Sunday, March 20, 2011

A Return To The Invisible College In Science 2

A continuation of the roles and connections between reductionism, science and academic freedom.


From the first article in this series we saw the long way that Science has travelled from its origins.  Some of the principles of research, including the scientific method, as is popularly known, have since then run into some paradoxes.


The Origin of Reductionism
Auguste Comte
Probably, the clearest definition of reductionism is a system of thought that believes that a complex system is nothing more than the sum of its parts.  Reductionism probably had its beginnings with a philosophy, which curiously enough, did not begin in the realm of science, but in the foundations of sociology with a man named, Auguste Comte.  Comte saw the scientific method as replacing any belief in the supernatural.  This of course resulted in critics who have since been labelled antipositivists, such as Max Weber, and Georg Simmel.  This opposing school of thought labelled positivism and scientism.  Scientism is the view that the methods used in natural science (by natural is meant that the universe follows natural laws and not magical or divine laws) is the best way to view not only science itself, but all other human endeavors.  This scientific view of the rest of human studies was thought to eliminate subjective experience from a proper course of inquiry.  This term was of course intended as an insult, and was used to refer to the use of scientific methods from 1910-1950s in all academic university studies.


Cafe Central, Vienna
This positivism, also called Logical Positivism, also thought that all "authentic" knowledge came from experience, sense, and positive verification.  Positivism as a philosophy, evolved into Logical Positivism in the early 20th century in the city of Vienna, Austria.  It was coined by the "Vienna Circle," a group of philosophers from the University of Vienna which met on a regular basis at the Cafe Central in Vienna.  The group consisted of Moritz Schlick, Gustav Bergmann Rudolph Carnap, Herbert Feigl, Phillip Frank, Kurt Godel, Hans Hahn, Tscha hung, Victor Kraft, Karl Menger, Richard von Mises, Marcel Natkin, Otto Nuerath, Olga Hahn-Neurath, Theodor Radakovic, Rose Rand and Frederich Waismann.  Many of the people in thie group would go on to make important discoveries that would affect Science and our modern view of the world.  The most prominent of them being Kurt Godel, who would later almost single-handedly destroy the foundations of Logical Positivism.
Tree of Knowledge by Gregg Henriques
These men believed that all the sciences formed a unified whole, broken up into hierarchical levels of importance with physics usually placed as the foundation of all other sciences.  There were foundational sciences and then, from there, would rise up all other human studies including even politics, hence the name political science.
This idea that all the sciences are unified has not been universally adopted.  Jerry Fodor is the most famous critic of it.
Although many believe that this kind of Positivism has fallen into disuse in the sciences, we do not agree.  We will list some of the tenets of what Hillary Putnam called the "received view."
A focus on science as a product, a linguistic or numerical set of statements;
A concern with axiomatization, that is, with demonstrating the logical structure and coherence of these statements;
An insistence on at least some of these statements being testable, that is amenable to being verified, confirmed, or falsified by the empirical observation of reality; statements that would, by their nature, be regarded as untestable included the teleological; (Thus positivism rejects much of classical metaphysics.)
The belief that science is markedly cumulative;
The belief that science is predominantly transcultural;
The belief that science rests on specific results that are dissociated from the personality and social position of the investigator;
The belief that science contains theories or research traditions that are largely commensurable;
The belief that science sometimes incorporates new ideas that are discontinuous from old ones;
The belief that science involves the idea of the unity of science, that there is, underlying the various scientific disciplines, basically one science about one real world.
This sounds very much to us as the view of most working scientists today.  The wikipedia article on Positivism goes on to say that, "By the end of the twentieth century, nearly every one of those claims or beliefs had been severely criticized or put into question, so much so that they can be regarded now as being untenable, or at least in need of many qualifications and caveats."  Although we believe this statement is true, we do not think that this information has filtered down to the average science lab or science instruction.  Our view is echoed by this same article when it glibly states, that this view is, "...still alive among many scientists and others who are not well-versed in, or knowledgeable about, what has occurred in technical philosophy since the 1950s."
The ultimate expression of this idea of the unity of science is in what has been called the "Theory of Everything" (TOE).  This approach attempts to unify all events in the entire universe to certain basic physical principles, represented by grand unifying mathematical formulas.  From Archimedes to Hawking, scientists have been trying to find this grand unifying theory to no avail.


Paul Feyerabend
Some like the famous philosopher Paul Feyeraband have questioned whether there ever can be any grand unifying theory.  He has gone even further to question the scientific method itself, as to whether it is the only or the best way of understanding nature.  He is on the radical end of the questioning of the scientific method as THE ultimate method for acquiring factual truth.  Here are two videos that read from his classic book Against Method.  These videos may be watched as wished.  But be warned they are dense and intense.  If given a chance, the words will show great truth.  If you cannot see the embedded video here is the link: http://bit.ly/hDzsFf.



The typical view of a reductionist nature is reflected in the view towards "variables."  The view espouses that the closer one gets to the most basic components the less "noise" or variables there will be.  These variables are seen as anomalies that do nit in the model or prediction.  Normally, they are associated with improper measurements in the experiment, or some other form or operator error.  At times, it is thought that with more precise measurements, more understanding of the MOST basic components of the thing being studied, these anomalies will disappear.

The whole point assumes that one does not wish variability in any experiment.  Thus the answer to this problem is seen as a further reductionist approach to be able to understand the missing, detailed components that will eliminate the noise in the experiment.  This is usually displayed in the desire for better instruments of measure, more powerful microscopes, huge colliders, etc.  The noise is seen as a distraction from what is really happening.


In our next part of this series we will mention some of the classic problems which are bringing down a reductionist approach in favor of complex systems.


4 comments:

Anonymous said...

You got great points there, that's why I always love checking out your blog.

My blog:
courtier en pret et Rachat de Credit ficp

Anonymous said...

You got great points there, that's why I always love checking out your blog.

My blog:
courtier en pret et Rachat de Credit ficp

Guillermo Santamaria said...

Monica thanks for the wonderful comments and insightful information. I will look at these videos carefully. It seems you and I are very much in agreement on many things.

Monica Anderson said...

I gave a talk called "Science Beyond Reductionism" at Future Salon a while back. Video available at http://videos.syntience.com/index.html . The alternative to Reductionism is to use Model Free Methods; the life sciences are increasingly doing this. The term Model Free Methods was first used in 1935 by Lionel S. Penrose. In the talk, I use the NetFlix competition as an example of a complex task that can be done both using Models and in a Model Free manner. I also provide a small zoo of Model Free Methods.

I also examine Reduction as a process in my second article for hplusmagazine.com . Its title is "Reduction Considered Harmful" and I make the limited claim that for the purpose of creating a true Artificial Intelligence (AGI) Reduction in the target problem domain is impossible. In the next article I will show that it is also unnecessary, and the fourth article will discuss Model Free Methods as the alternative. As of posting time that second article is still in the publishing queue.

Reductionism has had an excellent run since the 17th century with peaks in the 17th and the 20th centuries. But it happens to be close to useless for what I call "the remaining hard problems" and the switch to Model Free Methods currently underway needs to be recognized as a necessary progression for all disciplines and not just the Life Sciences. Artificial Intelligence Research is (for predominantly historical reasons) both one of the most Reductionist disciplines and the one that most needs to change; the mismatch between problem domain constraints (or lack of them :-)) and methodology in AI research is appalling and fatal and pretty much obvious to any observer outside the field.