The Simulation of Free Will

Kari Gwen Coleman

In this paper, I argue that a deterministic simulation of Robert Kane's libertarian free will would be adequate to ground what might be called a 'practically' free will (as opposed to an 'absolutely' free will), one which is good enough to enable the design and construction of morally responsible computers. I begin (section 2) by appealing to prototype theories of categorization to argue that a perfect instantiation of incompatibilist free will may be unnecessary for morally responsible computers. After summarizing Robert Kane's libertarian account of free will (section 3), I argue (section 4) that the notion of screening off can replace inadequate appeals to 'could have done otherwise' in defending the success of Kane's account. Although this screening off is best and most clearly exemplified by the indeterminacy in Kane's account, I argue that this indeterminacy can be adequately simulated with deterministic randomness, thus reconciling the deterministic nature of computers with the free will they would need to be morally responsible.