Why a Child With ADHD Performs Unevenly: A Few Slow Trials

What the research found

The task is built to be dull. A child sits at a screen with two keys under their fingers and presses one of them each time a letter or a shape appears, several hundred times over. Every response is timed to the millisecond, and in children with ADHD the resulting record is unusually uneven. That unevenness is among the most durable findings in the field.

A 2013 synthesis by Michael Kofler and colleagues, drawing on 283 studies, 9,780 individuals with ADHD and 12,024 typically developing comparisons, found an overall difference in reaction time variability of g = 0.71, where g expresses the gap between two groups in standard deviations. Age moderated the effect, which reached 0.76 in children and adolescents and 0.46 in adults.

What produces the unevenness was the question taken up in a 2014 review in the Journal of Child Psychology and Psychiatry. Sarah Karalunas and Joel Nigg, at Oregon Health and Science University, worked with colleagues in Amsterdam, Aachen, Juelich and Dresden to run two new meta-analyses on studies that had taken response times apart using a mathematical model of decision making. The model treats a simple two-choice decision as evidence gathering. Information accumulates until it crosses a threshold, and then the key is pressed.

Three quantities fall out of that account. Drift rate is the speed at which evidence accumulates toward the answer. Boundary separation is how much evidence a person requires before committing, which amounts to how far they favour caution over speed. Nondecision time covers everything outside the decision itself, perception at one end and the movement of the finger at the other.

Across five studies containing six independent samples, all of children and adolescents, drift rate was slower in the ADHD groups by g = 0.63, an effect the authors describe as moderate to large. Boundary separation showed no group difference whatever, g = 0.01. Nondecision time was slightly faster in the ADHD groups, g = -0.32, a small effect. The children were requiring the same weight of evidence as their peers before answering, and gathering it more slowly.

The slowness concentrates in the tail of the distribution, in a small number of responses that arrive very late while the rest fall in ordinary range. Calling those late responses attention lapses names an event without identifying a mechanism, and the statistical parameter that captures the tail has been read as reflecting decision processes and motor processes both. Decomposing the distribution did not sharpen the picture either, since drift rate distinguished the groups no better than the plain standard deviation of response times, g = 0.65.

What it means for you and your child

At home the same pattern appears without the millisecond record. The homework problem solved in a minute on Tuesday and abandoned on Thursday, the instruction followed at once and then the identical instruction met with nothing. Averaged across a week, the performance reads as a child who is capable and intermittently unwilling. The average is the wrong statistic.

The boundary separation result speaks directly to that reading. A child answering carelessly would show a lowered threshold, committing on thinner evidence, and a child stalling for time would show a raised one. Neither appeared in these samples. The children were holding themselves to the same standard of certainty as everyone else in the room, and on a minority of trials the evidence arrived too slowly to reach that standard.

Karalunas and her colleagues connect slow drift to difficulty separating a signal from the noise around it, and to low alertness and arousal. The authors call their conclusions preliminary, note substantial unexplained variation between studies, and observe that this class of model applies only to forced two-choice tasks. Within that narrow frame, both descriptions place the difficulty before any choice the child makes.

These are decision measures taken from a laboratory screen, and the honest account stops at the edge of the school day. They show a slow accumulation of evidence on some proportion of trials in groups of children with ADHD, which is a statement about groups and cannot be turned into a statement about one child. The inference a parent may reasonably draw, and it is an inference, is that the same slow accumulation operates in the moments when their own child stalls.

What follows from that is patience of a specific kind. The delay these measures locate sits before any choice the child makes, which makes it a poor target for urgency. Most responses in these tasks fell in ordinary range and a minority did not, and that is a pattern which repeats within a single sitting and will go on repeating at the kitchen table. Your patience through the slow trials is the response that matches the mechanism.