Human brain cells in a dish learned to play Pong

Human brain cells in a dish learned to play Pong

A Melbourne-led team has for the first time shown that 800,000 brain cells living in a dish can perform purposeful tasks – in this case, the simple tennis-like computer game ‘Pong’. Published in the journal neuronthe team’s so-called “DishBrain” system is proof that even brain cells in a dish can exhibit inherent intelligence, changing their behavior over time.

Dr. Brett Kagan, chief scientific officer of the biotech startup, was the study’s lead author. Cortical laboratories, which is dedicated to building a new generation of biological computer chips. He worked with collaborators from 10 other institutions on the project, including Monash University, RMIT University, University College London (UCL) and the Canadian Institute for Advanced Research.

“In the past, models of the brain were developed based on how computer scientists think the brain might work,” Kagan said. “It’s generally based on our current understanding of information technology, such as silicon computing.

“But in truth, we don’t really understand how the brain works.

“From worms to flies to humans, neurons are the starting point of generalized intelligence. So the question was, can we interact with neurons in ways that tap into this inherent intelligence? »

To perform their experiment, the research team took mouse cells from embryonic brains as well as some stem cell-derived human brain cells and cultured them on microelectrode arrays that could both stimulate them and read their activity. . While scientists have been able for some time to mount neurons on multi-electrode arrays and read their activity, this is the first time that cells have been stimulated in a structured and meaningful way.

Electrodes to the left or right of an array were drawn to tell DishBrain which side the Pong ball was on, while the distance to the paddle was indicated by the frequency of the signals. The feedback from the electrodes taught DishBrain how to return the ball, making the cells act as if they were the paddle themselves.

“We never got to see how cells act in a virtual environment,” Kagan said. “We’ve managed to create a closed-loop environment that can read what’s happening in cells, stimulate them with meaningful information, and then interactively modify the cells so they can actually alter each other. others.”

The researchers monitored neuron activity and responses to this feedback using electrical probes that recorded ‘spikes’ on a grid. The spikes got stronger the more a neuron moved its paddle and hit the ball. When the neurons were missing, their playing style was critiqued by software created by Cortical Labs. This demonstrated that neurons could adapt activity to a changing environment, in a targeted manner, in real time.

“The magnificent and pioneering aspect of this work lies in endowing neurons with sensations – feedback – and above all the ability to act on their world,” said co-author Professor Karl Friston, neuroscientist theory at UCL.

“Remarkably, cultures have learned to make their world more predictable by acting on it. It’s remarkable because you can’t teach this type of self-organization, simply because, unlike a pet, these mini-brains have no sense of reward and punishment.

“An unpredictable stimulus was applied to the cells, and the system as a whole reorganized its activity to better play the game and minimize random responses,” Kagan added. “You may also think that just playing the game, hitting the ball, and getting predictable stimulation inherently creates more predictable environments.”

The theory behind this learning is rooted in the free energy principle, developed by Friston, which states that the brain adapts to its environment by altering either its view of the world or its actions to better adapt to the surrounding world. surrounds it.

“We faced a challenge when we worked out how to order cells to follow a certain path,” Kagan said. “We don’t have direct access to dopamine systems or anything else that we could use to provide specific prompts in real time, so we had to dig deeper into what Professor Friston is working on: information entropy – a fundamental level of information about how the system can self-organize to interact with its environment at the physical level.

“The free energy principle proposes that cells at this level try to minimize the unpredictability of their environment.”

Kagan said an exciting finding was that DishBrain didn’t behave like silicon-based systems. “When we presented structured information to disembodied neurons, we saw that they changed their activity in a way that was very consistent with their actual behavior as a dynamic system,” he said.

“For example, the ability of neurons to change and adapt their activity as a result of experience increases over time, consistent with what we see with cell learning rate.”

By constructing a model of the living brain from basic structures in this way, scientists will be able to experiment using real brain functions rather than faulty analog models like a computer. Kagan and his team, for example, will then experiment to see what effect alcohol has when introduced into DishBrain.

“We’re trying to create a dose-response curve with ethanol — basically, get them ‘drunk’ and see if they play the game worse, just like when people drink,” Kagan said. This would potentially open the door to whole new ways of understanding what happens in the brain, and could even be used to better understand debilitating conditions such as epilepsy and dementia.

“This new ability to teach cell cultures to perform a task in which they are sensitive – controlling the paddle to return the ball via sensing – opens up new possibilities for discovery that will have profound implications for technology, health and society,” said Dr. Adeel Razi, director of the Computational and Systems Neuroscience Laboratory at Monash University.

“We know that our brains have the evolutionary advantage of being tuned for hundreds of millions of years to survive. Now it seems we have our fingertips where we can harness this incredibly powerful and cheap biological intelligence.

The results also raise the possibility of creating an alternative to animal testing when studying how new drugs or gene therapies react in these dynamic environments. According to Friston, “The translational potential of this work is really exciting: it means we don’t have to worry about creating ‘digital twins’ to test therapeutic interventions. We now have, in principle, the ultimate biomimetic “sandbox” in which to test the effects of drugs and genetic variants – a sandbox made up of exactly the same computational (neuronal) elements found in your brain and mine. »

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

Image caption: A microscopy image of neural cells where fluorescent markers show different cell types. Green marks neurons and axons, purple marks neurons, red marks dendrites, and blue marks all cells. When multiple markers are present, the colors are merged and usually appear yellow or pink depending on the proportion of markers. Image credit: Cortical Laboratories.

Please follow us and share on Twitter and Facebook. You can also subscribe for FREE to our weekly newsletters and our bimonthly magazine.

About Leslie Schwartz

Check Also

PROFILE: Philip Ewell ’01, the cellist shaping the racial calculus of music theory

2022 Wilbur Cross recipient and music theorist Dr. Philip Ewell sat down with the News …