Understanding reality through algorithms | MIT News

Fernanda de la Torre still has a few years of postgraduate studies left, but is already dreaming big when it comes to what the future holds.

“I dream of one day opening a school where we can bring this world of cognition and perception into places we’ve never touched,” she says.

It’s such an ambitious idea that De La Torre, a PhD student in the Department of Brain and Cognitive Sciences at MIT, has gotten to this point. De la Torre, who was recently awarded the prestigious Paul and Daisy Soros New American Fellowship, has found at MIT a supportive and creative research environment that allows him to delve into the cutting-edge science of artificial intelligence. But she is still driven by her innate curiosity about the human imagination and her desire to bring that knowledge to the communities in which she grew up.

An unconventional path to neuroscience

De La Torre’s first exposure to neuroscience wasn’t in the classroom, but in everyday life. During her childhood, she watched her sister suffer from epilepsy. When she was 12, she illegally left Mexico to reunite with her mother, exposing her to a whole new language and culture. Once in America, she had to contend with her mother’s changing personality in an abusive relationship. It made me want to better understand how it works,” says de la Torre. She feels something very different.

But finding an outlet for that intellectual curiosity was difficult. As an undocumented immigrant, her access to financial assistance was limited. She was also underfunded in her high school and had no electives. But along the way, mentors encouraged her aspiring scientist, and through her school’s program she was able to take community college courses to meet her basic educational requirements.

Her education took a lot of effort, but De La Torre attended Kansas State University, majoring in computer science and mathematics. At Kansas State University, she was able to get her first real taste of research. De La Torre said of her experience working in the Visual Perception Lab and discovering the field of computational neuroscience, “I was just fascinated by the questions they were asking and this whole space I had never encountered before. I was there,” he said.

Kansas State University didn’t have a dedicated neuroscience program, but her cognition research experience led her to a machine learning lab led by computer science professor William Hsu. There De La Torre became fascinated by the possibility of using her computations to model the human brain. Her Sue’s support also convinced her that her career as a scientist was a possibility. “He always made me feel like I could tackle big problems,” she says affectionately of hers.

After gaining confidence at Kansas State University, De La Torre joined Massachusetts Institute of Technology (MIT) in 2019 as a post-bachelor’s student in the lab of Tommaso Poggio (Eugene McDermott Professor of Brain and Cognitive Science). As a researcher at the McGovern Brain Institute. De la Torre began working on deep learning theory with Poggio, who is also director of the Center for Brain Mind and Machine. This is a branch of machine learning that focuses on how artificial neural networks modeled after the brain recognize and learn patterns.

“It’s a very interesting question because it’s starting to be used everywhere,” says De La Torre of Neural Networks, citing examples ranging from self-driving cars to healthcare. “But at the same time, we don’t fully understand how these networks can output anything meaningful out of just a bunch of ignorant numbers.”

Her experience as a postback was De La Torre’s first real opportunity to apply the technical computer skills she had acquired as an undergraduate to neuroscience. It was also the first time she was able to fully focus on her studies. “It was the first time I had health insurance and a steady paycheck. “But on the research side, it was really scary at first. I was insecure and unsure if I should be here.”

Fortunately, De La Torre says that both an unabashed enthusiasm for the field and the support of Poggio and other colleagues in MIT’s Brain and Cognitive Sciences Division helped overcome these fears. increase. When the opportunity arose to apply to her PhD in the department, she jumped at it. “I knew there were these kinds of mentors here and they cared about the students,” De La Torre said of her decision to stay at MIT for her graduate studies. talking about decisions. “It was really meaningful.”

Expanding the concept of reality and imagination

De La Torre’s work during the previous two-year graduate program extended the understanding of neural networks and their applications to the study of the human brain. Working with Gwangwoo Robert Yang, Associate Fellow at the McGovern Institute and Assistant Professor of Brain and Cognitive Science, Electrical Engineering and Computer Science, she explored more philosophically how to develop a sense of self. It addresses what she describes as a serious problem. as an independent entity. She is interested in how that self-consciousness develops in her and why it helps.

However, De La Torre’s primary advisor is Professor Josh McDermott, who heads the Laboratory for Computational Audition. De La Torre is working with his McDermott to understand how the brain integrates sight and sound. Combining sensory inputs may seem like a basic process, but much is known about how our brains combine multiple signals to form coherent impressions and perceptions of the world. There is an open issue of Many questions are posed by the audiovisual illusion that what we hear changes what we see. perceives the disks as bouncing off each other rather than passing each other. Given ambiguous images, simple auditory cues can produce different perceptions of reality.

“An interesting thing is happening that our brains are receiving two signals that tell us different things, but we have to combine them in some way to make sense of the world,” she says. .

De La Torre uses behavioral experiments to explore how the human brain understands multisensory cues and constructs specific perceptions. To do so, she created different scenes in which objects interacted with different sounds in her 3D space and asked study participants to describe the features of the scenes. For example, in one experiment, participants were asked to estimate the roughness of a surface by combining blocks moving across a surface at different velocities with different scraping sounds. She hopes to eventually bring her experiments into virtual reality. There, participants physically push blocks according to how rough the surface feels, rather than simply reporting what they have experienced.

Once the data are collected, we proceed to the modeling stage of our research to assess whether our multisensory neural networks perceive illusions in the same way humans do. “What we want to do is model exactly what’s going on,” says De La Torre. “Why do we these he receives two signals, integrates them, and at the same time uses all prior knowledge and reasoning about physics to really understand the world?”

Her two studies with Yang and McDermott may seem different, but she sees a clear connection between the two. Both projects aim to understand how artificial neural networks work and what they can tell us about the brain. At a more fundamental level, she says, how the brain perceives the world from various sensory cues can be part of what gives people a sense of self. is about building a cohesive, unified sense of the world from multiple sources of sensory data. Similarly, she claims that:

It’s a fitting sentiment for De La Torre, who has worked to understand and integrate different aspects of her life. For example, working at the Computational Audition lab, she began her experiments by combining electronic music with folk music from her native Mexico, uniting “two worlds,” she says. Having a space to undertake this kind of intellectual exploration, and having colleagues who encourage it, is one of De La Torre’s favorite parts of her time at MIT.

“Besides professors, there are a lot of other students whose thinking surprises me,” she says. “I see a lot of goodness and excitement in science, but it’s not nerdiness, it’s a love of very niche things. I love that.”