Mark Iosim
Posts: 27 Joined: Oct. 2007
|
I am back to this lovely discussion forum to report the results of my homework regarding Genetic Algorithm (GA). A few weeks ago I asked to refer me the statistical analysis demonstrating that random mutations are sufficient to cause adaptive changes in biological systems. Respond came from oldmanintheskydidntdoit who suggested to start with http://www.talkorigins.org/faqs/fitness/. In my following skeptical assessment of “Weasel program” I stated that it cannot be a model for a Natural selection, but for an artificial selection only. Later in Wikipedia (“Weasel program”) I found the same assessment made by Dawkins “…Shakespeare model is useful for explaining the distinction between single-step selection and cumulative selection, it is misleading in important ways. One of these is that, in each generation of selective 'breeding', the mutant 'progeny' phrases were judged according to the criterion of resemblance to a distant ideal target…”. I agree with Dawkins’ assestment of “Weasel program”, but I would defend the role of “distant ideal target” in the natural evolution process. I will return to this point later in this post.
Responding to my skeptical assessment I was told that the “weasel program” is just a tutorial example that demonstrates a difference between random changes and accumulative selection, but to understand mechanism of RM+NS I should make myself more familiar with Genetic Algorithm (GA). I followed this advice and spent some time learning about GA. Below is a couple definitions that in my opinion accurately define GA:
[qoute]http://en.wikipedia.org/wiki/Genetic_algorithm “A genetic algorithm (GA) is a search technique used in computing to find exact or approximate solutions to optimization and search problems. …GAs cannot effectively solve problems in which the only fitness measure is RIGHT/WRONG, as there is no way to converge on the solution. (No hill to climb.) In these cases, a random search may find a solution as quickly as a GA.”
http://www.talkorigins.org/faqs/genalg/genalg.html# … a genetic algorithm (or GA for short) is a programming technique that mimics biological evolution as a problem-solving strategy. Given a specific problem to solve, the input to the GA is a SET OF POTENTIAL SOLUTIONS to that problem, encoded in some fashion…[/qoute] By reading and thinking about GA over two weeks I think I got enough basic understanding to appreciate a power and limitation of GA to respond to the following challenges addressed to me in the past:
swbarnes2 Quote | How is this is a problem (threshold of usefulness) for a GA? Enough random starts and you will pass any threshold you like. | By “threshold of usefulness” I mean a minimum performance of evolved “digital organism” that is recognized by the fitness function as a potential solution. The small mutation in the evolving “digital organism” into direction to solution need to be detected and promoted by Fitness Function, otherwise prohibited amount of simultaneous mutation must occur in order to produce a “jump” to a better solution. If this explanation still doesn’t make sense to you try this: Quote | http://www.inf.ed.ac.uk/publications/thesis/online/IM050329.pdf “… What we do not always take into account, however, is whether evolution can provide a logical means of finding that peak (solution) through the gradual process required by an EA. If there is no “incentive” to evolve through the earlier stages required, as in the case of 2% of a wing, our evolution will get nowhere. |
swbarnes2 Quote | The hard fact is that GA's have and do succeed in solving problems that humans don't know the answers too. |
GA succeeds in solving problems the same way as hand held calculator does. They are both just a tools in our hands. You probably don't know the answers for x = 23.7E20, but a calculator “does”.
oldmanintheskydidntdoit Quote | “What do you make of http://ic.arc.nasa.gov/projects/esg/research/antenna.htm The fitness function used to evaluate antennas is a function of the voltage standing wave ratio (VSWR) and gain values on the transmit and receive frequencies. VSWR is a way to quantify reflected-wave interference, and thus the amount of impedance mismatch at the junction. VSWR is the ratio between the highest voltage and the lowest voltage in the signal envelope along a transmission line.
The two best antennas found, one (ST5-3-10) from a GA that allowed branching and one (ST5-4W-03) from a GA that did not, were fabricated and tested. Antenna ST5-3-10 is a requirements-compliant antenna that was built and tested on an antenna test range. While it is slightly difficult to manufacture without the aid of automated wire-forming and soldering machines, it has a number of benefits as compared to the conventionally-designed antenna. |
This is a typical application for GA, as an optimization TOOL by which potential solutions (changing geometry of wires) were RECOGNIZED as less or more effective by fitness function. Following landscape of more effective solutions the program eventually determines one of the optimum geometry of wire antenna. See a visual illustration for a similar process: http://www.obitko.com/tutorials/genetic-algorithms/search-space.php Wesley R. Elsberry Quote | Artificial life systems, such as Avida, can be configured such that the digital organisms contain -- and make modifiable -- the code that performs the self-replication process, making them an instance of evolution, not a simulation of evolution. Given an ancestral digital organism capable of reproduction and nothing else, Avida provides an experimental platform to do precisely what last page's rant said was the issue: examine the process of adaptive change by means of selection. And they do adapt. | Wesley provided me with the link to the article “The Evolutionary Origin of Complex Features” published in NATURE in 2003. This topic also featured in the Discovery (and probably other) magazine. It took for me a while to understand what this work about, but eventually I learned a few important things I would like to share with the rest of the folks. Quote | http://myxo.css.msu.edu/lenski/pdf/2003,%20Nature,%20Lenski%20et%20al.pdf Abstract A long-standing challenge to evolutionary theory has been whether it can explain the origin of complex organismal features. We examined this issue using digital organisms—computer programs that self-replicate, mutate, compete and evolve. …These findings show how complex functions can originate by random mutation and natural selection. |
It is indeed a long-standing challenge to evolutionary theory, what Behe refers as Irreducible Complexity. My first impression was that this article addressed and solved this challenge using rigorous scientific method, but it was a wrong impression. Instead the article is flooded with phenomenological details that in my opinion do not help to justify its title.
By using digital organisms of Avida program, the authors traced the genealogy from an ancestor that could replicate only to descendants able to perform multiple logic functions requiring the coordinated execution of many genomic instructions. To demonstrate that complex systems evolve from simpler precursors authors set up small rewards for simpler operations and bigger rewards for more complex ones and this way provide an “incentive” to evolve through the gradual process. However when the researchers took away rewards for simpler operations, “digital organisms” never found a final solution. However to come to this conclusion we do not need the sophistication of artificial intelligence”, because the “Weasel program” demonstrates the same result: if the small improvements in the “Hamlet line” would not be rewarded the final line from Hamlet never evolve.
Conclusion of this article: Quote | Our experiments demonstrate the validity of the hypothesis, first articulated by Darwin and supported today by comparative and experimental evidences that complex features generally evolve by modifying existing structures and functions. Some readers might suggest that we ‘stacked the deck’ by studying the evolution of a complex feature that could be built on simpler functions that were also useful. However, that is precisely what evolutionary theory requires, and indeed, our experiments showed that the complex feature never evolved when simpler functions were not rewarded. |
How does this conclusion addresses the biggest challenge of Darwinism – the phenomenon of Irreducible Complexity, by which a “simpler function” is useless? Apparently this article ignores not Irreducible Complexity only, but any analysis of complexity at all. This article, pretending to explane the emergence of complexity from simplicity, avoids defining which of the dozens different and often mutually exclusive definitions of complexity authors have in mind. This is very unusual, especially for scientists from Department of Computer Science and Department of Philosophy where scientific concept of Complexity is the “bread and butter. The conclusion is also a classical example for the circular logic between RC+NS incorporated into Avida program that bechave like RC+NS and this way this proves the validity of RC+NS.
Another pearl from Discovery magazine: Quote | http://discovermagazine.com/2005....t:int=1 “…When the Avida team published their first results on the evolution of complexity in 2003, they were inundated with e-mails from creationists. Their work hit a nerve in the antievolution movement and hit it hard. A popular claim of creationists is that life shows signs of intelligent design, especially in its complexity. They argue that complex things could have never evolved, because they don’t work unless all their parts are in place. But as Adami points out, if creationists were right, then Avida wouldn’t be able to produce complex digital organisms. … “What we show is that there are irreducibly complex things and they can evolve, says Adami… |
Holly Molly!!! So the article in Nature indeed addressed and solved the problem of Irreducible Complexity, I just missed it. Try to read this article for you self and may be you will be more successful reconciling the mutually exclusive statements from Nature and Discovery.
I have been wondering if Neo-Darwinism is theory, hypothesis, paradigm or religion. But this article makes me think that Darwinism is an ideology of Darwinian political party. I am not ID supporter, but after comparing the statements from Nature and Discovery, I want to join an opposition party to expose the foolishness, blindness and dishonesty of some “Darwinians”. Unfortunately ID party wouldn’t tolerate me either. If anybody knows about existence of Independent party in Evolution Biology, please let me know.
After being disillusioned with the sophisticated GA, I came back to “Weasel program” that in my opinion is not a tutorial example, but one of the simplest and the most transparent tool that could demonstrate the power of accumulative selection and limitation of random mutation as good as any of its more sophisticated GA cousins. The “Weasel program” was criticized by ID supporters (and this critic was accepted by Dawkins) because of the “final target sequence chosen in advance” (the exact line from Hamlet) is seen as a weakness because it could be interpreted in favor to ID. I think that the “target sequence” isn’t a weakness of this model, but one of the “incarnations” of the real fitness function. The phenomenon of mimicry or camouflage could be an example of living system that matches the target “existing in advance”. The complexity of the “natural targets” is often exceeding the complexity of “line from Hamlet”. The evolution driven by “warfare” with other organisms presents example of this sophisticated “fitness function”. I am defending the practical usefulness of the “Weasel program” and against tutorial example only, because I plan to put it to work helping me solve one problem.
The gradual, step-by-step changes are the most important concepts without each Darwinism wouldn’t be able to explain evolution. The main argument of Behe against gradualism is that it is impossible to define existence of the “appropriate fitness function” that would provide gradual evolution of the so called “Irreducible complex” systems. Dawkins’s counterargument is that regardless that we can’t reproduce these conditions now, it doesn’t mean they couldn’t exist millions or billions years ago. Dawkins expects that sooner or later the “appropriate fitness function” will be found and he has tried to demonstrate that this is not an impossible proposition.
I don’t share Dawkins’ optimism, but I wouldn’t waste time attempting to prove the nonexistence of these conditions billions years ago. How one can prove a non-existence of any thing at all, including the non-existence of God? We may prove a non-existence only in the absolutely defined area of knowledge. For example I may prove to my self the non-existence of a wallet in my empty packet, but only after thorough searching it and even then I may have some reservations.
Therefore I wish Dr. Dawkins a lot of lock in the reproducing of what may happen on Earth million a million years ago, but until these evidences are not discovered, the Neo-Darwinism, in its current form, do not deserve to be called Theory, but a controversial Hypothesis instead.
While evolutionists are working hard stretching their imagination about what may happen millions years ago, I would like to try a different approach by looking in what is happening now, before our eyes. The “Weasel program” (or any other GA tool) that already helped us to rule out the single step selection as unrealistic scenario and shown that an accumulative multi-step model is a much promised one, could help us again. I would like to use “Weasel program” to evaluate RM+NS mechanism in the development of drug resistance in bacteria. For example, in the process of developing drug resistance the particular segment of bacteria’s’ (or virus’) DNA mutates from form A to form B during N generation per mutation rate X etc. Using “Weasel program” (by replacing letters with nucleotides) we can modeling the DNA evolution from form A to form B and see if the model able to reproduce a “target line of nucleotides” within reasonable time frame. I expected to find plenty of published experiments that spell out the changes in bacteria (virus) DNA segment during drug resistance development. However I am having problem to locate these publications. Can anybody help me in this search?
|