Brain collapses and the secrets of critical neurological conditions are revealed
DishBrain reveals how human neurons work together to process information.
New research shows that when neurons are given information about the changing world around them (task-relevant sensory input), they change the way they behave, putting them on edge such that small inputs can then trigger a “torrent” of brain activity. Which supports a theory known as the critical brain hypothesis.
Researchers from Cortical Labs and the University of Melbourne used DishBrain, a collection of 800,000 human neurons that learn to play ping-pong. The study was recently published in the journal Nature Communications
It is the strongest evidence yet supporting the controversial theory of how the human brain processes information.
According to the critical brain hypothesis, large complex behaviors only become possible when neurons are on such edge that small inputs can trigger “avalanches” of brain activity.
This state of delicate balance is known as the “neurocritical” state, and lies between two extremes: the unbridled excitation seen in disorders such as epilepsy, and the comatose state in which signals cease.
“This not only shows the reorganization of the network to a near-critical state where it is fed structured information, but reaching this state also leads to better task performance,” says Dr. Brett Kagan, chief scientific officer at biotech startup Cortical Labs. Who created DishBrain.
“The results are amazing, far beyond what we thought we would achieve.”
The research adds a vital piece to the puzzle of the critical brain hypothesis.
Main findings and implications
To date, there has been little experimental evidence showing whether criticality is a general feature of biological neural networks or whether it is related to information load.
“Our results suggest that sub-critical network behavior emerges when the neural network is engaged in a task, but not when it is left unstimulated,” says Dr. Kagan.
However, Dr. Kagan’s research shows that salience alone is insufficient to drive learning through a neural network.
“Learning requires a feedback loop, where the network is given additional information about the consequences of the action,” says Dr. Kagan.
The latest research confirms DishBrain’s ability to help uncover the secrets of the human brain and how it works, something that is not possible with animal models.
“Usually to study the brain, especially at the level of neurons, researchers have to use animal models, but in doing so, there are a lot of difficulties and one can only address a limited number of topics,” says first author Dr. Forough Habibulahi. , Research Fellow at Cortical Labs.
“So when I saw DishBrain’s unique ability to answer different types of questions in a way that no one else could, I was very excited to start this project and join the team.”
Applications and future possibilities
Doctors also see great potential for research to help discover treatments for disabling brain diseases.
“The DishBrain Critical Project has been an amazing collaborative experience between Cortical Labs, biomedical engineering and neuroscience,” says paper author Dr Chris French, leader of the Neurodynamics Laboratory in the Department of Medicine at the University of Melbourne.
“The critical dynamics of DishBrain neurons should provide key biomarkers for the diagnosis and treatment of a range of neurological diseases, from epilepsy to dementia,” he says.
By building a living model brain, scientists will be able to experiment with using real brain functions rather than flawed computer-like models to not only explore brain functions but also test how drugs affect them.
Professor Anthony Burkett, author of the paper and head of the Department of Biosignals and Biosystems at the University of Melbourne, says the research also has the potential to solve challenges facing brain-computer interfaces that can restore functions lost as a result of neurological damage. Department of Biomedical Engineering.
“A key feature of the next generation of neural prosthetics and brain-computer interfaces that we are currently investigating involves the use of real-time closed-loop strategies,” he says. “So the results of this study could have important implications for understanding how these control and motivation strategies interact with neural circuits in the brain.”
“This field of biological modeling of the brain is still in its infancy, but it opens the way to a whole new field of science,” says Dr. Kagan.
Reference: “Critical dynamics arise during the presentation of structured information within embodied neural networks in the laboratory” by Forough Habibulahi, Brett J. Kagan, and Anthony N. Burkett, Chris French, August 30, 2023, Nature Communications.
Critical dynamics arise during the presentation of structured information within embodied neural networks in the laboratory
Forugh Habibulahi, Brett J. Kagan, Anthony N. Burkett, and Chris French
Understanding how the brain processes information is a very difficult task. Among the metrics that characterize information processing in the brain, observations of near-critical dynamic states have aroused great interest.
However, theoretical and experimental limitations associated with human and animal models have prevented a definitive answer about when and why neurocriticality arises with links from attention to perception to consciousness.
To explore this topic, we used a laboratory neural network of cortical neurons that were trained to play a simplified game of “Pong” to demonstrate artificial biological intelligence (SBI).
We demonstrate that critical dynamics emerge when neural networks receive task-related structured sensory inputs, reorganizing the system to a near-critical state. In addition, better task performance is associated with proximity to critical dynamics. However, importance alone is insufficient for a neural network to demonstrate learning in the absence of additional information regarding the consequences of previous actions. These results provide convincing support that neural salience arises as a fundamental feature of processing incoming structured information without the need for higher-level cognition.