Tuesday, 8 October 2013

An example of transdisciplinarity: expert systems and evolutionary cladistics 6. Looking under the hood

Before anything else, it has to be said that cladistics is a classification system and not a type of diagram showing provable genetic inheritance. The selection of characteristics presented, and the order in which they are placed are the creation of a person or persons and are intended to show the pattern of inheritance of such characteristics. As such, cladistics becomes a circular argument. This does not mean that it is wrong or right, it just means that it is a classification system based on observances of what seems to be structural changes over time that are caused by genetic inheritance.

The selection of a characteristic to be included is where problems can first emerge, but we usually have no way to prove that a particular characteristic descends genetically from the previous characteristic in the chart. It thus becomes a matter of faith -- something that Wolfgang Pauli would have severely criticized. So let us envision some sort of cladistic chart that follows the methods used in real examples and then break it. If we pick something from conchology -- the study of sea shells, the task becomes very easy. The classification of sea shells is much like any other sort of biological classification, in that it follows the Linnaean class, order, genus, species, and variety. The determination is made on the characteristics of the shell. Conchology studies shells and the creatures who live inside those shells are sorted by similarities and differences in the shell structure and appearance. However, there is another subject that studies these same animals, and that is malacology. You can take two examples of shells that are classified either as varieties of each other, or closely related species because of their shell features (conchology) and then study the mollusk inside (malacology). Remember, these species are biologically classified by their shell features. I once spoke to a malacologist who was looking at two "closely related" examples of the family: Cypraeidae (cowries). Upon inspecting the mollusks living inside these shells, he found that the soft tissue differences between the two were so considerable that no one would have called them "closely related' had they ignored the shell completely. Nature can produce two virtually identical structures that have no immediate genetic connections whatsoever.

If we place enough time between two of these features then the problem becomes obvious: a whale, for example, genetically descends from mammals that once walked on the land, yet all mammals descended, genetically, from sea life much earlier than that. As so much time has passed between the whale ancestors first leaving the sea and then returning to it, we cannot say that a specific feature of some sea life is always passed to future generations through a genetic process. Let's take some sort of streamlining -- something that can make a fish -- or a whale, move quicker through the water. As creatures develop such characteristics it gives them a greater chance of survival in that arena, than their relatives who do not develop such characteristics. Another creature might depend more on ferocity, poison or camouflage for survival against predators. We cannot take the fish-like streamlining of the whale and say that it descended, genetically, from a fish we are looking at, yet both whales and blue fin tuna have such evolved features.

 I have mentioned nature's habit of using shortcuts, and an expert system both uses shortcuts and has a cladistic structure. The difference is mainly that we invent the expert system, but the evolutionary cladistic chart is drawn through the observation of characteristics that are assumed to be hereditary. So let us start by right-clicking on this link and selecting "open link in new window". This will open the paper version of my expert system and then I can walk you through its creation and also link to the on line version that hides its inner structure so that the experience of using it is similar to that of using a cladistic evolutionary chart -- where, of course, the inner workings are unknown. Remember, I did not build this expert system to determine the order of the coin dies. I already knew that -- with the exception of  a few instances where there were no evolutionary features in adjacent dies which I included by placing their variable sequence within a parenthesis thus: 1, 2, 3, (4, 5, 6) 7, 8, 9. -- the order given of 4, 5 and 6 is necessarily arbitrary -- it could be any other combination like, (6, 4, 5) but any one of them comes after 3 and before 7 because all of the numbers not contained within parentheses are in exact chronological order as determined by overlapping evolutionary features.

At the top of the chart, I ask the question "Are there reins between the pony's head and chest" and show a diagram.

On line version:


Normally, you would try to divide the data into two equal parts at this stage. I could better have done that by picking the next question on the right: "Is driver's head like this?" which selects out my series Z which is a "special case", but in both of the possible second questions I selected out "the strange ones" -- "Class VI, Group A" is the rarest of all types and hardly ever seen. Series Z is represented by approximately half of the specimens in the large hoard that I used in my study (54.4% of 9,254 coins), but it only mimics an evolutionary process. I wanted to direct the main flow of the chart toward areas that were closely linked in the previously discovered "genome". I also embedded the earlier classification system which refers to classes even though it is confusing by having the chronological order: VI, V, IV, I, III, II (Colbert de Beaulieu).

To find the "fatal flaw" in using cladistics to determine evolutionary patterns follow through the next links to the on line expert system while comparing each step with the paper chart:


The final link asks if there is a small head above the pony. If you say "yes" the coin belongs to Group L. However, if you happened to have a Coriosolite coin in your hand, and it has a small head above the pony, and then you glance at the paper chart and see the question, you would answer "yes" without going through the system steps -- this actually happened once with someone. Sadly, though, Groups H through L all have this small head above the pony. It is just that all but Group L were eliminated because of other features earlier in the expert system.

What this demonstrates is that the structure of the expert system cannot be favorably compared to a genome. Instead, it utilizes the short cuts that are prevalent in evolution. The expert system seeks only to find the shortest navigation to the solution just as nature allows a series of consecutive thoughts to be combined in an immediate reaction to a stimulus -- the animal avoids being eaten by the predator and lives to pass those genes to its progeny.

Wolfgang Pauli was able to detect a, then unknown, feature of epigenetics which  delivers to an organism in stress, a bundle of pro survival features which might avert the problems. The numbers had told him that random genetic mutations could not explain the time line assigned to them and that some sort of pro-survival evolutionary mechanism thus had to exist.

When you look at a cladistic chart depicting evolution, there is no possible way to detect the existence of such short cuts that might "pose" as being genetic links. Yet an expert system has exactly the same structure but will always include such short cuts by its very nature. Biology, too, uses short cuts in a myriad of ways, some of them very similar to those used in expert systems, or perhaps even in Java programming language -- we can pick any subject that has been invented by us to find parallels in subjects where the details of authorship are absent, but otherwise, the known structures are the same. Each enters the transdisciplinarity state at the T expression: T1, T2, T3, and so on.

This ends the series. I'm sure I will come up with something new tomorrow!

No comments:

Post a Comment