Your Doing
Press a key, hear a tone a quarter of a second later, and report when each one happened. Your answers will be wrong in two opposite directions, and the shape of the wrongness is the feeling that you did it.
A dot goes around a dial once every 2.56 seconds. At some point something happens: you press a key, or a tone sounds. Afterwards the dot stops and you say where it was at that moment. That is the whole task, and you will do it twenty two times.
Sometimes your press causes nothing. Sometimes it causes a tone 250 ms later. Sometimes the tone arrives on its own with nothing of yours behind it. The physical intervals never change. What changes is whether the thing was your doing.
When it is yours, two errors show up, and they point at each other. You report your own press as later than it was. You report the tone as earlier than it was. The quarter second closes up. This is intentional binding, published by Patrick Haggard and colleagues in 2002, and it is the closest thing anyone has to a measurement of the sense of agency.
Everything else in this lab measures a sense pointed at the world. This is the first one pointed at you: not what happened, not when, but whether you were the author.
- 0️⃣ One practice round so the dial and the report stop being strange. Nothing is scored.
- 1️⃣ Twenty two trials in five kinds, shuffled: your press alone, the tone alone, your press causing a tone (reporting the press), the same trial reporting the tone, and a tone caused by the dial instead of by you. Plus two check trials.
- 2️⃣ Two numbers: how far your press moved and how far the tone moved, each against its own baseline, each with an exact permutation test.
- 🔊 Sound required. Headphones are better than speakers, and a wired connection is much better than Bluetooth.
- 👁️ Watch the dot the whole time. Do not decide in advance when to press, and do not count. Both wreck the measurement in ways the page cannot detect.
- Nothing here is scored as an absolute. Everybody misreads a moving dot, by a lot, in their own direction. Every number on the result page is the difference between two conditions that share the dial, the speed, the report, the hand and the screen, so a constant bias of any size cancels exactly.
- The conditions are interleaved trial by trial, not run in blocks. Getting tired, getting bored and getting better all land evenly on baseline and operant trials, so none of them can invent a shift.
- One condition is a tone caused by the dial rather than by you. Michael Buehner argued in 2012 that this binding is about causation and not agency at all. If your flash trials shift as hard as your press trials, you have replicated Buehner instead of Haggard, and this page will say so rather than hide it.
- Two check trials where the dot itself turns red. Ground truth is visible on the screen. Miss those by more than 200 ms and the run gets printed instead of scored.