Free will survives physics, three scientists argue for different reasons

The Short Version

Three working scientists, in three separate conversations with Lex Fridman, end up defending free will, and none of them does it by bending the laws of physics. Theoretical physicist Sean Carroll says free will is emergent but real, "just as real in tables and chairs," and that calling it an illusion misuses the word [1]. Astrobiologist and theoretical physicist Sara Walker says free will comes from being a deterministic structure in a random environment, with enough history bundled into you to select futures, and that it works over time rather than in an instant [2]. Astrophysicist Adam Frank says "free will is a given," something experience hands us that science has to account for, not explain away [3]. Their reasons differ, and so do their deeper pictures of reality, but together they reject the popular claim that physics has settled the question against us.

Table of contents

What They Said

Sean Carroll: free will is real in the way a table is real

Carroll comes at free will as a committed physicalist. He says he sees no obstacle in principle to machines that think like humans, and holds that only the natural world exists [1]. So he won't defend free will by adding any extra ingredient to physics. On consciousness, the closest neighbor to free will in his conversation, he says: "I do not claim to have any special insight on how consciousness works other than I see no reason to change the laws of physics to account for it" [1].

What he does is block the leap from made of physics to "not real. Lex asked whether consciousness is an illusion, a kind of shorthand for a feeling, and Carroll objected to the word itself: I almost never use the word illusion. Illusion means that there's something that you think you're perceiving that is actually not there [1]. His example is a desert mirage: Like an oasis in the desert is an illusion. It has no causal efficacy. If you walk up to where the oasis is supposed to be, you say you are wrong about it being there" [1].

An emergent thing is a different kind of thing: "That's different than something being emergent or non-fundamental, but also real. This table is real. Even though I know it's made of atoms, that doesn't remove the realness from the table" [1]. Then comes the line that anchors this whole insight: "I think the consciousness and free will and things like that are just as real in tables and chairs" [1].

Lex pushed back. Doesn't the oasis still have some causal effect, on your thirst, say, or by inspiring people to act? Carroll's answer gives his real test for when something counts as real: "your understanding of the world in a way that gives you power over it and influence over it is decreased rather than increased by believing in that oasis. That is not true about consciousness or this table" [1]. So for Carroll, a concept is legitimate if using it makes you better at understanding and acting in the world. Talking about choices, decisions and agency passes that test. Talking about a mirage fails it.

This fits his broader philosophy, which he calls "poetic naturalism." In it, only the natural world exists, but there are many valid ways of talking about it, from quantum fields to tables to judgments of beauty and morality [1]. He says the "more obvious respectable duty" of the idea "is that tables are real. Even though you know that it's really a quantum field theory wave function, right, tables are still real. There are different way of talking about the underlying deeper reality of it" [1]. Free will lives at one of these higher levels of description. It isn't written into the equations for quantum fields, and it doesn't need to be.

Carroll applies the same move to subjective words. Just as "the hardness or the brownness of this table" are "words that we attach to certain underlying configurations of ordinary physical matter," so "sadness and redness or whatever are words that we attach to you to describe what you're doing" [1]. On normative talk (beauty, morality) he goes further: "They might not be universal, they might be subjective, but they're not arbitrary, and they do have a role in describing how the world works" [1]. He's careful about where this stops. Science, he says, can't tell you what's right or wrong: you choose your goals and science helps achieve them, even horrible ones [1]. The choosing agent has a real place in his picture.

Carroll also draws a line that shows how he uses "illusion. He is critical of panpsychism, the view championed by Philip Goff that consciousness pervades all matter, and he rejects the part of Donald Hoffman's view that calls spacetime an illusion. He does, though, consider belief in God to have been a very reasonable illusion that later understanding replaced [1]. So Carroll isn't simply generous with the word real." In his usage, God was an illusion, while tables, consciousness and free will are not. The difference is whether deeper understanding dissolves the concept or explains it.

Sara Walker: free will comes from being a structure that is huge in time

Walker comes at the question as a physicist of life. Her central claim is that "life is the process of how information structures matter over time and space," and that any individual is just an instance of lineages structured across time [2]. She argues that living things are among the largest structures in the universe, but large in time rather than space: "The universe is far larger in time than it is in space," and "the modern technosphere is the largest object in time in the universe that we know about" [2]. Things only look emergent, she tweeted, because we can't see time. If we could, we'd see each object's entire causal history trailing behind it [2].

Her account of free will grows directly out of this. She is not a many-worlds person: in her view, all those quantum branches are bundled into the deterministic, macroscopic object that is you, carved out of a random quantum background [2]. Free will, she says, comes from being a deterministic structure in a random environment, with enough history bundled in you to select futures. It operates over time rather than instantaneously, like deciding to be in another state tomorrow [2].

Walker's picture has three parts. First, the base of reality is random. She argues that physics' reliance on initial conditions plus fixed laws, inherited from Newton, is a good approximation for systems studied so far but will fail at more basic levels, and that the universe is random at its base [2]. Second, living things are deterministic structures carved out of that randomness. Her description of the origin of life is a structure that reinforces its own existence against the randomness of chemistry and quantum mechanics, after which everything is holding on for dear life to the causal structure that put it there [2]. Third, what gives such a structure freedom is history. Intelligence, she says, is the ability of objects with a lot of selection embedded in them to select possible futures [2]. The more history an object carries, the more futures it can steer toward.

This turns the usual worry upside down. The common argument says that if we are deterministic machines, we can't be free. For Walker, being a deterministic structure is part of what makes freedom possible, because a structure with no stability couldn't select anything. Randomness alone doesn't give freedom either, since in an unconstrained random system nothing supports anything else [2]. Freedom sits in the combination: a stable, history-rich object inside an open, random world.

The time dimension also changes what free will is. Walker says it operates over time rather than instantaneously [2]. Her example is deciding to be in another state tomorrow, which is a choice spread over a long stretch of time, not a snap decision at one instant. The search for free will inside a single moment of neural firing is, on her view, looking in the wrong place, just as defining life by a thin slice of the present misses most of what a living thing is [2].

Her views on consciousness run in parallel. She resists assuming consciousness is fundamental or that it came before life. Her tentative view is that we have conscious experience because we are temporally extended objects with a huge inner space, and that it is private because no one else occupies the same thread of time [2]. Like Carroll, she is an avowed physicalist who holds that nothing exists outside the universe [2]. Also like Carroll, she doesn't think being physical makes something less real. She treats even mathematics and the multiverse as things that exist here, in human minds and the biosphere [2].

One more piece of her thinking matters here: she rejects the idea that reality has a single foundation from which everything else can be derived. She doesn't believe there is a theory of everything, quoting David Krakauer: "a theory of everything is a theory of everything except those things that theorize" [2]. What we call fundamental particles, she argues, are really the boundaries of what our technology can resolve, and she proposes that the truly fundamental things may be constructed objects whose histories and construction we can actually understand by taking them apart [2]. If that's right, the claim that physics at the bottom rules out free will is aimed at a foundation Walker doesn't think exists.

Adam Frank: free will is a given, and the equations are the map

Frank's route runs through philosophy of science. His book The Blind Spot, written with Marcelo Gleiser and Evan Thompson, argues that science has quietly erased lived experience from its picture of reality, producing paradoxes in consciousness, time and quantum mechanics [3]. When Lex asked whether free will could be shown to be real by integrating experience back into physics, Frank refused the framing: "What I would say is that free will is a given" [3].

He grounds this in Alfred North Whitehead: "it's not the job of either science or philosophy to account for the concrete. It's the job to account for the abstract. The concrete, what's happening between us right now, is just given" [3]. You're allowed to look for an explanation, he says, "but the explanation actually doesn't add anything to it" [3]. On his reading, free will is part of the concrete: "free will in some sense is the nature of being an agent," and "agency and autonomy are sort of the two things that are, you know, they're equivalent" [3]. "To be an agent is to be autonomous," so the real scientific question becomes whether "you have an account for agency and autonomy that captures aspects of its arising in the world? Or the way it and the world sort of co-erise?" [3].

He also diagnoses why the free will argument keeps coming up: "the reason why we argue about free will often is because we already have this blind spot view that the world is deterministic because of our equations, which themselves, we treat the equations as if they're more real than experience" [3]. The equations, he says, "are a thinner, you know, representation. As we like to say, don't confuse the map for the terrain" [3]. And: "What's happening between us right now, and this, you know, all the weirdness of it? That's the terrain. The map is what I can write down on equations and then in the workshop do experiments on super powerful, needs an account, but experience overflows that" [3].

His "parable of temperature" illustrates how the blind spot forms. Science began with the felt experience of hot and cold, pulled out shared structural invariants (philosopher Michel Bitbol's term) through thermometers, and built thermodynamics. Then it forgot its roots and started treating 60 degrees as more real than feeling cold, an "amnesia of experience" [3]. The free will debate, for Frank, is the same forgetting. We build deterministic equations out of experience, then use those equations to deny the experience of choosing.

Frank traces the deterministic picture to its history: "there was a reason why people adopted this third person, God's eye, deterministic view. This view of sort of like, yeah, the perfect clockwork of the universe, yeah, totally made sense. But by the time you got to the beginning of the 20th century, science itself was telling you like, eh eh" [3]. The strongest signal came from quantum mechanics and its measurement problem [3]. He says the Born rule says measurements matter. Many-worlds invents infinite unobservable universes to avoid this, while QBism (quantum Bayesianism), developed by Chris Fuchs and colleagues, puts the agent and information at the center [3].

He names the bundle of assumptions behind the blind spot: reductionism, the God's-eye objective frame, physicalism, and "the reification of mathematics, that mathematics is somehow more real than this" [3]. Together, he says, "what they all do is they end up pushing experience out and saying experience is an epiphenomena" [3]. He deliberately avoids the word consciousness in favor of experience, "because it's a verb kind of an way or it's verb, it's verb like" [3]. Elsewhere he puts it this way: experience "is not a thing you have, it is a thing you do" [3].

For Frank this goes beyond academic philosophy. It creates a crisis of meaning: "science wants to tell us that we're insignificant, we're not important, we're just biological machines," while cosmology and quantum mechanics keep putting observers at the center [3]. The responses he lists, scientific triumphalism, science rejection "on the left and the right," and pseudoscience, have in his view been "detrimental to society as a whole" [3]. Lex added that the arrogance of triumphalism fuels the rejection [3]. So the claim that physics disproved free will is, for Frank, a small, clear example of a larger cultural mistake.

Frank also thinks agency is the frontier of physics. He believes the biggest question in physics is what life is, and specifically how an autonomous agent arises [3]. Life's defining feature is that it innovates. Given a star's mass and composition, physics can predict its whole life, but no one could predict from a single cell that billions of years later there would be a kangaroo that can punch you in the face [3].

Where They Agree, and Where They Don't

Shared ground

All three agree on the headline point: physics as it stands does not rule out free will. Each of them treats the you are just particles, so choice is fake argument as a mistake, though each diagnoses the mistake differently.

For Carroll, the error is a confusion of levels. Something can be non-fundamental and still real, and an illusion is something with "no causal efficacy," which free will doesn't lack [1]. For Walker, the error is looking for freedom in the wrong place, in an instant and at the bottom of physics, instead of across time in a structure that carries history [2]. For Frank, the error is treating the equations "as if they're more real than experience" [3].

They also agree on some supporting points:

• No one invokes anything supernatural. Carroll sees "no reason to change the laws of physics" [1]. Walker is an avowed physicalist who holds nothing exists outside the universe [2]. Frank rejects physicalism, but he also rejects idealism, since both assume an outside, third-person view that is a useful fiction [3]. None of them appeals to a soul.
• Levels and scales matter. Carroll's poetic naturalism says there are many valid ways to talk about one world [1]. Walker says life is high-dimensional, you can be alive in some dimensions and not others, and sees life at every scale, from cells to the planet [2]. Frank endorses David Krakauer's work showing that a liver cell is an individual that is also part of a liver, a person, a language community, ecosystems and the biosphere [3]. All three expect agency to show up at the level of organized wholes, not single particles.
• Life and agency are the live scientific question. Carroll has shifted his own research toward how complexity emerges, including minds that imagine futures [1]. Walker says the frontier of new physics is the problems of life and intelligence [2]. Frank calls what life is, and how agency arises, the biggest question in physics [3]. Their overlap here goes beyond this one conversation: Frank co-wrote on planetary intelligence with Walker and David Grinspoon, and he quotes her line that "Life doesn't happen on a planet. It happens to a planet" [3].

Real disagreements

The agreement is about the conclusion, not the foundations. On the foundations they differ sharply.

Physicalism. Carroll and Walker are both self-described physicalists [1][2]. Frank lists physicalism as one of the philosophical views the blind spot bundles onto science, and he rejects it [3]. So when Carroll says free will is real because it's a legitimate high-level description of physical processes, Frank would accept the conclusion but not the premise that physical processes are what's ultimately there. For Frank, experience comes first and physics is abstracted from it. For Carroll, words like redness or sadness "describe what you're doing," which is a physical process [1].

Is free will something to explain? Walker offers a mechanism: determinism of the structure, randomness of the environment, history that enables selection [2]. Carroll treats free will as a valid higher-level description that coexists with the lower-level physics [1]. Frank says the concrete "is just given," and an explanation "doesn't add anything to it" [3]. Frank doesn't rule out explanation entirely, though. He asks whether we can account for how agency and autonomy arise, or how agent and world "co-erise" [3]. The difference is one of priority: Walker starts with the physics and works up to freedom, while Frank starts with freedom and asks what physics has to look like to fit it.

Determinism. Walker builds determinism into her account: the deterministic, macroscopic object that is you is what does the choosing [2]. Frank treats the deterministic view of the world as part of the problem, the thing "we already have" because of the blind spot [3]. They can partly be reconciled. Walker's determinism applies to the macroscopic structure, not to the universe as a whole, which she says is random at its base [2]. Frank's target is the global clockwork picture. Still, the word plays opposite roles in their arguments.

Many worlds. This is the clearest split. Carroll is one of the most prominent defenders of the many-worlds interpretation. He thinks cautious physicists are "wimping out," and says of visualizing it: "If you can't visualize it, so much worse for you" [1]. Walker is not a many-worlds person and folds the branches into the single macroscopic object that is you [2]. Frank says many-worlds invents infinite unobservable universes to avoid the idea that measurements matter, and prefers QBism, which puts the agent at the center [3]. So for Carroll, even a universe that branches endlessly leaves free will as real as tables. For Walker and Frank, the way you read quantum mechanics shapes where the agent stands.

Fundamentals and time. Carroll's picture is layered on top of a fundamental quantum description, a "quantum field theory wave function" [1]. Walker denies there is a ground or a theory of everything and wants time treated as fundamental [2]. Frank, following Henri Bergson's challenge to Einstein, says the block universe of relativity is a powerful map, but lived time, like hearing music, has a different structure [3]. Walker's and Frank's views on time are close. Carroll's stated commitments don't require that shift.

These differences are why the agreement matters. Three people with such different metaphysics arrive at the same practical verdict. That suggests the verdict doesn't depend on any one controversial philosophy, and that physics has disproved free will fails under several different, well-developed views of what physics is.

Why It Works

The sources give three distinct mechanisms for why free will survives physics.

Emergence without illusion (Carroll)

Carroll's mechanism is about language and levels. A table is "a quantum field theory wave function," and it is also a table [1]. Both descriptions are correct, and the higher one is the one you need to put a cup down. His criterion is practical and causal: an illusion is something whose belief leaves you with less "power over it and influence over it," like the oasis that isn't there when you walk up to it [1]. Concepts like consciousness and free will don't disappear when you look closely. They keep doing work in explaining what people do, so they count as real at their level.

This fits his broader view of how science treats theories. The summary notes that he thinks a good theory that fits the data means we should provisionally attribute reality to what it describes, and that he accepts structural realism, the idea that good predictions point to something true even when our deepest description changes [1]. A good description of human decisions, one that predicts and explains, earns the same provisional reality.

His work on complexity also points toward where agency comes from physically. Complexity starts low, rises and then falls as entropy increases. Life isn't fighting the second law: "We are surfers riding the wave of increasing entropy" [1]. The stages he describes, from configurational complexity to stars burning fuel to life using information, to minds that imagine futures, place choice inside the physical story as a late product of it [1]. He cites Malcolm MacIver's theory that when fish moved onto land and could suddenly see far ahead, it opened the door to planning and imagination [1].

Determinism plus randomness plus history (Walker)

Walker's mechanism is structural. Start with a random background [2]. Some structures happen to reinforce their own existence, once a loop closes, those objects can be made in high abundance [2]. Over billions of years these structures accumulate history, so that a living thing is an object huge in time, carrying its entire causal lineage [2]. An object with a lot of selection embedded in it can select possible futures [2]. Free will is that capacity in a deterministic structure, exercised against a random environment and over a span of time [2].

Assembly theory gives the history part measurable form. It tracks two quantities: copy number and assembly index, the minimum number of recursive steps to build the object from basic building blocks [2]. Cronin's team has shown that for organic chemistry, molecules above roughly 15 steps show up in high abundance only when life produces them [2]. That doesn't measure free will, but it's an example of the kind of physics Walker thinks is missing: a way to quantify how much history an object carries. Her view that "the most significant struggle for existence in the evolutionary process is not among the objects that do exist but between the ones that do and those that never have the chance to" also frames choice as the steering of what gets to exist [2].

The map is thinner than the terrain (Frank)

Frank's mechanism is epistemic. Science begins in experience, abstracts invariants from it, and builds models [3]. The models are extremely powerful: the rigor behind them "is why cell phones work" [3]. But they are "a thinner, you know, representation" [3]. When the deterministic model appears to contradict the lived fact of choosing, the problem is in the model, not in the experience: "experience overflows that" [3].

He points to quantum mechanics as evidence that physics itself is pushing back against the purely third-person view. The Schrödinger equation works like Newtonian physics, with initial conditions cranked forward to predict the future [3]. Measurement, though, still has to enter separately, and the Born rule says measurements matter [3]. QBism takes this seriously by putting the agent and information at the center [3]. Frank sees the same structure in his Templeton-funded work on semantic information in living cells [3]. He is also interested in informational organization, work on causal closure, and Robert Rosen's contested argument that living systems cannot be fully captured by algorithms [3]. That last idea, as the summary itself notes, is contested.

Where It Applies, and Where It Doesn't

Where it applies. The shared argument answers one specific claim: that modern physics has shown free will to be impossible or illusory. All three scientists reject that inference [1][2][3]. It applies when someone moves from everything is made of particles obeying equations to therefore nobody chooses anything. Carroll's reply is that being made of atoms "doesn't remove the realness from the table" [1]. Walker's is that the chooser is a structure extended in time, not a slice of particles [2]. Frank's is that the equations are the map and experience is the terrain [3].

It also applies to how we talk about ourselves. Frank argues that the "we're just biological machines" story feeds a real crisis of meaning, and the rejection of science that follows from it [3]. Carroll's poetic naturalism holds that talk of beauty, morality and choice is subjective but "not arbitrary," with "a role in describing how the world works" [1].

Where it doesn't apply. None of the three offers a laboratory test showing free will exists, and each is careful not to overclaim. Carroll says he has no "special insight on how consciousness works" [1]. Walker is tentative about consciousness and says we need better ways to think about information materially [2]. Frank's claim that free will "is a given" is a philosophical stance, not an experimental result [3]. On AI consciousness, Walker says we need a testable theory of consciousness before the question can leave the gray area [2], and the same caution applies here.

The argument also doesn't settle which kind of free will you get. Walker's version works over time rather than instantaneously [2]. That's different from a picture in which every momentary decision is uncaused. Carroll's version is a real higher-level pattern, not an exception to lower-level physics [1]. Frank's version is agency and autonomy understood as the same thing [3]. If what you mean by free will is the power to break the laws of physics, none of the three is defending that. Carroll explicitly sees no need to change those laws [1].

Their views also don't transfer automatically to machines. Carroll sees no obstacle in principle to machines that think like humans, but warns that language models exploit our strong bias to attribute intentions to anything humanlike [1]. Frank doesn't rule out machine agency, but doubts current LLMs are there, favoring an embodied, embedded, enacted and extended view of cognition [3]. Walker sees LLMs as a genetic system for language and cautions against treating AI as a separate species, since the technosphere is both human and technological [2]. Whether any machine has the kind of agency these arguments protect is still open.

Finally, the disagreements limit what the consensus can be used for. It can't be used to argue for a particular interpretation of quantum mechanics: Carroll defends many-worlds, Walker rejects it, and Frank prefers QBism [1][2][3]. It can't be used to argue for or against physicalism either [1][2][3]. It establishes a negative point, that physics hasn't disproved free will, while leaving the positive account open.

What To Do With It

1. Stop at the word "illusion. When someone calls free will an illusion, use Carroll's test: an illusion is something that you think you're perceiving that is actually not there, like an oasis that has no causal efficacy" [1]. Ask whether believing in your own choices leaves you with less power to understand and act in the world, or more. If more, the word doesn't fit [1].

2. Separate "made of" from "not real. Keep Carroll's table in mind: Even though I know it's made of atoms, that doesn't remove the realness from the table" [1]. Ask what level of description a question belongs to, since a quantum field description and a talk about decisions can both be correct about the same event [1].

3. Look for your freedom across time, not in one instant. Following Walker, treat decisions as things that play out over time, like deciding to be in another state tomorrow [2]. That makes big choices about habits, direction and who you'll be the central cases of free will, rather than split-second reactions.

4. Take your history seriously as a resource. Walker ties the power to select possible futures to a lot of selection embedded in an object [2]. The history you carry, from your lineage, your culture and the technosphere you're part of, is part of what makes choice possible, not a cage around it [2].

5. Don't let the map overrule the terrain. When a model of the world seems to contradict your direct experience, remember Frank's line: "don't confuse the map for the terrain" [3]. Ask what the model was abstracted from, as in his parable of temperature, before deciding which one to trust [3].

6. Watch for triumphalism, on both sides. Frank identifies scientific triumphalism, science rejection and pseudoscience as three bad responses to the crisis of meaning [3]. When you hear science proves you're just a machine, or science can't be trusted, consider that both may come from the same blind spot [3].

7. Hold the positive account loosely. The three scientists disagree about physicalism, many-worlds and the role of determinism [1][2][3]. Take the shared negative conclusion with confidence and treat any particular theory of how free will works as open.

8. Follow the frontier questions. If you want to go deeper, the sources point to specific areas: Carroll's work on complexity and emergence, including his planned book Complexity and Emergence [1]; Walker's assembly theory and her book Life as No One Knows It [2]; and Frank's The Blind Spot, QBism and research on agency and semantic information in cells [3].

9. Argue fairly. When you disagree with someone about free will, take Carroll's advice about disagreements: state the substantive disagreement instead of psychoanalyzing the other person [1].

Sources

1. Sean Carroll: General Relativity, Quantum Mechanics, Black Holes & Aliens · Lex Fridman Podcast
2. Sara Walker: Physics of Life, Time, Complexity, and Aliens · Lex Fridman Podcast
3. Alien Civilizations and the Search for Extraterrestrial Life · Lex Fridman Podcast