Sunday, April 10, 2011

On 'The End-Relational Theory of 'Ought' and the Weight of Reasons'

This paper1 by Daan Evers begins with an explanation of Stephen Finlay's view of the meaning of 'ought,' followed by an extension of that view, and finally a criticism. Readers of this blog are encouraged to read Finlay's own explanation in 'Oughts and Ends,'2 my condensed explanation,3 Evers' condensed explanation,1 or — for bonus points — all three! But for now...

End-Relational Theory: Extra Short Version

'Ought' — when used in an imperative rather than merely predictive sense — breaks down into an implied end (the goal kind of end) and a claim that a particular action is the most likely available way to realize that end.

Suppose I say, "You ought to look both ways before crossing the street." There is an implied end here: not being hit by an automobile while crossing the street. Further, I'm claiming that looking both ways is the most likely available way to not be hit by an automobile while crossing the street.

If that sounds ridiculously obvious, good!

Reasons to Look Both Ways

According to Evers, 'You ought to look both ways' entails 'There are reasons for you to look both ways' as well as 'The collective weight of the reasons for you to look both ways is greater than the collective weight of reasons for you not to look both ways.'

Also, we should be able to start from 'The balance of reasons favors that you look both ways' and validly conclude 'You ought to look both ways.' (I'm inserting the street-crossing example into Evers' more abstract language.)

Sounds reasonable enough to me. Evers goes on to show how an end-relational understanding of 'ought' can be translated into 'the collective weight of reasons.' But then he questions whether it can be made to work in the other direction, i.e. whether we can start with 'the collective weight of reasons' and validly conclude with an end-relational 'ought.'

Ought —> Weight of Reasons

Evers constructs a theory of reasons in steps I won't reproduce here. Instead, I'll skip to his definition of what it means to say a person has 'most reason' to do something relative to an implied end. (Again, I'm paraphrasing and making his abstract language into more specific terms):
You have most reason to look both ways (relative to the goal of not being hit by an automobile while crossing the street) iff the reasons to look both ways collectively raise the probability of not being hit to a value higher than alternate options do.
So the claim "You ought to look both ways..." can be translated into "You have most reason to look both ways..." in a way that retains end-relational and probabilistic aspects.

Weight of Reasons —> Ought

Now suppose we start with:
You have most reason to never point a gun at a person you don't intend to shoot (relative to the goal of you not accidentally shooting a person).
Can we translate this to an 'ought'? Yes!
You ought to never point a gun at a person you don't intend to shoot (in order that you not accidentally shoot a person).
But if this seems to work, then what criticism is Evers raising?

A Problem?

Without paraphrasing this time, I'll let Evers introduce the problem...
So Most Reason generates intuitively plausible results for the weights of reasons relative to a common end. However, our reasons derive from many different ends, and a theory of weight should allow for comparisons of the strength of reasons derived from different ends. As I will argue now, Most Reason fails here.
Oh. He's appealing to the intuition that the weight of reasons to take one action rather than another is often not relative to a single end, but relative to a multiplicity of ends. Evers offers two strategies to save the theory of weight he previously explained:
  1. Deny that reasons properly 'derive their status as reasons from different ends.'
  2. Claim we can always construct a 'superend' composed of lower-level ends, and so preserve the single-end semantics which didn't cause a problem.
Evers doesn't think either strategy is 'feasible,' but then he only argues against (2) without explaining why he thinks (1) is infeasible. Maybe he just didn't think anyone would go for (1), but I genuinely do think (1) is the correct solution.

My argument for (1) will be by analogy.

Philosophers have, at times, argued for the existence of a simple ought or 'ought simpliciter' or 'ought all things considered.' Though playing the rebel and rejecting this notion, Finlay's end-relational theory 'reveals remarkable unity both within normative language, and between it and related nonnormative constructions.' Additionally, as shown in this paper, end-relational theory 'generates intuitively plausible results for the weights of reasons relative to a common end.'

So why not consider playing the rebel again and reject the notion of simple reasons or 'reasons all things considered'? Or, more to the point, 'all ends considered'? In other words, why not affirm that having most reason for an action implies a specific end, just as affirming that one ought to perform an action implies a specific end?

3. (preprint PDF)


  1. Hi Garren,

    I am largely sympathetic to your comments. Small point: I meant to argue that the entailments between ‘A ought to X’ and ‘The balance of reasons favours that A X-es’ fail in both directions. But you’re right that I did so by appealing to the intuition that reasons appear to derive from many different ends. It seemed to me that an E1-based reason F1 to X could be stronger than an E2-based reason F2 to X, but not necessarily because the first explained that X was more conducive to a single end (or combination of ends) than the second reason (imagine that E1 and E2 are both fundamental but not simultaneously realizable ends). It seemed to me that the relative importance of E1 and E2 should matter to the weight of these reasons. But I think you’re right that the reasons for taking seriously Finlay’s theory are also reasons to take seriously the idea that what you have most reason to do is a matter of being most conducive to some (preferred) end or combination of ends. The sense that F2 still has some weight (not zero weight, which seems to be predicted by the hypothesis that its weight depends on the extent to which it explains that X is conducive to E1) might then (perhaps) be explained by a shift in contextually salient ends.
    But there is also another option: the weight of reasons may be determined by the expected utility of acts (as suggested by Finlay in a footnote of 2009): i.e. not just by the extent to which acts are conducive to ends, but also by the extent to which the subject cares about the ends. I have a paper in which I argue that this still prevents logical entailments between ‘A ought to X’ and ‘The balance of reasons favours that A X-es’. But at least there would be a strong pragmatic relationship between the two: speakers will normally say that A ought to X because A’s X-ing has more expected utility for the speaker than other relevant alternatives. I think that may be good enough. (Finlay is currently working on a book manuscript where he addresses many of these issues.)

    Best wishes,

  2. Thanks for stopping by, Daan!

    I started to write a response, but decided to shape it into a more general post instead.