Lab-grown brains can now read Braille and guide robot dogs

Guided by millions of years of evolution, our own brain cells know a thing or two about how to navigate a chaotic world where learning the correct movement is a matter of life and death.

Biohybrid computing is an emerging technological field that combines living neural tissue with conventional electronics and software. Instead of simply mimicking the brain with silicon chips, the approach uses real neurons to perform computational work.

The potential is considerable. Biological neural networks can adapt quickly and process information in parallel using little energy. Harnessing these capabilities could eventually help computers and robots learn from experience without the enormous computational resources that other computing systems demand.

But turning living cells into reliable computational components remains a serious engineering challenge.

In a study published this month in Frontiers of artificial intelligenceResearchers at the University of Bristol investigated whether small laboratory-grown models of the human forebrain, known as organoids, could process tactile information represented in Braille.

Arrangements of some 10,000 nerves and supporting cells, one millimeter wide, sat on microelectrode arrays that allowed researchers to stimulate the tissue and record its activity. A touch sensor scanned Braille characters and converted the measurements into electrical stimulation patterns. The sensor sent signals to the organoids and the researchers then analyzed their responses using a conventional machine learning algorithm.

The algorithm was able to identify the correct letter with 61% accuracy based on the response of a single organoid. Combining the processing of three organoids increased the system’s accuracy to an impressive 83%. The combined approach also proved to be more robust against noise that the team artificially introduced.

The findings suggest that living tissue can transform tactile information in ways that preserve useful patterns.

The organoids were not taught to recognize Braille through experience. They used a fixed method to encode the letters and a computer algorithm to classify the neural responses. However, other researchers are finding that biohybrid systems have significant potential when it comes to learning through stimuli.

Melbourne-based biological informatics company Cortical Labs, in collaboration with the Honda Research Institute Europe, recently demonstrated that cultured neurons can improve their performance through interactions with a simulated environment.

Biological computing faces a major obstacle: neurons do not respond to electrical inputs as predictably as conventional digital hardware. This makes sense, given that our meat circuits are confusing things that depend on complex and changing biological dynamics. The Embodied Neurocomputation Framework treats this as an engineering problem, systematically optimizing the information that enters the neural network.

Research to be presented at this year’s NeurIPS conference in Sydney, Australia, demonstrates the benefits of a process for making biological computing more systematic, known as the Embedded Neurocomputing Framework.

Using Cortical Labs’ CL1 biological computers, the collaboration connected living neural networks to a virtual agent that navigates a grid-like world.

The agent had to move toward a target guided by a simulated odor gradient, choosing whether to move forward, turn left, or turn right to find a “food” target. The researchers tested approximately 1,300 configurations of electrical signals used to communicate with neurons, accumulating more than 4,000 hours of real-time interaction in dozens of biological computing units.

In the end, 12 configurations consistently demonstrated an adequate level of learning across multiple episodes. In comparable interactions, this handful of optimized biological agents outperformed conventional reinforcement learning benchmarks in navigation.

So why bother using biology when silicon doesn’t even require such optimization?

The appeal of biohybrid computing lies partly in the extraordinary energy efficiency of our brains. Your gray matter performs astonishing feats of information processing using little more than the equivalent of a relatively bright globe of light.

Neurons also adapt their connections and activity in response to experience. This plasticity provides an alternative route to learning, which differs from the approaches used by neural networks that rely on extensive numerical processing.

If researchers can reliably take advantage of these features, biological computing could one day be the best option in applications where systems need to adapt to unknown conditions in a field far from home.

In a separate demonstration by Cortical Labs, the frequency at which cells in a biological neural network fired was used to direct the movements of a robot dog in a small space.

While a waltzing dog is a step short of a Robocop-worthy canine unit, neural choreography represents a critical link between adaptive wet software and the programmable actuators, sensors, and gears on which we increasingly pin our hopes for reliable monotonous work in the real world.

“Robots that are becoming remarkably capable machines under your physical control still struggle with the higher-order control needed to quickly learn or adapt when conditions change,” says Brett Kagan, co-founder and CEO of Cortical Labs.

“Walking, running and doing stunts is amazing, but determining where and when to perform the actions remains a challenge with traditional methods.”

It is far from the first example of using neurology to drive a robot’s movements. However, each experiment brings us closer to a system that can learn about novel changes in its environment and make informed decisions about what actions to take next.

Despite the progress, none of these demonstrations suggest that biological computers will soon serve in Amazon warehouses or take you around town in a Waymo. Keeping the tissue alive is a problem yet to be solved. Cultured brain cells require carefully controlled conditions and a continuous supply of nutrients. Although they are energy efficient, all that supporting hardware consumes power. The total energy cost of a practical system must be evaluated throughout the entire process, not just the neurons.

Still, the field is moving beyond whimsical demonstrations of organoids playing computer games. Companies are developing frameworks to measure performance, optimize interfaces, and test whether biological computing can deliver consistent results across tasks.

The goal is not to build a computer made entirely of brain tissue. It’s about discovering whether living neural networks can really complement silicon. If successful, biohybrid computing could offer a way to build systems that learn and adapt efficiently in places where our own brains cannot be present.