Bliss Chapman

Brain-interface software lead at Neuralink, responsible for the decoding pipeline and user-facing application that turn a patient’s neural signal into cursor control. He personally led the onboarding of Neuralink’s first human participant, Noland Arbaugh, monitoring live signal during the implant surgery itself and running the earliest sessions distinguishing which imagined movements produced usable motor-cortex activity.

Core positions

Chapman’s central technical claim is that a decoder trained to predict a user’s higher-level intention — the goal a movement is aimed at — outperforms a decoder trained on literal limb kinematics, even in monkey studies where the true intended action is independently known. This intention-over-action thesis is, in his account, the reason Neuralink’s software can give a paralysed patient effective cursor control despite reading no actual movement at all.

He describes his design philosophy as ‘UX is how it works, not decoration’ — treating interface details (adjustable cursor gain and friction, bias correction, features like Quick Scroll) as core to the product rather than cosmetic. He set Neuralink’s internal Webgrid benchmark record at 17 bits per second, well above the pre-Neuralink human record of roughly 4.2–4.6 BPS, and reports Neuralink’s end-to-end latency from neural spike to cursor movement at about 22 milliseconds — faster than the roughly 75 milliseconds a signal takes to move a human hand by muscle.

In the wiki

• Elon Musk and the Neuralink Team on Brain-Computer Interfaces, the N1 Implant, and the First Patient — explains the neural decoder, the intention-versus-action distinction, calibration, latency, and the Webgrid benchmark
• Brain-Computer Interface — concept; the software/decoding side of the mechanism