AI predicts brain behavior, starting with the humble fruit fly

Researchers from the Cold Spring Harbor Laboratory (CSHL) have developed an artificial intelligence (AI) model that can predict the movements and behaviors of fruit flies.

Professor Benjamin Cowley collaborated with Pillow and Murthy neuroscience labs at Princeton University to create an AI model that understands and predicts how neurons work in fruit flies.

“We can actually predict neural activity computationally and ask how specific neurons contribute to behavior,” Cowley said. “This is something we couldn’t do before.”

The research focuses on visual projection neurons that link the optic lobes to the brain. While researching how neurons function in fruit flies, the AI model revealed a distributed code in the fly’s brain, Reddit News said. This code transforms visual perception into different behaviors.

The team developed their model through something known as “knockout training.” The researchers first observed the male flies mating behavior. They then “genetically silenced specific types of visual neurons in the male fly and trained their AI to detect any changes in behavior,” CSHL said.

Furthermore, the researchers used genetic tools that silenced these neuron types and then trained the AI model on normal data and the silenced neuron data, Reddit News said.

This process was repeated multiple times with varying neuron types until the AI was able to accurately predict the behavior of the fruit fly when it encounters a female.

“With their new AI, the team discovered that the fruit fly brain uses a “population code” to process visual data. Instead of one neuron type linking each visual feature to one action, as previously assumed, many combinations of neurons were needed to sculpt behavior.”

Many may be wondering whether this tech could be used to better understand the human brain, and it very well could be, but a fly only has 100,000 neurons, and the human brain has almost 100 billion neurons.

With many decades of work and further understanding of fly computations, Cowley believes that “we can build a better artificial visual system.” Most prominently, the professor believes that “we’re going to better understand disorders of the visual system in much better detail.”