ChangLabUCSF
@ChangLabUcsf
Followers
3K
Following
12
Media
44
Statuses
144
Chang Lab at UCSF, human brain, speech, brain-computer-interfaces, neurosurgery
San Francisco, CA
Joined August 2020
Our latest work on the neural mechanism for stopping speech production is published! See a brief summary below and the original paper linked to the post at the bottom.
In natural conversations, people can stop speaking at any time. How? Using high-density electrocorticography, Zhao et al. find a distinct neural signal in the human premotor cortex that inhibits speech output to achieve abrupt stopping. @ChangLabUcsf.
1
13
53
Our models also showed stable performance without retraining for ~2 months and these results were achieved ~4 years after ECoG implantation. We hope these findings can be scaled to more patients in the near future!
0
2
3
We leveraged this finding to demonstrate transfer learning across languages. Data collected in a first language could significantly expedite training a decoder in the second language, saving time and effort for the user.
1
2
4
Cortical activity instead represented the intended vocal-tract movements of the participant, irrespective of the language. This allowed us to train a model that generalized across a shared set of English and Spanish syllables.
1
2
2
Despite learning English later in life, we found that cortical activity was largely shared across languages, with no clear differences in magnitude or language-selective neural populations.
1
2
4
Our system decoded cortical activity, measured with ECoG over the IFG and SMC, into English and Spanish sentences. The intended language was not set by the user, instead it was freely decoded from cortical activity and language models.
1
4
10
Speech decoding has primarily been shown for monolinguals but half the world is bilingual with each language contributing to a person’s personality and worldview. There is a need to develop decoders that let bilinguals communicate with both languages.
2
3
3
Work led by @asilvaalex1 with mentors and co-authors @jessierliu , @SeanMetzger5 , @bhaya_ilina , @KayloLittlejohn , @AtDavidMoses , Max Dougherty, Margaret Seaton, and Edward Chang.
1
2
3
Excited to share our work on developing a bilingual speech neuroprosthesis that decodes cortical activity into English and Spanish sentences in a person with paralysis. Out today in @natBME!
3
24
133
Reminder to those at #Cosyne2024 that @asilvaalex1 will be presenting his work on ***a bilingual speech neuroprosthesis*** ! Poster Session 2 #131 at 12:30!.
0
0
6
Chang lab is at #Cosyne2024!.🗣️Thurs, posters 1 #083: @ItzikNorman neural oscillations during speech production. 💬Fri, posters 2 #131: @asilvaalex1 bilingual speech neural prostheses. 🤖Sat, talk session 9: @shaileeejain modeling single neurons in humans with deep neural nets.
0
3
22
Contrary to prior work, we found no evidence for a music-region. Instead, populations selective for music were intermixed with other sound-responsive populations across a broad swath of auditory cortex. [7/7].
0
0
3
Neural populations that selectively responded to music (over speech) were driven by the encoding of expectation, indicating domain specialization for representing the statistical structure of melody. [6/7].
1
0
5
To what extent is this encoding specialized for music versus general auditory processing? To answer this, we also recorded neural activity while the same subjects listened to speech. [5/7].
1
0
1
Within the superior temporal gyrus, distinct dimensions of melody were represented across a spatial map – where the pitch, pitch-change, and expectation of notes were encoded in separate neural populations. [4/7]
1
0
1
What is the neural code that underlies our perception of melody? To answer this, we recorded high-density intracranial activity from the cortical surface as subjects listened to musical phrases. [3/7].
1
0
1
Melody is a core component of music in which discrete pitch-units (notes) are serially ordered to convey emotion and meaning. Humans perceive not only the pitch of notes, but pitch-changes between them and their statistical expectation given prior notes. [2/7].
1
0
1
Excited to share our work characterizing the encoding of melody using direct recordings from the human auditory cortex! Out today in @ScienceAdvances. Led by @nsankaran72 together with mentors @matt_k_leonard, @FrdricTheuniss1 and Eddie Chang [1/7].
4
29
130
This is just the beginning… Neuropixels and similar technologies will help us crack the neural code of language. [6/6].
0
1
12
At the same time, each vertical ‘column’ of neurons also contained a surprisingly diverse set of neurons tuned to many different types of speech features. [5/6].
1
1
3