The foregoing cases suggest that specific guarantees in the Bill of Rights have penumbras, formed by emanations from those guarantees that help give them life and substance. See Poe v. Ullman, 367 U. S. 497, 516-522 (dissenting opinion). Various guarantees create zones of privacy.
Lotfi Zadeh's seminal article in fuzzy sets was published in 1965. It is extremely unlikely that any of the Justices had read the article or had even heard of the article or the theory. So there is no direct lineage from Zadeh's theorizing to the Supreme Court's theorizing in Griswold v. Connecticut. Nonetheless, ...
The dynamic evidence page
Evidence marshaling software MarshalPlan
It's here: the law of evidence on Spindle Law. See also this post and this post.
7 comments:
Peter,
I hope you are not losing brain function beyond the loss that is expected for people in our cohort. If I may be permitted a small observation, you have over the last year or so developed a tendency to say that all sorts of reasoning are acceptable if they are what people do. Woolly wishy-washy bollux. Academic pish-posh. And now fuzzy logic. All fuzzy logic seems to say to lawyers is that we don't know everything. What does it add usefully to what Bayesian analysis can tell us about conditional probabilities and logical errors? To say that courts are reasoning in a way that makes fuzzy logic useful for lawyers is unnecessary, as it is only to say that courts recognise that we can't be sure to the necessary (pragmatic) standard about some things.
It seems to be an academic disease to state the obvious and then to lay claim to that statement as the foundation for a new school of thought. If scholars made the effort to translate their writing into language understandable to intelligent lay people they would realise that there is little new or astonishing in what they have to say. But you know that.
You can help us by using legal judgments and rulings to illustrate how various thinking tools could be applied to improve reasoning. That seems to me to be an important contribution that the academy should make. Your fact investigation course seems to be doing exactly that, but it should use only the best analytical tools that people have devised. I know that fuzzy logic is thought to be useful in some fields, but in law is it really an advance on Bayesian reasoning? If not, don't get distracted by it.
Regards,
Don.
Don, I would have to write an article to respond fully (if, that is, my brain functions were good enough to allow me to write an article). Let me make just a few short observations:
1. In my fact investigation course (and I thank you for the compliment) I don't have my students do Bayesian analyses either. In my view the proper structuring of inferential problems and arguments must precede the attempt to assess the force of an evidential argument. Cf. Schum's book on the evidential foundations of probabilitic argument.
2. Yes, I do think normative inferential theory must be tied to "actual cognitive activity" (or something like that). I have never accepted the thesis that normative inferential theory and descriptive inferential theory are or should be radically disjoint. And I don't think that paying careful attention to "actual inferential activity" (or something like that) makes logic immaterial. Cf. Tim van Gelder. And I make some comments about this point in my paper "Are There Universal Principles or Forms of Evidential Inference?" (2008 2011).
3. I have long thought that fuzzy logic is not mush. But theorists who accept the validity of fuzzy logic have not yet explained how fuzzy theory can explain or improve evidential inference in legal proceedings. I hope that this situation will finally begin to change. Perhaps the workshop in Lisbon http://justice12.quantius.org/ will take a step in that direction. (Zadeh will participate by videolink.) And Kevin Clermont has an interesting draft paper that carves out some new ways of thinking about fuzzy logic and factual proof in legal proceedings. How fuzzy logic should be used (if at all) either to dissect actual inference in trials or improve it is an extraordinarily tricky question that I can't answer very well at this point.
4. I think fuzzy logic has gone beyond being a "deviant logic" (compare Susan Haack). Fuzzy logic and rough logic are truly revolutionary developments in logic. The core idea underlying them is valid. (The question of whether various branches and offshoots of fuzzy and rough set theory make sense is, however, far beyond my ken.) At this point fuzzy logic in law seems best suited to describing the "natural behavior" of words and concepts such as "extremely hostile." I can't yet see how fuzzy logic should be used to _prescribe_ the "behavior" of such words. (Yes, I know, I know, I said that descriptive and normative inferential theory are not radically disjoint. But I'm not being inconsistent. The question here is whether some premises for some inferences are sometimes irreducible.)
That's a start?
Don, I had a discussion with a couple of people about fuzzy logic over at Google+ . Here, taken out of context, is one of my comments:
"On the question of whether fuzzy logic is a new or distinct logic: I try to have a healthy respect for my ignorance. But let me say a word or two. The question of whether fuzzy logic is fundamentally different from probability theory reminds me of a discussion I had about seven years ago about the relative merits and demerits of Pearl's causal Bayes nets (as he calls them) and David A, Schum's networks, which leave causal hypotheses "outside," or "alongside," inference networks. In response to suggestions that Schum's sometimes portrayed inferences better (more perspicaciously?) than causal nets, a stout defender of causal nets argued that causal nets could "handle" any problem of inference that other methods (such as Schum's) could. I did not and do not question (because I am unable to do so) that any inferential argument could be recast in the form of a causal Bayes inference network. But it seemed to me then and it seemed to me now that doing so sometimes amounted to making pigs fly: it was wondrous that the pig could fully but it wasn't an elegant spectacle.
"More specifically about fuzzy logic: Part of the reason I was drawn to fuzzy logic (and, later, to rough sets) was because the legal theory I was taught in law school included Edward Levi's INTRODUCTION TO LEGAL REASONING and the HART-FULLER debate. In each of these places there was much talk, effectively, about the fuzzy or elastic boundaries of legal concepts -- penumbras, H.L.A. Hart called them. (This was also a theme in all of the courses I took in the first year of law school.) I bought this party line hook line and sinker -- and I still do. Now one thing always struck me: It seemed to me then and it seems to me now that it does not do justice to to notion of a legal concept's penumbra to say that some situation has x probability of being within the boundary of some concept and that it comes closer to the mark to say that some situation seems to fall to x degree within the legal concept or category in question. Now although fuzzy and rough logic does not capture everything that is going in the relationship between specific facts and legal categories, it does seem to capture an aspect of this relationship that must elude statements cast as probability statements. If I am correct about this, a host of questions arises, which I will leave for discussion ... harrumph ... to the people (including Zadeh) participating in the workshop on "causality and inference" in Lisbon May 22-24, 2012. See the soon-to-be-final web site at http://justice12.quantius.org/ (It should be in final form by Monday, January 9 or Tuesday, January 10, 2012.)"
Don, a final note: Just as I wouldn't give a jury Bayesian equations and ask the jury use them to compute, e.g., posterior probabilities, I wouldn't give the jury numerical expressions of fuzziness and ask the jury to apply or manipulate such expressions. However, just as on some occasions I might give the jury visual representations of uncertainty that rest on Bayesian logic, on some occasions I might also give the jury visual representations of fuzzy words or concepts (but such visual representations would have to be developed by someone who is far more adept in fuzzy reasoning than I am).
.....
.... Dave, I'm afraid: my brain grows ever more fuzzy ... it's filled with fuzzy ideas ... Dave, I'm afraid ...
http://tillerstillers.blogspot.com/2012/01/hart-fuller-debate-post-modernized.html
Peter,
Thanks for your considered replies.
Re your first reply:
Bayes should help with evidence gathering, for example by pointing investigators to the right populations from which to gather data. Responding to assertions about the relevance and probative value of evidence should be an aim of trial preparation. If students are educated about Bayes - somewhere, even if not in your course - they will be better evidence analysts. Protecting them from the effort needed to acquire the necessary skills does not seem like progress. That sort of insulation may be why hundreds of years have to pass before people become conversant with this sort of thinking skill.
Hum yes.
You didn't mean to say this, but I read in a new "fuzzy theory" in addition to "fuzzy logic". Horror. Fuzzy logic is really the practical guy's way of doing calculations while standing on a ladder with a screwdriver in one hand, a light socket in the other, and the electric cable between his teeth. A vague approximation about the strength of current should be enough to get him down. It's like pretending pi is 4, just to be safe, if you are an architect who wants his building to stand up on its own.
If a person was "extremely hostile", he may have been in the class of offenders who committed the relevant offence. Does fuzzy logic put him in or out of that class? There must be other evidence to support either choice. If he had an opportunity to commit the offence, there might be a small fuzzy inclination to put him in the class, and if he had a motive that inclination might increase. But does this go sufficiently beyond common sense to warrant being recognised as a useful sort of logic?
Re your second reply:
Yes I agree with your point. Not everything is a matter of probability, which is appropriate for either/or questions but not for questions about composite qualities unless one needed to know their exact composition and unless that composition could be expressed as percentages or something else mathematical. This is why in criminal law we don't ask if the defendant is innocent, and we don't require certainty about guilt. Guilt has to be either/or because a verdict is required. In cases where guilt depends on composites, such as recklessness, it would be wrong to require guilt to be proved to a specified probability. (Incidentally, leaping ahead in time, I note James Franklin's reply to your next posting. I think a judge would say "I don't care if it poops in the street, it's still a f****** machine!")
Re your remaining replies:
I probably agree, but I haven't read your paper yet. I am supposed to be on holiday. I remember Hart-Fuller. Wasn't Hart a spy or something exciting in MI5?
Don.
I really should be in bed...
I do not claim or think that Bayesian reasoning never plays an important role during investigation. (It's hard to say everything at once!) I do think that during exploratory investigation inferential problems often have to be "configured" before Bayesian logic can fruitfully be applied. (That's not _always_ the case; sometimes Bayesian logic helps to configure evidential problems, or develop investigative possibilities. But it's often the case.) Toward the end of the first semester in my course in fact investigation course I do ask my students to start developing some fairly simple inference networks, and this task brings them to the brink of Bayesian reasoning.
I find nothing bizarre in thinking about words or concepts as being fuzzy and I think there is nothing necessarily fuzzy about thinking of words etc. as being fuzzy. (Let's try to think precisely, or carefully, about fuzzy entities.) For example, even during preliminary fact investigation legal rules can play a very important role. It is useful and important to point out to students (and investigators) that legal concepts such as "foreseeable" or "reckless" are fuzzy, often extremely so. Investigators have to find a rational strategy for dealing with such fuzziness. If they don't develop such strategies, there is in principle no limit to their investigation. Perhaps a picture or two of a rough set [a set that can, so to speak, both expand and contract] can help investigators manage what can seem to be (and what may be) an infinite number of possible pertinent factual hypotheses. Is this idea absurd? I think not.
For purposes of the kinds of problems I deal with I often use an extremely loose notion of logic. Put it this way: I am interested in structuring thought about factual (and legal) questions. This is why in MarshalPlan (my evidence marshaling software) I actually have a stack called "loose thoughts." Loose thoughts are important in exploratory investigation. The trick is to try to tame them so that they play a fruitful role instead of causing cognitive chaos.
Yes, H.L.A. Hart did do some work for a British secret service - for MI5, I think it was. See the biography of Hart by Nicola Lacey. (In her book she referred to Hart as "Henry." Too familiar?)
Interesting discussion!
Post a Comment