Tuesday, January 11, 2011

Varieties of Consequentialism

It's interesting to think about what questions serve to distinguish various forms of consequentialism. Here's my stab at some of the central questions -- but feel free to add any others that I've missed!

(1) Is it a theory of 'morality' (narrowly construed) or 'practical reason' more broadly?

(2) Direct or indirect? (That is, act/global vs. rule/motive consequentialism)

(3) Deontic or Scalar? (i.e., is significance accorded to a sharp division between the 'permissible' and 'impermissible'? Or is the more fundamental normative status just that of having more or less reason to perform certain actions?)

> 3a - If deontic: where do you draw the line? (E.g. maximizing, satisficing, dual rank views, etc.)

(4) Agent-neutral (impartial) or agent-relative (partial) value?

(5) Theory of welfare: hedonism, preferentism, or objectivism?

(6) Welfarism vs. Perfectionism

(7) Simple aggregation vs. value holism

(8) Precisely comparable vs. imprecise values.

My answers are:
(1) practical reason, (2) direct, (3) scalar, (4) lean impartial, (5) anti-hedonist; lean objectivist, but some sympathy for preferentism, (6) lean slightly perfectionist, (7) value holism, (8) lean precise.

How about you? (Or, for deontologists in the audience, if you had to pick a form of consequentialism which would it be?)


  1. Unsurprisingly, I share all your answers but consider several of your questions to be misguided. 3, in particular, seems to me to be the sort of question that is very easily answered by putting on an AI hat instead of a philosophy hat.

  2. It's interesting to think of how classical utilitarianisms fit into this scheme. Mill's utilitarianism, for instance, would end up being something like (1) practical reason (2) both (because of how he thinks of practical reason it can depend on the context) (3) both (because the Art of Life has different branches, one of which, morality proper, is deontic, and the other two of which, namely aesthetics and policy, are scalar) and maximizing in the deontic branch (4) agent-neutral (5) hedonism (6) welfarism (7) value holism (or at least tending that direction because of issues related to his liberalism) (8) precise. (There's room for controversy over one or two of these, of course, because on some of them Mill also says things that could be taken a different way.)

    I'm neither utilitarian nor deontologist, but with the exception of the hedonism and the welfarism, Mill has always seemed to me to have had the right idea as to how to build a utilitarianism.

  3. What about questions of distribution -- whether it makes the consequences better if the welfare or what-have-you is distributed equally, or with priority to the worst off?

    As for your questions, I say: 1) TOO HARD!!! Leaning morality, 2) Direct, 3) Ambiguous (see below), 4) agent-neutral if answer to 1) is morality, agent-relative if answer to 1) is practical reason, 5) Not sure where my view fits (see below), 6) Welfarism, I think, maybe..., 7) Holism, 8) Imprecise, I think (if the old Andrew could see me now!).

    On 3): I don't know what kind of significance you have in mind. I think deontic concepts play a distinct role in our practical reasoning (not sure yet how to characterize this role), but I don't think, e.g., whether you subjectively ought to do A depends on the probability of A's being objectively permissible as such; I think it only depends on the probabilities of states of affairs and the strength of reasons supporting A in those respective states of affairs.

    On 5): I'm attracted to a view on which welfare depends on the subject's conscious mental states, but that non-pleasureful aspects of those states can affect the subject's welfare. Feeling unloved, e.g., is really bad, over and above the displeasure of it. The kind of vertiginous feeling Kanye West is expressing in his song "Power" isn't necessarily pleasant, but it's welfare-increasing. The feeling of trying to get others to like you isn't so unpleasant, but it's massively welfare-decreasing. The feelings associated with being erotically bitten, or having your face pressed against a wrestling mat by an opponent as you strain every ligament struggling to extricate yourself are both painful, but welfare-increasing. Preferentism, or anything having to do with desires, is the devil, IMHO.

  4. Pluralist here, I am sympathetic to both deontology and consequentialism, and believe either one to be irreducible to the other.

    (1) Practical reason more broadly. (But there are different kinds of rationality, one having to do with the rationality of pursuing goals, another with that of commitments, and yet another with extending one's present concerns by means of sympathy. Both prudence and impartiality are constituted by these different kinds of rationality.)

    (2) Direct. (Indirect consequentialism is unstable, and in any case even if we choose to follow rules on consequentialist grounds, once we acquire conative habits to follow these rules our commitment to them becomes strong enough to provide independent justification for conforming to rules, without the need for instrumentalist scaffolding. So it's not rational to be a sensible knave.)

    (3) Deontic. (Maybe "dual rank"? Protecting certain important negative rights has absolute priority over promoting welfare. But I could allow that, e.g., in situations where protecting the negative rights of one involves violating the same rights of many, consequentialist considerations can kick in to minimize rights violations.)

    (4) Pass (seems to me both impartialists and partialists may advocate universal concern, but the former assigned equal weight to all persons and the latter greater weight to the near and dear... both seem plausible, and maybe that can be explained by the properties of a hyperbolic discount curve).

    Have run out of steam so won't address 6~8, but great questions, and I've been reflecting on these issues more carefully thanks to your posts.

  5. Some extra questions that seem very important:

    1.Does the theory rank alternatives on a ratio, cardinal, or ordinal scale, or on some other scale?
    2.Are uncertain prospects ranked in terms of their expected value (risk-neutrally), with some risk-aversion, or in some other way?
    3.Is the theory's value function bounded, lexical, or Archimedian? Or is there no value function?
    4.On the simple aggregation/holism question, lots of ways to go on population ethics: total, critical-level, average, variable value, prioritarian, mixtures of those, other.

    I find it interesting that so many of the questions on your list are defined in opposition to classical utilitarianism. (This is not your idiosyncrasy, I recognize, but a feature of how this discussion usually takes place.) The effect is to make the alternatives highly incomplete (this is especially true of 4-7). We pretty much know how classical utilitarians rank outcomes and alternatives, but saying, “I'm a holist about aggregation” lets us make essentially no predictions about how alternatives or outcomes are ranked, since the space of possible holist views is so large (even if one gave the classical answer to every other question). I have been rather unimpressed with every concrete consequentialist alternative to the classical picture that has been given. In my view, every concrete average, critical-level, variable value, prioritarian, etc. welfarist theory fares worse than the classical view (variations on the theory of well-being possibly excepted). My impression is that non-classical answers get much of their plausiblity because of their lack of specificity; since population ethics is riddled with paradoxes (a fact about which Gustaf Arrhenius's impossibility theorems constantly reminds us), it is easy to point out problems for any complete position. Incomplete alternatives effectively get a pass, since we can't extract any consequences from them. For this reason, non-classical views should take little comfort from their ability to avoid certain familiar problems, such as the Repugnant Conclusion.

    In any case, since the “other” category includes such broad classes of views, it is hard not to be more confident that some version of these other views is more likely than classical utilitarianism, at least for 5-7. But, given how badly all genuinely concrete alternatives to classical util have done, and I will continue to place more credence in classical util than any concrete alternative version of consequentialism. (One exception: I'll take the scalar answer.)

  6. Why isn't (5) just 'theory of value' instead of 'theory of welfare'? Must a consequentialist hold that welfare is valuable? I agree that it would be implausible to deny that welfare is valuable, but it doesn't seem that consequentialism commits one either to welfarism or to perfectionism.

    I would change (4) to 'personal-value, impersonal-value, or rational-desire?'

    Egoism is a version of personal-value consequentialism, utilitarianism is a version of impersonal-value consequentialism, and my commonsense consequentialism is a version of rational-desire consequentialism (where the permissibility of actions depends, not on the value of their outcomes, but on how much reason the agent has to desire that the various available outcomes obtain).

    I would add the following central questions to the list:

    (9) Single-ranking or dual-ranking?

    Traditional act-utilitarianism is a version of single-ranking act-consequentialism, whereas Ted Sider's Self/Other utilitarianism is a version of dual-ranking act-consequentialism.

    (10) How do we assess both larger and smaller sets of actions? Do we accept agglomeration, distribution, or across-the-boardism? Fred Feldman's world utilitarianism is distributive. First, we assess the deontic status of maximally large sets of actions (what he call entire life histories) and then we assess the permissibility of smaller sets actions according to whether or not they are contained within some permissible maximally large set of actions. Then there could be agglomerative consequentialism, where we first assess basic actions and then hold that the deontic status of non-basic actions agglomerates over conjunction, such that if P(x) & P(y), then P(x & y). But, as most people formulate act-utilitarianism, they seem committed to across-the-boardism: we use the principle utility to assess everything from the most basic action to maximally large conjunctive actions.

    (11) Actualist, possibilist, or other?

    My views are:

    (1) both
    (2) direct
    (3) deontic
    (4) rational-desire
    (5) objectivism
    (6) agnostic
    (7) lean holism
    (8) agnostic
    (9) dual-ranking
    (10) distribution
    (11) Other (viz., securitism)

  7. Hi Doug, I was thinking that any minimally plausible consequentialism will give significant consideration to welfare. Question 6 then allows the respondent to clarify whether welfare is the exclusive consideration. ('Perfectionism' is perhaps excessively specific, though it at least evokes the dominant alternative.)

    In light of the fitting attitudes analysis of value, I don't see how 'rational desire' accounts provide a distinct alternative in Q4. This may be terminological, e.g. if you wish to restrict 'value' to those things we have impartial reason to desire, but you also think we have partial reasons for desire. That sort of view is agent-relative in the sense that matters.

  8. Hi Richard,

    Because of the partiality challenge to (as well as the wrong-kind-of-reasons problem with) the fitting pro-attitudes account of value, I think that the right kinds of reasons will be restricted to impartial/impersonal reasons. And you're absolutely right that rational-desire consequentialism will be agent-relative. But so will personal-value consequentialism. And I think that it's important to not only to distinguish between agent-relative and agent-neutral consequentialisms, but also between two types of agent-relative consequentialism: that is, between personal-value consequentialism and rational-desire consequentialism.

    That's my thought anyway.

  9. I'm going to side with Yang Zhu on this one, and pick the answers to the ones that his work seems to answer.

    (1) It is a theory of 'practical reason' more broadly?
    (2) It's hard to say. I'd say that it's act/global consequentialism, but he also outlines them as general rules that determine one's happiness or misery.
    (3) It's scalar.
    (4) It's psychologically egoistic and engages "clan mentality", so it's agent-relative.
    (5) It's hedonistic.
    (6) It's welfaristic.
    (7) Yang Zhu does not care about contributory value, and argues that measuring the value of our lives in this way is a horrible mistake.
    (8) It's precisely comparable. Yang Zhu claims that universal aspects of human nature (that is, human psychology or anatomy) exactly show what is valuable.


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)