"Impairment" is not one dial that turns down. It's a set of separable abilities, and each cause of impairment dims a different subset. This map is the frame every later lesson builds on.
When people say someone is "impaired," it sounds like a single global state — as if the whole brain dropped a few percent. It isn't. Cognition is made of separable domains: distinct capacities that can be knocked out independently of one another.[1] A drunk person and a sleep-deprived person are both "impaired," but their profiles differ — different domains, different amounts, different time courses. We'll call that pattern an impairment fingerprint.
This is the single most important reframe for your product: a task doesn't detect "impairment" in the abstract — it detects degradation in one specific domain. So your battery is only as good as the match between the domains your tasks probe and the domains your target impairments actually hit.
There are formal taxonomies — the DSM-5 lists six neurocognitive domains;[1] the NIH Toolbox and CANTAB batteries operationalize overlapping sets.[2] Here is the working subset that matters most for impairment detection:
| Domain | What it is | Signature task | Why you'd care |
|---|---|---|---|
| vigilance | Sustained attention — staying responsive over time | PVT (reaction time to random cues) | First thing fatigue erodes; the most validated signal |
| divided attn | Doing two things at once | Dual-task / tracking + cue | First thing low-dose alcohol erodes |
| speed | Processing / psychomotor speed | Digit-Symbol (DSST), choice RT | Sensitive to almost everything — great screen, poor at naming the cause |
| working mem | Holding & manipulating info for seconds | n-back | Hit by alcohol, cannabis, benzodiazepines |
| inhibition | Suppressing an automatic response (executive control) | Go/No-Go, Stroop | Alcohol degrades it sharply → impulsive errors |
| time | Sense of elapsed time | Time-estimation | Distorted distinctively by cannabis — near-signature |
"Executive function" is not a single domain you can cover with one clever task. The field-standard model (Miyake) splits it into three separable pieces — inhibition, updating, and shifting.[4] If your assessment claims to test "executive function" with a single game, it's measuring one slice and missing two. Precision about domains protects you from over-claiming.
Retrieval, not recognition — try to answer before you read the options.
Sleep deprivation degrades which domain first?
Why is the PVT such a strong impairment signal?
A single task that probes only processing speed is a weak diagnostic because it...
You can now say: impairment is degradation across separable domains; each task is a bet on one domain; and a battery's power comes from covering the domains your target impairments actually hit. Hold that frame — Lesson 2 fills in the fingerprints (alcohol vs. cannabis vs. fatigue), and Lesson 3 turns to the measurement science that decides whether your detector can be trusted.