Neuroscientists uncover how the brain maps behavioral sequences

A new study published in Nature has identified brain cells that act like “map makers,” helping animals track their position within sequences of behaviors. These neurons, located in the medial frontal cortex, can encode abstract patterns of progress, enabling the brain to manage complex tasks such as planning, reasoning, and decision-making. The researchers found that these neurons function similarly to a music box, capable of flexibly organizing sequences of actions to adapt to changing goals.

Human behavior is highly structured, often involving elaborate sequences of actions to achieve specific goals. Whether it’s cooking a meal or solving a complex problem, these sequences require careful coordination. When tasks share common elements, the brain forms generalized frameworks called schemata, allowing it to adapt and learn new behaviors more efficiently.

While past research has implicated the medial frontal cortex in mapping task structures, forming schemata, and switching between tasks, the precise biological algorithms behind these functions have remained unclear. The researchers aimed to uncover how neurons encode abstract progress in complex, multi-goal tasks, helping to explain how the brain organizes and flexibly executes sequences of actions.

“Every day we solve new problems by generalising from our knowledge. Take cooking for example. When faced with a new recipe, you are able to use your background knowledge of similar recipes to infer what steps are needed, even if you have never made the meal before. We wanted to understand at a detailed cellular level how the brain achieves this and also to infer from this brain activity the algorithms being used to solve this problem,” explained Mohamady El Gaby, the first author on the study and postdoctoral neuroscientist in the Behrens lab at the Sainsbury Wellcome Centre at University College London and Nuffield Department of Clinical Neurosciences, University of Oxford.

The research team designed an experiment using mice and a structured maze task. The task required the mice to navigate a 3×3 grid maze to collect water rewards located at four goal positions (labeled A, B, C, and D) arranged in a repeating loop. Once the reward at location D was collected, the sequence reset, and the mouse needed to return to location A to continue the loop.

Although the spatial locations of the rewards changed between tasks, the overall sequence structure remained consistent. This setup allowed researchers to investigate whether the mice could learn an abstract framework (the sequence) independent of specific spatial layouts.

The study involved 13 mice, which were trained across two phases. In the first phase, the mice were allowed to perform as many trials as needed to master each sequence. In the second phase, the mice were introduced to a “rapid-task regime” involving three new tasks daily, with only limited trials per task. This phase tested the mice’s ability to generalize the sequence structure and perform efficiently without extensive practice.

To examine the neuronal activity underpinning these behaviors, researchers used silicon probes to record activity from neurons in the medial frontal cortex. The probes allowed researchers to monitor how individual neurons fired in response to task progress, goal states, and other behavioral markers. By analyzing these patterns, the team could infer how the brain organizes information about tasks and sequences.

The researchers discovered that neurons in the medial frontal cortex encoded the mice’s progress toward specific goals in a sequence, a feature termed “goal-progress tuning.” These neurons fired in response to the animal’s position in the abstract task structure, rather than physical variables like time elapsed or distance traveled. This allowed the mice to maintain a flexible understanding of their progress, regardless of changes in the maze layout.

Additionally, a subset of neurons exhibited “state tuning,” meaning they were specifically active at certain points in the sequence (e.g., at goal A or B). These state-tuned neurons were organized into clusters or “modules,” with each module acting as a memory buffer for a specific part of the sequence. These modules allowed the brain to track and predict the sequence’s structure, enabling rapid adaptation to new tasks.

When the sequence structure was modified to include a fifth goal (ABCDE), the same neural systems adapted seamlessly, demonstrating the brain’s ability to generalize its task maps. This showed that the medial frontal cortex uses flexible, reusable “building blocks” to represent abstract task structures, rather than creating entirely new representations for each task.

“We found that the cells tracked the animal’s behavioural position relative to concrete actions. If we think of the cooking analogy, the cells cared about progress towards subgoals such as chopping the vegetables. A subset of the cells were also tuned to map the progress towards the overall goal, such as finishing preparing the meal. The ‘goal progress’ cells therefore effectively act as flexible building blocks that come together to build a behavioural coordinate system,” said El Gaby.

The researchers also identified a hierarchical organization in the neural activity, where simpler goal-progress signals were used to build representations of more complex task structures. These findings were modeled computationally using a framework called the structured memory buffer (SMB) model. According to this model, neurons are organized into modules that encode progress relative to specific behavioral steps. These modules form a dynamic network that can store and compute sequences of actions, allowing the brain to adapt quickly to new tasks.

While the study provides important insights, it is not without limitations. The findings are based on animal models, which, though highly informative, may not fully capture the complexity of human behavior. Further research is needed to confirm whether similar mechanisms operate in the human brain. Preliminary evidence suggests that equivalent circuits are active in healthy humans, but more studies are necessary to explore this connection.

Additionally, the study focused on relatively simple task structures. Future research could investigate how the brain handles more complex, multi-layered sequences or combines separate sequences into larger frameworks. Understanding these higher-order processes could bridge the gap between basic neural algorithms and the sophisticated behaviors seen in humans.

The researchers are also interested in how these patterns of brain activity emerge during development and learning. By examining how the brain builds and refines its task maps over time, scientists hope to uncover new strategies for enhancing learning and adaptability.

The study, “A cellular basis for mapping behavioural structure,” was authored by Mohamady El-Gaby, Adam Loyd Harris, James C. R. Whittington, William Dorrell, Arya Bhomick, Mark E. Walton, Thomas Akam, and Timothy E. J. Behrens.