George Ellis — a cosmologist who co-authored the Hawking-Ellis The Large Scale Structure of Space-Time and worked closely with Stephen Hawking — argues that reductionism is “patently false” and that causation operates at every level of a complex hierarchy, not just at the bottom. Physics enables outcomes but does not determine them; context, constraints, and higher-level structures do. This episode covers his views on top-down causation, the evolving block universe, the non-existence of a wave function of the universe, moral realism, and why he parts ways with Roger Penrose on consciousness and cosmology.
Causation Is Real and Multi-Leveled
Causation is demonstrated by intervention: make a change and observe a reliable outcome. This applies from Galileo dropping balls to smoking causing lung cancer.
Causation does not require an agent to exist — it is happening all the time — but an agent is needed to demonstrate it experimentally.
Ellis identifies multiple distinct types of causation:
Physical causation — governed by physical laws
Purposeful causation — present in all living organisms and technology
Symbolic causation — the basis of language and society
Abstract causation — the basis of computing, where abstract symbols drive physical outcomes
Social causation — decisions by people causing events (e.g., the war in Ukraine)
Historical causation — events from centuries ago causing present outcomes (e.g., the Troubles in Northern Ireland)
Imaginative causation — imagining something that does not yet exist and then bringing it into being (e.g., Steve Jobs and the iPhone)
Top-Down Causation: Physics Is the Servant, Not the Master
In a computer, a high-level algorithm (e.g., Python code) ultimately determines what electrons do in transistors. The electrons enable the computation but do not determine it. The algorithm tells the physics what to do.
Ellis’s core claim: physics enables but does not decide outcomes. The context — the structure, constraints, and goals — decides what happens.
A thermostat is a clear example: setting the dial (macro-level) causes molecules to move faster (micro-level). The physics does not determine the temperature; the user does.
Reductionists like Sean Carroll claim everything reduces to microphysics. Ellis responds that Maxwell’s equations or Newton’s laws by themselves do nothing — they only do something in a context, and the context is not itself derivable from those equations.
Top-down causation operates through:
Setting constraints — e.g., the length of a pendulum determines its swing; a battery and switch in a circuit determine whether electrons flow
Creating, modifying, or destroying lower-level elements — e.g., gene regulatory networks create proteins; developmental biology turns pluripotent cells into specific cell types; apoptosis (programmed cell death) separates fingers during hand development
These same principles apply in organizations: Boeing hires (creates membership), trains (modifies), and fires (destroys membership) employees — all top-down control of lower-level components.
Modular Hierarchical Structure: The Universal Architecture of Complexity
Every sufficiently complex system — biological, technological, organizational — is a modular hierarchical structure. This is not a metaphor; it is a design principle discovered by computer scientists and engineers who actually build complex systems.
You decompose a complex task into simpler modules until each is simple enough to implement linearly, then build back up.
Causation takes place at every emergent level, and every level is necessary for the whole to function.
In the brain: thought and emotion occur at the level of neural circuits, enabled by action potentials, enabled by ion channels, enabled by electromagnetic interactions between protons and electrons — every level is needed.
Three kinds of emergence operate in biology:
Functional emergence — second-by-second operation
Developmental emergence — growth from a single cell over ~21 years
Evolutionary emergence — over millions of years
All three are integrated: evolution creates developmental systems, which create functional organisms, which enable further evolution.
Universal Biological Principles
Life on other planets will also be based on carbon — nothing else can do the job — and will use DNA for information storage.
Any living organism anywhere will require:
Information processing (to survive, it must assess its environment)
Decision-making (agency — evaluating possibilities and choosing actions)
Metabolism (energy processing)
Feedback control (homeostasis)
Adaptive selection (constant adaptation to environment)
Ellis defines life not by any single criterion but as the integrated bundle of all these features working together.
Biological Relativity and the Work with Dennis Noble
Ellis collaborated with Dennis Noble, a physiologist who discovered the heart’s pacemaker mechanism and who also started as a reductionist before realizing it was untenable.
Noble’s work shows that in the heart, pacemaker cells control pumping — causation operates at the cellular and systems level, not just the molecular.
Ellis and Noble developed biological relativity: the principle that there is no privileged level of causation. Every level in a hierarchy has its own causal laws, and all are necessary.
Ellis’s contribution was the philosophical and structural framework from general relativity and cosmology; Noble’s was the experimental and physiological evidence.
Ellis challenges Richard Dawkins’s claim that genes are replicators: DNA in a Petri dish with all necessary chemicals does nothing. The cell is the replicator — it replicates DNA, organelles, metabolic systems, and structures not controlled by DNA. The cell interior is a place of molecular chaos (billions of collisions per second), and molecular machines like kinesin extract energy from this chaos.
Where Ellis Agrees and Disagrees with Roger Penrose
Ellis admires Penrose’s transformative work on black holes and gravitational topology, which laid the foundation for the Hawking-Ellis book.
Agrees with Penrose: Inflation does not solve the smoothness problem — it assumes smoothness before it starts.
Disagrees on consciousness: Penrose claims it arises from quantum effects in microtubules. Ellis says this is simply the wrong scale — consciousness arises from circuits at the level of the whole brain.
Disagrees on conformal cyclic cosmology (CCC): Penrose proposes successive “eons” of the universe connected through conformal infinity. Ellis’s objection: infinity is not a very large number — it is larger than any number that can exist. If the boundary between eons is truly infinite, any signal from one eon is diluted by an infinite amount, so precisely zero information transfers. The transition between eons is therefore impossible.
Ellis wrote a paper titled “The Physics of Infinity” on this: physicists routinely treat infinity as a shorthand for “very large,” which leads to errors. The universe will never be infinitely old — no matter how long you wait, you have not taken even the first step toward infinity.
The Evolving Block Universe
The standard argument for the block universe (from special relativity) claims that because different observers disagree on what constitutes “space” vs. “time,” all of space-time must exist equally — past, present, and future.
Ellis rejects this for multiple reasons:
Special relativity does not describe the universe; general relativity does.
Every actual cosmological model (e.g., Robertson-Walker) has preferred time sections (surfaces of constant cosmic time). The real universe is not context-free.
The ADM formulation of general relativity writes gravity as space-time evolving in time — space itself grows, starting from the Big Bang.
This establishes a global direction of time: from the fixed beginning toward the moving present. Local arrows of time (thermodynamic, electromagnetic, gravitational wave) all derive from this global direction.
Ellis calls his model the evolving block universe: the past is fixed and real, the future is open and does not yet exist, and the present is the leading edge of becoming.
The Future Is Open and Laplace’s Demon Fails
Ellis argues the future is not determined by initial conditions. His iPhone argument:
Given perfect knowledge of the last scattering surface (or any early-universe data), you could not predict the iPhone.
There is no data on the last scattering surface that encodes the iPhone.
If you insist the iPhone was entailed by initial conditions, you must assume some non-physical intelligent agent wrote that information in — a “nerd’s form of intelligent design.”
The correct explanation: the universe allows evolution, which produces humans with intelligence and free will, who design things like iPhones.
Laplace’s demon fails because we are open systems. New, unpredictable information is always coming in (e.g., a car accident changes everything in a driver’s brain from that moment). Evolution has made our brains predictive processing systems precisely because we are open systems.
You cannot deduce Darwin’s theory of evolution from physics alone — physics has no concept of life, death, or evolution. These are emergent, biological concepts.
No Wave Function of the Universe
Ellis agrees with Carlo Rovelli: there is no wave function of the universe.
The Schrödinger and Dirac equations are linear. The real universe is nonlinear. A single global wave function cannot produce nonlinear physics.
The solution: there are local wave functions everywhere, covering space-time the way an atlas of coordinate charts covers a manifold in general relativity. No single chart covers the whole space-time; no single wave function covers the whole universe.
This undercuts the quantum multiverse (Everettian) program: if there is no wave function of the universe, there is no basis for saying the universe “branches.”
Ellis supports objective wave function collapse (with collaborator Barbara Drossel): collapse is real but always context-dependent — the context determines when and how collapse occurs.
The Multiverse Is Metaphysics, Not Physics
Ellis is prepared to accept that chaotic inflation might produce different expanding universe domains, but he cannot test this.
He is a conservative scientist: he believes science is about what can be observed or tested. Other universes are beyond the visual horizon — you cannot see them, test them, or falsify claims about them.
Talking about the multiverse is metaphysics, not physics.
This also undermines fine-tuning and anthropic arguments: they assume a probability distribution over constants, but we have only one universe — one data point. You cannot infer a distribution from a single example.
Dualism and Consciousness
Ellis is a dualist: the mind is enabled by the brain but is not reducible to it. We have no idea how action potential spike chains produce consciousness (the hard problem).
Consciousness enables symbolism, which gives humans an enormous evolutionary advantage — the ability to plan, cooperate, and build technology is why humans dominate the Earth.
He distinguishes between functional causation (moment-to-moment), developmental (learning to fly, growing up), and evolutionary (why flight evolved) — all three scales always operate in concert.
Michael Levin and Bioelectricity
Ellis highlights Michael Levin’s work on bioelectricity as revolutionary and potentially Nobel Prize-winning.
Levin has shown that bioelectric signals are at least as important as genetics in developmental biology — a level of control absent from standard textbooks.
More radically, Levin’s experiments suggest developmental processes appear to have goals — they seem to “know” what the outcome should be (e.g., salamanders growing second heads when bioelectric patterns are altered). This challenges Darwinian orthodoxy, which insists development has no foresight or plan.
Moral Realism as Data About the Universe
Ellis is a moral realist: some things are evil as a matter of fact, not opinion. Threatening to wipe out an entire civilization is evil — full stop.
He distinguishes morality (what is actually good or evil) from ethics (what a particular society believes is good or evil, which varies by time and place).
He argues that everyday life — love, home, consciousness, free will — is data about the universe just as much as telescope observations are. Steven Weinberg’s claim that the universe looks “pointless” ignores all the evidence from ordinary experience.
Ellis’s moral realism is informed by his decades fighting apartheid. He believes the objective reality of good and evil is a deep fact about the nature of the cosmos, even if it cannot be proven.
