Scientists have figured out how the brain learns to do things on autopilot

A new study by Georgetown University scientists shows how this can happen in the brain. After thousands of repetitions, a well-learned task relies less on frontal areas, which are responsible for control and decision-making, and moves more strongly to other parts of the brain. Simply put, it's as if the brain puts the skill into autopilot mode.

Details

Researchers studied how the brain learns to quickly sort similar images of cars. Participants had to distinguish between slightly altered images and put them into one of two categories.

At first, such a task requires attention. The brain needs to compare details, make a decision and control the answer. At this point, the prefrontal cortex, an area associated with executive functions, planning and conscious control, is active.

Then began a long training session. Participants performed more than 30,000 attempts over 5-10 weeks using an app on their phones. The scientists scanned their brains before and after the training using fMRI and EEG - techniques that allow them to observe which areas of the brain are involved in a task and how their performance changes. The study description also states that after initial training and after long practice, they compared which areas of the brain supported category recognition.

After lots of practice, the brain began to work differently. The task began to rely less on the prefrontal cortex and more on areas of the visual and temporal cortex that are associated with recognising complex objects. These areas became better at distinguishing the right categories of images.

Simply put, the brain stopped solving the task from scratch every time. It began to recognise the right type of image more quickly and automatically - in much the same way that an experienced driver no longer thinks separately about each movement of the steering wheel and pedals.

The most important result has to do with multitasking. The more the trained task "left" the prefrontal cortex and required less conscious control, the better participants were able to perform another task in parallel. The researchers call this bypassing the "frontal bottleneck" - a situation in which the prefrontal cortex is overloaded and cannot effectively conduct several complex processes at once.

Why it matters

We're usually not good at true multitasking. It often feels like we're doing two things at once, but the brain is actually switching between them quickly. This takes time and increases the risk of errors.

A new study shows there's an important exception. If one task is very well trained and has become almost automatic, it takes up less of the conscious control system. Then the brain has more resources left for the second task.

This helps explain everyday skills. It is difficult for a novice driver to drive and hold a conversation at the same time. An experienced driver can drive and talk because some of the actions have become habitual. But that doesn't remove the limitations: if the second task also requires vision, attention and quick decisions - like texting on the phone while driving - it remains dangerous.

Background

Scientists have long talked about the "bottle neck" of attention. The idea is simple: the brain can process a lot of information quickly, but tasks that require conscious choice and control often compete for the same resources. The prefrontal cortex plays a particularly important role in this.

Research on multitasking shows that when two complex tasks occur, the brain often queues them up rather than performing them completely in parallel. For example, a 2025 paper in Nature Communications provided neural evidence in favour of the fact that when two tasks overlap, information can be processed sequentially in a frontoparietal network associated with cognitive control.

The new study adds to this picture. It shows that long-term training can change the way a task is processed: some of the work moves away from general control to more specialised circuits. So the brain doesn't so much "become multitasking in general" as it learns to take a familiar action out of the centre of attention.

Source

Study: 'Extensive Experience Remodels Neural Task Circuitry to Escape the Frontal Bottleneck and Increase Automaticity of Categorisation', Journal of Cognitive Neuroscience, 2026.