From Value to Rightness

Consequentialism, Action-Guidance, and the Perspective-Dependence of Moral Duties

Vuko Andrić

• Publisher: Routledge
• Year: 2021
• ISBN:
  9781000405446
• Edition: 1
• Topic: Philosophy
• Subtopic: Ethics & Moral Philosophy

This book develops an original version of act-consequentialism. It argues that act-consequentialists should adopt a subjective criterion of rightness.

The book develops new arguments which strongly suggest that, according to the best version of act-consequentialism, the rightness of actions depends on expected rather than actual value. Its findings go beyond the debate about consequentialism and touch on important debates in normative ethics and metaethics. The distinction between criterion of rightness and decision procedures addresses how, why, and in which sense moral theories must be implemented by ordinary persons. The discussion of the rationales of "ought" implies "can" leads to the discovery of a hitherto overlooked moral principle, "ought" implies "evidence", which can be used to show that most prominent moral theories are false. Finally, in the context of discussing cases that are supposed to reveal intuitions that favour either objective or subjective consequentialism, the book argues that which cases are relevant for the discussion of objectivism and subjectivism depends on the type of moral theory we are concerned with (consequentialism, Kantianism, virtue ethics, etc.).

From Value to Rightness will be of interest to scholars and advanced students working in normative ethics and metaethics.

Part 1
Decision-Guidance

1
The Problem of Decision-Guidance

This and the next part of the book examine considerations that are related to action-guidance in order to adjudicate between objective and subjective consequentialism. The term “action-guidance” will be explained and elaborated at several places. In this and the next chapter, I am concerned with decision-guidance, which I understand to be a special kind of action-guidance. This chapter introduces what I call the problem of decision-guidance. Followers of objective consequentialism face this problem when they do not know how to decide what to do because they do not know which of the actions available to them maximizes actual value and is, hence, obligatory according to objective consequentialism. What is needed in order to solve the problem are rules that tell agents with cognitive limitations how they ought to decide under ignorance and uncertainty.

According to some philosophers, subjective consequentialism is superior to objective consequentialism because the first is decision-guiding but the second is not. However, I think that the problem of decision-guidance is not a problem exclusively for objective consequentialism. The problem also concerns subjective consequentialism, as Fred Feldman has shown. I will present and defend Feldman’s argument in the next section.

The plan of the chapter is as follows. I elaborate on the problem of decision-guidance in Sections 1 and 2. Section 3 determines the relevance of the problem of decision-guidance for moral theories. The problem of decision-guidance will be distinguished from other problems that concern the action-guidingness of consequentialism in Section 4. Final remarks follow in Section 5.

1. The Problem of Decision-Guidance for Objective and Subjective Consequentialism

Here are two quotes – the first from Shelly Kagan (1998, p. 64), the second from Fred Feldman (2006, p. 50) – which introduce the problem of decision-guidance for objective consequentialism:

Perhaps the most common objection to consequentialism is this: it is impossible to know the future. This means that you will never be absolutely certain as to what all the consequences of your act will be…. In fact lacking a crystal ball, how could you possibly tell what all the effects of your act will be? So how can we tell which act will lead to the best results overall – counting all the results? This seems to mean that consequentialism will be unusable as a moral guide to action.

Utilitarians are attracted to the idea that an act is morally right iff it leads to the best outcome. But critics have pointed out that in many cases we cannot determine which of our alternatives in fact would lead to the best outcome. So we can’t use the classic principle to determine what we should do. It’s not “practical”; it’s not “action-guiding”.

Objective consequentialism’s problem of decision-guidance seems to be this: Since you do not know the actual consequences of the acts available to you (after all, you would have to take into account all consequences until the end of days of any of the countless acts available to you), you do not know which of them are right and which are wrong. And therefore you cannot base your decisions on the objective consequentialist criterion of rightness, i.e. you do not know how to choose a right act. So essentially, the problem is that you do not know how to decide; this is why I call it the problem of decision-guidance.

James Lenman (2000) has illustrated how serious the ignorance underlying objective consequentialism’s problem of decision-guidance is. As a starting point, notice that all the consequences of an act are relevant according to objective consequentialism, however distant they may be. Two considerations are especially significant in this context. First, there are many identity-affecting actions. “These are actions that make a difference to the identities of future persons and these differences are apt to amplify exponentially down the generations” (Lenman 2000, p. 346). These actions have massive causal ramifications that cannot be predicted. Second, apparently insignificant actions can have huge impacts, so-called butterfly effects, because there are causal systems that are extremely sensitive to very small and locally triggered changes (ibid., p. 347). Such systems include the weather and the finance markets.

After illustrating the relevance of identity-affecting actions and sensitive causal systems by means of some examples, Lenman concludes:

The worry is not that our certainty [about the consequences of our actions] is imperfect, but that we do not have a clue about the overall consequences of many of our actions. Or rather – for let us be precise – a clue is precisely what we do have, but it is a clue of bewildering insignificance bordering on uselessness – like a detective’s discovery of a fragment of evidence pointing inconclusively to the murderer’s having been under seven feet tall. We may not be strictly without a clue, but we are virtually without a clue.

(Lenman 2000, pp. 349f)

Let us now turn to subjective consequentialism. In his article “Actual Utility, the Objection from Impracticality, and the Move to Expected Utility”, Fred Feldman (2006) argues that the often-suggested move to expected value does not provide a cogent reply to the objection that objective consequentialism is not action-guiding. ¹ I think that Feldman is right: subjective consequentialism has its own problem of decision-guidance and many philosophers do not seem to be aware of this. ² Feldman’s basic thought is this: In order to base your decisions on the subjective consequentialist criterion of rightness, you would have to proceed in any given choice situation in something like the following way:

1. List your alternatives.
2. List all the possible outcomes of the first alternative.
3. For each outcome of the first alternative, write down its value.
4. For each outcome of the first alternative, write down its probability given the alternative on your evidence.
5. For each possible outcome of the first alternative, multiply the value of the outcome by its probability on the first alternative.
6. Sum these products. This sum is the expected value of the first alternative.
7. Repeat Steps 2–6 for each of the other alternatives.
8. Identify the acts that maximize expected value.
9. Perform one of those acts. ³

The problem is that every one of these steps is problematic. The first step is problematic, as Feldman (2006, p. 58) explains, because there are millions of alternatives available to you in almost every choice situation. If you cannot list your alternatives, then you cannot perform the first step.

Step 2, Feldman (2006, pp. 58f.) argues, is not less problematic. An ordinary act might have thousands of possible outcomes. But agents certainly cannot list thousands of outcomes of each available act. Notice also that subjective consequentialism in this regard seems to be worse than objective consequentialism. For while the objective consequentialist only has to list the consequences each act would in fact have, the subjective consequentialist has to list these consequences plus the other consequences each act could have.

Feldman presents similarly motivated worries for the other steps. Feldman is right that no human agent could follow the aforementioned instructions in order to determine how she ought to decide. However, Hilary Greaves (2016) has recently argued that Feldman’s argument fails nonetheless:

Feldman (2006) argues that in fact we are more clueless about subjective c-betterness [i.e., about an act’s expected value being higher than the expected value of another act] than about its objective analogue. But Feldman’s argument assumes that in order to estimate which of two acts has the higher expected value (and by how much), we need to estimate what the expected value of each act is. This latter task is indeed massively more demanding, but (pace Feldman) is unnecessary.

(Greaves 2016, p. 318, footnote 12)

Greaves is right that in order to estimate which of two alternative acts has the higher expected value, we do not necessarily need to calculate any expected values. However, this criticism changes the subject. If I base my decision on my estimation about which of two acts has the higher expected value, then I am not basing it on the subjective consequentialist criterion of rightness. Rather than following the rule “Do what maximizes expected value!”, I follow a rule along the lines of “Do what maximizes expected value according to your estimation!”. Instead of defending the usability of subjective consequentialism against Feldman’s objection, Greaves in fact suggests a different decision rule. ⁴

I conclude, with Feldman, that subjective consequentialism has its own problem of decision-guidance.

2. A General Analysis of the Problem of Decision-Guidance

How can we spell out the problem of decision-guidance in general terms? The first thing to notice is that the problem of decision-guidance results from factual, rather than normative, ignorance. The complaint is that someone who accepts and understands objective consequentialism does not know the consequences her options would in fact have. The problem is not how or why to become a follower of objective consequentialism in the first place. Nor does the problem concern details of the moral theory in question such as the contents of the best theory of value.

Secondly, the person knows that she ought to do what would in fact have the best consequences, but she cannot tell or determine which of the acts available to her fits this description, that is, she cannot identify, among her alternatives, the one that would in fact have the best consequences. It is illuminating, I think, to point out that objective consequentialism has this problem qua moral theory. Objective consequentialism states the right-making features of acts in general terms: objective consequentialism tells you what it takes for acts in general to be right (namely, to maximize actual value), but it does not tell you which of the acts available to you in a concrete situation are right (that is, it does not tell you which of the available acts maximize value). In order to find out whether a particular act a₁ that you can perform in a concrete situation is obligatory, you have to subsume a₁ under the general formula of a moral theory in the following way:

• Subsumption
• (1) [Moral theory:] An act is obligatory iff the act has feature F.
• (2) [Application of the theory to the concrete choice situation:] a₁ has F.
• 
• (3) [Information necessary and sufficient to make a decision:] a₁ is obligatory.

Let us locate the problem of decision-guidance within the subsumption steps. If you know that a₁ is obligatory – that is, if you are able to infer (3) – then there is no problem of deciding what to do. Now since, as I have said earlier, we are dealing with factual rather than normative ignorance, it cannot be step (1) that the problem of decision-guidance is about. Rather, the problem of decision-guidance must be about step (2): the problem results from your inability to assign ought-making features to your various alternatives.

Thirdly, the problem of decision-guidance results from factual ignorance rather than false factual beliefs. Holly Smith (2018) recently proposed a solution to the problem of how to implement moral theories that attempt to take into account both problems (ignorance and false belief) simultaneously. In my opinion, this is a mistake. When you are mistaken about which of your alternatives have ought-making features, then you cannot, at the same time, be unable to assign ought-making features to your alternatives. A false assignment of ought-making features is, after all, an assignment of ought-making features! This suggests that the best way of solving the problem of decision-guidance is different from the best way of solving the problem of false factual belief.

Fourthly, since the problem of decision-guidance is to choose among one’s available acts given one’s inability to assign ought-making, right-making, and wrong-making features to them despite one’s acceptance and understanding of a given moral theory, the problem can be expressed by means of this question:

Q: Given that I do not know the ought-making, right-making, and wrong-making features of the acts available to me, which act ought I to choose?

Perhaps you do not feel the pull of Q. It may be that you already accept some rules you use for decisions when you do not know the right-making or wrong-making features of your alternatives. The problem might thus appear to you to be trivial. Or you might wonder whether you really understand it, or whether there is a problem at all. For these reasons, I think the problem of decision-guidance can best be illustrated by means of a hypothetical scenario.

Suppose you are an AI programmer. Your job is to get an intelligent robot going. Since you are a good person, you will equip the robot with what you think is a decent moral code. Let us say that the code consists of rules ab...