Our brain is bombarded with information every day. Imagine standing in Times Square in New York City, with its flood of sights, sounds, and smells. According to one estimate, we receive about 11 million bits of information per second, which is about the same amount of information needed to stream a high-definition movie. Acting in this world – for example, when looking to find a taxi – requires us to use our attention to filter this flood of input and focus on the information that is relevant to our goals. For example, when looking for taxis in Times Square, we must use our attention to focus on yellow cars. This considerably reduces the amount of information we need to process to find an available cab. Disrupting one’s ability to attend, such as in attention deficit disorders, leads to difficulties in focusing on the appropriate information, which disrupts learning and behavior. 

Over the last several decades, neuroscientists and psychologists have figured out many of the details of how attention works in the brain. We know a set of brain regions in prefrontal and parietal cortex are important for directing attention to specific locations or to specific stimuli (e.g., the yellow cab). And we know that attending to a stimulus magnifies its representation in sensory cortex, making it easier to make decisions and learn about that stimulus. However, this research has focused on how attention works when you are told what to attend to (e.g., someone tells you to ‘look for yellow cabs’). In the real world, things are more complicated.

For example, if you travel to Berlin, taxicabs are beige. So, looking for yellow cars will impair your ability to find a taxi in Berlin. Instead, you must adapt your attention to fit the situation -- you must learn that the ‘attentional template’ to find a taxi has changed to ‘beige car’. While critical to operating in the real world, the mechanisms used by the brain to learn new attentional templates have been largely unknown. 

In a new research article published in Cell, Jahn and colleagues studied this mechanism in monkeys by asking them to repeatedly learn new attentional templates while recording from different areas of their brains. Monkeys performed a task very similar to looking for a cab in a city – they were presented with a display of three colored squares and had to pick the one that was most similar to the current ‘best’ color. Importantly, just like in the real world, the monkeys were not told what color was the ‘best’, rather they had to learn it through trial and error. In this way, the monkeys learned to form an attentional template that allowed them to search for the best color on the screen. After a while, the best color would change (as if they went to a new city) and the whole learning process would begin again.

By analyzing brain recordings, Jahn et al found prefrontal and parietal cortex were important for representing the attentional template on each trial (i.e., what color was ‘best’). Interestingly, these representations were structured – just like in the real world, where ‘pink’ is similar to ‘red’, the template for ‘pink’ and ‘red’ were represented in similar ways in prefrontal and parietal cortex.

Next, Jahn and colleagues built a mathematical model to understand how the monkeys learned new templates. This revealed the templates were changed a little bit on each trial, depending on what feedback the animal received. When they chose a color and got a bigger reward than expected, the template moved towards the chosen color. But when they got a smaller reward than expected, the template would shift away. In this way, the attentional template the animal used to search the space would slowly change on every trial, moving towards the ‘best’ color. This is just like when you go to a new city and initially look for yellow taxis before trying out other colors and, eventually, settling on the correct car color. Jahn and colleagues were then interested in how the animals used the template. In other words, once you know the correct color car to look for, how do you test whether a new car matches this template (or an old one)? Interestingly, Jahn and colleagues found the brain calculated a ‘value’ for each stimulus, where high value meant the stimulus had a color that was close to the ‘best’ and low value meant that the color of the stimulus was far away from the best. The authors found that this helped the brain use one mechanisms for making the decision about selecting a stimulus, regardless of the current attentional template. This may help the brain efficiently adapt to the ever-changing world. Overall, the article presents new insights into how the brain learns to control attention and how humans can flexibly adapt to new environments.