Behind the Brain

Two weeks ago, I had the awesome opportunity to interview Professor Takaki Komiyama, a leading expert in neuroscience research from UC San Diego. Professor Komiyama’s work dives deep into the mysteries of how our brains learn and adapt. His lab uses techniques like live brain imaging and optogenetics to explore how experiences shape our brain’s structure and function, shedding light on everything from memory formation to behavioral flexibility. The transcript is below, with everything in green being what I said, and everything in black

Q: How did you become interested in neuroscience?

A: Middle school was really where I got interested in science in general. And then once both my grandmothers got Alzheimers, I knew I was interested in neuroscience.

Q: How did you know which fields of neuroscience? Like, how did you get that interest? Because I was reading a couple of your papers, and it seems like you have an interest in the olfactory bulb, how that’s regulated. And then also, I think motor functions was another thing I saw in your papers.

A: That’s right. No, that’s also a gradual process. You know, initially, I did my PhD. I chose my topic for my PhD, in the way that many many students do, which was just a continuation of the undergraduate studies. So in undergrads, you know, people students typically find the research opportunity that is available, not necessarily a very informed choice. And then, you know, you learn about the field, and then you, you know, pretty much any field in in biology is interesting if you know enough about it. So, you know, undergrads learn to, be interested in that field, and then they continue their study in grad school.

That that’s a very typical thing that happens in many other many students that go from other in grad school. But, during my PhD, which was 5 years long, it’s, I have, I tried to go to many seminars. I tried to learn about all the fields of neuroscience and and to try to really decide what I wanna spend my career on and during and what task take me the most and what’s what task take me is, ability of the brain to adapt to the new and arbitrary challenges and or changes in the environment. So so that’s really the the main curiosity that’s driving my research, since then. And you you’re right.

I’m impressed that you found. They say, you know, we study part of the lab cycle factron, part of the lab cycle, breast that you found, they say, you know, we study part of the lab studies on facts on, part of the lab studies motor system, and, another part of the lab studies, well, what’s called value based decision making. It’s a form of, adaptive decision making behavior. Yeah. All of these are just, you know, different ways to probe the basic function of the brain, which is the ability, to adapt to changes in the environment, ability to learn.

So, you know, and and the brain can learn different things, and we study olfactory learning. We study motor learning, motor skill learning, and also we study reinforcement learning in the context of digital banking. So those are all intellectually very much related in my mind. Yeah. So, like, going a bit more into the, like, the the truth or the field.

Q: Is there any similarities or differences you see between how the brain adapts to, you know, changes in, their experiences, how it impacts the brain’s factory system versus motor system versus decision making. Are there any differences you’ve seen when you’ve conducted studies, or are there any similarities?

A: There was many similarities and some differences. You know, similarities are the the basic cellular cellular mechanisms are, you know, generally maintained or conserved. So, you know, changes in new neuronal connections, a or synaptic connections between neurons.

That’s what really underlies a lot of long term learning and, you know, different parts of the brain do it for different contexts and different types of cells do it, but the basic, fundamental mechanisms are very similar. There are some differences. You know, for example, in the sensory system, like olfaction, the goal might be to have a sort of a stable static representation of the external, environment, like, other environment. And, goal of the motor system is very different. Right?

The goal of the motor system is to generate the dynamic movement of the body. So, that changes things a little bit. So, you know, we think of the sensory system to sort of try to have a more static representation of the world while the motor system has to learn to generate an internal dynamics that change over time, you know, in a short period of time during the movement.

Q: And while reading some of the other papers, I was also noticing you do, I think, a little bit of a lot of detail on the PPC, the posterior parietal cortex. And then you do a lot of research on that. So I just want to know how throughout your journey from the University of Tokyo through becoming a professor at UCSD, how did you focus in on certain specific areas in the brain?

A: Yeah. It really depends on the biological question that, we are asking. So in in the, in the studies that we looked at PPC, that was an educated guess based on previous literature. So in the, in that line of work, we, became interested in how the brain learns from recent experience to guide the next decisions. And that is, that, type of behavioral function has been found, at least partially done by, PPC. And this a lot of this was working, nonhuman primates, actually.

So based on the all the literature like that, we decided to hone in on PPC. But it really depends, you know, so for example, in a very similar, line of studies, we are focusing on an area of the brain that’s right next to PPC, retro spleenoid, RSC. And the reason we ended up focusing on that was because, we were looking for a particular, activity pattern and, in the brain, and we searched for it across many brain areas and found it in RSC. And then we, focused on that.

So in that in that in that context, we did, almost like a screen across brain areas to find a key area. So depending on the behavioral function that we’re studying, we would be looking at different parts of the brain. Mhmm. It feels like a kind of personal question. But when you’re talking about the PPC, you’re talking about how it, kind of mixes history and bias information.

Q: I was confused about what the difference between those two is because I feel like bias could stem from history or, like, historical, like, past experiences.

A: So yeah. So so so bias is, like so, you know, say you have different options and okay. In that particular context, we’re studying so, okay, so you’re supposed to be choosing the options based on some instruction.

Like, that’s like blue light, go go right, red light, go back to the light. Right? But, you know, sometimes you if you make it a little hard to discriminate, let me get purple. Right? And then you might make a guess kind of randomly.

I mean, that you know, in a situation like that, it sort of, reveals your internal bias. So maybe you’re more likely to go right, anyway. You know? Yeah. So that’s the kind of, bias we’re talking about.

And what we showed in that study was that at least in that context, the bias, like, for mice formed based on their recent history. So what happened to them recently, changed their bias in a predictable manner. Okay. Does that make sense?

Yeah. That makes sense.

Q: And then you’re dealing, like, a lot of research studies involving mice. Right? So are there any, like, significant differences between the nervous system of mice and the nervous system of humans that, like, could impact the results of the research you’ve done?

A: Absolutely. Yeah. Mice and humans are different. Right? Yeah. Although I would argue that there are a lot many more similarities and differences, the basic organization of the brain is remarkably similar between, mice, cats, rats, monkeys, humans, you dig it.

And, you know, it’s all, many, many neurons, millions, billions of neurons forming connections through various, special structures called synapses. Yeah. And, you know, different functions are in different parts of the brain. Motor functions tend to be in the more frontal area. Visual and sensory areas tend to be in the back of the brain.

All these things are conserved across these species. So so, yeah, there are a lot of, the similarities. There are, of course, differences as well. The scale is very different. The number of neurons could be off the off the top of my head.

I think, they’re different by a 100 fold or something. So, you know and and also there are some minor, intrinsic differences. We’re we’re learning more and more about this recently, but, there are some differences about how neurons individual neurons might perform computations, in different species. So so, yeah, we have to be careful. You know?

Like, even if you if we understand the mouse brain completely, that doesn’t mean that we understand the human brain. So, yeah, we have to be careful about what we can conclude and what we cannot. But there are so many, experiments that we can only do in animal models Yeah. To get to a really, you know, detailed mechanistic understanding of the brain. That’s just not possible in humans for many reasons, including ethical reasons. So, that’s why we study, other animals like mice.

Q: And then, are there any, like, big projects that you’re allowed to talk about that you’re kind of excited about or, like, any studies that you’re doing currently that you, are particularly optimistic about?

A: Yeah. Good question. So, you know, in the recent years, decades, couple decades, the ability of researchers to record the activity of many neurons in behaving animals, that’s improved tremendously. And so now we can record many the activity of many, many neurons in behaving animals. So we can sort of do, you know, study to correlate the, neural activity and behavior and infer the, functions of those neurons, and how the how they might have, control behavior. But what we’ve been really wanting to do is to also manipulate in a very specific way, the neural activity so that we can do, we can study causal functions of neurons. You know?

You hear about correlation versus causality. Right? Yeah. Like, even if two things happen at the same time, that doesn’t mean that one’s causing the other neuron. It doesn’t mean that the neurons are causing the behavior.

To do that, we have ideally, we wanna be able to, you know, artificially recreate the neural activity, and we’ll see if that can actually generate the behavior. Right? Yeah. And so, you know, it’s almost like, reading the brain activity versus writing specific activity into the brain. You know?

Okay. And this, writing part, the the technology has very recently, been maturing. And so that’s that’s that’s something that I’m very excited about. So we have, 2 microscopes in the lab that can do that. So, you know, we can we we use optical techniques imaging, but we can image the activity of neurons and then find some neurons that are, that show interesting activity patterns.

And we can selectively activate those neurons and see what happens to the behavior. You know? So this is an approach that is very new in the field, and, I think it’ll be a very powerful, approach. And I’m very excited about using that. Yeah.

We are in the last couple years, we’ve been using that very effectively, and we’re learning a lot about the brain.

Q: So then, like, one of my other questions related to research is, what do you say is, like, the biggest, discovery or paper, like, your proudest paper kind of or discovery or study that you’ve done, like, throughout your whole career research career?

A: I cannot pick 1. I love every one of them.

And, also, the problem is that I’m always more excited about new things that are happening in the lab. And and that’s reasonable because, you know, we get the deeper and deeper understanding as we work on the a problem for, many years. So, yeah, I’m always more excited about what’s ongoing in the lab than what I did 5, 10 years ago. But then, I mean, that’s a I I think that’s a good thing. It just keeps me keeps me excited, keeps me passionate about my research.

Q: And then for, some of your studies too, like, you used, in vivo two photon calcium imaging to keep, in mice. So could you just describe, like, how that kind of works? Like, how it’s able to image activity, specifically?

Yeah. Well, so first, we manipulate, the brain of an animal in such a way that we force the neurons to express this specific protein that changes its brightness based on the activity of the neurons. K? So we can, you know, introduce this, this protein in a number of different ways.

But the outcome is that so these neurons express this protein that when when it’s artificially excited, it, emits fluorescence, emits light. But the brightness of the light depends on the activity of the neurons. K? Yeah. And then using a 2 photon microscope, which is a sophisticated fluorescence microscope, we can identify the neuron that is active at the moment. And we can do this, over time and find, you know, okay. At this point in time, these neurons are active.

And 30 milliseconds later, these neurons are now active, and so we can see a very dynamic view of, which neurons are active at any given moment. So that’s one way of recording, the activity of many neurons at the same time. Yeah. So, also, there was one paper where you talked about how the RSC, like, kept value related, like, population activity. It was the one where your the mice you had the mice, and you were training them to do certain tasks based on the history, like the light color kind of that one.

Q: Also, you talked about also diversity in the strength and temporal stability of these signals. So I was a bit confused because I thought those two contradict each other, but both of those are, like, part of your results in that paper. I don’t know if I’m describing it well, but it’s basically, you’re saying how the RSC maintained persistent value related population activity that kept the updates on a trial by trial basis, but but then, you also said that there was diversity or, like, changes in the strength and temporal stability of history and value related signals across different brain areas.

A: So yeah. Great question. So within RSC, we found this, very stable, representation of subjective value. Mhmm. But, the way value was encoded was very different across different brain areas.

So I think that’s what you are talking about in terms of diversity. So the way that, neurons encoded certain information, was different across areas, diverse across areas. But within RSC, we found this, unique, stable encoding. Okay. So RSC is a brain area, but we also looked at other areas like PPC, m one, b one, etcetera.

Q: Another question… This is a bit unrelated from, like, your lab research kind of. I was reading another, interviewer article about you. And in that one, you mentioned about how you used to go on, trips with your father where you collected seawater samples kind of with your father who was, like, a chemical engineer studying coral reefs.

So where where did you find that? I haven’t talked about that in many years. Let’s see.

At this time, you were 35. This was November 2014.

Oh, wow. Okay. Yeah.

Q: So I just want to know how much you think that kind of impacted you. Because they also talk about how you went on to study biochem in University of Tokyo. Right? So I just wanna know how much that, like, experience kind of impacted you, how it helped you kind of.

Let’s see. So, yeah, my father was a researcher, and he at that moment, he was studying, coral meats.

And I have tagged along to his research group and even helped sample collection a few times. So I guess that’s why you, read about it. I guess what I did was that, you know, the career in research, became, something that was concrete. You know, it’s just, it’s not something that it’s something that I knew something about. You know, I met the people doing it.

I met, I’ve been in the lab. And and so, you know, for I think for many, many younger people, you know, making, research your profession is sort of a strange thing. You know? It’s not it’s not very intuitive. Yeah.

Right? Because you don’t see those people, around you very much. Like, if you walk around in the city, you might see restaurants, grocery stores, etcetera. You don’t see a lab. Yes.

But, you know, for me, it was not a strange thing. So it made it, not a huge leap to become a researcher. So I think that, I, you know, I thought about potential other carriers as well, but this was always one of the potential things that one could do. You know? So so, yeah, I think that helped.

And Bio Chem, you had the major. I think I ended up choosing the major mostly because the lab I wanted to work in. So one is which was professor Sakano’s lab, which was doing very cutting edge neuroscience at the moment. Mhmm. He happened to be in the biochemistry department, and that’s why I ended up declaring the major so that I could work in his lab.

So yeah. Yeah. By that time, I was already very much interested in neuroscience, but, biochem major was the most direct path for, for me at that moment based on who’s in which department in that university. You know? Yeah.

But, obviously, I mean, there are there were many other neuroscientists that, I could have worked with. I just didn’t know. As an undergrad, I knew very little. So, you know, just, I just happened to choose the one, lab that I knew most about.

And then, you did your postgraduate at Stanford. Right?

Graduate school.

Yeah. Sorry. Graduate graduate at Stanford.

And then during that time, you went to, the lab, right, in Virginia?

Well, after my PhD. So I did my PhD at Stanford, and then, for my postdoctoral work, I went to lab.

Q: So I was just wondering how or why did you decide to go to work in that lab specifically? What what was in that lab that you really wanted?

Yeah. So this relates to one of the first things you asked. So, you know, during grad school, I became very certain that I wanted to be a a professor.

And but, you know, it’s a it’s a long time of your life that you spend, doing research. So I wanted to make sure I will be studying something that I am truly passionate about. So, you know, I was doing a lot of, I was trying to learn about different fields in neuroscience. One field, that really particularly fascinated me was this area about how the the dynamics and learning. And so this lab had, this approach that I thought would be very powerful in studying that.

So, you know, it that’s a approach that I end end up using still in my lab. So this, you know, you you mentioned 2 photon calcium imaging. Using that to study behaving animals, that was completely new at the moment. This is in 2006, and, was just starting to become available, and this lab was one of the best to act. And so, you know, I wanted to go there, learn the technique, learn how to do that kind of neuroscience so that I can start my own lab using that technique.

So that’s why I chose a lab. At that point, at that time in my life, I decided to, only think about the the science and not the location. I decided to take the location out of the equation. So it didn’t matter whether he was in Virginia or somewhere else. I was looking for the lab I wanted to join, and he happened to be in Virginia.

Q: And how how would you say, like, teaching has impacted you or your ability to, you know, write papers and stuff kind of or do research. How’s being a professor or teaching impacted you kind of or impacted your ability to do research or, you know, stuff like that?

Absolutely. I mean, compared so I’ve been a professor for 14 years, I think. Yeah.

I think I’ve been a professor for 14 years, and I’ve learned a lot about how to do science, and I’ve become a better scientist. And a lot of that comes from working with, more, junior younger people in my lab. Many of them are graduate students and postdocs. And I work very closely with them. We are doing, you know, every day I’m having, like, really detailed project discussion with, someone in my app. And during that, during that kind of interactions, I have really learned, and I’m still learning how to, move a project forward, how to prioritize which aspects of, the project, how to work with different people.

Those are the things that I’ve really learned, and I’m still learning. I’m still becoming better. Yeah. It’s a it’s a lifelong journey, I think. And then, so, like, a bunch of papers.

Q: Like, I’m interested in neuroscience, so I sometimes read a couple of papers. A bunch of them, they use, like, AI models or mathematical models. So have you have you ever used those? And do you prefer those, or are you neutral about those? Because some, people I’ve spoken to, they think they should not be used at all, and some people think they’re really good. So I just wanted to know where you stand on that.

Yeah. Absolutely. We should use everything we can to try to understand the brain. And we are we have been using AI techniques. So for it’s machine learning techniques.

So for instance, we had a paper recently where, we trained mice to perform a certain behavioral task, and then we also trained an AI agent to perform the same task. K? And then we recorded the activity of the mouse brain, and also we looked inside of the AI, neural network, and we compare the similarities between these 2, very different things that are doing the same thing. Right? So they’re the behavior looks very similar.

You know? So they’re doing the same behavioral task. And the mouse is doing it using the mouse brain. AI is using it in, using, artificial neural network, and we found a lot of similarities. And I think that’s interesting.

And, also, it seems like, AI is sort of using the same solution that the brain has come up with over 1,000,000, 100 of 1,000,000 of years of evolution. And, it and I guess the fact that AI sometimes uses very similar solutions suggest, you know, this is one of the optimal solutions. So and and also, you know, in AI, it’s really easy to manipulate. So, you know, you can do all sorts of manipulations when you have AI agents that are smoother to the brain and learn about the dynamics of that network and perhaps maybe try to reproduce some of the results in the in the end model model. So so this is, you know, some of the ways we can use AI to aid, neuroscience research.

Q: Okay. And then I’ll I’ll actually have another question, which is basically I think you already answered it, which was, do you see any similarities between AI models and our human brain? I think you answered it. So the other another question I had was about one of the papers about, how our ability to sense different smells and stuff is impacted based on our, environment. Like, if we’re awake, if we’re asleep, basically, like, kind of you know?

And, I think in this one, like, in the findings, it said, when we’re under anesthesia, odor representations, they, like, kind of overlap. So we’re not able to detect as many smells. So I I was a bit confused here because, like, when I was reading the paper, like, from the beginning, I thought they would be the opposite that when we’re asleep, our brain has evolved to be able to smell stuff quicker because because we’re not able to see. So I was wondering, do you have any theories on, like, why when you’re asleep, odor representation goes down? Is it just that you’re, like, more tired or your brain is resting?

Or It’s actually the opposite that we found. Well, I mean, yes and no. So what we found was that during anesthesia, the odor representations are stronger. K? But they were not so the representations of different odors were not very different.

So, you know, whatever you’re smelling, you get similar, like, similar representations. Oh, so you can you can easily detect that something some smell is there, but it’s tougher to detect what type of smell it is? Exactly. Yeah. Yeah.

Okay. At at least based on the part of the brain we were looking at. Yeah. Okay. Yeah.

That makes sense. And then in that same paper, you talk about granule cells that which, shape like I hope I’m pronouncing this right. Mitral cell responses? And then, your paper mentioned that those granular cells were GABAergic interneuron. So when I googled that, it basically told me those are, like, neurons that relate to or interact with, we get GABA. But then I know that the role of GABA changes sometime, in the, like, our development years from an exit to excitatory, neuro, excitatory neurotransmitter to inhibitory in I can’t pronounce. Inhibitatory neurotransmitter.

Q: So when those type of switches occur in GABA, does that impact like, does the opposite switch occur in the cell as well, or does that impact the function of those cells too? How does that work? Yeah.

No. No. This is great. But, yeah. So so the, developmental switch you’re talking about is very, very early development. So in the adult animals, you you know, that’s pretty consistent to that the GABA GABA is inhibitory. Okay. So, yeah, in the, you know, normally functioning brain, GABA in almost all of the brain is, inhibits other neurons. Okay.

So you talked about how, you like, the next big thing you see happening in, like, neuroscience, like, new like studying neuroscience is, how we’re able to cause certain neurons to do things so that we can explore causality instead of correlation kind of. Right? Other than that, are there, like, any extremely big no. I I won’t say, like, advances, but anything there were instead of finding any treatments or methods for treatment you see for any neuro neurodegenerative diseases that you’ve heard about or you know? Yeah.

There’s so much, effort, so much, research in, at many different levels, trying to treat any diseases disorders, you know, psychiatric, neuro, neurodisgenerative, diseases. Yeah. So I think, new treatments are coming online, you know, every year. You know, my lab started to work on Huntington’s disease model mice, and also and and we have some potential, treatment that might be making might be making these mice, better. But, yeah.

So, you know, this is just one of many, many efforts. Yeah. Hopefully, some of this, you know, our basic science knowledge will translate to better treatment of, patients and and all of us. And that’s what’s indeed happening. I don’t know.

Do you have a last thing you wanna talk about?

I just wanted to ask about mirror neurons because that was one thing I’m really interested in. So I just wanted to know if mirror neurons could explain anything that you were seeing about how activity of, like, observed activity and the correlation between that and learning. That was kind of my last question I wanted to ask if mirror neurons could explain that.

Yeah. Mirror neurons are interesting. Right? In so so they respond to your own versus, other individuals’ same action. So, you know, that’s interesting.

And I think the way the the people find it interesting is because they it sounds like, more concept related. You know? It’s not just controlling your own behavior, but it’s a concept about that behavior because, you know, it doesn’t matter who does it. So, yeah, so it seems like a more high high level cognitive function, but, you know, exactly what they’re good for, what they’re doing, we have no idea, in my opinion. And, you know, the so here, another causal manipulation would be really, effective.

Right? So, you know, we find these neurons, but what happens if we, block their activity? What happens if we artificially cause their activity? You know, can we, induce a perception of some behavior? So that’s the type of experiment that the that is, that they’re they’re in for.

Summary

In conclusion, Professor Komiyama started out by talking about the beginning of his research journey and education. Then he talked about certain research topics and techniques he used throughout his research. Finally, he talked about future goals, research projects, etc.