the free will debate – a broad outline of positions

I recently did some research on free will and so thought I’d outline the (very) broad, basic positions held in the debate. I will make little comment on the strengths and criticisms of each view in this post.

Much of the fuss over free will is because of the concept’s importance for our traditional understanding of our own influence in the universe, and, more commonly focussed upon, its apparent undergirding of our belief that people can (and should) be held morally accountable for their actions. It seems arbitrary to hold people morally accountable for the events they cause, and not dogs, or hurricanes, if people aren’t free, or responsible, in a more robust sense than those other things.

Although free will is an important concept, there are challenges to believing in its reality. One such challenge comes from understanding how free will relates to determinism.

Determinism is the belief that all events are casually determined by antecedent conditions, typically, prior events (which were themselves caused). If determinism is true it seems hard, at least on first glance, to see how agents could be self-determining, or free, or responsible in any robust sense. Therein lies the challenge for free will from determinism.

Although the above understanding of determinism implies an infinite regress, this is not usually considered an important observation in the free will debate. One could conceive of a deterministic reality as causally infinite, but one could also conceive of a deterministic universe that comes to exist non-deterministically (the rationality of believing such a view is not at issue here). What matters is the assertion that things are deterministic at least local to human affairs.

Note that determinism is not exclusively (or, arguably, even necessarily) tied to materialism. One could believe that there are more than physical entities, and still believe in determinism. Indeed, one of the reasons why the problem of determinism looms over the free will debate is because determinism seems like the consequence of taking metaphysical intuitions about causality seriously. Surely, we think, the laws of causality apply, no matter what kind of substances are involved.

It is tempting to think that the micro-world of quantum mechanics, with its inherent uncertainties, might shatter the scientific plausibility of determinism, but things aren’t that simple. First, quantum physics is still open to interpretation, and one might think that the uncertainties involved are purely epistemic, not actual. Secondly, even granting that there is real uncertainty at the quantum level, it is questionable how much difference that makes at the macro level, let alone whether the difference can account for any robust exercise of free will. Besides the uncertainties are within certain ranges – they are not completely boundless. As such, quantum events still happen “mechanistically”, even if not with absolute precision.

So is determinism the massive problem for free will that it seems to be?

Naturally there are those that think it is. This view is called incompatibilism. Proponents of this view believe that there are necessary conditions for the existence of free will which cannot obtain on determinism. The proposed conditions typically centre on the principles of alternative possibility and ultimate responsibility.

The principle of ultimate responsibility is the view that for an agent to be responsible for an action, that action must have had its ultimate causal origin in the agent.

The principle of alternative possibility (often shortened to PAP) is the view that for an agent do be responsible for an action, an agent must have been able to do otherwise than what she did.

PAP is the condition most focussed upon in the literature, but not all incompatibilists think it is the most important condition, or even a necessary condition, of free will. While an incompatibilist thinks that there is at least one condition that is incompatible with determinism, he/she might think others are compatible with determinism, or think that other proposed conditions are not actually truly necessary conditions for free will at all.

Note also that an incompatibilist need not think that free will actually exists. An incompatibilist qua incompatibilist is committed to the claim that free will is incompatible with determinism, not to the truth or falsehood of determinism itself. As such an incompatibilist can be a determinist, thinking that free will does not exist. This position is called hard determinism.

Another approach to the relation between determinism and free will is the belief that the two are in harmony. This position is called compatibilism. Proponents of this view argue (obviously) that free will is compatible with determinism, or at least, that the kind of free will that is compatible with determinism is robust enough to allow for the moral responsibility of agents, or anything else that we generally consider valuable about free will.

Compatibilists might deny that PAP and other such principles are truly necessary conditions for a robust notion of free will, or might argue that the intuitions driving these principles are still satisfied even given determinism. Accounts of how this is so differ among compatibilists, but often (at least in regard to PAP), centre around counterfactual considerations.

As with incompatibilism, compatibilism itself is neutral as regards to whether determinism is true or not. One could be a compatibilist and think that determinism is false, or be agnostic toward it. As it happens, most compatibilists think that determinism is true, and thus hold the position of soft determinism.

If one thinks that determinism is false and is an incompatibilist, one probably thinks that agents possess libertarian free will (LFW for short). LFW can be understood as any conception of free will that is not compatible with determinism. There are generally two strands: event-causal libertarianism and agent-causal libertarianism.

Event-causal libertarians believe that agents are free when events pertaining to the actions and relevant mental states of the agent, occur indeterministically. Agent-causal libertarians believe that agents are free when they cause their actions and relevant mental states fundamentally as a substance. They propose that agent-causation is a different type of causation to event-causation. That is, agents qua agents possess some property or power to directly cause events in a controlled manner, and this property is defended as being more than just the explanatorily useless property of having ‘whatever it takes to satisfy free will’. If this is hard to grasp it is because this position (at least as typically offered) is hard to outline, and has been subject to the criticism of metaphysical obscurity and conceptual emptiness – that it is unintelligible.

In addition to these views, another is that free will is incompatible with indeterminism (or more clumsily, incompatible with incompatibilism). Indeed, the compatibility of free will with indeterminism is just as much an issue as the compatibility of free will and determinism. Such ‘double incompatibilists’ think that if indeterminism is true then an agent’s actions boil down to luck, and as such, are not within the agent’s free control. Or, a proponent of this view might argue in some other way that LFW is logically incoherent. Peter Van Inwagen, a prominent defender of LFW agrees that it is just as hard to see how LFW fits with indeterminism as determinism, and that proponents of LFW have a hard time (as an understatement) making a positive case for LFW, rather than merely arguing for the insufficiency of soft and hard determinism.

And there you have it, a very broad outline, with much subtlety missed. To show my hand, I am (currently) an agent-causation libertarian.

Advertisements

moral accountability argument (extended and reworked)

1. X is any instance of an agent (A) being animated that is morally relevant.*
2. X is either caused or uncaused.
3. If X is uncaused then A didn’t cause X.
4. If A didn’t cause X then A is not responsible for X.
5. If A is not responsible for X then A is not morally accountable for X.
6. From (3), (4) and (5), if X is uncaused then A is not morally accountable for X.

7. If X is caused then X is preceded by a causal chain.**
8. All causal chains are either infinite, finite, or circular.
9. From (7) and (8), if X is caused then X is preceded by either an infinite, finite, or circular causal chain.

10. If X was preceded by an infinite causal chain then the causal chain either occurred entirely within A’s agency, or extends outside of it.
11. It is impossible for there to be an infinite causal chain within A’s agency.
12. From (10) and (11), if X was preceded by an infinite causal chain then it extends outside A’s agency.
13. If X was ultimately caused by things outside of A’s agency then A is not ultimately responsible for X.
14. If A is not ultimately responsible for X then A is not morally accountable for X.
15. From (12), (13) and (14), if X was preceded by an infinite causal chain, then A is not morally accountable for X.

16. If X was preceded by a circular causal chain then X is ultimately the cause of itself.
17. It is impossible for X to be the cause of itself.
18. From (16) and (17), it is impossible for X to be preceded by a circular causal chain.

19. If X is preceded by a finite causal chain then the causal chain had a beginning.
20. All causal chains that have a beginning have a first cause.
21. From (19) and (20), if X’s preceding causal chain is finite it has a first cause.
22. If X’s preceding causal chain has a first cause, that cause is either A or not A.
23. If the first cause in X’s preceding causal chain is not A then A is not ultimately responsible for X.
24. From (14) and (23), if A is not the first cause of X then A is not morally accountable for X.
25. From (2), (6), (9), (15), (18), (21) and (24), if A is not the first cause in a causal chain preceding X, then A cannot be morally accountable for X.

26. Necessarily, the first cause in a causal chain cannot itself have a cause.
27. All contingent causes are themselves caused.
28. From (26) and (27), if A contingently causes X then A is not the first cause in the causal chain preceding X.
29. From (25) and (28), if A is not necessarily the first cause in a causal chain preceding X then A cannot be morally accountable for X.

*The language of being “animated” in “morally relevant” ways is used to refer to an agent’s action without using language that already presupposes that the agent is the responsible party. Look at the difference in language, say, between “an agent moving his hand” and “an agent’s hand moving”. The first implies that the agent is responsible, the latter leaves that open. In the same way being animated in morally relevant ways refers to an occurrence that is morally significant (say murder, or rape), without suggesting an a priori commitment to what the responsible party is, i.e. it could be an alien tapping into the agent’s brain, or indeed the agent herself.

**To some people “causal chain” might suggest that there is definitely more than one preceding cause. I don’t take that phrase to have any quantitative input in this argument (there could just be one preceding cause – it would still quality as a causal chain here).

we are morally accountable for our choices so long as we necessarily make them.

1. X is any way an agent (A) can be animated that is morally relevant.
2. A is morally accountable for X iff A is responsible for X.
3. A is responsible for X iff A is the ultimate cause of X.
4. From (3) and (4), A is morally accountable for X iff A is the ultimate cause of X.
5. A cannot be an ultimate cause of X unless A causes X.
6. From (4) and (5), if A does not cause X then A is not morally accountable for X.

7. If A causes X then A does so either necessarily or contingently.
8. Of metaphysical necessity, any contingent occurrence has itself got a causal explanation.
9. If an occurrence has itself got a causal explanation then that occurrence cannot be the first cause in a causal sequence.
10. If an occurrence is not the first cause in a causal sequence then it is not the ultimate cause of the causal sequence.
11. From (4), (8), (9) and (10), if A causes X contingently then A is not morally accountable for X.

12. From (6), (7) and (11), A can be morally accountable for X iff A causes X necessarily.

(Note: The language of being “animated” in “morally relevant” ways is used to refer to an agent’s action without using language that already presupposes that the agent is the responsible party. Look at the difference in language, say, between “an agent moving his hand” and “an agent’s hand moving”. The first implies that the agent is responsible, the latter leaves that open. In the same way being animated in morally relevant ways refers to an action that is morally significant (say murder, or rape), without suggesting an a priori commitment to the responsible party.)