Scientists Discover How Your Brain Holds Multiple Thoughts at Once

Have you ever tried to remember a phone number while also recalling a shopping list? Your brain manages this feat through a clever trick: it uses rhythmic brain waves to keep multiple items in mind at once. Now, scientists have found that artificial neural networks, when trained to do similar memory tasks, spontaneously develop the same trick. This discovery, published in a recent study, sheds light on how our brains might work and could lead to smarter AI.

The researchers trained recurrent neural networks (RNNs)—a type of AI modeled loosely on the brain—to remember sequences of items over varying delays. After training, the networks began to use oscillatory phase coding (OPC), where each remembered item is represented by a different phase of a shared rhythm. Think of it like a choir singing a round: each voice enters at a different time, but they all follow the same melody. In the brain, this would be like different neurons firing at different points in a brain wave cycle to hold different pieces of information.

The team found that three conditions were necessary for OPC to emerge: adaptation in neural dynamics (neurons adjusting their activity over time), variable delays in memory retention (the network had to hold information for unpredictable periods), and flexibility in reading out information (the network could access the memory in different ways). They also identified two types of neurons in the network: content cells, which encode what the item is, and time cells, which encode when it occurred. These cells connect in structured ways, with time cells linked by phase and content cells grouped by selectivity.

This research not only helps us understand how biological brains might implement working memory but also provides a blueprint for building AI that can juggle multiple tasks and remember sequences more effectively. It could inspire new treatments for memory disorders and lead to more human-like artificial intelligence.