Some people seem to think that legal researchers who fiddle with matters such as Bayes' Theorem are attempting to construct algorithms or some such things that describe how factual inference in trials or other legal proceedings works. It is possible that this is the aim of some legal researchers, but this is generally not what I am after when I try my hand at mathematical analysis of evidence: when I fiddle with Bayes' Theorem, fuzzy logic, or whatnot, I am not attempting to depict how the institution or practice of factual proof in legal proceedings such as trials actually works. I am trying to understand inference, but this is not the same thing as trying to describe how the legal system manages evidential inference.
What might a mathematical (or logical) theory of evidence in or for legal proceedings do? If one's aim in studying the mathematics of evidence is not to describe the legal management of evidence in law, why study the mathematics of evidence (and inference)?
I have asked myself this question before. I now ask the question again because it may be particularly important if one suspects that fuzzy logic says something interesting and important about factual inference in law. A fair question is, "What possible good would a fuzzy explanation of factual proof in law do?"
One answer (one that I have sometimes also given) is that a mathematical account of inference is a valuable heuristic procedure: it is a procedure that reveals to us the implications of our own (logical?) thinking.
This answer has some force, at least sometimes. It has particular force when the heuristic procedure conforms to our untutored intuitions about the workings of sound inference: the answer -- it's all about heuristics, stupid! -- has force, for example, if we already believe that our good thinking takes a Bayesian form and a mathematical account -- in this instance, a Bayesian account -- spells out for us clearly what we roughly but imperfectly already think. This kind of use of mathematics takes us, so to speak, where we already want to go. But this heuristic justification for mathematical analysis of evidence and inference sometimes runs into trouble ...
First, there are those pesky people who refuse to concede that the basic structure of their sound thinking is Bayesian. Well, let's put those silly people to one side for now. But there are other problems ...
Second, sometimes the mathematical calculations seem to run on, so to speak, by themselves -- to such an extent that even a person who thinks that the right logical procedure is being used might feel compelled to say, "I can't honestly say that those calculations represent what I think. I think the conclusion is correct -- I have to say this because I think the method of argument used here was correct and the premises, I think, were correct -- but I can't honestly say that the calculations here portray what I already, if only faintly, thought."
The difficulty with fuzzy logic may be related to this second difficulty: the procedure does not merely elucidate what is already in someone's head. Even if one can do the calculations and personally does the calculations, the procedures and calculations do not seem to be the calculator's. So in what sense (if any) is the mathematical (or logical) procedure "heuristic"?
The law is reluctant to allow legal reasoners -- e.g., jurors, judges -- to surrender their reasoning processes to other agents or mechanisms. The law permits this to happen sometimes, but not often. This is probably one reason why the law is particularly uneasy about analytical procedures that outrun the intuitions of its authorized reasoners.
But there may be a deeper reason why the law is uneasy about -- or uninterested in -- fuzzy logic. The law may take the view that (i) certain things -- e.g., ancillary generalizations, evidential hypotheses, warrants -- must be part of any good reasoning from evidence to possible facts and (ii) fuzzy logic does not make a place, at least not in any obvious way, for such essential features of evidential inference.
"So what?," a logician might say. You are confusing logic and psycho-logic, the logician might say. For example, the logician might add, contentiously, "Just because the law has the deluded notion that ancillary generalizations -- those things you call "evidential hypotheses" -- are necessary to sound argument and inference doesn't make it so. The job of logic is not to make the human psyche comfortable.The inestimable job of logic is to devise procedures that lead to correct answers!"
It has often been noted that people have a tendency to believe the things that please them and to disbelieve matters that make them unhappy. Is it so with law as well? Does the law like to think that its logic is a good logic because thinking so makes the law and lawyers comfortable with what they do and the (deluded) way they think? Is the apparent demand of law, lawyers, and (some) legal theorists that outsiders give lawyers a transparent logic attributable to this?
Another possible answer, of course, is that the intuitions of legal folk have been right all along and that good logic must have the characteristics that legal people have thought all along that good argument must have, and logicians only now (from a longer perspective) are beginning to appreciate this. Hence, on this view, the development of theories of argumentation, the elaboration of Toulmin's theory of argument and logic, and the like are welcome and long overdue developments and are the direction that future studies of the mathematics and logic of evidence in law should take.
But this kind of justification for mathematical or logical analysis of evidence and inference seems to leave fuzzy logic out in the cold. I am not yet prepared to do that. Question: Is there not a place for a mathematical account of evidence and inference that, even though not transparent (even mildly so) to most legal professionals, sheds light on some key features of legal argument, evidential inference, and human knowledge in general? For example, even if the mathematics of fuzzy logic escapes most of us -- and will continue to escape most of us -- is it possible that it is worth studying (but why?) because it grapples with a central(?) feature of at least some legal problems -- matters such as "partial existence"? And is it possible that even if fuzzy logic and its offshoots could and would never be used by professional protagonists in courtrooms, it might nevertheless say something very important about, e.g., factual inference and the way that forensic proof ought to be conducted or managed.(We have our theories about the behavior of plants and we think that some of those theories are correct even though most of us think that the plants themselves do not have in their heads -- do plants have heads? -- the theories that we think describe how they grow etc. Perhaps legal actors such as trial lawyers are sometimes the equivalent of headless vegetables.)