Neurons: Shocked, the neurons in the dish have learned to play games!

*For medical professionals only

What, the neurons in the dish have learned to play games?

Although humans have always been keen to teach various animals and AI to play games, AlphaStar is even already playing StarCraft, but this is a cluster of neurons in a petri dish! Does this work?

Recently, the journal Neuron published research from Cortical Labs in Melbourne, Australia, and the researchers called "DishBrain" neuronal cultures, actually learned to play a simple game [1]! The journal Nature also published a commentary interpreting the findings
.

The game chosen by the researchers in this study is a classic ping-pong mini-game called Pong, which was first released in 1972 and was one of the first arcade games.

Scientists have successfully taught pigs, monkeys and AI to play the game before, and the hand-mutilated cake said he didn't want to know their results
.

PongImage source wiki on the arcade

The principle of this DishBrain is not difficult to understand
.
What the researchers used to grow cells was not an ordinary medium, but an electrode array chip, which "hinted" where the neuronal globules were located through an external system, and the signals output by the neurons controlled the movement
of the racket.
If the ball is hit, the system will give a specific signal feedback, and if it does not hit, it will give a random feedback
.

Neurons on the chipImage source Cortical Labs official website

Neuronal culture on a chip where axon bundles at red arrows and individual axons at blue arrows can be seen

It is well known that neurons respond to external stimuli, which is the basis of
all animal learning.
So, just like that, the brain in the plate quickly "learned" to play table tennis!

This is what neurons play, 1.
5 times faster

Brett Kagan, the study's corresponding author, argues that if intelligence can be defined as the ability to collate information and apply it to adaptive behavior in a given environment, then it is clear that this study demonstrates that neurons in a dish are able to learn and exhibit basic intelligence
.

However, despite its name, the neurons are far from the brain, let alone conscious
.
Singularity Cake feels that this learning process is somewhat similar
to training machine learning algorithms.

Of course, in the future, DishBrain can not only be used to play games, after reasonable training, it can be fully applied to new drug testing, integrated biological computers and other application scenarios, which requires us to further explore the specific connection
between neuronal behavior and environmental stimuli.

Resources:

[1]  style="letter-spacing: 0px;font-size: 12px;color: rgb(160, 160, 160);box-sizing: border-box;"> ADtaxi

The author of this articleDai Siyu