|
Author
|
Topic: Evolutionary Logic
|
William A. Dembski
Member
Member # 7
|
posted 27. September 2002 13:09
I recently posted a frivolous piece on "evolutionary logic" here.
Keying off this piece but taking evolutionary logic seriously, John Wilkins wrote the following serious response on talk.origins. (I would simply include the URL, but things are trickier with citing individual posts on newsgroups.)
quote:
This is something that I am actually interested in - evolutionary logic.
It seems to me that it would have a number of nonstandard features:
Instead of true or false, it would be a many valued logic and the value assigned would be "fitness". This could be a continuous or discrete scale.
Fitness would not be a fully transitive property. Not only would it be contextual, but it would have something akin to a neo-Hebbian decay rate.
It would be a logic that used Sorites predicates, AKA vague predicates. In a fully evolutionary system, the extension of a predicate would be specified by the external relations of the instances of the predicate with other instances in the system. Potential fitness would be defined in terms of these relations - fitness functions would not be absolute or fixed. Neither would fitness functions have a single value over time, and they may even be indiscernable (below the evaluative resolution of
the system).
Systems could become sparsely instantiated, leading to subsystems that effectively behave like independent systems.
All of this is in direct contravention to classical bivalent logics and the treatments that fail to deliver evolutionary accounts (even despite the logician's own wishes) very often do so because they prohibit one or more of these assumptions. In short, classical logic cannot deliver evolution because evolution contravenes the axioms of classical logic.
I say this because it seems to me that many people cannot conceive of evolution being true because they are set in the logic of classical bivalence and exact predicates. Another term for this is "essentialism", although that term has cognitive and psychological implications as well. Strict essentialism is the view that membership in a set is either-or and that all members of a set have some defineable property; AKA intensional definition.
I think that if Dembski is authentic in his beliefs, this may very well be a reason why - he simply *cannot* conceive of a world that doesn't
work according to classical principles.
Let me raise a few questions:
- Is such a distinction between classical and evolutionary logic valid?
- Does classical logic truly presuppose essentialism?
- Assuming that biological reality is essentialistic (certainly a logical possibility), would evolutionary logic necessarily lead us astray?
- How do we determine which logic is appropriate for biology and evolution? Would we have to appeal to some overarching logic to decide which logic to employ? Or would classical bivalent logic be enough (in the sense of "you've either got to use classical logic or evolutionary logic")?
IP: Logged
|
|
|
|
Lizard
Member
Member # 118
|
posted 27. September 2002 22:56
Question for Dr. Dembski,
What is "ID logic?"
Evolutionary pathways are hypothesized based on other known pieces of evidence from other related scientific disciplines. For instance, a biological explanation must fit what is known from geology, paleontology, etc.. If an evolutionary explanation does not fit, then others in various fields will point it out (peer review).
I am keen to know on what scientific evidence an ID hypothesis (thus-far undefined, as far as I know) would be supported.
What *is* a testable ID hypothesis? Behe has proposed growing organisms for X number of generations in a lab and seeing if they grow flagella, but that is both impractical and pointless. Laboratory conditions cannot replicate nature. Whether the organisms did or did not grow flagella in a Petri dish would neither validate nor invalidate the theory of evolution.
I see many on this list and others getting hung up in minute details, while the elephant in the middle of the living room floor is ignored. What *is* ID? What is the "intelligent designer?" Could there be multiple intelligent designers? If not, why not? Why won't ID promoters propose a testable ID hypothesis? Is it because it is impossible?
In my opinion, it is. I am looking forward to seeing some brave soul prove me wrong. Let's quit speculating and arguing about it and test this contraption.
[ 27. September 2002, 22:57: Message edited by: Lizard ]
IP: Logged
|
|
Elend
Member
Member # 326
|
posted 28. September 2002 10:07
First, I'm not sure if you view "evolutionary logic" as just name for the logic of evolutionists (as your piece seems to suggest) or as a formal mathematical apparatus (as Wilkins reply seems to suggest).
If by classical logic you mean bi-valued logic (everything can be either true or false), then it is true that multi-valued logic (3-valued true, false, don't know, fuzzy logic, etc.) has been more successful in modeling reality. Fuzzy logic seems to be especially succesful (just a short reminder for those who do not know what we are talking about): based on fuzzy sets, where objects are not classified as definitely in one set, but according to a certain degree between 0 (false, not in the set) and 1 (true, in the set). In principle classical logic may be viewed as a special, restricted case of fuzzy logic.
Now to specific points:
1. There is a significant difference between classical logic and "evolutionary logic" (say fuzzy logic). Modeling power. Classical logic is elegant and relatively easy to reason with, but hardly of definite practical use - there are simply things that we do not know if they are true and false. Expert systems based on fuzzy or multi-valued logic are quite succesful, where those based on pure classical logic are even hard to imagine.
2. Since essentialism definitely classifies objects as in or outside a certain set, I believe it is reasonable to claim that classical logic presupposes essentialism. There are no perfect "cubes" out there, only better or worse prisms conforming within some error to the definition of "cube". The same can be said for almost every abstraction.
3. Since "evolutionary logic" (in my view) is a generalization of classical logic, I don't think it could lead us away. Whatever results can be obtained by classical logic can be obtained by multi-valued logic and more.
4. See 3. Fuzzy or multi-valued logic does not invalidate classical logic or the other way around. Classical logic is just a restriction of the others. It is usually easier to reason with classical logic (means faster), but multi-valued logic has more practical use. Note that neither is complete, neither can offer proofs for all imaginable questions - thank Goedel for that.
IP: Logged
|
|
John Wilkins
Member
Member # 418
|
posted 29. September 2002 20:57
Dr Dembski is a trained and experienced mathematician, and I am not. So I am somewhat surprised that he has therefore never heard of deviant logics. The "evolutionary logic" I somewhat speculatively threw out is just a case of thos deviant logics, including many-valued logics, such as Susan Haack has developed. Just as Riemann wondered what might develop if he did not accept the Euclidean postulate of parallel lines never meeting, deviant logics are based on what we might achieve if we do not assume bivalent truth conditions.
I am equally surprised that Dr Dembski does not realise that if we develop logics based on more than many truth values, or fitness values as I call them, then by definition a two-valued logic must be included in them. Two is a number contained within the set of all cardinal numbers, I believe. There is no need for a "metalogic" to decide which one to use - use the one that best fits the phenomena under discussion. Logics are abstract languages that may, or may not, have any application, and it would seem that an evolutionary logic would prove itself one way or the other fairly quickly.
Evolutionary biology is well modelled using differential equations, systems dynamics and recently of course, through simulated life systems like Tierra. While these do not require a novel logic, they are hard to fit into the Aristotelian view of bivalent and standard logics, and the nature of vague predicates is a current problem in logic. This is not a new set of problems - vagueness has existed in formal logic since the beginning, under the term "Sorites heaps" - exemplified by the grains of sand that at some point are a heap, but where there is no sharp line. These are present topics, and I have to wonder why Dr Dembski saw fit to introduce this in this way - does he think that these are not problems in modern logic?
And asking whether or not evolutionary logic would deliver a false view if life is essentialistic is a very silly thing to say. Life is not essentialistic, as the past two hundred or more years of biology have shown. This is something that Agassiz, no friend to transmutational (evolutionary) views, held back in 1842. This is something that recent systematics debates - the so-called phenetics debate of the 1960s - showed very clearly. Not only are taxa not defined by essential characters, different measures will deliver different taxonomic distinctions.
To ask, what is the "essential nature" of a species has been justly called "squaring the circle" by Paul Griffiths, a leading philosopher of biology. The sole essential answer that can be given is: common ancestry. If Dr Dembski believes there is still room in biology for essentialism, perhaps he should start looking at the biological evidence a bit more. If Darwinian sources are not to his liking, then perhaps Bateson's Materials for the Study of Variation (1894) might serve. Bateson did not believe that continuous variation could deliver new species (and was a saltationist) but he could not, and did not, deny that there was variation in essential characters.
In sum, Dembski is either unaware of some fairly standard treatments in logic and some fairly basic facts in biology, or he was not interested in a discussion and sought merely to disparage me. This also occurred in the Brainstorms session "Response to Howard van Till" where he accused Elsberry and I of saying that no design inference of any kind could be achieved, when in fact we said explicitly (indeed it was the point of our article) that one kind - the ordinary kind - could be. He has of this date still not apologised or responded. Rhetorical tricks of this kind do nothing to improve one's opinion of intelligent design arguments.
IP: Logged
|
|
William A. Dembski
Member
Member # 7
|
posted 29. September 2002 22:07
Briefly to Wilkins,
I posed a series of questions. Wilkins, it seems, thinks that in merely posing them I've committed myself to their answer. I haven't. Also, he infers from my raising those questions that my knowledge of vagueness and multivalued logics is deficient. I've actually done a fair amount of study in this area. Indeed, my most recent contribution to Progress in Complexity, Information, and Design is precisely on the topic of vague predicates -- have a look here.
As for responding to Wilkins's concerns about my reference to his and Elsberry's work in my response to Van Till, I expect to get to it sometime. Just for the record, I do take my critics seriously. But I have so many of them that between trying to get my own work done and responding to my critics, I need to set priorities. Right now Allen Orr, Elliott Sober, and Jeff Shallit are ahead in the queue.
[ 29. September 2002, 22:17: Message edited by: William A. Dembski ]
IP: Logged
|
|
John Wilkins
Member
Member # 418
|
posted 30. September 2002 01:34
I am more than happy to take a subordinate position in the queue to Sober, Orr and Shallit. ![[Smile]](smile.gif)
However, I do not think that your knowledge of logic, especially deviant logic and vague predicates, is deficient (nor even as meagre as my own). That was rather the point of asking why you commenced the thread that way. Since you demonstrate that your expertise is, in fact, quite sophisticated, my inference must be that you were attempting something else.
[ 30. September 2002, 01:36: Message edited by: John Wilkins ]
IP: Logged
|
|
|
Printer-friendly view of this topic