How many brain cells are required to play a videogame?
No, really. It’s not a joke and it doesn’t have a punchline. A neural network system called DishBrain provides a real answer.
If you play Pong, there are around 800,000.
Their slow-moving, one-sided strategy of digital table tennis will not see them win any esports championships in near future. However, it does show the potential for fusion living tissues with silicon technology.
frameborder=”0″ allow=”accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture” allowfullscreen>
This is the first synthetic biological intelligence experiment that shows neurons can adjust their activity to perform a specific task – and, when provided with feedback, can learn to perform that task better. This is amazing stuff that could have applications in computing as well as in studying all aspects of brain stuff, including how drugs and medication affect brain activity, and how intelligence develops.
“We have demonstrated we can interact directly with living biological neurons, causing them to alter their activity. This leads to intelligence. Brett Kagan, neuroscientistCortical Labs, a biotech startup in Australia.
DishBrain is a fusion of neurons from human embryonic neurons and those from embryonic mice. Stem cells. These cells were grown on microelectrodes which could be activated in order to stimulate the neurons. This provided sensory input.
Microelectrodes were placed on either side to indicate whether the ball is to the left or the right of the paddle for a game. The frequency of signals was used to relay the ball’s distance.
DishBrain, with just this setup, is capable of moving the paddle towards the ball but performs poorly overall. Neurons need feedback in order to play the game properly.
DishBrain would miss the ball, so the team created a software that could provide feedback via electrodes. The system was able to improve its playing of Pong and the researchers were able to observe the learning process in just five minutes.
“The beautiful and pioneering aspect of this work rests on equipping the neurons with sensations – the feedback – and crucially the ability to act on their world,” Karl Friston is a theoretical neuroscientistUniversity College London, UK
“Remarkably the cultures were able to predict their world by taking action. This is remarkable because you cannot teach this kind of self-organization; simply because – unlike a pet – these mini brains have no sense of reward and punishment.”
A Few years agoFriston created a theory known as the Energy free principle, which proposes all biological systems behave in ways that reduce the gap between what is expected and what is experienced – in other words, to make the world more predictable.
Friston claims that DishBrain does its best biology by adapting its actions in order to make the world less predictable.
Pong is a simple and familiar game that we chose to play. Pong also was one of our first games. Machine learning” Kagan says.
“An unpredictable stimulus was applied the cells and the system as the whole would reorganize itself to better play the game. This would minimize random responses. It is possible to think of playing the game and hitting the ball as creating predictable environments.
These are some of the most intriguing possibilities available, especially in artificial intelligenceComputers and computing. Our best computers are unable to duplicate the human brain’s power, which is around 80 to 100 trillion neurons. This is our best effort to dateIt took 82,944 processors and 40 minutes to reproduce just one second of activity from one percent of the human brain.
If the architecture is more like that of an actual brain – perhaps even a synthetic biological system like the one developed by Kagan and colleagues – this goal may not be quite so far out of reach.
However, there are also other, possibly more immediate consequences.
DishBrain may be able, for example, to help chemists understand how medications affect the brain at a cellular level. One day, it might even be able to tailor medications to patients’ specific biology using neurons grown from stem cells that have been reverse engineered from the patient’s skin.
“The tremendous potential of this research is to translate into therapeutic interventions. We don’t need to worry about creating a ‘digital twin’ for that purpose.” Friston. “We now have, in principle, the ultimate biomimetic ‘sandbox’ in which to test the effects of drugs and genetic variants – a sandbox constituted by exactly the same computing (neuronal) elements found in your brain and mine.”
The next step is to determine how alcohol and drugs affect DishBrain’s ability to play Pong. “We’re trying to create a dose response curve with ethanol – basically get them ‘drunk’ and see if they play the game more poorly, just as when people drink,” Kagan says.
In other words, a dish of brain cells rolls into a bar…
The research of the team has been published in Neuron.
