In July 2006, a paper in Nature described how a paralyzed man with a chip implanted in his brain used his mind to move a computer cursor and a robotic arm. The chip is one of the most successful examples to date of a neural prosthetic. Such devices pick up neural signals from a part of the brain involved in a given activity, such as the neurons in the motor cortex that fire as a person imagines moving a computer mouse by hand. Then they interpret those signals and direct a physical action accordingly–say, moving a cursor to the left.
Neural prosthetics promise to empower people with neurodegenerative diseases and spinal-cord injuries. But because they can involve many combinations of brain regions and hardware, each new prototype has needed its own software. Designing new algorithms from scratch slows development, says Lakshminarayan Srinivasan, SM ‘03, PhD ‘06, a neurosurgery research fellow at the Massachusetts General Hospital and a medical student in the Harvard-MIT Division of Health Sciences and Technology. So Srinivasan is developing general algorithms that could lead to software compatible with all such devices.
Technologies for detecting brain activity include functional magnetic resonance imaging (fMRI), electroencephalography (EEG), and functional near-infrared spectroscopy–each of which generates different kinds of data. Output devices might include TVs, computers, or robotic arms, each of which has different user commands.
Srinivasan’s algorithms apply to any combination of devices because they operate at a higher level of abstraction. Instead of working just with EEG inputs reflecting electrical activity or MRI inputs showing blood flow in the brain, the algorithms treat all neural activity as either continuous or binary (a simple on/off switch) and translate it into continuous or discrete commands. (In driving a car, for example, gradually pressing the accelerator is a continuous command; shifting gears is a discrete command.)
Srinivasan says these algorithms should help researchers develop new neural prosthetics and quickly repair any problems that crop up.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
How to fix the internet
If we want online discourse to improve, we need to move beyond the big platforms.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.