Wednesday, October 30, 2013

What Does Neuroscience Tell About Free Will? The Libet and Other Experiments, and Interpretations

Introduction and Context:
How do you think we "decide" to act? The common sense (and our experience) explanation is we (1)  make a conscious decision to do something (2) our brain activates whatever neuro-pathways are required for the action, the (3) we perform the action.  Libet's famous experiments give strong evidence showing that this is NOT the order of how our actions come about.

For many people, the famous Libet experiments show that we don't have free will.  Free will is only an illusion.  Our brains have already decided what we're going to do, then, after the fact, we only have the experience of deciding what we'll do.  Watch the video yourself and think about what the experiment shows.

In case it wasn't clear from the video, here's how the experiment goes:  The subject observes a timer thingy (the type of timer varies from experiment to experiment).  The subject is asked to raise their finger whenever they want.  By looking at the clock, they are also supposed to note the time at which they first became aware of their (conscious) desire to move their finger.  The subject is also wired up to an EEG (electroencephalography) sensor which measures the electrical potentials around the scalp coming from the part of the brain responsible for motor activity.  There's also an EMG (electromyography) sensor that measures the exact time the finger moves.

The results of the experiment show that there is a ramping up of brain activity .550 seconds before the subject's consciously aware of their decision or desire to move their finger.  This "ramping up" activity is called readiness potential (RP).  So, the order of events is (1) RP, (2) conscious willing to move the finger, (3) finger movement.   In theory, because readiness potential happens before conscious awareness of a decision, Libet can tell us we are going to move our finger before our own conscious awareness of our decision to do so!  Mind=blown.

Wegner's Interpretation
Libet's experiments seem to give compelling evidence in favour of determinism.  Our conscious experience of choice is an illusion.  Our body's physical systems have already "decided" what to do and our consciousness of what we will do occurs only after this happens.  Our conscious selves are merely along for the ride.  "Voluntary" actions don't go: (t1) "hmmm...I'm going to move finger now", (t2) *finger moves*.  They go like this (t1)  brain initiates preparations for moving the finger (t2) meta-brain says "I decide to move my finger" (t3) *finger moves*.

From these experiments it doesn't seem like we consciously will our actions.  Our dictator brain has already begun preparations for what you will do before you are even conscious of it.  Our consciousness selves just think they're making a decision.  Curse you, evil brain! I want to be free!

Libet's Interpretation
Libet's own interpretation was different.  He thought that rather than free will, we had "free won't".  He thought, yes, the brain initiates urges and intentions but we have a window (about .1-.2 seconds) to consciously override the brain's urge.

To test his hypothesis he set up the following experiment.  He set things up similar to the original experiment but this time he told the subjects to plan to move their finger at a set time on the timer and then to "veto" the intention to move their finger.

Results:  RP started about 1 second (vs .550 sec. in the original version) before the set time.  Then at about .1-.2 seconds before the subject was to move their finger RP flattened out.

Interpretation:  The brain generated the unconscious desire to move the finger but when this desire entered into consciousness "free won't" was able to veto the urge.   In other words, our desires and intentions are generated unconsciously but when they enter consciousness we have the ability to over-ride them.

Problem: What if the process that generated the 'free won't' is also unconscious?  Doh!

Mele's Interpretation
Alfred Mele be like, "whatchu talkin' 'bout Willis? That ain't no proof of determinism!"  Poor Libet. He doesn't have a philosopher's training and is therefore blurs some important distinctions.  In Libet's interpretation of the results, he uses the words "intention," "decision," "wish," and "urge" interchangeably.  Unfortunately for Libet, he never had the good fortune of taking philosophy 101 at UNLV where he would have learned that you can't just go around willy-nilly using words without first specifying what they mean.  Lets look at some of the important distinctions and see how they apply to interpreting Libet's experiments.

Wanting/Urges to vs Intending/Deciding to
You can want to do A without having settled that you are actually going to do A.  I want to live on a ranch with a herd of wiener dogs but I don't intend to do it (right now, anyway).  I can want to eat all the donuts in the bakery but still not form the intention to do so...

We can further see the distinction when we have competing wants.  I want to finish my grading by 9pm but I also want to finish writing my lecture by 9 pm.  I can't do both.  The one that I end up doing is the one for which I formed an intention.  When you make up your mind about a course of action between competing wants then you can say you intend to do it.  In short, wanting to A is simply having the desire to A.   Intending to A requires making a decision to A.

Distal vs Proximal Intentions
We can also distinguish between distal and proximal intentions.  A proximal intention is when I intend to do something that is temporally close.  A distal intention is when I intend to do something in the more distant future.  For example, on Saturday I intend to take my dogs for a hike.

Ok, back to Libet.  Libet says that the process that produces the urge to move the finger (the 'act now' process) is occurring before conscious awareness to decide to move the finger.  This process begins at around 550 msec before the finger moves.  Also, the urge that initiates the 'act now' process creates a proximal intention to flex the finger.   So far, we can agree with Libet that the "'act now' process is initiated unconsciously, [...] conscious free will is not doing it"; i.e. conscious free will is not initiating the 'act now' process.

However, why should we suppose that the role of conscious free will is to produce urges or causally contributes to urges?  Typically, free will is thought to apply to situations where the agent is deliberating between between possible courses of action or whether they should or should not act.  Free will is not thought to have the role of producing urges, rather, it is about choosing.

Free will does this:

Processes Have Parts
Free will doesn't create the urge.  The origins of the urge are unconscious.  However, the process that begins with an unconscious urge can give rise to a conscious intention to act or not act in accordance with the urge.  The conscious intention is temporally closer to the final act (move finger) and so it seems as though it is the conscious intention rather than the unconscious urge that is causally responsible for the act.

Issue:  What is the relationship between temporal distance and causal power?

Other Objections/Issues to Deterministic Interpretation of Libet Experiments

Issue:  Do these results generalize?  Lab conditions vs Real life.  Do the results generalize to all types of decisions/intentions?

Objection: Of Course There's Prior Brain Activity!
If brain events underlie mental events then we shouldn't be surprised that there is brain activity prior to a conscious decision.  Why should we suppose that the production of conscious decisions doesn't involve prior brain activity to lead up to the brain state that is a conscious mental state?  Having no prior brain activity to a conscious decision would be the surprising finding.  Not that there is prior brain activity.

Objection:  The Meta-State
Consider that you've been reading this post for the last minute or so.  The entire time that you were reading or watching the video were you actively conscious of the fact that you were reading or watching the video?  Or were you reading and watching without the awareness "I'm reading/watching".

The argument here is that the Libet experiment measures an awareness of a conscious state; i.e., a meta-consciousness.  Most activities that we do, we aren't actively aware.  When we drive, read, walk, etc...often doing so isn't part of our immediate conscious awareness, yet no one could seriously say that we aren't conscious when we do these things.  Only when something draws our attention to our activity do we become aware of what we are doing;  this is the meta-conscious state.

The Libet experiments measure the meta-conscious state about a prior awareness of our decision to move our finger, not the immediate state of awareness that we want to move our finger.  The time delay between the primary state and the meta-state is what accounts for the effect.

Further Studies that Might Prove Determinism True:

In further studies using an fMRI machine, scientists in lab coats have been able to predict a subject's decision of up to 7 SECS before the subjects own awareness of what she will choose.  Ho.  Lee.  Crap. If this doesn't sound like evidence against free will, I don't know what is!

Objection 1:  The Media Isn't Reporting the Whole Story and Sensationalizing (Surprise!)
What the data actually shows is that the scientist can predict your decision to move your finger at a rate 7% better than chance.  If you were to make predictions blind, over the long run, you'd be right about 50% of the time.  The fMRI data allows you to get it right around 57-58% of the time.

Reply:  Yes, but 7-8% above chance is still a significant result.  If you were given these odds in a casino, you'd be a fool not to take them.

Counter Reply:  True dat.  However, this result may be a consequence of how the experiment was set up.  If subjects were incentivized to try to fool the experimenters, this predictive power might disappear. (If you have an fMRI machine, please do this study!)

Further Study:  There may be newer studies that use better equipment and more sophisticated models that have better predictive power.  There doesn't seem to be any a priori reason to suppose in the future 100% predictive power could be achieved.

Preferences Vs Free Will
Suppose you frequently go out to dinner with someone you know very well.  You've eaten with them many times.  You know what their preferences are.  You go to a new restaurant and based on what you know about them, you successfully predict they will order the T-bone steak.  Does the fact that we can accurately predict someone's actions tell us anything about free will?

The Mechanistic Brain Vs Free Will (Adina Roskies)
So maybe being able to predict someone's behaviour--be it from fMRI scans or from preferences--isn't sufficient to imply determinism is true.  Do these studies provide evidence for any other challenges to free will?

One thing these (and subsequent) studies make clear is that the brain is mechanistic.  We can identify which parts of the brain and which neurons are responsible for certain actions and behaviours.  In short, the brain behaves in mechanistic law-like ways.  So, the difficulty is to explain how we get free will out of a mechanistic law-like system.  Consider a computer.  It performs its functions in mechanistic law-like ways, yet we don't attribute to it free will.  How are we different?  Are we really all like 2Chainz showin' up to scene with our top down but not consciously deciding to do so?  Is it cuz we're made of meat rather than metal and silicon?  What's so special about meat?

The underlying worry is that because we are meat-based mechanisms we don't have free will.  But Adina Roskies suggests maybe this conclusion needn't follow.  If we suppose that having a mind is necessary for free will then maybe having a better understanding of the brain's mechanism gives us a better understanding of mind.

For example, most theories of free will tell us that there certain mental capacities are required for free will:  the capacity for rational deliberation, the capacity to assign moral value to certain outcomes, the capacity to put judgments into action.  So, while at first blush it may seem that neuroscience undermines free will, in fact it doesn't, it gives us a better understanding of the brain mechanisms, functions, and states that underlie the mental capacities that are integral to free agency.  This type of study can inform us of things that can happen to the brain that can impede capacities for free agency.

An often cited example in the literature is a patient that led a perfectly normal life up to a point when he started to have pedophilic urges and eventually couldn't control himself.   When he was sentenced to jail, he complained of a headache.  When they scanned his brain they found a large tumor.  When they removed the tumor his urges completely went away.

Later he started to feel the urges again and when they scanned his brain, they found a tumor in the same place.  Once the tumor was removed, the urges disappeared again.  This is fairly strong evidence for a causal relationship between the tumor interfering with normal brain activity and the ability to exercise one's will.

Epiphenominalism: The Role of Consciousness in Decision-Making
Epiphenominalism is the idea that our conscious experiences don't play any causal role. They just "ride on top" of whatever our brains our doing. They're superfluous. We encountered this idea earlier with Chalmers' Zombies, and blind-sight.

So, if our brains are causing us to act before we have any conscious awareness of what we're going to do, then why should we think that consciousness plays a role in decision-making? First of all, it looks like there are at least some areas where consciousness does play a causal role. Conscious experiences (memories) can inform decisions event if those decisions proceed unconsciously.

For example, my memory of the salad bar being really slow causes me to choose something else like subway for lunch. The decision might be unconscious, but a conscious state plays a causal role in the decision.

Maybe Libet's results support limited epiphenominalism about decision-making.  Consciousness doesn't cause the decision but it doesn't make consciousness irrelevant. My conscious experiences figure into my unconscious decisions.  But the decision isn't caused by consciousness.

Should non-determinists be worried?
From a neuro-science point of view, Libet's findings make sense. The decision-making system does its job first, then the conscious monitoring system does its job. Of course the decision has to come first, otherwise there'd be nothing for the conscious monitoring system to monitor!

Possible Compatibilist interpretation: The urges/wants I have are going to be a consequence of my values and preferences. In that sense, they represent what "I" want rather than being totally random.

No comments:

Post a Comment