[Date Prev][Date Next][Thread Prev][Thread Next]   [Date Index] [Thread Index] [Author Index

Modeling relations and semantics



Carlos,
 
There was no contradiction between what Tim wrote and what I wrote: nobody could possibly write all there is on any given subject when it comes to human intellectual interaction with the world.
 
Measurement is not only a technological activity. All observations are forms of measurement, having been "encoded" into perceptions which are based on sensory input by our various combined powers of detecting some of these emanations from the world around us (the "ambience"). Observations further become "formalized" measurements as they are translated (encoded) into some sort of record via a particular language. So, for example: as soon as you write down a description of some observation you've made, you've created and encoded a measurement or an entire set of measurements. The description may be simply a prose-based chronological account of what you observed; or it may be a visual representation as your eyes/mind saw it (a painting or sketch); or it may be represented in a more traditional scientific language such as via numbers...
 
These all represent ways of recording information about an observation-- and it's worth noting that each of these ways of encoding will record different information-- different aspects of the observation.  We have created technologies to perceive and measure all sorts of things that our own natural senses can't detect, like radiation or certain wavelengths of light, sound, etc. However, any mode of measurement is going to do a few things: 1) It's going to leave information about the system out. 2.) It's going to create new information, pertaining to the observer/measurer and possibly about other aspects of the measurement mode (the way of encoding information) which will become an innate aspect of the measurement itself. In other words, all modes of encoding/measuring will add information to the measurement that doesn't actually pertain to the system. Some modes of measuring/observing create more new information than others in any given situation or context. Some modes leave more information out than others, depending on the situation and context. And some modes actually destroy that which they measure, which creates vast amounts of new information as part of "the measurement" (E = MC2 comes to mind). 
 
Clearly, one of the skills of science is to know about this aspect of measurement and to assess the modes at one's disposal so that one is able to make intelligent choices about which modes to apply in any given situation and so forth. Science often records the measurement mode as well as the measurement, but often this aspect of measurement/recording is forgotten or ignored. This is why it is so important not to equate the actual system with any set of measurements of it. In other words, any formalized version we create of a natural system is going to be incomplete and it may also be contaminated with new information that does not pertain to the system, itself. Another skill that science has to develop, according to Robert Rosen, is the ability to assess the nature of the system we want to learn about and to know what sorts of measurements to take and which modes of measurement are appropriate-- for the type of system under study. His view was that we need to be able to assess the organizational type, before we can make any other choices, and we need ways to measure the nature of relational aspects in order to do that. He suggested several ways he had come up with for achieving this and he pointed out avenues for many more ways to be developed. We have discussed quite a few of them on the list, in fact. 
 
This is a critical issue, because complex systems require entirely different scientific treatment than simple systems do and they also require different modeling techniques.
 
A model is a collection of various measurements of all sorts, reassembled into a new system that is intended to be analogous to the original. What my father did was to analyze what this process is actually intended to do: to recreate the entailment patterns of the system (at least, insofar as they are part of the aspects being modeled). So, his analysis of the process of creating a good, working model that will commute with the system being modeled was the diagram that you are describing:
RR wrote: "The crucial ingredients are the arrows 2 and 4, which
I call encoding and decoding, respectively. (I have
discussed the anomalous features of these arrows in
more
detail in Life Itself, section 3H.) They do not fit
entirely inside either the object system or the model;
they do not represent entailments, nor are they
themselves entailed. They manifest what Einstein (with
Infeld; 1938:33) once called "free creations of the
human mind," on which he believed science depends.
They introduce an obvious further semantic element
into the model, over and above what semantic (e.g.,
nonformalizable)
features may already be present in the model."
He is pointing out that any effort to model a natural system is going to have to go through this process. The part of the process that creates the model is the encoding process. The part of the process that tests the model against the original system to see if it "commutes" is the decoding process. Are you with me, so far?
 
Now this is where the arrows in his diagram come into play:
 
Step 1.) First, think about the various well-known models employed by science and medicine and industry... maybe it's the use of the laboratory rat as an experimental model of a human being... maybe it's the use of actuarial tables for use in the life insurance industry to predict how expensive you're going to be to insure, based on their predictions of when you're going to die... maybe it's the weather system models, which all disagree with one another on the evening news...
 
Step 2.) The encoding arrow: None of these models tell us how they were created or decided upon. In fact, it would take a lot of detective work to find out and sometimes it's just impossible. but it may be critical information. However, that's not information we can get from the model; it has to come from the modeler/s. Similarly, no natural system tells us how to best model it. That is information that is not in the natural system. We've developed that on our own, as a means to an end; namely to try and learn more about a system by recreating its entailment patterns. "They manifest what Einstein (with Infeld; 1938:33) once called "free creations of the human mind," on which he believed science depends." And so, the process of encoding that is used to generate a model is one that is not entailed by the system and not entailed by the model.
 
Once we achieve the creation of a model, we want to use it to learn more about how the actual system will behave, IF.... ; In other words, we want to predict future causality, by running certain exercises through our model.
 
Step 3.) The decoding arrow: This situation, again, refers to testing the model against the original system. The process for doing so is not information that is part of the model, nor is it part of the actual system. In other words, neither the model nor the actual system ENTAILS the decoding process; the decoding process is another "free creation of the human mind".
 
Encoding and decoding pertain to science. Science is a human creation. Do you see what I'm getting at? It's not that the encoding and decoding arrows are not entailed at all, it's that they are not entailed in the diagram-- not entailed by either the system or the model.
 
Does he believe that the act of measurement is a "free
creation of the human mind" and "introduce an obvious
further semantic element into the model, over and
above what semantic (e.g., nonformalizable) features
may already be present in the model"? Suppose I want
to calculate the translational kinetic energy of a
body, using the simple formula Et = 1/2m*v^2. Is Rosen
suggesting the measurement instrument I use in order
to determine the mass of a body introduce "an obvious
further semantic element"?
You bet. He most certainly IS suggesting that it introduces a further semantic element into the model or the measurement. It absolutely does. "The measurement instrument" and the measurement process, and the aspects of the system you have chosen to measure that will give you a number that you label "mass"....., these are ALL semantic elements that are not part of the system itself. These tell us as much about the science involved as they do about the system. 
 
Judith