|
Author
|
Topic: The Characterization of Intelligent Causation
|
miosim
Member
Member # 4541
|
posted 29. July 2007 22:15
quote:
LE:
… as was also discussed, your explanation does not actually fit the available data…
The standard in real science or hard science is independent testing …
I am not ready to go deep into discussion about diversity of human decision making mechanisms, but I think that the mechanism you are putting forward (LE: “a human needs to use a reliable and highly validated decision making algorithm and they need to be sure that the input data used in the decision process is accurate and reliable”) is too formal and rather is applicable to computers than to humans.
In the Wikipedia, I found a DECISION THEORY that is concerned with how real and ideal decision-makers make or should make decisions, and how optimal decisions can be reached.
Your proposal is similar to Normative and Descriptive type of theory, which is concerned with "identifying the best decision to take, assuming an ideal decision maker who is fully informed, able to compute with perfect accuracy, and fully rational”. This sort of decision making is aimed at finding tools and practical methodologies to help people make better decisions.
However in the “Choice under uncertainty” section it is emphasized that in actual human decision-making peoples “are more focused on changes in their utility states than the states themselves and estimation of subjective probabilities is severely biased by anchoring”. This brings us to the point that in a real life more than one solution to the same problem is proposed and to choose the best one is not a fully rational process. So the only choice is left to accept a solution chosen by a majority (if authority in the group is distributed evenly). If a proposed solution could be tested, only then its correctness can be conformed or rejected, but often, especially in case of complex problems, an initial decision is separated by years from an actual testing. In many cases a solution for a scientific theory or a paradigm cannot be tested at all and theirs fate is fully depends on the subjective opinion of a majority.
[ 30. July 2007, 04:53: Message edited by: miosim ]
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 30. July 2007 08:54
SELECTING THE BEST AVAILABLE THEORY
INTRODUCTION
To return to a topic I started to discuss earlier, if you want to develop useful scientific definitions and useful scientific theories relating to the subject of intelligent causation you have to perform formal scientific analysis of systems and behaviors that clearly exhibit intelligence. The most obvious starting point for such analysis is human beings and intelligent human behavior. More specifically, if you want to formulate useful scientific definitions and theories you need to perform formal scientific analysis of a form of human behavior that clearly involves intelligent causation.
The purpose of this set of postings is to discuss the formal analysis of one type of intelligent causation- “human select the best scientific theory decision making”. There can be little justification for arguing that this behavior does not involve intelligent causation.
GOAL OR PURPOSE OF DISCUSSION
The goal or purpose of the discussion here is to present the results of a particular set of scientific experiments. The goal or purpose is not to necessarily convince anyone of the validity of the conclusions, but rather to provide sufficient detail to permit a properly trained scientist to perform independent experiments to test and confirm (or falsify) reported findings.
The methodologies, concepts and techniques used in the studies discussed here are standard highly validated methodologies, concepts, and techniques or relatively modest variations on established methodologies and concepts. However, as has been established from discussions here and elsewhere, the fact that the methodologies and concepts used are largely ‘standard and highly validated’ does not mean that the methodologies and concepts are known and understood by even a majority of likely readers.
As viewed here, in line with standard scientific practice, the responsibility for learning and understanding established methodologies rests with the reader and reviewer. While someone presenting new materials has a responsibility to clarify points that may be ambiguous, the presenter of innovative scientific concepts can not be held responsible for making readers and reviewers understand what is presented.
Again, the goal of this discussion is to present conclusions developed from the formal scientific analysis of human ‘select the best available scientific theory’ decision making. The intent is to provide sufficient detail so that knowledgeable and trained scientists can perform independent experimentation to test and confirm or falsify the presented conclusions.
CONCLUSIONS
To help the reader better anticipate the rationale behind the experiments described the key conclusions of this analysis are presented at the outset. The following five conclusions can be viewed as key.
1. INTELLIGENT CAUSATION IS OR CAN BE VIEWED AS DYNAMIC
As should be reasonably obvious, intelligent systems learn, adapt, find new solutions and are generally dynamic or changeable. The relevant conclusion here is not that intelligent causation is or appears dynamic. The relevant conclusion is that it is useful and productive to view intelligent causation as dynamic in performing scientific analysis.
It will be noted that many attempts to analyze intelligent causation are based on the assumption that scientific explanations of intelligent causation must be based on the discovery of some permanent force or process or algorithm that could explain the changes produced by intelligence. In direct contradiction of this assumption, the conclusion reached here is that intelligent causation should be viewed and analyzed as dynamic. The scientific requirement for permanent and universal laws or theories can be satisfied even when intelligent causation is viewed as dynamic.
2. POINT IN TIME INTELLIGENT BEHAVIOR (DECISION MAKING) CAN BE USEFULLY MODELED AND ANALYZED BY DETERMINISTIC PROCESSING ALGORITHMS- As anyone familiar with modeling should recognize, the fact that point in time intelligent behavior can be modeled by deterministic algorithms is obvious or trivial. The significant conclusion or finding here is that modeling point in time intelligent causation in this manner is useful and productive.
3. DYNAMIC CHANGES IN INTELLIGENT BEHAVIOR CAN BE USEFULLY MODELED OR SIMULATED BY CHANGES IN PROCESSING PROGRAMS OR ALGORITHMS- Again, the significance here is that this approach is scientifically useful and productive.
4. CHANGES IN PROCESSING ALGORITHMS RESPONSIBLE FOR THE CHANGEABLE OR DYNAMIC FEATURES OF INTELLIGENT CAUSATION CAN BE USEFULLY EXPLAINED OR MODELED AS CAUSED BY EXTERNAL INPUT (CALLED INTELLIGENT AGENCY)- This is arguably one of the key discoveries of the analysis being discussed. This can be separated into two different conclusions. First, dynamic changes in intelligent causation can be explained and modeled in terms of external intelligent causation or agency. The approach or perspective is useful in performing scientific analysis of intelligent causation.
5. THE CHANGES IN INTELLIGENT CAUSATION OR PROCESSING ALGORITHM PRODUCED BY EXTERNAL AGENCY ARE NEVER COMPLETELY EFFICIENT- The fact of inefficiency in intelligent causation should not be particularly surprising. The importance of this finding is that it is very useful in performing scientific analysis of intelligent causation.
6. INTELLIGENT CAUSATION CAN ONLY BE FIT TO FAMILIES OF TELEOLOGICAL THEORIES- Intelligent decision making relating to scientific theories can be fit to a family of teleological theories and no other type of scientific theory fits the data with respect to the intelligent behavior of selecting scientific theories. It appears or is proposed that any type of intelligent causation can be analyzed using the methods being discussed here, any such analysis can be fit to a family of hard science teleological theories and no other type of non-trivial predictive theory will be compatible with the data.
To be continued. Any questions or comments on the above?
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 30. July 2007 09:18
Miosim,
Quote: However in the “Choice under uncertainty” section it is emphasized that in actual human decision-making peoples “are more focused on changes in their utility states than the states themselves and estimation of subjective probabilities is severely biased by anchoring”.
Under ideal or optimal conditions humans are capable of making near optimal decisions. Algorithms for making near optimal decisions are surprisingly well known.
Under non-ideal conditions, people make non-optimal and often highly inefficient decisions. Why? Is it uncertainty? Is it lack of appropriate information? Is it lack of intelligence?
People not infrequently make bad, highly efficient decisions even when the decision is in direct conflict with their self interest, when the information to make efficient decisions is readily available and when they clearly have the intelligence and technical skills to make near optimal decisions. Analysis of “Selecting the best available scientific theories” decision processes illustrates many if not all the factors involved in inefficient decision making.
The challenge is not simply to throw out vague verbal descriptions of ‘why people make stupid decisions’. The scientific challenge is to come up with testable predictive theories that fit and explain both decision making under ideal conditions and decision making under non-ideal conditions. There is a substantial body of data available on efficient and inefficient decision making. The challenge is fitting that data to predictive theories. The data clearly is not compatible with your simple force of intelligence type theory.
Or to put it another way, if you intend to promote the force of intelligence type theory, then you need to explain how it fits the available data on efficient and inefficient human decision making.
IP: Logged
|
|
IF
Member
Member # 1904
|
posted 30. July 2007 18:23
LE,
quote:
But there appear to be factors in addition to knowledge constraints that scare people away from real scientific analysis of intelligent causation.
I think the reason is not fear but, as stated by aiguy, the "fact" that intelligent causation is not properly characterized or defined.
IP: Logged
|
|
miosim
Member
Member # 4541
|
posted 30. July 2007 21:10
quote:
LE:
The data clearly is not compatible with your simple force of intelligence type theory.
What kind of data? What is “the force of intelligence type of theory”?
Let me just repeat what is my hypothesis about. It is about intelligence that is the property of any material elementary particle in form of elementary Problem Solving Ability (PSA). This is the most fundamental property of matter and all other properties that appear to us as physical properties (mass, inertia, forces, energy, etc) are reducible to the particle’s elementary PSA same way, as social phenomena, described in the terms of socio-cultural field, social forces, etc) are reducible to a human intelligence (collection of specific PSA). If matter would loose its intelligence, its physical properties would disappear also
The emergence of matter and its intelligence is outside of the scope of this hypothesis. This hypothesis also doesn’t intend to explain how exactly material systems are solving problems (logically, intuitively, based on the previous experience, etc.)
This hypothesis predicts that systems are capable to accumulate a PSA of its elements into higher level of system PSA. The existence of hierarchical organization of matter from elementary particles to human society is a series of levels of intelligence, which became more and more sophisticated from one hierarchical level to the next.
quote:
LE:
… then you need to explain how it fits the available data on efficient and inefficient human decision making.
According to my hypothesis PSA is defined, as a probability to find a correct answer, therefore any problem solving process could yield two outcomes: correct or incorrect answer. Because PSA value has a range from 0% to 100%, you can express it in the less demanding terms of “Efficient or inefficient human decision making”
quote:
LE:
…the responsibility for learning and understanding established methodologies rests with the reader and reviewer. … the presenter of innovative scientific concepts can not be held responsible for making readers and reviewers understand what is presented…
I think that this is really hurt effectiveness of your communication and undermines the ultimate goal to be understood. We chosen this audience by engaging in this discussion, therefore it is our responsibility for a clear and appropriate to the reader’s level message. Otherwise why do we waste ours and reader’s time?
[ 31. July 2007, 05:43: Message edited by: miosim ]
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 31. July 2007 06:26
IF.
Quote: I think the reason is not fear but, as stated by aiguy, the "fact" that intelligent causation is not properly characterized or defined.
Aiguy's argument was never anything more than "No definition of intelligence is acceptable unless it is acceptable to my personal subjective opinion". And nothing would ever satisfy his requirements because he insisted that the definition of intelligence must both 1) involve a mental component and 2) not involve an undefinable mental component.
Aiguy represents the type of authoritative game playing that characterizes all subjective political efforts to suppress and win a competition against science. The question I am asking is whether the anti-intelligence authoritarian positions are just standard anti-science positions, or if there is something unique about the opposition to scientific analysis of intelligent behavior.
Clearly the scientific analysis of human intelligence and human decision making is a serious and direct threat to the pecking order in every human social structure. It seems likely that humans have some type of bias in favor or defending existing social structures unless or until the flaws have been very clearly demonstrated. Methodologies that would anticipate risks and flaws in social structures and in the behavior of leaders before they actually create a disaster would be opposed.
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 31. July 2007 08:21
RESOLVING THE “IS INTELLIGENCE DEFINED” QUESTION
As anyone involved in the study of either designing computer systems or evolutionary biology is well aware, there is massive and seemingly irrational political resistance to even the idea that goal directed intelligence could be involved in scientific explanations of evolution or the behavior of computers.
As endless discussions on endless websites must convince most people, it is not and will never be possible to resolve the issue based on general discussions and arguments from authority. The vast majority of individuals will support arguments supporting their subjectively desired conclusion with little or no concern for the illogic of the arguments presented. Aiguy, as one simple example, spent a good time here presenting illogical and unsound arguments favoring subjectively rejecting all proposed definitions of intelligence. Despite the rather obvious game playing involved in his arguments, numerous opponents of intelligent causation are happy to site his as a valid authority on the subject. There is nothing particularly unique about aiguy and IF, but human beings in many if not most situations accept arguments of the general type “illogical blither by apparent authority produces desired conclusion” as valid support for a desired conclusion.
There is one and only one available methodology for scientifically ‘resolving’ the issue of defining intelligence. The phrase “scientifically resolved”, as used here means “a consensus among individuals with both the competence to understand hard science analysis and the willingness to accept the results and conclusions of hard science analysis”. A scientific resolution on a definition of intelligent causation is achieved if a proposed definition is useful in formulating the best current predictive hard science theories relating to intelligent causation”. The ‘best current theories” are in turn determined using a formal ‘near optimal’ decision process utilizing formal hard science standards.
Academic scientists appear to be in almost total agreement that their personal beliefs that they mislabel science should not and can not be subject to this type of formal objective scientific review. Why? Do academics like aiguy and IF know or understand that their beliefs would lose under this type of formal analysis? Or is theirs a simple knee-jerk reaction to any test that might threaten what they perceive as their status or position as authority? Or are they defending what they perceive might be a threat to some external authority structure that they view as protecting their current position and status?
I often get the impression that the opponents of intelligent causation are like drones blindly defending their hives. In the world of academic science they sacrifice themselves in blind and often irrational defense of a system that does not and will not come to their defense and will, when the time is right, simply reject them as not-scientifically-competent.
But whatever the motivation behind the opposition to the recognition of goal directed intelligence in science, the method of defeating such opposition involves the use of near optimal select the best theory decision processes.
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 31. July 2007 09:28
Miosim,
Quote: I think that this is really hurt effectiveness of your communication and undermines the ultimate goal to be understood. We chosen this audience by engaging in this discussion, therefore it is our responsibility for a clear and appropriate to the reader’s level message. Otherwise why do we waste ours and reader’s time?
One participates in for the sake of 1) what one hopes to teach and communicate to others and 2) what one hopes to learn from the postings and responses of others. I must admit that I am far more interested in trying to learn about and understand the readers here than I am concerned with attempting to promote my ideas.
There is such a massive and fundamental difference between 1) the communications the occur among productive analysts and systems designers and 2) the communications that occur on websites and with many academics.
One of the key differences is the technical skill and competence of discussion participants. Productive and highly efficient communications exist among productive analysts. One of the key reasons for this productive communications is the success in limiting participation to individuals with the appropriate skills and appropriate commitment to sound and rigorous analysis. [Communications in a business or design environment quickly become unproductive if ‘political types’ without the appropriate technical skills are permitted to participate.]
The first step in creating a productive discussion is therefore to attempt to limit discussion to individuals who either have the appropriate knowledge and skills or who are seriously interested in acquiring them. It is rather surprising, at least initially, what a very large percentage of those interested in scientific topics lack a working knowledge of such fundamental skills as defining and quantifying variables.
But attempting to limit Internet discussion to individuals with appropriate technical skills and knowledge does not appear to be effective. There are quite a number of posters who attempt to fake their level of technical competence or who try to avoid having their technical competence tested or evaluated. The questions of interest are why do these individuals attempt to disrupt discussion and how do they expect to get by with such behavior over an extended period of time.
As to your proposed approach, it is an old and oft discredited approach that keeps coming back in different forms. The current effort to promote your approach is the agent systems approach. Supposedly, collections of simple agents operating under simple processing rules can produce complex ‘intelligent behavior’. In fact such systems are just the latest form of misleading analysis being passed off as science.
IP: Logged
|
|
Luther Von Ruckerson
Member
Member # 6040
|
posted 31. July 2007 11:31
I'm trying to find the logic in these arguments. I'm not having much luck.
There's like some guy out there that started the universe?
IP: Logged
|
|
IF
Member
Member # 1904
|
posted 31. July 2007 14:51
LE,
quote:
Quote: I think the reason is not fear but, as stated by aiguy, the "fact" that intelligent causation is not properly characterized or defined.
Aiguy's argument was never anything more than "No definition of intelligence is acceptable unless it is acceptable to my personal subjective opinion".
I disagree simply because he is not the "decider" in these matters! He is only pointing out the difficulties with the word "intelligence" so that we understand the controversy better. This is similar to what was going on with the word energy with regards to the 17th/18th century's investigators in the context of steam. Other similar problems involved definitions of a gas (aether/spirits, etc), then entropy, and many many similar words/concepts.
quote:
The question I am asking is whether the anti-intelligence authoritarian positions are just standard anti-science positions, or if there is something unique about the opposition to scientific analysis of intelligent behavior.
What anti-intelligence authority? The only thing unique about it is that it is not as hot an issue as you would like it to be. I guess the reason is that it is on the back-burner until someone like Carnot or Boltzmann (per my examples above) comes up with something worthwhile.
IP: Logged
|
|
miosim
Member
Member # 4541
|
posted 31. July 2007 19:32
quote:
LE:
As to your proposed approach, it is an old and oft discredited approach that keeps coming back in different forms ... Supposedly, collections of simple agents operating under simple processing rules can produce complex ‘intelligent behavior ...
No, collection of complex and intelligent agents produces even more complex and intelligent behavior.
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 01. August 2007 16:50
SELECTING THE BEST THEORY PAGE 2
MODELING INDIVIDUAL POINT IN TIME DECISIONS
The starting point for the analysis of ‘select the best theory’ is modeling and simulating individual point in time decisions. As anyone who has studied the subject is aware human decision making is very complex and can be very difficult to simulate. The problems in modeling human decision making is that there significant differences between people in how they make decisions and there are lots of changes over time in how an individual makes decisions. Efforts to fit all these dynamic changes into a single decision making algorithm have not been particularly successful. Focusing on or at least starting with point in time decision making defers the problem of decision making dynamics.
There exists a reasonably well known and commercially used method of modeling and simulating point in time decision making. This methodology, the one used here starts with the concept or assumption that decision making has, or can be modeled and simulated, using the general format or concept or perspective “select the best of the available options using the best available information’.
Using this approach, the analyst needs to identify and define 1) the goal of the decision process, 2) the set of possible options being evaluated, 3) the factors or variables used in making the decision, 4) the quantification rules used to assign values for each combination of factor and option, 5) the weight or significance given to each decision factor and 6) the environmental conditions under which the decision is made. Using this approach, a point in time algorithm can be modeled and simulated using an algorithm of the form “the select the option with the highest value where value is calculated as the sum of the weighted values of the factors used in the decision making.
To clarify, the above methodologies are used to develop mathematical models of individual point in time decision making by individual decision makers. The analysis of human decision makers then becomes a process of analyzing the changes and differences among different point in time decision models.
Again, the above is an established methodology for modeling and simulating human decision making. The methodology is both practical and effective for modeling and simulating point in time decision making. Some rather silly objections to this approach have been raised to the effect that the models are imperfect or “don’t necessarily reflect what is actually happening”. Such views simply reflect ignorance of mathematical modeling concepts and principles. The scientific standard for modeling is not perfection, but rather usefulness in performing analysis. A model is not supposed to be and can’t be identical to the phenomenon modeled. It is also logically impossible for any model to completely represent the phenomenon modeled.
DIVERSITY AND DYNAMICS
Comparing different point in time models reveals a rather incredible degree of diversity and dynamics in human decision making. Different individuals making the same decision in a common environment will make decisions using 1) different sets of goals, 2) considering different sets of available options, 3) using different input sources with different input values, 4) using different decision factors (assigning different weights or significance to different decision factors. Similar levels of diversity or dynamics can be seen if a single decision maker is observed making the same type of decision over a period of time. Again the amount of diversity and dynamics observed in human decision making is somewhat unexpected.
DEPENDENCE ON OTHERS
A second key feature of human intelligence and human decision making demonstrated by analyzing point in time decision models is the dependence or reliance on others for the information used in decision making. People make large numbers of decisions. To thoroughly analyze all the data relevant to these decisions would require far more time and analysis than could be performed by an individual. In order for it to be practical for an individual to make the large number of complex decisions associated with human social behavior, individuals rely heavily on information generated by others and readily available for use in decision making.
For many of the more technically complex decisions, people delegate essentially the entire decision making process to specialized decision making experts. In crossing a bridge, we routinely rely (implicitly) on the opinions of the engineers who design and test bridges. We delegate governmental decisions to groups of elected representatives.
The extent to which human decision making depends on information communicated between people is significant for a number of reasons. First, this means that some portion of the factors or variables involved in human decision making (intelligent causation) arise outside the individual mind or brain. Second, it means that at least potentially human decision making (and thus intelligent causation) can be at least influenced or partially controlled (engineered) by external control and manipulation of input used in decision making. Third, and most immediately relevant, the heavy reliance of human decision making on information communicated by people, means that the a significant portion of the causal variables associated with human decision making (human intelligent causation) can be readily observed and scientifically measured.
Contrary to initial expectations, and as a direct result of the social or group features of human decision making, human decision making turns out to be easier to observe and study than many apparently simpler forms of intelligent causation.
It should once again be noted that the techniques of 1) modeling point in time decision making and 2) performing analysis by comparing differences in these point in time models is a reasonably well documented and validated methodology. Anyone interested in gaining some first hand experience with this methodology would benefit from developing models and analyzing decision making the widely recognized decision making phenomenon of diversity and reliance on information generated by others.
To be continued. Any questions?
IP: Logged
|
|
LifeEngineer
Member
Member # 3446
|
posted 01. August 2007 16:53
Miosim,
Quote: No, collection of complex and intelligent agents produces even more complex and intelligent behavior.
Nonsense. The behavior of a brain is more intelligent and complex than the behavior of a neuron. The behavior of a biological system is more complex and intelligent than the behavior of any individual in the biological system. Human social behavior is vastly more complex and intelligent than the behavior or any single individual.
IP: Logged
|
|
Daniel Smith
Member
Member # 3004
|
posted 01. August 2007 18:49
miosim:
quote:
This hypothesis predicts that systems are capable to accumulate a PSA of its elements into higher level of system PSA. The existence of hierarchical organization of matter from elementary particles to human society is a series of levels of intelligence, which became more and more sophisticated from one hierarchical level to the next.
This hypothesis seems to run on the assumption that a human society is more complex and efficient (sophisticated) than any other "lesser" system. The truth is: Human societies are not nearly as complex and efficient as cellular systems. If cellular systems functioned at the level of efficiency shown by human societies, most organisms would self destruct!
I work in a manufacturing plant run by humans. We often run into problems of communication, tardiness, laziness, lack of training, etc., all of which cause our products to be late, defective or misdelivered. Cellular systems, on the other hand, communicate virtually flawlessly, with just the right "product" delivered at just the right time and to just the right place. Human societies could learn a lot from biological systems! quote:
In any case, a human society is emerging and developing as a result of self organization (down-top mechanism) and all “tremendous coordination among its systems doesn’t need “an infinite (all knowing) intelligence”.
That's precisely because of the above stated reason: Human societies do not require infinite intelligence to work because they are extremely clumsy and inefficient when compared to cellular systems. In fact, I'd argue that the amount of sophistication exemplified by cellular systems vs. human societies gives us a direct comparison between the level of intelligence available to us humans and the higher level of intelligence required for the planning and implementation of life's systems.
IP: Logged
|
|
miosim
Member
Member # 4541
|
posted 01. August 2007 21:17
quote:
miosim:
… collection of complex and intelligent agents produces even more complex and intelligent behavior.
LE:
Nonsense. The behavior of a brain is more intelligent and complex than the behavior of a neuron. The behavior of a biological system is more complex and intelligent than the behavior of any individual in the biological system. Human social behavior is vastly more complex and intelligent than the behavior or any single individual.
Do you think that your examples are in full agreement with my statement?
IP: Logged
|
|
|
Printer-friendly view of this topic
