|
Author
|
Topic: Speciation
|
warren_bergerson
Member
Member # 262
|
posted 30. April 2004 07:31
Claire,
Quote: How can an argument about a new idea have worth, when brainstorming in a forum, and it be all based on logical consistency and be supported
Brainstorming is not unlike following Alice through wonderland. Experts routinely pop up insisting that new ideas satisfy a wide range of logically impossible criteria. My favorite is the requirement that for a new idea to be accepted you must prove that the idea has already been recognized and accepted in the literature. Existing or established ideas, by these same authorities, are not subject to logical requirements. Note the above discussion where the fact of a logical inconsistency was recognized but the existence of the logical consistency was denied because it was not presented in a format someone was willing to accept.
Brainstorming has this Alice in Wonderland appearance because most human beings give a much greater weight to authoritative opinions than they do to either facts or logic. For most people, brainstorming is simply a demonstration of conventional ideas beating up on unconventional ideas by invoking the power of authority.
John,
Quote: It is the only conceivable position for those that refuse to accept the possibility that there has been purpose in a process which culminated with a creature capable of ultimately understanding the process itself. Stephen Jay Gould summarized the Darwinian position perfectly when he arrogantly stated - "Intelligence was an evolutionary accident". For the Darwinian, everything was an evolutionary accident. Don't ever believe it.
I certainly agree with your view that it is important to distinguish the process of evolution from the various Darwinian and neo-Darwinian models of evolution. Evidence that evolution occurred is very different from evidence supporting one of the many versions of Darwinian evolution. I also agree that the concept of purpose is essential to any scientific understanding of evolutionary processes.
It may be somewhat unfair to Darwin to suggest that he rejected the idea of evolution driven by a goal or purpose (a purpose such as survival). I read recently that there is even a current movement to reject the anti-teleological views of the modern synthesis and to return to the Darwinian goal-directed form of evolutionary theory.
IP: Logged
|
|
nosivad
Member
Member # 767
|
posted 30. April 2004 08:09
Darwinism, like Lamarckism, is fundamentally flawed. It was abandoned years ago by some of the greatest minds of the twentieth century.
"No sadder proof can be given by a man of his own littleness than disbelief in great men."
Thomas Carlyle
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 30. April 2004 13:00
warren_bergerson wrote quote:
Experts routinely pop up insisting that new ideas satisfy a wide range of logically impossible criteria. My favorite is the requirement that for a new idea to be accepted you must prove that the idea has already been recognized and accepted in the literature.
I don't recall ever seeing that requirement imposed. Can warren_bergerson provide some actual examples of it, complete with URLs for the examples?
I do recall (since I have frequently done it) asking for real data implying, supporting, or otherwise somehow relevant to many of warren_bergerson's assertions and claims. A critical function of new ideas is to account for observations and data already in the literature, but not once have I seen warren_bergerson supply even a sketch of such an account stemming from any of his various "new ideas."
I also recall pointing out on occasion that warren_bergerson's claims about what researchers in biology and allied disciplines (e.g., evolutionary computing) know and do are at variance with the the facts, and I have repeatedly asked for substantiation of those claims by reference to actual papers describing research in biology or evolutionary computing, again to no avail. In my memory, warren_bergerson has not even once established any kind of link between his claims and assertions, and actual data or research.
RBH
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 30. April 2004 15:39
All I need to do, RBH, not that it’s a simple problem to solve, is enable my creatures with complex problem solving capabilities and I’ve insured their indefinite evolution. I need to provision, or enable, pre-adapt, them. Design them to be adaptable and evolvable.
Intelligent designers design intelligent designs, and by that I don’t mean they simply impress us as particularly “artful” designs. I mean the designer imbues them with intelligence. They emulate the designer. The “ideal design” is exactly that, the very object of AI and machine learning.
But that’s absurd on the dumbed-down “Darwinian” paradigm. Life forms don’t evolve by learning how to evolve. (And they certainly weren’t designed to learn to evolve. Or even to learn anything.) They evolve by sheer dumb luck.
In Neo-Darwinism a logically and empirically unsupportable theoretical distinction is made between “learning” and “evolving.” Thus we have a marked dichotomy between the cognitive-behavioral sciences, often characterized as “Lamarckian,” and those sciences that are characterized as “Darwinian.” (Whatever they are. Seems to be more the view of physicists; "universal Darwinism" and all that). Plainly, Darwin made no such distinction. He was as “Lamarckian,” if not more so, as he was “Darwinian.”
The two are only consistently synthesized, in the post-Neo-Darwinian era, in the applied sciences, such as evolutionary computation. (See how Henry Plotkin’s views have “evolved” over the years.)
Effective problem solving techniques often exploit proven techniques (and it hardly needs to be stated). Biologists often point to the fact that genomes, especially in eukaryotes, are composed mostly of “partial solutions,” which they traditionally interpret as relics or failures. Relic or not, what they plainly represent is a consistent use of design problem-solving techniques of patterning and reuse. We call this “tinkering,” “dithering,” “refining,” or whatever. The terms can hardly trivialize what is being done is complex problem solving.
In euks especially we observe an extensive and quite sophisticated rewrite or editing system. (The expression profile of the human genome is said to be 95-98% regulatory, i.e. controlling expression of constitutive genes. Fits nicely with my thesis that adaptation—and design—are fundamentally issues of control, i.e., “teleology.”)This is also why their genomes appear to be so “messy.”
Of course, life forms don’t have to satisfy a “Darwinian” evolutionary aesthetic. All they have to do is work, and any solution that works is to be (and is) preferred to an “elegant solution” that doesn’t work. (No such alternative solutions are proposed in evolutionary biology, except by reference to "how God would do it.")
Life is messy, grubby, gooey, oozy, smelly, and it aint’ pretty. But it works. (Actually I think life is beautiful and all the more so, because its intelligent. And I would say “Darwinism” is ugly and stupid, but I expect Mr. Moderator would rebuke me for saying so.)
The “messiness” (the complications) is why biologists, to borrow from Copernicus, can’t compute the length of the day. They cannot agree upon the variability of the genome or the patterns that are detected in that variability, e.g., and more importantly what are the causes and how to weight them.
One factor (not that it is “one”) that characterizes complex problems is a latent semantics (sets of related solutions) that must be discovered in a highly-dimensioned problem (design) space. Intelligent problem-solvers often exploit effective strategies for exploring or searching this space, such as indexing, partitioning, decomposing, classifying, etc.
It has been shown [!] that the DNA-code implements such an effective search strategy. (And it has been shown that reduced codes do not work nearly as effectively. If at all.)
The idea that I have to “front-load” my creatures with all the solutions (presuming a knowledge of all the problems), all that information, is certainly not realistic, as RBH indicates. So, now that we’ve dispensed with the strawman maybe we can consider something a little more realistic?
(There are two common uses of the term “latent:” those factors are latent that are eliminable because they are understood to be “artifacts” introduced in the modeling process; and those factors that are unknown, but are not necessarily even presumed to exist. The latter is just the strategy of “cover your ass.” Given the biologists failure to “compute the length of the day,” I will assume that such factors exist. I do not presume however that they are not known. I only presume that existing theory fails to identify them… for what they are. My use of the term "latent" sets it equal to that information that remains to be discovered. It is realized there, in the genome or the mind, as it is discovered. That information exists, but not necessarily in the genome. More importantly, to me, is that it does not exist, i.e., exists unrecognized in the minds of the principal observers.)
Evolution is just that process of discovery.
Traditional theory has considered only that discovery is an “accident.” But I prefer the “Kuhnian” view: discovery is intimately related to invention. (Also an evolutionary process.)
On a certain theological (and biological) view God has no “problems.”
My Mother is as close as I can get physically to “God,” and as long as I have known her (LOL) she has done at least one crossword puzzle a day. It strikes me as a pointless exercise (especially as I am so bad at it—she solves all of them). But upon reflection it is a very interesting problem (a crossword puzzle) to solve. Nevertheless all my problems have been “of the sort.”
Claude Shannon noted how the particular characteristics of (the English) language restricted the set of soluble (crossword) problems. I.e., there are some problems which cannot be solved within the conventions of the language—which was also the thesis of Godel. (Maybe its because Shannon, practically minded engineer that he was, didn’t further reflect on that fact and so wasn’t driven “mad” like Godel? My impression is that all “pure” mathematicians are a little crazy. Look at Dembski! Guy’s obviously a looney!)
To me it is endlessly intriguing that there is a universal genetic code (language). There are many adaptations that are shared, but there are no others that are universal. Let me put it plainly: there is a particular “language” that has, supposedly, in ~4-billion years, solved every “crossword puzzle.”
We are pretty smart (adaptable) at communicating. But we have never discovered any code or language that does this.
Here’s a test: Discover a code or language that does it better.
Irreducible code?
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 30. April 2004 19:06
J@MIT wrote quote:
All I need to do, RBH, not that it's a simple problem to solve, is enable my creatures with complex problem solving capabilities and I've insured their indefinite evolution. I need to provision, or enable, pre-adapt, them. Design them to be adaptable and evolvable.
I'm not so sure they need to be provided with complex problem solving capabilities. Pretty simple stuff goes a long way.
I agree that the distinction between learning and evolving is artificial - I like to think of them in terms of memory representations of various kinds and durabilities. Genetic memory is a lot more durable than the particular set of memories stored in my dog's skull, but the palimpsest that is DNA is not all that far removed from the overlaid representations in a brain that are accumulated during the lifetime of an individual, those representations themselves being overlaid on the DNA palimpsest.
RBH
[ 30. April 2004, 19:08: Message edited by: RBH ]
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 01. May 2004 08:12
It should be noted that no one has actually offered a resolution to the major inconsistency between the genotype to phenotype mapping assumed in neo-Darwinian theory and in GA demonstrations and the mapping assumptions supported by observation of developmental processes. Efforts to shift the burden of proof or avoid addressing the issue should be recognizable as debating tactics.
Of more general importance is the fact there non-Darwinian explanations which can reconcile the 1)deterministic nature of inherited traits and 2)the observed complexity of developmental processes.
The non-Darwinian reconciliation starts by noting that self-development or self-assembly processes can be modeled and simulated by complex goal directed programs that rely on environmental input as well starting point codes. These programs suggest that developmental processes depend upon or are determined by three types of factors: 1) genetic materials, 2) developmental programs and 3)environmental input. This is very different than the neo-Darwinian assumption of dependence only on genetic codes.
If you start with developmental processes defined in terms of genes, goal directed developmental programs and environmental input, then you can model evolutionary change in terms of goal directed programs that modify both genetic materials and goal directed developmental programs using environmental input. These ‘goal directed’ evolutionary programs involve processes well beyond mutation and natural selection.
I am not personally an expert on self-assembly programs and adaptive or evolutionary programs that operate to change or evolve self-assembly programs. I am, however, aware of the existence of this general type of program. If you prefer, I offer it as a prediction that such programs provide a resolution to the ‘observed deterministic heritability’ and observed goal directed nature of developmental processes.
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 01. May 2004 13:11
I'm sorry, Dr. Davison, your topic was closed, but I have a question (which I believe you can help me with), but I'm not sure its exactly topical (although it may be), so with warren_bergerson's kind indulgence I'll ask it: What is the "hotspot paradox"? Or is there such a thing?
(LOL It's not as if I've had anything topical to contribute so far.)
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 02. May 2004 06:29
Warren, most treatments of evolution are in terms of "selective advantage", which is an average over the entire population. Whether this is lowered because of incomplete penetrance (i.e. manifestation of many-to-many mappings) or some other reason, the math still works out the same way. You may have a less accurate estimate of selective pressure, in which case your estimate of fixation probability and such varies more. It doesn't completely overturn all methods that are used.
So I suspect that the reason nobody has bothered to respond to the "inconsistency" is because it really isn't a problem.
As far as I can tell, the essential features of the genotype-to-phenotype mapping are adequately captured in the theory. If you think otherwise, it would be helpful if you could explain how and why in detail--or refer to some published source that talks about the problem--rather than continuing to refer to a simplification as an inconsistency.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 02. May 2004 12:50
Rex,
`It seems contrary to elementary logic to deny that the difference between static one or one mappings and dynamic many to many mappings constitutes an apparent inconsistency. The existence of a major apparent logical inconsistency would seem to require either 1)explicit demonstrations that the apparent inconsistency can be resolved, or 2)an admission that we currently do not have sufficient knowledge to either reconcile the inconsistency or to demonstrate that the inconsistency can not be resolved. I am under the impression the current view in biology is that there is insufficient knowledge to resolve the apparent inconsistency.
I am not only suggesting the existence of an apparent logical inconsistency, but I am also proposing a resolution of the issue. I am proposing that both developmental processes and evolutionary change processes be viewed as ‘behaviors under the control of goal directed intelligence’. This approach, I suggest 1)replaces the genetic determinism assumption with the intelligent teleological determinism assumption and 2)eliminates the apparent logical inconsistency created by the genetic determinism assumption.
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 02. May 2004 15:59
But you have supplied a "resolution" to a problem that you have not shown exists, with less detail than current resolutions. This isn't very helpful. It is slightly more helpful than a flat statement that "Pixies did it" without further explanation, but in its current state, only slightly.
Can you please try to apply your hypothesis to a concrete example problem? Key quantities relevant to evolution include effective population size, selective advantage, and mutation rate, and from these things one can at least predict the rates of appearance and disappearance and fixation of point-mutant alleles. What can your theory offer?
Note: I want example (a simple one is fine) that is worked out in detail. Not repeated statements of "it seems natural to assume that" and "it should not be difficult to", because your theory does not seem natural to me, and I am baffled by how to do the things that you claim shouldn't be difficult.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 03. May 2004 07:24
Rex,
Just as you seem to be having a hard time visualizing the approach I am proposing, I am having some difficulty in understanding why you can’t see the distortions created by an overly simplistic, biologically inaccurate genotype to phenotype mapping assumption.
I think possibly the explanation lies with the difference between 1) after the fact descriptive models and 2) predictive models and hypotheses. If you looking after the fact at an evolutionary change that involves both genetic changes and phenotype changes, then you can always fit the data to model using a simplified genotype to phenotype mapping. In other words, using a simplified genotype to phenotype mapping assumption, you can always make the after the fact claim that the data fits your model.
However, overly simplistic mapping assumptions are problematic if you are attempting to create predictive models that are biologically realistic. To illustrate this point, it might be useful to start by looking developmental processes. The simplistic mapping assumption suggests that development processes can be modeled in the form: F(genetic code)= phenotype
The concepts and techniques used to create automated assembly processes provide a more complex and biologically realistic models of the genotype to phenotype mapping. Such a model would include components such as: Assume the self assembly process consists of 1) assembly instructions (equivalent to a genetic code ), 2)an assembly program to implement the instructions, 3)feedback processes or mechanisms, 4)the physical environment in which the assembly process operates, and 5)physical materials used in the assembly. [Note that automated information processing systems used a wide range of business applications have this same basic logical structure. My personal experience is with such business systems, not with automated assembly processes, but the logic and programming concepts are, I believe, similar.]
Both simplistic mapping models and the automated assembly models can be used to model evolutionary change processes. The two approaches are logically equivalent if the only thing that produces evolutionary change is changes in assembly instructions or genetic codes. However, the automated assembly model suggests that it might be possible to produce evolutionary change by modifying or evolving non-genetic factors such as processing programs, feedback mechanisms, or the assembly environment.
If, in fact, evolutionary change involves changes in factors other than genetic code, then the simplistic genetic mapping assumption produces an unrealistic or distorted model of evolutionary change.
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 03. May 2004 13:44
I thought evolution was about the "fittest." When did it become about the "averagist"?
The map G->p->G is the most significant outstanding general problem in evolutionary biology. That is evolutionary biology! Coming to an understanding of the "->" part.
The idea that it is largely solved is... what?!
[No takers on "hotspots"? If recombination intiation loci are systematically degraded upon each round (by conversion), then it could be because they are very highly "movable." (Is there much information upon RNA-mediated transposable elements that are native to the sequence? Just curious.) In which case, they may be missed (moving) over experimental windows (?).
Maybe genomes evolve more like immune systems?]
[ 03. May 2004, 13:54: Message edited by: Janitor@MIT ]
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 03. May 2004 21:23
Well, I'm now confused about whether Warren is proposing the "radical" change that evolutionary biologists do exactly what it is that they are already doing: trying to work out the mechanisms of genetic change.
That makes sense and is being done. It's the bits about "processing programs" and such that seem almost completely unconstrained (and therefore not very useful). Biologists are looking for such things all the time. But until you've found one (or found evidence of one), what can you do?
Of course you introduce distortions when using a simplified model. And most modern evolutionary biologists are aware of that. That's one reason why there are still lots of open questions.
[ 03. May 2004, 21:33: Message edited by: Rex Kerr ]
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 04. May 2004 07:25
The issue being discussed is the logical inconsistency between a simple one to one type mapping assumption and the observable, developmental process, relationship between genotype and phenotype. If we recognize that developmental processes (the transformation from genotype to phenotype) can be modeled and simulated by self-assembly type models, then we recognize that the genotype to phenotype transformation depends upon or is determined by factors other than genotype. Of particular importance, development depends on environmental conditions, processing programs, and environmental input in addition to genetic code.
Developmental processes are not logically consistent with a strict or narrow genetic determinism assumption. What remains might be labeled a genetic influence assumption. What specific assumption are biologists using to replace the strict genetic determinism assumption? What implications does a genetic influence assumption have for neo-Darwinian theory? Do you or even can you have a predictive scientific theory of evolution using the genetic influence assumption?
I am proposing that the genetic determinism assumption be abandoned because it is not compatible with observations of developmental processes. I am proposing that as an alterative we view developmental processes and evolutionary processes as ‘behaviors controlled by goal directed intelligence’. If you prefer, we can view developmental processes and evolutionary processes as having the appearance of behaviors controlled by goal directed or purposeful intelligence’. Viewed from this perspective, I suggest, we can construct predictive hypotheses using the intelligent teleological determinism assumption.
As numerous posters remind us, modern biology has made numerous modifications to the original neo-Darwinian assumptions and theories. Modifications that are necessary because the observed facts are not consistent with the original assumptions and theories. The question is whether there is enough substance left in the original neo-Darwinian approach to actually formulate predictive hypotheses. Or is what remains of the neo-Darwinian approach only sufficient to permit the development of after the fact descriptive hypotheses?
It is somewhat interesting to note that the original Darwinian approach was based on the concept of heritability rather than the concept of genetic determinism. The approach I propose, while it is not compatible with neo-Darwinism and the genetic determinism assumption, does appear to be compatible with the Darwinian concept of heritability. In fact, if you replace the natural selection concept with a more general concept of selection, then the approach I advocate is compatible with the general Darwinian approach. But that is a subject for another thread.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 04. May 2004 07:40
Quote Rex: It's the bits about "processing programs" and such that seem almost completely unconstrained (and therefore not very useful). Biologists are looking for such things all the time. But until you've found one (or found evidence of one), what can you do?
Processing programs are equivalent to complex causal processes. As with all causal processes, processing programs are not directly observable. We infer processing programs from observed behavior. In addition, we can create systems that operate on specified processing programs and we demonstrate that the behavior of the observed system parallels the behavior of the created system.
We do not need to know the actual processes or mechanisms involved in order create models and simulations of processing programs. Models and hypotheses of processing programs are generally based on observations of behavior. Such models are tested or validated based on their ability to predict behavior.
IP: Logged
|
|
|
Topic Closed
Printer-friendly view of this topic
