Carlos Ponce just started a new position as Member of the Faculty at Harvard Medical School. His lab, which studies visual recognition, is moving here from Washington University in St. Louis. Ponce received his MD PhD degree from Harvard Medical School and his Bachelor of Science from the University of Utah. He is originally from Chihuahua, Mexico.
Portrait photo by Anna Olivella
How does it feel to be back at HMS?
It’s fantastic. It’s interesting being back home. I spent a long time here, so it is great to see so many of the people that were important in my formative years. But I do come back a little bit changed. Having had the experience of running a lab at Washington University in St. Louis, with new responsibilities, developing a working laboratory environment and a team, now I have to bring that into the familiar comfort that I have with HMS. Hopefully I can now help advance my new home institution, using some of the experience that I got out there.
How would you describe your work for someone who is not a neuroscientist?
I study visual recognition, that is, how neurons in the brain allow the recognition of objects in the world, objects like faces and cars. This is an exciting time to be studying vision, because there are new tools in machine learning and artificial intelligence that we can use to advance our understanding of the brain.
For example, you might have seen so-called ‘deepfakes’ out in the world of the internet? These are fascinating videos, where you can see of individuals saying words and the exact same speech pattern being transposed and enacted on the faces of famous folks like President Obama. You can say anything and have the avatar repeat those words. That new type of technology is based on so-called generative networks, which are basically image generators – learned neural networks, that can create images that look a lot like those in the real world.
One of the things that we can now do in neuroscience is to use that technology to extract information encoded in the brain. We can pair these generators in a closed loop with neurons in the brain of an animal. The generators are guided by neuronal activity, using an algorithm, into creating images that simulate information the neuron has learned to encode about the world. That tells us which types of information the brain finds most important in the world, a kind of internal model. That will help us identify or develop better automated systems for visual recognition.
What drew you to this topic?
It’s hard to stay away from trying to understand vision, it is so important to our daily experience and perception. I’ve always been interested in vision, since before my PhD, but the integration of computational approaches is new for me. After I finished my PhD and I started working back in the hospital to finish my MD, I came back as a postdoc years later and I realized that machine-learning techniques were becoming a huge part in neuroscience. Neural networks were becoming very, very good, and neuroscientists were bringing them into their research programs. I had to spend my postdoc taking classes and reading books on machine learning just to catch up, but I’m glad I did.
That’s fantastic. You did medical training in addition to the PhD, right?
Yes. I did some medical training, finished my MD and did two years of residency training, although I focus now on research a hundred percent of the time.
Does that training influence your perspective as a researcher now?
Certainly. One of my first exposures to automated visual recognition technology was in my pathology residency. In one of the rotations, in cytopathology, I got to see automated visual recognition systems that help screen for tissue samples with concerning signs. Those get flagged, and then passed onto cytopathologists to observe. I was fascinated by that kind of technology. I wondered why it wasn’t used in more regions of pathology, or in medicine in general, like radiology.
I realized that some of those systems are promising and powerful, but they’re not up to the performance that we see in humans. So, one question coming back to a student of the visual system, like myself, was to understand, why not? What’s missing? I think part of the exciting path forward now is to understand what the brain does, and to take those lessons and try to implement them in automated systems. Systems that can eventually do pattern recognition in medicine, but just faster, and 24 hours a day.
What’s the hardest part about what you’re doing right now in the lab?
I think that the hardest part in general is having time to pursue several interesting questions out there. This is an exciting time for vision. Many developments that occur in machine-learning may have potential impact to neuroscience. There are many ways in which these two threads can come together and help each other. Now running a lab like mine means having individuals that are interested in both worlds, and have the expertise to be able to put them together, and the time to read, and ponder, how all these different subjects can interact and advance the goals of neuroscience, and medicine, and machine-learning? It’s a lot of work, and not enough time.
What did you want to be when you were a child?
I grew up in Mexico, on a farm. I had no understanding of the kind of academic and medical careers that might one day become available to me. I knew that I loved animals. Once my family moved to the United States and I started high school, I understood more about careers. I knew that biology was going to draw and sustain my attention and passion. I think I came a little bit late into understanding the professional world, but it’s interesting to me to look back at the seeds of my interests develop. Now I have my son, who is almost seven years old, and he too loves animals. I wonder if something as basic as an inclination to envelop oneself in nature, and to observe how cool creatures are, can eventually develop into hopefully, a fulfilling career for him too.
What is the trait you admire most in others?
Compassion. Just the ability to go through life without any need to cause harm to other humans, or to animals. As somebody who works with animals, it’s interesting having to find that balance between doing what society has deemed to be important, in terms of advancing health using animals, while at the same time, going back to being the kid on the farm, and feeling like animals are fascinating and deserve our respect. I think that balance is aided by compassion.
What was your favorite animal when you were on the farm?
Well, that’s embarrassing. It’s chickens. They’re lovely creatures. I remember there was one that never would run away when I approached. I’d greet her every day after school – I thought she was friendly. In retrospect, I think this was maladaptive, as there came a day when my grandma decided to make chicken soup…well, you can imagine how that story ended.
What do you do for fun?
Science is what I do for fun! But in addition to science, it’s taking my kids to museums or to the movies. I discovered during the lock-down of the pandemic, I never did this before, it sounds ridiculous…but I also re-discovered hiking. Specifically, just how easy it is to find animals out there in the world, just going out there and lifting up a rock, finding insects everywhere. I lived in Utah and all my friends would always try to get me to go hiking. I was like, “but that’s just walking outside.” If only somebody had pitched it in terms of animals.
And now hiking for fossils! Oh my gosh. Living in Missouri, I learned, there used to be an ocean there. The Western Interior Seaway, I believe it was called. A hundred million years ago. So there are marine fossils everywhere in Missouri. Now the cool thing is you go hiking and find fossils out on the trails. Now I can’t wait to start hiking here in Massachusetts. Also, for fun, I tried cooking and making cocktails. During the pandemic we did an Airbnb virtual experience where we learned how to make Caipirinhas from a bartender in Brazil. And churros from a chef in Mexico City. That’s stuck with us.
Sounds delicious! Are there things I haven’t asked that you’d like to share?
I just want to say that I’m really thrilled and excited to be back. To re-experience the intellectual energy and enthusiasm, and overall kindness of people here in the HMS community, and of the community at large at Harvard. I just can’t wait to start re-engaging with the community, at least as safely as we can, given COVID cautions. That will be great. And recruiting students! I just can’t wait to get started