Free Will
I've ended up reading about three different items in the past week on this particular debate, which has raged for centuries, if not millenia. Essentially, it goes like this:
- The universe obeys certain fixed laws.
- Our minds function in our brains, according to those laws.
- The laws of the universe dictate the functioning of our brains.
- Since our brains work by those laws, so do our minds.
- Since our minds work by fixed laws, our choices are determined by fixed laws.
- Since our choices are determined by fixed laws, we cannot have chosen otherwise.
- Therefore, free will does not exist.
The point of this argument is to show that free will is incompatible with determinism. In other words, it tries to show one of the following:
- Free will does not exist, or
- Some part of our humanity exists outside the realm of the physical.
Usually it's those on the left trying to show item (1), with those on the right trying to prove item (2). The above argument has existed in multiple forms ever since scientific thought actually began to show some success in working out the way the world actually functions.
A variation on this argument, of course, goes roughly like this:
- God is omniscient (all-knowing).
- Thus, God can see the future with certainty.
- If God can see the future, he can see the outcomes of our choices.
- Therefore, God knows what we will choose.
- Therefore, we cannot choose otherwise.
- Therefore, no free will.
Let's take a simple example. Let's say you're trying to decide whether to eat dinner, or skip it and go straight to bed. This isn't a huge decision in the grand scheme of things, but enough to illustrate our point.
You're hungry, but you're also tired. Plus, there's no real food in the house. And you ate a really late lunch, so you aren't all that hungry. Based on all of this, you decide to go to bed and just get up a little early and have a good breakfast.
Now. The opponents of free will argue that since you chose to go to bed, you could not have chosen otherwise. Their evidence? You didn't choose otherwise. Of course, had you chosen to eat a light supper before bed, the would have argued just as strenuously that you could not have chosen to go to sleep -- because you didn't.
Scientific Determinism
Determinism is the notion that, given knowledge of the initial state of a system, we can work out the later state at any point in time you care to name. Because the laws of Nature of known, anything that obeys those laws must be predictable.
When it comes to free will, though, there are several possible problems with this.
Chaos Theory
The first is chaos theory. To summarize, chaos theory works with iterative systems, which are systems where the "next" state is based in some way on the "current" state. In theory, any set of equations can be written in an iterative form. The exact methods are beyond the scope of this post, of course.
A system is said to be chaotic if the outcome can vary widely based on small variances in the initial conditions. This is best explained by contrast. If I throw a ball, and I'm wrong by, say, 0.1 percent about how hard I throw it, then in general I'm going to be wrong about where it lands by roughly 0.1 percent. It might be a little more, might be a little less, but in general: small initial error, small error in result.
A chaotic system doesn't work like this. It says that if I'm the tiniest bit wrong about how hard I throw the ball, the ball could end up in China instead of across the street. Now, naturally, balls don't work like this. But some systems do. The weather is a perfect example -- small disturbances in the atmosphere can, given the right circumstances, turn into a thunderstorm. Or not. It's hard to tell exactly what will happen.
All the information we have indicates that the brain is just such a chaotic system -- that a small question concerning whether or not a neuron quite reaches the threshold needed to fire can determine whether or not a whole train of thoughts or sensations occurs. It's hard to tell.
Chaos theory is still deterministic! It is still based on fixed rules. What it says, though, is that in practice we can never make good enough measurements to truly predict the outcome of a system down the road. That even though we may have a snapshot of how the brain is right now, that predicting your thoughts a week or even a few minutes later may well turn out to be impossible.
Quantum Theory
Quantum theory was developed during the first half of the twentieth century, and there have been a lot of people who have misinterpreted it to attempt to prove philosophical points. However, I won't be using it to try to prove deep fundamental principles, but instead simply to make some comments about chemistry which, after all, underlies the way the brain works.
Quantum theory, fundamentally, says that once you get down below a certain size, that events become basically random. Science can no longer tell you what will happen, but instead can only give probabilities. This is based entirely on the particle's mass, and electrons definitely fall within this mass range. This is called the Heisenberg Uncertainty Principle and is a core principle of quantum theory and the standard model, and has been verified by a huge amount of experimental data.
This is not just a limitation on our ability to measure things! As far as we can tell so far, this doesn't come from us just not having good enough instruments, but instead from a fundamental limitation of the physical laws of the Universe.
If everything's based on quantum mechanics, and quantum mechanics says that we can't predict the behavior of individual particles, then how can we predict anything? The answer comes from probability theory.
Let's say you flip a coin. I can't tell you much of anything about the outcome of that particular flip. But if you flip it 1,000 times, I can tell you that you're going to get about 500 heads and about 500 tails. I can even (assuming it's a fair coin) give you a pretty decent idea of how much away from 500 it's likely to get.
Likewise, we can't tell you what a single electron is likely to do. But when we have a million of them, their behavior becomes quite predictable and orderly. The chances of them doing something odd becomes vanishingly small -- small enough that we can ignore it. And when we're dealing with the everyday world, where we're talking about billions of billions of atoms making up a baseball, we can pretty much ignore the uncertainty principle and just use good old Newtonian physics to describe the path of a fastball.
The more we learn about the chemistry of the brain, however, the more it appears that quantum effects may play a role. By being poised on the edge, so that a few electrons either way can make the difference between a neuron firing and not firing, a certain degree of uncertainty enters the calculations. Note, of course, that they are still deterministic -- these are fixed laws describing our universe and all the matter within it. But we can't narrow it down any better than to a given probability.
The exact implications of this are still being researched, but it does not body well for being able to make firm predictions about a person's thoughts from first principles.
Emergent Behavior
Emergent behavior is a characteristic of complex (and chaotic) systems that involves a feedback loop of some kind. Emergent behavior means that a system can implement some very simple rules and from those rules come up with astonishingly complex patterns, patterns that might not be readily apparent from those rules.
A group of engineers decided to implement an evolutionary model of circuit design using a computer. The idea is that they wanted a circuit that would generate one result when it was fed a signal at, say, 1 KHz, and another result when it was fed a signal at 10 KHz. This is a fairly simple circuit.
They programmed a computer to basically build circuits and test them. At each stage, it would introduce random variations into the circuit, and it would then test each of those, and pick the one that it considered to be at least moving in the right direction. All the circuits were simulated in the computer, using standard electrical engineering rules.
The end result? After about 1000 iterations, the computer produced a circuit that functioned flawlessly, and was basically identical to the circuit that an engineer would have produced. Nothing huge there, right?
However, here's where it gets fun. Another researcher read about this, and realized that since the computer was working with theoretic circuits, that they came with a lot of theoretical assumptions. So he set up his own experiment. In this experiment, he used PLA (Programmable Logic Array) devices -- real, physical devices that can be 'rewired' under computer control. It used that as the test-bed, but used the same evolutionary strategy as before.
This time, the computer took a lot longer to come up with the result. But the result it came up with was very surprising. It worked flawlessly -- and used roughly a quarter the number of gates as the equivalent design that a human engineer created. Analyzing the circuit, though, gave no insight into how it worked. In one test, the circuit that was produced had a loop of a few gates that weren't connected to anything else in the circuit, nor to any of the inputs or outputs. It didn't appear to do anything -- but if it was removed, then the circuit stopped working.
The theory, of course, is that that loop of gates formed a 'resonance' component inside the circuit that was used to recognize the pulses. It's something no human would have done (because if you tried to use a different brand of PLA, it wouldn't work). But in this case, it represented a very efficient solution to the problem. Nobody could have predicted it, and it was still hard to figure out how the circuit actually worked.
It appears that the brain is just such an emergent system -- that our minds develop based on similar sorts of feedback loops that develop over time. The complexity of such systems can be many times that of the individual components, and just understanding how the components work can give you no real insight into what is really going on.
This, also, does not bode well for being able to predict the development of a mind.
Prediction versus Determinism
So what if we can't predict the outcome? We're still obeying fixed rules, and that means we could not have chosen otherwise.
The trouble, though, is that that definition of free will, in itself, outlaws determinism. By saying that we could not have chosen otherwise, it essentially demands that our choices be utterly random. The only time that free will exists, by that definition, is if our decisions are based on no reason whatsoever!
It's a circular argument. If you chose option (1), then your choice was fixed. If you decided to just be contrary and choose option (2), then your choice to be contrary was beyond your control. If you decided not to be contrary and choose option (1), then your refusal to be contrary was beyond your control. And so forth.
In other words, the only evidence they have that you were required to choose as you did was that you did choose as you did. Your inability to change the past after the fact is taken as evidence of a lack of free will. That hardly seems fair.
The usual response is that in order for us to have free will, then we must have a soul, something that is outside the usual definitions of space and time and not subject to the laws of matter and energy. That it must be metaphysical and distinct from our brains. Needless to say, religious types like this argument a lot. So do others intent on showing that there is a spiritual dimension to things. And the best part is, of course, that they don't have to show how it's possible, how this soul could work -- because the instant we can show how it works, it becomes subject to physical laws, and thus, non-deterministic.
Another response, of course, is that there is no soul, and therefore our choices are not governed by free will -- and thus, it makes no sense to punish, for example, criminals for their actions. Is it morally right to punish those who make wrong decisions, or reward those who make right ones, if there's no free will? This is used as the argument for various kinds of socialist agendas.
My Response
There are basically three rebuttals I can give against this argument, which, as I said, is employed by both sides, each anxious to prove their point.
Practicality
Because of the following:
- We certainly believe we have free will.
- Other people appear to have free will.
- We cannot, in practice, predict much of anything about what somebody will think based on deterministic priniciples,
It makes sense to say that, in effect, it doesn't really matter if we have free will on the metaphysical, esoteric plane. We certainly seem to act that way, and while certain external events affect our choices (e.g. I don't tend to eat a lot of food when I'm not hungry), and yes, certain events on our past may make the 'right' choices easier or harder, until such time as science can come up with a good, solid mechanism for predicting people's thoughts, acting as if people have free will is probably the simplest course.
And, based on what I said above, the trick of figuring out what people will think based on scientific principles may be a lot harder than it might first appear. The brain is a chaotic, emergent system with at least some degree of quantum effects. Don't expect to see any solutions to this problem anytime soon. We may be able to identify influences, but that's a long way from being able to say that we can 100 percent predict anybody's thoughts.
In addition, one must also consider that even if we do not, in the metaphysical sense, have true 'free will', having a system of rewards and punishments is certainly a powerful incentive structure that affects people's choices. Whether or not we have free will is irrelevant to the fact that, were this structure torn down without something that we know would work better to replace it, the results would probably not be desirable.
Execution
This is a fundamental question that goes to the heart of free will, and the definition that gets used. I'm going to argue here that the definition that is being used is flawed, in the sense that it doesn't mean what people think it's supposed to mean.
To begin with, let us talk about the nature of a decision, of a choice. Let's say we program a computer in such a way that a light will turn off if a number fed to it is greater than or equal to zero, or on if it is negative. This is a very simple rule, and in fact can be solved without the use of much of any components (hint: for standard integers, wire the most significant bit to an LED in series with a 100-ohm resistor to control current).
Feed this thing numbers, and the light will obediently turn on or off. Has a decision been made?
This is arguable. It's the same level of "decision" as a light bulb "deciding" to turn on or off when a switch is flipped.
However, let's extend the definition. A decision means that something must alter its behavior based on the outcome of that choice.
Let's make the computer more intelligent. Let's assume that if it receives a negative number at its input, then it will begin to flash a series of lights from right to left. A positive or zero number causes it to flash the series of lights from left to right.
Now, did it make a decision? It has altered its behavior based on an external input. Did it make a choice?
By the old definition of free will, it did not. Its programming was such that it could not have chosen otherwise. Therefore, it did not choose. Why? Because, based on the programming, the other choice would have been incorrect.
Let's assume that instead, what the computer does is subtract two numbers it is fed. One is stored in memory, and represents a temperature. The other comes from a thermometer. Every so often, the computer subtracts the second number from the first, and if the result is negative, it sends a signal out that, in turn, inserts control rods to shut down a nuclear reactor. In other words, if the temperature recorded by the thermometer is above a fixed value, shut down the reactor.
Now, let's assume that those events do indeed happen. Did the computer decide to shut down the reactor? Once again, the advocates of the old definition of free will would say, no. The computer did not make a decision.
But I argue that it did, very definitely. The decision may have been hardwired, but the simple fact is that a device changed its behavior based on a series of external events. Whether or not it could have chosen otherwise is irrelevant. An input was received, and based on that input, the computer changed its behavior. That is all that is required.
Likewise, if you punch two numbers into a calculator, does it truly add those numbers? In reality, the circuits go through a series of oscillations based on their design, and once they settle down they control a series of LED elements (or LCD elements) that we interpret as a number. I argue that even though the device is following its constraints, it nonetheless took two pieces of information and combined them to reach a result. It added the numbers, and it made a decision.
Now, I am not arguing, of course, that computers have free will. But what I am saying is that just because something is deterministic does not mean it cannot make choices. And when you are dealing with a system that is chaotic, emergent, and potentially exhibits quantum randomness, it is capable of integrating a lot of inputs, transforming them in strange and difficult-to-predict ways, and changing its behavior based on them. And if it isn't free will, then it's something so close that we can't tell the difference.
Useless Definition
Finally, I would argue that the traditional definition of free will is such that anything can be included or excluded at the whim of the person arguing. That makes it a very poor definition, because one criterion of a useful definition is that a third party should be able to tell you whether something does or does not fit into a category based on its definition.
The definition of free will as "the ability to have chosen otherwise" applies, in effect, to nothing. The only things that, by that definition, would undeniably fit would be hypothetical entities that could, on a whim, reverse time and change their decisions, and find a way to provably show you that this had, in fact, occurred.
In fact, amusingly enough, the only things that we know of in this world that fit the criterion are those objects small enough to be subject to the rules of quantum mechanics. In certain cases, an electron can retroactively change its spin in such a way that that decision seems to propagate backward in time, retroactively. Discussing these experiments and what they actually mean is well beyond the scope of this article, but it makes for fascinating reading.
Humans, though, like other macrophysical objects, make decisions based on their perceptions. If you blindfold me and then tell me to guess a card you are holding up, my "decisions" will be no better than chance. And I fail to see how a disembodied soul could do any better. Having the soul outside the body doesn't even solve the problem of free will -- after all, one could argue that the soul's decisions, having been made, could not have been otherwise. One just can't involve the rules of science to try to prove it.
It all comes down to the old question of proving sentience. How can we determine if a computer is sentient, an alien race, or an animal? Come to think of it, how can we prove that any other humans are conscious and experience a sense of self like our own? We assume that they do -- they are people, we are people. But it was not always the case. How long ago was it that some people regarded those of a different race as subhuman and lacking true feelings?
The best definition of free will comes from Alan Turing, widely regarded as the father of computer science, who devised an experiment to determine if a computer program could be regarded as sentient. Basically, you connect somebody via a terminal to the program on one hand and to a real person on the other, and you have the person text message back and forth, with suitable delays as needed. If the person can't tell which is the computer and which is the person (or if a good-sized group of people cannot, on average, do better than chance), then we must regard the computer program as sentient.
There is no other objective criterion. If it acts sentient, then we must consider it to be. And likewise, if it appears to exhibit free will, then we must consider it to possess it.
Implications
This basically applies to anyone who would like to use the free will argument to prove something and recommend policy decisions based on it. It applies to religious types who want to try to show that the soul exists outside our bodies, and thus proves the existence of God. It applies equally well to religious types who want to say that our having free will contradicts God's plan. And it applies to those left-wingers who want to make a case against punishing crimes on the basis that the criminal could not have chosen otherwise.
And finally, it also defies anyone who wants to defend indoctrination programs, whether for conservative or liberal values, who uses the defense that all behavior is programmed and there is no free will, so it's not about whether to brainwash but simply the content of the brainwashing that is under debate.
