14/10/2022
The human brain is more adaptable and learns faster than current artificial-intelligence (AI) systems. Could it be a better model for the next generation of computers?
Mini-brains were first produced in 2013, to study microcephaly, a genetic disorder where the brain is too small, and have since been used for research into brain development.
But this is the first time they have been plugged into, and interacted with, an external environment, in this case a video game.
The research team:
grew human brain cells grown from stem cells and some from mouse embryos to a collection of 800,000
connected this mini-brain to the video game via electrodes revealing which side the ball was on and how far from the paddle
In response, the cells produced electrical activity of their own.
They expended less energy as the game continued
But when the ball passed a paddle and the game restarted with the ball at a random point, they expended more recalibrating to a new unpredictable situation
The mini-brain learned to play in five minutes.
It often missed the ball - but its success rate was well above random chance.
Although, with no consciousness, it does not know it is playing Pong in the way a human player would, the researchers stress.
The researchers say their mini brains are more adaptable than current AI systems and so might provide the basis for more adaptable robots
Dr Kagan hopes the technology might eventually be used to test treatments for neurodegenerative diseases such as Alzheimer's.
"When people look at tissues in a dish, at the moment they are seeing if there is activity or no activity. But the purpose of brain cells is to process information in real time," he says.
"Tapping into their true function unlocks so many more research areas that can be explored in a comprehensive way."
Next, Dr Kagan plans to test the impact alcohol has on the mini-brain's ability to play Pong.
If it reacts in a similar way to a human brain, this would underscore just how effective the system might be as an experimental stand-in.
