Background

Neurological conditions such as epilepsy are linked to unusual patterns of activity in the brain. When brain cells become either too active or poorly coordinated, it can affect how information is processed, stored, and recalled. Studying these changes directly in people is difficult, so researchers often use computer-based models to explore how brain networks behave under different conditions. 

In this study, we used a type of artificial intelligence (AI) called a spiking neural network. Unlike most AI systems, spiking neural networks are designed to communicate in ways that more closely resemble real brain cells, making them useful for investigating how changes in brain activity might affect learning and information processing. 

Research

We built a spiking neural network that learned to compress and reconstruct simple handwritten digit images. This task allowed us to test how well the network could process and transmit information. 

We then altered the network so that its artificial neurons behaved in a way similar to hyperexcitability—a state in which nerve cells become overly responsive, and fire signals too easily. Hyperexcitability is often observed in epilepsy. 

As expected, this change made the network less effective at learning and reconstructing images. However, we found something unexpected. When we added a small amount of carefully controlled random variation (known as noise) to the input data, the network’s performance improved. Rather than disrupting learning, the noise appeared to help the network recover some of its lost function and process information more effectively. 

Potential Impact

This research shows how brain-inspired AI systems can be used as experimental testbeds for studying how changes in brain activity affect information processing. The findings suggest that, in some cases, small amounts of randomness may help stabilise networks that are functioning poorly. 

Although this study does not provide a direct model of epilepsy or a new treatment, it offers a safe and controlled way to investigate ideas about how neurological disorders affect brain function. In the future, this type of research could help scientists better understand the mechanisms underlying conditions such as epilepsy and explore new ways of restoring healthy patterns of brain activity.