Theories of Everything with Curt Jaimungal - This Quantum Physicist Says The Wave Function Isn't Real: Rob Spekkens

Episode Date: February 2, 2026

From Ancient Egypt to Leibniz... Brand‑new interview out with Robert Spekkens of the Perimeter Institute, one of the sharpest minds working on quantum foundations. In 2004, he constructed a classical toy theory where your maximum knowledge is always incomplete—and out popped the no-cloning theorem, teleportation, and interference effects Feynman deemed impossible to reproduce classically. Spekkens compares our situation to Egyptian hieroglyphs before Champollion: a category mistake where we treat quantum states as descriptions of reality when they actually describe knowledge of reality. If you’re interested in the topics above, you’ll love this podcast. As a listener of TOE you can get a special 20% off discount to The Economist and all it has to offer! Visit https://www.economist.com/toe SUPPORT: - Support me on Substack: https://curtjaimungal.substack.com/subscribe - Support me on Crypto: https://commerce.coinbase.com/checkout/de803625-87d3-4300-ab6d-85d4258834a9 - Support me on PayPal: https://www.paypal.com/donate?hosted_button_id=XUBHNMFXUX5S4 JOIN MY SUBSTACK (Personal Writings): https://curtjaimungal.substack.com LISTEN ON SPOTIFY: https://open.spotify.com/show/4gL14b92xAErofYQA7bU4e TIMESTAMPS: - 00:00:00 - Defining Quantum Innovation - 00:06:40 - Realism vs. Empiricism - 00:12:12 - Leibnizian Methodological Principle - 00:23:40 - Causal Explanations of Correlations - 00:30:24 - Epistemic Quantum States - 00:41:00 - Foil Theory Methodology - 00:54:00 - Causal Influence vs. Signaling - 01:07:27 - Thermodynamics and Ignorance - 01:15:00 - Conceptual Understanding in Physics - 01:21:00 - Philosophy of Physics Utility - 01:30:00 - Speckins' Toy Theory Origins - 01:40:13 - Perimeter Institute's Ambitious DNA - 01:52:00 - PBR Theorem Implications - 02:05:40 - Ontic Separability Assumptions - 02:17:40 - Hieroglyphs and Category Mistakes - 02:29:00 - Revolutionizing Modern Physics - 02:37:20 - Unscrambling Causation and Inference LINKS MENTIONED: Journals, papers, books: - https://www.rwspekkens.com - https://pirsa.org/speaker/Robert-Spekkens - https://arxiv.org/pdf/2507.01122 - https://arxiv.org/pdf/quant-ph/0401052 - https://arxiv.org/abs/0706.2661 - https://arxiv.org/abs/quant-ph/0406166 - https://arxiv.org/pdf/2207.11779 - https://amazon.com/dp/1108066488?tag=toe08-20 - https://www.jstor.org/stable/687269 - https://plato.stanford.edu/entries/qm-copenhagen/ - https://plato.stanford.edu/entries/identity-indiscernible/ - https://www.fourmilab.ch/etexts/einstein/specrel/specrel.pdf - https://plato.stanford.edu/entries/spacetime-holearg/ - https://www.sciencedirect.com/topics/mathematics/hidden-variable-theory - https://www.nature.com/articles/299802a0 - https://arxiv.org/pdf/2011.01286 - https://link.springer.com/article/10.1007/BF02058098 - https://arxiv.org/abs/2005.07161 - https://www.sciencedirect.com/topics/engineering/maxwells-equation - https://www.einstein-online.info/en/spotlight/equivalence_principle/ - https://perimeterinstitute.ca/ - https://amazon.com/dp/9810241054?tag=toe08-20 - https://journals.aps.org/pr/pdf/10.1103/PhysRev.47.777 - https://arxiv.org/abs/1111.3328 - https://www.smithsonianmag.com/history/rosetta-stone-hieroglyphs-champollion-decipherment-egypt-180980834/ - https://www.sciencedirect.com/science/article/abs/pii/S0160932707000282 Videos: - https://youtu.be/gEK4-XtMwro - https://youtu.be/YWbjI-QsH2E - https://youtu.be/fU1bs5o3nss - https://youtu.be/NKOd8imBa2s - https://youtu.be/6I2OhmVWLMs - https://youtu.be/Tghl6aS5A3M - https://youtu.be/HIoviZe14pY - https://youtu.be/bprxrGaf0Os - https://youtu.be/4MjNuJK5RzM - https://youtu.be/c8iFtaltX-s - https://youtu.be/9AoRxtYZrZo - https://youtu.be/uOKOodQXjhc - https://youtu.be/3mhctWlXyV8 - https://youtu.be/gsSJPLX-BTA - https://youtu.be/FFW14zSYiFY - https://youtu.be/HhWWlJFwTqs Learn more about your ad choices. Visit megaphone.fm/adchoices

Most of my best ideas don't happen during interviews. They come spontaneously, maybe in the shower or while I'm walking. And until Plaud, I kept losing them because by the time I write it down, half of it's gone. I've tried voice capture before, like Google Home and just cuts me off in the middle of a thought. And I don't know about you, but my ideas don't come in these 10-second short sound bites. They're ponderous, they wind, they're often five minutes long. And Apple notes, Google Keep, the transcription is quite horrible, and you even have to do multiple taps to get to it.

Plod lets me talk for as long as I want to. There's no interruptions. It's accurate capture, and it organizes everything into clear summaries, key takeaways, and action items. I can even come back later and say, okay, what was that thread that I was talking about about consciousness and information?

My personal workflow is that I have their auto flow feature enabled, and it sends me an email whenever I take a note. I have the note pin S for the shower, and then I carried this one around me in the apartment, and I love them both. very much, especially this one. The fact that I can just press it and it turns on instantly and starts recording without a delay is an extremely underrated feature. And it's battery. I haven't had to charge this since I received it. Over one and a half million people use plot

around the world. If your work depends on conversations or the ideas that come after them, it's worth checking out. That's p-l-a-u-d-a-I-slash-T-O-E. Use code T-O-E for 10% off at checkout. Ultimately, I don't think any of these models are the right picture of reality. We need to try something completely different. Feynman famously said interference is the essence of quantum theory. My work showed that that's just not true. Feynman declared interference as the essence of quantum theory, with, quote, absolutely no way to reproduce this classically.

However, Robert Speckens of the Perimeter Institute proved him wrong. In 2004, he found a theory of a classical world where your maximum knowledge is always incomplete. What happens is out pops the no-cloning theorem, teleportation, and even those interference effects that Feynman deemed impossible. We need to invest in people who have very different ideas of how things are going to go, and some of those are going to be right and revolutionary. I'm Kurt Jemungle.

On this channel, Theories of Everything, we don't sidestep the recondite technicalities of a theory, as that's the whole point of this channel is to go research-level deep, So allow me to explain some of the prerequisites for those who aren't physics professors. Speckins is trying to understand what exactly quantum mechanics is. What features of quantum mechanics are irreducibly quantum and not classical? As you'll see in this episode, he argues it's bell inequality violations. These are statistical correlations between distant measurements that exceed any classical bound.

Most physicists conclude nature is non-local. Speckins disagrees. In 1716, Leibniz had a debate against Clark saying, basically, look, if you move every particle, say a few meters in absolute space, it leaves every single observation unchanged. Thus, that displacement isn't real. Einstein actually wielded this principle. Twice. Number one, eliminating the ether, and number two, identifying gravity with spacetime curvature. This Leibniz principle is called the identity of indiscernible.

Speckens says that superluminal influences violate the same principle. They're constructed to be undetectable. And if there is no empirical difference, then there's no ontological difference. Quite a controversial claim. Now, for Speckens, his target is the entire commodious framework of hidden variable models. That's that apparatus that underlies sciontic interpretations. Now, sciantic is just jargon, meaning that the wave function is real. Like Bohemian mechanics.

It refers to an element of reality. and this contrasts with what's called sci epistemic, which is, again, just jargon meaning that the wave function represents knowledge. Speckins wants to reject this dichotomy altogether. He compares our situation to Egyptian hieroglyphs. Toward the end of this conversation, Speckens says there's a similar category mistake we're making that we used to make when interpreting Egyptian hieroglyphs,

a mistake we're continuing to make in interpreting our most fundamental theory, quantum mechanics. You challenge something quite sacred. So many popularizers of science love to talk about how quantum mechanics is mystical or magical, and no one understands it, bro. And there's some assumptions that physicists, many physicists have, that these quantum quirks or bizarre aspects of quantum mechanics are unique quantum features like interference, superposition, entanglement. They're said to be non-classical.

Build for me the case that it is non-classical just so that people can get some context. And of course, explain any jargon along the way so that no one's lost. And then we'll get to why you protest. Yeah, I guess the standard assessment of, you know, is this classical or not. So first let me put out there that I think it's an important question. So sometimes it gets labeled as, oh, that's just semantics, you know, how we're going to use the language. But ultimately what's at issue here is, you know, what's the real innovation that quantum theory has brought relative to classical theories? I think that's a really important foundational question. So to look at the phenomenology of quantum theory and say, okay, what in there is truly innovative is important.

And traditionally I would say, you know, there was a lot of phenomenology that kind of looked new relative to our classical theories. And so all of that was said, okay, well, that's sort of what's distinctive about quantum theory. And that's fair enough. But a deeper question is, you know, whether something really challenges the principles of classical theories, right? So I might have some very particular classical theories in the past, you know, that theory of particles, the theory of fields. And then somebody might come along with something new, like say a theory of strings. And it's not overthrowing the principles of classical physics. So the principles of classical physics might still hold. So it might be a

the dynamics is still described by a Lagrangian and all the usual stuff still holds, it's just in a slightly different context. So the more interesting question to some extent is which of the quantum phenomenology challenges the principles of classical physics. So yeah, the standard list

is kind of things like quantum interference, non-communativity of measurements, the fact that there's a no cloning theorem, a whole bunch of phenomenology, of quantum information theory. So, you know, the fact that you cannot discriminate non-orthogonal states,

the fact that there's this property called entanglement, and entanglement has various features like monogamy. It can't be shared between more than two parties. And there's just a long list of things that you would put, say, most people think that this is quintessentially quantum phenomena. Okay, but? So, yeah, the, The but is that I think most of that long list is not actually violating the principles of classical

physics. So I have a different attitude. And what I really want to do is sort of find what more subtle features perhaps of that operational phenomenology actually challenges our classical principles. So to do that, I think I need to take a step back and kind of tell you about what my principles are, where I'm coming from, because otherwise none of it makes sense. So often in the discussion, of foundations of quantum theory, a lot of somebody's research program has to do with philosophical principles. And so if we don't start there, it won't make sense. So let me take a detour and then we can get back to sort of which phenomenology, which of the phenomenology of quantum theory is truly distinctive. Please. So the most important thing I think in discussing

interpretations of quantum theory is, you know, where do you stand on the dichotomy and the philosophy of science between empiricism and realism? Okay. So the empiricist point of view, is that it's the job of a theory to describe what we observe in experiments, and it shouldn't really go beyond that, right? So that ultimately all of our descriptions should reduce to statements about what's happening in an experiment. Sorry, just a moment. Would the empiricists say that it shouldn't go beyond that or that it just doesn't happen to go beyond that, necessarily?

I think a real empiricist would say that it is, you know, somebody like mock, would say it's fundamentally inappropriate to ask for more than just a formalism that allows you to kind of compress the data, reproduce the phenomena. So yeah, there's a sense that wanting something more like an explanation or some deeper understanding is fundamentally inappropriate. And in particular, the sort of conceptual building blocks

ought to be facts that you can't really dispute. like facts about experiments. So the philosophical tradition, I think, goes back to people kind of wanting certainty. So the fact that you can't really be mistaken about the things you've observed means that if you sort of define everything in terms of what you observe,

you will immunize yourself against being wrong because, you know, you're wedded to things that you just can't doubt. So that's kind of the background. I think that's where empiricism came from. And realism, of course, is looking, is granting that there's, could be concepts that are sort of not manifest, they're not defined in terms of sort of what we

observe. We still have to recover what we observe, but, you know, the deeper description might be in terms of concepts that are abstract, and we have a kind of story about reality. And so that's a big dichotomy in views about quantum theory. And the Copenhagen interpretation, you know, that means many things to different people, but there's a kind of common approach which says, look, let's just be good empiricists. Let's be, in the quantum context, we call it, you know, operationalists. Let's just grant that quantum theory gives us an algorithm for predicting the outcomes of experiments when they're well described. And that's all a scientific theory should really do. So we should be happy. There's nothing more to want. Whereas the realist is looking

for explanations of those predictions, something more. And so from my perspective, I'm a realist. So I'm on the realist side of that debate. And the thing that I see wrong with empiricism as a kind of fundamental philosophy of science, is that, first of all, it's sort of inappropriate to ask for certainty. So the kind of philosophical tradition that led to it was kind of based on a fallacy, I think, the idea that we could possibly kind of believe things and not risk being wrong. But more importantly, I think it's just an illusion that we can kind of go to the world, get data, and compare it to some theory, because all observations are ultimately theory-laden. So you can't really say what did the experiment yield, you know, what did it tell

me, without bringing your realist, you know, presuppositions to bear. So if I'm looking at the, you know, positions of stars on the horizon at night through my telescope, I'm bringing a theory of optics to bear. I'm taking into account how those light rays are refracted by the atmosphere and my hypothesis. And a whole bunch of theory is going into the interpretation of those observations. So I think it's an illusion that, you know, we, we, we, we, can kind of agree on what the observations are because our interpretations, those observations are always infected by our views on, you know, what's really going on. So nonetheless, I think, you know, to your point earlier, that there is a kind of weaker type of empiricism that's sort of

more a methodological principle that says, look, as long as we don't really understand what's going on, it's a good idea to try to describe what we know in kind of minimalist terms that most of us can agree with. So if we just talk about the statistics, the relative frequencies of outcomes and an experiment, then typically we can sort of agree on that despite our interpretational differences. So it's sort of a minimal statistical interpretation of what's going on that we can all agree with. And it's a good way of freeing yourself from some presuppositions about what reality is like. So Einstein famously used a kind of operationalist methodology when he started thinking about synchronization of clocks, for example, in developing special relativity.

So he wasn't denying the importance of having a realist interpretation, but he was using, kind of restricting his set of concepts to operational ones in order to make progress, in order to sort of free himself of all the realist presuppositions of the theories that came before. So I think in quantum foundations, you know, that's something, even the realists should be using that tool, you know, thinking about, okay, well, what is it that we have to explain? Answer, it's the operational predictions of quantum theory.

That's sort of our goal of what we're trying to reproduce. with our realist models. So, you know, Bell famously, in his theorem, connected sort of what you would see in an experiment to some deep principles about the nature of reality. Anyway, so for me, I kind of like a middle road. So realism kind of informed by operationalism. And there's a specific constraint that I want to impose

on the realist interpretation. So for me, this is the most important principle in my research program. And I think most physicists appeal to this principle. It sort of shows up. And if you've done a degree in physics, you're sort of familiar with the reasoning. But strangely, we don't usually talk about it as a principle.

So I call it the Leibnizian methodological principle, because it goes back to Leibniz. So the way I want to articulate it, it's a version of what Leibniz called the identity of indiscernible. So the idea is, you know, I like to say it's the on. ontological identity of empirical and discernibles. So what does that mean?

Say I'm a theorist. I have some realist theory of the world, and I imagine two scenarios, two distinct physical scenarios. And I noticed that those two scenarios predict that what you would observe empirically are exactly the same. In that case,

if those are considered to be ontologically distinct, like they describe different states of affairs according to my theory, I should reject that theory, because that would be a situation where you have no ability to empirically distinguish these two scenarios, and yet your theory says they're ontologically distinct. So Leibniz, for example, criticized Newton's approach with its positing of absolute space

by saying, look, if I took every particle in the universe and moved it over by five feet in absolute space, all observational data would be exactly the same. And therefore, there isn't some real degree of freedom associated with the positions of particles in absolute space. there's only the relational degrees of freedom amongst the powers. That's the only thing that's real. So there he was appealing to the principle that you should eliminate any aspects of the ontology that don't have some empirical consequence. So I think that that principle shows up everywhere in physics and we should use it. But the main reason for thinking it's a good principle, I would say, is the use that Einstein put to it. So if you look at Einstein's work,

I would say there's strong evidence that this was the principle that really guided him. So Einstein, of course, knew about Leibniz's work, and he was influenced by Mock and Poincarre, who were big fans of Leibniz. But if you look at his 1905 paper on Special Relativity, for example, so there, he's criticizing

ether theories. And he says, look, in the ether theory, you know, you can imagine, say, like a coil and a bar magnet, and you can imagine the coils at rest with respect to ether, and the bar magnet is moving. And then here's a different scenario. It's the bar magnet that's at rest with respect

to the ether in the coil that's moving. And so in the case, the first case, you get a current in the coil, and it's because, you know, the time-varying magnetic field that the electrons in the coil see has a force associated with it. Time-varying magnetic fields,

introduce an electric field, which means a force on the electron, so you get a current. In the other picture, you have a static magnetic field, but now the electrons in the coil are moving through that static magnetic field, so they feel a Lorentz

force, and that generates a current. But Einstein notes the current's exactly the same in the two situations. There's no way to arrange this kind of experiment in, or any electromagnetic experiment, that responds to anything except the relative motion of the two. So to imagine that there's some significance, some ontological significance to the actual state of motion relative to the ether is to deny Leibniz's principle. So he says we have to come up with a way of thinking about things where it's only, you know, we have to enforce liveness as principle. It must be that it's only the relational motion that's important. And the formula of special relativity is Einstein's attempt to say,

okay, here's how we can have a picture of reality where, you know, any two ontologically distinct scenarios correspond to empirically distinct observation. He does it again in general relativity. If you look at what's the, you know, the famously's most beautiful idea, it's the equivalence principle. following Galileo's thought experiment in a ship, but with Einstein, it's an elevator.

You know, on the one hand, you have an elevator that's at rest in a gravitational field associated with some acceleration. On the other hand, you have the same elevator that's accelerating through free space, absent of any gravitational fields, with a rate of acceleration that corresponds to the force of the gravitational field. And then famously, you see no differences. Any experiment you do looks exactly the same. So Einstein says, well, any theory of gravity that says that accelerating through free space and being in a gravitational field are distinct runs afoul of liveness of principle because there's no empirical difference. No experiment I can do can tell the difference between those two situations. So he says we have to have a picture of reality that does away with that distinction. And that's what general relativity achieves. It says, oh, the local curvature of space time is exactly the same in those two circumstances. And that's all that matters for the empirical phenomena.

There's one other example where, like, the very last step in 1915 before he got to general relativity, he was trying to enforce general covariance. And he had this thing called the whole argument where there was sort of diphthomorphism you could do in just one region of space. And it looked like his theory was sort of underdetermined that there was, you know, the predictions you would make would depend on whether this dipheomorphism had been applied or not. And so that kind of held them back. And then he suddenly realized, oh, wait a second, the only thing that's empirically significant are point coincidences, like when two particles arrive at the same point. And that is invariant under diffiomorphisms. So he suddenly realized, oh, these diphomorphisms, I shouldn't think of them as ontologically distinct situations, right? Because there's no empirical consequence for these things. So then when you realize that, and he said, okay, the only ontologically significant thing is sort of what's invariant under diffiomorphisms. because of Leibniz, he wanted to enforce that.

He was able to see how to impose general covariance, and then he had general relativity. So for me, this principle has been extremely effective, you know, in the development of special and general relativity. I think it's very plausible. I think it's a great principle, you know, for us to be using, and I think we use it all the time. So to some extent, it's the thing that I hold on to.

And I want to ask the question, you know, what interpretation of quantum theory led to if the principle on which you don't compromise is this Leibnizian principle. You know, Einstein, of course, could have made very different decisions at various points when he, you know, encountered difficulties. He could have given up on Leibniz and tried something else. So I see a lot of his success as being due to sort of never compromising on that principle. And so that's kind of dictates my research program as well, that, you know,

what do we get to if we don't want to compromise on Leibniz? Now, are there counter examples to this Leibniz principle? Well, it's not a principle that you can go and test. It's rather a principle that constrains your kind of realist theories that you ought to construct for a given set of data. Sorry, I'll give you an example. My understanding is that French and Redhead in 1988 or so, they showed if you have two bosons in a symmetric state, maybe you've seen this already, that they can share all their properties that neither has some definite position and they're all symmetric under some relational properties,

yet we still would call it a member of a two-particle state in Fox space, and not just one. With Amex Platinum, $400 in annual credits for travel and dining means you not only satisfy your travel bug, but your taste buds too. That's the powerful backing of Amex. Conditions apply. Yeah, I think you're referring to the place in physics where people usually talk about liveness as principle of the identity of discernibles is when we're talking about

particle statistics, bosons, ruminons, that sort of thing. So I'm, you know, the first thing to emphasize is I'm using in a very different way as a kind of methodological principle for like, what kinds of realist theories should you entertain? So it's, yeah, I don't really have a strong view on what should we think about, you know, the identity of two particles. I think that's, that's kind of a different direction in which to take Leibniz's ideas. So I'm more, I was led to it because when you study non-contextuality in quantum theory, you can ultimately justify it via Leibniz's principle. So that's sort of one of the reasons it appealed to me that is sort of the principle

behind non-contextuality. But ultimately it's, it is a principle, right? So I think we can argue about whether it's something we should impose or not. It's not something we can really test. I brought up Einstein in the historical success of the principle because I think in physics, it is wise to sort of look at the history and say what kinds of principles have tended to lead to progress, you know, things like, you know, energy conservation is a great principle where, you know, you ought not to give it up if you can avoid it. That generally progress comes by being very

conservative at the principle level. Most of the principles that have proven to be effective in past physical theories survive in the new theory. And it's usually something more subtle that changes, something about the background assumptions, not these principles. So that's the spirit in which I'm using it. So it was quantum contextuality that led you to think about this modified Leibniz principle? Yeah, I think, I mean, I knew, I guess, something about the literature on kind of relationalism in the study of gravity. And I was sympathetic to the, the relationalist program, which really sort of comes from Leibniz, ultimately, Leibniz and Poigens. So I think I was already open to that. And maybe it was my sort of knowledge of, you know,

like Leibniz articulates this pretty well in the correspondence with Clark, which is basically Newton's spokesperson. So I knew about that. And then I, I guess when I was studying non-contextuality, I sort of saw that same pattern. And so I made the connection. And yeah, that's, so now I would say it's Leibniz I'm really committed to. And it's the thing that, underwrites my belief in things like non-contactuality and locality and other principles. So, anyways, that's principle number one. And then maybe the second thing to say about my views on realism is, realism is kind of nebulous, like what counts as a realist interpretation of quantum theory. And so where I've landed is that there's one aspect of realism that I think is important.

So this is the kind of realism I want for quantum theory, which is that correlations should have causal explanations. So, you know, if somebody says, hey, there's this experiment and there's some correlations, but I don't feel any need to answer the question of whether, you know, these two variables are correlated because one influences the other or because there's a common cause if they just say, no, I don't care about why I just say they're correlated. And for me, that's a kind of anti-realism. Because, you know, in the context, so I spent a lot of time thinking about modeling things

causally, and there's this field of research called causal inference, which people in statistics and computer science study. And, you know, if I have data that tells me, you know, taking some drug is correlated with recovering from some condition, well,

that doesn't mean that the drug was effective, right, because there are other ways of explaining that correlation. So it could be that, you know, gender influences both whether you take the drug and whether you recover. So men are more likely to recover from this condition than women are, but men are also more likely to go out and get this particular treatment or medication. So in that case, it might be that there's no causal connection at all. And when you learn

that somebody took the drug, you do update your probability about whether they recovered. You say it's more likely that they recovered, but not because of the causal effectiveness of the drug. Rather, you say, well, because they took the drug, I infer that they're more likely to have been a man, and because they're man, they're more likely to have recovered. And so that's the explanation of the correlation. So in that context, you know, you wouldn't want to say it doesn't matter what the causal explanation is because obviously, you know, whether it's a common cause or cause effect entirely determines whether you would want to take this drug or not. That's really the question you want to know, is the drug causally affected? So I think across science, what we're often

looking for is causal explanations of the correlations we see. And I think it's no different in quantum theory that to understand what's really going on, we need a causal account. And so I'm willing to give up on certain aspects of classical realism or the conventional framework for thinking about realism. But at the end of the day, I want to insist that it has to provide causal explanations of correlations. So that's kind of where I'm coming from. And ultimately, these two things are in tension with one another. So there's commitment to liveness as principle and wanting causal explanations and correlations, there's a strong sense in which their intention, if you want to understand quantum theory. And that's because we have these no-go theorems that basically say,

if you believe the quantum predictions are correct, and you subscribe to the conventional framework for describing realist theories. So I often call this the ontological models framework. It's sort of a picture of reality where you say, okay, variables are just things that take a set of possible values. Dynamics is basically functions. My ignorance of the values of things. Well, that's described by Bayesian probability theory. And I can use that to calculate, you know,

what sorts of relative frequencies we should see in experiments. So this is, you know, exactly the kind of thing that used to be called a hidden variable model that Bell used and Cotian Specker used and proving their no-go theorem. And what you can show is that this commitment to the Leibnizian principle, actually, it's strong enough to get you to Bell's notion of local causality, to get you to to a caution and speckers notion of non-contextuality, which we may talk about later. And so really all these no-go theorems can be thought of as

if you subscribe to the conventional framework for realism and you believe in this Leibnizian principle, you're going to get a contradiction with the quantum predictions. So, you know, what are you going to do in the face of that? So I would say a common response is to kind of give up on Leibniz, right? So when people say, let's imagine that there are superluminal influences that explain Bell inequality violations in a Bell experiment. Or let's imagine that there's some hardwired context dependence

in my hidden variable model. They're giving up on liveness. And I'm unwilling to do that. So what I would rather say is, well, there's aspects of the conventional framework for realism that we can relax. We don't have to do it the conventional way. So if we all were demanding is that we have causal explanations and correlations,

we can have certain kinds of realism. They're just not going to meet the, you know, they're not going to be described by the classical framework. And so that's sort of my attempt to make sense of quantum theory is that, you know, roughly speaking, the innovation is going to come in how we think about causation, how we think about inference, right? So those are going to be two really important elements that we need to unscramble. Like, what's about reality and what's about our knowledge of reality? So the drug trial example was a really, you know, nice example of that, which is like, you want to distinguish whether, you know,

learning about whether somebody takes the drug merely informs you about whether they're going to recover. You know, so that's just the inference. Right, right. Or whether, you know, the taking of the drug actually causes them to recover. So a lot of what I do in quantum mechanics is trying to, you know, what Jane's called the omelet of epistemology and ontology in quantum theory. So I want to know what's about causation, what's about inference. And so roughly speaking, you know, for me, you know, there's going to be quantum notions of causation and inference, and they are going to be innovations relative to the classical notions of causation and inference, much like, you know, if you think about the notions of space and time,

relativistic notions of space and time are innovations on the pre-relativistic notions. So they're still recognizably about space and time, but there's certain aspects of the conventional notions that we have to give up. And so it's similar here, that there are certain aspects of the notions of causation and inference that we're going to have to give up when we go to a quantum theory.

So that's the background, the research program. So maybe now is a good time to come back to your question about, you know, why the usual list of surprising quantum phenomena, I don't consider to be surprising. And the reason is that if you, you can write down some models that are classical, you know, they obey classical principles. And the only real innovation relative classical theories is that these models say there's a restriction on how much you know. So there's sort of like, if you ask what's the maximum amount of knowledge you can have about some physical state? The answer is it's still incomplete knowledge. So, so, okay, so for me, this came out of thinking about the nature of the quantum state that I'd heard some, you know, a long time ago, I heard about arguments in favor of the epistemic nature of quantum state.

So that means quantum states are representing knowledge. And I was sort of exploring, okay, how much of the phenomenology of quantum theory can be explained if that's the view we take? And it became apparent that you can explain a lot. So, you know, what's a good example? The, like, okay, in quantum information theory, we have this property that if you give me two non-orthogonal quantum states, then there's no measurement that allow you to determine with certainty which one it was. So if they're orthogonal, then there's a measurement that will distinguish them. But if they're not orthogonal, so in the block sphere picture, they're not antipodal,

then there's no quantum theory predicts. There's no measurement that allow you to discriminate them with certainty. If you take the view that the quantum state just describes all the properties of the physical system, so that's the view that is sort of a complete description of the reality of that system, then it's kind of surprising. It sort of suggests that, oh, there's a limit to the kinds of measurements we can do. if you take the view that the right way of thinking about quantum state is that it's epistemic, then a classical analog of a quantum state is not a point in a physical state space,

but rather probability distribution over the physical state space. So one of these quantum states might be a distribution like this, another could be a distribution like this, and being non-orthogonal course costs to those distributions overlapping somewhere. Right. If you say, okay, let's think about state discrimination in quantum theory from that perspective. So now the perspective is, you know, I give you a system, and I tell you, look, it was either sampled from this probability distribution over here or it was sampled from this distribution over here. Your job is to figure out which it was.

Now, these distributions overlap. So some of the time, the physical state I'm sending you is in the region of overlap. And in that case, obviously, you're not going to be able to tell which distribution was sampled from. If you figure out that it's in a region that is only consistent with distribution one, you can say, oh, it was definitely a distribution one or similarly with distribution two. But if it's in the region of overlap, you'll have to say, I don't know. And so you expect, if quantum states are interpreted this way, that you ought not to be able to tell which distribution was sampled from.

And indeed, that's what quantity is. Okay, let me small this out with a different analogy for some of the regular viewers of this channel. men and women's heights are normally distributed, but men tend to be slightly taller. So if you were to look at people who are seven foot and above, and you knew that seven foot and above were only men, and I told you, okay, I'm just going to tell you the height. I'm not going to tell you the sex of the person. I want you from the height to infer the sex. Right. 7.1 foot. What is the sex of this person? You say, well, it's a male because only seven foot and above are male. And then I say, well, five foot, five. Then you're like, well, I have no clue.

I mean, 40% of them could be, or 60% could be women and 40% could be men. You could say something like that. So you have, in the overlap, you have some uncertainty. And you could also have a, in this case, you could have, it's not just, you're completely uncertain, you can have a model for it, like 60% women, 40% men. Yeah, yeah, the, I like that example. Maybe somebody could say, look, there's always a non-zero probability that there could be a woman who's seven foot tall.

Yes, sorry, let me modify that. It's not a normal distribution that, what I just said. Otherwise, there would always be a non-zero probability, right? Yeah, it's like that, I mean, if I have one deck of cards and I say, look, there's only hearts and spades here, and here, there's only hearts and clubs, right? And now I draw a card and I give you one, and I say, well, do you know which deck it was drawn from? Well, if it's a spade or a club, you're like, yeah, I know exactly which deck's from.

But if it's a heart, you're like, well, that's consistent with both, so I can't tell them apart. So that example makes it clear that, okay, there's zero. probability of drawing a spade from here, and there's zero probability of drawing a club from here. So when I get those results, I'm certain. And that's kind of how these models work, that you have classical probability distributions that assigns zero probability to certain outcomes. So that's one example of like the phenomenology of quantum theory that you cannot discriminate non-orthogonal states arises in models where the pure quantum states get represented as

probability distributions that are not point distributions, right? So they have some spread, right? They're consistent with many physical possibilities, and they tend to overlap. So you can explain, you know, just that simple idea gives you intuitive accounts of a lot of phenomena. So if you look at the no cloning theorem, for example, and you ask, well, let's phrase it in terms of probability distribution. So what is no cloning? It says, look, if you give me a system and you tell me it was either quantum state psi 1 or side 2

and they're not orthogonal and now you have another system and you're trying to sort of copy the quantum state onto the other system. So you're if it was si 1, you want to be si 1 tens or sci 1 at the end, right? Two copies of si 2. If it was si 2 you want tensor si 2 at the end, two copies of

side 2. So it's the same. If I say, well, the way you ought to understand that is somebody's giving you a sample from one of two distributions and they overlap. Now please make it so that at the end of the day, your knowledge, your state of knowledge about the pair of systems is just, you know, whatever the probability distribution pertained initially, it now pertains to both. But there's no way of doing that. If I copied the physical state, I would generate correlations

between the two, and that's not represented by a product distribution. And there's that, you know, there's always these zontic states that are in the overlap of both. So you cannot figure out what the distribution was by, you know, measuring it. So you can prove that there's no way of doing this with classical probability distributions. It's just a very natural fact about distributions. And so you start to realize that when you look at what happens with quantum states, so, you know, they can't be cloned, they can't be discriminated, you can teleport them in a certain protocol, you can steer them in the Einstein-Padolsky-Rosen steering experiment.

You can sort of look at all these things and say, is it the case? that that sort of stuff happens with probability distributions as well. And the answer comes back, yes. And in particular, if you have a model of a classical world where there's a fundamental limit on how much you can know, so your probability distributions are never too narrow, they're never down to zero entropy, then you can do a really good job of reproducing

the qualitative aspects of this long list of things that occur in quantum theory. So that's what my toy. theory from 2004 does. It sort of shows what is all the phenomenology you can reproduce. And that's what led me to say, well, these are, the sort of standard list is not the thing we should really be focused on if we want to know where the real innovation of quantum theory lies. So it's not that I think these models are empirical competitors to quantum theory. They're not. Because there are things that they can't reproduce, right? So Bell inequality,

violations provably, like I mentioned earlier, there's no-go theorems that say if you subscribe to the conventional views on how to think about realism, which these toy models do, and you believe in this Leibnizian principle, which means that you have to have locality and non-contextuality, and the toy models have all that. Then there's some phenomenology in quantity that you will not reproduce, and Bell inequality violations are one of them. So this is a particular kind of set of correlations you see in a Bell experiment. So from the very beginning, it was sort of clear that these models aren't competitors to quantum theory. But we study them as foils.

So I like to talk about the methodology of foil theories. So a foil is something that you study because it contrasts with the thing you care about. So we care about quantum theory. We want to understand it. But part of understanding of theory is knowing what are the other possible ways the world could have been?

So my toy theory is, you know, you can think of it as a way the world might have been, it's different from quantum theory. Okay. But by studying it, I realize, you know, what is truly distinctive about quantum theory? So this toy theory is very conservative in terms of like it imagines the classical reality

and a relatively modest change to our classical theories, which is we can't know everything. And yet, it reproduces a whole bunch of the phenomenology. So what is it that's truly distinct about quantum theory? I can learn a lot about that

by studying these kinds of foils. So ultimately, you know, here I study these, these foils as a way of saying, well, what are the phenomena, the operational phenomenology of quantum theory that really do resist the kind of conventional framework for realism together with the Leibniz and principle? They cannot be interpreted in that way.

Because that's your best guide to how to make progress in a research program where you're trying to hold on to Leibniz and innovate on the framework for realism. Distinguished for me Toy Models versus Toy Theory. Well, I think the way they get used is often synonymous, but personally I like to use the word model when I'm modeling something. So a model of quantum theory might be like a particular account of how you reproduce the quantum predictions, whereas a theory could be an alternative to quantum theory that makes different predictions. So this toy theory I was talking about is an example of that. It makes different predictions from quantum theory, so in certain circumstances.

And so I don't think of it as a model of quantum theory. It's rather a different theory. And the word toy theory was meant to just kind of a list, make it clear that this is really being viewed, not seriously as an empirical competitor to quantum theory, but as a foil, as something that will help us learn in particular what principles might underlie quantum theory. So there's this research program,

where you imagine a landscape of possible ways the world could have been. So people use different formalisms for talking about points in this landscape. One is called the framework of generalized probabilistic theories. And essentially it's sort of like saying, look, the operational stuff that we can all agree on, like what are the statistics we see in experiments, that's what can vary as I vary over this landscape. And, you know, one of the things that people in quantum foundations like to do is try pick out some axioms that might find locate quantum theory in that landscape. It's the only theory that, you know, abides by these particular principles. So, like, you know, you might have said,

oh, well, the only theory that allows you to violate bell inequalities while still not signaling super luminally is quantum theory, right? So that would be a conjecture. Well, that conjecture is false because you can find a foil theory, another point on the landscape, where you violate bell inequalities more than quantum theory does, but you still can't send any signals. Okay, so this was famously proven by Pepescu and Roerlich, and you can develop those ideas into a whole theory, which is sometimes called box world. So that's like a point in the landscape, different from quantum theory. Right. And so, you know, if somebody thought that those principles could get you to quantum theory, you know, box world shows no.

No, there's alternatives to quantum theory that satisfy those principles. So you haven't captured the true essence of quantum theory yet. Similarly with these toy theories that, you know, if you thought that, I don't know, no cloning and teleportation and some other list of phenomena was enough to get you to quantum theory, these toy theories show you no. Like that that phenomenology is reproduced in other theories distinct from quantum theory. So it cannot pick out what's truly distinctive about quantum theory. I was guilty of multiple skin care crimes. Two counts of sleeping and makeup.

One count of using disposable wipes. I knew my routine had to change. So I switched to Garnier-Missler water. It gently cleanses, perfectly removes makeup, up and provides 24-hour hydration. Clear away the evidence with the number one Missler Water worldwide by Garnier. Okay, so if I'm understanding you correctly, there are various features of quantum theory that I mentioned earlier, such as superposition entanglement and interference, but there are

others like Bell correlations and maybe four more here, or 20 more, whatever. The point is that you then wonder, okay, of these features that we think of as being non-classical, can we develop a classical theory that triggers some of them? And you have found, yes, for interference, entanglement, and superposition. Yes, the answer is yes. But you have not been able to find it

for bell correlations and so forth and so forth. And then I heard you say, or I thought I heard you say, that we have shown that you can never recover some of these fingers here that I have on screen, classically. but it sounds to me

like the latter part of what you're saying is that we just have not found a point that can actually recover them. So one is that there doesn't exist something versus that we have not found it. So it's the first one that it doesn't exist. So the key is that

when we're showing that something can't be explained under some set of principles, we're proving a no-go theorem. So you might say take the conventional framework for realism, which I like to call the ontological models framework. It's about hidden variables, Bayesian probability theory, that sort of thing. Assume something called local causality, which is Bell's way of talking about things being local.

I like to think of it in a causal model. There's only a common cause between the two wings of the experiment. Relative to those assumptions, you can prove that the only correlations you get will satisfy certain inequalities. The quantum formalism violates those. inequality so you can say, right, I now have a contradiction from those assumptions. And so I know that, you know, one of my assumptions has to be wrong. So it, you know, might be the local causality. And a lot of people say that was the wrong assumption that actually there are superluminal influences going on, you know, that that's what's innovative about quantum theory. But I'd rather say, no,

it's still a common cause explanation, but it's this conventional framework for thinking about realism, for thinking about causation and inference. That's what's going to change. But it's, it's, it's It's not that we don't know whether there's some model that can reproduce this. It's rather we've proven a theorem, right? So there are certainly models that can abide by the conventional framework for realism and violate Bell inequalities. It's just that those models run afoul of Leibniz's principle. And so that's why I don't care about them, right?

So let's say I try to build a model that really has superluminal influences in it. So that's how I explain what's going on in Bell experiment. I say, okay, the set, you know, we have Alice and Bob, there's a pair of particles that gets prepared at a source, and each one gets a particle. And then Alice has a setting variable that determines what she's measuring in an outcome, similarly for Bob. So if I allow Bob's outcome to depend not only on, you know,

the physical state of the particle he got and the setting variable, but Alice is setting variable as well. So that would have to be a super luminal influence because they can do these measurements at space-like separation. So that influence is going to have to go fast and speed of light. if I allow that, then yes, I can explain the bell inequality violation. Okay, but why should we be dissatisfied with this? Well, you know, the usual story you hear is, you know, people who trust relativity say,

well, it just feels like relativity is not merely saying that signals can't travel fast and the speed of light. It's saying that influences can't travel fast in the speed of light. So you might just say, well, it seems intention with relativity to say that there could be an influence that travels fast and speed of light. but you can do better than that and you can just say, look, if you believe in Leibniz's principle, then this is clearly not Leibnizian because, you know, imagine, you know, Einstein's elevator thought experiment. I go to Bob's lab and I do every experiment you can imagine looking to get some information

about what Alice's setting was. And lo and behold, no matter what I do, according to relativity theory, because there's space like separated and there's no signaling, nothing I do in Bob's lab can teach me anything about what's happening inside Alice's lab. So, yeah, what I do in Bob's lab can't teach me about Alice's lab. So at the empirical observed phenomena scale, there's no evidence of any influence, no signals for any experiment that you could do. But to imagine that there's nonetheless an influence, well, that's violating blindness. That's saying that there's an ontological difference, but you cannot see it, right?

There are no experiment you do will see that difference. So that's the sense in which Bell's notion of local causality fall, from Leibniz's principle. And that's why, you know, I would not want to give that up. So it's the conjunction of Leibniz's principle together with this conventional framework for realism that leads you to contradiction. And so that's the sense in which I could say, look, there's a certain kind of, you know, there's no toy theory-like construction, you know, that it'll satisfy Leibniz, it'll be in this

conventional framework that's going to reproduce the Bell inequality violation. that's clear. Now, more generally, that's my take, but more generally, what I'd say is, you know, in quantum foundations, it is good to follow this methodology of no-go theorems, right? So, like, if we just argue about, oh, I got this model, I want to call it classical,

I think it's nice. If I look at what Feynman wrote about things like interference, so, you know, Feynman famously said, interference is the essence of quantum theory. So some of my work shows that, well, that's just not true. Like, I can reproduce. You know, Feynman said,

there's absolutely no way to reproduce this classically. I say that's not true. I can have a classical model reproduces the phenomenology. And the problem with what Feynman said was that he didn't prove a no-go theorem. He didn't say, this is what I mean by a classical explanation, you know. Right, right, right. Here are the predictions of quantum theory. Here's the contradiction.

If Feynman had done that, then all we could do in response is maybe dispute those assumptions. We might say, well, that's not how I think about classicality. So I think we escape the contradiction by giving up assumption one. And that's what's so nice about the methodology of no-go theorems in quantum foundations. Like we all researchers in quantum foundations tend to disagree about what the fundamental principles are. But what's great about a no-go theorem is that if you agree that it's a logically valid argument, you have to say which of the assumptions or many of the assumptions that you disagree with and why. And so it focuses your attention on the principles and how should we articulate these principles

and are they reasonable or not. So, yeah, that's the sense in which somebody could always say, all right, you've proven a no-go theorem, but here's why I don't think it's very significant because I think the assumptions you put into it aren't natural assumptions. So most people agree that Bell's assumptions and Cotian Specker were pretty natural in some sense.

And so those no-go theorems are significant, but, you know, other no-go theorems might not be because they're built on assumptions that are just not natural that nobody really thinks are plausible. Can you explain to a mathematician, what is special relativity such that, so the mathematician may say something like,

it's invariant under S-O-1-3, some theory that's invariant under that. Now, of course, there may be more conditions, but either way, once you've specified what special relativity is, what the heck is the difference between something being non-locally influenced and then, some people will say,

well, special relativity allows non-local influences, it just doesn't allow you to communicate non-locally or superluminally. Where is that said in special relativity? Like, explain to a mathematician. Yeah. I mean, if you take, say, Maxwell's theory of electrodynamics, then it's sort of clear that you have neither any superluminal signals

nor any superluminal influences. So it has the Laurentian symmetry, you know, it has the right symmetry to be consistent with special relativity, and it has neither of these features. It's only when we start asking about quantum theory, characterized totally operationally, where we're like, okay, we just have these statistics of outcomes of experiments.

And what Bell asked was, is there any picture of reality, you know, could be Maxwell's theory, it could be fields, it could be anything? I don't care what the underlying theory is. Is there any theory that respects the following aspect of relativity, namely that influences do not propagate fast and speed of light. So he was assuming the structure of space time that's in relativity.

So I can find two regions that are space-like separated, meaning that there's no influences between them. Well, that's what you're stipulating. If you imagine that all influences have to propagate within light cones, then I can find two regions that are outside each other's light cone. I can use that for the Bell experiment, right? The settings on Alice's side are outside the backward light cone of Bob's outcome, and therefore, if I believe that influences subluminal, there can be no influences there.

So it's not the full apparatus of special relativity that's appealed to in Bell's argument, but rather just this aspect that says there is a fundamental speed limit, and therefore there can be two regions of space time that are not ordered causally. they're sort of causally disconnected. That's the only thing that Bell's really using in that argument. Okay, what's meant by influence? And is influence the same as causal influence? Yeah, I've been using influence to mean causal influence.

Is there such a thing as a non-causal influence? Like, help me understand what influence means. Influence, just a good word to talk about a cause-effect relation. So, you know, variable X, influences, variable Y. that means that, you know, X is a cause of Y. So far in a discussion, I've been trying to distinguish influence from signaling. So you could say, yeah, maybe here's the best example to see how they really come apart. They're not the same idea.

It's from cryptography. So, okay, let me get there in a moment. So, you know, what's the way in the field of causal inference that, causal influences are mathematically modeled. You can think of it as just its functional dependence. So the key is unscrambling this omelet of inference and influence. And so if I think of like projectile motion, right, I can write down a formula that relates, say, the maximum height of the cannonball, the initial velocity of the cannonball when it came out of the cannon and the angle of the cannon.

So I can write down a formula that relates them. and if you give me any two of those variables, I can infer what the value of the third is, right? So I can say, oh, you give me the maximum height and the angle, I'll figure out what the velocity must have been. So that's undergraduate physics. Allows me to write that equation and solve for any of the three variables.

But the only one of those equations that really describes a cause-effect relation is the one for the maximum height in terms of the velocity and the angle, right? Because if I modify the initial velocity or modify the angle, that will impact what the maximum height is. but if, you know, the canon has been shot and I go and I move it up a bit, well, that's not going to influence the canon.

So we all kind of intuitively understand the asymmetry of causation, right? The inference is symmetric. I can make inferences between lots of different variables, but influence, causal influences, is asymmetric. Another good example would be, like, if you think about the rooster crowing and the sun rising, there's a correlation there. So you might say, okay, well, what's the causal relation? well, if I make the rooster crow in the middle of the night,

it doesn't cause the sun to rise. But if I move the rooster eastward on the earth so that the sun rises later, it crows later, right? So I can figure out what the causals are based on interventions. But also our theories of physics teach us something about, you know, what are the causal relations, like in the cannonball example. So effectively, causal relations are described by functional dependencies,

but they're very specific functional dependencies, right? Because inferences can also be functional dependencies. There are functional dependencies where we imagine that if we, you know, say, say Y is a function of X, and were I to vary X, Y would vary, that's effectively a causal influence. Okay, so now, why is that distinct from signaling? So here's the example. It's from cryptography. So there's something called the one-time pad in cryptography, or it's sometimes called the Vernum cipher.

So say I have some plain text, some sequence of bits, zeros and ones. I want to communicate them to you. And we share a key, so some secret string of zeros and ones. And that key has been generated at random from all possible sequences. Then I can add the key to the plain text, send it out. That's the cipher text. When I add them up, mod two, send it out in the open. You get it.

You add the key, you get the plain text back. So in the simplest example, I'm sending you one bit, that's the plain text. The key is just one bit. I add them together, mod two. That's the cipher text. Now, the way this is designed is that the key is sampled at random from a uniform distribution. It's either zero or one with equal probability. And therefore, the ciphertext, if you say, well, what do you learn about the plain text?

Well, if the key was zero, then the cipher test is equal to the plain text. And if the key is one, it's just the flip of the plain text. And the eavesdropper gets the cipher text and says, my estimate of the probability of the plain text being zero is a half, and the probability that is one is a half. And so the eavesdropper has exactly zero information. It's just as if you were guessing. right? So information theoretically, they learn nothing about the plain text. That's the whole

purpose of the protocol. Now, if you ask about signals, right, so say I vary the plain text, and I say, well, what's the distribution over ciphertext that results from that variation? That the ciphertext is always uniformly distributed 0, or 1, because the key is uniformly distributed 0.1. So it's equally likely to be 0.1. And if I'm sending a different message, it's also just going to be the uniform distribution over 011. So that's an instance where there's no signal from the plain text to the cipher text. Changing the plain text is not going to change the distribution of the ciphertext at all. That's no signaling, right?

There's no change in the distribution. But of course, there's a causal influence. Because if there are no causal influence, there'd be no way that the recipient could decode the message. There has to be causal influence. So the vernon cipher shows you that there can be causal influences where you, you kind of wash out the influence. You don't get to see it because you've sort of added noise. So signaling is stronger than influence. It's like, oh, there's an influence. And furthermore, it's not being washed out by a noise source. But I can have influence with no signaling. So that's the

sense in which they're different concepts. And so in Bell's theorem, when you look at models like boamine mechanics and other inverable models that managed to reproduce Bell inequality violations using superluminal influences, they use exactly this trick that they sort of wash out. They say, we have influences, but we can't use them to signal. Why?

Because we're sort of washing out the influence by adding local noise. So it's the Vernum cipher trick happening in all these invariable models. That's how you make it consistent that there's no signaling and yet influence. And so that works fine, unless, of course, you want to stick to Leibnest's principle,

in which case, the fact that nothing you ever do can see that signal means that you should not be assuming that there's an influence there. Perfect. Okay. Okay. I want to talk about causation. Okay. Let's go to how physics is ordinarily said. I'm not talking about the equations on the board. Most of the time when people are talking about physics talking, they'll speak about physics in terms of causal accounts. I'll throw the ball and that will cause you to fall down to the ground when it hits your face or what have you. Then when they're drawing equations on the board, they're just equations with evolution. And then you could always ask the professor, where is causation here?

Like what variable is causal? What's going on? And then they may say, well, ultimately in physics, fundamentally, there is no causation. That causation is the story. And even in this other account of a projectile being thrown or the rooster crowing, Emily Adlam, who I spoke to a few months ago and I'll place the link on screen, my recollection of or my understanding of her theory is that there is no causation and you can also tell a story where the end is somehow constraining the beginning part. So these variables that you

mentioned with the maximum height and the angle and so forth, you could just give the maximum height and you don't have to talk about causal stories. Then when I heard you speak about causation, you said interventions. Now interventions to me imply agents. And so if one was to give a fundamental account of causation, it would seem to me that you're placing an agent or an intervention fundamentally in physics. And I just don't see that there, so I need help to understand. So I don't subscribe to the view that we need agents and the notion of intervention to define causation. There are some researchers who certainly do push for that view of causation that it's sort of define in terms of interventions. I would rather say that a good way to get a handle, a good way to get

evidence about what the true causal structure is with intervention. So if I can afford to do a randomized drug trial, that's a kind of intervention where I give somebody a drug or a placebo, not based on their preferences, but just based on the outcome of a coin flip. And then I can be sure that when I see a correlation with recovery, it's by virtue of the causal effectiveness of the drug, because there was no common cause to the outcome of the coin flip and their recovery, Unlike an example where, you know, health awareness or something might influence somebody's decision to take the treatment and influence their likelihood of recovering. Right. So interventions are a great way of getting a hold of what's going on causally, but I don't think it's part of the definition of causation.

I would say rather we have to reach for just, you know, the notion of counterfactuals that, you know, when I think about classical physics and I say, well, what does it mean to have the law of projectile motion? In what sense is it a law? It's not just an account of every, you know, cannonball flight that ever was in history. No, it's a law in the sense that it allows you to reason about counterfactuals, to make inferences. Had the velocity of the cannonball been different, this is what the maximum height would have been.

You know, it allows us to make all those inferences, right? So it basically stipulates a functional dependence of some effect variable on the variables that are the causes of that effect. Local news is in decline across Canada, and this is bad news for all of us. With less local news, noise, rumors, and misinformation fill the void, and it gets harder to separate truth from fiction. That's why CBC News is putting more journalists in more places across Canada,

reporting on the ground from where you live, telling the stories that matter to all of us, because local news is big news. choose news, not noise. CBC News. So yeah, I think generally we don't kind of formalize it. That's all kind of in the words that go along with things. So if I'm doing a problem on projectiles of cannonballs

and I get asked, you know, what happened if you went and you modified where the cannonball went by grabbing a hold of it and moving it, you know, would that impact the cannon? No, of course not. We know enough to understand what the cause-effect relations are in those circumstances. So I think that generically in our physical theories, there is part of the theory that tells us what the causal mechanisms are. I think it makes sense to imagine causes always proceeding from past to future, not in the other direction. So I don't see any reason to depart from that.

And yeah, so in other words, I think causation is a fundamental primitive of description of physical theories. And even though you're right that, you know, a lot of people, including some famous people, have said causation doesn't belong in physics. I disagree. It seems to me that it's a mistake to try to exclude notions of causation from physical theories. for me it's quite the opposite. So since the late 80s and 90s, this subfield of statistics called causal inference came along, and they've done quite a nice job of actually adding some mathematical formalism that allows you to talk about causation. And what's remarkable is how well that slots into puzzles in physics. So like in quantum theory in particular, I realize that the ability to talk formally about causal relations actually helps very much in sort of understanding

what's going on in physics. To me, it was a big mistake to try to say that causation is not a part of physics. I think it is, and I don't see any reason why we should go in for backward in time influences

or any kind of exotic notions of causation. Right now, it seems to me that I think we can get away with very conventional notions of what causation is, while still being required to innovate them a bit

to be able to cover the quantum case. So, in other words, like, you know, when I think about trying to explain the Bell experiment, I think I can get away with a common cause.

I don't need anything exotic in terms of what's the structure where you do need something exotic is that you're not going to be updating your knowledge by Bayesian probability theory. You're going to have a more exotic formalism for doing that sort of thing. You said you think you can get away

with a common cause? Why not you can? Why do you think it? So I don't have an interpretation of quantum theory on the table to tell everyone this is how you should think about quantum theory. It's a research program. So there's principles that I think I can secure in a realist interpretation of quantum theory, but we're not there yet.

So the situation is that I think there's a lot of evidence in favor of this being the right direction, but we're not at the end yet. We don't have a story that cleanly separates what part of the quantum form. is describing reality, how do we describe our knowledge of that reality? We don't have that. So I like to say that it's, that the existing interpretations of gone theory are sort of like a built house, but I look at them and I think they're all built on shaky foundations because they're all non-Libnizian in one way or another. What I've got is only the foundation, but I think it's a secure foundation and we're trying to build up the rest.

It's a great analogy. So my understanding of causal inference is that pearl? Yes. Pearl's work. Okay, so perhaps I didn't get deep enough into Pearl's work. Perhaps it just started out this way and maybe it's evolved. I remember there were due variables. But in fundamental physics, there is no do variable. Operationally, yes, sure. But it's a do seems to imply an agent, at least to me. So has the theory of causal inference moved beyond due to something more fundamental or something that doesn't require an agent? let me try to answer at a kind of more general level

and then we'll come back to this, which is, I would say, the definition of causation in the field of causal inference doesn't need to appeal to intervention. It doesn't need to appeal to these due operations. Like I was saying a moment ago, that I think you can define causation without referring to agents at all.

But nonetheless, when we're talking about, you know, what evidence do we have, what can I infer about what's going on causally, often we want to avail ourselves of the fact that I intervened and this is what I saw. And clearly there is a difference about the probability of recovery, given that I made someone take a drug or a placebo, versus the probability that they recovered merely after having observed that they took the drug, you know, and not in a randomized controlled trial, right?

Because in those two cases, I could have very different inferences. So it makes sense to talk about the agent-interested. terms of what you can learn. More broadly, I would say, you know, discussions of agents tend to leave many physicists feeling like, oh, this all feels a bit new age, like consciousness is important or something like that, and I don't go in for that sort of stuff. But I think that's really the wrong attitude. The way I would put is that there's a long tradition of pragmatism in physics. If I look at the second law of thermodynamics, you know, where did it come from? It came from Carnot studying heat engines and asking, you know, what's the most energy I can get out of a steam engine?

So that's a very pragmatic concern. It's about, you know, it's about achieving the most efficiency we can. And ultimately, that led to the second law thermodynamics, which is, you know, very fundamental in physics and thinking about the evolution of the universe and the evolution of life and places where there are no agents and no steam engines, you know, it has applications there. We came to it by asking questions about not, you know, what just happens if I set up these initial conditions and let it go, but rather what are the fundamental limits on what I can achieve, you know, for a certain type of thing, if I'm really trying to engineer it to do as well as possible. But that's also something that's constrained by the laws of physics, right? So, so like the limitation

on the speed of any influence could be thought of as a limitation on how quickly I can send signals. You know, limitations on how much work and I can extract from some heat baths tells me something about the laws of physics. Even though it's about these sort of parochial concerns of mine as to like how much energy I can get, it can still tell me something really important about the laws of physics. So similarly, when we start talking about a theory of inference, a theory of knowledge, you might say, oh, there's an agent there. But, you know, look at thermodynamics. So there was a time when we had very phenomenological thermodynamic laws and we didn't really understand them. And then statistical mechanics came along. So people like Boltzmann had the

kinetic theory of gases, and they recognize that we don't have knowledge of the microscopic configuration of all these particles. So what we're going to do is imagine a probability distribution over all the possible physical states, representing our ignorance of what's going on. And then we're going to ask, okay, how does that probability distribution evolve over time? And what does it allow me to infer about various macroscopic quantities like temperature volume pressure, things like that. But all along, there's an acknowledgement that we are ignorant and we have to model our ignorance quantitatively in physics to describe these situations. So it's not as if, you know, talking about quantum states as states of knowledge is something

highly innovative. It's been there in statemeck forever. We're always talking about probability distributions and our ignorance of the microscopic degrees of freedom. There's nothing about it that is sort of new agey or appeals to consciousness. It is just being, you know, quantitative about your ignorance, right? So using the formalism of Bayesian probability theory to talk about your ignorance and doing it quantitatively. So it's really the reverse, actually.

It's ironic because when you try to understand the quantum state as being something real, you often find that you have to go in for some idea like, oh, consciousness collapses the wave function and learning about something late influences the past, you know, like in a wheeler's delay choice experiment, that sort of thing. And it's only when you properly formalize, you know, what it would be for a quantum state to represent mere ignorance, that these things become very conventional. Like if I learn something in the future, I might

update my knowledge about the past. That's not exotic or new agey in any way. It doesn't involve consciousness. It's just I find some fossils in the ground and I infer something about dinosaurs in the past. I haven't caused dinosaurs to exist. I've updated my information about dinosaurs. So a lot of the odd things in quantum theory or things that people point to as mysterious, when you bring the agent in and you describe their ignorance,

they become much more innocuous. Times Arrow. Does that presuppose causation or does causation presuppose Times Arrow? Or are these independent? Yeah, I think the view that causation is a primitive notion in physics

basically presumes that there's an arrow to time, yes. It's the situation is as false. People often point to the fact, let's take Newton's theory as an example. Here's evidence that people will sometimes point to against time zero.

They say, look, if you give me the configuration at you know, time zero, then it will completely fix the configuration at some later time, just evolve, but similarly, if you give me the state at the final time, I can evolve it back and get the situation at time zero. So there's this time symmetry there,

and so there doesn't feel like any asymmetry, and so that means that, you know, why should I say that there's an asymmetry between the past and the future? To me, that feels more like a fact about inference. So just like inferring the existence of dinosaurs from fossils, in Newton's mechanics, if you tell me the configuration particles at one time, I can infer what the past was. And I can also infer the future from the past.

But if you ask me, if you imagine a change to the configuration at one time, what do you think that means? And there I think the intuition is strong that a change at this time implies a change in the future, but it does not imply a change in the past, right? Because that's the conventional way we think about causation. Certainly, like when we go through the world and we intervene, we understand that our interventions are going to have effects in the future and they're not going to have effects in the past.

So I like to think of the standard arguments for time symmetry as really being arguments for time symmetry of inference, right? Yeah, you know, the theory of inference doesn't care whether you're making inferences from the past to the future or the future of the past. but I think causation does care. And because I think it's a primitive notion, I think there is an arrow of time in our physical theories.

Since we're talking about the philosophy of physics, often philosophers of physics will say that many physicists don't have a conceptual understanding of what they're speaking about, whether it's GR, or most of the time it's quantum mechanics from my conversations on-air and off-air. So what does it mean when someone understands something conceptual, and how does one go about

if one wants to increase that muscle, how does one increase their conceptual understanding muscle? That's a really good question. Yeah, okay, so let's see. Certainly I would say, from the kind of quantum foundations perspective, increasing your conceptual understanding

is about, I think, understanding what's going on beyond just saying I can predict what's going to happen. So there's this kind of operationalist point of view, which is all you should ask of a theory is that it makes predictions. So I think that's a very poor image of science. Because if you gave somebody an Oracle

that would just answer yes, no questions, it would serve the same purpose. Like I could ask it a question and, you know, is this thing going to happen if I do this experiment, yes or no, it answers it? I get everything that an operationalist says a scientific theory should give me. But the problem is that's not really what I want out of a scientific theory.

So if I want to cure cancer, I don't even know what question to ask, right? I don't know where to start. I need to have a model of like, what is the, where do I start on this problem? And I can't just reduce that to answers to yes, no questions. It's because I have like a conceptual image of what's going on. And so in other words, scientific theories don't just answer questions. They provoke questions, right? They tell you what the next question you should be asking is.

So people who have different views on the interpretation of quantum theory tend to ask very different kinds of questions in their research. They tend to predict that we'll see, like if they happen to believe that there's going to be a breakdown in quantum theory at some point, they'll predict where that breakdown is going to happen very differently because they have very different conceptual understandings of what's going on, beyond what an operationalist would say. So I think, yeah, conceptual understanding, you know,

that part of that could be, you know, having a realist interpretation of what's going on. You know, often in a smaller context, like when you're looking at some phenomena and you sort of, you know, you follow the derivation and you say, okay, I guess I sort of understand where that result came from. But, you know, to really understand it,

like I'll often tell my students, like, if you can show another example of this phenomenon in a different context, but then you're on to something. Then, you know, you've got some, you've generalized the idea from the context in which you saw it to a broader context, that indicates some understanding.

So, yeah, often, like, generalization is a good proxy for a deeper understanding. Yeah, it's unification, situating this phenomena as cognate with this other phenomena or another example of something, that's also important for understanding. So I think there's a lot of signatures that, you know, you don't just follow the derivation,

you really understand what's going on. Something I want to do with this channel, and I hope I have been doing and I hope to do, is to have a positive influence on the research scene in the places that I, on the themes that I interview about. So whether it's biology with Mike 11 or developmental biology in that case, or the philosophy of physics, for instance, with you, Jacob Barnett, Tim Modlin, Sean Carroll, and so forth. And one of the ways to have a positive influence is to articulate something that hasn't been said much to the public. So the public has heard from people like Neil deGrasse Tyson that philosophy, maybe even the philosophy of physics, but let's just say philosophy is a sure, it was useful in the past, and sure it maybe gave rise to science in some sense. Natural philosophy became physics. But then at some point, it loses its ROI. And I'm sure you've heard similar, I'm sure you've had, even privately, perhaps even publicly, arguments about this.

Can you make the case for me? Make my life easier. Why should we not discount the philosophy of physics? What is it providing us? I would say maybe rather than focusing on the academic discipline, like the departments of philosophy that employ philosophers of physics, let's kind of focus on the culture of philosophy of physics. So there's people in physics departments who are doing philosophy of physics, right?

So often people studying the foundations of quantum mechanics are taking a much more philosophical approach to physics than typical physicists do. So I want to maybe defend that methodological stance, which is like let's ask the kinds of questions that are typical of natural philosophy. You know, why is that useful to do in physics? And I think the, again, like when we're talking about methodological questions, you know, it's not as if evidence data will solve this, but we can look to history and say, well, what sorts of approaches have been beneficial in physics?

And so if I look at the really big revolutions in physics, they all came along with, you know, they were instigated by people that were thinking very philosophically. And they often came along with radical changes in how we even think about science. So a lot of revolutions in physics also meant revolutions in the philosophy of science. Okay, so, you know, Newton and Einstein in particular with relativity and quantum theory. I think that how we even thought about what a scientific theory is changed as a result of what they were doing. And of course, they were very interested in philosophical questions.

So, you know, on the other side is when you look at some of the best progress, you know, in the 20th century and more recently, it was sort of technical sophistication that led to that progress and not thinking about, you know, deep questions about, you know, the nature of locality or relationalism or anything like that. And so that's sort of the kind of the bias we might have these days in physics department is that, you know, so, some of the recent progress, you didn't need to think deeply about philosophical issues to make progress. But what I like to tell students who maybe come with that attitude and say, oh, I don't know why foundations of quantum theory is so important, I would say, well, when you compare the magnitude of the revolutions, you ask, well, what are the revolutions that were like really significant? They're the ones that I think touched on deep philosophical issues. That's my sense of it, that, you know, revolutions in physics

that, you know, change what the nature of space and time are, you know, that philosophers have talked about the nature of space and time for a long time. That's a very philosophical issue.

I think they are more interesting, more significant, than mere technical innovations. So yeah, I guess I would just say to students that, you know, it's good to be philosophically savvy

if you really want to be part of revolutionary physics, right? You don't need to be philosophically savvy if you want to do good physics work, but revolutionary work in physics tends to require that kind of savvy. The other thing I would say, so that's sort of maybe one kind of answer. The other kind of answer is just that, you know, so I did a degree in philosophy. I did a joint degree in physics philosophy during my undergrad. And what I say, would say I took away from my philosophy degree is not, you know, the details of any particular philosophy, that's not what was important for me later, I would say in philosophy there's a certain

discipline about what constitutes a good argument. So you learn in a philosophy training, basically, if you're doing analytic philosophy, how to make an argument bulletproof. So how to define your terms carefully and how to explain things well and how to see the errors in your own analysis and how to go looking to criticize your own thinking. And yeah, there's a certain way of approaching argumentation, characteristic philosophy. And at the time, when I was doing both, I noticed that in physics, everything was much looser, that, you know, we're going to make these approximations, not worry too much. Let's try to get an answer. There was a lot less rigor about, you know, is this argument a good one? What are the premises that this argument ultimately rests on? You know,

can I articulate better the principles? Can I, you know, dispense with some of the, axioms that I'm using or the principles that I'm using this argument. There's a lot of things that in philosophy you tend to do to try to get to the essence of something. So you sort of learn to try to get to the essence of something. That, I think good physicists do that as well, but it's less a part of the training of a physicist and it's much more the part of the training of a philosophy. So I think there's also just some skill sets from philosophy that translate well to the foundation's physics. The scorebed app here with trust. So, it stats and real-time sports news.

Yeah, hey, who should I take in the Boston game? Well, statistically speaking. Nah, no more statistically speaking. I want hot takes. I want knee-jerk reactions. That's not really what I do. Is that because you don't have any knees? Or...

The score bet. Trusted sports content, seamless sports betting. Download today. 19 plus, Ontario only. If you have questions or concerns about your gambling or the gambling of someone close to you, please go to Connixonterio.ca. So you did your undergrad at McGill, correct?

Right, okay. I didn't even know they had, so I did mine from the University of Toronto, which you did later for your master's in PhD. Yes. Okay, I didn't even know McGill had a philosophy of physics undergrad. Well, no, they don't. So the situation was, so I started at McGill in physics, and even though I was sort of, you know, my loves were physics and philosophy at the time, there wasn't really that option. But I met somebody in my first year at McGill, and he had done most of a physics degree, but, you know, wanted to change course, and he'd become interest in philosophy. And he noticed that McGill offered a joint honors degree in philosophy in English, philosophy in political science, philosophy and linguistics.

Like philosophy gave you joint degrees with almost every other discipline, but there was no philosophy in physics. And so he went to the faculty of arts and science and said, why can't there be a joint honors degree in philosophy and physics. And so they went to the department of physics and said, will you relax your credit requirements to set up this joint degree? And the physics department said, absolutely not. There's no way we can relax any of these credits. And so then this guy went back and said, well, what if I did it in five years rather than four?

And the physics department said, yeah, that'd be okay. So they made this five-year program that was basically the physics curriculum, the whole physics curriculum, together with what was effectively a philosophy of minor. And so when I learned that that thing existed, because he basically created it, I was like, yeah, that's what I want to do. So I enrolled for that. Super interesting. Okay. Now you mentioned analytical or analytic philosophy. Was there anything from continental philosophy, which I'm sure you had to study some that influenced your physics? Not really. No, I didn't have to study much continental philosophy. I think I naturally gravitated to the analytic philosophy courses. because they were sort of more clearly just, yeah, useful and in line with how I was thinking. Also, there was a professor in the philosophy department.

His name was Paul Petroski. He was an amazing lecturer, and for whatever reason, he taught so many different courses while I was there. So I took his philosophy of logic and philosophy of language and epistemology and philosophy of mind and a long list of courses, and they were great. So yeah, I didn't actually have that much room for many more courses. But it was a great degree. I also got exposed for the first time to the foundations of quantum theory in a seminar course on metaphysics. So it was the philosophy department where that kind of ignited my interest in the foundations of quantum theory.

because the seminar course, basically we were reading primary material on things like the quantum measurement problem and Bell's theorem. And I remember reading these things. And then going back to my physics professors and saying, look, there's a real problem here. And they wouldn't acknowledge that there was a problem. And they told me that I was mistaken and I just didn't understand what was going on. So, yeah, the physics department really tried to persuade me not to work on these kinds of issues. but it was too late. It was already kind of hooked

by virtue of my exposure in the seminar course. What surprised you in the philosophy of language course? That was a long time ago. Yeah, I thought it was interesting, the whole kind of analysis of language turn

in analytic philosophy. Yeah, I mean, it had elements of kind of logic, you know, when you take an ordinary language English sentence and you try to decompose it into something that's more logically precise. You know, there exists a person such that that person is the king of France and they are bald. You know, like that, those kinds of exercises, you know, they're a small part of, you know, what I think is useful to do, which is, you know, often we'll have ideas and we'll try to

express them clearly and being able to sort of parse English sentences into things that are more precise and get clear on exactly what you're trying to say. You know, often the things we're trying to say, you know, often the things we're trying to say, in a paper on the foundation of physics are quite nuanced. And so you want to use mathematics. Mathematics is a great tool for saying things very precisely. Ordinary language is another great tool, especially if you want to be understood.

And in between, there's sort of a mix of some signposts of ordinary language and then some formalism that is maybe the most effective that's sort of really communicating. And so I think a lot of those courses and logic and philosophy of language helped me a little bit in terms of like, how do you make arguments understandable? Now, I want to understand Rob Moore. Okay.

And I want the viewers to understand Rob Moore. So now it would be a great time for you to take us through. What did you used to believe as an undergrad? How did that change as a grad? How did that change as you became a seasoned researcher? So take us through that timeline. Okay. All right.

So as an undergrad, uh, that's what I, first got exposed to foundations of quantum theory. I became interest in it. And I really wanted to work on that as a grad student, but there weren't really many opportunities to study the foundations of quantum theory. So I went to the University of Toronto for my master's in PhD, and I started out as a high-energy experimentalist for a short time. and then I realized it wasn't for me

and I looked into things I could do that were theory and there was a prof there by the name of John Sype who had a student who was working on decoherence theory and so it was sort of adjacent to foundational problems in quantum theory Sype? Yes, John Sype. Yeah, yeah, I remember him.

Did he also teach the combinatorics course or you have no idea? I don't think he taught a combinatorics course So he was in the physics department, and for many years, he taught a course on the interpretations of quantum theory. So he was interested in the foundations of quantum theory. And so I remember realizing that there's an opportunity here. I could work with John,

do the foundations of quantum theory, which is what I really wanted to do. But I knew that it could very well end my career in physics, because I knew there weren't really any real prospects. There weren't faculty jobs and the foundations of quantum theory. There might not be many postdoc opportunities either. And so I had to make a tough choice because I could have done something sort of more conventional

and sort of bided my time and when you have tenure, you can do what you like, versus do what I was interested in right then and risk ending my career. So in the end, I decided I would work on foundations of quantum theory. I sort of made my piece with the idea that this could very well end my career in physics. And, you know, mostly it was I thought I wouldn't want to spend, you know, the way I say is that you need a lot of motivation.

Success in theory and in physics often comes down to can you, you know, do the hard work. Because there's a lot of hard work involved in making progress. and you have to be motivated to do that hard work. So if you're working on something that you really care about, you're motivated to do that work. So I decided, yeah, I'm going to do Foundation's Guam Theory. And just as I was graduating, Perimeter Institute opened, and that's just down the road from Toronto and Waterloo.

And they were, you know, one of the founding fields, so there was only four fields represented at Permanitor in the early days. it was quantum gravity, fields and strings, quantum information, and foundations of quantum mechanics. So they were looking for postdocs who had expertise in the foundation of the quantum theory. And the situation was that not many people in the world had decided that they were going to risk their livelihood in physics on studying this stuff. So there wasn't a lot of competition, I would say, but it fit very well. So I could move into a postdoc position of perimeter. and that's, so I would say the place where I became,

so I worked with John on, we were, he gave a course with interpretations of quantum theory, so I had familiarity with all the different interpretations of quantum theory, and I was dissatisfied with all the existing interpretation. Now, a friend of mine, his name is Joseph Emerson, so we did undergrad together at McGill, he had gone to do a PhD with Leslie Ballanty, who's at Simon Fraser University,

in British Columbia. And Valentine famously has a textbook on quantum theory, and he also has some articles arguing for what he calls this, the ensemble interpretation of quantum theory. Anyways, it's very similar to this view that quantum states represent states of knowledge. So my friend Joe visited me in Toronto and his project with Valentine, what was really pointing out that a lot of features of quantum, quantum theory, make more sense if you make the analogy between quantum states and probability

distributions over classical states. So they were looking in particular at the Aaronfest limit, so the idea that, you know, I can recover a particle trajectory by the sort of expectation value of the position. And, you know, in situations where, let's say, you have a wave packet and it hits a barrier and it splits into two, the expectation value of the position goes right through the center of the barrier. And that kind of doesn't make sense in terms of like that particle picture. But if you think of that wave function as just analogous to a probability distribution, then you say, well, with some probability the particle goes left, with some probability the particle goes right, you have a very similar phenomenology for distributions over face-space.

They compare much better to wave functions than individual trajectories do. And they were studying these kinds of questions for chaotic dynamics, and he had a sort of body of evidence that we should think of quantum states in this way. And at the time, I wasn't persuaded, but I thought that if this is wrong, this evidence needs to be explained away somehow, right? Like, there has to be, if it's wrong to think that quantum states are epistemic, then there's some reason that these phenomena suggest it, but it's fundamentally wrong. And so then around the same time, so this is like late, late in my PhD, I started hearing talks by Christopher Fuchs at conferences and he is one of the main proponents of quantum basinism.

Well, now it's called cubism at the time. It was quantum basinism. And so he also was providing evidence for the correct interpretation of quantum states to be states of knowledge or states of belief. And so the evidence started mounting, so I felt like there was more things that, you know, I felt needed to be explained

if the quantum state wasn't a state of knowledge. But then I kind of got interested in just, trying to figure out for myself whether other phenomena could be explained with this view of quantum states. And suddenly I was able to add to the list of quantum phenomena that you could naturally explain in this way. And so I started changing my own view. So I had worked a lot on interpretations of quantum theory where the quantum state was describing something real. So things that were a bit like booming mechanics. They were called modal interpretations.

I'd worked on those. But yeah, my views were shifting. I was starting to see, you know, the evidence and it was compelling. And then what was very decisive was I went, I think it was 2002, I went to a conference in Oxford on the philosophy of physics. And I was talking to various people about, you know, isn't this interesting that there's all this phenomenology that can be reproduced if you take the view that quantum states are like probability distributions

and doesn't this need explanation? And what do you think? Like, how do we explain this? And everyone was very dismissive. And they said, no, that's just clearly wrong and it can't explain this. I can't explain that. And I came away very sort of frustrated at the dismissive attitude that people had. And so I just said, well, they say it can't explain this, that, and the other thing, but I think it can.

And so I remember I sat down after the conference in the airport to work out how to explain it. And on the flight back, I basically worked out what became the toy theory. Like all the, I started, I think, with the quantum teleportation protocol. I showed how you could explain that in one of these classical theories. So that was sort of the moment where I saw, oh, okay, this. really works. And it was very impactful in the sense that all I put in was this idea that, you know, there's a classical physics and your maximal knowledge is not complete, but everything else is just conventional classical physics. And out came all this phenomenology

that was, you know, recognizable as things that you see in quantum theory and quantum formation theory. And I hadn't put that in. I just like every time I would ask a question, like, oh, is there a no cloning theorem? The answer would come back. Yes, there's a no cloning theorem. And so, you know, once you've got a long list of things that you haven't put in and they're just coming out and they look very much like the quantum phenomena, you say, well, there's something right about the principles that went into this toy theory. And so for me, that was like this analogy between pure quantum states and states of incomplete knowledge. Like that struck me as the key innovation of this sort of approach relative to things like Bowman mechanics and other hidden variable models, which were not like that. So that sort of set me down this, you know, path of trying to pick out what aspects of the phenomenology really are surprising. They can't be explained in this way. So that's, you know, I was thinking about contextuality a lot, the non-locality, and trying to understand, well, what else do we have to vary?

What new principle do we need to really reproduce quantum theory? And, yeah, after coming to perimeter, so I did a postdoc at perimeter. that's when I wrote the toy theory work. I also worked on contextuality. And then I went to the UK. I had a Royal Society fellowship, and I spent three years at Cambridge in the quantum information group.

And yeah, I would say, for many years, I was struggling to find that kind of extra principle, some direction as to how to get a proper realist interpretation of quantum theory. And it was only when I came across the field of causal inference, and I started, you know, I had always been interested in sort of unscrambling the omelet of what's about ontology and what's about epistemology, that the toy theory

sort of pulls those things apart. But I started thinking about the ontology from this causal lens and the epistemology as a theory of inference. And I, yeah, eventually I realized, okay, I think this new framework that comes from the causal interest community is really the right way of thinking about quantum theory, and it's an innovation in the language of this framework that I'm really looking for. So that's sort of, I mean, there's maybe a lot of other details of the story, but in terms of my views on quantum theory, that's how I got to where I am today. What's it like being at the faculty at the perimeter? Tell me about the culture, the philosophy of physics culture, and how it's viewed, how you are viewed, and how the collaborations are like, and tell me about it. Yeah, it's very,

collegial. We don't have philosophers here per se. We have physicists. I would say the thing I appreciate the most about being at perimeter is that it started with very much the vibe of a startup, very ambitious, and from the very beginning, the idea was that we want breakthroughs in physics, we want to revolutionize physics. And in particular, Lee Smollin, created this culture. We spoke to Lee, especially in the early days. That's our objective. We need to work on the big problems.

We need to revolutionize physics. And I think having that as your mission is very good because there's a temptation always to sort of do the thing that you can do because you can get some result. And you might drift away from the really significant question. but if there's this constant reminder that no you're you're supposed to be working on the really tough problems then you're more disciplined you might still work on smaller problems but it's all in the service of this bigger goal so i definitely feel that you know that so i would say my own research program is as ambitious as it is

probably in part because i'm here and that was kind of the DNA of this place is that we should always be ambitious. So I appreciate that a lot. It should also be said that because Permanitor supported the foundations of quantum theory from the very beginning, which was the year 2000, you know, it was one of the very few places in the world that was doing that. So, you know, back then, you didn't find many groups in the world for working on the foundation of the quantum theory. And as a result, you know, over the years, it's been a great place to support. support the foundations of physics, foundations of quantum theory. We've had, you know, as visitors, all kind of representatives, all different views come through. So you get to talk to the top

people in the field on a regular basis because they come through. But you can also attract really talented postdocs and students who are interested to come. So you have a really high talent density, which is obviously, you know, any individual researcher benefits a lot from having really smart people around who have thought deeply about these things. So, yeah, I'm very appreciative of that aspect of Perimeter. As I said, I mean, the groups, the different groups of Primiter are very collegial. You know, we sometimes collaborate together, especially like the closest groups to quantum foundations would be the quantum information group and the quantum gravity group.

So, yeah, we have the strongest interactions with those two. groups, that works out very well. Yeah, I would say it's a really good place to do the foundations of quantum mechanics. It's one of the bigger groups in the world, right? So, I mean, there are some that are bigger, but it's big enough that generally we kind of have a sense of sort of everything that's going on in the world and quantum foundations. If there's some new result, somebody here will inform themselves about it, we'll have a talk about it. So we can sort of be up to speed

on what's happening in the field. And so it's nice to have that overview of the field. It's, you know, rather than becoming very focused on, you know, your own particular thing, which can happen if you're in a smaller group, you get a sense of everything that's going on and what are the big trends and the ideas. And so, yeah, it's a really nice place to,

if one is interested in the foundation of quantum theories, to sort of see what's out there, understand what's going on. And what makes it different than other physics departments or physics institutes, the Perimeter Institute? Well, there are no labs, so it's a theoretical physics institute, so we don't have experimentalists at Perimeter. Some of us do work with experimentalists,

so I have an experimentalist colleague at the University of Waterloo, which is just down the street. And I work with his group doing quantum foundations experiments. But, you know, the founding, you know, the institute was set up by Mike Lazaridis. He gave most of the money to set up the institute. Nowadays, it's funded partially by the federal government, partially by the provincial government, and partially by private donors. I can give you some context behind my question. Just so you understand my mind frame.

So Mike, who you referenced, Mike Lazaridis, he invited me recently a few months ago to his home. Beautiful. I don't know if you've ever been, but it's the most beautiful home that I've ever seen. And he sent this private driver to come pick me up all the way from Toronto to this place. And I was speaking to the driver. And the driver said, he's like, bro, you have to see this place. You've never been. You have to see this place.

It's the, Eddie Hale, he just kept going on and on at saying that, oh, bro, I've driven for Drake. I've driven for this, but this guy's home, he didn't even know the guy's name. It's just, this guy's home tops it. And so I was so hyped. Anyhow, I spoke to, or Mike wanted to ask me about a variety of topics, my opinions on them. Something that became clear to me was that many of the faculty at perimeter have expressed to Mike that there's been a change in the culture over the past 20 years. and I just wanted to know if you've seen a change in how,

I don't know, in how the foundational research is done and how, well, I just want to know if you've seen a change either the positive, neutral, I just want to know, what is it like? Well, I think that maybe the biggest change over the years is that Permanent Institute, by virtue of being very successful, has grown a lot, and any institution necessarily kind of changes as it gets much bigger. So in the early days, we didn't need a lot of rules and policies and things like that because it was very small and everything could be decided on a kind of one-off case.

So it really had that feeling of a startup. And with very few people and a large endowment, you have a lot of flexibility in terms of what you want to do. And then you become mature and now you're doing a lot more science. but it also comes with you know challenges just in terms of administering a large institute so I would say yeah maybe the most noticeable changes

are just the kind of standard changes that come from any institute that grows from something small to something much larger I still feel like we haven't lost our DNA of you know we're here to make breakthroughs and we have to stay

focused on that. So we've certainly expanded the number of fields of theoretical physics. So as I mentioned earlier, you know, it started out with just quantum foundations, quantum information, quantum gravity, and quantum fields and strings.

And since then, you know, we now have a condensed matter physics group, a particle physics group, a cosmology group, a strong gravity group, a mathematical physics group. Right? So it's expanded in terms of the scope of areas of physics. which is good because, you know, there's connections between all these different areas.

But, yeah, it also means that, you know, in the early days, if you wanted to, you could go to every seminar, and there's no way anyone could attend every seminar anymore. So you have to pick and choose who you're going to interact with. So it's a good problem to have. Boarding for flight 246 to Toronto is delayed 50 minutes. Oh, what? Sounds like Ojo time. Play Ojo? Great idea.

Feel the fun with all the latest slots in live casino games and with no wagering requirements. What you win is yours to keep groovy. Hey, I won! Feel the fun. Morning will begin when passenger Fisher is done celebrating. 19 plus Ontario only. Please play responsibly. Concerned by your gambling or that if someone close, you call 186653310 or visitcomnetontera.ca. Is there a piece of advice that you consistently give your students?

Hmm. Yes, I mean there's many. Maybe on the side of being understood, I think it's a particular challenge on foundations. Because many researchers come with very different ideas, it's sometimes hard to communicate effectively. So it's really important if you want to get your idea across to learn how to communicate effectively.

And so it's, it's, I tell them, you know, it's, it's all about the narrative, right? That you need to have a strong narrative. So if you're giving a talk, you need to, like, make it really clear what you want people to come away with. So, or when we're writing papers. So, so there's a, can you give an example of that, an example of how that's done poorly and then how you would sharpen it, tighten it, improve it? Yeah. So, like, maybe something that would be done poorly is, like, if, you know, if, you would, If I'm making an argument for something and I just sort of jump into, here's a fact, here's another fact, together they imply this and they imply that, and we arrive at this result. That's worse than, you know, here's what we're interested in. And here's a result. Here's a theorem. Now I'm going to tell you everything you need to understand what this theorem says. I'm going to give you the definitions of the term so that you understand the content of the theorem before I try to persuade you that it's true, right? Before I try to give you the proof. And often,

in talks, you know, trying to give the proof is definitely the wrong thing to do because the audience won't have the time to absorb it. But spending the time so that they understand what the result is saying, what the definitions of the terms on. That's really what you want to focus on. So you want people to be able to come away and say, I understood what the claim was, you know, they claim to have proven this theorem. I have to go to the proof to make sure that it's right, but I understand what they're claiming. So things like that where it's about what you should be emphasizing, what you should be spending your time on, you know, spend more time on motivating it because people don't see why what you're answering is interesting.

They're not going to pay attention. So you've got to spend a lot of time motivating why you're trying to answer this question. That's extremely important and you don't want to try to contract that. So advice, like, I feel like that's the kind of advice I feel comfortable giving a very kind of small-scale advice about how to present well, how to write well. I actually give a talk to the grad students on how to give a good talk, and I've given talks on how to write a good paper. So this is something I do think a lot about,

like what are the tricks you can use to write well, to present well. I also sometimes offer advice about how to do physics, but there, you know, it's harder because different people with different strengths are going to make progress in different ways. So there isn't necessarily sort of like one kind of root to making progress. People who have different styles of research, that can be very useful.

So you don't necessarily want to force somebody to do research in any particular style. Are you often misunderstood? I think so, yes. Yeah, it's surprisingly so, that it is a frustration in a way that I spend a lot of time and energy, for example, writing papers in a way

that I hope that this cannot be misunderstood and giving talks where I hope I'm sort of clear about what I'm arguing for. But yeah, I will often learn that people don't understand what I'm arguing. I've got to summarize my views in a way

that's surprising to me because it's sort of the opposite of what I believe. So yeah, I'll often, in trying to, like the students, for example, come back to the students, like they will complain about how long it takes to write a paper with me

because I can always, we see opportunities for improvements and places where we could be clearer. And I often argue that, you know, in spite of putting in so much effort to be clear and going over this text many, many times till it feels like it could not possibly be misunderstood, nonetheless, it will be. So that's one of the models I like to say is that don't write so that it's possible to be understood, right so that it's impossible to be misunderstood, that it has to be totally clear what you're saying.

And even when you set that as your goal, you're still going to be misunderstood. So if you don't set that as your goal, you'll definitely be misunderstood. I see. So why don't you give an example? How about this? What is it that Rob is most misunderstood

about that frustrates him the most, like the intersection of those two? Okay. Yeah. So I think if, So a lot of my articles are set in what here I've called the sort of conventional framework for realism or the ontological models framework, a kind of classical view of how to be a realist. And I, you know, so my toy theory is set in that framework.

And I derive kind of all the things you can do and the limitations, like what you can't do. And I do that because I'm studying it as a foil. I hope that it'll teach me something about quantum theory. and from the very beginning I've noted that I'm not proposing this as the correct interpretation of quantum mechanics this is just meant to be an argument

in favor of a particular interpretation of the quantum state but the correct interpretation of quantum theory is not going to be this sort of thing in fact I think that we have to abandon this framework and come up with a new framework but that's a sort of a subtle point

that you might study something that you know to be wrong to learn something about where you're trying to go. So, you know, the analogy I like to use is, if you look at Picasso's early paintings when he was 15 years old, he painted in the classical style. He could paint like a master at age 15.

But I don't think Picasso could have sort of transcended the norms of art if he hadn't mastered the orthodox approach, right? I think a lot of what he did required a mastery of the orthodox approach. So a lot of what I do is like, look, hidden variable models weren't really we don't fully understand what they're capable of reproducing

if we take hidden variable models where we can have some limit on how much we know there's a lot of phenomenology we can reproduce and we should study that to know what the limits of what we can understand are because that will tell us anything that's beyond those limits is what is really surprising and needs to be understood.

So I kind of study these as a foil as something that will prepare me for the more radical step that will get me to a proper realist interpretive quantum theory. But often people think I just endorse this framework, that what Rob wants is a ontological model where the quantum state is epistemic, and that's not what I want. I want to reject the framework of ontological models altogether.

So yeah, that's one of the things that I feel like I can never, no matter what I say, people will think that I'm endorsing this ontological models framework. I very much hope I did not misinterpret you at all during this conversation. I think I've been studying you for four weeks, your work, and then also other work that is related to yours, and dates stick out to me. So I know 1994, Roorlich, I believe, Roarlich came up with a non-local correlation bound larger than Bell's correlations, and then that the PBR theorem is supposedly excluding epistemic interpretations, but you have your responses to that. So it's difficult for me to formulate a question

that isn't loaded with a misunderstanding. A loaded question may have a misunderstanding. So I try to ask them more, okay, well, what is the PBR theorem? And tell me what you think about it, rather than me saying, look, the PBR theorem excludes your theory because of reasons A and B, even though you've said C about it. That's false because of D. Yeah. In fact, Rob, part of the reason that I do this podcast is so I can understand or ensure that I'm understanding the person that I'm speaking to, the interlocutor's theory, by recapitulating it, and then hopefully they agree. They say, yes, that's what I meant. That's what I said. And then it's only then that I realized I didn't misconstrue them. Otherwise, it's super easy for me to rapidly misunderstand something,

and I'd rather slowly understand something. So anyhow, I hope I haven't misunderstood you, or misrepresented you. What do I say about the PBR theorem? So the way the PBRR theorem is often understood and the way it's maybe the popular understanding of that is that shows that ontological models that are

what I call sci epistemic, so I can come back to define that in a moment, are ruled out. They cannot reproduce predictions of quantum theory. So, okay, what does it mean to be sci epistemic? So after the, I showed this toy theory, I did further work and you can find sub-theories of quantum theory. So you can find, so what do I mean by a sub-their of quantum theory? If I take quantum theory and I say, let me just consider some subset of the preparations

and the measurements and the transformations that are allowed by the full theory, I can get sub-theories, right? So they have a lot of features in common with quantum theory, but they just don't span the full set of states and measurements. measurements and transformations. And so there's these certain sub-theories, which I like to call Clifford sub-theories, because the transformations that are reversible form the Clifford group. And they get studied a lot in quantum formation theory,

and there's continuous variable versions of them. But anyways, it's a part of quantum theory, and you can show that these things admit of Leibnizian ontological models. So they're local, they're non-contextual. You can often define a Vigner representation of the states and measurements, and the Vignor representation gives you

a probabilistic interpretation on a classical face base for everything in these theories. So it's another angle on how do you understand what part of quantum theory is sort of easy to explain from a classical point of view, while these Clifford sub-theories don't have anything truly innovative about them. Now, when you think about those kinds of models,

there's still, in a way, hidden variable models because you're imagining that there's some ontology and some probability distributions over it, but they're very different from something like Bowman, mechanics or any other pilot wave theory. And so the way I tried to articulate that difference in a paper was by defining a dichotomy for ontological models between sciontic and sci epistemic. So, boamine mechanics is an example of a sciontic model. These classical theories of the Clifford sub-theories of quantum mechanics are examples of sci-epistemic.

systemic models. So what's the definition? And why this terminology? So it comes down to the following. If I think of a variation of the quantum state in the full theory or in one of these sub-theories, does imagining a variation in the quantum state forced me to imagine that the real physical state of the world has changed, or might it be consistent with nearly a change in my knowledge of the world? So if, for example, the quantum state, it's like if I think about the physical state space, like in Bowman mechanics, the quantum state, the universal wave function is sort of part of the physical state. And then there's these particle configurations. So if I say, well, what do I know when I know the quantum state?

Well, I have a kind of delta distribution on the axis, which runs overall quantum states, and then some non-delta distributions along the other axes. But those distributions don't overlap. As I vary the quantum state, I move to a completely disjoint distribution, right? So there's no overlap between those distributions. Whereas the kinds of models I was telling you about earlier, where one quantum state gets represented by a distribution like this, and our non-orthogonal quantum state gets represented by a distribution that overlaps with it. So now I can imagine a variation from this quantum state to this quantum state,

but the physical state has not changed. It's in that region of overlap. But that kind of model means I can imagine variations in quantum state that don't necessitate a variation in the real physical state. So those models are si epistemic. Okay, so that quantum state must be interpreted as a state of incomplete knowledge in those models. So that's the dichotomy. Among the sci-ontic models, there are some that are si-complete, which means there's only the quantum state, there's nothing else to say.

So that's like some orthodox interpretations of quantum mechanics that do not introduce any hidden variables are basically those si-complete models. And then there's some that I call si-supplemented, which is like, oh, the quantum state is real, but there's other stuff as well. So BOMU mechanics is an example where there's these particle configurations that are real as well. But being Psi incomplete, you know, sye not being a complete description, that can happen with si being ontic or epistemic, right? One way in which side could be incomplete is if it's merely describing probability distributions and then there's some deeper reality below it. So roughly speaking, the old debate, you know, the one that goes back to Einstein, Podolsky, and Rosen is the quantum state complete or incomplete? That was the language they used in that paper.

but this newer debate is is it ontic or is it epistemic so the toy theories and these classical theories where there's a restriction on how much you can know to my mind the fact that they can reproduce so much of the phenomenology of quantum theory

gives evidence that it is better to think of the quantum state as being epistemic but strictly speaking this dichotomy I just defined for you was all in this framework of ontological models So, you know, I want to reject that framework, but I still think it's useful to study, you know, sci-a-epistemic ontological models. What can they explain? So that's the sort of spirit of all this, which is like, ultimately, I don't think any of these models are the right picture of reality.

But by studying them, like kind of mastering what's possible, we can learn something about, you know, how to reject the ontological models framework and move on. So I think, you know, Bell's theorem and the Cotian Specker theorem, which are the two big no-go theorems in quantum theory, in this language, what they show is, it doesn't matter if you're sci-autic or sci-epistemic. If you believe in local causality, you can't reproduce the quantum predictions. If you believe in non-contextuality,

you can't reproduce the quantum predictions. Being sci-epistemic is not going to save you, right? It's just can't be done. And because I think that the Leibnizian principle underlies local causality and non-citectuality, to me, those no-go theorems are just saying, take the conventional framework of ontological models, and the Leibnizian principle,

you're going to get a contradiction with quantum mechanics, so you should give up the ontological model's framework, not Leibnizian principle. So to my mind, those are the good no-go theorems, because they teach us something about why this framework is failing. So that's kind of the key bit, which is like, why do I study these hidden variable models that I don't believe in?

Because I want to know precisely where they fail. If I have wrong ideas about the failings of the ontological model's framework, like if I have some no-go theorems that are just, you know, that the logic is wrong, or their assumptions are unnatural, then that will be a poor guide to sort of being a revolutionary and replacing this framework with something better.

I need to know what is it about the phenomenology of quantum theory that really resists explanation? So that's why I studied. Okay. But, okay, so coming to PBR now, so the PBR theorem is a theorem that says we make certain assumptions at, so I can get into what they are,

together with the idea that the quantum state is epistemic, derive a contradiction with quantum predictions. And so the conclusion that is drawn is that we don't have a sci-epistemic ontological model. We have to have a sci-ontic ontological model. So the first thing I say in response to this is, I kind of don't care because I am not a proponent of sci-epistemic ontological models.

I want to get rid of the ontological models framework. And so, you know, a theorem that says you can't have that, I already knew that. Bell's theorem already tells me, I can't have my Leibnizian principle even if I take Sy to be epistemic, right? So Cotian Specker and Bell have already taught me this. But nonetheless, in the spirit of,

it's good to have no-go theorems to guide the way, you can ask, is the PBR theorem something that is likely to teach us how to go beyond the ontological models framework? Is it likely to be something that will help me in my research program? And so that's where my kind of criticism of the details of the argument come in. Okay, so we have to get into the details

of what is the PPR at the arm. So it makes two key assumptions. I'm going to sort of give two passes through what those assumptions are. The first pass, well, you'll see why I'm going to go through it twice. But the first assumption in the popular understanding is what I will call ontic separability.

So it's the idea that if I have a composite system, A and B, and I ask, what is the ontic state space of that composite? So an ontic state is just a description of all the properties that the system might have. What is the ontic state space of the composite in terms of the ontic state spaces of the components? And the answer is it's just the Cartesian product, which means if I want to tell you everything about the properties in the composite system, it's sufficient for me to tell you all the properties of A and all the properties about B, and that's it. So in other words, it's kind of a commitment to there being no holistic properties.

It's a reductionist model where the composite, all of its properties come from the properties of its components. Okay. So that's acceptability. And then the second assumption I like to call like independence preservation, right? So I have, at the operational level, I might come up to system A and do a preparation procedure on it and come out to B and do some independent preparation procedure. So I imagine I can vary how I prepare A independently of how I'm preparing B. and in quantum theory, we model those situations by quantum states that factorize, right? So there's a density operator for A and a density operator for B, and I can vary them independently.

And the idea of independence preservation is that in one of these ontological models, these quantum states are represented by probability distributions that factorize across these two ontic states bases. So in other words, there's no correlation between the properties of A and B when the preparations of this type. Okay. So I think that those are very reasonable assumptions. And in particular, this independence preservation property actually follows from Leibnizium principles, right?

So I can give an argument that says, look, if it weren't the case, if there were correlations between the properties of A and B, even though as I modify the preparation on A, no measurement I do on B can tell that change, well, that would be a failure of Leibniz because it would be a ontological distinction. I could learn something about the ontic states of B by, you know, changing or learning about A, but there would be no empirical consequences to those differences. So that would be compromising Leibniz. So if, you know, my commitment to Leibnett says, yes, independence preservation is a great assumption.

And ontic separability also strikes me as a totally natural assumption. And it's true that you can take the structure of the PBR argument and show that those assumptions, together with sci epistemicity, the fact that implies a contradiction. So I have no problem with that. The problem I see is that it also rules out sciontic models. So the way I've just phrased those assumptions,

I can say that sciontic ontological models are also inconsistent with the quantum predictions. And I can do it very trivially because if I look at an entangled state and I ask, does it satisfy the principle, is the principle

of ontic separability satisfied if I think quantum states are real and I grant that there's entangled states in the world, no, an entangled quantum state

cannot be understood as being just a product quantum state. It is generally understood by proponents of the sciontic view as invoking some holistic properties

of the pair of particles. So if ontic separability is saying no holistic properties, well, you know, you can't, to be sciontic and hold on to ontic separability. So the way I think of it is, ontic separability is the assumption, then here's the easy no-go theorem. You can't have a sci-ontic ontological model. That's trivial. It just follows from the existence of entangle state.

And then there's a hard no-go theorem, which is the one that uses the PBR construction. But at the end of the day, if these are your assumptions, then what you show is that just all ontological models, whether they're sci-ontic or sci-epistemic, cannot do justice to the quantum predictions. And so I like to think of that as the rehabilitated PBR theorem. It's It's like Bell, it's like Cotian Specker, and in my view, the right lesson is that you have to give up on the framework of ontological models. Okay, but in fact, the assumption that they make in the papers is not ontexapability precisely. I think it is what most people think they're assuming.

Most people who look at the PBR argument, I think believe that what they're assuming is what I've called ontexceptrability. but in fact what they need to assume to get their conclusion, right? So what they're trying to show is that si-epistemic ontological models are ruled out, but sci-ontic ontological models are not ruled out. They're still ruled in. To get that, you need to make a much more refined nuanced assumption, which is that I allow holistic properties in the support of distributions that represent entangled states,

and I only disallow holistic properties in the supports of distributions that correspond to product column states. Okay, so that's the assumption they need to get their conclusion. I call that the entanglement holism link, right? So when you have an entangled state, then there are holistic properties in that circumstance. Now, the thing is that most people look at that refined, so it's usually not clear that that's the assumption, but even if that becomes clear, I think most people look at that

and say, yeah, that's something I believe, so that's all good. But in fact, it's only plausible. if you already believe that the quantum state is a state of reality. So if you think that a quantum state is ontic, then you think that an entangled state is a strange way of being. It's a strange way for A and B to exist. It's a strange properties of A and B, and there's some sort of connection between them

and some sort of holistic properties. But with these sci-epistemic ontological models that I've worked on, the way they represent entang state, it's just a correlated probability distribution over ontic separable state spaces. So every entangled state just means, oh, if I update my knowledge about A, I'll learn something about B,

but there are no holistic properties anywhere. So from the perspective of a sci-epistemic ontological model, somebody who comes from that position, and you ask them, do you want to take on board this assumption that's natural for your opponents? The answer is no, it's not natural for a sci-epistemic perspective. So it makes the argument kind of circular to say,

oh, well, if you grant that this is natural, which you'll only do if you already believe the quantum state is real. Well, then we have an argument that establishes that the quantum state's right. Right. So my challenge to those who want to use the PBRR theorem as an argument in favor of sciontic ontological models is, you know, please provide a motivation for this entanglement holism link that doesn't just rely on your sciatic intuitions. You know, it isn't just a reaffirmation of your beliefs that an entangled state is a funny way of being that includes holistic properties. can you

motivate it in terms that will be appealing to somebody who believes in a sci-epistemic ontological model

to begin with where an entangled state is just a way of knowing about a pair of properties, a pair of systems.

And that's the challenge that hasn't been met. So yeah, my, you know, the main response I have is sort of like it's about

ontological models and I don't think we need to keep these. But I also don't think it's like Bell's Theorem or Corson-Specker

insofar as it it's not making natural assumptions and ruling out only sci-epistemic models. The natural assumptions, if you make them, rules out both sci-epistemic and sciotic models. And so in that regard, it's like Bell, it's like Caution Specker. It's another particular proof that we're going to have to go beyond the ontological models framework. So my understanding is in 2014 or so Leifer, if I'm pronouncing that correctly, as I only read it, lifer or leifer. Right, great. So that these sci-epistemic models are exponentially not good at explaining distinguishability.

So how do you respond to that? Well, yeah. So again, it's a question of like the assumptions that go into these theorems, are they natural assumptions? So there's many no-go theorems that try to make an assumption about a particular way of explaining the indistinguishability of certain non-orthogonal quantum states, for example. And there are some assumptions. So there's a particular assumption called maximal sci epistemicity that says that when two distributions are indistinguishable,

it should be entirely explained by kind of the magnitude of their overlap. So if you tell me how much overlap two distributions have, then sort of classical probability theory will impose a bound on how well you could possibly discriminate them. and I think that is a reasonable assumption precisely because you can derive it from the Leibnizian principle. It's an instance of non-contextextuality. And so for me, if you think that's why it's a good assumption,

then really you should assume the deeper principle, the Leibniz's principle. And there's no way to salvage a sciantic model if you assume that principle. We know we can rule it out with Bell's theorem and the quotient spectrum. So the game becomes, can you give me any reason for believing this, you know,

assumption about how to discriminate states that isn't just an appeal to Leibniz, because if it's an appeal to Leibniz, then, you know, I can't salvage sciontic ontological models. So again, the challenge is like, right, you've got this assumption. Please give me a motivation for this assumption that doesn't just boil down to appealing to Leibniz's principle. So, I mean, there's a lot of details I've left out. But that's essentially the state of things. that, you know,

here's a kind of subtle point, which is that it's often thought that if you derive a no-go theorem from logically weaker assumptions, that somehow that's better. Like, okay, two Nugo theorems. One has some assumption X, and here's another one,

and it uses assumption Y, that's logically weaker than X. And so the claim would be like, okay, well, that's somehow impressive. But I think that's just a mistake. And the example I use is that, suppose assumption X is, you know, Newton's laws hold. And assumption Y is Newton's laws hold on Tuesdays,

and I'm not going to tell you anything about whether it holds or not the rest of the days of the week. Well, that's logically weaker. It commits me to less. It's not physically more plausible. If anything, it's physically less plausible. So its logical weakness has not added to its physical plausibility at all, if you see what I mean.

Yes, interesting. And so the mere fact that you can set down an assumption that's implied by Leibniz's principle but doesn't imply Leibniz's principle is not a reason for thinking that assumption is more physically plausible. It's merely logically weaker. That doesn't make it more physically plausible. To me, the success of Leibniz's principle

in the work of Einstein is what makes it a good principle. and if you can derive some particular assumption like local causality or non-contactuality from it, those are plausible as well. But if you have something and you want to assume only that consequence but not the deeper principle, I'm going to ask,

well, motivate that for me, please. Is that like Newton's laws on Tuesdays with no commitment for the rest of the week? Because if so, I don't see the physical plausibility. It's only logically weaker and that's not an argument in favor of it. So in your mind, in your framework,

your model, whatever you want to call it. If the wave function is representing a knowledge, our knowledge about what? Yeah, that's a good question. That's the right question to be asking. And unfortunately, the situation is that, you know, this is a research program, and I would say we don't have a good answer to that question.

But that doesn't mean that, you know, we don't have evidence that a quantum state might be epistemic. So my argument is that you can find facts about the phenomenology of quantum states that lends evidence to them being epistemic without necessarily be able to answer the question of exactly how do I formulate quantum theory in a way that sort of unscrambles what's reality and what's knowledge of reality and then tell me exactly what that reality is. I mean, I have ideas about that, but they're speculative. So that's, the kind of tricky thing. Does that even make sense to say that you can argue for the kind of status of something without having a concrete model of what it means? So let me try to explain that

by an analogy. So my claim about, you know, why does quantum theory appear so mysterious for so long, I think it's mostly due to a category mistake, and that category mistake is thinking that a quantum state describes reality when in fact it describes knowledge of reality. So the analogy I want to draw is to my favorite example of a category mistake, prior category mistake,

that kind of held back progress. And this is the example of the decipherment of Egyptian hieroglyphs. Yes, I've heard this. Yeah. I like it, yeah, please. So hieroglyphs, you know, the last person who could, you know, write and read hieroglyphs probably died around 500 AD.

And then, you know, 1,400 years later, people were trying to decipher it, you know, trying to say, well, what did these inscriptions mean? And it was, you know, long forgotten how to make sense of it. And the prevailing idea at the time was that each glyph was an ideogram, which means it represents a concept directly, not through the medium of some spoken language. An alternative is that the glyph is phonetic. It represents the sound in some spoken language, so it might represent a vowel or a consonant or a syllable.

But the prevailing idea was that it's an ideogram. So it's like a sign at the airport, where no matter what language you speak, it'll be clear what this means. And so people were trying to decipher high-glysses under that ideographic idea. idea of what kind of category a glyph was. And people wrote books about what all these inscriptions actually said under various

interpretations. And in retrospect, you know, so Champolillon famously deciphered hieroglyphs. One of the big moments was sort of the Rosetta Stone being discovered. I think that was 1799. And so a few years later, we got the decipherment. And in retrospect, it was phonetic, right? The glycerophonetic. So the ideographic interpretation was just a category mistake.

And all the kind of interpretations ended up just being nothing to do with what the real meaning was. And it's quite surprising that it's phonetic. It looks like it's no to you? Well, I mean, surprising maybe because we use an alphabet where none of the letters look like anything anymore. But if, you know, if you ever do those puzzles where it's... It's like, you know, a picture of a bee and a picture of a leaf, and it spells out belief. You know, we could do writing with, you know, symbols that represent sounds as well. It's just that, you know, in the early days of writing, because the Egyptian Hargloose was the very first writing systems,

that was kind of the natural way of representing sounds. And over time, they evolved into things that were more abstract. So from a modern perspective, it's a bit odd. I'm a time-saving person. There's so much energy that has to go into creating an image to represent a B sound. or Lee or that's just, it's just, I'd rather just not write. Yeah, yeah. I'll just communicate with my voice.

Yeah, I think we're accustomed to being able to write quickly, but that was an innovation that happened, you know, thousands of years after the invention of writing systems. So in any case, the, if you look at the story of decipherment, you know what, I'm trying to remember where, I'm coming back to what point again, where was your question? Well, I want to know what is it about, what is the si-epistemic, if it's epistemic, it's about what? Yes, and then you were going there.

Right. Thank you. Good. Okay, so yeah, the analogy is going to try to answer that question. So the way the story goes, after the discovery of the Rosetta Stone, right? So the Rosetta Stone had an inscription that was in sort of three different scripts. One was Greek, which we knew how to read. What was Demotic, which was a particular version of the Hierocene.

glyphs. And so we knew what the inscription said because we knew how to read the Greek. And so we knew what the hieroglyphs were saying. We just sort of didn't know how they were saying it. I like that because it's analogous to how I think about quantum theory, where it's like, okay, I might know what it's predicting. But what I really want to understand is how is it making those predictions, right? Like different interpretations will give different stories about how the formulas of quantum mechanics makes the predictions about relative frequencies. Okay. So people at the time had the Rosetta Stone. They had a few other examples of Egyptian hieroglyphs in multiple scripts. And a key thing was that Thomas Young, the same guy, famous for the Young double-slit experiment, realized, so he knew that the Greek script

on the Rosetta Stone had the names of pharaohs in it. And so the Egyptian hieroglyphs had certain symbols that they were called cartouches, they were kind of circled. And he thought, well, maybe those are the names of the pharaohs, because that's important. And then he realized there was an assignment of phonetics, two different symbols that would reproduce the names of the pharaohs that appeared there. And then people went and looked at other scripts and found that they could reproduce the names of other pharaohs, older pharaohs from times before the Rosetta Stone,

using this phonetic interpretation that Young came up with. Okay, so the strange thing is that people didn't say, oh, okay, so it's all phonetic. They continued to try to interpret everything in terms of ideograms, And they had an explanation for why proper names are special. They would have to be sounded out phonetically. There's no way you could communicate them with concepts. And it was only when Champoliano finally said,

okay, well, let me try. He was frustrated and he said, let me try interpreting all of it phonetically, see what happens. And if the language spoken by the Egyptians had died out and there was no trace of it in Champoliano's time, he would have been able to make progress because he would be able to maybe

sound out things, but he wouldn't correspond to any language that we knew. But fortunately, that hadn't happened. So the Egyptians, the kind of common language spoken by the Egyptians at the time did die out as sort of a language that people used on a regular basis after the

Arabic invasion, and Arab became the language they spoke, but it survived in the Christian Coptic church. So there was a liturgical tradition, and that language was basically the language of the ancient Egyptians.

And so it just so happened that Champolillo had studied Coptic. He knew the language Coptic. And so when he tried this phonetic interpretation, he started seeing kind of the pronouns and the grammatical structures of Coptic, which he knew. And so that was his Eureka moment. It's like, okay, this is a phonetic interpretation and it's Coptic.

Okay. So what's the analogy here? So what's the analogy here? So I see the fact that, you know, we have these sub-theories of quantum theory, what I call the Clifford sub-theories, that can admit, they can abide by the Knitzen principle, using just a conventional framework for realism. That's like these cartoches, right? So the phonetic versus ideographic is kind of like the sci epistemic versus sciontic.

And we can have a perfectly local, non-contextual understanding of a part of quantum theory. And I look at that and say, okay, that's good evidence that all of it is going to have to be interpreted in a way where the quantum state is epistemic. But to answer the question, you know, knowledge about what? is a bit like knowing the Coptic language. So the nice thing about the names of the pharaohs is that they sounded the same whether it was spoken by a Greek person

or an Egyptian person. So they were sort of common. You didn't need to know the language spoken by the Egyptian scribes to interpret the cartusias. But you did need to know Coptic to be able to decipher the rest. And so we're kind of in that situation. that I think, you know, in the decipherment of hieroglyphs, there was, you know, good reason for people to try phonetic everywhere

and then figure out, okay, what's the language relative to which this phonetic interpretation works? And we're kind of there. We need to sort of find a way of formulating facts about causation and inference, a new formalism that's a modification from the Costco one, much like relativistic space and time or different from non-relativistic space and time, pre-relativistic space and time. So we need an innovation to the notions of causation and inference, and that will give us sort of the answer to the question,

what's this knowledge about? But we're not there yet. So we need to find our Coptic. The thing that I think is also lacking on the side of people who want to defend the view that quantum states are states of reality is that I've never heard a good answer to the question why is it, if it's true that quantum states are states of reality, why is it that these sub-theories of quantum theory

admit of these very natural interpretations, interpretations that preserve liveness's principle, but wherein the quantum state epistemic. You can only preserve liveness as principle if you make the quantum state epistemic. So it's a bit like how prior to Champolone, people did have a reason for saying why it could be that the cartouches were phonetic,

but everything else was ideographic. They said, well, it's the special status. of proper names. You know, proper names would naturally have to be sounded out phonetically, unlike other words in the language. And I feel like it would be nice if, you know, if we could have, if somebody wants to persuade me that I'm wrong, what they really need to do is say, well, why is it that the sci epistemic point of view worked so well in these sub-theories of quantum theory? Why is that just an illusion? Why have you been misled by that? You know, here's the explanation. So anyways, that's my attempt to answer, you know, what it means,

how to think about a research program where you know that a quantum state is knowledge of something, but you haven't quite figured out what it is. So we both got to get going, and I only have a couple of questions left. Something I want to know. Tell me about some concept, something you were trying to understand for months, maybe even years. You kept banging your head against the wall, and then all of a sudden, it clicked. Whether it was mathematical, it could be philosophy related, it could be anything. I just want to know about that whole process and what was it that made it click.

Hmm. I can certainly give you an example where maybe I don't go into the details, but more kind of like a sociological example. So back in 2005 or so, I wrote a paper about extending the notion of non-contextuality to preparate. and so usually it's just talking about measurements. So it was extending it to preparations and measurements that are not projective and to transformations. And then a few years later,

I showed how you could do an experimental test of this notion, which I call generalized non-contextuality. But those experimental tests only leveraged non-contextuality as it applied to preparations, and they didn't leverage it for, measurements. And I couldn't see how to do it. And that's sort of how it stood for many years.

I sort of just didn't quite know how to do an experiment where I would make use of this assumption for measurements. And so, yeah, maybe it was 10 years later or something. There was a student, Ravi Kunjwell, who was visiting perimeter. And I said, I think this would be a good problem to come back to. and so I went to the board to explain what the problem was. And when you are forced to explain something and what the obstacles to progress are, you often articulate your ideas more clearly than you would if you didn't need to explain it to somebody.

So the amazing thing was that as I found myself trying to communicate properly what the problem was, I immediately saw the solution. Interesting. It turned out that all that really needed to, to be done to kind of get through the impasse was to explain clearly to somebody what the problem was. That's happened several times in my life as well. Even with personal problems, just, well, what's frustrating you? And then I say it and then I say the solution as I'm speaking it.

Yeah. Yeah, it was probably the clearest example of that because as I was sort of, I got to a certain point in my explanation, I was like, ah, there isn't really a research problem here because I've now seen, you know, this is what I've explained to you is actually there's the resolution. And so there's some details to be worked out, but the big question is answered now. So we can move on. What's the most inspirational advice you've received? That is a good question. I think, as I was saying earlier, Lee Smolin, in the early days of Permanent Institute,

would really communicate very forcefully the kind of impact. importance of trying to revolutionize physics, which kind of sounds like a lot of hubris, like, oh, who am I to revolutionize physics? But in the end, when you think about it, it's like, well, yes, that is the goal you should set yourself. It's not overly ambitious. It's not arrogant to set yourself that goal. So I think Lee's advice is probably some of the more inspirational advice I've ever gotten, which is like set your sights high, have audacious goals, try to revolutionize physics. What is it about Lee that's different? I can be more specific. Many of the physicists that I, most of the physicists I speak to on this channel are the sort that

are thinking in terms of how can they contribute strongly to the field of physics. But many of the physicists that I speak to off air, they almost disdained this attitude of revolutionizing, one because as you mentioned there's hubris behind it or maybe it's because of something else they've tried and they've failed and they're the elephant that has tried and failed and so don't try anymore i forget the whole parable but it could be that it could just be it's too difficult of a problem it's it could be it doesn't need a revolution it needs a slight modification it could be that hey which is generally true almost all the knowledge production in science comes from incremental change anyhow. So don't discount the non-revolutionary because the non-revolutionary is the 95% important part. So there must be something different about how Lee thinks, as you just mentioned, it was one of the most inspirational pieces of advice. So it doesn't come from many people. It came from Lee. What is it about him? Yeah, Lee probably is distinctive in that he values much more highly than most physicists.

the very innovative approaches. So like new ideas, new approaches. He understands that most of them are probably wrong, so he sort of describes it as a high-risk, high-reward kind of situations, sort of like an investor will invest in a lot of high-risk. So I think he sees physics as we need to invest in people who have very different ideas

of how things are going to go, and some of those are going to be right and revolutionary. And so he's very supportive of researchers who go their own way and try something completely different. So he's, yeah, very tolerant of that in a way. So if you're more conservative and less and more risk-averse, you might say, well, a little bit of that's fine, but, you know, let's also stick to kind of what is guaranteed to make progress.

but Ali is somehow happy for people to be doing their own thing. He's happy for lots of mistakes to be made on the way to progress. So I think that's why he, yeah, that's what's different about the kind of advice he gives, that he certainly wants you to pursue your own particular path, and he sees that as the way that the community can make faster progress. I think to some extent that that's certainly true. It is difficult. I mean, the sociological reality is that you try something completely new.

It might be a very long road before you've made progress. And so how do you get a career going when you haven't got much to show for yourself? So everybody kind of needs to build their career. They can't be taking too long to make any progress at all. So there kind of has to be a balance. You have to sort of show that you're capable in, in ways that don't require you to have sort of finished your research program. But I definitely think that, you know,

more tolerance for original ideas is needed. And certainly at Permanentor, our strategy for hiring postdocs over the years has very much been preferential hiring of people who really do have original distinctive research programs. We don't hire them into a group to be mentored or supervised by somebody, but rather as independent researchers.

and that's a good recipe for, you know, generating some really novel and innovative results. Thank you for spending so much time with me. That's been a pleasure. Now, my last question, as you, that's a quick one, an easy one. There's an oracle behind you or whatever, there's an oracle, and you can ask it any one question about quantum mechanics. What do you ask it? And it's obviously, it's a truthful oracle. yeah okay um i'm going to interpret that as follows that let's let's say the oracle knows the correct

realist interpretation of quantum theory and uh i'm going to get a clue from it that will help me in my research program um so i think what i would really want to know is i i feel strongly that it is possible to unscramble the omelette of causation and inference in the quantum formalism so i kind of know how to do that classically where the causal stuff is basically, you could formalize it using category theory. You can say, okay, the objects, the causal relata are sets and causal dependencies are functions. So I have sort of a category of sets and functions. And then on the side of inference, I have a category of substochastic matrices, which basically represent conditional probability distributions and these categories interact. And so

when I believe that certain things are causally connected, then I can sort of propagate what I believe about one to updating my knowledge about the other. And I have a kind of whole full formalism for how you disentangle these notions of causation and inference. And we're trying to build that formalism in the quantum case, but there's a lot of things that I don't know how to do. So I think, you know, the most valuable clue to me right now would be, all right, just tell me how to unscramble causation and inference. I think if I knew how to unscramble causation and inference, you know,

in the mathematical formalism, I could then make, you know, progress towards, okay, what is the principle that distinguishes classical theories from quantum theories? And I could ask the next question. But there's this formal thing that we haven't sorted out yet, which I feel would be a great help moving forward. Thank you, Rob. Hi there. Kurt here. If you'd like more content from theories of everything

and the very best listening experience, then be sure to check out my substack at Kurt Giam. mongle.org. Some of the top perks are that every week you get brand new episodes ahead of time. You also get bonus written content exclusively for our members. That's c-U-R-T-J-A-I-M-U-N-G-L.org. You can also just search my name and the word substack on Google. Since I started that substack, it somehow already became number two in the science category. Now, substack for those who are unfamiliar is like a newsletter, one that's beautifully formatted, there's zero spam, this is the best place to follow the content of this channel that isn't anywhere else. It's not on YouTube, it's not on

Patreon. It's exclusive to the substack. It's free. There are ways for you to support me on substack if you want, and you'll get special bonuses if you do. Several people ask me like, hey, Kurt, you've spoken to so many people in the field of theoretical physics. a philosophy, of consciousness. What are your thoughts, man? Well, while I remain impartial in interviews, this substack is a way to peer into my present deliberations on these topics. And it's the perfect way to support me directly.

Kurtjymongle.org or search Kurtjymongle substack on Google. Oh, and I've received several messages, emails, and comments from professors and researchers. saying that they recommend theories of everything to their students. That's fantastic. If you're a professor or a lecturer or what have you and there's a particular standout episode that students can benefit from or your friends, please do share.

And of course, a huge thank you to our advertising sponsor, The Economist. Visit Economist.com slash tow, T-O-E to get a massive discount on their annual subscription. I subscribe to The Economist, and you'll love it as well. To is actually the only podcast that they currently partner with. So it's a huge honor for me, and for you, you're getting an exclusive discount. That's Economist.com slash tow, T-O-E.

And finally, you should know this podcast is on iTunes. It's on Spotify. It's on all the audio platforms. All you have to do is type in theories of everything, and you'll find it. I know my last name is complicated, so maybe you don't want to type in Jiamengal, but you can type in theories of everything, and you'll find it. Personally, I gain from re-watching lectures and podcasts. I also read in the comment that toe listeners also gain from replaying,

so how about instead you re-listen on one of those platforms, like iTunes, Spotify, Google Podcasts, whatever podcast catcher you use, I'm there with you. Thank you for listening.