|
Author
|
Topic: Designing Complex Causation
|
warren_bergerson
Member
Member # 262
|
posted 09. January 2004 11:59
From the discussion here and elsewhere surrounding the my proposed definition of intelligence, it became apparent that you can not understand intelligence unless you first understand what I call artificial or engineered complex causation. As outlined in the following draft, the concepts and principles of complex causation are much easier to understand than is sometimes suggested.
ARTIFICIAL OR ENGINEERED COMPLEX CAUSATION
INTRODUCTION
The analysis of complex causation, in its simplest form, comes down to the question "What happens if two simple forms or causation or two simple causal relationships interact?". If I can answer the question regarding the interaction of two simple causal relationships, then, in theory, I can ultimately answer the question "What happens if I combine very large groups of simple causal relationships using very complex structures?’.
If I have 1)the capacity to combine simple causal relationships to form complex causal relationships, and 2)the knowledge of what will happen when large numbers of simple causal relationships are combined, then, it follows logically, I have the ability to design and create artificial or ‘engineered’ forms of complex causation. Somewhat surprisingly, as will be discussed, humans have long had a general understanding of what happens when simple causal relationships are combined. Furthermore, humans have been manipulating simple causal relationships to create artificial or engineered complex causation for a very long time.
MATHEMATICS, LOGIC AND THE ‘LAWS’ GOVERNING THE INTERACTIONS OF SIMPLE CAUSAL RELATIONSHIPS
We know that simple causal relationships can be expressed or modeled as simple set theoretic relationships or functions. Furthermore we know, or so I have been led to understand, that any complex type of mathematical relationship can be constructed or derived from such simple relationships. This means, for example, that any complex computer program can be constructed from or derived from sets of simple relationships or the type used to model simple causal relationships.
If complex mathematical algorithms and complex computer programs can be constructed from elementary set theoretic functions, then it seems reasonable to propose that some complex causal processes could be derived or constructed in the same manner.
PRINCIPLE OF COMPLEX CAUSATION: The construction or derivation of complex causal relationships, subject to real world constraints, follows the concepts and principles of mathematical and deductive logic.
THE CAPACITY TO DESIGN AND CONSTRUCT COMPLEX CAUSATION
If the principle of complex causation is valid, then it should, taking into account real world constraints, be possible to construct systems which manifest the complex causal processes constructed or derived from logic and mathematics. The electronic computer provides one direct demonstration of this. We can design complex algorithms or programs representing complex causal processes and we, recognizing real world constraints, we can create real world computer programs which manifest these complex causal relationships. The demonstrations are a bit more complex, but it seems apparent that all complex designed or engineered systems represent this same phenomena.
CONCLUSIONS AND NEXT STEPS
Subject to convincing evidence to the contrary, it appears reasonably obvious that complex causal relationships, subject to real world constraints, follow the concepts of deductive and mathematical logic. It further appears obvious that humans have the capacity to design and construct some systems which manifest ‘artificial and engineered’ complex causal relationships constructed using mathematical techniques and logic.
Human beings are a natural part of the universe. If humans can design and construct ‘artificial and engineered complex causal processes’ then we must logically conclude that such artificial complex causal processes can be generated by nature. A logical next question is whether there are processes and systems other than humans capable of designing and constructing such ‘artificial complex causal processes’.
IP: Logged
|
|
Jack
Member
Member # 265
|
posted 09. January 2004 14:50
Warren<< Human beings are a natural part of the universe. If humans can design and construct ‘artificial and engineered complex causal processes’ then we must logically conclude that such artificial complex causal processes can be generated by nature. A logical next question is whether there are processes and systems other than humans capable of designing and constructing such ‘artificial complex causal processes’. >>
The Darwinist will say yes. The process of random variation and natural selection.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 10. January 2004 13:24
Jack,
I would be interested in hearing which Darwinist’s have actually addressed the questions of complex causation and man-made or engineered or designed complex causation. Such analysis would, at the very least, require significant clarifications of the Darwinist position on any number of issues.
Defined in terms of complex causation, RM&NS even is a complex process made up of a very large number of component processes. Many or most of these component processes and the interactions among such processes have not been identified or defined by biologists. Also, if you define RM&NS in terms of complex causation, RM&NS is a dynamic process which means there exists other complex processes capable of modifying the RM&NS processes.
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 10. January 2004 13:40
For a good and recent thesis which addresses many of your questions I suggest you check out
M. Toussaint (2003): The evolution of genetic representations and modular neural adaptation
Discussed on this ISCID thread
quote:
Neutral genotype-phenotype mappings can be observed in natural evolution and are often used in evolutionary computation. In this article, important aspects of such encodings are analyzed.
First, it is shown that in the absence of external control neutrality allows a variation of the search distribution independent of phenotypic changes. In particular, neutrality is necessary for self-adaptation, which is used in a variety of algorithms from all main paradigms of evolutionary computation to increase efficiency.
Second, the average number of fitness evaluations needed to find a desirable (e.g., optimally adapted) genotype depending on the number of desirable genotypes and the cardinality of the genotype space is derived. It turns out that this number increases only marginally when neutrality is added to an encoding presuming that the fraction of desirable genotypes stays constant and that the number of these genotypes is not too small.
Other resources include
EVOLUTION OF BIOCOMPLEXITY AND ROBUSTNESS at the Santa Fe Institute
or the paper published in Nature
The Evolutionary Origin of Complex Features
Chris Adami's Publications
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 10. January 2004 14:09
Pim,
Do the references you listed support or contradict the suggestion that the combining of sets of simple causal relationships to form complex causal relationships follows the principles of mathematical logic? Do the references support or contradict the claim that humans can design and construct complex causal processes? Do the references lists support or contradict the existence of complex causal processes capable of modifying complex evolutionary processes? If as you suggest the references listed are relevant to the subject being discussed, could you please clarify the relevance. Thanks in advance for your assistance.
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 10. January 2004 14:34
These links are relevant to your suggestion that
quote:
A logical next question is whether there are processes and systems other than humans capable of designing and constructing such ‘artificial complex causal processes’.
Check'em out.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 11. January 2004 10:55
Systems designers routinely write programs with hundreds or thousands of lines of code where each line of code is reducible to large number of logical operations. Running such a program can easily involve millions or billions of logical operations. I propose that such programs can be viewed as engineered complex causation and can be reduced to complex sets of elementary causal relationships.
If you write a GA program to model and simulate an evolutionary change process, what have you accomplished? Viewing or defining programs as engineered complex causation would suggest you have demonstrated that evolution is a form of ‘engineered complex causation’? Since the GA program needs to be modified to model or simulate different occurrences of evolutionary change, this suggests that there exists complex causal processes not defined by evolutionary biology which are capable of modifying evolutionary processes.
Pim provides references demonstrating the existence of GA programs. Such references do not address the issue of what such programs mean for the causal nature of evolution.
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 11. January 2004 15:52
Warren: Pim provides references demonstrating the existence of GA programs. Such references do not address the issue of what such programs mean for the causal nature of evolution.
I suggest that Warren reads Toussaint for instance before he makes these claims. Of course he is correct that such programs do not address that it happened merely that it could have happened. What is fascinating to me is how our GA program experience has found what evolution seems to have found naturally namely the importance of evolvability.
Toussaint's thesis provides for a solid mathematical foundation
The experiments by Adami, Lenski, Ofria et al show how evolutionary processes may increase complexity and information in the genome.
Surely they address the question raised by Warren logical next question is whether there are processes and systems other than humans capable of designing and constructing such ‘artificial complex causal processes’
Toussaint
quote:
Evolutionary search explores the search space by means of mutation and recombination, which defines a variational topology on the search space. Intuitively, this kind of search will be efficient if the fitness function is not completely random but more or less continuous with respect to this variational topology such that continuous evolutionary progress becomes possible. The continuity of fitness on the variational topology has been captured by the definition of strong causality (Rechenberg 1994; Sendhoff, Kreutz, & von Seelen 1997), a concept which in turn may be used for the design of mutation operators that allow for more continuous progress.
However, in natural evolution mutation operators are not designed by some intelligence. A central question arises: What does it mean to “learn” about the problem structure and exploit it? How in principle can evolution realize this? The answer we will give is that the implicit process of the evolution of genetic representations allows for the self-adaptation of the “search strategy” (i.e., the phenotypic variability induced by mutation and recombination). To some degree, this process has been overlooked in the context of evolutionary algorithms because complex, non-trivial (to be rigorously defined later) genetic representations (genotype-phenotype mappings) have been neglected by theoreticians. This chapter tries to fill this gap and propose a theoretical framework for evolution in the case of complex genotype-phenotype mappings focusing at the evolution of phenotypic variability. The next section lays the first cornerstone by clarifying what it means to learn about a problem structure.
For more background check Here
Also the following paper
Causality in Genetic Programming (1995) Justinian P. Rosca, Dana H. Ballard
may be helpful for the discussion
quote:
Causality relates changes in the structure of an object with the effects of such changes, that is changes in the properties or behavior of the object. This paper analyzes the concept of causality in Genetic Programming (GP) and suggests how it can be used in adapting control
parameters for speeding up GP search. We first analyze the effects of crossover to show the weak causality of the GP representation and operators. Hierarchical GP approaches based on the discovery and evolution of functions amplify this phenomenon. However, selection gradually retains strongly causal changes. Causality is
correlated to search space exploitation and is discussed in the context of the exploration-exploitation tradeoff. The results described argue for a bottom-up GP evolutionary thesis. Finally, new developments based on the idea of GP architecture evolution (Koza, 1994a) are
discussed from the causality perspective.
Proceedings of the Fifth International Conference (ICGA95) Morgan Kaufmann, San Franc...
Another interesting webpage on a causal theory of evolution
Causality and teleology in evolution may be quite similar concepts. In another posting I have argued that evolutionary processes can explain teleology in nature.
James Barnham has collected BIBLIOGRAPHY OF NATURAL TELEOLOGY AND NON-DARWINIAN EVOLUTION . I would argue that his claim of non-Darwinian evolution may miss the point namely that selection was but one of Darwin's contributions and even Darwin did not consider selection to be the exclusive mechanism. In fact as we know now, neutrality is an important concept as well.
In another ISCID paper Jakob Wolf addresses issues of causality and teleology. And although I disagree with many of his statements such as "In his book Darwin’s Black Box, Michael Behe demonstrates with a wealth of empirical detail
that irreducibly complex biochemical systems exist." (no evidence of IC systems has been presented so far (See also Mark Perakh "Unintelligent design, Prometheus Press")), but he attempts to deal with the issue of causality. That IC systems are argued to be unexplainable by natural processes mirrors the problems with the definition of IC namely the issue of 'begging the question'.
While the concepts of causality raised by Jakob are interesting, I would argue they suffer from many of the same problems as the concept of ICness.
[ 11. January 2004, 16:28: Message edited by: Pim van Meurs ]
IP: Logged
|
|
RBH
Member
Member # 380
|
posted 11. January 2004 16:47
Pim wrote quote:
In another ISCID paper Jakob Wolf addresses issues of causality and teleology. And although I disagree with many of his statements such as "In his book Darwin's Black Box, Michael Behe demonstrates with a wealth of empirical detail that irreducibly complex biochemical systems exist." (no evidence of IC systems has been presented so far (See also Mark Perakh "Unintelligent design, Prometheus Press")), but he attempts to deal with the issue of causality. That IC systems are argued to be unexplainable by natural processes mirrors the problems with the definition of IC namely the issue of 'begging the question'.
I have to disagree. Behe did demonstrate that irreducible systems in his restricted DBB definition exist. What he did not demonstrate was the (erroneous) conclusion he drew from that fact, namely that IC systems cannot evolve via the various 'Darwinian' mechanisms. They demonstrably can.
RBH
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 11. January 2004 17:22
Pim,
Neither your comments nor your references are addressing the issues being discussed. I am not disputing that GA’s are capable of modeling and simulating some aspects of evolutionary change processes. The questions here are 1)what is the logical/causal nature of a GA, and 2)what are the implications for evolutionary theory. I am saying that a GA is a ‘dynamic or programmable engineered complex causal process’. Nothing you have said or referred to addresses the question of ‘what is the logical/causal nature of a GA’. Until one of your quoted experts either accept the proposed interpretation, or provide an acceptable alternative interpretation, you are going to have a difficult time explaining what implications GA’s have for the logical/causal nature of evolutionary processes.
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 11. January 2004 18:35
RBH
quote:
I have to disagree. Behe did demonstrate that irreducible systems in his restricted DBB definition exist. What he did not demonstrate was the (erroneous) conclusion he drew from that fact, namely that IC systems cannot evolve via the various 'Darwinian' mechanisms. They demonstrably can.
I stand corrected. Thus the stronger conclusion has been replaced by the much weaker 'no detailed Darwinian pathways exist'
quote:
In The Origin of Species Darwin stated 6:
If it could be demonstrated that any complex organ existed which could not possibly have been formed by numerous, successive, slight modifications, my theory would absolutely break down.
A system which meets Darwin's criterion is one which exhibits irreducible complexity.
Link
I meant my comments to reflect the statement that "An irreducibly complex system cannot be produced gradually by slight, successive modifications of a precursor system, since any precursor to an irreducibly complex system is by definition nonfunctional"
Warren objects
quote:
Until one of your quoted experts either accept the proposed interpretation, or provide an acceptable alternative interpretation, you are going to have a difficult time explaining what implications GA’s have for the logical/causal nature of evolutionary processes.
WHy would they have to accept your intepretation when they have given a much more acceptable interpretation of causality and GA's? If Warren wants to ignore the comments and resources provided by me that is his choice of course.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 12. January 2004 09:10
GA programs or systems are clearly 1)man-made or designed, 2)dynamic or modifiable(programmable), 3)expressible as complex mathematical operations or functions which can be reduced to sets or collections of interacting elementary operations or functions, and 4)capable or controlling the real world behavior of a computer. Simple or elementary causal relationships of the form "A always causes B’ can be expressed or modeled as mathematical functions. It is therefore reasonable to suggest that complex causal processes can be expressed as interacting sets of simple causal processes. It is further reasonable to suggest that computer programs including GA programs are expressions of complex causation. To date, no one here or on any other forum discussing the subject has questioned the reasonableness of this approach.
Starting with the reasonable view that GA programs are manifestations of complex causation leads to some interesting and useful conclusions regarding evolutionary processes. To begin, it is useful to note that if you start with a GA program, you can first reduce the program to a set of lines of computer code, and each line of code can be reduced to a set of logical operations. In analyzing a GA program in terms of logical operations, you will find no single simple elementary operations which represent either ‘random mutation’ or ‘natural selection’. A GA program is not an expression of the interaction of elementary variation and selection processes. Variation and selection are at most emergent properties of GA programs. This would in turn suggest that variation and selection are emergent properties of complex evolutionary processes rather than fundamental or elementary causal processes.
Not only are variation and selection, emergent rather than fundamental properties of GA’s, but they are highly modifiable emergent properties. You can create GA programs that use forms of variation and selection that vary substantially from the basic concepts of ‘random variation’ and ‘natural selection’. Not only are departures from these basic concepts compatible with GA programs, but many of the modifications significantly improve performance of GA search routines.
Three points are worth noting here. First, the findings derived from analyzing GA’s in terms of complex causation are the same as the results that would be obtained analyzing GA’s in terms of elementary logical or mathematical operations. Second, if anyone is interested, it is not difficult using this type of analysis to find number of apparent inconsistencies between the logical mathematical nature of GA’s and common Darwinist beliefs. Finally, the primary purpose in recognizing that computer programs are manifestations of ‘dynamic, engineered, complex causation’ is not to address Darwin, but to address the modeling, simulation and formulation of predictive theories relating to the ‘intelligent’ behaviors of biological systems.
IP: Logged
|
|
Pim van Meurs
Member
Member # 541
|
posted 12. January 2004 18:36
Warren: Second, if anyone is interested, it is not difficult using this type of analysis to find number of apparent inconsistencies between the logical mathematical nature of GA’s and common Darwinist beliefs.
Such as? Pray do tell us.
[ 12. January 2004, 18:36: Message edited by: Pim van Meurs ]
IP: Logged
|
|
Rex Kerr
Member
Member # 632
|
posted 13. January 2004 02:33
Warren wrote that selection and variation are non-elementary processes, and said: "This would in turn suggest that variation and selection are emergent properties of complex evolutionary processes"
(emphasis mine).
Variation and selection are indeed emergent properties, but why suppose that they are emergent from an evolutionary process? It seems simpler to say that evolutionary processes are emergent from certain sets of more fundamental properties/qualities/rules/etc..
[ 13. January 2004, 02:34: Message edited by: Rex Kerr ]
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 13. January 2004 09:26
Rex,
The point of the paragraph you quoted from was that it does not appear possible that an evolutionary process can be constructed or defined from elementary variation and selection processes.
The suggestion here is that ‘an evolutionary process’ like a GA program is a complex causal process which, at least in theory, is reducible to elementary functional or causal elements. As in computer programming, reducing a complex causal process to elementary logical operations is not useful or informative.
It is also suggested here that just as ‘all GA programs’ are not identical, not all evolutionary processes are identical. This in turn suggests that that analysis of evolutionary processes involves the analysis of sets of classes of related complex causal processes. This would in turn suggest that it is not appropriate to suggest that ‘evolutionary processes are emergent from certain sets of more fundamental properties/qualities/rules/etc..’ .
IP: Logged
|
|
|
Printer-friendly view of this topic