|
Author
|
Topic: Evolution in Single and Dynamic Environments
|
James A. Barham
Member
Member # 50
|
posted 19. May 2002 21:36
Charlie D. wrote:
-----------------------
"It is not clear to me if the "dynamic environment" hypothesis proposes that a different type of resulting vector be used to measure evolutionary outcomes, but if the goal is to demonstrate the "the inadequacy or incompleteness of Darwinian models and theories", unless that resulting vector is something different than relative reproductive success, it will still fall underneath Darwin's theoretical umbrella, just with a more sophisticated mathematical treatment. "
------------------------
Explanation is a relative concept.
Kepler discovered a superficial, descriptive ("phenomenological") relation governing the orbits of the planets. Nevetheless, this was highly explanatory in relation to Ptolemaic models. Later, Newton provided a far more sophisticated and unifying formalism which was a far deeper explanation of the observed phenomena than Kepler's. Later still, Einstein provided an explanation of the same phenomena that was deeper still.
According to Charlie D.'s logic, Newton and Einstein both just fell under Kepler's "theoretical umbrella." (Or, come to think of it, they were all of them really just following in Master Ptolemy's footsteps . . .)
IP: Logged
|
|
charlie d.
Member
Member # 159
|
posted 19. May 2002 22:41
James:
as you say yourself, Kepler just provided an accurate description, not an explanation, of the planets' orbits, while Darwin (whether you agree with it or not) provided an explanatory paradigm for evolution, so I don't really see your point there.
My point is that as much as one can make the mathematical treatment of adaptive evolutionary change as complex as one chooses (or can manage), if the predominant explanatory principle is differential reproductive success, that's a darwinian view of evolution. Do you disagree?
Therefore, trying to stay on subject, I was just asking whether warren's "dynamical environment" thing had to do with a different explanatory principle, or just with higher math capable of dealing with multiple adaptation vectors simultaneously. His original post made a straw-man "one-organism-one-environment-one-adaptation" argument which no biologist would subscribe to, but I failed to see what his theoretical criticism of darwinism actually boiled down to. Maybe you understand his hypothesis enough to be able to tell me? Otherwise, I'll just have to wait for Warren's reply.
IP: Logged
|
|
James A. Barham
Member
Member # 50
|
posted 20. May 2002 08:23
Charlie D.:
I will let Warren speak for himself, but I will be glad to reiterate why I believe the Kepler/Newton = Darwin/?? analogy is a sound one.
The reason is that I regard natural selection as a superficial or phenomenological description of the evolutionary process, not as a deep explanation.
What do I think would constitute an explanation of evolution? A deeper physical understanding of life itself. In nonlinear dynamics, we have both many-to-one ("equifinal") and one-to-many ("bifurcation") types of behavior. I view living systems as instantiating a particular (not generic) kind of equifinal behavior (specifically, govered by low-energy triggers that provide correlations between the timing of functional actions and external circumstances). Thus, evolution will consist of symmetry breaking events giving rise to new attractors, usually encompassing additional functions. This can all be redescribed as natural selection, to be sure, but I believe that symmetry breaking and bifurcation are far deeper and more explanatory than "selection."
Any change whatsoever can be described as "selection" (and there is no lack of philosophers who take such a pan-selectionist view). For instance, we could "explain" the process of building up gold inside stars as a process of "selection". But I think you would agree that this is a superficial view of the matter. Quantum mechanics and astrophysics provide a much deeper explanation.
Similarly, it is not that I believe Darwinian selection theory is "wrong," so much as I view it as superficial. But it is the insistance that this is all there is to the story---that "selection" explains everything about life, and all else is "tinkering"---that I regard as willfully blinkered and obstructionist. It is this Dennett-style ultra-Darwinian metaphysics that I (and many others here at ISCID) mainly have in mind when discussing the flaws in Darwinian logic, etc.
[ 20 May 2002, 08:25: Message edited by: James A. Barham ]
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 20. May 2002 09:22
Charlie D (and others)
As you point out, the dynamic environment assumption, it is not an hypothesis, would only produce a more complex model of Darwinian evolution unless in some way it requires a major revision in Darwin’s concept of Natural Selection. It is also generally true that finding fault in one theory will not be widely accepted unless there is a positive, and acceptable alternative. In starting this thread, I was looking for at demonstrating the incompleteness and inadequacy of current theoretical formulations, but the dynamic assumption also leads to ‘positive alternatives’.
To begin, while it is true that scientists recognize that life forms exist in a dynamic environment, Darwinian ‘theory’ is based on the single environment assumption. It is generally assumed, that by adding non-evolutionary processes and mechanism you could create a model that would work in a dynamic environment. My claim is that 1)contrary to the generally accepted assumption, Darwinian theory breaks down when try to apply it in a dynamic environment and 2)solving the dynamic environment ‘challenge’ leads to a positive alternative to Darwin (this is the part I haven’t previously discussed).
To begin with, the single to dynamic environment transition can easily be studied with mathematical models. Genetic algorithms provide good examples of single environment models. To create a dynamic environment model, you simply create a model that 1)solves for multiple adaptive solutions, and 2)that can change between different adaptive solutions as the environment changes. As far as I know, there has been some, but not a lot of work in this area. The following comments are based on my unpublished analysis.
First, and fairly obviously, in order to build a model that will work in a dynamic environment you need to add operations and functionality in addition to the neo-Darwin mutate-select or the more general Darwinian diversity-selection-heritability. [This introduces the whole issue of ‘how do you precisely define and scientifically model’ complex causal relationships like evolution?’ Is it acceptable to simply name a couple of key processes like mutation and selection, or do scientific explanations require defining the connective mechanisms? IMO, a scientific model, theory or explanation of a complex process like evolution requires a mathematical model capable of simulating evolutionary change.]
Second, there are some fairly simple yet powerful ‘adaptive’ processes that work very well in a dynamic environment. I call them ‘progressive equilibrium’ processes, but there are other names which could be used. The thermostatic or equilibrium maintaining mechanisms is the simplest example. If the temperature drops, a source of heat is activated. If the temperate is above a certain point, the heat source is deactivated. The progressive equilibrium process is simply a thermostat that learns or changes.
Progressive equilibrium processes and models, have two ‘interesting’ features. First, biological systems involve lots of these mechanism, ‘maintaining equilibrium’ seems to be a better description of biological adaptation than ‘find the optimal solution’. Second, viewed in terms of ‘equilibrium maintenance’, evolution would operate on ‘equilibrium maintaining processes, not of ‘heritable traits’. This observation raises the questions 1)what does evolution change and 2)what does ‘selection’ operate on? The genetic approach offers an answer the question of what evolution changes, but it leaves open the question of ‘what does selection operate on?’ [ I am only offering you rough sketch of how the dynamic environment assumption leads to both identifying problems with Darwinian theory and leads to a possible alternative. Analysis of progressive equilibrium processes lead to the suggestion that concept of ‘heritable phenotype trait’ is not precisely defined. ]
Third, analysis of progressive equilibrium processes leads to the conclusion or at least suggestion that the Darwinian concept of Natural Selection is fundamentally flawed. The analysis of progressive equilibrium processes suggests that ‘adaptive change’ involves a hierarchy of processes operating on processes operating on processes. Selection or ‘select the option consistent with survival’ is occurring at every level in the hierarchy. The high level ‘between generation’ Darwinian selection processes are only a part, and in post instances, only a relatively minor part of evolutionary/adaptive change.
Changing from the single environment to the dynamic environment assumption is only important if it leads to a viable alternative to Darwinian evolution. In raising the issue of changing or challenging the single environment assumption, I was asking for comments on the first step of a trail I think leads to an alternative to Darwin. The discussion above, is a brief outline of where the trail leads once you challenge the single environment assumption.
IP: Logged
|
|
Janitor@MIT
Member
Member # 125
|
posted 20. May 2002 13:09
warren_bergerson might find some interesting pre-prints on the Web by Tony Hirst and Jonathan Rowe: “Evolution as Signal Processing: On the Interpretation of Analytic Models of Evolution in Cyclically Changing Environments.”
Standard quantitative genetics models of the population dynamics under cycling and “fast” environmental regimes are translated into 1) a low-order Butterworth filter and a 2) Finite Impulse Response filter. I found these articles while looking for information on extended and adaptive Kalman filters (which are simple “in principle,” but very effective), because I thought populations would implement more sophisticated filter technologies to track targets that move faster than typical generational spans. (The authors see the problem and even consider briefly a “learning model” of genotype<>phenotype mapping.)
The authors note that (in particular, the FIR analysis) “... is hard to interpret in a common-sense way, identifying such conditions as possibly falling outside the scope of the quantitative genetics model... In addition, the setting of the filter parameters by evolution raises interesting questions under the adaptationist view…”
Specifically the difficulty (for a traditional interpretation, a “Darwin filter”) is the selective setting of filter parameters recursively , i.e., by “implementation” rather than by “imposition,” if you know what I mean. Accordingly, the authors reverse the traditional causal ordering, which I find attractive because I think traditional biology has it exactly backwards: calling causes what are in fact effects.
Additionally, the authors note how the setting of the performance band (its depth and width) poses a number of difficulties for received theory and leads to an interesting hypothetical adaptive strategy of “virtual populations.” (Isn’t that what a phenotype is?—the genotype’s “theory” of its own future evolution?)
Anyway, I’m rambling. Very interesting article though.
IP: Logged
|
|
|
Printer-friendly view of this topic