A Conversation with Prof. Julio Martinez-Trujillo
Date Posted:
March 25, 2019
Date Recorded:
March 22, 2019
CBMM Speaker(s):
Diego Mendoza-Halliday Speaker(s):
Julio Martinez-Trujillo
All Captioned Videos Scientific Interviews
Description:
On March 22, 2019, CBMM Postdoc Diego Mendoza-Halliday took the opportunity to sit down and chat briefly with Prof. Julio Martinez-Trujillo of the Robarts Research Institute.
[MUSIC PLAYING] DIEGO MENDOZA-HALLIDAY: Hi, my name is Diego Mendoza-Halliday. I am a post-doctoral research fellow here at the Center for Brains, Minds, and Machines at MIT. And I'm here with Julio Martinez-Trujillo, our guest speaker. Julio is a professor of pharmacology, physiology, and psychiatry at Western University and principal investigator of the cognitive neurophysiology lab at the Roberts Research Institute and at the Brains and Minds Institute. As one of the many scientists who had the honor to train as a PhD with Julio, it is a great pleasure for me to host him today at our conversation series. Welcome, Julio.
JULIO MARTINEZ-TRUJILLO: Thank you very much, Diego. It is really a great pleasure to be here today.
DIEGO MENDOZA-HALLIDAY: Julio, you began your career training as a medical doctor, but later you chose to leave the world of medicine and became a neuroscience researcher. What led you to make this career change?
JULIO MARTINEZ-TRUJILLO: Well, that's a very good question, Diego, and I ask myself the same thing all the time. The conclusion is that I was trained as a medical doctor, and then I did a specialization in clinical neurophysiology. So I had the opportunity to work as a clinical neurophysiologist for about three years in Bogota, Colombia in an institution where many of the children that they had epilepsy. And some of them, they had developmental disorders.
I was very impacted by my inability actually to treat the children with the current pharmacological treatments, so most of the time I would use drugs that would decrease the intensity or duration of seizures, or frequency. But most of the time, actually, it was almost like a trial and error thing. In some children, that would work. In others, it didn't work, so I'd have to try a different drug. And something that impacted me very deeply was that some of the children with developmental disorders had deficits of attention. So the childrens were actually-- which is defined as ADHD.
And some of them, they had very strange EEGs, and some of them, they had kind of normal EEGs. So my conclusion was that we just really didn't have enough understanding of what was happening in the brain of those childrens to start prescribing medication. Of course, we had to do it, and the medication works many times.
So I decided to leave my career as a clinician and to go on, pursue a PhD in research in Germany in this case at the University of Tuebingen and trying to find out a little bit about the structure and function of the brain. So that was a very critical decision that I made in my life. And the way that I actually look at myself-- I say, well, maybe if I learn a lot about this, and I might be able to actually come up with alternative solutions that they're more like engineering-based. Knowing the system exactly how it is, maybe we can actually find ways to remedy some of the problems in the system. I might be able to contribute more than a single doctor does. As a single doctor, you may see thousands of patients in your life. If you are able to contribute with something basic, or some knowledge that allows to design interventions, you might impact the life of millions of people.
DIEGO MENDOZA-HALLIDAY: And this is how you decided to go do a PhD with Stefan Treue and got into the field of attention.
JULIO MARTINEZ-TRUJILLO: That's correct. So I pretty much left my practice, and I had very wonderful clinicians I was working with. Some of them understood my decisions. Others, they didn't. But that was exactly how I decided to do this.
DIEGO MENDOZA-HALLIDAY: And Stefan Treue you decided to study the mechanisms of attention at the level of single neurons in animal models, and from that work came a very influential study published in Nature in 1999 with Stefan Treue. Could you tell us a little bit what the impact or the influence of that paper was on our field?
JULIO MARTINEZ-TRUJILLO: Yeah, that was actually when I was finishing my PhD. We had this experiment, and we were looking at the neural basis of attention. So we were training animals to attend to a stimulus and ignore the same stimulus in another condition. We were looking at how neurons react in the individual cortex.
The view of attention at that point was that attention was like a spotlight, like a spotlight that kind of illuminates or highlights the attended items, so like what you see, for example, in the theater or something like that. So in our case, what we looked at was something called feature-based attention. In feature-based attention, what you do is you train the animal to attend to a certain feature. In this case, it was motion direction, but would be-- for example, color, red. And then we explore how attending to a certain features could influence the processing of the same features across the whole visual field or even in the opposite [INAUDIBLE] field.
What we found was that if you attend to color red, that will facilitate the processing of red items across the visual field. So basically, it kind of breaks the notion of the spotlight. It's mostly like a feature-based mechanism in which, when a feature is relevant to you, what your brain automatically does is enhance the processing of the same features anywhere in the visual field.
Now, it's not like a spotlight. It is what-- we call it the feature similarity gain model, which was a theoretical model that we develop in which space wasn't essentially the dimension in which attention could enhance deep processing of stimuli.
DIEGO MENDOZA-HALLIDAY: But attention does act in space as well.
JULIO MARTINEZ-TRUJILLO: Absolutely. Space was one of the most dimensions of attention.
DIEGO MENDOZA-HALLIDAY: And you contributed by saying, it's not just about space. It's about features as well that are in spatially independent frame.
JULIO MARTINEZ-TRUJILLO: That's correct. So I have to correct that. In psychophysics and behavioral, it was already evidence of that. They had the cocktail party effect, for example, that you hear your name, and suddenly it attracts your attention. But it was no neurophisiological evidence of that phenomenon at that point, not in the way that we show it under very well-controlled conditions. I think that that was the main contribution, and that paper was essential for my career. So basically, from that point on, I decided, well, I can do this. This is something that I enjoy. I'm always a very strong curiosity driven person, and I just decide to pursue a career in research at that time.
DIEGO MENDOZA-HALLIDAY: So you gave that insight into attention, but there's many other questions that are still unanswered in the field of attention. What are some that you would like to target in your future?
JULIO MARTINEZ-TRUJILLO: I think that at this level we have an idea of how attention enhances the processing of stimuli-- so by enhancing the firing rate, improving the signal to noise at the level of single neurons and populations. We have some idea of the areas that control attention. For example, areas in the prefrontal cortex seems to be critical for what we call voluntary attention, which is allocating attention to stimuli in space or in other dimensions.
Now, I think that what I would love to do, and I think that we are ready for that, is to start looking more at the circuit level mapping of attention, and not only the circuits when you look at connectivity between areas, and the ways that the areas implement attention, the brain is a network, but also at the mesoscale level. So what are the cell types that contribute to circuit operations, for example, at the level of a cortical column or at the level of a given area? So what are the units that are fundamental in terms of neuronal types, in terms of response properties, and how the brain implements these at different levels?
So meso-level scale and macro level scale, I think that we are more-- we have more [INAUDIBLE] at the macro level scales. At the mesoscale level, it's a little bit more difficult, but I would love to go into these kind of questions. So that what I plan to do in the future.
DIEGO MENDOZA-HALLIDAY: So your career path has taken an interesting direction in that, having started with a focus on attention, with time, your interests have greatly broadened to the point that attention just makes a small fraction of the current work you're doing. Could you tell us a little bit what other fields you have incorporated into your lab work, and why did you decide to explore those realms?
JULIO MARTINEZ-TRUJILLO: Well, I have a couple of fields that I have incorporated over the last years. One is I have some research in the lab going into, exploring the mechanics of memory, short-term memory and long-term memory formation, and I'm going to tell you a little bit about that in a second. And the second thing-- I have been very interested to look at specific symptoms of very specific diseases like autism, and this is something that I have a personal interest in.
So in terms of the memory, that has a very strong connection to attention. So when you are looking for something that is red, or if we're looking at a particular person in a crowd, you have to have a mental image of what you're looking for. And that becomes very interesting when some people start talking about the attentional template.
So I read a paper by Bob Desimone and John Duncan. I think that is in 1995, this review that they talk about this attentional template signal. And this attentional template, it has to reside somewhere in your brain. Even if the stimulus is not in the visual field, you have to have a representation of that template.
And what I realized is that this signal looks like a working memory, or short-term memory representation. That was another field of study in neuroscience. So I became very interested in finding this source of the attentional template or the working memory representation and that is when you came to the lab, and we started actually exploring the mechanism of short-term memory. And I expanded research in that particular field, and I think that now we has advanced to a point where we have a good mapping of where those signals exist in the brain, and we kind of started looking into questions like, how are these signals, for example, influencing processing in visual cortex or in sensory cortices? Or how these signals could get converted into long-term memory representations?
So we started doing also research into specifically that field, how the short-term memory signals are converted into long-term memory presentations, and we started doing recordings in the hippocampus. The other thing that motivated me very strongly was trying to run the experiment in a naturalistic way, so to see that incorporates visual stimuli, eye movements, like people doing a naturalistic fashion.
DIEGO MENDOZA-HALLIDAY: To shift away from the classical paradigm, where the subject is in front of a computer screen, where there's very minimal, stimuli, and you control every parameter to something more complex and that looks like the real world.
JULIO MARTINEZ-TRUJILLO: Yeah, absolutely, and I think that we were ready for that a couple of years ago when the video game engines started coming along. With these video game engines, we could recover frame by frame the stimuli that we had on the screen, and if-- we did some research in the lab, and we developed a couple of softwares that allows us to map the eye movement patterns, for example, into those frames. You have a precise timing of signals with the video engines, and to make the virtual environments as complex as we wanted, or as simple as we wanted.
Then, at this point, we started running experiments that were like the experiments that we do in 2D with computer screens, with objects, and we started developing another series of experiments on parallel in virtual environments. So to see how much actually-- how the neurons behave in one situation or another, I believe that the naturalistic way to running experiments is the future. However, it is complex.
The reason is if you have a neuron that is firing to a particular stimulus, this neuron could change completely the firing rate, or at least partially the firing rate, when this is other stimuli around or when the subject is engaged in a situation, where you have to navigate through an environment, and that's the way that we do. When we go to a supermarket, you don't have a computer screen in front of you that say, oh, bread, and then I'm going to choose the bread from a shelf that is just empty. And there is only the bread in the shelf, or the bread and some of product.
So you have a multitude of products there. You have to walk to the supermarket you have to keep the working memory signals alive while you walk, or you have to retrieve it from long-term memory to do that kind of things.
DIEGO MENDOZA-HALLIDAY: In the real world, all these signals are working together. Sensory inputs, cognitive operations-- all these are working at the same time. The approach that most people have done of isolating one or a couple of these components of what happens in the real world is to be able to experimentally manipulate them and control them. Now, by letting all these things vary, you obviously are shooting yourself in the foot in that now you have to deal with this complexity. How do you deal with that complexity now?
JULIO MARTINEZ-TRUJILLO: Well, first of all, I don't want to discredit the experiment using reductionism. Scientists, we're reductionists by nature. That's what we do to understand a system, and then we add layers of complexity as the system goes. But I want to tell that the virtual environments allow you to do that. You can your reductionist experiment, and then you can add layers of complexity and to see how these layers of complexity affect the results that you had in the first place.
I think that a condition that was necessary was to explore this, the neurons and the brain in this reductionist kind of approach, using this reductionist kind of approach. I think we are ready now, at least in some fields and with some paradigms, to move on and to try to add layers of complexity. I also have to tell you that other techniques-- like computer power has increased over the last years. Techniques like machine learning, multivariate analysis-- so they're available now there, and there is a generation of computer scientists that they're coming with incredible programming skills that actually could take all this data and try to make sense out of those computer using models.
The best way to do this is you build your layers of complexity around your reduction reduced world, and then you start adding. And then you start planning of having a model of how that would work, maybe with predictions, and now you can test all that kind of thing. So I think that we are ready to go this way, but I could totally understand that many of us feel that we still haven't gained insight into the brain using these reductionist approach, and we may have to be cautious in that kind of thing.
DIEGO MENDOZA-HALLIDAY: What are some of those questions that you couldn't answer with the classical approach that you can now answer with this one?
JULIO MARTINEZ-TRUJILLO: Absolutely, for example, I mean, I-- a lot of evidence and people doing research in rodent hippocampus that you have placed cells that kind of change when you change the spatial layout of a maze. So these cells are-- it's almost like a [INAUDIBLE] or an empty changing when you change your configuration of stimuli, something that is actually conceivable to have. We will never know the answer if we don't make those changes. So I understand that we need to go step by step, but I feel that with the development of computer games and computer game engines, I think that we are ready to take over some of those tasks.
DIEGO MENDOZA-HALLIDAY: So Julio, besides studying the fundamental mechanisms of cognition, part of your research has been dedicated to study what goes wrong in those mechanisms in patients suffering from some neurological disorders such as autism and attention deficit and hyperactivity disorders. Could you tell us a little bit about your approach to studying these disorders, and particularly how studying these disorders in animal models can help us understand what happens in humans?
JULIO MARTINEZ-TRUJILLO: Yeah, that's a very good question, Diego. I'm very interested. As a physician by training, I like to help people, and I believe that animal models could help us a lot to understand what's going wrong in some of the diseases. An example is autism. So my approach to this mental disease is a very complex thing.
And actually, the diagnosis of many of the categories of psychiatric disease is done through symptoms. So we don't have biological markers for those things, for those kind of diseases. For example, autism is just characterized for deficit in social interactions, communication, language, and [INAUDIBLE] repetitive movements. So my approach to this is to take one of the components of autism and try to test in an animal model if we can isolate the circuits that actually would be responsible for that specific behavior. And what components of the circuit actually could be affected to a certain degree to do this?
It's a little bit complicated because I'm doing the reverse. I'm taking the whole, taking a part of the whole and trying to reverse engineer this part of the whole. I know that at some point we have to put it back into the whole phenotype of autism, but I think, for example, in my lab we're trying to understand gaze avoidance. Gaze avoidance is a symptom that you see almost 100% in people with autism.
So now we know a lot about [INAUDIBLE] motor system in the macaque monkey because primates are probably one of the few animals that they use gaze as a means to establish social interactions and to interpret the gaze of others. So this is very prominent in primates, and the macaque monkeys is a good model, as so other primates, marmosets perhaps. Now, we know a lot of about [INAUDIBLE] system, the [INAUDIBLE] LIP, all the [INAUDIBLE] nuclei.
DIEGO MENDOZA-HALLIDAY: Which seem very preserved to a certain extent in--
JULIO MARTINEZ-TRUJILLO: That's correct. It's very preserved. So the first thing is to understand that this is a good animal model to study that symptom.
The second thing is to start actually studying the circuit behind that and trying to reverse engineer. How could we get to a gaze-avoidant phenotype, for example, from that circuit perspective? So I'd isolate the circuit, the symptom, and I target that specific component.
There is a disadvantage to that, which is, well, you don't have the whole phenotype, but the problem is that the complexity is such that I don't think that we can understand the whole phenotype at once. So that's my approach to actually establish circuit models of symptoms of psychiatric disease.
DIEGO MENDOZA-HALLIDAY: As opposed to complete models--
JULIO MARTINEZ-TRUJILLO: That's correct.
DIEGO MENDOZA-HALLIDAY: [INAUDIBLE]
JULIO MARTINEZ-TRUJILLO: I believe that the genetic could give you, actually, phenotypes. Actually, in animal models, it's very hard to have the whole phenotype. Sometimes you find a deficit. You don't find it, and so actually, what we are doing is not that different. The only thing is that we're trying to probe the circuit with tools like optogenetics, or tools like chemogenetics, or things like that, that you can actually note down or enhance the functioning of certain parts of the circuits, and you can generate very clear hypotheses about how this phenotype for that specific symptoms arises.
So I think that that's my approach to that, is kind of a reductionist this approach. It's a very different approach where we were saying with the virtual environments, but I believe that. I just don't see another way to attack the complexity of mental disease.
DIEGO MENDOZA-HALLIDAY: What are some of the most exciting new tools that you have been able to incorporate in your work that were not available back in the '90s when you started as a researcher, and what kind of questions can we now answer with those?
JULIO MARTINEZ-TRUJILLO: It's very exciting to live in these times and to do neuroscience in these times, where you have a lot of new tools coming out. So some of them are older. Some of them are newer. For example, some of the new tools, like being able to record from many, many neurons at the time in the brain of a behaving animal, or even a human in some of the cases when they undergo epilepsy surgery that you can record those signals, is a very exciting thing. There is also a lot of signal analysis processing tools that are right out there that allows you to process large number of data.
Some of the exciting tools-- for example, I got very excited about optogenetics and calcium imaging. The reason why I got excited about this is because, as a neurophysiologist that does electrophysiology, and we listen to the responses of neurons in the brain-- we record it-- I also felt that when I started my PhD I was listening to the radio. And in fact, that's what I did. I would just put my headphones, and I would just listen to how the neuron spikes.
But it's almost the same thing as listening to people talking and you don't see the faces. Now, with the new tools like calcium imaging, you would be able actually to see the activity of the neurons and actually to visualize neurons in different layers and different type of neurons, depending on your genetic tools too, which has been a big advance in the degeneration of those genetic tools, whether it is transgenic animals or viruses to load actually your fluorescent markers into the neurons. So that's very exciting to me.
It's almost like now I'm watching TV. It's still a very low resolution TV because calcium imaging doesn't have the resolution of the electrophysiology that we do, but you still could see some of the faces and actually--
DIEGO MENDOZA-HALLIDAY: Yeah, see the actors and know who they are.
JULIO MARTINEZ-TRUJILLO: You see the actor and who they are. It's almost like when people actually were just listening to the radios, I don't know, the last century, and then suddenly the TV came out. Well, the TV was black and white at the beginning. Then it just became the color TV. I think that we're going to see that over the next year, that the resolution of those calcium tools are going to become really higher, and also they opto.
The optogenetic allows you actually to manipulate neurons' responses with a very high resolution, time resolution. So those are things that could be combined actually to study circuits, to probe circuits, whether you're studying a physiological function or a specific symptom that you want to produce, or a specific phenotype. Those are very exciting times.
DIEGO MENDOZA-HALLIDAY: While neuroscience has made tremendous advances in understanding the neural mechanisms of cognitive functions, some neuroscientists believe that those advances have been mostly underwhelming and that we should have expected more from the field. What are your thoughts on that statement? And do you believe that neuroscience has all the tools it needs already to answer those questions, or do you believe that there is a need for a revolutionary change in terms of technologies available, theoretical approaches, or experimental paradigms in order to give a full answer to those questions?
JULIO MARTINEZ-TRUJILLO: Yes, that's a difficult question, but I'll gives you my thoughts. So I believe that neuroscience has progressed enormously from when I started my PhD to now. I mean, we were recording single electrodes, one neuron a day, and trying to repeat the tasks again and again, and spending years training an animal to do a task. And now you have multi recording probes. You can't record from multiple layers.
You have improved the resolution, and the power of those techniques have improved. And the results of the emergence of new techniques, like what I was telling you in microscopy-- the marriage of microscopy on electrophysiology with these calcium imaging techniques is just amazing. Genetic has also made contributions that are very large to that, so I believe that neuroscience is not stuck, so neuroscience is definitely progressing, and we are getting better and better at using those tools to understand the neural mechanism of cognition.
There are two things that actually pushes development forward. One is the human ability, how smart we are to come up with those things. The second thing is how many resources we dedicate to this. I mean, it is good that institutions like NIH dedicates a large amount of resources to developing techniques and to fund researchers, and many of us actually-- still, many researchers are underfunded for whatever reason.
DIEGO MENDOZA-HALLIDAY: So do you believe that there should be more funding to advance at the level at which our society should expect?
JULIO MARTINEZ-TRUJILLO: I believe that there is-- at some point there is a balance between how humans can achieve with our brainpower and the amount of resources that you put into it. I don't think that we are at that balanced stage. I think that we could put more resources into it, and we would move faster. That's definitely something that I-- I deeply believe that.
Now, do we have all the tools that we need to understand? No, of course not. I mean, in fact, we have very few non-invasive tools to do research in humans, and that's the reason why we use animal models because actually we can access many of the processes in the human brain without hurting the brain so, actually, without being invasive. So that's the reason why we use animal models, so I don't believe so. So there is a breakthrough in which we can actually visualize and record the activity of neurons in an intact brain, and going through the skull, and do all that experiments that we do.
That would be terrific. It just doesn't happen. It doesn't exist at this point. I mean, there is functional imaging, but the resolution is not the same, so it's very low spatial and temporal resolution.
There is EEG, but then the EEGs just allows information in the superficial layers of the cortex. Its spatial resolution is bad. Yes, we need breakthroughs, and probably there are going to come through engineers, physicists, biologists all working together. And of course, you have to dedicate resources to develop that.
We are not ready yet, I think, to say, well, we're going to stay with the knowledge that we have, and we're going to do experiments with the techniques that we have. And with that, we are going to develop mathematical models of the brain, and now we're going to predict what happens during a disease, and we are going to find cures for that. I don't think we're ready for do that.
Yes, we need new developments, but we have progressed a lot. I think scientists worked very hard to achieve this progress. So it would be-- to say that we haven't made enough progress, actually, it's partially truth because what I was telling you. You cannot actually get to do experiments in humans, and there is abilities like language and things like that, that you have to go to the human to test that, but we have made progress. Those are my thoughts. I don't know if this is a clear answer, but this is actually what is in my mind.
DIEGO MENDOZA-HALLIDAY: Julio, this has been a very insightful conversation. Thanks for joining us today.
JULIO MARTINEZ-TRUJILLO: You're welcome.
DIEGO MENDOZA-HALLIDAY: Julio will be giving a talk today at MIT titled "Probing Memory Circuits in the Primate Brain from Neurons to Neural
Networks," and this talk will be available in our CBMM channel. Thanks for watching.
have an interactive transcript feature enabled, which appears below the video when playing. Viewers can search for keywords in the video or click on any word in the transcript to jump to that point in the video. When searching, a dark bar with white vertical lines appears below the video frame. Each white line is an occurrence of the searched term and can be clicked on to jump to that spot in the video.
