The Unexpected Reasons Scientists Are Making Mushroom Computers

Believe it or not, our brains are much better thinkers than our personal computers are. But scientists hope to change that by creating "neuromorphic computers" to emulate the efficiency of the human brain. IBM, Intel, and a handful of other companies have already broken ground on the endeavor; their neuromorphic computing chips are currently in use for AI processing, but the capabilities are limited. It turns out the hardware involved in manufacturing a brain-on-a-computer-chip device is wildly difficult — and expensive — to create.

Researchers from Ohio State University (OSU) have proposed a cost-effective solution; rather than pouring tons of money and rare earth minerals into creating neuromorphic hardware, why not just use common mushrooms? After all, mycelium, the root-like part of a fungus, already grows in branching networks that are structurally similar to the neural networks in our brains. Furthermore, both brains and mycelium exhibit emergent properties from their individually simple components, such as the ability to solve novel, complex problems.

The study was led by OSU's John LaRocco, a neural engineer whose work explores the interface between the circuitry of the brain and the circuitry of electronic machines. LaRocco and his team connected the mycelium of common shiitake mushrooms to a computer circuit using wires. Then, they toggled the circuit's voltage. After testing different attachment points, the team observed that the "fungal chips" were able to transmit signals and switch states at incredible speeds. Even more remarkable, the fungal networks adapted to repeated charges by adjusting their resistances to optimize the signal speed. From a computing standpoint, the shiitake mushrooms were learning.

Currently, the building block of the modern computer is the all important transistor. Transistors are tiny switches that, when connected together by the billions, process the binary language that computers use to think. They're analogous to the neurons in our brains. However, transistors lack an important feature of the neuron: individual memory. Computer processors can only execute a task by drawing from data stored elsewhere, while neurons are able to learn by retaining memories of past connections. Give a transistor the ability to remember, and you'll have a "memristor," which is functionally an artificial neuron.

Now, John LaRocco and his team are trying to build memristors with fungus. At their most optimized, the shiitake "fungal chips" were able to transmit signals at a frequency of 5.85 kHz. This equates to 5,850 signals per second, which is comparable to the capabilities of the earliest silicon-based memristors. Considering this was the first experiment to use mycelium for such an application, that's quite promising. But why would we use fungus in place of tried-and-true computer hardware in the first place? The answer is (mostly) money.

Manufacturing memristors from silicon and rare earth metals is a delicate, costly process. Companies like Intel and AMD have poured billions of dollars into AI research and development, and a large portion of that funding goes towards neuromorphic computing. Obviously, shiitake mushrooms would be a cheaper, eco-friendly alternative. However, there are other benefits to using fungus computers, as well. LaRocco's team writes, "Shiitake has exhibited radiation resistance, suggesting its viability for aerospace applications." In other words, shiitake mushrooms may be a critical component in the future of space travel. The technology has a long way to go, but such imaginative innovations reveal that evolution is a better engineer than we are.