An Interview with Dr. Jessica Ausborn

By Nancy Mack

Dr. Ausborn is a recently appointed instructor in the department and conducts her research in the Rybak lab where she was a post-doctoral fellow. In this conversation, as she shares her academic journey, she offers unique insight into the world of computational neuroscience, showing us that computational modeling is not so different from experimental work.

Do you remember when you first thought of becoming a scientist?

It was a gradual thing. Initially, I studied to become a high school teacher in Germany. The state in Germany where I studied didn't have a particular program for that, so they put prospective high school teachers in the regular classes with bachelors and masters students and made them take some additional classes in pedagogy. I studied math, biology and computer science to teach high school students. When I finished my studies, my supervisor for my final thesis asked me to continue in his lab for a PhD. Back then we didn't have programs that you could apply to for a PhD--you would just agree with your prospective supervisor to work on a PhD project in their lab. So, when this offer came about, I thought I'd get a PhD and then go back to teach high school students. But somewhere along the way I got stuck.

What was the topic of your dissertation?

I worked on a computational model of the locust flight system. I was looking at a sensory organ and its interaction with the locust flight central pattern generator. It was a really nice, small study which used computational model to produce some predictions and directly tested these predictions in the animal. We did some electrophysiological experiments where we, in real time, replaced the sensory organ with a computer simulation of that sensory organ and fed it back into the biological system. The locust flight system is one of the pioneer model systems to study central pattern generators and my thesis was mostly concerned with general questions about sensory-motor integration. I really enjoyed this work.

You have done research at multiple institutions around the world, from Germany to Sweden to the United States. Have your scientific experiences varied across these locations?

It's hard to say because it was also a progression from being a student to being a PhD student, to being a post-doc and so on. So, it absolutely differs, but it's hard to know if it was the different position I was in or the different countries I was in. For example, I feel like regulations have increased more and more, but that could be over the years or from country to country. I also changed model systems and approaches vary dramatically, so it's very hard to compare.

You have a background in both experimental research and computational modeling. Do you have a preference or like different things about one or the other?

I absolutely like different things about each one. I did exclusively experimental work in Stockholm, Sweden, where I was working with zebra fish. Originally, I had planned to go there to start experimental work and to develop computational models of the data. But in the end, it wasn't realistic to do both well. I think there are certain model systems where you can do that, but with this one I only ended up doing experiments. I liked planning and coming up with experiments. I even liked setting up the experiments and maybe even performing the first few experiments of each type. But somehow, I'm not very good at doing repetitive tasks. As soon as I had an idea of what the outcome could be, it was really frustrating for me to continue with the experiments. In computational work, you obviously have much less of that. You have an idea and there are techniques and skills involved in implementing that idea, but it's not as repetitive as experimental work. What I like the most is working with concepts and ideas and discussing data. What we do here at Drexel is perfect for that. We work very closely with experimental biologists and we very often work closely with them while they're actually performing the experiments. We discuss with them what experiments to do and what types of simulations we would need, for example, to understand how the system works. So I get to do all of this planning, and I just don't have to do the experiments.

I personally study the activity of a single circuit, and even just looking at a single connection between two neurons can be so complex. I'm fascinated by the idea of modeling an entire network of connections. Can you describe what goes into the process of making a computation model?

I think it sounds more complicated than it actually is. If you look at a single connection, you understand a lot about that single process in exquisite detail, and if you were to scale that detail into a larger model it would be very complex. However, it doesn't really make sense to model a network of neurons at such a level of complexity. You have to abstract—you have to sideline detailed processes such as vesicle release. You have to ask yourself, “What question am I interested in?” Very often, what our group is interested in is how different populations interact with each other. To try to understand how the system behaves, we often don't model all neurons individually but rather model the whole population as one unit. With that being said, we also develop models with individual neurons, especially if we want to know how neurons within a population interact. For example, if it's important for the neurons in this population to be connected amongst each other, then synchronization of action potentials within the population plays a big role, and that is hard to model in a more abstract way. We are trying to go as abstract as possible to still answer the questions that we're interested in, only adding detail if it's necessary for the question or the functioning of the model. What we're not trying to do is replicate biology. I think many people assume what computational modeling is doing is taking all the data that's available, putting it into some equations, and hoping something comes out that resembles biology. But that never works.

There are some really interesting studies in the stomatogastric nervous system of the crab. The core of that system comprises only 24 individual neurons, each responsible for a specific task. This is particularly advantageous since the same neuron is often identifiable from animal to animal. Researchers were able to characterize all of the ionic currents in a specific neuron by measuring over many animals, blocking other channels, and using voltage and current clamp recordings. The groups working on the stomatogastric system then averaged all their recordings for a specific neuron across many animals and made a computation model, but the model didn't do at all what this neuron was supposed to do. This sparked a whole field of studies that essentially found that if you average all of the measurements that you do, you won't get an average neuron but something that's nonsensical. This suggests for computational models that there's no point in measuring all the data and trying to make a model out of averages. Instead, we have to have a certain question and try to come up with possible solutions. We test our models by saying, if our hypotheses are right, if the biological system is really set up like that, then the perturbations we use in our model to elicit a change should produce similar behavioral change in the animal. For example, if we take out a certain population of cells, the network rhythm speeds up in the model. We can then tell experimentalists to inhibit or knock out this population and see if that rhythm really gets faster. If our model prediction is confirmed in the animal, that brings us one step closer to understanding the biological system.

I actually see myself as a part of the experimental process, not as something separate. If you look at your own experimental data, you make a mental model in your head and you have this idea about how things should work out. The benefit of computational models is that you don't just make an intuitive model in your head. You can't just say I think it's like this. You actually have to put it into mathematical equations and these equations then have to produce a biologically relevant output.

What is one of the biggest challenges you've faced thus far during your scientific career?

It's very hard to come up with something unique and personal. I'm at a point where I think about where my research focus is going to go. I'm working with Ilya and he obviously has an established research direction and I need to find something independent that will fuel my own research path. I think that is the most difficult part, especially if you do computational neuroscience the way we do, where we heavily rely on experimentalists. I need to find people going in the research direction I want to go in to collaborate with me. I can't just say this is what I want to study. I am always dependent on experimentalists providing data. Many people, including us, often use published data to make models, but being involved in the planning of the experiments is the most fruitful part for me. So currently my biggest challenge is finding something to make my own and identifying collaborators for it.

What do you like to do when you are not doing science?

I'm pretty crafty. I like to draw, crochet, knit, decorate my apartment and things like that. I also like to go for walks and meet with friends.

If you had to choose an alternative career path to science, what would you choose and why?

Well, I like teaching. I've always liked teaching and that's never gone away. I don't think I'd go back and teach high school students, for various reasons. But I do like teaching, and I think it's always a viable option.

If you could go back and give yourself advice when just starting your doctoral training, what would it be?

As I said, I started my PhD without actually planning to continue on in academia. If I had the foresight of knowing where I would end up now, I would probably just tell myself to keep doing what you're doing. I've had so many fortuitous opportunities and changes. I never applied to come here, for example. I was visiting Kim Dougherty—we have been friends since Stockholm. She was setting up her lab and I had some time and vacation to spend, so I took two weeks off to see where she lived and help her solder some cables, get things running in her lab. And since I was already here, I gave a talk. One thing led to another, and Ilya offered me a position in his group. I was looking for a new job at the time but hadn't decided if I wanted to switch to computational work completely or if I wanted to do something in combination. Most of my career wasn't planned. I have often just followed new opportunities that came up, and I think that's a good way to do it as well.