Rapid spatial cognition in mice, with and without neocortex and hippocampus
This study demonstrates that mice lacking both the neocortex and hippocampus can rapidly learn, generalize, and retain complex 3D spatial navigation tasks, indicating that these sophisticated cognitive abilities do not strictly depend on these brain structures.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
To understand how an animal learns its way through a new world, scientists have long focused on two specific parts of the brain: the hippocampus and the neocortex. The hippocampus is often thought of as the brain's internal mapmaker, helping creatures remember where they have been and where they need to go. The neocortex, the outermost layer of the brain, is believed to handle complex thinking, allowing animals to figure out rules and adapt to new situations. For decades, the prevailing view has been that rapid learning—the ability to solve a problem after just a few tries—requires these two sophisticated structures. If an animal lacks them, the assumption was that it would be stuck in a loop of trial and error, unable to form a mental picture of its surroundings or remember a successful path. This idea has shaped how we understand intelligence, suggesting that quick thinking is a luxury of complex brains.
However, a new study challenges this long-held belief by asking a simple question: what happens when you remove the very parts of the brain we think are essential for learning? Researchers at the California Institute of Technology designed a complex, three-dimensional maze for mice to navigate, a structure resembling a grid of city streets but stacked in layers. They then tested mice that were born without a neocortex or a hippocampus. These animals, known as acortical mice, possess only the older, deeper parts of the brain. The goal was to see if these mice could learn the maze, remember the solution, and apply what they learned to new versions of the puzzle. The findings reveal that the brain's ability to learn quickly does not strictly depend on the structures we once thought were indispensable.
The researchers built a maze called the Manhattan Maze, a reconfigurable labyrinth made of two layers of corridors. Imagine a grid of horizontal tunnels on the bottom and a matching grid of vertical tunnels on top, separated by a plastic sheet with holes. By swapping this sheet, the researchers could instantly change which tunnels were connected, creating thousands of unique layouts without moving the physical maze. The mice started with a simple version to learn that water rewards were available at two specific ports, one at the "home" end and one at the "out" end. They had to run back and forth between these ports to drink. Once they mastered this basic routine, the researchers introduced them to a complex maze with a long, winding path that required nine correct turns to reach the reward.
Normal mice learned this path remarkably fast. Within just a few hours and after only a handful of rewards, they stopped wandering aimlessly and began taking direct routes. They didn't just memorize a single sequence of turns; they learned a general rule about how to move forward without backtracking, a skill they could apply to new maze configurations the very next day. They also showed an ability to remember the solution for weeks. Most strikingly, they could figure out which specific decision point in the maze was the most important, even if that decision was far away from the reward in both space and time. This ability to assign credit to a distant choice is a hallmark of sophisticated learning.
The researchers then turned their attention to the mice lacking a neocortex and hippocampus. At first, these animals struggled. They explored the maze inefficiently, running back and forth in the same corridors and taking much longer to find the water than their normal counterparts. They seemed unable to build a mental map of the space during their initial exploration. However, once they finally found a reward and began to learn, something surprising happened. The acortical mice improved just as quickly as the normal mice. They learned to take direct routes, remembered the path for days and even weeks, and could generalize their knowledge to new maze layouts. They even managed to identify the critical decision points in the complex maze, showing that they could assign value to a choice made far from the reward.
The study suggests that the delay in the acortical mice was not a failure of learning, but a failure of exploration. Without the higher brain structures, they were less efficient at gathering information about the maze before they found a reward. But once they had that first piece of information, their remaining brain circuits were fully capable of processing it, forming a memory, and using it to navigate. The researchers used computer simulations to test how different learning algorithms might explain this behavior. They found that a model based on building a map of the environment could explain the mice's rapid success, while a model based purely on trial-and-error reinforcement could not. This supports the idea that the subcortical brain, the older part of the brain, is capable of constructing a cognitive map and learning from very few experiences.
This discovery reshapes our understanding of where intelligence lives in the brain. It shows that the ability to learn quickly, remember a path, and adapt to new situations is not the exclusive domain of the neocortex and hippocampus. These functions can emerge from the deeper, more ancient parts of the brain, provided the animal has the opportunity to explore and find a reward. The study does not claim that the neocortex and hippocampus are useless; rather, it suggests that their primary role in this context might be to facilitate the initial exploration and organization of information, rather than being the sole seat of learning itself. The brain's capacity for rapid spatial learning appears to be a more fundamental and widespread trait than previously thought, residing in circuits that have existed long before the evolution of the complex outer layers of the mammalian brain.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.