Two Types of Moral Dilemmas

PETER VALLENTYNE
TWO TYPES OF MORAL DILEMMAS
ABSTRACT. In recent years the question of whether moral dilemmas are conceptually
possible has received a fair amount of attention. In arguing for or against the conceptual
possibility of moral dilemmas authors have been almost exclusively concerned with
obligation dilemmas, i.e., situations in which more than one action is obligatory. Almost
no one has been concerned with prohibition dilemmas, i.e., situations in which no
feasible actions is permissible. I argue that the two types of dilemmas are distinct, and
that a much stronger case can be made against the conceptual possibility of obligation
dilemmas than against the conceptual possibility of prohibition dilemmas.
1. INTRODUCTION
A moral dilemma arises when an agent is in a choice situation in which

he/she cannot satisfy the dictates of morality. Suppose, for example,
that breaking a promise is absolutely forbidden, i.e., under no circum?
stances is it permissible to break a promise. Suppose that this morning
I promised my wife that I would phone her exactly at 5:00, but that
(due to a lapse in memory) I later promised a friend that I would
phone him exactly at 5:00. Here I am just before five o'clock, and I
have only one phone in front of me. I can phone my wife or I can
phone my friend, but I can't phone both at exactly 5:00. Since
promise-breaking is absolutely forbidden, and I have promised to
phone each at exactly 5:00, no matter what I do I will fail to satisfy the
dictates of morality. I am, it seems, in a moral dilemma.
In this example I find myself in a dilemma because of my previous
actions (making two promises which it is often impossible to jointly
satisfy). Dilemmas can (at least apparently) arise without being due to
an agent's previous actions. Suppose, for example, that it is forbidden
to kill one's parents and forbidden to allow them to die. A dilemma
would arise in a situation in which unless one kills one's mother, she
will kill one's father. In such a situation it would be forbidden to kill
one's mother, but also forbidden to do anything else (since that would
allow one's father to die).
In recent years the problem of moral dilemmas has received the
attention of a number of philosophers. Some authors1 argue that moral
Erkenntnis 30: 301-318, 1989.

? 1989 Kluwer Academic Publishers. Printed in the Netherlands.
This content downloaded from 128.206.9.138 on Wed, 08 Feb 2017 15:10:03 UTC
All use subject to http://about.jstor.org/terms
302 PETER VALLENTYNE
dilemmas are not conceptually possible (i.e., that they are incoheren
given the nature of the concepts involved) because they are ruled o
by certain valid principles of deontic logic. Other authors2 insist t
moral dilemmas are conceptually possible, and argue that there
the principles of deontic logic that rule them out must be rejected
In arguing for or against the conceptual possibility of moral dile
mas authors have been almost exclusively concerned with oblig
dilemmas, i.e., situations in which more than one action is obligato
Almost no one has been explicitly concerned with prohibition dilem
mas, i.e., situations in which no feasible action is permissible? I sha
argue that the two types of dilemmas are distinct, and that a m
stronger case can be made against the conceptual possibility of
gation dilemmas than against the conceptual possibility of prohibit
dilemmas.
2. THE POSSIBILITY OF PROHIBITION DILEMMAS

FOR ACTION TOKENS
Talk about actions is ambiguous as to whether it concerns actio

tokens or action types. Action tokens are particulars, they are p
formed by a particular agent, at a particular time (or interval of tim
can be performed at most once, and are not the sort of things that a
instantiated. Action types (such as "a going to the store") are un
versal in that in general they can be instantiated by a number
different action tokens.4
Let us start by focussing our attention on the deontic status
action tokens. Later we shall consider the deontic status of act
types.
Token prohibition dilemmas are choice situations in which no feasi?
ble action token is permissible. The choice situations described at the
beginning of this paper are, at least apparently, examples of token
prohibition dilemmas. In the first example, every feasible action token
is a promise-breaking, and therefore forbidden. In the second exam?
ple, each feasible action token is either a killing or an allowing to die
of one of one's parents, and is therefore forbidden.
The question, then, is whether token prohibition dilemmas are
conceptually possible. My claim is that there is nothing in the logic of
the deontic concepts that rules them out. That this is so can be seen by
reconsidering the first example of a moral dilemma, and taking the
TWO TYPES OF MORAL DILEMMAS 303
relevant normative system to be - not that of morality, but rather -

that of the rules of a particular club. Because it is conceptually
possible that promise-breaking be absolutely forbidden according to a
club's rules, and that someone make conflicting promises, club rule
prohibition dilemmas are conceptually possible. The fact that situa?
tions can arise in which no feasible action token is judged permissible
is an unattractive feature of the absolute prohibition against promise
breaking. On careful reflection we might not choose such a rule to
govern our behavior. Still, there is nothing inconsistent about it. It
does not issue contradictory directives. It does not, for example, hold
that in the above choice situation some action token is both per?
missible and not permissible (it holds that no action is permissible). It
merely holds that no feasible action is permissible in the given choice
situation. Thus, since the above choice situation is conceptually pos?
sible, it is conceptually possible for there to be token prohibition
dilemmas, at least for club rules.
Now, this does not show that moral token prohibition dilemmas are
conceptually possible. It may be that there is something about morality
(as opposed, e.g., to club rules) that rules out their possibility. It may
be, for example, that there is something about morality that rules out
the conceptual possibility of promise-breaking being absolutely for?
bidden. The logic of deontic concepts does not, however, rule out this
possibility, and so the burden of proof is on those who wish to argue
against the conceptual possibility of moral token prohibition
dilemmas.
One argument against the conceptual possibility of moral token
prohibition dilemmas might go as follows. An action token is morally
permissible just in case it is a best (most reasonable) feasible action
from a certain (e.g., fully informed, impartial, benevolent) point of
view. (Different moral theories will give different accounts of what it is
that makes an action a best action.) Action tokens that are not best
actions are forbidden. Because, in any given choice situation there will
be at least some feasible action tokens that are best, there will be at
least some actions that are permissible. Token prohibition dilemmas, it
seems, cannot arise on this account.
One problem with this account is that if it is to rule out token
prohibition dilemmas, it must be assumed that action tokens of a given
choice situation are always comparable in the sense that for any two
actions, at least one of them is at least as good as the other. For if
some action tokens are incomparable, there is no guarantee that

is at least one best action (i.e., action that is at least as good
alternative). It might be that there are only maximally good
(i.e., that are not less good than any alternative) incomparable to
other. If this were to be so, then in some choice situations no
token would be permissible on the above account. Of course, it m
be replied that the proper account of morality is rather that an
token is morally permissible just in case it is a maximally good
from the specified point of view. On this account to be perm
actions need only be such that no alternative action token is b
They need not be at least as good as any alternative. Becaus
will always be at least one maximally good action, this accoun
rule out token prohibition dilemmas. But what needs to be argu
why this and not the previous account is the correct one.
Of course, there are all sorts of other accounts of morality wh
not rule out moral prohibition dilemmas. To mention but one m
still in vein similar to the above, it might be suggested that an
token is morally permissible just in case there are all things cons
good reasons from the specified point of view for performing i
account does not rule out token prohibition because it does n
out the possibility of there being some choice situations in which
are no action tokens for which all things considered there ar
reasons to perform them. There might be some choice situat
which there are good reasons, all things considered, against perf
ing each of the feasible action tokens. Such situations would be
prohibition dilemmas.
The general point here is that although some conceptions of m
ity rule out the possibility of token prohibition dilemmas, many d
Because the nature of morality is a highly contested issue, it
simple matter to determine whether moral token prohibition dile
are conceptually possible. What is clear, however, is that the
cannot be settled simply by appealing to the logic of deontic con
It is rather the nature of morality that is relevant.
Because it will be relevant below, let us note that it is conceptu
possible for there to be choice situations in which no action - fe
or not - is permissible. For consider a club that has a ru
prohibits men from sitting when a woman is in the club r
Suppose, that the members realize that the rule is sexist, and
not only to repeal it, but - to break the members' old habits - t
a new rule prohibiting men from being in any position other than a
sitting position when a woman is in the club room. Suppose further
that they pass the new rule, but forget to repeal the old one. Then,
when a man is in the presence of a woman in the club room every
action - feasible or not - is prohibited, since every action either puts
or keeps the man in a sitting position, or puts or keeps him in some
other position. Again, these rules do not issue contradictory directives
(e.g., that some action is permissible and that it is not). It merely
prohibits everything in that choice situation. So the logic of deontic
concepts do not rule out even this strong form of prohibition dilemma.
Token prohibition dilemmas, then, are conceptually possible for at
least some normative system (e.g., club rules). There may be some?
thing special about morality that makes moral token prohibition
dilemmas conceptually impossible, but so far we have been given no
compelling reason to believe that this is so.
3. THE IMPOSSIBILITY OF OBLIGATION DILEMMAS

FOR ACTION TOKENS
A token obligation dilemma is a choice situation in which more tha

one action token is obligatory. I shall argue that token obligatio
dilemmas are not conceptually possible.
First, however, I need to distinguish obligations from what I shall
call 'quasi-obligations'. An action token is obligatory just in case it is
permissible and no alternative to it is permissible (i.e., it is permissib
to perform it, and wrong to omit it). An action token is qua
obligatory just in case no alternative to it is permissible (i.e., it i
wrong to omit it). Unlike obligation, quasi-obligation does not enta
permissibility. Quasi-obligatory actions are obligatory only if they ar
permissible.
A token quasi-obligation dilemma is a choice situation in which
more than one action token is quasi-obligatory. I shall argue that
token quasi-obligation dilemmas are conceptually possible (at least for
some normative systems), but that token obligation dilemmas are not.
The distinction between obligation and quasi-obligation collapses, if
one assumes that in every choice situation there is at least one
permissible action. For if there is at least one permissible action, and
some action is quasi-obligatory (such that no alternative is per?
missible), then the quasi-obligatory action must be permissible (since
some action is permissible). Consequently, the quasi-obligatory actio

is obligatory. We have seen, however, the logic of the deon
concepts does not guarantee that there is always at least
permissible action, and so the distinction between quasi-obligation a
obligation is genuine.
Given that it has been generally assumed that there is always at l
one permissible action, the distinction between obligation and q
obligation has not been clearly made. I shall argue below that
above definition of obligation best explicates our intuitive notio
obligation, but even if this claim is rejected, the distinction betwee
the two sorts of obligation-like dilemmas remains important. For t
argument that I shall give below will establish that obligation-
dilemmas are not conceptually possible for the strong concept
obligation (what I call 'obligation'), but that they are possible for th
weak concept of obligation (what I call 'quasi-obligation'). M
defenders of "obligation" dilemmas, I suggest, are defenders of qua
obligation dilemmas, not of obligation dilemmas (in my stipul
sense).
That token quasi-obligation dilemmas are conceptually possible
follows immediately from the conceptual possibility of token pro?
hibition dilemmas, for the two are equivalent whenever there are at
least two feasible actions. For, assuming there are at least two feasible
actions, any choice situation in which all actions are prohibited is a
choice situation in which each action is quasi-obligatory (since all its
alternatives are prohibited), and vice versa. Since token prohibition
dilemmas are conceptually possible, for at least some normative sys?
tems, so are token quasi-obligation dilemmas. Again, there may be
something about morality that rules out the possibility of such dilem?
mas for morality, but so far no compelling reason has been given.
Token obligation dilemmas, however, are not conceptually possible.
For if the agent has an obligation to perform an action token, acl, and
an obligation to perform an alternative, ac2, then, by the first obliga?
tion, acl is permissible and no alternative to acl is permissible, and so
ac2 is not permissible.5 But by the second obligation ac2 is per?
missible, which yields a contradiction. Since the assumption that token
obligation dilemmas are conceptually possible yields a contradiction,
this shows that they are not conceptually possible.
Agents can, of course, have conflicting prime facie obligations.
Clearly, situations can arise in which relative to some subset of the
relevant moral considerations one action is obligatory, and relative to

some other subset of the relevant moral considerations another action
is obligatory. That is not the issue here. We are concerned with
obligations in the sense of that which all things considered the agent
ought to do. There can be no token obligation dilemmas in that sense.
This impossibility result follows immediately from the definition of
obligation as "permissible and no alternative action is permissible". It
might be objected that our intuitive notion of obligation is not
captured by this definition, and so the impossibility result is irrelevant
to the issue at hand. There are two relevant possibilities here: either it
is denied that obligatory actions must be permissible, or it is denied
that they must be such that no alternative action is permissible. Let us
consider these separately.
Can an action be obligatory but not permissible? Not if our intuitive
deontic classification scheme has a structure parallel to that of our
classification scheme for possibility and necessity. The category of the
possible has the two mutually exclusive and exhaustive subcategories
of the contingent and the necessary. All necessary things are possible
things. Our category of the permissible has the two mutually exclusive
and exhaustive categories of the optional and the obligatory. An action
is optional just in case it is permissible to perform it and also permiss?
ible to perform some alternative. An action is obligatory just in case it
is permissible to perform it but not permissible to perform any alter?
native. To deny that obligatory actions must be permissible is to deny
that the optional and the obligatory are two mutually exclusive and
exhaustive subcategories of the permissible, and that is not plausible.
This point can be made in the following slightly different manner.
To say that an action is obligatory is to say that all things considered
the agent ought to perform that action. How could it possibly be true
that an agent ought to perform some action that is not permissible? If
it is not permissible, surely it's not the case that the agent ought to
perform it.
So the first line of attack fails. What about the second line of attack,
which denies that obligatory action tokens must be such that no
alternative action token is permissible? The most plausible (although
not the only logically possible) defense of this denial seems to be the
following. On my definition of obligation (permissible and no alter?
native action is permissible) obligation is a comparative matter in that
whether or not an action is obligatory depends on the permissibility of
the actions that are its alternatives. But that, it might be claime
wrong. Obligation is a non-comparative matter. Obligations a
"directly", so to speak, to actions and not merely "derivative
those cases where it contingently turns out that no other act
permissible. Those actions that are obligatory are so, it migh
claimed, independently of the moral status of their alternatives.
The problem with this line of attack is that of making sense of
claim that obligations attach "directly" to actions. Consider,
example, the injunction "Always keep your promises!". Does ob
tion as judged by this principle attach directly to an action m
because it is a promise-keeping! I think not. To see this, suppose t
have promised my wife to phone her exactly at 5:00, and tha
before 5:00 there are two phones in front of me. I can phone her
the black phone, or I can phone here on the red phone. Are ea
these action tokens obligatory merely because they are way
keeping of promise? Surely not! The injunction "Always keep
promises!" does not mean that every action token that fulfi
promise is obligatory. It only means that any action token w
violates a promise is forbidden. Often there will be many wa
fulfilling a promise, and to think that each of them is obligatory
absurd. The injunction "Always keep your promises!" doe
"directly" attach obligations to any particular actions. It only att
obligations contingently in those cases where there is only one ac
token that fulfills the promise.
This is, of course, but one example, but the general idea shou
clear. Moral injunctions do not directly determine which act
tokens are obligatory. They only determine which actions tokens
permissible. An individual action is obligatory just in case it is th
permissible action token in the choice situation.
Stated somewhat differently the argument is this: An action tok
obligatory only if it is wrong to omit it. To omit an action token
perform one of its alternatives. So, an action token is obligatory o
all its alternatives are forbidden.
Thus, the second line of attack against my argument that t
obligation dilemmas are not conceptually possible also fails.
about the examples of moral dilemmas with which I began the pa
Are those not examples of token obligation dilemmas? They ar
In those examples no action token was permissible, and theref
action token was obligatory. The examples are examples of prohib
dilemmas, and of quasi-obligation dilemmas (since all the alternatives

to each token are prohibited).
So token obligation dilemmas are not conceptually possible for any
normative system, whereas token quasi-obligation dilemmas are con?
ceptually possible at least for some systems. The importance of this
result remains even if one denies that the given definition of obligation
captures our intuitive notion. Whatever they are called, strong obliga?
tion-like dilemmas (what I call 'obligation dilemmas') are not possible,
whereas weak obligation-like dilemmas (what I call 'quasi-obligation
dilemmas') are not.
With respect to tokens, then, both prohibition and quasi-obligation
dilemmas (which are equivalent whenever there are at least two
feasible actions) are conceptually possible, but obligation dilemmas
are not.
4. THE POSSIBILITY OF PROHIBITION DILEMMAS

FOR ACTION TYPES
So far we have been considering the possibility of dilemmas for actio

tokens. Most of the discussion that has taken place in the literature ha
concerned action types, so let us turn our attention to that issue.
A type prohibition dilemma is a choice situation in which no feasibl
action type is permissible. Because the logic of deontic concepts does
not guarantee that in any choice situation some feasible action i
permissible, it follows that it does not rule out the conceptual po
sibility of type prohibition dilemmas. For, since an action type i
feasible just in case some action token of that type is feasible, and an
action type is permissible just in case some action token of that type i
permissible, it follows immediately that type prohibition dilemmas are
conceptually possible just in case token prohibition dilemmas are.
Since the latter are conceptually possible, so are the former. Further
more, since strong token prohibition dilemmas (i.e., situations in whi
no token - feasible or not - is permissible) are conceptually possible,
so are strong type prohibition dilemmas (i.e., situations in which not
action type - feasible or not - is permissible).
This has important implications for deontic logic. For a principle of
standard deontic is P(p) v P(~p), which just says that (in any give
choice situation) some action is permissible. As we have seen, if w
were devising a normative system, we would try to make it satisfy th
principle, but the examples of this paper illustrate that the notio
permissibility does not guarantee the truth of this principle.
must give up this part of standard deontic logic.6
Thus, type prohibition dilemmas are conceptually possible
respect to club rules (e.g. if they forbid promise-breaking). Wheth
not they are conceptually possible for the moral case depend
whether there is something about the concept of morality that r
them out. As for the case of action tokens, no cogent argume
yet been given to rule them out.
5. OBLIGATION DILEMMAS FOR ACTION TYPES
A type obligation (quasi-obligation) dilemma is a choice situati

which there are two action types, each of which is obligatory
obligatory), the conjunction of which is not feasible. I shall argue
quasi-obligation dilemmas are conceptually possible, at least fo
normative systems, but that - with an important qualification - o
tion dilemmas are not.
Here and below the following assumptions will be used conce
action types. (1) An action type is feasible for a given agent in a
choice situation if and only if some action token of that type is f
for that agent in that choice situation.7 (2) An action type is
missible (forbidden) for a given choice situation if and only if
(no) action token of that type is permissible in that choice situatio
An action type is obligatory for a given choice situation if and onl
is permissible and its negation is not.8 (4) An action type is
obligatory for a given choice situation if and only if its negation
permissible. (As for action tokens obligation implies quasi-obligat
but not vice versa.)
The arguments provided above for the claim that my stipu
definition of obligation captures our intuitive notion of obligatio
apply here, and so I shall not repeat them. Again, anyone unconv
by these arguments can simply substitute 'obligation in the s
sense' for what I call 'quasi-obligation'.
Type obligation dilemmas are situations in which more tha
action type is obligatory, but the conjunction of the actions type
not feasible. A strong type obligation dilemma is a choice situatio
which the obligatory action types are conceptually incompatible (
as "phoning Toronto" and "not phoning Toronto"). A weak
obligation dilemma is a choice situation in which the obligatory a
types are conceptually compatible, and merely incompatible given the

circumstances (such as "keeping my promises" and "not phoning
Toronto", when I have promised to phone Toronto). Let us consider
each case separately.
5.1. The Impossibility of Strong Type Obligation Dilemmas
Consider first the case where there are two obligatory action types,
one of which is the negation of the other. Such dilemmas are not
conceptually possible, since if a given action type, i, is obligatory, it is
permissible, and its negation, ~ t, is not. So, ~ t cannot also be obliga?
tory, for if it were, ~ t would be permissible, which is a contradiction.
So, strong obligation dilemmas where both t and " t (for some i) are
obligatory are not possible. The argument here is exactly parallel to
that against the possibility of obligation dilemmas for action token.
But what about the case where both t and s obligatory, and t and s
are conceptually incompatible, but s is not identical with ~t (e.g.,
suppose s is equivalent to ~t&r, for some r). The standard argument,
which I shall endorse, against the possibility of such obligation
dilemmas is this, where 'Pos' and 'Nee' designates conceptual pos?
sibility and necessity, respectively:
Al. Ob(i) & Ob(r) & Tos(i & r)

A2. ~Pos(?& r)-*Nec(i->~r)
A3. [Ob(i) & Nec(i^ >)]-* Obfr)
A4. Ob(r)^>P(r)&~?(~r)
A5. P(f)&~P(r)
A1-A4 lead to a contradiction as follows: From A1-A3 (focussing
on Ob(0) we get Ob(~r). Applying A4 to Ob(~r) (substituting '~r' for
V in A4) we get P(~r). But from Al and A4 (focussing on Ob(r)) we
get ~F(~r), which give us the contradiction. So one of the premisses
must be given up.
Al is the supposition that a strong obligation dilemma obtains, and
must, I shall argue be rejected. A2 is acceptable, since it is the
standard principle of modal logic that says if t & r is not conceptual
possible, then it is conceptually necessary that if t holds, then " r holds.
A3 is demonstrably acceptable by the following argument: If t is
obligatory, then some token of type t is permissible and no token of
type ~t is permissible. But if being of type t conceptually entails being
of type ~r, then it follows that some token of type " r is permissib
no token of type r (i.e., ~~ r) is permissible. It follows, that is, th
obligatory. So A3 is acceptable. Finally, A4 follows immediately f
the definition of obligation. Consequently, it is Al, the suppo
that a strong type obligation obtains, that must be rejected.
type obligation dilemmas are not possible.
Strong quasi-obligation dilemmas, on the other hand, are co
tually possible. Note first that the above argument does not app
quasi-obligation, since the counterpart of A4 is false for quasi-ob
tion. Quasi-obligation does not imply permissibility. Second, the
sibility of type quasi-obligation dilemmas, at least for some norm
systems, follows immediately from the possibility of strong
prohibition dilemmas. In such situations no token is permissib
consequently every action type, t, is quasi-obligatory (i.e., such t
token of type "t is permissible). In particular, both t and " t
quasi-obligatory. So, once again, we see that quasi-obligatory
mas are conceptually possible for some normative systems. W
moral quasi-obligation dilemmas are possible depends on wh
moral prohibition dilemmas are possible, and the logic of de
concepts does not settle that issue.
6. THE QUALIFIED IMPOSSIBILITY OF WEAK TYPE

OBLIGATIO DILEMMAS
So far we have been considering the possibility of two con

incompatible action types being obligatory. A more interestin
that where two conceptually compatible action types are obliga
are empirically incompatible given the laws of nature and the
stances. Let us now consider the case of such weak type ob
dilemmas.
The usual argument against the conceptual possibility of typ
gation dilemmas takes the following form, where 'Feas(i)' d
that type t is feasible (i.e., some token of that type is em
possible for the agent in the circumstances):
Bl: Ob(il) & Ob(?2) ~Feas(?l & ti)
B2: [Ob(?l)&Ob(?2)]-?Ob(?l & ti)
B3: Ob(?l & ?2)-*Feas(?l & ti)
B4: Feas(?l & ti) & ~Feas(?l & ti)
Bl is simply the supposition that a weak type obligation dilemma

obtains. B2 follows from the principle of deontic distribution, an
axiom of standard deontic logic, according to which the conjunction
of two obligatory action types is also obligatory. B3 follows from the
principle that ought implies can. B4 is a contradiction derivable
directly from B1-B3.
Since the conjunction of B1-B3 yields a contradiction, one of the
three premisses must be given up. Opponents of the conceptual
possibility of obligation dilemmas take B2 and B3 to be unobjection?
able, and therefore conclude that Bl - the supposition that an obliga?
tion dilemma arises - is demonstrably false. The crucial part of their
argument is therefore the claim that B2 and B3 are unobjectionable.
Let us examine them.
To see that B2 is unobjectionable, it suffices to remember how the
deontic status of action types if related to that of action tokens. An
action type is permissible just in case some action token of that type is
permissible. An action type is obligatory just in case that type is
permissible and its negation is not. Thus, if il is obligatory, some
action token of type il is permissible, and no action token of type ~ il
is. Likewise, if i2 is obligatory, some action token of type i2 is
permissible, and no action token of type ~ tl is. So, if both il and i2
are obligatory, then no action token of type ~(il & i2), is permissible.
Furthermore, from that, the fact that some token of type il is
permissible, and the fact that some token of type i2 is permissible, it
follows that some action of type il & i2 is permissible. Therefore,
action type il & i2 is permissible and action type ~(il & i2) is not;
that is, action type il & i2 is obligatory. Thus, if il and i2 are each
obligatory, then so is their conjunction. So B2 is okay.
The situation with B3 is much more complicated. The truth of B3
depends crucially on the point of view that the deontic concepts
represent. Here we are concerned with the realistic point of view, that
is, with what the agent may or ought to do given the circumstances of
a given choice situation (as opposed to what ideally may or ought to
be the case). Some might claim that there are at least two different
realistic points of view. From the nomic point of view an action is
permissible in a given choice situation just in case it would not violate
the moral law given the circumstances - whether or nor it is feasible.
From this point of view neither "may" nor "ought" implies "can".
From the deliberative point of view an action is permissible just in case
it is feasible and would not violate the moral law given the circum
stances. From this point of view "may" and "ought" do imply "can"
Which of these two points of view corresponds to the intuitive
realistic point of view? One argument in favor of the deliberative
viewpoint is that it does not seem appropriate to say 'You may [or
ought] to do X [e.g., stop an oncoming train with your bare hands]
when it is not feasible for you to do X. If it is permissible or obligatory
it must, it seems, be feasible.
The problem with this argument, it might be suggested, is that i
overlooks the possibility that the feasibility of an action may merely b
a conversational implicature of the speech act of claiming that th
action is permissible - as opposed to part of the content of the clai
itself. It overlooks, that is, the possibility that the feasibility of the
action is merely something that must be supposed to be true, if th
speech act is to be appropriate (e.g., relevant) - not something that is
literally said.9
This general point raised by this objection is a good one. It is
unclear, however, that it will work in the present case. It assumes tha
appropriate sense can be made of the notion of permissibility from the
nomic viewpoint. But what exactly might it mean to say that in a given
choice situation an infeasible action does (or does not) violate the
moral law?
This notion should not be confused with that of counterfactual
permissibility, according to which an action is counterfactually per?
missible just in case if it were feasible, then it would (or might) b
permissible. This is of no help here, because from the realistic view?
point we are concerned with permissibility given the circumstances.
That an action is permissible in some other choice situation - no
matter how similar it is to the given one - is of no relevance to it
permissibility in the given choice situation. The problem is that in
general the permissibility of a given action depends on the exact
circumstance of the choice situation, and that determines which
actions are feasible. Therefore there does not seem to be any basis for
claiming that certain infeasible actions do or do not violate the moral
laws, and so it is far from clear that appropriate sense can be made of
the nomic viewpoint.
The issue is, of course, rather complex. Although I am skeptica
about the prospects of adequately explicating the notion of the nomic
point of view, I cannot here defend that skepticism. The best we can
here conclude is therefore B3 is true from the deliberative viewpoint
but not from the nomic viewpoint.
Where, then, does this leave us with respect to the argument

[B1-B4] that type obligation dilemmas are not conceptually possible?
Because B1-B3 lead to a contradiction, and for the deliberative
viewpoint, B2 and B3 are conceptually true, we can conclude that Bl
- the supposition that a type obligation dilemma arises - must be
conceptually false from the deliberative viewpoint. For the nomic
viewpoint - assuming appropriate sense can be made of that viewpoint
- B3 is not a conceptual truth, and so the above argument does not
show that type obligation dilemmas are not conceptually possible from
the nomic viewpoint.10
Note that the examples with which we started (where promise
breaking, and the killing or allowing to die of one of one's parents are
repectively forbidden) are not examples of type obligation dilemmas
from the deliberative viewpoint. In these examples no feasible action
token is permissible. Thus, since from the deliberative viewpoint an
action type is permissible just in case some feasible action token of
that type is permissible, no action type is permissible, nor, a fortiori,
obligatory. Consequently, there are from the deliberative viewpoint no
conflicting obligatory action types.
They are, however, cases of type quasi-obligation dilemmas.
Keeping my promise to my wife is quasi-obligatory (since not keeping
my promise to her is forbidden), as is keeping my promise to my
friend. But since neither of these actions types is permissible, neither
of them is obligatory. The quasi-obligation counterpart of argument
B1-B4 is unsuccessful, because B3, the principle that ought implies
can, is not true of quasi-obligation - even from the deliberate view?
point. Too see this consider a situation in which a token prohibition
dilemma arises (i.e., where not feasible action token is permissible). In
such a situation the action type "moving at less than or equal to the
speed of light" is not permissible (since no token of that type - or any
other feasible type - is permissible). Consequently, the type "not
moving at less than or equal to the speed of light", i.e., "moving at
faster than the speed of light", is quasi-obligatory. Given (we may
assume) that it is not feasible for an agent to move at faster than the
speed of light, this shows that something can be quasi-obligatory, but
not feasible - even from the deliberative viewpoint.
To sum up: Strong type obligation dilemmas (where two concep?
tually incompatible action types are each obligatory), are not possible.
Nor are weak type obligation dilemmas (where two conceptually
compatible, but empirically incompatible, action types are each obli
gatory) from the deliberative viewpoint. Weak type obligation dil

mas may, however, be conceptually possible from the nomic viewp
(if sense can be made of that viewpoint). Strong and weak qu
obligation dilemmas are conceptually possible - for at least s
normative systems - from both the nomic and the deliberative vi
point.
7. CONCLUSION
In summary, except for the case of weak type obligation dilemmas

from the nomic viewpoint, obligation dilemmas - token or type, strong
or weak - are not conceptually possible for any normative system.
Prohibition dilemmas and quasi-obligation dilemmas (which are
effectively equivalent), on the other hand, are not ruled out by the
logic of deontic concepts, and so for at least some normative systems
(e.g., club rules) they are conceptually possible. There may be some?
thing about morality that rules them out for the moral case, but so far
no compelling argument has been given to this effect.
The key move in my argument was to establish that it is concep?
tually possible for at least some normative systems for no action token
to be permissible in a given choice situation. Having established that,
we saw that the notions of obligation and quasi-obligation come apart
in exactly such situations. Because this has not been generally recog?
nized, proponents of "obligation" dilemmas have not adequately dis?
tinguished between the two sorts of "obligation" dilemmas. Dis?
tinguishing between the two allows us to see that two principles that
proponents of dilemmas often reject - "ought implies may", and
"ought implies can" - are true (from the deliberative viewpoint, at
least) of obligation, but not of quasi-obligation. Most proponents of
moral dilemmas are best understood, I suggest, as proponents of
quasi-obligation dilemmas. But - assuming there is more than one
feasible action - quasi-obligation dilemmas arise when and only when
prohibition dilemmas arise. So the case for the possibility of moral
dilemmas ultimately rests on the possibility of moral prohibition
dilemmas.
Given that (except perhaps from the nomic viewpoint) obligation
dilemmas are not possible for any normative system, far too much
attention has been focussed on moral obligation dilemmas. The key
question for morality is whether prohibition (or, if you like, quasi
obligation) dilemmas are possible. Unlike the case for moral obligation
dilemmas, the conceptual possibility of more prohibition dilemmas

cannot be settled simply by appealing to the logic of deontic
concepts.11 The issue is much broader: it concerns the very nature of
morality.12
NOTES
1 For example: Terrance McConnell, 1976, 'Moral Dilemmas and Requirin

Impossible', Philosophical Studies 29, 409-413; Terrance McConnell, 1978, 'M
Dilemmas and Consistency in Ethics', Canadian Journal of Philosophy 1978,
287; Earl Connee, 1982, 'Against Moral Dilemmas', Philosophical Review 91,
2 For example: E. J. Lemmon, 1962, 'Moral Dilemmas', Philosophical Review
139-158; Bernard Williams, 1965, 'Ethical Consistency', Proceedings of the Aristot
Society, Supplementary Volume 39, pp. 103-24 (reprinted in Bernard Williams,
lems of the Self); Roger Trigg, 1971, 'Moral Conflict', Mind 80, 42-55; Bas
Fraassen, 1973, 'Values and the Heart's Command', Journal of Philosophy 70, 5-1
Ruth Barcan Marcus, 1980, 'Moral Dilemmas and Consistency', Journal of Philo
77, 121-136.
3 Two notable exceptions: Patricia S. Greenspan, 1983, 'Moral Dilemmas and Guilt',
Philosophical Studies 43, 117-125, and G. H. von Wright, An Essay in Deontic Logic
and the General Theory of Action (Amsterdam: North Holland Publishing Company,
1968), pp. 78-81.
4 I remain neutral on the exact nature of act tokens. I assume only that, whatever their
nature, act tokens can be distinguished from act types.
5 Two action tokens are alternatives only if they have the same agent, the same time of
performance, and are incompatible. Note also that because action tokens are particulars,
it does not make sense to apply logical operations (negation, conjunction, etc.) to them
(just as it does not make sense to apply, for example, negation to a particular chair).
For further discussion of these matters, see Lars Bergstr?m, 1976, 'On the Formulation
and Application of Utilitarianism', Nous 10, 121-144; J. Howard Sobel, 1971, 'Value,
Alternatives, and Utilitarianism', Nous 5, 373-384; and J. Howard Sobel, 1972, 'The
Need for Coercion', in Coercion, edited by J. R. Penncock and J. W. Chapman (New
York: Aldine-Atherton, 1972); 148-147, esp. sec. 1.1.
6 In Peter Vallentyne, 1987, 'Prohibition Dilemmas and Deontic Logic', Logique et
Analyse 117-118, 113-121,1 discuss in detail the implications of revising deontic logic
so as not to rule out prohibition dilemmas.
7 An action token is feasible in a given choice situation just in case, given the actual
circumstances and laws of nature, it could be performed by the agent.
8 Note that an action type can be obligatory without any token of that type being
obligatory; namely, when more than one token of that type is permissible, but no token
not of that type is. Thus, the fact that there cannot be token obligation dilemmas does
not entail anything about the possibility of type obligation dilemmas.
9 After writing the first draft of this paper, it was pointed out to me that Walter
Sinnott-Armstrong, 1984, "'Ought' Conversationally Implies 'Can'", The Philosphical
Review 93, 249-261, has argued just this point. For the reasons that follow in the text, I
remain neutral on the question of whether 'ought' implies 'can'.
10 Note that my earlier arguments with respect to the possibility of dilemmas for action
tokens applies for both the deliberate and the nomic viewpoint. There can be no
obligation dilemmas for action tokens because there cannot be more than one action
token that is obligatory - whatever the viewpoint. Likewise, the logic of deontic
concepts does not rule out prohibition dilemmas for action tokens because they do not
guarantee that at least one feasible action token is permissible - whatever the viewpoint.
Note, however, that from the nomic viewpoint there can be obligation dilemmas of a
different sort than defined. From the nomic point of view it is conceptually possible that a
choice situation arises in which an obligatory action is not feasible. This might be called
'an infeasible obligation dilemma'. The obligation dilemmas defined in the text would be
better called 'conflicting obligation dilemmas'. From the nomic viewpoint infeasible
obligation dilemmas are conceptually possible for both action tokens and action types.
11 Geoff Sayre McCord, 1986, 'Deontic Logic and the Priority of Moral Theory', Nous
20, 179-97, argues that the possibility of moral dilemmas cannot be ruled out by deontic
logic, on the ground that there are no neutral principles of deontic logic (they all reflect,
he claims, substantive conceptions of morality). My position is thus intermediate
between Sayre McCord's and the more usual one: some principles of deontic logic are
neutral and do rule out certain types of moral dilemmas (e.g., obligation dilemmas), but
other principles (e.g., Pp\j P"p) are not so neutral, and so cannot be used to rule out
other types of moral dilemmas (e.g., prohibition dilemmas).
12 For helpful comments on an earlier version of this paper I want to thank: Robert
Audi, Kurt Baier, John Baker, Bob Binkley, Dick Bronaugh, Earl Conee, Don Hubin,
Andrew Jones, Shelly Kagan, Geoff Sayre McCord, Mark Vorobej, and Michael
Zimmerman.
Manuscript received 20 March 1987
Department of Philosophy
Virginia Commonwealth University
Richmond, VA 23284
U.S.A.

Two Types of Moral Dilemmas

Uploaded by

Copyright:

Available Formats

Two Types of Moral Dilemmas

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Two Types of Moral Dilemmas

Uploaded by

Copyright:

Available Formats

PETER VALLENTYNE

TWO TYPES OF MORAL DILEMMAS

A moral dilemma arises when an agent is in a choice situation in which

Erkenntnis 30: 301-318, 1989.

2. THE POSSIBILITY OF PROHIBITION DILEMMAS

Talk about actions is ambiguous as to whether it concerns actio

relevant normative system to be - not that of morality, but rather -

some action tokens are incomparable, there is no guarantee that

3. THE IMPOSSIBILITY OF OBLIGATION DILEMMAS

A token obligation dilemma is a choice situation in which more tha

some action is permissible). Consequently, the quasi-obligatory actio

relevant moral considerations one action is obligatory, and relative to

dilemmas, and of quasi-obligation dilemmas (since all the alternatives

4. THE POSSIBILITY OF PROHIBITION DILEMMAS

So far we have been considering the possibility of dilemmas for actio

5. OBLIGATION DILEMMAS FOR ACTION TYPES

A type obligation (quasi-obligation) dilemma is a choice situati

types are conceptually compatible, and merely incompatible given the

5.1. The Impossibility of Strong Type Obligation Dilemmas

Al. Ob(i) & Ob(r) & Tos(i & r)

6. THE QUALIFIED IMPOSSIBILITY OF WEAK TYPE

So far we have been considering the possibility of two con

Bl is simply the supposition that a weak type obligation dilemma

Where, then, does this leave us with respect to the argument

gatory) from the deliberative viewpoint. Weak type obligation dil

In summary, except for the case of weak type obligation dilemmas

dilemmas, the conceptual possibility of more prohibition dilemmas

1 For example: Terrance McConnell, 1976, 'Moral Dilemmas and Requirin

Manuscript received 20 March 1987

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.