As a young child maturing in the former Soviet Union, Evelina Fedorenko PhD ’07 learned several languages, including English, as her mother aspired for her to have the opportunity to eventually relocate overseas for enhanced prospects.
Her linguistic education not only aided her in crafting a new existence in the United States as an adult but also sparked a lifelong fascination with linguistics and the mechanisms by which the brain processes language. Now an associate professor of brain and cognitive sciences at MIT, Fedorenko examines the brain’s regions responsible for language processing: how they develop, the extent to which they share functions with other cognitive processes, and how each area aids in language understanding and production.
Fedorenko’s foundational work was instrumental in pinpointing the precise areas of the brain involved in language processing, and she has continued to expand upon that research to gain insights into how distinct neuronal populations in these areas execute linguistic computations.
“It took time to cultivate the methodology and determine how to swiftly and reliably identify these areas in individual brains, considering the common issue of brain variability among individuals,” she explains. “Then we just persisted in asking inquiries such as: Does language intersect with other similar functions? How is the system structured internally? Do various components of this network perform different roles? There are numerous inquiries you can pursue, and many avenues we have explored.”
In some of her more recent investigations, she is examining how the brain’s language-processing areas develop early in life, through research involving very young children, individuals with atypical brain structures, and computational frameworks known as large language models.
From Russia to MIT
Fedorenko spent her childhood in Volgograd, Russia, which was part of the Soviet Union at the time. Following the dissolution of the Soviet Union in 1991, her mother, a mechanical engineer, lost her position, and the family faced difficulties in making financial ends meet.
“It was a truly intense and challenging period,” Fedorenko reminisces. “However, one enduring source of stability for me was the immense love I received from my parents, grandparents, and aunt and uncle. That was incredibly meaningful and instilled in me the confidence that if I put in the effort and had a goal, I could realize my dreams.”
Fedorenko dedicated herself to her studies, focusing on English, French, German, Polish, and Spanish, and also participated in mathematics competitions. At the age of 15, she spent a year attending high school in Alabama, as part of a program that connected students from the former Soviet Union with American families. Initially, she had contemplated applying to universities in Europe but revised her plans upon discovering that the American higher education system provided greater academic flexibility.
After gaining admission to Harvard University with a comprehensive scholarship, she returned to the United States in 1998 and attained her bachelor’s degree in psychology and linguistics, all while juggling multiple jobs to remit funds to support her family.
While at Harvard, she also enrolled in courses at MIT and ultimately decided to apply to the Institute for her graduate studies. For her doctoral research at MIT, she collaborated with Ted Gibson, a professor of brain and cognitive sciences, and subsequently, Nancy Kanwisher, the Walter A. Rosenblith Professor of Cognitive Neuroscience. She began her research with functional magnetic resonance imaging (fMRI) to investigate brain areas that seemed to respond preferentially to music, but soon redirected her focus to studying brain activity in response to language.
She discovered that collaborating with Kanwisher, who specializes in the functional organization of the human brain but had not extensively researched language previously, helped Fedorenko craft a research program free from potential biases inherent in some of the early studies on language processing in the brain.
“We essentially started from the ground up,” Fedorenko states, “integrating the expertise in language processing I gained from working with Gibson and the rigorous neuroscience techniques that Kanwisher established while examining the visual system.”
After completing her PhD in 2007, Fedorenko remained at MIT for several years as a postdoctoral researcher funded by the National Institutes of Health, furthering her investigations with Kanwisher. During that period, she and Kanwisher devised methods to identify language-processing areas across various individuals and uncovered new evidence that specific brain regions react selectively to language. Fedorenko then spent five years as a research faculty member at Massachusetts General Hospital before receiving an invitation to join the MIT faculty in 2019.
How the brain processes language
Since establishing her lab at MIT’s McGovern Institute for Brain Research, Fedorenko and her team have made numerous findings that have advanced neuroscientists’ understanding of the brain’s language-processing areas, which are distributed across the left frontal and temporal lobes.
In a series of experiments, her lab demonstrated that these regions show a high degree of specificity for language and are not activated by activities such as listening to music, reading computer code, or interpreting facial expressions, all of which have been argued to share similarities with language processing.
“We’ve distinguished the language-processing mechanism from various other systems, including the general fluid reasoning system, as well as the systems for social perception and reasoning, which assist in processing communicative signals, such as facial expressions and gestures, as well as reasoning about others’ beliefs and motivations,” Fedorenko clarifies. “So, that was a noteworthy discovery; this system is indeed its own distinct entity.”
More recently, Fedorenko has redirected her focus towards determining, in greater detail, the specific roles of different components of the language processing network. In one recent investigation, she identified distinct neuronal populations within these areas that seem to showcase various temporal windows for processing linguistic information, extending from just one word to six words.
She is also examining how language-processing circuits emerge in the brain, with ongoing research wherein she and a postdoctoral researcher in her lab are using fMRI to scan the brains of young children, observing how their language regions function even before the children have completely mastered speech and comprehension.
Large language models (similar to ChatGPT) can facilitate these developmental inquiries, as researchers can gain better control over the language inputs to the model and have ongoing access to its capabilities and representations at various stages of learning.
“You can train models in diverse ways, on various types of language, following different training regimens. For instance, training on simpler language initially and then progressing to more complex language, or on language coupled with some visual inputs. Then you can analyze the performance of these language models on a range of tasks, while also investigating changes in their internal representations throughout the training process, to assess which model most accurately reflects the trajectory of human language acquisition,” Fedorenko explains.
To gain another perspective on how the brain develops linguistic abilities, Fedorenko initiated the Interesting Brains Project several years back. Through this initiative, she is researching individuals who experienced some form of brain damage early in life, such as a prenatal stroke, or structural abnormalities due to congenital cysts. In some cases, these individuals sustained damage to or significantly altered the customary language-processing areas of the brain, yet all of them are cognitively indistinguishable from those with typical brain structures: they managed to learn to articulate and comprehend language as usual, and in some instances, they were unaware of their atypical brain conditions until they reached adulthood.
“That research centers around brain plasticity and redundancy, seeking to uncover what the brain can accommodate, and how” Fedorenko remarks. “Are there multiple pathways to construct a human mind, even when the neural basis appears so different?”