This whole discussion seems to me to be quite decoupled from evolutionary quantitative genetics theory and the rigorous mathematical framework developed by Russel Lande and Stevan Arnold in their famous 1983-paper in "Evolution" (one of the most cited papers in evolutionary biology which is a must-read for anyone who serious claims to say anything in the topic and be taken seriously).

I would also recommend Stevan Arnold's paper "The Adaptive Landscape as a conceptual bridge beteween micro- and macroevolution" from 2001 (Genetica) as well as the edited volyme by myself and Ryan Calsbeek "The Adaptive Landscape in Evolutionary Biology" (Oxford University Press, 2012). In all these references, you will find explicit mathematical definitions of "natural selection", more specifically defined as the covariance between trait value and fitness.

Claims by Emilio Cervantes and Darryl Smith above that "natural selection does not exist" seems more based in faith and conviction, rather than in solid mathematical arguments. There is a well-developed theoretical and mathematical framework that explicitly defines the concepts of natural selection and evolution by natural selection, and the discussion above would benefit if we start from what we actually know and what has done before, rather than painting strawmen and using purely verbal arguments based on misunderstandings and caricatures.

A couple of additional points:

1. The term "survival of the fittest" was not Darwin's invention, but Herbert Spencer. This is well-known by all who knows a minimal amount of the history of evolutionary biology and has read the litterature.

2. The claims of tautology, using "survival of the fittest" was first raised by science philosopher Karl Popper, but has been discussed and dismissed over and over again, so that these claims come up over and over again only says that scholarship is poor among many self-appointed experts on natural selection theory. If you are not convinced, read some of Elliot Sobers books and check up what Karl Popper said about the topic in late life (he admitted that we has wrong).

3. The first sentence in Ronald Fishers book "The Genetical Theory of Natural Selection" is "Natural selection is not evolution". He clearly separated the process of selection from the evolutionary response, and any serious discussion must have that as the starting point, if one really does not want to re-invent the wheel or propagate misunderstandings, errors or claims that have been dismissed a long time ago.

A level of selection is not necessarily the same thing as a unit of selection. For example, group structure may contribute to individual fitness, and so including both a group-level term and an individual-level term maybe appropriate for computing overall fitness, but that doesn't mean the group itself is replicating. The best overview is Okasha's 2006 book, "Evolution and the units of selection."

I would also suggest the paper by R. Lewontin, "the unit of selection", Annual Review of Ecology and Systematics, 1:1-18(1970).

This is a question on which I recently spent much time and wrote a paper, submitted to a biological journal, but now possible to find at:

http://kie.vse.cz/english/new-pavel-pelikans-working-paper/

Its abstract with the main points of the answer reads as follows:

This paper proposes a conceptual model of evolutionary and developmental processes as a source of insights into the old, but still controversial issues of group selection, multilevel selection, and the origins of order. Its pillars are two elementary principles of information-processing – one linking information to choices, and one pointing out the need for basic instructions – and a well-defined distinction between evolutionary selection and developmental selection. It generalizes and modifies the gene-centered view of biological evolution by centering on basic instructions, which are more than genes; and by focusing on their instructing of ontogeny, instead of replicating during phylogeny. It directly applies to socioeconomic evolution, finding the basic instructions in the genomes of individuals and the institutional rules of societies. The main insights obtained are: (1) biological evolution involves several levels of developmental selection, but only one of evolutionary selection; (2) in socioeconomic evolution, the evolutionary selection is not of groups, but of genomically compatible institutional rules for the forming, developing and operating of groups successful in developmental selection; (3) self-organization generates order at more levels than evolutionary selection, but this produces more of new information.

I believe that this could clarify much of the confusion about this question - including that caused by Gould and Lewontin - but am most interested in receiving objections.

PP

In principle any level of organization can be a level of selection. The thing to think about is if there is selection there must be an "among" and a "within". Thus, we can have selection among individuals within a population; selection among cells within an organism etc.. This precludes several levels as being important. For example, the Gaia hypothesis makes no sense from this perspective. For this to be valid we would need selection among planets within the universe. It is hard for me to conceive how this happens. On the other end of the spectrum, I have trouble with genic selection. Does this mean selection among alleles within a locus? Again, doesn't make a lot of sense to me.

I suggest taking a look at contextual analysis. This method, which, incidentally, uses the same math as the direct fitness approach, assigns fitness at only one level, which may be arbitrarily assigned by the investigator (see my upcoming chapter in Bouchard and Huneman's book "from groups to individuals"). A level of selection is important if there is a significant effect of membership in the units at that level on (individual) fitness. While CA seems wrong and counter-intuitive to the uninitiated, in fact, relatively thorough analysis has shown that it is a very good way, and possibly the best way, to think about multilevel selection.

Quick gut-feeling answer, not backed up by citations:

I feel, this is in fact two questions: What is being selected (out of several alternatives)? And, second: What does selection act upon?

My answer to the first question would be: the allele, as it is the bit of long-term inheritable information that is changing its frequency against its alternatives in the course of evolution.

My answer to the second question would be: the phenotype, as this is the interface with the environment (including, of course, the same organisms body), where performance differences (reproductive success) result from varying selective pressures. I am not sure, if a phenotype equals an individual body here, as a phenotype may have relevant emergent qualities also at higher levels like kin, group, population, up to ecosystem. So, selective pressures act on phenotypes, ultimately resulting in changing allele frequencies, but a given allele may be part of different genomes, hence phenotypes, and receive different pressures according to the phenotype it is caught in.

Therefore, the relevant level of selection may be the one a given selective pressure acts most strongly on in a given case.

This answer is most probably heavily influenced by the earlier works of Dawkins, Gould, and Dennett´s "Darwin´s dangerous idea" ;-)

edited 5 min later for clarity

A topic of much controversy and misunderstandings. I agree with the first part Charles Goodnight's answer above. The "among" and "within" is essential. However, I don't see the problem with genic selection. Certainly there can be selection for different kinds of allelic variants at a locus. Some variants confer an advantage and others do not. The among/within is among alleles within a population.

May I bother all of you again with the solution proposed in the new paper I mentioned a while ago? I believe that the entire lasting confusion about multilevel selection is due to not distinguishing selection of cells, organism, groups, etc, which are cases of what I call "agents," from selection of genomic instructions, according to which the agents are formed, developed and operated. Allow me to paste in three paragraphs from the paper's introduction:

The model puts basic instructions in the center between evolutionary processes, which use trial-and-error searches for producing them, and developmental processes, which are guided by them. The outcome is the forming, developing, and operating of certain acting and interacting entities, termed “agents.”

Both evolutionary and developmental processes are recognized to involve selection, but of substantially different types. Evolutionary selection concerns basic instructions, whereas developmental selection concerns the according to them formed and developed agents. The agents need not be long-lived and the outcomes of developmental selection therefore need not last. But they are the key commands for evolutionary selection, whose outcomes, if successful, can last: select and retain the basic instructions of those agents that are succeeding in the developmental selection.

Intuition may be helped by thinking of agents as possibly multilayer cakes, and of basic instructions as the recipes for making them. The developmental selection may then be thought of as tasting the possibly several layers of the cakes and selecting the layers and the cakes that taste good. The evolutionary selection may then by seen as choosing and keeping the recipes for the good cakes, of which some may have only one layer and others several, while each recipe is written on one single piece of paper.

I wish someone read it and agreed or objected. PP

I do not have anything to add to Charles' explanation (except that he is wrong about genic selection :-), but here are some additional references:

SOBER, E. (1984). The Nature of Selection: Evolutionary theory in philosophical focus. Cambridge: Bradford/M1T Press.

Wilson & Sober. 1989. Reviving the superorganism. Journal of Theoretical Biology. 136:337.

In particular, pay attention to Sober's argument about selection of vs selection for.

I am sorry, but my down-to-earth mind, originally educated in computer construction and later turned to evolutionary economics, has difficulties distinguishing "selection of" from "selection for." It can only see that selection is always both OF something, and FOR some reason (e.g. success according to some objective function). May someone enlighten me? Thanks!

To find selection between genes, in my opinion a gene should be able to reproduce itself. Which genes are able to do this?

Our lab group is currently reading "Evolution and the Levels of Selection" by Samir Okasha. So far, I'm finding it to be a good read. Worth looking into. Also Ben Kerr has looked into this as well.

@Veiko

thought experiment: If a newly mutated allele leads to relative overexpression of a pigment relevant for the coloration of display feathers, its host might be more likely to be chosen as a mate. Therefore, all else equal, the allele gets reproduced more often (via its host) and increases its frequency of occurrence on this locus troughout the population over time. Couldn´t one say, there is selection working between the alleles for this gene? Did I get your question ?

Or did you mean two genes that are not alleles of each other "competing" for the same function (via mutation leading from the original, different, to very similarly behaving enzymes, for example)? That is, if for example a mutated duplication of gene B is performing better in a new role than the original gene A performing the same function, gene A may, and will, degrade and ultimatley vanish, without reducing the fitness of the host. I could imagine, this would be, by selection via the inclusive fitness of their hosts, very rare, but not completely impossible. I think I remember some shifts in developmental pathways were explained by similar scenarios.

Interestingly, there seems to be a gap in selection levels: While this may be described as selection between genes on the population level, it does happen between genes in different individuals, and thus might also be described as selection between individuals that differ in their genome. The way of describing the process, or deciding on a single relevant level, might be a matter of perspective.

I feel that the levels of allele, individual and population are very strongly linked by functional interaction in my understanding of selection as an evolutionary process.

The relevant level depends on what am I interested in,and/or what can I observe.

Thnx everyone for their reading suggestions, appreciated.

I also liked Goodnight’s response. I noticed there wasn’t much mention of community level selection. Bill Swenson had a couple of studies showing this was theoretically possible, at least in the laboratory. See the following three:

Swenson et al. 2000. Artificial selection of microbial communities for 3-chloroaniline biodegradation. Environmental Microbiology 2:564-571.

Swenson et al. 2000. Artificial ecosystem selection. Proceedings of the National Academy of Sciences, USA 97:9110-9114.

Wilson and Swenson. 2003. Community genetics and community selection. Ecology 84:586-587.

The first paper was inspired by the discovery that a bacterial community sampled near a dry-cleaning store had evolved to break down dry-cleaning waste but that no one species could do it alone. The following paper found something similar:

Radianingtyas, H., G. K. Robinson, and A. T. Bull. 2003. Characterization of a soil-derived bacterial consortium degrading 4-chloroaniline. Microbiology 149:3279-3287.

Response in three points to Veiko Krauss, who wrote: "To find selection between genes, in my opinion a gene should be able to reproduce itself. Which genes are able to do this?"

1) Of course, genes cannot reproduce themselves directly, they are only memories of instructions. But longtime ago, if one believes the most recent theory about the origins of life, the instructions in self-catalyzing RNA molecules could do it: they were also the agents. After that there was some "division of labor" - the instructions started to be more and more backed-up by DNA memories, and the agents consisted more and more of the proteins synthesized according to the instructions.

2) Back to my intuitive example of cakes and the recipes for them. The recipes need not reproduce by themselves, yet the ones for good cakes can be selected and retained. OK you need some readers for interpreting them, some ovens for baking the cakes, and someone tasting the cakes to say which ones are good. But please pay attention to the information contents. The readers and the ovens are largely standard (ribosomes ..etc), it is the instructions are special (cf. record-players vs. records with special music).

3) My distinction between "instructions" and "agents" is nothing very special. Just a slight generalization of those made by Dawkins (replicators vs. vehicles) and Hull (evolvors vs. interactors). But maybe you dislike these authors ...

PP

@ Pavel Pelikan

I read your paper, very interesting, and a good account for the generality of the evolutionary process and for transitions in individuality.

The differentiation between agents and instructions is itself one answer for the question of the difference of "selection of", and "selection for".

"Agents" traits are "selected for", because they are ones who interact with environment differentially, and by consequence "instructions" are "selected of". Other possibility is the change in frequencies of byproduct traits due causal selection of other traits.

But since "instructions" interacts in non-linear ways and are context dependent, better than talk of recipes, is more accurated consider changes in heritable phenotypic traits as what is been selected for, and hence the unit of selection.

In my opinion, the agent concept is better captured not by Dawkins or Hull vehicles/interactors, but with Griessemer (2000) Reproducer.

Using Griessemer’s concepts, different levels of selection can have, in your terminology, different b-particles and C-particles properties.

Many thanks for helpful comments! But I still see both the developmental selection of agents and the evolutionary selection of basic instructions to be both "of" and "for". The former is OF agents FOR their abilities to succeed in their environmental performance tests, and the latter is OF instructions FOR their abilities to lead to such developmentally successful agents.

Now I am sorry I did not express earlier my agreement with Goodnight's statement: "In principle any level of organization can be a level of selection." As many other discussants expressed agreement with his views, this may be a good starting point for trying to obtain more agreement with mine.

What I need to add is only that selection of different levels of organizations is what I propose to term "developmental selection" of agents. Cf. my favorite ant example : organelles, cells, ants, ant societies. All these are agents/organizations, all tested for their "fitness" - somewhat like the multiple quality controls along the production line of complex products. But then ask the question, where do the basic instructions for their forming, developing and operating come from? In the case of ants, they are all written in their genomes (NB: now known to be more than just genes!), which are the outcomes of a single level of evolutionary selection.

But in the case of human organizations (from small groups to large societies), the genomic instructions of Homo sapiens do not suffice, but must be complemented by additional instructions in the form of institutional rules, which require another level of evolutionary selection - the socioeconomic one (more on that in my paper ...).

May someone agree?

@Jeff

Thank You very much for literature about community selection.

@Paavo

Genes could produce a phenotype only together within many other genes within a genome of an distinct organism under distinct environmental conditions. Only all these conditions together are able to produce phenotypes, which could be selected. An exception are transposons ("jumping genes") because they are able to reproduce itself within a host genome. Thus, they could be selected against other transposons.

@Pavel

Artificial theories about selection might have no relation to biological reality. Dawkins replicators for instance are simply not to grasp (see the "definition" of it). If evolution suppose to be a form of information-processing, and genes suppose to be instructions, I would like to see strong evidence for this proposals.

@Veiko

1) All theories are artificial and can only imperfectly be related to reality. Whether they are useful or misleading depends on the questions asked. As you now ask about levels of selection, you certainly need some good artificial theory that would define all the notions involved with sufficient simplicity to allow our minds to work with them, and sufficient realism to correspond to real-world phenomena. Such theory is obviously missing, otherwise this question would no longer be interesting to debate. OK, I find my artificial theory helpful for my questions about socioeconomic evolution (including "group selection"), and the ways in which this builds on, and is constrained by, the biological evolution of Homo sapiens. But I fully agree, and even say so in the concluding remarks of may paper, that it may not be interesting to pure molecular biologists.

2) Much of what you say about genes may be said about instructions in computer programs: each instruction needs many other instructions, and they all together need a computer to be executed.

3) I hope my paper takes a clear distance from Dawkins, whose definitions I also consider fuzzy, but I find his approach as a convenient point of departure for explaining mine.

4) The evidence for considering genes (and let's not forget the other "instructing" parts of genomes - such as the ncDNA coding for regulatory RNA) to carry information in the form of instructions for guiding chemical reactions towards forming, developing an operating a certain organism can only be logical, stemming from the definition of information as a factor helping to determine choices. To understand that the ovum of a cat must have some instructions to form, develop and operate just a cat, and not a dog or a fly, can only be seen as a logical consequence of a certain understanding of the notions of "information" and "instructions." Not to overestimate the influences of environments, note that a cat's ovum will become a cat, and not a dog, in a broad variety of environmental conditions.

5) I now have a fourth point to my answer to your earlier question:

"To find selection between genes, in my opinion a gene should be able to reproduce itself. Which genes are able to do this?" This is to paraphrase it as follows: "To find selection with lasting outcomes between individuals, in my opinion an individual should be able to live unlimitedly long. Which individuals are able to do this?"

In general I greatly enjoy this debate and, as an evolutionary economist trying to understand evolutionary biology, and the thinking of evolutionary biologists, I am learning a lot. Thanks!

@Pavel

You are right, every theory is artificial. My argument was badly chosen. However, which novel insight did You gain by modeling biological evolution in a similar way as the development of human societies? To me, information represent the objective reality within the mind of a conscious being, i.e. it does not exist independently of such beings. An instruction is a kind of information. Thus, both terms made no sense if seen independently of our minds.

Selection is concrete and concerns entities which successfully reproduce or not. Organisms do that, their genes only depending on these organisms. To be selected, no entity has to exist for a long time. Selection is in its essence only the consequence of the relative abundance of two events: reproduction and death. And no entity is able to exist unlimitedly long. Genes, for instance, are endlessly changed (also deleted) by mutations.

I agreed that the term instructions or information can cause confusion in a sense that imply cognition, although the majority of authors that use it clarify that issue well (John Maynard Smith or Eva Jablonka are some that I remind now).

I think that the term INHERITANCE units are enough, and some cultural inheritance (behavioral or symbolical) are more like real instructions, others, like genetic, epigenetic, citoplasmatic or ecological legacies are less or almost nothing like.

I forgot to post a link for Griessemer's paper that I cited above, here it goes:

http://innovation.ucdavis.edu/people/publications/22%20Griesemer%202000%20Selection-1.1-67-80.pdf

@Veiko

I agree that instructions, and information in general, may exist only in relation to certain agents, but these are not necessarily only minds of conscious beings. In the view of modern theories of information-processing (well, not so modern, this view is over a half-century old), the agents may also be computers (I am sure you know of the use of the term "instruction" in computer programming, and are only momentarily forgetting it), and also all organism receiving inputs form environments ("information-data") and responding to them according to the "information-instructions" they posses. So this is only a question of sufficiently broad interpretation of terms. I try to express this general view of information on p.6 of my paper:

"... information ... may meaningfully exist only in relation to a certain choice problem of a certain choice-making agent – such as a cell, an organism, a society, or a technical device."

The entire section 2 of my paper is trying briefly to explain two very elementary principles of information-processing that are valid for a broad variety of agents, including brains, computers, organisms, and societies, which I believe important just for understanding the issue of multilevel selection.

Of course, if you are used to narrowing the meaning of the terms "information" and "instructions" to human minds, I understand that you may find the more modern broader view a little difficult to accept. Allow me to express my conviction that accepting it may be fruitful even for a molecular biologist, but I do not want to insist.

I also agree that no entity is able to exist unlimitedly long, and that even genomic instructions are changing (which is of course the very way the biological evolution proceeds). But I still see an enormous difference between the life-span of the genomic pool of a certain species (=collection of their species-specific basic instructions), and the life-span of any of its individuals. (OK, I do understand that it is the currently existing individuals that also serve as the carriers and the memory of the pool ...)

Some kind of selection might arise at any organizational level, and among these levels, where members (genes, chromosomes, mitochondria, cells, individuals, taxa...) compete for available resources and space.

@ Pavel

"The former is OF agents FOR their abilities to succeed in their environmental performance tests, and the latter is OF instructions FOR their abilities to lead to such developmentally successful agents."

Yes, I’m sure this is a fair definition and I agree with your point in cases like that, but the other case, byproducts selection is not explained here.

Giving a simple example, polar bear coat was/is selected by the properties of warming the bear. Together with that, the coat is a heavy tissue that was not selected for that, so heaviness is a byproduct of selection of other feature.

There was selection "of" a coat, but only" for" warming properties.

@tomáš

You are right in principle, but this very general answer, which puts very different things into one big bag, does not get you far. To gain more clarity on the multilevel selection issue, you also need to specify some meaningful relationships among these things. In particular, I believe it important to specify that (i) any success of organisms in the competition for resources in their environments is fundamentally due to their genes and other genomic instructions, while (ii) the success of these instructions is due to their abilities to enable their organisms to be so successful. I believe it also that clarity requires recognizing that the selection of organisms and the selection of genomic instructions, although interrelated, nevertheless belong to two logically different categories, and therefore to separate hierarchies of levels of selection. This is what I mean by the distinction between evolutionary selection of genomic instructions and developmental selection of possibly multilevel organisms. But, as I said before, I welcome objections.

In my view, the difference between a self-conscious being ( a subject, of course not necessarily a human) and any other real object is a very critical one. So, it is not useful to think that information exists independently of such beings. Computers are build by us to processing information for us. As long as computers are not aware of itself, the concept of information does not make any sense for computers itself.

Again, Pavel, which novel insight did You gain by modeling biological evolution in a similar way as the development of human societies? The building of a new model of reality should serve in some way a better understanding.

I think that chromosomes (except of the so-called B chromosomes) and taxa (except of species) are not able to be selected against each other. Is there evidence against this assumption?

Is valuable to remember that units of selection (or heredity) are conceptually different from levels of selection.

Genes and Chromosomes (and other molecular and cultural units) are units of selection (or heredity, or inheritance). They compose the heritable phenotypic traits that changes in frequency. But the trait is not a level of selection, since he is a part of whole individuals.

When genes are segregation distorters or B-chromosomes, they became also a level of selection on their own, but otherwise they are inherited because of his effects on selection on the whole organism.

In any level of selection we can have "cheaters" that breaks the fitness integrity of the individual and become a new level of selection (e.g. cancerous cell, laying egg bees), like they were before the transition.

Here we may have some fundamental difference between two views, or approaches, both of which may in some way be legitimate, but which may hinder us from better understanding each other.

I simply adopt the view of abstract information theory - and I know I am far from alone - that admits information to exist even for "computers," or "regulation devices," that have not been constructed by humans. Living organisms are perhaps the best empirical examples. Thus, I am willing to say, and as far as I know in agreement with many modern biologist , that genes "instruct," or "carry the information for," the synthesis of proteins - even if neither the genes, nor the ribosomes that receive the information (after a simple translation) have not been constructed by humans. It is your right to stick to a strict anthropocentric view and refuse to admit existence of information outside ourselves and our constructions, but for me, this view is highly restrictive, obscuring important relationships in the real world.

A little remark about self-awareness of computers: I do have a simple model of a computer that can be said to be self-aware! Can describe on demand.

Now, what I have personally learned is in the first place a better understanding of the crucial difference between the evolution and the development of economies, with important implications for economic policies and reforms. Then I am also finding intellectual satisfaction in discovering similarities between apparently different processes. This has lead me to what believe is a very clear view of how socioeconomic evolution builds upon, and continues, one special branch of biological evolution - and thus how both constitute a unified process of the evolution of life on earth. But I admit that such intellectual satisfaction may be just a personal fancy, shared with perhaps only a limited number of theorists looking for "unified theories," and may not be the taste of everyone.

@veiko

Daniel posted his contribution before I finished my answer to you. Now I only want to make it clear that my answer about the two views of existence of information concerned your contribution.

@Pavel

Thank You for Your detailed and patient answers! I like like emphasize that I see my view not as anthropocentric, I like only to differentiate between subject and object, i.e. between mind and reality. If You have an self-conscious computer (not only a model), this will be very interesting and ethical problematic at the same time. Besides "constructed by humans" - which other subjects You know are able to construct something? Organisms were evolved, not constructed (construction implies some sort of a goal).

@Daniel

My view: SOME genes and chromosomes are units of selection. Individuals are units of selection. Supposedly also species and communities of species. Units of selection are not necessarily units of heredity. All things which could be selected (or selected out) are units of selection. Where selection occurs, there is the level of selection. So what should be the difference between a unit and a level of selection?

Hi Krauss,

I agree that levels of selection are "where" selection acts, which, to me, like you, is not the same of what's been selected i.e. what is changing in frequencies, the units of selection. And a unit of selection is not the same of, as it was to Dawkins, a unit of inheritance.

Units of selection are always evolutionary “individuals” (organisms are obvious individuals, but we have to remember that a multicellular organism can't be taken as given, because in the past it was formed by selection upon groups of unicellular organisms - different levels of selection).

Because organisms are ephemeral (at least more than genes), their traits are inherited due some material inherited particles (genetic and epigenetic) and cultural traits are responsible for cultural inheritance themselves, those are the units of inheritance.

To me, the whole thing is in understand the origin of individuality in the called “major transitions”, and the difference between multi-level selection 1 and 2 (Samir Okasha worked a lot in these concepts, and is the clearer explanation on the subject as far as I know).

A DNA fragment, is always (recombination aside) a unit of inheritance, and sometimes can also be a unit of selection since it's a level where selection can always operate, but normally does not, because organismal evolved cohesiveness (individuality) and the epistatic interaction between then and other factors in development.

A group is always a level of selection, as a context (MLS-1), but not a unit of selection, unless the group through evolution became a new individual (MLS-2).

I think the main questions made in this subject are:

What's been inherited? UNITS OF INHERITANCE

What's surviving/reproducing differentially? UNITS OF SELECTION (INDIVIDUALS)

Which levels are been beneficiated? LEVELS OF SELECTION

@veiko

I will be short, packing for my return to Prague tomorrow. I may add a few more point on Tuesday. But now only two little questions:

1) In a humanly constructed thermostat, you admit that what the sensor is sending to the temperature-changing effector is information. Why do you refuse to use the same term when a logically similar thermostat is part of a mammal?

My basic working hypothesis is that this is also information, and, more generally, that studies of thermostats and other automatic regulation devices, whether or not constructed by humans, have nothing to do with mind and consciousness. These are interesting philosophical, or perhaps even transcendental issues, but in my mind entirely irrelevant to such studies, and even to studies of levels of selection

You asked me what I learned from developing my artificial theory of levels of selection. Actually it was more than I wrote last time, and I may add a few more points next time. But now I only wish ask you a similar personal question:

2) Veiko, why did you ask your question about levels of selection? What useful for yourself did you hope to learn? And what, if anything, you have actually learned that you had not known before?

Well, I see that they are actually three questions, but I hope you will excuse me. And in any case: many thanks for asking this question, it allowed to take part in a very interesting debate.

@daniel

My simple model considers only levels of selection, which it classifies into two categories: (i) developmental, selection for phenotypes (possibly multilevel), and (i) evolutionary, selecting genomes. But i do not explicitly consider units of selection.

Your question makes me realize - thank you! - that biologists like them and that I should therefore also try to speak of them. My first idea is that a level of selection is more general, possibly involving different, or differently large units of selection. For instance, at the genomic level, the units may be single genes, or "cooperatives" of several genes, or even combinations of genes and non-genic DNA coding for regulatory RNA. But they are always made of DNA (or RNA), although of possibly differently large packages of it.

I am afraid my dealing with inheritance is rude. I consider it to be only one special way of realizing the memory of important instructions (excuse me Veiko). I can conceive of general evolutionary-developmental processes where such instructions are preserved over time by memories based on other "technologies." But I recognize that this requires a heroic generalization and abstraction, which many biologists may not be very willing to accept

*Beneficiated = benefited,

Sorry.

@Pavel

Units of selection are an important distinction in multi-level selection's differentiation. MLS-1 deals with groups as context (as Inclusive Fitness Theory), and what is surviving/reproducing differentially are the organisms, but the groups are benefited as consequence. MLS-2 is for cases where whole groups survive and reproduce, hence became individuals on their own. This is an important distinction, and the lack of it caused years of debate in biology literature, and now is cleared.

Your visions on "information" are shared by biologists and philosophers that seek similarities between biological and cultural evolution, and Eva Jablonka's Symbolic Inheritance System is the better definition of what you been dealing. But there are some crucial differences as well between those two "forms" of evolution that needed to be accounted. As a biologist I'm indeed interested in those similarities and differences.

@Veiko

Some animals besides us make constructions e.g. Beaver's dams, termites mounds. Not deliberated as humans, but can.

Thank You, Daniel and Pavel, for answers and questions.

Construction. Generally, I am very careful to use typically human-bound terms for animals in a anthropomorphic way. Did the beaver have a plan of a dam during his activities? I don't know. In case of termites, I am sure that there is no goal, only a result. Remember that the combs of bees are build according to the bodies of the bees itself. Insects are certainly not self-conscious.

Units/level of selection. If You search for positive selection in molecular evolution, You will end up with probabilities and single nucleotide changes. Irrespective of this, You need the whole organisms acting in certain environments to end up with a selection coefficient which depends of many parameters of this whole system. Thus in my view the level of selection is the unit of selection, as many parameters may change without to change the selection coefficient (Transposons and B chromosomes are special cases, i.e. parasitic genomic elements).

But my question was about EVIDENCE of different levels of selection, not about units of selection or something else. As yet only Jeff has provided an answer for me. Thus, if I will have read the papers which he proposed, I will have learned something about the levels of selection.

"1) In a humanly constructed thermostat, you admit that what the sensor is sending to the temperature-changing effector is information."

This was a misunderstanding. I didn't admit that, but this is good example. If I am cold, my body will react with muscle tremor and goose skin. Its not information for me. Its reality. But now I could give You the information that I am cold. But if I look at the thermometer and see "-10°C", that's an information. For me. Not for the thermometer.

Veiko:

The question you ask is most important for me but it is too huge for prompt discussion. My understanding of evolution is a bit different from that of Richard Dawkins. I give here citation from “Biological species as a form of existence, the higher form”, published as a book chapter in “The Species Problem: Ongoing Issues” (InTech), 2013. Try also:

Shcherbakov V.P. Stasis is Inevitable Consequence of Every Successful Evolution. Biosemiotics 2012; 41-49. Shcherbakov V. P. 2010. Biological species is the only possible form of existence for higher organisms. Evolutionary meaning of sexual reproduction. Biology Direct 5:14.

The reductionistic methodology is not an adequate tool for operation with the hierarchically organized world of life. The idea of gene as a replicator is bewildering. Gene is not a living entity. It is not a self-replicator. It is replicated. It is a replica or a template. Genes are manufactured by cell, just like all the other cell constituents: RNAs, polypeptides, organelles. Only self-reproducing substantive entity can serve as a unit of selection. In case of asexual lineages it is individual organism (not gene of course). In case of obligatory sexual organisms, deme (or group) selection is the only meaningful level of selection. Deme is the lowest substantive entity that reproduces itself. Opponents of the group selection reject it as a too slow process: “lower-level selection easily trumped a higher-level selection. First, the group selection may be rapid enough: a generation of population (the unit of reproduction) is of the same longevity as an individual organism; second (and uppermost), the lower-level selection in itself is the destructive side of the overall process. If it is not trumped by the higher-level selection, the group simply will not go through (with all its constituent organisms).

The mechanism of species evolution (microevolution or species ontogenesis) presumes a group selection. Successful groups may prosper, while less successful shrink and extinct.

Competition and struggle between the groups may play little role in their fate: they parish or prosper primarily because of their own merits, their ability to reliably reproduce themselves.

The idea of multilevel selection has now received a substantial support. It is a step in

the right direction (as compared with selfish gene theory). However, I think that here remains some inconsistency. Given that the higher level of selection operates, the selection at the lower levels must be forbidden because it can produce nothing but casualties like a parasitic DNA or a malignant cell. The so called ultra-selfish genes are factual parasites with a net harmful effect on the host. They, along with other parasites and harmful mutations, are representatives of the destructive force of nature. Evolution in action is an unending struggle against this force. Successful evolution ends in stasis. Biological species is organization, which is the cooperation of individual organisms. A hierarchical organization presumes submitting behavior of parts in favor of the whole. Selection presumes the selection of genes (just as when a poet creates a verse it ‘selects’ words) but the vector of selection is “for the good of species”, not “for the good of gene”! A gene is not a living entity, not organization with the function of survival! It is strange for me to insist on such a self-evident statement. These two goods coincide. If they not coincide, the gene will be rejected “for the good of species”. The reverse (rejection of species for the good of gene) is nonsense.

@ Veiko,

I'll thank you too for the excellent topic and opportunity to contribute and learn;

I'm also think is good to be careful to use typically human-bound terms for other life forms, in a anthropomorphic way, since this generates some misunderstandings the can be only, in the end, semantic.

And I see "construction" only as results of behaviors as well, that can come to being instinctively, by trial and error or responsiveness of environmental changes, and sometimes deliberated and self-aware, but I guess humans are the only that posses the last. Niche construction perspective deals with how organisms actively change the relation between then and environment, changing their niches and selective pressures, but no intentionality is assumed, only when dealing to our own species.

I think is important differentiate units and levels of selection when we have a multilevel selection scenario. Selection acting in two levels (e.g. Organisms/ Social Groups or Unicellular Organisms/ Multicellualar Organism), initially we have the organism as level of selection because they that are surviving or failing to. So, two Levels but one Unit.

For evidences of multilevel selection, look for paper of Richard Michod who studies Volvocacine algae, in a transition from uni-to-multicellilarity, when two levels of selection operates.

The book "The Major Transitions in Evolution" From Maynard Smith and Szatmary, is an excellent theoretical account but full of citations from empirical literature.

@ Victor

I think that your insigths on gene selection are very useful, and I agreed with you so far. They are units of inheritance, but not selection.

But:

"In case of obligatory sexual organisms, deme (or group) selection is the only meaningful level of selection."

I agreed that the individual is not the focal level, but the mating pair, thinking in light of MLS-1. A deme is also another level where selection can operate in cases of sexual reproduction, but not always, cheaters individuals are diferent from cheaters genes because they, by the same reason you pointed, can reproduce thenselves. And with selection on species level is the same. Demes of a same species can be geograficaly isolated in a way that they can't contribuite to species benefit.

@veiko (and partly also daniel)

Due to differences in our education, or specializations, we may describe the same empirical things in different terms, and perhaps even ask about them different questions. For you, Veiko, information is only what people can say to people - whereas whatever is treated in thermostats, computers, robots, within brains ... etc., is not information. For me, in contrast, what these devices use is also information - and, I am afraid, this understanding of information is now widespread. But OK, I am definitely in favor of the rights of minorities, it is only important to realize the differences in the uses of same words. Thus, for you, what's circulating in your nerves is X, and not information; information is only what you can say about X. For me (and today many others), X is information, and what you call information is already "information about information" - if you like multilevel things, you might perhaps be tempted to consider joining us? :-)

We may also disagree on how to call the X that all beavers have in their heads that make them construct dams, and more fundamentally, what they have in their genomes that make their brains develop with this X. It is only that I find it very convenient and enlightening to describe both this X and its genomic basis in terms information and instructions, or programs, or algorithms. How to call it, we may disagree, but the fact that this ability of theirs must stem from some special order in their genomes via their brains is hardly deniable - unless expecting that each beaver, when constructing its dam, is guided by some supernatural being.

Concerning different levels of selection: I also found the selection of communities of different bacteria, as referred to by Jeff, to be a very interesting case, that also deserves to be considered. I general, I fully agree that there are many levels of selection of phenotypes, their parts, and possibly also their groups, communities and societies.

It is only that I wish to go deeper, and find out WHY - that is, what is the source of the abilities that make all those different acting and interacting "agents" of different levels so successful? This is what leads me to genomes, and to the information, or the basic "instructions," that these must contain. Just like energy does not fall from the sky, neither does information - this is perhaps the main advantage of the information theoretical view that demands, like Sherlock Holmes, when an "agent" (machine or organism) systematically does A and not B, from where does she/he/it have the information telling it/her/him to do so.

And it is when I explore the WHY that I arrive at their genomes. There, I found only one level of selection within biology. Next level of selection is the one on institutional rules for human groups and societies. Even the communities of different bacteria must thank for whatever success they obtain, to the genomes of their members (including the possibilities that they laterally switch parts of the genomes). This is why I find it important to separate the levels of selection of phenotypes, which may be many, from the selection of genotypes and parts of genotypes, which are very few.

Allow me to remind once more all of participants of this debate of my new paper, where all this is explained more clearly and systematically (on only 23 pages)?

http://kie.vse.cz/english/new-pavel-pelikans-working-paper/

The paper is submitted to a respectable biological journal, but will certainly have to be revised, which means that all comments are still welcome and will be most gratefully acknowledged. Many thanks in advance!

@veiko

You are right, of course, a gene does not build a phenotype. It does not even build a character. It only influences it in cooperation with the rest of the genome (at least).

I was only wondering if that would not be enough to fit the definition, i.e. permit selection between genes. Thank you for clarifying, I should have been more careful not to confuse terms here by sloppy writing.

@Paavo & Veiko

I am glad there is at least one point on which Veiko and I agree: the statement "a gene builds a phenotype" is wrong. But I would like to add a few details of what I have learned as an amateur biologist about what guys are the real builders, and why they build what they build, and not something else. Please correct me, you professionals, if there is anything I got wrong.

So, for my down-to-earth engineering mind, the basic builders are ribosomes, and not genes, and not even any gene-cooperatives. Genes only contain the instructions - or the recipes, if some biologists, like Veiko, may dislike the term "instructions" - for the building, by ribosomes, of linear proteins, which self-fold into an operational form, and thus become building blocks of a possibly multilevel self-assembling puzzle. Its result, obtained after a possibly long process of ontogeny, is a cell, or a multilevel organism, or even a society of organisms (although not human). Many of the recipes are for quite universal bricks that may be used in the building of many different organisms. The Hox genes carry the recipes for special proteins that act as organizers of the building site. And there are now known even higher-level organizers that are not proteins, but long RNA regulators, coded for not by genes, but by non-genic DNA (part of what was earlier believed to be junk-DNA). Moreover some additional instructions may also be supplied by non-DNA epigenetic markers. And of course all this may be influenced by inputs from environments - although only withing genomically determined limits. E.g. they cannot make a cat's ovum develop into a dog, or a dog's brain develop into a brain able to understand multilevel selection.

Now, unless some of this I got wrong - please tell me! - I remain convinced that the distinction between EVOLUTIONARY SELECTION of genomic instructions (possibly of differently long DNA segments), and DEVELOPMENTAL SELECTION of phenotypes, their parts, and possibly also their groups and societies, is the only way to avoiding confusion about levels of selection. The levels do not correspond to each other - there may be many more levels of the latter than of the former, which in non-human biology, is just one. (Again, more about it is in my paper.)

@Pavel

Scientific language is filled with metaphors. This is often unavoidable, but we should be always aware of this simple fact. So, "genetic information", "genetic instructions", "genetic recipe" or "gene-cooperatives" are metaphors for genes or genomes, not genes or genomes itself. If we build models out of metaphors, these models will fail to explain any part of reality.

Neither ribosomes, nor genes or genomes are "basic builders" of cells. Cells emerge today only from other cells, that is, from a dynamic, compartmented, nevertheless highly ordered assemble of many different macromolecules. Life is an emerging property of this complex interaction which results of course in development (and in evolution in the long run). No single molecule or molecular complex could be named a builder of a cell. Besides, non-coding RNA is transcribed from non-coding RNA genes within the genome, and neither non-coding RNA nor epigenetic markers add "additional instructions", they rather modify the existing regulation of gene activities.

There are different levels of selection, but not different sorts of it. Only complete phenotypes could be selected, the genes and genomes will be selected according to these phenotypes.

@veiko

I agree with the following paraphrase of the beginning of your last message (NB the differences!):

Scientific language is filled with WORDS (or other symbols). This is ABSOLUTELY unavoidable, but we should be always aware of this simple fact. So, "genetic information", "genetic instructions", "genetic recipe" or "gene-cooperatives" are WORDS referring to genes or genomes, not genes or genomes itself. If we build models out of WORDS, these models will EXPLAIN some parts of reality, and fail to explain other parts ... - depending on the choice of the words, and more precisely, on the categories into which we choose to partition the observable "reality," and to which we then ascribe words.

Personally I don't know how you could build and communicate models from REAL genes and genomes. I find the idea of communicating by sending each other pieces of REAL DNA a little bizarre.

If we agree that both you and I must use words corresponding to categories of observable things, then our basic disagreement is that we partition the world into different categories. You like to treat life as a single big blob "... an emerging property of this complex interaction which results of course in development (and in evolution in the long run)." These WORDS are for my taste too vague and fuzzy. I prefer to cut LIFE into smaller pieces, and VIEW genes and other parts of DNA and RNA as carrying information and instructions. You are of course free to refuse this view, but your understanding of multilevel selection will be correspondingly affected.

Without denying you this freedom, allow me to note three arguments for this view:

1) Longtime ago, in his Introduction to Cybernetics, Ashby divided the questions that biologists can ask about a chicken's ovum into two categories:

- A) Assume that the ovum knows how to develop into a chicken, and ask: where from does it get all the needed materials and energy to do it?

- B) Assume that all all the needed materials and energy are available, and ask: why does it develop into a chicken, and not a frog or just a tumor?

Ashby then shows that Question B leads to information problems, and tries to convince biologists (half-a-century ago) to VIEW their problems also as information problems.

I find this view fruitful in economics, and believe (and know of many biologists who also believe it) that it is also fruitful in biology.

2) The information contents of strings of DNA - or in this case rather RNA - most sharply appears in observations of REAL viruses attacking REAL bacteria (or other cells). What I have learned is that they simply take over the REAL building guys of the cell (i.e. the ribosomes etc), and give them INSTRUCTIONS to build copies of themselves (the viruses), and not parts a similar cell. Doesn't this REALITY make you claim that "Cells emerge today only from other cells" a bit oversimplified? Could you accept the following qualification?

What emerges from a cell depends on the genomic instructions actually obeyed by its ribosomes - which may be a similar cell, or a mutated cell, or a bunch of viruses.

3) As far as I know, most of modern studies of phylogeny have switched from recording it in terms of actual organisms with their anatomic features - which sometimes appeared to make unexplainable jumps - to recording it in terms of their genomes (DNA and RNA sequences), in which all such jumps appear smoothly explainable. This switch implies the understanding that the DNA and RNA sequences carry instructions for the forming, developing and behaving of the organisms.

I believe that these arguments clearly indicate that the view in which information contents is considered separately from the material carriers used is fruitful - but of course you may remain unconvinced.

Allow me to add that I see your claim "There are different levels of selection, but not different sorts of it." only as a result of your partitioning of reality. In fact, you should say "I can see ..." and not "there are .... In contrast, in may partitioning of reality, I CAN SEE two sorts of levels of selection - those of genomes ("instructions"), and those of according to them formed, developed and operated phenotypes ("agents"). I believe this view to be clearer and more fruitful - which is why I am submitting it for consideration to biologists. But I am far from imposing it on anyone. I have full respect for biologists who are happy with their current view, in which genomes and organisms are mixed together an put into one big bag.

And more thanks for the continuation of this for me highly enlightening debate!

Pavel, I am afraid that You are not willing to understand me. I was speaking of metaphors, not of words. I was telling You that cells (today!) only could emerge from other cells, not about what cells will became later on. And selection occurs only at phenotypes, not genotypes, so there are not a sort of selection on genotypes and another one at phenotypes.

And besides:

" I find the idea of communicating by sending each other pieces of REAL DNA a little bizarre." Thats Your idea. I have never mentioned such nonsense.

An advice: If You like to know and not only to talk about evolution, please study biology. Science is fruitful if I can say "there are ..." and not only "I can see...", because the second phrase is independent from knowledge.

Veiko, I am afraid that YOU are not willing to understand ME. For me, metaphors are words, or abstractions put in words, or more "scientifically," theoretical models. The problem only is that we abstract in different ways. So, if you still ask about levels of selection (= you are still unclear about them), my way of abstracting cannot help you, for you are impermeable to it - so just forget it.

Concerning your way of abstraction, I believe I can see it, but find it unsatisfactory to my more precision demanding mind. Communicating by pieces of REAL DNA was a joke - but in fact a strictly logical implication ad absurdum from your downgrading of my efforts to communicate by means of an abstract conceptual model, and not REAL genes and genomes.

I thank you for your advice to study biology, but it comes a little bit too late. If you only looked at the beginning of my paper, instead of judging it unworthy of your attention, you would find there the information that I this is exactly what I have been doing - as well as I could - during the last 20 years, of which the last 10 benefited from help by my daughter, who got a couple of years ago her PhD in biotechnology. In fact, I have the immodest impression that I referred to more specific biological facts - e.g., the instructing by the genome of a virus the ribosomes of the by it invaded cell. You just answered by a big and for me obscure package that life is too many things together which cannot be disentangled by some "artificial" abstractions (models, "metaphors").

I am afraid it is not my lack of knowledge of biological facts, but rather may way of drawing abstractions from them and theorizing about them, that makes me incomprehensible to you. You simply make other abstractions and theorize in other ways about the same facts. I know several biologists with whom I find mutual understanding - but obviously you are not one of them. But I wish you good luck to progress with your way of abstracting (and I claim you must abstract even when you conduct the most concrete experiments with REAL LIFE), and I'll try to progress with mine.

@Pavel,

I understand what you mean, but to me, there's a flaw in your abstraction because what you call information (but could be called heredity as well), are not a level of selection, only units of inheritance (or informative particles if you will).

Viruses are composed of units of inheritance; but are selected due their phenotypic effect on the parasitized cell and for have independence as an individual (unit of selection).

But cellular "cooperative genes" are not units of selection, only the cells are.

What's selected is different from what's inherited. I think that is useful to distinguish that, but, like Veiko, I can't agree that there's two kinds of levels of selection.

There's just one kind, with multi-levels of phenotypic expression and organization.

And there are some different units of inheritance as well.

Http://plato.stanford.edu/entries/selection-units/#3.4

@veiko

I apologize for the low precision of my answer. To increase it, I begin by agreeing with you that there are indeed many empirical facts about which science has enough reliable evidence to be able to conclude "they ARE."

But I insist that there is also some space for relative subjectivity of how we classify them, what importance we assign to their different aspects, and whether we consider them sufficiently similar or significantly dissimilar. It is on this that we may only say "I see ... " or "I cannot see ..".

In our case: I agree with you that both organisms and genomes ARE, but we disagree on the importance of the differences between the two. You SEE them similar enough to classify their possibly different levels into one hierarchy, whereas I SEE them different enough to classify the levels separately.

The difference that I SEE, and you do not, is basically the one brought to light by Ashby (cf. my earlier reference to him): in my abstraction, I SEE genomes to carry the crucial information-instructions for the forming, developing and operating of a certain specific organism, and not another. I see the difference similar (OK, this is only a superficial analogy, but it appears often to help) to the one between cakes and the recipes for making them. In this analogy, what you do is mixing recipes with cakes, while I believe it fruitful to consider them separately. OK, your alleviating circumstance is, that organisms are special cakes, as each of them has in-baked a copy of its recipe. But I still SEE merit in reserving for the recipes a different logical bag from the one for the organisms.

@daniel

we do not yet understand each other well enough. In my abstraction (conceptual model), heredity is only a technology for preserving key instructions over time - in other words, a way of realizing a memory. It is a material necessity in biological evolution, but not in socioeconomic evolution: some of its key instructions may be preserved by other technologies.

As to the units of selection, I prefer to avoid specifying them to sharply, not to fall into the same trap as Dawkins: they are not always ONLY genes, and sometimes even NOT genes at all (e.g. the ngDNA coding for regulatory RNA that appears to determine much of the overall architecture of a complex organism). I simply do not know precisely what they are, but only claim that they must be more or less long segments of DNA or RNA.

True, the tangible and directly observable selection is only the one of the resulting organisms - like tasting the ready-baked cakes. But organisms die and cakes are eaten, and what can LASTINGLY be selected are only the RECIPES for making them again and again. But to SEE this selection requires a certain abstraction, which perhaps not everyone likes to make. But if you make it, you may then view organisms (and cakes) only as testing grounds for their recipes.

Pavel,

1)"heredity is only a technology for preserving key instructions over time - in other words, a way of realizing a memory. It is a material necessity in biological evolution, but not in socioeconomic evolution: some of its key instructions may be preserved by other technologies."

I understanded your abstraction. Besides the commom properties that you rightly presented, I think you fail to distinguish some differences in your objects of study. I disagree that biologically speaking, inheritance = selection. "Information"(genetic/epigenetic) is inherited, not selected; only phenotypes are selected.

And what I using as a concept of units of selection is not similar to Dawkins. I'm talking of well defined "individuals" (as conceived in major transitions literature – link below).

Http://plato.stanford.edu/entries/selection-units/#3.4

Dawkins replicators are what I'm calling units of inheritance (That are inherited, not selected).

In cultural evolution (which may include economics) phenotypes themselves can be inherited, and in this case selection = inheritance, because the” information” is the phenotype itself (e.g. learned behaviors, symbolisms).

Those differences are important.

2)" But if you make it, you may then view organisms (and cakes) only as testing grounds for their recipes."

This is the same as Dawkins distinction between replicators and vehicles, and if one is not cautious enough can't realize the crucial empirical flaws of it. (Better explained also in the link, above in that answer).

.

Daniel,

I am far from claiming "inheritance = selection". Instead, I like to refer to the 1965 generalization of Darwinism by Campbell in terms of "variety, selection, retention" and to see "inheritance" as one of possibly several technologies of "retention."

For instance, I consider it to be the critical error by Hodgson and Knudsen, in their generalization of Darwinism for economists, that they reduced the Campbell general triple to the more biologically specific one "variety, selection, inheritance." This form, which is much less applicable to socioeconomic evolution, has turned many economists, even some evolutionary ones, away from Darwinism.

In my paper, I am quite explicit about the similarities and differences between Dawkins and myself - esp. on pp 3-4. So, emphatically, my model is definitely NOT THE SAME as Dawkins's.

Thank you for the link, it is very instructive, and it may even be practically useful to me if the reviewers of may paper demand more references. But I do not see my model to fit very closely any of the views considered and criticized in it (I already know many of them, but not all). I also believe to be cautious enough to keep my model free of all the empirical flaws of Dawkins's. But I may be mistaken - this I always expect - so if you see some, please tell me. But so far, as I can see, our differences have been only terminological, giving the same words more or less different meanings..

Pavel

I think that maybe, as you said, is a matter of terminology, but say that change in frequences of "information" (genetic/epigenetic) is not a kind of selection.

I think this has lead to many issues in biology literature and would be helpfull to avoid that.

In cultural variants (behaviors/symbols), witch would be more related to economics, this distinctions are not a rule and the information sometimes is itself the realized trait, the is in fact selected and inherited by the same process.

I think you should consider this difference between biological and "cultural" evolution.

Daniel,

Thanks, you are right, I should make the differences between biological and socioeconomic evolution clearer. One of the problems that is important for evolutionary biologists, but less so for evolutionary economists is the one of frequencies, and its relationship with evolutionary success. In biology, the number of successful agents is crucial for the evolutionary success of their genomic instructions: the more of them, the greater chance the instructions have to be lastingly retained. In economics, you cannot have zillions of economies. The evolutionary success or failure of different institutional rules must be deduced from relatively few cases.

But I believe that this problem can formally be solved by including, among the success criteria for the developmental selection of agents & the evolutionary selection of their instructions, the ability to multiply - BUT only in those environments where this ability is crucial, which is the case of most of living organisms, but not of social and economic organizations. Does this make sense to you?

Yes, it does.

And I think is crucial do differentiate selection from inheritance (in biological systems). Genes and other molecular inheritances are only inherited due selection on the sum effect of then and environment (phenotype). So is less valuable talk of selection of information in biological systems, only sorting. In the same way we don't talk of selection of our Metazoa cells, only in cases of intra-organismal selection(e.g. cancer).

Could anyone provide really good evidence (original articles or good reviews) for group or community selection? I am not convinced by this concepts.

Some that I know:

http://www.pnas.org/content/early/2012/01/10/1115323109.abstract

http://www.annualreviews.org/doi/abs/10.1146/annurev.ecolsys.39.110707.173510?journalCode=ecolsys

http://www.pnas.org/content/73/12/4604.abstract

@daniel

I agree with the importance of distinguishing selection from inheritance. Once more, following Campbell, "selection" must be distinguished from "retention," and "inheritance" is only one special technique of retention.

But I do not know why selection of information would be unimportant in biology. It depends, of course, what we SEE AS information. Following Ashby, we may see information in genes and and other segments of genomes - in other words, these may be seen as carriers of information (instructions) on how to form, develop and operate phenotypes. But this is may not (yet) be the way in which the notion of information is generally understood in biology (despite Ashby's efforts half-century ago). I came across several texts in which biologists tried to apply information theory, and they got quite confused about it.

@veiko

Daniel's references are very interesting (I have read their abstracts), but they all concern groups of cells. For social groups, I believe a good source of references is the long paper by two Wilsons in the QRB of December 2007. As an economist, however, I prefer to avoid the term "group" and speak instead of "organizations" - which I define to include both firms within an economy, and entire economies. They may both more or less prosper or fail, as can be found documented in many economic statistics. I then try to identify the causes of their results in their institutional rules (= the basic instructions at that level). For instance, the fall of the Soviet type economies can be found largely due to their forms of property rights and their methods of central planning. And perhaps the future failures of some capitalist economies will be possible to find in their wrong regulations of financial markets, or the wrong rules of social security ...

@daniel

A second thought on

http://www.pnas.org/content/early/2012/01/10/1115323109.abstract

- the first of your references.

The authors start with "Multicellularity was one of the most significant innovations in the history of life, but its initial evolution remains poorly understood. Using experimental evolution, we show that key steps in this transition could have occurred quickly. ..."

I started to wonder: if it could occur quickly, why did the evolution take over three quarters of its total time - approx. 3 billions of years out of 4 - to find it? At least this is what I read it had done.

Pavel,

You said:

"@veiko,Daniel's references are very interesting (I have read their abstracts), but they all concern groups of cells."

You are wrong, one is of groups of cells, another from groups of organisms (deme level) and the other from species selection.

Multicellulality could emerge experimentally today, first because we are looking for organisms(cells) that are already in transition phase, but we can't take that as given, the history of their evolution has to be consider. Another crucial point is ecological conditions. For example, multicellularity rise is often correlated with the oxygenation of the atmosphere by cyanobacteria, that changed drastically all niches on earth, but that took time.

Pavel,

Another theoretical point to consider is the emergence of eukaryotic cell by endossymbiosis, before multicellularity, with an expanded capacity to produce energy.

Thank You very much, Daniel, for Your references. I am mostly interested on experimental evidences (as You provided). As You rightly point out, major transitions in evolution are critically dependent of some changes of environmental conditions.

Daniel,

I apologize for my misinterpretation of two of the three abstracts you provided. From my limited perspective of an economist, all appeared to be a matter of cells, or even more exactly, of their genomes ... But obviously this was too imprecise. My problem is that when I hear of selection of groups or communities, I tend to think of groups of humans - such as considered in sociobiology, e.g., by Wilson & Wilson. So thank you for pointing this out to me.

Veiko and Daniel,

I would like to know more precisely - in the statement "major transitions in evolution are critically dependent of some changes of environmental conditions" - the meaning of the notion "critically dependent": what depends on what and how?

I may again be mistaken, but in biological evolution, I see environments only as raising problems, to which uninformed ("random") variations (mutations, recombinations ... ) of DNA or RNA must search, by trial-and-error, solutions, and finding them may take a long time. But of course, if the environments change, this raises new problems, which require a new trial-and-error search for new solutions.

The question would then be: Was the long time (approx 3 billions of years) before the first multicellular organisms started to appear due to the difficulty of finding the needed genomes, or was there, just then, such a change of the environments that from their previous hostility to multicellular life they suddenly became hospitable to it?

PS: Daniel, my mistake of putting all your three references into one bag was that I spoke of "cells" - i.e., in my terminology "agents," subject to "developmental" selection, which may indeed be multilevel even within biology. As you rightly pointed out, this was not the case of two of the references.

But I believe I could put all the three into one bag if I considered "genomes." subject to "evolutionary" selection, of which biology has only one level. Correct me, if I am wrong, but I believe that all cases in which "groups" of cells, including endosymbiosis, succeed in a higher level of "developmental" selection, must be due to their genomes. E.g., the cell and its mitochondria have each their genome, and it must be thanks to these genomes that the cell is able to use the services of the mitochondria, and this is able to live so happily within the cell.

@Pavel,

an organism and the environment of this organism modify each other permanently during organismal development. At an higher level, the same occurs during evolution of biological diversity. If You like to learn more about evolution, please read the great books "The origins of genome architecture" by Michael Lynch and "The origins of evolutionary innovations" by Andreas Wagner. Evolution is not an optimizing process but an open-ended one. So, in reality there are no "solutions" and no "needed genomes", only lineages of organisms (contingency!) with more or less reproduction under contemporary conditions.

@Pavel,

Veiko's answer above said most of what I would.

I understand that your unilateral view of environments in evolution is also due the history of biological though itself, but you should take a look on niche construction perspective for a more dialectical view (but with some issues in their own), which can be very helpfull for economics as well.

http://lalandlab.st-andrews.ac.uk/niche/pdf/NicheInheritBiolThr.pdf

In cases of environments in major transitions, ecological factors change depending on the transition itself, and there's no recipe for all.

The 3 Billion years delay for multicellularity could be due the gradual oxygenation of atmosphere, could be for origin of eukaryotic cells and mechanisms that promoved low conflict in and betwenn then, but probably both factor are necessary, although I can't say they were sufficient.

Besides Veiko's suggestions, you should look on Jonh Maynard Smith and Eors Szatmary Major Transition in Evolution (1995) book, as well in Sterelny and Calcott Major Transition in Evolution Revisited (2011).

On philosophical grouds, I'm reading Peter Godfrey-Smith's Darwinian Population and Natural Selection, that seems to have some interesting on that subject, but is too early to say.

@veiko

You write that "an organism and the environment of this organism modify each other permanently during organismal development." I wish to call attention to the fact - implied by the elementary principles of information-processing, which I try partly to recall in may paper - that organisms are MODIFIABLE only to the extent allowed by their genomes. These are the source of all abilities of organisms, including those of letting themselves be modified. More precisely, an organism can modify its behavior with the same genome, if this provides for sufficient learning abilities; otherwise it must wait for a happy mutations, which may take one year or a million of years, that would enable it to enlarge or modify its abilities.

@daniel

I do not see environments "unilaterally." I fully admit that organisms may also modify them, e.g., by products of metabolisms (or industrial production). But the key question then is, whether the organisms will remain able to cope with the by them modified environments, and this I see again depend on their genomes, and the adaptability for which these provide.

I happen to know the arguments by John Maynard Smith and E. Szatmary, and find them interesting, but not deep enough, just because they do not get all the way down (as far as I could see) to the genomic sources of all the abilities of organisms (and their possible groups), including the abilities to learn and adapt.

Pavel,

as You probably know I do not think that so-called "elementary principles of information-processing" has any meanings for organisms. Genomes consist of DNA or RNA, and they do not "allow" something. The organism develops in permanent interaction with its genome(s) and its environment.

I think that You suspect that there are at least some common, evolutionary rules which are valid in both biology and economics. I know that some other people think similarly as You. However, I am sure that economical rules are completely different from the rules of evolutionary biology, in the same sense as physical rules are completely different from those rules. Thus, it makes no sense for You to discuss with us about evidence for different levels of selection in biology.

Pavel,

"I fully admit that organisms may also modify them [environments], e.g., by products of metabolisms (or industrial production) "

But (maybe) you are failing to see how those changes in environments (made by organisms or not) can also be causal to changes in "information" origin themselves, which the paper I suggests deals.

Veiko and Daniel,

I believe to follow most of today's biologists by respecting the Weissman barrier: Environments cannot directly cause adaptive changes in genomes. There is a certain one-way traffic: mutations or other genomic variations must guess, and environments can only approve or reject.

Veiko,

Whether you see it or not, informational constraints are just as general as energy constraints - equally bothersome for machines, organisms and societies. To thank you for your recommending me to read more about biological evolution, allow me to recommend you to read more about information problems. You may begin by old Ashby - it is now freely available on internet (cf. the site I give in my paper). Gödel, Escher, Bach by Hofstadter (1979) is also an excellent mind-blowing reading (you may jump the chapters with formal logic).

It is absolutely essential to realize that no machine or organism can do more - including learning from, and adapting to, environments - than what the basic instructions with which it has been endowed allow it to do. This is as simple and as important to realize as realizing the conservation-of-energy law. Until you realize it, and no longer claim some miraculous exceptions for organisms, you cannot get very far in your search for understanding multilevel selection.

Pavel,

No serious biologist today thinks environment only approves or reject.

Take one example of niche construction behaviors: A beaver can influence the environment of a long time, future generation of beavers through ecological legacy of building dams. Environments coevolve with organisms, and sometimes, adaptively. Some traits are selected (and "information" inherited) not only to survive to an environment, but to influence and modify that environment for development and survival of future generations of organisms.

[See link on my answer above on niche inheritance]

Even in the absence of knowledge of epigenetic inheritance, the majority of biologists knew that environments are causal on development and hence in evolution. A genome has a "norm of reaction" that confers plasticity for traits in different manners to different environments.

Besides that, plants (through asexual reproduction) and unicellular organisms don't show Weismann barrier. More amazing, some matazoa and lots of sexual metaphyta, with the expected Weissman barrier, presented epigenetic inheritance of environmentally biased traits, even in further absence of inducing environment. Several mechanisms can influence these outcomes: RNA interference, methylation, hormonal influences, etc.

Environments can influence and bias rates of mutation, and mechanism that regulate genes activities.

Jablonka and Raz review lots of those empirical works on transgenerational epigenetic inheritance, and it is worth to read.

http://ts-si.org/files/JablonkaQtrRevBio2009.pdf

Daniel,

Many thanks for the interesting reference. But, as you may have noted, my paper does not ignore epigenetics: non-DNA epigenetic markers are there duly included - together with genes and non-genic DNA coding for regulatory RNA - into the broad category of "basic instructions" that guide the forming, developing and operating of phenotypes ("agents"). This is one of my departures from the narrow gene-centered view, e.g., by Dawkins. My instruction-centered model thus fully accommodates even epigenetics. But I recognize that the paper by Jablonka and Raz contains many interesting details that were new to me, and I am grateful to you for recommending it to me.

But we still have some misunderstanding about the roles of environments. Here again, I see the information approach (which unfortunately rings no bells in Veiko's mind) to offer clarity. Admitting, as you point out to me, that environments also have ways to influence the basic instruction of an organism (be these DNA or epigenetic markers), the key question is:

(A) do these influences contain RELEVANT INFORMATION that helps the organism to succeed in these environments?

or (B) are they only another kind of stochastic animators that increase the rate of experimentation with the instructions, but without giving any specific clue on which of the experiments will succeed?

An example of (B) is the increased rate of mutations in bacterias that the environments deprive of their usual food, which triggers the search in panic for a DNA modification instructing the synthesis of an enzyme able to use some food that is available. Many fail, but the few that succeed will multiply - and may even laterally communicate the by chance found successful gene to their neighbors.

(A) would be pure Lamarckism, which not even Jablonka and Raz imply (as far as I could see). They disturb classical neo-Darwinism by pointing out that not all of the "instructions" (in my terminology) that effectively contribute to the ontogeny of phenotypes are written in DNA or RNA, but many may consist of epigenetic markers - which is OK with me. But as far as I could see they do not imply the existence of any good spirit in the environments that would give gentle hints on HOW to change the instructions IN ORDER to make the organism successful.

Hi Pavel,

Those are relevant questions, so pardon me for the extensive answer that will follow:

You questioned:

“(A) do these influences contain RELEVANT INFORMATION that helps the organism to succeed in these environments? (A) would be pure Lamarckism”?

“or (B) are they only another kind of stochastic animators that increase the rate of experimentation with the instructions, but without giving any specific clue on which of the experiments will succeed?”

This point of environment relevance is clear by comparing development between plants (i.e. Coffee - http://www.scielo.br/pdf/babt/v47n6/a05v47n6.pdf) raised on full-sun or under shade. The development is plastic and with the same genome having two different developmental outcomes that are adaptative to each environment. So environmental cue are obviously relevant. But that don't mean that environmental information is apart from genome, is an interaction, and environment’s instructive contribution are a product of previous selection on plastic and responsive development, but that's not enough to say that we should see environment as only what approves or reject.

Analyzing your possibilities:

First is necessary to understand that many cases of environment cues, results on noise on development and are deleterious, but that doesn’t ruin the possibility of some been beneficial. Is like genetic mutation.

(B) – Stochastic events: Environments are only approving or rejecting in, e.g. stress responses? No! They play a function of "inform" to genome that things are bad out there”. The environmental cues in that case don't determine the outcome as you rightly said, but it raises the probability of it happens in a population. It’s RELEVANT "information" and the capacity to “receive” it is probably adaptative.

(A) – Lamarckism: I know that you recognize other kind of “informations”. But Dawkins too recognize other replicators (Dawkins, 2004). The problem of this is deeper. Is not only that they ARE epigenetic that is intriguing, but that they can be environmentally biased.

Eva Jablonka considers those as a kind of Lamarckian inheritance (which is debatable, since Lamarck’s theory involves lots more than soft inheritance), other authors don’t. But what is obvious is that is a case of soft inheritance, where environments influence what information is achieved and inherited. This can be achieved due selection of more plastic and responsive developments (though genetic and epigenetic inheritance). If soft inheritance is sufficient to account something Lamarckian, than bacteria, that have inheritance of acquired traits shows that this happens. I don’t seem necessity of account that as Lamarckian.

Anyway, I think you should pay more attention on development. In biological development and cultural learning, the limits of environmental “information” is related to genomes capacity to interact, of course, but that don’t mean that they are not important or “informative”, nor that it can’t change the future rules of environmental approval / rejection .

I think that a thermodynamic view may resolve some of the problems and paradoxes mentioned above. Life is an irreversible thermodynamic process. As such it is dependent on the dissipation of a generalized thermodynamic potential. The most important thermodynamic potential available to life today is solar photon potential. The overwhelming thermodynamic work that life performs is the dissipation of visible and UV light into infrared light. This is done primarily by the organic pigments in cyanobacteria and plants in contact with water. This production of entropy is the "driving force" behind life, the reason for its existence. Life "strives" to augment the entropy production of Earth in its solar environment.

The evolutionary history of life on Earth has been one of increases in entropy production overall, and per unit biomass. Take for example the succession of ecosystems. The older the ecosystem, the greater its entropy production (lower albedo and lower emitting temperature). Onsager's principle of the coupling irreversible processes to augment entropy production is in line with this.

Within this thermodynamic view, selection is based on overall global entropy production. However, the non-linear, non-equilibrium nature of life means that evolution to higher entropy production is a historical process, with many available meta-stable stationary states. The actual path taken by life will depend sensitively on initial conditions and subsequent external perturbations, however, the general trend is to increasing entropy production, and this is what selection is all about.

This selection will occur on any level, from the genome to the biosphere and will involve the biotic with the abiotic. Please read my article entitled "Biological catalysis of the hydrological cycle; life's thermodynamic function" http://www.hydrol-earth-syst-sci.net/16/2629/2012/hess-16-2629-2012.html

You may also find interesting my thermodynamic view on the origin of life;

http://www.earth-syst-dynam.net/2/37/2011/esd-2-37-2011.html

@Karo

Your thermodynamic view is undoubtedly interesting, but I am not sure I understand it clearly enough. Especially I cannot judge the chemical plausibility of your claim that "RNA and DNA ... can reproduce ... without the need for enzymes." I only know that some RNA is now widely recognized capable of doing so - cf. the self-catalyzing RNA molecules now hypothesized to be the first forms of life. But I guess that there may be more problems with DNA. However, because of my imperfect understanding what thermodynamics has exactly to do with the origins of life, besides supplying the energy needed for some Darwinian "agitating and experimenting" (I confess that I have similar difficulties with understanding, on this point, even Prigogine), I cannot relate it very closely to the issue of multilevel selection or the one of genomes vs. environments.

@Daniel

Concerning the latter issue, I fully agree with you that an organism's "development is plastic and with the same genome having two different developmental outcomes that are adaptive to each environment. So environmental cue are obviously relevant." I even recognize it very explicitly in my paper, where I give the example of the food that determines whether the bee will develop into a queen or a worker. But I also emphasize that whatever development alternatives there are, they must all be genomically prepared, including the specification of the inputs on which the choice among them will depend. In other words, environments may only chose through genomically prepared inputs from genomically prepared lists of alternatives.

A communication problem may be that all today's biologists are certainly very competent users of their computers, but probably have not yet reflected enough on the general architecture of the information processing that the computers allow them to conduct. I strongly believe that reflecting more on it could usefully clarify what information processing can do, and what it cannot. A more detailed (although still very elementary) explanation is in the Section 2 of my paper (allow me to recall its site: http://kie.vse.cz/english/new-pavel-pelikans-working-paper/).

But here, for the genomes vs. environments issue, it may perhaps suffice

(1) carefully distinguishing between information INPUTS, and the PROGRAMS (networks of instructions) for receiving and interpreting them

(2) realize that there cannot be inputs without some pre-existing and for them making room programs.

The key questions about environments then is: (i) do they provide the organism with HELPFUL ADDITIONAL INSTRUCTIONS for its genome, or (ii) are they only INPUTS treated according to some of the already existing genomic instructions?

Clearly, only answer (ii) is Darwinian. Answer (i) is again pure Lamarckism. Note that both your example of alternative development of plants and mine of alternative development of bees correspond to answer (ii).

Daniel, this distinction may also help you clarify your statement: "In biological development and cultural learning, the limits of environmental “information” is related to genomes capacity to interact, of course, but that don’t mean that they are not important or “informative”, nor that it can’t change the future rules of environmental approval / rejection." As is, I consider this statement very fuzzy or even confused. But with the above distinction, you may clarify the meaning of "informative" and decide whether (i) or (ii) - I expect you to choose (ii).. or?

Pavel,

Maybe you and I will just agree to disagree, but I think my point was not made clear, so some considerations can be made:

1) I read your paper since you had posted it, and I know well your positions. I find it very interesting and the reason I keep debating is that I think that you can do even better, and im my humble opinion, your view is lacking some crucial differentiation between some phenomena.

When making a generalization like in your paper, is a double-edge sword strategy. We can unite and have a big picture of a general process, but as well we'll have the risk to bypass important distinctions. In the end we have to measure our ratio of being a lumper/splitter.

2) I appreciate the analogy with computers and information, and many times catch myself thinking parallels between that and biological "information". But analogies, besides been powerful tools for reasoning, leads us to tempting mistakes if some empirical distinctions of differences are not made. And that’s the reason to some biologists be skeptical of using it, like Veiko, and I see a good point in his skepticism.

Computer’s information is more suited analogous to learning and communicating, so cultural evolution has not all the problems that biological evolution has when compared to technological information, and that is expected since we used our understanding of information (nervous systems based) to make computers work alike.

3) Biological entities have developmental and life cycles, computers don't. There's no computers reproduction, with material overlapping between generations, and that's not how information is passed by computers. Because of that, there's no way of compare how environments act on biological development using computers.

That being said, I can answer your question between (1) or (2) - In the case of plant development, bee castes development, is obviously an (2), because (2) are a evolved adaptative development, but we can’t take it for granted because environments can cause non -adaptative changes in development by breaking canalization, and those changes can be deleterious, neutral or beneficial, in the last case, selection will work out to produce a (2), but it begins with a simple stimulus that genomes "know nothing about yet", and this initial stimulus can be (1), soft inheritance, via epigenetic inheritance. In the paper I suggest, Raz and Jablonka, you will see empirical cases that are non-adaptative, hence, is not logical to say that this change in development is due previous selection on genome always. Selection can came after, and with environmental information "triggering" a new developmeny AND a new selective regime upon a responsive genome to cope with it.

Environment is not only approving or rejecting.

This is not Lamarckism because is not a “determined instruction” emerged by necessity, which is what Lamarck really wrote about, but cleary is soft inheritance and selection can evolve it to a responsive, adaptative plasticity.

Daniel,

I most appreciate your answers and am very grateful for them. I am also learning a lot from them, and already have some ideas on how to improve, thanks to them, my paper. If it is published I will not forget to thank you for them.

I do not think we must end up by agreeing to disagree, but strongly believe that a good agreement between us IS possible. That we have not yet reached it is undoubtedly my fault - my pedagogic abilities are not yet sufficiently adapted to biologists. But this is precisely what I want to learn, and hope that I am making some progress.

One thing I should express better is that my recommendation to think of computers is NOT to draw DIRECT ANALOGIES (or metaphors) with living organisms, but to get a helpful glimpse of the GENERAL PRINCIPLES OF INFORMATION-PROCESSING, from which not even living organisms can go free.

I suppose that all biologists now agree that both machines and organisms are subject to the same energy constraints - roughly: neither can use more calories than what they can get. The point I am trying to explain is that they are both equally subject to severe information constraints.

The difficulty is that these constraints are more complex, involving loops and branches, in which it is easy to get lost. The general advice I may now think of - but don't know if it is understandable enough - is: for each piece of information, search not only for its sources, but also for the needed pre-existing instructions that allow it TO BE INFORMATION.

It is this search that ultimately leads, as I try to explain in my paper, in the case of computers to their hardware instructions, and in the case of living organisms, to their genomic instructions.

But these instructions are only seeds from which complex information structures, with many higher levels of instructions, and to a large extent thanks to inputs from possibly changing environments, may be growing. But without these seeds nothing would ever grow, and it is also these seeds that imply, be it only very indirectly, the limits to which they may in any environments, possibly grow (cf. the cat's ovum that no environments can convince to develop into a dog ...).

What I find insufficiently clear is to say, as many biologists appear to do, that "organisms and their environments co-evolve." I admit that their respective changes may be interdependent, but I find it necessary, to avoid confusion, to make it clear that there are important INFORMATION ASYMMETRIES. All the informational influences of environments cannot but DEPEND ON the organism's genome, which limits both their admissible scope and their possible effects.

But now you are right that epigenetics complicates this picture, which I did not take sufficiently into account - whence your pessimism about our possible agreement. But what I argue - and hope you will agree - is that all the complications can be taken into account, yet the main principle fully preserved. It suffices to admit that the environmental information is not limited to inputs of DATA, treated according to some of the organism's actual instructions (which may be genomic, or developed from these) - to which I limited attention last time. What is also necessary to admit is that the input information may have the meaning of PARAMETERS that the instructions will ASSIMILATE and start using. But I believe that the main principle holds: even such assimilating ultimately depends on, and its possibilities are constrained by, the initial genome.

Intuitively: you cannot hang new information in the air, but need a solid hanger on which possibly many other pieces of information already hang, but you just cannot do without it.

Pavel,

My pessimism about our possible agreement maybe was rushed and I too think we can agree, at least in some more points.

Some things, the way you putted above, are more clear and dealing with more necessary complexities. I just want to comment more 2 points, and made final suggestions:

1) Organisms and Environments coevolution - I also appreciate the perception of the "information" asymmetry that we have there and I also have a question on RG on niche construction, which to me is a too broad concept, and sometimes ignores that and lump things together. When I said about the need to see environments as more than a selector/disposer, is not that I think they have a real evolution and heredity, but only that, though life's activities we can change environments in ways that would change life too.

Think in the example of gene-cultural coevolution of lactose absorption: Human cultures that lived in regions with milk availability in the past, today shows a different genetic composition about lactose absorption. With culture (of drinking milk) the populations changed the selective environments for our genes the allowed us to do so. And human evolution has been that for a while - change environments to suits us, today and tomorrow. Beavers dam's do it too, as plants activities in soil and some other interesting examples. Selective pressures and environmental resources are dynamically changed too.

2) I also agree that genomes are the main ”information” reference that all life has as background condition and environment, through epigenetic, cultural and other ways influence development. But is useful think that even that information is dependent of cellular apparatus that make then come to being. So in the same way environments need genomes to express influence, genomes need cells components, and that1s why I think your "hardware analogous" should not be genomes but cells as a whole (with information and something to read than) and developmental systems in multicellular organisms. Genes need an evolved "reader" to make sense, as cultural “information” needs brains.

And why I don't think that we can see that all evolution involves" information" selection:

Technological and Cultural information are usually transferred through replication, and can be independent of reproduction (material overlap). This will apply to computer data, ideas, symbols and behaviors.

Some organic entities can does the same, replicate, through copy, without reproduction with material overlap, e.g viruses, prions, horizontal gene transfer, transposons, etc. In those cases the units of information/heredity are the same as the units of selection.

The "common" inheritance of genes and epigenetic factors that are only transmitted from parent to offspring, between organisms, cannot be equated to the replication form described above, because there's real reproduction, and the "informational" or inheritance units change in frequencies only due the sum effects on the level of organisms, and in this case the units of selection are the organisms. Information is only sorted by downward effect.

The both cases involve information change, but only the former could be called selection.

Daniel,

I agree with all you say, but believe all of it is included as special cases in my general conceptual framework.

An example of what you say: "But is useful think that even that information is dependent of cellular apparatus that make then come to being." - Yes, but I see the relationship between the two to resemble the one between a DVD and a DVD-player. The important information is on the DVD, but you need the player, which is rather standard, to get it. That the player is standard can be seen from the earlier example of the cellular apparatus that can play the "DVD" of a virus - and produce viruses ...).

Another example: concerning the epigenetic factors you mention, I would emphasize that they can mainly block parts of existing genomes, which may then slowly be lost (e.g. the genes for enzymes making possible the digestion of milk), but cannot invent new genomic instructions.

Pavel,

I think you haven’t understood the point of my examples. Yes, they are special cases, but what I'm saying is that these special cases are the ONLY cases with selection on the level of inheritance particles themselves/information itself! Virus, selfish dna, prions, cultural variants, computer information are transmitted or inherited directly, like the information in DVDs. This are cases that we don't have true reproduction, only replication. In a multi-level selection perspective, those are the only cases where, e.g. genes are selected for their effects on the genic level. When genes are inherited though effects on selection on phenotypes of organisms or social groups, is NOT gene selection that are happening! This is ONE of the points why Dawkins gene eye view (replicator eye view to be more precise) has limitations as a heuristic method.

My point on transcription/translation apparatus was NOT for virus-like inheritance. Like digital information, they really work like the DVD information, but "cooperative genes" from genomes don't suit the metaphor, because they usually don't have independent "information", due epistasis and heterosis (and most of times, due shared fates through reproductive bottlenecks), and are inherited together with the apparatus (the dvd-player is copied along with the dvd, and the dvd just shows images that are composed together with other dvds - this point shows how the analogy breaks, there are no dvd images that are compose of various dvds).

The idea that ALL selection has a description on the level of genes or another information replicator is flawed, and would be true only if genes or epigenetic factors possessed a direct and additive effect on phenotype, which is not the case, the most common are epistatic effects that are “visible” to selection as a compost phenotype, only on the level of organism or above.

The other point you mentioned:

New epigenetic variants can arise and are called epimutations, introducing novelties like a mutation, they have to beguin sometime don't they? And they not only block part of genomes, methylation and histones works like that when they already exists in a state X and change to state Y, but metabolic loops and RNAi don't. A epimutation that occurred on a parental organism can be inherited as a novelty due its effects on phenotype of offspring, even if no gene related to that trait is inherited. The persistence of a given epimutation will depend on how selection will proceed, it can be lost, or maintained, and even accommodated in development due a future mutation.

To a clearer difference between selection base and detection base epigenetic inheritance:

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3116147/

Daniel,

Many thanks for more of interesting cases of selection, from which I gratefully learn new exciting details, but which I still see to fit the basic logic of the distinction between (a) selection of INSTRUCTIONS and (b) selection of according to them formed, developed and operated AGENTS.

I find it indeed interesting to note that some instructions may be eliminated as such, e.g., because they are wrongly written, without ever getting the chance to try to govern the forming of any agent. But this is OK with me. In my favorite recipe-cake example, I see them comparable to confused recipes that are thrown right away, as according to them no cake can be baked.

I also find it interesting to learn that there are many (at least three) ways in which epigenetics can work. But I see all of them basically to belong to the category of instructions, or changes of instructions. Some of them may indeed be more than just markers glued on parts of the genome, but may even affect genomes deeper, and thus be one of many possible sources of genomic mutations. But again, unless they could be shown intentionally helpful - issued by a good spirit in the environments, helping the genome to lead to an organism successful in these environments - they are just that: another source of uninformed mutations.

Once more, I am most grateful for all the interesting detailed examples, which have increased my still imperfect knowledge of biology. But how do YOU relate them to the distinction between (a) and (b)? Do you see them to contradict it, or do you agree with me that the distinction remains basically sound and useful?

If Veiko still follows, I hope that he will note that the article to which you refer speaks of "epigenetic INFORMATION channels" ...

Pavel,

No need to thank - is a good exercise and makes me rethink and re-study some points as well, what's always good.

1) On information and agents:

A) Since you prefer to describe the process like that, just think that in my "special examples", that the information ARE the agents themselves, and in the default cases selection act on phenotypes(agents) and change frequencies of genes and epigenetic factors (information).

B) I keep thinking that in default cases, the word selection is inappropriate to describe what’s happening to single replicators/information. Is not that the recipe is a "bad" one, even if is only an alternative good one, the "words" only have meaning together with other "words", so the cake will be selected, the "words" in the recipe will be sorted, change in frequencies, but not be selected. The causal process of selection is on the "cake". Different from the special cases where "words" are their own "cakes", and hence, are selected. I don't disagree that "words" will change in frequencies, but selection is not only that, so I can't equate sort and selection.

A good paper on the subject:

http://www2.hawaii.edu/~khayes/Journal_Club/fall2006/Vrba_&_Gould_1986_Paleobiology.pdf

2) On epigenetic inheritance:

"But again, unless they could be shown intentionally helpful - issued by a good spirit in the environments, helping the genome to lead to an organism successful in these environments"

- We'll not find it, unless by coincidence, which is very unlikely.

That leads back to your question of (a) or (b).

There are two different cases of epigenetic inheritance that seems to occur:

Selection base: Is like you said, works like a random mutation and are subject to natural selection to be fixed or lost. (But it is a source of information, like mutation is, as you was not recognizing).

Detection base: Because o prior selection on the past, organisms has a plasticity and can respond to environments cues adaptively. Notes that it is only possible by previous selection on development, so, it’s not Lamarckism, but environment contributes too with information cues that can be fixed epigenetically and genetically after or not.

No one of those cases fits the "good-spirit", otherwise, it would be Lamarckism indeed, but as evidences shows, and even logic by all we know, this can only happen through the two routes above, so both evolves selection, after and/or before environmental stimuli.

Daniel

I am glad that our learning is mutual. But I am afraid that we still see some things quite differently. In your last response, there are several points which I do not quite follow - but you may have your reasons to see them as you do. I cannot consider them all, just a couple:

- "... the information ARE the agents themselves": In some very general sense, this might be said, but this would not get you very far. I consider most fruitful to focus on the information in the instructions (genomes) according to which the agents (organisms) are formed, developed and operated. Then the equation appears to be true only for the self-catalyzing RNA molecules, with which life is now believed to have started - they may indeed be said to be both the instructions and the agents. But the following "division of labor" between the DNA storing the instructions, and the proteins formed according to them and then forming the agents, calls for a clear distinction. At least I would get confused if I mixed the two.

In fact, this is precisely why I get confused when reading the paper by Gould and Vrba, who mix in the same multilevel hierarchy genes and organisms. This is also what I see to be the main reason why they cannot get (for my understanding) the story very clear. I appreciate Gould very much as a knowledgeable palaeontologist and an excellent story-teller, but not as a logical thinker.

".... Because of prior selection on the past, organisms has a plasticity and can respond to environments cues adaptively." I prefer to be more precise and say first that the "plasticity" (=learning abilities, adaptability) of each organism is due to, and LIMITED BY, its genome (possibly including some epigenetic markers). And only then observe that the genome results from the past evolutionary selection from the variety of the genomes that have actually been tried - i.e., given the chance to guide the forming, developing and operating of organisms, submitted to developmental selection.

Allow me to note that it was an economist, Armen Alchian, who made the simple but important point, which not all biologists appear always fully to realize, that "selection can work only on what has actually been tried."

Pavel,

I agree fully with your 2nd paragraph, but I agree because that’s not the point.

The point of Vrba and Gould's paper was to show you that sorting (change in frequencies) is different from selection (direct selection).

A gene who change is frequencies because of their effects on the gene itself (e.g. transposon), has its phenotypic effect on the level of the bearer of information. A transposon is in the same time the molecule of heredity (info) and a phenotypic expresser of it (agent). The transposon is selected because a causal process on gene level. HERE the gene is selected, directly - and it's fitness depends basically of his information.

A gene that changes in frequencies due its effects on an organism, has its phenotypic effect on a different level from the information bearer. The gene is the molecule of heredity (info) and the organism is the phenotypic expresser (agent). The gene is selected because of a causal process on the level of organism. HERE the gene is "sorted" or indirectly selected, if you wish - and its fitness depends on all other genes on that same organisms.

That's all I'm saying. You agree?

About epigenetics: We are saying the same think, don’t you agree?

Adaptative plasticity exists BECAUSE prior selection, and genetic and epigenetic past selection limits it, because is something already tried. Environment enters like an activator cue.

And epimutations are like mutations, random due effects on fitness. Environments can acts like mutagens, inducing epimutations.

But, the reason that we began to disagree was another: is not right to say, like you said before, that environments only accept or reject. I think you can now apreciate that don't you?

Daniel

I do not see the conceptual difference between "sorting" and "selecting" to be so great - I simply see the former to take gradual steps that may eventually lead to the latter (e.g., decreases of certain frequencies until these drop to zero ...).

The words I describe "plasticity" and environmental "cues" differ from yours, but describe basically the same phenomenon - i.e., I prefer to say that the development of an organism proceeds according to its genomic instructions, which are open to (have conditional continuations in function of) certain, by the instructions prepared, environmental inputs (your "cues").

I do appreciate your reminder that environments do more than either accept or reject, but with a qualification. I forgot indeed to say that environments may be, and often indeed are, sources of mutations that propose new variants of genomes for the selection tests. But this is rather obvious - even if you are right: I should say it. I nevertheless continue to argue that environments DO NOT do more than "either accept or reject" as sources of RELEVANT information. We agreed that they do not contain any good spirits giving clues on which mutations would be beneficial So, regardless of how their influences might be labeled - be it as "epigenetic" or otherwise - they fall into only two broad categories "turning the roulette" and "saying yes or no" to the randomly produced results.

I do not have more comments. I thank you very much for providing me with challenging examples that at first sight may have appeared not to fit my conceptual model - but now, after I have thought of them, I can see that all of them do fit. I'll leave it up to you to see whether my model can serve you as well as it can serve me (and hopefully also some biologists), or whether you need a less abstract model that distinguishes more notions as entirely different, instead of seeing them as special cases of one more general category.

In any case, good luck with all your research projects,

Pavel

Pavel,

I'm afraid that you didn't fully get what's the difference of sorting and selection in a multilevel scenario, taking by your description, but maybe I'm currently unable to explain better my points, and if you wish to further read on those questions, I'll suggest you two books on levels of selection that helped me clarify those issues, if you don't have read then yet, or even if you did, take a look sometime, maybe will change your view, maybe don't.

http://www.amazon.com/Evolution-Levels-Selection-Samir-Okasha/dp/0199556717

http://www.amazon.com/Darwinian-Populations-Natural-Selection-Godfrey-Smith/dp/0199596271/ref=sr_1_1?s=books&ie=UTF8&qid=1365426294&sr=1-1&keywords=darwinian+populations

Good luck for you too, in whatever projects that you have, and besides our disagreements, and even most for then, was a pleasure to change ideas here.

.

Daniel

I must add a little clarification. There are certainly many aspects of biological selection that I do not get, and actually am not trying to get. As I put hopefully clearly in my paper, the purpose of my model is not to explain multilevel selection in all relevant details, but only to offer a few additional insights into it.

I know the main lines of both Okasha's and Godfrey-Smith's arguments, and recognize that they raise and discuss many fine points that I do not take into account. My only point is that they both suffer from one omission, which I believe makes their arguments more complicated than necessary - namely, they do not systematically go all the way down to the basic instructions according to which all the possibly multilevel subjects of the multilevel (developmental) selection are formed, developed and operated (true, some attention is paid to "the gene's point of view," but this is not enough). In biological evolution, this would lead them to the SINGLE LEVEL of genomic instructions (possibly complemented by epigenetic markers) , and in socioeconomic evolution, to a combination of that level with the one of the institutional rules of human groups and societies.

This is in essence the "instruction-centrism" of my model that I believe may help simplify and clarify parts of the multilevel selection issue. But certainly not all parts, and perhaps not for everyone.

Pavel

They do address the agent/information view, which is a refinement of vehicle/replicator view.

Okasha thinks that the referred framework can't accommodate epigenetics and cultural evolution, which I disagree for cultural evolution, and also for epigenetic marks (methylation on genes). But when thinking of some other epigenetic mechanisms, like changes on histones disposition, structural inheritance of proteins and metabolic self-sustained loops, thinks complicate, because the effects are on development, not on genes like methylation marks. I think an agent/information can cover those cases too, but missing some things that as you said, you are not trying to get to your model.

Godfrey-Smith thinks that the general framework of agents/information, is very good abstract way of see biological and cultural evolution, but lacks some consideration, e.g. a gene besides been information, is a physical entity, a materialistic reality that are inside your developmental levels, different from culture for example. I don’t think this is important for your model, but it’s for biologists interested in the real life proces.

Thank You for participation, Pavel and Daniel.

However, I like to suggest to focus on evolutionary biology and my original question about concrete levels of selection. Are there any additional comments on this topic?

Daniel,

I agree with what you see as limitations of the Okasha and Godfrey-Smith approaches. I would refer to these two authors in a similar way, if the reviewer of my paper asked me to include them.

But I would then also add emphasis on how many people use the word "information," while meaning more or less different and not always well-defined things. In this respect, I believe Section 2 of my paper, with its two elementary principles of information-processing, to be helpful.

Veiko,

Thank you for raising such a highly interesting question, just a few days after I started my adventurous excursion into biology with a paper just about it. I apologize for driving the discussion to spheres (to avoid confusion I do not use the term "level" in this methodological context) that you consider too abstract - or not enough concrete.

The problem with me (and many economists) may be that I cannot understand "concrete" things without some good "abstract" framework (a clear logically consistent "model"), into which they would reasonably fit.

The most concrete way I may think of selection levels in biology - and this may be only a very naive view - starts bottom-up, with the embryogeny of an organism instructed by its genome (using certain environmental inputs). The genome itself is a result of the Darwinian "lottery" cum EVOLUTIONARY SELECTION, which I suggest to consider apart. But once it is there, I imagine (from what I have read) the embryogeny and ontogeny to unfold as a multilevel "production" process, in which there appear to be strict quality controls, with DEVELOPMENTAL selection and rejection, at virtually each level - form the assembly of proteins through organelles, cells, individuals, and possibly even societies of individuals. You have certainly much more concrete knowledge of what concrete biochemistry these levels may involve. My point is only very general and simple, yet not always properly realized: all these "developmental" levels unfold from the single genome.

I now wonder: do you find this ABSTRACT distinction between the two logically different kinds of selection of any help in your search for the CONCRETE levels of selection, or do you still find it entirely irrelevant. And may you give me an example of what you see as a CONCRETE level of selection - I would very much like to consider it.

Pavel,

additional discussions between us would be purposeless. Obviously, You are not so interested in biology, and I am not so interested in unrealistic models. Reality does never fit into a model (as You said), the opposite is true: A model have to fit into reality in at least some important aspects to improve knowledge.

Veiko,

Allow me to object a little: I am more interested in biology (among other things, thanks to my daughter with her PhD in biotechnology, which I followed quite closely) THAN, and my model is not as unrealistic AS, YOU THINK. In any case, I do not see it more unrealistic than most of the other theoretical models of evolutionary biology. How realistic or unrealistic a model is SEEN TO BE often depends, much like the proverbial beauty, on the EYE of the observer. This in turn depends on the abilities to abstract - such as seeing similarities between apparently different things and/or differences between apparently similar things. I immodestly believe that my model does fit into reality in SOME important aspects, know of some biologists who believe it too, and only regret that you are not among them. But be more modest and limit your verdicts about any theoretical models to what YOU can see in them, without making sweeping statements on what they ARE.

In any case, I regretfully conclude that my way of theorizing about evolution is definitely nonsense to YOUR eye and mind. On the other hand, however, I remain interested in what YOU may see as CONCRETE levels of selection! When you find some, please let me know, I promise to try to understand them, and will not bother you in return with any of my for you too abstract reflections. Be as concrete as you wish - in the worst case, my daughter will help me understand it!

Good luck with your concrete research,

Pavel

Veiko,

I apologize for my stray from your main question, but I think this subject, multi-level selection demands some abstraction even in the experimental works.

This paper may interest you (and others here), is theoretical but full of interesting, concrete examples of selection operating in at least 2 levels, from cases where the transition is beginning to cases where is fulfilled.

Queller, using the concept of "organismality" (degree of being an organism), analyses them comparatively, availing degrees of cooperation and conflict.

I found most interesting because the abundance of examples of cases where is possible apply a multilevel selection perspective and the references leads to some practical works for the cited examples.

Http://rstb.royalsocietypublishing.org/content/364/1533/3143.full

Daniel, thank You very much for the literature.

I apologize for my often very direct speech which may seen to be sometimes disrespectfully. However, one of the reasons why I try to condense my text is to improve understanding. Pavel, You do not follow my arguments, and I do not follow Yours. Your PRIMARY interest seems to be to establish some general (i.e. not exclusively biological) models consisting at least in part not of scientific termini, but of metaphors of it (We have discussed this already). So You are not MAINLY interested in biology. Fine. I am not interested in unrealistic models. (1) As I suppose that your models as discussed here are unrealistic as based upon invalid terminologies and (2) my time is limited, I decided to be furthermore not interested in Your models. Besides, the potential (and real) levels of selection I see I have already pointed out in the question and in some of my comments.

Daniel, science consists of validated models and needs always abstraction, but essentially with strong relations to reality. Otherwise, You will not be able to made successful predictions about matter.

By the way, I am now especially interested in arguments and evidence for and against species selection.

Veiko

What is "exclusively biological" is now an open question - an increasing number of social scientists now consider the evolution of human societies to be a natural continuation, and indeed an "organic" part, of the evolution of life on earth.

I did try to follow your argumentation, but could never quite get your point - with the exception of your sharp refusal to admit any "abstract" or according to you "unrealistic" notions into some by your delimited "exclusively biological" area. And then you tried to convince me that in biology, things are too complicated, with too many things influencing each other, and you sounded as if they could not be disentangled - e.g. genomes and organisms, or organisms and environments ... Fortunately, I did some reading of other biologists, who see these things, and the possibilities of disentangling their mutual influences, more clearly than you.

Which leads me to a question by which I would like to check whether your abstraction abilities have at least a little progressed since the beginning of our discussion. If you remember, in the beginning you did not even want to admit "information" into biology. For you, information was only what intelligent humans tell each other.

So now, what do you make out of the paper on epigenetics, recommended by Daniel, entitled "epigenetic information channels"? Have you progressed, so that now you do admit the notion of "information" into biology, or do you still wish to keep this notion out of it, and therefore consider also the authors of this paper not to be concrete biologists, but only "unrealistic" modelers?

And please, don't forget to inform me about concrete levels of selection, when you find some. I am still interested in your arguments, if you make it clear what you really want to say.

Veiko,

"Daniel, science consists of validated models and needs always abstraction, but essentially with strong relations to reality. Otherwise, You will not be able to made successful predictions about matter,"

That's something I fully agree, and that's what we have to seek as biologists.

It's becoming clearer that the separation between evolution and development through the Modern Synthesis did some harm to the reality of our heuristic models of selection, that's why biologist are more interested today in less abstract models. When the distinctions made by Dawkins and company were made, the replicator/vehicle model covered lots of gaps in our understanding that the classical model was failing to do.

Today, the classical model was improved with adding development, expanded inheritance and a multi-level perspective, filling those gaps, and the replicator/vehicle was improved by the information/agent model. Both can be useful as heuristic models, and as you said, some biologists will prefer your way to see.

I think the classical models improvement are closer to the reality of biological systems, and the Reproducer (Grisemer, J. 2000) and Darwinian Populations (Godfrey-Smith, P. 2009) concepts are dealing with that better. The big point is not that if you call inheritance information or not, but if you deals with it in an ontogenetic way.

The three last paragraphs are for Pavel. I'm sorry, I forgot to indicate that.