Tufts team open-sources HAL, showing rat cortical cultures learn to steer network bursts
MacrinePhD · x · 2026-10-25
New bioRxiv research from Michael Levin, Wesley Clawson and teams at Tufts' Allen Discovery Center, Harvard, and Google introduces HAL (Holistic Agential Learning), an open-source closed-loop platform showing that dissociated rat cortical cultures can learn to control where network bursts originate through minimal reinforcement. Key findings: burst initiation zones shift toward the goal side over days while propagation directions are preserved; trained networks recover better from incorrect states rather than just making fewer errors; and performance gains consolidate across days rather than within sessions.
More from Embodied
- Foldable iPhone Duo units reportedly breaking en masse, displays failing along the hinge crease — flngr · 2026-10-25
- Open-source virtual human body: 700 muscle actuators, 206 joints, 90 segments — mtizard · 2026-10-25
- Open-source model does live single-camera 3D scene reconstruction at ~20 FPS, wows SLAM veteran — facontidavide · 2026-10-25
- $399 robot duck rakes in $2.6M in preorders within 24 hours — CodeByPoonam · 2026-10-25
- Powerhouse hosts free Shenzhen program for hardware founders building in China — DominiqueCAPaul · 2026-10-25
- China Mobile demos brain-controlled humanoid robot with dexterous hand via non-invasive BCI — CyberRobooo · 2026-10-25