In my philosophical folklore post last week I asked about other tidbits of philosophical folklore, and commenter Ray Ingles gave one example:
The “is-ought fallacy” is another recurring ‘folk philosophy’ phrase – meaning “you can’t derive an ‘ought’ from an ‘is’”, after Hume.
This is a very interesting one, and it is undeniably common — even the exact phrase “you can’t derive an ‘ought’ from an ‘is’” returns something like 48000 hits on Google, and when you start adding variations, the number explodes. The principle is sometimes called ‘Hume’s Guillotine’, a label that seems to go back to philosopher Max Black in the 1960s. Others call it ‘Hume’s Law’, the source of which I have not been able to trace, although it does seem to be both more recent and less useful, given that there are plenty of other things that have also been called ‘Hume’s Law’. As is often the case with things that reduce to a slogan, it seems to be used in very different ways. Here are some various formulations that often get thrown around when talking about the ‘Is-ought fallacy’ or ‘Hume’s Guillotine’:
You can’t derive an ‘ought’ from an ‘is’.
You can’t derive an imperative from an indicative.
You can’t derive value judgments from factual judgments.
You can’t derive normative claims from factual claims.
You can’t derive evaluative claims from non-evaluative claims.
But oughts, imperatives, and value judgments are all very different things. ‘Ought’ statements, for instance, are generally indicative statements. What adds to the confusion is that all of these, even if they are often true, seem to have obvious counterexamples, yet they are all treated as absolute statements. There are many intriguing puzzles here, and the question is sometimes even raised as to whether the use of the principle is self-defeating. As a friend of mine, James Chastek, once joked, “We can’t derive an ‘ought’ from an ‘is’; therefore we ought not to try.”
Perhaps we should go back and look at the source of this slogan, David Hume (1711-1776).
It may help to give some very general background. There were in Hume’s day two major approaches to ethics, which we might call moral rationalism and moral sentimentalism. They’d overlap in many of their details, but they had very different conceptions of what moral judgments were. On the moral rationalist view, moral judgments were perceptions of relations between ideas. On this account, cultivating good moral judgment is like learning mathematics. As you learn mathematics, you develop the ability to ‘see’ necessary and eternal relations between ideas; these were sometimes called ‘relations of equality’. Moral rationalism holds that cultivating good moral judgment is very much like this: you develop the ability to ‘see’ necessary and eternal relations between ideas, which were sometimes called ‘relations of perfection’. Moral claims are either true or false, and they are necessarily so. This was a very popular view in the eighteenth century, but the most widely read proponents of it were Nicolas Malebranche and Samuel Clarke, both of whom we know Hume read very closely.
Moral sentimentalism, however, denied that moral judgment was in itself a matter of reason. Rather, it was a matter of orderly sentiment. Whereas the moral rationalist thinks that moral judgments are like mathematical judgments, the moral sentimentalist thinks that moral judgments are like judgments of good taste. It’s important to understand that this did not mean that they thought that moral judgments are mere expressions of taste or arbitrary opinion; they thought that they were like judgment of good taste, taste that was informed, sympathetic, based on wide experience, capable of making fine distinctions. The person of good taste was held to have a better understanding of whatever they were discussing, capable of backing up their judgments with good reasons. In short, moral sentiments see the person of good moral judgment as a moral connoisseur, able to distinguish good and bad action in the same way that a connoisseur of wine is able to distinguish good and bad wine.
Hume is a moral sentimentalist, and the passage from which the “Hume’s Guillotine” derives occurs at the end of his attack on moral rationalism (Treatise of Human Nature, Book III, Part I, Section I):
I cannot forbear adding to these reasonings an observation, which may, perhaps, be found of some importance. In every system of morality, which I have hitherto met with, I have always remarked, that the author proceeds for some time in the ordinary way of reasoning, and establishes the being of a God, or makes observations concerning human affairs; when of a sudden I am surprized to find, that instead of the usual copulations of propositions, is, and is not, I meet with no proposition that is not connected with an ought, or an ought not. This change is imperceptible; but is, however, of the last consequence. For as this ought, or ought not, expresses some new relation or affirmation, it is necessary that it should be observed and explained; and at the same time that a reason should be given, for what seems altogether inconceivable, how this new relation can be a deduction from others, which are entirely different from it. But as authors do not commonly use this precaution, I shall presume to recommend it to the readers; and am persuaded, that this small attention would subvert all the vulgar systems of morality, and let us see, that the distinction of vice and virtue is not founded merely on the relations of objects, nor is perceived by reason.
This is a passage that has been interpreted in many different ways by Hume scholars. My own approach to interpreting the passage is to argue that context is important. Hume is not arguing about ethics in general. In fact, he goes on in the very next section to present an account of obligation, or ‘ought’, based on moral sentiment. Rather, he is finishing up his attack on moral rationalism. Indeed, he as much as says so. What conclusion does he want us to draw? “That the distinction of vice and virtue is not founded merely on the relations of objects, nor is perceived by reason.” In other words: the claim only affects positions, like moral rationalism, that take moral judgments to concern relations perceived by reason. Hume’s own account of moral judgments is not affected by this because he is a moral sentimentalist, and thinks that obligations are neither relations nor perceived by reason, but are the objects of a kind of moral sentiment, something we feel. What is more, it is not put forward as a necessary or obvious truth, but as a challenge to the moral rationalist. It depends on the arguments that Hume has made in the rest of the chapter, in which he argued that nothing can have the features that moral rationalists like Clarke attribute to moral relations. Hume is here challenging the moral rationalist to prove him wrong. To be sure, it’s a challenge he doesn’t think the moral rationalist can meet, based on the arguments he’s already given — but it’s merely a challenge.
Regardless of whether this is the best way to interpret Hume, we can nonetheless see that there’s a great deal more complexity lurking beneath this apparently simple slogan. Ray Ingles, in the comment mentioned above, had noted some of that complexity:
And it’s sort of true. You can’t derive an ‘ought’ from an ‘is’. But you can derive an ‘ought’ from an ‘is’ and a goal.
For example, given the ‘is’-es of the rules of chess, and the goal to win the game, it follows pretty quickly that you ought not sacrifice your queen early in the game.
I wouldn’t put it exactly the same way, in part because I think statements of goals clearly fall under the ‘is’ category, but I think Ray’s approach to the matter is essentially right. We use the word ‘ought’ when we’re talking about decisions, plans, strategies — practical matters. So it makes sense to see ‘ought’ statements as identifying solutions to potential problems. If the problem is to build a bridge that won’t collapse in the wind, it follows from the claims of material science and engineering that there are things you ought to do and things that you ought not to do. Faced with the problem of designing an experiment that will test a hypothesis, a good scientist can derive from available facts how the experiment ought to be designed. When you are faced with the problem of how to act rationally, there are facts about reason that undeniably force the reasonable person to draw conclusions about what he or she ought to do.
Now, of course, given that the slogan seems to be used in so many different ways, it’s entirely possible that it’s being used correctly in a particular context. But the next time someone brings it up, press them to explain what they actually mean.
January 10th, 2013 | 12:02 am
Near the end you write, “I think statements of goals clearly fall under the ‘is’ category.” But this cannot be the case. An “is” statement is a statement describing an actual state of affairs, while a goal is an aspiration, an object of the will. What makes it a goal is that it has value, it is perceived as good. It is what one ought to strive for.
So then Mr. Ingles is correct, that an “ought” can be derived from an “is” and a goal, because a goal is an “ought.” But the “ought” could not be derived from the “is” alone.
January 10th, 2013 | 1:54 am
suppose i make a promise (is). the fact that i’ve made a promise (is) places me under an obligation to keep said promise (ought). at least my wife thinks so.
January 10th, 2013 | 4:36 am
In discussing Hume’s argument, Miss Anscombe examined the relationship of “is” to “owes.”
“Suppose that I say to my grocer “Truth consists in either relations of ideas, as that 20/- = £1, or matters of fact, as that I ordered potatoes, you supplied them, and you sent me a bill. So it doesn’t apply to such a proposition as that I owe you such-and-such a sum.”
Now if one makes this comparison, it comes to light that the relation of the facts mentioned to the description “X owes Y so much money” is an interesting one, which I will call that of being “brute relative to” that description. Further, the “brute” facts mentioned here themselves have descriptions relatively to which other facts are “brute” – as, e.g., he had potatoes carted to my house and they were left there are brute facts relative to “he supplied me with potatoes.” And the fact X owes Y money is in turn “brute” relative to other descriptions – e.g. “X is solvent.”
January 10th, 2013 | 8:45 am
Matt Shadle: This is not right, though. Goals may be aspirations, but aspirations are actual states of affairs — we do actually aspire, and what we aspire to is a factual matter capable of being stated in an ‘is’ statement.
Likewise, it is simply false to claim that a goal is “what one ought to strive for”; this may or may not be true. Goals as such are not ‘oughts’ at all; you would be committed to claiming that anything willed is something that ought to be willed, which is nonsense. Further, your argument requires committing the mistake I note at the beginning of the post: you conflate value, good, and ‘ought’, although it is certainly false that values and ‘oughts’ are the same.
January 10th, 2013 | 9:03 am
Michael PS:
An interesting approach. In effect, since (in Humean terms) moral rationalism takes moral judgments to state relations of ideas and moral sentimentalism takes them to state matters of fact, this is jumping through the horns of the dilemma and advocating a third way. And, indeed, one of the problems of simply taking Hume to have established once and for all that Ought can’t be derived from Is, is that in context his argument clearly is based on Hume’s assumptions about the distinction between relations of ideas and matters of fact.
January 10th, 2013 | 9:45 am
Over on the biologos blog, Calvin DeWitt has a series of posts on science, ethics, and praxis. He distinguishes science (how does the world work?), ethics (what is right?),and praxis (what then should we do?).
I assume that “is” in Hume’s guillotine is roughly equivalent to DeWitt’s “how does the world work?” But is Hume’s “ought” more like DeWitt’s “ethics” or his “praxis?”
January 10th, 2013 | 10:52 am
As Michael PS says, read Anscombe’s short, profound and brilliant paper On Brute Facts. “Owes” expresses a kind of “must”, and even has a certain moral thrust, so it serves to refute Hume. Her argument’s upshot is that if one takes “I owe the grocer for the potatoes he delivered” to be non-factual, that one is thereby forced to call a statement like “A bag of potatoes was delivered to me” (a “more brute” fact in relation to the “owes”) non-factual (I haven’t given this argument here: you need to read the paper), which is contrary to our common-sense talk of facts.
One can fruitfully regard her discussion as showing that the notion of “fact” is much richer than what people (including Hume) are often inclined to think.
January 10th, 2013 | 10:54 am
There are also the conditions necessary for observation.
In order to know what is we must be able to observe what is. So, we ought to live, and since we are limited in our ability to observe we ought to promote having more observers in the world. Science is predicated upon multiple living observers, so a life affirming ‘ought’ is a necessary precursor to any inquiry.
January 10th, 2013 | 12:35 pm
Nickp
As Miss Anscombe also notes, “The terms “should” or “ought” or “needs” relate to good and bad: e.g. machinery needs oil, or should or ought to be oiled, in that running without oil is bad for it, or it runs badly without oil.”
This raises some interesting questions about what, exactly, we mean, when we say that an unjust man is a bad man, or an unjust action a bad one, with an obvious bearing on Hume’s contrast of is/ought.
January 10th, 2013 | 12:42 pm
Apparently, those who embrace this slogan offer it so that one ought not even try to derive an ought from an is.
January 10th, 2013 | 2:06 pm
David Oderberg has a brilliant 6-page (brief but persuasive) demolition of the fact-value distinction in one of his books. Must reading. The title eludes me, but it’s not in his *Real Essentialism,* which also deserves a wide reading.
January 10th, 2013 | 2:07 pm
If you want to stipulate that goals are a type of fact, then “Hume’s Law” would perforce be wrong. But you could rephrase it as, “You can’t derive an ‘ought’ without a goal-type ‘is’.” And I’d say that version would be correct… but of very limited scope.
January 10th, 2013 | 2:48 pm
Picking up on Brandon Watson’s point that ought statement are generally indicative, I am reminded of a discussion about the senses of ‘ought’ that began with Gilbert Harmann’s 1975 paper “Moral Relativism defended”. He distinguishes between four types of claims using “ought”.
1. the ought of expectation
2. the ought of rationality
3. the ought of normativity
4. the ought of morality
Sample judgements that correspond with the above typology are:
1. Oscar ought to have arrived by now.
2. The thief ought to wear gloves when attempting to crack a safe.
3. No child ought to die from starvation.
4. I ought to keep my promises.
Consider the claim that *No child ought to die from starvation*. This judgement can be genuinely normative, even if it expresses no purported requirement on the part of a subject to act or not act – in a better world, no child would starve to death, but, for all that, it is not necessarily the case that any subject who concludes thus should bring about the associated state of affairs (it may be impossible for the agent to achieve this, regardless of how better an outcome it is).
The fourth type of ‘ought’ is the prescriptive or imperative term, according to Harman. Presumably, a subject who judges that she should keep her promises is committed to acting that way in the relevant circumstances. So this is an action-guiding ‘ought’, and, since morality in part concerns the nature of actions, this is a moral ‘ought’ (it is also a normative ought, but narrower in scope than the third ‘ought’).
January 10th, 2013 | 3:00 pm
Brandon, I would respond that a factual statement of what my aspiration is, is not the same thing as the aspiration itself. I will try to illustrate this in a couple of ways.
First is the process by which we discover them. If I were to say, “I want to attend the University of Iowa,” how would you discover if this was true or not? You could search my desk to see if I had literature from the university, search my browsing history on the internet, etc., and if you found enough information, you could reasonably conclude that the statement was true. That is because the statement, even though it is about an aspiration, is an “is” statement. But that is not the process one would use to make the judgment about which university one wanted to attend. Yes, I would gather facts about both the University of Iowa and Iowa State, but the facts by themselves would not tell me which one should be desirable to me. I would also have to draw on prior judgments about desirability, such as desiring a certain career, desiring to be with my girlfriend, or whatever.
Second, if I am stating my aspiration in the form of an “is” statement, then it seems very likely that the purpose of this action is to communicate that information to someone else. So if I told you, “I want to attend the University of Iowa,” your reception of that knowledge, by itself, does not entail any action on your part. But, except in rare exceptions like if I forgot where I wanted to go to college, if I make the same statement to myself, even though the words are the same, the purpose is not to communicate information but to articulate my own judgment that attending the the university is desirable. And this knowledge does entail me taking a certain course of action. Again, the judgment itself and the communication of what inf act the judgment is are distinct, even though they might use the exact same wording…
January 10th, 2013 | 3:02 pm
Andrew, that is only because you are keeping the “ought” premise hidden.
Premise 1: Promises ought to be kept. (The hidden one)
Premise 2: I made a promise to my wife.
Conclusion: I ought to keep the promise to my wife.
An “ought” can be derived from an “is” only when the latter is combined with a prior “ought.”
January 10th, 2013 | 4:14 pm
Matt Shadle,
It’s true that factual statements of aspirations are not aspirations, but as far as I can see this is irrelevant, since we’re talking about what a conclusion with an ‘ought’ — which will be a judgment, proposition, or statement — derives from, and derivation is of propositions (or statements) from propositions (or statements). But from statement of the aspiration (I don’t think all goals are aspirations, but if you prefer ‘aspiration’, we can use that) plus factual statements about various constraints, we can get statements about strategies that will meet those goals, including those that are required for them or inconsistent with them; and statements about strategies required for or inconsistent with goals are statements about what one ought, or ought not, to do to achieve those goals. Thus oughts from ises.
What one ought to do in an actual situation will depend on one’s actual aspirations; but talking about oughts or obligations does not require actually having this or that aspiration; we can make hypothetical statements about aspirations and conclude to ‘ought’ statements on the basis of those and other relevant facts. Either way, statements about aspirations are either categorical or hypothetical statements about facts, either what, in fact, the aspirations are, or what the aspirations might be in fact under whatever circumstances. It is positing goals (whether they really are goals are not) that plays a role in plan-formulation, strategy-making, and the like. But goals themselves are not obligations, even weak ones, nor does it generally make sense to express goals in terms of ‘ought’ statements.
Despite common wisdom, then, it is at least not obvious that an ‘ought’ can be derived from an ‘is’ only when the latter is combined with a prior ‘ought’, and should not be treated as such.
January 10th, 2013 | 4:20 pm
Ray,
Agreed. As far as I can see, the reformulated principle, “You can’t derive an ‘ought’ without a goal-type ‘is’,” is equivalent to saying you can’t reach a judgment about means required to achieve an end without presupposing a judgment about what the end is. And that does seem to be true and obvious. It’s also quite important, although not, I think, usually in dispute.
January 10th, 2013 | 4:41 pm
Matt,
Thinking about your response to Andrew, I think you are over-quick. It is certainly true that one way to derive C from P2 is by taking it to be an enthymeme with an ought-premise, like P1, as the implicit premise. But this is not strong enough. What you would need to show is that it is the only way. You would need to be able to rule out, for instance, the possibility that you can also get C from P2 directly, simply given the meaning of terms like ‘promise’, or that there is an account of ‘ought’ that allows you to cash out an ‘ought’ in terms of more fundamental ‘is’ statements of the right sort, which can then be used instead of any ought-premise.
January 10th, 2013 | 6:41 pm
Brandon wrote: “Either way, statements about aspirations are either categorical or hypothetical statements about facts, either what, in fact, the aspirations are, or what the aspirations might be in fact under whatever circumstances.” But you seem to be assuming that all statements of fact are “is” statements. But “one ought not murder” is also a statement of fact. To deny that is to fall into the mistake of G.E. Moore and A.J. Ayer, who believed that ethical judgments were either intuitions or emotions, and therefore the statement “I want to go to the University of Iowa” is nothing more than a factual statement about a feeling I have toward that university. So just because we have a statement about what someone wants, we still don’t know what kind of statement it is until we know what purpose the statement is serving.
So I assume you are imagining a scenario like this:
“Brandon wants to go to the University of Iowa,”
“And going to college costs a lot of money,”
“Therefore Brandon ought to save up a lot of money.”
The first two statements appear to be “is” statements, and therefore an “ought” appears to be derived from two “is-es.” But I think to get to the conclusion, we need to analyze what we mean by “wants” in the first statement. In order for the conclusion to be valid, one of two things must be true. Either we define “want” in such a way that it already includes a moral imperative, therefore meaning that the first statement is not a simple description of a state of affairs after all, or we admit that there is a missing premise here: “Barring any good reasons not to, one ought to do what one wants.” If neither of those are acceptable, then we are reduced to Moore and Ayer’s perspective, that “want” simply refers to a feeling, and then we have no way to reach the conclusion.
As for the case of promises, for the…
January 10th, 2013 | 10:29 pm
It is paradoxical that some philosophers take “no ought is derivable from an is” as an existent law of ethical argument, and say that we “ought” to obey that law. Hume himself in context was referring to some contemporary ethicists who derived moral conclusions from what he called “external” facts. He himself bases his approach to ethics on “internal” facts — the basic inclinations connected with human nature, and he shows the relation of morality to these internal states. Fr. Frederick Copleston, because of the similarity of Hume’s procedure to e.g. Thomistic natural law theory, classifies Hume as akin to a natural law theorist.
January 11th, 2013 | 4:25 am
The “moral ought” has problems of its own.
In her 1958 paper, “Modern Moral Philosophy,” Miss Anscombe famously argued that it was a hangover from the Law Theory of Ethics and, outside that context, in which most modern philosophers no longer believe, it has a merely emotional and incantatory force. Compare Aristotle’s criticism of the notion of a general “good” in NE 6:1, where he acknowledges that which what is required under any virtue may be regarded as required in the same way, if we regard the acts of the various virtue as all obligatory under some system of law (hence an act of courage and of justice are both alike in being “what is law-abiding.”
In a Virtue Theory of Ethics, such as Aristotle’s, “ought” is merely the instrumental ought, based on a concept of human flourishing. But, can we arrive at such a concept? Philippa Foot’s efforts in that direction were not encouraging.
St Thomas, of course, married the two, when he says that God is only offended when we act against our own good [Non enim Deus a nobis offenditur nisi ex eo quod contra nostrum bonum agimus, ScG III. 122]
January 13th, 2013 | 7:13 am
Matt.
But you seem to be assuming that all statements of fact are “is” statements.
No, but for obvious logical reasons, all statements of fact can be converted into literal “is” statements. That’s one reason why it’s never been very helpful to talk in terms of Is and Ought; ‘ought’ is logically a modality, not a copula. The claim that you can’t get an ‘ought’ from an ‘is’ has never literally been true; it’s a figure of speech, and if it’s going to do the sort of work it does, the figure of speech needs to be cashed out much more precisely. Hume did; almost no one else does.
In the example you use,
“Brandon wants to go to the University of Iowa,”
“And going to college costs a lot of money,”
“Therefore Brandon ought to save up a lot of money.”
The reasoning is certainly incomplete, but it’s not because of any need for a hidden “moral imperative”. In this case, the problem to which the conclusion is a solution has not been fully developed — there are a lot of things a person may want to do without actually deciding to do them, and there are other alternative paths to doing it anyway. The first could be handled by simply stipulating that ‘want’ here indicates an actual goal for action, and not the more amorphous kind of wanting, but the second can’t be stipulated away except by adding definite information not already included in the argument. Thus the premises are inadequate, not for lack of a moral imperative, but because there are many more options open for getting the goal, and therefore it can’t be ruled out that something else ought to be done to get the goal. If we put it in a game-theory matrix we’d see that there is enough information lacking that no single strategy rises to the fore, and thus no particular conclusion for action can be derived.
Links
Blogs
Find Us
Contact