|
Author
|
Topic: Modeling Creative Intelligence
|
warren_bergerson
Member
Member # 262
|
posted 20. May 2002 16:14
ID, in simplistic terms, is the ‘hypothesis’ that biological complexity is the result of a very powerful creative intelligence. Within ID, there are those who believe that this creative intelligence arises from external sources and those that believe it arises from something within biological systems. Of those who believe that creative intelligence exists ‘inside’ biological systems, I am probably alone in having developed a generalized logic machine program the appears capable of modeling, simulating and/or explaining the creative intelligence which could have produced most known occurrences of biological complexity. The generally accepted view is that there is no known model capable of simulating creative intelligence. The purpose of this thread is to introduce for comment and discussion the basic concepts, perspectives, techniques and assumptions that I suggest make it possible to observe, measure, analyze, model, simulate, and explain creative intelligence.
BACKGROUND
The approach I am proposing arose not from the study of creative intelligence or ID, but from modeling and analyzing human decision making. As will be discussed, I discovered a couple of techniques (tricks?) that make it practical to measure and model the ‘logical programs’ or ‘decision logic’ people use to make decisions. Expanding on the initial discovery, it proved practical to observe, measure and model the processes involved in changing the ‘decision programs’ people use to make decisions. People, it is generally recognized are capable of creative decision making or using creative intelligence in decision making. By observing the processes associated with this creative decision making, it was possible to develop a mathematical model of the process. [Note: This is just a high level overview to provide a perspective on where these ideas can from. It is obviously not an explanation of the ideas or a description of how they were developed.] Given a model of how creative intelligence forms complex patterns of human behavior, it was then possible to generalize the processes and models of creative human intelligence to the creation of other types of biological complexity.
PERSPECTIVE
As might be expected, the proposals presented here are based almost entirely on reinterpreting existing information, facts, and research rather than a substantial body of new research. The subject of ‘allowable perspectives’, ‘required perspectives’, and ‘useful perspectives’ will likely arise in any discussion of creative intelligence.
THE STARTING POINT
The claim here is that there are mathematical or logic machine programs that can model, simulate, and ‘scientifically explain’ creative intelligence. In discussing this program it will be useful to start with a simple genetic algorithm or mutate-select program. Although, as viewed here, GA models can’t simulate creative intelligence, they do provide a useful starting point in defining some of the terms and concepts involved in modeling intelligence.
Building a system capable of simulating creative intelligence, can start by making the following 3 modifications to a GA model:
1. DYNAMIC ENVIRONMENT - This is being discussed elsewhere. Rather than solving one adaptive problem at a time as with basic GA models, an ‘intelligent system’ is assumed to face a rapidly changing environment requiring many different adaptive solutions and requiring the ability to distinguish which adaptive solution applies when.
2. DYNAMIC FITNESS LANDSCAPES- Intelligent systems, like GA simulations, search for ‘optimums values in a fitness landscape’. Unlike basic GA simulations, the fitness landscape for an intelligent system is assumed to be dynamic or changeable.
3. MULTIPLE AND CONTINUOUS INTERACTIONS - GA simulations are assumed to involve a limited number of processes or mechanisms involving interaction with the physical environment. Intelligent systems involve more numerous and continuous interactions with the external environment.
As viewed here, GA models are rather simplistic systems which very slowly perform a very limited range of functions, in a very (artificially) limited environment. Intelligent systems, by contrast, operate in a complex and unstable environment where it is important to be able to adapt rapidly to both previously experienced and novel conditions. This I suggest is the type of environment in which human creative intelligence operates. By observing and modeling how humans operate in such an environment, I developed a model or logical system which appears to be capable of creative intelligence.
If it is acceptable to the moderators, I will present additional detail on this intelligent system and the techniques used to develop it. From what I have read here, creative intelligence is a central concept in ID. I also get the impression that many believe it is not possible to simulate creative intelligence. I would interested in any comments people might have one this subject.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 21. May 2002 15:12
[The first step in modeling creative intelligence is to precisely define and model what is changed. The following is a brief overview of the ‘objects’ changed by human intelligence and techniques for modeling these objects. ]
FINDING A SMALL UNIT TO OBSERVE AND MODEL
Creative intelligence is characterized as a change processes. Human creative intelligence operates to creatively change human behavior. The first step, as viewed here, in studying creative intelligence is to identify ‘What is changed by creative intelligence?’ Ideally, the identified object of change would be 1)simple or elementary, 2)contain all the key features of change objects and 3)be easy to observe, analyze, and model. The model developed here is based in large part on 1)identifying the elementary units changed by human intelligence and 2)discovering a practical method of observing and modeling these elementary units.
The starting point in identifying this elementary unit is the observation that the ‘decision’ or ‘voluntary decision’ plays a central role in controlling or determining voluntary human behavior. It is possible to model complex human social behavior in terms of sets of voluntary decisions. [Probably the most sophisticated known technique for modeling complex behavior in terms of voluntary decisions is double entry accounting. I will probably loose half my audience with the claim that something as lowly as accounting is actually a powerful modeling tool. Contrary to popular belief, calculus is not the only useful form of mathematics.]
Human decision making is a very complex process. For the analysis here, this complex process is separated into ‘preliminary information processing’, ‘decision implementation’, ‘learning how to make a decision’, and ‘final decision making’. In attempting to isolate and model the final decision making process, I found an interesting and useful type of mathematical algorithm which will be labeled here as a ‘decision making algorithm’ or DMA. The DMA has the following five useful features[I will the algorithm available to anyone interested in the mathematics.]
1. PROGRAMMABLE PROCESSING ALGORITHM- A DMA is an information processing algorithm which translates the values of input variables into the value of the an output variable. A DMA can be reprogrammed to process any(or essentially any) possible logical/functional relationship between input and output.
2. PROGRAMMING- A DMA is programmed or reprogrammed by changing input and output variables, not by changing the processing algorithm or logic. This is ‘useful’, because while it is possible to observe and measure input and output, it is not possible to directly observe what is going on ‘inside the black box’.
3. MODELS FINAL DECISON MAKING FOR ANY VOLUNTARY DECISION- It appears that the DMA can be used to model the final decision making for any voluntary human decision. This appears to apply equally to very simple and very complex decisions.
4. OBSERVING VARIABLES- The variables, and the values or contents of the variables, associated with making a voluntary decision, (the variables in a DMA) can be independently observed and verified.
5. OPTIMIZATION- It is possible(it appears possible) to identify the optimal or near-optimal form of DMA for a specified decision in a specified environment. This is useful in testing the validity of the models developed.
SUMMARY
If you have a operation O that exhibits creative intelligence, then Starting with A0 and applying O generates a creative result A1. or O(A0)= A1. The claim is made here that "Human creative intelligence applied to voluntary decision logic will produce creative decision making". The discussion above is an outline of a process to precisely define and model A0 and A1.
The generally accepted view is that it is not ‘possible’ to define or model or measure how people make voluntary decisions. The claim here is that these decisions can be modeled by a DMA. The first three characteristics of a DMA as discussed above involve mathematical modeling techniques or tricks which make modeling practical.
The ‘surprising’ feature of DMA’s is that the input and output involved in human decision making are publicly observable. The ability to model voluntary decision making depends on the fact that the information used in human decision making is ‘publicly available and publicly measurable’. This rather surprising finding can be demonstrated experimentally.
Using the DMA, and publicly available information, it is possible to identify for a specific decision and a specific environment, an optimal or near-optimal decision making. If human decision makers are using information or information processing logic beyond that defined by the near-optimal DMA, then the human decision maker, particularly the ‘expert human decision maker, should be able to outperform the DMA logic. The experimentation to date indicates human’s generally can not outperform near-optimal DMA logic. [ I will be glad to demonstrate the phenomena for anyone interested in performing such an experiment.]
The above discussion is, I know, heavy on general concepts and short on specific detail. I can provide at least some additional details either here or privately for anyone interested.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 22. May 2002 12:41
As viewed here, human creative intelligence is primarily responsible for modifying human behavior and this primarily involves modifying voluntary decision processes. Having proposed DMA’s as precise models of voluntary decision making, it is possible to start defining "O" the process or operation responsible for the transition from some DMA0 to some DMA1 or O(DMA0)=DMA1.
BASIC PROCESS
The basic concept or principle underlying the process O has been known since the time of the ancient Greeks. In simplified terms applied to decision making, purpose or teleological causation is the result of the a process where "Given 2 forms of decision logic DMAc (current form) and DMAn (new form ) then ‘O’ will replace DMAc with DMAn if and only if the expectation of achieving goal G is higher for DMAn than for DMAc"
Human decision making, like most features of biological systems, is purposeful or teleological. Purposeful or teleological relationships or causation are the result of processes which replace relationships with a lower expectation of achieving a goal G with relationships with a higher expectation of achieving the goal G. The model offered here to explain creative intelligence is simply a refinement of this basic concept of purposeful causation.
It is, IMO, one of the ironies of the evolutionary debate that Darwinian theory is clearly a teleological theory. Darwinian theory, despite its numerous and obvious flaws, has survived for 150 years because it is a teleological theory, and because its supporters refuse to permit consideration of alternative teleological theories. In terms of teleological models, the model proposed here could be viewed as an enhancement of the Darwinian model. The two major changes in the proposed enhancement are 1)a far more complex description of the change process, and 2)recognition of a wide range of teleological change processes in addition to Natural Selection. Darwinian Natural Selection, it will be noted, is assumed to play no direct role in modifying human behavior.
SPECIFIC CHANGE PROCESSES AND MECHANISMS
In order to produce a specific detailed model of the change process which exhibits ‘human creative intelligence’, it is necessary to answer the following three questions:
1. What, exactly, is changed when a DMA changes?
2. What are the logical processes involved in producing specific changes? And
3. What are the observable real world mechanisms responsible for the claimed logical change processes?
1. WHAT CHANGES? Changes in DMA’s can be grouped into three types of changes as follows:
A. STIMULI- changes in input values- as mentioned earlier, such changes involve not only changes in input values, but also changes in parameters that control the logic or program controlling decision making.
B. RESPONSES- Decision making can be defined in terms of ‘select a response option’. Changes in the set of options to which selection applies represents a change in decision making.
C. GOALS- As viewed here, the goals of human decision making don’t change, but the ‘expectations’ that particular DMA’s will achieve those goals can change. This is the dynamic fitness landscape assumption.
2. WHAT ARE THE LOGICAL OPERATIONS ARE ASSOCIATED WITH THE ABOVE TYPES OF CHANGES?- As viewed here, there are four basic logical operations or processes performed on DMA’s. These are a)preservation, b)execution, c)modification and d)creation(of new DMA’s). The combination of these four operations and the three types of changes listed can logically model or simulate any change from any DMA0 to DMA1. A generalized or idealized model of these change processes has been developed and is available for review. (This is called the Life Force Simulator or LFS). [ Note: The assertion that changes in decision making, including creative changes, can be modeled or simulated by the LFS is similar to the claim that genetic changes can be modeled or simulated by mutate-select or GA models. There are important philosophical differences between ‘could be simulated’, ‘can be simulated’ and ‘can be scientifically explained by’. As viewed here, genetic change ‘could be’, but ‘can’t be’ simulated by mutate-select. The claims are made here that changes in human decision making, could be, can be, and can be scientifically explained by the LFS. As with mutate-select, evaluating these claims is in itself a very complex process.]
3. WHAT ARE THE OBSERVABLE REAL WORLD MECHANISMS RESPONSIBLE FOR CHANGE? - Given the complexity of the LFS, it is not, in the space available here, feasible to list all the real world mechanisms associated with changing DMA’s. However, the major process or type of process is readily identified. As will be recalled, one of the major components of the teleological paradigm or process is selection. Specifically, teleological causation involves selecting between different types of relationships. In the case of decision making, this involves selecting between DMAc and DMAn. Human decision making, as defined by the LFS, involves several different types of variations on the selection process. The real world mechanisms for all the different types of selection are, however, very similar. The selection mechanisms are voluntary decision processes. [The idea of ‘voluntary decision processes’ operating to change the logic of ‘voluntary decision processes’ may at first be ‘intuitively confusing’. ]
SUMMARY
The above is a brief overview of the processes or operations which, it is claimed here, are responsible for changing voluntary decision making. These change processes, again it is claimed, are capable of producing creative changes in decision making. The process defined is obviously both complex and unconventional. However, despite the complexity involved, it is believed that the entire process is based on ‘hard science’ standards and subject to hard science review, testing and validation.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 23. May 2002 11:17
Over the last three days I have outlined an approach which, I claim, makes it possible to model, simulate, and explain human creative intelligence. I am proposing that creative intelligence is a property of a human change process. The elementary subject or object of this human change process is modeled by a DMA. The change process which operates on the DMA is modeled by a LFS.
The development of techniques for modeling, simulating and explaining human creative intelligence has both practical and theoretical significance. The techniques developed are scientifically verifiable because they produce testable predictions regarding the nature of human intelligence which directly contradict generally accepted views on the subject. The two most glaring departures from conventional wisdom are the predictions that:
1. Human intelligence is a group or social phenomena, not a characteristic of the individual. And
2. Human behavior and human nature are dynamic and ‘currently programmable’.
Although human creative intelligence is important in itself, there is evidence that there are other forms of creative intelligence operating in biological systems. If the techniques developed here can model human creative intelligence, the obvious question arises, could the techniques be used to model, simulate, and explain other forms of creative intelligence operating in biological systems? The suggestion or hypothesis offered here is that ‘although the physical mechanisms or manifestations are very different, all forms of creative intelligence exhibited by biological systems involve the same logical structure identified for human creative intelligence’.
The initial evidence for this hypothesis is the general ‘intelligent design’ observation that the complexity and designs observed in nature are, or appear to be similar to the complexity and design created by human intelligence. As should be apparent, the assertion that ‘creative intelligence in biological systems can be modeled, simulated, and explained in terms of LFS’s operating on some type of DMA’ is far more precise and testable than the general claim that there are similarities among different types of creative intelligence.
If the hypothesis that ‘occurrences of creative intelligence can be modeled by LFS’s and DMA’s is valid, then 1)it must be possible to identify the physical manifestation of the DMA, and 2)it must be possible to identify the physical manifestations of storage, communications, and processing operations suggested by the LFS.
It is believed here that there are only a very limited number of biological phenomena that meet the requirements to be modeled by DMA’s. The human decision process is one. The neuron is a second such unit. It is believed that 1-3 such mechanisms exist at the cell level and are responsible for ‘cellular intelligence’.
In order to be modeled as a DMA, a physical phenomena must satisfy the criteria for a ‘dynamic causal relationship’ or ‘programmable information processing unit’(PIP). To qualify as a dynamic causal relationship, a physical phenomena must 1)generate a reaction or response to input or stimuli, 2)must have the capacity to preserve the input output relationship or program, and 3)the capacity to be modified or reprogrammed. It can be shown that human ‘decision processes’ have these properties. It can also be demonstrated that neurons have these properties.[Note: more details on the neuron as ‘programmable information processing unit can be made available.]
Adaptive/evolutionary change in organisms without nervous systems provide an important test of the ‘intelligence in biological systems’(IBS) hypothesis. Darwinian evolution(DE) suggests one explanation of adaptive/evolutionary change in such systems, IBS suggests a dramatically different process of change. DE suggests that genes store the information that defines an organisms, and over time mutation and selection operate change this information.
IBS suggests there are ‘programmable information processing units’(PIP’s) in cells. These units (there may be one type of unit or several types of units) would be responsible for operations such as 1)constructing organic chemicals, 2)assembling simple organic chemicals into more complex structures, and 3)coordinating activity of different cells or groups of cells in multi-cellular organisms. Unlike DE, the IBS would suggest these PIP’s can undergo several types of modifications, and that there may be a variety of mechanisms responsible producing the expected changes. The IBS concept can not predict exactly what physical mechanism are involved in cellular intelligence, but it does predict the ‘types of processes’ to be expected, and it predicts the existence of mechanisms that would directly contradict DE. [Note: One ‘interesting’ possibility would be ‘Lamarkian’ mechanisms. It is known that experience can change genetic material (from the study of neurons) and it is known that there are mechanisms by which the cells of multi-cellular organisms communicate. It is therefore at least possible that the ‘adaptive experiences’ of one group of cells could change the genetic coding in reproductive cells. I don’t have any evidence that such mechanisms actually exist, but contrary to popular belief, such mechanisms are possible. The IBS would describe how such mechanisms might work, and thus might make it easier to identify them. Just a thought.]
SUMMARY
DMA’s and LFS’s appear to provide models, simulations and explanations of how creative intelligence works in humans. These models suggest how creative intelligence might work in non-human biological systems. This ‘suggestion’ can be labeled the intelligence in biological systems or IBS hypothesis. The claim that DMA’s and LFS’s could explain cellular intelligence could be viewed as an IBS prediction. The validity of these predictions provide a useful test of the DMA and LFS models.
IP: Logged
|
|
warren_bergerson
Member
Member # 262
|
posted 24. May 2002 09:51
This is the fifth and final posting outlining my approach to modeling creative intelligence. As stated above, it is my claim or hypothesis that all creative intelligence in biological systems can be modeled, simulated and explained in terms of LFS’s operating on DMA’s. I further claim that these models of causal relationships provide the basis for a predictive, fully testable, hard science theory of adaptive/evolutionary change in biological systems.
Given the scope, the complexity, and the number of unconventional features of the proposal being offered, evaluating the validity of the proposal is not a simple task. Leaving the evaluation of evidence for later, I would like to consider some of the implications ‘if’ the proposal offered has merit.
If the offered proposal has merit, then the ID assertion that biological systems show signs of ‘designs created by intelligence’ are fully validated and explained. The often repeated request to produce a testable scientific ID theory will also have been fulfilled.
The adaptive change is even useful to those who support the concept of an external designer. Evidence for an external designer is based on ‘unexplained discontinuities’. As Dembski himself has stated, such discontinuities are relative to a specific set of known or hypothesized naturalistic mechanisms (not relative to all possible naturalistic mechanisms). The adaptive change approach provides a specific and detailed set of naturalistic mechanisms against which such discontinuities could be identified. [No claim is made here that unexplained phenomena must be interpreted as supporting the proposed theory.]
From an historical perspective, the adaptive approach might be viewed as an extension or refinement of Darwin’s teleological theory. The adaptive approach can be characterized as Darwinian evolution with a lot more processes, complexity and power. The adaptive approach can not, however, be interpreted as compatible with or a part of Darwinian theory. The adaptive approach is clearly and unambiguously different from and logically inconsistent with the basic concepts and principles of Darwinian theory. The adaptive approach claims that evolution/adaptive change depend on a variety of mechanisms not included in Darwinian evolution. The adaptive approach even suggests that Darwin’s Natural Selection, while it clearly exists, performs a role very different than the role assumed by Darwin.
The adaptive approach is also clearly incompatible with current Darwin based theories of genetic change. Genes don’t function as originally hypothesized, mutation doesn’t operate as hypothesized, and genetic change doesn’t operate as hypothesized. As viewed here, genetic theory is currently in a state flux. The old Darwin based theories are known to be incompatible with the knowledge currently available, but no one has been able to formulate a viable alternative. Defining genetic change in terms of DMA’s and LFS’s is a dramatic departure from prior genetic theories, but it does appear to be consistent with recent discoveries in genetics.(As I stated earlier, genetic change and cellular intelligence provide a useful opportunity to test the adaptive approach.)
The adaptive approach can be characterized as both 1)a dramatic departure from traditional techniques of constructing scientific theories, and 2)a dramatic return to the fundamental principles of scientific theory construction. This apparent paradox is, IMO, the key to understanding the adaptive approach.
Scientific theories can be defined as mathematical/logical models of causal relationships meeting certain specified standards for consistency with observed data, testability, and reliable, useful predictions. Traditionally, scientific theories have expressed relatively simple, permanent and universal causal relationships. Models such as F=MA and E=MC2 can be described as expressing ‘simple’ permanent and universal causal relationships or laws of nature. The LFS represents a major departure from tradition in that it models a set or collection of causal relationships many of which are not expressed in the permanent and universal format. Attempting to construct a scientific theory or model from complex sets of causal relationships (called paradigms) can be described as a major departure from traditional scientific approaches.
Somewhat surprisingly, at least to me, this rather dramatic change in format results in theories which fit the traditional ‘hard science’ standards for precise definition, testing and reliable predictions. Theories constructed from the LFS appear to ‘fit’ hard science standards much better than traditional life science theories. As viewed here, biological systems involve complex causation. Traditional attempts to force these complex relationships into simple models and simple theories has resulted in seriously flawed scientific theories. By constructing theories which actually model the complexity that exists, it is possible to develop theories which actually meet rigorous hard science standards. [Note: This finding makes intuitive sense if you recognize that ‘the scientific paradigm’ developed from or is grounded in engineering or applied science, not theory construction or pure science. Engineering models have always involved complex causation. ]
As should be obvious, if the adaptive approach and the techniques associated with the adaptive approach prove sound, the approach and techniques will impact essentially all areas of the life sciences.
I would like to thank the moderators for allowing me to post this rather lengthy, yet very sketchy, description of a technique for modeling and analyzing creative intelligence. I am well aware that the generally accepted view is that is not possible to model, simulate, and explain creative intelligence in humans. The suggestion that a model of human creative intelligence might also produce a model, simulation, and explanation of ‘creative intelligence in cells’ must appear, shall we say, at least a bit crazy.
The white man who ‘discovered’ Pike’s Peak in Colorado, is said to have predicted that no human would ever reach the summit. The thousands of tourists who each year drive to the top of the mountain seem to enjoy the prediction almost as much as the view. I clearly have not build the road to the top of ‘modeling creative intelligence’. I hope, however, I can leave you with the impression that there might be a possible ‘naturalistic’ path to the top of ‘intelligent design in nature’.
IP: Logged
|
|
|
Printer-friendly view of this topic