Neurons in a Dish: The Fascinating Study of Pong and Intelligence
Written on
Chapter 1: Introduction to the Pong Experiment
Have you ever played Pong? This iconic game, often credited as one of the earliest video games, features two paddles moving vertically on either side of the screen, attempting to intercept a bouncing ball. I recall playing it on our family's first computer at the age of eleven, unaware that decades later, scientists would be teaching this game to brain cells cultivated in a laboratory. This surprising development might even connect to another intriguing study.
Neurons Learning to Play Pong
Before we dive deeper, let's discuss the Pong experiment. This groundbreaking research was recently published on October 12, 2022, in the journal Neuron. Brett Kagan, the chief scientific officer at Cortical Labs in Melbourne, Australia, and his team, dubbed their project DishBrain, given that the neurons are cultivated in a Petri dish.
Kagan clarifies that while the DishBrain is not an actual brain, these cells exhibit a form of intelligence, defined as "the ability to gather information and apply it adaptively in a specific environment." According to Kagan, "We have demonstrated the capability to interact with living biological neurons in a way that encourages them to adjust their activity, leading to something that resembles intelligence."
This research has the potential to revolutionize how we study brain function, possibly reducing our reliance on animal testing. Instead, researchers could cultivate the necessary cells for study and combine them with AI models to explore brain activity.
The Research Methodology
The neurons for this experiment were sourced from human stem cells and embryonic mouse brains. Once gathered, the researchers placed them on microelectrode arrays, stimulating the cells and monitoring their responses.
By applying electrodes on either side of the array, the team trained the neurons to react to the Pong ball's position, represented by stimulating the neurons along the ball's trajectory relative to the paddle they controlled. Kagan's hypothesis suggested that neurons would repeat actions that created a predictable environment. To verify this, the neurons were stimulated at consistent frequencies and locations to "hit the ball." Conversely, when the stimulation varied, the neurons "missed" the ball.
Over time, the neurons adapted to hit the ball in response to predictable stimulation patterns rather than random ones. Adeel Razi, a co-author and director at Monash University's Computational & Systems Neuroscience Laboratory, stated, "This newfound ability to train cell cultures to perform tasks that exhibit sentience—controlling the paddle to return the ball by sensing—opens new avenues for discovery with significant implications for technology, health, and society."
Kagan's findings are not the first instances of neurons demonstrating learning capabilities. In 2008, neuroengineer Steve Potter and his team at Georgia Institute of Technology revealed that rat neurons exhibited goal-directed behavior and learning. However, the technology used in the Pong experiment is more advanced, as each DishBrain contains thousands of electrodes, compared to Potter's setup with only dozens.
Yet, we must remain cautious not to make assumptions. The neurons' apparent learning may stem from reasons we have yet to uncover, emphasizing the need for further research.
Next Directions
Kagan presents an intriguing perspective on future research possibilities. He notes that current textbooks mainly view neurons through the lens of human or animal biology. They are often overlooked as information processors, despite their remarkable ability to process information in real-time with minimal energy consumption.
Moreover, Kagan points out that previous brain models were developed based on how computer scientists believed the brain functioned, which may limit our understanding of both brain function and information technology.
Kagan's team is also planning an entertaining yet scientifically significant experiment: "drunk neuron Pong." The aim is to create a dose-response curve using ethanol to observe whether the neurons play worse when "drunk," paralleling human behavior. This simple experiment could shed light on the spectrum of learning capabilities across different brain regions and cell types, potentially leading to astonishing discoveries.
A Shift in Perspective
It's hard to ignore the notion that neurons in a dish appear capable of learning and exhibit basic signs of intelligence. Kagan defines intelligence as the capacity to collect information and adapt behavior accordingly.
Recent research suggests that consciousness may not be exclusive to humans or animals; rather, it could exist at varying levels in all matter. This raises the possibility that consciousness might reside within the atomic structures of everything, including neurons in a Petri dish learning to play Pong.
To see the neurons in action, check out the video demonstrating the game Pong being controlled by cultured neurons. For those interested in a bit of nostalgia, here's a link to play Pong yourself.
Chapter 2: Videos on Neurons Playing Pong
The first video titled "Human brain cells in a dish learn to play Pong" showcases the remarkable ability of brain cells to interact with the game.
The second video, "Researchers teach brain cells in a dish to play Pong," further explores the implications of this research, highlighting the innovative methods used by scientists.