Fruit Fly Brain Replicated in a Computer... Now Driving, Gaming, and Investing [Tech Talk]

3D Fruit Fly Brain Connectome Released

Neural Network Connected to Programs for Training

Still an Experimental Curiosity... Boosting Brain Research

Editor's NoteWe break down complex technologies that are essential yet unfamiliar to our daily lives—such as AI, semiconductors, telecommunications, and biotechnology—and explain them in an easy-to-understand way.

What if you could replicate a living creature's brain inside a computer? Thanks to Google, what once seemed like science fiction has now become a reality. They have released for free, as open source, map data that connects the neural network of a male fruit fly's brain.

Even more astonishing is that developers, with the help of artificial intelligence (AI), have connected the fruit fly brain to various tasks. The virtual fruit fly brain, running on a computer, can now be used to drive, play games, and even invest in Bitcoin.

Game Connected to ‘Fruit Fly Brain’ Completed Over 10 Years

A screen connecting the Drosophila brain map 'Connectome' with the rhythm game 'Beat Saber'. Screenshot by X

Google, in collaboration with the Howard Hughes Medical Institute (HHMI) in the United States, the University of Cambridge in the United Kingdom, and other research teams, has recently unveiled a 'brain map' that digitally recreates more than 166,000 brain neurons and 125 million synapses connecting them in a male fruit fly. This type of mapped biological neural network data is referred to as a "connectome."

This fruit fly connectome is the culmination of a research project that took a full 10 years and holds the distinction of being the largest brain map ever created by humans to date. Google has made this data available for free as open source.

Afterwards, a developer conducted a unique experiment. By linking the fruit fly brain map to video games, they used the virtual fruit fly replicated in a computer to test game control. In fact, the fruit fly successfully completed complex games such as the rhythm game "Beat Saber," the classic shooting game "Doom," "Super Mario," and "Minecraft."

A Programmable Fruit Fly... Sparks a Trend in Training

An image showing the connection of a fruit fly connectome to autonomous driving software. Photo captured from X

How were developers able to teach the fruit fly brain to control a game? First, they set up the sensory neurons in the fruit fly brain to activate in response to changes on the game screen. The resulting neural activities were then mapped to actions such as movement or attack keys. They also implemented a reward system based on reinforcement learning, giving certain stimuli when the fruit fly brain successfully performed in the game.

This approach is applicable to more than just gaming. It has been proven feasible for various activities that require specific responses to changing environments, such as autonomous driving or investing in Bitcoin. As a result, there is currently a boom among developer communities experimenting with programming using the fruit fly brain.

Could a Human Brain Be Transferred to a Computer Like a Fruit Fly's?

If it is possible to train the fruit fly brain map in games, driving, and investing, could a day come when the human brain is also digitally replicated? While the fruit fly connectome experiment is fascinating, transferring the actual consciousness of a fruit fly into a computer is a completely separate issue.

While the fruit fly brain can control games, manually linking each neural network to a game is actually work performed by humans. The fruit fly is not performing tasks through its own intelligence. Rather, it would be more accurate to say the researchers used the fruit fly nerves as if they were a computer chip circuit diagram.

Male fruit fly connectome released as open source. Howard Hughes Medical Institute (HHMI)

The connectome itself also has clear limitations. Although the connectome visualizes an organism’s highly complex brain network as a three-dimensional map—and is a valuable tool to deepen our understanding of brain mechanisms—the really critical variables remain out of reach, such as specifically how each brain area interacts and how neurotransmitters are actually applied.

AI Deepens Academic Understanding of the Brain

Scientists have been constructing connectomes for decades. Currently, organisms that have connectomes mapped include Caenorhabditis elegans (a nematode worm) and the fruit fly, relatively simple creatures with far less complex neurons compared to advanced animals like humans.

Nevertheless, because the brain itself is such an incredibly complex organ, it took a full decade just to complete the fruit fly connectome. That is because creating a connectome requires slicing the brain into millions of ultra-thin sections, photographing each section with an electron microscope, and then stitching together these flat images in a computer to reconstruct a three-dimensional model—a painstaking process.

Thanks to the implementation of a detailed connectome, the research team was able to identify the differences in neuronal cells between male and female fruit flies. Google Homepage

Google introduced a new AI technology called the "Flood-Filling Network" during the implementation of the fruit fly connectome. This technique automatically tracks the root of each neuron as thin brain cross-section images are merged together. It was instrumental in reproducing irregularly shaped and hard-to-count three-dimensional neurons as one unified model.

Going forward, Google's goal is to create a connectome of the zebrafish brain. Like the fruit fly, the zebrafish connectome is a project that research teams have been working on for over a decade. By leveraging accumulated research results and new AI technologies, the zebrafish is considered the leading candidate for the next completed connectome after the fruit fly.

This content was produced with the assistance of AI translation services.