Attention Drifts Most During Long Waits: Evidence From ADHD

What the research found

Finding where in an hour a mind drifts means interrupting the work while it runs. At roughly one-minute intervals the task stopped and put a question on the screen: “Where was your attention just before this probe?” Two answers were available, “On task” and “Off task.” Thirty of these interruptions fell across each of two half-hour tasks, and the proportion answered off task became the measure of mind wandering.

Natali Bozhilova and colleagues at King’s College London ran that procedure with 27 adults diagnosed with ADHD and 29 without, aged 18 to 65 and matched for age, sex and IQ. The work appeared in the Journal of Attention Disorders in 2021. Every participant was an adult.

Across both tasks the adults with ADHD answered off task on 50 to 70 percent of probes, and the comparison group on 10 to 20 percent, with standardised differences of 1.5 to 2.5 where 0.8 is conventionally counted as a large effect.

The study was built around context regulation, which means adjusting the rate of wandering to what the task requires, so that the mind drifts less as the work grows harder. The two tasks raised demand in different ways.

The first varied load on memory. Its easy condition asked participants to match a shape on the screen; its hard condition required them to hold the previous trial’s shapes in mind and answer about those. When that memory load arrived, wandering in the ADHD group fell from 66 percent of probes to 45 percent, a reliable drop. The comparison group moved from 16 percent to 10 percent, short of significance, which the authors attribute to a floor: their rate already ran far below the 50 to 55 percent recorded for unscreened samples on the same task.

The second task asked only for speed. A counter of milliseconds appeared and participants pressed a button as fast as possible, usually after a predictable one-second gap and sometimes after unpredictable delays of two, five or eight seconds. The longer the wait, the more sustained attention it required. In the ADHD group, wandering rose from 53 percent of probes after a two-second delay to 69 percent after five seconds and 68 percent after eight. The comparison group registered 13, 13 and 18 percent, with no significant movement across the delays.

Adjustment to demand therefore held in one task and failed in the other, in the same people on the same afternoon. A memory load pulled the ADHD group back toward the work, and an unfilled interval let them go. The selectivity is what the paper adds, and it is a finer result than the overall rate.

What this means for you and your child

The gap is the demanding part. An eight-second wait in a laboratory compresses something that recurs all day: the pause after one question is answered and before the next is asked, the minute while a class settles. Nothing is asked during those intervals except that attention be held ready, and holding it ready is the operation this study found hardest.

Several things restrain how far the finding can be carried. Fifty-six people is a small sample, and the authors note that it could detect only medium to large effects. Fourteen of the 27 ADHD participants took medication, withheld for 48 hours before testing. Once the researchers controlled for working memory capacity, the group difference across the delays weakened to the edge of their correction threshold and no longer survived it. Applying any of this to a child is an inference from adults, and the authors do not make it.

The method rests entirely on what a person says about their own attention one second after being interrupted, and no external check on that answer exists. What argues for trusting it is how closely the probe scores tracked measures collected separately, with standardised coefficients between 0.62 and 0.77 against interviewed ADHD symptom counts, executive function ratings and functional impairment in daily life.

The same variable accounted for the performance results. Adults with ADHD were slower, more variable in their response times and more likely to miss a stimulus altogether, and once wandering frequency was entered as a covariate, those group differences were no longer statistically significant. The data come from a single sitting and establish association rather than cause, though the association runs through everything the tasks measured.

The eight-second delay, which looks like the easiest thing in the session, was the condition under which this group came apart, while the condition that looks demanding was the one they met. A child who fades in the quiet stretch and returns when the work resumes is following the pattern these adults produced under measurement.

The research offers no method for shortening those intervals and makes no promise about any child, but it supplies a reason to treat the drift in the gap as the expensive part of the hour, and to allow the interval its minute before calling attention back. Even at their highest rate of wandering, close to a third of probes found these adults still on task.