Deontic

The Philosopher

Contrary-to-Duty Obligations and Deontic Analysis of Moral Dilemmas

Hamid Karimianpour

This paper is concerned with the logical structure of moral dilemmas. A purported moral dilemma is a situation where an agent is faced with two (or more) conflicting ought-claims, either because one of the claims is simply the negation of the other or because some contingent aspects of the world make it impossible to satisfy both (or all) of the claims. A standard deontic analysis of moral dilemmas yields a contradiction. The classic analysis include as premises two ought-claims, where according to the principle of standard deontic logic �ought� implies �can�. It also includes a modal premise stating that it is not possible to fulfil the conjunction of both of the claims. From these premises and by the aggregation principle of deontic logic � which conjoins two ought-claims into one (complex) ought-statement � a contradiction is inferred. Various strategies have been adopted to overcome this contradiction. Bernard Williams abandons the aggregation principle of deontic logic. Lemmon, van Fraassen, Marcus, and Foot reject the principle of �ought� implies �can�. Both of these strategies allow the possibility of conflicting ought-claims. McConnell denies the existence of genuine moral dilemmas. Accordingly, he believes one ought-claim overrides all other claims in situations which appear to be dilemmatic. There are good arguments in favour as well as against each of these moves. However, this discussion extends beyond the scope of the present paper. For the purpose of this paper it suffices to say that each of these strategies discards one or more of the principles of deontic logic, each of which makes sense intuitively when considered individually.

This paper presents an alternative solution, without rejecting any of the principles of the standard deontic logic, by drawing a parallel to Hilpinen�s analysis of contrary-to-duty obligations (CTD). As it will be shown later, Hilpinen�s analysis of CTD obligations introduces a distinction between obligations in perfect worlds and obligations in imperfect worlds. The CTD does not present a dilemma, but when it is analysed according to the standard system of deontic logic, a paradoxical situation arises. It will be demonstrated that Hilpinen�s solution to overcome the inconsistency of CTD can also help solve the inconsistency problem of the deontic analysis of moral dilemmas.

The CTD model applies to situations, where an agent�s illegitimate breach of a duty implies a contrary-to-duty obligation. The logical structure of a CTD situation is typically of the form:

?) Victoria ought to go to the assistance of her neighbours.

?) If she does go, then she ought to tell them she is coming.

?) If she does not go, then she ought not to tell them she is coming.

?) Victoria does not go.

If L is defined as L={?, ?, ?, ?}, then it seems that L is both logically consistent and non-redundant in that the sentences of L are logically independent. However, Roderick M. Chisholm demonstrated that a paradox arises from L, if it is analysed according to the semantic of the standard deontic logic (SDL). The semantic is given in terms of possible worlds by a model structure following the normal modal logic of the type KD: M = <W, I, R>. W is interpreted as a set of possible worlds {u, v, w,..}. I is the standard valuation function which assigns the values true or false to sentences at a world in W. For any u ?W, the truth of a sentence A at u under M is expressed �M, u? A�. R is a 2-place relation on W, called the relation of deontic alternatives. Each member of R takes a member of W and returns a subset of W. A sentence A is logically true, if and only if (iff) it is true at every world u ?W for any interpretation M. B is a logical consequence of A iff every interpretation M at any world u such that if M, u? A then M, u? B. Formally, the condition for the concept of obligation can be formulated as:

M,u? OA iff v? A for every v ?W such that R(u,v)

R is a serial relation, i.e.:

For every u ?W, there is a v ?W such that R(u,v)

In SDL, L is translated in the following way:

?) Oa

?) O(a ? b)

?) �a ? O�b

?) �a

where:
a means: �Victoria goes to the assistance of her neighbours�
Oa means: �It ought to be that a�
b means: �She tells them she is coming�

Following SDL, (?) and (?) entail:

?) Ob

and (?) and (?) entail:

?) O�b

and applying the consistency rule (O�A ? �OA) on (?) entails:

?) �Ob

by propositional logic, (?) and (?) entail:

?) Ob & �Ob

Thus the given translation of (?) � (?) in SDL leads to a contradiction. Translating (?) to:

?) a ? Ob

or (?) to:

?�) O(�a ? O�b)

will solve the inconsistency problem, as (?) and (?) do not entail (?), and (?�) and (?) do not entail (?). However, the new translations violate the non-redundancy requirement of L, as (?) is derivable from (?), and (?�) from (?).

As already indicated above, Hilpinen proposes a way out of this paradox. This proposal will be examined in some detail later. It needs to be stressed that although Hilpinen�s proposal seems helpful for the analysis of moral dilemmas, the latter (moral dilemmas) differ structurally from the typical CTD situation. Following (?) and (?) in Chisholm�s example, the execution of b is contingent upon whether or not Victoria fulfils a. To do b is required, if Victoria carries out a, and �b is required, if she fails to carry out a. Another classical example is the obligation to apologise, if an agent i fails to keep her promise. Here, keeping the promise completely releases i from the obligation to apologise. Hence, the classical CTD obligations may be regarded as provisions to minimize the damage, just in case a duty, which ought not to be violated, is in fact violated.

However, moral dilemmas may often involve obligations, the execution of which may not release the agent from other obligations. These obligations are fully independent, but competing obligations when applied to the same situation. Say, for instance, Oscar Bradley ought to do p (in symbol: Op) and ought to do q (in symbol: Oq), where he can only do p iff he does not do q. An example is that he ought to go to his office to meet his student Gareth due to a prior appointment to see him. Before leaving home Oscar realizes that his daughter Veronica has suddenly become ill. Oscar can either go to his office and see Gareth or stay home and look after Veronica. Some may claim that in this situation Oscar�s obligation to look after Veronica overrides his obligation to see Gareth. Conversely, a Kantian view would draw a distinction between Oscar�s perfect obligation to honour his promise to Gareth and his imperfect obligation to look after Veronica. According to this view Oscar ought to go to his office to see Gareth. Both of these claims reject the idea that the above example is a genuine moral dilemma. Note that the Kantian account of perfect and imperfect obligations operates within one and the same world, i.e. the actual world. Hence, it should not be confused with moral theories which draw a distinction between obligations in perfect and obligations in imperfect worlds.

The feasibility of the above views will not be scrutinised here. Whatever the status of the obligation to see Gareth and the obligation to look after Veronica may be, they cannot presumably both be fulfilled in that specific situation. Hence, in compliance with the definition of moral dilemmas given in this paper, the contingent aspect of the world � i.e. the situation where Veronica�s illness coincides with Oscar�s appointment to see Gareth � makes it impossible to fulfil both of the obligations. The fulfilment of one of the obligations requires the non-fulfilment of the other. Thus the example above is taken here as a genuine case of moral dilemma.

The underlying structure of (purported) moral dilemmas can be brought out in SDL in the following way:

1) Op       Premise
2) Oq      Premise
3) p ? �q        Premise

Premise (3) seems to be the source of the inconsistency of the standard deontic analysis. It represents the incompatibility of two ought-claims. In the classic deontic analysis of moral dilemmas this incompatibility is represented as a conjunction of the fulfilments of two incompatible claims � in symbols: �?(P & Q), where �?� is a modal operator denoting �possible that� and P and Q are placeholders for action descriptions. Premise (3) represents, however, this incompatibility as a biconditional, where the fulfilment of one claim is conditional on the non-fulfilment of the other and vice versa. This is based on the assumption that under specific circumstances, e.g. when Veronica�s illness coincides with Oscar�s appointment to see Gareth, there is an inverse sufficient and necessary relation between �Oscar does p� and �Oscar does q�, i.e. the two sentences are inversely materially equivalent. In reality, of course, this may not be the case. Oscar may do p and q still does not obtain, vice versa. In this case no paradox emerges from the SDL analysis, hence there is no puzzle to be solved. However, a paradoxical situation similar to the case of CTD obligation arises, if premise (3) is assumed. The basic structure of premise (3) is a conjunction of p ? �q and �q ? p. However, (3) has been written in biconditional format for the ease of analysis.

Following the rule of necessitation for deontic logic, if (p ? �q) is provable, then it is provable that:

4) O(p ? �q),

and (4) and the rule of equivalence distributivity of ought (O(A ? B)/(OA ? OB)) entail:

5) Op ? O�q,

from (2) and (5) it follows:

6) O�p,

applying the consistency rule on (6), entails:

7) �Op,

finally, by propositional logic, (1)-(7) entail:

8) Op & �Op.

Thus, SDL analysis of (1)-(3) leads to a contradiction. In English (8) means:

�Oscar ought to do p� and �It is not that Oscar ought to do p�.

Solution to the problem

As is the case with Chisholm�s example, the source of the paradox which arises in SDL is that Op and Oq are based on a deontically ideal or perfect world. In such a world the fulfilment of one ought-claim does not require the non-fulfilment of another. A deontically perfect world is a world u ?W, where every ought-claim is obtainable independently. Premise (3) is not applicable in this world, thus (2) is redundant. Premise (2) is effective in a deontically imperfect world, where (3) is the case and (1) violates (2) under condition (3).

To solve Chisholm�s paradox, Hilpinen suggests replacing the truth-functional conditional in premises stating contrary-to-duty obligations with Lewis-type intensional or subjunctive conditionals . A subjunctive conditional, symbolized �?>?�, states �were ? true, ? would have been true�. This suggestion helps avoid the inconsistency in CTD analysis by introducing a division between a perfect world, where Victoria fulfils the obligation to go to the assistance of her neighbours, and an imperfect world, where she fails to fulfil that obligation. This analysis will not be pursued any further in this paper. The matter under investigation here is the plausibility of subjunctive conditionals in the analysis of moral dilemmas based on the interpretation that �were the world perfect, it would have been the case that Op�. This interpretation follows directly from the general R-relational principle of SDL, which holds that for any consistent proposition A, there is a nonempty set of deontic alternatives.

Conditionals are evaluated in terms of closeness of possible worlds. Subjunctive conditionals present counterfactual possible worlds. ?>? is true iff ? obtains at every member of some class W of ?-worlds such that every member of W is closer to (more like) the actual world than is any ?-world that is not in W.

It is an important philosophical question as to whether moral dilemmas can be analysed in terms of perfect-imperfect worlds. Christine Korsgaard and Carla Bagnoli have developed theories which take into account the division between the perfect and imperfect or ideal and non-ideal worlds. Korsgaard, building on Kant�s principle of Universal Law, argues that the maxim of an action can be universalized under a specific situation, where its efficacy is preserved. For Bagnoli, moral dilemmas occur due to an agent�s deliberation under constraints. The details of this discussion lie, however, outside the scope of an analysis of the formal structure of moral dilemmas. For the formal analysis here it suffices to say that the application of a subjunctive conditional presupposes the model of a perfect-imperfect ethics. What constitutes the specific situation for Korsgaard or the constraints for Bagnoli is represented as the antecedent of the material conditional, which operates in the imperfect (actual) world. The ideal world of Korsgaard and Bagnoli and Ross�s world of prima facie duties are represented as the antecedent in the subjunctive conditional in the perfect (counterfactual) world.

Applying a subjunctive conditional, if the strategy is adopted in the case of moral dilemmas, (1) can be rephrased in this way:

1Sa) �(p ? �q) > Op

1Sb) (p ? �q) ? O�p

The new premise (1Sa) says �where it not the case that p iff �q, it would have been the case that Op�. Premise (1Sb) says �if p iff �q, then O�p�. On this reading (3) and (1Sb) entail (7), but (3) and (1Sa) do not entail (1), hence no contradiction obtains.

Thus the analysis of moral dilemmas, which uses the standard SDL, leads to a contradiction because it involves an ideal sense of �ought�, i.e. involves obligations in a deontically perfect world, whereas the last analysis deals with actual deontically imperfect situations based on factual premises. Hence the application of a Lewis-type subjunctive conditional solves the inconsistency problem which arises in the formal analysis of moral dilemmas in SDL.

Cowans C. W. Moral Dilemmas, Oxford University Press 1987, p. 3.
Williams B. Ethical Consistency, printed in Proceedings of the Aristotelian Society, 1965, Vol 39 pp. 103-24. Reprinted in Cowans 1987, pp. 115-137.
Lemmon E. J. Moral Dilemmas, in Philosophical Review 70, 1962, pp. 139-58,
Van Fraassen B. C. Values and the Heart�s Command, in Journal of Philosophy, 1973, pp. 5-19,
Marcus R. B Moral Dilemmas and Consistency, in Journal of Philosophy 77, 1980, pp. 121-36,
Foot P. Moral Realism and Moral Dilemma, in Journal of Philosophy 80, 1983, pp. 379-398
Reprinted in Cowans 1987, chap: 5, 7, 10, and 13 respectively.
McConnell T. C. Moral Dilemmas and Consistency in Ethics, in Canadian Journal of Philosophy 8, 1978, pp. 269-87, in Cowans 1987, pp. 154-173.
   Chisholm, R. M. Contrary-to-Duty Imperatives and Deontic Logic, in Analysis, 1963, 24, 34 � 35.
Hilpinen R. Deontic Logic, in Goble L. The Blackwell Guide to Philosophical Logic, Blackwell Publishing, 2001, p. 162. See also F�llesdal D. and Hilpinen R. Deontic Logic: An Introduction, in Hilpinen, Deontic Logic, D.Reidel Publishing Company, 1971, pp. 15-19.
Goble, 2001, p. 168.
Goble, 2001, p. 161.
Goble, 2001, p. 161. See also �qvist L. Deontic Logic, in Gabbay D. and Guenthner F. Handbook of Philosophical Logic Vol II, D. Reidel Publishing Company, 1984, pp. 619-620 and 636-637.
Goble, 2001, pp. 171-173.
Goble, 2001, p. 172
Bennett, J. A Philosophical Guide to Conditionals, Clarendon Press Oxford 2003, pp. 165-166
Korsgaard C. M. Creating the Kingdom of Ends, Cambridge University Press, 1996, reprinted 1999, pp. 135-137.
Bagnoli C. Il dilemma morale e i limiti della teoria etica, summary in English: Moral Dilemmas and the Limits of Ethical Theory, LED, Milano, 2000.