How our brains build mental maps of the world

Our brains constantly create internal maps of the world, allowing us to navigate spaces, recall past experiences, and plan for the future. These cognitive maps enable us to recognize patterns, anticipate changes, and adapt our behavior accordingly. 

Whether it’s finding the right floor in a hotel or realizing we’ve stepped out onto the wrong one, these maps help us make sense of our surroundings.

Neuroscientists have long studied how neurons respond to specific locations, but how these mental maps form over time has remained largely unknown. 

Now, researchers at HHMI’s Janelia Research Campus have uncovered a step-by-step process for how these maps develop in the hippocampus, a brain region critical for memory and spatial navigation.

Tracking neural activity over time

By recording the activity of thousands of neurons in a mouse’s hippocampus over days and weeks, the research team, led by the Spruston Lab, systematically documented how the brain constructs cognitive maps as an animal learns. 

The study focused on how the hippocampus differentiates between two visually similar environments – much like different floors in a hotel – with each offering a reward in different locations.

“We are mapping out the step-by-step process of cognitive map formation, which is such an important concept,” said Weinan Sun, an assistant professor at Cornell University and a co-leader of the research. 

“But there is also a second contribution: The result of watching that process gives us a hint about the underlying computations, and we get a little bit closer to understanding what the brain is doing in making these maps.”

The findings not only provide new insight into memory formation but could also improve treatments for neurodegenerative diseases like Alzheimer’s and advance artificial intelligence (AI) systems by helping them reason more like biological brains.

Brain maps of similar spaces

To study cognitive map formation, researchers used high-resolution imaging to track the neural activity of a mouse navigating a virtual environment. The mouse learned to recognize two nearly identical corridors – one offering a reward at a near location and the other at a far location – based on visual cues.

At first, the mouse had no understanding of the task, and its neural activity looked similar for both corridors. But as learning progressed, distinct patterns began to emerge. 

Initially, the mouse learned where not to search for rewards, ignoring locations where it had previously been unsuccessful. Next, it recognized that only one reward was given per corridor and adjusted its behavior accordingly. 

Finally, the mouse learned to suppress its response at the near reward location in the corridor where the reward was placed farther ahead. As the animal learned, its brain gradually separated the two corridors into completely distinct cognitive maps.

“Initially, the brain activity is very similar, and with learning, the activity becomes more and more different until they are orthogonal,” Sun explained. “In the end, each neural pattern of activity will encode a hidden state that reflects the true nature of the task.”

This ability to encode hidden states – aspects of an environment that are not immediately visible – is crucial for navigating complex and ambiguous spaces in the real world.

The brain extracts hidden information

The researchers identified a special type of neurons, which they call “state cells,” that help encode hidden information and distinguish between similar environments. These cells allow an animal to recognize whether it is in one corridor or the other – even if the visual surroundings appear identical.

“The brain cares about the immediate sensory input but interprets it in the context of the hidden state the animal is in,” Sun said.

This process mirrors how humans learn to differentiate between similar locations. When staying in a multi-floor hotel, for example, the hallways on each floor might look nearly identical. 

At first, we might struggle to tell them apart, but over time, our brains encode subtle differences – such as the layout of furniture or the feel of the carpeting – allowing us to recognize our floor effortlessly.

Computational principles of brain maps

Beyond mapping neural activity, the researchers wanted to understand the computational principles behind these maps.

The experts tested various models and found that the best fit was a Clone-Structured Causal Graph (CSCG) – a type of state machine that infers hidden states based on observed patterns.

This finding suggests that the brain builds cognitive maps much like a computer program that predicts an outcome by piecing together hidden clues.

“One of the ultimate goals of neuroscience is to connect observed behavior with the cellular, molecular, and computational processes that produce it,” saID Janelia Executive Director Nelson Spruston, the study’s senior author. 

“We are getting at the algorithmic level – arguably the hardest to pin down – which helps us connect the dots of how the brain operates to form these computations.”

AI and neuroscience

Understanding how the hippocampus constructs cognitive maps could lead to advances in artificial intelligence and the treatment of memory disorders. AI systems, such as large language models (LLMs), struggle with tasks requiring reasoning, planning, and long-term memory. 

By incorporating biological principles from the hippocampus, AI could become better at contextual learning and decision-making.

“Neuroscience and AI can learn a lot from each other,” said Johan Winnubst, lead scientist at E11 Bio, who co-led the research. 

“What large language models are able to do is very impressive, but they also fail in a lot of very obvious ways. Some of that has to do with reasoning and long-term planning. Maybe we can introduce lessons from the hippocampus to improve these models.”

Additionally, insights from this study could help researchers design better treatments for neurodegenerative diseases like Alzheimer’s, where memory formation and cognitive mapping are severely impaired.

Future research directions

The researchers have made their data publicly available through an interactive visualization tool, allowing scientists worldwide to explore the neural patterns behind cognitive map formation. 

Moving forward, they plan to expand their research to more complex environments, testing whether the same principles apply to different types of learning and decision-making.

By connecting behavior, neural activity, and computational models, this study takes an important step toward unraveling one of neuroscience’s greatest mysteries: how the brain builds the maps that shape our perception, memory, and intelligence.

The study is published in the journal Nature.

—–

Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates.

Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.

—–