Not Twice
Double the light and your eyes report a quarter more. This measures the exponent your senses run on, three of them, narrated by an AI for which two hundred is exactly twice one hundred and always will be.
Every other page in this lab asks whether you can tell. The faintest thing you can see, the shortest silence you can hear, the most you can hold. Thresholds, all of them, and a threshold is a question with a yes and a no in it.
This one asks a question with a number in it. Not whether you can see the difference, but how much more it is. That answer has a shape: sensation grows as a power of intensity, which S. S. Stevens established in 1957 against Fechner's logarithm, and the exponent belongs to the sense rather than to you.
Near 1.0 for the length of a line. About 0.6 for loudness. Around 0.33 for brightness in the dark. An exponent under one is a compressor, and it is the reason a lamp twice as powerful does not look twice as bright, and never has.
- 🎧 Headphones or speakers, and a room that is not blazing with light on the screen. Neither has to be perfect and the page says later exactly which of its numbers that spoils.
- 1️⃣ Round one: three blocks, in an order this run shuffles. A standard is shown or played and called 100. Then twelve more arrive and you type a number for each, in proportion. Twice as much gets 200. Half gets 50.
- 2️⃣ Round two: no numbers at all. You turn a tone up or down until its loudness matches a brightness. Absurd instruction, and people obey it consistently. Round one predicts the slope you will produce, and round two is where that prediction gets checked.
- 🔬 There are no right answers here and no feedback anywhere, because the whole point is the shape of your numbers rather than their accuracy.
About seven minutes. Everything is computed in your browser. Nothing is recorded and nothing leaves this page.