How neural networks help the brain predict future events

On 17 October 2024, Martina Lamberti publicly defended her Ph.D. thesis, “Know the Past to See the Future: Memory and Prediction in In-Vitro Cortical Neurons,” at the University of Twente. Her research sheds light on how the brain predicts future events and forms memories. These findings could open new doors in understanding neurological conditions like dementia.

While memory has been extensively studied, much less is known about how the brain uses past experiences to predict future events at a cellular level. Scientists have long theorized that prediction is an actual brain function, but experimental evidence on this function and its connection with memory has been limited. Lamberti explored these cellular mechanisms involved in both memory and prediction and in the link between the two processes.

Lamberti’s work demonstrates that neurons are capable of predicting future inputs, showing that prediction is a general function of neural networks. Her research highlights how memory not only stores past experiences but also helps the brain anticipate what happens next. This discovery could help explain why patients with neurodegenerative diseases like Alzheimer’s struggle with both memory and everyday decision-making.

Mimicking brain conditions

One of the key breakthroughs is the creation of an in-vitro model of neural networks. This model mimics the brain’s conditions, allowing researchers to study how neurons form memories and make predictions outside of a living body. In this model, neural networks were shown to form long-term memories and use recent past information to predict future stimuli.

Lamberti cultured neural networks on multi-electrode arrays to record activity from the neurons and to stimulate them. She used two types of stimulation: focal electrical stimulation, activating a specific subgroup of neurons, and global optogenetic stimulation. In this case, neurons were treated with a virus to make them respond to light.

This setup allowed her to study the formation and consolidation of memory and the relationship between prediction and short- and long-term memory. She discovered that in general, prediction depends on short-term memory, but if long-term memory is formed, as by focal electrical stimulation, the reliance on short-term memory decreases. In contrast, upon global optical stimulation, when no long-term memory is formed, prediction remains fully dependent on short-term memory.

Advancing neuroscience

By developing this system, Lamberti has provided scientists with a valuable tool to investigate the detailed processes behind memory formation and prediction. This research is especially promising for exploring how brain disorders impact these functions and could eventually lead to improved treatments for memory-related conditions.

Lamberti’s findings open new avenues for understanding the brain’s predictive abilities and memory processes. By showing how closely linked memory and prediction are at the network level, her research helps to clarify the brain‘s complex decision-making functions. This work not only enhances our understanding of how we interact with the world around us but also provides a solid foundation for future research into neurological diseases.