Lately I’ve been thinking about taxonomy of philosophical views, especially in the area of normative ethics.
Here’s a fragment of a common taxonomy of ethical theories. (This will be a bit rough; but the details aren’t important here.) To begin, we distinguish consequentialist from non-consequentialist views, according as they accept or reject the claim that we ought to maximise the good. Then, within the category consequentialist, we distinguish between utilitarian and non-utilitarian views, according as they accept or reject the claim that the good is overall wellbeing. Then, within the category utilitarian, we distinguish between henonistic and non-hedonistic views, according as they accept or reject the claim that wellbeing consists solely in pleasure. Then, within the category hedonistic, we distinguish between phenomenological and non-phenomenological views, according as they accept or reject the claim that pleasure is just a mental state with a particular phenomenology. And so on and so forth.
So, here’s what I’m wondering about. Can we go on like this forever, adding distinction after distinction without end? Or will we at some point find that we cannot distinguish any further? Will we eventually reach the “basic unit” of ethical theory? And if so, how will we know when we get there? What would be wrong with stopping at some more general level? Why not, say, simply distinguish consequentialism from non-consequentialism and leave it at that?
I might need more help focusing in on the question you have in mind, but let me say what comes to mind.
We will not run out of distinctions that we could make, but we will quickly run out of broadly morally interesting distinctions.
The only thing that seems wrong about stopping the work of making distinctions quite early is that we will miss broadly interesting ethical distinctions, and so our discussion will run together ideas that should be separated if we are to make progress in thinking about ethics.
But I fear that likely these sort of answers do not address what you have in mind so perhaps, at least for me, you might say more about that.
Quick Question on what I realize is a side-issue. Where do you (or do you) make a distinction between views that allow goodness to be agent or time relative, versus theories that don’t? In the way I talk, a theory is not consequentialist until it is on the agent-neutral side of that division within theories that say you should maximize the good. But Doug and Jamie (and I think Michael S) at least some of the time seem willing to apply the term ‘consequentialist” to theories that allow the betterness ordering of outcomes to vary between individuals or across different times (or what I think is equivalent to allow the outcomes to be individuated in agent or time relative ways).
On the main question, I think I agree with David, you can always make further distinctions, but lots of those distinctions won’t correspond to anything a plausible moral theory will try to work in. (Maybe the viability of particularist theories should make me worry about that though. Hmm.)
I wonder if the taxonomies of ethical theories are more like explanations and less like natural kinds. That is, aren’t they interest relative? We want to draw attention to certain contrasts with them for some practical purposes without them being written in stone. We can go on making the distinctions as far as we want if they are needed for argumentative purposes like David says. In addition there is a number of ways of making the distinctions at any level. For instance, the satisficing consequentialists would be non-consequentialists in your taxonomy so if we need to account for them we might want to make that distinction in some different way.
This doesn’t answer the question, but I doubt that any completely general taxonomy will be so neat. I think it often happens that interesting questions cross-cut each other in such a way that of views A,B,C, and D, A and B pair together against C and D for some purposes, but A and C pair against B and D for others. So which division makes it earlier into the tree? Surely that just depends on which question we are using the tree in order to clarify.
For example, non-consequentialists can engage in the same debate as consequentialists about which things are good – which will be particularly important for them, if they believe that better outcomes are generally supported by pro tanto reasons. It’s just because in fact, ‘axiology’ is more often a matter of infighting amongst consequentialists, that it gets appended to the tree further down than the dispute between consequentialists and non-consequentialists.
You might think that there is some justification for thinking that one of these questions ‘really’ comes first. But I’d be highly suspicious. For actual philosophers in actual time, of course, one view often comes before, and leads to, another, and there might be patterns in how it goes. But it’s highly suspect to think that it can be ruled out a priori that pressure could come from the other direction.
For example, the plausibility of consequentialism might itself turn on the plausibility of the kind of theory of value that would be needed in order to substantiate a non-counterintuitive consequentialist view. I know that Campbell has explicitly denied in a draft of ‘Consequentialise This’ that it is possible to have intuitions about what is good that are not informed by the consequentialist project, but that just looks obviously false. Any deontologist can have intuitions about whether murder is a good or a bad thing.
So it looks to me like here, as with most everywhere in philosophy, the different questions bear on and cross-cut one another, and different ones will ‘come first’ for different thinkers at different times, and different ones will seem more fundamental, depending on which ones interest us more.
Thanks everyone for your comments. A couple of replies:
Dave,
Let me try to focus the question a bit more. Here’s a quote from Peter Railton, which I happened to read yesterday (from the alienation paper):
This seems right. Within the category of consequentialist moral theories there is a diverse range of options. For example, some consequentialist theories permit eating meat, while others forbid it. So, if all you say is that you endorse consequentialism, then your moral view is quite indeterminate, it leaves many moral questions unanswered.
Suppose, however, that this is generally true: whenever it’s possible to divide a category of moral theories into further subcategories, if you merely endorse the larger category, while remaining neutral on the subcategories, your moral view is indeterminate. Then if, as you suggest, it’s always possible to go on dividing categories into subcategories, we’ll never arrive at a fully determinate moral view. Some moral questions will never be answered. That seems troubling, for some reason.
What do you think about this? Take a silly moral view: an act is right iff it’s done on a Tuedsay. That seems fairly determinate; except perhaps for borderline cases of acts done around midnight, it is perfectly clear whether a given act is right according to this view. So what is it about consequentialism that makes it less determinate than “Tuesdayism”?
Mark vR,
I’m happy to go either way on whether the term “consequentialism” is reserved for agent-neutral views. My hunch is that one is more likely to be misunderstood if one uses “consequentialism” in the broader sense, allowing both neutral and relative varieties, without making it explicit that this is what one is doing. Rightly or wrongly, people tend to associate “consequentialism” with agent-neutralism. I have a paper where I use “strict consequentialism” for the narrow sense (i.e. neutral only), and “general consequentialism” for the broader sense. I think Michael sometimes says “big ‘C’ consequentialism” (narrow) and “small ‘c’ conseqeuntialism” (broad).
I’m not so sure that I fully appreciate the worry, or the inquiry. If the concern is that a normative moral theory may fall short of answering every moral question because this sort of probing/distinction-izing can (let’s assume) continue ad infinitum, then that doesn’t seem terribly problematic. Physics is a good example of a discipline with this sort of ‘problem’ and they continue on just fine. Newtonian physics could not account for many physical happenings, but that did not make it useless, or worrisome, just rather incomplete. The following Einsteinian physics helps answer more of those unanswered questions, but it too can not account for certain happenings in space, that at the time of Einstein were not even known to exist. That doesn’t make for a worrisome situation in physics, it simply means there’s more work to be done. Why should moral theory be any different? The changing world is going to open up new frontiers for moral interaction, bringing with it a need to further analyze moral intuitions and the theories that are meant to account for them. The moral question of stem cell use certainly did not cross anyone’s mind prior to the past decade. If this is the worry, then ethicists are no worse off than any theorists, and neither is their subject.
In response to your most recent question comparing Tuesdayism to consequentialism, as usual, the devil is in the details. Consequentialism fails to give us answers to many (important) moral questions without further articulation of what good is to be maximized. Tuesdayism does this simply based on the boundaries of the (adapted) Roman calendar, and whatever definition of action that it adopts insofar as an action’s done-ness is central to its moral status. But this goes hand-in-hand with the reasons we reject theories like Tuesdayism, they simple get the moral facts wrong. To bring it back to the overall concern, these moral facts are getting more complicated (seemingly) as moral agents interact in new and interesting ways. As they evolve philosophers are required to account for this evolution in our moral theorizing if we are hoping to get the moral facts right, which is why the taxonomy in ethics is getting more complicated.
Or is the worry something else entirely…?
Campbell – Now I see what worries you. But I’m not worried. Consequentialism is no less determinate than Tuesdayism. So long as there are determinate facts about what brings about the most good, there are determinate facts about what consequentialism requires people to do.
Of course, people disagree about what those determinate facts are, which creates a complication. People do not, usually, disagree about whether Tuesday comes after Monday, but they might. Similarly, even among those who agree that Tuesday comes after Monday, there might be disagreement about whether Monday comes after Sunday, or in the middle of the week.
I take it your question is not the metaethical one about why it is that people are more likely to disagree about what is good than about where Tuesday falls in the week.
It looks to me like the ‘indeterminacy’ in consequentialism is a result of good dialectical practice in philosophy, in a way that is related to the discussion of what is at stake between expressivists and others in a recent thread. Suppose that two philosophers, X and Y, can agree that if the facts about what is good are way A, then consequentialism jars shockingly with moral common sense, and is thereby rendered very implausible, but if the facts about what is good are way B, then consequentialism is not thereby rendered implausible. And suppose X believes that the facts are way A, and hence rejects consequentialism, while Y thinks the facts are way B, and hence is not deterred from her consequentialism.
If X and Y are right in their commonsense moral views, and the facts about what is good really are way A, then consequentialism is false, and determinately false, no matter what Y thinks. But it’s poor dialectical practice for philosopher X to offer that to Y as her argument that consequentialism is false, because Y rejects the argument’s key premise. In general, a view can be shown false because of a sound argument against it, even though that view denies one of the premises. But no such argument is one that Y need take seriously, unless independent evidence is offered for A over B. So in looking for a neutral way of framing the debate that both sides can accept, we _treat_ consequentialism as being less than fully determinate, because we are interested in what we can use arguments to commit each other to.
And rightly so – any view needs to be assessed on its own merits, given _all_ of its commitments. If we allow ourselves to argue against conjunctive views by means of arguments against one conjunct which appeal to the negation of the other conjunct, then we’ll very easily find ourselves unable to engage in productive discussions with one another, and very unlikely, I think, to converge on the truth. I suspect that dialectical considerations like this, rather than genuine indeterminacy in the content of consequentialism, is what is at stake in the selection you quote from Railton.
Mark,
Good — now you’re scratching my itch (to use a Sobelism). But I’m still worried. You say this:
But isn’t this at odds with the common view articulated, for example, in the above quotation from Railton. He says, to repeat, “one has adopted no morality in particular in adopting consequentialism unless one says what the good is.” The idea here, as I understand it, is as follows. Consequentialism itself is not a particular moral theory. Rather, consequentialism is a “family” or “class” of distinct particular moral theories. These theories all have something in common: they all hold that we should maximise the good. But they differ over what the good is.
Suppose that C1 and C2 are two particular consequentialist theories. C1 holds that the good is such that eating meat is compatible with maximising the good, whereas C2 denies this. It follows that consequentialism is indeterminate with respect to whether eating meat is permissible. One consequentialist view, C1, implies it is permissible; another, C2, implies it isn’t. Hence, merely to adopt consequentialism is not to take a stance one way or the other on the permissibility of eating meat.
On your view, however, consequentialism is not indeterminate in this way. Consequentialism has determinate implications regarding the permissibility of eating meat, and these are fixed by facts about the good. For example, suppose it’s a fact that eating meat isn’t compatible with maximising the good. Then consequentialism implies that eating meat is impermissible.
But this view has quite odd implications. For example, given our supposition about the facts, it follows that, not only is C1 false, but it’s inconsistent with consequentialism. It’s fine to say, I think, that whether C1 is true depends on the facts about what is good; but it would be quite odd to say that whether C1 is consequentialist depends on those facts. Many people hold, for example, that utilitarianism includes a false theory of the good; but no one infers from this that utilitarianism is not a form of consequentialism. Suppose we’re ignorant of the facts about what is good; in particular, we don’t know whether eating meat is compatible with maximising the good. Then it follows, on your view, that we cannot know whether C1 or C2 is consequentialist (though we do know at least one of them isn’t).
Don’t you think that’s all a bit odd?
Chris,
It seems you accept what Mark rejects: that Tuesdayism is more determinate than consequentialism. I want to ask, why is this so? What is it about consequentialism that makes it less determinate? Presumably, it has something to with the notions of “good” and “Tuesday”. Philosophy tolerates competing theories of the good, but it wouldn’t tolerate competing theories of Tuesday. Good is up for debate in a way that Tuesday is not. Why is that?
I think the notion of ‘Tuesday’ is up for debate in many ways. Imagine a galaxy far far away… A planet much like ours revolving around couple of suns, having completely different times of daylight, rotation periods, and so on. So, our calender as such would be inapplicable there. Imagine also that they have human like beings living in similar societies doing similar stuff than us. We would want to say that their actions have the same moral properties than ours. Could Tuesdayism say this? It’s not clear it could. When would it be Tuesdays there? When Tuesdays here in GMT? East cost time? Tokyo time? Given the theory of relative the idea of precisely at the same time somewhere completely distant also I take it is going to give.
Also possibilia is supposed to have moral properties too. So, imagine the previous planet in a possible world that lacks Earth’s counterpart in terms of the planetary constellations. The acts there still would be wrong and right. When would it be Tuesday there? Do possible worlds happen at the same time as our world?
Consequentialisms in many forms seem to do much better with the regards to these cases.
I just wonder if we are ever going to be able to give completely determinate moral theories. At some point, we are going to need to use vague concepts anyway. Wouldn’t suffice to have enough of determinacy in order for the views to have a practical role in being able to guide our actions?
The concept of Tuesday may be well up for debate, but let’s assume Tuesday is not up for debate and Tuesdayism is determinate. Consequentialism doesn’t share the property of determinacy with Tuesdayism precisely because Tuesday is not up for debate. How does Tuesday differ from the good? Well, any competing definition of Tuesday would be as arbitrary as the functional one. The names days have serve merely functional roles that we arbitrarily agree on so as to ease communication. What the good is doesn’t seem to be an arbitrary matter at all, but a rather important one. We tolerate disagreement about the good because we’re working toward an agreement, much like the one we have for Tuesday, but one that isn’t arbitrary; which is why we don’t agree with one another on what the good is, there are good reasons not to.
But this doesn’t seem like a very satisfying answer, which makes me believe I still don’t fully appreciate the worry.
Hambone,
You wrote:
Suppose, however, that this is generally true: whenever it’s possible to divide a category of moral theories into further subcategories, if you merely endorse the larger category, while remaining neutral on the subcategories, your moral view is indeterminate. Then if, as you suggest, it’s always possible to go on dividing categories into subcategories, we’ll never arrive at a fully determinate moral view. Some moral questions will never be answered.
I don’t see this yet. Suppose, as I do, that there is a finite number of intrinsically interesting moral distinctions. Then the fact that our moral view admits of further distinctions is not to say that it is not fully morally indeterminate. Our theory, once we have exhausted the distinctions we think are intrinsically morally important, claims that further distinctions are not intrinsically morally significant. This could be false, but there is no lack of determinacy, or necessary lack of determinacy, that I see.
Campbell –
Distinguish one sense in which consequentialism really _is_ just a class of theories. There are lots of ways of trying to make formal sense of alternatives and induce an ordering on them. For example, how fine-grained must they be? Are alternatives possible act-tokens, or elements in a partition on accessible futures, or …? And does the ranking come from the single way the world will be if the agent acts in that way, or some kind of expected value, or does it derive from the best or worst possible outcome that could result from that action, given future choices of the same agent? Different answers all lead to different views, which all seem to deserve to be called ‘consequentialist’. So in that sense, it is a family of theories.
But holding fixed the answers to all of those questions, amd ony allowing for variation in what we hold to be good: no, I don’t think you’ve made my view that consequentialism has determinate results look unintuitive. I take it your theory C1 consists of at least the following three theses (I simplify slightly):
1) Any action is permissible just in case it brings about the best result.
2) Eating meat always brings about the best result.
3) Eating meat is always permissible.
You say: suppose that thesis 2) is false. Then C1, which is committed to 1), 2) and 3) is not just false, it is inconsistent with consequentialism. I don’t know what you mean. Thesis 1) _just is_ consequentialism. So C1 is certainly committed to consequentialism, even if C2 is false. Is it also, somehow, committed to denying consequentialism? No – even if thesis 2) is false, there is no way to derive a contradiction from the three theses. So what can you mean?
You must mean this: that if 2) is false, then 1) and 3) can’t both be true, and so someone who accepts 1) and 3) accepts something that cannot be true, if consequentialism is. Hence, you allege, her view is ‘inconsistent with’ consequentialism. This is no mystery – since this theorist believes thesis 2), she does _not_ think that 1) and 3) are incompatible. Rather, she thinks that 3) follows from 1). There is no logical mistake in this. There is only a mistake about the truth of 2), which is a substantive question in moral philosophy.
Compare a mundane example, to illustrate: Suppose that we are wondering who will be inaugurated president of the US in January, 2017. Jon says that it will be whoever is inaugurated as vice president in January 2009. Mary agrees. Jon and Mary agree on a definite, testable view (although it will take some time to test it), with determinate content. Now suppose that Jon also happens to think that Barack Obama will be inaugurated vice president in January 2009, while Mary thinks that Tom DeLay will be inaugurated vice president in January 2009. If they follow through on their own logic, then Jon will think that Barack Obama will be inaugurated president in January 2017, and Mary will think that Tom DeLay will be inaugurated president in January 2017.
Now suppose that we wait until January 2009, and it turns out that Barack Obama is inaugurated as vice president. Will we say: ‘omigosh, not only was Mary’s view false, she didn’t even think that whoever was inaugurated vice president in January 2009 would be inaugurated as president in January 2017!’ No – we’d just say that not only was Mary’s belief that Tom DeLay would be inaugurated vice president in January 2009 false, at least one of her other beliefs must be false, too – either that whoever is VP in 2009 will be P in 2017, or else that DeLay will be P in 2017. And then we’d wait until 2017 in order to find out which.
The same goes for consequentialists. Railton’s point is that consequentialism doesn’t logically commit you to results on cases, because you can have a logically consistent view, with different theories of the good. (The same thing goes for the VP theory.) As I was trying to explain before, I don’t see whay I can’t accommodate that. It looks like it is about dialectical considerations, which make perfect sense and are well-motivated.
Mark,
We’re making progress. But I think we still disagree.
Let me state my argument a bit more precisely. Let E be the proposition that eating meat is always permissible. Let C be consequentialism. And assume that C1 and C2 are two particular consequentialistic theories such that C1 implies E, and C2 implies ~E. Since C1 is consequentialistic, C is consistent with C1. So C is consistent with E. And since C2 is consequentialistic, C is consistent with C2. So C is consistent with ~E. Therefore, C is indeterminate with respect to E; that is, C doesn’t imply E, nor does it imply ~E.
Perhaps you will accept my conclusion but maintain that it doesn’t contradict what you’ve said above. The kind of determinacy that I’ve shown C not to have, you might say, is not the kind that you’ve claimed C does have. So let’s talk about kinds of determinacy.
Here’s what I have in mind. A proposition P is indeterminate with respect to a proposition Q iff P is consistent both with Q and with ~Q. And P is consistent with Q iff it’s possible that P and Q. But what kind of possibility? Since there are different kinds of possibility, there are, correspondingly, different kinds of consistency and different kinds of determinacy. For example, corresponding to logical possibility we have logical consistency and logical determinacy.
Now, you might argue that all my argument shows is that C is logically indeterminate with respect to E. And that, of course, is no big deal. No one should expect to get from consequentialism and logic alone to any conclusion about the morality of eating meat. However, I think the argument shows something stronger than logical indeterminacy: what we might call “conceptual” indeterminacy.
Let Anti-C be the proposition that we ought to maximise the bad. According to Anti-C, an act is permissible only if it has the worst possible outcome. Surely, Anti-C is not consequentialistic. Yet it is logically consistent with C. So being logically consistent with C is not sufficient for being consequentialistic. However, Anti-C is inconsistent with C in a weaker sense. If you grasp the concepts of good and bad, best and worst, then you can see that C and Anti-C cannot both be true. So, we might say, C and Anti-C are conceptually inconsistent.
Suppose, then, that being conceptually consistent with C is a necessary condition of being consequentialistic. Then my argument shows that C is conceptually indeterminate with respect to E.
So this, I think, is where we disagree. Right at the end of your last comment you say:
I think many philosophers, Railton included, would accept something stronger: not only does consequentialism not logically commit you to “results on cases”, but it doesn’t conceptually commit you to such results either.
Campbell –
I don’t know what conceptual commitment is. But suppose, as you say, that it has something to do with what anyone would see followed, who ‘grasped the concepts’ involved. Then I agree – consequentialism is ‘conceptually indeterminate’ with respect to E. But I never meant to deny that, either.
Obviously in that case, conceptual commitment would be what is dialectically important, which was what I’ve been claiming all along is the sane thing to think is at stake in the Railton quotation. It is about what we can use arguments to get someone to acknowledge commitment to – not about the conditions under which a view is false. My point all along has been that consequentialism is a _determinate view_, in that it has perfectly precise determinate content, even if it is conceptually coherent to disagree about what that content is, and hence, even if it makes perfect sense for philosophers to unknowingly accept things that cannot be true if consequentialism is, along with their consequentialism.
So go back to Jon and Mary and the vice president inaugurated in 2009. Their view that this predicts who will be inaugurated president in 2017 is conceptually indeterminate with respect to who is inaugurated president in 2017, too, by your definition. But that in no way makes it an indeterminate view. It is a perfectly precise determinate view, and subject to a straightforward test, albeit one that it will take us approximately a decade and a few weeks to perform.
If consequentialism is only ‘indeterminate’ in the way that Jon and Mary’s view is, then the claim that it is indeterminate is true but apparently banal.
Mark
I think I agree with all that (except perhaps the last sentence). Though I’d like to clarify that I never meant to suggest that consequentialism is indeterminate in the sense of lacking a truth-value. I think consequentialism has a truth-value: it’s true. Perhaps “indeterminate” was a poor choice of word.
Maybe I’m merely making mountains out of molehills. But I have a few lingering doubts. It still seems to me that there’s some interesting difference between consequentialism and Tuesdayism in the neighbourhood I’ve been clumsily stumbling around in. Though I’m having trouble putting my finger on it.
Anyway, thanks for your comments. They’ve been helpful.