Training Arena · beta
Choose one task. Practise deliberately.
Start with Flick practice, or choose a goal below. Five existing visual drills, separate task metrics — not one universal brain or gaming skill score.
Choose your goal
Each link opens a setup, not an active trial or saved result. Fixed benchmarks can be provisionally ranked; optional Ghost is adaptive, unranked practice.
Target acquisition · Flick Grid
Move the pointer to each target and click. A hit or miss advances.
24 targets · finishes by attempts, not a fixed time
Read controls and set up →Continuous tracking · Lock-On
Keep the pointer inside the moving target. No clicking required.
20 seconds of tracking
Read controls and set up →Choice response · Snap Decision
Click the lit zone among 2, 4 or 8 choices.
40 trials · 1.5-second response window each
Read controls and set up →Go / no-go discrimination · Bomb Squad
Click the green sphere (go). Do not click the red cube (no-go).
40 trials · 1.5-second response window each
Read controls and set up →Conflict response · Focus
Press ← or → for the CENTER arrow, ignoring the surrounding arrows. Left/right pads are an alternative.
40 trials · 1.5-second response window each
Read controls and set up →Recommended first · target acquisition
Beginner Flick practice
Move the pointer to each target and click. Hits and misses advance. Mouse or trackpad recommended; pointing is required.
Two sets of 24 targets, with a six-second automatic breathing break before set two. Finishes by attempts, not a promised session time. Optional hit-feedback sound starts off.
Adaptive Flow practice · not ranked. Anonymous raw player telemetry is submitted on completion for server validation; this is not local-only. No Ghost opponent. The resumed action's timer restarts after an already-shown stimulus, so this is not an unpaused fixed benchmark.
Current level …. The next level adjusts target size from hit rate; different levels are not like-for-like movement-time comparisons. Feedback is same-setup relative, not a universal ability grade.
Recent Arena sessions · optional
History belongs to this browser and website address. Records saved on another domain are not copied here.
Keep up to 30 summaries in this browser only. No account, sync, raw cursor paths or personal identifier. Shared-browser users can see these records.
History does not change the run's separately disclosed server upload. Same browser does not prove the same hardware or input. Protocol version and full-session setup are not recorded, so there are no personal bests, trends or performance comparisons.
Loading local history…
Compare like with like
Flick throughput uses nominal target width, not click-spread effective width. Lock-On reports on-target sample proportion, not time-weighted duration. Decision speed belongs beside correctness. Game points are mode-specific feedback; none of these metrics is bare reaction time.
The engine renders seed-derived scenes; native browser input timestamps and pointer paths are scored locally and submitted for server validation on completion, except explicitly local-only practice. Server outcome is shown separately. Browser/display/input delays are not physical accuracy guarantees, and not every metric has a confidence interval.
Ghost starts from population/anchor priors and adapts within a run. It is not a persisted prior-self replay. All five tasks are visual; optional Flow hit sounds are feedback, not an auditory drill.
Arena board
Loading this board.
Only accepted runs recorded as human, non-adaptive and level 0 are included. Runs are re-scored server-side and screened for implausible paths and timing; this does not verify a player's identity. Ranks are provisional, not a normalized comparison across input setups or protocol versions. Rows and population are separate reads, not one atomic snapshot; ranks and percentiles can change. See how we measure.