Brain Organoids Advance: Learning, Monitoring, and Insights into Skill Acquisition

Recent research has significantly advanced the understanding and study of brain organoids – lab-grown tissues that mimic aspects of brain development and function.

Scientists at the University of California, Santa Cruz, successfully trained brain organoids to perform a goal-directed task, demonstrating their capacity for learning through a bioelectrical interface.

Concurrently, a collaboration led by Northwestern University developed a novel 3D electronic framework for comprehensive monitoring and manipulation of neural activity within these organoids. These developments contribute to fundamental neuroscience and the potential for new approaches to studying neurological conditions.

Goal-Directed Learning Demonstrated in Brain Organoids

UCSC Organoids Master the "Cart-Pole Balancing Problem"

Researchers at the University of California, Santa Cruz (UCSC), successfully trained brain organoids to solve the "cart-pole balancing problem." This task, also known as the "inverted pendulum," is a standard benchmark in robotics, control theory, and artificial intelligence, assessing a system's ability to adaptively process and respond to information by balancing an upright pole hinged to a movable cart. Their findings were published in the journal Cell Reports.

The methodology involved growing brain organoids from stem cells, resulting in tiny pieces of brain tissue containing networks of several million neurons. These organoids were placed on specialized chips that enabled observation and stimulation of neurons via an electrophysiology system. A closed-loop bioelectrical interface was established, allowing the organoid to receive electrical signals indicating the virtual pole's angle and respond with signals interpreted as forces to balance it. An artificial coach, utilizing a reinforcement learning algorithm, provided targeted training feedback to specific neurons each time the pole fell.

Key findings from the UCSC study include:

• The success rate for balancing the pole increased from 4.5% with random training signals to 46% with consistent adaptive training using reinforcement learning.
• This marks the first rigorous academic demonstration of goal-directed learning in lab-grown brain organoids.
• Organoids exhibited short-term learning, with performance returning to baseline after approximately 45 minutes of inactivity following 15 minutes of training. Researchers suggested that more complex organoids, potentially incorporating multiple brain regions, might be necessary for long-term adaptive performance retention.

The research aims to clarify how information transmission through neural electrical spiking facilitates learning. This understanding has implications for basic science and health research, potentially offering a tool to study how neurological conditions such as Alzheimer’s disease, dementia, stroke, concussion, autism, schizophrenia, Parkinson’s disease, dyslexia, and ADHD impact the brain’s learning capacity. To aid further research, Ash Robbins developed BrainDance, an open-source software tool designed for biologists to conduct neural simulation learning experiments and analyze results without extensive coding.

Researchers emphasized that the goal is to advance brain research and treat neurological diseases, not to replace robotic controllers or computers with lab-grown animal brain tissues, particularly noting ethical considerations for human brain organoids in such applications.

Keith Hengen, an associate professor of biology at Washington University in St. Louis, commented that the tissue demonstrated sufficient plasticity and structure to solve a control problem when given targeted electrical feedback, suggesting an intrinsic capacity for adaptive computation in cortical tissue.

Advanced 3D Bioelectronic Framework for Organoid Monitoring

Northwestern Unveils Novel 3D Electronic Mesh

A team of scientists from Northwestern University and Shirley Ryan AbilityLab, in collaboration with Tsinghua University and the University of Illinois Chicago, developed a new 3D electronic framework to monitor and manipulate neural activity in human brain organoids.

This technology, published in Nature Biomedical Engineering, addresses previous limitations in comprehensively studying these millimeter-sized, lab-grown brain-like tissues.

The new system features a soft, porous 3D electronic mesh that conforms to the organoid's spherical shape, providing near-complete coverage with hundreds of miniaturized electrodes, each approximately 10 microns in diameter. This design allows for the necessary flow of oxygen, nutrients, and waste products, maintaining the organoid's viability, and aims to overcome the limitations of traditional flat, rigid instruments that could only sample a small fraction of an organoid's neurons.

The functional capabilities and observations using this technology include:

• Extensive Recording: A 240-channel array recorded synchronized oscillatory waves across entire organoids, generating detailed 3D maps of electrical activity.
• Coordinated Communication: The system detected signals originating in one region and rippling across the network, identifying split-second delays indicative of coordinated neural communication.
• Drug Response Detection: The platform identified changes in neural network firing in response to various drugs. For instance, 4-aminopyridine increased neural signaling, while botulinum toxin disrupted coordinated activity.
• Neural Stimulation: The system can deliver precise electrical pulses to trigger responses in specific regions, influencing neural activity.
• Organoid Shaping: The device can also guide organoid growth into non-spherical geometries, including hexagonal or cubic shapes.

This advancement is intended to bring organoid research closer to accurately modeling human brain development, function, and disease, potentially reducing reliance on animal models. The ability to map activity across nearly the entire organoid is considered crucial for assessing how potential regenerative treatments rebuild functional circuits. Researchers anticipate using this technology to study brain disorders, evaluate drug responses, and assess experimental regenerative strategies.

Insights into Biological Skill Learning Mechanisms

MIT Uncovers Neuron-Specific Feedback in Skill Acquisition

Complementing these advancements in organoid research, new research from MIT indicates that the brain can precisely adjust individual neurons during skill acquisition by sending targeted feedback to each neuron.

This finding aligns with a core principle in modern artificial intelligence, where AI systems learn by comparing their output to a target, calculating an "error" signal, and fine-tuning network connections.

A team led by Mark Harnett, an investigator at MIT's McGovern Institute for Brain Research, identified these instructive signals in mice. The mice were trained to control the activity of specific neurons via a brain-computer interface (BCI). To achieve success, accompanied by a sugary reward and visual feedback, some neurons needed to increase activity while others needed to decrease it. Researchers monitored the target neurons daily, focusing on dendrites where feedback signals were suspected to arrive, and tracked activity in the parent cell bodies.

The study concluded that the two groups of neurons controlling the BCI in opposing ways received opposing error signals at their dendrites during learning. Some dendrites were instructed to increase activity, while others were instructed to decrease it. When these instructive signals in the dendrites were inhibited, the mice failed to learn the task. This represents the first biological evidence that vectorized, neuron-specific, signal-based instructive learning occurs in the cortex. This discovery is expected to foster greater collaboration between neuroscientists and machine learning researchers, offering new avenues to investigate parallels between the brain and machine learning, and potentially leading to improved brain-inspired artificial intelligence models.

Broader Context and Future Outlook

The development of brain organoids has evolved since the isolation of pluripotent stem cells in 1981 and 1998, and the creation of the first 3D brain cell culture in 2013 by scientists led by Madeline Lancaster. These miniature brains, derived from stem cells, provide a platform for research into brain development, neurological diseases, and drug testing.

The recent demonstrations of goal-directed learning and advanced neural activity mapping significantly expand the capabilities of this research field. Scientists are also exploring complex simulations, including preparing to simulate the human brain on a supercomputer, aiming for a more comprehensive understanding of brain function and disease.