Your brain is an incredibly complex network. The firing circuits of your brain and nervous system shape everything from what you’ll eat for breakfast, to keeping your heart beating while you sleep. But what happens when the firing of those circuits goes awry? Well, researchers are learning how to fix them: because if we better understand the circuit, maybe we can change the outcome.
Your brain is an incredibly complex network. The firing circuits of your brain and nervous system shape everything from what you’ll eat for breakfast, to keeping your heart beating while you sleep.
But what happens when the firing of those circuits goes awry? Well, researchers are learning how to fix them: whether it’s using vagus nerve stimulation to safeguard people with epilepsy from the worst outcomes, or exploring the hidden machinery behind confidence, decision-making, and how dopamine shapes the way we learn - there’s a lot we still don’t know about these networks. The research spans different fields and specialties, but shares a strong common thread: if we better understand the circuit, maybe we can change the outcome.
Featured guests:
Dr. Ana Suller Marti, Neurologist at London Health Sciences Centre and Associate Professor of Neurology at Western University.
Dr. Paul Masset, Associate Professor in the Department of Psychology at McGill University and Associate Academic Member at Mila Quebec AI Institute.
This work is supported by The Catherine and Maxwell Meighen Foundation and the Azrieli Foundation.
[theme music]
Dr. Ana Suller Marti 00:06
My heart was trying to take care of patients with epilepsy who unfortunately don’t respond to medications, and beyond medications options we have the stimulation. And that just brings me here almost 10 years later when I started looking into that, “How many questions I had? How many questions I could not find anywhere?” So, I thought it was an excellent path.
Fiona 00:35
This is Bold Minds: Future Leaders in Canadian Brain Research. I’m your host Fiona Sanderson. I work at Brain Canada, where our mission is to bring together funders and researchers to enable health innovations for Canadians. The Future Leaders program is made possible thanks to an anchor gift from the Azrieli Foundation and matched by Brain Canada through the Canada Brain Research Fund. Come along with me as we journey into the bold minds and labs of researchers who are redefining our understanding of the brain. [music ends]
[rousing music] Your brain is an incredibly complex network. The firing circuits of your brain and nervous system shape everything from what you’ll eat for breakfast to keeping your heart beating while you sleep. Today we’re zooming in on what happens when those circuits are activated and when they go awry and how are researchers learning to fix them. Whether it’s using vagus nerve stimulation to safeguard people with epilepsy from the worst outcomes or exploring the hidden machinery behind confidence, decision-making and how dopamine shapes the way we learn, there’s still a lot we don’t know about these networks.
I’m joined by two researchers at the cutting edge of brain science in two very different arenas but sharing one powerful idea; if we better understand the circuit maybe we can change the outcome. So, let’s get into it. This is Bold Minds. [music ends]
Today I’m joined by….
Dr. Ana Suller Marti 02:09
Ana Suller Marti. I’m an epileptologist and associate professor at Western University in London, Ontario.
Fiona 02:16
Her team uses targeted brain stimulation therapy along with brain scans and electrical recordings to treat people with epilepsy, improve their health outcomes and understand the neurobiology behind these processes.
[whooshing] I’m also joined by….
Paul Masset 02:32
Paul Masset. I’m an assistant professor at McGill University in the Department of Psychology and an associate academic member at Mila-Quebec AI Institute.
Fiona 02:42
His team is interested in understanding what happens in our brains when we sense our environment, make decisions, gain confidence and process rewards or deterrence. They use rodent models’ complex behavioral tasks and brain recordings together with AI to understand how differences in the brain circuitry contribute to cognition and health and disease. Ana, Paul, welcome to Bold Minds.
Paul Masset 03:06
Thank you.
Dr. Ana Suller Marti 03:07
Thank you.
[theme music]
Fiona 03:11
So, Ana, let’s get started with you. You study epilepsy and about a third of people who have epilepsy have a drug-resistant form, meaning their seizures are not effectively treated with a currently used drug. So, can you walk us through exactly what’s happening in the brain when someone’s having a seizure? What are the long-term implications or risks for someone with untreated epilepsy?
Dr. Ana Suller Marti 03:33
So, epilepsy can be simplified as an abnormal electrical activity in the brain, and this interferes with cognitions, in mood, autonomic functions during the seizures, of course, and also in between the seizures. Furthermore, there is the chance possibility that they suffer what we call sudden unexpected death in epilepsy. It is a fatal complication called SUDEP. So, we need to be extremely careful and diligent to manage our patients. That’s the reason we offer different types of strategies, including neurostimulation as one of my main areas of research. It’s a technique that applies electrical stimulations to some different parts of the body, mainly in the nervous system. It could be in the vagus nerve or inside the brain. So, we are using some of those techniques to try to modulate the seizure activity, to try to reduce this abnormal electrical activity that is happening. Epilepsy is a disease that presents as recurrent seizures, but it’s associated with changes in neurobiology, psychological, social, and cognitive. So, what it means is in patients with epilepsy, they can struggle with other things like maybe memory deficits, with mood disorders like depression, anxiety, or even sleep problems or feeling isolated. So, all these, what we call comorbidities, directly impact to the life of patients with epilepsy. At some point, some of the patients even refer that it’s even worse having severe depression than having even recurrent seizures. So, that’s something that most of the efforts of the epilepsy community have been focused in, seizure control, and has been kind of ignoring all these other conditions going on in our patients. In the most recent years, when the epilepsy group has started to think, “Oh, we should start paying attention to the other things and how we can help.” So, not easy tasks, but we are trying our best to improve lives.
Fiona 05:44
So, you mentioned that dysregulation of the autonomic nervous system leads to an increased risk of sudden death in epilepsy or SUDEP. So, could you walk us through this? Like what is the autonomic nervous system? What does it mean when it’s dysregulated? What can be done about that for patients?
Dr. Ana Suller Marti 06:00
So, this is a brand-new area of research that we started exploring through collaborations with other scientists. And, of course, as we know that the brain is connected and regulates different organs or all the organs of the body and how these organs may respond differently to our brain. So, the implications still it’s a chapter that I’m trying to understand better, but perhaps that could affect the response to, for example, the heart response or the breathing response. And, of course, after a seizure, many of these organs get affected. So, the idea is to understand if normalizing the autonomic nervous system, we will be reducing the risk of dying after one of these seizures. So, it’s not an easy chapter, but I hope that through some of these research we can answer more of your questions.
Fiona 06:59
So, Paul, you’re also looking at a sort of cascading of neural signals through circuits, but this time from one part of the brain to another and for a very different purpose that has to do with dopamine learning and confidence in our decision making. Tell us a bit about your research.
Paul Masset 07:15
So, my research is sort of quite diverse. It sort of bridges the computational neuroscience part where we’re sort of thinking about, what are the computations that brain circuits are doing and the experimental, you know, systems in neuroscience? Which is let’s have, in our case an animal model, do a task, see what the neurons are doing while the behavior is occurring. And it’s trying to bridge that gap to, what is a circuit? What are the neurons doing while there’s a behavior? To thinking about, what are the computations and what are they doing while this behavior happens?
And one of the key sort of directions in the lab is thinking about, what is the role of dopamine in learning, as you mentioned? And dopamine is very important because a large amount of psychopathologies and also some neurodegenerative diseases, right? Like Parkinson’s disease are basically linked to dysregulation of dopamine signaling. Basically, dopamine signaling is not occurring as it should, and that leads to disease. But it leads to very different sort of issues. And so, one of the key questions in the lab is what is sort of the function of the heterogeneity, right? You have this neurotransmitter dopamine, and we’re sort of really interested in how it acts in a part of the brain called the striatum, where dopamine is sort of released there and it helps modulate the synapses and basically learn associations between events. But even though the striatum is sort of one structure, it is not homogeneous. It has different inputs. Different parts of the striatum gets input from different parts of the brain and different parts of cortex. And the question is, how does dysregulation in one part versus the other leads to different symptoms? Right? Why is disrupting this specific part of the pathway leading to this specific type of sort of symptoms and disease? And that will help us understand basically what goes wrong in that disease. And so, it can come from different origins, right? For example, in Parkinson’s disease, it’s really the neurons that sort of produce the dopamine and release the dopamine in the striatum. These neurons themselves die, so obviously there’s less dopamine around. And for example, one of the sort of therapies is to sort of give dopamine precursor. And so, you increase all the levels of dopamine and so you improve on the symptoms, but you also have side effects. Because you didn’t just change dopamine back to sort of vaguely normal levels in the pathway you’re interested, you also increased it in other areas where it was not disrupted. And so now it is disrupted in some ways and so that leads to other side effects, right? And so, understanding basically how the same circuit, the same architecture, the same types of neurons do completely different things depending on, for example, the inputs they’re getting, right? So, one of the ideas, for example, is that you’re getting inputs and you’re learning about something that is motor, right? And so, you can think in Parkinson’s, you’re learning about something that is motor actions. You can have the same architecture, the same way of learning. But now if your inputs are not motor, they’re, for example, sensory, then you could have another problem and you have issues about—learning about sensory things.
And it’s thought—there’s starting to be evidence that part of the striatum that is more linked to sensory areas might be involved in, for example, hallucination like behaviors in schizophrenia. Right? And so, you have the same dysregulation at the local level, but because the circuit is in a different place in the brain it just leads to completely different outcomes. And so, understanding the fundamental computation might be absolutely key to understand how to treat these super different sort of pathologies as not one, but at least having some common principles.
Fiona 10:59
That’s interesting. So, there are a lot of different characteristics or personality traits that are commonly ascribed to people, and I think many can be boiled down to our decision making. So, things like people who are risk takers versus risk averse ,or those that crave a short-term reward versus those that like a slow burn or a long-term satisfaction. So, are these things all related to the way our brains process dopamine? Are they more or less hardwired into our brains?
Paul Masset 11:25
So, whether they’re hardwired, that’s a very difficult question to answer. Another type of diversity that I uncovered in my postdoctoral work is, like, looking at these dopamine neurons, we identified that different dopamine neurons actually do this learning, and they basically compute the error that is needed for learning, right? The discrepancy between what you expect and what you get, and that’s what the dopamine neuron computes. We showed that this error signal is computed at different time scales. Because basically the neurons, they compute the value of things or the signal they’re computing about over some time horizons. Like how good are things in the near future? Right? But what is near? Near has very different definitions. Can be the next second, the next 10 seconds or something like that. And so, what we identified is that different neurons have different time scales. And so, from that, you have this very tempting hypothesis to say that, “Oh, okay. If you have more short time scale neurons, you’re going to be sort of impulsive because you’re going to only learn about short time scales thing. If you have long time scale neurons, it’s going to be more long term.” So, this is, like, a very tempting hypothesis, but it’s sort of a working hypothesis at the moment. But we do see that the neurons are doing this. So, there’s starting to be this mechanism that potentially could implement this sort of things. And there is ideas that some of these things like, you know, what you’re talking about, time interval, the idea of temporal discounting. How much do you devalue things in the future, does change in addictive behavior, for example. And so, if you take sort of cohorts of patients that have a specific addiction versus not, they tend to potentially have sharper discounting. They’re sort of more myopic in the temporal horizon. They think about more near term things. The question is [about 13:11] the chicken and egg, right? Are they addicted because they have this thing and that they cannot sort of plan ahead and see that it’s not good to do something because it’s going to lead to sort of issues in the future, or is it the addiction that created that? There’s still very much open questions, but there’s starting to be ways to measure this and potentially quantify this in a way that is quite sort of, I want to say, computational and we have sort of a grounded theory of, “Why would you even do this sort of behavior?”
Fiona 13:38
I’m curious, when someone has a seizure or has a series of seizures over their lifetime, so this flurry of electrical activity that’s happening in the brain, does that change the neurotransmitter milieu? Like when you have all of these synaptic connections being activated, what happens there?
Dr. Ana Suller Marti 13:55
To learn that, it’s a bit tricky because we would like to see over time. And one of the ways that we can assess some of those things is using MRIs. And we can see that over time some areas may shrink or become more atrophic than other areas. And, of course, having less tissue may be associated with less function in some of those areas. It doesn’t mean that patients who suffer seizures through the entire life will have an atrophic brain. But in some cases, we have seen in some specific conditions like mesial temporal sclerosis, other specific types that you can see a change. And of course, that, as I say, impacts function and the function is likely related with some neurotransmitters change. So—but again, we cannot generalize as epilepsy is a very heterogeneous disease. And from presentation, like as you say, you may have two seizures in your lifetime or you may have tons of seizures through your life, so that’s very different scenarios and as well as the response to medication and other comorbidities or the reason why you have epilepsy could be different and that could accelerate some of those processes or maybe and nothing changes. That’s one of the biggest challenges, this complexity in the disease presentations, evolution. So, that’s the reason we always are looking for more answers and more options for our patients.
Paul Masset 15:23
I think one thing you pointed out is—and I think is very important, in neuroscience is the idea of heterogeneity, right? It’s like these things, we classify them as the same, but they’re very, very different, right? Like if you think in sort of more molecular biology, right? Like cancers, you have a cancer of something, they’re actually very different. And we understand a bit better there because we can sequence how different they are. Right? But in neuroscience, we’re still quite far from that. I mean, we’re getting there, but we’re still far and we still classify as one thing things that are intrinsically different. They just manifest themselves in potentially the same way, but they’re like intrinsically very, very different.
Dr. Ana Suller Marti 16:02
Yeah, I completely agree that in epilepsy at least it’s having one single model doesn’t work. You can have one single model with one type of epilepsy and despite that two patients with the similar presentations will have very different evolutions, different response to medication. So again, it’s a bit challenging, this field.
Fiona 16:25
So, you’re both study neural circuitry that has distal effects, so meaning something happening in one spot that has an effect some distance away, be it all the way from the gut or the heart to the brain or from one part of the brain to another. Does understanding these far-reaching effects, both literally and metaphorically, hold promise for the future of non-invasive modulation therapeutics? So, you know, that could be to control a brain disorder or even just to boost how we learn and form memories?
Dr. Ana Suller Marti 16:54
Again, the dimension of these therapies is still unknown. And part of my research is—of course, in this project was to assess the autonomic impact of the vagus nerve stimulation. But in other areas of research that I’m doing is looking into other comorbidities very frequent in people living with epilepsy, so it could be from mental health to sleep disorders. And of course, this goes back to the initial questions about how the autonomic nervous system maybe modulates all of these functions. And it’s something that we cannot try to put in compartments. We focus in mental health, we focus in sleep, but perhaps there is an interrelation with all these functions and how this could be modulated. So, how then at least I see neuromodulation has been proposed, for example, vagus nerve stimulation for—in clinical settings for different conditions, like from severe depression, from obesity. It’s hard to puzzle how this same function works in so many different diseases, but perhaps there is some route that it’s all indirectly responding to the same way. So, I think we still don’t have the right answer. We hope to have it in the future. But the interconnections and the distal effect, it’s obvious that we are seeing.
Paul Masset 18:17
Yeah. So, I think the key thing there is the invasive versus non-invasive. Right? So, I think specifically if you think about dopamine signaling, for Parkinson’s disease you have sort of deep brain stimulation, so it’s invasive. But you are modulating basically those nuclei and then it’s, like, making the effects on the learning. You’re sort of boosting the activity of the neurons. And it does work really well for Parkinson’s disease and for other diseases as well, right? But this is very invasive techniques. They’re sort of last resort things.
One of the things with the dopamine system, right? Like the dopamine nuclei in the ventral tegmental area, the substantia nigra, they’re very deep in the brain, so it’s sort of hard to reach them without, like, going inside. The stratum is also a bit deep. So, all of that part of the circuitry is hard to access, that’s where you would need, really, to understand, how is the learning happening? Because now you’re not changing the learning itself, you would be changing sort of, what are you learning about?
Fiona 19:15
It’s so interesting that, you know, all of these things are controlled very physically in our brains. So, I want to talk a bit about how you got where you are. Paul, how did you end up at the intersection of neuroscience, AI and psychology?
Paul Masset 19:29
Yeah, so that’s a good question. So, I started sort of my undergrad as an engineer thinking I was going to go in some sort of, you know, heavy engineering things like, you know, designing airplanes or something like this. And then I got brought into the more computer science side of things, I got interested in that. And then I had some very interesting lectures sort of people who are exactly at this interface; who are in the engineering department, but who are neuroscientists. And so, who basically were talking about bringing ideas of sort of Bayesian inference, how do we compute probabilities correctly and how can that basically help us understand motor control, visual perception and all of that? And so, that sort of got me into that part. And then I continued—I did a master’s in sort of computational neuroscience to learn more about the neuroscience as well. Right? Because I had done some classes, but not that many. And that really got—sort of got me into this. But then I’ve had the feeling that I couldn’t stay on the computational side only. I had to actually understand what was happening. And so, for my PhD, I looked for a place where I would be able to sort of be in the experimental world a bit more and do experiments. And so that’s where I did some sort of, you know, rat behavior, decision making, recording neurons and prefrontal cortex, looking at this idea of representations of confidence while the animals are behaving.
Also, AI has changed tremendously in the past 10, 15 years, and so there’s all these ideas about—it’s not necessarily that the brain is not doing what the algorithms are doing, but the algorithms show us how information can be transformed from inputs to pretty complex outcomes. And so, there’s at least ideas there that we should look at to try to understand how the brain can do it.
Fiona 21:16
So, it’s not just that we’re using the brain as a model for what AI should do, but now we’re doing the opposite. We’re looking at how AI does things to learn about how the brain might be doing them.
Paul Masset 21:26
Exactly. So, the idea is to also go the other way to think about how the brain is built and how can that help us do better algorithms. But in a way that has worked at a very conceptual level. It’s not like you go to school for 20 years and then you stop learning. That’s something that actually algorithms are not great at doing at the moment. You sort of train them, put them out. And so, there’s clearly inspirations that should be looked at, but what is the right level of abstraction, I think, is where the big question is at the moment.
Fiona 21:56
Ana, your research background was in headaches and migraines. So, what drew you to epilepsy and neuromodulation therapeutics?
Dr. Ana Suller Marti 22:03
I became interested in epilepsy very early on in my residency. So, during residency, I already started doing, of course, [passion 22:12] about the patients, but also starting to do some small projects in epilepsy and presenting in different national and international meetings. But unfortunately, when you finish residency, it’s not what you want. Sometimes it’s the options that you have. So, a close field that I already enjoyed was also headache. So, during my master’s, I focused on headache. And the lucky part is with my mentor in headache, he was an expert in neuromodulation, so he was applying neuromodulation or stimulation to manage headaches. And we published a few papers about that, and I managed a few of those devices. But unfortunately, my heart was trying to take care of patients with epilepsy, so that’s when I decided to make the big jump and moved to Canada and pursued a fellowship in epilepsy. And then I got to learn about these patients who unfortunately don’t respond to medications and beyond medications options, we have the stimulation. And that just brings me here almost 10 years later, how much we don’t know. When I started looking into that is, how many questions I had, how many questions I could not find anywhere? So, I thought it was an excellent path learning more about these devices that commonly we are using.
Fiona 23:33
I think this is like a quintessential research experience. There’s no shortage of questions, you just have to find the path that sort of speaks to you, right? What you find to be the most interesting is where you’re going to have the biggest passion and answer those questions. So, I love hearing that.
You know, you’re both doing really cutting-edge research with pretty novel techniques. Is there something you believed about the brain early on or when you started that you since changed your mind about with the advent of new tech or new information?
Paul Masset 23:59
So, I think one thing in the field I am, right? Where we sort of look at animal models and we record the neurons—activity of single neurons, one of the sort of transformational technologies in the past, like, five, 10 years, when I started you basically were recording from, like, a few neurons at a time. Right? You had an animal, you record like four neurons at the same time, you’re super happy, right? It got better. You had some slightly better methods. You could get 20 neurons, maybe, at the same time. And so, now you’re starting to get several single neurons and see how they behave together. People had some ideas about what to do and they were looked at as independent entities to some extent. And now basically there’s imaging techniques, recording techniques. You can record, like, thousands—tens of thousands of neurons, single neurons, right? At the same time from different brain areas while an animal is doing a complicated task where they see something, they have to decide what it is and then make a choice. And so, now you have this population level view of things, right? I always felt because I came from the more computational side and that—before I got into the experimental, that the single neuron view was obviously not the right thing completely. Like it’s very important because in the end single neurons are the ones implementing the computation. They are sort of protein machines, and when the gene goes wrong, when something goes wrong, it goes wrong in the single neuron to some extent. But what is happening at the level of the circuit, right? That was hard to know what it was going to be, and I think that’s the big change where now we can record this and how these neurons are talking to each other.
Dr. Ana Suller Marti 25:36
Some of the things that Paul mentioned, I can make also a correlation in humans, that, of course, trying to record from brain activity it was nothing new. Like from 1950s, they were already recording brain activities to try to determine where the seizures are originated. But over time, we have been improving the technique, being able to record for longer days and also better understanding of these recordings. So, I think it’s quite revolutionary that the spy is not a new technique, for example, for epilepsy; has been modernized and the ways that we are looking into the data is completely different. Similarly, the epilepsy surgery is not a new technique, but we have been improving significantly how we do it. We use the new techniques as stimulation, that this has been more in the last two decades. How that field has grown dramatically and how this can improve lives of patients, but also how this may help not just seizure reductions or the comorbidities that we’ve seen in our patients.
Fiona 26:46
Really better tools to understand what’s going on so that you can ask the questions and, you know, get that information that you need to feed into the research. That’s really important.
If you could explain one insight from your work that you wish everyone understood, what would it be?
Paul Masset 27:01
I think one thing is that learning is about reducing the error between what you expect and what you get. Right? The learning signal and sort of what you’re getting has to match what you expect to some extent, right? And so, that’s why it’s a very adaptive process and something that works for someone may not work for someone else. I think about it in a very sort of narrow sense, right? That when you have a learning in a [neural 27:28] sort of, you know, network or in the brain, right? You’re getting inputs from the world, like you’re getting something, and you are expecting you have an internal model of what’s happening. So, these two have to be different if you want learning, but they can’t be too different, right? Otherwise, there’s no learning. And so, I think this idea of learning as a sort of a prediction error is very important and thinking about it that way, and that is incremental steps that allow you to build much more complex sort of understanding of the world.
Dr. Ana Suller Marti 27:57
Epilepsy is more than seizures, that there are many comorbidities associated with epilepsy. And when we try to treat, again, one of every three will not respond to normal treatments. We need to offer other, more invasive techniques that, of course, for general populations seems a bit dramatic to implant devices or do surgeries but this has a drastic impact in their lives from just not the quality of life only, but also the risk of dying. How this works, again, this is kind of a black box still. That we know the input, we know we are looking for the output, but still how the brain is responding to that is still unknown. So, again, this is still work in progress. This is unfortunately the reality that we are living. And some of these more invasive techniques are used because there is no other options right now.
Fiona 28:54
I know you both have almost brand-new little ones at home, so thank you so much for sharing some pretty valuable time with us. And, you know, I also have kids, not so little anymore, but parent-to-parent, what do you hope your research can unlock for the next generation?
Dr. Ana Suller Marti 29:10
At least from the epilepsy field, there is a lot of questions and a lot of answers that need to be made. So, the idea is to improve life of the people living with epilepsy, reducing comorbidities and reducing the risk of dying from epilepsy. That will be a good goal. But, of course, this, as I say, through the complex disease is not an easy task.
Paul Masset 29:37
I think one thing is, like, my work specifically is a bit more, you know, theoretical and sort of experimental, so we’re not that close to the clinic at the moment. But I think the dream is to see some of these ideas really making an impact in terms of how we think about disease, potentially designing new diagnostics or new therapeutics. We can see a path to that, help design a new drug, help design a new therapy, behavioural therapy. And so, like seeing some of these ideas come through to actually change people’s lives, that could be an amazing thing. And that’s the hope, right?
Fiona 30:09
So, thank you so much Ana, Paul. And to all of our listeners today, thank you all for joining us on Bold Minds.
Paul Masset 30:15
Thank you so much. It was a great discussion.
Dr. Ana Suller Marti 30:18
It was a pleasure to be here. Thank you for also helping us to learn about our research going on in Canada.
Fiona 30:25
[theme music] Bold Minds is a Brain Canada production with support from the Azrieli Foundation. Our executive producers are Jillian Donnelly and Kate Shingler. Our lead producer is Jess Schmidt, with editing by Morgane Chambrin. Thanks for listening.
If you enjoyed this episode, we’d appreciate it if you could send it to a friend. If you want to learn more about Brain Canada and our Future Leaders program, please visit our website at braincanada.ca. [music ends]