@JakeGusman

Postdoctoral research scientist at @BrainGateTeam. I post science and shower thoughts.

Joined July 2012
  • Tweets50
  • Following208
  • Followers107
15Media
I’m excited to share that one of the manuscripts from my PhD work, “Observation-Related Activity in the Human Motor Cortex Increases with Effector Anthropomorphicity,” is now published in @PNASNews. (1/12) Link: pnas.org/doi/10.1073/pnas.25… #NeuroTech #BCI @BrainGateTeam
5
5
2
28
4,030
more replies
Within seconds of T11 “figuring it out”, neural responses to watching the dot stimulus became much stronger – and more hand-like. This suggests that mirror-like activity is not just a bottom-up process but can be substantially affected by top-down contextual cues as well. (11/12)
1
3
68
Huge thank you to the BrainGate participants, especially T11 and T17, and to all of my co-authors, collaborators, and mentors. 🙏🧠 (12/12) pnas.org/doi/10.1073/pnas.25…

Observation-related activity in the human motor cortex increases with effector anthropomorphicity |...

Neurons in the motor cortex can be engaged not only in motor execution but also during observation of movements performed by other anthropomorphic ...

pnas.org
3
52
Congrats to @JakeGusman, Carlos Vargas-Irwin, @neuroleigh, et al. on their publication in @PNASNews , “Observation-related activity in the human motor cortex increases with effector anthropomorphicity”! pnas.org/doi/10.1073/pnas.25… @brownengin @MGH_CNTR @CarneyInstitute #BCI #neuro

Observation-related activity in the human motor cortex increases with effector anthropomorphicity |...

Neurons in the motor cortex can be engaged not only in motor execution but also during observation of movements performed by other anthropomorphic ...

pnas.org
1
11
440
The Latch decoder enabled major improvements in a challenging closed-loop multi-gesture drag-and-drop task. Participant T11 dragged icons across a screen with no gesture misdecodes in 79% of trials - compared to just 3% if a Gesture decoder was used on its own. (6/9)
1
2
76
more replies
This makes the Latch decoder a versatile and generalizable approach for neural control of sustained actions — with potential to enable not only better 'drag-and-drop' with a cursor, but also more intuitive and reliable operation of assistive and rehabilitative robotics. (8/9)
1
2
62
Huge thanks to co-first author Tommy Hosman and the rest of my excellent co-authors, collaborators, and reviewers. And of course to the participants who make this exciting work possible. 🙏🧠 (9/9)
2
56
Decoding sustained actions like grasping and holding an object can be a challenge for iBCIs. Intracortical signals from motor cortex tend to reflect transient, dynamic phases of movement — not stable postures or holds — making tasks like “drag-and-drop” of a cursor tricky. (2/9)
1
1
78
more replies
So we developed the “Latch” decoder, which uses two parallel LDA-HMM decoders: 1️⃣ A Gesture decoder that classifies the specific gesture being attempted 2️⃣ An Attempt decoder that latches the output to the initially decoded gesture as long as an overall ‘attempt’ is decoded (5/9)
1
2
49
Sweeping a standard, multi-class LDA decoder over trial durations revealed a marked decrease in the discriminability of gestures sustained beyond 1s. But a binary attempt/no attempt LDA was more reliably accurate over longer gesture attempts. (4/9)
1
54
We evaluated neural activity from the precentral gyrus in two BrainGate participants using a “Gesture Hero” task — a game-like paradigm that asked participants to hold up to 7 different hand gestures for different durations. (3/9)
1
1
61
Load more