Neuroscientists discover the brain’s memory center starts “full” and prunes itself down to optimize learning

A newly discovered developmental process reveals that the brain’s primary memory center starts out with an excess of tangled, random connections that get pruned away to form a highly structured, efficient network as an animal grows. These physical and functional changes optimize the brain’s capacity to store and retrieve memories over a lifetime. The study detailing this transformation was recently published in the journal Nature Communications.

The hippocampus is a seahorse-shaped region deep within the brain that handles memory formation and spatial navigation. Within this region lies a specialized circuit called the CA3 network. This area acts as an autoassociative memory system, meaning it helps the brain recall a complete memory from just a tiny fragment of information. For example, the network allows a person to remember an entire childhood kitchen just from the smell of a single spice.

To accomplish this feat, nerve cells in this region communicate through electrical and chemical junctions called synapses. The brain’s elasticity allows these connections to grow stronger or weaker over time as an animal learns new things. While researchers understand how this network operates in an adult brain, the way it physically takes shape after birth has remained unclear.

Researchers at the Institute of Science and Technology Austria wanted to understand how this vital memory network develops. Two competing philosophical and biological models framed their approach. The first model is the tabula rasa, or blank slate theory. This concept suggests that the brain starts with very few connections and slowly builds them up as the animal experiences the world.

The opposing model is the tabula plena, or full slate theory. In this scenario, the brain begins with an overabundance of connections that are gradually trimmed away, leaving only the most necessary pathways. Neuroscientist Victor Vargas-Barroso led the investigation to find out which model correctly describes the developing hippocampus. Vargas-Barroso worked alongside Peter Jonas, a professor of life sciences at the institute, to map the circuitry of the mouse brain.

To trace the wiring of this brain region, the research team examined mouse brains at three distinct developmental stages. They looked at mice shortly after birth, during adolescence, and in adulthood. The team used an advanced recording technique called multicellular patch-clamp recording. This method allowed them to monitor the tiny electrical signals of up to eight individual nerve cells at the exact same time.

By stimulating one cell and listening to the responses of the others, the team could map exactly which neurons were communicating with one another. To physically see these connections, they filled the cells with a specialized dye. This allowed the researchers to reconstruct three-dimensional models of the nerve cells using high-resolution microscopes. In total, they tested more than seven thousand potential connections between nerve cells.

Through this mapping process, the researchers observed a massive shift in how the cells were linked. In the youngest mice, the nerve cells were densely packed with connections that seemed to form at random. As the mice grew into adulthood, the total number of connections dropped. The network shifted from being highly localized and dense to becoming sparse, widespread, and highly structured.

“This discovery was quite surprising,” Jonas said in a press release. “Intuitively, one might expect that a network grows and becomes denser over time. Here, we see the opposite. It follows what we call a pruning model: it starts out full, and then it becomes streamlined and optimized.”

The physical shapes of the neurons also changed in unexpected ways. Nerve cells have two main types of branches: axons that send signals and dendrites that receive them. The team found that the signal-sending axons actually grew shorter and less complex as the animal aged.

In contrast, the signal-receiving dendrites continued to grow and develop more receptor sites. This means the trimming process was heavily one-sided, mainly affecting the outbound communication cables of the nerve cells. In the adult mice, the remaining axon connections became clumped into specific, specialized patches rather than being spread evenly.

Beyond the physical wiring, the team discovered a major shift in how the cells transmitted electrical messages. In the immature brain, single connections between cells were incredibly strong. A single signal from one neuron was often enough to make the receiving cell fire off its own electrical pulse. Researchers refer to this as a near-detonating effect.

In adult mice, these individual connections became much weaker. A single incoming signal was no longer enough to trigger a response in the receiving cell. Instead, the adult nerve cell required simultaneous signals from multiple different neighbors to reach the threshold for firing.

This shift means the mature brain relies on integrating multiple pieces of information at once, rather than reacting strongly to a single input. The brain moves from a system that relies on timing to a system that detects spatial coincidences. To see how these physical changes affected memory, the researchers built a computer simulation of a network with one hundred thousand neurons.

In the computer simulation, they tested how different types of connectivity influenced the network’s ability to recall stored patterns. The results showed that moving from a dense, highly reactive network to a sparse, weaker, and wider network actually maximized the system’s memory storage capacity. A specialized network of weaker connections proved much more efficient at retrieving information.

Jonas suspects that starting with a massive tangle of connections allows neurons to link up quickly in early development. This brain region does not just store visual, smell, or sound information in isolation. It links all these sensory inputs together into a cohesive memory.

“That’s a complex task for neurons,” Jonas explained in the release. “An initially exuberant connectivity, followed by selective pruning, might be exactly what enables this integration.” Without this dense starting point, neurons would be too far apart to find each other, making communication nearly impossible.

While these results offer a new understanding of brain maturation, the study has a few limitations. The researchers conducted their mapping on thin slices of brain tissue. Because they only looked at specific slices, they could not observe long-distance connections that run across the entire brain. The exact biological mechanisms that drive the physical trimming of the axons also remain unidentified.

Future research will need to explore what triggers this selective pruning process. Scientists suspect that specialized immune cells in the brain might act as the biological shears that cut away the excess connections. Specific types of inhibitory nerve cells that develop during adolescence might also play a role in this physical remodeling.

Researchers also want to use live imaging techniques to watch individual synapses appear and disappear over time in a living animal. This live imaging could reveal whether the structured patterns seen in adult brains are pre-programmed by genetics or carved out by the animal’s daily life experiences. Currently, it is not clear how the brain knows which connections to keep and which to cut.

The findings might also shed light on human development. Humans develop episodic memory relatively late, around two years after birth. This developmental milestone roughly matches the timeline of the adolescent mice in this study.

This timeline might help explain infantile amnesia, which is the inability of adults to remember events from their earliest years. The memories might form in the dense early network, but they become inaccessible once the brain prunes its connections. Further studies in the human brain will be necessary to confirm if our own memories follow this exact same developmental path.

The study, “Developmental emergence of sparse and structured synaptic connectivity in the hippocampal CA3 memory circuit,” was authored by Victor Vargas-Barroso, Jake F. Watson, Andrea Navas-Olive, Alois Schlögl, and Peter Jonas.