|
Author
|
Topic: Evolution and Design: a synthesis
|
charlie d.
Member
Member # 159
|
posted 25. March 2002 11:32
Evan:
thanks for the quick reply!
Actually, I don't think you would need to witness these events while they are happening. You can work it out backwards, genetically. Let me see if i can explain it.
The first way would be to analyze known speciation events and study the genetic heterogeneity of the diverging populations: you'd tend to expect bottlenecks (e.g., "mitochondrial Eves", or "Y-chromosome Adams") for darwinian mechanisms, while bottlenecks should be rarely detected in the case of widespread mutation events. This is fairly obvious.
The other way, not involving speciation, would be to take a mutation with known adaptive value (such as this, discussed in the Ka/Ks thread, but there are dozens of others), and analyze whether linkage disequilibrium is there.
"Linkage disequilibrium" is a measure of the association of a particular allele with other alleles at other loci (in most cases, on the same chromosome). For instance, imagine a population in which a gene with sequence X1 occurs on chromosomes bearing different alleles at a nearby locus (alleles Y1 and Y2). If a new mutant allele at gene X (allele X2) occurs only once in a population's history, on a chromosome bearing allele Y2 at gene Y, if you sample the chromosomes in the population, you'll find that X2 will be preferentially "associated" (in a statistically measureable way) with allele Y2, at least until recombination between the loci (which depends on their distance) will re-equilibrate the distribution (assuming of course that combinations X2-Y1 and X2-Y2 have the same fitness).
This is usually what is observed for very recent mutations, or for very closely linked loci (for instance, molecularly detectable point mutations within the same gene, recombination between which is extremely rare).
So, if an adaptive mutation were to occur by darwinian mechanisms (a single event, followed by positive selection), linkage disequilibrium is expected for close loci and/or very recent mutations (e.g., antibiotic or pesticide resistance in natural populations, mutations in the HIV virus in an individual patient), while if the same mutation had been introduced by "intelligent" mechanisms, multiple contemporaneous mutation events would be expected, presumably on genetically heterogenous chromosomes, and therefore no linkage disequilibrium would be observed.
Of course, both hypotheses could be compatible with a low rate of opposite outcomes: for instance, in your theory, adaptive change could still occur by teleologic coordinated mutation of many individuals, but as long as this happens in a small and genetically homogeneous population, the result will be apparent bottlenecks effects and linkage disequilibrium (and vice-versa for darwinian mutations: early crossing-overs could apparently disrupt linkage disequilibrium, or rapid spreading of a mutation in a heterogenous population would eliminate genetic bottleneck effects).
Overall, however, darwinian mechanisms would predict one type of outcome over the other, and your theory the opposite. As they say for the X-files, the answer is out there. Now, all you've got to do is hit the nearest scientific library. ![[Wink]](wink.gif)
[ 25 March 2002, 11:35: Message edited by: charlie d. ]
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 25. March 2002 12:17
Very interesting.
One of the reasons I started this thread was to offer a more concrete hypothesis about ID than I had seen anyone offer elsewhere, in order to turn the subject to positive ideas about what ID perhaps is rather than negative ideas about what evolution perhaps isn't.
One of the reasons this needs to be done is to generate testable hypothesis.
So you are pointing out here that if ID works more or less simultaneously on whole populations, perhaps there would be a testable difference between this and natural evolutionary processes.
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 25. March 2002 13:20
You are right, although I should probably specify "your version of ID-inspired teleological evolution appears testable". (I think, however, many IDists would disagree with several of your assumptions). In fact, as I said, I am quite sure the data is already out there, you should just go looking for it.
In principle, I personally don't see why ID, like any other teleological theory, would not be testable, provided that proper assumptions about mechanisms are made. Over at ARN I have been discussing another teleological, neo-lamarckian theory with mturner (he calls it endogenous adaptive mutagenesis, or EAM), which I also think is potentially testable - you should take a look at it, some of his ideas are quite close to yours.
Anyway, once you have a set of testable hypotheses, you are, whether you like it or not, in the science business.
[ 25 March 2002, 13:26: Message edited by: charlie d. ]
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 25. March 2002 14:12
Charlie d writes, "You are right, although I should probably specify "your version of ID-inspired teleological evolution appears testable". (I think, however, many IDists would disagree with several of your assumptions). . . . once you have a set of testable hypotheses, you are, whether you like it or not, in the science business."
Yes. One of my purposes is to put forward enough concrete detail about a version of ID that we could follow through to some kind of testable hypothesis.
If other IDists might disagree with some of my assumptions, the question I would put to them is what assumptions can they put in their place, and are their assumptions ones that might also lead to testable hypotheses?
There are two challenges here, i think.
One is be concrete enough about one's version of ID that testable hypotheses can be generated.
The other is to be firm about sticking with inferences, hypotheses, and eventually conclusions (presumably) that rest just on the evidence, and don't slide over into metaphysical speculations.
I would be very interested in hearing of other ID hypotheses, from those that might not be satisfied with the proposal I have put forth. Frankly, I am surprised that people in the ID movement have not spent more time thinking about the questions that Tom Stalnaker asked that got this thread started - what happens when ID enters the world, how would we know it has happened, where does it happen, and how would things look different from (as well as the same as) they do according to evolutionary theory.
These questions cannot be answered unless some concrete proposals are put forth.
Even then, the ultimate test of design is improbability, and the difficulty will be that the testable hypothesis will need to include events whose naturalistic probablities are well enough known that the overall probability of the event not happening could be calculated. Having 1000 very similar genetic changes happen in a popuation of 1000 individuals in one generation might qualify.
IP: Logged
|
|
Jules
Member
Member # 181
|
posted 25. March 2002 14:44
Evan,
You write:" Evolution proceeds slowly, whether it is by natural evolutionary microevolution or designed
macroevolution. Whichever happens, the changes at each generation are small, and must
at least not produce a distinct disadvantage in terms of survival. The designer can make
such changes so as to produce viable offspring that are still enough different so as to
continue a multi-generational trend."
Is this part of your hypothesis, or are you stating it as known fact? If the latter, then I, at least, would dispute it. For example, it's not clear that microevolution could have produced the bat's wing or the avian lung, let alone the flagellum. And I'm not sure there are functional intermediates between the wingless and the winged bat.
Charlie,
I wish I could have followed you on that last post. It looked interesting.
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 25. March 2002 16:08
quote:
Jules:
I wish I could have followed you on that last post. It looked interesting.
I am not sure if you are poking fun at me, or you are serious (in which case, you are the first person I know who finds linkage disequilibrium interesting... ![[Big Grin]](biggrin.gif) )
Nevertheless, let me try again. Imagine a population with 50% "blue" chromosomes and 50% "red" ones. Then, imagine darwinian evolution: an advantageous mutation X occurs on a single blue chromosome, and its frequency increases because of positive selection. After a while, if you sample the chromosomes in the population, you should find that all the genes carrying the X mutation are still on blue chromosomes, and none on red ones (we assume for now that chromosomes as such remain intact through generations, for simplicity). Note also that this is independent on the frequency of the mutation in the population (even if at the end you have 90% blue chromosomes, most of which bear X, and 10% red ones, the "preference" for chromosomal location of X still stands). Statistically, you'll be able to say that the X mutation is significantly associated with the blue chromosome. That's linkage disequilibrium.
On the other hand, if the mutation occurs contemporaneously in many individuals, as per Evan's hypothesis, you would expect stochastically that 50% of the X mutations would fall on blue chromosomes and 50% on red ones. Thus, when you do your sampling you would find no preference, that is no linkage disequilibrium.
Now, instead of colored chromosomes, imagine some genetic marker (another point mutation you can track) distinguishing "red" from "blue" ones, close enough to the X mutation so that genetic recombination (you know, physical exchange of DNA material between chromosome pairs, crossing-overs and such) does not take place (or is exceedingly rare), and you have the real thing (well, almost, but you know what I mean). Any clearer now?
[ 25 March 2002, 16:42: Message edited by: charlie d. ]
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 25. March 2002 17:36
If anything is a miracle, life is definitively so. I can certainly appreciate that. But my take is very different from Evan’s, which I have, maybe unfairly, referred to as seeking validation for faith in miracles. But I think that the more often we require the designer to “intervene,” the more haphazard appears the “design,” and the more it resembles exactly what I would expect if the received theory were true. I consider the designer the “irreducibly complex” component. So I want to know what are those aspects of design that can be reduced and those that cannot be reduced.
With that as a premise let me suggest an alternative: The “totopotentiality” of adaptive design strategies, which explicitly seek to limit “appeals to the Oracle,” are a principle interest of mine. This article (referred to me by an ARNie) explains well some aspects of adaptive design without getting all technical on ya, “Intelligent Control” by Panos J. Antsaklis (accessed online 9/30/01). (But it is very important to note, as the author does, the distinction between adaptive control very generally, and its precise meaning in control theory, because I think that is exactly where the received [biological] theory of adaptive evolution breaks down.)
Also of interest are AI/machine learning solutions to the dilemma of the representational limits of the “Darwin algorithm” (natural-selection.com is a good source of interesting articles, including a good review on the evolution of executables by Peter Angeline), and the related areas of constrained function optimization (Behe’s challenge) and least path analysis (which can be abstracted from many articles online about molecular phylogeny and structural biochemistry Markov models + DPA’s). Richard Bellman’s classic “Dynamic Programming” (1957) is still the best, and includes the theorems of observability, controllability, and realizability, which are as fundamental to the applied sciences as they are to the natural sciences. (Constant “intervention” robs design of observability and controllability, and reduces it to a realization of “randomness,” or at least something that becomes increasingly indistinguishable from randomness. “Randomness” enters design in some fascinating ways that should be more carefully considered.)
So you can see where I’m going with this. I think one of the failures of the “Paley program” was that it focused on designs as finished products and ignored the fact that design is a process, an experimental and developmental process. Living designs are not so many sculptures collecting dust in a natural history museum. Nor are they, as Paley emphasized, chimeras. Apart from the broader cultural construals, an evolutionary conception had an obvious appeal to 19th century naturalists. But the broader conception of evolution, which considered Darwinism problematic well into the 20th century, was suppressed by the Neo-Darwinian hegemony. I’m sympathetic to anyone who questions this stultifying hegemony, Evan, so I’m not as snotty as I sound.
Like everything else, designers seek to perfect the design process itself, to reduce its ad hoc, pre-paradigmatic aspects. In doing so, his intent is to limit his role (strange as that may sound) to exactly what cannot be achieved “mechanically.” And that is creativity.
Much of design can be “mechanized,” and I believe that a truly intelligent designer will implement these mechanical aspects of design in the designs themselves. In other words, the designs will actively participate in their own design in so far as that can be mechanized. This frees the designer to do what designers can do and no mechanical process can: Imagine! Create! Make those “windmills in the mind” material! Perfectly natural processes are good at "improvization." Darwin traded on this aspect of nature to good effect. But it is exactly in the limits of "improvization" that his theory collapses. See what I mean?
(I might also pose a question about the utility of traditional evolutionary categories from a design perspective: microevolution, macroevolution, species, mutation, fitness, etc. Some of these terms were never very well defined in the traditional paradigm, so maybe it behooves us to examine them, and a host of other such terms and concepts, more closely.)
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 25. March 2002 22:29
I do not follow all of what Janitor is saying, but I would like to respond to two things he wrote.
First, he writes, “But my take is very different from Evan’s, which I have, maybe unfairly, referred to as seeking validation for faith in miracles.”
I don’t believe that the position I am describing in this thread has anything to do with “faith in miracles.”
In my beginning post, I emphasized that my hypothesis is that the designer can influence the world in a non-miraculous fashion - one that doesn’t contravene the flow of natural law. See Dembski’s “ID Coming Clean” for some very pertinent comments on this. I could explain more if that would be useful.
Secondly, I am emphasizing that the only evidence we can really rely on for design is improbability, and that is a very difficult thing to show. So there is not much being taken on faith here. I have proposed a hypothesis, not a conclusion, and we are discussing both the possible implications of the hypothesis and ways that it might be tested.
Secondly, Janitor writes, “Constant ‘intervention’ robs design of observability and controllability, and reduces it to a realization of ‘randomness,’ or at least something that becomes increasingly indistinguishable from randomness.”
Yes, I agree. In fact one of my main points is that any single act of the designer would in fact be indistinguishable from a naturally occurring random event. The only way we could ever infer design would be by looking at the cumulative result of many such events, having realistic estimates of the probability of each event happening by natural means, and then calculating that the cumulative result of all the individual events was improbably small, less than 10 ^ -150.
This is why Charlie D’s idea is interesting. If the same mutation would show up simultaneously in 1,000,000 individuals, and we could provide indirect evidence that had happened, as opposed to a mutation in one individual spreading throughout a population, we might be able to “do the math” and infer, or not, design.
IP: Logged
|
|
Jules
Member
Member # 181
|
posted 26. March 2002 01:17
Charlie,
I wasn't joking. I find all of this fascinating. If I had had a good high school biology teacher and not been a total klutz in lab, I might have gone into biology.
Now if I understand what you've said about gene? chromosomal? disequilibrium, and if I understand Evan's hypothesis, the designer would introduce the mutation to all (or most) of the members of a population at the same time? And then we wouldn't notice any disequilibrium? Do I have it right?
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 26. March 2002 10:32
quote:
Jules:
Now if I understand what you've said about gene? chromosomal? disequilibrium, and if I understand Evan's hypothesis, the designer would introduce the mutation to all (or most) of the members of a population at the same time? And then we wouldn't notice any disequilibrium? Do I have it right?
"Linkage" disequilibrium (linkage being a measure of the proximity of genes on a chromosome). The formally correct name is actually "gametic association" (because in some cases it has nothing to do with linkage proper, but with preferential combination of alleles at genes on different chromosomes, for functional rather than "geographic" reasons).
As for the answer to your question: pretty much, assuming Evan's hypothetical mechanism. I should stress that this would apply not just to a designer, but essentially to any teleological mechanism (such as a vitalist and/or lamarckian process) that would introduce change simultaneously in a large, heterogeneous population, rather than in a single individual (or small group of individuals). And, of course, it only applies under the assumption that the mechanism would operate by introducing gene mutations of the type we are accostumed to see in nature (as opposed to, for instance, adding or replacing whole chromosomes, etc).
Finally, I should also say that although I did my graduate training in genetics, I am neither a professional geneticist, nor an evolutionary biologist. Any thoughts expressed here are therefore just my personal opinions, which may well be wrong for theoretical reasons I can't see right now, and certainly are not necessarily those of the American Society of Human Genetics, the Genetics Society of America, or any of their members or underwriters. LOL...
[ 26 March 2002, 10:33: Message edited by: charlie d. ]
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 28. March 2002 08:15
I’ve been thinking about Charlie D’s thoughts on the consequences of the designer working with a whole population at once, which would distinguish design from natural evolutionary processes and lead, perhaps, to some testable distinctions.
Let us suppose that we did test and establish that the evidence did not support that whole populations changed somewhat in concert. This would change, I think, our concept of the some of the characteristics of the designer.
My original assumption, that the designer could work on whole populations, correlated with an assumption that the designer (while not omnipotent) could stand back and observe the external world somehow, so as to be in contact with multiple individuals and therefore coordinate change.
If the evidence were to show that this did not happen, then perhaps this would be evidence that the designer is not in any way a transcendent entity or force (one who can stand back and survey the world), but rather an immanent force within each living thing that works to experiment with itself, so to speak, separate from any knowledge or care as to what other individuals are doing.
This view might account more realistically for the amounts of time it takes for presumably designed events to occur (because even short amounts of time geologically are very long amounts of time, and large numbers of generations, for organisms.) Also, of course, design that emanated out from individuals, as opposed to groups, would look much more like natural evolutionary processes, and therefore would make more difficult the job of eventually showing what was in fact designed.
A second issue arises that complicates the issue. The designer is a “intelligent designer”, not in the sense that he does everything perfectly (Dembski has addressed this issue in some of his essay), but in the sense that he is a creative agent, not a mechanical force of nature.
What that means is that there is not necessarily any rhyme or reason to the designer’s actions - no regularities by which we might pin down design. Even if the designer can coordinate design across populations (and therefore across populations of multiple species simultaneously), that doesn’t mean he will, or that he ever does. Similarly, even if the designer works only through individuals, that doesn’t mean he will move an organism from one feature to another by causing maximumly feasible change in a succession of generations.
The designer might make a small change and then do nothing for many generations, perhaps to see how things turn, perhaps because he is in no hurry, perhaps because he is whimsical, and so on. In theory at least, we have no way of knowing.
In fact, if we found that there are regularities to what appears to be design events, that would be evidence that in fact some unknown natural process is at work, and not a creative agent.
The whole issue, then, seems to me to continually go back to the core claim of the explanatory filter: those things which can shown to be highly improbable are designed - period. And yet, as the conversation in the thread “Is Complexity Relative or Absolute” is discussing, showing that the things are improbable (hence designed) is really dependent on methods being developed to show which things are probable (hence evolved).
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 28. March 2002 14:20
Hi Evan:
of course you hit the nail right on the head. Potential falsifications of your hypothesis do not falsify Intelligent Design, because the idea of a Designer per se is unfalsifiable, as you so craftily demostrate.
Your hypothesis, on the other hand, can be tested and falsified because you have tried to do science. And if you still believe in Design after your first hypothesis is falsified, make another one according to a different model, and test that one too. In science, however, you'll never find you are right just on the basis of your unassailable logic (you know, the "beautiful theory slained by ugly facts" thing). That's how the job goes. As a scientist, you get used to being wrong: it's usually just philosophers who NEED to be right.
Now, would you mind please sending this whole thread to the Ohio Board of Education?
[ 28 March 2002, 14:23: Message edited by: charlie d. ]
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 29. March 2002 16:22
I appreciate your desire to develop a testable theory of intelligent design, Evan. But consider the results and implications of an already existing and very highly developed theory of intelligent design:
On received theory environmental parameters provide the “selection criteria.” Accordingly, the population is “adapted to its environment.” This is a process of discrete mapping of the population to a set of continuously changing parameters. Now, as stated, an engineer sees a problem. A fundamental and well-tested deduction from the engineering theorem of observers (as well as a basic principle in higher mathematics) is that a non-linear observer is required for a non-linear system. A population evolving discretely in response to a continuously changing environment is not what an intelligent designer would design. Accordingly, there is either something very wrong with our theory or something very wrong with the design. The theory however contradicts itself: populations are considered to be optimally adapted/adapting. The problem is obviously with the theory and not with the design.
On engineering design principles alone, no population evolving under these conditions could ever be expected to be well adapted or optimized. Even if given an initial state of stability the system cannot be expected to remain stable, and certainly cannot be expected to adapt to change. Its time evolution (in evolutionary time or under any sets of conditions changing out of phase with the discrete mapping rate) is predictably divergent or wildly oscillatory. It crashes. It will spin out of control and fly off into the oblivion of bad designs. (Actually, I think about three theorists have recognized this problem. For most biologists, a certain 19th century theologian solved all problems in advanced 21st century technology in principle!) The obvious conclusion is that populations are in fact continuously adapting.
This now becomes a problem in system identification. Given the complexity of the system, it is a very difficult problem. The system is considered as a “black box” and the components of the system are deduced from the I/O’s, from the mapping of inputs to outputs. A considerable analytical problem is that living designs do not satisfy simplifying minimality and irredundancy criteria. This means that the mapping I > O is not unique. This is indeed a general result for all complex adaptive systems. (It is well understood in biology that this mapping is many to one, the inverse is rarely considered, and in any case this whole process of mapping an “abstract” genotype to its material instantiations is a significant unsolved problem in biology.) But one thing is certain even to begin with: the “black box,” the implementation, is not simply “digital.” In fact we already know enough about what’s in the “black box” to say that it is a “hybrid” system. The control system is digitally implemented/encoded but its operations must be (and are) continuous. We have ever reason to believe, given the inherent instability and inadaptability of the traditional conception, that the time-independent state (its dynamic stability) and its continued evolution/adaptation is due to the continuous (adaptive) operations. Again, biologists recognize that this is exactly what is occurring, except on theory it is not occurring in evolution.
Let me say that this is not in any way intended as a “tentative” hypothesis. It is a conclusion based on a considerable body of experimentally tested theory. If the theory didn’t work then you and I would be living in a world of 19th century technology. So I think my confidence in the conclusion is well warranted. Obviously, the conclusion has serious implications for received theory. It is to be rejected. It also has consequences for any theory of ID that posits either “random” or “frequent” interventions by the designer. The system is plainly not designed for that. It is a perfectly fair question to ask: when is the designer required to intervene and how? What is required of us is to abstract the answer to that question from the design principles themselves, and not presume that the design ever requires any intervention.
Let me suggest a "hypothesis," with far less confidence than the conclusion (above) it is derived from: any complex adaptive system, such as we have, will implement comparatively simple and effective machine learning techniques. The designs themselves will be “intelligent.” (This is indeed a “hypothesis.” It does not necessarily follow from what we know. E.g., the genome could implement an effective recursive filtering technology, which would not necessarily be perceived as particularly “intelligent.” Although “filtering” is an important element of “intelligence.”) Even though the whole field of machine learning/AI is, even after 50+ years, still in its infancy, it has much to recommend here. In so many words, the problem of biological evolution is a problem in the evolution of intelligence at all levels of biology. Not just the human brain.
Wrt the question of “interventions,” or as I like to think of it, “appeal to the Oracle” (LOL): we might consider the minimal implementation necessary (to save the phenomenon), and identify how its intrinsic limitations on induction/recoding of the genomic representation cause it to “halt” or not to “halt,” if you know what I mean. These instances would be the opportunity for a particularly observant, even if scientifically indifferent or detached, designer to intervene.
Do you follow me so far?
(Since you are not averse to “theological” speculations, Evan, you might want to consider the many cases when the Designer might have intervened but didn’t. Could it be that a designer, either a purely mundane, or a truly “Cosmic” one, might choose not to intervene? Can you think of any good reasons why this might be the case? It’s not simply a “theological” question…)
[Insert my usual disclaimer and apology to Mr Moderator for long, digressive, and obtuse postings. Did I ever tell you how much I admire your long-suffering patience and indulgence? LOL]
IP: Logged
|
|
Moderator
Administrator
Member # 1
|
posted 29. March 2002 17:46
Janitor, you made the following comment:
quote:
[Insert my usual disclaimer and apology to Mr Moderator for long, digressive, and obtuse postings. Did I ever tell you how much I admire your long-suffering patience and indulgence? LOL]
As the Moderator I find your posts refreshing and creative and look forward to reading them. I think they add a lot to the discussion here at Brainstorms. Keep up the great work!
IP: Logged
|
|
Evan
Member
Member # 164
|
posted 29. March 2002 17:53
Janitor, I again have trouble follwoing what you are saying.
But you write, "(Since you are not averse to “theological” speculations, Evan, you might want to consider the many cases when the Designer might have intervened but didn’t. Could it be that a designer, either a purely mundane, or a truly “Cosmic” one, might choose not to intervene? Can you think of any good reasons why this might be the case? It’s not simply a “theological” question…) "
I don't think I have made any "theological" speculations. I have specifically said that my hypothesis does not include an omniscient or omnipotent designer..
But if the designer is creative, then, yes there is the problem that there are no rules the designer necessarily has to follow within the constraints of his power, so we don't know, and can't know I think, anything about when the designer doesn't intervene.
I again emphasize that according to Dembski we can only infer design when improbable events happen, and so even individual interventions themselves can't be detected and don't count as design. (And the interventions aren't miraculous in the sense of breaking the flow of cause-and effect.) So I don't think I'm making theological speculations.
IP: Logged
|
|
|
Printer-friendly view of this topic