Reaction Time Test with Camera

Traditional online reaction tests display a colored box or circle, and you click or tap as fast as possible when it appears. The measurement is: time from stimulus appearance to click event. This sounds straightforward, but the method has significant biases.

Click tests measure the total time for you to see the stimulus, decide to respond, move your finger to the button, and press. Experienced mouse users often "pre-hover" their cursor over the button, reducing travel time to near zero. Touchscreen users develop micro-movement habits that shave time. The result is scores that reflect familiarity with the specific test format as much as genuine reaction speed.

Worse, anxious participants can achieve sub-100 ms scores by pressing slightly before the stimulus appears — a form of anticipatory cheating that the test cannot reliably distinguish from genuine fast reactions.

Our hand tracking system uses a neural network trained on millions of hand pose images to identify the position and shape of your hand in real time at 30 frames per second. The AI continuously monitors whether you are displaying the 6-7 gesture, and logs the precise timestamp when the gesture is first detected after the stimulus appears.

Because the gesture requires physically moving your hand and forming a specific shape, there is no equivalent of "pre-hovering." You cannot subtly pre-position your hand into the gesture without the AI registering it. Anticipatory false-starts are easier to detect and discard.

The result is a measurement that more closely reflects the neuromuscular chain involved in real physical reactions: catching a ball, responding to a traffic event, defending in a sport, or executing a quick movement in a performance context.

Camera hand-tracking scores are typically 25–50 ms higher than click test scores from the same person, because physically moving your hand and forming a gesture involves more motor steps than pressing a pre-positioned finger.

This does not mean the camera test gives you a "worse" score — it gives you a more honest score. Think of it like the difference between measuring your running speed on a track versus riding a fast bicycle. The track number is lower, but it more accurately reflects your own physical capability.

When comparing your hand-tracking score over time, the improvement trend is what matters most. A drop from 270 ms to 230 ms on our test represents the same proportional neural improvement regardless of the absolute numbers.

Camera access can feel invasive, so it is worth being explicit: our hand-tracking system processes all video data entirely within your browser using WebAssembly. No video frames, images, or streams are ever transmitted to our servers or stored anywhere. You can verify this by monitoring network traffic in your browser developer tools — no video data leaves your device.

For optimal tracking, ensure good frontal lighting (avoid strong backlighting from a window behind you), keep your hands within the camera frame, and use a plain background if possible. The AI performs best when your hand contrasts clearly with the background.