It has become quite popular lately to view the notion of free will as a misconception to be ‘debunked.’ To be sure, if we really do not have free will, we should be prepared to face that fact. But is it really a fact? I will argue here that key arguments against robust free will are seriously overrated. These are:
(1) Physical theory implies a block world (i.e. all events exist, including future events).
(2) Physical law, including indeterministic quantum physics, is inconsistent with free will.
Concerning (1), this very widespread misconception has been refuted in the literature. See, e.g., Sorkin (2007) and Kastner (2012), Chapter 8.
Concerning (2), it seems clear that if the world were fully deterministic, then all our actions would be fully determined by prior causes, so in that case there would be no room for a robust form of free will. (Caveat: there is an approach called ‘compatibilism’ that argues that free will is compatible with determinism; I find this approach decidedly unconvincing, but it’s something readers can look into.) However, in most interpretations, quantum theory implies that the world is genuinely indeterministic: given well-defined conditions, it is impossible to predict with certainty what will follow from those conditions. Nevertheless, free will skeptics such as Ted Sider (2005) have argued that free will must violate even the statistical laws of quantum theory. His argument basically assumes that a free agent, considered as a quantum system, would make choices that would violate the quantum statistical laws applying to the outcomes of his actions.
There are two serious problems with this argument. First, as noted by Clarke (2010),
“probabilistic laws of nature also do not require, for any finite number of trials, any precise distribution of outcomes. The probabilities involved…are the chances that events of one type will cause, or will be followed by, events of another type…These probabilities, we may assume, determine single-case, objective probabilities, or propensities. Actual distributions can diverge from proportions matching these probabilities.”
Thus, a statistical law is not ‘violated’ unless very large numbers of precisely repeated experimental runs yield statistically significant deviations from expected mean values, where even ‘statistically significant’ can be a matter of context and degree. Highly unlikely strings of outcomes may occur, and yet a statistical law may still not be violated. The point here is that the demonstration of a real violation of a statistical law requires a very high hurdle of empirical evidence.
The second problem is in trying to apply the quantum statistical law – the Born Rule – to human agents, which are macroscopic biological systems. In order to predict empirically useful probabilities of outcomes with the Born Rule, one must have a clearly defined system and a clearly defined observable being measured on that system. A definition of a system must specify how many degrees of freedom (usually considered as ‘particles’) are in play, and exactly what the initial state of that system is. A definition of an observable must specify exactly what forces are acting on the system and what sort of ‘detection’ constitutes each outcome of the observable being measured. These requirements may be straightforwardly met for microscopic systems in the laboratory, but it is a highly nontrivial matter as to whether they may be met under conditions obtaining in the context of human behavior.
Sider essentially argues that a human agent governed by the Born Rule should be able to make choices that would observably deviate from the Born Rule. But this assumes that one could set up repeatable experiments in which the agent could be precisely defined as a ‘quantum system’ whose applicable observable was so tightly defined as to allow detection of such deviations. It is only if such deviations were in principle detectable that there could be a violation from the statistical laws of quantum mechanics, as observed in Clarke’s remark quoted above. However, there are very good reasons to think that this is not the case.
For one thing, as noted above, one has to be able to perform precisely repeatable experiments. Does exposing a given human agent to repeated opportunities to make a choice constitute a precisely repeatable experiment of this type? Why should we think so? The human agent is an open system, continually exposed to variable influences from his or her environment: air currents, radiant energy, etc; as well to internal fluctuations (number of blood cells in the brain, number of activated neurons, etc.). Assuming the brain is the most relevant bodily system concerning the choice, the state(s) and the number of relevant degrees of freedom in the brain are in continual flux. No matter how tightly one might attempt to control the agent’s environment, one is dealing with an enormously sensitive, complex and ill-defined system, from a quantum-mechanical perspective.
At the level of individual instances, the Born Rule gives only propensities for outcomes. A human agent might instantaneously be subject to those propensities; yet, given quantum indeterminism, could still have room to make a free choice– one that would not violate any statistical law. This is because another instance outwardly presenting the same choice to the agent is in fact highly unlikely to constitute an identical repetition of the relevant initial conditions: i.e., the agent is almost certainly not in exactly the same state that he or she was just prior to the previous choice. Therefore, the Born Rule propensities are likely not really the same as in the previous instance. Even if the experiment is repeated many times, a resulting set of outcomes in which so many parameters are ill-defined and subject to change cannot be used to determine whether a statistical law is being violated.
Thus, it is a highly nontrivial matter to try to apply the Born Rule to macroscopic biological systems; yet claim (2) presumes without argument that one can straightforwardly do so. If this is not in principle possible due to the intrinsically ill-defined and/or ever-changing nature of the macroscopic physical system constituting the choosing agent, then there is no necessary violation of the Born Rule. This is so even if the agent’s choices are governed by the Born Rule, in terms of propensities, for each individual instance.
The bottom line: rather than see quantum theory as falling under yet under ‘physical law’ that is supposedly violated by free will, we can view quantum theory as being precisely the kind of physical law that allows for free will.
Clarke, R. (2010). “Are we free to obey the laws?”, American Philosophical Quarterly 47, pp. 389-401
Kastner, R. E. (2012). The Transactional Interpretation of Quantum Mechanics: The Reality of Possibility. Cambridge University Press.
Sider, T. (2005). “Free Will and Determinism,” in Riddles of Existence, by Earl Conee and Theodore Sider (Oxford: Clarendon Press), pp. 112-133.
Sorkin, R. D. (2007). “Relativity theory does not imply that the future already exists: a counterexample,” in Vesselin Petkov (editor), Relativity and the Dimensionality of the World. Springer. Preprint version: http://arxiv.org/abs/gr-qc/0703098
 The approach known as ‘compatibilism’ holds that free will is compatible with determinism. I will not address that here. However, I do think that compatibilism yields a very impoverished notion of free will.