I just wanted to point out that the phrase "free will" comes up in discussions of Bell's theorem (which is a proof that quantum mechanics implies correlations between distant measurements that cannot be realized using classical, non-quantum, means without violating Special Relativity). There, it isn't really a philosophical condition, and doesn't necessarily have to do with conscious human beings at all. Instead, it's a kind of independence assumption.
In an EPR type experiment, you have two distant experimenters, Alice and Bob. Each of them has a device that has a number of possible settings. A pair of particles is created at some location between Alice and Bob, and one particle is sent to each of the experimenters, who then perform a measurement using the device with the chosen device settings. The free will assumption amounts to something along the lines of:
For any property \lambda of the twin pairs at the time of their creation, and for any pair of settings \alpha (Alice's setting) and \beta (Bob's setting), there is a possible run of the experiment in which the twin pairs have value \lambda and Alice's setting is \alpha and Bob's setting is \beta.
So it doesn't actually involve Alice or Bob's "free will". What this assumption rules out is the possibility that the choice of \lambda constrains the possible values of \alpha or \beta. The free will assumption seems pretty innocuous in this limited form. If Alice and Bob use some kind of random number generator to decide what setting to use, then that would satisfy the "free will" condition without saying anything about the nature of human consciousness--a robot can use a random number generator as well as a human can. If the free will assumption is false, then it implies that, in a sense, there are no random number generators, that any two supposedly random number generators produce results that are correlated in a specific way.
