Living brain cells learn to play pong in real time

Human and mouse neurons living in a shell learned to play the video game pong. Scientists reported the achievement in the journal Oct. 12 neuron. Their fascinating experiments show that even brain cells in a dish have inherent intelligence and can change their behavior over time.

“From worms to flies to humans, neurons are the starting point for general intelligence,” says first author Brett Kagan. “So the question was, can we interact with neurons to harness this inherent intelligence?” Kagan is Chief Scientific Officer at Cortical Labs in Melbourne, Australia.

First, the scientists connected the neurons to a computer in such a way that the neurons received feedback on whether their racket hit the ball during play. Using electrical probes that recorded ‘spikes’ on a grid, the researchers were able to monitor the neuron’s activity and responses to that feedback.

This video shows the game of Pong controlled by a layer of neurons in a shell. Photo credit: Kagan et. al / neuron

The more a neuron moved its paddle and hit the ball, the stronger the spikes got. When neurons missed, their playstyle was criticized by a software program created by Cortical Labs. It was shown that the neurons were able to adapt their activity in real time to a changing environment.

“We have chosen pong because of its simplicity and familiarity, but it was also one of the first games to be used in machine learning, so we wanted to acknowledge that,” says Kagan, who worked on the project with collaborators from 10 other institutions.

“An unpredictable stimulus was being delivered to the cells, and the system as a whole would reorganize its activity to better play the game and minimize a random response,” he says. “You may also think that playing the game, hitting the ball, and receiving predictable stimulation inherently creates more predictable environments.”

This is a visual representation of the simulated pong environment, with neuron activity reflected in the tiles growing in height. Photo credit: Kagan et. al / neuron

The theory behind this learning is rooted in the free energy principle. Simply put, the brain adapts to its environment, either by changing its worldview or its actions to better suit the world around it.

pong wasn’t the only game the research team tested. “You know when the Google Chrome browser crashes and you get this dinosaur you can jump over obstacles (Project Bolan). We’ve done that and seen some nice preliminary results, but we still have more work to do to create new environments for custom purposes,” says Kagan.

Future directions of this work have potential in disease modeling, drug discovery, and expanding the current understanding of how the brain works and how intelligence emerges.

“This is the beginning of a new frontier in understanding intelligence,” says Kagan. “It not only touches on the fundamental aspects of what it means to be human, but also what it means to even be alive and intelligent, to process information and to be sentient in an ever-changing, dynamic world.”

For more information on this research, see Watch human brain cells learn to play pong live.

Relation: “in vitro Neurons learn and exhibit sentience when embodied in a simulated game world” by Brett J. Kagan, Andy C. Kitchen, Nhi T. Tran, Forough Habibollahi, Moein Khajehnejad, Bradyn J. Parker, Anjali Bhat, Ben Rollo, Adeel Razi and Karl J. Friston, October 12, 2022, neuron.
DOI: 10.1016/j.neuron.2022.09.001

Financial support was provided by Cortical Labs.