Wednesday, July 18, 2007
A Really Good Analogy is Like Finding a Ten Point Match in Fingerprinting
There the question was whether Laurence Tribe's article using curved space was a good analogy for constitutional law issues. And Professor Dorf called into question the power of Judith Jarvis Thomson's well-known analogy to the famous violinist mysteriously attached to one's circulatory system as a way of looking at the morality and legality of abortion. (I read the essay in Tom Grey's jurisprudence class in the spring of 1979 and I remember thinking the analogy was powerful, but that is somewhat beside my point here.)
Professor Dorf says:
The point of an analogy is to take something fairly complicated and compare it to something simpler that the reader/listener already understands.
I am not sure if "complicated-simple" is the point of an analogy. What is critical, it seems to me, is the pre-cognitive (abductive?) recognition of patterns by which we say if A is sufficient similar to B, then if A leads to C, B will lead to something similar to C (call it C prime). To make a point about analogy with an analogy, the power comes from the extent to which A matches B, like a five point or a ten point match in fingerprinting, even though it is still no more than an explanatory theory about the causal relationship between A and C, on one hand, and B and C prime, on the other. ("Complicated-simple" seems to me to have more to do with a model than an analogy, both of which bear some resemblance to a metaphor, but that takes us into "family resemblances" among words, and I don't want to go there because there is a whole class of people who will stop reading anything that includes a reference to W............)
We sense there is something fundamentally different between social structures like the Constitution, or a corporate entity, and physical structures, like space, so it weighs against the analogy, but we see the pattern nevertheless, so the analogy still has some power. I used to sit in board meetings in which non-lawyer board members discussed transactions, and it would drive me nuts when they would get asset sales and stock sales mixed up. So I had this explanatory analogy. Think of the corporation as an egg carton, and the assets as the eggs. The rule is that you can't cut up the egg carton. You can sell interests in the egg carton, or you can take the eggs out and sell them, but you can't cut the carton. Then some smart-ass would say, "okay, how do you explain mergers?" And I would respond "well, if you have two egg cartons, you can magically superimpose one upon the other and now instead of a carton that holds a dozen eggs, you have one that holds two dozen eggs, except that it still looks just like the first egg carton." At which point somebody would say "uh, guys, can you take this off-line and let us get back to discussion of the deal?"
My intuition is that a good analogy is something like finding a ten point match in fingerprinting, but it breaks down because good analogies are not necessarily as quantitative as that analogy implies. If your head is not spinning yet (note that is a metaphor, not an analogy), go get some more coffee and be glad you are not me trying to put this together into an article, which I was supposed to be doing instead of writing this post.
TrackBack URL for this entry:
Listed below are links to weblogs that reference A Really Good Analogy is Like Finding a Ten Point Match in Fingerprinting:
I agree that "complicated-simple" is not the (typical) point of an analogy. Rather, analogies are used to help us come to grips with that which is novel or unfamiliar "or surprising situations that do not readily fit into known patterns, [situations in which] we no longer feel we are seeing the world as it actually is." As Thagard and Holyoak point out in Mental Leaps: Analogy in Creative Thought (1995), analogies are, roughly, subject to three constraints: similarity, structure, and purpose (they examine each of these in some detail). After getting a handle on analogical reasoning we might take a look at "conceptual blending:" Gilles Fauconnier and Mark Turner, The Way We Think: Conceptual Blending and the Mind's Hidden Complexities (2002). And we might appreciate such works while bracketing or not necessarily subscribing to philosophy of mind presuppositions and assumptions that some of these authors adhere to (much like we can learn from Lakoff and Johnson's work on metaphor without taking on board their reductionist picture of our mental life), presuppositions and assumptions I think are not necessary to appreciating what they have to teach us.
Posted by: Patrick S. O'Donnell | Jul 18, 2007 11:21:23 AM
Comparing one fingerprint to another fingerprint does not seem to me to be a felicitous example for illustrating what a good analogy is like if only because the "source" and "target" of the analogy are already of the same type (as it were).
Posted by: Patrick S. O'Donnell | Jul 18, 2007 11:30:44 AM
I think analogies, particularly to nonlinear dynamical systems theory, may be of significant use in constitutional interpretation. In fact, my first law review article is on precisely this point. It's not on SSRN, but here is a link (PDF).
FWIW, I'm currently working on a paper applying NLDST to disease causality and public health policy. Analogies are just that, but the expansive power of metaphor indicates, at least to me, how useful they can be as meaning-making constructs. Rorty and Keller are excellent on this point, as are some sociologists of science and technology (Latour leaps to mind).
Posted by: Daniel S. Goldberg | Jul 18, 2007 11:42:59 AM
Right: complicated/simple is not an important distinction for analogy. What analogies are good for is extending our knowledge from a better understood situation (source) to a less-well (or novel) situation (target). In order for an analogy to be truly useful, the two situations should have a similar underlying causal structure, not just be similar on the surface.
We know this from our legal training. Two cases can involve the same types of parties (i.e., be superficially similar) but that is irrelevant. What matters is that the legal relations between the parties are similar (i.e., the cases have similar structure). Only then can we predict what will/should happen in the second (novel/less-well understood) case.
I once wrote an article with Keith Holyoak (see Lipshaw posting) called "If Saddam is Hitler who Is George Bush?" It was about the first Persian Gulf War and, so, GHWBush. One point we made there is that you can "push around" what looks like a good analogy by adding information but some structural things stay in place. So, for example, if you think that Bush I was like FDR then you think the US of 1991 was like the US of 1941. However, when faced with the realization that in 1941 the US didn't jump into the war but rather waited until it was attacked, then Bush I was like Churchill (well, that's what the subjects said) and the US of 1991 was like Britain in 1941. The second analogical mapping has less superficial similarity but more structural similarity when certain facts are emphasized over other facts.
Most analogies are not perfect. Remember learning that "electrons go around the nucleus of an atom like planets go around the sun"? That's a horrible analogy. The only thing it captures is "going around" -- but the path, and the quantum nature, and the force at work are all so wrong that you can't take anything else you know about planets use it to learn something about atoms. (Note that the size difference, a superficial feature, is NOT what makes it a bad analogy.) On the other hand, if you learn that blood vessels are like water pipes, you will be able to use that analogy to guess at a lot of features about blood vessels even though the analogy will eventually break down because blood vessels aren't rigid whereas pipes are.
So, I don't think that the 10 point match is really a good analogy for good analogy. On the pro side, there is a similar underlying causal structure (i.e., the same finger would cause the same pattern) and the prints might not look exactly the same (due to angle or what it was touching) but the points would bear the same relation to each other in both. On the other hand (no pun intended), it's a bad analogy not only because it is looking for identity (i.e., whether two things are exactly the same) whereas analogy is looking for important similarities but also because the information is "flat" -- there is a finger that "caused" the two prints and all the points were generated the same way in the process. It doesn't capture the richness -- the interplay of surface and structural and pragmatic features -- that a good analogy does.
Posted by: Barbara A. Spellman | Jul 22, 2007 12:06:49 AM
My fingerprint analogy has proved Barbara's point about analogies. What struck me was the analogy to the number of times there is a similarity, so that a ten point match is better than a six point match. The more structural points you have matching.... Well, whatever.
The deeper meta-question in Barbara's comment is the distinction between explanation as a matter of physics, like the nature of the atom, and a search for meaning or intelligibility, as in the Bush exercise.
Posted by: Jeff Lipshaw | Jul 22, 2007 6:25:15 AM
The comments to this entry are closed.