Patrick Kaifosh
More actions
Patrick Kaifosh is a neuroscientist and a co-founder of CTRL-labs, the startup that developed a wrist-worn surface electromyography (sEMG) input device and was acquired by Facebook in 2019.[1][2] He was the company's chief scientist, and by 2025 he was director of research science at Meta's Reality Labs.[1][3]
Kaifosh and Thomas Reardon are the equal-contribution lead authors and the corresponding authors of "A generic non-invasive neuromotor interface for human-computer interaction", published in Nature in July 2025. The paper describes an sEMG wristband and decoding models, trained on data from thousands of participants, that generalize across people and work for new users out of the box.[4] Meta's research on sEMG led to the Meta Neural Band, the wristband sold with Meta Ray-Ban Display glasses from 30 September 2025.[5]
Education and neuroscience research
According to the Mathematics Genealogy Project, Kaifosh earned a PhD from Columbia University in 2016 with a dissertation titled "The neural circuit basis of learning", advised by the theoretical neuroscientist Laurence F. Abbott.[6] A 2016 paper in Neuron gives his affiliation as Columbia's Department of Neuroscience and Center for Theoretical Neuroscience, and states that he was supported by a Howard Hughes Medical Institute International Student Research Fellowship and by NSERC PGS M and PGS D awards during that work.[7]
Several of his graduate publications were written with Attila Losonczy's laboratory at Columbia. Kaifosh was first author of a 2013 Nature Neuroscience brief communication that used two-photon calcium imaging to record the axon terminals of GABAergic neurons projecting from the medial septum to hippocampal area CA1 in awake mice, and of a 2016 Neuron modeling paper, written with Losonczy, proposing a circuit model in which nonlinear interactions between separate dendritic input pathways of CA3 and CA1 pyramidal cells support memory.[8][7] He was also second author of a 2014 Science paper by Matthew Lovett-Barron and colleagues, which found that dendrite-targeting interneurons in CA1 suppress aversive sensory input to pyramidal cells and are needed for contextual fear learning in mice.[9]
With Jeffrey Zaremba, Nathan Danielson and Losonczy he wrote SIMA, an open-source Python package for analyzing fluorescence imaging data. It corrects motion artifacts in laser-scanning microscopy, segments images into regions of interest and extracts their signals, and comes with a graphical tool called ROI Buddy for editing regions and matching them across imaging sessions.[10] Outside the hippocampus, he co-authored two 2014 studies with Nathaniel Sawtell's group on how weakly electric fish predict the sensory consequences of their own motor commands, one of them also with Abbott.[11][12]
CTRL-labs
Reardon, Kaifosh and Tim Machado founded CTRL-labs after doctoral work in neuroscience at Columbia; VentureBeat gives the founding year as 2015.[2][13] In a 2018 interview with TechCrunch, Reardon said the three were wrapping up their graduate studies when they decided to start a company together, and he introduced Kaifosh as the company's chief scientist. The same article includes a late-2017 CTRL-labs demonstration of Kaifosh playing Asteroids with neural control.[1]
US patent assignment records show an earlier company name. An application filed in July 2017 by Kaifosh, Machado, Reardon, Erik Schomburg, Joshua Merel and Steven Demers was assigned to Cognescent Corporation in August 2017, and Cognescent changed its name to CTRL-Labs Corporation in October 2017. That patent, granted in May 2021 as US 11,000,211 B2, covers adapting the mapping from wearable neuromuscular sensor signals to device commands using a model of the user's intent.[14] Kaifosh is also a named inventor, with Alexandre Barachant and Daniel Wetmore, on US 10,772,519 B2 (granted September 2020), which describes deriving control channels from the activity of individual motor units or groups of motor units and using them to operate a device.[15]
In a February 2019 talk at the Redwood Center for Theoretical Neuroscience at UC Berkeley, where he was listed under Columbia University and CTRL-labs, Kaifosh described the company's wireless forearm sEMG system with dry electrodes that needs no skin preparation. According to the talk abstract, it detects individual motor unit action potentials and predicts wrist and hand joint angles, muscle tensions and forces in real time. The abstract argues that neuromotor interfaces can add bandwidth and reduce latency because sEMG signals "precede forces and movement by tens of milliseconds".[16]
Facebook announced on 23 September 2019 that it would acquire CTRL-labs and bring it into its Facebook Reality Labs division. Andrew Bosworth, Facebook's vice president of AR and VR, said the goal was to build this kind of technology at scale and get it into consumer products faster.[17][2] The two patents above were assigned from CTRL-Labs Corporation to Facebook Technologies in January 2020; the assignee was renamed Meta Platforms Technologies in 2022.[14][15]
Meta Reality Labs
As of July 2025, Kaifosh was director of research science at Reality Labs at Meta, and The Register described the sEMG team as led by Kaifosh and research vice president Thomas Reardon.[3] Their Nature paper was received on 23 February 2024, accepted on 6 June 2025 and published online on 23 July 2025, with the author byline "Patrick Kaifosh, Thomas R. Reardon & CTRL-labs at Reality Labs". The two are listed as equal contributors and as corresponding authors, and the author contributions credit Kaifosh with senior leadership, manuscript drafting, experimental design and execution, and work on data processing, deep learning architectures, model training and data analysis.[4] A preprint with a nearly identical title appeared on bioRxiv in February 2024.[18]
The paper reports closed-loop median results for test users of 0.66 target acquisitions per second in a continuous navigation task, 0.88 gesture detections per second in a discrete-gesture task, and handwriting at 20.9 words per minute. Personalizing the handwriting models improved decoding performance by 16%. The authors describe it as, to their knowledge, the first high-bandwidth neuromotor interface with out-of-the-box generalization across people.[4] With the paper, Meta released a public dataset of more than 100 hours of sEMG recordings from over 300 participants, and said it had prototyped an sEMG wristband with Orion, which it calls its first pair of true augmented reality glasses.[19]
In press coverage of the paper, Kaifosh told The New York Times that the system "can work with a new user it has never seen data for", as quoted by the Spyglass newsletter.[20] Euronews quoted him saying he expected "the technology to go a great deal farther", and that "these scaling curves continue as you get more people, more participants, data, it gets better and better."[21]
Meta describes the Meta Neural Band as "the product of years of surface electromyography (EMG) research" involving close to 200,000 participants. It went on sale with Meta Ray-Ban Display at limited US retailers on 30 September 2025, starting at US$799 for the glasses and band together.[5]
Selected publications
| Year | Title | Authors | Venue |
|---|---|---|---|
| 2013 | "Septo-hippocampal GABAergic signaling across multiple modalities in awake mice" | Kaifosh, Lovett-Barron, Turi, Reardon, Losonczy | Nature Neuroscience, vol. 16[8] |
| 2014 | "Dendritic Inhibition in the Hippocampus Supports Fear Learning" | Lovett-Barron, Kaifosh, et al. | Science, vol. 343[9] |
| 2014 | "SIMA: Python software for analysis of dynamic fluorescence imaging data" | Kaifosh, Zaremba, Danielson, Losonczy | Frontiers in Neuroinformatics, vol. 8[10] |
| 2014 | "A temporal basis for predicting the sensory consequences of motor commands in an electric fish" | Kennedy, Wayne, Kaifosh, Alviña, Abbott, Sawtell | Nature Neuroscience, vol. 17[11] |
| 2016 | "Mnemonic functions for nonlinear dendritic integration in hippocampal pyramidal circuits" | Kaifosh, Losonczy | Neuron, vol. 90[7] |
| 2025 | "A generic non-invasive neuromotor interface for human-computer interaction" | Kaifosh, Reardon, CTRL-labs at Reality Labs | Nature, vol. 645[4] |
See also
References
- ↑ 1.0 1.1 1.2 Alice Lloyd George (2018-11-01). "Thomas Reardon and CTRL-Labs are building an API for the brain". TechCrunch. https://techcrunch.com/2018/11/01/thomas-reardon-and-ctrl-labs-are-building-an-api-for-the-brain/. Retrieved 2026-10-13.
- ↑ 2.0 2.1 2.2 Kyle Wiggers (2019-09-23). "Facebook acquires Ctrl-labs to advance its brain-machine interface efforts". VentureBeat. https://venturebeat.com/ai/facebook-acquires-ctrl-labs-to-advance-its-brain-machine-interface-efforts. Retrieved 2026-10-13.
- ↑ 3.0 3.1 Lindsay Clark (2025-07-23). "Meta eggheads demo Bluetooth wristband that decodes muscle twitches for UI control". The Register. https://www.theregister.com/2025/07/23/meta_wristband_ui_control/. Retrieved 2026-10-13.
- ↑ 4.0 4.1 4.2 4.3 Patrick Kaifosh, Thomas R. Reardon, CTRL-labs at Reality Labs, et al. (2025-07-23). "A generic non-invasive neuromotor interface for human-computer interaction". Nature, vol. 645, no. 8081, pp. 702-711. https://doi.org/10.1038/s41586-025-09255-w. Retrieved 2026-10-13.
- ↑ 5.0 5.1 "Meta Ray-Ban Display: AI Glasses With an EMG Wristband". Meta Newsroom. Meta Platforms. 2025-09-17. https://about.fb.com/news/2025/09/meta-ray-ban-display-ai-glasses-emg-wristband/. Retrieved 2026-10-13.
- ↑ "Patrick William John Kaifosh". Mathematics Genealogy Project. https://mathgenealogy.org/id.php?id=273492. Retrieved 2026-10-13.
- ↑ 7.0 7.1 7.2 Patrick Kaifosh, Attila Losonczy (2016). "Mnemonic functions for nonlinear dendritic integration in hippocampal pyramidal circuits". Neuron, vol. 90, no. 3, pp. 622-634. doi:10.1016/j.neuron.2016.03.019. https://pmc.ncbi.nlm.nih.gov/articles/PMC4860028/. Retrieved 2026-10-13.
- ↑ 8.0 8.1 Patrick Kaifosh, Matthew Lovett-Barron, Gergely F. Turi, Thomas R. Reardon, Attila Losonczy (2013-08-04). "Septo-hippocampal GABAergic signaling across multiple modalities in awake mice". Nature Neuroscience, vol. 16, no. 9, pp. 1182-1184. https://doi.org/10.1038/nn.3482. Retrieved 2026-10-13.
- ↑ 9.0 9.1 Matthew Lovett-Barron, Patrick Kaifosh, et al. (2014-02-21). "Dendritic Inhibition in the Hippocampus Supports Fear Learning". Science, vol. 343, no. 6173, pp. 857-863. https://doi.org/10.1126/science.1247485. Retrieved 2026-10-13.
- ↑ 10.0 10.1 Patrick Kaifosh, Jeffrey D. Zaremba, Nathan B. Danielson, Attila Losonczy (2014-09-23). "SIMA: Python software for analysis of dynamic fluorescence imaging data". Frontiers in Neuroinformatics, vol. 8. https://doi.org/10.3389/fninf.2014.00080. Retrieved 2026-10-13.
- ↑ 11.0 11.1 Ann Kennedy, Greg Wayne, Patrick Kaifosh, Karina Alviña, L. F. Abbott, Nathaniel B. Sawtell (2014). "A temporal basis for predicting the sensory consequences of motor commands in an electric fish". Nature Neuroscience, vol. 17, pp. 416-422. https://doi.org/10.1038/nn.3650. Retrieved 2026-10-13.
- ↑ Tim Requarth, Patrick Kaifosh, Nathaniel B. Sawtell (2014-11-26). "A Role for Mixed Corollary Discharge and Proprioceptive Signals in Predicting the Sensory Consequences of Movements". The Journal of Neuroscience, vol. 34, pp. 16103-16116. https://doi.org/10.1523/JNEUROSCI.2751-14.2014. Retrieved 2026-10-13.
- ↑ Emil Protalinski (2019-11-20). "Ctrl-labs CEO: We'll have neural interfaces in less than 5 years". VentureBeat. https://venturebeat.com/ai/ctrl-labs-ceo-well-have-neural-interfaces-in-less-than-5-years. Retrieved 2026-10-13.
- ↑ 14.0 14.1 "US11000211B2 - Adaptive system for deriving control signals from measurements of neuromuscular activity". Google Patents. 2021-05-11. https://patents.google.com/patent/US11000211B2. Retrieved 2026-10-13.
- ↑ 15.0 15.1 "US10772519B2 - Methods and apparatus for providing sub-muscular control". Google Patents. 2020-09-15. https://patents.google.com/patent/US10772519B2. Retrieved 2026-10-13.
- ↑ "CTRL-labs: non-invasive neural interfaces for human augmentation". Redwood Center for Theoretical Neuroscience, UC Berkeley. 2019-02-13. https://redwood.berkeley.edu/seminars/patrick-kaifosh-february-13/. Retrieved 2026-10-13.
- ↑ Lucas Matney (2019-09-23). "Facebook buys startup building neural monitoring armband". TechCrunch. https://techcrunch.com/2019/09/23/facebook-buys-startup-building-neural-monitoring-armband/. Retrieved 2026-10-13.
- ↑ CTRL-labs at Reality Labs, David Sussillo, Patrick Kaifosh, Thomas Reardon (2024-02-28). "A generic noninvasive neuromotor interface for human-computer interaction". bioRxiv. https://doi.org/10.1101/2024.02.23.581779. Retrieved 2026-10-13.
- ↑ "Control Shift: New Reality Labs Research on sEMG Published in 'Nature'". Meta Blog. Meta Platforms. 2025-07-23. https://www.meta.com/blog/reality-labs-surface-emg-research-nature-publication-ar-glasses-orion/. Retrieved 2026-10-13.
- ↑ M.G. Siegler (2025-07-24). "A Mere Thought of a Flick of the Wrist". Spyglass. https://spyglass.org/meta-mind-wristband-ctrl-labs/. Retrieved 2026-10-13.
- ↑ Roselyne Min (2025-07-29). "Meta's new wristband allows you to type without touching a screen. Here's how the AI device works". Euronews. https://www.euronews.com/next/2025/07/29/metas-new-wristband-allows-you-to-type-without-touching-a-screen-heres-how-the-ai-device-w. Retrieved 2026-10-13.