Can the brain of a housefly resolve slow reaction times in AI and robots 

Ever had difficulty swatting a pesky housefly? New evidence suggests that houseflies combine temporal motion and visual signals, which gives them close to ‘predictive’ vision. 

By understanding how tiny fly brains can process information at high speeds, scientists hope that these findings may be adapted to AI systems and robotics to improve efficiency.  

Traditional models of neural processing assume that information flows through fixed pathways. This is similar to how current AI systems are modelled, relying on large-scale data processing pathways that are slow and energy intensive. Meanwhile, houseflies are able to react to stimuli in milliseconds, sometimes before the visual signal has been fully delivered. This allows them to react rapidly, even whilst they are already moving and flying around quickly.  

A new study, conducted by the University of Sheffield and Queen Mary University, explains how houseflies maintain visual accuracy during fast motion.  

“Patterns [of light] give a very high information rate in the photoreceptors… we found what we call frequency hopping,” said Jouni Takalo, a neuroscientist and physicist from the University of Sheffield.  

He describes ‘frequency hopping’ or ‘frequency jumping’ as a mechanism where visual signals to the brain shift to a higher frequency during fast movement. This means information is sent to the brain much more quickly, allowing insects to move in sync with what they see. 

While human eyes have a single lens, flies have compound eyes which are made up of thousands of tiny lenses. This allows them a wide angle of vision and excellent movement detection. All animal eyes – including insects and humans – operate through saccades. These are voluntary movements of the eyes in reaction to your surroundings, bringing different points of interest into focus.  

Behind the lens, are cells called photoreceptors which convert light signals into electrical signals. These travel along inter-neurones to the brain, where the signals are processed leading to vision. 

The team replicated the visual conditions experienced by flies during rapid flight by using diverse, fluctuating light patterns to induce saccades. They then measured signal frequency from photoreceptors to inter-neurones, which drastically increased during saccades. 

These findings challenge current insect models of vision, which assumed that compound eyes are fixed, and low in visual accuracy. “The [current] simple filter model doesn’t give the accuracy which is needed to explain what is actually happening,” explained Takalo. Instead, he proposes a new model of the eyes, in which the photoreceptors shift in time with saccades and signal frequency increases. 

Takalo believes this movement-driven, adaptive processing system identified in insects could eventually be used in AI systems and robotics, for example in self-driving cars, to make them faster, less energy-intensive and less expensive.  

“I talk from time to time with people in robotics and they have all kinds of problems with their algorithms, for example varied levels of light… insects have solved these problems,” he said.  

By Marina Milsum, July 5, 2026.

Edited by Kazuma Oura.