalence classes of
worlds. Te sensitivity to time or available acts may also be captured by treating indices as world-
time pairs or situations. I remain neutral on these alternative implementations. For discussion of
timeless contrary-to-duty cases, see, e.g., Pv.xxi & Sivco1 1o, 1,.
1
See Fv.x 1o, GiUv1s ioo, vo Fi1ii & I.1viuoU ioo,, Lisiii ioo, Sw.so io1o,
Siix io1o. I leave open what kind of reading the posited higher necessity modal is to be given.
11
Tis correctly predicts that (1o) is true at AM1, andthus consistent with(1,): roughly,
(1o) is true at AM1 iyou have to meet Alice is true at AMi i{(hl), h, l} l.
. Outweighing and undercutting
Wiicu1iu vvomisis was a case where one consideration was outweighed by an-
other conficting consideration. Now consider a case where the applicability of one
premise undercuts, or excludes, the applicability of another premise.
1,
Suppose Betty
is a cadet, and her Captain orders her to clean the barracks. Ordinarily, this would
imply that Betty has to clean the barracks. But the Major, who outranks them both,
orders Betty to ignore the Captains command. Intuitively, (i1) is true.
(i1) Betty doesnt have to clean the barracks.
Te Majors command, it is ofen claimed, isnt an ordinary, weightier frst-order
reason; rather, it undercuts the consideration about the Captains command from
bearing on Bettys deliberation.
Let b be the proposition that Betty cleans the barracks, c be the proposition that
the Captain ordered Betty to clean the barracks, and m be the proposition that the
Major ordered Betty to ignore the Captains command. Let CM be the world as it
is described by the case, and CM be an otherwise similar world in which the Major
doesnt order Betty to ignore the Captain. We cancapture the contents of the relevant
norms at play with an unsaturated premise set with the following properties:
(ii) G
CM
=
G
CM
= {b}
Te normative import of the Captains commandis refectedinGs assigning a premise
set that includes b to (c m)-worlds in which the Major doesnt interfere. Te un-
dercutting role of the Majors command is refected in Gs assigning a premise set
that fails to include b to (c m)-worlds. Tis correctly predicts that (i1) is true in
the given context (i.e., at CM): {c, m} b.
Tat the Captains command is undercut, and not outweighed, is refected in the
fact that the Major neednt forbid Betty from cleaning the barracks or order her to
performsome alternative action. He might just want to undermine the Captains au-
1,
Compare the distinction between rebutting defeat and undercutting defeat in the litera-
ture on epistemic reasons (Poiiocx 1,o, 1,), and the notions of exclusionary reasons (R.z
1,,/1o, G.s 18o, Pivvv 18) and overridden requirements (Cuisuoim 1o, 1,) in
ethics.
1i
thority. Te undercutting role of the Majors command can be further reinforced by
considering a minor extension of the case. Adapting an example fromHov1v (io1i:
111), suppose that the situation is as before, but Betty also received an order
from her Lieutenant to do drills. Since the Captain outranks the Lieutenant, ordi-
narily Betty would have to obey the Captains command and not do drills (assuming
she cant both clean the barracks and do drills). But given the Majors command to
ignore the Captain, there is nownothing excluding the Lieutenants command from
applying. Te contents of the relevant norms can be refected as follows, where d is
the proposition that Betty does drills, l is the proposition that the Lieutenant ordered
Betty to do drills, and Lxx is an l-world.
(i) G
LCM
= {d}
G
LCM
= {b}
G
LCM
=
Tis correctly predicts that (i1) and (i) are true in the revised context (i.e., at
LCM): {c, m, l, (b d), d} entails d but doesnt entail b.
(i) Betty must do drills.
More generally, the contrasting ways in which considerations can be undercut
and outweighed is represented in terms of what premise sets are assigned at certain
minimally dierent worlds. Let a and b be relevant conditions, AB be a relevant
(a b)-world, and AB be a relevant (a b)-world. (For simplicity, suppose that
G
AB
and G
AB
are each consistent, and bracket the role of F.) Suppose that G
AB
,
refecting that given a and absent some defeating condition b, is necessary. Te
premise that , with applicability condition a, is outweighed if there is a premise
with applicability condition b, where = , such that G
AB
and G
AB
.
By contrast, the premise that is undercut by a background condition b simply if
G
AB
.
. Epistemic readings
So far we have been focusing on root modals. Te proposals for capturing priorities
among premises, and for distinguishing outweighing from undercutting, apply to
epistemic readings of modals as well. First, consider a familiar epistemic analog of
wiicu1iu vvomisis. Suppose you are looking at a ball. It seems red to you, but a
peer tells you that the ball isnt actually red. Your sense perception is general reli-
able; typically, the fact that an object seems red is good evidence that the object is
1
red. But your peer is eminently trustworthy and might have access to information
that you dont for example, perhaps there are unusual lighting conditions. Te re-
liability of your peer, lets suppose, is even greater than that of your sense perception.
Intuitively, this outweighs your reason for thinking that the ball is red indeed, it
gives you reason for thinking the ball is not red. Suppose that there is no other rele-
vant possible evidence that may bear on the color of the ball. (i,) seems true in this
scenario.
(i,) Te ball must not be red.
We can capture this as follows. Let l be the proposition that the ball looks red; r be
the proposition that the ball is red; and t be the proposition that your peer told you
that the ball isnt red. Consider the worlds LT and LT, characterized in the expected
way as above. We can capture the priorities among the relevant evidential norms
with an unsaturated premise set with the following properties.
(io) G
LT
= {r}
G
LT
= {r}
Te evidential import of the objects looking redis refectedinGs assigning a premise
set that includes r to t-worlds where the ball looks red. Te priority of your peers
testimony is refected in Gs assigning a premise set that includes r to t-worlds. Tis
predicts that (i,) is true in the given context (i.e., at LT): {l, t, r} r.
Nowconsider a case of epistemic undercutting. Suppose that rather thanhearing
fromyour peer that the object is not red, you realize that you have taken a drug that
makes everything look red. Intuitively, this undercuts your reason for concluding
that the ball is red. (i,) is false in this scenario.
(i,) Te ball must be red.
Let d be the proposition that you took the drug, and consider the worlds LD and
LD, characterized in the expected way. We can capture the probabilistic information
encoded in the relevant epistemic norms with an unsaturated premise set with the
following properties:
(i8) G
LD
=
G
LD
= {r}
Te evidential import of the objects looking redis refectedinGs assigning a premise
set that includes r to d-worlds in which the ball looks red. Te undercutting role of
1
taking the drug is refected in Gs assigning the empty set to d-worlds. Tis predicts
that (i,) is false in the given context (i.e., at LD): {l, d} r.
Modals and comparatives
An adequate general treatment of priorities must extend to the case of comparatives,
like (i).
(i) It is better for me to keep my promise to my mother than for me to keep
my promise to Alice.
Indeed, in the passages fromL.ssi1iv io11 cited in 1, Lassiters central criticismof
the implementation in Kv.1ziv 181a, 11, io1i is that its proliferation of incon-
sistencies in premise sets leaves it unable to capture the truth of such comparatives.
Te primary aim of this paper has been to capture how intuitions about priorities
among considerations can aect the truth conditions of modal sentences. Te pro-
posed strategy is compatible with various views on the semantics of comparatives
like (i) and the semantic relation between modals and comparatives. Neverthe-
less I would like to briefy mention one way of extending the account fromi to
capture certain data involving comparatives.
I suggest that comparatives like (i) have a kind of counterfactual element to
their meaning: in considering whether an option is better than another option ,
one looks at relevant possibilities in which or is necessary (in the relevant sense),
and assesses whether is necessary in those possibilities. Informally, (i) seems to
mean something like If I had to keep my promise to my mother or to Alice and
conditions were otherwise normal, expected, or as desired I would have to keep
my promise to my mother. Roughly, is better than is true at w i for all closest
(maximally similar) relevant worlds w