Scientists teach brain cells to play video game Pong

Cells were able to play longer rallies over time, with researchers now planning 
to see how they react when drunk

By Tory Shepherd  Thu 13 Oct 2022
https://www.theguardian.com/australia-news/2022/oct/13/scientists-teach-brain-cells-to-play-virtual-pong


Scientists have taught brain cells how to play the video game Pong, and they 
now plan to experiment with getting them drunk.

As the cells, taken from humans and mice, played the game, they learned. They 
played longer rallies and were aced less often.

That the 800,000 cells have mastered a version of the cult 1970s video game is 
evidence they can exhibit sentience, the Melbourne-based researchers said.

Their peer-reviewed study has been published in the journal Neuron.

Pong was an arcade game released in 1972 where two players use paddles to bat a 
ball back and forth – making a “pong” noise upon contact, hence the name.

Pong was hugely successful, and extremely simple. Its simplicity and iconic 
status prompted the researchers to choose it for their initial test.

The team, from Cortical Labs and universities including Monash University, the 
University of Melbourne and University College London, used human cells derived 
from stem cells and mouse cells derived from embryonic cells.

They put them on to the “DishBrain”, a multi-electrode array that can sense 
cell activity and stimulate the cells, then gave the cells feedback on whether 
the paddle was hitting the ball.

It worked, and within five minutes the cells started to “learn” how to play the 
simulated game using a shared “language” of electrical activity.

Over time they were less likely to be aced, and were able to have longer 
rallies – and the human cells could rally for longer than the mice (humans and 
mice were not pitched against each other).

The cell groups played 486 “games”, with researchers testing their reaction to 
different stimuli or lack thereof. For example, in some games they were given 
feedback either through giving the cells information or by removing 
information. In other games there was no feedback, so cells could not learn how 
their behaviour affected the environment.

Those with feedback learned from it.

Now the researchers will see how the cells perform when they are drunk or given 
medicines. They hope to use the DishBrain to learn more about conditions such 
as epilepsy and dementia.

The study’s lead author, Dr Brett Kagan, is the chief scientific officer of 
biotech startup Cortical Labs. He said by manipulating the neurons and showing 
that they change behaviour based on feedback, the team has shown “something 
that resembles intelligence”.

“This is the new way to think about what a neuron is,” he said.

“It’s not necessarily a computer, but it’s a small biological device that can 
process information and respond intelligently with incredible speed, low power 
consumption and flexibility.”

DishBrain is a model of the brain that would allow them to experiment with 
different inputs.

“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 said.

Dr Adeel Razi, the director of Monash University’s computational and systems 
neuroscience laboratory, said the ability to teach cell cultures to “perform a 
task in which they exhibit sentience … opens up new discovery possibilities 
which will have far-reaching consequences for technology, health and society”.

“We know our brains have the evolutionary advantage of being tuned over 
hundreds of millions of years for survival,” he said.

“Now, it seems we have in our grasp [the ability to] harness this incredibly 
powerful and cheap biological intelligence.”

Another co-author, Prof Karl Friston from UCL, said the pioneering work allowed 
neurons to experience sensations.

“Remarkably, the cultures learned how to make their world more predictable by 
acting upon it,” he said.

“This is remarkable because you cannot teach this kind of self-organisation; 
simply because – unlike a pet – these mini brains have no sense of reward and 
punishment.”

That has allowed them to design a “sandbox” made of the same elements in an 
actual brain where they can test interventions.

The authors wrote in the Neuron article that a synthetic biological 
intelligence, “previously confined to the realm of science fiction”, could now 
be within reach.

--
_______________________________________________
Link mailing list
[email protected]
https://mailman.anu.edu.au/mailman/listinfo/link

Reply via email to