In parallel to recent developments in machine learning like GPT-4, a group of scientists has recently proposed the use of neural tissue itself, carefully grown to recreate the structures of the animal brain, as a computational substrate. After all, if AI is inspired by neurological systems, what better medium to do computing than an actual neurological system? Gathering developments from the fields of computer science, electrical engineering, neurobiology, electrophysiology, and pharmacology, the authors propose a new research initiative they call “organoid intelligence.”
OI is a collective effort to promote the use of brain organoids—tiny spherical masses of brain tissue grown from stem cells—for computation, drug research and as a model to study at a small scale how a complete brain may function. In other words, organoids provide an opportunity to better understand the brain, and OI aims to use that knowledge to develop neurobiological computational systems that learn from less data and with less energy than silicon hardware.
The development of organoids has been made possible by two bioengineering breakthroughs: induced pluripotent stem cells and 3D cell culturing techniques.
Taking the existing field of neuromorphic computing, where the structure of neurons and the connections between them are studied and mimicked in silicon architectures, OI extends the engineering analogy with the opportunity to directly program desired behaviors into the firing activity of animal brain cell cultures.
Organoids typically measure 500 microns in diameter—roughly the thickness of your fingernail. As organoids develop, the researchers say, organoids’ constituent neurons begin to interconnect in networks and patterns of activity that mimic the structures of different brain regions. The development of the organoids field has been made possible by two bioengineering breakthroughs: induced pluripotent stem cells (IPSCs) and 3D cell culturing techniques. IPSCs are stem cells–notably capable of developing into any cell found in an animal’s body–that are created by turning an adult cell back into the stem cell. These induced stem cells are then biochemically coaxed into the specific neurons and glia needed to construct a given organoid. More recently developed 3D-scaffolding methods allow biologists to grow iPSC-derived neural tissues vertically as well as horizontally, allowing organoids to develop the interneuronal networks seen in an animal’s brain. Scientists have studied 2D-cultures for decades, but monolayer tissues are not able to grow into brain-like networks in the ways the organoids can.
Networks make organoids a powerful model for understanding and potentially exploiting the dynamics of brain activity. Jens Schwamborn, a Professor of Cellular and Developmental Biology at the University of Luxembourg, is using organoids to study the development of neurological disorders like Parkinson’s disease. “We’ve recapitulated the key features of the pathology. We can see the loss of dopaminergic neurons, we see the appearance of protein aggregates that are relevant to the disease,” said Schwamborn, whose lab has developed an organoid model of Parkinson’s. These platforms allow them to study, on a small scale, Parkinson’s development in a cellular network context that monolayer cultures cannot: “That’s the major advantage. We can see features of the disease that we know are happening in patients but so far have been unable to recapitulate in the lab. Now, finally, we can do that.”
“We aren’t teaching the cells how to do it. [Organoids] end up with the organization of structures in the brain. I think that’s the power: the computational power comes from that organization.”
—Alysson Muotri, University of California, San Diego
Just as organoids themselves are the product of bioengineering advances, their utility as models for neurological function is the product of several other biochemical innovations—electrophysiology and microfluidics. Researchers can now guide organoid development more reliably and precisely than they could even half a decade ago, and can use that specificity to create organoids that mimic the network structure and cellular composition of specific cortical and subcortical structures. Alysson Muotri, a professor of pediatrics and molecular medicine at the University of California in San Diego, believes that these structures may provide them with the information processing capabilities of brain tissue. “In 3D, you see all this additional organization that you don’t see in 2D. This is genetically encoded. We aren’t teaching the cells how to do it. They end up with the organization of structures in the brain. I think that’s the power: the computational power comes from that organization.”
Having consistent, sustainable organoids also allows scientists to take meaningful measurements of neuron activity within them. Multielectrode arrays (MEAs) are panels of tiny electrodes capable of measuring and stimulating the electrical activity of neurons near an organoid’s surface. Flexible MEAs that can wrap around an organoid mass are capable of recording from the entire surface, instead of just the bottom layer of neurons in contact with the petri dish. By analyzing those recordings, scientists can infer how all those neurons are talking to each other. Through a set of signal processing techniques called causal modeling, researchers can produce maps of connections between neurons that compose networks of organoid functional structure. These network maps can then be used to trace how information is processed by the developing mass of neural tissue.
By conditioning neuron populations within organoids to consistently and predictably respond to set electrical inputs, scientists hypothesize that they can turn organoid systems into organic processing units that may leverage the apparent information processing capabilities of neural tissue to create flexible and powerful computing systems.
Cortical Labs, a Melbourne-based biotech startup, is launching Dishbrain, the first such trainable neurobiological computing platform. The company aims to provide programmable, monolayer 2D neural cultures —which have already been shown to reliably learn digital input/output patterns such as playing the classic video game pong—to end users as a cloud service. Brett Kagan, the company’s Chief Science Officer, says the company plans to have the service running by year’s end: “We should have, before the end of this year, a beta system for people to be able to, either by cloud or by partnering with us for in-house use, log on and be able to run very basic environments,” he said.
While similar organoid-on-chip computing systems are not yet available, the OI team is optimistic about their rate of progress. Professor Muotri thinks we may see organoid computing systems developed within the decade: “We might see a prototype in the next two or three years,” he said. “For things to become more reproducible, with all the tools we’ll need—that’ll be 5 or 10 years.”
