pith. sign in

arxiv: 2606.03384 · v1 · pith:EGXY4NO6new · submitted 2026-06-02 · 🧬 q-bio.PE · math.ST· stat.TH

Evolution as a Process of Causal Inference

Jacopo Iacovacci This is my paper

Pith reviewed 2026-06-28 07:35 UTC · model grok-4.3

classification 🧬 q-bio.PE math.STstat.TH
keywords causal inferencenatural selectionquasispecies equationFisher's fundamental theoremmutationsreplicator dynamicspotential outcomesevolutionary dynamics
0
0 comments X

The pith

Evolution by natural selection is a process of causal inference where each mutation acts as a natural experiment screened by fitness effects.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper reframes evolution for haploid replicators in static environments as causal inference instead of Bayesian learning, because mutations break the pure selection analogy to Bayes' theorem. It maps the Neyman-Rubin potential-outcomes framework onto biology by treating each parent as the control unit and its mutant offspring as the treated unit, so that selection identifies the causal effect of the mutation on fitness. The central result uses the unnormalised quasispecies equation to prove that intergenerational change in mean fitness decomposes exactly into a selection term that recovers Fisher's Fundamental Theorem plus a mutation term equal to the fitness-weighted average of the cumulative causal effects of all mutations across parental genotypes. Under suitable assumptions this decomposition extends to the generalised replicator-mutator equation, and matched parent-offspring frequencies update in proportion to the average causal effect. A sympathetic reader would care because the mapping supplies a formal language for interpreting evolutionary change in terms of identifiable treatment effects rather than raw frequency shifts.

Core claim

Using the unnormalised quasispecies equation, the intergenerational change in mean fitness decomposes exactly into a selection term recovering Fisher's Fundamental Theorem plus a mutation term that corresponds to a fitness-weighted average of the cumulated effect of all mutations over all parental genotypes. The frequencies of populations of matched parents-offspring update in proportion to the average causal effect of mutations on fitness. This formalizes evolution as causal inference within the Neyman-Rubin potential-outcomes framework, where the core identification assumptions map onto evolutionary biology for haploid replicators in static environments.

What carries the argument

The exact decomposition of mean fitness change from the unnormalised quasispecies equation into a selection term and a mutation term given by the fitness-weighted average of causal effects of mutations.

If this is right

  • The decomposition of mean fitness change extends to the generalised replicator-mutator equation under suitable assumptions.
  • Matched parent-offspring frequencies update in proportion to the average causal effect of mutations on fitness.
  • Natural selection retains mutations whose causal effect on fitness is non-negative.
  • The mapping allows mutations to be treated as natural experiments whose effects are identified under the standard causal inference assumptions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Empirical sequences of parent-mutant pairs could be analyzed with causal-inference estimators to quantify average treatment effects of specific mutations.
  • The perspective implies that evolutionary trajectories in constant environments are shaped by the distribution of causal effects rather than by fitness differences alone.
  • Relaxing the static-environment assumption would require extending the framework to time-varying treatments to model causal inference during environmental change.

Load-bearing premise

The core causal identification assumptions hold exactly for populations of haploid replicators in static environments so that mutations can be treated as natural experiments with identifiable effects.

What would settle it

Direct measurement in a controlled haploid population with known mutation rates and fitness values showing that the observed change in mean fitness fails to equal the sum of the Fisher's theorem selection term and the predicted fitness-weighted average causal effect of the mutations.

read the original abstract

Recently, the mapping of the replicator equation onto Bayes' theorem has been recognised, leading to an analogy between evolutionary dynamics and Bayesian learning. However, this analogy holds only for pure selection in infinite populations and breaks down when mutations -- a central mechanism of evolution -- are introduced. Here I propose that evolution by natural selection, at least for populations of haploid replicators in static environments, is best understood not as a learning process but as a process of causal inference. Each mutation event constitutes a natural experiment in which the parent serves as the control and the mutant offspring as the treated unit. Natural selection screens the causal effect of the mutation on fitness, retaining mutations with non-negative effects. I formalise this view within the Neyman-Rubin potential-outcomes framework. I first develop the general theory using a generic fitness outcome and show how the core identification assumptions in causal inference (Stable Unit Treatment Value Assumption, Consistency, Unconfoundedness, Positivity) map onto evolutionary biology. Using the unnormalised quasispecies equation, I prove that the intergenerational change in mean fitness decomposes exactly into a selection term -- recovering Fisher's Fundamental Theorem -- plus a mutation term that corresponds to a fitness-weighted average of the cumulated effect of all mutations over all parental genotypes. I show that this decomposition extends, under suitable assumptions, to the generalised replicator-mutator equation and that the frequencies of populations of matched parents-offspring update in proportion to the average causal effect of mutations on fitness.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The paper claims that evolution by natural selection, for populations of haploid replicators in static environments, is best understood as a process of causal inference in the Neyman-Rubin potential-outcomes framework. Mutations are treated as natural experiments with parents as controls; the core identification assumptions (SUTVA, Consistency, Unconfoundedness, Positivity) are mapped onto evolutionary biology. Using the unnormalised quasispecies equation, the intergenerational change in mean fitness is shown to decompose exactly into a selection term recovering Fisher's Fundamental Theorem plus a mutation term that is a fitness-weighted average of the cumulated causal effects of all mutations over parental genotypes; the decomposition extends to the generalised replicator-mutator equation under suitable assumptions, with population frequencies updating in proportion to average causal effects.

Significance. If the algebraic decomposition holds exactly as stated, the work supplies a parameter-free bridge between replicator dynamics and causal inference, recovering a classic result (Fisher's theorem) as a special case while interpreting mutational change in terms of identifiable causal effects. The explicit mapping of identification assumptions and the absence of free parameters or ad-hoc axioms in the core derivation are strengths that could facilitate cross-application of causal tools to evolutionary questions.

minor comments (3)
  1. [Abstract] The abstract states that the decomposition extends 'under suitable assumptions' to the generalised replicator-mutator equation, but the specific assumptions required for that extension are not listed; adding a concise enumeration in the abstract would improve accessibility.
  2. [Introduction or Methods] The first appearance of the unnormalised quasispecies equation should include its explicit mathematical form and equation number to assist readers who may not recall the standard form.
  3. [Main decomposition section] Notation for the fitness-weighted average in the mutation term should be introduced with a clear definition immediately after the decomposition is stated, rather than relying on subsequent prose.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive assessment of the manuscript, including the recognition that the algebraic decomposition recovers Fisher's theorem as a special case and provides a parameter-free bridge to causal inference. The recommendation for minor revision is noted. No specific major comments were listed in the report.

Circularity Check

0 steps flagged

No significant circularity; derivation is algebraically self-contained

full rationale

The paper's central result is an explicit algebraic decomposition of intergenerational mean-fitness change under the unnormalised quasispecies equation into a selection term (recovering the known Fisher's Fundamental Theorem) plus a mutation term expressed as a fitness-weighted average of causal effects. This decomposition is derived directly from the model dynamics without fitting parameters to data or redefining quantities in terms of the target result. The mapping of Neyman-Rubin assumptions (SUTVA, consistency, unconfoundedness, positivity) onto haploid replicators in static environments is presented as an interpretive correspondence rather than a load-bearing premise that forces the algebra. No self-citations, ansatzes smuggled via prior work, or uniqueness theorems imported from the same authors appear in the derivation chain. The result therefore stands as an independent re-expression of the dynamics rather than a tautology.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

The paper relies on standard causal inference identification assumptions mapped to biology rather than introducing new free parameters or entities; the quasispecies equation is used as a starting point from prior literature.

axioms (3)
  • domain assumption Stable Unit Treatment Value Assumption holds for the population
    Mapped to no interference between individuals in the evolutionary setting.
  • domain assumption Unconfoundedness holds for mutation events
    No hidden factors confound the effect of mutation on fitness.
  • domain assumption Positivity and Consistency assumptions hold
    Every genotype has positive probability of mutation and the observed outcome matches the potential outcome under the mutation.

pith-pipeline@v0.9.1-grok · 5795 in / 1450 out tokens · 29968 ms · 2026-06-28T07:35:31.782987+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

38 extracted references

  1. [1]

    P., (1980)

    Bird, A. P., (1980). DNA methylation and the frequency of CpG in animal DNA.Nucleic Acids Res., 8, 1499–1504

    1980

  2. [2]

    V., & Payne, J

    Cano, A. V., & Payne, J. L. (2020). Mutation bias interacts with composition bias to influence adaptive evolution.PLoS Comput. Biol., 16, e1007523

    2020

  3. [3]

    et al., (2000)

    Capy, P. et al., (2000). Stress and transposable elements.Genetica, 107, 149–159

    2000

  4. [4]

    Cambridge

    Charlesworth, B., (1994).Evolution in Age-Structured Populations(2nd ed.). Cambridge

    1994

  5. [5]

    Coulondre, C., et al. (1978). Molecular basis of base substitution hotspots inE. coli.Nature, 274, 775–780

    1978

  6. [6]

    Towards a mechanistic foundation of evolutionary theory.eLife, 6, e23804

    Doebeli, M., et al., (2017). Towards a mechanistic foundation of evolutionary theory.eLife, 6, e23804

    2017

  7. [7]

    W., (1991) A constant rate of spontaneous mutation in DNA-based microbes.Proc

    Drake, J. W., (1991) A constant rate of spontaneous mutation in DNA-based microbes.Proc. Natl. Acad. Sci. USA., 88, 7160–7164

    1991

  8. [8]

    W., et al

    Drake, J. W., et al. (1998). Rates of spontaneous mutation.Genetics, 148, 1667–1686

    1998

  9. [9]

    Eigen, M., & Schuster, P. (1977). The hypercycle.Naturwissenschaften, 64, 541–565

    1977

  10. [10]

    jeopardize

    Fitzgerald, D. M., et al. (2017). What is mutation? A chapter in the series: How microbes “jeopardize” the modern synthesis.PLoS Genet., 13, e1007003

    2017

  11. [11]

    Foster, P. L. (2007). Stress-induced mutagenesis in bacteria.Crit. Rev. Biochem. Mol. Biol., 42, 373–397

    2007

  12. [12]

    Feschotte, C., & Pritham, E. J. (2007). DNA transposons and the evolution of eukaryotic genomes.Annual Review of Genetics, 41, 331–368

    2007

  13. [13]

    Gillespie, J. H. (1973). Natural selection with varying selection coefficients — a haploid model. Genet. Res., 21, 115–120

    1973

  14. [14]

    J., & Lewontin, R

    Gould, S. J., & Lewontin, R. C. (1979). The spandrels of San Marco.Proc. R. Soc. Lond. B, 205, 581–598

    1979

  15. [15]

    L., & Keightley, P

    Halligan, D. L., & Keightley, P. D. (2009). Spontaneous mutation accumulation studies in evolutionary genetics.Annu. Rev. Ecol. Evol. Syst., 40, 151–172

    2009

  16. [16]

    Hamilton, W. D. (1964). The genetical evolution of social behaviour.J. Theor. Biol., 7, 1–52

    1964

  17. [17]

    Hastings, P. J. et al. (2009). DNA repair: A link between cancer and aging.DNA Repair, 8, 1258–1267. Hernán, M. A., & Robins, J. M. (2006). Estimating causal effects from epidemiological data. Journal of Epidemiology and Community Health, 60(7), 578–586

    2009

  18. [18]

    Holland, P. W. (1986). Statistics and causal inference.JASA, 81, 945–960

    1986

  19. [19]

    G., & Halloran, M

    Hudgens, M. G., & Halloran, M. E. (2008). Toward causal inference with interference.JASA, 103, 832–842

    2008

  20. [20]

    Lanciano, S., & Mirouze, M. (2018). Transposable elements: all mobile, all different and emerging actors.Curr. Opin. Plant Biol., 42, 44–51. 22

    2018

  21. [21]

    Lenski, R. et al. (1991). Long-term experimental evolution inEscherichia coli. I. Adaptation and divergence during 2,000 generations.Am. Nat., 138, 1315–1341

    1991

  22. [22]

    C., & Cohen, D

    Lewontin, R. C., & Cohen, D. (1969). On population growth in a randomly varying environ- ment.Proc. Natl. Acad. Sci. USA, 66, 1056–1060

    1969

  23. [23]

    E., & Delbrück, M

    Luria, S. E., & Delbrück, M. (1943). Mutations of bacteria from virus sensitivity to virus resistance.Genetics, 28, 491–511

    1943

  24. [24]

    Lynch, M. (2007). The frailty of adaptive hypotheses.PNAS, 104, 8597–8604

    2007

  25. [25]

    Lynch, M. (2010). Rate, Molecular Spectrum, and Consequences of Human Mutation.Pro- ceedings of the National Academy of Sciences, 107, 961–968 Maynard Smith, J., & Price, G. R. (1973) The logic of animal conflictNature, 246, 15–18

    2010

  26. [26]

    Metz, J. A. J., et al. (1992). How should we define “fitness”?TREE, 7, 198–202

    1992

  27. [27]

    Otsuka, J. (2016). Causal foundations of evolutionary genetics.BJPS, 67, 247–269

    2016

  28. [28]

    Pearl, J. (2010). Causal inference.Causality: objectives and assessment, 18, 39–58

    2010

  29. [29]

    Radman, M. (1975). SOS repair hypothesis. InMolecular Mechanisms for Repair of DNA, pp. 355–367. Plenum

    1975

  30. [30]

    E., et al

    Rebollo, J. E., et al. (2012). Transposable elements: An ancient and diverse source of genetic innovation.Nature Reviews Genetics, 13, 491–502

    2012

  31. [31]

    Robins, J. M. (1986). A new approach to causal inference in mortality studies.Math. Model., 7, 1393–1512

    1986

  32. [32]

    Rosenbaum, P. R. (2002).Observational Studies(2nd ed.). Springer

    2002

  33. [33]

    Rubin, D. B. (1974). Estimating causal effects of treatments.J. Educ. Psychol., 66, 688–701

    1974

  34. [34]

    J., & Martienssen, R

    Slotkin, D. J., & Martienssen, R. (2007). Transposable elements and the epigenetic regulation of the genome.Nature Reviews Genetics, 8, 272–285

    2007

  35. [35]

    Stoltzfus, A., & McCandlish, D. M. (2017). Mutational biases influence parallel adaptation. Mol. Biol. Evol., 34, 2163–2172

    2017

  36. [36]

    Tenaillon, O., et al. (2016). Tempo and mode of genome evolution in a 50,000-generation experiment.Nature, 536, 165–170. Tchetgen Tchetgen, E. J., & VanderWeele, T. J. (2012). On causal inference in the presence of interference.Stat. Methods Med. Res., 21, 55–75

    2016

  37. [37]

    Wade, M. J. (1978) A critical review of the models of group selection.Quarterly Review of Biology, 53, 101–114

    1978

  38. [38]

    Zhang, F., et al. (2003). Segmental duplication and the evolution of the primate genome. Nature Reviews Genetics, 4, 761–773. 23

    2003