Brainscan breakthrough: Working robot limbs come closer
Brains in jars controlling cyber colossi, too, of course
Canadian brain boffins are chuffed to announce a new breakthrough in the tricky business of machine mind-reading: they have developed a method of working out from brain scans what physical action a person is about to carry out – before he or she can actually do it.
"This is a considerable step forward in our understanding of how the human brain plans actions," says Jason Gallivan, PhD student at Western Ontario uni's Centre for Brain and Mind, who was heavily involved in the new research.
Using functional magnetic resonance imaging (fMRI), Gallivan and his fellow boffins scanned the bonces of test subjects as they performed a range of hand movements. The team found that by using the signals from many brain regions, they could predict which action a subject would perform seconds before he or she did so.
"Being able to predict a human's desired movements using brain signals takes us one step closer to using those signals to control prosthetic limbs," notes Gallivan.
In this context it may be worth noting that US military boffins are trying at the moment to produce a mind-to-limb interface with Billy-the-Kid style latency of under 10 milliseconds: if connected to a sufficiently speedy arm this would let cyber-limbed troops beat normal un-augmented fleshies to the trigger in most cases (as normal human reaction time is around 200-250 milliseconds).
Gallivan and his colleagues' new fMRI technique might allow a robotically-enhanced gunman – or of course a disembodied brain in a bubbling jar controlling a powerful robotic colossus of some sort – to act before it had even realised it was going to, by the look of things.
The new Canadian research is published here in the Journal of Neuroscience. ®
Sponsored: Optimizing the hybrid cloud