Through extensive testing, researchers at the Beckman Institute for Advanced Science and Technology at the University of Illinois Urbana-Champaign established a gold standard to determine how inhibitory control changes between individuals and within an individual.

Inhibitory control tests are often used in neuropsychological studies to measure an individual’s ability to override automatic responses to complete a specific goal.

Deficits in inhibitory control are seen in many neuropsychiatric disorders but these tests aren’t currently reliable enough to use as clinical markers or to track individual differences.

For example, in the Stroop task, participants are asked to identify the color of a word. A congruent test will show the word “blue” in matching blue ink. An incongruent test will show the word “blue” in nonmatching red ink, making it more difficult to identify the ink’s color due to the automatic tendency to read the word.

During these tests, researchers can see the congruency effect when a participant’s speed and accuracy change based on whether the word matches the ink color. On average, participants demonstrate faster processing speed and accuracy during a congruent condition than an incongruent condition.

Caterina Gratton, a professor in the Department of Psychology at Illinois, and co-first authors Hyejin Lee and Derek Smith, led the research team. Lee is a postdoctoral researcher in the Department of Psychology at Illinois and Smith is a researcher in the Department of Neurology at the Johns Hopkins University School of Medicine. Their work appears in the journal Nature Human Behavior.

The congruency effect has been extensively tested, thoroughly replicated in lab settings and has been shown to be successful. Congruency tasks are stable across time and likely represent a trait-like feature of an individual.

These characteristics have spurred investigations of individual-level inhibitory control behavior both within and between individuals and the relationship between behavior and neural brain activity.

However, poor reliability in measures of inhibitory control limits the use of these tests for clinical applications, in predicting self-regulation in real world scenarios and may interfere with theoretical interpretations to other control functions.

“Establishing reliability for individual-level inhibitory control was a major goal of this research,” Gratton said.

While the National Institutes of Health Toolbox Flanker Inhibitory Control and Attention Test only suggests 40 test trials per individual, Beckman researchers estimate that individuals require about 1,000 trials to increase reliability and obtain clinical-level precision.

For reference, it takes a participant about one hour, on average, to complete 1,000 trials.

The team collected data from nine participants across 36 sessions during which they each completed four inhibitory control tasks, three of which included congruency effects. Two of the four inhibitory control tasks were completed per session. Using this dataset, called the EPIC or Extended Precision measurement of Inhibitory Control, researchers determined the number of tests required per individual to reach maximum precision and how feasible that level of testing would be.  

“We’ve shared this dataset as a public resource, and we’re excited to see other researchers use it to further examine related questions,” Lee said. 

In addition to the EPIC dataset, the team used two other public datasets as well as simulations to demonstrate that collecting more trials affects both estimates of between-participant variability and the performance of advanced modeling approaches.

Through their investigation, the team found that congruency effects can be successfully measured with high precision and that extensive data collection is feasible. Repeated measures reduce variability in congruency estimates and high trial sampling size stabilizes between-participant variability.

The team also found that within-participant errors can bias between-participant estimates, indicating that obtaining an appropriate level of trial counts per individual is more important to increase reliability than the number of participants in the study. Additionally, these findings support that predictive machine learning models benefit from increased within-participant data.

“This type of testing has been used for decades and is relevant to many psychological and practical applications,” Gratton said. “Our work builds on a rich foundation of work.”

In the future, the research team wants to examine how inhibitory control changes in relation to aging and to improve the accuracy of machine learning models that predict cognition and behavior based on brain activity and neural network data gathered from noninvasive medical imaging techniques like magnetic resonance imaging.

“I’m excited to have more discussions on how to improve these measures,” Lee said.

Editor’s Note: The publication titled “Precise individual measures of inhibitory control” can be accessed online at https://www.nature.com/articles/s41562-025-02198-2.

This research was financially supported by funds from the National Science Foundation CAREER grant, National Institutes of Health, National Institute of Neurological Disorders and Stroke, National Institute of Mental Health and the Therapeutic Cognitive Neuroscience Fund.