Context of the discussion: what Dancy ('Moral reasons') believes a fully fledged non-consequentialist theory must account for:
In the context of this discussion, Dancy considers the sort of defence of a non-consequentialist theory that implies explaining the existence of deontological constraints. There can be theories that are not consequentialist but neither recognize deontological constraints. Assuming that consequentialism has no space for agent-relative values (a thesis criticized by Sen), non-consequentialism can (arguably) be refuted by arguing that there are "agent-relative" reasons or values, that is to say reasons (or values) that justify the agent's relative lack of committment for realizing the aim of the morally best state of affairs, when accomplishing this aim entails too big a sacrifice of the agent's own special concerns (such as projects, friendships, respect of non - moral ideals, etc). It would represent already a significant departure from (neutralists) versions of consequentialism to argue that a person has the right to give more weight in one's moral calculation to sacrifices regarding one's own projects or frienships, than the analogous sacrifices faced by another person. (This is the kind of view defended by Scheffler, who cannot find a rationale for deontological constraints, while can find a rationale for resisting the demands of the sort of impersonal morality in question.)
Dancy refuses this characterization of a non-consequentialist outlook. Dancy sides with Foot's claim that a rational defence of a good (non-consequentialist) moral theory should explain what may be called "options" and "constraints" with equal ease.
Notice that this discussion with proceed upon the assumption that a consequentialist theory is an act-consequentialist one, that is to say, a theory giving each agent the same moral aim, that of producing states of affairs with the greatest ethical value, and identifying the act the agent ought morally to do with the act that would, in each instance, and among the ones the agent is not prevented to perform, bring about that aim.
(I am not considering rule or motive utilitarianism here, for the following reason. Rule or motive utilitarianism can be seen as either an act theory or an agent theory- that is to say, either as a criterion for the wrongness or rightness of actions, or as the theory saying what act an agent ought morally to perform. If rule utilitarianism is conceived as an act-theory, then it seems undermined by its own justification. In fact, if what makes a rule a good rule is its leading on the whole to the achievement of the best state of affairs, how can an act be right that fails to reach this very objective? If, on the other hand, rule utilitiarianism is conceived as an agent theory, and act-utilitarianism as the act-theory, there will be many cases in which the act that is right (in virtue of the act theory) will not be the one that the agent ought morally to perform, according to the agent theory.
I have doubts that this picture is coherent, for it seems that a committment to act-utilitarianism as an act-theory and a committment to follow the dictates of rule-utilitarianism in practical deliberation can hardly coexist in the same agent. )
Let us consider what may justify the existence of deontological constraints. According to Dancy, deontology oblidges me to choose A2 over A1 in the following case:
A1. I allow J to be killed by K
A2. I kill N.
(N and J are innocent human beings)
I have to choose between A1 and A2: (¬(A1) → (A2)) & (¬(A2) → (A1)),
We are supposing that both A1 and A2 are facts I am aware of. I am, that is, is wholly conscious that by failing to kill J I will let K be killed.
This leads to a whole host of difficulties explaining how can A2 have more value than A1 from my point of view; it brings to the introduction of agent-relative reasons and agent-relative value. It leads to talk about "moral cost to the agent" (what explains the difference in value between A1 and A2 from my point of view.)
I have suspects about calling the violation of a constraint as "a moral cost". If it would be a moral cost, then our theory should imply that the agent has the right to discount this cost. (If the cost of killing an innocent is one that I should pay, I should be praised, not blamed, if I choose to pay it.)
The wrongness of violating a deontological constraint is conceived as a "cost" to the agent, we cannot make sense of a similar deontological restriction, like the following.
I believe that, just as an agent should morally choose not to kill an innocent even when this would lead to let another innocent be killed, a person should choose morally not to kill a person now, even when this would certainly lead to kill another person in the near future.
Letting someone die can be as bad as killing, when letting die qualifies as an agent act in some sense. What is required in order for an omission to count as an agent's act is something difficult to tell. But we clearly have the sense that when we have full power and control over letting someone die (say, when we have the technical means and we could prevent this at an insignificant cost to ourself) letting someone die can be as much a morally objectionable act as killing. We do not make the best sense of deontological prohibitions, I think, by conceiving them under the rule that omissions are always to be preferred to acts.
In order to gain some meaningfulness to the idea deontological perspective, I think, one must asks what deontological prohibitions are for. The answer, I think is something like the following:
if we did not perceive that there would be something morally problematic in killing a person for the sake of saving another one, we would be morally allowed to take on ourself the responsibility of deciding the life of death of many individuals. This is risky, because there are all sort of factors that may militate against this sort of self-ascription of responsibility to myself, such as personal biases.
Moreover, a person must have a strong disposition against killing innocents in general in order to refrain from this when his own interest is at stake; and a person with the right sort of disposition will have troubles killing an innocent even when a. the state of affairs that would result from his failing to do so will be equally bad in consequentialist terms, and b. the state of affairs that would result from his failing to do so will be worse in consequentialist terms.
What follows from the idea that what appear in first-person thinking as deontological constraints is objectively the effect of a moral outlook, acquired from a good moral upbringing, that takes away from the agent the responsibility of performing the sort of moral calculation that act-consequentialism requires from him?
The effect of having a deontological constraint as a part of one's moral outlook produces a shrinking of the "deliberative sphere" (the range of considerations that are relevant for the choice of action). This must not be understood as implying that, if one adopts a deontological morality, certain facts about the consequences of an act would become irrelevant (say because that act is prohibited and nothing else can be said in his favor)
Rather, according to my theory, deontological constraints are rules directing the agent's search of facts that bear upon the morality of an action. Having a deontoligical constraint about killing means, first of all, reaching the conclusion that the action is wrong in a straighforward manner, without interrogating oneselves too much about what shall be the other consequences of forgoing that act.
This explains why talking about an act /omission asymmetry is not the best way to understand a deontological outlook. When the possibility of causing the death of an innocent by omission has already entered the deliberative sphere, the agent cannot pretend that the omission causing the death of this person is somehow less wrong qua omission.
.... To be continued....