First self-replicating RNA molecule

Bill postulated the first emergence of life in the form of self-replicating molecules that were capable of self-repair, a fundamental form of resistance to environmental disturbances.

evolution_purpose1995.pdf (179.7 KB)

Among evolutionary biologists, one strand of talk of this nature refers to the ‘RNA world’ hypothesis. This article says that at the time of its publication (2012) the evidence for RNA being that first molecule was as yet weak, and that there might have been some precursor self-replicating molecule.

Robertson MP, Joyce GF. The origins of the RNA world. Cold Spring Harb Perspect Biol. 2012 May 1;4(5):a003608. doi: 10.1101/cshperspect.a003608. PMID: 20739415; PMCID: PMC3331698.

This recent article reports some success in the laboratory.
Papastavrou, Nikolaos; Horning, David P.; and Joyce, Gerald F. (2024). RNA-catalyzed evolution of catalytic RNA. PNAS 121.11:e2321592121. DOI: 10.1073/pnas.2321592121

A lay account is in this Washington Post article (my PDF image:
1st-replicator.pdf (1.4 MB)

Looking at this, the obvious became evident: self-repair can be a subsequently evolved ability. At this first stage, repair is a population phenomenon rather than an individual capability; that is, those that need repair to survive do not survive to replicate, and those that can survive and replicate without repair, do. In evolutionary terms, environmental disturbances are a major source of variants and therefore a major driver of evolutionary change.

MoL note: same principle in the individual.
Reorg_evolution2009.pdf (290.2 KB)

This is great stuff Bruce. But I’ve got to admit that I don’t understand a lot of it. One thing I’m having a problem with is this part of Bill’s 2009 paper that you posted:

Clearly, the analog of E. coli’s “swimming” has been observed in the genetic drift of organisms, and the phenomenon of reorganization as current visualized has been seen: “… a mutation that changes a DNA sequence and causes natural selection, also causes a new pattern of genetic drift among organisms that carry that mutation.” The “mutation” is, of course, a tumble, and the “new pattern of genetic drift” is a new direction of change in hyperspace

I don’t understand how a mutation that changes a DNA sequence can cause a new pattern of genetic drift that is analogous to a change in the direction of E. coli in hyperspace. I understand how E. coli “drift” in hyperspace works. I guess I don’t really understand what genetic drift is and how it relates to the effects of a single mutation in the DNA of a single individual. Can you help me out?

Best, Rick

For e. coli the direction in 3D space is an observer’s perception, we presume that is not something that the bacterium perceives. What it perceives is stuff in the fluid environment that it either wants more or less of. If a direction of swimming ‘works’ for its control of those variables, it persists; if not, it tumbles to a new direction.

In the evolution analog, the direction is a change in genotype. This is imperceptible to the parent or the offspring, but it may make a change in the (phenotypic) ability of the offspring to perceive and control stuff that might matter. If it ‘works’ in the sense of those offspring being able to control successfully, and consequently produce another generation of offspring under conditions such that they in turn come to reproductive maturity (that’s what ‘survival’ means), then that genotypic ‘direction’ persists in the hyperspace of ongoing generational genotype/phenotype changes. If it doesn’t work, the analog of a ‘tumble’ is some other genetic change that is more successful in its phenotypic realizations.

All analogies leak.

Gary Cziko’s books have some nice discussion of a neo-Darwinist view of learning.

What is missing from this (and also from Bill’s description of E. coli evolution in his From Reorganization to Evolution paper) is the mechanism that does what the evolutionary equivalent of the E. coli model does; that is, a mechanism that computes am “error” – a discrepancy between a perception of the result of a mutation (tumble) and what that perception should be. As in the E. coli model, the size of this error determines the duration of the delay until the next random mutation occurs. Without including this mechanism, the reorganization model of evolution seems little different than the Darwinian model of natural selection.

Bill and I wrote a paper entitled Random Walk Chemotaxis where we showed that the E. coli model of the learning process is much more efficient than a reinforcement model. This fact is also demonstrated in my Selection of Consequences demo. In the paper we noted that the E. coli model of learning could make a more efficient model of evolution and we pointed to some recent (at the time) evidence (the work of Cairns, Overbaugh & Miller, 1988) that a process like E. coli reorganization might be the basis of evolutionary change. Another nice piece of evidence is the apparently “punctuated” nature of the fossil record.

So I think we are in a position of having some evidence that evolution occurs by a process analogous to E. coli reorganization, we just don’t have a good idea of how it is implemented in actual populations of organisms. Any thoughts?

Thanks Rick and Bruce for interesting topic and references.

My quite layman opinion is that there really are (or must be?) differences between A) the evolution of species system and B) reorganization of an organism system. For example, the time scale is totally different and so A might be much slower and less efficient than B which is (assumedly) evolved from A. B can be thought and modelled as a simple control system as in Rick’s and Bill’s article, but in A there is no single control system (if we do not like to use a god hypothesis). Perhaps A could be a case of virtual control? In addition, there seems to be many different mechanisms in A. First there is the mutation mechanism which both the change rate and change direction seem (perhaps) mainly random. Then there are at least two different selection systems which cause some of consequences of the mutations to spread or not to spread in the population . One that is called natural selection simply makes some consequences of the mutations to die out before they reproduce themselves. The other is called sexual selection (is some species) which increases the probability of reproducing of some consequences of the mutations. This mechanism seems somewhat like control, and it creates something like directions of evolution, for example the development of the fan of the peacock. In addition, there are suggested still different mechanisms like e.g. in ​pdf-kuvakeKull Adaptive evolution without natural selection.pdf.

I’m not at home. Do you have a PDF of Bill’s From Reorganization to Evolution paper?

Maybe I’m missing something, but I agree with Cziko (Without Miracles; some meaty refs in this review) that learning and evolution have the same fundamental mechanism: variation and selective retention. That is, generation of variants and death (in evolution) or abandonment (in learning) of variants that do not result in successful control. In evolution, the variants are simultaneous (loosely speaking, i.e. generationally) in a population; in learning, the variants are successive. In evolution, those in the population which survive in the sense of producing offspring which/who reproduce determine what the species ‘learns’ generation to generation; in learning, generation of variants continues until control improves adequately. KISS.

Mechanism. What is needed to start and persist the generating of variants in evolution is biologically innate mechanisms for successful propagation plus a ‘drive’ (if you will) to exercise those mechanisms, and disturbances to the integrity of DNA resulting in mutations of DNA and disturbances of the DNA’s cellular and intercellular environment affecting its phenotypic expression.

What mechanism starts and persists the generating of variants in reorganization? I have guessed that what is needed is something in the intercellular environment of those control functions in an organism which are not controlling successfully. Given the observation that neurons isolated in vitro spontaneously brachiate and ‘seek’ to make connections with one another, it is possible that in a nervous system there is something in the intercellular environment when control is good, which wanes when control deteriorates. This is of course only speculation about a topic for research in neuroscience.

Here’s a pointer to the From Reorg to Evo paper!

And I found it here.

I’m working and will be flying cross-country back home in a couple of days but I’ll get back to this as I can.

A few years ago, I copied the following from B:CP.

This reorganizing system may prove to be no more than a convenient fiction; its functions and properties may some day prove to be aspects of the same systems that become organized.

Since this is the most generalized control system so far considered, it will also operate on the slowest time scale of all—a point to keep in mind as we consider how this system reacts to various events. To the reorganizing system, the disturbances associated with a single trial in an experiment may be as the blink of an eye—barely noticeable.

we will imagine a device that senses the set of quantities in question, and reports them in the form of one or several perceptual signals. Perception is a risky term here, however. Let us merely call such perceptual signals intrinsic signals, saying that they play the role of the reorganizing system’s inner representation of the organism’s intrinsic state. Postulating such signals is a convenient fiction, serving the same purpose as “temperature” serves in representing the kinetic state of molecules in our thinking.

To represent the fact that each intrinsic quantity has a genetically preferred state, we will provide the reorganizing system with intrinsic reference signals. These signals are also convenient fictions, representing the presence of stored information defining the state of the organism (as represented by intrinsic signals) that calls for no action on the part of the reorganizing system. Action is called for only when the intrinsic signals differ from the intrinsic reference signals. This stored reference-signal information may prove to be a message carried in our genes.

When there is a difference between sensed intrinsic state and the intrinsic reference signals, some device must convert this difference into action. As before, we insert a comparison function (a comparator) into the system, a device which emits an intrinsic error signal that drives the output of the system. The intrinsic error signal (perhaps multiple) will be zero only when intrinsic signals representing the state of the organism are all at their reference levels. Thus, the output of the system is driven by a condition of intrinsic error, ceasing only when intrinsic error falls to zero.

In the reorg and back paper, Bill says the ‘reorganization system’ concept “simply the old idea of “trial and error” reified, brought up to date, and described in terms suitable for modeling. The concept is irrefutable.”

In post #8 of this topic, I merely rearticulated the same ‘irrefutable’ concept of variation and selective retention. Comparing evolution and learning the domain of variation differs (generational populations vs. successive variations in the configuration of control systems. In evolution variation doesn’t stop when the currently most successful variant continues to be successful (the others just continue to be sporadic and less successful); I suspect the same is true in learning if only because of internal conflict (a common cause of ‘mistakes’) and because of environmental disturbances, both of which can disclose new possibilities for conscious consideration.

I have pursued how “its functions and properties may some day prove to be aspects of the same systems that become organized” by asking “what’s in it for the cell?” I touched on that in post 8 and won’t try to rehash those speculations here. That line of investigation suggests specific mechanisms, and accounts both for the restriction of reorganization to those systems which aren’t working well and for the spread of reorganization if the restricted scope of changes is inadequate.

In the 2009 ‘reorg-evolution-back’ paper, Bill alludes to the surmise that attention directs reorganization, which arose in connection with clinical experience with MoL. He sets this aside.

If awareness tended to seek out problem areas, we then had at least one way to keep reorganization focused where it was needed. But this introduced another bit of magic: awareness and its mobility. While those phenomena clearly exist, they are wild cards in any explanatory theory since we can’t explain them. We do not want any more wild cards in our explanatory theories that we absolutely have to have. Even when we have no alternative, they never stop nagging at the theoretician’s conscience.

The error is related to the post hoc ergo propter hoc fallacy and the truism that correlation does not mean causation.

He then proposes that reorganization begins with control systems in the somatic branch for intrinsic variables, and that when they are not able to regain control by reorganization, then reorganization spreads to just those systems in the behavioral branch which are affected by their loss of control.

In the example he gives, however, somatic sensations are inputs to a (behavioral) systems that construct a perception that we call ‘hunger’ and which control the acquisition and consumption of food. Though he does not say so, this is exactly parallel to somatic sensations (‘feelings’) as inputs to systems in the behavioral branch which construct perceptions that we call emotions (in Bill’s model of emotions). This is not reorganization, because those systems are already in place.

The abrupt transition at that point in the paper could do with an explicit reference to the prior discussion of the arm reorganization demo. Reorganization builds new systems from the bottom up, as in his 14-factorial demo of an arm-control system gaining control by pruning unneeded cross-connections for controllers of each of 14 joint angles specifications. The universe of systems subject to reorganization is constrained to controllers of just those behavioral variables which are not controlling well. How?

Attention is from the point of view of control by higher-level systems which observe behavioral outputs and their consequences, which are also active in deliberate learning and practice of skills. (Because they are slower, if they intrude they can disrupt the smooth operation of control systems which are normally well integrated: try analyzing your movements while riding a bicycle and observe yourself wavering and losing your balance.) When control which is normally ‘automatic’ falters, such higher systems ‘pay attention’. A stumble brings your attention from your conversation to the rough terrain you have entered. At higher levels, reorganization and planning can be complementary. Variations proposed by reorganization can be tested in imagination by those systems.

It’s a wonderful book and it would be great if Gary could join in this discussion since he’s the expert. But Ill just say that I think what you are missing is the difference in how selection is done in a Darwinian versus a Powersian model of evolution. In Darwinian evolution (as in reinforcement theory), selection is done by the environment; in Powersian (E. coli type) evolution (reorganization theory) selection is done by processes in the organism itself.

I think there are two methods for generating variants in any model of evolution: sexual reproduction (producing new genes via random combinations of alleles) and mutation (resulting in random changes in the genetic code). I think only selection of the latter type of variant can result in new species.

As I understand it, selection in both cases is a function of properties of the organism in relation to properties of the environment. In one case, control of a particular perception by particular means fails; in the other case, that failure to control prevents reproduction.

There’s a lot of new understandings of epigenetic inheritance. Bill was reading a book about this when I stayed with him during the last Colorado conference. I have a copy at home, but I don’t remember the title. Google will turn up more recent stuff. There’s mention in at least one of these articles on aging that I have read recently.

Two other researchers discussing the field of aging and longevity:

  1. Eric Topol (Scripps Research) interviews Colleen Murphy on the science of aging and longevity.
  2. Matt Reynolds (Wired) interviews Venki Ramakrishnan on the science of aging

That’s very interesting. Powers’ 1995 paper The origins of purpose: the first metasystem transitions fairly screams of epigenetics. Were you staying with Bill at about that time?

The epigenetics can be seen in Powers’ discussion of the effect of disturbances on the accuracy of gene replication. Powers assumes that replication is a control process that acts to prevent disturbances from affecting the accuracy of replication. These disturbances are presumably chemicals in the substrate of the cells (particularly the gametes, I presume) in which replication occurs.

If an organism is unable to control (behave) well in its environment, the (intrinsic) control error could presumably have an effect on the chemical composition of the substrate of the gamete, resulting in disturbances that cannot be resisted by the existing replication accuracy control systems. The result would be more replication errors, which means more mutations.

So this is what I think is Powers’ proposal regarding the epigenetic effect of behavior (control) on the gene: poor behavioral control results in an increase in mutation rate, which provides the raw material for the E. coli type process of evolution.

That does sound plausible, Rick, I agree. (Sorry for the long delay.) Here are some excerpts from a grad student’s summary of a weird case that has come up here previously: Cordyceps fungus modifying neurochemical control structures in ants of a particular species.

“Paraphrase: The parasite works to accomplish something. The host works to suppress it. The resulting behavior is the compromise position between the two. Often that behavior is exactly what the parasite really needed, it just had to work against the host to get it. … This [case] is a big deal because it’s definitely the presence of the parasite which makes this happen. This is behavioral alteration on a large scale, introducing a new behavior to an animal that doesn’t really learn new behaviors. Much more dramatic and visible than a parasite which gives its host stomach cramps and diarrhea, or perhaps makes the host take more risky behavior as evinced by questionnaires and statistics.”

I stayed with Bill in the summer of 2011, after that paper was written. Our conference and the biennial Linguistic Institute were both on the University of Colorado campus at Boulder that year.

In this article:

We drew the analogy between the genotype and the control system properties; and between the phenotype and the emergent behaviour in the environment during control (involving acting against disturbances).

It is the phenotype that is selected by the environment but the genotypic code that is transmitted across generations.

It is the control in the environment that is selected, but the control system properties (completely internal to the organism) that are transmitted to the next iteration of control.

So I still see the processes of evolution and reorganisation as analogous but clearly operate of different timescales. There is also the contentious issue of memetic evolution, which appears to allow the selection of information across individuals within a species, especially humans. I think this is also analogous to evolution and reorganisation but operates on a third level of organisation - the propositional/symbolic level.

I also agree with the potential of a PCT model of epigenetics you are both discussing which is about an interaction between the processes of control system reorganisation (or lack of its capacity in extreme cases to reduce intrinsic error) and genetic evolution.

Yes, Ted Cloak has been especially interested in ‘memes’ and memetic evolution. His study of the wheelwright’s craft is a lapidary piece.

I love Ted’s description of how spoked wheels are made. But I don’t see how Ted came to the conclusion that Darwinian evolution can account for the development of the spoked wheel. According to Ted, the evolution of the wheel occurred via the wheel-making apprenticeship process: “Wheelwrights who make worse wheels get to train fewer apprentices”. This would be a Darwinian process if the variations in the wheels produced by different wheelwrights were random. But I think they are intelligently produced; they are non-random. When weight or reliability or durability or smooth ride are required by the people who buy wheels, and the technology exists for addressing these requirements, then smart wheelwrights will develop wheels that are lighter, more reliable, more durable and give a smooth ride. And as soon as one smart wheelwright figures out how to do produce wheels the meet a requirement the other wheelwrights will imitate them (until someone invents patents).

So I see the evolution of the spoked wheel as an instance of intelligent design, the intelligence coming from from that species that Shakespeare (whose birthday is today), via Hamlet, described as being “In apprehension how like a god”.

Yes, very interesting description, although the strange special terms made it difficult and somewhat superficial for me to follow the process.

I agree that there are (and must be) differences between natural evolution of species (described by Darwin) and cultural evolution of such innovations like wheels. I see three nested levels: first is the natural evolution where biological species are developed mainly through mutations and natural selection, the second is learning / reorganization in animals - and even conscious “intelligent planning” in the most developed animals, and the third is the social cultural evolution of memes and innovations.

What I think is crucial in Ted’s theorization is that all these processes are incremental and that chance has big (although perhaps somewhat different) role in them and finally such new features are selected to remain in existence which help living and control.

DNA and RNA provide an obvious analogy for cultural inheritance and for social stability and change, but it is only an analogy.

Analogies and metaphors are useful. They propose ways that we might control perceptions in a novel or unfamiliar domain, constraining the scope of trial and error. We also fit inputs to the model, imagination fills in gaps and we shade, neglect, or deny misfits that imagination can’t override.

Psychologists went to steam-engine hydrodynamics for metaphorical drives, pressures, releases, etc. and more recently to digital computing and information processing, because they don’t know of anything better. Anthropologists went to linguistics seeking an -emic vs. -etic account of cultural ‘sememes’ bringing the notion of memes into public common usage, and social sciences generally look to DNA metaphors, again because they know of nothing better than analogy and metaphor.

We have something better in both cases. We’re pretty well familiar with what PCT offers individual psychology. For the social sciences we also have something better. It’s called collective control, the theoretical consequences of including other people and their artifacts in the environment for PCT models. We have only begun modeling the simplest cases.

The principles for extending the scope of the environmental part of what is to be modeled are simple enough. It’s still a matter of individuals controlling their idiosyncratic perceptions, yes, but those perceptions include perceptions of the actions and attributed CVs and reference values of others (their ‘motivations’), perceptions of principles by which they expect others to be guided, perceptions of aspects of the environment that facilitate or impede control, etc. And too long neglected in the basement of PCT is the mammalian brain which perceives ‘attractants’ and ‘repellants’ and has a strong influence in the settings of reference levels and of gain for control in the larger system of which it is part. Some individuals cynically exploit all such perceptions as means of controlling how others affect their CVs. Most of us are more or less alert to perceptions of others’ motivations, perhaps even as among the environmental ‘artifacts’ that facilitate or impede control.

A hall of mirrors is simple, reflections in it are not. But that’s only an analogy.