In my experience, in scientific discussions, particularly those with strong world view implications, the use of Occam’s Razor to “settle” questions of dispute is fraught with problems.
Occam’s Razor, sometimes called the Principle of Parsimony, is used in inductive logic to prefer a simpler hypothesis over a complex one.
One of the best discussions of this principle is an article by Roald Hoffmann et al. So why is the use of this logical test often a problem and why is it frequently abused? In the first place, the users of Occam’s Razor often forget that this is an extra evidential test. That is to say if one invokes Occam’s Razor to distinguish between hypotheses (or two theories), one has already conceded that both hypotheses explain all of the available data and so one needs to look elsewhere for the distinguishing test. Thus one should select a third order polynomial to fit a curve even though a fourth order fits the curve as well since the third order polynomial is the simpler function (i.e. fewer parameters) that explains all of the data. However since Occam’s Razor is an extra evidential test, the next data point you measure may force you to switch to a fourth order polynomial. At best you have established a temporary logical preference, always waiting for the next data point to make you change your mind.
A second problem that often arises has to do with defining “simpler explanation”. The only sure fire test that one hypothesis is simpler than another is satisfied if one hypothesis is a subset of the other as the set of all third order polynomials is a subset of all fourth order polynomials. Often the simplicity is in the mind of one of the debaters when in fact two completely different hypotheses are being compared.
The third and perhaps most serious problem occurs when Occam’s Razor is applied to discount data in order to leave a preferred theory intact. Consider a thought experiment where an observer reports seeing a ghost. A skeptic analyzes the situation and shows that in a controlled experiment, a similar combination of light and shadow can fool some people into believing they have seen a ghostly figure when none is present. Therefore the skeptic says quid est demonstratum and claims to have shown that the first observer has not seen a ghost and has merely been fooled into seeing a ghost. If you ask the skeptic “why?” and “how do you know?”” the answer will be “Occams’s Razor”. Disbelieving the anecdotal data of the first observer is deemed simpler than revising the skeptic’s view of the reality. In this case Occam’s Razor is used inappropriately to eliminate data (the first observers anecdotal evidence) from consideration.
Finally I want to consider Occam’s Razor and its use as an argument for atheism. Generally the argument is formulated as follows:
(1) There are two competing hypotheses:
a. A Universe and a God Who Created It
b. A Universe
(2) Hypothesis (b) is simpler than (a)
(3) Therefore based on Occam’s Razor (b) is held to be correct since there is no unequivocal evidence for (a).
One could of course point out that arguing there is no unequivocal evidence for (a) is begging the question since all of the discussion is about the validity of evidence for (a), but let’s look at this question from the perspective of Occam’s Razor.
First of all do both (a) and (b) explain all of the data out there? I would say absolutely not. Every anecdote about a personal religious experience is a data point that cannot be explained by (b). So it is inappropriate to apply Occam’s Razor since the argument really centers about what one considers as valid data and what one does not (so often how we define “valid data” will end up determining the world view we end up supporting with our arguments). Occam’s Razor should not be used because we have not passed the evidential test.
What about the question of simplicity. Even here we have a problem. We’re fooling everyone with the word “universe”. Universe does not have a precise meaning and the universe in (b) is not the universe in (a). For example in a previous post [http://peterkazmaier.com/?p=18], I pointed out Steven Weinberg acknowledged that for life to exist in this space-time, some physical constants cannot vary by one part in 10120. He rationalized this amazing coincidence by postulating the existence of multi-verses (an infinite number of universes all with differing fundamental physical and cosmological constants) and so by survivor bias we happen to only know about the one where the constants are consistent with supporting life. The word “universe” is like a variable that has to change to make the hypothesis tenable and so it is no longer possible to define what is simpler since each has its own unique complexities.
In summary Occam’s Razor is at best a temporary tool for comparing hypotheses that explain the same data set. The next data point that is measured may force us to accept the more complicated hypothesis. Think carefully when it is brought out to dismiss an argument or a hypothesis.
Thanks for reading,
Peter
No hay comentarios:
Publicar un comentario
COMENTE SIN RESTRICCIONES PERO ATÉNGASE A SUS CONSECUENCIAS