DARPA: read 10^6 neurons, write 10^5 neurons
DARPA wants a dream neural interface. They can wait 4 years, but they want a complete plan now (abstracts due Feb 25, full proposals April 14).
This is a direct quote from the announcement for the Neural Engineering System Design (NESD) program.
Successful NESD proposals must culminate in the delivery of complete, functional, implantable neural interface systems and the functional demonstration thereof. The final system must read at least one million independent channels of single-neuron information and stimulate at least one hundred thousand channels of independent neural action potentials in real-time. The system must also perform continuous, simultaneous full-duplex interaction with at least one thousand neurons. While DARPA desires a single 1 cm3 device that satisfies all of these capabilities (read, write, and full-duplex), proposers may propose a design wherein each capability is embodied in separate 1 cm3 devices. Proposed implementations must not require tethers or percutaneous connectors for powering or facilitating communication between the implanted and external portions of the system.
P.S. They want FDA pre-submission happening in about a year or so into the project.
(alternative link, in case the one above doesn’t work)
I am terrify.
Thanks for pointing out this interesting call for submissions…
To me, the project sounds completely unrealistic, almost every single detail. Is this normal for a Darpa call for projects? And I wondoer who is responsible for writing this kind of document …
I cannot imagine a researcher having the chutzpah to write “I can deliver this device” in his proposal …
Funnily, the most precise detail that I found in the document was this: “For example, voltage-sensitive proteins, the …” – He could at least mentioned calcium sensors instead. VSPs do not really work yet in regular in vivo applications, and to my knowledge nobody has experience with viral delivery of activity indicators or rhodopsins to human brains. Seems like somebody lives really far from reality…
Phillip Alvelda (MobiTV) is the lead on the program. Here are a couple of YouTube videos you can watch to get an idea of how he thinks about this:
https://www.youtube.com/watch?v=z3AuOVyHFCE
https://www.youtube.com/watch?v=vjac3RBoK1c
Did anything happen there? The proposal is obviously completely nuts. It even requires it done in human cortex, which is even more difficult. Seems to be particularly aimed at some Schnitzer/Inscopix device… Since it’s never going to work: are there consequences for not delivering?