Irrelevant visual details interfere with decisions under uncertainty, brain research shows

Cognitive flexibility, the ability to adjust one’s way of thinking or behavior as a situation changes, is known to gradually decline with age or due to some conditions, including Alzheimer’s disease, Parkinson’s disease, schizophrenia and substance use disorders. One aspect of cognitive flexibility is being able to switch between different tasks without becoming distracted and losing focus.

Researchers at the University of Chicago recently carried out a study involving both people and monkeys to understand how suddenly changing the rules of a task influences how the brain processes available information. Their findings, presented in a paper published in Nature Neuroscience, suggest that uncertainty increases the likelihood that irrelevant visual features will interfere with decision-making.

“Even healthy young people feel stretched when they juggle several tasks, and a half-century of psychology research shows that we become slower and more error-prone right after a switch,” Cheng Xue, a co-first author of the paper, told Medical Xpress. “However, we still didn’t know what caused that effect in the brain.

“I was trained as a physicist, so personally I tend to think of the brain as a dynamical system whose activity is shaped by an underlying landscape, like water flowing over the bedrock of a creek. My starting intuition was that the type and timing of our mistakes say a lot about that landscape.”

Testing performance on tasks when rules are unclear

Xue and his colleagues wanted to find a possible neuronal explanation for why switching between different tasks can reduce performance. Specifically, they wished to understand what brain changes are associated with mistakes during basic perceptual judgments when people and nonhuman primates are unsure about what task they should be performing.

To shed light on these neural changes, the researchers conducted experiments involving humans and monkeys. Both species were shown images as they performed one of two visual tasks.

The tasks required participants to focus on one feature of an image while ignoring another. The rule governing which visual feature mattered changed during the experiment, requiring participants to determine what task to perform at different times.

“Their performance on the task they chose to do indeed became worse, as we expect from the well-documented task-switch cost,” explained Xue. “To find out why the brain makes such mistakes, we first taught a model to err in the same way. Instead of training an artificial neural network to perform the task well, we trained it to reproduce the animals’ actual choices, mistakes included. The network then became a specimen we could take apart to generate precise hypotheses.”

After training a machine learning model to reproduce the choices made by monkeys during the experiment, the researchers trained a second model to make correct choices on the same tasks. This allowed them to compare the performance of the two.

“Analyzing the model showed that mistakes during task switching are likely driven by the interference of the task-irrelevant information with task-relevant information,” said Xue. “We tested the model’s predictions with new behavioral experiments in humans, analyses of the neuronal recordings while the animals performed the task, and small artificial perturbations of the neuronal activities in the animals’ visual cortex.”

The team’s findings suggest that the additional mistakes made by both humans and monkeys under uncertainty are not random. Instead, the mistakes appeared to follow a specific pattern, which the team dubbed “feature interference.”

“When we are unsure which rule applies, the brain holds on to information it should be ignoring,” explained Xue. “The brain’s representations of different features are not perfectly separate; they sit at an angle to one another. The lingering irrelevant information therefore leaks into the relevant information and biases decisions in a predictable direction.”

The results gathered by Xue and his colleagues indicate that, under uncertainty, mistakes may not come from a shortage of mental resources. Instead, mistakes could arise, at least in part, when mental representations of different visual features interfere with one another.

“We confirmed this in several ways,” said Xue. “When unsure of the rule, people remembered the ‘irrelevant’ feature better. The value of the ignored feature pushed choices in a consistent direction. In V1, the first stage of visual processing in the cortex, one feature could be read out from neural signals that normally carry the other. Stimulating a single V1 site shifted both judgments together in the direction the model predicted.”

Toward a better understanding of mental flexibility

This study offers new insight into how uncertainty, particularly a change of rules, affects the performance of humans and monkeys on visual tasks. As some psychiatric and neurodegenerative conditions are linked with deficits in cognitive flexibility, the team’s findings could pave the way for new disorder-specific research.

“If each condition leaves its own signature in the pattern of errors, those patterns could eventually support earlier diagnosis and more targeted interventions,” said Xue. “Meanwhile, it is at least comforting for me to know that when I struggle to multitask, it isn’t a failure of willpower or discipline. The brain is prone to relying on the wrong information when switching gears. Perhaps the solution is simply to finish one task, then give yourself a short buffer with something undemanding, like tidying your desk, before starting the next.”

Eventually, these initial results could help to improve existing neurocognitive models of mental flexibility. Xue and his colleagues plan to continue their research in this field, focusing on patterns of feature interference that may be linked with specific neuropsychiatric conditions.

“We want to find out whether error patterns can serve as fingerprints for specific disorders,” added Xue. “In a related recent study by colleagues in our lab, Douglas Ruff, Drew Sheets and others (PNAS, 2026), they found similar interference in an animal model of Alzheimer’s disease, even without task uncertainty.”

Who’s behind this story?

Ingrid Fadelli

Freelance journalist with BSc Psychology and MA International Journalism. Covers AI, robotics, neuroscience, and astrophysics since 2018.

Full profile →

Gaby Clark

MA in English, copy editor since 2021 with experience in higher education and health content. Dedicated to trustworthy science news.

Full profile →

Robert Egan

Bachelor’s in mathematical biology, Master’s in creative writing. Well-traveled with unique perspectives on science and language.

Full profile →

© 2026 Science X Network