Mathematically Defining Functional Information In Biology

_{January 28, 2009

Intelligent Design}

Share: Facebook; Twitter; LinkedIn; Flipboard; Print; Email

Lecture by Kirk Durston, Biophysics PhD candidate, University of Guelph

[youtube XWi9TMwPthE nolink]

Click here to read the Szostak paper referred to in the video.

HT to UD subscriber bornagain77 for the video and the link to the paper.

Comments

tribune[200], Data about politicians can be used to assess the behavior of politicians. I wouldn't use them to assess the behavior of the designer of the universe.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
01:04 PM
1
01
04
PM
PDT}

One major difference between such cases and “life & universe” is that we have empirical data to assess our hypotheses in the first place (cheating politicians have been known to exist). But the point is that we do have empirical data that design exists and has quantifiable characteristics. Hence, it seem rather arbitrary not to allow these characteristics (or the methodology used to find them) to be applied to determine an aspect of the nature of life. Another ironic point -- especially for Mark Frank to consider -- is that to require a benchmark in universe creation to determine the likelihood of design is a science stopper. You seem to be saying it is unfair to calculate the chance of something happening so we must assume it was chance that caused it to happen. As Mr. Spock would say: ruling out probability calculations to determine whether something happened by chance is highly illogical. :-)tribune7_{February 1, 2009
February
02
Feb
1
01
2009
12:58 PM
12
12
58
PM
PDT}

Jerry[194]. That's double daftness, dude.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
12:45 PM
12
12
45
PM
PDT}

#190 (Professor Oloffson, Mark Frank, or whomever), assume you’re an ID advocate, what would you do differently than Dembski, Behe or Durston when it comes to calculating the probabilities or supporting a design inference? I would concentrate on creating a genuine design hypothesis to compare to evolution. Who, when, how, why. Then you can start to point to try and find positive evidence.Mark Frank_{February 1, 2009
February
02
Feb
1
01
2009
12:38 PM
12
12
38
PM
PDT}

kairosfocus, thanks for the link and the log function explanation; could you try to answer one more question for me? Are they actually getting a pure measure of functionality in information her?e i.e. is it a, across the board approximation of 3-D functionality to information? I ask this because this anomaly caught my eye in this paper: “Measuring the functional sequence complexity of proteins” “although we might expect larger proteins to have a higher FSC, that is not always the case. For example, 342-residue SecY has a FSC of 688 Fits, but the smaller 240-residue RecA actually has a larger FSC of 832 Fits. The Fit density (Fits/amino acid) is, therefore, lower in SecY than in RecA. This indicates that RecA is likely more functionally complex than SecY. Thus as you can see, me being unfamiliar with the math as I am, that this appears that they may actually being getting a (somewhat?) true measure of functionality of a 3 dimensional structure translated into bits that may be used to firmly establish the principle of Genetic Entropy. Thanks again for your helpbornagain77_{February 1, 2009
February
02
Feb
1
01
2009
11:58 AM
11
11
58
AM
PDT}

Seversky, I will, but I doubt I can say anything interesting or meaningful. I usually assume that it is reasonable to think of genes as carrying information.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
11:52 AM
11
11
52
AM
PDT}

Patrick[190], You write
So, in regards to Kirk’s argument essentially your position comes down to asserting that “intelligence” as a foundational starting point is not qualified to your satisfaction?
Not at all. Where do you get that from? My fundamental criticism is outlined in my post 150(f), and there is some basic reading material in post 185. Kirk writes in boldface that he compares an "empirical probability" to a "Bayesian probability." This statement is unclear until we learn what type of inference he intends to use. There is a discrepancy between what he says in the video and what he explains in this thread. As for answering questions, I usually try to answer all that are serious and expressed in a civil manner. I may occasionally overlook some, so remind me please. Participants who have previously bombarded me with insincere questions or personal insults will no longer get replies. Not that I'm overly sensitive but it gives me a criterion to decide how to spend my time!Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
11:44 AM
11
11
44
AM
PDT}

Prof_P.Olofsson, We can be daft together and go watch Daffy Duck to get inspiration.jerry_{February 1, 2009
February
02
Feb
1
01
2009
11:39 AM
11
11
39
AM
PDT}

I realize this is tangential to the subject of this discussion but I would be interested in Professor Olofsson's comments on Australian philosopher John Wilkin's argument here and here that it is misleading to think of biological systems like genes as containing information.Seversky_{February 1, 2009
February
02
Feb
1
01
2009
11:38 AM
11
11
38
AM
PDT}

Pattick[190], I didn't mean to ignore you, sorry about that. There's a lot going on. I did actually address your questions in my post 185, in response to Robbie[181]. OK then, here is Olofsson the ID prononent: I would use strict elimintion, but not the explanatory filter as it is too strict (we can NEVER rule out all chance hypotheses). I would formulate a collection of chance hypotheses that would be acceptable to the biological community, taking into account reproduction, mutation, and natural selection. Next, I would form a reasonable rejection region consiting of observed outomes together with other possible outcomes (think flagellum here which is not the only conceivable motility device). Then I would compute the probabilities of the rejection region under the chance hypotheses and show that they are all ridiculously small. If anybody would object that I cannot assess the likelihood of the hypotheses themselves, I would have to agree but argue that there is nothing else we can do.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
11:24 AM
11
11
24
AM
PDT}

PO [188] I agree. I should have phrased this as "it means something to estimate the probability of an outcome given current evolutionary theory". The results will be incredibly unreliable, and for many (most?) outcomes the task is impossible, but where it can be done the estimate will at least pose the question "why is this estimate wrong?" which is useful for directing research. To put it another way - the difficulty of doing this type of estimate is solved through biology - not metaphysics.Mark Frank_{February 1, 2009
February
02
Feb
1
01
2009
11:13 AM
11
11
13
AM
PDT}

Prof O. (and others) So, in regards to Kirk's argument essentially your position comes down to asserting that "intelligence" as a foundational starting point is not qualified to your satisfaction? That seems more like a philosophical objection rather than mathematical. Personally, when it comes to "intelligence" and the sciences I've always treated it like gravity. We can observe it functioning but we do not know exactly how it works. I'd also like to highlight this:
(Professor Oloffson, Mark Frank, or whomever), assume you’re an ID advocate, what would you do differently than Dembski, Behe or Durston when it comes to calculating the probabilities or supporting a design inference?
I've actually asked this myself many times of various people (and at least once for Prof O.) and I've been repeatedly ignored.Patrick_{February 1, 2009
February
02
Feb
1
01
2009
11:03 AM
11
11
03
AM
PDT}

Footnote to [185]: You can use Bayes' formula without doing a Bayesian analysis. The former is a formula about probabilities in general; the latter is a particular way of doing inference.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
10:58 AM
10
10
58
AM
PDT}

Mark[186], Now I must perhaps slightly disagree with you: I think it is very difficult to estimate probabilities given current evolutionary theory as this theory is often not quantitative in nature. For example, I think Dembski's "shopping cart model" for the flagellum is unreasonable and not motivated by biology, but I do not know with what to replace it. I re-read my post and what I mean is that P(E|C) is the only probability we can even hope to compute.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
10:53 AM
10
10
53
AM
PDT}

As for "daft," yes, I jested in honor of Jerry the Celt. After consulting various online dictionaries, I conclude that "daft" does not describe my opinion of probabilistic analysis.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
10:28 AM
10
10
28
AM
PDT}

Re #182 and #183 I agree (as usual) with PO. The logic of probability should be applicable everywhere. However, when you start to ask questions about the prior probability of ID or CHANCE then this is so meaningless it is useless. In the Caputo case, for example, it is quite meaningful to ask ourselves how likely was he to have cheated even before we saw the evidence (knowing politicians - quite likely). I still find Bayesian and comparative likelihood logic useful for thinking about ID in a structured way. But you can't do the actual calculations - except - as PO says - you can estimate the probability of an outcome given current evolutionary theory.Mark Frank_{February 1, 2009
February
02
Feb
1
01
2009
10:21 AM
10
10
21
AM
PDT}

Robbie[181], Sure, ID is possible, but if we want to base our inference on probabilities, we must quantify it. How do we quantify "possible"? As for the gentlemen you mention, Behe argues mostly qualitatively from the field of biochemistry. I cannot judge those arguments. His forays into probability and statistics have been fairly rudimentary. If I were an ID proponent and were to choose a general inference strategy, I would probably go along the lines of Dembski's eliminative approach, but without making arbitrary uniformity assumptions that no biologist would support. For a simple probabilistic analysis (returning to the big questions about life, the universe, and evolution) we have two competing hypotheses: C for chance and D for design, and some evidence E. There are then the following probabilities involved, in the usual notation for conditional probability [P(A|B) means the probability of A if we have access to the information given by B]: P( C) and P(D) -- how likely are the hypotheses without any prior knowledge? P(E|D) and P(E|C) -- how likely is the evidence to occur assuming each hypothesis? P(D|E) and P(C|E) -- how likely is each hypothesis given the evidence? These probabilities relate to each other through Bayes' formula. In a likelihood comparison, you compare P(E|D) to P(E|C), and in a Bayesian analysis you compare P(D|E) to P(C|E). In a strictly eliminative analysis, you compare P(E|C) to some preset number that you think is small enough to warrant rejection. There are problems with all 3 approaches but my opinion is that only P(E|C) can be computed or estimated. Dembski's "explanatory filter" is also based on this premise. It is currently unclear whether Durston wishes to do a likelihood or a Bayesian anlysis (see my post 150). I think Mark's point with the planets was to illustrate that you can always bias your analysis in favor of intelligent design by assuming that P(E|D)=1. Such an assumption is tantamount to assuming "guilt" in a court case because that is the one explanation that confers certainty on any circumstantial evidence.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
10:18 AM
10
10
18
AM
PDT}

----Professor Olofsson: "Daft: (2) Scottish : frivolously merry." Oh come now my good professor. Surely, you jest. Reading Mark's comments in context we can safely conclude that the remark was meant to convey "daft" as "daffy," not frivolously merry. Still, I admire your deft, though daft, foray into damage control.StephenB_{February 1, 2009
February
02
Feb
1
01
2009
10:16 AM
10
10
16
AM
PDT}

tribune[182], Yep. I assume you have been with us when we beat up the Caputo case which is a good example. One major difference between such cases and "life & universe" is that we have empirical data to assess our hypotheses in the first place (cheating politicians have been known to exist).Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
09:51 AM
9
09
51
AM
PDT}

Professor & Mark, Leave aside life & the universe etc. Are probability considerations useful in determining the existence of design in other phenomena?tribune7_{February 1, 2009
February
02
Feb
1
01
2009
09:39 AM
9
09
39
AM
PDT}

Mark Frank (#114) " Assuming random inclinations of planets the probability of such an alignment happening by chance is one in several million (depending how you calculate it). This is nothing compared to the zillions you are working with, but quite big enough to make the point." Isn't the size of the number(s) precisely the point? As Prof Olofsson observes (#132): {...} all we are left with, and all that is hidden in Kirk’s computations, is calculating the probability of chance occurrence of various features in nature. Probably that’s all we can reasonable do, but then we need to be very careful with assumptions." What's reasonable here? Why is, for example, jerry (#130) being unreasonable when he says, {...} ID is possible as an explanation of life. If that is granted the real battle will be easier. Those who defend the extremely unlikelihood of naturalistic methods must by definition deny an intelligence prior to life or else their whole world view falls apart. {...} [Life: designed or not]At what point would the inference be reasonable? Are you (anyone) saing an ID inference is unreasonable (uncalculable) or you saying that the evidence contradicts it? (Professor Oloffson, Mark Frank, or whomever), assume you're an ID advocate, what would you do differently than Dembski, Behe or Durston when it comes to calculating the probabilities or supporting a design inference? Surely these guys deserve some credit; it's not exactly an easy thing to do. I can sometimes appreciate the idea that ID can never be big-T Theory - as Dr. Dembski has noted elsewhere, it's not mechanistic - but certainly the inference is sound or reasonable. No? Why not? Is the idea that life and its origins are the result of blind processes reasonable? Why?Robbie_{February 1, 2009
February
02
Feb
1
01
2009
09:22 AM
9
09
22
AM
PDT}

jerry[179], Daft: (2) Scottish : frivolously merry.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
08:52 AM
8
08
52
AM
PDT}

"I actually think the whole probability approach is daft " One would have to if one denies ID. That is because the logic and reasoning is so obvious and there seems no defense against it except to call it "daft." How is one to explain these incredibly complex and incredibly rare functional proteins? Probability is one obvious route to assess the difficulty by which all paths have to be evaluated. Whether it is being applies in the best possible way can be an argument but to call it daft is daft. Maybe you should suggest some other ways since modern evolutionary biology hasn't a clue. So step up and earn your Nobel prize because that is what is awaiting anyone who can do it.jerry_{February 1, 2009
February
02
Feb
1
01
2009
08:34 AM
8
08
34
AM
PDT}

JayM, It seems like one has to dot every i and cross every t in these discussion when the comments are made in a hurry and you think the meaning is obvious. I was talking about non life and the inability of nature to create FCSI. There has never been a case. That is why I delineated it that way. Both life and humans as part of life and are also part of nature. As far as life is concerned this thread and the other thread on Dembski's two papers are essentially about the ability of nature to create new FCSI in life or from scratch. All you are pointing to are slight modifications of current genomes and no really new FCSI. If you ask some of the regulars here they will point out that I am probably one of the most vigorous proponents of micro evolution on this site and its power to provide new variants of life and eventually new species. But these new organisms are not very different from their predecessor or original gene pool no matter how long the time period is.jerry_{February 1, 2009
February
02
Feb
1
01
2009
08:27 AM
8
08
27
AM
PDT}

Mark[169],
I actually think the whole probability approach is daft
I agree, as far as the Bayesian or likelihood analyses are concerned. The only probability we can ever reasonably try to estimate is that of the evidence assuming a chance hypothesis, P(E|C). For this reason, I have repeatedly pointed out in this thread that I agree with Dembski: only elimination is at all possible. Not that P(E|C) is easy to estimate either, but at least it is conceptually clear. Thanks for your very clear and accurate explanations of basic probability theory. If we are to discuss these issues at all, we should all learn the basics.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
07:46 AM
7
07
46
AM
PDT}

kairosfocus[174], Hello again my insular friend! I gave a very brief answer to bornagain[25] in my post [27], but I think it may have been overlooked. At any rate, your explanation is more complete and the key issue is the desirable additivity. Welcome to this thread. We are awaiting Kirk's reply. I hope all is well in the worst soccer nation in the world!Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
07:34 AM
7
07
34
AM
PDT}

DaveScot[170],
According to the laws of physics there could be any number of elementary particles in a universe ranging from zero to infinity.
OK, but what probability distribution do the laws of physics give us? There is no information about that distribution in the numbers 10^80 and 10^20.Prof_P.Olofsson_{February 1, 2009
February
02
Feb
1
01
2009
07:23 AM
7
07
23
AM
PDT}

Jerry @161
The first example is in all life and the other two examples are of human activity. As stated above this phenomena exists no where else in nature and this comment itself is an example of functional complex specified information.
So far we're in agreement. The observations are that human intelligence generates CSI and that CSI exists in biological systems. There is one immediate problem, however. So far no ID researcher has demonstrated how to calculate CSI for a real world biological organism, or even a component of a cell. That's got to be the next step.
Now since no one can never say never there might be a time sometime in the future where someone demonstrates that nature can produce functional complex specified information.
This does not follow from your observations. Biologists have observed information being channeled into the genomes of populations of organisms via MET mechanisms. Unless you're denying the evidence for microevolution, this can't be disputed. That means that some level of CSI can be created by natural processes. The question becomes, how much? In order to answer that, we need the ability to compute CSI rigorously, an understanding of the limits of MET mechanisms, and an understanding of the topology of viable genome space. As Robert Heinlein said: "If it can't be expressed in figures, it is not science; it is opinion." JJJayM_{February 1, 2009
February
02
Feb
1
01
2009
06:20 AM
6
06
20
AM
PDT}

BA 77: Been busy on other things, so I didn't notice this thread. I see your, @ 25 : I am disappointed that no one here at UD has elucidated why the -log function was necessary. I am fascinated that this particular function is required. --> Why not look at my hopefully more or less 101 level discussion here? --> In a nutshell, - log2[X] converts X from a probability metric to an information one in bits [base e = 2.7182818 . . . would be in "nats," etc], with the properties we want such a metric to have, e.g. additivity so Info A plus Info B is Info (A plus B) --> Probabilities are inherent in information measures [and that is why there is so much selectively hyperskeptical noise against them above . . . but since for instance coding in DNA strands is independent of the chaining chemistry, we can in fact do a very simple calc as the presenter envisions(MF I have a calculation here on Dembski's related metric . . . )] --> What was done in the presented paper is to simply assess the fraction of a contingent space that exhibits function [net area of the archipelago of function], and take the ratio to the whole space [area of space as a whole] as a probability metric, per Laplacian indifference. --> By definition, information can only be stored in a contingent system [think states of alphanumerical characters: no state variability from a program or a law, no info storage; and random arrays are for sufficiently complex funciton, maximally unlikely to hit on function. Monkeys hitting keys at random do not Windows 7 make -- and MS is proof positive that less than perfect design is still design.] --> So, if you want to say some version of pre-life NS and some sort of pre-life chemical ladder leads up to life in some sort5 of pre-biotic soup [realistic empirical evidence, please . . . ?], you are in effect saying that the contingency was displaced from the pre-life chemistry to the underlying physics that makes for the chemistry. --> that info is not going to come out of lucky noise, all you have done is say the keys are pressed to get the characters on the screen and once the keys are pressed [physics is set up], the characters will form [pre-biotic chem ladder]. --> Thus, you are now looking at having implied that the cosmos is fine-tuned for life to emerge. And, as John Leslie has shown, that leads into the sort of local fine-tuning that makes a multiverse hypothesis irrelevant as an escape. --> And that is what the presenter raised and implied. Information, per massive experience, comes form somewhere; and that ain't from lucky noise. Trust that helps. GEM of TKIkairosfocus_{February 1, 2009
February
02
Feb
1
01
2009
04:59 AM
4
04
59
AM
PDT}

# 171 Dave - probabilites are always numbers between 0 and 1 - consult any text book. You seem to be talking about odds which are different. In any case if you were talking about odds then there is no relationship between the odds of e given ID and the odds of e given non-ID. The odds of a black swan dying before its 10th birthday are very high. The odds of non-black swan doing the same are almost identical.Mark Frank_{February 1, 2009
February
02
Feb
1
01
2009
02:05 AM
2
02
05
AM
PDT}

Prev 1 … 3 4 5 6 7 … 11 Next

You must be logged in to post a comment.

Leave a Reply