Sustained Attention Test Online

A hit is a response to a target. A miss is a target with no response. A false alarm is a response to a non-target. A correct rejection is a non-target that receives no response. Hit rate divides hits by targets; false-alarm rate divides false alarms by non-targets.

Speed should not be interpreted alone. Responding to every digit would produce quick responses but many false alarms. A useful reading keeps target detection and response restraint together, then uses median response time only for correctly detected targets.

Response time is measured with performance.now() from the appearance of the active digit. The primary timing result is the median of correct target responses, which is less sensitive to one unusually slow response than a mean. Responses under 150 milliseconds remain recorded and receive an anticipatory-response flag.

RT variability uses median absolute deviation, or MAD. First find the median correct response time, calculate each hit's absolute distance from that median, and take the median of those distances. A smaller MAD means correct responses were more tightly grouped in this session; it is not a general stability trait.

The trial stream is split near its midpoint. Hit rate and false-alarm rate are calculated separately for the first and second halves. Time-on-task hit-rate change is second-half hit rate minus first-half hit rate, expressed in percentage points.

This comparison is descriptive, not diagnostic. A difference can reflect the random location of targets, a brief distraction, learning the rhythm, fatigue, or chance—especially in a short one-minute practice. Repeat matching sessions before treating one direction as a pattern.

The response handler ignores repeated keydown events, so holding Space cannot create multiple responses. Each digit accepts at most one response. If the page becomes hidden, the task pauses and adds a focus-interruption flag; resuming restarts timing for the visible stimulus.

A restarted session is not stored as a normal completion. Completed sessions preserve anticipatory and focus-interruption flags so the history does not hide conditions that may reduce comparability.

A lower target rate creates longer periods without a required response. A longer session increases monitoring time. Those changes affect both opportunity counts and rhythm, so a 60-second 25% session should not be compared directly with a five-minute 15% session.

Keyboard, touch input, display refresh, browser scheduling, power-saving mode, and background activity can change response timing. Keep the same device and input method for personal comparison and pay more attention to broad patterns across several sessions than to a small millisecond difference.

This task is not an ADHD test, diagnostic interview, standardized continuous performance test, age norm, or clinical cutoff. It uses an original digit stream and page-specific scoring. A result cannot establish whether attention in school, work, driving, or daily life is normal or impaired.

Use it to observe how this particular rule works and to compare your own matching sessions. If attention concerns affect daily life, seek an appropriate qualified professional rather than interpreting this browser score as medical evidence.

The completed record includes settings, aggregate counts, response-time summaries, timestamps, and validity flags. It does not store the keys pressed outside the task, personal identity, or a device fingerprint. Clear the current tool's history at any time.

Tool results remain in localStorage. Ordinary site analytics described in the privacy policy are separate and do not receive this task's per-stimulus response list.