On "Oughts and Ends"

In his paper "Oughts and Ends,"¹ Stephen Finlay explains how normative ought-statements can be broken down into non-normative components. In other words: how we can understand statements like "You ought to X" or "You ought not X" without necessarily basing them on a prior 'ought.'

Background

In the early eighteenth century, David Hume pointed out that people making moral arguments often jump from is-statements to ought-statements without justifying the sudden change.² A modern day example might be an observation that urban expansion is likely to drive a particular species to extinction, followed by "therefore we ought to create a nature reserve." But there's a missing step! To make the logic work, it seems something like "We ought to avoid causing extinctions" is needed.

The is-ought problem comes from worrying that ought-statements might always require a prior ought-statement, or that some ought-statements are mysterious brute facts.

Demystifying 'Ought'

Finlay's approach is to defend a plausible interpretation of ought-statements which has the nice side effect of dissolving the is-ought problem (to my satisfaction, anyway).

Must, Should, Could, Shouldn't, Can't

Let's talk about these five other words first! These are all called modal auxiliary verbs because they change the way (or mode) in which the main verb is meant to be understood. For example, they can add information about probability:

"All mortals must die."
"Jones is a punctual guy, so he should arrive any time now."
"The new candidate could win this election."
"According to the weather report, it shouldn't rain tomorrow."
"Dogs can't recite epic poetry."

Or they might add a normative tone to the main verb:

"Drivers must stop at red lights."
"Mary should study for her math final."
"You could take the subway or hire a cab."
"Defendants shouldn't represent themselves in court."
"Congressional candidates can't make jokes like that in public!"

Notice how the first set of sentences passively report on the probability of things, while the second set have that distinctively action-guiding feel of normativity. Isn't it a little curious how a whole range of modal verbs just happen to have both probabilistic and normative forms? Or, wait, maybe this isn't a coincidence at all!

Watch what happens when we explicitly mention goals (or ends) for the second set:

"[In order that they avoid violating the law], drivers must stop at red lights."
"[In order that she passes the test], Mary should study for her math final."
"[In order that you make it to your appointment downtown], you could take the subway or hire a cab."
"[In order that they are adequately represented], defendants shouldn't represent themselves in court."
"[In order that they be elected], Congressional candidates can't make jokes like that in public!"

Can you see how the modal verbs are tied into probability again? Drivers who don't stop at red lights certainly violate the law. Mary has the best chance of passing her math final if she studies for it. Either the subway or a cab are ways for you to most probably make it to your appointment downtown. It's unlikely defendants will be adequately represented if they represent themselves. And there's no chance Congressional candidates will be elected if they make those kind of jokes in public.

So maybe these five modal verbs always have to do with probability, and optionally relate to some goal. When they do relate to a goal (or end), they gain normative tone! This is Finlay's end-relational theory. It accomplishes, as he puts it:

"a straightforward analysis of instrumental normative language, unifying the language of ordinary modality and normativity, and providing a univocal semantics for two isomorphic sets of terms."

Back to 'Ought'

The previous section was a bit of a trick. It turns out we've already been discussing 'ought'! It works about the same as 'should.'

"Jones is a punctual guy, so he ought to arrive any time now."
"Mary ought to study for her math final."
"[In order that she passes the test], Mary ought to study for her math final."

The first sentence uses 'ought' in a probabilistic, non-normative way. The second uses a normative 'ought.' And the third sentence reveals how the normative 'ought' was formed by relating probability to an end.

The Many Flavors of 'Ought'

If 'ought' is end-relative, then the full meaning of 'ought' varies from instance to instance, as far as ends may vary. A critic may claim this is a terrible disadvantage compared to another theory which assigns 'ought' a single, full meaning all the time. "However," Finlay says,

"although the end-relational theory recognizes a multiplicity of ways 'ought' is relativized, it also gives a universal semantics for 'ought' itself. 'Ought,' on this view, is no more semantically ambiguous than attributives like 'real', comparatives like 'big', or indexicals like 'here', which despite their complex interaction with context are not difficult to interpret."

Some...Maybe All?

If you're ready to acknowledge that some 'ought's gain their normativity in the manner described by the end-relational theory, progress has been made in demystifying normative 'ought'! Finlay takes things one step further by suggesting normative 'ought' always presupposes ends. He claims the end-relational theory can account for the empirical data and shows how some likely objections fail to show otherwise.

A critic might, for example, try to categorize the end-relational theory as an instrumental theory of 'ought' in the sense that "if you want to do X, you ought to do Y." This would be an inadequate account because we very often use 'ought' in a way that goes against a particular agent's desires. But the end-relational formula of "in order that X, you ought to do Y" avoids this limitation by not binding the meaning of 'ought' to a particular agent's desires.

The end-relational formula may sound too weak to provide practical guidance. If you tell me that I ought to do Y, but I realize the end it would serve is X — and I don't care about X — then I could shrug and go about my business. But it may be the case that X is especially important to me, so you could guide my behavior through an explicitly relativized 'ought.' From the paper:

"'In order that I don‘t kill you, you must come with me' may be end-relational, but this 'must' is not lightly ignored."

Alternatively, you may influence my attitudes or behavior by taking advantage of (exploiting) the elliptical nature of ought-statements. If you forcefully tell me what I 'ought' to do without making any end apparent, I might go along with you! In short, Finlay believes it is "plausible that categorical uses of 'ought' express demands and attitudes just as expressivists claim." 

Hume Revisited

(Going beyond the scope of the paper here...)

As I understand Hume, he was pointing out that something is missing between (I) and (III) in cases like this:

I. Urban expansion is likely to drive a particular species to extinction.
II. ???
III. Therefore, we ought to create a nature reserve.

And as I understand the is-ought problem, the worry is that (II) must either contain an infinite regress of ought-statements or a regress far enough back to hit a brute fact ought-statement. But what about something like this instead:

I. Urban expansion is likely to drive a particular species to extinction.
III. Therefore, [in order that biodiversity is maintained] we ought to create a nature reserve.

There is no separate (II). Instead, we understand the 'ought' in (III) to presuppose an end which is most likely to be brought about by creating a nature reserve. It's a qualified 'ought' brought to life from probability and a goal. The force this goal has on us relies — as Hume would no doubt agree — not ultimately on reason but on desire.


1. http://www-rcf.usc.edu/%7Efinlay/OughtsandEnds.pdf (Preprint PDF)
2. A Treatise of Human Nature III.i.i (last paragraph)