Uncommon Descent Serving The Intelligent Design Community

# Phillip Johnson and Bayesian Priors

Share
Flipboard
Print
Email

For scientific materialists the materialism comes first; the science comes thereafter.  We might more accurately term them “materialists employing science.”  And if materialism is true, then some materialistic theory of evolution has to be true simply as a matter of logical deduction, regardless of the evidence.  That theory will necessarily be at least roughly like neo-Darwinism, in that it will have to involve some combination of random changes and law-like processes capable of producing complicated organisms . . .

Phillip Johnson

Johnson’s observation came to mind when I read HeKS comment to a prior post.  That comment recasts Johnson’s observation in terms of Bayesian priors.  It would be cumbersome to put everything in block quotes.  All that follows is HeKS:

“If the probability that mind is responsible for some effect is preemptively set at zero for methodological purposes and only one naturalistic explanation, in its rough outline, is logically possible, then is the probability that the naturalistic explanation is correct actually 1? And does it necessarily remain 1 in spite of what the evidence may tell us?”

I’m not looking for a formula to make a calculation. I’m asking a philosophical question about the effect of presupposition when you are trying to explain some effect where two causes seem logically possible, but the one that is actually known to be causally adequate is ruled out by an a priori philosophical or methodological presupposition.

That said…

p(theory) = [p(theory|naturalism) * p(naturalism) ] + [p(theory|not-naturalism) * p(not-naturalism)]

On Methodological Naturalism (MN), it seems this would be:

P(theory) = [P(theory|naturalism) * 1] + [P(theory|not-naturalism) * 0]

Without knowing the value of P(theory|naturalism) (or, the probability that the theory is true given naturalism is true), we can’t give a final probability percentage. We can, however, recognize that [P(theory|not-naturalism) * P(not-naturalism)] = 0.

On MN, it doesn’t matter what the probability is that the theory is true given naturalism is false, because the probability of naturalism being false is already determined to be 0.

of course, if a naturalistic theory explains data well then it makes p(naturalism) higher:

where

p(observations|naturalism) is p(observations|theory) (or perhaps summed over all naturalistic theories that might explain the observation)

Wouldn’t that be…

p(naturalism|observations) = p(observations|naturalism) * P(naturalism) / [p(observations|naturalism) * p(naturalism)] + [p(observations|not-naturalism) * p(not-naturalism)]

…or am I missing something?

In any case, the problem here is that on MN, P(naturalism) = 1

Methodological Naturalism is the methodological implementation of a philosophical presupposition, not the conclusion of a Bayesian probability calculation. But if we incorporate MN into a Bayesian calculation for the probability that the ‘only viable naturalistic theory’ (ovnt) for some effect is correct, the result is entirely predictable.

P(ovnt) = [P(ovnt|naturalism) * P(naturalism)] + [P(ovnt|not-naturalism) * P(not-naturalism)]

Becomes…

P(ovnt) = [1 * 1] + [x * 0]

[Given naturalism is true (i.e. P(naturalism) = 1), the probability that the only viable naturalistic theory is true is, by logical necessity, 1 (i.e. P(ovnt|naturalism) = 1). Meanwhile the probability that the theory would be true on not-naturalism is ultimately irrelevant because the probability of not-naturalism is 0]

So, this becomes…

P(ovnt) = 1 + 0

Becomes…

P(ovnt) = 1

It doesn’t seem to me that this is the kind of scenario that Bayes Theorem was intended for.

On MN, if we find a singularly viable naturalistic theory to explain something, a Bayesian calculation that takes MN into account will always reveal to us with 100% certainty that the theory is correct.

You obviously get the exact same result if you replace ‘only viable naturalistic theory’ (ovnt) with ‘some naturalistic theory’ (snt). We get certain conclusions we can hold with certainty without the need for any supporting evidence at all. We only need to deduce them from the principle of Methodological Naturalism.

For example, on MN, what is the probability that the Origin of Life came about through purely naturalistic causes without any role played by a mind? Simple. The probability is 1. On MN we can know this with complete certainty even if we never figure out how it could possibly happen.

I’m no expert in Bayesian analysis, but this seems philosophically problematic to me.

@wd400 #17
I really don’t think there is any problem at all with saying “theory x is the best science to explain some part of nature, but it’s not every good”.
Neither do I, at least not under normal circumstances, if what you mean by "best science" is "best naturalistic explanation". That's not the problem. The problem, or one of the problems, is that no matter what an objective consideration of the evidence may indicate, Methodological Naturalism makes it impossible to say, "This artificial (teleological) explanation currently provides a better explanation than the currently best naturalistic explanation." But this is not the only problem...
For instance, at the moment there are not satisfactory scientific accounts of the origin of life on earth.
Correct. There is not. But here's the issue: On Methodological Naturalism, can we be certain that the origin of life is explicable purely by reference to natural causes? If you say no, you're mistaken. As I've said, Methodological Naturalism is simply applied Philosophical Naturalism. Methodological Naturalism assumes a priori that naturalism is true, which means natural (non-artificial) causes are all that exist. Hence, anything that exists is necessarily explicable by reference to the only causes that exist, which are natural ones. Hence, since life exists, its origin is necessarily explicable purely by reference to natural causes and there can be no possible role for artificial (teleological) causes because they do not exist. These are logically necessary deductions from Methodological Naturalism. The evidence is ultimately irrelevant because if it is consistent with a naturalistic explanation it will be claimed in support of the naturalistic explanation, but if it points away from any naturalistic explanation or is inconsistent with naturalism itself, it will simply be ignored, labeled as illusory, or be made to fit a naturalistic picture through the use of ad hoc auxiliary hypotheses.
Science will keep working on that probably, looking for naturalistic explanation as they are the only testable ones.
October 3, 2014
October
10
Oct
3
03
2014
11:31 AM
11
11
31
AM
PDT
There are a lot of questsions, I hadn't noticed yours. I don't think the origin of life requires a miracle, I don't think it's impossible for there to be a testable theory for the origin of life that involves "mind interacting with matter", I just think no one has proposed anything that gets close to that.wd400
October 3, 2014
October
10
Oct
3
03
2014
10:38 AM
10
10
38
AM
PDT
WD, I noticed that you've ignored my question at 13.Barry Arrington
October 3, 2014
October
10
Oct
3
03
2014
10:24 AM
10
10
24
AM
PDT
HEKs, I really don't think there is any problem at all with saying "theory x is the best science to explain some part of nature, but it's not every good". For instance, at the moment there are not satisfactory scientific accounts of the origin of life on earth. Science will keep working on that probably, looking for naturalistic explanation as they are the only testable ones. In the mean time, reasonable people don't conclude any of the existing scientific theories are True, since that's all naturalism has given us so far. They either see it as evidence of a supernatural origin of life, or that science doesn't have a good answer to this question (yet). (in fact, I guess it's always evidence for a supernatural origin, though many will think it not enough evidence to move their prior very far).wd400
October 3, 2014
October
10
Oct
3
03
2014
09:56 AM
9
09
56
AM
PDT
wd400:
there is nothing supernatural about archaeology or SETI
ID doesn't require the supernatural. BTW the design is natural and as such can be detected and studied.Joe
October 3, 2014
October
10
Oct
3
03
2014
03:14 AM
3
03
14
AM
PDT
wd400, What is the probability that a naturalistic explanation is true if naturalism is false?Mung
October 2, 2014
October
10
Oct
2
02
2014
03:52 PM
3
03
52
PM
PDT
@wd400 #7
I really don’t know why people are having such a hard time with this…
I don't think anybody is having a hard time with this other than you. You don't seem to understand what Methodological Naturalism actually is. As I said in the other thread in my latest comment, Methodological Naturalism is simply applied Philosophical Naturalism. Methodological Naturalism does, indeed, say, "Assume Philosophical Naturalism is true when doing science." It says, "Assume naturalism is sufficient to explain any observed effect." MN does not leave science in a position to say, "This observable effect in nature does not seem to be capable of being explained under naturalism (i.e. without reference to an intelligent agent) so we shall simply refrain from making any assumptions that it is capable of being explained under naturalism.' Rather, it assumes from the outset that all observable natural phenomena are capable of being fully explained by reference to purely naturalistic causes, regardless of whether we ever discover those naturalistic causes or not. In other words, MN gives science the ability to employ an endless naturalism-of-the-gaps argument that is never threatened by evidence for intelligent agency, no matter how overwhelming the evidence for that intelligent agency might be. Apparently you don't see any problem with that. Most of the rest of us do.HeKS
October 2, 2014
October
10
Oct
2
02
2014
03:22 PM
3
03
22
PM
PDT
WD@11:
there is nothing supernatural about archaeology or SETI.
I infer there is an unspoken "as opposed to ID" at the end of this sentence. Thus, you believe for an agent to design a living thing a miracle would have to occur. This leads to the obvious question, what miracle do you believe is necessary for life to occur?Barry Arrington
October 2, 2014
October
10
Oct
2
02
2014
01:53 PM
1
01
53
PM
PDT
wd400 you claim,, "there is nothing supernatural about archaeology or SETI" If you really believe that, Dr. Nelson wants to talk to you,, Do You Like SETI? Fine, Then Let's Dump Methodological Naturalism - Paul Nelson - September 24, 2014 Excerpt: "Epistemology -- how we know -- and ontology -- what exists -- are both affected by methodological naturalism (MN). If we say, "We cannot know that a mind caused x," laying down an epistemological boundary defined by MN, then our ontology comprising real causes for x won't include minds. MN entails an ontology in which minds are the consequence of physics, and thus, can only be placeholders for a more detailed causal account in which physics is the only (ultimate) actor. You didn't write your email to me. Physics did, and informed you of that event after the fact. "That's crazy," you reply, "I certainly did write my email." Okay, then -- to what does the pronoun "I" in that sentence refer? Your personal agency; your mind. Are you supernatural?,,, You are certainly an intelligent cause, however, and your intelligence does not collapse into physics. (If it does collapse -- i.e., can be reduced without explanatory loss -- we haven't the faintest idea how, which amounts to the same thing.) To explain the effects you bring about in the world -- such as your email, a real pattern -- we must refer to you as a unique agent.,,, some feature of "intelligence" must be irreducible to physics, because otherwise we're back to physics versus physics, and there's nothing for SETI to look for.",,, http://www.evolutionnews.org/2014/09/do_you_like_set090071.htmlbornagain77
October 2, 2014
October
10
Oct
2
02
2014
01:24 PM
1
01
24
PM
PDT
there is nothing supernatural about archaeology or SETI. As I answered in the other thread, the limit on the supernatural is only a extension of the more general rule that, in order to work, science must limit itself to testable hypotheses. I don't see how its possible to make testable claims when you let supernatural causation slip inwd400
October 2, 2014
October
10
Oct
2
02
2014
12:50 PM
12
12
50
PM
PDT
WD@7:
we limit science to naturalistic explanations.
Take your pick of examples: archaeology, SETI, forensics, etc. Does your methodological naturalism allow a scientist to infer that an artifact found in the real world may have been caused by purposeful intelligent action, as opposed to purely natural causes like chance and necessity?
You must have overlooked the question, because you did not answer it. Now's your chance, because it seems to me like you've just relegated archaeology, SETI, forensics and cryptography to the realm of "non-science."Barry Arrington
October 2, 2014
October
10
Oct
2
02
2014
12:14 PM
12
12
14
PM
PDT
I was just about to leave the site after posting, but saw that post from wd400:
we limit science to naturalistic explanations.
I really wonder who has the authority to limit science? Isn't science supposed to be the search for the "truth"? If observations defy (or has the potential to defy) the known natural laws, then why SHOULD we limit science to naturalistic explanations? As I said above, as long as observations are repeated, that's fine. (BTW, why a "natural law" itself is taken to be "natural" is another story. Some "thing" called law exists, outside time and space, and it's natural.. Yeahh, sure :)) However, for unique (especially historical) events, if natural laws are not (or it is possible that they may not be) sufficient, then why should we limit ourselves to naturalism?CuriousCat
October 2, 2014
October
10
Oct
2
02
2014
11:00 AM
11
11
00
AM
PDT
Thanks for the very interesting post Barry. I think that p(N) = 1 may not be the only problematic assumption here. I'll try to explain below. I hope I'm not mistaken. Let's consider the equation you have written: (Here, N = Naturalism, O = Observation, N' = not naturalism) p(N|O) = p(O|N) * P(N) / [p(O|N) * p(N)] + [p(O|N') * p(N')] It may be easier to analyze this when we write in the simplified form as p(N|O) = p(O|N) * P(N) / p(O) One of the interpretations of Bayes' rule is that it "maps" p(N) to p(N|O) via multiplying with p(O|N) / p(O). If we can make repeatable observations (ex. if you do not hold the ball, it will fall down), under similar conditions, p(O) will tend to unity. And if we have a naturalistic theory to explain this observation (e.g. Newton's eqn. of motion and gravity), then p(O|N) -> 1, which means that if naturalism is "true" under the given conditions, you are almost guaranteed to make this observation. Hence, the ratio of p(O|N) to p(O) will be equal to unity. So, p(N|O) ~ P(N). That result shows that if p(N) = 1, then p(N|O) = 1, as you pointed out. Moreover, it also shows that N and O are independent events from each other!!! [ p(N|O) = p(N) is the definition of statistical independence] So how can this "belief" about naturalism be refuted? p(N) should be decreased. How? The coefficient p(O|N) / p(O) should be lowered. Since p(O) consists of summation of two terms, and since we cannot change the belief that p(N) -> 1, so the only remaining strategy is to decrease p(O|N) with respect to p(O). (Here I should add that increasing p(O|N') using explanations from non naturalistic theories would not work, since p(N') is assumed to go to 0). How is this possible that p(O|N) will be decreased? By showing that naturalistic theories cannot explain observations, e.g. cambrian explosion, irreducible complexity, fine tuning of the universe, etc. So, p(N|O) will be less than p(N), and N and O become correlated again. As a result, for a believer in naturalism (p(N) = 1), it would be difficult to support naturalism in the light of the existing observations (p(N|O) less than unity).CuriousCat
October 2, 2014
October
10
Oct
2
02
2014
10:49 AM
10
10
49
AM
PDT
FWIW, here is the central mistake in the above. Given this equation P(ovnt) = [P(ovnt|naturalism) * P(naturalism)] + [P(ovnt|not-naturalism) * P(not-naturalism)] It would be true that P(ovnt|naturalism) =1 , but methodolocical naturalism doesn't require that p(natrualism) = 1, only that we limit science to naturalistic explanations. If, having applied that limit,we find the ony naturalistic explantion (or, more sensibly, the sum of all possible naturalistic explanations) is improbable then that's evidence for non-naturalism. I really don't know why people are having such a hard time with this...wd400
October 2, 2014
October
10
Oct
2
02
2014
10:17 AM
10
10
17
AM
PDT
Also of interest to this topic of Methodological Naturalism, i.e. the apriori assumption of materialism as true in science, is that the experiments of quantum mechanics have a consistent history of showing local realism, i.e. materialism, to be false.
Bell-inequality violation with entangled photons, free of the coincidence-time loophole - Sept. 2014 Abstract In a local realist model, physical properties are defined prior to and independent of measurement and no physical influence can propagate faster than the speed of light. Proper experimental violation of a Bell inequality would show that the world cannot be described with such a model. Experiments intended to demonstrate a violation usually require additional assumptions that make them vulnerable to a number of “loopholes.” In both pulsed and continuously pumped photonic experiments, an experimenter needs to identify which detected photons belong to the same pair, giving rise to the coincidence-time loophole. Here, via two different methods, we derive Clauser-Horne- and Eberhard-type inequalities that are not only free of the fair-sampling assumption (thus not being vulnerable to the detection loophole), but also free of the fair-coincidence assumption (thus not being vulnerable to the coincidence-time loophole). Both approaches can be used for pulsed as well as for continuously pumped experiments. Moreover, as they can also be applied to already existing experimental data, we finally show that a recent experiment [Giustina et al., Nature (London) 497, 227 (2013)] violated local realism without requiring the fair-coincidence assumption. http://journals.aps.org/pra/abstract/10.1103/PhysRevA.90.032107
So would quantum mechanics not be considered 'science' to the methodological naturalists since it disconfirms his notion that only material causes are in play?bornagain77
October 2, 2014
October
10
Oct
2
02
2014
06:43 AM
6
06
43
AM
PDT
Of supplemental note to the Wigner 'consciousness' quote I cited in post 3,,, “It will remain remarkable, in whatever way our future concepts may develop, that the very study of the external world led to the scientific conclusion that the content of the consciousness is the ultimate universal reality” – Eugene Wigner – (Remarks on the Mind-Body Question, Eugene Wigner, in Wheeler and Zurek, p.169) 1961 – received Nobel Prize in 1963 for ‘Quantum Symmetries’ It is interesting to note that many of Wigner's insights have now been verified and are also now fostering a 'second' revolution in quantum mechanics,,, Eugene Wigner – A Gedanken Pioneer of the Second Quantum Revolution - Anton Zeilinger - Sept. 2014 Conclusion It would be fascinating to know Eugene Wigner’s reaction to the fact that the gedanken experiments he discussed (in 1963 and 1970) have not only become reality, but building on his gedanken experiments, new ideas have developed which on the one hand probe the foundations of quantum mechanics even deeper, and which on the other hand also provide the foundations to the new field of quantum information technology. All these experiments pay homage to the great insight Wigner expressed in developing these gedanken experiments and in his analyses of the foundations of quantum mechanics, http://epjwoc.epj.org/articles/epjconf/pdf/2014/15/epjconf_wigner2014_01010.pdfbornagain77
October 2, 2014
October
10
Oct
2
02
2014
05:33 AM
5
05
33
AM
PDT
BA & HeKS: Imposing an a priori commitment subverts any Bayesian revision calculation. All you have done in the end is to spell it out in that language, up to a possible typo or two. If someone were to come to a corporate board with such a calculation in an investment analysis, he would be dismissed. (Of course, similar question-begging occurs all the time, just not so blatantly.) KFkairosfocus
October 2, 2014
October
10
Oct
2
02
2014
04:17 AM
4
04
17
AM
PDT
One thing that is strange about the word naturalism in the term Methodological Naturalism (MN) is that consciousness is considered non-natural in MN. Only matter and energy are considered 'natural'. A methodological naturalist could just as well say, 'if I cannot physically touch it, see it, taste it, smell it, or hear it, it is not real and/or 'natural' to me.' Yet the methodological naturalists forgets that the very first thing he is presupposing, in his argument that only material is 'real', is a subjective, conscious, 'I' that is able to do the sensing of the material world in the first place.
What drives materialists crazy is that consciousness cannot be seen, tasted, smelled, touched, heard, or studied in a laboratory. But how could it be otherwise? Consciousness is the very thing that is DOING the seeing, the tasting, the smelling, etc… We define material objects by their effect upon our senses – how they feel in our hands, how they appear to our eyes. But we know consciousness simply by BEING it! - APM - UD Blogger
Dividing the world into that which can be sensed and that which can do the sensing is known as the measurement problem in quantum mechanics
How observation (consciousness) is inextricably bound to measurement in quantum mechanics: Quote: "We wish to measure a temperature. If we want, we can pursue this process numerically until we have the temperature of the environment of the mercury container of the thermometer, and then say: this temperature is measured by the thermometer. But we can carry the calculation further, and from the properties of the mercury, which can be explained in kinetic and molecular terms, we can calculate its heating, expansion, and the resultant length of the mercury column, and then say: this length is seen by the observer. Going still further, and taking the light source into consideration, we could find out the reflection of the light quanta on the opaque mercury column, and the path of the remaining light quanta into the eye of the observer, their refraction in the eye lens, and the formation of an image on the retina, and then we would say: this image is registered by the retina of the observer. And were our physiological knowledge more precise than it is today, we could go still further, tracing the chemical reactions which produce the impression of this image on the retina, in the optic nerve tract and in the brain, and then in the end say: these chemical changes of his brain cells are perceived by the observer. But in any case, no matter how far we calculate -- to the mercury vessel, to the scale of the thermometer, to the retina, or into the brain, at some time we must say: and this is perceived by the observer. That is, we must always divide the world into two parts, the one being the observed system, the other the observer. In the former, we can follow up all physical processes (in principle at least) arbitrarily precisely. In the latter, this is meaningless. The boundary between the two is arbitrary to a very large extent. In particular we saw in the four different possibilities in the example above, that the observer in this sense needs not to become identified with the body of the actual observer: In one instance in the above example, we included even the thermometer in it, while in another instance, even the eyes and optic nerve tract were not included. That this boundary can be pushed arbitrarily deeply into the interior of the body of the actual observer is the content of the principle of the psycho-physical parallelism -- but this does not change the fact that in each method of description the boundary must be put somewhere, if the method is not to proceed vacuously,,," John von Neumann - 1903-1957 - The Mathematical Foundations of Quantum Mechanics, pp.418-21 - 1955 http://www.informationphilosopher.com/solutions/scientists/neumann/ The Measurement Problem in quantum mechanics – (Inspiring Philosophy) – 2014 video https://www.youtube.com/watch?v=qB7d5V71vUE
Quote:
"It will remain remarkable, in whatever way our future concepts may develop, that the very study of the external world led to the scientific conclusion that the content of the consciousness is the ultimate universal reality" - Eugene Wigner - (Remarks on the Mind-Body Question, Eugene Wigner, in Wheeler and Zurek, p.169) 1961 - received Nobel Prize in 1963 for 'Quantum Symmetries'
bornagain77
October 2, 2014
October
10
Oct
2
02
2014
03:21 AM
3
03
21
AM
PDT
I've just finished the 1927 book by Arthur Eddington (The Nature of the Physical World), pretty much the first English-speaker to read, understand and promote Einstein's theory, and one of the first to demonstrate it experimentally. He's answering different questions, and I don't think Bayesian logic was even invented then. Nevertheless, from an unrivalled understanding of the structure of physics and the science associated with it he says:
The mind has by its selective power fitted the processes of Nature into a frame of law of a pattern largely of its own choosing; and in the discovery of this system of law the mind may be regarded as regaining from Nature that which the mind has put into Nature.
It seems to me that's a different expression of the same thing - and one reason he rejected materialism as a sufficient explanation for reality.Jon Garvey
October 2, 2014
October
10
Oct
2
02
2014
12:14 AM
12
12
14
AM
PDT
Hi Barry, Just a quick note. A few of the lines you quoted were from wd400 that I was replying to. Here are the relevant portions for reference with his comments in the blockquotes and mine outside of them. ------------ That said…
p(theory) = [p(theory|naturalism) * p(naturalism) ] + [p(theory|not-naturalism) * p(not-naturalism)]
On Methodological Naturalism (MN), it seems this would be: ------------ And... ------------ On MN, it doesn’t matter what the probability is that the theory is true given naturalism is false, because the probability of naturalism being false is already determined to be 0.
of course, if a naturalistic theory explains data well then it makes p(naturalism) higher: where p(observations|naturalism) is p(observations|theory) (or perhaps summed over all naturalistic theories that might explain the observation)
Wouldn’t that be… p(naturalism|observations) = p(observations|naturalism) * P(naturalism) / [p(observations|naturalism) * p(naturalism)] + [p(observations|not-naturalism) * p(not-naturalism)] …or am I missing something? In any case, the problem here is that on MN, P(naturalism) = 1 ------------HeKS
October 1, 2014
October
10
Oct
1
01
2014
10:34 PM
10
10
34
PM
PDT