New research shows brain activity travels between a small number of states based on what we are thinking and paying attention to

July 5, 2023 (last updated on January 8, 2024)

The brain state optimal for focus depends on the task, according to a paper from the Cognition, Attention, and Brain Lab

By Sarah Steimer

Using functional MRIs, researchers found that one brain state consistently predicted periods of inattention. However, different brain states occurred when people were attentive to tasks and movies. For the first time, this shows that the brain state optimal for focus depends on what we are doing, whereas the brain state associated with suboptimal attention is shared across contexts.

The research, published in eLife, originated in the Cognition, Attention, and Brain Lab and follows previous research on narrative engagement from the lab.

Monica Rosenberg
Monica Rosenberg


“One of the things that our lab studies more broadly is how we pay attention and how our attentional states change over time,” says Monica Rosenberg, an assistant professor in the Department of Psychology. “Why, at some moments, are we engaged in what we're doing, and at other moments we're distracted or we're mind-wandering.”

Previous studies have shown that the brain tends to transition between clusters or patterns of activity that are relatively constrained, rather than all parts of the brain changing randomly in all possible ways. It tends to be the case that brain activity cycles between one pattern of activity, or brain state, to another. What the lab was interested to learn was how the small number of brain states that we cycle through maps on to the various ways that we pay attention to the world.

“The way our brain activity changes over time can be simplified as the brain jumping across a small number of states in a latent space,” says doctoral student Hayoung Song. “We were interested in what drives transitions across brain states, and whether that has any relevance to how our attention constantly changes over time in various situations.”

The team designed a study that used functional MRI to measure changes in large-scale brain states as study participants’ attention and engagement shifted in four task contexts: performing a complicated task in which they had to constantly exert their attention; watching comedy sitcom episodes; watching a boring educational documentary; and looking at  a central dot on a screen while resting.

“You may easily be engaged in a funny episode, whereas you may have to force yourself to stay awake when watching an educational  documentary.” says Song. “So we purposely picked shows that were on different ends of the spectrum in how engaging they were.”

Hayoung Song
Hayoung Song


With the brain data they collected, the researchers used machine learning algorithms to capture brain states that recurred during the various activities. They found four canonical brain states that occurred regardless of whether a person was watching something boring or something fun, or whether they were doing a task or simply resting.

Though the four brain states commonly appeared in all conditions, the probability of whether a certain brain state was likely to appear differed across conditions. They found that different brain states occurred when people were engaged in a comedy and performing well on a complicated task. However, there was just one common state indicating low attention in all of the contexts. This state recurred when people were disengaged in the show and performing poorly on the task.

In addition to their research findings, the team says the study underscores that even a coarse, broad perspective of brain dynamics — in this case, focusing on just four states — can provide insight into what a person is thinking and how they’re paying attention. Plenty of questions can be answered by studying the brain from this bird’s-eye view.

“It motivates work that investigates what the right scale is for asking the question we're interested in,” Rosenberg says. “Hopefully this spurs future work not to just look at high spatial resolution patterns, but to also consider these large-scale states and ask how they relate to other mental processes as well.”