Bad beginnings.

• Author: Alexander, Larry
• Position: Analogical reasoning in law

Like those movie monsters who, when dealt what surely should be mortal wounds, manage to revive themselves and jump out at the prematurely at ease, so too analogical reasoning in law -- ARIL as I shall call it -- is back again in apparent triumph after having been most recently attacked by Peter Westen(1) and Fred Schauer.(2) ARIL's virtues were proclaimed recently by both Cass Sunstein(3) and Scott Brewer(4) in the pages of the Harvard Law Review.

My aim in this Article is to convince you that, like those movie monsters, ARIL is a fantasy. We can reason in law and about law in a variety of ways, but none of those ways is ARIL. Moreover, like those movie monsters, ARIL, were it to exist, would be deformed. It would be deformed by the same process that creates human monsters: bad beginnings.

In discrediting ARIL I mean also to discredit the notion that law is an autonomous discipline with a distinctive form of reasoning. Some who argue for law's autonomy are mystics:(5) they assert that the legally trained can directly grasp relevant similarities and differences among cases without any further reasoning, that is, that legal training produces a special (and presumably superior) way of seeing things. Others are less mystical; they attempt to describe ARIL in terms of familiar forms of reasoning such as deduction, induction, and abduction.(6) Nonetheless, these rationalists, like the mystics, believe ARIL is a distinctive and legitimate form of reasoning. My critique of ARIL extends to both its rationalist and mystical proponents and to the autonomous methodology of law that they seek to advance.

Finally, I attempt to show how the most promising account of ARIL is really Ronald Dworkin's account of law.(7) Dworkin's jurisprudence is an explanation and justification of both ARIL and law's autonomy. Dworkin's account of ARIL is, however, untenable.

In case I be misunderstood, I should make clear at the outset that in attacking ARIL, I am not attacking either normatively or descriptively courts' reliance on precedent. Courts do cite precedents and claim to be bound by them, and my argument discredits neither the sincerity nor the tenability of that claim. I have previously analyzed precedential constraint and concluded that precedent cases can justifiably constrain later ones to the extent the former lay down rules to govern the latter.(8) My argument here is not that courts cannot follow precedents, but that ARIL is not a satisfactory methodology for doing so, either descriptively or normatively.

1. What Is ARIL? The Views from the Literature

   We first must get a clear grasp of ARIL's nature. ARIL is frequently invoked, but infrequently described with any rigor or care. We know that ARIL has something to do with comparing possible resolutions of some present legal issue with past resolutions of "similar" legal issues, and that ARIL is operating when we resolve the present legal issue in a particular manner because of the way similar issues were resolved in the past. But how do we determine whether issues and their resolutions are "similar" or "dissimilar"? And does ARIL operate only in common law decisionmaking, or does it operate as well in decisions invoking canonical texts, such as statutes and constitutions? We must answer these and other questions before we can recognize our target.

   1. Sunstein's Account of ARIL

      Cass Sunstein recently described and praised ARIL in an article in the Harvard Law Review(9) and in a book.(10) According to Sunstein, when ARIL is employed in common law decisionmaking, a court looks at precedent cases, grasps the norm implicit in them, and then applies the norm to the case at hand.(11) The norm need not be the rationale(s) given in the precedent cases.(12) Rather, the court employing ARIL is free to interpret the precedent cases differently from the way in which the precedent courts themselves understood those cases. That in turn means that the new rationale given for those cases by the present court is itself only provisional for, and revisable by, a future court. Cases are precedents only for the results reached (the holdings), not for the rationales given (the dicta).(13) It is the rationale that the present court finds immanent in the past results, not the express rationales given for those results by the deciding courts, that provides the basis for finding cases to be similar or dissimilar.

      Sunstein not only describes ARIL but also applauds it. He sees ARIL as a method by which members of a polity who hold quite different moral and political theories can resolve conflict. Members can come to agree that Case A is relevantly similar to Case B but not to Case C, even in the presence of high-level theoretical disagreement, because they can agree on the low-level -- i.e., less abstract and general -- norms that provide the criteria for judgments of similarity and dissimilarity.(14) These low-level rationales are what Sunstein calls "incompletely theorized agreements,"(15) and they allow us to avoid the stalemates that our conflicting views of the Good and Right would otherwise produce.(16) In other words, even if we cannot agree among ourselves about the relative merits of Nozick, Rawls, and Bentham, we can agree that finding a brooch in the window frame of a requisitioned house is more like finding banknotes on a shop floor than like finding gold rings at the bottom of a pool.(17) ARIL is the legal method of choice for the modern multicultural liberal state.

      Nor does Sunstein restrict ARIL's domain to common law decisionmaking.(18) For him, ARIL can be legitimately employed in both statutory and constitutional cases. If a statute forbids dogs in restaurants, but expressly excepts guide dogs, how should the statute be applied to the airport cop who brings his drug-sniffing dog into an airport restaurant? Sunstein says this is a case for ARIL.(19) Likewise, if the Constitution forbids punishing political protest but does not forbid punishing draft card-burning, how should the Constitution be applied to laws against flag-burning? Or if the Constitution demands a trial-type hearing prior to revocation of welfare benefits, but not prior to the revocation of disability benefits, what does it demand prior to suspension from public school? Again, for Sunstein, these are cases for ARIL.(20) Indeed, even in easy cases of straight deduction from clear canonical rules, interpreting those rules almost always involves applying ARIL: we see that the green Honda before us is "like" the blue Ford the legislature was picturing when it banned cars in the park.(21) ARIL, and the low-level, incompletely theorized rationales it produces, is truly the most important legal methodology for Sunstein.

   2. Scott Brewer's Account of ARIL

      Whereas Sunstein's actual account of ARIL is quite spare and conclusory -- we somehow just grasp the rationales and the judgments of sameness and difference they produce -- Scott Brewer, in a recent article, gives us a fuller account of ARIL and places that account within the broader framework of reasoning by analogy generally.(22) Both aspects of Brewer's piece merit close attention.

      1. Argument Types, Logical Forms, and Rational Force

         Brewer takes ARIL to be a special case of a more general type of argument, which he calls "exemplary" or "analogical" argument.(23) (Brewer prefers the term exemplary to analogical, but nothing hangs on the terminology.) Argument by example -- analogical argument -- requires that we make judgments of relevant similarity and dissimilarity. How we make these judgments, however, is typically left "largely mysterious and unanalyzed."(24)

         We are left, then, with explanations of analogy that tend to fall

         into either of two roughly divided camps: in one camp are those

         who are deeply skeptical about the argumentative force of analogical

         argument; in the other camp are those who evince an almost

         mystical faith that, even though analogy does not have the rational

         force of either induction or deduction, it still has some ineffable

         quality that merits our entrusting it with deep and difficult matters

         of state.(25)

         Brewer believes that analogical argument does have rational force and that it is not mystical. In support of that belief, he taxonomizes the various argument types, explains the logical form and rational force of each type, and locates analogical reasoning within that taxonomy.

         1. Deductive Argument

            One standard type of argument is the deductive or syllogistic argument.(26) In terms of logical form -- the relation between the truth of the premises and the truth of the conclusion -- the truth of the premises of a deductive argument guarantees the truth of the conclusion.(27) If it is true that all men are mortal and that Socrates is a man, then it is true that Socrates is mortal. Thus, deductive arguments have the most rational force of any argument type.

         2. Inductive Argument

            Brewer describes inductive arguments as being of two types: inductive generalizations from particular instances and inductive analogies from particular instances to other particular instances.(28) He illustrates these argument types as follows:

            Imagine a chicken that reasons inductively ... about a sequence

            of events: on each of 500 successive days [days are the particular

            individuals in the inductive premises], when the chicken hears a bell

            (characteristic F), she comes out of her coop and gets fed

            (characteristic G). Thus the reconstructed inductive inference on day 501

            looks like this:

            (1) [x.sub.1] is a hear-bell day and [x.sub.1] is a get-fed day.

            (OR: [x.sub.1] is an F (hear-bell day) and [x.sub.1] is a G (get-fed day))

            (2) [x.sub.2] is a hear-bell day and [x.sub.2] is a get-fed day.

            ([x.sub.2] is an F and [x.sub.2] is a G)

            (3) [x.sub.3] is a hear-bell day and [x.sub.3] is a get-fed day.

            ([x.sub.3] is an F and [x.sub.3] is a G)

            ....

            (500) [x.sub.500] is a hear-bell day and [x.sub.500] is a get-fed day.

            ([x.sub.500] is an F and [x.sub.500] is a G)

            The two typical...