In The Naturalistic Case for Free Will: The Challenge, Christian List opens a series of posts at The Brains Blog discussing some key ideas from his book ‘Why Free Will is Real’ (Harvard University Press, 2019).
But if Free Will is defined, even just on a “first gloss”, as “an agent’s capacity to choose and control his or her (or its?) own actions”, then surely it does exist – even for an agent as simple as a programmed light switch (which chooses whether to switch on or off depending on what it perceives by way of motion and/or warm objects in its environment). So the “Free Will” that is denied by some must certainly be more than this (though rather than deny its existence, I would rather say that I have never seen a coherent definition of it).
What neither the light switch nor the human agent possesses is the capacity to change its own programming for past decisions. A human, or even a suitably programmed learning agent, can of course modify the part of its “program” that will make future decisions, and even may modify its own learning algorithm; but neither he, she nor it has the capacity to modify the program with which it started.
What is relevant for ethical and legal discussions though is not some mythical mystical property that is possessed uniquely by humans, but rather the property of responsibility – by which I think we should mean the capacity to modify ones future behaviour on the basis of responses (praise, censure, punishment, reward, etc.) that one receives, or perceives as being received by others as reactions to past decisions.