Across the CPT validation literature, intra-individual reaction time variability (RTV) — typically operationalized as the within-subject standard deviation of RT, or coefficient of variation (SD/mean) — emerges as one of the most consistent discriminators between ADHD and non-ADHD groups. In many studies, RTV outperforms mean RT, omission rate, or commission rate as a single-metric classifier.

What RTV measures

Mean RT captures average processing speed. RTV captures moment-to-moment fluctuation — the difference between the patient's fastest and slowest responses to identical stimuli within the same task. A patient with mean RT of 420 ms and SD of 60 ms is responding consistently. A patient with the same mean and SD of 140 ms is having dramatic attentional fluctuations — fast responses on some trials, slow lapses on others.

Why ADHD elevates RTV

The leading theoretical accounts converge on attentional fluctuation:

  • Default-mode network intrusion. Functional neuroimaging studies show that DMN activity (the resting-state, mind-wandering network) intrudes during task performance more frequently in ADHD. These intrusions produce slow trials interspersed with normal-speed trials.
  • Tonic-phasic alertness fluctuations. Sustaining task-relevant alertness requires continuous noradrenergic input. ADHD-related differences in noradrenergic function may produce uneven alertness across a task.
  • Effortful control gaps. When the task is monotonous or under-stimulating, top-down control of attention becomes effortful; lapses produce slow trials.

Practical implications

For initial assessment

RTV is a useful indicator of attentional fluctuation that can persist even when mean performance looks normal. A patient with reasonable accuracy and average mean RT but markedly elevated RTV is showing exactly the lapsing pattern characteristic of ADHD. This is particularly relevant in higher-functioning adults whose intelligence or strategy keeps mean performance in the typical range.

For titration

Stimulants typically reduce RTV more reliably than they affect mean RT. Some clinicians use pre-/post-medication RTV change as a within-subject index of response: e.g., a 30% reduction in RT-SD on medication suggests good neurochemical engagement, independent of how the patient feels subjectively that day. Non-stimulants also reduce RTV but typically at smaller magnitude and over a longer time course.

For longitudinal tracking

RTV is a stable trait at the population level but sensitive to acute state factors (sleep, caffeine, fatigue). Within-subject changes across visits — when administered at consistent time-of-day under similar conditions — provide a quantitative outcome measure that complements rating scales.

Operationalization differences across tools

Different CPTs report variability differently:

  • Standard deviation of RT (RT-SD) — the simplest. Sensitive to outlier slow trials.
  • Coefficient of variation (CV = SD/mean) — normalizes for processing speed differences. Useful for cross-population comparisons.
  • Ex-Gaussian decomposition. Some research-grade analyses fit RT distributions to mu (Gaussian center), sigma (Gaussian SD), and tau (exponential tail). Tau — the long-tail parameter — captures the "lapsing" component specifically and shows the strongest ADHD effects in some studies.
  • Block-by-block change. Some tools report RT or RT-SD change across task blocks (e.g., first third vs. last third). Greater within-task increase in variability suggests vigilance decrement.

Confounders to rule out

  • Sleep deprivation. Acute sleep loss elevates RTV substantially and mimics ADHD-pattern fluctuation.
  • Sub-effortful engagement. RTV elevation can reflect lack of motivation or effort. Look at consistency of accuracy alongside.
  • Anxiety states. Acute anxiety can produce performance fluctuation through attentional capture by worry.
  • Substance influence. Caffeine, sedatives, and alcohol all affect RTV in different directions.
  • Fatigue / time-on-task. RTV often grows over the course of a CPT in everyone; ADHD groups grow faster.

When RTV is most clinically useful

  1. Adults presenting with attention complaints but mean performance in typical range. Elevated RTV with normal accuracy is a common ADHD pattern in higher-functioning patients.
  2. Treatment response evaluation. Pre-/post-medication RTV change is a within-subject quantitative measure that doesn't depend on patient self-report.
  3. Differentiating ADHD from primarily slow processing speed. Slow processing speed elevates mean RT but typically doesn't elevate RTV proportionally.
  4. Establishing a baseline before structural changes. Even if you don't intend to use CPT routinely, a baseline RTV at intake gives you an objective comparator if treatment response becomes ambiguous later.

RTV is reported by every major CPT tool — comparison of how each operationalizes it is in the comparison table.