Stanford Neurosciences Building, John A. and Cynthia Fry Gunn Rotunda, E241
290 Jane Stanford Way, Stanford, CA 94305
View map
This event is over.
Event Details:
Abstract
The highest performing speech BCIs to-date have leveraged neural activity in motor cortex evoked by attempting to physically produce speech. For people with paralysis, attempting to speak can be fatiguing and slow communication; if the paralysis is partial, it can also produce distracting sounds and breath control difficulties. Here, we demonstrate that inner speech (i.e., purely imagined words) also evokes robust and repeatable patterns of neural activity in motor cortex recorded using microelectrode arrays. We show that an intracortical inner speech BCI can decode self-paced imagined sentences with error rates as low as 14% for a 50-word vocabulary and 26% for a 125,000-word vocabulary. We found that inner speech remains decodable even at true conversational rates (145 words per minute). In contrast, none of our participants were able to reach conversational speed using attempted speech, with maximum attempted speeds varying between 59-97 words per minute. We further analyzed the representation of inner speech and find it is highly correlated with attempted speech, though we also identified a neural “motor-intent” dimension that differentiates the two. Finally, we show that even spontaneous inner speech (e.g. when counting objects) can be detected in motor cortex. We demonstrate software design solutions to ensure that private inner speech is not unintentionally decoded, with novel approaches for both attempted speech or inner speech BCIs.
Erin Michelle Kunz
Erin is currently a postdoc in the Stanford Neural Prosthetics Translational Lab (NPTL) with Professors Jaimie Henderson and Frank Willett. Her research focuses on developing high-performance communication neuroprostheses for individuals with impairments resulting from neurodegenerative diseases or stroke. She also aims to better understand the neural mechanisms underlying speech and language.
Erin received her PhD in Electrical Engineering with Professors Shaul Druckmann, Jaimie Henderson and Krishna Shenoy, also at NPTL. She earned her B.S. in Mechanical Engineering and minor in Electrical Engineering & Computer Science (EECS) from UC Berkeley and Master’s in Electrical Engineering at Stanford. Prior to joining NPTL she worked as a software engineer in autonomous vehicle development at General Motors.
About the Center for Neural Data Science Seminar Series
The Center for Neural Data Science Seminar Series is a platform for trainees across campus to share insights and innovative approaches that bridge the gap between neuroscience and data science.
As neuroscience continues to generate vast amounts of data—from intricate neural circuit maps to large-scale brain activity recordings—the need for interdisciplinary expertise in data science, statistics, and engineering has never been more critical.
The Center for Neural Data Science mission is to advance brain research through the development of cutting-edge analytical methodologies and collaborative approaches. Stanford's affiliates are invited to join this vibrant community dedicated to transformative discoveries.
This seminar series is only offered in person. Please note the location change (Gunn Rotunda) for this event only.
Happy Hour/Reception for Attendees:
Seminars attendees are invited to stay for a happy hour reception after the seminar to continue the conversation with the speaker.