Adam Marblestone——AI 对大脑的认识缺失了某个根本要素

The big million-dollar question that I have, that I've been trying to get the answer to through all these interviews with AI researchers: How does the brain do it? We're throwing way more data at these LLMs, and they still have a small fraction of the total capabilities that a human does. So what's going on?

This might be the quadrillion-dollar question or something like that. You can make an argument that this is the most important question in science. I don't claim to know the answer. I also don't think that the answer will necessarily come even from a lot of smart people thinking about it as much as they are.

My overall meta-level take is that we have to empower the field of neuroscience to make neuroscience a more powerful field, technologically and otherwise, to actually be able to crack a question like this. Maybe the way that we would think about this now with modern AI, neural nets, and deep learning is that there are certain key components of that.

There's the architecture. There are maybe hyperparameters, like how many layers you have, or properties of that architecture. There is the learning algorithm itself. How do you train it? Backprop, gradient descent—is it something else? How is it initialized?

If we take the learning part of the system, it still may have some initialization of the weights. And then there are also cost functions. What is it being trained to do? What's the reward signal? What are the loss functions, the supervision signals?

My personal hunch within that framework is that the field has neglected the role of these very specific loss functions, very specific cost functions. Machine learning tends to like mathematically simple loss functions. Predict the next token, cross-entropy—these simple computer-scientist loss functions.

I think evolution may have built a lot of complexity into the loss functions, actually: many different loss functions for different areas, turned on at different stages of development. It's basically a lot of Python code generating a specific curriculum for what different parts of the brain need to learn. Because evolution has seen many times what was successful and unsuccessful, evolution could encode the knowledge of the learning curriculum.

In the machine-learning framework, maybe we can come back and talk about where the loss functions of the brain come from. Can different loss functions lead to different efficiency of learning? People say the cortex has got the universal human learning algorithm, the special sauce that humans have. What's up with that? This is a huge question, and we don't know.

The cortex typically has this six-layered structure, with layers in a slightly different sense than the layers of a neural net. Any one location in the cortex has 6 physical layers of tissue as you go in layers of the sheet. Those areas then connect to each other, and that's more like the layers of a network.

I've seen versions of that where what you're trying to explain is just, “How does it approximate backprop?” And what is the cost function for that? What is the network being asked to do if you are trying to say it's something like backprop? Is it doing backprop on next-token prediction, or is it doing backprop on classifying images, or what is it doing? And no one knows.

But one thought about it, one possibility about it, is that it's just this incredibly general prediction engine. So any one area of the cortex is just trying to predict basically any subset of all the variables it sees from any other subset. That's omnidirectional inference, or omnidirectional prediction.

Whereas an LLM is just seeing everything in the context window, and then it computes a very particular conditional probability, which is, “Given all the last thousands of things, what are the probabilities for the next token?”

But it would be weird for a large language model to say, “The quick brown fox blank blank the lazy dog,” and fill in the middle versus doing the next token if it's doing just forward. It can learn how to do that stuff at this emergent level of the context window and everything, but natively it's just predicting the next token.

What if the cortex is natively made so that any area of the cortex can predict any pattern in any subset of its inputs, given any other missing subset? That is a little bit more like probabilistic AI.

A lot of the things I'm saying, by the way, are extremely similar to what Yann LeCun would say. He's really interested in these energy-based models, and something like that is the joint distribution of all the variables. What is the likelihood or unlikelihood of just any combination of variables?

If I clamp some of them and say that definitely these variables are in these states, then I can compute, with probabilistic sampling, for example—conditioned on these being set in this state, and these could be any arbitrary subset of variables in the model—can I predict what any other subset is going to do and sample from any other subset, given clamping this subset? And I could choose a totally different subset and sample from that subset. So it's omnidirectional inference.

There could be some parts of the cortex—there might be association areas of the cortex that predict vision from audition. There might be areas that predict things that the more innate part of the brain is going to do.

Because remember, this whole thing is riding on top of a lizard brain and lizard body, if you will. And that thing is worth predicting, too. You're not just predicting, “Do I see this or do I see that?” Is this muscle about to tense? Am I about to have a reflex where I laugh? Is my heart rate about to go up? Am I about to activate this instinctive behavior?

Based on my higher-level understanding, I can match somebody telling me there's a spider on my back to this lizard part that would activate if I were literally seeing a spider in front of me. You learn to associate the 2 so that even just from somebody hearing you say, “There's a spider on your back”—

Well, let's come back to this.

This is partly having to do with Steve Byrnes' theories, which I'm recently obsessed about. But on your podcast with Ilya, he said, “Look, I'm not aware of any good theory of how evolution encodes high-level desires or intentions.”

I think this is very connected to all of these questions about the loss functions and the cost functions that the brain would use. And it's a really profound question, right? Let's say that I am embarrassed for saying the wrong thing on your podcast because I'm imagining that Yann LeCun is listening and he says, “That's not my theory. You described energy-based models really badly.”

That's going to activate in me innate embarrassment and shame, and I'm going to want to go hide and whatever. That's going to activate these innate reflexes. That's important because I might otherwise get killed by Yann LeCun's marauding army.

The French AI researchers are coming for you, Adam.

So it's important that I have that instinctual response. But of course, evolution has never seen Yann LeCun, or known about energy-based models, or known what an important scientist or a podcast is.

Somehow the brain has to encode this desire to not piss off really important people in the tribe or something like this in a very robust way, without knowing in advance all the things that the Learning Subsystem of the brain—the part that is learning, the cortex and other parts—will encounter.

The cortex is going to learn this world model. It's going to include things like Yann LeCun and podcasts. And evolution has to make sure that those neurons—whatever the Yann-LeCun-being-upset-with-me neurons are—get properly wired up to the shame response or this part of the reward function.

And this is important, right? Because if we're going to be able to seek status in the tribe, or learn from knowledgeable people, as you said, or exchange knowledge and skills with friends but not with enemies, we have to learn all this stuff. It has to be able to robustly wire these learned features of the world, learned parts of the world model, up to these innate reward functions, and then actually use that to learn more.

Because next time I'm not going to try to piss off Yann LeCun if he emails me that I got this wrong. We're going to do further learning based on that. In constructing the reward function, it has to use learned information. But how can evolution, which didn't know about Yann LeCun, do that?

The basic idea that Steve Byrnes is proposing is that part of the cortex, or other areas like the amygdala that learn, are modeling the Steering Subsystem.

The Steering Subsystem is the part with these more innately programmed responses and the innate programming of these series of reward functions, cost functions, and bootstrapping functions that exist. There are parts of the amygdala, for example, that are able to monitor what those parts do and predict what those parts do. How do you find the neurons that are important for social status? Well, you have some innate heuristics of social status, for example, or you have some innate heuristics of friendliness that the Steering Subsystem can use.

The Steering Subsystem actually has its own sensory system, which is crazy. We think of vision as being something that the cortex does. But there's also a subcortical visual system called the superior colliculus, with an innate ability to detect faces, for example, or threats. So there's a visual system that has innate heuristics, and the Steering Subsystem has its own responses.

There'll be a part of the amygdala or part of the cortex that is learning to predict those responses. What are the neurons that matter in the cortex for social status or for friendship? They're the ones that predict those innate heuristics for friendship. You train a predictor in the cortex and you say, “Which neurons are part of the predictor?” Those are the ones that—now you've actually managed to wire it up.

This is fascinating. I feel like I still don't understand—I understand how the cortex could learn how this primitive part of the brain would respond to… Obviously, it has these labels on: “Here's literally a picture of a spider, and this is bad; be scared of this.” The cortex learns that this is bad because the innate part tells it that. But then it has to generalize to, “Okay, the spider's on my back. And somebody's telling me the spider's on your back. That's also bad.” Yes. But it never got supervision on that. So how does it…?

Well, it's because the Learning Subsystem is a powerful learning algorithm that does have generalization, that is capable of generalization. The Steering Subsystem—these are the innate responses. You're going to have some built into your Steering Subsystem, these lower brain areas: hypothalamus, brainstem, et cetera. Again, they have their own primitive sensory systems.

So there may be an innate response. If I see something that's moving fast toward my body that I didn't previously see was there, and that's small, dark, and high-contrast, that might be an insect skittering onto my body. I am going to flinch. There are these innate responses. There's going to be some group of neurons, let's say, in the hypothalamus, that is the “I-am-flinching” or “I-just-flinched” neurons in the hypothalamus.

When you flinch, first of all, it's a negative contribution to the reward function. You didn't want that to happen, perhaps. But that's a reward function that doesn't have any generalization in it. I'm going to avoid that exact situation of the thing skittering toward me. Maybe I'm going to avoid some actions that lead to the thing skittering.

That's a generalization you can get—what Steve calls downstream of the reward function. I'm going to avoid the situation where the spider was skittering toward me, but you're also going to do something else. There's going to be a part of your amygdala, say, that is saying, “Okay, a few milliseconds, hundreds of milliseconds, or seconds earlier, could I have predicted that flinching response?”

It's going to be a group of neurons that is essentially a classifier of, “Am I about to flinch?” And I'm going to have classifiers for that for every important Steering Subsystem variable that evolution needs to take care of. Am I about to flinch? Am I talking to a friend? Should I laugh now? Is the friend high-status? Whatever variables the hypothalamus and brainstem contain: Am I about to taste salt? It's going to have all these variables, and for each one it's going to have a predictor.

You're going to train that predictor. Now the predictor that it trains can have some generalization. The reason it can have some generalization is because it just has a totally different input. Its input data might be things like the word “spider,” but the word “spider” can activate in all sorts of situations that lead to the word “spider” activating in your world model.

If you have a complex world model with really complex features, that inherently gives you some generalization. It's not just the thing skittering toward me; even the word “spider,” or the concept of “spider,” is going to cause that to trigger. This predictor can learn that. Whatever spider neurons are in my world model—which could even be a book about spiders or somewhere, a room where there are spiders, or whatever that is…

The amount of heebie-jeebies that this conversation is eliciting in the audience… Now I'm activating your Steering Subsystem—your spider-hypothalamus subgroup of neurons for skittering insects is activating based on these very abstract concepts in the conversation. If you keep going, I'm going to put in a trigger warning.

That's because you learned this. The cortex inherently has the ability to generalize because it's just predicting based on these very abstract variables and all this integrated information that it has. Whereas the Steering Subsystem only can use whatever the superior colliculus and a few other sensors can spit out.

By the way, it's remarkable that the person who's made this connection between different pieces of neuroscience, Steve Byrnes, is a former physicist. For the last few years, he's been trying to synthesize—he's an AI safety researcher. He's just synthesizing. This comes back to the academic incentives thing.

I think that this is a little bit hard to say. What is the exact next experiment? How am I going to publish a paper on this? How am I going to train my grad student to do this? It's very speculative. But there's a lot in the neuroscience literature, and Steve has been able to pull this together.

And I think that Steve has an answer to Ilya's question, essentially, which is: How does the brain ultimately code for these higher-level desires and link them up to the more primitive rewards?

Very naive question, but why can't we achieve this omnidirectional inference by just training the model to not just map from a token to the next token, but remove the masks in the training so it maps every token to every token, or come up with more labels between video and audio and text so that it's forced to map one to each one?

I mean, that may be the way. It's not clear to me. Some people think that there's a different way that it does probabilistic inference or a different learning algorithm that isn't backprop. There might be other ways of learning—energy-based models or other things like that—that you can imagine are involved in being able to do this, and that the brain has.

But I think there's a version of it where what the brain does is crappy versions of backprop to learn to predict through a few layers, and that it's kind of like a multimodal foundation model.

LLMs are maybe just predicting the next token. But vision models maybe are trained to learn to fill in the blanks or reconstruct different pieces or combinations. But I think that it does it in an extremely flexible way.

If you train a model to just fill in this blank at the center, okay, that's great. But what if you didn't train it to fill in this other blank over to the left? Then it doesn't know how to do that. It's not part of its repertoire of predictions that are amortized into the network.

Whereas with a really powerful inference system, you could choose at test time what is the subset of variables it needs to infer and which ones are clamped?

Okay, 2 sub-questions. One, it makes you wonder whether the thing that is lacking in artificial neural networks is less about the reward function and more about the encoder or the embedding. Maybe the issue is that you're not representing video and audio and text in the right latent abstraction such that they could intermingle and conflict.

Maybe this is also related to why LLMs seem bad at drawing connections between different ideas. Are the ideas represented at a level of generality at which you could notice different connections?

Well, the problem is these questions are all commingled. If we don't know if it's doing a backprop-like learning, and we don't know if it's doing energy-based models, and we don't know how these areas are even connected in the first place, it's very hard to really get to the ground truth of this.

But yeah, it’s possible. I think that people have done some work. My friend Joel Dapello actually did something some years ago where he put a model—I think it was a model of V1, specifically how the early visual cortex represents images—as an input into a ConvNet, and that improves some things.

There could be differences. The retina is also doing motion detection, and certain things are getting filtered out. There may be some preprocessing of the sensory data. There may be some clever combinations of which modalities are predicting which, and so on, that lead to better representation. There may be much more clever things than that.

Some people certainly do think that there are inductive biases built into the architecture that will shape the representations differently, or that there are clever things that you can do. Astera, which is the same organization that employs Steve Byrnes, just launched this neuroscience project based on Doris Tsao’s work. She has some ideas about how you can build vision systems that basically require less training. They build into the assumptions of the design of the architecture things like objects are bounded by surfaces, surfaces have certain types of shapes, and relationships of how they occlude each other and stuff like that. It may be possible to build more assumptions into the network.

Evolution may have also put some changes in the architecture. I think the cost functions and so on may also be a key thing that it does.

1. What the genome actually encodes

I want to talk about this idea that you just glanced off of, which was amortized inference. Maybe I should try to explain what I think it means, because I think it’s probably wrong, and this will help you correct me. It’s been a few years for me too.

Right now, the way the models work is that you have an input, it maps it to an output, and this is amortizing a process—the real process, which we think is what intelligence is. It’s that you have some prior over how the world could be, what are the causes that make the world the way that it is. Then, when you see some observation, you should be like, “Okay, here’s all the ways the world could be. This cause explains what’s happening best.”

Now, doing this calculation over every possible cause is computationally intractable. So then you just have to sample: “Oh, here’s a potential cause. Does this explain this observation? No, forget it. Let’s keep sampling.” Then eventually you get the cause, the cause explains the observation, and this becomes your posterior. That’s actually pretty good.

Bayesian inference in general is this very intractable thing. The algorithms that we have for doing that tend to require taking a lot of samples, as in Monte Carlo methods. Taking samples takes time. This is like the original Boltzmann machines and stuff. These techniques are still used with probabilistic programming and other types of methods often.

The Bayesian inference problem, which is basically the problem of perception, is: given some model of the world and given some data, how should I update my model? What are the missing variables in my internal model? I guess the idea is that neural networks are hopefully doing this. Obviously, mechanistically, the neural network is not starting with, “Here is my model of the world, and I’m going to try to explain this data.” But the hope is that instead of starting with, “Hey, does this cause explain this observation? No. Did this cause explain this observation? Yes,” what you do is just: observation to cause. What’s the cause that the neural net thinks is the best one? Observation to cause.

So the feedforward goes from observation to cause and then to the output. You don’t have to evaluate all these energy values or whatever and sample around to make them higher and lower. You just say, approximately, that process would result in this being the top one or something like that.

Exactly. One way to think about it might be that test-time compute—inference-time compute—is actually doing this sampling again. You literally read its chain of thought. It’s actually doing this toy example we’re talking about, where it’s like, “Oh, can I solve this problem by doing X? Nah, I need a different approach.”

This raises the question. Over time, it is the case that the capabilities which required inference-time compute to elicit get distilled into the model. So you’re amortizing the thing which previously you needed to do these rollouts, these Monte Carlo rollouts, to figure out.

In general, maybe there’s this principle that digital minds, which can be copied, have different tradeoffs which are relevant from biological minds, which cannot. So in general, it should make sense to amortize more things because you can literally copy the amortization, or copy the things that you have built in.

This is a tangential question where it might be interesting to speculate about. In the future, as these things become more intelligent and the way we train them becomes more economically rational, what will make sense to amortize into these minds, which evolution did not think was worth amortizing into biological minds? You have to retrain every time.

First of all, I think the probabilistic AI people would be like, “Of course you need test-time compute,” because this inference problem is really hard, and the only ways we know how to do it involve lots of test-time compute. Otherwise, it’s just this crappy approximation. You’d have to use infinite data or something to make this work.

I think some of the probabilistic people will be like, “No, it’s inherently probabilistic, and amortizing it in this way just doesn’t make sense.” They might then also point to the brain and say, “Okay, well, the brain—the neurons are stochastic, and they’re sampling and they’re doing things. So maybe the brain actually is doing more like the non-amortized inference, the real inference.”

But it’s also strange how perception can work in just milliseconds or whatever. It doesn’t seem like it uses that much sampling. So it’s also clearly doing some baking things into approximate forward passes or something like that to do this.

In the future, I don’t know. Is it already a trend to some degree that things that people were having to use test-time compute for are getting used to train back into the base model? Now it can do it in 1 pass.

Maybe evolution did or didn’t do that. I think evolution still has to pass everything through the genome to build the network, and the environment in which humans are living is very dynamic. So maybe, if we believe this is true, there’s a Learning Subsystem, per Steve Byrnes, and a Steering Subsystem, where the Learning Subsystem doesn’t have a lot of pre-initialization or pretraining. It has a certain architecture, but then within a lifetime it learns.

Then evolution didn’t actually amortize that much into that network. It amortized it instead into a set of innate behaviors and a set of these bootstrapping cost functions, or ways of building up very particular reward signals.

This framework helps explain this mystery that people have pointed out and I’ve asked a few guests about, which is that if you want to analogize evolution to pretraining, well, how do you explain the fact that so little information is conveyed through the genome?

So 3 gigabytes is the size of the total human genome. Obviously, a small fraction of that is actually relevant to coding the brain. Previously, people made this analogy that evolution has found the hyperparameters of the model—the numbers which tell you how many layers there should be, the architecture, basically how things should be wired together.

But if a big part of the story is that increased sample efficiency aids learning, generally makes systems more performant, it’s the reward function, it’s the loss function—and if evolution found those loss functions that aid learning—then it actually makes sense how you can build an intelligence with so little information. Because the reward function, in Python, is literally a line. So you just have 1,000 lines like this, and that doesn’t take up that much space.

Yes. It also gets to do this generalization thing with what I was describing, where we were talking about the spider, where it learns just the word “spider,” which triggers the spider reflex or whatever. It gets to exploit that too. It gets to build a reward function that actually has a bunch of generalization in it just by specifying this innate spider stuff and the Thought Assessors, as Steve calls them, that do the learning.

That's potentially a really compact solution to building up these more complex reward functions that you need. It doesn't have to anticipate everything about the future of the reward function. It just has to anticipate what variables are relevant and what heuristics are useful for finding those variables. And then it has to have a very compact specification for the learning algorithm and basic architecture of the Learning Subsystem.

And then it has to specify all this Python code for all the stuff about spiders, all the stuff about friends, all the stuff about your mother, all the stuff about mating and social groups and joint eye contact. It has to specify all that stuff. So is this really true?

I think that there is some evidence for it. Fei Chen and Evan Macosko, along with various other researchers, have been doing these single-cell atlases. One of the things that scaling up neuroscience technology—again, this is one of my obsessions—has done through the BRAIN Initiative, a big neuroscience funding program, is basically go through different areas, especially of the mouse brain, and map where the different cell types are.

How many different types of cells are there in different areas of the cortex? Are they the same across different areas? Then you look at these subcortical regions, which are more like the Steering Subsystem or reward-function-generating regions. How many different types of cells do they have? Which neuron types do they have?

We don't know how they're all connected, exactly what they do, what the circuits are, or what they mean, but you can just quantify how many different kinds of cells there are by sequencing the RNA. And there are a lot more weird, diverse, and bespoke cell types in the Steering Subsystem, basically, than there are in the Learning Subsystem.

For the cortical cell types, it seems like there's enough to build a learning algorithm up there and specify some hyperparameters. And in this Steering Subsystem, there's a gazillion—thousands of really weird cells—which might be the one for the spider flinch reflex and the one for “I'm about to taste salt.”

Why would each reward function need a different cell type?

Well, this is where you get innately wired circuits. In the learning algorithm part, in the Learning Subsystem, you specify the initial architecture and a learning algorithm. All the juice is happening through plasticity of the synapses, changes to the synapses within that big network.

But it's a relatively repeating architecture in terms of how it's initialized. It's just like how the amount of Python code needed to make an 8-layer transformer is not that different from what makes a 3-layer transformer. You're just replicating.

Whereas all this Python code for the reward function—if the superior colliculus sees something that's skittering and you're feeling goosebumps on your skin or whatever, then it triggers the spider reflex—that's just a bunch of bespoke, species-specific, situation-specific crap. The cortex doesn't know about spiders; it just knows about layers.

But you're saying that the only way to write this reward function is to have a special cell type?

Yeah, well, I think so. I think you either have to have special cell types, or you have to somehow otherwise get special wiring rules that evolution can use to say, “This neuron needs to wire to this neuron,” without any learning. And the way that is most likely to happen, I think, is that those cells express different receptors and proteins that say, “Okay, when this one comes in contact with this one, let's form a synapse.” So it's genetic wiring, and those need cell types to do it.

I'm sure this would make a lot more sense if I knew 101 neuroscience, but it seems like there's still a lot of complexity—or generality, rather—in the Steering Subsystem. So if the Steering Subsystem has its own visual system that's separate from the visual cortex, different features still need to plug into that vision system. The spider thing needs to plug into it, and the love thing needs to plug into it, et cetera, et cetera. So it seems complicated.

It's still complicated. That's all the more reason why a lot of the genomic real estate on the genome, in terms of these different cell types and so on, would go into wiring up the Steering Subsystem, pre-wiring it.

Can we tell how much of the genome is clearly working? So I guess you could tell how many are relevant to producing the RNA that manifests, or the epigenetics that manifests, in different cell types in the brain. Right?

Yeah. This is what the cell types help you get at. I don't think it's exactly like, “Oh, this percent of the genome is doing this,” but you could say, “Okay, in all these Steering Subsystem subtypes, how many different genes are involved in specifying which is which and how they wire? And how much genomic real estate do those genes take up versus the ones that specify visual cortex versus auditory cortex?”

You're just reusing the same genes to do the same thing twice.

Yes, you're right. They have to build a vision system, and they have to build some auditory systems, touch systems, and navigation-type systems. Even feeding into the hippocampus and stuff like that, there are head-direction cells. Even the fly brain has innate circuits that figure out its orientation and help it navigate in the world.

It uses vision, figures out its optical flow—how it's flying—and how its flight is related to the wind direction. It has all this innate stuff that I think, in the mammal brain, we would all lump into the Steering Subsystem. There's a lot of work. So all the genes that basically go into specifying all the things a fly has to do, we're going to have stuff like that too, just all in the Steering Subsystem.

But do we have some estimate of how many nucleotides or how many megabases it takes?

I don't know. I think you might be able to talk to biologists about this. We have a lot in common with yeast from a gene perspective. Yeast is still used as a model for some amount of drug development and stuff like that in biology.

And so much of the genome is just going toward you having a cell at all: it can recycle waste, it can get energy, and it can replicate.

And then what do we have in common with a mouse?

So we do know, at some level, that the differences between us and a chimpanzee or something—and that includes the social instincts and the more advanced differences in the cortex and so on—come from a tiny number of genes that go into this additional amount of making an 8-layer transformer instead of a 6-layer transformer, or tweaking that reward function.

This would help explain why the hominid brain exploded in size so fast. Presumably—tell me if this is correct—but under this story, social learning or some other thing increased the ability to learn from the environment. It increased our sample efficiency. Instead of having to go and kill the boar yourself and figure out how to do that, you can just be like, “The elder told me this is how you make a spear.” Now it increases the incentive to have a bigger cortex, which can learn these things.

Yes, and that can be done with relatively few genes, because it's really replicating what the mouse already has, making more of it. It's maybe not exactly the same, and there may be tweaks. From a genome perspective, you don't have to reinvent all this stuff.

So then how far back in the history of the evolution of the brain does the cortex go back? Is the idea that the cortex has always figured out this amortized inference thing, that it's been a solved problem for a long time? Then the big unlock with primates is that we got the reward function, which increased the returns to having amortized inference?

It's a good question.

Or is the amortized inference also something that took a while to unlock?

I'm not sure that there's agreement about that. I think there might be specific questions about language. Are there tweaks, whether that's through auditory and memory, some combination of auditory-memory regions?

There may also be macro-wiring, where you need to wire auditory regions into memory regions or something like that, and into some of these social instincts to get language, for example, to happen. But that might also be a small number of gene changes to be able to say, “Oh, I just need from my temporal lobe over here, going over to the auditory cortex, something.” There is some evidence for Broca's area and Wernicke's area.

They’re connected with the hippocampus and so on, and the prefrontal cortex. So there may be some small number of genes enabling humans to really properly do language. That could be a big one.

But is it that something changed about the cortex and it became possible to do these things? Or is it that the potential was already there, but there wasn’t the incentive to expand that capability and then use it, wire it to these social instincts, and use it more? I would lean somewhat toward the latter.

I think a mouse has a lot of similarity in terms of its cortex to a human. Although there’s Suzana Herculano-Houzel’s work on how the number of neurons scales better with weight in primate brains than it does in rodent brains. So does that suggest that there actually was some improvement in the scalability of the cortex?

Maybe. I’m not super deep on this. There may have been changes in architecture, changes in the folding, changes in neuron properties and stuff that somehow slightly tweak this. But there’s still a scaling either way.

That’s right. So I’m not saying there isn’t something special about humans in the architecture of the Learning Subsystem at all. But I think it’s pretty widely thought that this is expanded. But then the question is, “Okay, well, how does that fit in also with the Steering Subsystem changes and the instincts that make use of this and allow you to bootstrap using this effectively?”

But just to say a few other things, even the fly brain has some amount of learning, even very far back. I think you’ve read this great book, A Brief History of Intelligence, right? I think this is a really good book. Lots of AI researchers think this is a really good book, it seems.

You have some amount of learning going back all the way to anything that has a brain. Basically, you have something like primitive reinforcement learning going back at least to vertebrates. Imagine a zebrafish. Then you have these other branches. Birds may have reinvented something cortex-like. It doesn’t have the 6 layers, but they have something a little bit cortex-like.

So after reptiles, in some sense, birds and mammals both made something somewhat cortex-like, but differently organized. But even a fly brain has associative learning centers that actually do things that maybe look a little bit like this Thought Assessor concept from Steve Byrnes, where there’s a specific dopamine signal to train specific subgroups of neurons in the fly mushroom body to associate different sensory information with, “Am I going to get food now?” or “Am I going to get hurt now?”

Brief tangent. I remember reading in one blog post that Beren Millidge wrote that the parts of the cortex which are associated with audio and vision have scaled disproportionately between other primates and humans, whereas the parts associated, say, with odor have not. And I remember him saying something like that this is explained by that kind of data having worse scaling law properties.

Maybe he meant this, but I think another interpretation of what’s actually happening there is that these social reward functions that are built into the Steering Subsystem needed to make more use of being able to see your elders, see what the visual cues are, and hear what they’re saying. And in order to make sense of these cues, which guide learning, you needed to activate vision and audio more than odor.

There’s all this stuff. I feel like it’s come up in your shows before, actually. But even the design of the human eye, where you have the pupil and the white and everything, we are designed to be able to establish relationships based on joint eye contact. Maybe this came up in the Sutton episode. I can’t remember.

But yeah, we have to bootstrap to the point where we can detect eye contact and where we can communicate by language. That’s like what the first couple years of life are trying to do.

2. What kind of RL is the brain doing?

Okay, I want to ask you about RL. So currently, the way these LLMs are trained, if they solve the unit test or solve a math problem, that whole trajectory, every token in that trajectory, is upweighted. What’s going on with humans? Are there different types of model-based versus model-free learning happening in different parts of the brain?

Yeah, this is another one of these things. Again, all my answers to these questions—any specific thing I say—it’s all just saying that directionally we can explore around this. I find this interesting. Maybe I feel like the literature points in these directions in some very broad way.

What I actually want to do is go and map the entire mouse brain and figure this out comprehensively, and make neuroscience a ground-truth science. So I don’t know.

But first of all, I think with Ilya on the podcast, he was like, “It’s weird that you don’t use value functions, right?” You use the dumbest form of RL, basically. Of course, these people are incredibly smart and they’re optimizing for how to do it on GPUs, and it’s really incredible what they’re achieving. But conceptually, it’s a really dumb form of RL, even compared to what was being done 10 years ago.

Even the Atari game-playing stuff was using Q-learning, which is basically a kind of temporal-difference learning. Temporal-difference learning basically means you have some kind of value function: what action I choose now doesn’t just tell me literally what happens immediately after this; it tells me what the long-run consequence of that is for my expected total reward, or something like that.

So you would have value functions like that. The fact that we don’t have value functions at all in the LLMs is crazy. I think because Ilya said it, I can say it. I know 1/100th of what he does about AI, but it’s kind of crazy that this is working.

But in terms of the brain, I think there are some parts of the brain that are thought to do something that’s very much like model-free RL. Those are parts of the striatum and basal ganglia. It is thought that they have a certain finite, relatively small action space.

The types of actions they could take, first of all, might be like, “Tell the brainstem and spinal cord to do this motor action? Yes or no.” Or it might be more complicated cognitive-type actions like, “Tell the thalamus to allow this part of the cortex to talk to this other part,” or “Release the memory that’s in the hippocampus and start a new one or something.”

But there’s some finite set of actions that come out of the basal ganglia, and it’s just a very simple RL. So there are probably parts of other brains and our brain that are just doing very simple, naive-type RL algorithms.

Layering one thing on top of that, some of the major work in neuroscience—like Peter Dayan’s work—and a bunch of other work are part of why I think DeepMind did the temporal-difference learning stuff in the first place. They were very interested in neuroscience.

There’s a lot of neuroscience evidence that dopamine is giving this reward-prediction-error signal, rather than just reward—“yes or no, a gazillion time steps in the future.” It’s a prediction error, and that’s consistent with learning these value functions. So there’s that, and then there’s maybe higher-order stuff.

We have the cortex making this world model. One of the things the cortex world model can contain is a model of when you do and don’t get rewards. Again, it’s predicting what the Steering Subsystem will do. It could be predicting what the basal ganglia will do.

You have a model in your cortex that has more generalization and more concepts and all this stuff that says, “Okay, these types of plans, these types of actions, will lead in these types of circumstances to reward.” So I have a model of my reward.

Some people also think that you can go the other way. So this is part of the inference picture. There’s this idea of RL as inference. You could say, “Well, conditional on my having a high reward, sample a plan that I would have had to get there.” That’s inference of the plan part from the reward part.

I’m clamping the reward as high and inferring the plan, sampling from plans that could lead to that. So if you have this very general cortical thing, it can just do that. If you have this very general model-based system and the model, among other things, includes plans and rewards, then you just get it for free, basically.

So in neural network parlance, there’s a value head associated with the amortized inference that’s happening in the—

Yes, or there’s a value input.

Oh, okay. Interesting. Yeah, and it can predict. One of the almost sensory variables it can predict is what rewards it’s going to get.

By the way, speaking about amortizing things, obviously value is like amortized rollouts of looking up reward.

Yeah, something like that.

It’s like a statistical average or prediction of it.

Tangential thought. Joe Henrich and others have this idea for the way human societies have learned to do things like: How do you figure out that this kind of bean, which almost always poisons you, is edible if you do this 10-step, incredibly complicated process, any one step of which, if you fail at it, will make the bean poisonous? How do you figure out how to hunt this seal in this particular way, with this particular weapon, at this particular time of the year, et cetera?

There’s no way but just trying shit over generations. And it strikes me that this is actually very much like model-free RL happening at a civilizational level.

No, not exactly. Evolution is the simplest algorithm in some sense. If we believe that all of this can come from evolution, the outer loop can be extremely not foresighted.

Right, that’s interesting. Evolution: model-free. So what does that tell you? Maybe the simple algorithms can just get you anything if you do it enough.

Right. Yeah, I don’t know.

So evolution: model-free. Basal ganglia: model-free. Cortex: model-based. Culture: potentially model-free. You pay attention to your elders or whatever. Maybe there’s group selection or whatever of these things that’s more model-free.

But now I think culture stores some of the model.

3. Is biological hardware a limitation or an advantage?

Stepping back, is it a disadvantage or an advantage for humans that we get to use biological hardware, in comparison to computers as they exist now? What I mean by this question is, if there’s “the algorithm,” would the algorithm qualitatively perform much worse or much better if inscribed in the hardware of today?

Here’s what I mean. Obviously, the brain has had to make a bunch of tradeoffs which are not relevant to computing hardware. It has to be much more energetically efficient. Maybe, as a result, it has to run at slower speeds so that there can be a smaller voltage gap.

The brain runs at 200 hertz; it has to run on 20 watts. On the other hand, with robotics, we’ve clearly experienced that fingers are way more nimble than we can make motors so far. So maybe there’s something in the brain that is the equivalent of cognitive dexterity, which is maybe due to the fact that we can do unstructured sparsity. We can co-locate the memory and the compute.

Yes.

Where does this all net out? Are you like, “Fuck, we would be so much smarter if we didn’t have to deal with these brains”? Or are you like—

I think in the end we will get the best of both worlds somehow. I think an obvious downside of the brain is that it cannot be copied. You don’t have external read-write access to every neuron and synapse, whereas you do with a digital computer.

I can just edit something in the weight matrix in Python or whatever and load that up and copy that, in principle. So the fact that it can’t be copied and randomly accessed is very annoying. But otherwise, maybe it has a lot of advantages.

It also tells you that you want to somehow do the co-design of the algorithm. Maybe it doesn’t even change it that much from all of what we discussed, but you want to somehow do this co-design. So, how do you do it with really slow, low-voltage switches? That’s going to be really important for energy consumption.

Co-locating memory and compute will be really important. I think that hardware companies will probably just try to co-locate memory and compute. They will try to use lower voltages and allow some stochastic stuff.

There are some people who think that all this probabilistic stuff that we were talking about—“Oh, it’s actually energy-based models,” and so on—is doing lots of sampling. It’s not just amortizing everything. The neurons are also very natural for that because they’re naturally stochastic.

So you don’t have to do a random-number generator in a bunch of Python code, basically, to generate a sample. The neuron just generates samples, and it can tune what the different probabilities are and learn those tunings. So it could be that it’s very co-designed with some kind of inference method or something.

It’d be hilarious. The message I’m taking from this interview is that all these people that folks make fun of on Twitter—Yann LeCun and Beff Jezos and whatever—I don’t know, maybe they got it right.

That is actually one read of it. Granted, I haven’t really worked on AI at all since LLMs took off, so I’m just out of the loop. But I’m surprised, and I think it’s amazing how the scaling is working and everything.

But yeah, I think Yann LeCun and Beff Jezos are kind of onto something about the probabilistic models, or at least possibly.

In fact, that’s what all the neuroscientists and all the AI people thought until 2021 or something.

Right.

So there’s a bunch of cellular stuff happening in the brain that is not just about neuron-to-neuron synaptic connections. How much of that is functionally doing more work than the synapses themselves are doing, versus whether it’s just a bunch of kludge that you have to do in order to make the synaptic thing work?

With a digital mind, you can nudge the synapse—sorry, the parameter—extremely easily. But with a cell modulating a synapse according to the gradient signal, it just takes all this crazy machinery. So is it actually doing more than it would take extremely little code to do?

I don’t know, but I’m not a believer in the radical idea that, “Oh, actually, memory is not mostly in synapses, or learning is mostly genetic changes,” or something like that. I think it would just make a lot of sense—I think you put it really well—for it to be more like the second thing you said.

Let’s say you want to do weight normalization across all the weights coming out of your neuron or into your neuron. Well, you probably have to somehow tell the nucleus of the cell about this and then have that send everything back out to the synapses or something. So there are going to be a lot of cellular changes.

Or let’s say that you just had a lot of plasticity and you’re part of this memory. Now that’s got consolidated into the cortex or whatever. Now we want to reuse you as a new one that can learn again. There are going to be a ton of cellular changes, so there’s going to be tons of stuff happening in the cell.

But algorithmically, it’s not really adding something beyond these algorithms. It’s just implementing something that, in a digital computer, is very easy for us to do: just find the weights and change them. In a cell, it literally has to do all this with molecular machines itself, without any central controller. It’s kind of incredible.

There are some things that cells do, I think, that seem more convincing. One of the things the cerebellum has to do is predict over time. What is the time delay? Let’s say that I see a flash and then, some number of milliseconds later, I’m going to get a puff of air in my eyelid or something.

The cerebellum can be very good at predicting what the timing is between the flash and the air puff, so that now your eye will just close automatically. The cerebellum is involved in that type of learned reflex.

There are some cells in the cerebellum where it seems like the cell body is playing a role in storing that time constant, changing that time constant of delay, rather than all of that being somehow done with, “I’m going to make a longer ring of synapses to make that delay longer.” No, the cell body will just store that time delay for you.

So there are some examples, but I’m not a believer, out of the box, in essentially this theory that what’s happening is changes in connections between neurons and that that’s the main algorithmic thing that’s going on. I think there’s very good reason to still believe that it’s that rather than some crazy cellular stuff.

Going back to this whole perspective of how our intelligence is not just this omnidirectional inference thing that builds a world model, but really this system that teaches us what to pay attention to, what the important salient factors are to learn from, et cetera, I want to see if there’s some intuition we can derive from this about what different kinds of intelligences might be like.

So it seems like AGI or superhuman intelligence should still have this ability to learn a world model that’s quite general, but then it might be incentivized to pay attention to different things that are relevant for the modern, post-singularity environment. How different should we expect different intelligences to be?

I think one way to think about this question is: Is it actually possible to make the paperclip maximizer or whatever? If you try to make the paperclip maximizer, does it end up just not being smart or something like that because the only reward function it had was to make paperclips?

I’d say: Can you do that? I don’t know.

If I channel Steve Byrnes more, I think he's very concerned that the minimum viable set of things in the Steering Subsystem that you need to get something smart is way less than the minimum viable set of things you need for it to have human-like social instincts and ethics and stuff like that. So a lot of what you want to know about the Steering Subsystem is actually the specifics of how you do alignment, essentially, or what human behavior and social instincts are versus just what you need for capabilities. We talked about it in a slightly different way because we were saying, “Well, in order for humans to learn socially, they need to make eye contact and learn from others.” But we already know from LLMs that, depending on your starting point, you can learn language without that stuff.

So I think that it probably is possible to make super-powerful, model-based RL-optimizing systems and stuff like that that don't have most of what we have in the human brain reward functions and, as a consequence, might want to maximize paperclips. And that's a concern.

But you're pointing out that in order to make a competent paperclip maximizer, the kind of thing that can build spaceships and learn physics and whatever, it needs to have some drives that elicit learning, including, say, curiosity and exploration.

Yeah, curiosity, interest in others, interest in social interactions. But that's pretty minimal, I think. And that's true for humans, but it might be less true for something that's already pretrained as an LLM or something.

So most of why we want to know the Steering Subsystem, I think, if I'm channeling Steve, is alignment reasons. How confident are we that we even have the right algorithmic conceptual vocabulary to think about what the brain is doing?

What I mean by this is that there was one big contribution to AI from neuroscience, which was this idea of the neuron in the 1950s—just this original contribution. But then it seems like a lot of what we've learned afterward about the high-level algorithm the brain is implementing—from backprop, to whether there's something analogous to backprop happening in the brain, to “Oh, is V1 doing something like CNNs?” to TD learning and Bellman equations, actor-critic, whatever—it seems inspired by this dynamic where we come up with some idea, maybe we can make AI neural networks work this way, and then we notice that something in the brain also works that way. So why not think there are more things like this?

There may be. I think the reason that I think we might be onto something is that the AIs we're making based on these ideas are working surprisingly well. There's also a bunch of just empirical stuff, like convolutional neural nets and variants of convolutional neural nets. I'm not sure what the absolute latest is, but compared to other models in computational neuroscience of what the visual system is doing, they are just more predictive.

You can just score CNNs, even pretrained on cat pictures and stuff: What is the representational similarity that they have on some arbitrary other image compared to the brain activations measured in different ways? Jim DiCarlo's lab has this Brain-Score. There seems to be some relevance there. Neuroscience doesn't necessarily have something better than that.

So yes, that's just recapitulating what you're saying: The best computational neuroscience theories we have seem to have been invented largely as a result of AI models and finding things that work. You find that backprop works and then say, “Can we approximate backprop with cortical circuits?” or something. There have been things like that.

Now, some people totally disagree with this. György Buzsáki is a neuroscientist who has a book called The Brain from the Inside Out, where he basically says all our psychology concepts, AI concepts, all this stuff is just made-up stuff. What we actually have to do is figure out what the actual set of primitives is that the brain uses. And our vocabulary is not going to be adequate to that. We have to start with the brain and make new vocabulary, rather than saying “backprop” and then trying to apply that to the brain or something like that.

He studies a lot of oscillations and stuff in the brain, as opposed to individual neurons and what they do. I don't know. I think that there's a case to be made for that. From a research program design perspective, one thing we should be trying to do is just simulate a tiny worm or a tiny zebrafish, almost as biophysical or as bottom-up as possible. Get the connectome, molecules, activity, and just study it as a physical dynamical system and look at what it does.

But I don't know, it just feels like AI is really good fodder for computational neuroscience. Those might actually be pretty good models. We should look at that. I both think that there should be a part of the research portfolio that is totally bottom-up and not trying to apply our vocabulary that we learn from AI onto these systems, and that there should be another big part of this that's trying to reverse-engineer it using that vocabulary or a variant of that vocabulary. We should just be pursuing both.

My guess is that the reverse-engineering one is actually going to work-ish or something. We do see things like TD learning, which Sutton also invented separately.

That must be a crazy feeling.

Yeah, that's crazy. This equation I wrote down is in the brain. It seems like the dopamine is doing some of that, yeah.

4. Why we need to map the human brain

So let me ask you about this. You guys are funding different groups that are trying to figure out what's up in the brain. If we had a perfect representation, however you define it, of the brain, why think it would actually let us figure out the answer to these questions?

We have neural networks which are way more interpretable, not just because we understand what's in the weight matrices, but because there are weight matrices. There are these boxes with numbers in them. Even then, we can tell very basic things. We can kind of see circuits for very basic pattern matching, like following one token with another. I feel like we don't really have an explanation of why LLMs are intelligent just because they're interpretable.

I think I would somewhat dispute it. We have some description of what the LLM is fundamentally doing. What that means is that I have an architecture, a learning rule, hyperparameters, initialization, and training data. But those are things we learned because we built them, not because we interpreted them from seeing the weights. The analogous thing to the connectome is seeing the weights.

What I think we should do is describe the brain more in that language of things like architectures, learning rules, and initializations, rather than trying to find the Golden Gate Bridge circuit and saying exactly how this neuron actually works. That's going to be some incredibly complicated learned pattern.

Konrad Kording and Tim Lillicrap have this paper from a while ago, maybe 5 years ago, called “What Does It Mean to Understand a Neural Network?” What they say is basically that you could imagine you train a neural network to compute the digits of pi or something. It's like some crazy pattern. You also train that thing to predict the most complicated thing you find, predict stock prices, basically predict really complex systems, computationally complete systems. I could train a neural network to do cellular automata or whatever crazy thing.

We're never going to be able to fully capture that with interpretability, I think. It's just going to be doing really complicated computations internally. But we can still say that the way it got that way is that it had an architecture, we gave it this training data, and it had this loss function. So I want to describe the brain in the same way. And I think that this framework that I've been laying out is that we need to understand the cortex and how it embodies a learning algorithm. I don't need to understand how it computes “Golden Gate Bridge.”

But if you can see all the neurons, if you have the connectome, why does that teach you what the learning algorithm is?

Well, I guess there are a couple of different views of it. So it depends on these different parts of this portfolio. On the totally bottom-up, we-have-to-simulate-everything portfolio, it kind of just doesn't. You have to make a simulation of the zebrafish brain or something, and then you see what the emergent dynamics are in this and come up with new names and new concepts and all that. That's the most extreme bottom-up neuroscience view. But even there, the connectome is really important for doing that biophysical or bottom-up simulation.

But on the other hand, you can say, “Well, what if we can actually apply some ideas from AI?” We basically need to figure out: Is it an energy-based model, or is it an amortized VAE-type model? Is it doing backprop, or is it doing something else? Are the learning rules local or global?

If we have some repertoire of possible ideas about this, just think of the connectome as a huge number of additional constraints that will help refine and ultimately produce a consistent picture of that. I think about this for the Steering Subsystem stuff, too—just very basic things about it. How many different types of dopamine signal, or Steering Subsystem signal, or thought assessor, and so on? How many different types of what broad categories are there?

Even this very basic information—that there are more cell types in the hypothalamus than there are in the cortex—is new information about how much structure is built there versus somewhere else. How many different dopamine neurons are there? Is the wiring between prefrontal and auditory the same as the wiring between prefrontal and visual? The most basic things, we don't know.

The problem is that learning even the most basic things through a series of bespoke experiments takes an incredibly long time, whereas learning all that at once by getting a connectome is just way more efficient.

What is the timeline on this? Presumably, the idea of this is, first, to inform the development of AI. You want to be able to figure out how we get AIs to want to care about what other people think of their internal thought patterns.

But interpretability researchers are making progress on this question just by inspecting normal neural networks. There must be some feature. You can do interpretability on LLMs that exist. You can't do interpretability on a hypothetical model-based reinforcement algorithm like the brain that we will eventually converge to when we do AGI.

Fair. But what timelines on AI do you need for this research to be practical and relevant?

I think it's fair to say it's not super practical and relevant if you're in an AI 2027 scenario. What science I'm doing now is not going to affect the science of 10 years from now, because what's going to affect the science of 10 years from now is the outcome of this AI 2027 scenario.

It probably doesn't matter that much if I have the connectome. Maybe it slightly tweaks certain things. But I think there's a lot of reason to think maybe we will get a lot out of this paradigm.

Then the real thing—the thing that is the single event that is transformative for the entire future, or singularity-type event—is still more than 5 years away, or something.

Is that because we haven't captured omnidirectional inference? We haven't figured out the right ways to get a mind to pay attention to things in a way that makes sense?

I would take the entirety of your collective podcast with everyone as showing the distribution of these things. I don't know. What was Karpathy's timeline? What's Demis's timeline? Not everybody has a 3-year timeline. But there are different reasons, and I'm curious which ones are yours.

What are mine? I don't know. I'm just watching your podcast. I'm trying to understand the distribution. I don't have a super-strong claim that LLMs can't do it.

But is the crux the data efficiency, or...?

I think part of it is just that it's weirdly different from all this brain stuff. Intuitively, it's just weirdly different from all this brain stuff, and I'm kind of waiting for the thing that starts to look more like brain stuff.

I think if AlphaZero, model-based RL, and all these other things that were being worked on 10 years ago had been giving us the GPT-5-type capabilities, then I would be like, “Oh, wow, we're both in the right paradigm and seeing the results a priori. So my prior and my data are agreeing.”

Now it's like, “I don't know what exactly my data is. It looks pretty good, but my prior is sort of weird, so I don't have a super-strong opinion on it.” I think there's a possibility that essentially all other scientific research that is being done is somehow obviated. But I don't put a huge amount of probability on that.

I think my timelines might be more in the 10-year-ish range.

If that's the case, I think there is probably a difference between a world where we have connectomes on hard drives and we have an understanding of Steering Subsystem architecture, where we've compared even the most basic properties—what are the reward functions, cost function, architecture, and so on—of a mouse versus a shrew versus a small primate, et cetera.

Is this practical in 10 years? I think it has to be a really big push. How much funding? How does it compare to where we are now?

It's low-billions-dollar-scale funding in a very concerted way, I would say.

And how much is on it now?

If I just talk about some of the specific things we have going on with connectomics, E11 Bio is our main thing on connectomics. They are trying to make the technology of connectomic brain mapping several orders of magnitude cheaper.

The Wellcome Trust put out a report a year or two ago that said to get one mouse brain—the first mouse brain connectome—would be a several-billion-dollar project. E11's technology, and the suite of efforts in the field, is trying to get a single mouse connectome down to the low tens of millions of dollars.

That's a mammal brain. A human brain is about 1,000 times bigger. If, with a mouse brain, you can get to $10 million, $20 million, or $30 million with technology, if you just naively scale that, a human brain is still billions of dollars just to do one human brain. Can you go beyond that? Can you get a human brain for less than a billion?

But I'm not sure you need every neuron in the human brain. We want to, for example, do an entire mouse brain and a human Steering Subsystem, and the entire brains of several different mammals with different social instincts.

With a bunch of technology push and a bunch of concerted effort, real significant progress, if it's focused effort, can be done on the hundreds-of-millions-to-low-billions scale.

What is the definition of a connectome? Presumably, it's not a bottom-up biophysics model. Is it just that it can estimate the input-output of a brain? What is the level of abstraction?

You can give different definitions, and one of the things that's cool is that the standard approach to connectomics uses the electron microscope and very, very thin slices of brain tissue. It's basically labeling. The cell membranes are going to show up and scatter electrons a lot, and everything else is going to scatter electrons less.

But you don't see a lot of details about the molecules—which types of synapses, or different synapses with different molecular combinations and properties.

E11 and some other research in the field have switched to an optical microscope paradigm. With optical microscopy, the photons don't damage the tissue, so you can wash it and look at fragile, gentle molecules.

With E11's approach, you can get a molecularly annotated connectome. That's not just who is connected to whom by some synapse, but what molecules are present at the synapse and what type of cell that is.

A molecularly annotated connectome is not exactly the same as having the synaptic weights. It's not exactly the same as being able to simulate the neurons and say what the functional consequence is of having these molecules and connections.

But you can also do some amount of activity mapping and try to correlate structure to function—train an ML model, basically, to predict the activity from the connectome.

What are the lessons to be taken away from the Human Genome Project?

One way you could look at it is that it was a mistake, and you shouldn't have spent billions of dollars getting one genome mapped. Rather, you should have just invested in technologies that have now allowed us to map genomes for hundreds of dollars.

George Church was my PhD advisor, and he's pointed out that it was $3 billion or something—roughly $1 per base pair—for the first genome.

Then the National Human Genome Research Institute basically structured the funding process right. They got a bunch of companies competing to lower the cost. The cost then dropped by roughly a million-fold in 10 years because they changed the paradigm from macroscopic chemical techniques to these individual DNA molecules, which would make a little cluster of DNA molecules on the microscope.

You would see just a few DNA molecules at a time on each pixel of the camera. It would give you, in parallel, a look at different fragments of DNA. So you parallelized the thing by millions-fold.

That’s what reduced the cost by millions-fold. By switching from electron microscopy to optical connectomics, potentially even future types of connectomics technology, we think there should be similar patterns. That’s why E11, the Focused Research Organization, started with technology development rather than starting with saying, “We’re going to do a human brain or something, and let’s just brute-force it.” We said, “Let’s get the cost down with new technology.”

But then it’s still a big thing. Even with new next-generation technology, you still need to spend hundreds of millions on data collection. Is this going to be funded with philanthropy, by governments, or by investors? This is very TBD and very much evolving in some sense as we speak.

I’m hearing some rumors going around of connectomics-related companies potentially forming. So far, E11 has been philanthropy. The National Science Foundation just put out this call for Tech Labs, which is somewhat FRO-inspired or related. You could have a tech lab for actually going and mapping the mouse brain with us, and that would be philanthropy plus government, still in a nonprofit, open-source framework.

But can companies accelerate that? Can you credibly link connectomics to AI in the context of a company and get investment for that? It’s possible. I mean, the cost of training these AIs is increasing so much. If you could tell some story of, not only are we going to figure out some safety thing, but, in fact, once we do that, we’ll also be able to tell you how AI works, that could be compelling.

You should go to these AI labs and just be like, “Give me 1/100th of your projected budget in 2030.” I sort of tried a little bit 7 or 8 years ago, and there was not a lot of interest.

Maybe now there would be.

But all the things that we’ve been talking about, it’s really fun to talk about, but it’s ultimately speculation. What is the actual reason for the energy efficiency of the brain, for example? Is it doing real inference or amortized inference or something else?

This is all answerable by neuroscience. It’s going to be hard, but it’s actually answerable. So if you can only do that for low billions of dollars or something to really comprehensively solve that, it seems to me that, in the grand scheme of trillions of dollars of GPUs and stuff, it actually makes sense to do that investment.

Also, there have been many labs launched in the last year where they’re raising at valuations of billions for things which are quite credible but are not like, “Our ARR next quarter is going to be whatever.” It’s like, “We’re going to discover materials,” and—

Yes, moonshot startups or billionaire-backed startups. Moonshot startups I see as on a continuum with FROs. FROs are a way of channeling philanthropic support and ensuring that it’s open-source public benefit, various other things that may be properties of a given FRO.

But yes, billionaire-backed startups, if they can target the right science—the exact right science. I think there are a lot of ways to do moonshot neuroscience companies that would never get you the connectome. It’s like, “Oh, we’re going to upload the brain” or something, but never actually get the mouse connectome or something. These fundamental things are what you need to ground-truth the science.

There are lots of ways to have a moonshot company go wrong and not do the actual science. But there also may be ways to have companies or big corporate labs get involved and actually do it correctly.

This brings to mind an idea that you had in a lecture you gave 5 years ago. Do you want to explain behavior cloning?

Actually, this is funny, because the first time I saw this idea, I think it might have been in a blog post by Gwern. There’s always a Gwern blog post. There are now academic research efforts and some amount of emerging company-type efforts to try to do this.

Let’s say I’m training an image classifier. I show it pictures of cats and dogs, and they have the label “cat” or “dog.” I have a neural network that’s supposed to predict the label “cat” or “dog.” That is a limited amount of information per label that you’re putting in. It’s just “cat” or “dog.”

What if I also had it predict, “What is my neural activity pattern when I see a cat or when I see a dog,” and all the other things? If you add that as an auxiliary loss function or an auxiliary prediction task, does that sculpt the network to know the information that humans know about cats and dogs and to represent it in a way that’s consistent with how the brain represents it and the kind of representational dimensions or geometry of how the brain represents things, as opposed to just having these labels?

Does that let it generalize better? Does that let it have richer labeling?

Of course, that sounds really challenging. It’s very easy to generate lots and lots of labeled cat pictures. Scale AI or whatever can do this. It is harder to generate lots and lots of brain activity patterns that correspond to things that you want to train the AI to do.

But again, this is just a technological limitation of neuroscience. If every iPhone was also a brain scanner, you would not have this problem, and we would be training AI with the brain signals. It’s just the order in which technology has developed: we got GPUs before we got portable brain scanners.

What is the ML analog of what you’d be doing here? Because when you distill models, you’re still looking at the final layer of the log-probs across all—

If you distill one model into another, that is a certain thing. You are just trying to copy one model into another. I think that we don’t really have a perfect proposal to distill the brain. To distill the brain, you need a much more complex brain interface.

Maybe you could also do that. You could make surrogate models. Andreas Tolias and people like that are doing some amount of neural network surrogate models of brain activity data. Instead of having your visual cortex do the computation, just have the surrogate model. So you’re distilling your visual cortex into a neural network to some degree. That’s a kind of distillation.

This is doing something a little different. This is basically just saying, “I’m adding an auxiliary loss function.” I think of it as regularization, or I think of it as adding an auxiliary loss function that’s smoothing out the prediction task to also always be consistent with how the brain represents it. It might help you with things like adversarial examples, for example.

But what exactly are you predicting? You’re predicting the internal state of the brain?

Yes. So in addition to predicting the label—a vector of labels like yes, cat; not dog; yes, not boat; a one-hot vector or whatever of “yes, it’s cat,” instead of these gazillion other categories, let’s say in this simple example—you’re also predicting a vector which is all these brain signal measurements.

Gwern, anyway, had this long-ago blog post of, “Oh, this is an intermediate thing. We talk about whole-brain emulation, we talk about AGI, we talk about brain-computer interfaces. We should also be talking about this brain-data-augmented thing that’s trained on all your behavior, but is also trained on predicting some of your neural patterns.”

And you’re saying the Learning System is already doing this through the Steering System?

Yeah, and our brain, our Learning System, also has to predict the Steering Subsystem as an auxiliary task. That helps the Steering Subsystem. Now, the Steering Subsystem can access that predictor and build a cool reward function using it.

5. What value will automating math have?

Separately, you’re on the board of Lean, which is this formal math language that mathematicians use to prove theorems and so forth. Obviously, there’s a bunch of conversation right now about AI automating math. What’s your take?

Well, I think that there are parts of math that seem pretty well on track to automate. First of all, Lean was developed for a number of years at Microsoft and other places. It has become one of Convergent Research’s Focused Research Organizations, to kind of drive more engineering and focus onto it.

Lean is this programming language where, instead of expressing your math proof on pen and paper, you express it in this programming language, Lean. At the end, if you do it that way, it is a verifiable language, so that you can click “verify,” and Lean will tell you whether the conclusions of your proof actually follow perfectly from the assumptions of your proof.

So it checks whether the proof is correct automatically. By itself, this is useful for mathematicians collaborating and stuff like that. If I’m some amateur mathematician and I want to add to a proof, Terry Tao is not going to just believe my result.

But if Lean says it's correct, it's just correct. So it makes collaboration easy, but it also makes the correctness of proofs an RL signal, very much as in RLVR. Formalized math proving—so “formal” means it's expressed in something like Lean and is verifiable—is now mechanically verifiable. That becomes a perfect RLVR task.

I think that is going to keep working. It seems like there is at least one billion-dollar valuation company, Harmonic, based on this. AlphaProof is based on this, along with a couple of other emerging, really interesting companies. I think this problem of RLVRing the crap out of math proving is going to work, and we will be able to have things that search for proofs and find them in the same way that we have AlphaGo, or whatever, that can search for ways of playing the game of Go. With that verifiable signal, it works.

So does this solve math? There is still the part that has to do with conjecturing new, interesting ideas. There is still the conceptual organization of math—what is interesting? How do you come up with new theorem statements in the first place? Or even the very high-level breakdown of what strategies you use to do proofs.

I think this will shift the burden so that humans don't have to do a lot of the mechanical parts of math: validating lemmas and proofs, checking if the statement in one paper is exactly the same as the statement in another paper, and stuff like that. That will just work. If you really think we're going to get all these things we've been talking about, real AGI would also be able to make conjectures.

Yoshua Bengio has a paper, more like a theoretical paper. There are probably a bunch of other papers emerging about this. Is there a loss function for good explanations or good conjectures? That's a pretty profound question.

A really interesting math proof or statement might be one that compresses lots of information and has lots of implications for lots of other theorems. Otherwise, you would have to prove those theorems using long, complex paths of inference. Here, if you have this theorem, the theorem is correct, and you have short paths of inference to all the other ones. It's a short, compact statement.

So it's like a powerful explanation that explains all the rest of math. Part of what math is doing is making these compact things that explain the other things. It's like the Kolmogorov complexity of this statement, or something—the complexity of generating all the other statements, given that you know this one. Or, if you add this, how does it affect the complexity of the rest of the network of proofs?

So can you make a loss function that adds, “Oh, I want this proof to be a really highly powerful proof”? I think some people are trying to work on that. So maybe you can automate the creativity part.

If you had true AGI, it would do everything a human can do. So it would also do the things that creative mathematicians do. But barring that, I think just RLVRing the crap out of proofs is going to be a really useful tool for mathematicians. It's going to accelerate math a lot and change it a lot, but not necessarily immediately change everything about it.

Will we get a mechanical proof of the Riemann hypothesis or something like that? Maybe. I don't know. I don't know enough details about how hard these things are to search for, and I'm not sure anyone can fully predict that, just as we couldn't exactly predict when Go would be solved or something like that.

I think it's going to have lots of really cool applied applications. One of the things you want to do is have provably stable, secure, unhackable software. So you can write math proofs about software and say, “This code not only passes these unit tests, but I can mathematically prove that there's no way to hack it in these ways, or no way to mess with the memory,” or the types of things that hackers use. You can prove that it has these properties.

You can use the same Lean and the same proof to do formally verified software. I think that's going to be a really powerful piece of cybersecurity that's relevant for all sorts of other AI-hacking-the-world stuff. And if you can prove the Riemann hypothesis, you're also going to be able to prove insanely complex things about very complex software. Then you'll be able to ask the LLM, “Synthesize software that I can prove is correct.”

Why hasn't a provable programming language taken off as a result of LLMs?

I think it's starting to. One challenge—we are actually incubating a potential Focused Research Organization on this—is the specification problem.

Mathematicians know what interesting theorems they want to formalize. Let's say I have some code that is involved in running the power grid or something, and it has some security properties. What is the formal spec of those properties? The power-grid engineers just made this thing, but they don't necessarily know how to lift the formal spec from it. And it's not necessarily easy to come up with the spec that you want for your code.

People aren't used to coming up with formal specs, and there aren't a lot of tools for it. So you also have this user-interface-plus-AI problem: What security specs should I be specifying? Is this the spec that I want?

So there's a spec problem, and it's just been really complex and hard. But it's only in the last very short time that LLMs have been able to generate verifiable proofs of things that are useful to mathematicians, and are starting to be able to do some amount of that for software verification and hardware verification.

But I think if you project the trends over the next couple of years, it's possible that it just flips the tide. Formal methods—this whole field of formal methods, or formal verification and provable software—is this weird, almost backwater part of the more theoretical side of programming languages and stuff. It's often very academically flavored, although there was this DARPA program that made a provably secure quadcopter helicopter and stuff like that.

Secure against… What is the property that is actually proved? Not for that particular project, but just in general. Because obviously, things malfunction for all kinds of reasons.

You could say that what's going on in this part of the memory over here, which is supposed to be the part the user can access, can't in any way affect what's going on in the memory over here, or something like that. Things like that.

So there's 2 questions. One is, how useful is this? Two is, how satisfying, as a mathematician, would it be? The fact that there's this application toward proving that software has certain properties, or hardware has certain properties—if that works, that would obviously be very useful.

But from a pure… Are we going to figure out mathematics? Is your sense that there's something about finding that one construction cross-maps to another construction in a different domain, or finding that, “Oh, this lemma, if you redefine this term, still satisfies what I meant by this term, but a counterexample that previously knocked it down no longer applies”? That kind of dialectical thing that happens in mathematics—will the software replace that?

Yeah. How much of the value of this sort of pure mathematics just comes from coming up with entirely new ways of thinking about a problem, mapping it to a totally different representation? Do we have examples? I don't know.

I think of it maybe a little bit like when everybody had to write assembly code or something like that. The amount of fun, cool startups that got created was just a lot less. Fewer people could do it; progress was more grinding, slow, lonely, and so on. You had more false failures because you didn't get something about the assembly code, rather than the essential thing of whether your concept was right. It was harder to collaborate and stuff like that.

So I think it will be really good. There is some worry that by not learning to do the mechanical parts of the proofs, you fail to generate the intuitions that inform the more conceptual, creative part.

It's the same with assembly. Right. So at what point does that apply? With vibe coding, are people not learning computer science, or are they vibe coding while also simultaneously looking at the LLM explaining these abstract computer-science concepts to them, and it's all just happening faster? Their feedback loop is faster, and they're learning way more abstract computer science and algorithmic stuff because they're vibe coding?

I don't know; it's not obvious. That might be something about the user interface and the human infrastructure around it. But I guess there's some worry that people don't learn the mechanics and therefore don't build the grounded intuitions or something. My hunch is that it's super positive.

Exactly. On net, how useful that will be, or how many overall math breakthroughs—or even math breakthroughs that we care about—will happen, I don't know. One other thing that I think is cool is the accessibility question. Okay, that sounds a little bit corny: more people can do math, but who cares? But I think there are lots of people who could have interesting ideas, like maybe the quantum theory of gravity or something.

One of us will come up with the quantum theory of gravity instead of a card-carrying physicist. In the same way that Steve Byrnes is reading the neuroscience literature and he hasn't been in the neuroscience lab that much, but he's able to synthesize across the neuroscience literature and be like, "Oh, Learning Subsystem, Steering Subsystem. Does this all make sense?" He's an outsider neuroscientist in some ways. Can you have outsider string theorists or something, because the math is just done for them by the computer? Does that lead to more innovation in string theory?

Maybe, yes. Interesting.

Okay, so if this approach works and you're right that LLMs are not the final paradigm, and suppose it takes at least 10 years to get the final paradigm in that world, there's this fun sci-fi premise where you have—Terence Tao had a tweet today where he's like, "These models are like automated cleverness but not automated intelligence." You can quibble with the definitions there.

But if you have automated cleverness and you have some way of filtering—which, if you can formalize and prove things that the LLMs are saying, you could do—then you could have this situation where quantity has a quality all of its own. What are the domains of the world that could be put in this provable symbolic representation?

In the world where AGI is super far away, maybe it makes sense to literally turn everything the LLMs ever do, or almost everything they do, into super-provable statements. LLMs can actually build on top of each other because everything they do is super-provable. Maybe this is just necessary because you have billions of intelligences running around. Even if they are superintelligent, the only way the future AGI civilization can collaborate with each other is if they can prove each step.

They're just brute-force churning out—this is what the Jupiter brains are doing. It's a universal language; it's provable. It's also provable from the perspective of, "Are you trying to exploit me, or are you sending me some message that's trying to hack into my brain effectively? Are you trying to socially influence me? Are you actually just sending me the information that I need, and no more, for this?"

So davidad, who's this program director at ARIA now in the UK, has this whole design of an ARPA-style program, a sort of safeguarded AI that very heavily leverages provable safety properties. Can you apply proofs to—can you have a world model, but that world model is actually not specified just in neuron activations, but it's specified in equations? Those might be very complex equations, but if you can just get insanely good at auto-proving these things with cleverness—auto-cleverness—can you have explicitly interpretable world models as opposed to neural net world models and move back, basically, to symbolic methods just because you can have an insane amount of ability to prove things?

Yeah, I mean, that's an interesting vision. I don't know in the next 10 years whether that will be the vision that plays out, but I think it's really interesting to think about. Even for math, I mean, Terence Tao is doing some amount of stuff where it's not about whether you can prove the individual theorems. It's like, let's prove all the theorems en masse, and then let's study the properties of the aggregate set of proved theorems.

Which are the ones that got proved, and which are the ones that didn't? Okay, well, that's the landscape of all the theorems instead of one theorem at a time.

6. Architecture of the brain

Speaking of symbolic representations, one question I was meaning to ask you is: How does the brain represent the world model? Obviously, it nets out in neurons, but I don't mean extremely functionally. I mean conceptually: Is it in something that's analogous to the hidden state of a neural network, or is it something that's closer to a symbolic language?

We don't know. There's some amount of study of this. There are things like face-patch neurons that represent certain parts of the face that geometrically combine in interesting ways. That's with geometry and vision. Is that true for other, more abstract things?

There's this idea of cognitive maps. A lot of the stuff that a rodent hippocampus has to learn is place cells: Where is the rodent going to go next, and is it going to get a reward there? It's very geometric. Do we organize concepts with an abstract version of a spatial map?

There are questions about whether we can do true symbolic operations. Can I have a register in my brain that copies a variable to another register regardless of what the content of that variable is? That's this variable-binding problem. Basically, I don't know if we have that machinery, or if it's more like cost functions and architectures that make some of that approximately emerge. But maybe it would also emerge in a neural net.

There's a bunch of interesting neuroscience research trying to study what the representations look like. But what's your hunch?

My hunch is that it's going to be a huge mess, and we should look at the architecture, the loss functions, and the learning rules. I don't expect it to be pretty in there.

Which is that it's not a symbolic-language-type thing?

Yeah, probably it's not that symbolic. But other people think very differently.

Another random question, speaking of binding: What is up with feeling like there's an experience? All the parts of your brain that are modeling very different things have different drives, and at least presumably feel like there's an experience happening right now. Also, across time, you feel like—

Yeah, I'm pretty much at a loss on this one. I don't know. Max Hodak has been giving talks about this recently. He's another really hardcore neuroscience and neurotechnology person. The thing I mentioned with Doris Tsao maybe also sounds like it might touch on this question. But yeah, I don't think anybody has any idea. It might even involve new physics.

Here's another question that might not have an answer yet. Continual learning: Is that the product of something extremely fundamental at the level of even the learning algorithm? You could say, "Look, at least the way we do backprop in neural networks is that you freeze the weights. There's a training period, and you freeze the weights. So you just need this active inference or some other learning rule in order to do continual learning."

Or do you think it's more a matter of architecture, how memory is exactly stored, and what kind of associative memory you have, basically?

Continual learning—I don't know. At the architectural level, there's probably some interesting stuff that the hippocampus is doing. People have long thought this. What kinds of sequences is it storing? How is it organizing and representing that? How is it replaying it back? What is it replaying back? How exactly does that memory consolidation work? Is it training the cortex using replays or memories from the hippocampus or something like that?

There's probably some of that stuff. There might be multiple timescales of plasticity or clever learning rules that can simultaneously be storing short-term information and also doing backprop with it. Neurons may be doing a couple of things: some fast-weight plasticity and some slower plasticity at the same time, or synapses that have many states.

I mean, I don't know. From a neuroscience perspective, I'm not sure that I've seen something that's super clear on what causes continual learning, except maybe to say that this systems-consolidation idea of the hippocampus consolidating the cortex—some people think it is a big piece of this, and we still don't fully understand the details.

Speaking of fast weights, is there something in the brain that's the equivalent of this distinction between parameters and activations that we see in neural networks? Specifically, in transformers, we have this idea that some of the activations are the key and value vectors of previous tokens that you build up over time.

There’s the so-called fast weights: whenever you have a new token, you query it against these activations, but you also obviously can’t query it against all the other parameters in the network, which are part of the actual built-in weights. Is there some such distinction that’s analogous?

I don’t know. I mean, we definitely have weights and activations. Whether you can use the activations in these clever ways—different forms of actual attention, like attention in the brain—is another question.

Is that based on, “I’m trying to pay attention”?

I think there are probably several different kinds of actual attention in the brain. I want to pay attention to this area of visual cortex. I want to pay attention to the content in other areas that is triggered by the content in this area. There’s attention that’s just based on reflexes and stuff like that.

So I don’t know. There’s not just the cortex; there’s also the thalamus. The thalamus is also involved in somehow relaying or gating information. There are corticocortical connections. There’s also some amount of connection between cortical areas that goes through the thalamus.

Is it possible that this is doing some sort of matching or constraint satisfaction—matching across keys over here and values over there? Is it possible that it can do stuff like that?

Maybe. I don’t know. This is all part of the architecture of this corticothalamic system. I don’t know how transformer-like it is, or if there’s anything analogous to that attention. It’d be interesting to find out.

We’ve got to give you $1 billion so you can come on the podcast again and tell me how exactly the brain works.

Mostly, I just do data collection. It’s really unbiased data collection so all the other people can figure out these questions.

Maybe the final question to go off on is: What was the most interesting thing you learned from the Gap Map? Maybe you want to explain what the Gap Map is.

In the process of incubating and coming up with these Focused Research Organizations—these nonprofit, startup-like moonshots that we’ve been getting philanthropists and now government agencies to fund—we talked to a lot of scientists.

Some of the scientists were just like, “Here’s the next thing my graduate student will do. Here’s what I find interesting: exploring these really interesting hypothesis spaces, all the types of things we’ve been talking about.” Some of them were like, “Here’s this gap. I need this piece of infrastructure. There’s no combination of graduate students in my lab, or me loosely collaborating with other labs with traditional grants, that could ever get me that.”

“I need to have an organized engineering team that builds the miniature equivalent of the Hubble Space Telescope. If I can build that Hubble Space Telescope, then I will unblock all the other researchers in my field, or some path of technological progress, in the way that the Hubble Space Telescope lifted the boats and improved the life of every astronomer.”

But it wasn’t really an astronomy discovery in itself. It was just that you had to put this giant mirror in space with a CCD camera and organize all the people and engineering and stuff to do that.

So some of the things we talked to scientists about looked like that. The Gap Map is just a list of a lot of those things, and we call it a Gap Map. I think it’s actually more like a fundamental capabilities map. What are all these things, like miniature Hubble Space Telescopes?

Then we organized that into gaps to help people understand or search that.

What was the most surprising thing you found?

I think I’ve talked about this before, but one thing is just the overall size or shape of it, or something like that. It’s a few hundred fundamental capabilities. So if each of these were a deep-tech-startup-sized project, that’s only a few billion dollars or something. If each one of those were a Series A, that’s only—it’s not like $1 trillion to solve these gaps. It’s lower than that.

So that’s one thing. Maybe we assumed that, and that’s what we got. It’s not really comprehensive; it’s really just a way of summarizing a lot of conversations we’ve had with scientists.

I do think that in the aggregate process, things like Lean are actually surprising, because I did start from neuroscience and biology, and it was very obvious that there are these “-omics.” We need genomics, but we also need connectomics. We can engineer E. coli, but we also need to engineer the other cells. There are somewhat obvious parts of biological infrastructure.

I did not realize that math-proving infrastructure was a thing, and that was emergent from trying to do this. So I’m looking forward to seeing other things where it’s not actually this hard intellectual problem to solve. It’s maybe slightly the equivalent of AI researchers just needing GPUs or something like that, along with focus and really good PyTorch code, to start doing this.

Which are the fields that do or don’t need that? Fields that have had gazillions of dollars of investment—do they still need some of those? Do they still have some of those gaps, or is it only more neglected fields?

We’re even finding some interesting ones in actual astronomy—actual telescopes that have not been explored. Maybe because once you’re above a critical-mass-sized project, then you have to have a really big project, and that’s a more bureaucratic process with the federal agencies.

I guess you just need scale in every single domain of science these days.

Yeah, I think you need scale in many of the domains of science. That does not mean that low-scale work is not important. It does not mean that creativity, serendipity, and so on—and each student pursuing a totally different direction or thesis that you see in universities—is not also really key.

But I think some amount of scalable infrastructure is missing in essentially every area of science, even math, which is crazy. Because mathematicians—I thought they just needed whiteboards, but they actually need Lean. They actually need verifiable programming languages and stuff. I didn’t know that.

Cool. Adam, this is super fun. Thanks for coming on.

Thank you so much. My pleasure.

Where can people find your stuff?

Pleasure. The easiest way now—my adammarblestone.org website is currently down, I guess. But convergentresearch.org can link to a lot of the stuff we’ve been doing.

And then you have a great blog, Longitudinal Science.

Longitudinal Science, yes, on WordPress.

Cool. Thank you so much.

Pleasure.