Philosophy, Film, Politics, Etc.

Wednesday, February 26, 2014

An Argument For Compatibilism

Compatibilism is the idea that there is no conflict between determinism and free will.  Incompatibilism is the idea that free will cannot exist in a deterministic universe.  There's been a lot of discussion over which view is correct.  What's remarkable about the debate isn't so much the stubbornness or passion which has been exhibited by this or that party, but the fact that the very terms of the debate are controversial.  There is a great deal of confusion about what the key issues in the debate are and how we should be talking about them.  As a result, there is a meta-debate within the debate itself.  You cannot engage in the debate without also engaging in a debate about the debate--about what issues are at stake and about how the issues should be framed.  So here's what I want to do:  I want to explain why I think incompatibilists are doing a very bad job of framing the debate, and also why compatibilism is the most reasonable option on the table.  It's an ambitious project, and I don't expect to win over many audiences with my arguments.  But I do hope to stimulate a bit of critical reflection and perhaps help guide others towards a more fruitful way of thinking about the issues.

First, we need a working definition of "free will."  One philosophically respectable way of defining it is as the ability of a rational agent to choose from among a variety of options in such a way as to satisfy the requirements for moral responsibility.  In other words, the extent that a person has free will is the extent to which they are morally responsible for their actions, where moral responsibility is predicated on their ability to make choices.  This is not the only possible definition, but it seems flexible enough to fit with everyday intuitions about free will.  For that reason, I will adopt it for now.  If it needs to be altered, so be it.

The compatibilist position, therefore, is this:  A deterministic universe can contain rational agents which are capable of making choices among a variety of options and therefore carry a burden of moral responsibility.  The incompatibilist position is that free will cannot exist in a deterministic universe.

Some self-identifying incompatibilists say that the very notion of free will is incoherent.  If that is the case, we should not say, "Free will is compatible with determinism," because the phrase "free will" does not have any denotation. Yet, we should also not say, "free will is incompatible with determinism," and for the same reason.  Thus, this formulation of incompatibilism is problematic.  If there is no coherent notion of "free will," then we simply cannot say anything about what is compatible or incompatible with it.  We either have to find an acceptable definition of the term or stop using it.  Therefore, incompatibilists who think "free will" is incoherent are in a bind.  They should not say that free will is incompatible with determinism.  They should rather say that they have no idea what "free will" means, or is supposed to mean.  They might rather call themselves noncognitivists about free will, instead of incompatibilists.  I will come back to this view later.

Other incompatibilists are cognitivists.  They think that there is a coherent notion of free will, and they think it is incompatible with determinism because determinism means you never really have a choice.  If your behaviour is determined by forces which are beyond your control--by states of the universe that existed before you were even born--then there was never any real choice.  There was only the illusion of a choice.  So we have, on the one hand, the idea that your choice is only real if it is not the inevitable result of forces which are outside your control.  A compatibilist, in contrast, will say that the inevitability of our behaviour does not mean we lack a choice.  What makes an action a choice is the involvement of rational deliberation.  This requires a sort of information processing which represents different patterns as choices.  We only have a choice in so far as we represent an option to ourselves as a choice.  Whether or not our behaviour is inevitable is not the issue.  The issue is whether or not our behaviour entails the rational deliberation of patterns which are represented as choices.

Now, incompatibilists (who believe "free will" is coherent enough to talk about) will say that this is not enough.  For a choice to really be a choice, it cannot be inevitable.  Here's why they are wrong.

If a choice is evitable, it means that it was not the result of deterministic forces.  It was, in essence, random or uncaused.  That means it was not caused by our beliefs or desires, or mental states of any kind.  At least, not necessarily.  It was the accidental or random outcome of some processes.  Yet, if the action were a random or accidental outcome, we would not be able to claim responsibility for it unless we were already responsible for the decision to act on that random outcome.  To deserve responsibility for an action (according to indeterminists), there must be some causal relationship between our action and our beliefs and desires.  It cannot all be random.

Dennett has discussed this in great detail in, for example, his paper, "On Giving Libertarians What They Say They Want."  He develops a two-stage model of free will (see here for a clear and concise analysis of Dennett's arguments.)  What he concludes is that we can imagine a scientifically respectable scenario in which human beings have the ability to make rational choices which are not inevitable, which do in fact involve a degree (however large or small) of random processes.  The idea is that the construction of our beliefs and desires can entail a large degree of randomization, while the choosing of our actions and judgments can be deterministic.

The point of this is not that incompatibilists are correct.  It is rather the opposite:  Even though we can imagine a scenario in which rational agents make choices from among a variety of randomly-generated options, the making of the choices is still deterministic.  It is still a result of forces outside the agent's control, whether or not we introduce non-deterministic elements into the system.  So there is no reason to say that it is not "really" a choice in the deterministic universe.  Either it is a choice in both scenarios, or it is not a choice at all.  Dennett has shown that, even if we give incompatibilists (of a certain stripe) what they say they want, they still have no basis for denying compatibilism.  Free will is just as compatible with determinism as it is with indeterminism.

The issue, then, is not whether we make choices, but whether we do so in a way which gives us moral responsibility.  The question comes down to whether or not our choices are sufficient to make us morally responsible for our actions.  That, I believe, is where the problem of free will needs to be resolved.  For now, let's take "morally responsible" to mean "deserving of punishment or reward."

Let's imagine a God's eye view, but without the supernatural baggage. Let the term "God" refer to any being powerful enough to create human beings, know whether or not they are morally responsible for their actions, and punish or reward them for their behaviour.  Let the term "soul" refer to that aspect of a person which is capable of being morally responsible and which God rewards or punishes.  God rewards or punishes according to rules, the source of which is irrelevant for our purposes.  Also, God rewards or punishes by measuring ultimate causal determination.  (Notice that I am intentionally avoiding any supernatural or theological language here, for the sake of coherence.)

Imagine God creates two people, Person A and Person B.  Person A follows God's rules, but Person B does not.  The question I want to consider is, what could justify God's decision to reward Person A and punish Person B?  That is, what could justify God's decision about ultimate causal determination?

If Person A chose God's path because of some inherent qualities that made them do good--i.e., if they had a good soul--then God would be the ultimate cause of their good behaviour.  As a result, Person A is not deserving of any reward.  Similarly, if God gave Person B a soul inclined towards evil, Person B is not the ultimate cause of their choice.  Therefore, if God is to be justified in punishing or rewarding anybody by appealing to ultimate causes, they must be created with a neutral soul.  The question then is, what could lead anybody to choose bad over good, or good over bad?

Two possibilities spring to mind.  The first is external circumstances.  However, these are out of the people's control, so they should not be punished (or rewarded) for how the environment has affected their souls.  The second is random choice:  People can flip coins and act accordingly.  In that case, it is chance (which is just another external factor) that is responsible.  In neither case is the person's soul the ultimate cause of their good or bad behaviour.

This seems to exhaust all of the options, which means that there is no case in which God is ever justified in punishing or rewarding anybody's soul, if we take ultimate causes to be the determining factor.  In other words, if we take a God's eye view and rely on appeals to ultimate causality, there is no such thing as moral responsibility.  And notice that I have not stipulated whether this is a deterministic or indeterministic universe.  Either way, the result is the same: Ultimate causality is not an adequate ground for moral responsibility.

When we want to determine if a person has moral responsibility--remember, this just means whether or not a person can be deserving of reward or punishment--it is a mistake to frame it in terms of whether or not they are the ultimate cause of their choices. If there is moral responsibility, it must be understood in different terms.

Whether or not a person is deserving of punishment or reward could be a psycho-social matter, a matter to be decided by human roles and relationships.  In that case, people can hold themselves morally responsible, but they are doing so as self-aware agents capable of taking up a moral attitude towards themselves.  It is a complex psychological phenomenon, and it is not resolved by appeal to questions of fundamental causality.  (Notice that I am remaining neutral at this point about what psycho-social justifications might be acceptable.  The important point to note is that, whatever justifications we find acceptable, we will not be able to entirely reduce them to statements about ultimate causality.)

If you accept that point of view, then the implications for free will are straightforward.  People do have an ability to rationally choose among options (to the extent that they can represent options to themselves and choose between them according to processes of rational deliberation) and this process can ground people's sense of moral responsibility in so far as it supports psycho-social justifications for holding them responsible for their actions. This is free will (by the definition given at the outset), and it does not depend on whether the universe is deterministic or indeterministic.  Therefore, it is a coherent (and I think compelling) version of compatibilism.

As for the noncognitivists who maintain that there is no coherent definition of "free will," I submit that I have just presented one.  They might respond that what I am talking about is not the free will that people are normally talking about.  People normally want free will that can be judged by a God's-eye view by appeals to ultimate causality, and this is impossible.  They say I'm therefore just changing the subject.  I'm replacing an incoherent notion with a coherent one.

I do agree that many people want free will from a God's-eye view and by appeal to ultimate causality, and they will not be easily satisfied by the psycho-social foundations of free will I have described.  However, that does not mean I am changing the subject.  I have not altered the definition of "free will" or the definition of "moral responsibility."  All I have done is shown that the foundation for moral responsibility people think they want is an impossibility.  People are mistaken about what could make them deserving of punishment or reward. It cannot be ultimate causality.  People might therefore conclude that there is no such thing as moral responsibility.  They might say that, if God doesn't exist, nobody is morally responsible for anything.  And, indeed, it would seem that there cannot be free will if there is no moral responsibility.  However, the way forward is not to simply claim that there is no free will.  The way forward is to explain why morality does make sense from a psycho-social point of view--why people should invest in their sense of moral responsibility.  Of course, you cannot argue that people should embrace moral responsibility without begging the question.  But what you can argue for is a coherent picture of the way moral responsibility actually works.  If a person can be convinced that moral responsibility does make sense in psycho-social terms, then they will have made room for belief in free will, and no definitions will have been altered.

My conclusion is this:  If we insist that there is no coherent definition of "free will", then compatibilism and incompatibilism are both a waste of time.  There is no sense in claiming that free will is or is not compatible with determinism.  If, on the other hand, we find sense in the meaning of "free will," then incompatibilism seems unjustified.  If there is such thing as rational choice and moral responsibility, the these things are compatible with determinism.  Therefore, if free will is a coherent concept, it is compatible with determinism.