The Origin of Life

Stochastic corrector

Eörs Szathmáry's "Stochastic corrector" model is seen by many as making the error catastrophe problem faced by early organisms less pressing.

Here I discuss the relevance of this model to the origin of life.

Motivation for the model

Szathmáry's model deals with a population of distinct self-replicating entities in dividing cells.

It deals with the problem of how information could be transmitted to offspring - in the form of the ratio between the constituent replicators - despite the fact that these replicators are not physically joined together - thus:

  • Stochastic forces present when the cell divides distributing them unevently in offspring;
  • Differences in the rate of reproduction of each sort of replicator within the cell;

Explanation of the model

The model is best explained with a diagram:

Stochaistic Corrector

Organisms are composed of a small number of different types of self-replicating agents: R1 and R2.

The organisms work best if there is some specified ratio between the replicators: in the diagram the preferred ratio is 1:1.

The cells grow and eventually split into two - with the replicators being distributed randomly between the offspring.

Then selection acts - destroying any cells that deviate too far from the optimal ratio of replicators within each cell.

The result is that the information content represented by the replicators - and their proportions in the cell - is preserved between generations.

The model can still work - even if some of the replicators reproduce faster than other ones - and thus there is between-replicator selection within each cell.

Good points

The model is useful - since it shows how a small collection of small molecules - none of which may be capable of independent replication could co-exist in a community - and help catalyse each others replication, and be inherited without being physically connected together or having their division orchestrated by some sort of controller.

Not so good points

However the model doesn't scale up very well - the more replicators are involved the greater the chance that stochastic variations will result in one type of replicator being omitted from any children - and the stronger selection is needed to maintain verbatim transmission of information between the generations.

Rewriting the model

I think the model makes a lot more sense if it is rephrased a bit:

Rather than consider the diagram as representing different sorts of replicator in a cell, consider it as representing different sorts of replicator in an ecosystem:

For example, if you have a "pool" filled with replicators, then the species in it may colonise another pool downstream.

If a key species gets wiped out - or fails to get transmitted to the new environment - then the new ecosystem will not flourish.

The idea that the proportion of replicators in each "cell" is somehow significant is abandoned in this model. That information is no longer strongly inherited.

However, the collective genomes of all the important species still get transmitted to the next generation - i.e. information about the existence of the different species is preserved.

This phrasing retains a key feature of the model - namely the possibilty of inheriting a lot more information than is present in any individual replicator.

It also still leaves open the possibility of a community of mutually-dependent organisms surviving in an environment where none of them could exist alone.

Overcoming differential reproductive rates

What about the possibilty that one sort of replicator will wipe out the other ones?

Szathmáry's original model invoked a kind of group selection to explain how the system could overcome this problem - suggesting the selection between the ecosystems would act to penalise ecosystems where one replicator does too well.

Selection between ecosystems is likely to have been an important factor.

Some other points may also be worth mentioning, though:

  • Independent niches
    • If the different replicators do not compete significantly for resources then they may be able to co- exist peacefully - without one wiping the other out.
      Niches might be partially independent in an ecosystem - the replicators in the system:
      • ...may have different nutrient needs;
      • ...may exist in different areas of the ecosystem.
  • Frequency-dependent selection
    • It's in all the replicators interests to prevent the population of any one replicator from dropping too low - since the loss of an essential player is terminal for everyone involved.
      This fact tends to favour cycles involving frequency-dependent selection - where the least-numerous critical replicator is preferentially synthesized.
  • Mutual dependency
    • It's possible that one replicator may depend on the products of another one to successfully reproduce.
      Such dependencies may produce frequency-dependent selection that favours the scarce replicators.

The membrane-free corrector

This reformulation no longer depends in any way on the notion of a membrane or cell - instead, the role of container could equally well have been played by the environment - which could have been as simple as a rock pool.

Membranous material is unlikely to be involved in the earliest living systems - since the organic material that composes most membranes tends to form sticky messes - that are incompatible with the process of crystallisation that is likely to be responsible for the replication of the genomes of the first organisms.


The idea presented here owes an obvious debt to the model presented in my earlier Increasing Complexity essay - i.e. it is basically the same idea wrapped up in different terminology.


  • Eörs Szathmáry and J. Maynard Smith - The Major Transitions in Evolution. Oxford, 1995.

  • Eörs Szathmáry and J. Maynard Smith - The Origins of Life, Oxford University Press, 1999;

  • Eörs Szathmáry and László Demeter - Group selection of early replicators and the origin of life. Journal of Theoretical Biology 128, 463-486, 1987; |