Wednesday, December 19, 2007

Some More Musings on Free Will

Someone once famously asked John Searle "If determinism were scientifically shown to be 100% true, would you accept it?" His response was "Think about what you're asking me: if it were scientifically proven that there are no free and deliberate actions, would you freely and deliberately accept that?" His response, while somewhat arrogantly dismissive (he's great like that), is pretty telling: no matter what we learn, we're never going to be able to escape the illusion (if it is an illusion) that our actions are free. In some sense, then, the problem of free will isn't really a problem at all--no matter what the answer is, nothing will change in our practical experience.

It's worth asking, then, why we even care about the problem in the first place, given the fact that even if we learn beyond a shadow of a doubt that we are not free, we will be unable to stop experiencing the world as if we are. One good reason to still care about free will, it seems to me, lies with moral responsibility. We have a powerful intuition that we cannot be held praiseworthy or blameworthy for anything we don't do of our own free will; if you put a gun to my head and demand that I rob a bank, it doesn't seem fair to blame me for robbing the bank. Similarly, if you put a mind control chip in my brain and force me to run around town fighting crime, it doesn't seem that I deserve to be praised for my law enforcement efforts. It seems that we need freedom to get responsibility.

This is the jumping off point for semicompatibilism, a view proposed by John Martin Fischer in his essay "My Compatibilism;" it is this idea that I want to explore today. The first question we should ask is "what sort of freedom do we think we need in order to get moral responsibility?" Our intuition, I think, is that we need alternative possibilities--if we fail to praise or blame someone for actions taken while under the effect of mind control, it is precisely because we see that the person had no alternative possibilities; in other words, it only makes sense to praise or blame someone for actions he takes when he could have done otherwise. The idea is that without alternative possibilities, we can't have moral responsibility; the two notions are stuck together in a fundamental way (this is called the Principle of Alternative Possibilites, or PAP). Let's see if we can pry them apart.

Let's suppose I kidnap you while you're asleep, and bring you to a locked room. When you wake up, you find yourself in strange, unfamiliar surroundings. On closer inspection, though, it turns out that the room you're in is full of all sorts of interesting things to explore and interact with (populate the room with whatever appeals to you). While you notice the closed door, it doesn't occur to you to try to open it, so taken are you with the diversions in the room. Let's suppose that you stay there for 45 minutes playing around before you try to leave the room, only to discover the locked door. You can't leave the room, nor could you have during the past 45 minutes.

Now, it seems to me that during the 45 minutes in which you were playing around in the room, you were staying there freely--that is, you chose to stay in the room, and any adverse consequences (say, if you forgot to pick up your kid from school) from your stay would be your moral responsibility. However, it turns out that you actually didn't have any alternative possibilities during your stay there--you couldn't have left the room even if you wanted to (think about this in relation to the conditional analysis discussed last time). It looks like we're getting somewhere here in our attempt to pry apart responsibility and alternative possibilities.

That last thought experiment was originally penned by John Locke, but critics quite reasonably pointed out that in that situation, you would have at least two distinct alternative possibilities--namely, to try to leave the room or not to try to leave the room. In response to this, philosophy Harry Frankfurt proposed moving Locke's Locked Room inside the head, giving rise to what are now called "Frankfurt cases." Here's a typical Frankfurt case.

Suppose that I, an evil neuroscientist, have implanted a control chip in your brain. This control chip is designed to let you take your actions normally when I'm not in control, but gives me the option to override at any time and take control of you. It also lets me know what your intentions are a moment before you actually take the action, so I have a chance to countermand you if I'm quick. Suppose further that in addition to being an evil neuroscientist, I'm also a devout Republican, and that I decide to test my chip in the upcoming election by forcing you to vote Republican. Election day rolls around, and you step into the booth, totally unaware that I'm watching my chip's readout with my finger on the button to make you vote Republican. After some time considering your options, you decide that the Republican candidate best represents your beliefs, and cast your vote for him. I take my finger off the button, satisfied (though perhaps a bit disappointed at not being able to test my invention).

Here we can see how Locke's room has been moved inside your head, and once again the results of this thought experiment are interesting. In the case I just described, it seems that it would be very reasonable to assign moral responsibility to you for voting Republican, even though you really had no alternative possibilities (even if you'd wanted to vote Democrat, you wouldn't have). It looks like we've succeeded in prying those two notions apart.

Fischer talks about two different kinds of control that we might have over our actions: guidance control and regulative control. The kind that we tend to think we have is regulative control--that is, regulative control implies true freedom, and the ability to do otherwise at any juncture. Guidance control, though, is the kind of control being exercised in the stories presented above. Here's another example that will let you see the difference more clearly.

Suppose I secretly install a computerized guidance system in your car before you go to work. I've been researching your route for a few days now, so I know that you always proceed to work in exactly the same way. I program the guidance computer to kick in as soon as you get ready to back out of your driveway, and to quit as soon as you park. The computer guides your car to work just as if you had driven it there; in fact, it is so convincing and well designed that you never even notice that anything is amiss--you simply feel like you're driving as normal, as the car responds appropriately when you brake, turn the wheel, or change gears.

In this case, Fischer (and I) would argue that though you lack regulative control (if you tried to diverge from the normal path to work, the computer wouldn't let you), but you do have guidance control. Just as in Locke's room and other Frankfurt-style cases, you can be said to have had a certain degree of control over your circumstances, even though you lacked regulative control (and alternative possibilities). Fischer contends that it is only this lesser notion of control that is necessary for moral responsibility, leaving the question of whether or not we actually have regulative control open.

I like this doctrine. It has problems, but I think it is parsimonious enough that a careful formulation will avoid many of them, particularly in light of the fact that it has nothing to say one way or the other about regulative control and alternative possibilities. It does seem to me that the major reason to care about the free will problem is for the sake of moral responsibility, and semicompatibilism shows us that we can be responsible without being truly free.


Michael said...

I like this doctrine.... It does seem to me that the major reason to care about the free will problem is for the sake of moral responsibility, and semicompatibilism shows us that we can be responsible without being truly free.

My reaction is 180 degrees from yours. What possible point does moral responsibility have if I have no free will? What possible benefit is it to the world or to myself or to you if you assign moral responsibility to actions over which I had not choice, because I choose not to flagellate myself about those actions? Such a thing would seem to be a sort of moral version of an untestable hypothesis. Absolutely nothing in the world is changed by your decision to assign moral responsibility to me for my actions, then if you didn't assign any responsibility.

The only positive purpose I can think of for assigning moral responsibility for actions is if my doing so might change my future actions in a future similar situation. But you have hypothesized that this is impossible. So why would you or I ever choose to assign moral responsibility for actions I cannot control?

I am really enjoying your blog and checking it every few days. If I am missing something important about this, please let me know. Also, if there are any reasonably accessible books on this topic I could read, I would love a recommendation.


Michael said...

If you have any good things to read on the libertarian point of view I would appreciate those. I am a PhD physicist who has worked a lot on quantum effects in superconductivity, and on quantum noise. When you write I've yet to see a libertarian account that impresses me as much of anything except wishful thinking (they also tend to make pseudoscientific style appeals to quantum mechanics) I have to tell you that it does seem possible that the loophole in determinism is in quantum mechanics.

I think it is important to realize that the "billiard ball" determinism of classical newtonian physics does not extend to QM. I believe it is at worst an open question, and at best clear enough, that QM is NOT deterministic. So if the physical universe is NOT deterministic, then free will creates no conflict with physics.

I don't understand the appeal of a theory that gives us free will in determining what we think of some actions, but no free will over any actual actions. First it is puzzling, if I ask you what you think, it is deterministic what you will tell me, since that is an action. Then it seems random whether your conversation about such a thing has any meaning at all. If you tend to be honest, then it is an amazing coincidence that your beliefs (over which you have free will) just happen to align over your predetermined statements about your beliefs.

I'm with Searle on this one (not that I know anything about him beyond what you said.) If you ask me if I have free will, I choose to say yes. It is hard to even parse what is going on if I "choose" to say no.

To me it seems clunky to separate free will about what I believe from determinism about what I do. And especially if the motivation to do so is what is at best an unproven and at worst a mistaken idea that physics, as far as we know, is deterministic. As far as I know, QM theories in which a wave function collapses on observation are NOT deterministic. Attempts to make them deterministic (for example, hidden variable theories) have failed.

Anyway, great stuff! Thanks!


That 0ne Guy said...

I don't think this is a good demonstration...

If I had a mind control chip in my brain, and the controller decided I needed to vote republican, and yet I chose to vote republican anyways, well that just tells me I have similar political views, and chose the same as the controller.
I understand I couldn't otherwise, but I don't think that it makes a very good argument.

Now on the flip side, if I had chosen to vote Democrat, and was forced to vote republican, doesn't mean I don't have regulative control, it just means I've been over powered by guidance control. Because when I'm forced to vote republican, doesn't mean I didn't want to vote democratic.

So tell me, what does guidance control have anything to do with the matter?