Methodology

CognitiveBench measures browser-based reaction time from a visible color change to your click, tap, or keyboard response.

What we measure

When a round becomes ready, the page records a high-resolution timestamp with performance.now(). When you respond, another timestamp is recorded and the difference is stored as an integer number of milliseconds. Early responses during the waiting phase are treated as false starts and are not included in the five valid rounds.

Factors that influence results

Browser timers are useful for personal comparison, but they are not laboratory equipment. Hardware differences, display refresh rate, input device latency, system load, browser choice, and background processes can all change measured times.

• Display refresh rate affects when a color change becomes visible.
• Mice, trackpads, and touchscreens have different input latency.
• Busy systems and browser extensions can add delay.
• Different browsers may schedule rendering and input events differently.

Comparing across devices

Scores from different phones, tablets, or computers are not directly comparable. Prefer tracking your own history on the same device and browser when possible.

How estimated percentiles work

Reaction Time, Stroop, and Memory Match convert completed results into an estimated percentile using initial hardcoded simulated distributions. Reaction Time uses the five-round average, with lower milliseconds ranking higher. Stroop uses the number of correct answers in 30 seconds, with higher scores ranking higher. Memory Match combines the percentile for move count with the percentile for completion time, weighting both equally.

These virtual distributions provide consistent product functionality before enough real, consented benchmark data is available. The displayed “test takers” comparison is not calculated from live CognitiveBench visitors, user accounts, or a continuously updated leaderboard, and it is not a validated population norm. Device latency, task familiarity, and testing conditions can shift scores. The hardcoded reference tables can later be replaced with aggregated real distributions without changing the result components.

Practice effects

Repeating the same task can improve performance on that task through familiarity and anticipation. Improvement here should not be treated as evidence of general intelligence gains or clinical change.

For entertainment and self-assessment context, see the Disclaimer. To try the tool, open the Reaction Time Test.