Daniel and Kelly’s Extraordinary UniverseHow Does Memory Work? (Featuring Dr. Michael Yassa)
Kelly talks to Dr. Michael Yassa about how memories are formed, comparing brains to computers, brain-computer interfaces, and more!
The analogy that I like to use for this is imagine you are at a club and there's lots and lots of noise going on, lots of things going on, lots of people talking to each other, there's loud music and so on, and you're trying to communicate something to somebody who's dancing alongside gm, and it's very, very difficult because of all of that noise. But now you get real close to them and you kind of, you know, hold your hand to their ears and you start to talk directly into their ears. Now they're going to receive that communication with much higher fidelity, be able to tune out noise and selectively attend to that communication. You've enhanced the communication between that person and the person they're talking to. That's what happens at synapses. There's lots and lots of synoptic firing, lots and lots of communication happening. But when cells start to attach to each other, they communicate much more preferentially. They can transmit signals that express that form of learning. In other words, if there's an experience that happens that is learned by the brain, the brain can express a form of plasticity or a form of memory in the strength of the connections. So if the connections grow stronger, that's a signal that this memory has been learned. And most of the information that we have about this comes from animal models, comes from slice recordings where we can see evidence for enhancements in the connectivity, enhancement of the communication between cells as a result of a learning experience.
You've reminded me why I hate clubs. I'm sorry, no one invites me to them anymore.
I share that with you. Hi, I'm Daniel.
I'm a particle physicist, and every day I rely more on my computer's memory instead.
Of my own.
Yeah, I am Kelly Leader Smith. I'm a biologist, and I'm having a lot more of those moments where you walk into a room and think why am I in here?
I walk around campus here at you see Irvine and a lot of folks go hey, Professor Whitson, and I go hey, and I think I have no idea who you.
Are and why we know each other.
And sometimes it's just because they were in a class I taught with four hundred people, and the relationship is a little axometrical. And sometimes it's just because my memory is terrible, and maybe we hit coffee and I've forgotten, So I apologize to folks out there who I'm pretending to recognize.
Yeah, yeah, no, I'm there also, And every once in a while be standing in the grocery store and I'll stand in front of the aisle for a little too long and my daughter will go, you forgot, didn't you. Yeah, I have no idea what I'm looking for right now. And hopefully this doesn't give us too much anxiety thinking about what the root cause of our memory problems are.
I have that as well. Is that a memory issue or is that a distraction issue? Like where you're looking for capers, but then you see a jar pickles and it makes you think about that last time you had a pickle, and then you're like, m I wonder if you can pickle at home. And then five minutes later you're dreaming about a whole pickling building in your backyard, and you've forgotten that you were looking for capers.
So I don't like pickles. So no, that's not my scenario. But sometimes it'll be like, you know, I see my reflection in a bottle, and I'll be like, oh, is that spot skin cancer? When am I gonna die.
There we go?
Yeah, exactly, but.
It is often distraction instead of forgetfulness. But yeah, sometimes it's hard to disentangle those things.
Well, I just learned something deeply troubling about you that you don't like pickles.
What you know, Daniel, You've made a really great point about your memory, because we have definitely talked to about and I think even on the show, we've talked about this.
That's embarrassing, but not as embarrassing as being closed minded to the wonderful world of pickles. I'm with you, like the big deal pickle, okay with the sandwich, but I'm not munching on one in general. But have you ever done home pickling, you know, like you can pickle cauliflower or carrots.
It's really wonderful.
The only pickled item I've ever enjoyed is Cowboy candy, which which is when you slice of really hot halopanos and put them in like a sugar pickle. Oh so good. That I love on sandwiches. Other than that, I have not met a pickle that I like.
I'm sorry, I have to work on that, all right.
Okay, all right, well, I'll keep my mind open. But today we got a wonderful question from our listener, Simon, who is interested in memory, and so let's go ahead and listen to Simon's question.
Now, Hi, Daniel and Kelly. This is Simon from New York. I was a huge fan of Daniel and joege Explain the Universe. However, I am loving the new show with Kelly and never missed an episode. Here's my question, I think mainly for Kelly. I am now into my seventies and not surprisingly find myself reflecting a lot on my life and the thousands, perhaps millions of memories that go into a lifetime. I have often wondered exactly how the human brain, using biological substances, chemicals, and electricity, actually doors memories. I have tried to read about this, but have not found anything to satisfy my curiosity or wonder Perhaps this blurs too much into the question of what is consciousness and sentience and it's not something you want to delve into. But if you would like to tackle it, I would love to hear about what insights biology and physics have to offer. I have very recently been reading a lot about computers, trying to understand how they really work and that seems somewhat related, but only somewhat anyway, Thanks to both of you for all you do. My weeks would not be complete without listening to your podcast episodes.
All right, Simon, we love your accent.
We do.
I grew up in New Jersey and I miss hearing that accent more often, so that was awesome, And thank you for this fantastic question. And we got really lucky. So folks, remember we talked in the past about whether or not tripped a fan from Turn he actually makes you sleepy on Thanksgiving and we interviewed Mark Mapstone for that and he suggested that if we were interested in memory, we should talk to his colleague, and his colleague was willing to come on the show. So on today's show we have doctor Michael Yassa. He's a professor at the University of California, Irvine, where he's also the director of the Center for the Neurobiology of Learning and Memory and is the director of the UCI Brain Initiative. So like clearly the perfect person for this topic.
Yes, amazing, and also it's one more notch on my personal goal to have my entire neighborhood on the podcast. For those of you who don't know people that you see Irvine. Many of us live in this faculty neighborhood right next to campus, so we're all friends and neighbors and we know each other. And by now we've had a significant fraction of that neighborhood on the podcast. Because I want to know, like, hey, who's an expert flight I'm like, oh, I know that guy who's on the next street. And so it's a fantastic resource.
That is pretty great. So let's go ahead and bring Mike on the show. But we should mention you were off telling people about your amazing research ideas, so you were not able to join us for this interview. So I flew solo with Mike.
Thanks very much for handling this while I was goofing around in the area my pleasure.
I had a blast. All right, Welcome to the show, Mike, Thanks for being with us today.
Thanks for having me.
So what got you interested in studying memory? Let's start by getting to know you of it?
Sure, So, I don't think that I really knew much about memory when I was an undergraduate. I was fascinated by the brain just by virtue of taking a couple of classes that kind of inspired that passion, that love for everything brain related. And one of the things that was really fascinating about it is that I felt, even at the time, this was in the late nineties, that we knew next to nothing. Unlike other classes that I took, where there was sort of a big body of knowledge, it felt like with the brain, there's just so much more that we really didn't understand. So that became really fascinating. As I started to dive a little bit more deeply into different aspects of how the brain functions, memory came about as one that was front and the foremost, particularly because we started to see or I started to see that memory loss is just very devastating, unlike any other cognitive domain that if you know, if you have attentional deficit or if you have a deficit with executive function, you know, that can be somewhat circumscribed. It's a contained kind of deficit, not memory loss, which just utterly devastating, you know, seeing patients with Alzheimer's disease, seeing patients with various forms of memory loss, that was really compelling, and I started to understand a bit more that memory is what makes us who we are. It's so fundamental to our core. It's the essence of our consciousness. Everything that we do we do because of some experience that we've had that we've been able to store, and it just became not just fascinating, but like entirely all consuming. So I focused on memory from all of its aspects. One is trying to understand it's fundamental inner workings, and too trying to understand how it breaks down in a variety of differ and conditions. And if we can do that, maybe we can help people. Yeah.
I've got a neighbor with Alzheimer's and it's been totally devastating for the two of them. So yeah, So this interview was inspired by a question that we got from the listener and one of the things that they asked about is does the concept of memory blur the line with consciousness and sentience? How do you view these concepts?
Yeah, you know, it's interesting. There's a somewhat related question, which is, you know you can have a computer have memory. When we talked about memory and a computing platform in a robot or a botic application, certainly when you think about chat GPT, well that can hold on to memory for some period of time and use that to guide how it responds to the us there and so on. So memory in and of itself may not be the thing that I would say as associated with sentiens. I think that it's the way that our memory works, not just as humans, but as sort of you know, live organisms. It's not like the way that you would do it in a computer. So let me elaborate. Memory is stored in a computer is very much one to one. Everything that you see and learn, your storing with incredibly high fidelity. You try to retrieve it twenty years from now, thirty years from now. It's exactly to say there's no degradation. But that creates a problem for a memory system, and that it's more difficult to extract generalities. It's more difficult to generate knowledge based on memory. But say as a human and you're encoding memories all the time. These memories are stored, but we know that there's blurriness of memories, there's forgetting of memories. There's all sorts of things that we tend to think of as memory problems, but in fact one could argue they're not bugs, they're features of the system. Because memory is not intended to be a super high fidelity kind of system. It's intended to get enough information in so that you can generalize knowledge, so you can learn from experience and be able to guide your future decision making. So, while a computer's memory is really about storage, with high fidelity, you don't want to write a filin word and then store and then later on have an abstract version of it rather than what you actually wrote. You want to have an accurate record of what you actually wrote. But for the brain, what you might want to get later on is that abstract version. It's just enough knowledge to be able to guide your future decision making. And that's the reality of how memory evolved in live organisms. And maybe the thing that makes it very different from nonsensient beings is that it never really evolved to think too much about the past. It evolved almost entirely to think about the future. So the reason why you might store something is because you want to use that knowledge to guide future decision making, to make sure that you do things that are adaptive to promote your survival. You're not going back to the same poison as berry bush. You know, to run from a bear out in the wild, as supposed to go up and say Hello, those kinds of things are based on memory, equipping us to make better predictions for the future.
So can we dig in a little bit more to this trade off? So why can't you remember everything perfectly and generalize when the time is right.
Yeah. So it turns out that you can mathematically and computationally model this and it gives you a pretty straightforward answer. And the way that it works is that if you were to encode every single experience that you have in a very high resolution and high fidelity kind of approach, then it becomes very difficult to generalize that to new situations. The representations in the brain become almost hyper specific. They're very very specific to those instances in which they were encoded. So being able to extract the generalities or the knowledge that you can apply to new things requires that there's enough blur, enough fuzziness across the different instances so you can generalize that knowledge. I'll give you an example. If I were to ask you what is the capital of the United States, you'd have a very very quick answer for me, which is Washington, d Z Exactly. Now, if I ask you, when did you first learn that?
Now don't know.
So now if I had asked you the day right after you first learned that, be in class or maybe from parents, you would have a pretty good memory for it. That's a pretty exciting thing that you just learned. But the reality is you've learned it so many times over so many different exposures. You've heard in a million times in many different settings. And what's most important is not so much the first time you heard it or the last time you heard it, but the fact that you extracted that piece of knowledge, that core piece of knowledge out and now that's part of your body of knowledge. So the specifics around how and when and where we encoded specific things may not be all that important from an evolutionary standpoint to hold on to as the memory for the actual knowledge is important. That's what's going to guide your future action. That's what's important to generalize the new situations. So in some ways there's no evolutionary pressure to hold on to the specifics over time. But it is an interesting point like why can't we have both? Well, you can think about it from an energetic standpoint. If you have a limited resource system, why would you invest your energy into storing the specifics when you know you're not going to use them down the line.
Yeah, no, fair enough. So you mentioned the analogy of the brain as a computer. Is there any analogy over history comparing a brain to something else that you think is a helpful analogy or the brain is just so different it doesn't make sense to try to compare it to anything we have experience with.
So that is a really interesting question. And I have thought about this before, and I keep coming back to the computer as the closest thing, and I always tell people I think it is possible that we will get to a point, perhaps with quantum computing and other types of things, where we might be able to approximate a human brain like functionality in a computer. To this day, we don't have that, even with the incredible advent of AI in large language models and all of those things. Still there are certain things that human brains can do, and mammalian brains in general can do that computers just are not capable of. But there is not another device out there that is capable of this level of processing that I can think think of to associate with that analogy. Particularly. I mean, I think one of the things that the brain does, don't get me wrong, like the computational kinds of things that have been built are incredible, and the amount of the minuscule amount of time that it takes for them to be able to process and provide an answer is just wild beyond the imagination. But there are still some things that brains can do that the computer scans, and in particular it has to do with the ability to extract knowledge, the ability to error correct, the ability to do the kinds of decision making that we do as humans that are very difficult to encapsuling in the computer. The ability to have emotion emotional reactions. Those things are still very difficult to model. While you can tell the computer all you want about what we think the human experience is, we still don't understand that well enough to be able to model it computer.
Yeah, fair enough, all right, So let's pull back to memory a little bit. So we've talked about how memory differs from sentience. Are there different kinds of memory and do our brains store different kinds of manator differently?
Yes, And oftentimes we just say memory as like one big blanket umbrella kind of term, But it is important to know that there are different memory types, different memory systems in the brain, and they serve different functions. So let me give you a couple of examples. One type of memory, which I tend to like quite a bit, we studied quite often in my research laboratory, is what we call episodic memory, memory for episodes, memory for events that happened to us, And we tend to kind of operationally define it as remembering what happened, where it happened, and when it happens, and whenever you have kind of like the collection or the conjunction of those three, you can label that as an episode and you have a memory for a particular episode in your life. Typically you think about that as also the root of our autobiographical memory, our memory for autobiographical experiences, things that happen to us. But that's very different from membering how to tie your shoe races, or how to ride a bicycle, or how to do something like your tennis swing or your golf swing. You know, those kinds of things are trained in the brain very differently. They involve very different systems. A lot of times, they require more trial and error kind of learning, and they tend to be a bit less accessible to consciousness. So there will call sort of implicit kinds of learning so if you look at how the brain is organized, pretty much every patch of cortex is capable of some form of memory. Another term that we typically use in neuroscience is plasticity. The idea that the brain is plastic means it's capable of change. So whenever you have experience, cells that are responding to that experience are capable of change, and that change typically is thought of as a change in the connections and the way that cells communicate with each other, but some change that reflects a record of the experience that you had. Now, those changes happen throughout They can happen in our visual cortex, our visual system, our auditory system. They can happen in the episodic memory systems in the brain, or they can happen in these more implicit memory kinds of systems in the brain that typically support more unconscious function like knowing how to ride a bicycle, like knowing how to swing the golf club and those kinds of things. Those are stored separately. And we know this to be true because we see patients that have deficits in one type of memory and not another, because they have maybe a focal stroke or some damage or some deficit that impacted one system and not the other. So they struggle with that one type of memory that's affected in that system, but everything else seems to be intact.
And by type of memory does that mean like category of memories like the tennis swing and the other sort of muscle memory things, or like I could forget high school if I had a stroke in the right place right.
Although that's actually increasingly difficult. So typically, if there is a stroke that is focal, it might affect motor memory. It might affect memory that allows you to kind of move your hands in the right way and be able to support that kind of function. But episodic memory is this really weird thing. Initially, it does depend on key regions of the brain, one of them being the hippocampus. That's a really important region for episodic memory. But over time memories start to become somewhat independent of the hippi campus they start to become stored elsewhere, and that's one of the reasons why in Alzheimer's disease, where we know the hippocampus is one of the earliest regions to degenerate. As that starts to go away, you see a loss of recent memories things that were recently acquired, maybe weeks, months, or a couple of years before the decline started. But things from long ago, like high school are preserved. And the reason they're preserved is that they've now been consolidated that sort of a technical term for made strength and made resilience to loss. And that's because their stored sort of in parallel throughout the brain. So just the focal deficit there is likely not going to wipe out those particular memories, but it's much more likely to wipe out say, yeah, your ability to have the right swing or ride a bicycle or anything like that. Those are the kinds of things that are much more focally stored.
Okay, and can we dig a little bit more into exactly how the brain stores memories. Is it like how do neurons connect with each other? Yeah, let's dig in.
Yeah. So brain cells are very very unique compared to other cells in the body. And provided this is again I always tell my students, you know, this is true ninety five percent of the time according to our knowledge today. I sometimes things you know, several weeks from now, months, years, things can get revised. So this is just according to our current knowledge today, brain cells are able to communicate with one another in a way that other cells in the body are not able to. And the way that they communicate with each other is using a combination of electricity and chemistry. So the transmission of signals within a brain cell is entirely electrical, and we can talk about that in a second, but the transmission from one cell to the next most of the time is chemical. It involves the release of a neurochemical that goes from one cell, binds to the other, and then initiates another electrical signal from the next cell to the next cell. So it goes really fast electrical, somewhat slower chemical, really fast electrical, someone slower chemical, and so on, and you have this sort of progression of communication between cells. And that's really important because these cells need to bring in signals that essentially encode the outside world and bring that knowledge into the brain to create some sort of representation of it, and then act on them. Allow us to move, allow us to avoid a threat, allow us to seek reward, all of those kinds of things. So that's how the brain typically communicates. But your question is how does memory happen? And there was sort of lots of answers over the years. We used to think, well, maybe it's encoded in the DNA and cells. Maybe it's encoded in the cell size, maybe it's encoded in the whatever else is happening to change cell shape. And the current answer is that it is much more likely to be encoded in the connections in the way that these cells communicate with one another. So let's say sell A is firing and cell B is receiving a signal from cell A. If I were to modify some how the frequency by which sell A communicates, would sell be by making it fire more, or increase the neurotransmitter release, or increase the number of receptors on the second cell that receives that neurotransmitter. I can make it so that the communication between those cells is enhanced. And the way that the analogy that I like to use for this is imagine you are at a club and there's lots and lots of noise going on, lots of things going on, lots of people talking to each other, there's loud music and so on, and you're trying to communicate something to somebody who's dancing alongside you, and it's very, very difficult because of all of that noise. But now you get real close to them and you kind of, you know, hold your hand to their ears and you start to talk directly into their ears. Now they're going to receive that communication with much higher fidelity, be able to tune out noise and selectively attend to that communication. You've enhanced the communication between that person and the person they're talking to. That's what happens at synapses. There's lots and lots of synaptic firing and lots of communication happening. But when cells start to attach to each other, they communicate much more preferentially. They can transmit signals that express that form of learning. In other words, if there's an experience that happens that is learned by the brain, the brain can express a form of plasticity or a form of memory in the strength of the connections. So if the connections grow stronger, that's a signal that this memory has been learned. And most of the information that we have about this comes from animal models, comes from slice recordings, where we can see evidence for enhancements in the connectivity, enhancement in the communication between cells as a result of a learning experience.
One. You've reminded me why I hate clubs. That's sorry, no one invites me to them anymore.
I share that what you do?
Yeah yeah, Okay. So we've got these connections, they get strengthened, but then it feels like there's another step between having this connection and then having a like specific memory. So you know, like we're not at the point where we could even in mice and correct me if I'm wrong about this, where we could like see which neurons are firing together and know they're thinking about food. Yeah, yeah, So where do we go from there?
So what you're talking about actually is a very very big problem that we deal with in neuroscience and in cognitive science, and it's the credit assignment problem. How do you know that a particular cell is assigned to a particular memory or a particular connection it's assigned to a particular memory. And it's a very challenging question and we don't know the answer yet, but we suspect There's been a lot of research on what's called mechanisms of allocation. In other words, how can you allocate particular synapses, brain cell connections and cells to a particular memory and not another right so how can we get the specificity that we need in the system. And there's been a flurry of work in recent years understanding there are certain proteins that are used to quote label synapses, label particular cells, and assign them to a one memory and not another. It's just really brilliant work by some colleagues in the field that is trying to really get at this specificity question. Now we're still early days. We don't have final answers yet, but I think we have some tentitive ideas that you can with specific proteins that are expressed in the synaps essentially label them or prime them to be the ones that are modified by this experience and maybe not another.
Wow, and that's pretty cool, right.
Yeah, because from the looment of the time, we thought, well, how can we ever have any specificity in our brains if memory just activates cells, how do you know which cells are the ones that are involved here? And this might actually re find us with some answers.
Wow. I feel like I recently heard about the connectome project, which I think is trying to figure out all of the neurons. So does this suggest that once we have a connectome next we need to work on the proteome that connects to the connectome to really understand how all of this works, or how helpful is this connect dome project going to.
Be very helpful? I think that every time we try it and map another home it is very helpful. At some point we all have an everything owned and you know, with that sort of large scale data effort, we're going to need also the AI, the machine learning, all of those tools people to pass through it and actually figure out what's going on. But it's interesting, you know, Kelly, When I was coming up as a student, there was always this question that was asked by faculty in my department and many other departments, And it's a theoretical question, and I want to pose that question to you, which is, if we were to map every single neuron in the brain and every single connection in the brain, would we have learned anything about how the brain actually functions? And whenever they asked that question, you know, some of us were attempted to say, well, yeah, of course you can have all the data, right, And the answer they wanted us to get to is no, you're no better off because you have a ton of data but no way to really test hypotheses and understand function. You have to have the right model, You have to have the right kind of strategy to go into that data and look for what's necessary. But I would argue that over the last twenty thirty years that thinking has evolved and now we can go in in a completely unsupervised without having a model, which means also we avoid some of the biases that might come from a model and ask what does the data tell us? Sure, there is an explosion of data, but there are patterns that are hidden in there, and if you train up AI enough, it can pick up on those patterns and maybe tell us that there's a new model. There's a different model. The way that we were thinking about the brain informing hypotheses may not have been right all along. So I go back to things like the connect home projects and trying to resolve the connect home, the proteome, the epigenome, all of those kinds of things as different layers of knowledge about the nervous system. And if we have all of that information, it stands to reason that we should be able to pick up on patterns, and patterns can maybe transform the way that we think about the brain. Maybe we're all along maybe the brain does quantum computing. Maybe all sorts of things that we just would have never imagined that are buried in that data.
It sounds like an exciting time to be in the field.
Oh. Absolutely, absolutely.
I'm going to pull us back a little bit. In our conversation. You were talking about how you can strengthen connections between neurons. Does that happen because you're thinking of the same memory over and over again? And so how do memories get strengthened and how do we lose some of them?
Yeah, So the strengthening of memory, or the idea of consolidation, which is exactly what it sounds like. It's making memories more solidified or more strengthened. It can happen because we are repeatedly rehearsing the memory, so we're thinking about it over and over. But the good news, Kelly, is that that happens completely incidentally, without you intentionally trying to do it. Your brain constantly brings up old memories and thinks about them, and even if you're not consciously aware of it, it happens while you sleep at night. Okay, so the brain is sort of on the back burner, constantly playing through these memories, replaying through these memories, and even when you go to sleep, it's replaying through these memories. So the strengthening act of memories doesn't have to be this intentional thing, which I think is a really powerful thing to tell students. Also, you don't have to sit here and like regurgitate and rehearse everything over and over and over. Just go through and understand and then get a good night's sleep, right, which is a lot of them really struggle to do. But during that period you might think, well, that's just rest. Actually the brain can be quite active during that period of time and can be playing through these memories and storing them and trying to make them morsistant to forgetting now. One thing to note also is that when you replay memories, when you bring back memories, you don't end up with the same thing being stored again. So memories are reconstructive in nature. If I were to bring back an experience, say that I have from a few weeks ago and talk about it with you, what I end up storing now and reinforcing is a somewhat altered version of that memory. It's not the same thing, because memory is reconstructed on piecing together the pieces. Other pieces just get incorporated because we're having a conversation about it. And then later on what I remember is some amalgamation of all of those experiences when I brought it back and changed it ever so slightly. My colleague here at you see, Aravin Beth Loftus built her career on studying false memories and how they arise, and they are very, very frequent. They arise all the time. We generate the more as we get older, and they happen just by virtue of our memory being a reconstructive system rather than a high fidelity video camera or a picture of reality. It's just a sort of a Hodgepodg construction of what that reality might have been. And again we say, why is this happening? Why can't we have a high fidelity version of things? And it's possible that we just don't need to. So even with these false memories arising, they're very artifactual, like you really need to remember exactly what happened and where it happened, that when it happened, Ergie, you just need to remember the core knowledge. So if we focus on, hey, the corenology is being remembered accurately. That's all that matters. Everything else and go to crap, and nobody's going to be less able to survive. So from a survival standpoint, it certainly doesn't matter that you have this strengthening, be a very high fidelity strengthening. It just matters that the core component knowledge is the thing that strengthened, Like the capital of the United States, Washington, everything else. Who cares.
But if the lawyer is you priding you for details in a court case, that's when you're in some trouble.
Absolutely, and you know the good news is, well, the somewhare good news is some lawyers, some federal judges, some jury get the lecture about false memories and understand that when you call witnesses to to stand and you're asking very very specific questions, that their right collection is going to be some combination of what actually happens, what the brain sort of reconstructed it to be, what kinds of questions are being asked, the pressure, any interrogation that happened earlier, All of those things sort of weave their way into that memory. You're never going to be able to get this beautiful, accurate, one hundred percent depiction of what happens, you're going to get some very of it that could be quite a bit more corrupted.
Gosh, there could be a whole podcast on that topic. So my co host wanted me to dig into how we learn about this kind of stuff. So you mentioned that there are animal models, but just focusing on people right now, how good are our techniques for watching how the brain works in living humans to sort of try to get a handle on some of this stuff.
Yeah, So when we started out, the discipline of neuropsychology had very, very poor tools available to it. So you have to develop cognitive assessments, which have come a long way. You can have the right kinds of cognitive assessments and so on, but you really have to work with patients and the clinic who come in presenting with memory problems, with a variety of different conditions, and essentially it's akin to leision studies. You're working with patients who might have circumscribed focal deficit in the brain. You can see that on structural MRI for example, and say this part of the brain is damaged or missing. Therefore they have this kind of function, So you might be able to say something about the function of this part of the brain in a healthy person. But our tools have evolved significantly since, so two major advances that I can tell you are still today. The chief ways by which we study this. One is functional MRI, and we tend to do a heck of a lot of that in my lab. So functional MRI operates on the principle that you can put somebody in the scanner totally intact brain and give them a game to play, or a memory task, or any sort of challenge that would engage the memory bits of their brain, for example. And what you're imaging with functional MRI is not neural activity directly. What you're imaging is blood flow. The idea being that if there's a patch of cortex patchup brain that is more active, that is engaged in this challenge, it's going to require oxygen and glucose and it's going to try to extract that out of the blood flow. So by mapping how much oxygenated blood and deoxygenated blood are going to different areas in the brain, you can generate a contrast because it turns out that the degree of oxygenation has a different magnetic signal. So that's sort of a little hack that we pull in MRI because we're changing ragnetic fields. So by measuring that contrast, we can get an indirect proxy to where neural activity might be by virtue of that blood flow change. So that's been a really, really helpful technique since the late nineties early two thousands. In the early days of functional MRI, people did a bunch of like really just awful studies because the technology was new and we didn't know what to do with it, and folks didn't really think beyond you know, the X marks the spot kind of approach, Right, I want to know what the fill in the blank part of the brain is. It even got as absurd as I want to know what the god part of the brain is. Right, So people start to do those kinds of studies, try to go in and say, X marks the spot, where's the stuff happening?
Is this the same system where they had that dead trout? Oh?
Yeah, you know that that trout study. Of course. So at the end of the day, it is a statistic approach to comparing activation, and yeah, that study is really compelling because you could show that you see activation essentially in something that is dead, and people every now and then will kind of make fun of this, and remember the old days of fMRI, when folks didn't really know what they were doing, and you could get something like this right, and in some cases you can even get it to be published, which is crazy when you think about it. But we've come a long way since. So the beauty of functional MRI now is that one we understand how to do the X marks the spot much much better. We now have much better handle on this historical challenges, the way to build the right contrasts, the way to correct for multiple comparisons, all the things that you tend to think of when you're doing large scale statistics. That discipline had to kind of come to functional imaging and inform it and that has happened, which is great. But the second part is I told you before that memory is all about the connections, and functional MRI initially was all about blobbology, right, try to find little hotspots in the brain. You pretty pictures the cover of science and all that with little hotspots in the brain, and that was the approach. But we know that memory is not in the hotspots, memories and the connections, so we've started to move much more towards connectivity analysis and asking about how are the different parts of the brain communicating dynamically and essentially coactivating with each other to support solving this challenge or doing this memory test or memory game while you're in the scanner. So that was the advent of functional connectivity kinds of approaches which are I think far more compelling, far more robust against some of the initial critiques of a functional MRI, and they reflect the true nature of how the brain works. The brain is one big dynamical system. It's not just regions working in isolation. Everything is connecting with each other, so we owe it to ourselves to try to understand it from that much more complex way. So functional MRI still remains a very very powerful tool, but now the analysis that we can do are just far more advanced. The other thing that I think is just incredibly powerful, aside from anim models, is the ability to directly record electrical activity from cells or from what it's called local field potentials that the areas around cells that have also electrical activity. It can be measured directly in patients, and these are typically patients that are going to undergo surgery for epileptic seizures. So the surgery is done to remove the part of the brain where the seizures are emanating from. And typically when they come into a hospital, they're in a hospital for about a week or so. They get implanted with electrodes to record from the parts of the brain where the clinician might suspect that epilepsy is happening, and they're taken off of anti epileptic medication and essentially they're waiting to induce a seizure, and once that seizure is induced, they attract the location. That's how they decide on a way to do the surgery. There's about a week or so while they're in the hospital with electrodes penetrating deep into their cortex and many of those electrodes directly into the memory bits of the brain, like the hippocampus. And those are just an incredible group of individuals because they also most of them want to help science and they understand the opportunity the scientists have while they're laying in a hospital bed for about a week to understand something fundamental about the brain. So every now and then they give us the opportunity to give them a challenge, maybe on an iPad or a computer while they're laying there and they try to solve this challenge, try to play this memory game or do this memory test while we're recording direct electrical activity from their brain cells, which is incredible. So it gives us almost the same degree of information that you can get in an animal model. Now in an animal model, in a road that you can stick more electrones, you can get higher fidelity. And with patients you have to do things that are only clinically warranted. So there's an ethical obligation, of course to make sure that nothing is being done that would ever put the patient an increase risk. So that also poses some limitations as to how you can record activity and get that data. But it's just incredible access that we have to the brain in partnership with these remarkable individuals, and we've learned a heck of a lot about how the brain works and how memory works from those direct electrical recordings.
Wow. I previously wrote a chapter on brain computer interfaces and I was reading about like, utah arrays. Is this the same thing or is this a different kind of electrode.
Yes, So, utah arrays are one way to do it. Utah arrays are a little bit more invasive. They involve several electrodes that are kind of going through the surface of the cortex. At the same time, they're no longer kind of the standard practice for most patients. They're still used in some cases where they're clinically warranted, but in many cases they're not because you suspect that what's happening is deep into the brain, so you stick direct single electrodes all the way down to where you suspect the action might be, and you avoid some of the potential damage that happens with UTAH rays. So they're used in some clinical contexts, but in many others, we can stick these much thinner, slimmer electrodes directly into the parts of the brain that we suspect yet epilepsy is ennything from so far less damage that way, and those patients typically have better outcomes than patients implanted with guitar rays. Now, your point about BCIs, and there's a number of companies out there that are trying to develop brain computer interfaces using these kinds of arrays. I think that's a particular challenge for those enterprises is how do you create a way to measure directly from the brain and to be able to stimulate and influence the brain without causing too much damage. Having electrones that are thin enough, that are made from the right material so that you don't cause a lot of tissue damage, because ideally, what you want to do is create an interface that helps people, so you don't want to inadvertently cause more damage.
Now, when we were thinking about UTAH rays and damage, you know, we thought about like a cup with jello in it, and you stick some needles in there, and as you move the jello around, if the needles are kind of staying in place, that would sort of mess up the brain. Is that a good way to think about it? Does it all move together?
Yeah? The brain certainly is as vulnerable maybe as a cup of gello. But the key is also flexibility. So you're right. When you have these electrodes, there's a bit of a compromise. So want them to be flexible so that they're moving with their brain. You're absolutely right, and that'll cause less tissue damage. But at the same time, flexibility you can come at a cost, which is what they're targeting, might change. So you want to make them flexible enough sort of they don't cause damage, but rigid enough so that they can continue to target the same region, so it is not an easy challenge at all. But there's been some developments recently in doing these kinds of arrays with animals, and we haven't yet pourted that over to humans and done the FD approval and all of those kinds of things. It's happening soon. There's already experiments that try to test out one of the technologies like neuropixel for example, or near epixels technologies. Those have been incredibly powerful for animal models for reading from road insight from non human primates and their small form factor. They're thinner, but they have a ton of electrode contacts on there, so it can really give you information from a lot of different sounds simultaneou sleep. And pointing that over to humans I think will be a really helpful thing to do. But that's only been done in some limited experiments and not widespread. You so hoping that some variants of those kinds of technologies will make us way to primetime soon.
I've watched some videos of people with brain computer interfaces that were able to do incredible things, But one of the things that sounded totally devastating to me was if I understand this correctly, it's over time, the brain has a response to those electrodes and like kind of walls them off and the connection gets less good. I don't know exactly what's happening. Do we have any progress in that area.
Well, so that's another thing that needs to be tackled. Also, with some of these newer silicon probes, they're less likely to have the inflammatory and the calcification kinds of responses that happened around electrodes, because remember, this is a foreign object entering the brain, and the brain's natural disposition towards foreign objects is attack it, right. That's why we have brains immune cells, we have microglia, we have a lot of cells that are dedicated to detecting and eliminating foreign objects. So you tend to see them of aggregate around electrodic contact locations and things like that. But there are ways with different substances to kind of maybe fool the brain a little bit into thinking this is okay. You can try to also reduce the brains immune response to some extent when these things are coming in. So there's approaches that are being developed to try to get better long term outcomes, but yet we're still very early.
In this game, And for listeners who are maybe not es familiar with brain computer interfaces, what are some reasons that people might get a brain computer interface.
There's a variety of reasons. So, for example, for someone who has lost the ability to control their limbs because of a stroke or a focal deficit, being able to have a brain computer interface shortcut signals so that they can still control their limbs and their body is remarkable. And patients who have had those kinds of approaches, it is just life changing. They go from a paraplegic or quadriplegic to being able to have use of their arms or their legs again. So there's incredibly utility there. For folks who might have epilepsy. For example, there is a brain computer interface that is a stimulator that is implanted in the brain that responds to the earliest science of epileptic seizures and that is able to with electrical stimulation essentially knock it out. So now instead of having to have the person be going in for surgery and lomping up parts of the brain or having them be devastated by epileptic seizures, you can have an implanted BCI that responds in a closed loop system, so it uses the responses of the brain itself to tell the stimulator what to do. And that's a completely closed loop, so it doesn't require any user interventionally outside. That allows it to do a much better job of helping the patients overcome seizures or epilepsy. So there's a number of different uses. I can imagine that for movement disorders, for a variety of different conditions where you might want to implant something that communicates with the brain and feeds at the right signal at the right time, there's going to be a huge use for a BCI. Then there's a whole other class of uses that may not work vire implantation. Right, So these external devices maybe devices that are communicating with the brain and external fashion portable or multiple, and they allow it to improve function for stroke rahabilitation, or improve function in some other way. There's a lot of folks also kind of you know, taking the transhumanist approach here and trying to develop PCIs to just improve our function. We just want to be better at something, right, So what if I can control this robotic arm to do some you know, whatever it is. So don't care too much about those but certainly the utility for helping patients is huge.
So getting a little more sci fi here, do you think we'll ever be able to know what somebody is thinking by having, you know, a cap on their head or electrodes in their brain or is that just way too far off?
I think we already do so. I think we are to some extent with an electrocapsule. Egen is really the technology that you're talking about, or there's other ways to do it also, various ultrasound technologies and so on. You can detect and stimulate pretty easily, but the problem is the kinds of things that you can get the system to do are still fairly rudimentary. So I can tell, for example, based on eg signal whether the person is going to move their hand or make some overt kind of gesture, and motor control is a somewhat simpler system than like memory, executive decision, or emotion or having very complex feelings like guilt, right or things like that are much more difficult to capture in the simple signals that we're capturing. The BCIs right now. So do I anticipate that at one point we'll be able to do one percent? There's no doubt the technologies there. It's just a matter of are we recording from enough are we able to build the sophisticated models to model this function, and we're making rapid progress in that arena. So from a sci fi perspective, I'm sure you've heard this before. The moving vehicle, the car was sci fi at one point until someone invented it, right, So the things that we think about a sci fi are just science that's not here yet.
All right, Yeah, the future is now. So let's talk about losing memory a little bit as we wrap things up. So we talked about how your brain is working overnight to strengthen memories. How do we lose memories over time?
So there's a number of different ways to do this. Memory can be lost because of decay, so just the passage of time a lot of times can make memories just harder to remember, harder to access. And we've known this since eighteen eighties. Herman Ebbinghouse was the first to kind of do this using experiments with himself. You would learn lists of nonsense syllables and then try to map his own forgetting curve. So how much forgetting happens over the first twenty four hours, next twenty four hours, next twenty four hours. Yeah, I know, experiments on yourself very very boring times in eighteen eighty so it didn't have too much else to do, so I did this with himself, but maps what's called the forgetting curve, which we still to this day. We can look at any sort of memory function or any memory task that we do in the lot and we see a very clear forgetting curve. A lot of forget the first twenty four hours. Then things to kind of taper off. So there's that, And the suspicion is that mostly it's decay, but sometimes it happens because of interference, because similar memories get in there and kind of interfere with one another, compete with one another, sort of the memory that you have for them get a little bit fuzzy. And again these are maybe features, not bugs, right where maybe the system is not intending to hold onto memories with high fidelity for long. So interference and decay help us extract what's most important and keep that over time, and then other things can kind of go away. So those are natural things that happen in every brain all the time, and there's nothing to be concerned about decay, interference, nothing to be concerned about. But as we get older, and by older I hate to say this, you know, talking about like forties.
And above that what I wanted to hear money, Well, I'm going to.
Say graduate in our fourth decade and then maybe a little bit more precipitously, over time, memory does get more difficult. So things do degrade, and we start to maybe lose to some extent, our ability to make new memories, encode new memories, our memories. For the old stuff is still there's still resilience, even though every now and then we might have a problem with access. So we get distracted, we lose retrieval cues, but you know it's there because if you get the right reminder, boom, it comes back. Right. It's just a matter of like tip of a tongue, you know, being able to remember exactly that the right queue for retrieval. That becomes more difficult when we're just more distracted. But as we get older, making new memories becomes harder. And then for some folks who might go down the trajectory to Alzheimer's disease, right then that becomes exceptionally more difficult. And that's one of the first things to go. And there's a very difficult line between what's normal I hate to say that word, maybe more typical age associated memory impairment, which you can expect in every brain, and that's something that may be a bit more costs for concern because it may be going down to about the Alzheimer's disease. Dissociating those two, say in the sixties and seventies, is actually very difficult. It's not very easy because both can start out as a form of forgetfulness. But with Alzheimer's disease or with dementia, it's very progressive, so it does get worse and worse and worse over time. That change in a healthy aging brain or a typical aging brain is far less deep, so you don't see too much changing over time. You don't see this degradation to the point where it becomes very noticeable by family, friends, neighbors, and so on. So those are the forms of memory loss or memory change that happened. Some totally innocuous, no cause for concern, they happen every day to everyone, to the best of us. And then some that are a bit more cause for concern.
As we get older and mechanistically, is it just the messages are not getting sent between those anymore, or this is.
Something we've done quite a bit of work on. Actually mechanistically, what seems to be the case is that first the part of the brain that's really important for encoding these episodic memories, the hippocampus, as I said, has a very interesting change and it's dynamic. So this is a massive information processing hub in the brain. Even though it's a small structure, it carries a big information processing load, and it kind of can shift its state from encoding new information to remembering old information. And there are a few changes that happen to the cells in that system as we get older that bias the system towards remembering old information and away from encoding new information. And there's lots of sort of reasons why that is, we tend to change the excitation inhibition balance, we tend to change neurotransmitter concentrations, all of those kinds of things as we get older that cause that change in the information processing balance. But the other thing that happens more in the context of Alzheimer's disease is you also start to deprive the hippocampus of its main input coming in from the rest of the brain. So there's a region that sits alongside the hippo campus called the antorhinal cortex, and that region shrivels up in Alzheimer's disease. It's one of the first regions to deposit what's called tangle pathology or tau tangles, and that's a marker of cell death. So there's massive cell loss that's happening in that's around cortex early on, which deprives the hippocampus of its principal inputs. So in computer science we have this old adage called garbage in garbage out, and that's essentially what's happening in the hippocampus. It's the quality of the information that's coming in starts to become much more degraded than the context of Alzheimer's c's, So we can't possibly expect it to do a good job encoding information and story with high fidelity if the information coming in is in some ways quote garbage. So that tends to be one of the things that is also mechanistically associated with memory loss in the aging brain.
Do we know why that region starts to degree in some people and not others.
You know, there's hypotheses and I have my pet hypothesies. In the field has its pet hypotheses and so on, But there's a variety of contributors. We suspect that in the context of Alzheimer's disease, it's a combination of inflammatory changes, so that the mirror immune system is disregulated in Alzheimer's disease, where the inflammatory response that normally is very healthy starts to kind of take a turn and become very pathological. There is vascular damage that happens as we get older, and we suspect that some of those early vascular insults, so related to blood flows, deifenning of the arteries and so on, can also contribute to that. There's metabolic regulation that changes, so the metabolic demands of different regions also becomes disregulated as we get older. And then one thing that we've worked quite a bit on is excitation inhibition balance and the notion that normally the brain keeps this dynamic beautifully between stop signals and ghost signals, and as we get older there actually is an overabunton of the ghost signals which can drive the brain pathologically more towards memory loss and not enough of the stop signals, so that imbalance can also be a contributor. So it's a multifactorial issue. There's lots of contributors. This is not just one single pathology kind of model, as much as some of the field likes to believe that. It's a bit more complex.
So as someone who just learned that they're in the old category, I.
Would necessarily call it that. I would just say, we're maybe done developing and we're now just kind of hitting that hump of we're going to start the aging process.
Got it, got it, I'll think of it that way. But I like that better. What kinds of science based things can we do to maintain our memory? So I see all these apps, pitch me, is there any science behind any of those things?
Well, the app stuff probably not. I will tell you that the four big things that are really really important, and when I say really important, I mean they are supported both by epideological data and by clinical trials. So these things are really really helpful. The first one is physical activity. We know that physical activity maintains brain health well into older adulthood. We know that it can delay the onset of Alzheimer's disease. It can make the outcomes better for patients and as protective it really is knocking out. So sedentary lifestyle is a risk factor. We've done work also to show that even activity as brief as ten minutes of walking can be helpful to memory. So it doesn't take a lot, you know, I would say thirty minutes of you know, mild to moderate activity like walking, risk walking, that's sufficient to be able to give people a way to handle that risk factory. The second thing that's really important, and this kind of goes back to your idea about apps and kind of cognitive engagement. It turns out that apps and brain games on all of that efficacy is a little bit tendless, so there's conflicting reports. But we know that social engagement is really important. So in other words, if you remove social engagement, if folks become isolated the same past retirement, that's a risk factory for sure. But if they're able to continue to be social build archer social networks in person, you know, volunteering, community centers, churches, synagogues, whatever it is, or being around people in general, and that can also combine with physical activity. So let's say it's dance class now, it's physical activity and social contact. Right, those kinds of things, even in interventional studies, have been shown to have very positive results. Then the third piece is diet. A heart healthy diet is a brain healthy diet. And the one that has been tried and true in clinical trials is the Mediterranean diet. So the Mediterranean which has lots and lots of variants, but I think very colorful, leafy, green vegetables and so on, healthy fats, and reducing you know, things like red meat and some on. So that one also has been tried in clinical trials compared to other diets and seems to be able to stave off risk. And then the last piece, which I think is one of the most important, to sleep. We all need good quality and quantity of sleep every night. It turns out that actually during sleep we go through a process of glymphatic clearance and we clear out on the life lot of the pathologies that can lead down the path to Alzheimer's. These or at least be contributors to it. And studies have shown that if you have sleep loss folks for example, who sleep less than six hours a night versus those who sleep more than seven or eight hours there's differences in their amyloid uptakes, so that emiloid pathology is one of the chief pathology in Alzheimer's disease. You see a lot more of it in those who sleep those fewer hours, and a lot less of it and those who sleep longer. So sleep disruption, sleep loss, that's a risk factor. The good news is most sleep problems are treatable, whether it's because of obstructive sleep apnea, or insomnia or any other reason, restless slag, all of those things. There's good treatments out there that will help people sleep better. So those are the four sort of chief things. There's many other smaller things, but those are the four ones that I like to lead with because they have just excellent data in their favor.
Well, I'm excited about the sleep thing. I like sleeping. Maybe it's time to get out and exercise a little more so. My co host is a physicist. He always ends on a aliens. So here we go. If an alien were to land on Earth today, do you think they would store memories in a similar way.
You know, if you had asked me that question ten years ago, I would have said, yes, I think our memory system is exceptional. It's wonderful, it's brilliant. Why not they should be like a model to strive to achieve. But I will take the opportunity since we've got a couple more minutes, Kelly, and tell you about something that changes my answer to this question. And that is the discovery that was made first by James Magaw who is the founding director of my center here at uc Aervine, and we continue to do work with this group of remarkable individuals who have what's called highly superior autobiographical memory. We spent a lot of time during this conversation talking about how memory is fallible. There's forgetting, there's interference. It's not meant to store everything with high fidelity because that's going to compromise your knowledge generation and so on. Well, these folks would beg to differ. And it's incredible because they do store things with high fidelity. They can remember everything that happened in our lives since that they were teenagers and tell you exactly what happened on what day of the week, what month, and someone. And they don't have a problem extracting generalities and knowledge. Also, So sometimes I'll jup with them and say I think you're like the X men of our generation, X people of our generation. It's remarkable and we still don't understand how they do it and how their brains are wired differently. So if aliens sufficiently advanced to aliens, I think, if they're reaching Earth before we reach them, they're probably far more advanced than us. They're more likely to have figure it out a way to do that which sort of combines the best of what we have built into our computers and what we have built into our brains. Just no compromise, no sacrifice.
Awesome. Well, I wish Daniel we're here to hear that, but I'm sure he'll enjoy hearing that explanation when he gets back. Thank you so much for your time, Mike. This was absolutely fascinating.
They're very welcome. I very much enjoyed to Kelly. Thank you.
Daniel and Kelly's Extraordinary Universe is produced by iHeartRadio. We would love to hear from you, We really would.
We want to know what questions you have about this Extraordinary Universe.
We want to know your thoughts on recent shows, suggestions for future shows. If you contact us, we will get back to you.
We really mean it.
We answer every message, email us at Questions at Danielandkelly dot org, or.
You can find us on social media. We have accounts on x, Instagram, Blue Sky and on all of those platforms. You can find us at D and K Universe.
Don't be shy, write to us
