Box 140, Item 1: Draft of Moral dilemmas and the logic of deontic notions ; Draft of Moral dilemmas ; Notes, correspondence and cutting on moral dilemmas and deontic logic

Box 140, Item 1: Draft of Moral dilemmas and the logic of deontic notions ; Draft of Moral dilemmas ; Notes, correspondence and cutting on moral dilemmas and deontic logic

Typescript of drafts of two papers, undated. Second paper with handwritten emendations, and first paper published, Routley R and Plumwood V (1984) 'Moral dilemmas and the logic of deontic notions', Discussion papers in environmental philosophy, 6. Dept. of Philosophy, Australian National University. Includes three letters addressed to Richard: 27 July 1984 (handwritten, 1 leaf (2 pages)), author unidentified re Moral dilemmas; Undated (handwritten, 2 pages), from Charles re New Studies in Deontic Logic and Castaneda's proposed deontic logic; Undated (handwritten, 7 leaves with annotations by Sylvan?), author unidentified re Moral dilemmas; and one letter from Richard Sylvan? to Philip, undated (handwritten (photocopy), 3 pages), re feedback on a paper. Includes incomplete draft paper, Other attempts to neutralise moral dilemmas (21 leaves), with handwritten emendations, undated.

Letters and cuttings redacted from access file (PDF) due to copyright restrictions. Unnumbered paper from collection, item number assigned by library staff.

The University of Queensland's Richard Sylvan Papers UQFL291, Box 140, Item 1

Antipodean Antinuclearism: (Re)constructing Richard Routley/Sylvan's Nuclear Philosophy

This item was identified for digitisation at the request of The University of Queensland's 2020 Fryer Library Fellow, Dr. N.A.J. Taylor.

For all enquiries about this work, please contact the Fryer Library, The University of Queensland Library.

[87] leaves. 64.35 MB.

Manuscript

MORAL DILEMMAS AND THE LOGIC OF DEONTIC NOTIONS

Richard Routley and Vai Plumwood^

The

§1.

paradoxes

of

deontic

logic

consistency and

the

and

modal requirements.

Especially since mid-century many systems of deontic logic have been advanced, which

logic of such deontic

attempt to explore the
prohibition,

and wrong.

right

all

Almost

which treat deontic notions as modal,

notions as obligation, permission and

modal

not

demand

functors,

All such systems are mistaken m a

For deontic functors such as obligation and prohibition are

but

are

more highly

of

degree

greater

a

systems,

i.e. as if strict or provable equivalents are

intersubstitutable within them preserving truth.
quite fundamental way.

systems have been modal

these

intensional than

propositional

modal

modal

than

discrimination

and so

functors,

logics

can

provide.

The damage caused by the mistaken treatment of deontic notions as modal shows up

in

various ways,

but we

shall be

primarily concerned with only one of these, the

consistency requirements and the exclusion thereby of moral dilemmas.
which the modal damage

appears is through a

Another way in

series of related paradoxes,

such as,

directly, the paradoxes of derived obligation, and, less directly, the paradoxes of
the Robber and the Good Samaritan.

The consistency requirements come out in two ways in mainstream deontic systems.
consistency

Firstly,

a

negation

are

requirement,

-

obligatory

in

the

that

usual

no

claim

such

is

symbolism,

Op

that

~0~p,

both it
i.e.

and its

~(0p

&

0~p)

— appears as a principle, frequently a postulate, in all mainstream absolute
systems.21 Such absolute systems, which are variants on alethic modal systems, are

characteristically obtained from corresponding alethic systems by replacing the modal

axiom,

that’ or

’it ought to be the case that’, or

the functor ’□ ’,
modal

(such as

logics

system demands
world

is

’it is obligatory

’it ought to be seen to that', replaces

’it is necessary that', of alethic systems. We know of course that

consistency requirement;

some

~0~p - where ’O’, read

Op D p, by the deontic axiom, Op

the doxastic

systems)

the standard

but it appears that the intended interpretation of deontic

the consistency

requirement.

or

encapsulates

acceptable

subsequently introduced,

can be designed without

For

its semantical equivalent

what

is

permitted

(i.e.

is that

in

symbols

(Px)STx) which is surely true (for let world x be determined

precisely by what is so permitted).

Moreover,

the

second

related

way

in

which

consistency

escaped by a modal formulation of deontic (or doxastic) logics.

figures

cannot

be so

For it seems clear,

2

and is widely agreed, that 0 satisfies the following distribution principle:
(1)

Op&OqiffO(p&q).

Since however 0 is modal, and p & ~p & q

p&~p (i.e.

-

the sides are strictly

equivalent),

0(p & ~p) iff 0(p & ~p & q)
Thence,

Op & 0~p implies 0(p & ~p), which implies 0(p & ~p & q),

and (2),

by (1)

which implies Oq;

(2)

if there are cases of moral dilemmas where Op and also 0~p,

i.e.

then obligation spreads so that any claim is obligatory.

This argument tells against all mainstream deontic logics;

rock-bottom system, minimal deontic logic.

for it holds in the

This logic MDS, presented by von Wright

and others (see e.g. Hilpinen 81, p.3 and p.5), adds to classical sentential logic CL
(with rules of Material Detachment and Substitutivity duly extended)

the one-place

connective P, of permissibility, subject to just the conditions

P(A v B)

e

(1’)

PA v PB

and
Intersubstitutivity of provably equivalent CL (i.e. truth functional) wff
Now

define

a

connective

-

in

fact

the

standard

obligation

Then, where A and B are any CL wff, 0(A & B)

OA =£j£~P~A.
negation and

It then follows,

given

that

(extensional)

principles to

De Morgan

= ~(~P~A & ~P~B).
follows),

0

moral

claim

is

dilemmas

obligatory,

the

consequence

as above

of

the

nothing

functor

)
*
(2

-

thus:

OA & OB, upon applying

of (1
)
*

that P~(—A & —B)

(or by variants thereupon that which

form

OA

&

0~A,

(extensional)

i.e.

any

permissible.

The

that
is

OB,

revealing variant argument applies the provable CL-equivalence A & -A = B & ~B to get

0(A & ~A) = 0(B & ~B).
d

OB & 0~B => OB.

Then (in condensed notation) OA & 0~A => 0(A & ~A) => 0(B & ~B)

That is, if there are any deontic moral dilemmas (of such forms as

OA & 0~A) then everything is obligatory.

other

forms;

But there are moral dilemmas, of this and

and not everything is obligatory.

incorrect. So too are all its extensions, and

Therefore minimum deontic

logic is

thereby all mainstream systems.

For

'most systems of deontic logic include the system’ of von Wright 'as a subsystem' (F
& H, p.13).

There is little appreciation of the
involve in the ethical

Hare,

having

set

up

a

force of these arguments or of what they

literature. But one prominent example will

dilemma,

and

rejected

one

logical

have to suffice.

route out,

proceeds to

pronounce that 'it is not very helpful to try to sort out these difficulties by some

relatively minor tinkering with the calculus of deontic logic.
The linguistic and
logical intuitions which give rise to them are all right so far as they go' (82,

3

intuitions behind the mainstream systems are not alright, but lead to

The

p.28).

deontic collapse and to a heap of paradoxes; and a bit of '’tinkering" is enough to
remove the worst difficulties.5

(2’) and

The trouble lies we shall argue, just with the modal requirement (i.e.
its usual extension to all wff).

that not everything is obligatory,

evident

For while it is

This leaves much to be defended.

on no more than the strength of a few

examples (such as that it is not obligatory that everyone visit the tallest eucalypt
every day), none of the other assertions is so uncontroversial.

Given that there are such moral dilemmas, as will be argued at length, and given

that sentential logic CL is admissible when applied to consistent situations (which

are all

that are

elsewhere

(in

arguments:

required

RLR

in

especially),

modal

the

namely,

the arguments
there

are

a claim that has been argued

above),

two

only

and

requirement,

one

fault the

which to

points

at

half

of

the

distributivity

principle, specifically
r

(O-adjunction).

OA & OB -> 0(A & B)

But it is bizarre to have to fail O-adjunction in order

OA & 0~A =5 B has
quite

especially when

OB to stand,

0(A & ~A)

to be

correctly

foundation.

And

obligation.

For,

not

it

contemplated

among

in

the

things,

is a

irrelevant mates
7
desperate measure ,

systems

since

it

intended

role

and

mainstream
against

entirely

other

is conceded that its

Faulting O-adjunction

rejected.
goes

it

to allow the paradox

O-adjunction

can

be

argued

is

without

meaning

for

from

of
the

semantics of 0, along these lines:- Roughly OA is true iff A holds in all acceptable

situations.

But if A holds in all such situations and B does also then so does A &

B, validating O-adjunction.

Accordingly, the trouble must be located in the modality

requirements.

This can

be

seen

from another

angle also.

obligation, which also underlies O-adjunction,

An

important logical

feature of

is its transmissibility, that what is

tightly entailed by what is obligatory is also obligatory; in symbolic form,

A ■+ B -o OA -> OB

But now,

(0-transmission).

if entailment is erroneously construed as strict implication, which is what

the modal requirement would have us do, then 0(A & ~A) -> OB, since by a paradox of

strict implication A & ~A
B. If there are moral dilemmas of the form 0(A & ~A), as
will be contended, then this result is sufficiently damaging on its own, and focusses
the trouble on the modal requirement.
properly maligned OA & 0~A ->0B.

Otherwise apply O-adjunction,

to obtain the

4

the

Both

and

consistency

requirements should

modality

be challenged

then by

anyone who claims that, for some p, both Op and 0~p (and that such judgements are not
merely prima facie, vanishing upon further reflection).

Many have argued that such

cases do occur, indeed are central examples of value conflict, and are of fundamental
importance in ethics.$ We want to go even further and claim, firstly, that such cases

occur quite frequently, as cases of conflicting obligations, conflicting rights, and

genuine moral dilemmas - so in particular there are many cases where both p and ~p
ought to be the case - and secondly, that adequate logical treatment of such cases
requires an underlying logic of a distinctively paraconsistent kind.

we shall

cases has,

argue,

with many important arguments and cases.

many of

logic

rendered deontic

Neglecting such

incapable of dealing adequately

The matter is particularly serious since

fiercely debated moral cases are cases of moral dilemmas,

lthe most

e.g.

limited resources, nuclear war, permissible war tactics. While our initial argument
will overlap previous work,9 it goes beyond it by adding a semantical treatment and
moral

of

explanation

semantical

dilemmas,

and

in

setting adequate

deontic logics

within the framework of paraconsistent logic.

simple

The

§2.

resolution

relevant

epicycling.

classical

versus

The

relevant

It has already been indicated.

solution to the base problem is disarmingly simple.

Abandon the modal assumption, that obligation, permissibility, and so on, are modal
This blocks

logical equivalents.

functors allowing replacement of modal

from OCA & ~A), or from OA & 0~A, to 0(B & ~B) and thence to OB.
giving

up

intersubstitutivity

are

replacement
coentailment,

not

simply

just

more

in

contexts;

deontic

rigorous

than

lax

strict equivalence.

intimately

- the Spread

with

What

ones.

is

permitting
required

is

While A & ~A and B & ~B are strictly

equivalent, they do not coentail one another.

connected

modal

It does not involve

conditions

the

the move

law,

That is as much of a fallacy as - one

A

that would have that a

& ~A -> B,

contradiction such as A & ~A entails any statement whatsoever.

An

alternative

satisfactory

distinctive
implciation

deontic

way

then

logic

of

should

explaining
be

in

particular

of

relevant

paraconsistent,

feature of relevant logics, their

and

the

is

solution,

simply

and

through

why

the

a

most

rejection of the paradoxes of strict

Spread.

For,

as

remarked,

one

logical

characteristic of deontic notions such as obligation is their transmissibility; i.e.
where A

entails B then OA is logically sufficient for OB.10 Then were entailment

strict implication it would follow at once by Spread that 0(A & ~A) -> OB, whence by
O-adjunction, OA & 0~A -> OB.

basis

of

interpretations

straightforwardly on

RLR II).

of

Transmissibility may be independently argued for on the
entailment

(e.g.

as

content

inclusion),

and

follows

the semantical analysis of obligation to be advanced (cf. also

5

A good deal

of effort has already been expended in trying to save mainstream

deontic theory from the worst of the difficulties
and consistency

requirements.

Most of

through its modal

it encounters,

the repairs

however

attempted amount

to

epicycling: trying to tack theory-saving devices onto the initial theory, rather than
questioning any fundamental features of it.

Intractable
exposed.

problems

is called

What

the

with

so-called

minimal

the standard system of deontic

system

already

have

been

logic is in an even worse

position than the minimal system, because it adds to the minimal system as an axiom
12

Then any standard system including a moral dilemma, 0(p^ & ~p ), is
13
o
o
simply inconsistent. For Op => Pp,
hence Pp^ & ~Pp-j where p1 is pQ & ~pQ.

~p(p & ~p).

Main marketed deontic logics are epicyclic variations on

(i)

standard deontic

logic,

being modal attempts

to repair

defects in earlier

prototypes.
They include
(ii)

reductions of

systems of

type

(i)

to alethic logics by introduction of a

constant ’o’ read, for example, ’the world will be worse off’ or 'punishment ought to

cr is the moral sanction or some bad state of affairs.

follow’:

is defined

Then *0'

EK-A^cr); i.e. A is obligatory iff the negation of A (non-fulfilment of A, it

OA

is said) necessitates the sanction.

It is usually postulated: -0^.

Such reductive

frameworks, since they include minimal deontic logic, repeat its defects.
(iii)

relativistic

systems

which

attempt

axiomatise

to

circumstances,

c',

conditional

of

reads ’p is obligatory given, or

permission and conditional obligation; thus ’0(p/c)
*

in

notions

underlying model being that of conditional

the

probability.

These systems were designed chiefly to avoid the paradoxes of derived obligation to
which systems of type
really

in

(as

and

(i)

probability

(ii)

theory)

lead.

to

But

the new operator, /,

compensate

for

implicational connective in the modal framework assumed.

so-called

"logics

of

conditional

e.g., van Fraassen, p.438

’...

obligation"

still

lack

the

is introduced

of

an

adequate

So, not surprisingly, these

exclude

moral

dilemmas

(thus,

it would be more apt to say that we have here a logic

of obligation that remains after obligational conflicts are resolved').

(iv)

systems which

conditional

d,

such

include a modal conditional functor superior to the material­
as

a

Stalnaker-Lewis-style

conditional,

conditional deontic notions like those of (iii) can be defined.

in

terms

of

which

But these systems,

since modal, do not escape modal limitations and problems (see further RLR).

(v)

systems based

on a

logic of change or action. These systems turn out to be

reducible to a combination of absolute systems with tense logics, i.e.
14
to multiply modal systems and so repeat the problems of modal systems.

None of

these modal

types are satisfactory,

none can be made

so,

they amount

because of

6

What is at

modal assumptions, like consistency requirements, that they all involve.

fault is not merely types of systems of this sort, which confine the framework to a

worlds

possible

defect enough,

one

thereby

and

but the

write

in

conditions,

consistency

that is

though

sweeping consistency interpretation of deontic

more

logic,

that ’principles of deontic logics are conditions of consistency for normative
systems’.^ The consistency interpretation - which is deeply entrenched in much
ethical

theory

(especially

making no adequate allowance

from Kant)

deriving

that

for what is essential

fundamentally mistaken,

- is

moral dilemmas and

to morality,

the familiar inconsistency of normative systems.

Documenting moral dilemmas. Cases of moral dilemmas are, perhaps unfortunately,

§3.

both familiar and inevitable.

perhaps not,

and

logic

theory

deontic

(including

is unfortunate

It

such

for standard deontic

approximations

rough

as

utilitarianism)

since they have ultimately to be dismissed as inadequate to account for the data.

is

also

often

extremely

On the other

dilemmas.

awkward

who

those

for

such dilemmas

hand,

probity, and are a main source of moral change.

A moral dilemma occurs when
considerations,

for

instance

cannot

T

situation

in

act

(where

such

a

of

which

shape

cannot, or

precisely

as

way

afford real tests of moral worth and
16

is confronted with moral

perhaps not directly inconsistent,
factual

the

probably cannot,

facts

satisfy

to

principles,

binding

obtain);

principles

that

is,

one

so

as

to

or

although

be satisfied

Then the situation T cannot but induce a moral fix.

considerations.

in moral

(but not only when) one

the

in

themselves enmeshed

find

It

in the

(probably)
meet

moral

This can come

about in two ways, either because of the logical limitations supposedly satisfied by
T,

in particular its consistency feature (that, whatever p,

does

not)

or

through

certain

contingent features

if p holds in T then ~p

of T. In

what

follows

we shall

concentrate on a major subclass of what are commonly accounted moral dilemmas, those
of (paraconsistent) logical interest where what is given or derived are pairs such as

OA and OB, WA and WB (i.e. A is wrong and so is B), or the like, where A and B are
And

incompatible.

of

particular

logical

importance

are

cases

where

deontically

incompatible pairs such as Op and 0~p, Wp and W~p, or the like are guaranteed.

The examples of moral dilemmas we consider provide only a small selection from
the

rich

variety

of

such

dilemmas,

but

hopefully

they

illustrate

that

variety.

Several of the examples involve more than one moral principle: however some commonly-

adopted moral principles generate dilemmas, of one sort or another,

their own.

pretty much on

Consider for instance first principles of the following type: Situations

involving the

(indiscriminate)

killing of numbers of (uninvolved) people should be

avoided, should not happen, are wrong - principle at very first approximation of the

form (p)(p satisfies condition C + Wp).

Applying such a principle, a dilemma results

7

where

and

occurs a large number of people will be killed,

if p obtains or

if ~p

obtains a large number of people will also be killed.

War situations provide obvious examples, with the V(ietnam) dilemma typical.

17

For example, people have described the situation in Vietnam as follows (not that it
really is this way):

If the Americans withdraw from Vietnam, a large number of people

If the Americans stay in (do not withdraw from) Vietnam,

will be killed.

number of people will

withdraw

from

C. Hence

under

be killed.

Vietnam',

Hence,

condition

satisfies

p

where p is
C

'the Americans withdraw from

~p

and

the principle both p and ~p are wrong,

similar is the company commander's dilemma.

of his troops will be killed.

a large

also

i.e.

satisfies condition
18
Wp and W~p.
Rather

If he invades the hamlet, a large number

If he does not invade the hamlet the prisoners held

Taking p as 'he invades the hamlet', p satisfies C and

therein will be killed, etc.

~p satisfies C. Therefore, Wp & W~p.

Both p and-not p are wrong; neither would occur

The killing does not however have to be of large numbers

in an acceptable situation.

of people to induce moral dilemmas.

variants on the commander's dilemma

Favourite

regularly wheeled out concern murdering a potential murderer to prevent him killing

several other
presented

people and

with

the

related constrained

option

killing

of

some

choice situations
smaller

special

person is

where a

group

of

people

(or

creatures) in order to avoid the certain killing of a larger or less favoured group
1Q
of creatures. 7

The pacifist's (or differently seen, patriot's) dilemma, which Lemmon draws from
on
Sartre, concerns the proposition p^: Z goes to war.
If Z does not go to war, he
help his friends and

fails to

fellow countrymen when they are in desperate need.

Thus ~p-j is wrong, W-p-j , on the principle that so failing to help one's friends and
fellow countrymen is wrong.
If Z goes to war, he will be involved in killing people
he has nothing morally or otherwise against; whence Wp^, on the principle that things

which involve such killing are wrong.

Therefore again Wp^ and W~pp

A different sort of example, again however involving the taking of life in one

way or another is the abortion dilemma (as conceived by Anglicans, say).
presented as a clash between conflicting rights.

to life
(i.e.

(on

that

the

one

part of the

should

not

foetus);

have one's

In one corner, there is the right

in the other, the right to bodily integrity

body used

against one's

detriment for someone else’s benefit) on the part of the mother.

where

a

particular

state

of

affairs

is

an

instance

of

both

so

(~p)

and

will

principles.

Hence Wp.

would deny the mother's right to bodily integrity.

Whence adjoining, Wp & W~p.

to one's

A dilemma results

performance of an abortion (p) would deny the foetal right to life.

not doing

This can be

The
But

Hence W~p.

8

Not so far removed from these examples, particularly when omission is conflated
with commission,

are environmentally-oriented

examples involving

sacrifice (broadly

construed),

for

resources.

Suppose that we have a duly qualified principle to the effect that things

which lead

to the feeding of starving people should happen.

example the

is a dilemma of limited

three cases:- First

following

Suppose also that we

have limited resources, and have decided to distribute these as follows: If the coin
comes down heads (p), then group A of starving people will be fed.

But if the coin

does not come down heads (~p) then group B of starving people will be fed.

Plainly

in an ideal situation both p and ~p will occur, that is the coin will come down heads
and it will not come down heads, so that both groups of people will be fed.

And if

the moral principle cited is correctly applied then a case of Op and 0~p results.
course minor

Of

involved leads to conclusions of the form

adjustment of the principle

W~p and W—p, whence Wp.

For an analogous conservationist * s (or humanitarian * s) dilemma consider the case

forest

where it ought to happen that the

is chopped down,

the benefit of the

for

presently starving (or etc.), and it ought to happen that the forest is preserved by
positive action

and

Op

dilemma

for the benefit of coming generations (or etc.).

0~p

emerges

the

from

of

application

distinct

In this case the

The

principles.

society dilemma, concerns the reform of primitive societies so that they

primitive

Consider the dilemma of a feminist environmentalist as

conform to outside standards.

regards Aboriginal

On the

women.

other

hand,

suppression of women in Aboriginal

society should be opposed; on the other hand major western interference in Aboriginal

affairs should be avoided; but changing the position of women would constitute major
interference.

Different examples involving
public

institution)

dilemma,

which

following:- On the one hand,

the

be restricted and the

funds will

principle

being

the

concerns

preposition

p2:

the

University

The Vice-Chancellor, or rather his advisor on logic, argues

opposes the government.
both

separate principles are the Vice-Chancellor * s (or

formulated

Wp2,

p2 is wrong,

students and

accordingly).

On

because the University’s

learning in general will suffer

the

other

hand,

W~p2,

~p2

(the

is wrong

because #p# will strengthen the present iniquitous status quo (the principle being

states

that

of

affairs

which

strengthen

the

present

status

quo

be

philosophers,

is

iniquitous

avoided). Therefore, conjoining Wp2 & W~p2»

A

signifi ant

source

of

moral

dilemmas,

long

observed

by

afforded by a relatively unrestricted institution of promising, and more generally by

parallel

institutional

arrangements

for

entering

perhaps worth distinguishing two classes of cases.
situations,

Z

is under

an obligation

into

(moral)

Firstly,

contracts.

It

is

in purely contractual

p, because of a contract or promise, and also

9

under an obligation q, because of a (likely different) contract, where q entails or

In the second sort of case, Z has promised to do ~p, and hence

commits one to ~p.

But p also satisfies conditions C,

0~p.

it emerges, and Z adheres to the principle

(q)(Cq ->0q), whence Op.

As an example of the

second type consider Lemmon’s simple and dated example,

adapted from Plato:

A friend leaves me with his gun saying he will be back for it in the evening,
and I promise to return it when he calls.
He arrives in a distraught
condition, demands his gun, and announces he is going to shoot his wife
because she has been unfaithful. I ought to return the gun, since I promised
to do so - a case of obligation.
And yet I ought not to do so, since to do
so would be to be indirectly responsible for a murder, and my moral
principles are such that I regard this as wrong.
I am in an extremely
straightforward moral dilemma, evidently resolved by not returning the gun
(62, p.148).
According to Lemmon the outcome is both Op, where p is ’I return the gun’, and also

The

0~p.

namely,

for

first

an

inference

concerning

promise-keeping;

every q, that q is a genuine promise entails Oq (for

genuine promises

follows

engender obligations),

from

and p

is a

immediate
genuine

promise.

inference of the following sort: for every q,

The

second

follows

that q really contributes to a murder

implies Wq, and that p really contributes to a murder; and moreover,

-> 0~q.

using an

for every q, Wq

Even if this last principle is repudiated a dilemma remains: for both Op and

Wp, i.e. I ought to return the gun and it is wrong that I return the gun.
that the initial principle sustains the conclusion 0~q as well.

(It may be

In any case,

it is

not difficult to amend Lemmon’s example so that it does, or more important, so as to
render the ensuing dilemma more convincing.)

Other dilemmas recur in variant form with contracts and promises.

For instance,

limited resources dilemmas reappear as follows: we promise all our surplus to a group
of poor only to have even poorer people turn up.
where

one's

deception,

employment calls
or

participation in

for

activities

Similarly work and role dilemmas,

which are

the production of nuclear

morally repugnant

(such as

or chemical weapons), may

arise either contractually, e.g. one signed up for the position, or from duty, e.g.
to provide for one's family.

Some of the moral dilemmas we have made look more determinate than perhaps they

should, or than they would be if the circumstances were filled out, or got filled out
slightly differently.

Then we should find that some of the dilemmas take the form of

Lemmon's more complex class, namely those where rather than OA and 0~A, what one has

is Prob OA and Prob 0~A, where Prob is the probability functor ’it is probable that'.
Lemmon describes

this class of "fully dilemmatic situations" in these terms:

’there

10

is some,

but not conclusive, evidence

that one ought to do something, and there is

some, but not conclusive, evidence that one ought not to do that thing’ (pp.152-3).
sort of dilemma is again drawn from Sartre: in

A real-life working example of this

brief, Sartre’s pupil is under some (sort of) obligation to stay with his mother, but
he

also probably under

is

an obligation,

felt only ambiguously,

to join the Free

French in England (for a fuller description see Lemmon, pp.153-4).

Lemmon suggests

that this class of dilemmas is, and perhaps should stay beyond deontic formalisation:
however the dilemmas are now within the scope of formalisation of a sort by multiple

functor intensional logics of a type not different
represent such principles as ’’ought implies can”.22

those already

from

needed to

Though we have presented a wide variety of dilemmas, some of them in some detail

- to

the

dispel

that dilemmas

common prejudice

are a rare

phenomena that can be

safely set aside or that need not be taken seriously in an ethical theory - we have
by no means exhausted the range and wealth of such dilemmas.

the

extent

of

analogues

practical

of

these

Nor have we indicated

the

dilemmas;

fact

is

that

similar

dilemmas are frequent in economics, over budgeting, devaluation (we ought to devalue
because ... and we ought not to devalue because ...), and so on.

Nor have we brought

out the political significance of the dilemmas; but the fact also is that some of the
deeper problems in moral and political theory revolve around dilemmas.
Familiar
examples

are

issues

of

(permitting

permissiveness

interests possibly harming or

licence

individuals

to

pursue

offending others) and freedom (restricting freedom as

regards actions which infringe others
*

freedom), of paternalism and state authority

versus civil liberties, and of democracy as against minority interests.

A

assortment

varied

accounts of the
where sources

of

examples

is

source of moral dilemmas,

for moral claims conflict,

contractual obligations.

Lemmon

also

in particular

and

promises,

is attracted

principles

as

those

morally

simplistic

that they generally arise

greatest

(p.150).

But he

fact that one and the same

conflicting directives

requiring

type,

since principle may conflict with

or duties, may conflict with one another

principle can deliver
such

removing

by an account of this latter

underestimates the extent of conflict, and neglects the
contract or

in

e.g. duty with principles, principles with

while realising that it will not always suffice,

principle

important

(e.g.

double maximization

happiness

of

the

greatest

number).

As well

a sufficient variety of tough examples helps remove the illusion that

moral dilemmas can one and all be escaped.

Admittedly, fortunately, there are ways

around some of the dilemmas we have outlined, especially those we have borrowed.
with

perhaps

further

around the dilemmas),

artifice

in

setting up

the cases

(to exclude

But

lateral moves

some of the dilemmas stand: from them (as is further argued in

11

residue

’To insist that there is in every case a solution without

is no escape.

§ 6) there

false

is

the

to

moral

facts’

p.132),

Marcus

(Barcan

a

theme

she

well

illustrates with the abortion dilemma.

§4.

deontic

Paraconsistent

logic

cannot be modal

makes

it

modal

treatment

the relevant development.

logic:

An adequate deontic

have to be paraconsistent.

but will

For

a modal treatment

to accommodate moral dilemmas - which are hard data

impossible

it

makes

as dilemmas

treat dilemmas

impossible to

(§3)•

A

without

proper

catastrophic collapse (as shown in §1).

The

of

principles

standard

deontic

logic

are,

however,

as

added to

readily

various paraconsistent logics as they are to modal or classical logic - at least this

(e.g.

OP),

in

(da Costa

negation-weakened

namely

logic and now distinguished

style)

logic

paraconsistent

and

For Jaskowski-style discussive logics there are some difficulties;

relevant logics.
in

styles of paraconsistent

two of the main

for

is true

particular,

the

given

of

rejection

can

principles,

adjunction

such

logics

satisfactorily include O-adjunction, OA & OB -> 0(A & B), which reflects adjunction in

for the possible-worlds sentential framework

It would seem not;

acceptable worlds?

logic offers no way of handling

of Jaskowski-style

the form 0(C & ~C) that result by O-adjunction.

properly accommodate dilemmas,
for

dilemmas concerning obligation, where

such as moral

clearly

principles

difficulties;

For these require impossible worlds

Generally, Jaskowski-style logics are in deep trouble trying to

where C & ~C holds.

adjunctive

full-strength moral dilemmas of

logics

Negation—weakened

hold.

they cannot deliver

expected deontic

also

are

interconnections, e.g.

in

the

deontic ’’square of opposition”, presented shortly, which depends crucially on normal

principles.

negation

In

difficulties.

Relevant

brief,

deontic

the reasons

types of paraconsistent logics

logics

can

both

avoid

choosing relevant logics over

for

(the substantial

sets

these

of

other main

reasons for which are given in OP)

are accentuated still farther as soon as applications, such as to deontic logic, are

Accordingly we will assume a relevant logical base, and a relevant semantical

made.

in what follows.

framework,

Even so, much of what we say and do can be adapted to

other logical bases, e.g. negation-weakened or positive-plus bases.

Often

a

combined

syntactical-semantical

is

approach

more

fruitful

and

illuminating than either a syntactical-axiomatic or a semantical approach carried out
separately. Each enriches and controls the other.23 Such is the case with relevant

logical theory,

logic.

Even

analogue

of

and so it is with relevant deontic logic, which builds on relevant
so,

a

standard

syntactical
deontic

standard superstructure,
relevant logic.

beginning

logic

we

normally added

can

is

easier.

simply

add

For
an

to

obtain

exact

a

analogue

relevant

of

the

to classical logic, to (what supersedes CL)

12

For minimal relevant deontic logic, MD, the one-place functor 0 - of obligation,
as usual

read

(e.g.

’It ought

suitable

some

to be

extending

system

that’ - is added

(the case)

B

system

basic

the

to some relevant logic

of

RLR

conforms to the usual formation rule: where A is a wff so is OA.

however

are

options

at

Functor, F read

’It is forbidden that’, is defined thus:

’It is permissible that’, is defined : PA =Df~FA.

FA =Df 0~A (cf. OED); and P, read
There

this

stage

(as

at

e.g.

stages),

can

morality,

be

argued,

from

directions

several

other

F

could
oh
introduced independently. But this leads to complications that are unnecessary.

It

0

The other familiar

functors of deontic logic, F and P, are defined in the usual way.

’It is wrong that’ or, less satisfactorily,

Functor

I).

(e.g.

from

the

rationality

be

of

from the meaning or semantics of obligation, from the idea of commitment),

that 0 satisfies at least the following requirements:[RM] A -> B -> OA -> OB,

i.e. where A + B is a theorem so is OA -> OB.

That is to say,

where A entails B is provable so is OA entails OB, obligation transmits over provable

entailment.

Hence

the claim that this principle (R7 of RLR II) is a transmission

principle.

[K]

OA

& OB

regards

the

i.e.

-> 0(A & B),
conjunction;

or,

a conjunction of obligations entails obligation as

to

adjunction (principle G of RLR II).

These two important

it

put
25

principles,

entailment and adjunction, make 0 a

a

little

differently

0

is

closed

which ensure that 0 is closed under

under

provable

systemic functor (in the sense of RLR II), and

ensure that it has a relational (i.e. first order) semantical analysis.

The postulates which correspondingly minimally characterise deontic functions F
and

P

now

follow

by

relevant

logic

upon applying

the definitions.

They

are as

follows:For wrongness:

A

For permissibility :

With

the

A + B -*>

functor

further

B —f> FByFA

Q,

FA & FB

P(A v BH, PAv

PA + PB

defined

F(A v B)

QA =Df ~0A,

a

reduced

pb

deontic

square

of

opposition follows:

0Ax ...............FA

PA^ .............. .QA

The opposites are
dotted peripheral

opposed,

as modally,

relations of the modal

in

the way the solid lines show.

But the

square are bound to be the lost given the

13

the modal

(on

fact of deontic dilemmas

square

Prior 62 p.220).

see

is worth

It

observing, however, that the peripheral relations can be reinstated through adoption
of the so-called consistency scheme

OA -> ~FA, i.e. OA + ~0~A;

[D]

and that the relevant deontic logic that results still excludes the rubbish brought

in

along

deontic

standard

necessary truths,

Nordamericano

of

that

course

not

(though

paradoxes,

essentially because paradoxical

etc.),

deontic

obligatoriness

the

of

spread principles are not

whereas

triviality),

incorporate dilemmas.

the

as

degenerate

dilemma-permitting

The problem with deontic logics which include [D]

also

cannot

they

well

as

logic,

Antipodean form we are advocating.

is

deontic

In short, relevant theory can deliver a clean-cut, clean-living, dilemma-

available.
free

(e.g.

logic

include

relevant

moral

dilemmas

systems

without

without

can

[D]

inconsistency
consistently

These various logical claims, as to the modal rubbish that is

not brought along, can all be made good by application of the semantical theory. The
semantics shows what does not hold, as well as further explaining why the principles

that hold do.

Semantics for MD result upon adding to modellings for relevant logic a two-place

relation S, on worlds, which is subject to the following condition
W.

b and Sbc then Sac,

Where a

i.e. where world a precedes b (indeed canonically a is contained in b) and world c is

accepted vis-a-vis b then c is accepted vis-a-vis a. Given that a is a subworld of b
this is evident enough.

Distinctive features of the underlying semantical modellings

for relevant logic - consisting of structures of the form <T, K, 0, R, *,
be

sufficiently

and,

well-known

any

in

are

case,

explained

v> - should

elsewhere.

These

structures differ from the best of those for modal logics, the nonnormal structures

<T, K,

N, R, v> for the Lewis systems S2 and S3,

in the following ways:- The class K

of worlds, to which the factual (base) world T belongs, is expanded to include a much
more comprehensive class of inconsistent and incomplete worlds; the class N of normal

worlds

K is

in

adjusted

theorems hold; relation R,

to

a

three-place

the class

to

v(p,

sentential

a)

relation

=

parameter

1)

regular worlds,

i.e.

worlds;

on

then

and

wff.

I(A, b)

=

1,

where all

on

is

subject

to

a

constraint

where I

is the

so

that

Specifically, where a < b and p holds in a

p holds in b,

p. Conditions

v

for every world a and b

modelling

structures

are

hereditariness condition extends to all wff; i.e. in symbols, where a
1 then

those

used in modelling implication, is relaxed from a two-place

inclusion of worlds is duly reflected.
(written

0 of

in K and every
such

that

this

b & I(A, a) =

interpretation function which extends v to all

Inclusion, <, can be added as an extra component of the model structures, but

in fact it can be defined thus:

a < b =Df (Px 0)Rxab.

Finally relevant modellings

14

add to modal modellings a new function, *,

in terms of which negation is evaluated,
27
which facilitates the inclusion of inconsistent and incomplete worlds.

same role and interpretation as in standard

relation S has the

The semantical

It is employed in the semantical evaluation of obligation functor 0,

deontic theory.

which is given by the following rule:

I(0a, a) = 1 iff, for every world b such

that Sab, I(A, b) = 1, i.e. OA holds at world a iff, for each world b accepted vis-avis a, A holds at b.28 The rule is a world relativisation of the idea that what is

obligatory is

what holds

in all

accepted worlds.

(morally)

It

is thus an evident

adaption of Leibnitzian prescriptions, given the linkages between obligation or moral
binding and moral "musts” or imperatives, on the one side, and morally possible and

or

acceptable

accepted

worlds,

the

on

other.

The

for

motivation

unrelativised

connections can be forged at base world T; for there
A is (morally) required, I(0A, T) = 1, iff for every b for which Mb

I(A, b) = 1 (where Mb =Df STb), i.e.
That is, what is required is

iff A holds in every (morally) accepted world.
what holds in all accepted or received worlds.

Other construals of the semantical apparatus are of course open.
tempting

ideal,

temptation

(a

succumbed

to

initially)

regard

the

accepted

worlds as

and Sab as ’b is ideal as seen from c’.

to read Mb as 'b is ideal’

i.e.

to

It is a little

But

that is like requiring supererogation in cases where the meeting of commitments or
obligations is enough: the worlds do not need to be that good but simply to pass, to
meet required standards.
is

worlds.

acceptable

through

A perhaps easier construal than by way of accepted worlds

construal

That

is

alright

provided

the

term

’’acceptable” is deployed, as it often is, nonevaluatively : for it is important in the
semantical assessment to

prescriptions

for

be able to make an evaluative step—down, to give meaning

evaluative

expressions

nonevaluative terms.

in

Naturally this

does not imply that evaluative notions are naturalistically eliminated, any more than

familiar
terms,

extensional

reductions,

semantical

extensional

analysis

of

intensional

implies that intensionality is removed: both are merely pushed down into the

not-further-explained apparatus of worlds.

Obligation is construed by way of what goes on in acceptable or accepted worlds,

not

what

happens

in

permissible

or

permitted

directly with the construal of permissibility.

one another,

but in

worlds,

because

this

latter

Given modal muddles the two reduce to

the more subtle relevant setting star shift is involved in the

transfer from accepted to permitted worlds.

The (derived)

semantical rule for P is

this:
KPA,

a) =

ties

1 iff for some b such

that Uab

I(A, b)

= 1,

where Uab

.
b
*
Sa

15

at base

Thus,

A is permissible,

world T,

i.e.

I(PA,

=

1)

iff A holds at some

1,

permitted world b; what is permissible is worldwise what is somewhere permitted.
modal equation of relations U and S

further

axioms,

deontic

the

(where

namely

if Sab

(in effect,

condition

then Sa
)
b
*

(Pb)Sab

The

corresponds to

e.g.

guaranteed,

is

modally by ~0(P & ~p)) by the scheme OA ^PA.

Now modal-strength deontic logics typically do include an axiom scheme of the
form
[D’J

OA

-> ~PA,

i.e.

from

(alethic)

is obligatory

is not

that

permissible,

under modal

is,

Indeed this used to be considered what distinguished deontic

~0~A.

assumptions, OA
logics

what

modal

on

logics

the

one

side,

which

include

a

stronger

principle analogous to OA -*■ A, and epistemic logics on the other, which were weaker
in

that

[D’J

failed.

However

[D’J, like

[D], is deontically incorrect

because it

rules out consistent inclusion of moral dilemmas; or, to turn matters around, cases

where OA & 0~A,
[D’]»

which

or OA & FA, hold counterexample [D].

may

seem

a

little

more

plausible

transform of it on usual deontic definitions.

than

Similarly these cases refute

[D],

but

is

a

definitional

For what is obligatory is not ipso

facto permissible where the obligatory item is also wrong, as in dilemmatic cases.

To distance relevant deontic logic from modal analogues, while at the same time
demonstrating that relevant theory can encompass moral dilemmas unproblematically , we

next assume that some moral dilemmas hold, that is that for some p, both Op and Fp.

Since however

propositional quantifiers much complicate the semantical apparatus we

shall introduce constants into the syntax, and let p^ be the first constant for which
a dilemma ensues. That is, a new axiom, extending the logic, is
[DD] 0p1 & Fpp i.e. p1 is dilemmatic.3°

The modelling condition for [DD], is
wd.

For each a in 0, for some world x, Sax and v(ppx) = 1= v(ppX
).
*

This condition, which does not involve T, does not impose nonclassical conditions on

normal worlds.

Accordingly [DD] is only a weakly paraconsistent requirement; it does

not imply that inconsistency protrudes into the factual

or regular

worlds.

Indeed

one advantage of the semantics is that it enable proof that dilemmatic judgements can
be included without inconsistency.

By contrast, modal theories cannot include

even one half of [DD]

do) without automatically excluding dilemmatic judgements.

consideration has a thesis 0p2 for some

(either will

Suppose the logic under

Then for every x in 0 there is some y

such that Sxy with p2 set to hold at y. But modally this is enough to guarantee that
[D] holds materially,
More

explicitly,

so ~(0A & 0~A) generally,

suppose

otherwise

that

in

i.e. there can be no such dilemmas.

some

model

the

material

form

of

[D]

failed;

I(OA,

then

=

T)

1

=

I(A~O,

T),

which

impossibility at y since I(A, y) = 1 i I(A, y).

syntactical

problems

earlier in

1).

even

for

Although [DD] implies Op^,

very weak

obligatory is

conjured

from the fact

some

for

STy

yields

y

This is a semantical analogue of the

deontic logics

(considered

type

of modal

it does not say, nor does it follow, that OT for any

Such modal stupidity as that every necessary

and every tautology T (e.g. C v ~C).

truth is

since

automatically removed

that relevant

in the

relevant wash.

can be

This

theory removes such paradoxes as p1

from which 0(A v ~A) would follow modally by 0—distribution (i.e.

A v ~A,

Again the

[RM]).

semantics can be applied to establish the point in proper detail.

primitive principles of minimal relevant logic are open to challenge,

Both the

and

[K]

*
Jennings

has

been

disputed

by

Schotch

(in

Jennings

and

81).

The

main

Schotch-

arguments against [K] — which induces what they call ’paradoxes of complete

*
aggregation

essentially

turn

—

automatically avoided

on

modal

deontic

by relevant

and

assumptions,

Their

theory.

accordingly

are

(p.152,

also

main point

p.154) is that [K] obliterates the distinction between [D] (i.e. Op

~0~p) and

[Con] ~0 , i.e. in effect ~0(r & ~r).
Their argument depends on construing implication in [D] as material-implication and

on strict replacement.

~0~p iff ~(0p & 0~p), i.e. iff ~0(p & ~p) using [K]

Then Op

and its provable converse,

i.e.

iff ~0(r & ~r) using p & ~p

- .

r & ~r and strict

According to the relevant approach the trouble is not with [K] but with

replacement.

the modal assumptions, which are independently objectionable (see

1).

Their further

argument begins from the premiss that ’we should certainly want to include among our
deontic principles the law of moral consistency [Con]
across

cuts

all

different things.

moral

theories’

(pp.153-6).

The

...

because that principle

premiss

appears

to

conflate

What cuts across all moral theories is the theme that some things,

such as a sufficiently evil state of affairs, are not obligatory, whence ~0s.

does cut across all

theories is the assumption that there are no moral dilemmas of

the form 0(A & ~A), which is what [Con], with
not

just

that

there

What

are

theories

like

properly defined, would imply.

those

of Lemmon

and Sartre

It is

which include

contradictory obligations; it is also that some theorists would want to insist that
there is something logically wrong with principles like [Con], e.g. that it doesn’t

make sense to talk of the nonobligatoriness of absurdity.

Schotch and Jennings do contend that we should allow for moral conflict of the

form OA & OB where A and B are inconsistent.
their

assumptions

there

[K]

stops

rejection

of

are

moral dilemmas

that

yielding

Since then B -> ~A,

of the

0(A & -A).

it follows that on

form OA & 0~A.

Yet

as

they

Certainly only

themselves

have

reported, the consensus view is that there are (in the end) no dilemmas of the form

OA & O~A: insofar,

all theories’,

’cuts across

then, rejection of dilemmas of the form 0(A & ~[~]A)

so, near enough, does rejection of those of the form OA & 0~A.

What

certainly does not emerge is that
there can

[K].

The

be no deontic logic which takes as a primitive law the principle

be

can

that

best

if

done,

subscribe to [K],

we

is to

formalise

ethical theories - namely those which do not allow moral

certain particular
conflicts (p. 156).

logics provide decisive counterexamples to this contention.

Relevant deontic

They

can take [K] as primitive: they allow for moral conflict; but they do so by allowing,
like Lemmon, for moral inconsistency.31

disconcerting

More

from

flowing

the

than

fundamental

difficulties

deriving

principles,

transmission

[K]

from

are

allegedly

those

difficulties

include

which

well-advertised paradoxes of deontic logic.

§ 5.

Resultant

on

impact

other

relevant paraconsistent shift.

and

puzzles

of

paradoxes

deontic

theory

of

the

Often cleaning up one logical puzzle in due technical

detail has a beneficial effect on surrounding puzzles and on associated problems in

the area.

Light is cast on, and

So it turns out also in the case of moral dilemmas.

Among them

sometimes solutions are delivered for, a number of other linked problems.

are the following puzzles generated by theorems of (relevant) deontic logic.
are,

the ’’paradoxes of commitment”. Where material implciation => is

in first place,

defined as usual, e.g.

T1.

There

~A v ~B, it is a theorem that

A ° B

0(B => A), whence if OA then 0(B 25 A),

OA

sometimes

read:

A

if

is

then

obligatory

anything

one

commits

to

A,

or:

’Doing

anything commits us to doing what is obligatory’;
T2.

FA

OCA => B), if 0~A then 0(A = B),

sometimes read: if A is forbidden then doing A commits one to doing anything at all,

or:

’Doing what is not permitted commits us to doing anything whatever’

readings are

proceed

from Prior

from

there

interpretation of

paradoxes

symbols,

the

The claims read off are paradoxical.

62, p.224).

to

little justification - reading ’0(A

commits one to doing B’.

the

in

and

(the quoted

in

logic

this case

B)’

as

depends,

always,

depends crucially

on -

But to

on

the

what has

as a commitment relation, e.g. as ’doing A

Such a reading which presupposes that material implication

is a satisfactory conditional can in no way be sustained (see RLR), and that is the

end of the problem.

that

A

then

paradoxical

guises;

and

it

If,

ought

to

for example, T1

be

features disappear.
these

commitment

the

case

is simply read: if it ought to be the case

that

either not

B or

A,

then

the alleged

However puzzles have a tendency to reappear in new

puzzles

are

no

exception.

Even

with

conditional

18

terms of obligation applied to a

properly defined in

obligation

conditional,

the

puzzles

in

reemerge

new

forms,

example,

for

genuine (relevant)
in

is

what

called

’’Ross’s paradox”, and in related transmission problems.

It has been claimed that the principle,

OA -> 0(A v B), yielding if OA then 0(A v B)

T3.

is

which

T1,

to

equivalent

is

Ross paradox principle),
-

paradoxical

to do either A or B,

it

may

be

More generally, given OA,

obligatory to read a letter but not to read or burn it.
there is an obligation

example,

for

because,

where the content of B may be totally

do

to

established.

(The comment goes beyond what T3 supplies, but that is not material).

the

in

circumstance

which

obligation

the

to

is

set

inappropriate

A

or

A similar objection is made to the analogous permissibility principle
PA ■> P(A v B),

T4.

Ross’s paradox,

on the alleged grounds that there are examples where A is permissible but A

For example,

be.

not

the following

it is permissible

that

to post a

takes

based

on

some

one

application of T4 is said to the paradoxical:

letter

false proposition

the

distance

implies that

it is permissible to post a

The assumption that there is some sort of paradox here is

letter or kill one’s aunt.
however

B may

the

towards

that the

permissibility of

of

permissibility

killing

the consequent

one’s

The

aunt.

proposition itself seems to be based on an application, never articulated and quite
inadmissible,

of Disjunctive Syllogism,

distribution,

P(~A &. A

For

PB.

(A v b) ■+ B, and what results from P-

it may certainly be permissible not to post

However it is assuming more than standard deontic logics will grant to

the letter.
go on

)

~A &

to the permissibility of not posting the letter and

letter

or

killing

disjunction.

one’s

aunt,

especially

given

the

also either

derivation

of

posting the

latter

the

Still going on remains tempting (and would presumably be warranted by a

Lukasiewicz deontic logic analogous to his 4-valued modal logic).

But so going on is

to no avail in a duly relevant setting, where Disjunctive Syllogism fails: there is
no

admissible

route

here

to

the

permissibility

of

killing

aunt,

one’s

however

desirable the act might otherwise be.

main

The

transmission

conditional obligation,
strict

implication,

as

problems

in

found

also modally derived.
an

adequate

deontic

logic

are,

like

those

of

They come from reliance on material or

representation

of

implication.

The

worst

manifestation of this is that already discussed, deontic trivialisation leading from
Op^ and 0~p.| to OB,

for

any B,

in virtue of the provability of p^ & ~p^ => B.

But

there are variety of lesser manifestations of the same underlying problem, of which

the following example (adapted from recent discussion) is just one sort:
p4.

We have an obligation to feed the starving poor D. There exist starving poor.

Therefore,

since

starving poor,

it ought to be

by O-transmission,

the case
it ought

that we meet our obligation to feed the
to be that there exist starving poor, a

19

that

is

far

from obviously true.

deontic logic,

for

the

simple reason

presupposition

relation,

conclusion

of

transmission problems of modal

the

that the

cannot

p4

Such analyses do not

entailment.

an

to

rather

or

implication,

(material-)

upgraded

be

plague relevant

The

many

systems all enamate from excessive O-distribution on

basis of implications will

not

them,

support

in the

as

following transmission

principles typical of orthodox deontic logics, all of which should be rejected:

OCA D B) D. OA 3 OB

A D B -> OA 3 OB

A => B =>. OA => OB.

With their rejection a thorny tangle of modally-derived paradoxes is also rejected.

There appears however to be an immense temptation to try to impose excessively

strong transmission principles; and the same phenomenon to be seen at work in modal
contexts can also be observed when a proper entailment is introduced.3^ This occurs,

particular,

in

attempts

in

to

’’normalise”

[RM],

to

strengthen

it

a

from

rule

connection to an entailment, for instance, to (the characteristic S3 form)
[OT] A + B +. OA + OB;

it occurs with analogous changes thereon,

and

e.g.

A-> B.

PA

PB and A

FB + FA, as well with as imported forms of these principles, e.g. OA & (A
and (A -> B) & FB

As an example of the

FA.

B

B)

OB

type of puzzles that [OT] produces

consider

p5.

Grannie killed the vicar.

Grannie repented for having killed the vicar

The

problem with putative

entailments such as p5

is that,

while we are ordinarily

(with the context duly flashed out) inclined to agree that Grannie ought to repent,
we

don’t want to agree

to have killed the vicar.

that Grannie ought

The problem

turns essentially however on the availability of [OT] in place of [RM], for p5 is not
a provable entailment of the underlying logical theory (and there is little case for

it one),

trying to make

sc

[RM]

does not apply to p5,

in contrast with

awkwardness [OT] appears to require, of having somehow
entailment, is thereby straightforwardly avoided.33

Most

the

of

paradoxes

falling

paradoxes and Robber paradoxes can be
paradoxes and usual

forms of these

resolution.

one’s

According

helping

a

man

under

such

to

deny

descriptions

as

similarly resolved or avoided.

variations

upon

[OT]. The

is

that

p5

Good

Samaritan

an

The original

them certainly succumb to such

to the Robber’s paradox if robbing a man is

forbidden then

The

Good Samaritan

whom

one has

robbed

is

also

forbidden.

paradox is a further turn on this paradox: not only the robber’s helping a man he has
robbed

is wrong,

so is someone else’s, a

good Samaritan’s,

helping the

robbed man

(see Prior 58).

-

be

It is undoubtedly paradoxical that it should - especially as a matter of logic
wrong to help a man who has been robbed: but do any fundamental deontic

20

principles commit us to such a thing, for appropriate substitutions on variables? No,
These puzzles all take the following form: B, which describes something

they do not.

admittedly wrong, gets in or implied as part of the description of something larger,
A, arising as a result of the situation in B; however A is not wrong but perhaps even

Nothing

required.
theory;

and

in

none of

shows

this

the

however

that

A

B

literature can

examples in the

the

provable in

is

it

There

be.

logical

is another

important feature, commonly neglected however, which is that A does not assert B. If

it did,

then A,

since amounting

to A & B, would appear to be wrong also, as B is.

This points to a difficulty also for the usual assumption that A entails B; for then

A (certainly modally) coentails A & B, whence the wrongness of A coincides with that

of A & B, the wrongness of which would seem to be guaranteed by B. If however A only
involves B in a weaker sense, [OT] is exonerated.3^

The paradoxes can be pushed another way, to contest this data, and bear against
systems as minimal as MD. For MD yields theorems like the following:

T5.

FB

T6.

0(

F(A & B), whence if FB then F(A & B).

OB, whence if 0(A & B) then OB.

&B)

theses,

Such minimal

it

is now suggested,

are

paradoxical

in

the

same way as the

Robber. This is not so: only gross confusion of decidedly distinct syntactical forms

- a practice formalisation can unfortunately encourage - could support such a
suggestion. If it is wrong to rob a man then it is wrong to rob him and help him;35
but it is not thereby wrong to help him, indeed to help him who has been robbed.

But

the relative and the tense shift are both important; and the whole relative clause in
no way reduces to a conjunction conjoining robbing the man.

The positivistic objection that the paradoxes of deontic logic show that there
is really no ’’logic of norms” has come even from a modern founder of deontic logic,

von Wright.

This he now contends that

There is no such thing as an "entailment” between a norm Op and a norm 0(p q)
- and Ross’s paradox makes us aware of the absurdity of talking as though it
existed. ... Ross was right when he offered his paradox as a proof that,
properly understood, there could not be such a thing as a logic of norms (82,
p.15, rearranged, emphasis suggested in text).

And

he

’essentially
(p.14).

(also

also

But

mistakenly)

of

variations

given

that

one

regards

and

the

von Wright

contains such normative tautologies Op

are

inevitable.

Thus

while

’’paradoxes’*

the

same,

viz.

wants also

the

of

deontic

so-called

to supply

logic

as

Ross’s Paradox’

a deontic

logic,

which

0(p q), some subtle and dubious distinctions

Op & ~0(p q)

is

normatively

inconsistent,

inconsistent; while Op can be thus said to ’’normatively entail” 0(p q)

it

is not

it does not;

and while Op comes rational commitment to Op, it does not carry logical commitment to

it.

the

Unlike

said

are

connections

associated

to

and

entailment

(p.15).

no paradoxes

induce

other

But

normative

these

relations,
what is

said

go

to

for

normative entailment can be said as well as regards entailment, that one (e.g. a law

giver)

is

Op

adopts

who

rationally

to

0(p q)

going on to endorse the entailment of 0(p q) by Op.

for

a oitrary

addition

away there is no puzzle in

associations are cleared

once mistaken modal

q. Indeed

committed

The ’’paradoxes” of deontic logic

are not a reflection of the illusory notion of a logic of norms but rather a product
of a modal logic of norm.

However von Wright has other significant reasons for insisting that (in a sense)

’norms have no logic or that normative discourse is ’’alogical”’
from

noncognitivism,

that

symbolising norms,

formulas’,

’prescriptive

standard

a

on

One derives

(p.4).

interpretation

deontic

’do not express true and false propositions.

It makes

no sense to speak of relations of contradiction or entailment between formulas when
thus interpreted ’

hold

nontruthvalued

regards

as

semantical

This is not so:

(p.4).

theory

relevant

as

such

discourse,

for

sketched

logical relations such as inconsistency

logic

deontic

carries

no

commitment to the position that normative discourse is truthvalued.
0 interpreted

In

imperatives.

fact

ineradicable

The values 1 and

by way of holding are not intended to be tied to truth and falsity,

except at the factual world T (see RLR1); and that exception is readily removed.

judgements,

instead normative
received,

logic

the

both

(the point

requirements

e.g.

Otherwise the

etc.).

adjustments

the

and

of the

form Fp,

semantical

can be said to be in

theory

its ’’semantical”

is developed,

For

force (or

is as before.

With such minor

theory conform

to noncognitivist

to meet von Wright’s argument,

in Morscher,

pp.97-8).
Von Wright’s other reason for coming out against a logic of norms is but grist

It is that ’there cannot be a logic of norms because a

for the paraconsistent nill.

could enjoin contradictory

legislator
(at

’’relative

least

serves"
Morscher,

...)

dilemmas.

it

but

p.98).

to

the

is

normal

not

states of affairs.

This might be irrational

purpose which the institution of giving orders

logically

excluded,

it

is

not

alogical’

(cited

in

That is, a legislator can set up (usually unintentionally) deontic

This is

certainly not

triviality,

by

modal-strength

indictment,

not

of deontic

deontic

logic,

It

impossible.

but

logics.

of a

is simply excluded,

Once

again

the

on

criticism

certain modally-dominated

pain of
is

an

approach to

it.36

§6.

Upsetting the moral consensus: unrealisable obligations, and inevitable wrongs.

The theme that there are moral dilemmas, like those cited, which stand, which do not

dissolve upon

rational reflection,

philosophers, 3? and

apparently runs agains the consensus among moral

certainly upsets dominant positions

in ethical

theory, notably

22

broadly Kantian positions (built into much contemporary British moral philosophy) and

utilitarian

positions

(which

pervade

Anglo-American

and

theory

value

thinking,

especially outside narrowly philosophical circles).

Kantian positions suppose that morality is above all a rational procedure of a

certain sort (which we tentatively grant), and that rationality requires consistency.
Or put differently, but

for instance, must be consistent.

Thus what is obligatory,

equivalent given modal assumptions, ought implies can, OA -> OA, for every A and some

Wherever an agent appears to be in a moral dilemma

suitable possibility functor 0.

this is a matter of appearance only, a moral illusion which will (and must) vanish on
further rational reflection.^ This Kantian assumption should be rejected:
moral

dilemmas are part of the data and no illusion, and no amount of reflection need budge
them.

By insistence upon consistency requirements, and through failure, consequently,

to recognise intractable moral dilemmas, deontological theories have been forced to

jump through a variety of different hoops and to weaken considerably their positions.
In particular,
temporary and

moral

firm principles are almost invariably weakened to something much more

provisional.

is

experience

While

important,

some revisability of principles

in the light of

such wholesale weakening of principles is decidedly
There is a point then in trying to

undesirable, on both logical and moral grounds.

expose some of the tangled assumptions on which the weakening is based, and to loosen
their grip. The assumptions involved on a Kantian approach include these:
H1.

Moral principles - in

those concerning

justice,

particular more

fairness and the

basic principles of obligation, such as
like - are to be explained at bottom as

principles of rationality, or rational behaviour for persons.
H2.

Rationality is a matter of consistency; certainly it implies it.

H3.

Consistency reduces to, or can be accounted for, through possibility.

Put together these assumptions yield the theme
H4.

What is obligatory must be possible - ought-implies-can , for short.

So if that gives,

so do some of the assumptions.

but on

moral dilemmas,

It does give, not just because of

So also then

independent grounds.

should the assumptions

give; and they fail not just on the grounds that they lead to themes like H4.

The deeper trouble is that they are based on a super-modal view of things which
combines

modal

assumptions

with

rational

strong

The

assumptions.

combination

is

important, for once the modal assumptions are rejected, it becomes fairly transparent

that

the

rationality assumptions,

whatever

their

merit,

cannot

such things as moral framework or system or course of action.
rational
case,

systems:

removing

much more
modal

is required

assumptions

-

a

than rationality

main

present

uniquely determine

Rather there are rival

for

uniqueness.

exercise

-

In any

destroys

such

rationality route to ought-implies-can.

Finally consistency is neither necessary nor

That it is not sufficient is indicated by a variety of

sufficient for rationality.

consistent worlds which do not conform to minimal rationality requirements.

is

converse

relevant

refuted

which conform

inconsistent worlds

by

meet rationality requirements (see further RLR and OP).

H3

fails,

paradoxes

induces

~0A -> ~(A ° B),

generally

is

not

just

that

the

Modality

removed.

of

paradoxes

with

entailment,

e.g.

anything

all.

from isolated

transmitted

to

are

inconsistent

is

arguments

Not only H2 (as a necessary

assumptions

parallelling

proposition

impossible

inconsistency

It

modal

when

consistency

of

an

fails,

also

is not
40
this indiscriminate way.

However

to rationality

to sufficient reason and the like), and by dialethic theories which

standards (e.g.

condition)

The more

impossibility in

collapse;

ought-implies-can

at

there are

distinct from those supplied by moral dilemmas.

counterexamples to the theme,

The

counterexamples concern obligations (or oughts) which are impossible, in some sense,

to realise,

counterexamples

Op & ~0p)

But

wrongdoings.

most

the

of

possibility

Derivatively then (substituting ~p for

unrealisable obligations.

that

wrongs

are

inescapable,

there are many determinates of possibility

examples

than

concern

also

mere

involve,

in their

logical

possibility

necessitated

(and necessity),

and

less anemic

notions of

correspondingly broader

notions of

original

(and

p in

setting,

necessity than a narrow logical

necessity).

perhaps strained, adaption, and,

in any case, they serve to undermine the point and

Nonetheless they sometimes admit of,

force of the ought-implies-can dictum (which is to link obligation to freedom to do

otherwise),

and

These examples all arise
nature,

circumstances,

its effective use against moral dilemmas.

seriously weaken

thereby

because agents are constrained in one way or another - by

fate,

the

Gods,

their

past

actions -

but

are

not

thereby

released from their obligations or exonerated from wrong-doing.

Consider first very down-to-earth types of counterexample to ought-implies-can,

before turning to more high-flown but not altogether dissimilar examples from our
barbaric past.
’Suppose that I say "I ought to go and see him, but I can’t because I

don’t know where he is”.

There need be nothing inconsistent in such a remark; and,

yet there would be if "ought” always implied ’’can'” (Hare 63, p.52).
same

with

situation

”’I
where

ought

I've

to

return

driven

the

this

car

I’ve

car
so

far

borrowed,

away

that

returning it in time to keep a promise’ (Thomason, p.175).

but I

It is much the

can’t”,

there is

said

no chance

in a
of my

Likewise the claim of the

person who cannot swim that she ought to rescue the child caught in the current but
can't.

One traditionally important class of counterexamples derives from the Catholic
doctrine of perplexity secundum quid.^ Donegan, in defending what he calls 'the

24

Hebrew-Christian tradition’, contends that once one has sinned one May be faced with
situations from which one cannot escape without further wrong-doing (77, p.152). That
♦

is, one may be so placed (through one’s own fault) that one ought not to perform any
of the available actions though one cannot do otherwise. So it is necessary that one

act in a way that is forbidden.

Protestant thinkers, particular Luther, appear to have considered the human race
involved in one grand secundum quid on account of original sin.
’God

punished

justly

sinners

who

not

could

otherwise

do

Luther believed that

than

sin’

that God

and

’provided men with a law they were unable to keep and then righteously damned them
for failing to keep it’. 242 The Calvinists also were committed - but through their
doctrine

of

predestination

predestined

to sin,

unpalatable

conclusion

-

to

rather

they cannot but

by suggesting

Because

conclusions.

similar

humans are

However Calvin attempted to escape this

sin.

that in

some suitable

sense

are not

sinners

compelled, that it is somehow possible for them to do otherwise than they do. J Given

however the theme of predestination this attempt to escape looks unconvincing.

Any stronger determinism that does not also obliterate morality is in a similar
Creatures will have obligations which they cannot fulfil because they

predicament.

are determined

and will

to do otherwise,

sometimes be

bound to do wrong.

Plainly

where this becomes a general and automatic matter, a serious erosion of moral notions

such as responsibility, normally linked to obligation, is bound to set in.

However isolated cases analogous to those of determinism can occur without such
erosion

of

moral

notions

(as

indeed

with

moral

dilemmas

themselves).

rigidifying personality and ossifying character are of this

gets older

Cases

of

A person as she

sort.

may build up a personality so rigid that she cannot escape from it.

At

some point it becomes impossible (at least psychologically) for her to act other than
she does,

yet on

some occasions

she

ought not

to act

in the way she does.

outcome resembles that with original sin though the route taken

Yet even

if we

grant

that the

person ought not to have got

still we are entitled to our moral judgements;

modern morality
*

The

is very different.

into such a position,

as Pigden puts it,

’an enlightened

can allow such violations of ought-implies-can.

Classical Greek morality certainly seems to have allowed for

such violations.

In Homeric ethics a man could be blamed for failings he could not avoid,

unavoidable wrong-doing.

and so for

Subsequently it was taken for granted that impossible moral

requirements could be placed upon a hero, yet he could be blamed for failing to meet

them. Thus the situation of Agamemnon and of Oedipus, both despite their struggles,
despite their awareness of their predicaments and their anguish, were bound, indeed,

fated, to act wrongly, and were duly blamed for their wrong-doing.

25

There have been various attempts to salvage the ought-implies-can theme, through
a residual Kantianism.

presupposes-can

Thus Hare proposes, but does not really work out, an ought-

a

with

theme

weaker

relation

presupposition

logical

replacing

implication (53, p.54); and both Barcan Marcus and Morscher suggest a model theoretic
connection: essentially that for OA to hold true A must be satisfiable in some world,

and so, modally, A is consistent. But within a relevant setting, the satisfiability
is hardly a

requirement

and does not ensure consistency.

constraint,

Once again,

consistency is only obtained at the cost of other more important connections, such as
Nor does Hare’s comparison with presupposition (on p.54) succeed.

O-adjunction.
the

analogy

such

situations

the

worked

implies-can.

as

practical

question

those

and

dilemmas

moral

But (pace Hare, p.55)

as

to

what

to

do

not

would

counterexamples

affording

If

arise

in

ought-

to

the question most certainly does arise in many

cases.

§7.

Resolving

proceed,

in

moral

any

dilemmas,

dilemma

so far

situation

be.

as can

easily

are

Recipes for

what

provided,

enough

decidedly poor recipes have been much canvassed in the literature.

how to

to,

indeed

several

Among these are

deontic ranking recipes, which would have us impose some type of priority ordering on
deontic principles, so that consistency is maintained and in every situation there is

a dominant rule; defeasibility recipes, which would have us weaken all deontic rules
to merely prima facie or other-things-being—equal principles and operate, when things

look

would

going

like

have

wrong,

utilitarian

by

and

assessment;

pure-chance

recipes,

which

situationally but arbitrarily by such methods as

us determine outcomes

coin toss.

Such recipes fail to meet important conditions of adequacy upon any account of

the way resolution is effected, if and when it can be, namely:
A1.

The account

should

sufficiently resemble what rational

choosers do when they

find themselves in a moral dilemma.

Many human choosers have moral principles,
have

not

rigorously ordered

in

such

which, as against ranking recipes,

a way

as

to

exclude dilemma

they

situations,

and

which, as against thorough-going defeasibility recipes, they are not prepared to give

Thus, in addition,

away or modify when they encounter snags.

A2.

The

principles

induce

which

moral

dilemmas

are

not

generally

abandoned

or

qualified, but very often stand.

Naturally

sometimes

moral

dilemmas,

especially

development and revision of principles.

if

they

recur,

do

lead

to

moral

But the logic of dilemma situations does not

force changes or adjustment and none may be made; and typically none will be made,

where dilemmas are resolved,
resolve dilemmas

which they may not be.

The assumption that we must

is probably a hangover from consistency assumptions,

that deontic

inconsistency must be removed - whereas all we have to do is act, or do nothing, in

26

the face of them.

Part of what is wrong with the idea of a predetermined ordering of principles,
the conflict, and also the

they remove

is that

as with utilitarianism procedures,

felt tension, and often anguish, that goes with it.

Even if the appeal to priority

succeed in some cases, it does not in fact work, and
45
general.
For with Lemmon’s complex dilemmas where

ordering of principles does

not

could

succeed,

in

uncertainties enter

be

can

there

preassigned

no

solution

moral

to

the

in

dilemma

virtue

of

higher-order principles or a given ordering of one’s duties and obligations
and the like, because part of the very dilemma is just one’s uncertainty as

to one’s actual situation (p.153).
In any event dilemmas do not go away as they should if the conflicting obligations
were merely prima facie.
Very often someone in a moral dilemma is in a fix and feels
it^^; and doesn’t say ”My God there isn’t a problem, I must simply look up the book

principles are ranked and what’s to be done’,

to see how these

not having been so

prepared.
in which we, as it were, merely

.... dilemmas in which we morally prepared,

have to look up the solution in our private ethical code, are rare ... and in

case

any

of

little

practical

Of

interest.

are those

importance

greater

dilemmas in this class where some decision of a moral character is required
Our predicament is here so described that, whatever we do, even if we do

...

nothing at

that might mean), we are doing

all (whatever

something which we

ought not to do, and so can be called upon to justify either our activity or

our inactivity (Lemmon, p.151).

The

usual

deontological

positions,

which

deontic

weaken

principles

to

provisional or prima facie principles to avoid problems with dilemmas, also play into

utilitarian hands.
reflecting

things

that

of

in

often

is

which

only

are

to act

overriding requirement
character

utilitarians want

For

the

for

such

to represent

most

a way

disguised

(a

sc

part

as to

of

obligation

statements

concerning

distributed, e.g.

are
the

taken

coherent

way

as

derivative;

net

utilities

position

from

would

their

own

the deontic
make

this

Deontic principles such as
reduce

somehow

distributed

type of action which tends to maximize utility.

accounts typically rule out moral dilemmas:
both

apart

or

to

tend

complex

to

be

an action is obligatory when it maximizes (expected) utility and it

is of a recognised

cannot

they
ar .

-

principles as

deontic

maximize utility,

maximisation principle so only for the most part also).

those

all

maximise

utility,

and

indeed

27

Such reductive

for example, an action and its opposite
if an action increases utility then its

opposite characteristically diminishes it.

if all moral injunctions

More generally,

answer back ultimately to just one standard, maximum utility, then there can be no
real conflict

(but

etc.).

for these

It is

apparent conflict through failure to calculate correctly,

sorts of reasons that

the considerable data now accumulated regarding moral

try to explain away,

to

utilitarians deny that there can be

So they are in the awkward position of having to dispute,

dilemmas.

genuine moral

and

only

dilemmas.

But how, when stuck in moral dilemmas, do we proceed to reach a decision? For a
47
decision is called for: those in dilemma situations still often have to act.
The

intuitive procedure in such dilemma situations is not difficult to grasp.

terminology is regularly employed, and is instructive.
to

those

the

in

-

dilemmas

weighed

assessed,

are

The same

Alternatives - outcomes open

up,

the balance.

put in

The

familiar scales picture, with the scale sensitive to evaluative weights, is helpful
- except

it

that

can

suggest,

weighed against another,
whereas there may be

as with

of justice

where one

thing is

two competing principles or

just

that there are

several.

scales

the

options,

In the light of such a weighing up of alternatives,

those in dilemmas then determine what they should do, what is best, or satisfactory,
’If ... we are to act here in good faith,we shall recognise

in the circumstances .

that the dilemma is what it is and make the best decision we can
*

There

is

an

attempt

to

then

intuitive

apply

(Lemmon 62, p.151).

methods,

decision

consist

which

in

assessing and weighing the alternative outcomes open, and maximizing or satisizing on

these

This

outcomes.

already

very

looks

consequentialist decision theory, which is what,

a

like

preanalytic

account

of

in a qualified way, we shall reach

It is in a qualified way for several reasons; not only may no best (or even

towards.

sufficiently good) outcome be determinable - there is nothing approaching a decision
procedure here - but the further decision theory methods apply within moral dilemmas,

so

they

supply,

at best,

a

second

best

resolution,

a

first best result requiring

removal of moral dilemma constraints and so being ruled out.

Those

in dilemma

unworkable

(and

situations usually do not resort to anything as

mostly arbitrary

and often

ineffective)

utilitarian

as

complex and

assessment

procedures. Nor do they surrender to pure chance procedures.

A3.

The account should not give way to irrationalism, but should adhere to rational

decision procedures, which are not necessarily, however, effective.

This requirement rules out pure chance methods, which are only appropriate where it

is a matter

of indifference which way the decision,

if any,

Even where some resort to

methods can give a worse outcome than might be obtained.
tie-breaking

devices

is

appropriate,

that

may

dilemma, but rather to some component of the

not

be

situation.

Otherwise such

goes.

to

the

final

Envisage,

outcome

of a

for instance, a

decision-tree for an issue where a tie-breaking device such as a coin toss is applied
to a high branch and that proves sufficient for rational decision methods to resolve

28

Then it would be irrational to try to resolve the whole issue

the issue concerned.

by coin toss.

Nonetheless

A4. The account should be appropriately situational.
That is, the account should be applicable to each situation on its own merits, taking

It should not simply be a rule­

detailed account of the features of that situation.

for in dilemma cases the rules have ceased to yield a resolution and

based account,

into a

are locked

be

will

fix position.

situational,

it will not

that of situational

be

involve resort to pure—chance methods.
by situation,

situational

most part,

by example)

or example

by virtue of A3 and A4, though the account

Thus,

procedures

ethics where

such ethics

Nor is a situational ethics (ethics situation
For the

of a broader type being recommended.

into operation where a moral

only go

dilemma is

reached; in general, deontic rule-based procedures apply.

The broad outlines of a satisfactory account are beginning to emerge.

Before we consider

apply rational decision methods to each given dilemma situation.

is

how this

to

be

so as

account,

to be said about the

needs

For

to meet apparent objections.

insensitive

those

accomplished more
to

in

switches

language,

both FA and

A,

some

options, and

yet that,

also F~A,

that

it

ought

also

be

that

so

B,

very

idea

of

it should be that B, where B entails

We do not however go on to

one of A and -A, so that not both A and ~A are ruled out.
claim

the

is both A and ~A are ruled out as moral

that

situations,

in certain

in

For are we not only claiming that

decision-theoretic resolution of moral dilemmas.

for

structure of the

things, there is, to

among other

contradiction

a

It should

that

not

both

A

and

~A

ruled

are

out

morally.49 What is being claimed that B, or an adequate thing, is the best thing to
do in the constrained circumstances, that B should be done in that sense, net that B
is

lifted

to

status

obligation

enough in the circumstances,
by

looking

at

you

does not

the

can

B is

In brief, <»B,

"second-best obligation").

determined

(though

imply OB.

acceptable

worlds

worlds where B holds which fail to be acceptable.

say,

if

you like,

that it

is a

the best that can be done or good
is not going to be

When <£>B holds
where B

When

holds,

because

there are

holds is not going to be

settled in general by falling back on deontic rules, because in the interesting cases

the

rules

are

jammed.

But

such deontic rules dees not

being beyond

imply being

beyond rational procedures, indeed procedures that may be represented by way of rules

(though not controlled by deontic rules).

The rational

procedures in

question

are essentially those of decision theory.

What is required is a decision as to what can be done in the circumstances; and this

can

be

determined

by

relevant alternatives.

be

of

something

weighing

up

the

expected

outcomes

and

selecting

among

The selectiin may be of the best in the circumstance;

satisfactory

in

the

circumstances.

The

approach adopted

pluralistic, allowing for both maximizing and satisizing procedures.

29

the

it may

can be

Maximizers will

of course argue that it would be irrational to select anything but the best among the
alternatives (given that it was conceded to be the best); But satisizers will resist

arguments.what

such

elaborates

is

what

upon

we

a

under

do

decision

done

normally

this:- Set out the relevant alternatives,

action

courses of

alternative.

to us

open

Select

a best,

i.e.

way

-

is

-

merely

which

in

outline

the results that the different

list

to.

analysis

intuitive

more

i.e.

lead

will

a

in

theory

Determine

the

of each

likely value

with maximum likely value,

an alternative

or

a

The basic procedure is childishly simple, but filling

sufficiently good alternative.

out the details may be extremely difficult or even impossible in a satisfactory way.
The

in

problems,

normally

order

increasing

alternatives open, establishing their probability,
rankings to them.

things

these

cases,

favourable

more

delineating

the

assigning values or desirability

techniques from decision theory however for doing all

There are

in

are:

difficulty,

of

and

in

less

circumstances

favourable

guestimates can be made or resort had to fuzzy methods.
To descend to concrete applications (where procedures may be clearer than at the

theoretical level), consider again the Platonic dilemma discussed by Lemmon, where I
ought to return my friend’s gun and also ought not to do so.

The alternatives are to

return the gun and have the wife shot, or to return the gun avoiding a shooting but

thereby breaking an undertaking.

the

alternative

first

wife

It is certain, in the circumstances, that under the

will

be

shot,

highly

a

undesirable

outweighing the lesser undesirability of the second alternative.

the

value of

second

Since the expected

first,

exceeds that of the

alternative considerably

much

outcome,

the

best

thing in the circumstances is to return the gun.

However
undertaking

goes

Lemmon

to

seriously

the

return

’to

gun

suggesting that one sacrifices one’s

astray in

utilitarian considerations’

No doubt

(p.152).

utilitarian considerations would bring out (could be massaged to bring out, rather) a

But the fact is that no consideration (or interpersonal comparisons)

similar result.
of

the

utilities

pleasures

and

pains

contemplated.

’’critical

involves

an

(whoever

Lemmon

utilitarian ones.

the

everybody

of

(involved

they

confuses

are),

the

in
is

made,

consequentialist

universe,

or

or

even

needs

whatever),

to

considerations with

be

their

remotely

much narrower

Similarly the recipe Hare suggests for resolving moral dilemmas at

level”,

namely

illegitimate

resort

to

assimilation

act-utilitarianism

of

(see,

consequentialist

e.g.

82,

considerations

p.43),

under

utilitarianisms.

In certain respects Lemmon’s Platonic dilemma is not a good example of a moral
dilemma;
tougher

in particular,
dilemmas,

the

it is too easily resolved, and so characteristic features of

pain,

emerge, moral dilemmas can

anguish,

indecision,

get omitted.

As

thus begins to

be divided roughly into two types, easy and hard.

30

Easy

dilemmas are those that are comparatively easy to find a way through eventually, and

But hard ones are those that leave people caught in the dilemma

so are resolvable.

at

least

uncomfortable

irresolvable.

they

whatever

to

choose

and

do,

are

so

in

a

good

sense

Though a way-through or a way-out may be charted by decision methods,

it is a way that leaves those involved morally uneasy, dissatisfied, or worse.

For

at the far hard end of the spectrum lie those dilemmas of tragic dimensions.

Although

consequential

decision

rationally within moral dilemma

theory

provides

a

for

model

how

proceed

to

situations, by no means everyone confronted

with a

moral dilemma will apply such procedures in an intuitive way, even though many will.

For

one

thing,

there is plenty of scope for

bad faith,

irrationality and

and more

generally, for non-rational procedures; and severe moral dilemmas often enough induce

non-rational and even paralytic or self-destructive modes in those who are caught in

them.

another,

For

there

are

indirect ways of obtaining the effects of decision

theory, analogue decision theory so to say.

Counselling, whether through a friend or

a professional counsellor, is an important analogue: there the discussive mode helps

to

bring

out

eludicating

of

and

assemble

alternatives,

undesirability of each.

the

and

elements

the

of

the

advantages

decision

and

approach,

the

desirability

and

theory

drawbacks,

Alternatively, or as well, these elements may be bought out

by reflecting on a range of similar situations and experience there using these cases
as

some

sort

of

precedent,

in

particular,

for

determining

value

rankings

of

alternatives open.

Moral dilemmas have corollaries for moral ways of doing things.
is that perfectibility of the moral scheme of things is impossible.

One corollary

Any sufficiently

rich set of rules risks inconsistency, and will be inconsistent if completed.

dilemmas are thus inevitable
be;

in the larger scheme of things.

Moral

This is as it should

for it is in this way especially that moral development occurs and moral stature

is obtained.

31

^A much earlier version of this paper was presented by the second author at the

Australian National University in 1975. Subsequently much of the paper was lost; what

survived is largely incorporated in the first three sections, especially §3. However
the new version takes over the leading ideas of the initial version. The authors are
much indebted for comments and criticism to C. Pigden, R. Goodin and G. Priest.

was

^This

even

true

Maliy’s

of

pioneering

system

of

1926

which

permitted

replacement of logical equivalents within the scope of obligation function 0 (see F &

Von Wright’s seminal work

H p.3).

(in 51),

which

set the pattern

for mainstream

deontic logic, was explicitly based on the modal comparison of deontic functors with
modal

functors,

and built modal

that break the modal

systems,

exceptional

assumptions

in

at the bottom of the

theory.

The

connection, are deontic systems based on

relevant logics on such systems see RLR, chapter 8.

^These paradoxes are explained,

obligation,

paradoxes of derived

and their

see

modal bases exposed,

On the

for some of the

p.224; and

Prior 62,

further

in §5.

considerable literature on paradoxes like those of the Robber and Good Samaritan, see
anthologies and

the Hilpinen

A detailed uniform treatment of

especially Vermazen.

all these and other deontic paradoxes in the framework of relevant logic is given in

RLR II.
4In its equivalent form, Op => ~Fp,

i.e. what is obligatory is not forbidden, it

is the dubbed ’’Bentham’s Law” by von Wright 81, p.2.
5In

a

way,

furthermore,

avoids

that

the

,

problems Hare’s

and

complexities

attempt to transfer a levels-resolution to deontic paradoxes would lead to if worked
out

paradoxes.

the

compare

properly:

any

In

counterintuitively,

case,

that

problems

of

semantical

if

we

were

even

’ought’

is

ambiguous

levels

say,

to
as to

Hare

as

levels,

avoid

to

some

semantical

proposes,

but

tinkering would

still be required to fix up a deontic calculus at each level.

Moreover, there are good grounds for doubting that the so far ill-characterised and

artificial distinction of levels that Hare wants to impose can be satisfactorily made
out.

For

one

thing,

critical hardly

the

holds up;

level
for

separation

intuitive

of the

paradoxes

intuitive
are

from the

typical of

rational or

act utilitarian

assessments at the upper level, and we can be rational and critical, and ask for why-

explanations, at both levels.
^There are various different
form, OA, OB ->

^It

is

this principle can take,

e.g. also a rule

0(A & B), and a rather useless material form, OA & OB =>. 0(A & B).

tried

criticised below.

forms

however

by

Schotch

and

Jennings,

p.152;

their

argument

is

It is also suggested by Barcan Marcus, p.134., and toyed with by

32

Williams
o

In particular Lemmon, taking off from Sartre, has argued this in two important

papers, neglected in mere recent philosophical literature on moral dilemmas and moral
conflict.

earlier

Recent British

work,

in

literature

that

it could

logical

further the next note).

on conflict of values has

particular,

profitably have

taken

into

ignored

account

(see

Such an uninformed approach is however not uncharacteristic

of British philosophy.

^Especially Lemmon’s work 62 and 65.

But although Lemmon suggests (in 62, e.g.

p.150) that there are no logical problems in accommodating moral dilemmas, including
those of the form 0(A & ~A), none of the systems of deontic logic he presents (e.g.

in 65), not even the very weak systems, can do so without trivialising.

The reasons

for collapse are the modal ones already explained.
Apart from the work of Lemmon and a few who follow him, work which at least suggests

the need

for a bridge even if it does not offer so much as good working plans for

one, the literature on moral dilemmas tends to bifurcate - into purely philosophical
material which

does not

address the

logical

issues,

and

largely

logical material

which characteristically avoids the crucial moral issue of dilemmas, which the logics

espoused cannot handle (and maybe because of that: otherwise the logical literature

On the regrettable lack of interaction between deontic logicians

disputes the data).

and ethicists, the latter being the worse offenders, see also Thomason’s notes.
10

In

fact

the latter

implication of O-transmission could be much weakened for

the purpose of the arguments, e.g. the form A

B - > OA D OB would serve to make the

point.

11

Gumanski 80 affords a good example of the complication resulting, complication

readily removed by relevant logic relocation.
12

The argument

thoroughtly

for this

fallacious.

’principle of deontic contingency’

Either

an

in F & H, p.13, is

illegitimate quantificational

conversion

(from

some to all) or a modal confusion (from p & ~p to q & ~q) is used to get from P(p &

~p) for some p - which is all the denial of deontic contingency gives - to Pq, that
is, to ’everything is permitted’.

1^For see axiom, (D1) Op =>~0~p, of the standard systems as formulated in F & H,
p.13, and use the interdefinability of P and 0, e.g. PB

4

1

~0~B.

There would

be little point in presenting features of the epicycling systems

of types (iii)-(v)

in any detail since all will be quickly dismissed. In any event,

details can be found

in the Hilpinen surveys

33

(71 and 81),

texts which are ir

fact

restricted to

epicycling positions

within

the dominant modal paradigm for

deontic

Even Schotch and Jennings, who do question some minimal principles, remain

logics.

staunchly

the

within

modal

sphere,

their

as

possible

worlds

and

semantics

their

substitutivity principles reveal.

15

F & H, p.8,
in

support

p.16. This consistency interpretation appears to have extensive

Scandinavia.

Of

course

if

consistency were meant

in a different

sense

(that of regularity), it would be different. But it is not: see the development in F
& H, p.l6ff.
16

Cf.

Lemmon:

philosophers

that

’It

...

does

the

not

to

seem

adoption

of

have

been

new morality

a

associated with the confrontation of a moral dilemma.

much
by

observed

an agent

ethical

by

is

frequently

Indeed, it is hard to see what

else would be likely to bring about a change of moral outlook other than having to

make a difficult moral

decision’

(62,

p155).

not

settled

without

so to speak

Similarly Barcan Marcus (p.12): ’although

idealistically, a source of moral motion.

dilemmas are

is,

Moral inconsistency

residue,

the recognition of their

reality has a

dynamic force.

17

In 1975 when the paper was first drafted and presented this example was still

topical.

However at most times in modern human history there are many candidates to

substitute for variable V.

18

In the narrow scheme of standard deontic logic, these would be recast in terms

of the forbidden functor F, defined as 0~.

However in relevant logic settings, it is

a further question, not settled by the framework whether W reduces to 0~.
*
^Williams

20

are

Jim-Pedro example (in 81) is of this type.

As there are other war dilemmas, and especially nuclear war dilemmas, so there

other

violence.

pacifist

dilemmas,

e.g.

the

problem

of

violence

to

prevent

greater

On nuclear war dilemmas, and on the main pacifist dilemma, see the War and

Peace series.
21

For many

of the

contractual arrangements,

isomorphism,

known

moral

dilemmas

we consider,

especially those

there are analogous legal dilemmas.

preserving

very many

important

logical

generated by

For there is a wellfeatures

between

moral

obligations and rights and wrongs and corresponding legal obligations and rights and
wrongs.

The complexity of legal

frameworks, the relatively independent design of using

parts and

divine

numerous

lack of design control

legal

of

likelihood

the

dilemmas

and

sources

decidedly

complexity

to

high.

analogous

An

is

dilemmas

moral

for consistency,

or checks

argument

considered

in

render the
fjrom

the

McConnell,

p.275ff.
pp

Lemmon’s final
treatment,

but

(fifth) type of ’’moral dilemma” does so far resist more formal

is

that

the

mainly because

class concerned

description and lacks even clear working examples.

so far

These are,

defies adequate

roughly, situations

‘in which an agent has to make a decision of a recognisably moral character though he

is completely unprepared ... by his present moral outlook’ and has somehow to fashion
a

new

moral

outlook.

is

There

so

far

of

here

nothing

especial

paraconsistent

concern.

By no means all the types of situations falling under Lemmon’s ’classification

of moral

dilemmas’

(62,

classifying types of
first class and

p.156)

are dilemmas.

situation’,

‘moral

the classification

starts out

is nothing dilemmatic about

there

and

which Lemmon

about the second,

little

Indeed

the

‘does not wish to call ...

moral dilemmas, because [as with the first] in all the cases our moral situation is

perfectly clear’

(p.143). The main types of moral dilemmas Lemmon discerns, the third

and fourth classes, are precisely those treated in the text.

The concurrence even of several different,

23()f course the control is limited.

perhaps apparently independent, logical approaches is no sure guide to correctness of
an

explication or

has to answer back satisfactorily to

The matter still

account.

preanalytic principles.

22*
In
dilemmas,

the initial

version of this paper

three

into

types

(0,

0~;

complication was based on a mistake,
collapse.

are

F,

a

rather

F~;

and

artificial classification of

0,F)

was

proposed.

But

this

that equation of 0~A and FA would lead to modal

And, apparently a more natural and simpler theory does emerge if 0~ and F

equated.

Naturally definitional

than coentailment.

More

highly

equation

intensional

in this context does not

differences

between

the

imply more

notions,

which

there are, remain unaffected.

What holds

for F does not automatically transfer

independently.
^The labels

and Jennings.

[RM] and

[K], convenient

to P. Perhaps P should be treated

in what

follows, are

taken from Schotch

Note that although symbolically virtually the same as the originals,

they symbolize different determinate principles.

^See as well as RLR I, Routley 72, Routley and Meyer 73 and OP.

35

the

2?Under

four-valued American plan

for

relevant logic

* can be

semantics,

avoided. But here the superior two-valued Australian plan is followed.

2&Though here we simply take the rule over from standard deontic semantics, it
can be argued in several ways, explained (though usually for the analogous necessity

functor) as

in RLR.

class of such A,
b

One way is this:- in situation a, where OA holds, consider the

i.e.

{C:

I(OC, a) = 1}.

This gives an initially acceptable set up

with respect to a; and indeed would yield the sought rule (with S improved to a

but for

1-1 relation)

fail,

conditions

the fact that bo may not be a world because required closure

conspicuously

We

primeness.

can

however

extend

though not

ba,

uniquely, so that world requirements are met. The picture looks like this:

S is the composition of the function (from a to b ) and the extension (to the b-’s).

The b.’s are the accepted worlds vis-a-vis a.
^The point and role of evaluative step-down is explained in Routley 83.

3°The

strategy

resembles

that

earlier

in

Routley

and

Meyer

75

in

Plainly there is room for other postulates,

formulating explicitly dialethic logic.
e.g.

exploited

In virtue of the model theorem Ot, where

0p2 where p2 is not dilemmatic, etc.

t is any tautology, such a postulate is guaranteed in standard deontic logic, but it

is rightly left as an option in relevant deontic logic, and also in Barcan Marcus.

third

31A

type

of

argument

for

rejecting

[KJ

is

implicit

in

Schotch

and

and also in Barcan Marcus, namely that without [K] both moral conflict and

Jennings,

the Kantian ought-implies-can principle

can be

retained.

That

supposes of course

that the Kantian principle is a desirable one, worth going considerably out of one’s

logical

way

disaster

to

retain.

[KJ

But

the

is not

from the combination of Kantian

only

in

principle involved

theory with moral dilemmas:

reaching

other equally

solid and respectable principles than [KJ could be considered for rejection.

^2At this

logic

and

point we

do detour

analyses of moral

beyond our

dilemmas,

to look

critique of modalisation of deontic

at some

separate puzzles

in deontic

logic - which have also been used against the idea of a formal treatment of ethics.

And

the

approach

we

take

to

these

further

36

puzzles

-

by

limiting

transmission

in the fashion of weaker, and deeper logics - is largely independent of

principles,

the reaminder

IN short, we could have taken a different approach to

paper.

of the

these transmission puzzles, without varying the rest of the paper.
33

Of course it can be argued that one is storing up problems for (much) later on

when the

logical theory

repenting have

fuller

a

better developed,

is much more fully developed, and predicates like those of
logical role.

so also will be

presumably when

But

the

logical

theory is

the distinctions between types of predicates and

those sustaining transmission.

34

It is common practice indict deontic logic by so stating the deontic ’'paradox"
that they smuggle in much more than a satisfactory entailment relation could warrant

and more even than a modal relation would grant.
paradoxes

will

be

presented

as

arguments

Here is an example. These [deontic]

against

Consider the Good Samaritan paradoxes formulated

the

"standard"

as follows:

deontic

logics.

1) A is obligated to

help an injured criminal
2) If A is obligated to do B then A is obligated to do whatever B entails.

(This is

a principle of the standard deontic logics.)

3) There could not be an injured criminal if someone did not commit a crime.

So 4) A

is obligated to bring it about that someone commit a crime. (From 1- 3 by the

standard deontic logics.)

The

So 5)

logics are

standard deontic

(For

flawed.

4)

is outrageous,

though its

premisses are not.)
Entailment principles do not justify the importing of 'bringing about' in clause 4).
35
And wrong to rob him and not rob him, and so on.
Conjunction is not here
connexive with one conjunct undercutting another - though connexive deontic logic is
what some seem to be seeking.

^Accordingly Morscher's retreat to "rational norms" and "rational legislating"
is

But

unnecessary.

presupposed

again

that a deontic

Morscher

the

grasps

key

point,

that

von

Wright

has

logic must contain certain consistency-type principles,

and that 'this presupposition need not be fulfilled' (p.98).
37

Schotch and Jennings,

community of moral

philosophers'

philosophy,

analysis.

where

claim 'widespread support for (a) among the

(a)

is the theme 'that cases of real moral

Most likely this claim, like most such sociological claims

conflict cannot occur'.

in

(pp.152-3)

is merely

impressionistic,

and

not based on

any sort of survey or

But we expect further investigation would tend to confirm it.

no
J Thus, to take just one example:
there appears

to

be

a

conflict

'One may be in a situation of course in which

of principles each of which appears categorically

37

binding - but natural reflection will ultimately show that one or other must prevail’

Benn (p.5). But examples abound.

The reason reflection removes inconsistency is that obligations do not collide.

We have it from Kant himself
Because however duty and obligation are in general concepts that express the
objective practical necessity of certian actions, ... it follows that a
conflict of duties and obligations is inconceivable (obligationes non
colliduntor).
(Metaphysis of Morals, cited in Barcan Marcus, p.323.)

But

does

it

world

the

(merely)

practical

follow

not

of

without

action

However

necessity.

sweeping

a

but

the worlds

concering

consistency

assumption

picked out

by function

not

of objective

these worlds are, moral dilemmas certainly appear to

show, sometimes inconcistent.

There are

depending what is

the explanation can go,

various different ways

and

contractual

arrangements a route like that of Hobbes through Rawls will be followed.

Given less,

taken

e.g.

granted.

for

only

grounded

example,

For

or

preferences

given

choice

individual

behaviour,

interests

a

semantical

route

might be

And so on.

tried.

110 The

inadequacy of modal

accounts of inconsistency is explained in detail in

RLR, chapter 5.
ji -1

These examples and most of those that follow were suggested by C. Pigden, who

also supplied the references.
HO

The

first quote

is

from Kenny,

p.75, the

second from Skinner,

p.7,

both of

which furnish additional details.
4^Cf. Kenny, p.78.

24/4 According

to

MacIntyre

66.

The

situation

of

Agamemnon

and

Oedipus

are

remarked upon in Williams 81, p.80.

There

there

are

is

not

a substantial

and

cannot

literature

be general

arguing,

not entirely

universally applicable

principles in all circumstances of their application.

conclusively,

rules

for

that

ranking of

Thus, for example, Ross, p.41,

also discussed in Barcan Marcus, p.124, Barcan Marcus, p.135, Hare 82,

pp.32-4.

It

seems that such a result would be proved, as regards effective rules, by methods of
reversion theory.

38

^The
relevant;

points

Hare

makes

concerning

anguish

remorse

and

(82,

pp.29-30)

are

they apply also against his own proposal to treat all principles as prima

facie.

But points concerning anguish and guilt are not decisive;

account

of

how

these

attitudes

arise

can

be

given

from

for one thing, an

modal

a

angle

(cf.

the

discussion in Barcan Marcus, p.127).

^By no means all dilemmas require resolution, e.g. those of no practical import
such

as

purely

theoretical

dilemmas

of

freedom,

etc.

liberty,

These

can,

in

principle, just stand.
48 As Lemmon

explains

(in

62),

really

following Sartre,

there

is considerable

scope for bad faith and opting out of our responsibilities here, e.g. pretending that

no decision is called

for, or

that another has preempted the decision, or that the

decision is otherwise determined or out of our own control.
49No doubt we could tough it out dialectically, but the basis for doing so is

Any

lacking.

adherence

contradictions

to

calles

for

especially

rigorous

justification

500n

satisizing

decision

theory,

and

the

dialectic

between

maximizers

and

satisizers, see MS.

Fuzzy decision theory is however only in the initial stages of development at
present.

5^This is just one of several reasons why Hare’s grand synthesis (in 82) of Kant

and the utilitarians,

initially appealing in its two level tidiness and simplicity,

goes down the philosophical drain. A main reason is that the synthesis is built upon,
and

incorporates,

consistency and

modal

assumptions,

and

so

is inadequate

moral data to be accounted for theoretically (see further War and Peace II).

to the

REFERENCES

R. Barcan Marcus,
121-136.

Journal of Philospohy 77 (1980)

’Moral dilemmas and consistency
,
*

S. I. Benn, ’Public and private morality: clean living and dirty hands’ in Public and
Private in Social Life (ed. S.I. Benn and G.F. Gaus), Croom Helm, London &
Canberra, 1983, 155-182.

A. Donegan, The Theory of Morality, Chigago University Press, Chicago, 1977.

D. Follesdal and R. Hilpinen, ’Deontic
pp.1-35; referred to as F & H.
B. C.

logic:

van Fraassen, ’The logic of conditional
Logic 1 (1972) 417-38.

an

introduction’

Hilpinen

in

obligation’. Journal

1971,

of Philospohical

L. Gumanski, ’On deontic logic’, Studia Logica 39 (1980) 63-75.
S. Hampshire (ed), Public and Private Morality, Cambridge University Press, 1978.

R.M. Hare, Freedom and Reason, Clarendon, Oxford, 1963.

R.M. Hare, Moral Thinking,
Conflicts’).

Clarendon, Oxford,

1982 (see especially chapter 2 ’Moral

R. Harrison, Rational Action, Cambridge University Press, 1979.

R. Hilpinen (ed). Deontic
Dordrecht, 1971.

Logic:

Introductory

and

Systematic

Readings,

Reide1,

R. Hilpinen (ed), New Studies in Deontic Logic, Reidel, Dordrecht, 1981.

A. Kenny, The God of the Philosophers, Oxford University Press, Oxford, 1979.

E.J. Lemmon, ’Moral dilemmas’, Philosophical Review 71 (1962) 139-58.
E.J.

Lemmon,’ Deontic
(1965) 39-71.

Logic

and

the

T. McConnell,
’Moral dilemmas and
Philosophy 8 (1978) 269-87.

logic

of imperatives’,

consistency

in

ethics’,

Logique

et

Canadian

Analyse 8

Journal

A. MacIntyre, A Short History of Ethics, Routledge & Kegan Paul, London, 1966.

of

39

A. A.

Martino (ed.), Deotnic Logic, Computational
Systems, North Holland, Amsterdam, 1982.

Linguistics and Legal Information

E. Morscher, ’Antinomies and incompatibilities within normative languages’ in Martino
82, pp.83-102.

Oxford English Dictionary, Oxford University Press, 1971; hereafter OED.
G. Priest and R. Routley, On Paraconsistency, Research
Australian National University, 1983; hereafter OP.

in

Papers

Logic,

#13,

A.N. Prior, Formal Logic, Second edition, Clarendon Press, Oxford, 1962.
A.N. Prior, ’Escapism: the logical basis of ethics'
(ed. A.I. Melden), Seattle, 1958.

in Essays in Moral Philosophy,

W.D. Ross, The Right and the Good, Oxford University Press, New York, 1930.

R. Routley, ’War and Peace I. On the ethics of large-scale nuclear war and war­
deterrence, and the political fall-out’, Discussion Papers in Environmental
Philosophy,
#5, Research School of Social Sciences,
Australian National
University, 1983.

R. Routley, ’On the alleged inconsistency and moral insensitivity and fanaticism of
pacifism’, Inquiry 26 (1984)
,
; also War and Peace II.
R. Routley, ’Maximizing, satisficing, satisizing: the difference in real and rational
behaviour under rival paradigms’, Discussion Papers in Environmental Philosophy,
#10, Research School of Social Sciences, Australian National University, 1984;
hereafter MS.
R. and V. Routley,

’The semantics of first degree entailment’, Nous 6 (1972) 335-59.

R. and V. Routley, ’Semantical foundations of value theory’, Nous (1983) 441—56.
R. Routley and R.k. Meyer, ’The semantics of entailment I’, in Truth, Syntax, and
Modality (ed. H. Leblanc) North-Holland, Amsterdam, 1973» pp.199-243.

R. Routley and R.K. Meyer, ’Dialectical logic, classical logic, and the consistency
of the world’, Studies in Soviet Thought 16 (1976) 1-25.
R. Routley and others, Relevant Logics and Their Rivals, Ridgeview, California, Part
I (1982) and Part II (1984); referred to as RLR I and RLR II respectively.

A. Ryan, The Idea of Freedom, Oxford, 1970.

P. Schotch and
pp.149-62.

R.E.

Jennings,

’Non-Kripkean

deontic

logic’,

in

Hilpinen

81,

40

Q. Skinner, The Foundations of Modern
University Press, Cambridge, 1978.

Political

Thought,

Volume

11,

Cambridge

R. M. Thomason, ’Deontic logic as founded on tense logic’, in Hilpinen 81, pp.165-175.
B. Vermazen, 'The
(1977) 1-71.

logic

of

practical

'ought’-sentences’,

Philosophical

Studies 32

B. Williams, Moral Luck, Cambridge United Press, 1981.

B. Williams, 'Conflict of values' in Williams 81, pp.71-82; also in Ryan 79.

G.H. von Wright, An Essay in Modal Logic, North Holland, Amsterdam, 1951.

G.H. von Wright, ’On the logic of norms/ and actions’, M. Hilpinen 81, pp.3-35.
G.H. ven Wright, ’Norms, truth and logic’ in Martino 82, pp.3-20.

TYRANT

MOPE ON NOPAL DILEMMAS

1.

ethics.

2.

Revolutionising

Possible heading, why the failure to comprehend moral dilemmas.

All these Reading force^'^^e off in other directions.

Undoubtedly a great deal of ethics has been messed up by a failure to understand

moral dilemmas.
' ...

The failure is often reported; for instance
human

since

justifiable claims ...

and

are

rights

claims,

they

be

outweighed

by

other

As in all conflicts of moral principles, human rights

reasons must be weighed

other moral

can

against

one another by an as yet

poorly understood and ill-defined process' (Rajfles, p.42).

The failure shows up on the apparent clash between doing ethics by way of principles
and

doing

ethics

consistency reasons,
principles.

vor principles break down,

through examples.
or dilemmatic

And

examples.

examples

it is

thought for

air^. thought

to

refute

A proper approach to dilemmas removes the tension between examples and

principles and allows a synthesis to be achieved.

3.

Moral

rectitude.

Now

is

notion

this

to

be unpacked0

mhere

are

competing

explanations, e.g.
1) in terms of obligation and deontics;

2)

by

way

of

consequentialist

rectitude decisive.

assessment.

Pettit

to

make

moral

mhen it requires assessment thus

3) Method 1) unless dilemma; otherwise method 2\
rectitude so it is absolute in terms of O.

4.

wants

This is impossible.

Mcjlly's system has no moral dilemmas, since ~fu

For it proves trivial deontically.

But he also wants to run

0~u^ - at least intentionally.

Aquinas denies certain moral dilemmas, e.g. Np &

F~p, in the doctrine concerning perplexus simpliciter F & N p.12.

5.

Investigate further the connection of permitted and accepted worlds,

and

commission?

This

'possibly permitted'

does

not

involve

discussed in NNL.

the

fallacy

of reducing

permission

*
'permissible

to

It does tie with the suggested connection of

commission and omission through the * function.

2

6.

Is problem of the individual in society, of a person in politics, an organisation

or a company of this type i.e.

dilemma

of

freedom?

Of.

a dilemma that can just stand, like the theoretical

also

Machiavellian

a

problem:

person

in

public

must

(morally?) act evilly sometimes.

7.

of PUBLIC

distinction

The

moralities.

VS

MORALITIES

PRIVATE

implies

that

there

are

two

The BASE ARGUMENT against is this:

But there are not two.

if we acknowledge this conception of morality as requiring some action as the
thing to do, how, logically speaking, could there be two moralities0

the thing to do when they conflict?
overriding reason for action?^

How could

morality then

What is

provide an

Similarly &n utilitarianism single maximisation recipe and results.

(/? DOEBN'01

REQUIRE KANTIANISM.)

However A can

to

be subject

conflicting

public

and

private claims.

Seen

this way

^public vs private just generates.heap of further moral dilemmas, and the 2 moralities

^is just

shoddy attempt

to escape

these dilemmas,

expedient (i.e.
would become n

shoves morality out
as in the Hinduism

resolution

dilemmas.

of

often in a way that’s politically

of the way for expediency'1.
The 2 moralities
* int
if
pursued this distinction-making attempt at
/
Something special about public/private case0
Or

differently, why are arguments from expedience
so much more acceptable in public sphere0

Be^n gives

r

interest

some reasons, but these oughtn't to justify a difference. | public interest

DIRTY HANDS PROBLEM

(PARADOX) is like GOOD SAM?

behalf of the statecommitting murder.

Sometimes it

is right

get one's hands dirty.

E.g. Doing such and

such on

OR just a straight dilemma?

to try to succeed,

and then it must also be right to

But one's hands get dirty from doing what it is wrong

to do.
And how can it be wrong to do what is right? Or how can we get our
hands dirty by doing what we ought to do9 Z M. Walzer, Philosophy and Public
Affairs,

moral

2,

10°^,

dilemma.

164.

Op

Walzer sees this as a tragic dilemma:
p-> q

Oq,

but

0~q.

Example:

but really a

torturing

captured

terrorist to extract information to save innocent lives.

The problems start from questions of relations between two different moralities.

Conclusion:

8.

Conflict

conscience 'that,

the

public

to

at its most liberal,

loyalty,

overriding moral

liberalism

in

end.

to a
On

between

will stand

the

individual,

tried

committed concern for its collective interest,

the

one hand

the

individual

as a

relate

this

ambiguous model

of the

a

individual and

as an

autonomous,

citadel,

on the other, the

overriding perverse, wilful obstacles in the way of the general will.
to

with

no gainsaying, and the claims of

possessed of liberties that only he himself can waive, by consent;

State,

equipped

society,

not

T have
only to

liberal conceptions of what belongs to the private and what to the public, but also
to the tension between a morality of principles and one of valued outcomes, which, in

my view, is very closely connected with the ambiguity of the model
.
*

("Renn?

9.
Argument for weak paraconsistent position.
Morality is a guide to action.
Action is consistent.
So have to r?l consistencize.
^Inconclusive"1.
Like law one
consistencizes (e.g.

for action^ alright.

But that confirms that it is inconsistent

in the first place.

10.

The

difference.

fought distinction

is,

it

It hands the opposition its case.

is objected,

a

Simply not so.

distinction

without a

The following have been redacted from access file (PDF) due to copyright restrictions.
•
•
•
•

•

Letter, unidentified author to Richard, 27 Jul 1984 re Moral dilemmas. (2 pages (1 leaf))
Reference card and handwritten paper, ? Alchourron and Bulygin (1 reference card + 5
leaves)
Letter, Charles to Richard, undated re New Studies in Deontic Logic and Castañeda's
proposed deontic logic. (2 pages)
Cuttings from unidentified publication, The paradoxes of deontic by Hector-Neri
Castañeda (page 46-47), and Some main problems of deontic by Jaakko Hintikka 9page
94-96. (2 leaves)
Letter, unidentified author to Richard, with annotations by Sylvan?, undated, re Moral
dilemmas. (7 leaves)

Ml
Although the sorts of examples we have described seem familiar

enough,

it might well be disputed that they are correctly described

as cases of Op and 0~p, or in the alternative representation as

cases of Fp & F~p.

The main ground for this is the belief that

19.
in any such case one of the parties to the conflict is less wrong,

and that this one can be claimed to be not really wrong at all.
There are however several objections to this move to neutralise
the dilemmas
(1)

First,

wrong party.

there is no guarantee that there always is a less

Both parties may be equally wrong,

no way of deciding which is less wrong.

Vietnam example;

To take

and there may be

our hypothetical

if we withdraw half a million people will die;

but if we do not withdraw the war will continue and half a million

people will die; clearly the example may be elaborated.

There

may be no grounds for choice.
(2) The fact that one party is less wrong than the alternative
is no grounds for saying that it is not wrong at all.

In fact

the case would normally be described as one of choice between " two

evils",

two wrong aciwsese©;

if one chooses the lesser of two evils,

then one still chooses an evil.
ing

One is simply not justified in claim1

that, where p involves killing 10 people, p is not wrong

simply on the grounds that ~p involves killing 15 people,

any more

than one is in the case of preference theory justified in saying
that p, which brings a gift of $10,

is not a good thing simply on

the grounds that ~p brings a gift of $11.

things;
(3)

They are both good

and in our moral case, both are wrong.

If we do say this, plainly it will be very difficult to find

any moral principles which can be asserted with any confidence
universally.

In fact it will probably

be possible to find

hypothetical counterexamples to any statement of the form "All

actions of type x are wrong",

just by producing a case where p is

an action of type x, but ~p is an even worse action.
that p,

although an action of type x,

is not wrong,

It will follow
so we will have

However,

a counterexample.

it seems to be important that we

should be able to assert such general principles, and doing so
gives us an essential way of assessing, morally,
(4)

Finally,

types of actions.

the proposed redescription of the case as involving

only one wrong element,

the one which is more wrong,

leaves out

an essential feature of such a case,

that it is a dilemma,

case where one cannot but act wrongly.

If we omit this element

the case is not properly described,

it is falsified;

a

we may as

a result find ourselves unable to explain many resulting features
of the situation which result from the conflict,

e.g. human

behavioural features >n such a situation involving hesitation,

doubt,

inaction,

perhaps anguish and guilt.

All these will appear

quite irrational.

(5)

Assigning a priority to our moral rules itself does

nothing to resolve the logical problem raised by the dilemmas,
(although it may solve the practical problem of how best to act).
%
For this logical problem iX
the problem of how to arrange our

logic so as to be able to describe such cases adequately and without

falsification or incon-sisteney.
Arranging a priority, so that
A
one alternative can be claimed to be less wrong, does not solve

the logical problem, which is that

whatever happens will be wrong,

that T cannot be identical with I.

The consequence that whatever

happens is wrong cannot be removed unless as well as a priority

assignment one also adopts the principle that what is less wrong
is not wrong at all.

But this principle leads to inconsistency

for then the less wrong action is both wrong and not wrong.

To

be consistent this method must introduce a new concept of wrong
in addition to the old one,

say "wrong2" , and the less wrong action

can be said to be not "wrong2".

But now the original problem is

21.
still not eliminated, because the problem was not that whatever
be
is
.
happened would /'wrong2" (although this/now false), but that
whatever happened would be wrong, which is still true.
sort of priority - dominated concept cannot replace,

may supplement,

the original concept.

And this

although it

Sf-

The problem of classifying these dilemmas.

z

mainly using the term ’wrong’.

We have stated our dilemmas

However, it may seem that this is ambiguous,

and that it is possible, or desirable, to make a further distinction, between
pi inciples which are obligatory and principles which are prohibitory.

For

wrongs can be due to two very different causes - om^iission or commission.

The distinction between obligation and prohibition principles can be
explained initially as follows:

an obligation principle of the form Op

read ’it ought to be the case that p’,Qt

is

f ?
L ' r

jT

f'.Jlz.f:? /t/tjyJ

/4-<

'it ought to come about that p'.

This says that something, p, ought to

should happen that p'A,

r,
happen,

an
ideal situation I.

that is^ that it is a member of

&
A situation^satisfies Op,

Op if p is not in it

then, if p is In it

( p eaJ;and violates

p's presence satisfies Op, whereas

(pZ S?) •

£ ["Obligation"
Z7 —:----- is
’

being used here

its omission or absence violates it.

in a somewhat unusual sense,

4

for Lemmon seems to be right in

urging that obligations are usually contractual in nature.
and contracts may be contracts to do something or to refrain from

However, we

doing something - hence both positive and negative.
must press

'obligation'

and

'obligatory'

and adjective corresponding to
it
*

into service as a noun

ought to be the case that'

in

order to avoid clumsiness and for lack of any other suitable words.
The other class of principles,

hand, we shall write

prohibitive principles, on the other

'Ep' and read

'it is forbidden that p'.

These principles claim that p is the sort of action which should
p is not a member of the ideal situation I.

be avoided

satisfies Fp,

situation a such that p Z

such that pea violates it.

a situation a

whereas

p's omission or absence satisfies

Fp, whereas pSoccurrence or presence violates it.
is paralleled on the

obligations)

The distinction

action level by the distinction between action

and by that between

and inaction,

A

'sins of omission'

and 'sins of commision'
A

(which violate

(which violate
prohibitions).

'I have done those things which I ought not to have done'
violating a prohibition but also,

and not equivalently,

not done those things which I ought to have done',

thus

'I have

thus violating

An obligation is satisfied by doing something,

an obligation.

a

prohibition by failing to do something, by avoiding something.

It is important not to confuse 0~p, which says that
should happen'

or

'that ~p

'It is wrong that p', with Fp which says ’that

p should not happen', or

it
*

should not

happen that p'.

This

distinction is clear, as a scope distinction,

in the English reading

However the two are continually confused in modal treatments of
But
deontic logic, since given Modal Muddle p / I = ~p g I.
that
to say/p should happen is not to say anything about what should not
happen

but leaves the issue as to whether ~p should or should

And to say that p should not happen is not say
and
0~p tells
anything about ~p,/whether it should happen or not.

not happen open.

aj
us that ~p is a member of I, where.Fp tells us that p is not a

member of 1.^
Dilemmas then seem to fall into three classes depending upon

whether the principles which lead to them are obligation or
prohibition principles^or both.
Fp and F~p,

Thus we obtain Op and 0~p,

and Op and Fp from our examples depending upon how the

principles involved are construed^

a<Tprohibitory or obligatory.

(A): Dilemmas obtained from prohibition principles,

Fp, % F~p1.

i.e.

cases of

We obtain cases of this sort given principles of the

*
form.
—

(1) Whenever P satisfies condition C,
(Action form:

(2)

it shouldn't happen that p.

action of type x are wrong)

it shouldn't happen that p.
A

Whenever p satisfies condition C',
The dilemma is

obtained where p satisfies condition C,

~p satisfies condition C'.

Jo O

F

and

-

■L Give the semantical reading of 0 and F?and the examples with
'wrong' .
Then mention the necessity for distinction between 0~p and Fp.

(1)

Wrongs divide into sins of omission and sins of commission.

(2)

These are semantically different.

(3)

Unless one distinguishes these concepts semantically
have the consistency assumption again.

Give the readings.

one will

10.

Examples can also be obtained from two distinct prohibitory
These are cases where the state of affairs described

principles.

(for reasons given by the first

by p is really indescribable
principle)

(for different reasons given by the second principle).

able,

*
(tndesir

For/(B)

and the state of affairs described by ~p is m^-ally

The second category of dilemmas are dilemmas obtained from

obligation principles i.e. cases of Op & O~p,

for some p.

We

obtain cases of this sort given principles of the form:(1)

Whenever p satisfies condition C,

(2)

Whenever p satisfies condition C',

it should happen that p
it should happen that p.

The dilemma is obtained when for a given p, p satisfies
condition C and ~p satisfies condition C':

(C)

Finally,

so Op and O~p.

examples derived from conflicting but not inconsistent

moral principles, which urge both that p should occur and that it
should not occur i.e.
and prohibited.

cases of Op and Fp, where p is both enjoined

These cases

come about where are principles of

the following sort:

(1)

Whenever p satisfies condition C, p should happen.

(2)

Whenever p satisfies condition C', p should not happen.

The dilemma results because one can find cases
one and the same state of affairs

C and condition C'?

e.g.

(p)

(in T)

where

satisfies both condition

as^our example derived from Lemmon's

uAcrw-tt

bolew, x gives back the gun

and helping a murderer.

(p)

is a case both of keeping a promise

But promises must be kept,

so Op holds.

However actions which help murderers are forbidden and should
not occur;

examples,

hence Fp also holds.
unlike the previous kind,

It seems that these sorts of
cannot be generated by a

single moral principle, but requires two principles which come into

conflict.

With a certain degree of artifice we can arrange our

examples into these categories, with f A) groups being prohibitions
the (fe)groups obligations,

and the

(C)

groups mixed.

It is necessary then to avoid the modal treatment and to

But though 0 principles and F principles

distinguish 0~ and F.
are

distinct, 0 and F provide alternatives bases for

(semantically)

The O-principles correspond semantically to

deontic system.

inclusion principles,the F formulated principles to exclusion
Thus e.g.

principles.

TRfH

5® where

iff

iff TROH
*

,

(UH) (TRfH oX(p,H)

= f)

In contrast,
I(Fp,T)

= t

iff

(UH)

TROH

d

Kp,H)

Thus the modellingswill mirror

f)

.

one another, with R

Rf
For the representation of principles then we can work

satisfactorily with an O-basis, or with an F-basis, but not with
a

mixed basis which uses both 0 and F principles.

Failure to

distinguish 0~p and Fp will provide a ground for the mistaken
consistency assumption:

iff p i H.

for it will follow from evaluation rules

12.
The cases described satisfy the characterisation

of moral

For the principles involved are not inconsistent, yet

dilemmas.

they are not satisfiable in T.

In the case of the first set of

leading to Fp and F~p,

dilemmas,

the principles could not be

satisfied unless neither p nor ~p occurred.

necessarily complete,

But since T is

they cannot be satisfied in T.

of the second set of examples,

Op and 0~p,

In the case

the principles could

only be satisfied if both p and ~p occur, hence since T is con­
sistent,

they cannot be satisfied in T.

different.

The third case is very

Here it cannot be satisfied because one and the same

proposition satisfies both conditions for occurring and conditions

for not occurring?

this proposition,

hence our principles direct, with respect to

that it should occur and that it should not occur.

This also cannot be satisfied in T,

since p is either a member of

T, or not a member of T.
The principles involved are not however inconsistent ones;

for

in each case a finite proposition modelling of the principles can
be provided.
In the third case,

directives,

despite the fact there are inconsistent

given a contingent proposition p^, which satisfies the

antecedent conditions, no inconsistency need result because such a
proposition p^ may not occur.

In other words,

though the principles

together with p^ form an inconsistent set, p-^ is essential and can
be deleted from the set since contingent;

and the set without p1

as the principles on their own are satisfactory

is consistent,

where no antecedents satisfy both,

a po£S»4ta-

case.
■ v

r

•

The principles
**

(a.

though inconsistent in application are not inconsistent^

i

In the

first two cases(they are T consistent?) also it is a contingent matter
that the principles come

i

...

into^ cannot be satisfied in T;

r;

for

detachment of a contingent truth is made, viz that the given
?

proposition satisfies the conditions.

13.

since they constantly arise““without sacrificing the generality
of moral principles.

For unless carefully treated dilemmas will/

as we shall see, provide counterexamples to moral principles.

Our moral dilemmas result from judgements which cannot be

satisfied in T,

so that whatever happens will be wrong.

Howe

is
there/still quite a crucial difference between the first two
classes and the third class of dilemma.

In the third case the

result of the dilemma on detachment of a contingent premiss that
p satisfies c and c'

simultaneously is an inconsistency,

cannot both belong and not belong to the ideal set-up I.

since p

Although

the principles upon which the judgement Op and Fp was based were not

themselves mutually inconsistent,
moral judgerent

they do result in a inconsistent

that p both ought to happen and ought not to happen

is prohibited from happening.

Given that p has both properties,

the moral principles concerned are in conflict.

This is brought

out by the fact that if p does satisfy these conditions,
cannot be any situation, T-like or not,

there

in which the principles are

satisfied.

Furthermore here we have a case where each moral judgement
provides a counterexample to the other.

If Op and Fp both

obtained, both the principles from which they were derived would
be false;
principle

for Op would provide a counterexample to the prohibitory

(p)

(if p satisfies c’

then Fp)

since Op

~Fp;

while

Fp would provide a counterexample to the obligatory principle
(if p satisfies c then Op).

(p)

J
\

14.
In view of these points it seems that a moral code which

resulted in this sort of conflicting judgement is defective;

the

principles involved should either be restricted by tightening one
or other of conditionsc and c', or else restated by reformulating

one or other of the principles,
(p) (Cp => F~p)

replaced by

(p) (C'p => 0~p) .

or

i.e.

either

(p) (C'p => Fp)

(p) (Cp => Op)

is

is replaced by

the case is reduced to one of the first

In short,

two cases.
In this case the dilemma arises because the principles

In the first two cases however

involved come into conflict.

the difficulty does not arise from conflicting principles, but is

simply a case where our

not

v/

judgements cannot both be satisfied in T,

because they are conflicting, but because of the peculiar
Where Op and 0~p both hold,

nature of T.

(or Fp and F~p)

what-

ever happens will be wrong, because p & ~p cannot both hold in T,

The dilemma is however simply explained

nor p & ~p fail to hold.

and the judg.ements accepted once it is understood that I,
situation,
action,

is not T-like.

The dilemma may produce a conflict in

because given the logical structure of T and the limitations

it imposes upon action,

both judgements.

it is impossible to act in accordance with

But this does not effect the judgements themselves;

as moral judgenents, assessments of the moral worth of p
of affairs)

conflict.

(a

state

of ~p^jthey are quite compatible and there is no
To say that p is wrong and that ~p is also wrong is

in no way to be inconsistent,

true.

the ideal

and both judgements may very well be

The absence of conflict also explains

the fact that such

cases can be produced from single moral principles,

such as

'Killing is wrong', which in no way inhibit internal conflict.

15.
Judgements such as

'p is wrong

of action or states of affairs;
as to how to act in T.
alternative in T,

Hence Op does not compare p with its

~p.

Such judgements are implicitly comparative,
not to ~p's worth, but to the

Thus if we reach situations as to moral

worth of its absence.

(using moral criteria for selection),

if for every situation H, H+p is worse than H-p.

in a situation

assess types

and do not simply give directions

but p's worth is assessed relative,

desirability

& ~p is wrong'

p is wrong

If p's

presence

makes the situation morally worse than it would be

if p were absent,

then p is wrong.

But now it is a straightforward

matter to claim that p is wrong and that ~p is also wrong;

one is

aHL

simply claiming that situations without p^orally preferable to
the

that

(should be chosen^etc.);

same situations with p

*

and also

situations without ~p are morally preferable to that same

situation with ~p in it.

These judgements in no way conflict,

and if we were in a position to choose a situation with neither of
them in, we could satisfy both judgements simultaneously.
is the choice must be regarded as a hypothetical one,

As it

and the

judgements as representing a kind of thought experiment, rather

than in giving directions as to how to behave in T.

The choice

embodied in such a judgement is one between two hypothetical situations
and not between alternative versions of T.

It is just for this reason that this type of moral judgement
is adequate for assessing morally different types of actions or

sorts of states of affairs.

Because we are concerned with choosing

between hypothetical situations, we can opine that all actions of
a certain specifiable

type and states of affairs satisfying a

certain condition are wrong

(not wrongyetc.), without risking

* In the case of Op & 0~p.
In the case of Fp & F~p one ismaking
a slightly different claim, that a situation from which p's
absent is morally better than some in which it is present.
Note
that the difference between 0 and F can be
to the difference
in directionthe rankings.

16.
falsification because of the contingent features of T.

Finally,

in such a judgement one is concerned with assessing the action
or state of affairs in itself rather than passing judgement about

the best way to act in a particular situation.

Dilemmas of this sort then, as opposed to dilemmas of the
conflict sort first discussed,

are not the result of defective

moral codes or principles and do not require any readjustment or
resolution

(beyond psychological acceptance of the fact that it is

sometimes inevitable that what happens is wrong and that perfectibility
is impossible).

They do not provide counterexamples to general
In contrast

moral principles based on contingent features of T.

codes containing the conflict type of dilemma are defective.

Any

code containing two principles of the form 'wherever p satisfies
Cl, Op'

and ’Whenever p satisfies C2, Fp’

where C1

and C2 are not

inconsistent, will be defective because there is a possible situation

(a hypothetical case)

with a p^ which satisfies both C1

Whence the inconsistent pair Fp,

potentially inconsistent.

& Op,.

and C2.

Any such code is at least

This sort of defectiveness of a moral

code might be excusable if it were inevitable.

But it is not.

can be avoided if we adopt either 0 or F as primitive,

and state

our judgements in terms of one or other of them, but not both.
alternative,

terms,

conflicting

is intolerable.

moral

It

The

principles stated in

No principle could be asserted with any

confidence, because they would hold or fail just according to whether
or not a suitable proposition with conflicting properties appeared
in T and provided a counterexample.

And since this is contingent

whether such principles hold or not would be entirely a matter of

good or bad luck.

Moral judgements on the other hand ought to

apply beyond T to possible situations, ought to be non-empirical

and apply to hypothetical cases,
matter of accident.

and ought not to hold just as a

4

C ‘S’ -X

f *

univeralisa'blemoral principle was a judgement

The general

‘

7^'

about types of action or states of affairs and compared p's presence

in a situation with its absence from the same situation on a scale
of moral desirability.

This concept of wrongness is essential to

be able to state general moral judgements.

But as well as this

concept we can distinguish another important concept which is

designed to assess not types of action but what is the best course
of action in a given situation.

This is obtained by comparing

p's presence in a situation not with its absence but with the presence
of its rivals in a given situation,

e.g. ~p in the case of T.

p will be said to be ’the best course of action'

~p are both wrong)

if a situation

Thus

(even where p and

(usually T), with p in, ranks

higher morally then the same one with ~p in.

A more practical

- concept still can be obtained by comparing p and its practical rivals,
say

will be the best course of action

practical alternative q)
f

Y
r

if a situation

(as compared with its

(usually T)

(morally) when p is in it than when q is in it.
between alternative ways of

extending

will rank higher

One chooses

T.

It is important as Lemmon points out to distinguish this
concept clearly from the more general concept earlier explained.
Lemmon is wrong in suggesting that this is not a moral concept at

all and does not give rise to moral judgements but to judgements
of expedience.

The judgements involved in this case are still

moral in the sense that the criteria for ranking situations are

22

specifically moral ones and not based simply,

say, on aesthetic

considerations or desirability considerations.
clear that this distinction

Yet

it is

(which we shall call the distinction

between differential obligation and intrinsic obligation)

must be

ignored and eliminated if obligation is treated as a modal functor
For give/} the principle of Modal Muddle

(H) (p e H = ~p e H)

semantical recipes we have given will be identical.

p's presence in H with p's absence

(p / H)

comparing p in H with ~p's being in H.

the

To compare

will be the same as

It is essential however

to distinguish this notion of 'best course of action' (differential

basic notion we have discussed, because it has quite different

properties.

*

It cannot replace although it can supplement,

The reasons are as follows

other notion.

02 does not allow for

(1)

the

moral dilemmas.

p with its practical alternative in T, e.g.

For since we compare
~p,

to say that Op &

0~p is to say that a situation H such p e H is morally better than
a situation H such that ~p e H, and also to say that a situation H

such that ~p e H is better than one such that p e H.

This is a

(Similar results will ensue when F is taken as a

contradiction.
basis).

(2)

02 cannot allow for general moral judgements of the form "All

actions

(states of affairs)

counterexample.

of type x are wrong" without fear of

For no matter how compelling our principle and

how repugnant the type of action described, we can almost always
find a case where the only alternative action is worse.

however repugnant p may be,

Hence,

it will not in this case be wrong,

a counterexample to one general judgement will be provided.

new notion then cannot do the same job as the old;
—-----------------------------------------

.

and

Our

it should be

,

We u^e such
as 'the best thing in the circumstances'
'the right thing', 'the correct course all
in all'
to capture
differential obligations .

23.
regarded as parasitic,

because it can be alternatively represented

as admitting the wrongness of the alternatives,

but going on to

weigh up the alternatives to obtain a recipe for practical action.

(3)

implies can" slogan

It is to this notion that the "ought

applies.

O2P o locjically possible p, because one has selected

p as the best course of action from logically possible alternatives
in T.

Indeed

’02p’

can be made to imply

‘p is physically possible'

(or even physically possible to the person being exhorted)

by

comparing p with its physically possible alternatives in ranked

situations.

In a similar way,

the concept of blame attaches to

failure to follow the best course of action,i.e.
if p is better as a course of action than its

and one fails to do it.

action.*

(practical)

alternatives,

One is not blameworthy for doing p merely

only as good as,

because p is no better than, i.e.
alternatives,

one is blameworthy

its remaining

it has to be worse than some alternative course of

It is important to note that blame attaches to this

second differential concept of obligation, rather than the first
intrinsic notion,

for the possibility and obvious unfairness of

someone's being blamed for doing wrong where both Op and 0~p
hold and he has no alternative which is not wrong would otherwise
be a powerful objection to cases of moral dilemmas.

(5)

Finally,

differential obligation makes a basically different

type of judgement from the intrinsic notion.
not of a type of action or state

It is an assessment
or
of affairs, the action/state of

affairs in itself, but of the situation in which it occurs,

action in this situation.

This is reflected in such phrases as

"the best thing in the circumstances"

"the right thing to do

the conditions obtained at the time)", etc.
* Oh well, you cannot be blamed.
circumstances .

the

(given

Hence it is not

You did the best thing in the

universalisable , and does not try to be.

It

“results in a recipe

for action, and is closely allied to advice or exhortation/orders ,

persuasion,

recommendation or other directed activities.

Intrinsic

obligation however/ frafc^s no account of possible action and does not

need to;

it assesses action-types regardless of how they occur in

a given situation.

This last point is important in connection with argument

about moral dilemmas.

For the confusion between our two concepts

has a disastrous effect on argument, particularly those hotly
disputed arguments which occur in the case of moral dilemmas.

The

following sort of argument, due to confusion between Oi and O2

is

all too common in such cases.

Actions of type x are wrong
(1')

(if p satisfies c then Oi~p).

(p)

This must be Oj because

it represents a general and exceptionless
(2)

p

is an action

(3)

~

02

i.e.

~p:

(state of affairs)

judgement about a type.
of type x.

p should not be done.

Therefore do not do p.
The argument is plainly unsatisfactory because the fact that

Oi~p

holds is not sufficient to establish that 02~p

To

holds.

do this one would have to show that the alternative, p, was not worse

or even just as good.
not be so.

And in cases of dilemmas this may very well

To establish 02~p

one needs to show not merely that p

is wrong but that there is some better

course of action.

The

fallacy derives from trying to base a recipe for action on the

intrinsic obligation notion, Oi.

s

25.
A closely related confusion holds that it is inconsistent to
claim that p is a wrong type of action,

that it should be done nonetheless

the best course of action.

August 6 197 x

(e.g.

is repugnant etc.,

but

- because it is nevertheless
Canberra Times Editorial,

on abortion reads as follows:

"It is significant,

and inconsistent with its official stand that the s ynod should
have qualified its recommendation to family doctors with the words
"however regrettable we feel this to be"."
to be inconsistentthen,

one states to be

What is being claimed

is that one should recommend a course of action

(morally)

there is no inconsistency.

regrettable or repugnant.

But of course

One is simply saying that ~p is wrong,

"Oi-p", but that it is nevertheless the best thing to do given the
circumstances because the alternatives are worse?i.e.

"O2P"•

The distinction between differential and intrinsic obligation,
then,

seems to be an important factor in assessing the validity of

argument where moral dilemmas are concerned.

/,

2

ft

I

z^z/ /
rt&jZy

7-,

I