|
Author
|
Topic: Royal Truman: Avida, a biologically unrealistic model for neo-Darwinian Theory
|
John Bracht
Member
Member # 5
|
posted 21. April 2004 11:59
I argued that Lenski et al's simulation is biologically unrealistic, and RBH responded by approvingly quoting Charlie d. as saying, in part,
quote:
Again: this is not meant as a simulation of biological evolution in any way, shape or form, but as a proof of principle that IC systems can evolve spontaneously through evolutionary processes.
Keep this in mind: Lenski et al were not simulating biological evolution in any way, shape, or form". Yet we get RBH's comment below (from the same post!)
quote:
Lenski, et al., showed that structures that meet Behe's definition of irreducible complexity can readily evolve in avida in a constrained context employing just a subset of the evolutionary mechanisms known to operate in biological systems.
But cannot anyone see the logical disconnect here? Because of the way "Lenski evolution" works, it is completely irrelevant to Behe's claims. Charlie d. even says that the types of mutations, selection pressures, etc., do not really matter, because the model is not trying to simulate reality. Yet RBH and other critics are claiming this as some sort of refutation of Behe's claims about IC. The whole point about IC is that it is a problem for biological evolution--not for some idealized, weird "Lenski evolution" that has no relevance to reality (as admitted and proclaimed by the design critics themselves!).
What I was trying to do earlier with my comments about swapping entire flagellar parts out (hook, rod, filament, etc) was to imagine what biological evolution, Lenski style, might look like. What if we make biological evolution operate analogously to Lenski's evolution? First off, we'd have high-level functional components that are pre-defined and ready to interact. All necessary components to build a flagellum would be pre-made and stored somehwhere in such a way that they could be swapped into bacterial genomes by "mutations". No "mutations" could ever disrupt the individual components--they would only swap in or out individual components (thereby sometimes generating nonfunctional composite structures, but out of fully functional components). Under these conditions, would anyone doubt that this odd sort of "evolution" could produce irreducible complexity? If evolution operated this way, Behe would never have made his claims. As things now stand, to refute Behe you need to address the systems he was referring to. It just so happens that he was referring to REAL bioloigcal evolution, which does, in fact have serious problems building irreducible complexity. Critics who want to use computer simulations need to ensure that they capture the essence of biological evolution in order to claim they've made a relevant simulation. Just the fact that a simulation incorporates heritable change and selection is not enough to ensure that relevant conclusions about the evolvability of IC systems can be made.
Given what has been said on this thread thus far, the critics need to start using a lot more qualifiers when they make claims about Lenski et al's simulation. Stop referring to it as some sort of refutation to Behe. It's disingenuous to claim that Lenski et al showed evolution can build IC structures--because to the layperson, your useage of the term evolution seems to be referring to biological evolution, not some sort of abstract, completely irrelevant "Lenski evolution". So, from now on, please refrain from making such misleading comments. RBH, your comment was
quote:
Lenski, et al., showed that structures that meet Behe's definition of irreducible complexity can readily evolve in avida in a constrained context employing just a subset of the evolutionary mechanisms known to operate in biological systems.
Behe's definition of IC is based on a biological idea of evolution, not on "Lenski evolution" so it's misleading to claim that it can "readily evolve". Furthermore, Lenski et al used very different mechanisms than biological evolution (eg, complex function swapping instead of mutations at the level of the DNA). Hence this claim is entirely misleading. In the future, I hope to see more honesty about the Lenski simulation from the critics, and less obfuscation and equivocation of terms.
Argon--I didn't mean to imply that genetic mutations are always single base changes. Obviously, I'm a biologist, so I know that. But it's inescapable that evolution involves changes at much lower levels than the Lenski et al simulation. Evolution clearly doesn't start with entirely pre-defined sets of functional components which are entirely protected from mutation and which get swapped in and out of genomes. Rather, those functional units have to be built up from somewhere (whether they get co-opted or not), before they can be swapped around within a cell by various types of genetic recombination. This point goes to RBH's claim that the components of the flagellum have functional homologs. Some do, but many components are entirely unique (for example, the proteins that make up the hook, rod, and filament are unique). Furthermore, the main homology is to the type III secretory system which is widely believed to have evolved from the flagellum, not the other way round. So for the vast majority of the flagellum, there is no viable protein to be co-opted into the structure. In any case, evolution didn't start with the components of the flagellum sitting in a bin someplace, and just start swapping them into and out of bacterial genomes (which is what the Lenski et al simulation does)--it had to first build the units themselves. Co-option doesn't accurately model what the Lenski simulation does, because co-option requires that the components be pre-existing within the cell; the Lenski simulation has the components defined externally and it pops them into and out of the simulated critters from "nowhere" (relative to the critters).
John
[ 21. April 2004, 12:05: Message edited by: John Bracht ]
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 21. April 2004 13:37
Bracht wrote quote:
But cannot anyone see the logical disconnect here? Because of the way "Lenski evolution" works, it is completely irrelevant to Behe's claims. Charlie d. even says that the types of mutations, selection pressures, etc., do not really matter, because the model is not trying to simulate reality. Yet RBH and other critics are claiming this as some sort of refutation of Behe's claims about IC. The whole point about IC is that it is a problem for biological evolution --not for some idealized, weird "Lenski evolution" that has no relevance to reality (as admitted and proclaimed by the design critics themselves!). (Emphasis original)
Then what on earth does Behe think he's doing by waving his bloody mousetrap around? His iconic example of an IC structure is not even a biological one! Nor do his several formal definitions of IC say anything specific about just biology: they are couched in generic terms about generic systems, not in specifically biological terms.
avida is not a simulation of a specifically biological system; it is an evolutionary system that shares some properties and mechanisms (populations of entities with imperfect replication, mutations to the heritable information-bearing representations, non-uniform selective environment, and competition for limited resources) with biological systems. Biologists have offered evolutionary accounts of allegedly IC biological structures and processes involving indirect pathways - gene duplication and cooption and scaffolding being processes that enable those indirect routes to exist. The Lenski, et al., paper shows that in another evolutionary system, the avida system, those indirect routes can indeed produce IC structures in a non-biological system that employs a subset of generic evolutionary mechanisms.
charlie d's response said nothing about "reality." The constrained avida world is just as "real" as the biological world. charlie d said that: quote:
Again: this is not meant as a simulation of biological evolution in any way, shape or form, but as a proof of principle that IC systems can evolve spontaneously through evolutionary processes. (Emphasis added)
If IDists can ground their case on the promiscuous use of a vague and informal analogy between the human design of artifacts and the putative "intelligent design" of biological structures and processes, why on earth can one not employ the much more detailed formal correspondence between the avida evolutionary system and biological evolutionary systems? After all, we can clearly identify specific mechanisms that are common to both avida and biology (see list above) and we can actually investigate them quantitatively, using the results from avida investigations to inform research in biological systems. That is a whole heck of a lot more than IDists have ever done with the ill-defined human design analogy.
RBH
[ 21. April 2004, 13:52: Message edited by: RBH ]
IP: Logged
|
|
Royal
Member
Member # 1060
|
posted 05. May 2004 16:39
What are Avida's "Instructions" and "Logic Functions"?
I agree with John Bracht ( contra RBH et. al. claims), that we need to establish whether Avida has any real world biological relevance.
Claims are being made in this discussion that IC has been disproved and evolution of complex new functions via intermediate stepping-stones established. I intend later to demonstrate why RBH's reported runs do not demonstrate what he claims (and I have a large number of my own runs to demonstrate in detail where his reasoning has gone astray). Therefor, let me justify my view that logic functions are meant to convey the notion of genes which provide the necessary information for biochemical processes.
Let the authors speak for themselves:
quote:
(1) http://arxiv.org/abs/quant-ph/0301075
Selective Pressures on Genomes in Molecular Evolution; C. Ofria, C. Adami, T.C. Collier, J. Theor. Biol. 222 (2003) 477-483.
Computations can be carried out by evolved programs if they develop sequences of code (“ genes ”) that perform logical, bitwise, computations on random numbers provided in their environment. Such genes evolve spontaneously if the performance is rewarded with bonus CPU cycles (their unit of energy). The complexity of the environment can be controlled by changing the number of logical operations whose performance is rewarded.
quote:
The development of the necessary code to perform such computations is, in the digital world, the analog of the evolution of a sequence or gene enabling the organism to catalyze exothermic reactions leading to faster replication.
quote:
(2) http://arxiv.org/abs/physics/0209081
Ab Initio Modeling of Ecosystems with Artificial Life; C. Adami, Natural Resource Modeling, 15 (2002) 133-146.
On top of that, CPU “bonus" time is given out for those programs that have developed computational genes. Such genes are stretches of program which read numbers from the environment, perform computations on them, and write them back out.
quote:
Because it is these computations which provide the organism with the “energy" (in the form of CPU time) it needs to replicate, we can think of this computational code as the genes that code for the organism's metabolism. To this extent, we are able to observe the emergence of metabolic genes in self-replicating organisms, and thus the evolution of complexity.
quote:
In a sense, these genes can be viewed as the analog of exothermal catalytic reactions that are being carried out by the organisms, as they allow a more “efficient" exploitation of the primary resource, CPU time.
I repeat an ealier observation: one can always design a computer program which generates an intended outcome. I did not provide nor discuss my own Avida runs in the original essay. This will be corrected.
Let us first agree, that the authors are claiming at least conceptual relevance to real genes, and real metabolic processes. Can we establish this as a minimal basis to continue?
IP: Logged
|
|
Royal
Member
Member # 1060
|
posted 05. May 2004 17:06
Avida runs: Relevance to Irreducible Complexity (IC)?
In several posts the claim was made that Avida runs disprove Mike Behe's concept of IC. Some of us, including John Bracht, pointed out that the probabilistic distance between logic functions is, biologically speaking, absurdly small. My computers have been working for weeks to quantify this rather obvious fact.
Now, the typical settings in the environment.cfg file are:
REACTION NOT not process:value=1.0:type=pow
REACTION NAND nand process:value=1.0:type=pow
REACTION AND and process:value=2.0:type=pow
REACTION ORN orn process:value=2.0:type=pow
REACTION OR or process:value=3.0:type=pow
REACTION ANDN andn process:value=3.0:type=pow
...
The simplest logic functions (NOT, NAND) do provide helpful infrastructure AND, ORN can build upon.
Let me point out something. In the usual parameter settings, lineages will have no problem at all developing e.g. AND, ORN without an intermediate rewarded NOT, NAND. Simply comment these out in the above file. The runs merely take a little longer.
Exactly why are some of you alleging these logic functions are by defintion IC? What are the necessary underlying functions / components(an instruction is not a function!) without which the complex logic function no longer works?
In Behe's examples multiple proteins are necessary concurrently for the described cellular process. You can't remove one component at the prcess still works. What exactly are you claiming can't be removed from the intermdiately complex logic functions?
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 05. May 2004 19:20
Truman wrote quote:
In several posts the claim was made that Avida runs disprove Mike Behe's concept of IC.
Nope. As far as I can see no one has made that claim. The claim is that structures - code sequences - that meet Behe's original definition of "irreducible complexity" can relatively easily evolve under specific selective and entironmental conditions using a minimal subset of the naturalistic mechanisms posited by evolutionary theory. There is no claim that Behe's concept of IC was "disproved," just the conclusion IDists draw from it, namely that IC structures cannot be produced by evolution.
Truman asked quote:
Exactly why are some of you alleging these logic functions are by defintion IC? What are the necessary underlying functions / components(an instruction is not a function!) without which the complex logic function no longer works?
This is simply incoherent. Truman is conflating functions on the one hand, and the structures and processes that perform functions on the other. Let me say it as clearly as possible: The function of the code sequences evolved in the avida study was to perform input/output mappings corresponding to the several logic functions comprising the selective environment. There are two different senses of "function" in that sentence, and Truman doesn't distinguish between them.
No one has "alleged" that the logic functions themselves are "by definition IC." Even Behe doesn't claim that functions are IC: he claims that biological structures (e.g., the flagellum) and processes (e.g., the blood clotting cascade) are IC. Those structures and processes perform functions, but functions are not what Behe thinks are IC. If it were, he'd be saying that bacterial motility (or catching mice) is IC, rather than saying that the flagellum that performs the function of motility (and the mousetrap that catches mice) is the thing that's IC.
What has been claimed about the Lenski, et al., study is that the evolved code sequences, the sequences that perform the input/output mappings corresponding to the logic functions, are IC. Reread the Lenski, et al., paper: their knockout analyses identified the instructions (which certainly are components!) that were necessary for performing those various mappings, exactly instantiating Behe's operational definition for determing ICness, the knockout procedure for identifying IC structures and processes.
Finally, Truman wrote quote:
In Behe's examples multiple proteins are necessary concurrently for the described cellular process. You can't remove one component at the prcess still works. What exactly are you claiming can't be removed from the intermdiately complex logic functions?
As a consequence of Truman's conflation of functions and structures, his question is irrelevant. To echo the first sentence, 'in the avida runs multiple instructions are necessary concurrently for the described logic-performing process.' The knockout analyses show that you can't remove one instruction and still have the process work. That's IC, kids.
RBH
[ 05. May 2004, 19:29: Message edited by: RBH ]
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 05. May 2004 21:59
Royal,
I apologize for coming into this discussion at such a late date, but after look over some of the exchanges, I wonder if you maybe you aren’t asking the wrong question. Avida is based in some manner on some form of neo-Darwinian theory. Maybe the question that should be asked is “Is neo-Darwin theory biologically unrealistic?” If you start by making the assumption is the theory more realistic, for example by using more realistic genotype to phenotype mapping, you end up with far more complex or more difficult or more unlikely evolutionary changes.
If you start with the more complex and more improbable (but more biological realistic) evolutionary changes you can still find mathematical change processes which can produce the realistic changes, but they will involve more complex processes than suggested by neo-Darwin theory.
I am suggesting that if start by defining biologically realistic evolutionary changes, then look for algorithms that can simulate biologically realistic changes, you will have a new model or theory of evolution.
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 07. May 2004 17:36
Truman asked quote:
Let us first agree, that the authors are claiming at least conceptual relevance to real genes, and real metabolic processes. Can we establish this as a minimal basis to continue?
Sure, if we're careful about claiming particular equivalences. For example, in the quotations Truman supplied, a "gene" is clearly being taken to be analogous to a sequence of code that performs a particular logic function. That is, "gene" is not analogous to an instruction but to a sequence of instructions that performs a function, the performance of which increases reproductive fitness. In the avida case, performing a logic function increases reproductive fitness and the sequence of instructions that performs that function is taken by the authors of the program to be analogous to a biological gene.
That would seem to make individual instructions in avida analogous to codons, though avida instructions are not divisible as are biological codons. So a substitution mutation in avida is analogous to the substitution of a whole codon, not a base pair.
However, these are analogies and not strict correspondences, and so caution is necessary.
RBH
IP: Logged
|
|
Scott
Member
Member # 1222
|
posted 07. May 2004 19:27
quote:
That is, "gene" is not analogous to an instruction but to a sequence of instructions that performs a function, the performance of which increases reproductive fitness.
RBH,
How does selection work in the avida model?
Wouldn't a closer analogy be one in which the logic function did not represent some individual gene, but rather the entire sequence which begins with a gene and ends with something in the phenotype, say a protein, something upon which selection can operate?
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 07. May 2004 21:06
The genotype of an Avida program is the sequence of instructions that encode it. The phenotype of the program includes what output a program produces when a certain input is made available to it.
The logic functions that are rewarded are ones performed at this input-output level--at the phenotypic level.
Just as biological organisms have a genotypic (instruction) and phenotypic (outcome) level, so do Avida programs.
You don't say that blue eyes is the entire sequence which begins with an allele of a gene and ends with certain eye pigments. That's not really even a coherent thing to say. "Blue eyes" is the phenotype, it's the outcome we observe. It's not the process of creating blue eyes. Likewise, the rewarded "logic function" is the outcome, not the process.
IP: Logged
|
|
Royal
Member
Member # 1060
|
posted 14. May 2004 15:59
RHB 07.May 2004:
quote:
For example, in the quotations Truman supplied, a "gene" is clearly being taken to be analogous to a sequence of code that performs a particular logic function.
Royal: Yes, no question about it. One could argue that generally several genes are needed for truly novel biological functions, but this is simply a statiscally matter which could be approximated by requiring that the Avida logic functions be (far, far) more distance in sequence space.
RHB:
quote:
That is, "gene" is not analogous to an instruction but to a sequence of instructions that performs a function, the performance of which increases reproductive fitness. In the avida case, performing a logic function increases reproductive fitness and the sequence of instructions that performs that function is taken by the authors of the program to be analogous to a biological gene.
Royal: Precisely.
RHB:
quote:
That would seem to make individual instructions in avida analogous to codons, though avida instructions are not divisible as are biological codons. So a substitution mutation in avida is analogous to the substitution of a whole codon, not a base pair.
Royal: Exactly right. Insertion and deletions in base pairs will code for new amino acids. And the redundancy of the genetic code means that not all base pair mutations produce a new codon, but most do. This minor detail is of no concern, since one can simply increase the Avida mutational rate by about 30% to compensate.
Now, we can all agree. Many of the papers are indeed claiming, as I've been alleging the whole time, that the reported Avida runs support the notion neo-Darwinian processes have produced novel genes, indeed complex biological features. It is my intention to demonstrate that as we attempt to extrapolate from the reported experiments to biological realities, such claims do not hold.
IP: Logged
|
|
|
Printer-friendly view of this topic