You mean, stop and not press the button? In this case, the light never appears. The switch must be pressed eventually in the future so that the circuit sends the message back in time. If you find out a way to "trick" the system by never pressing the switch, it never sends the message in the first place.

The mechanism doesn't work by predicting whether you will press the button in 1 second. It works after you have actually pressed the button!

Then the light doesn't come on (at least until you start to entertain the idea of hitting the button after all...).

My point is that this device proves more than non-existence of free will. It requires inability to react to visual stimulus within 1 second timeframe.

I mean I can construct a robot which does the same - moves its arm towards the button and only stops when either the button is pressed or the light comes on.

You say you can construct such robot, but in fact, you can't. That's because if you could, then it would cause a paradox with a predictor. So you can't (though it may be fun watching you try).

Don't mess with time travel.

My point exactly - the existence of the Predictor proves more than lack of free will. It's lack of free will, so we are automatons that respond to stimuli in a deterministic way, sure.

But there are also some strong restrictions on the set of reactions that are allowed, for example none of them can lead to a functioning counter-predictor robot.

I spent way too long thinking about this and getting annoyed before remembering that Predictors aren't real.

They are real and this is the future of advertising.

> So what happens if I train myself to slowly move my hand towards the button and only stop when seeing the light?

Mu[1]. There is no such a device, so there is no way to train against its intended purpose.

[1] https://en.wiktionary.org/wiki/mu#Interjection