Highscores for DishBrain, the Lab-grown Intelligence A scientific story in October 2022 left a lot of people speechless. Dr Brett Kagan, Chief Scientific Officer at Cortical Labs, in a peer-reviewed article in the journal Neuron, claimed to have created the first sentient lab-grown brain in a petri dish. His research team had taught it how to play the arcade game, Pong and next, they intend to get it drunk. By Richard Forsyth
H
elping disembodied neurons to play a video game and get drunk may seem like a comedy science fiction plot, but the ground-breaking research by Cortical Labs has profound implications for our understanding of the brain, AI and for the future of healthcare. The Australian-led team mounted neurons on multi-electrode arrays to read the activity. The neurons were cultivated in a nutrient-rich solution and grown across a silicon chip with pins in it, that send electrical impulses into the neural lattice, as well as receive impulses back.
00 38
For the first time, some 800,000 brain cells were stimulated in a structured way with the singular goal, to play the 1970’s tennis-like arcade game, Pong. Kagan said: “In the past, models of the brain have been developed according to how computer scientists think the brain might work… That is usually based on our current understanding of information technology, such as silicon computing… But in truth, we don’t really understand how the brain works.” To fathom the mysterious workings of brain matter, they had to let its nature reveal itself by giving it something to react to, to provoke its awareness of a real-time changing situation. What better context for this than a simple arcade game?
EU Research
Ready Player One Growing brain tissue on electrodes is not new but the breakaway from convention in this study was in the feedback loop. In the experiment, mouse cells from embryonic brains and human cells from stem cells were grown on top of the microelectrode arrays which could both stimulate them and read activity from them. Electrodes on the left for example, highlighted for the brain matter, which side the ball was on in the game of Pong, and distance from the paddle was shown by the frequency of signals. Feedback from the electrode taught DishBrain how it could return the ball by making the cells act as if they were themselves, the paddle. The cell cultures controlled the paddle to return the ball, essentially, via sensing. Hitting the ball gave it a predictable response, missing gave it random chaos. It preferred the former and this led to learning to move the paddle to hit the ball. This was a mini-brain in action and startlingly, it was labcreated biological intelligence.
The great benefit of building structures of brains in this way means scientists can experiment on real brain function, as opposed to analogous models like a computer. “We’ve never before been able to see how the cells act in a virtual environment,” said Kagan. “We managed to build a closed-loop environment that can read what’s happening in the cells, stimulate them with meaningful information and then change the cells in an interactive way so they can actually alter each other.” The research seemed to confirm that these cells were trying to minimise the unpredictability in their environment. Kagan was excited by the revelation that DishBrain did not behave like silicon-based systems. “When we presented structured information to disembodied neurons, we saw they changed their activity in a way that is very consistent with them actually behaving as a dynamic system… For example, the neurons’ ability to change and adapt their activity as a result of experience increases over time, consistent with what we see with the cells’ learning rate.”
“In the past, models of the brain have been developed according to how computer scientists think the brain might work… That is usually based on our current understanding of information technology, such as silicon computing… But in truth, we don’t really understand how the brain works.”
Brett Kagan, Ph.D
Hon Weng Chong, M.D.
www.euresearcher.com
00 39
You could say, DishBrain essentially understood and responded to the game presented to it. Cortical Labs call their modern-day Frankenstein of technology and neural matter a bIOS (biological intelligent operating system). In the era of AI, the research team have an advantage, working with organic matter this way. Whilst AI teams around the world try to emulate nature and mimic the mastery in processes that define smart behaviour, working with self-programming neurons, gives the team a head start in developing an intelligence, and the learning occurred fast. In their written summary, they reported there was ‘apparent learning within five minutes of real-time gameplay’. Frequently, after a successful hit, the paddle would position to where the ball would eventually end up on the return. The data showed that the experimental cultures improved their performance by reducing how often they missed the initial serve and they learned to sustain longer rallies. “We have shown we can interact with living biological neurons in such a way that compels them to modify their activity, leading to something that resembles intelligence,” said Dr Kagan. In the article in Neuron, they state: “We show that supplying unpredictable sensory input following an ‘undesirable’ outcome and providing predictable input following a ‘desirable’ one significantly shapes the behaviour of neural cultures in real time”. It is particularly remarkable because the self-organising cultures learned to make their world predictable without some of the mechanisms for learning that we might assume are necessary for brains.
“The beautiful and pioneering aspect of this work rests on equipping the neurons with sensations — the feedback — and crucially the ability to act on their world,” says co-author Professor Karl Friston, a theoretical neuroscientist at UCL, London. “Remarkably, the cultures learned how to make their world more predictable by acting upon it. 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,” he stated. It was concluded by the team that DishBrain ‘demonstrated that a single layer of in vitro cortical neurons can self-organise activity to display intelligent and sentient behaviour when embodied in a simulated game-world’. It was a first and made for bewildering headlines for all the major news broadcasters around the world. “This new capacity to teach cell cultures to perform a task in which they exhibit sentience – by controlling the paddle to return the ball via sensing – opens up new discovery possibilities which will have farreaching consequences for technology, health, and society,” said Dr Adeel Razi, Director of Monash University’s Computational & Systems Neuroscience Laboratory. “We know our brains have the evolutionary advantage of being tuned over hundreds of millions of years for survival. Now, it seems we have in our grasp where we can harness this incredibly powerful and cheap biological intelligence.”
“We know our brains have the evolutionary advantage of being tuned over hundreds of millions of years for survival. Now, it seems we have in our grasp where we can harness this incredibly powerful and cheap biological intelligence.”
Dishbrain visualisation video screengrab. © Cortical Labs
Cortical Labs Chief Scientific Officer, Dr Brett J. Kagan (seated), and Chief Executive Officer, Dr Hon Weng (standing), conducting cell work on multielectrode arrays in a biosafety hood httpsbit.ly3SLgbAy © Cortical Labs.
00 40
EU Research
Dendritic Network Scanning Electron Microscope image of a neural culture that has been growing for more than six months on a high-density multielectrode array. A few neural cells grow around the periphery. © Cortical Labs
Image showing ‘DishBrain’ at work. © Cortical Labs
“DishBrain offers a simpler approach to test how the brain works and gain insights into debilitating conditions such as epilepsy and dementia.” Chips and Booze So, what next? With a ‘sentient’ creation in their hands, it’s time to get it drunk. Kagan and his team want to understand how alcohol affects DishBrain. As Kagan puts it: “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.” This experiment points to the great potential in healthcare and pharma for such a lab-grown brain-on-a-chip. It has implications for being able to experiment more freely on living / reacting brain matter, providing a fast track for new treatments as well as being a more palatable alternative to animal testing. Experiments can be carried out on how the brain might respond to medical drugs and perhaps gene therapies. There will no longer be a need to create ‘digital twins’ for testing therapeutic interventions. It provides a way to experiment on neuronal elements efficiently, with high accuracy and without some of the moral dilemmas and technical complexities researchers can face. “DishBrain offers a simpler approach to test how the brain works and gain insights into debilitating conditions such as epilepsy and dementia,” explained Dr Hon Weng Chong, Chief Executive Officer of Cortical Labs. Whilst the research is incredibly exciting there are some important facts that need to be understood before we get carried away into
www.euresearcher.com
the realms of science fiction. DishBrain is not equivalent to a human brain, it has a similar number of neurons to a bumblebee, with close to a million – a human has 100 billion neurons in comparison. It’s true, you don’t need a degree to play Pong. Nor, should the term ‘sentient’ be mixed up with the concept of consciousness. Kagan was referring to ‘sentient’ by its strictest definition, and whilst it can respond intelligently to the environment presented to it, it would not be aware in its broader context. This is the activation of a sensory interchange, which a lot can be learned from. If anything, the research might stir up arguments around definitions of life and intelligence and indeed, that provocative word, sentient. What is clear, is that there is an enormous road of potential to travel ahead. What is really edifying in this work by Kagan and his team is that it marks the beginning of essentially a new branch of science and biological chips. The start-up isn’t being greedy either, with a desire for collaboration with other scientific teams in order to expand and scale this kind of research. For themselves, their stated aim is to create machines that possess biological intelligence. That’s quite a goal but by enabling brain matter on a chip to learn how to be a competent game player, they have proved it is closer than we could have imagined.
https: //corticallabs.com / 00 41