Just a note that I have put up a new post at Panda’s Thumb in response to a post by Eric Holloway at the Discovery Institute’s new blog Mind Matters. Holloway declares that critics have totally failed to refute William Dembski’s use of Complex Specified Information to diagnose Design. At PT, I argue in detail that this is an exactly backwards reading of the outcome of the argument.
Commenters can post there, or here — I will try to keep track of both.
There has been a discussion of Holloway’s argument by Holloway and others at Uncommon Descent as well (links in the PT post). gpuccio also comments there trying to get someone to call my attention to an argument about Complex Functional Information that gpuccio made in the discussion of that earlier. I will try to post a response on that here soon, separate from this thread.
J-Mac,
Try this https://youtu.be/NGVP4J9jpgs
DNA_Jock,
For the sake of argument I will grant you your theoretical. It may or may not have a real empirical case. My argument would be that function should be defined as how a normal eukaryotic cell behaves given the state of the life cycle it exists in. In a mature cell the WNT/beta-catenin pathway should be down regulated. In early embryo development it should be peddle to the metal.:-)
My comment asking about information in science in general was poorly thought out. I should have focussed on biology and on Shannon information in particular.
Here is a fun usage for physics; check Sean C’s comments at the start:
Is “information is physical” contentful?
And the Shannon information work on coding/compression has had an application or two in engineering.
The entropy question was Mung humor or Mung contentiousness, right?
If not, when you have the free energy, check that out.
I’m going to break my word on shutting up and offer another way of seeing this discussion that I find helpful.
This is the key question: What does ‘normal’ mean in your sentence? In particular, normal according to what standard?
I think there is a connection between this discussion and the objective morality issue which has come up from time-to-time at TSZ. Here is what I mean.
Just as in the discussions of morality focusing on a single, universal morality, I believe you are looking for a single, universal definition of ‘normal’ function. Further, I read you as thinking this universal standard must come from the outside, from an intelligent designer perhaps (I don’t know your view for sure, that is just a guess).
I (and DNA-Jock as I understand him) think ‘normal’ is defined by perspective. In this case, the perspective of the researcher. I would add further that we judge which definition is appropriate in science by the success of the resulting research program. Here the definition of success depends on the goals of science, which are explanation, prediction, control of the world we humans live in and experience.
But is there a way of talking about normal biological function that is independent of any human or other outside, intelligent perspective? That in general is the philosophical question of naturalizing norms. In the case of biological function, the current best approach defines ‘normal’ from the point of view of the organism* in achieving (or having achieved evolutionary) success in its goals of continuing to live and to reproduce.
Now it is true that is an instrumentalist approach to understanding normal. That is, it focuses on achieving a set of goals which are agreed to as a pre-condition for the rest of the discussion which, given that agreement, goes on to define ‘normal’. That’s likely the best a naturalist can do when it comes to norms. (Consider eg Leiter’s Normativity for Naturalists)
And to circle back to morality, there is also a naturalistic approach in philosophy of morality which starts from a goal of human flourishing. That issue of course has been discussed ad nauseum here at TSZ, so I will hasten to say that in no way am I suggesting that particular conversation be re-opened!
———————-
* Dawkins might prefer the gene’s perspective to the organism’s.
I agree.
The only sensible approach is to consider the phenotype. Genes are sustained by individual organisms, their survival is utterly dependent on them being passed on through viable phenotypes. Dawkins wasn’t wrong exactly but overemphasized as a reaction against group selectionists. Nor does group selection work as an idea. Adaptation, niche, phenotype! (Kin selection notwithstanding!)
From SEP on Fitness
—————————————————
Some mathemeticians agree, at least for mathematics.
When it comes to science, you run into that pesky issue of having to link the formalism to predicting something about the world. Then again, there is string theory, so maybe you are onto something about sticking to math only.
What I said about information and mathematics in earlier post was intended to refer just to scientific uses of information. For science in general, there are many formalisms used in explanation and theories, eg mechanisms, diagrams, algorithms, physical models; whatever works for the relevant scientific community.
As normal for me, I include a possibly relevant SEP link .
——————–
*ETA on ‘Darwinian’ in the SEP quote: Not meant to start the discussion down the road of arguing about that word, please and thank you, Mung.
I just read the PS post about Demski’s second order information, which I had not known about. Some eerie similarities to what I posted. Definitely unintentional on my part.
However, information has no role in my post.
It works.
Thanks Bill!
BruceS,
I am using normal in the same way we would say it is normal for a human to have 2 eyes and a nose. Not all humans do but missing any one of the three it is the exception not the rule.
In most mature animals cells the, WNT/Beta catenin pathway is down regulated, reducing cell division rates, reducing vascular growth, increasing immune system activity.
So you want to interpret normal as what most of the examples have in common?
In that case, why not focus on what most cancer cells have in common*. In particular, if one’s goal is to treat or at least explain cancer, focus on what makes them successful in the body’s environment?
Perhaps that means a portion of the genome that is common for non-cancerous cells has changed to something that is common in cancer cells. That is an empirical question. But the answer depends on the interests of the investigator. And that is what is matters for Szostak FI.
I don’t know the biology so I am focusing on generalities, which I think suffice for the point I am trying to make.
——————————-
* The deeper question is of course how it got to be common. For example, all blood is red. But being red is not blood’s function.
Except when it’s blue. 🙂
BruceS,
Most cancers are caused by the up regulation of embryonic pathways. WNT, Sonic hedgehog, Notched, and Nanog are all embryonic pathways. This can happen without any change in FI but having low blood levels of vitamin d. It can also happen with a reduction in FI as defined by a healthy (non cancerous) cell.
Are you trying to blind me with biology? If so, you have succeeded.
All of these might constitute valid definitions of Szostak FI. I cannot say. Regardless, that would not change the fact that so would Samidass’s approach to cancer FI.
This is interesting Bill!
Has it ever occurred to you that the main reason why scientists can’t really find the underlying cause of cancer is because the don’t really know all the sources of information in the embryonic pathways?
BruceS,
I don’t agree with Joshua that there is such a thing as cancer FI. There is cellular FI that cancer modifies once DNA repair is disabled. I believe that the FI that is involved in cancer pre existed in the cell prior to cancer.
This whole thread seems to have “devolved” down to whether cancer cells have increased functional information when they are selected to grow more rapidly. So let me comment on that.
If we consider functional information of whole organisms, keep in mind that improvement of fitness of some of them my be bad for others (predators versus prey, parasites vs. hosts, etc.) In those cases we do not consider improved fitness, say improved hunting ability of a predator species, as lower FI because it is bad for the prey. So I think that this speaks in favor of considering higher growth rate of cancer cells as an increase of FI, as it is considered from their point of view.
Joe Felsenstein,
Maybe there’s a law of total fitness lurking in there. One population’s increase in fitness is offset by another’s decline. The total fitness of the environment remains unchanged! 🙂
There are some rather-artificial ecological models in which it works that way. But in general a predator has to eat much more than 1 gram of prey in order to gain 1 gram.
Joe Felsenstein,
So I should have said total fitness can never increase!
No, because improvement in resistance to predation, on the part of the prey, leads to a greater total biomass, the larger numbers of prey more than offsetting the smaller numbers of predators.
Apologies, Joe. I was just alluding to Dembski’s law of information (and the second law) not seriously.
Although I might add that the resources of this planet are not infinite.
Joe Felsenstein,
I agree if you define fitness carefully as short term cell survival that this is possible but isn’t FI related to higher sequence specificity? Can you have a gain in fitness and a loss of FI? The FI that is allowing elevated cell division and other up regulation already exists in a normal cell. It is how the cell behaves during embryo development.
colewd,
What about streamlining?
Not necessarily. Gpuccio seems to think so, but there is no reason that the high function genotypes would be very similar to each other.
FI can (in principle) be calculated for any property of an organism that is positively correlated with fitness. Such as rate of reaction of the malate dehydrogenase enzyme, or an ability to carry oxygen of hemoglobin beta, or even fitness itself. There is not just one thing called function. There is thus no
general description of it. It might be the flying speed of a bird, or its acuity of vision. It just depends on what we want to talk about.
Anyone who thinks that FI is some general value for an organism does not understand how people like Szostak and Hazen wanted to use the term.
Actually when we talk about fitness in this context, it is really relative fitness we are discussing. The values on the axis used in calculating FI are relative fitness. Because there are also population density-regulating mechanisms, like scarcity of resources, which will reduce or increase the relative fitnesses when making them into absolute fitnesses, keeping their ratios the same but changing their absolute sizes.
I agree with Joe: Szostak FI depends on first taking a point of view (POV) regarding the definition of the function. There is no POV-free Szostak FI.
You are adding to the concept. You say “there is cellular FI”. That means you are saying there is a single correct viewpoint. There is nothing inherently wrong with you claiming that. But it is different from Szostak FI.
Durston engages with this issue at PS There he compares three approaches to taking a POV on function: the inteligent observer, the non-cancerous cell, the cancerous cell. He accepts that third case is valid, but thinks that we can eliminate it if it depends on some intelligent observer declaring it to be a function when he thinks it really is not. So he seems to be making the same point as you.
Swamidass ignores that long aside by Durston, and simply declares in his next post that “First, I am not interested in “non-functional” p53, we are interested in carcinogenic p53”. He simply declares his POV without a long aside on its correctness.
So is there a correct, single way to understand function here? Not for science. In science, all that matters is whether the POV leads to a successful research program.
There does seem to be a misreading of Szostak (which I have admittedly only skimmed). His example function is RNA-GTP binding. It is quite clear that evolution can increase, and decrease, such a function. It can do it most readily if that functional change is associated with an increase in fitness, but it doesn’t have to be.
Joe Felsenstein,
BruceS,
Here is one of Kirk’s examples in his discussion with Joshua:
In this case the functional information of an apoptosis inducing protein appears to be less then in its normal state as measured in bits and appears to be independent of how function is defined. The way functional information is measured appears to be determined by the number of different AA arrangements that will perform the specified function and the length of the sequence measured in bits.
If I defined cancer as a disease that reduces functional information in the cell can anyone make an empirical case against this hypothesis?
Examples of Hypothesis:
1. If I replace the battery in my car, then my car will get better gas mileage.
2. If I eat more vegetables, then I will lose weight faster.
3. If I add fertilizer to my garden, then my plants will grow faster.
4. If I brush my teeth every day, then I will not develop cavities.
5. If I take my vitamins every day, then I will not feel tired.
6. If 50 mL of water are added to my plants each day and they grow, then adding 100 mL of water each day will make them grow even more.
http://www.softschools.com/examples/science/hypothesis_examples/104/
Try again?
If I defined cancer as a disease that reduces functional information in the cell then what?
Can you actually say something that can be disconfirmed I wonder? That’s why ID persists. Nobody ever puts anything testable out there.
Why isn’t the contradiction in this obvious to you?
“the functional information… [is] independent of how function is defined”
“functional information is measured … by the number of different AA arrangements that will perform the specified function”
So it depends on the function you specify.
What hypothesis? You haven’t specified a hypothesis. You have given a (useless) definition of cancer.
Ninja’d by Rumraket, on both counts.
I’m not sure who you are replying to; if it me, let me say I think there are two different issues.
1. What should be considered functional? The 2007 paper Hazen et al is most explicit is saying this is a separate decision
2. Separately, an examination of how that predetermined function contributed to fitness.
As Rumraket says, first you have to provide a hypothesis detailed enough for empirical test. If you want to use the concept of Szostak FI to describe your hypothesis, then you have to specify the function involved, presumably some protein specified by genome of non-cancerous cells that is no longer specified by cancer cells.
Now you develop a theory and mechanisms and set of experiments to test your function and go through the usual scientific process for review.
But my point is that you pre-specified the function. Saying you “define” cancer by your function changes nothing. It’s still your POV.
[end of serious comment]
Now I wonder exactly what benefit the concept of FI has to carrying out that research. It seems more like philosophy than biology to me. But what do I know? I know no biology worth mentioning but I do know a bit of philosophy. So maybe that is just my POV of FI as a nail for my philosophical hammer.
BruceS,
No, I wasn’t responding to you; it was a general comment on people who confuse an increase in FI (wrt a particular function) and an increase in fitness.
I agree that it depends upon one’s choice of function; no one sequence can increase one function without a likely decrease in another.
BruceS,
Thanks to you and Rum for these thoughts.
A cell that goes through a life cycle from embryo development to a mature adult must be able manage the full life cycle which requires lots of functional information for the proteins that control the signal pathways to the ubiquitin system that allows for the variation the cell needs to properly function through this life cycle. Most cancers are when the cell loses the ability to go through this life cycle having the right functions at the right time in the cycle. If an embryo pathway is inactive in early development this is very problematic for the organism. In the same way if an embryo pathway is active in a mature animal cancer can develop. The key functions controlled by the ubiquitin system (embryo pathway regulation) are.
-Cell cycle control including DNA repair and apoptosis regulation
-Immune system regulation
-Protein degradation regulation (ubiquitin system)
All these process require functional information to be properly regulated. Loss of regulation of these systems can cause cancer. I believe that loss of regulation is the result of loss of functional information. You and Rum brought up how you test for this and that is a very good question. As we move into gene editing technology such as crisper I think it may make sense to understand and measure functional information as Szostak and Hazen have defined it.
colewd,
Funny how you seem to be unable to grasp the problem. Try this for size:
Rous Sarcoma Virus is an oncovirus that causes sarcoma in chickens. A virus goes through a complex life cycle in which it invades a cell, has its RNA genome reverse transcribed to DNA, and then has its cDNA genome inserted into the host genome. Expression of the viral genes causes uncontrolled proliferation of the host cells (tumors) and the release of new virus particles, completing the cycle. The genes in the viral genome (gag, pol, env and src) need to properly function to complete this life cycle, and if any of these are rendered inactive this may prove to be very problematic.
Suppose this virus suffers a mutation that renders it incapable of completing this complex life cycle. Was functional information gained or lost? Both? Neither?
Why do people allow threads like this to be derailed by talk of FI?
What does FI have to do with the OP?
Yes. Probably both.
Mung,
I’ve submitted the information theory argument that flabbergasted Felsenstein. If you are an admin, perhaps you can approve the article, so we can get the discussion back on track.
Yup. And when Hazen et al. took up Szostak’s idea of functional information, they put “functional information” in scare quotes, and characterized it as a complexity measure.
“Functional information” is not information in the sense of Shannon.
Corneel,
This is a great idea and why FI may be relevant. I think that this is very unlikely to cause uncontrolled proliferation alone without subsequent loss of FI.
Tom English,
How would you compare it with Shannon’s concept of mutual information?
It seems to me that Bill Cole has reified the abstraction of functional information. That is, he consistently refers to it as though it were something that’s “really out there,” to be measured by us, when it’s purely an explanatory construct (the utility of which has yet to be established, in my opinion).
I’m saying that “functional information” not comparable to anything in Shannon’s mathematical theory of communication. There’s merely a resemblance in mathematical expression: “-log” put in front of a fraction “p” lying between 0 and 1. That superficial resemblance in expression does not imply that there’s any substantive connection. It is NOT the case that “-log p” is categorically information, though the expression comes up in Shannon’s mathematical model of information in communication. In other contexts, the significance of “-log p” generally will be different than in the context of communication.
Tom English,
I didn’t say that very well, and I won’t be around to straighten it out with further comments. Sorry if I just added to the confusion,.
Because I haven’t been able to figure out how to calculate CSI myself, Mung 🙂
It represent the (minus log of) the tail probability of drawing a random genotype that meets the level of specification (I am guessing fitness) of the genotype that is being evaluated. But what is the “random” pool from which I am drawing? Don’t I need to know the allele frequencies of this random population to be able to do that?
Not there yet. How about you?
EricMH,
Done.
Tom English,
Why don’t you think you can measure it?