Pith. sign in

REVIEW 3 major objections 4 minor 44 references

How Downwards Causation Occurs in Digital Computers

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A running digital computer is a working case of downward causation: the program and its data, not the underlying physics alone, decide which electrons flow through which transistors.

desk verdict A philosophically serious and well-informed argument for downward causation in digital computers, with a technical error in the central mechanism of §4.3 that is correctable and does not sink the broader open-systems point. read the letter →

arxiv 1908.10186 v2 pith:IV4W3GPE submitted 2019-08-15 cs.OH quant-ph

George Ellis , Barbara Drossel This is my paper · ORCID
classification cs.OHquant-ph
keywords EmergenceDownwardcausationDigitalcomputersLogicalcontrolTransistorsSupervenienceCausalcompleteness
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that a digital computer in operation is an example of downward causation: the abstract algorithm, the program, and the data determine which electrons flow through which transistors at which moments. The authors' counterfactual test is simple—change the program and the electron flows change, with the hardware unchanged. They locate the mechanism in the transistor gate: a time-dependent voltage $V(t)$ set by the machine code enters the electron Hamiltonian as a time-dependent potential, so the usual uniqueness theorems for fixed Hamiltonians do not apply. They conclude that the bottom-level physics of a computer is not causally complete by itself, and that synchronic supervenience (one microstate, one macrostate at a given time) can hold while diachronic supervenience (earlier microstate fixes later outcomes) generally fails.

What carries the argument

The load-bearing mechanism is the time-dependent gate voltage $V(t)$ applied to a MOSFET transistor. It is a higher-level variable—determined by bit patterns in machine code—that enters the microscopic electron Hamiltonian as the potential term $H_V(t) = \sum_i e V(\mathbf{r}_i, t)$. The paper's pivotal claim is that this time-dependent term changes the Hamiltonian itself, so the usual existence-and-uniqueness theorems that make a microstate's future unique do not apply; the electron dynamics is thus open to constraint from above. A second piece of machinery is the disconnect between levels: because conduction-band electrons, phonons, and depletion zones cannot be obtained from Hamiltonian (2) without approximations that smuggle in classical and statistical assumptions, the higher-level description has its own logic and is not a coarse graining of the lower one.

What would settle it

A complete derivation starting from the microscopic electron–ion equations that produces a transistor's threshold and current–voltage behaviour without adding lattice, band, phonon, or classical-electrostatics assumptions would refute the claimed disconnect. Also, a demonstration that the exact microstate of a running computer at an earlier time logically determines the later program state and outputs (including interactive inputs) would refute the paper's rejection of diachronic supervenience.

Watch

Extended reading notes

Core claim

The central claim is that the logical hierarchy of a computer—application program, compiled code, machine instructions—reaches into the physical implementation hierarchy through the time-dependent gate voltage of each transistor. A field-effect transistor is an ON/OFF switch that turns on when $V(t)$ exceeds a threshold, and this $V(t)$ is itself the machine-code representation of the running algorithm. Because $V(t)$ appears in the electron Hamiltonian as $H_V(t)$, the Hamiltonian changes in time in a way set by higher-level logic; the authors therefore state that standard existence-and-uniqueness results for a fixed Hamiltonian no longer determine the outcome from the initial microstate. The argument is supplemented by a disconnect claim: the phenomenological solid-state models used to describe transistors (band structure, phonons, depletion zones, resistance) cannot be derived from the microscopic electron–ion Hamiltonian without auxiliary approximations and classical elements. If correct, this means higher-level variables are not eliminable descriptions but occupy real causal roles, and physics per se is not causally complete.

Load-bearing premise

The case rests on two premises that must both hold: transistor behaviour cannot be derived from the microscopic physics of electrons and ions without extra classical and statistical assumptions, and counterfactual dependence (change the program, change the electrons) is enough to earn the word cause; if either fails, the downward-causation conclusion loses its footing.

Editorial extensions

If this is right

  • If the central claim is right, the causal completeness of physics at the bottom level fails in an everyday engineered system: a complete microphysical description of a working computer is not enough to tell you which electrons will move; you also need the program.
  • Diachronic supervenience is false in general for computers: the microstate at an earlier time does not fix the high-level state at a later time, because programs, data, and interactive inputs enter along the way.
  • Abstract entities—algorithms, programs, and data—have causal powers in the paper's counterfactual sense: changing only the abstract algorithm changes physical outcomes at the transistor level.
  • The same pattern, applied to biology in the paper's companion argument, implies a general scheme for top-down causation: higher-level needs act down by altering time-dependent constraints on lower-level dynamics, although the concrete mechanism differs between transistors and biomolecules.
  • All levels in the logical and implementation hierarchies are equally real and causally effective; compilers and interpreters are the explicit machinery that transmits logic downward.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper does not test this, but the mechanism suggests a general criterion: a higher-level variable is causally effective when varying it alone changes the lower-level trajectory on a fixed physical substrate. Programmable circuits or reconfigurable logic could be used to test this criterion directly.
  • The paper does not address it, but if one rejects the counterfactual theory of causation, the same facts might be redescribed as control or constraint rather than cause; the force of the conclusion shifts with that philosophical choice.
  • The non-derivability claim is empirical in principle: a future multiscale derivation that reproduces switching behaviour from the electron–ion Hamiltonian without hand-added approximations would narrow the disconnect and weaken the case, even if it did not touch the time-dependent-voltage mechanism.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper argues that digital computers provide a concrete case of downward causation: high-level abstract programs and data control electron flows in transistors, so the microphysical dynamics is not causally complete on its own. The proposed mechanism is that time-dependent gate voltages V(t) add a term H_V(t) to the electron Hamiltonian (Eq. 7), altering band structure and depletion zones, and the authors claim that this invalidates existence and uniqueness theorems, so outcomes are not fixed by initial data. The paper then uses this to deny diachronic supervenience and to argue for the causal power of abstract entities. The argument is presented through a detailed hierarchy of logical and implementation levels, with examples such as Bubble Sort and interactive programs.

Significance. If the argument were correct, it would provide a rigorous mechanistic example of strong emergence and downward causation in a well-understood physical system, potentially relevant to debates on causal completeness, supervenience, and the mind-body problem. The paper usefully organizes well-known facts about computer architecture and transistor operation, and it also acknowledges open-system effects (heat baths, cosmic rays) that complicate simple Hamiltonian descriptions. However, the central technical mechanism as stated is false, and the paper's broader philosophical conclusions depend on that mechanism. The significance is therefore conditional on correcting the technical claim and clarifying the sense in which V(t) is not determined by lower-level physics.

major comments (3)
  1. [§4.3, 'The crucial dynamic'] The claim that a time-dependent bias voltage V(t) added to the Hamiltonian as H_V(t) 'so usual existence and uniqueness theorems do not apply: outcomes are not determined by initial data' is technically incorrect. For a Hamiltonian H0 + H_V(t) with V(t) a bounded, piecewise-continuous, self-adjoint potential, the time-dependent Schrödinger equation is well-posed: Kato's theorem guarantees a unique unitary propagator, and the state at any later time is uniquely determined by the initial state and the function V(t). The presence of a time-dependent potential does not introduce non-uniqueness. What the argument needs is the separate claim that V(t) itself is not determined by the microscopic initial state of the system alone (e.g., because it is imposed by external program logic). That claim may be defensible, but it is not what the paper states, and the current formulation directly undermines the diachronic-supervenience argument in §6.
  2. [§4.2 and §6.6] The paper's reliance on the non-derivability of transistor properties from the microscopic Hamiltonian (2) is asserted largely through the extended Leggett quotation rather than through a worked argument. Since the central conclusion that 'outcomes are not determined by initial data' depends on this alleged disconnect, the manuscript would be strengthened by an explicit statement of what would count as a successful derivation and a careful defense of why the approximations (Born-Oppenheimer, band structure, Boltzmann equation, etc.) are not merely practical but in-principle barriers. As written, the appeal to authority leaves the load-bearing premise under-supported.
  3. [§1 and §7.1] The paper explicitly adopts a counterfactual theory of causation (reference [11]) and then in §7.1 uses counterfactual dependence to conclude that abstract entities 'have causal powers.' This is a legitimate philosophical stance, but the paper does not engage with alternatives or with the objection that counterfactual dependence may be a symptom rather than a cause. Since the conclusion is framed as a result about the world ('abstract entities have causal powers'), the argument would be more persuasive if it distinguished the counterfactual-dependence reading from a stronger productive-causation reading, and indicated which one is intended in the main claims.
minor comments (4)
  1. [References] Reference [11] has a typo: 'Menzies, P) Menzies, P' should be 'Menzies, P.'
  2. [§4.3] The phrase 'the electron dynamics is no longer unitary' is imprecise: the reduced dynamics of an open system is non-unitary, but the full system, including the heat bath, still evolves unitarily. Clarifying this would avoid a potential misunderstanding.
  3. [§6.7] The causal-exclusion argument (quoted from the Stanford Encyclopedia) is dismissed by saying that the instantiation of M by P is itself explained by downward causation. This is a circular response in the present context, because the existence of downward causation is precisely what is at issue. The manuscript should acknowledge this circularity concern explicitly.
  4. [General presentation] The manuscript contains some rough edges, such as the header 'Version: 2019/10/18/Foundations of Physics' at the end, and a few awkward typographical artifacts (e.g., 'M I' appears as 'MI'). These should be cleaned up before final submission.

Circularity Check

1 steps flagged · score 4.0 of 10

Crucial uniqueness premise is imported from the authors' own prior paper via footnote 3 of §4.3.

  1. uniqueness imported from authors [§4.3, 'The crucial dynamic' paragraph and footnote 3]
    "The crucial dynamic: When a time dependent bias voltage V(t) is applied to a transistor gate, it alters the underlying Hamiltonian through the time-dependent potential term H_V(t), see (7), so usual existence and uniqueness theorems do not apply: outcomes are not determined by initial data. [Footnote 3:] See the Appendix of [14]."

    This is the paper's designated mechanism for downward causation: the lower-level electron dynamics is said to be non-unique because the gate voltage adds a time-dependent potential, and hence 'outcomes are not determined by initial data.' The only support offered for this non-uniqueness is a footnote to [14], a paper by the same authors (Ellis and Kopel 2019). The present paper gives no derivation or external mathematical citation for the claim. Since the conclusion that physics is not causally complete depends on this premise, the crucial step reduces to a self-citation rather than to an independently established result.

full rationale

No fitted-input prediction loop or self-definitional derivation is present: the paper does not calibrate any parameter to data it then 'predicts,' and the central claim is not obtained by renaming a known result. The transistor hierarchy, compiler logic, Bubblesort example, and open-system examples are independent, non-circular content. However, the 'crucial dynamic' asserting that a time-dependent gate potential breaks existence/uniqueness of the microscopic dynamics is load-bearing, and the only citation is to the authors' own prior work ([14], Ellis & Kopel 2019). As presented, that premise is not independently established in the paper; the argument's key step is thus supported by a self-citation. A second self-citation, [31] (Drossel & Ellis 2018), supplies the arrow-of-time mechanism but is peripheral to the computer-specific argument. Overall score reflects one load-bearing self-citation while acknowledging substantial independent content elsewhere.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No parameters were fitted. The paper's argument is conceptual; its main premises are non-derivability of condensed-matter models, counterfactual causation, and the correctness of the time-dependent Hamiltonian claim (which is flawed). It also invokes the authors' contextual wavefunction collapse for irreversibility.

assumptions (4)
  • domain assumption The phenomenological models of solid-state physics (Born-Oppenheimer, Boltzmann equation, phonons) are not in principle derivable from the microscopic Hamiltonian.
    Stated in §4.2 via the Leggett quote; essential to establish a real discontinuity between levels.
  • ad hoc to paper Contextual wavefunction collapse resolves the arrow of time via heat baths.
    Invoked in §4.3 citing [31]; non-standard, from the authors' own prior work, but not central to the main argument.
  • domain assumption Counterfactual dependence is an adequate account of causation.
    Stated in §1 and §3: 'we rely on a counterfactual view of causation [11]'. The whole argument uses counterfactual tests.
  • ad hoc to paper Existence and uniqueness theorems for time-dependent Hamiltonians do not apply when H(t) is time-dependent.
    Stated in §4.3; this is technically incorrect because the time-dependent Schrödinger equation has unique solutions for given H(t). The point is that H(t) itself is not specified by initial data, but the theorems still apply.

how reviews work

0 comments
Cite this review

Pith. "Pith review of How Downwards Causation Occurs in Digital Computers." pith.science (2026). https://pith.science/paper/IV4W3GPE

@misc{pith2026190810186,
  author       = {Pith},
  title        = {Pith review of: How Downwards Causation Occurs in Digital Computers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IV4W3GPE}},
  note         = {Machine review of arXiv:1908.10186}
}
read the original abstract

Digital computers carry out algorithms coded in high level programs. These abstract entities determine what happens at the physical level: they control whether electrons flow through specific transistors at specific times or not, entailing downward causation in both the logical and implementation hierarchies. This paper explores how this is possible in the light of the alleged causal completeness of physics at the bottom level, and highlights the mechanism that enables strong emergence (the manifest causal effectiveness of application programs) to occur. Although synchronic emergence of higher levels from lower levels is manifestly true, diachronic emergence is generically not the case; indeed we give specific examples where it cannot occur because of the causal effectiveness of higher level variables.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 44 canonical work pages

  1. [11]

    Counterfactual Theories of Causation

    Menzies, P) Menzies, P (2017) "Counterfactual Theories of Causation", The Stanford Encyclopedia of Philosophy , Ed. E N Zalta (ed.), URL <https://plato.stanford.edu/archives/win2017/entries/causation-counterfactual/>

    work page 2017

  2. [1]

    More is different

    Anderson, P. W. (1972) “More is different”Science 177:393-396

    work page 1972

  3. [2]

    Tanenbaum, A S (2006)Structured Computer Organisation(Prentice Hall, Englewood Cliffs), 5th Edition

    work page 2006

  4. [3]

    On the nature of research in condensed-state physics

    Leggett, A. J. (1992). “On the nature of research in condensed-state physics”Foundations of Physics 22:221-233

    work page 1992

  5. [4]

    Hohwy J and Kallestrup J (Eds) (2008)Being Reduced(Oxford University Press)

    work page 2008

  6. [5]

    Humphreys, P (2016)Emergence: A Philosophical Account(Oxford University Press)

    work page 2016

  7. [6]

    F., and Lancaster, T

    Gibb, S., Hendry, R. F., and Lancaster, T. (Eds.) (2019).The Routledge Handbook of Emergence. Routledge

    work page 2019

  8. [7]

    Downward causation: an integrating theme within and across the sciences?

    Ellis G F R, Noble D, and O’Connor T (2011) “Downward causation: an integrating theme within and across the sciences?”Interface Focus2 23 November 2011. 19 How Downwards Causation Occurs in Digital Computers 27

    work page 2011

Show all 44 references
  1. [8]

    Ellis G (2016)How can Physics Underlie the Mind: Downward Causation in the Human Context (Springer: Berlin/Heidelberg, Germany)

    2016

  2. [9]

    A theory of biological relativity: no privileged level of causation

    Noble, D (2011) “A theory of biological relativity: no privileged level of causation”Interface focus 2: 55-64

    2011

  3. [10]

    (2011).Nine algorithms that changed the future: The ingenious ideas that drive today’s computers

    MacCormick, J. (2011).Nine algorithms that changed the future: The ingenious ideas that drive today’s computers. Princeton University Press

    2011

  4. [12]

    Simon, H. A. (1996).The sciences of the artificial(MIT Press)

    1996

  5. [13]

    Mellisinos, A. C. (1990).Principles of Modern Technology(Cambridge: Cambridge University Press)

    1990

  6. [14]

    The Dynamical Emergence of Biology From Physics: Branching Causation via Biomolecules

    Ellis, G and Kopel J (2019) “The Dynamical Emergence of Biology From Physics: Branching Causation via Biomolecules”Front. Physiol., 25 January 20

    2019

  7. [15]

    The constraint interpretation of physical emergence

    Blachowicz, J. (2013). “The constraint interpretation of physical emergence.”Journal for general philosophy of science44:21-40

    2013

  8. [16]

    Pearson Education India

    Booch G (2006)Object oriented analysis and design with application. Pearson Education India

    2006

  9. [17]

    Top-down causation by information control: from a philosophical problem to a scientific research programme

    Auletta, G., Ellis, G. F., & Jaeger, L. (2008). “Top-down causation by information control: from a philosophical problem to a scientific research programme”.Journal of the Royal Society Interface5:1159-1172

    2008

  10. [18]

    Multiple realisation: Keeping it real

    Anthony, L. M. (2008) “Multiple realisation: Keeping it real”. InBeing Reduced, Ed J Hohwy and J Kallestrup (Oxford University Press), 164-175

    2008

  11. [19]

    Multiple Realizability

    Bickle, J (2019) “Multiple Realizability”,The Stanford Encyclopedia of Philosophy, E N. Zalta (ed.), URL = https://plato.stanford.edu/archives/spr2019/entries/multiple-realizability/

    2019

  12. [20]

    (2014).The Java virtual machine specification

    Lindholm, T., Yellin, F., Bracha, G., and Buckley, A. (2014).The Java virtual machine specification. (Pearson Education)

    2014

  13. [21]

    Lafore, R (2002)Data Structures and Algorithms in Java(SAMS: Indiana)

    2002

  14. [22]

    Ross Ashby, W (1952).Design for a Brain(Chapman and Hall)

    1952

  15. [23]

    V., Lam, M S, Sethi, R., & Ullman, J

    Aho, A. V., Lam, M S, Sethi, R., & Ullman, J. D. (2006).Compilers, principles, techniques, and Tools. Addison Wesley

    2006

  16. [24]

    Knuth, D. E. (1973).The art of computer programming, Vol. 1: Fundamental algorithms

    1973

  17. [25]

    Simon, S. H. (2013).The Oxford Solid State Basics(Oxford University Press)

    2013

  18. [26]

    (2012).Advanced solid state physics(Cambridge University Press)

    Phillips, P. (2012).Advanced solid state physics(Cambridge University Press)

    2012

  19. [27]

    (2007).Quantum mechanics(Springer Verlag Berlin Heidelberg)

    Schwabl, F. (2007).Quantum mechanics(Springer Verlag Berlin Heidelberg)

    2007

  20. [28]

    (2009).fundamentals of the physics of solids Volume II: Electronic Properties (Springer Verlag Berlin Heidelberg)

    Solyom, J. (2009).fundamentals of the physics of solids Volume II: Electronic Properties (Springer Verlag Berlin Heidelberg)

    2009

  21. [29]

    Emergence in exact natural science

    Primas, H. (1998). “Emergence in exact natural science”Acta Polytech. Scand.91, 83–98

    1998

  22. [30]

    (2014).Reductionism, emergence, and levels of reality (Springer Verlag Berlin Heidelberg)

    Chibbaro, S., Rondoni, L., Vulpiani, A. (2014).Reductionism, emergence, and levels of reality (Springer Verlag Berlin Heidelberg)

    2014

  23. [31]

    Contextual wavefunction collapse: An integrated theory of quantum measurement

    Drossel, B., and Ellis, G. (2018) “Contextual wavefunction collapse: An integrated theory of quantum measurement”New Journal of Physics20:113025

    2018

  24. [32]

    Abelson H and Sussman J S (1990)Structure and Interpretation of Computer Programs (MIT Press)

    1990

  25. [33]

    A tutorial on the free-energy framework for modelling perception and learning

    R Bogacz “A tutorial on the free-energy framework for modelling perception and learning” Journal of Mathematical Psychology76 (2017) 198-211

    2017

  26. [34]

    (1996).An Introduction to Genetic Algorithms

    Mitchell, M. (1996).An Introduction to Genetic Algorithms. Cambridge, MA: MIT Press

    1996

  27. [35]

    Ghirardi, G. (2007). Sneaking a Look at God’s Cards: Unraveling the Mysteries of Quantum Mechanics(Princeton University Press)

    2007

  28. [36]

    Field testing for cosmic ray soft errors in semiconductor memories

    O’Gorman, T. J.et al (1996). “Field testing for cosmic ray soft errors in semiconductor memories”.IBM Journal of Research and Development40:41-50

    1996

  29. [37]

    Mental Causation

    Robb, D and Heil, J, "Mental Causation", The Stanford Encyclopedia of Philosophy (Summer 2019 Edition), E N. Zalta (ed.), URL = <https://plato.stanford.edu/archives/sum2019/entries/mental-causation/>

    2019

  30. [38]

    Physics, complexity and causality

    Ellis, G. F. R (2005) “Physics, complexity and causality”.Nature 435: 743

    2005

  31. [39]

    Watson, J. D. (2004).Molecular biology of the gene.(Pearson Education India). 28 George Ellis, Barbara Drossel

    2004

  32. [40]

    (2014)Cell Signalling Biology.(London: Portland Press)

    Berridge, M. (2014)Cell Signalling Biology.(London: Portland Press)

    2014

  33. [41]

    Development of multiscale models for complex chemical systems: from H+ H2 to biomolecules

    Karplus, M. (2014). “Development of multiscale models for complex chemical systems: from H+ H2 to biomolecules”Angewandte Chemie International Edition53:9992-10005

    2014

  34. [42]

    Physics, complexity and causality

    Ellis, G F R (2005) “Physics, complexity and causality”Nature 435:743

    2005

  35. [43]

    (2019).Coders: Who they are, what they think, and how they are changing the world.(Picador)

    Thompson, C. (2019).Coders: Who they are, what they think, and how they are changing the world.(Picador)

    2019

  36. [44]

    (2019).ZUCKED: Waking Up to the Facebook Catastrophe

    Bissell, T. (2019).ZUCKED: Waking Up to the Facebook Catastrophe. (Penguin Random House) ————————————————————————————————— ——— Version: 2019/10/18/Foundations of Physics

    2019

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.