A recent study published in PLOS One suggests that adults with attention-deficit/hyperactivity disorder (ADHD) exhibit a distinct pattern in how they visually process information over time. This pattern differs enough from that of neurotypical adults that a machine learning algorithm was able to accurately classify individuals with ADHD based on these visual traits with over 90 percent accuracy. The same approach was also able to distinguish whether a person with ADHD regularly takes stimulant medication. These findings indicate that ADHD may involve a consistent, underlying difference in how the brain handles brief moments of perception.
ADHD is a neurodevelopmental condition characterized by symptoms such as inattention, impulsivity, and hyperactivity. It affects around 3 to 4 percent of Canadian adults and about 2.6 percent of adults worldwide. While researchers have known for some time that ADHD influences attention, memory, and executive functioning, less is known about how it affects the brain’s handling of incoming sensory information—especially how that processing changes from one moment to the next.
Previous studies using brainwave recordings have shown that people with ADHD often display different patterns of electrical activity, particularly in the alpha and theta frequency bands. However, these findings have not always been consistent. To better understand the functional impacts of such oscillations, the researchers behind this new study used a method called random temporal sampling, which allows them to track how efficiently someone processes visual information across tiny slices of time.
Their goal was to explore whether ADHD is associated with a distinct rhythm or timing in visual perception, which could reflect underlying neural oscillations. If such a rhythm exists and is consistent across individuals with ADHD, it could provide a new behavioral marker for identifying the condition.
“In light of the relatively high incidence of ADHD, there is surprisingly little that we know about it for sure,” said study author Martin Arguin, a professor of psychology at the University of Montreal and director of the Neurocognitive Vision Lab.
“This is especially true of the neural bases of the disorder. My lab had recently brought to maturity the technique of random temporal sampling, which serves to capture temporal variations in perceptual efficiency. Given that these temporal variations can be assumed originate from oscillatory neural activity, we thought that examining ADHD from this perspective might bring a positive contribution to our knowledge of the disorder.”
The researchers recruited young adult participants from two colleges in Quebec. A total of 49 people were included in the final analysis: 26 were neurotypical controls and 23 had received a formal ADHD diagnosis. Among those with ADHD, 17 regularly used stimulant medication while six did not. All participants had normal or corrected vision and completed a battery of visual tasks under controlled conditions.
The central experiment involved briefly showing participants a series of five-letter French words overlaid with visual noise. The visibility of each word varied rapidly during the 200 milliseconds it appeared on the screen. Participants were asked to read the word aloud, and researchers recorded whether they were correct. The contrast of the noise was continuously adjusted so that each participant would average about 50 percent correct responses, ensuring a balanced challenge across individuals.
What set this experiment apart was how the noise varied. It wasn’t static; instead, the signal-to-noise ratio changed throughout the 200-millisecond window in a pattern composed of multiple sine waves at different frequencies. This random fluctuation allowed researchers to analyze how well participants could extract the word at each moment in time, depending on the frequency of the noise changes. The technique generated what are called classification images—maps showing how efficiently each person processed the visual input at different times and frequencies.
By comparing the classification images of those with and without ADHD, the researchers could examine whether there were consistent group differences in the rhythm of visual processing. They then used a machine learning model trained on features extracted from these classification images to see if it could distinguish between the two groups.
The classification images revealed consistent differences in visual processing patterns between participants with ADHD and those without. Although the general structure of visual processing looked similar across both groups, certain frequencies showed marked differences. These included processing oscillations at 5, 10, and 15 cycles per second (Hz), particularly when the noise in the stimulus oscillated at 30 to 40 Hz.
“The immediate implication of our results is that, in ADHD, there seems to be a systematic divergence in visual function from that of individuals with typical development,” Arguin told PsyPost. “This divergence points to a difference in brain function that has yet to be clearly determined.”
When these patterns were fed into a machine learning algorithm, it was able to classify individuals as either ADHD or neurotypical with 91.8 percent accuracy, using only about 3 percent of the total features. The sensitivity (the algorithm’s ability to correctly identify ADHD participants) was over 96 percent, while its specificity (the ability to correctly identify neurotypical individuals) was 87 percent.
“The literature largely emphasizes the individual differences among persons with ADHD as a potential indicator of varied causes for the disorder,” Arguin said. “Our findings rather indicate that we can actually classify 100% of our participants into their respective group (i.e. ADHD vs typical development) from their individual data patterns pertaining to perceptual oscillations; thereby pointing to a possibly unique cause.”
“This high classification rate in itself was also rather unexpected. However, at the time we carried out the data analyses, we had already completed a similarly designed project comparing perceptual oscillations across young vs elderly participants with normal cognitive functioning (Lévesque & Arguin, 2024). In that study too, we found an accuracy of 100% in the classification of individual participants into their respective group based on their patterns of perceptual oscillations. This somewhat contributed to raise our expectations for the outcomes of the ADHD study.”
The researchers also examined whether the algorithm could tell apart individuals with ADHD who were on stimulant medication from those who were not. Despite the small sample size, the model performed well, achieving 91.3 percent accuracy. It was particularly sensitive to identifying those taking medication, with 100 percent accuracy in that subgroup. Only a few features were needed to make this distinction, suggesting that regular medication use has a measurable effect on the timing of visual processing.
Interestingly, when the researchers directly compared the classification images of medicated and unmedicated participants with ADHD, they did not find statistically significant differences using traditional statistical methods. However, the machine learning approach was able to detect consistent patterns, indicating that the medication status leaves subtle but consistent traces in the visual processing rhythms.
In summary, the findings indicate “that there is a particular pattern to perceptual oscillations that seems to uniquely characterize most, if not all, persons with ADHD,” Arguin explained. “This fact suggests that there may be a single atypical brain mechanism underlying ADHD. This is at variance from a large portion of the relevant literature, which seems to assume a variety of potential causes in attempting to account for sometimes significant individual differences among ADHD sufferers.”
Although the findings are promising, the study has some limitations. The sample size was modest, especially for the comparison between medicated and non-medicated ADHD participants. Larger and more diverse samples are needed to confirm the generalizability of the results. The participants were mostly young adults, so it is not yet clear whether similar patterns would be found in older adults or children.
Another issue is that while the classification method was very effective, the specific link between the observed visual processing patterns and underlying neural mechanisms remains unclear. The classification images are thought to reflect brain oscillations, but this assumption has not yet been directly tested using brain imaging techniques. Future studies could combine this behavioral method with electroencephalography or functional MRI to identify the brain regions and networks involved.
The researchers also point out that the task they used—a repetitive word recognition task with limited cognitive flexibility—may have made it especially sensitive to the effects of ADHD and medication. It remains to be seen whether similar processing differences would appear in more complex or varied tasks.
Despite these caveats, the findings suggest that random temporal sampling may offer a powerful and objective tool for identifying ADHD. The ability to classify individuals with such high accuracy, based on a brief visual task, suggests that there may be a single core difference in perceptual timing that characterizes the condition. This idea stands in contrast to much of the current literature, which tends to emphasize individual differences and multiple potential causes for ADHD.
If the technique proves reliable in broader populations and under different testing conditions, it could eventually be adapted into a clinical tool.
“Based on our investigation in adults with ADHD, we are now pursuing a related study to examine whether we can replicate its findings in children in the age range where an assessment for possible ADHD is most often sought (10-14 year olds),” Arguin explained. “If so, it would indicate that random temporal sampling could constitute an excellent test for the assessment of ADHD. Such a test could eventually prove very useful in the current context where: 1. access to the relevant specialists is often very difficult; 2. assessment for possible ADHD may be very costly; 3. the diagnosis of ADHD is based on symptomatology.”
The study, “Visual processing oscillates differently through time for adults with ADHD,” was authored by Pénélope Pelland-Goulet, Martin Arguin, Hélène Brisebois, and Nathalie Gosselin.
