33 | James Ladyman on Reality, Metaphysics, and Complexity
What this covers
James Ladyman, Professor of Philosophy at the University of Bristol, speaks with Sean Carroll about the foundations of reality and how metaphysics should be done in the light of contemporary science. The conversation centers on three interconnected ideas: that what counts as real are real patterns indispensable to prediction and explanation rather than fundamental particles or abstract entities; that structural realism preserves continuity across scientific revolutions at the mathematical level even when old theories are abandoned; and that emergence in complex systems arises straightforwardly from interaction and feedback among parts, requiring no mysterious extra ingredients. Throughout, Ladyman argues for naturalized metaphysics—the view that questions about the nature of reality cannot be answered from the armchair but must be grounded in what current science actually tells us.
The discussion ranges across how we should think about objects and composition, the status of physicalism as a scientific discovery rather than a prior truth, and what "fundamental" means differently in physics versus biology. Ladyman pushes back against what he calls the domestication of science in analytic philosophy—the tendency to distort scientific results to fit commonsense intuitions rather than accepting the mathematical and relational character of our best theories. The conversation also touches on Platonism in mathematics, quantum indeterminism and free will, and the role of feedback in human cognition and social development. A recurring theme is that more is different—that understanding complex phenomena requires engaging with how they actually work in nature, not retreating to first principles or assuming that everything reduces without remainder to microscopic descriptions.
Ladyman argues that what is real is best understood as 'real patterns' indispensable for prediction and explanation, and that metaphysics must be naturalized—grounded in current science (structural realism, emergence in complex systems) rather than done a priori from the armchair.
- To be is to be a real pattern—reality is tied to non-redundant indispensability in prediction and explanation
- Structural realism preserves continuity across theory change at the mathematical/structural level even when fundamental claims are abandoned
- Emergence in complex systems arises from interaction and feedback among parts, not extra magic ingredients
Serious philosophy requires engagement with science, not armchair theorizing divorced from empirical knowledge.
- Serious epistemology and philosophy of mind cannot be done in ignorance of science—epistemology should engage Bayesianism, confirmation theory and statistical inference, and philosophy of mind should engage cognitive science—because, as Quine said, 'philosophy of science is philosophy enough.'
“the idea of doing epistemology and not engaging with that kind of work is a bit odd. And likewise with philosophy of mind, I think come on, you've gotta look at cognitive science.”
- Many of the great philosophers were also great scientists who sometimes spent more time on science than philosophy (Aristotle being an example), so if reincarnated, figures like Aristotle, Kant, Descartes or Leibniz would prioritize catching up on centuries of scientific knowledge over reading pure philosophy.
“if you reincarnated Aristotle or Kant or Descartes or Leibniz, they would be... The last things on their reading list would be pure philosophy... they would be wanting to catch up with the scientific knowledge”
- The ultimate test of naturalized metaphysics is that it must actually contribute to and help scientists interested in how the world is; if it never plays such a role, we should stop talking about it (though no time limit is set for this).
“this should actually help scientists who are interested in how the world is, and it should play some kind of role. And if it doesn't, we should stop talking about it.”
- Naturalized metaphysics rejects a priori 'first philosophy' that theorizes about reality from the armchair without scientific knowledge, because questions like 'in virtue of what do parts form a whole' have no generic a priori answer—cells form an organism through specific interactions, atoms form a table through physics, knowable only by studying the specific systems.
“we point out that in science, composition is dynamical. It's not a static thing... the table is a dynamical entity”
- Naturalized metaphysics is the attempt to say something about the world as a whole in the light of all the science we have—how macroscopic relates to microscopic, biology to chemistry to physics—and is essentially about how science as a whole hangs together, i.e. the unification of science.
“naturalized metaphysics is the attempt to say something about the world as a whole in the light of all the science that we have... naturalized metaphysics is about saying how science as a whole hangs together. It's about the unification of science”
- The engagement between philosophy of physics and physics is very close and healthy and improving: more physicists now recognize that philosophers of physics know what they are talking about, and many younger researchers hold dual PhDs in physics and philosophy.
“more and more physicists are realizing that philosophists of physics know what they're talking about... some of them have got two PhDs.”
Reality consists of patterns indispensable for prediction and explanation, not fundamental stuff independent of our conceptual needs.
- The real-patterns criterion gives a unified account of reality across common sense and science: a table is real because I need to talk about it to get by, and the Andromeda galaxy is real because I need to talk about it to systematize my observations of the universe.
“it gives a unified account of reality between common sense, everyday life, and science. So, for example, let's say the table's real... I can also say... the Andromeda galaxy is real, because I need to talk about it to systematize the observations I have of the universe.”
- The interesting and useful distinction marked by 'reality' is the everyday and scientific one between things we posit but reject (unicorns) and things we accept exist (horses); appeals to some never-found 'fundamental level' as the only real thing fail to capture that distinction and can be dismissed as 'real schmeal.'
“What I'm interested in is the distinction that we make in everyday life and in science, between things that we hypothesized but then decide don't exist... and the things that we do think exist, like horses.”
- What is real is what is indispensable to talk about in order to capture information about the world and make predictions and explanations; on this view it makes no sense to say something is real but cannot be used to predict or explain anything, tying reality to non-redundant information efficiency.
“we use this idea of Dan Dennett that to be is to be a real pattern... what's real is, what it's necessary to talk about if you want to capture information about the world and in particular... want to make predictions and explanations.”
- On the real-patterns view the pattern is primary and the 'thing' is just a way of identifying or talking about the pattern; whenever we investigate actual things they turn out to be collections of relations among other things, so relations may go all the way down, and we should not assume otherwise.
“the pattern is primary, the thing is the way of identifying the pattern, the way of just talking about it.”
- The parts of a table compose a real object because they hang together dynamically and let you track behavior efficiently, whereas an arbitrary compound (e.g. a left earlobe plus Miles Davis' last solo plus the largest elephant in Namibia) gives no extra purchase on the world over treating its parts separately, so it does not count as a bona fide object.
“You combine those things, and you get no extra purchase on the world than you have by talking about each of them separately. And so for that reason, that wouldn't count as a bonafide object.”
Emergence arises through interaction and feedback among system parts without requiring magic, explaining how higher-level structure forms.
- Genuine emergence requires no extra magic ingredient; it arises through interaction and feedback among the parts of a system, as exemplified by ant colonies that, with no central controller, build bridges and even optimize the tradeoff between building a bridge and going the long way—behavior no individual ant computes.
“it arises through interaction and feedback between individual ants. And that's the thing you've got to understand if you want to understand emergence”
- Emergence is precisely the name for whatever goes on when we get biology as well as chemistry as well as physics—the production of higher-level structure out of the basics—making complex systems the area of science giving us the most understanding of how the levels of science relate.
“emergence is precisely the name for whatever goes on when we have biology, as well as chemistry, as well as physics.”
- In science, composition is dynamical rather than static: parts form a whole because of how they are interacting, often on very short time and length scales relative to the scale of interest, so even a table is a dynamical entity rather than a fixed set of parts that happen to be a whole.
“we point out that in science, composition is dynamical. It's not a static thing... the table is a dynamical entity”
- There can be emergent Darwin-like laws applying to particular classes of complex living systems (e.g. the Price equation), but there will not be a single law of all complex systems, because on a liberal view condensed matter, the internet and the economy are all complex systems with too little in common.
“I don't think we're gonna have something like a law of all complex systems.”
Mathematical representation is primary in our best scientific theories and cannot be fully translated into object-property language.
- Many of our best theories—in physics, economics, and mathematical biology—are not easily translatable into object-property talk, so mathematical representation is primary and we should not expect to translate our best theories without residue out of that mathematical representation.
“kind of mathematical representation is primary and we shouldn't expect that we can translate our best theories without residue out of that mathematical representation.”
- Within the mathematical community, Platonism—the view that mathematics has an independent, mind-independent reality that we discover—is not the dominant view; many mathematicians are content treating mathematics as an invented formal system where you set up rules and see what follows.
“within the mathematical community, the Platonism is not... the dominant view of mathematicians, and that's quite interesting.”
- There is something misleading about saying the number two is real and equally misleading about saying it is not real; the real-patterns account is only a partial account of concrete empirical reality, and abstract objects like numbers require a separate treatment.
“there's something misleading about saying that the number two is real, and there's also something misleading about saying the number two is not real.”
Physicalism is a contingent empirical discovery, not an a priori truth, and is compatible with epistemic humility about reality-in-itself.
- Materialism (physicalism) has two sides: a positive claim that everything is ultimately physical or emerges from the physical, and a negative claim that there is no supernatural stuff; physicalism is the updated term used because the classical early-modern conception of matter is not apt for contemporary physics.
“there's a kind of negative side to materialism and a positive side to materialism. And the positive side is some claim about, 'All there is is the physical'... But the negative side is in no supernatural stuff”
- Physicalism about the mind was not an a priori truth but a contingent empirical discovery: in the 19th century it was reasonable to wonder whether there were special physiological or chemical (vital) forces, and it simply turned out there was no need to posit them once chemistry was understood predominantly in terms of electromagnetism.
“David Papineau... chronicled how physicalism is a view about the mind, and more generally was a contingent empirical discovery. There was no reason to think it had to be true a priori.”
- The basic Kantian insight is that the way we conceptualize things conditions our representations, so our theories reflect both how the world is and how we are constrained to think; because we cannot escape our own contingent, evolved cognition, a degree of hedging and agnosticism about describing the fundamental nature of reality as it is in itself is appropriate.
“the basic insight we get from Kant is that there's the possibility that the way that we conceptualize things conditions our representations and that plays a role to some extent, as well as how things are”
Naturalism permits humaneness toward people and does not entail cold disregard; understanding human conditions neurologically improves ethical treatment.
- There is no necessary connection between a naturalistic, materialist understanding of human beings and a cold, clinical attitude toward them; in fact, the more we naturalize our understanding (dementia, autism, schizophrenia as brain conditions rather than moral failings or possession), the more we tend to treat people humanely.
“the more that we've naturalized our understanding of human beings, the more we've been inclined to treat people humanely.”
- Quantum indeterminism does not provide free will because stochastic events do not make actions free; what we want is for actions to be caused by our own deliberations, so a Humean compatibilism—being free when you can tell a plausible story about why you acted in terms of your deliberations—fits best, and free will admits of degrees rather than being all-or-nothing.
“that doesn't make me free, I want my actions to be caused. I just want them to be caused by my deliberations”
- Feedback and interaction are the basis of human cognition: a human who never interacts with other humans cannot learn language or develop many aspects we regard as essential to being human, so it is essential to our natures that we develop in interaction—raising the worry that interaction increasingly mediated by technology or AI could make us into something different.
“If you have a human being who doesn't interact with any other human beings, then they will not be able to learn language... it's essential to our natures that we develop in interaction.”
- There is a saying that human life is based on three illusions: the self, free will, and love.
“The self, free will, and love.”
- A second error in analytic metaphysics is the 'domestication' of science: taking scientific results and distorting them to fit a commonsense, manifest-image picture (e.g. little particles bumping into each other) that physics has rendered outdated, then pretending to talk about scientific objects while not describing them as they actually are.
“take science, and then distort it so it fits your commonsense way of thinking about things and then kind of pretend that you're talking about the scientific objects, but you're not really because you're not talking about them as they are.”
- There is a lot of evidence that pro-social behavior correlates strongly with religious belief, and much of the most important charity work in the world is done by people with strong religious commitment—so being a naturalist does not make one a nicer person than a religious person.
“there's a lot of evidence that pro-social behavior correlates strongly with religious... Religious belief.”
- If there is nothing supernatural and each person has only this one life, then the suffering and curtailed options of, say, a child born in poverty matter a great deal more, whereas believing in an afterlife where the virtuous are compensated could leave one less motivated to address injustice now.
“if you think that kid has only got this one life, then it matters a hell of a lot that their life options are so curtailed... if you tell yourself, 'Well, yeah, but don't worry because they're virtuous. God will make up for it later,' you might be less motivated to do something about it.”
Scientific theories preserve real patterns across revolutions at the structural level even when fundamental claims are abandoned.
- Structural realism is a form of scientific realism compatible with radical theory change: when we abandon theories, continuity with current theories is much easier to see at the mathematical or structural level than at the level of fundamental claims about reality, so what is preserved across revolutions are the real patterns at some level of approximation.
“It's much easier to see the continuity with our current theories at the mathematical or structural level, than it is at the level of the fundamental claims those theories make about reality.”
- Newtonian physics is false in its fundamental claims (absolute time, intrinsic mass, instantaneous action at a distance) yet still captures a real pattern: it remains the low-energy limit of general relativity, accurate within a domain, and just as good at predicting the return of comets as in the 17th and 18th centuries.
“Newtonian physics says there's absolute time. We don't think there's absolute time... And yet I would say there's a kind of reality to Newtonian forces.”
Complexity lacks a single measure or definition; different systems exhibit distinct features each quantifiable separately, not holistically.
- Some complexity measures really just measure order, but order could have been produced by a central controller, which is not what we associate with complex systems; a measure tells you how much order is present, not where it came from, so order is only a proxy for complexity when you already know it was produced by a complex system.
“order could have been produced by a central controller and that's not what we associate with complex systems. The measure itself can't tell you whether the order that you've got where it came from.”
- There is a distinctive cluster of phenomena worth calling complexity, but there is no single thing called complexity and no single true measure of it; complex systems have different features (nested structure, robustness, modularity/division of labor, network and information-processing universalities) and each proposed measure quantifies one such feature rather than complexity per se.
“there is a kind of distinctive phenomena here, but there isn't a single thing called complexity.”