Computational Neuroscience and the Joy of Discovery With Dr. Kanaka Rajan
"As you progress in your career as a scientist, it’s impossible to stagnate. You stay young, creative and excited about your work." – Dr. Kanaka Rajan.
Complete the form below to unlock access to ALL audio articles.
Kanaka Rajan, an associate professor of neurobiology at Harvard University and founding faculty member at the Kempner Institute for the Study of Natural and Artificial Intelligence, works at the intersection of two fast-paced fields – artificial intelligence (AI) and neuroscience. Where other subfields of biology have shed their secrets over decades of in-depth investigation, brain science still hums with mystery.
Kanaka is a leader in a new wave of neuroscience that has leveraged vast volumes of computing power as a tool for understanding the brain. Where the three-pound lump in our head remains stubbornly resistant to precise measurement, code can be dissected and stitched back together. Simulated neurons can be tracked and manipulated freely, whereas even a single biological brain cell requires fastidious protocols to sample from.
This blend of disciplines requires more than a mastery of molecular biology. Kanaka has a degree in biotechnology from Anna University in India’s southernmost state, Tamil Nadu. Her doctoral work mixed ideas from matrix theory and physics. Her thesis, published in 2009, pinned down the chaos of neural networks with a general mathematical framework.
If this approach once placed Kanaka on the fringes of traditional neuroscience, it’s clear that the field has now rushed forward to embrace her methods. When Technology Networks saw Kanaka speak at the recent Society for Neuroscience conference in Washington DC, her “meet-the-expert” session on the “allure” of computational neuroscience was not only packed out but ran over time by a full hour as a band of enthusiastic young researchers peppered Kanaka with questions and requests for advice.
In this interview, Kanaka explores her proudest work, her motivations and shares advice for young female scientists entering an area of research that remains male-dominated.
RM: What motivated you to pursue a career in science?
KR: I was always drawn to the STEMM route. As a young student in India, I followed the typical academic path through my undergraduate program: studying math, physics and engineering. My interest in neuroscience specifically arose after my undergraduate degree when I took an internship position related to mental health research. It was a new scientific environment for me, but it also felt familiar and personal. My grandmother – who helped raise me – lived a challenging life with a schizophrenia diagnosis and other family members have had struggles with mental health too, so I knew firsthand what a direct difference understanding the brain could make in people’s lives.
After that internship, I decided to study neuroscience in graduate school. This path was a surprising detour for me, but I was inspired to make a difference in brain health for the benefit of those who live with devastating mental illnesses. With my earlier mathematical and physical sciences training, I could see myself combining my love of problem-solving and curiosity about what makes things tick with my personal interest in brain health by looking at the brain through a computational lens.
RM: Could you tell us more about your current research interests and area of expertise?
KR: I lead a computational neuroscience lab at Harvard, where we bring together the fields of brain research and AI/machine learning to figure out how the brain works. We use mathematical and computational models based on data collected from neuroscience experiments to design artificial systems – essentially “in silico facsimiles” – that can perform realistic behaviors like animals and humans do, using only the machinery the biological nervous system has access to (e.g., neurons and synapses operating at a fast timescale). We build such facsimiles, and then we “reverse engineer” them to reveal the operating principles of the real brain.
More specifically, I build theories grounded in the question of how key cognitive processes such as learning, remembering and deciding – which unfold over minutes, hours, days and even years – are accomplished by the rapid, cooperative activity of neurons and synapses in the brain. Working at the interface of neural network theory, AI and experimental data analysis, I theorize how neural circuits bridge this gap in time scales to produce adaptive behavior.
My models use the properties of synapses and neurons, which are powerful enough to extend beyond a single experiment, task or brain area. They reveal unexpected “design principles” of the biological brain, particularly the mechanisms responsible for producing or mediating key behaviors that vary over time, which are inaccessible from measurements and may be conserved or uniquely divergent across different species.
RM: You’ve studied computational neuroscience in a period where our understanding of the brain and of informatics has advanced rapidly. Can you please highlight one or two of the most significant developments since you started working in the field?
KR: I think one of the major changes since I started in computational neuroscience is that we have grown to work with much larger-scale data. We first started looking at hundreds of neurons in one brain area, and now we’re looking at thousands of neurons across multiple brain areas in many different species. We’re not just looking at more cells at a time across more functional areas, but we’re also looking at that activity during increasingly naturalistic behaviors over longer and longer periods of time, such as during learning of a task or skilled behavior, or during development.
The spatial scale is larger, sure, but the time scale is also much longer, and the behavioral context is much less tightly task- or lab-based, but rather, more like in nature. This massive shift in scale has given us more flexibility to study electrical activity in the brain from a systems perspective, contextualized by behavioral data, which is a much more effective approach to mapping what connects to what, and why that pattern of connectivity gives rise to the dynamics we observe in the neural activity and behavior of the whole animal.
Alongside this volume and depth of neuroscience information, we’re also able to build super-powerful mathematical and computational models that are nonetheless flexible enough to map with these kinds of rich, large, biological data sets. These types of flexible, large-scale modes, like those borrowed from the field of AI, were previously out of reach in neuroscience, but they’ve become much more accessible now that academia and industry are working more closely together to solve fundamental problems facing mankind together, like what principles makes animals and humans intelligent, and what happens when the mechanisms and pathways responsible for adaptive behaviors go awry.
RM: What’s your favorite thing about having a career in science and what would you say are your proudest achievements?
KR: My life as a scientist is a little like being in a video game. I get to simulate biological systems – animals and people – as a computational theorist, so my team and I have total freedom to build and play with the systems we make and apply them to answer scientific questions that interest me. We know so little about the biological brain, so there are constant small wins every time we learn something new.
Even when there are disappointments or frustrations in my work, the joy of discovery is always within reach for scientists in my field. Neuroscience isn’t just theoretical, it’s related to real, living systems and there is also that satisfaction of seeing the real-world impact of our work.
Unlike physics or other theoretical disciplines, we get to engage with the messiness of life and see the way our findings can change people’s lives. I think that it’s an incredible way to spend a career.
I wish more people had access to my field. The thing I’m most proud of right now is actually a current project. I’m working with a few artists to create science comics. They’ll cover important abstracts and larger neuroscience concepts in a more accessible, engaging and jargon-free way. The way our manuscripts and papers are written can be pretty opaque for people who don’t already have a running mental glossary of neuroscience jargon and a lot of practice reading that type of writing. We hope that it will make the entry point to this field less daunting for young people, and lower the barrier to entry to minoritized individuals, who may otherwise be too intimidated to give a career in this field a real shot.
RM: Computational neuroscience, compared to neuroscience as a whole, remains a male-dominated field. What can be done to address this gender imbalance and encourage young female scientists into the field?
KR: As both a woman and a minority, I know all too well how hard this path can be. Those of us who have been able to carve out a space in this world can use that position to make the path easier for others – to create the network and resources that I wish had been available to me during my own training.
In New York, I organized and hosted a women’s group for theorists in the greater New York City area. We call ourselves, humorously, “WTF”, which stands for “Women on Two Floors” because, at Columbia, where I did my doctoral training, the theory center is located on the 5th and 6th floors of that building. We have ~35 members and have been meeting monthly since February 2021. We have dinner, discuss our work, give advice and offer support. It’s a great group, and I think we all find the opportunity for community, sharing and connection valuable. I think this type of support from within our community has real power.
Creating a community among women scientists at the professor/investigator level facilitates spaces where young scientists feel confident enough to share their ideas and find mentorship. Community support like that allows young women in science to see a path forward for themselves in research.
RM: If you could give one piece of advice to a young woman who is considering a career in science, what would you say?
KR: I would say that this field has a lot to offer in terms of flexibility, freedom and joy of discovery. Women in general don’t always have the necessary flexibility and agency in their chosen careers; not many careers allow for it. Super-regimented jobs – even when highly lucrative – don’t allow for flexibility of life and outside obligations, whereas a career in science is less linear, and isn’t penned in by the restrictions of a 9-to-5 job. There is also an unusual amount of agency in what you get to work on day to day. If you have questions about the natural world based on your own experiences, that means that you can choose those problems you have a personal connection to and are passionate about, and then get paid to study them.
It’s not all upsides and freedom though. Science is a high-variance path with lots of ups and downs. The steps along the way can be unpredictable. When I was starting out, I wish I knew how circuitous a career in science ends up being and how behind the traditional checklist of adulthood I would be compared to my peers.
Each stage of a career in science is long, between school and other training. You won’t move at the same speed as people of the same age. You’re still early in your career as a student or trainee when friends of the same age may be mid-career. I don’t think that's always a bad thing.
As you progress in your career as a scientist, it’s impossible to stagnate. You stay young, creative and excited about your work because you’re constantly surrounded by younger colleagues, who are seeing these questions with fresh eyes.
They constantly bring new perspectives and energy, and that is contagious. Despite the ups and downs, I would say if you are a curious person, who appreciates freedom in their work and doesn’t mind a somewhat circuitous career path, science is a career well worthwhile.
Dr. Kanaka Rajan was speaking with Ruairi J Mackenzie for Technology Networks.
About the interviewee:
Kanaka Rajan is an associate professor of neurobiology at Harvard University and founding faculty member at the Kempner Institute for the Study of Natural and Artificial Intelligence. Her research seeks to understand how important cognitive functions – such as learning, remembering and deciding – emerge from the cooperative activity of multi-scale neural processes.
