2025 Gruber Neuroscience Prize

As a medical student at the University of California, San Francisco, Edward Chang was struck by how little was known about the brain mechanisms underlying language. “In our first-year neuroscience course, it dawned on me that we understood so much about how we breathe or how the heart pumps blood, but almost nothing about how we produce words,” Chang recalls. “It seemed like magic—and a deep mystery.” That mystery became a personal challenge. During medical school, he completed a two-year research fellowship under neuroscientist Michael Merzenich, PhD, a pioneer in brain mapping techniques.

After earning his medical degree, Chang continued at UCSF for his neurosurgical residency, specializing in the care of patients with severe epilepsy and brain tumors. “I was drawn to neurosurgery for three reasons,” he says. “First, I wanted the close relationship with patients that comes from treating serious neurological conditions. Second, I enjoyed the technical challenge of operating on delicate neural structures. And third, I was deeply curious about how the brain works—and I wanted to help advance our understanding of what makes us human.” His training included postdoctoral research under renowned mentors such as neuroscientist Robert Knight, MD, and neurosurgeons Mitchel S. Berger, MD, and Nicholas Barbaro, MD.

A cornerstone of Chang’s research is functional brain mapping, using high-density electrode arrays to identify regions responsible for specific functions. His team has mapped the cortical areas involved in both speech comprehension and production, identifying how the brain processes phonemes (vowels and consonants), vocal pitch and intonation, syllable rhythm, and the filtering of background noise. They also charted the motor areas that control the vocal tract—including the larynx, tongue, and lips—and those involved in planning and sequencing speech.

These discoveries laid the scientific groundwork for a first-of-its-kind clinical trial: a speech neuroprosthesis designed to restore communication for people with paralysis. “We created a system that can safely record from the speech areas of the brain and translate those signals into words,” Chang explains.

The neuroprosthesis uses surgically implanted electrodes on the brain's surface to detect electrical patterns linked to speech. When a patient attempts to speak, the system decodes those patterns and converts them into synthesized words and sentences. Chang’s team demonstrated this approach successfully with a man who had been unable to speak for over 15 years due to a brainstem stroke. Since then, the technology has improved, enabling faster speech synthesis and even avatar-based communication. Other research groups have now replicated key findings using similar methods and algorithms.

“What allowed us to build this technology was a decade of foundational research on the neural code for words,” says Chang. Advances in artificial intelligence were also key, helping to decode brain signals by filtering noise and matching them to intended speech. As a result, patients with severe paralysis have been able to communicate again—simply by trying to speak.

Looking ahead, Chang aims to push the boundaries of resolution in brain mapping. “Over the last fifteen years, we’ve recorded from the brain’s surface at about millimeter resolution,” he says. “Now, we have tools that let us study thousands of individual neurons simultaneously. The science is evolving to a level of detail we've never had before.”