Understanding Reality Through Algorithms | MIT news

Although Fernanda de la Torre still has several years ahead of her postgraduate studies, she is already dreaming when it comes to what the future holds for her.

“I dream of one day opening a school where I can bring this world of cognition and perception to places that could never have a connection to this,” she says.

This kind of ambitious thinking is what got de la Torre, a doctoral student in the Department of Brain and Cognitive Sciences at MIT, to this point. Recently awarded the prestigious Paul and Daisy Soros Fellowship for New Americans, De La Torre has found MIT a supportive and innovative research environment that has allowed her to delve into cutting-edge AI science. But she is still driven by an innate curiosity about the human imagination and a desire to bring that knowledge to the societies in which she grew up.

An unconventional path to neuroscience

De la Torre’s first exposure to neuroscience was not in the classroom, but in her daily life. As a child, she watched her younger sister suffer from epilepsy. At the age of 12, she crossed the border into the United States from Mexico illegally to meet her mother, exposing her to an entirely new language and culture. Once in the United States, she had to deal with her mother’s changing personality in the midst of an abusive relationship. “All these different things I was seeing around me made me want to better understand how psychology works, to understand how the mind works, how we can all be in the same environment and feel very different things,” says de la Torre.

But finding an outlet for that intellectual curiosity was difficult. As an undocumented immigrant, her access to financial assistance has been limited. Her high school was also underfunded and lacked elective options. Despite this, the mentors along the way encouraged the aspiring scientist, and through a program in her school, she was able to take courses at the community college to fulfill basic educational requirements.

It took a great deal of dedication to her education, but de la Torre transferred to Kansas State University for her undergraduate studies, where she majored in computer science and mathematics. In Kansas, she was able to get her first real taste of research. “I was fascinated by the questions they were asking and this whole space that I didn’t encounter,” De La Torre says of her experience working in the Visual Perception Lab and discovering the field of computational neuroscience.

Although Kansas State did not have a specialized neuroscience program, her research experience in cognition led her to a machine-learning lab led by William Hsu, a professor of computer science. There, De La Torre became fascinated by the possibilities of using computation to model the human brain. Hsu’s support also convinced her that a scientific life was possible. “It always made me feel like I could handle the big questions,” she says fondly.

With her trust at Kansas State, de la Torre came to MIT in 2019 as a post-baccalaureate student in the lab of Tommaso Poggio, Eugene McDermott Professor of Brain and Cognitive Sciences and an investigator at the McGovern Institute for Brain Research. With Poggio, who is also director of the Center for Brains, Minds, and Machines, De La Torre began work on deep learning theory, an area of ​​machine learning that focuses on how artificial neural networks modeled on the brain can learn to recognize patterns and learn.

“It is a very interesting question because we are starting to use them everywhere,” de la Torre says of neural networks, listing examples from self-driving cars to medicine. “But, at the same time, we don’t fully understand how these networks can go from knowing nothing to being just a bunch of numbers to making things out of sense.”

Her post-baccalaureate experience was De La Torre’s first real opportunity to apply the technical computer skills she developed when she was an undergraduate in neuroscience. It was also the first time she could fully focus on research. This was the first time I had health insurance and a fixed salary. This was, in and of itself, a kind of life changer, she says. But in terms of research, it was very intimidating at first. I was worried, and I wasn’t sure I belonged here.”

Fortunately, de la Torre says she has been able to overcome these insecurities, through an unabashedly growing enthusiasm for the field and through the support of Poggio and other colleagues in the Department of Brain and Cognitive Sciences at MIT. When I had the opportunity to apply to the department’s PhD program, I jumped at it. “It was just knowing that these kinds of mentors are here and that they care about their students,” de la Torre says of her decision to stay at MIT for graduate studies. “This was really helpful.”

Expand the concepts of reality and imagination

During her two years so far in her graduate program, De La Torre’s work has broadened the understanding of neural networks and their applications to the study of the human brain. Working with Guangyu Robert Yang, associate investigator at the McGovern Institute and assistant professor in the departments of Brain and Cognitive Science and Electrical Engineering and Computer Science, she has engaged in what she describes as more philosophical questions about how one develops a sense of self. as an independent object. She is interested in how this self-awareness developed and why it might be beneficial.

However, de la Torre’s principal advisor is Professor Josh McDermott, who leads the Computational Testing Laboratory. With McDermott, De La Torre is trying to understand how the brain integrates vision and sound. While incorporating sensory input may seem like a basic process, there are many unanswered questions about how our brains integrate multiple signals into a coherent impression or perception of the world. Audiovisual illusions raise many questions in which what we hear changes what we see. For example, if one watches a video clip of two disks passing each other, but the clip contains the sound of a collision, the brain will perceive that the disks are bouncing, rather than passing through each other. Given the ambiguous image, this simple auditory cue is all it takes to create a different perception of reality.

“There’s an interesting thing going on where our brains receive two signals that tell us different things, however, we have to somehow combine them to make sense of the world,” she says.

De La Torre uses behavioral experiments to explore how the human brain understands multisensory signals to construct a specific perception. To do this, she created different scenes of objects interacting in a three-dimensional space on different sounds, and asked the research participants to describe the characteristics of the scene. For example, in one experiment, it collected images of a mass moving across a surface at different speeds and different scraping sounds, and asked participants to estimate how rough the surface was. Ultimately, she hopes to bring the experience into virtual reality, where participants will actually push blocks in response to how hard they perceive a surface, rather than just reporting what they are experiencing.

Once you’ve collected data, you’ll move on to the modeling phase of research, evaluating whether multisensory neural networks perceive illusions the way humans do. “What we want to do is model exactly what’s going on,” says de la Torre. “How do we perceive these two signals, integrate them, and at the same time, use all our prior knowledge and inferences of physics to truly understand the world?”

Although my two threads of research with Yang and McDermott may look different, she sees clear connections between the two. Both projects are about understanding what artificial neural networks can do and what they tell us about the brain. On a more fundamental level, she says, how the brain perceives the world through various sensory cues may be part of what gives people a sense of self. Perception is about building a coherent and unified sense of the world from multiple sources of sensory data. Likewise, she argues, “The sense of self is really a combination of actions, plans, goals, emotions, all these different things that are components of their own, but somehow create a unified being.”

It feels right to De La Torre, who is working to understand and integrate different aspects of her life. Working in a computer test lab, for example, she began experimenting with combining electronic music with folk music from her native Mexico, connecting her “world,” she says. Having the space to do these kinds of intellectual explorations, and the colleagues who encourage them, is one of De La Torre’s favorite parts of MIT.

“Apart from the professors, there are also a lot of students whose way of thinking amazes me,” she says. “I see a lot of the good and the excitement of science and a little bit of that — it’s not intelligence, but a love of very specialized things — and I kind of like that.”

Leave a Comment