Thursday, February 11, 2016

Replaying life's tape

Intelligent Design Creationists are promoting a structuralist view of evolution [What is structuralism?]. The idea is that the history of life as we know it was largely preordained by the initial laws of physics and chemistry. According to this view, once life got under way it was inevitable that it would eventually evolve the way it did resulting in humans or something that closely resembles humans.

Structuralism is based on the idea of intrinsic forms that severely limit evolutionary pathways. These forms are constrained, and defined, by the physics of matter and energy. Some creationists like this idea because they believe that god created the universe and fine-tuned it for life. According to their faith, once the original laws of physics and chemistry were set up it was just a matter of time before humans evolved. These creationists can make their belief in a creator god compatible with evolution as they see it.

(Let's not forget that there are many structuralists who are legitimate scientists and some of them are atheists. Structuralism is not a creationist invention.)

The conflict between structuralists and others is brought into focus by Stephen Jay Gould writing in Wonderful Life.
I call this experiment "replaying life's tape." You press the rewind button and, making sure you thoroughly erase everything that actually happened, go back to any time and place in the past—say, to the seas of the Burgess Shale. Then let the tape run again and see if the repetition looks at all like the original. If each replay strongly resembles life's actual pathway, then we must conclude that what really happened pretty much had to occur. But suppose that the experimental versions all yield sensible results strikingly different from the actual history of life? What could we then say about the predictability of self-conscious intelligence? or of mammals? or of vertebrates? or of life on land? or simply multicellular persistence for 600 million years? (pp. 48-50)
Gould strongly believes that if you replay the tape of life it will not turn out the same. He thinks that the history of life is unpredictable for several reasons such as ...
  • mutations are "random" and the timing of mutations can greatly influence the path that a lineage follows
  • not all changes are due to adaptation and natural selection—many are just accidents
  • there's no evidence that the forms we see are the only possible forms or Baupläne
  • there's more to the history of life than just evolution and these other events, such as asteroid impacts and plate tectonics, play a major and unpredictable role in the history of life
The major emphasis in Gould's writing is on "historical contingency." What he means is that the path that lineages take through time is not constrained by any intrinsic forms but mostly by what happened before. He also means that some historical events—such as the K/T mass extinction—have nothing to do with evolutionary theory which means that the tape of life could have many different endings. This is connected to his claim that there's more to macroevolution (history of life) than just lots of microevolution (natural selection and random genetic drift).

One of those possible endings is a planet with no vertebrates, or no mammals, or no humans. Indeed, if you start the replay early enough, there may never be any multicellular species, no photosynthesis, no endosymbiosis leading to eukaryotic cells.

Gould revisited this idea in The Structure of Evolutionary Theory devoting many pages to challenging the structuralist ideas of D'Arcy Thompson who wrote Growth and Form, the "Bible" of structuralism. He also critiques the more modern views of of Stuart Kauffman and Brian Goodwin. He devotes two long chapters1 to the problem.

Chapter 10: The Integration of Constraint and Adaptation (Structure and Function) in Ontogeny and Phylogeny: Historical Constraints and the Evolution of Development.

Chapter 11: The Integration of Constraint and Adaptation (Structure and Function) in Ontogeny and Phylogeny: Structural Constraints, Spandrels, and the Centrality of Exaptation in Macroevolution.

Most Sandwalk readers will never delve into those chapters to extract the essence of Gould's idea so let me try and explain it. Keep in mind that this is a difficult topic and if you are really interested in criticizing Gould's view of life there's no excuse for avoiding the hard work needed to justify your opinion.

Here's an example of Gould's thoughts on the predictability of the history of life (page 1156)
If a group of Martian paleontologists had visited earth during the Eocene epoch, they would have encountered two coexisting, and scarcely distinguishable species of the genus Hyracotherium. If they had then followed the subsequent history of the lineages, they would have watched one species differentiate into the clade of rhinoceroses and the other into the clade of horses. But if a modern commentator then concluded that horses and rhinos had existed as distinct designs in their modern form (lithe runners vs. horned behemoths) since the Eocene, we would laugh at such a silly confusion ...

With respect to the Cambrian Explosion (the subject of Wonderful Life), the issue is whether anyone could have predicted which body plan would survive and give rise to the species in modern phyla. He points to species like Xidazoon (right) as an example of a body plan that didn't leave any descendants and asks how that fits into structuralist notions.

But even more profound is the idea that the bilateran body plan itself may not have been the only possibility.
If the basic developmental patterns of bilaterans arose quickly, and have remained fixed in form since then, do these historical invariants represent a set of mechanistically limited and excellent, perhaps even optimal, designs that natural selection would have established in much the same way at any time and under any ecological of geological regime? Or do they represent just one possible solution among many numerous and entirely plausible alternatives of strikingly different form, each yielding a subsequent history of life entirely different from the outcome actually experienced on earth? ....

Historical constraint based on developmental homology assumes great importance in either case, but if the particular constraints that actually set the channels of bilateran diversity could only have arisen within a narrow range of basically similar and workable states, then much of life's pageant unfolds by predictable regularities of natural selection. If, however, the developmental plans actually established during the Cambrian explosion—albeit eminently workable, and therefore exploited by natural selection to build the particulars of life's later successes and failures—represent only one contingently-achieved set among a broad range of alternatives (each "equally pleasing" to natural selection), then life's actual pageant on earth becomes highly unpredictable, and the happenstance of a realized beginning (the historical constraints of bilateran developmental homology) assumes a far more prominent role in shaping the subsequent history of life.
What he means is that just because the basic body plans of modern animals all evolved from a bilateran ancestor doesn't mean that this pattern was preordained by the immutable laws of physics and chemistry. It could be a accident of history that was locked in by constraints on subsequent evolution (historical contingency) just as the genetic code may have been a "frozen accident."

The gedanken experiment (thought experiment) of replaying the tape of life is an important way of focusing attention on divergent views about the history of life. As we have seen, Gould doesn't believe that the replay would look anything like the original.

It's worth reminding readers of the play on words in the title Wonderful Life. It not only refers to the wonderful life of the Burgess Shale but also the well-known movie "It's a Wonderful Life starring James Stewart and directed by Frank Capra. In the movie, an angel takes the hero back in time to show him how different life would have been if he had never been born. It's an example of replaying the tape of life and getting a different result.

Gould's analogy is presented in the form of a thought experiment but that doesn't mean the conclusion is based on thought alone. In fact, there's a great deal of scientific evidence to support the idea that the history of life is not the entirely result of predictable laws of form.

Most people don't know that Gould published a series of papers on the predictability of evolution back in the 1970s. He collaborated with his colleagues David Raup and Thomas J.M. Schopf to investigate computer simulations of evolution to see how they matched up with the actual fossil record. The collaboration began in the summer of 1972 at the Marine Biological Laboratory (MBL) in Woods Hole, Massachusetts (USA) and the resulting model was known as the MBL Model. E.O. Wilson's student, Dan Simberloff was part of the original group.

Later on the group was joined by Gould's a graduate student Jack Sepkoski whose thesis topic was the construction of a large database documenting the history of life who worked on coding the simulations. The history of this history has, in turn, been well-document by Sepkoski's son David Sepkoski an historian of science at the Max Planck Institute for the History of Science in Berlin (Germany). See Sepkoski (2016) for a concise account.

The work culminated in a number of papers showing that the actual fossil record looked very similar to simulations where the survival and extinction of clades was essentially random (Gould et. al., 1977
Raup, D.M., Gould, S.J., Schopf, T.J., and Simberloff, D.S. (1973) Stochastic models of phylogeny and the evolution of diversity. The Journal of Geology, 525-542. [PDF]

Raup, D.M., and Gould, S.J. (1974) Stochastic simulation and evolution of morphology-towards a nomothetic paleontology. Systematic Biology, 23(3), 305-322. [doi: 10.1093/sysbio/23.3.305]

Schopf, T.J., Raup, D.M., Gould, S.J., and Simberloff, D.S. (1975) Genomic versus morphologic rates of evolution: influence of morphologic complexity. Paleobiology, 63-70. [PDF]

Gould, S.J., Raup, D.M., Sepkoski, J.J., Schopf, T.J., and Simberloff, D.S. (1977) The shape of evolution: a comparison of real and random clades. Paleobiology, 3(01), 23-40. [doi: 10.1017/S009483730000508X]
The results are nicely summarized in the abstract of Gould et. al. (1977)
The history of life is replete with apparent order. Much of this order may reflect the deterministic causes conventionally invoked, but we cannot be sure until we measure and subtract the order that arises in simple random systems. Consequently, we have constructed a random model that builds evolutionary trees by allowing lineages to branch and become extinct at equal probabilities. We proceed by dividing our simulated tree into clades and by comparing their sizes and shapes with the patterns exhibited by “real” clades as recorded by fossils.

We regard the similarity of real and random clades as the outstanding result of this comparison. In both real and random systems, extinct clades arising after an “ecological barrel” had been filled have their maximum diversity at the midpoint of their duration; clades arising during the initial “filling” reach an earlier climax during this preequilibrial period of rapid diversification. However, some potential differences also emerge. Clades still living are much larger than extinct clades. We may attribute this to the morphological superiority of survivors, but we can also simulate it in a model that chooses the originators of clades at random. Real clades undergo greater fluctuations in diversity than do random clades, but the effect is not marked.
This can be simplified by saying that the actual history of life looks random and unpredictable except for some effects due to historical contingency.

The idea of replaying the tape of life dates back to a letter Gould published in response to Niles Eldredge in 1976 after Eldredge had criticized the totally random interpretation of the fossil record (Eldgredge, 1976; Gould, 1976). Gould agreed with the criticism and that's where historical contingency met with randomness to make replaying the tap of life the better analogy.
Eldredge, N. (1976) Differential evolutionary rates. Paleobiology, 2(02), 174-177. [doi: 10.1017/S0094837300003456]

Gould, S.J. (1976) The genomic metronome as a null hypothesis. Paleobiology, 2(02), 177-179. [doi: 10.1017/S0094837300003468]
Here's part of Gould's letter. We see two themes here that come up again and again in Gould's writing. The first is that we must follow the data and not our intuition. The second is that we must have a clear null hypothesis and this null hypothesis cannot beg the question.
... if we could replay the tape of life, might coelacanths become the teleosts and beetles the monoplacophorans? In short, empirical differences in rate of speciation exist—how could it be otherwise in a non-typological world. But we must know whether these differences are an essentially random scattering about a mean rate (the metronome hypothesis) or a determined and predictable feature of certain morphologies and environment. ... (Frankly, I must confess that my own intuition leads me to Eldredge's conclusion—that the differences are predictable consequences of functional anatomy and environmental space. But I cannot vindicate this intuition with any satisfactory data. And scientists must learn to mortify their intuitions when they arise from the philosophical baggage that we all carry as products of Western intellectual traditions—i.e. a basic belief in determinism, mechanism, directionalism, etc.)
Note that this letter was published two years before the Spandrels paper was presented (Gould and Lewontin, 1979) and it echoes some of the same ideas in that paper.

And it's not just historical contingency that makes evolution unpredictable ...
Another point in favor of the metronome and one not yet honestly faced by Darwinians—is the possibility that a relatively large amount of the enormous genetic variation in natural populations is not controlled by selection. It matters rather little whether this variation be strictly" neutral" (sensu Kimura and Ohta) or "slightly deleterious" (sensu Selander, so that mutation rate balances the small disfavor of selection to produce an equilibrium). In either case, the control of genetic change would lie with stochastic factors (genetic drift, mutation rate) unrelated to any deterministic controls on varying rates of speciation among higher taxa. These non-Darwinian factors almost surely act to decrease the among-group variance in genomic rates of evolution.
The point here is that the concept of unpredictability in the history of life is based not on wild speculation but on solid evidence from the fossil record and computer simulations and a firm understanding of modern evolutionary theory and history.

If you replay the tape of life there's a good chance that no intelligent life will evolve. There's no evidence that the actual history was pre-determined by any fundamental laws of form and there's certainly no evidence of purpose.

You can believe that life was purposefully designed if you like but such a belief flies in the face of (i.e. conflicts with) science [see Full House: The Spread of Excellence from Plato to Darwin].

If you want to learn more about Gould's view of life you can't do better than to read Stephen Jay Gould: Reflections on His View of Life.

If you want something shorter try his essay Eight (or Fewer) Little Piggies in which he says,
Never apologize for an explanation that is "only" contingent and not ordained by invariant laws of nature--for contingent events have made our world and our lives. If you ever feel the slightest pull in that dubious direction, think of poor Heathcliff, who would have been spared so much agony if only he had stayed a few more minutes to eavesdrop upon the conversation of Catherine and Nelly (yes, the book wouldn't have been as good, but consider the poor man's soul). Think of Bill Buckner who would never again let Mookie Wilson's easy grounder go through his legs--if only he could have another chance. Think of the alternative descendants of Ichthyostega, with only four fingers on each hand. Think of arithmetic with base eight, the difficulty of playing triple fugues on the piano, and the conversion of this essay into an illegible Roman tombstone, for how could I separate words withoutathumbtopressthe spacebaronthistypewriter.


Photo Credit: The photo of Jack Sepkoski and David Raup is from the University of Chicago and was published in David M. Raup 1933-2015.

1. All of the chapters in that book are long. It took me the better part of a day just to re-read two chapters.

Gould, S.J., and Lewontin, R.C. (1979) The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proceedings of the Royal Society of London. Series B. Biological Sciences, 205(1161), 581-598. [doi: 10.1098/rspb.1979.0086]


Sepkoski, D. (2016) “Replaying Life's Tape”: Simulations, metaphors, and historicity in Stephen Jay Gould's view of life. Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences. [doi: 10.1016/j_shpsc.2015.12.009]



58 comments :

  1. Still trying to understand structuralism. Is it just an extreme point on the continuum of evolutionary constraint, in which the other end is the adaptationist caricature that anything can be selected and that the environment is a perfect molder of form? Sensible people will be somewhere in the middle with Gould. Yes, there is constraint, and it arises alike from physical possibilities, the existence of possible pathways from where you are to somewhere else, and the existence (or not) of variation leading along any given pathway.

    ReplyDelete
    Replies
    1. I agree...I must be sensible.
      I think the proper way to pose the thought experiment is not to ask what we'd see if we replayed the tape of life once, but what we'd see if we replayed it 1000 times. Off the cuff, I'd say if we replayed it from 600 million years ago to the present we'd see recognizable chordates 90% of the time, vertebrates ~75%...of those land dwelling verts 100% etc.
      I think most of the randomness comes from the major extinction events but even these aren't mostly random.
      I think its relevant that Goulds ideas on this were tied to his belief that many of the unusual animals of the Burgess were members of bizarre extinct phyla. But most of the animals have since then been assigned to extant phyla.

      Delete
    2. Your statement is ambiguous. What is your presumed starting point 600 million years ago? In your opinion were there already chordates at that point? Vertebrates? We don't actually know.

      And even with a given starting point we have very little data from which to estimate probabilities of outcomes.

      Finally, what do you mean by "chordate" and "vertebrate"? Do you mean organisms with certain key features, perhaps a notochord or bones? If so, what about iodine-binding cells in the floor of the pharynx? Would a taxon lacking those still count as a chordate to you?

      Delete
  2. Just a small quibble:
    Later on the group was joined by Gould's graduate student Jack Sepkowski whose thesis topic was the construction of a large database documenting the history of life.
    Sepkoski (note the spelling error in the text) started constructing his Compendium after his thesis work. The references are
    J. J. Sepkoski, Jr. 1982. A compendium of fossil marine families. Milwaukee Public Museum Contributions in Biology and Geology 51:1-125.
    J. J. Sepkoski, Jr. 1992. A compendium of fossil marine animal families, 2nd edition. Milwaukee Public Museum Contributions in Biology and Geology 83:1-156.
    J. J. Sepkoski, Jr. 2002. A compendium of fossil marine animal genera. Bulletins of American Paleontology 363:1-560.

    It's also worth taking note of
    Gould, S.J.(2001), "Contingency" in Briggs D.E.G. & Crowther P.E. (eds), "Paleobiology II", 195-197, Wiley-Blackwell, Oxford(UK).
    which is a rather concise technical version of his central argument in wonderful life - which means that it does give in text references and does reflect changes in systematics, which are somewaht of an issue when reading WL from a modern perspective (it was after all published just before the first solid molecular phylogenies for metazoans came around and as a result some of the discussion of the Burgess fauna is quite dated).

    ReplyDelete
    Replies
    1. Thanks.

      I was relying on the words of Jack Sepkoski's son who says ...

      So, in 1973, he hired a young graduate student, Jack Spekoski, to begin compiling an actual database of fossil marine invertebrate orders, families, and genera ...

      Delete
    2. Double-checking I found that Sepkoski was part of the earlier MBL meetings, but this had nothing to do with his thesis work (his advisor wasn't Gould, but Kummel). At the MBL somebody came up with the idea to produce a compendium like database (it's not clear who) and Sepkoski said that he'd like to do it, while Gould thought he could find a way to fund it. Because he had a thesis to complete and even before he was finished with it got hired as assistant professor it took until the late 70s to get a version he could work with and then another couple of years until he felt that the database itself was worth a publication.

      Delete
  3. Just a small point. Jack Sepkoski was not Gould's student (despite what it says on Wikipedia), but Bernie Kummel's.

    Also, Jack may have begun his work compiling the Fossil Record database with Raup in 1973, after leaving Harvard (before the completion of his PhD), but this was not the subject of his dissertation which was on the Early Cambrian small shelly fauna

    Christine (aka Mrs Sepkoski)

    I regret that Jack did not get to see my take down of Meyer.

    Steve Meyer's primary position
    Is to practice the lie of omission
    He omits the small shelly fauna
    Don't say that I didn't warn ya.

    ReplyDelete
    Replies
    1. He also omitted ichnofossils. Don't forget that.

      Delete
    2. Hello Christine,
      can you check my post above in terms of accuracy? And do you know who came up with the idea of the compendium at the MBL meetings?

      Delete
    3. Christine Marie Janis says,

      Just a small point. Jack Sepkoski was not Gould's student (despite what it says on Wikipedia), but Bernie Kummel's.

      Thanks for clearing that up. It's not a big deal but it's a shame that wikipedia is wrong and it's a shame that David Sepkoski wrote in 2005 ...

      Among those most centrally involved in this process were Gould, Thomas Schopf, David Raup, and Gould’s graduate student Jack Sepkoski, all of whom made major contributions in theoretical and quantitative analysis of the fossil record and evolutionary history.

      And it's too bad the error was propagated in the New York Times eulogy which says,

      ''No one had ever done it before,'' said Dr. Stephen Jay Gould, a paleontologist at Harvard University who was Dr. Sepkoski's graduate school adviser. ''He compiled a completely consistent data set on all groups, terrestrial, marine, single-celled, multicelled, animals and plants, everything.''

      Even Gould refers to him as his "former student" in Wonderful Life (p. 61).

      I found one reference saying that Jack Sepkoski completed his Ph.D. working with Kummel and Gould. I suspect that this is a (not uncommon) case of someone who was formerly a student of one professor but did a lot of research with another who may have been on the student's committee.

      Delete
    4. Jack was certainly friendly with Gould, but Bernie Kummel was his main adviser. I don't know if Gould was on his grad committee, -- possibly. Jack did start doing his data analysis while at Harvard (on some free computer time), but it wasn't part of his thesis. (I know this from talking to Jack later ---- although I did know him at Harvard I wasn't his spouse then.)

      I will say that Jack was amused by the 'replaying the tape' that Gould did to make him (Jack) his student in retrospect. (Before the NYT eulogy, obviously.)

      Delete
  4. Well, the whole thing is slightly more complicated. Steve did indeed ask Jack to begin compiling the database in 1973, after the first MBL meeting. The idea wasn't really discussed at the MBL meeting--Raup remembers that the major conclusion was that there was no way they were getting anything done with existing compilations.

    It was certainly not his thesis topic--that was traditional paleoecology/stratigraphy (with Kummel). The "database" evolved over the next decade (it technically wasn't a database at all until it was computerized by a colleague in 1982) from a mere compilation of existing data sources to an original resource.

    But his involvement with MBL had nothing to do with this. He was brought in on later papers to code the simulation program.

    - David Sepkoski

    ReplyDelete
    Replies
    1. Thanks. I've corrected this point in my post.

      Frankly I'm surprised that most of the comments so far have been on this relatively insignificant point. It's nice to get the history exactly right but I had no idea so many people cared.

      Delete
    2. (it technically wasn't a database at all until it was computerized by a colleague in 1982

      Some of us old fuddy-duddies actually remember a time when databases could be printed on paper or stored as index cards.

      Delete
    3. Frankly I'm surprised that most of the comments so far have been on this relatively insignificant point.

      I'm sure my love of Gould's writing in general (though I haven't taken on Structure) and Wonderful Life in particular affect my view, but I think this is one of your best posts. It's on a truly fascinating topic.

      The view that life's "tape" is primarily the result of contingency is something I've had since Gould asked the question at the end of Wonderful Life. Of course once contingencies occur and patterns are established, those established first are the most difficult, perhaps impossible, to change, and thus constrain the available directions of form thereafter. To use a perhaps overly simplistic analogy, it's impossible to start renovating a house by ripping out the foundation.

      What will be very interesting will be to see how low a level this argument between structuralism and contingency can be waged on, and perhaps even settled. Did the particular biomolecular changes that created morphological change occur primarily as a result of simple contingency, or to what extent did the rules of molecular biophysics constrain the potential direction(s) of mutation?

      Delete
    4. It seems possible that the constraints were laid down in bacteria, long before the Cambrian.

      Delete
    5. "Frankly I'm surprised that most of the comments so far have been on this relatively insignificant point."

      Well, you have his relatives paying attention!

      Delete
  5. Darwin in books insisted loudly that the millions of events of selection were behind the evolution of results in biology.
    So a rewind would never repeat what we now have.
    yet Darwin was wrong.
    Physics, I guess other non biological influences, constrains conclusions.
    You can't get a horse running on deer legs1 It must have a equation that allows swiftness and bulk management. Big deals.
    There are not many options. in fact i read about a running rhino, in old N america, that had its legs very much like a horse leg. In fact a horse leg is not a horse leg but a equation of leg horses just have too.
    likewise options for flight in birds and insects has its boundaries.
    In fact convergent evolution must constantly be employed to explain such like answers to like needs in biology.
    YEC/ID expect this and more as more study is done.
    You will not find the diversity one should if random evolution was going on and had so many options.
    There is a ceiling on options in biology.
    One will find this everywhere and do.
    The good criticism from creationists IS that biology does not show the option curve as it would be. The fixed rules of options CURVE is apparent and is evidence for structural constrain and implies these structures are from a common blueprint.
    yet its a option even evolutionism would have these boundaries.
    Physics etc etc is a real thing in biology and must be obeyed.
    Darwin missed this.

    ReplyDelete
    Replies
    1. "There are not many options. in fact i read about a running rhino, in old N america, that had its legs very much like a horse leg."

      Yup. Both horses and rhinos are perissodactyls (odd-toed ungulates). Not at all surprising that an Oligocene rhino (Hyracodon) would have legs rather like a contemporaneous horse as they both had 3 toes due to common ancestry.

      Let us know when you find a one-toed rhino.

      Delete
  6. I have nothing of particular value to add, but I would just like to say that I really enjoyed this informative and insightful post.

    ReplyDelete
  7. - mutations are "random" and the timing of mutations can greatly influence the path that a lineage follows

    - not all changes are due to adaptation and natural selection—many are just accidents


    The religious try to get away with claiming that humans are inevitable by addressing these as follows:

    They sometimes claim that on the macroscopic level God had it all planned and we were inevitable. Now, from what little I understand about the theory of inflationary cosmology, that would be a deeply problematic claim, but as I said, I understand very little and it has not been conclusively confirmed anyway, so let's leave that alone for a second.

    Of course having an explanation on the macroscopic level does not address the problem with mutations. So then the next deus ex machina is rolled out and now mutations are directly influenced by God, who acts on the "quantum" level, of course in invisible and undetectable ways.

    That the two scenarios do not readily fit together is swept under the rug.

    But there is a problem at a level in between these two. Mutations have to not only arise but get fixed. And that happens on the level of chromosomes, which are inconveniently bulky for God to influence them on the "quantum level" yet also too small for their segregation to be straightforwardly though of as affected by macroscopic processes in deterministic way. Which creates a kind of conundrum that I haven't seen addressed.

    ReplyDelete
    Replies
    1. That the two scenarios do not readily fit together is swept under the rug.

      But there is a problem at a level in between these two. Mutations have to not only arise but get fixed. And that happens on the level of chromosomes, which are inconveniently bulky for God to influence them on the "quantum level" yet also too small for their segregation to be straightforwardly though of as affected by macroscopic processes in deterministic way. Which creates a kind of conundrum that I haven't seen addressed.


      To address it would take the intersection of the set of Creationists and the set of those who truly understand quantum physics. I would guess, allowing for contingency and quantum tunneling, that at any given time this intersection would be the null set.

      Delete
  8. Normally, I now usually skip over Larry’s amateur stabs at philosophy and wearisome attacks on Intelligent Design and restrict myself to his elucidations of biochemistry in particular and science in general.

    This blogpost is the exception! The reference to Gould caught my eye!

    High School students (not to mention teachers) often fall into the trap of “structuralism” by failing to properly understand the nature of Genetics which for the most part is decidedly non-Mendelian!

    Consider the evolution of “adapted” melaninized Rock Pocket mice on dark lava flows:

    http://www.hhmi.org/biointeractive/making-fittest-natural-selection-and-adaptation

    Needed quick reminder - different mutations were responsible for the same phenotype in different Rock Pocket Mice populations when adapting to the new habitat.

    Second needed quick Reminder - by definition, we were unable to observe the most frequent outcome, namely the extinction of those populations that failed to adapt.

    The difficult point for high school students presented the over-simplified fallacy of “Bean Bag” Genetics would be that Genes do not operate in isolation from each other. As different mutations accumulate - the forward trajectories of evolution for different populations becomes quite different and quite unpredictable. Ultimately, there comes a point that replaying the "tape of life" will generate significantly different outcomes when played over larger timescales.

    This blog post is excellent. Thank you!

    Of course, this also brings up again Gould’s excellent treatise on NOMA, but let’s leave that sleeping dog lie!

    ReplyDelete
  9. Gould emphasizes his doubts that *intelligent* life would always appear if, in his metaphor, the tape of life were rerun. I had my doubts about that. Of course, my criteria for intelligence aren't nearly as high as Gould's were. (Possibly because my IQ isn't nearly as high Gould's was?) It's certainly true, however, that, say, ravens, chimpanzees, and octopuses are all more intelligent than their last common ancestors. So, intelligence is apparently a property that will get selected for under many conditions.

    The event that seems to me more likely to be a one-off fluke is the development of the eukaryotic cell. If the "tape of life" were rerun multiple times, how often could you expect to get that outcome? (And without it, how complex, let alone intelligent, would life ever get?)

    ReplyDelete
    Replies
    1. I think during any general arms race of predator vs prey there is always some sort of strong selection pressure for enhanced intelligence in predators. Frankly, I am amazed at the intelligence of Cnidarians.

      That said, I doubt here is some evolutionary niche for "higher intelligence" as advocated by Kenneth Miller.

      Your point of " one-off fluke is the development of the eukaryotic cell..." is well taken and eloquently reiterated in Nick Lane's book The Vital Question

      Delete
  10. Multicullular life is 90 percent based on genes invented by bacteria. It's possible that this foundation biases what is possible or likely.

    ReplyDelete
  11. I purchased Wonderful Life as a young and impressionable lad trying to maintain my currency it the field I had previously studied. You can imagine my disappointment when I get to the last couple chapters and find out that the publisher had messed up and duplicated a couple of the previous chapters.

    Did I miss anything important?

    ReplyDelete
  12. If you replay the tape of life there's a good chance that no intelligent life will evolve. There's no evidence that the actual history was pre-determined by any fundamental laws of form and there's certainly no evidence of purpose.

    This post seems more than a little relevant to the previous post on the Fermi Paradox. One of the failings of the paradox, speculative and qualitative as it must be, is that it tends toward the presumption that where there is life, intelligent and technologically advanced life will be a nearly inevitable or at least common outcome.

    The paradox may well say something about the frequency of intelligent life arising in the universe. But I would bet that the paradox as often conceived (why haven’t we been visited, why hasn’t earth been colonized, why haven’t we at least heard the distant muttering of intelligent aliens) is an exceptionally poor metric for speculation about the frequency of life itself arising in the universe (in so much that the rarity of intelligent life begets the conclusion that life itself must also be extremely rare).

    ReplyDelete
  13. What a fine post!
    There is a clarity in the summation of the argument, the historical tidbits are interesting making the total very instructive and fun to read.
    It has helped me clarify an objection I have.

    Before I get to the objection I want to say I understand the contingency argument better from this post than I have before and my objections aren’t about that aspect of the argument.

    My problem is with the conclusions about the tape of life based on the computer simulations. I’m probably missing something.

    I read this paper-
    Gould, S.J., Raup, D.M., Sepkoski, J.J., Schopf, T.J., and Simberloff, D.S. (1977) The shape of evolution: a comparison of real and random clades. Paleobiology, 3(01), 23-40

    I read part of Gould’s letter about the ’null hypothesis’ and ‘eight (or fewer) little piggies’…

    It seems he is making the argument that because the tape of life can be modeled as a stochastic process (in the computer), we can conclude it is inherently random and without direction.

    That conclusion does not follow. Many things that are not inherently random can be modeled as stochastic processes.
    Stochastic processes can have direction.

    I’m lost. Help!

    ReplyDelete
    Replies
    1. Question being asked: Are stochastic processes sufficient to produce the patterns we see in real biodiversity, or are additional processes needed? The simulations allow Gould et al. to conclude that the stochastic processes are sufficient to produce the patterns we see.

      Delete
  14. bwilson295-
    Yes, but a stochastic process can also describe your heart rate over time but we don’t conclude the heart operates at random without direction because of that.
    I’m not questioning a stochastic process can mimic what we see, I’m questioning the conclusions about the system based on that fact.

    ReplyDelete
  15. bwilson295-
    Thanks. I responded too quickly. It’s been a while since I studied the math and my memory finally arrived.

    Stochastic processes are generally related to two types of systems- random walk and dynamic equilibrium.
    Each life form exists in dynamic equilibrium, homeostasis.
    It seems each ecosystem exists in dynamic equilibrium as well.
    Why wouldn’t the ‘tale of the tape’ be one of dynamic equilibrium?

    I’m imagining this possibility has been discussed at some point, but I can’t find the reference.

    It’s an odd situation- a ‘random walk’ is generated by coin flips. ‘dynamic equilibrium’ exists when a system that 'wants' to be in a certain state is subjected to outside influence.

    The systems are quite different at the core, but we can model them the same way with the math.

    ReplyDelete
    Replies
    1. I think you have things backwards here. A stochastic process in general is a function X(t), where X(T) is a random variable for all T. That's very very general (and it's worth noting that all ordinary functions are a subset of this, in these cases all X(t) are dirac-distributed).

      The MBL paper uses what's basically a Galton-Watson branching process, where extinction and origination rates are identical it is more specifically a Glaton-Watson branching process in the critical case. Where rates of origination and branching were affected by standing biodiversity this defines a new process.
      The key thing here is assuming uniform rates of speciation and extinction does serve as a useful null, i.e. we can try to reject it by finding cases where there are significant deviations from uniformity.

      Delete
  16. Simon-
    I agree, you are correct about the math behind a stochastic process and the program they ran.
    I agree assuming uniform rates makes a good null.

    But that still leaves open the question about what sort of system we are observing. The two types of systems most associated with this type of process are those that are inherently random (coin flips) and those in dynamic equilibrium, like a heart rate.

    I note that Gould called the fossil record one of ‘punctuated equilibrium’ in 1972, before the computer simulations took place. I’m imagining a discussion occurred about the possibility the results from the computer simulations were indicative of a system in dynamic equilibrium as opposed to a system in random walk.
    I can’t find anything on that yet.

    ReplyDelete
    Replies
    1. The work by Gould et al. doesn't PROVE that evolution of biodiversity is a stochastic process. It does indicate that evolution can be a stochastic process and that therefore the burden of proof lies with someone who thinks it isn't. The person who thinks it isn't may be right, but has to provide good evidence.

      Delete
    2. bwilson295-
      The evolution of biodiversity can be modeled as a stochastic process.
      Since there are at least two types of systems that model that way I thought Gould and the others would have left a documented discussion of this.
      Apparently I am wrong about that.

      It’s probably moot-

      http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4413263/
      “Paleontologists have long argued about what the fossil record call tell us about how species evolve over long periods. Simple models such as stasis and random walks have been used to explore evolutionary patterns, but these have not always adequately captured the ways in which traits change over time in fossil species. Here we find that models with complex evolutionary dynamics are often favored, especially for fossil series that sample many populations, suggesting that the underlying evolutionary reality is likely more complex than represented by simplified—though still useful—models of trait change.”

      You might as well know- Your heart rate can be modeled as a stochastic process. But your heart isn’t a ‘random number generator’, it is a highly specialized machine attempting to keep the body in a proper state. It turns out that systems in dynamic equilibrium model as random walk quite nicely.
      This does not mean your heart is a supernatural entity created by a man in the sky with a magic wand. Your heart is a perfectly natural thing that exists in this universe and is made of atoms.

      My positing the biosphere exists in dynamic equilibrium and the evolution is a tale of dynamic equilibrium punctuated by contingency as opposed to random walk in no way implies a magic wand or that an historical figure loves you.
      It is an acknowledgement that ‘punctuated equilibrium’ would fit the observations and that contingency plays a large role in any case.

      From the article I linked it seems ‘it’s more complicated than that’ will win the day (as usual), so this discussion with regards the history of life is largely moot.

      Delete
  17. The only way the rewound tape would come out differently is if the earth/atmosphere/environment formed differently.

    That ultimately would mean that you would have to go all the way back to a change in the big bang to get a different replay(Asteroids IMO would only have a temporary effect).

    ReplyDelete
    Replies
    1. The only way the rewound tape would come out differently is if the earth/atmosphere/environment formed differently.

      That ultimately would mean that you would have to go all the way back to a change in the big bang to get a different replay


      Riiight, since in the 9 *billion* years between Big Bang and the formation of Earth, that little stuff that was going on, like creation of the elements, galaxy and star formation, black holes, quasars, etc., couldn't possibly have affected the formation of one single smallish planet around one average-size star among hundreds of billions in an outer spiral arm of one average-size spiral galaxy among a hundred billion or more - that was an absolute certainty.

      :rolleyes:

      Delete
    2. But we can get a pretty good idea what happens when the "tape of life" is replayed by looking at what happens after major extinctions. Invariably, the large, dominant species don't return. Instead, some modest "also ran" species (or clade) evolves into a large number of new species, including the newly dominant top predator species. There are some long-term trends. (E.g., larger brains and "warm" blood evolved more than once.) But the trilobites, T rex, and so on have never returned.

      So, we do know something factual about what happens when the "tape of life" is replayed. And it hasn't led to humankind on every, single replay-- just the latest one.

      Delete
    3. Steve-
      Modern physics allows us to make predictions about the future given the current state. But the predictions are probabilistic and so the tape could be quite different because the physics isn’t completely deterministic.
      Small changes at the beginning- due to the randomness of physics- could lead to large effects down the road.
      I’m thinking the evidence thus far indicates the chances of the tape replaying exactly are as close to zero as one can get and still be non zero. My opinion is based on the physics and observations about the current experiments.
      What is your opinion based on?

      Delete
    4. @Jack Jackson
      I’m thinking the evidence thus far indicates the chances of the tape replaying exactly are as close to zero as one can get and still be non zero

      There is no such thing as "as close to zero as one can get and still be non-zero". The probability of the tape replaying exactly is precisely 0, which renders it almost impossible. Here almost is used in its technical sense. To give a similar example, the probability of having n fair coins come up heads in a row is 2^-n, which goes to 0 as n goes to infinity. Yet "all heads" is clearly an allowed result for an infinite series of coin tosses. And the probability for any particular sequence of infinitely many coins is also 0. Yet, obviously the probability that the mth coin in such a sequence is heads is 0.5. The key feature of probability theory (indeed of measure theory) that allows this is that probability is sigma-additive, but not straight up additive. I.e. if Ai i€Iare mutually exclusive events, then p(union Ai)=sum p(Ai) if I is a countable set, but not if I is uncountably infinite.
      So, how much does this matter? It turns out that every random variable can be expressed as a function on an infinite coin toss model. And while for some types of RV this behavior goes away, it remains for a large class of RVs, including most of the physically relevant ones (exponential distributions, normal distributions, gamma distributions to name but 3).

      Delete
    5. Simon-
      You are assuming the sample space is infinite.
      I think you are mistaken in this case.

      Delete
    6. I even assume it's uncountable. And that's the certainly the case if time is continuous.

      Delete
    7. Your statement, ‘if time is continuous.’ implies you are considering a level of fidelity to the original time line that goes beyond what I was.
      There is a level of fidelity where the possibilities become uncountable and the probabilities become zero.
      But if we demand that level of fidelity have we really said anything interesting when we say the tape did not replay exactly?
      I’m not sure.

      If the sample space is uncountable, then the probability of any one outcome is zero.
      If the sample space is countable, then there is a smallest nonzero probability.

      I take it you are a frequentist.

      Delete
    8. But if we demand that level of fidelity have we really said anything interesting when we say the tape did not replay exactly?

      Well, no. The interesting questions there are really the core of theoretical morphology:
      Given the space of potential organisms, how large is the subset of viable organisms and how large is the subset of realized organisms. But when faced with ideas like Steves that started this thread, then I think it's worth pointing out that an exact replay is almost impossible to start with.

      To nitpick further:
      If the sample space is uncountable, then the probability of any one outcome is zero.

      Not necessarily. Not all probability distributions on uncountable sample spaces are absolutely continuous.

      I take it you are a frequentist.

      I try to avoid taking metaphysical positions. Both the strong law of large numbers and Bayes theorem follow from the Kolmogorov axioms. I find it amusing that both have given rise to interpretations that claim that they define what probabilities "really mean" in a mutually exclusive fashion and wish that people were at least a bit creative with these things. Where are the Chebychev inequalists? The Lindeberg-Fellerists?

      Delete
    9. Simon-
      You are correct about the distributions. You are asking for a precision in these descriptions that i have become too lazy to maintain. Thank you for that.

      I don’t know what probabilities ‘really mean’ either.

      Using the the Chebychev inequality to come up with a definition for probability is a brilliant idea and has advantages- one could show the frequentists and the baysians are wrong. Then we could spend years arguing if the Tchebysheff version is correct or if the Chebychev version is correct. Something constructive…

      Delete
  18. Hi Simon

    I wonder if it is a good time to resurrect one of Joe Felsenstein's observations?

    http://tinyurl.com/hhqtllz

    ReplyDelete
  19. Maybe it’s time to bring back this paper from an earlier thread-

    http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3509945/

    "In other words, although multiple evolutionary trajectories are often accessible, evolution is strongly constrained and the part of the fitness landscape available for exploration is highly variable but typically small. Thus, if we actually could replay the tape of evolution, the outcome could have been considerably more similar to the existing diversity of life forms than Gould expected."

    ReplyDelete
  20. I think this result would be part of an argument that the ‘tape of life’ would replay similarly to how it has played-

    http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1005926
    “The universal genetic code is redundant, with some amino acids being encoded by up to six different codons. Synonymous codons are not used randomly and typically one codon is used more frequently than others. This biased use of codons has been observed in all branches of life. Codon bias is the result of long-term selection and is presumed to confer an evolutionary advantage.”

    If the different codons yield different survival rates, then one would expect the ‘tape of life’ to favor those over the results of ‘drift’.

    Is this sensible?

    ReplyDelete
    Replies
    1. Not very much. Codon bias, for example, varies over taxa and genomic region, and thus also with time, so it's another factor open to contingency. And codon bias may be interesting when thinking about molecular evolution, but it has no effect on the morphological features that "replaying the tape of life" is all about.

      Delete
    2. "Based on these data we determined the average strength of selection that acts on codon choice in highly expressed genes and show that codon usage bias is selected to optimize the speed of protein production."

      Sounds like optimization inevitably leads to 'can't evolve no more'. All highly conserved, and nowhere to go.

      Delete
    3. "Sounds like optimization inevitably leads to 'can't evolve no more'. All highly conserved, and nowhere to go."

      But it's just overselling an idea. Nothing more.

      Delete
    4. "If the different codons yield different survival rates, then one would expect the ‘tape of life’ to favor those over the results of ‘drift’."

      There's a couple mistakes in the way you're thinking this. First, the codon bias is not a bias "determined" or "predetermined" by anything but something of a happenstance. After all, different organisms have different codon biases. Second problem, it's a tendency, not a definitive outcome. The bias might change with time, and therefore the direction of the tendency might change. It's a constantly "moving target."

      Delete
    5. Jack,

      "If the different codons yield different survival rates..."

      The less than optimal codons didn't affect survival rates. It was about essentially immeasurable 'fitness', and barely detectable 'selection'.

      "By systematically exchanging multiple codons at the same time we were able to increase the fitness effects of codon usage to a measurable level."

      Delete
    6. It's well known that selection coefficients small enough to be undetectable in the lab can still result in changes in natural populations.

      Delete
    7. John Harshman
      You make a differentiation I overlooked.

      How’s this—

      Which codons eventually win might be determined by which operate faster and that is a matter of chemistry.
      Which body types eventually win will be determined by which reproduce faster and that is effected by the environment and that is largely contingency.
      contingency and chemistry- the dance goes on.

      Delete
    8. I would point out the results these experimenters got are in line with the estimates made earlier based on the mutation, selection, drift models.

      Delete