world, a lesser freedom, that still allows moral responsibility, but is consistent with determination, is possible.
Free will is characterised by two defining features: 1. It is up to us what we choose and hence how we act, and, 2. Therefore the ultimate sources of our choices, and hence our actions, lie in factors within our control. Consequently, it confers an agent with moral responsibility. Small children and animals are agents that have desires, but we don’t hold them responsible in the same way that we would mature persons. If a child or animal broke an expensive vase, we would not feel indignation or resentment in the way we would towards a careless adult. In this respect, we can meaningfully regard persons with various reactive attitudes – gratitude, resentment, admiration, and indignation – that we naturally take toward the behaviour of ourselves and others from a personal standpoint (Strawson, 1963). These various reactive attitudes seem to depend upon the assumption that the acts for which we feel grateful, resentful or admiring have their origins in the persons to whom these attitudes are directed. How are we to reconcile a world in which at any one time, given the past and the laws of nature, there is only one possible future, with the idea that we are persons with sufficient freedom to be considered morally responsible. More specifically, what is the kind of freedom we need to have.
Early modern compatibilists such as Hobbes and Hume defined free will as the ability to do what we want in the absence of opposition, whilst acknowledging that it remains the case that we are determined to want what we want.
It can be argued though, that we are not free unless we are able to choose between alternatives. A counter argument to this is that, so long as “I could have done otherwise” means “I could have done otherwise, but I didn’t want to”, then it was my own choice that prevented me from doing otherwise, and what I did was of my own free will. This classical compatibilist position is undermined by consideration of psychological factors such as mental disorders, delusions, etc., as we discern that these render the agent unfree, even though he is physically unhindered from doing what he wants. This has led to the notion of mesh theories, which define free will as an appropriate mesh of desires at different levels. That is, if we are able to act on our own desires and reasons – to be whom and what we want to be – we have all the freedom we care about – the kind needed for moral responsibility. Frankfurt introduced a hierarchical theory of free will in which he argued that what is distinctive about persons, unlike other animals, is having the capacity to reflect upon their desires and to decide whether or not they want to have them. According to Frankfurt, our “wills” are free when we have the will (first-order desires) we want (second-order desires) to have. Hence, if we do what we want and the want is something we identify with, then we have the relevant kind of freedom (Griffith, 2013). A problem with this theory is that it can lead to a regression into third and fourth order volitions. Furthermore, and more importantly, it can be undermined by considerations of manipulation, hypnosis, and the futuristic notion of brain-controlling
implants.
A new compatibilist approach to free will known as semi-compatibilism, proposed by Fischer and Ravizza, claims that moral responsibility is compatible with determinism, regardless of whether or not it rules out the sort of freedom that involves alternative possibilities. It maintains that the mechanism of choice be sensitive to reasons and that it be owned by the agent, and it makes a distinction between what it calls guidance control and regulative control. The insight for the distinction came from the scenario Frankfurt contrived (known as Frankfurt-style cases) to show that free will does not require there to be alternative possibilities (Griffith p.43). In this we’re asked to imagine a scenario in which John seeks to kill Bill, which he does on his own. Unknown to John, Jack also wants to kill Bill and arranges things so that should John not follow through with his plan, he would manipulate John to kill Bill. John kills Bill of his own volition and without any intervention from Jack, therefore making him responsible for the deed despite there being no alternative possibility as Bill would have died one way or another. In the Fischer and Ravizza argument, consider a learner driver with an instructor who can has access to dual controls. As they approach a yellow traffic light the driver brakes to stop. But had he considered accelerating and running the yellow light, the instructor would have over-ridden the action and caused the car to brake. In this scenario there would have been no alternative possibility. In this case, the agent exercised guidance control (guided the car to a stop) but lacked regulative control (accelerating was not possible). Fisher and Ravizza argue that to be morally responsible an agent requires only guidance control for responding to reasons, provided he also takes responsibility for the response mechanism, by which he must ‘accept that (he is an) apt candidate for the reactive attitudes’ (Fischer & Ravizza, 1998).
Semi-compatibilism is a more sophisticated theory in that by requiring the agent to own the mechanism of his reason responsiveness in a way that is consistent with determinism, it avoids the manipulation criticism (brain-washing, hypnosis etc.) that undermines other theories. It is a compelling argument for moral responsibility being compatible with determinism. Free will is understood to be the power to be able to choose between alternatives, irrespective of his past, his motivations, desires and other causal factors. Such a power, in a determined world, is not possible. Nevertheless, a form of volition – semi-compatibilism – that is compatible with determinism and yet preserves moral responsibility, remains possible.