« The Worst Jurisprudential Article of the Year? | Main | Why don't American philosophers of law talk about Kelsen? »

September 30, 2007

An Epistemological Puzzle about Legal Positivism?

MOVING TO FRONT FROM AUGUST 26, as I have finally got around to replying to the excellent comments.


The puzzle is due to Mark Greenberg (UCLA), who suggested it in conversation the other day (when we were discussing this paper of mine).  (Greenberg is spending the calendar year 2007 at UT Austin.)  I'll give my rendition of it, and then sketch what I take to be some possible answers.

According to the positivist, if a judge says, "I have a legal obligation to X," that statement is true if there is a convergent practice of behavior by officials to X, and officials accept the rule describing that practice from an internal point of view.  But--contra Dworkin in "The Model of Rules II"--the fact of convergent behavior is not the justification for taking oneself to have the legal obligation (the judge will not appeal to the convergent behavior in explaining the obligation).  Since the truth-conditions of the statement are not the justification for the statement, how can a judge know that he has a legal obligation?  That's the puzzle.

Two possible answers, in my view.  First:  for an externalist reliabilist, one can know one has an obligation if there is some reliable process causally connecting the truth-conditions for the judgment with the making of the judgment, even if the maker of the judgment knows nothing about those truth-conditions.  I don't think this is the route Hart would go, but externalist reliabilism about justification would be one way of solving the puzzle.

The other route is truer to Hart, I think:  namely, to deny that the judgment, "I have a legal obligation to X" is in fact a cognitive judgment, i.e., that it is knowledge at all.  We have known for a long time that Hart took for granted the then-prevalent non-cognitivism about ethics, and in a quite important recent paper on"Hart's Expressivism and his Benthamite Progress" which appeared in Legal Theory in 2005, Kevin Toh gave the first really systematic account of how the non-cognitivism figures in his theory of law.   If judgments about legal obligations are really non-cognitive--really expressions of acceptance of a norm which licenses the claim about legal obligation--then there is no question about how a judge "knows" that he has a legal obligation. 

But, of course, that also means that it can't be true that he has a legal obligation to X, doesn't it?  Or perhaps it is just "minimally" true (vide Wright, Truth and Objectivity)? 


Posted by Brian Leiter on September 30, 2007 in General Jurisprudence | Permalink


I have a more interesting puzzle about Legal Positivism? Can the following proof that it does not exist be refuted?
1) There are no historical intellectual movements or schools of thought without members, without people who belong to them.
2) Nobody has ever espoused the view that "I have a legal obligation to X" is true if there is a convergent practice of behavior by officials to X, and officials accept the rule describing that practice from an internal point of view.(by diffuse general knowledge)
3) According to the positivist "I have a legal obligation to X" is true if there is a convergent practice of behavior by officials to X, and officials accept the rule describing that practice from an internal point of view.(by Leiter, and reportedly Greenberg)
4) Ergo, there is no legal positivism
Happy end

Posted by: Joseph Raz | Aug 27, 2007 7:45:44 PM

Is it true that judges in the United States have a legal obligation to invalidate unconstitutional statutes? I imagine it is. If so, what makes it true?

Posted by: Brian Leiter | Aug 27, 2007 8:09:29 PM

Convergent conduct together with acceptance from the IPV are at most necessary conditions for having a legal obligation, even on a positivist account. (Bentham and Kelsen add sanctions; Hart adds some features of the content and character of the rule (supposed importance, categorical nature...)) They don't distinguish obligations from other reasons for acting, and they don't distinguish legal obligations from other obligations supported by law.

When one says that the truth conditions of 'there is a legal obligation to do X' are not its justification conditions, one is presumably not denying that they are the conditions that justify believing, knowing, etc that one does indeed have such a legal obligation. One is denying, I take it, what Dworkin asserts: that these conditions must be at the same time the presumptively sufficient conditions for the legal obligation to do X being morally legitimate (by which he means that it is morally justified to coercively enforce them).

Posted by: Leslie Green | Aug 28, 2007 8:57:56 AM

I am confused. Is this a substantive legal obligation you are talking about? Did not Hart suggest that accepted practice is NOT what makes substantive legal obligation?

Posted by: Nicos Stavropoulos | Aug 30, 2007 7:09:42 AM

I'm not sure what Nicos means by a "substantive" legal obligation. (As opposed to what other kind?) But Hart did *not* suggest "that accepted practice is NOT what makes ... legal obligation."

To begin, "accepted practice" does not mean *a practice generally accepted by the norm-subjects*. That mistake is made by Dworkin in LE, 34 line 14ff. (This distorts the whole character of Hart's account, for reasons I explain in 94 Michigan Law Review 1687, 1700-02.) The relevant practice may be limited to a subset of the population.

Second, for a legal obligation to V to be practice-dependent it is obviously not necessary that anyone have *a practice of V-ing*, or even requiring that someone V. Practice-dependence is secured provided only that there exists *some* practice, R-ing, such that, if there were no accepted practice of R-ing, there would be no legal obligation to V.

At CL 2e 85, Hart writes, "The statement that someone has or is under an obligation does indeed imply the existence of a rule...." (the other conditions follow at 86-88). Hart also holds that the existence of a rule requires as a necessary condition "that their behaviour in fact converges" (55)(something it shares, of course, with a mere habit--the sufficient conditions follow at 55-57).

At CL Postscript 255-6 Hart concedes that this account explains only conventional social rules. He maintains, however, that ultimate legal rules *are* of this sort. Non-ultimate legal rules, including some that create legal obligations, *need not* be of this sort. They may be binding because they so recognized by at least one other legal rule that is binding only because of accepted practice. They are binding "before any occasion for their practice has arisen and the practice theory is not applicable *to them*."(256-emphasis added)

That is why I wrote that convergent conduct and a certain attitude are "at most necessary conditions conditions for having a legal obligation." *Necessary* because in no case can one conclude that there is a legal obligation from the mere fact of convergent and accepted conduct. *At most*, because some legal obligations are binding before the relevant conduct and attitude to them has arisen, or could have arisen.

But it does not follow, and it is not true, that no legal obligations are dependent on accepted practices.

In any case, Hart's theory of obligation is wrong--but that is another story.

Posted by: Leslie Green | Aug 30, 2007 10:34:15 AM

Interesting puzzle. Just to make sure I've got hold of the right issue: the question is about whether a judge's belief about the existence of some particular legal obligation could amount to her *knowing* that there is such an obligation. (Given the non-cognitivist option, "claim" or "judgment" would be more neutral than "belief"; but pretend for the moment that we're cognitivists.) And the worry, I take it, is about whether the judge's belief can be based in the appropriate way on the facts that make the belief true. The background intuition is that for a belief to be knowledge, the belief must be based in the appropriate way on the facts that make the belief true; and the worry is that, if positivism is right about the truth-conditions for claims about legal obligation, then judges' beliefs aren't so based (typically, at least).

Brian suggests that the positivists who want to stay in the cognitivist camp could go the reliabilist route. But perhaps there's another possibility.

It's true that when judges attempt (in judgments or in argument, say) to justify their claims about what the legal obligations are, they don't explicitly appeal to the sorts of things that positivist go on about: the convergent behaviour of officials, the internal point of view, etc. Instead judges appeal to legislation, previous court judgments, principles of the common law, and so on. Call these the "professional facts". But maybe a positivist could argue that appeals to these professional facts *are*, in some sense, appeals to those social facts which, on the positivist story, are the truth-makers for the obligation claim. The thought would be that claims about legal obligations which are soundly based on the relevant professional facts *are* claims which are appropriately based on the truth-makers for the obligation claim. That would explain why the judge's beliefs amount to knowledge.

There's quite a bit of hand-waiving in there, but I think the general line has something going for it. There's a tendency for discussions of positivism to focus on the very abstract parts of the truth-makers for claims about legal obligation: the convergent behaviour of officials, the internal point of view, etc. This can encourage the idea that the "real" truth-makers are foreign to ordinary legal practice. But the complete truth-maker for any judgment about a particular, concrete legal obligation will have to include more specific social facts -- facts about, eg., what particular pieces of legislation say, or what courts have said in the past. And these things are central to ordinary legal practice. So we can at least say this: a judgement about a particular, concrete legal obligation which is soundly based on the relevant professional facts will be a judgment which is based on *part* of the complete truth-maker for the judgment.

In the end, this proposal may just be a variant on the reliabilist story (soundly basing one's judgements on the relevant professional facts is after all quite a reliable way of producing true legal judgements). Still, it seems to me to have a slightly less externalist flavour. For one thing, judges are certainly aware of -- and presumably often know -- the relevant professional facts on which their judgments about obligations are based.

Posted by: James Morauta | Aug 30, 2007 2:20:36 PM


"But ... the fact of convergent behavior is not the justification for taking oneself to have the legal obligation (the judge will not appeal to the convergent behavior in explaining the obligation)."

By "the justification," do you mean the judge's reason for thinking that he has a legal obligation to X? Your appeal to what the judge would "appeal to" makes it sound that you're speaking not of the judge's reasons, but rather what the judge would cite in conversation as support for his belief. But of course these two things--the "real" reasons and the "discursive" reasons, let's call them--could come apart.

It is certainly possible that the judge (i) accepts legal positivism and (ii) believes that the relevant pattern of behavior obtains, and so (iii) infers that he has a legal obligation on the basis of those two things. Even if the judge wouldn't cite those factors as discursive reasons, they still might be his real reasons.

Does it matter for your purposes that the judge *could* come to hold the relevant belief in this way? Or is it sufficient that the judge would not, as a contingent matter of fact, come to hold the belief this way?

Here are a couple other ways the judge might come to know, assuming that legal positivism is true:

(A) The judge gets it on testimony.
(B) The judge has an intuition that he has a legal obligation to X.

Posted by: John Turri | Aug 30, 2007 2:46:17 PM

Is this puzzle any different from what confronts, just for an example, the mathematical Platonist? According to the platonist, the truth conditions for 2+2=4 are just the facts that those numbers stand in the stated relations. But the justification conditions involve the wiggling of fingers, etc. Logicists would have a similarly hard time explaining how, prior to Frege, anyone knew that 2+2=4. This is just off the top of my head, but aren't there numerous philosophical doctrines in which there is little if any overlap between truth conditions and justification conditions?

Posted by: Eric W. | Aug 30, 2007 3:45:43 PM

I just noticed something, which is probably minor. In the setup, Brian says, "According to the positivist, if a judge says, "I have a legal obligation to X," that statement is true IF there is a convergent practice of behavior by officials to X, and officials accept the rule describing that practice from an internal point of view."

Was that 'IF' supposed to be an 'ONLY IF' or even an 'IFF' (plus, perhaps, some commitment to metaphysical determination from right-to-left)? If it's just a sufficient condition, then I don't see any puzzle arising.

Judging from my limited knowledge of legal positivism, I'm guessing it's supposed to be an 'iff' with determination from right-to-left.

Posted by: John Turri | Aug 30, 2007 10:21:26 PM

The practice theory is supposed to explain an obligation to X by reference to an accepted practice of X-ing. It is therefore not meant to explain what I call substantive legal obligations, i.e. obligations imposed by what Les calls non-ultimate legal rules. For a given obligation to X imposed by such rules, it is not true that the obligation is explicable by reference to an accepted practice of X-ing—such a practice need not obtain. These rules owe their force to validity, not acceptance. This is clear enough from Hart’s original book, and is nicely confirmed in the passage from the postscript Les quotes in his response: ‘the practice theory is not applicable to them’.

Accepted practice is of course in play in this case too, albeit in the background: substantive legal obligation to X is ‘dependent’ on accepted practice of what Les might call the ultimate legal rule(s). This is after all what validity is all about. But in this case, the practice in play is not a practice of doing that which it is a substantive legal obligation to do, i.e. X, but a practice of doing something very different.

Brian’s description of the challenge put to him does not involve ‘dependence’ of an obligation to X on a practice of Y-ing. It involves a practice of X-ing making it true that an obligation to X exists. I suspect that Brian has in mind what is sometimes called judicial obligation. That is obligation the content of which is determined by official practice at what we might call the ‘ultimate’ level, where the practice theory is applicable. The challenge invented by Joseph again involves an obligation to X and a practice of X-ing. It is different in that it is about legal obligation, unqualified. Of course, neither Hart nor anyone else has ever espoused the view Joseph describes, namely that a legal obligation to X exists if it is the accepted practice of officials to X.

Posted by: Nicos Stavropoulos | Aug 31, 2007 11:58:05 AM

Why is there a puzzle specifically here? In general, it is true that truth conditions are different from justification conditions, not only in philosophy but with respect to virtually any scientific theory. One of the truth conditions for plate tectonics is that there are massive subterranean plates moving about beneath the earth's surface. No one tests plate tectonics by looking for the plates but instead by looking at surface features predicted by that theory. One of the truth conditions for the general theory of relativity is that space is warped in the vicinity of matter. What justified the acceptace of the GTR was observing light bending near the sun, not observations of the curvature of space per se. It seems to me a perfectly general rule that, as soon as one starts talking about postulated or theoretical entities, justification conditions diverge from truth conditions. If they did not, empirically-grounded theoretical speculation would be impossible.

I'm also midly perplexed by the talk about a judge knowing that he has a legal obligation. Why is that the pertinent propositional attitude for posing the puzzle? Surely, rationally believing would be enough, would it not?

Posted by: Larry Laudan | Aug 31, 2007 2:34:54 PM

First, let me say thanks for an excellent and illuminating set of comments. And, second, my apologies for taking so long to post some replies. But better late than never (perhaps!).

Nicos Stavropoulos, in his second, long, comment correctly states the issue involved in the puzzle and thus the crux of the response to Joseph Raz's amusing first comment (and also to Les Green, though Les's elaboration on my quick statement of the truth-conditions for ultimate obligations are, of course, correct and welcome).

James Morauta also gives a nice restatement of the puzzle in his comment, though I am inclined to the view (as he suggests) that his version of a response is in the "reliabilist" family. I'm not certain about its "externalist flavor," or lack thereof: I take it that even on James's proposal judges can be ignorant of the truth of legal positivism, even though we can see how the justification-conditions he proposes are derived from the actual truth-makers. But that is probably not enough for the account to be an internalist one.

John Turri is surely right that the "real" reasons and "discursive" reasons could come apart, but I don't see any reason for thinking that is true in the judicial case. John's "possible" justification scenarios are, indeed, intelligible, but they aren't actual, or, in any case, I see no reason for thinking they are actual.

Eric W. and Larry Laudan both make versions of the important point that this "puzzle" is not peculiar to this context--and, more importantly I take it, they're not so sure it's really puzzling, given how common truth-conditions and justification-conditions come apart. (As a side note, I don't see that the "puzzle," if it is a puzzle, would be any different if we substitute "rational believing" for "knowing," per Larry's worry). If there is a special reason to be worried about this in the case of the positivist theory of law, it must be this. On the positivist account, law is, at its foundations, a product of human conventions of a certain kind. Do we have the notion of a practice whose content is fixed by human conventions where the truth-conditions and justification-conditions for knowledge of (or rational belief about) that content come apart? Maybe that is the really crucial question here. Thoughts?

Posted by: Brian Leiter | Sep 30, 2007 5:20:11 PM

The comments to this entry are closed.