Inside the minds of 28 epilepsy patients fitted with intracranial electrodes, neuroscientists from Beijing Normal University, the Chinese Academy of Medical Sciences, and University College London have discovered something remarkable about how the brain solves brand-new problems: it shuffles the old pieces around. When you face a situation you've never encountered before, your brain doesn't start from scratch. Instead, it reaches into the library of everything you've learned and rapidly recombines familiar building blocks into fresh configurations—a process that happens so fast it was nearly invisible to science until now.
This capacity to flexibly reassemble past knowledge into new mental structures is how humans navigate an unpredictable world. A chess master recognizes a novel board position by seeing it as a new arrangement of familiar patterns. A surgeon adapts a known technique to an unfamiliar anatomy. A parent finds creative solutions to problems their own parents never faced. Understanding how the brain performs this feat has long intrigued neuroscientists, who suspected two regions—the hippocampus and the medial prefrontal cortex (mPFC)—played central roles. The hippocampus handles memory formation and spatial navigation; the mPFC orchestrates decision-making, planning, and reasoning. But how exactly do they work together?
He Li, Xiongfei Wang, and their colleagues set out to solve this puzzle by asking their 28 patients to complete two LEGO-like inference tasks while their brain activity was monitored in real time. The electrodes captured high-resolution intracranial recordings from both the hippocampus and cortex, a level of neural detail typically unavailable in human studies. What they found, published recently in Nature Neuroscience, points to a specific mechanism: brief, high-frequency bursts of neural activity called hippocampal ripples.
These ripples appear to be the brain's way of hitting pause and reorganizing. During a ripple event—a window lasting mere milliseconds—the hippocampus replays sequences of past experiences, essentially shuffling familiar information blocks into new candidate sequences. Think of it as the brain rapidly testing different combinations: What if this block went here? What if this one came next? The replayed sequences aren't random; they're targeted toward solving the specific problem at hand. Simultaneously, the medial prefrontal cortex shifts its activity patterns in synchrony with these ripples, updating its representations to encode the newly inferred solution as a compositional structure.
The timing is crucial. Replay is strongest during ripple periods, tightly coordinated with mPFC activity, and predictive of how efficiently people actually solve the task. This suggests that hippocampal ripples and replay function as a dynamic online system, constantly updating cortical representations to support real-time inferential behavior. The brain isn't simply retrieving a stored solution; it's actively constructing one.
What makes this discovery particularly significant is that it reveals the neural substrate of human flexibility—that distinctly human ability to think in new ways by recombining the familiar. As the researchers note in their paper, the human brain excels at solving novel problems "by flexibly recombining a limited set of familiar elements, often through the internal planning of sequences that assemble these elements into new configurations." Now we know part of how it does it.