In the book “Freedom Evolves,” Daniel Dennett takes up the task of providing an account of free will in a naturalistic world. His crown jewel is the discovery that determinism does not imply inevitability:
In this paper I take up two tasks. First, I will claim that if determinism is true, there is nothing we, or any agent, can avoid. Secondly, trying to understand why Dennett is mistaken, I will suggest that Dennett should have taken another path in the argumentation and declare that concepts such as “inevitability” and “avoidance” do not make any real sense, if we are determinists. I also want to nag about Dennett’s (ab)use of the concept of “complexity.”
Can we avoid determinism?
In this section, I aim to disprove the claim that avoidance is possible in a deterministic world.
First, I am going to provide a brief explanation of the analogy that Dennett gives us as a thinking tool to help us understand determinism-the three Dennettian concepts of different stances-and help us imagine how we can reconcile determinism and avoidance. Then I want to define determinism and to take any given event in the world as a test to see whether it is avoidable.
What is the Life World?
The Life World is a simulation game made up of simple pixels. According to some predefined (deterministic) “transition rules,” the pixels can turn on or off. Depending on the beginning configuration and the transition rules, the simulation can develop in a variety of different ways. The interesting part is that we can adopt three different points of view regarding what happens during this game:
The physical stance is a description at the lowest level, dealing with the transition rules and the initial status (on or off) of the pixels.
The design stance is a bird’s eye view of some of the situations that develop during the course of the game. Some pixels group together, giving the impression of well-defined entities. At times, these entities give the illusion of movement, development and disappearance. Because the game becomes very complex, we can gain a better understanding from the design stance as opposed to explaining everything in terms of the transition rules and describing pixel positions.
At an even broader bird’s eye view, we could describe the shapes in the game as entities that have desires, beliefs, purposes, intentions. This is the intentional stance.
When designing worlds, the players of the game attempt to create beings that avoid harm. When they talk to each other, they speak only of beings, and generally adopt only the design stance or the intentional stance.
Dennett affirms that it only makes sense to discuss avoidance or inevitability when one adopts either the design stance or the intentional stance. At the physical level, this kind of talk would be improper. Dennett gets irritated by sentences like the following:
He continues with an explanation of why:
According to Dennett, any discussion of concepts like avoidance and inevitability belong at the design and intentional level and not at the physical level. But is this claim true in a deterministic world?
First of all, let’s have a simple definition of determinism. According to Dennett, we can define determinism as something like:
Starting with this definition let’s see whether any event can be avoided. Let’s say that next Tuesday there is going to be a terrorist attack. Is this event avoidable?
Staying at design level, I can describe this in a number of different ways. For instance, consider the following:
Description 1 (or D1)
“On Tuesday at 9:30am, John presses a red button which detonates an atomic bomb above London.”
But nothing keeps me from describing the event in greater detail, like the following example:
Description 2 (or D2)
“John takes a deep breath, thinks about all the innocent babies that will die, puts his index finger at an 45-degree angle above the red button and presses it with a power of two Newton.”
In fact, it does not matter which degree of detail I use to describe that event. I could even choose to describe the position of all the particles in the Universe, their energies, speed, direction of movement and whatever else may be relevant to an exhaustive description of that particular instant. I will name this exhaustive description GOD (Good old Okham’s Description).
If I say that I want to avoid the terrorist attack next Tuesday, there is no difference between the different possible descriptions of that event. If the descriptions are accurate, there should be no change in the truth of the description. D1, D2 or GOD should all possess the same truth value. They are interchangeable in my speech, and I would not be telling a lie if I chose to use one description rather than another.
It might not be within the scope of my powers to describe the world as with GOD, but that is irrelevant. I could choose some short cut formula or the sentence “the exhaustive description of the world at the moment when atomic bomb is detonated.” It does not matter-it is all about that very same instant.
Well, between now and GOD there are a finite number of other intermediate state descriptions. These state descriptions follow from one to another, if we assume that the laws of nature do not sneakily change in the meantime. What do I have to do to avoid GOD? Once we reach the GOD instant, it is too late, because I am not God. This means that I must change something between now and GOD in order to avoid GOD.
I cannot change the current state description, because it is already here and has already happened. Could I change the very next state description? According to my definition of determinism, I cannot. The transition rules (the laws of physics) determine exactly which state description follows the current state description. Thus, I cannot change the next state description. I cannot avoid it.
From the next state description, there will necessarily follow another state description that I am powerless to change or avoid. This process continues all the way up to the GOD, up to next Tuesday at 9:30am when London ceases to exist. That means that I am unable to avoid any state description of the world. Or, in other words, any state description of the world is inevitable, from the beginning of the Universe until such a instant as it finally perishes.
But how is that possible? Dennett sustained that there are avoiders in the world, even skillful avoiders, who avoid certain things. How could I come to a different conclusion? Dennett might object and say that I played unfair. I might have cleverly changed the stances, going from the design stance or the intentional stance to the physical stance. No, I would swear, I am still at the intentional stance: I am speaking of some kind of agent, myself, who is trying to prevent something from happening. It does not matter how I describe the event that I want to avoid.
In conclusion I repeat that some kind of mix between the design/intentional stance and the physical stance is possible. From the design/intentional stance it is perfectly acceptable to choose to avoid a physical state description of the world. But, no state description is avoidable.
Complexity works in mysterious ways
But where might Dennett’s argument go astray? In this section, I suggest that Dennett might have incorrectly assumed that complexity leads to freedom.
In anticipation of objections that his Life World example is not an example of “real avoidance,” Dennett says that we are “perhaps being misled by the simplicity of the imaginary examples… There is a contrast between simple, ‘hard-wired’ avoidance responses and fancier varieties, but you can’t use it to contrast real-world avoidance with Live world avoidance.” Well, I’m not misled by the simplicity of his example because I am a kind of fancy agent, and yet I am still unable to avoid any outcome in the “real-world.”
Perhaps is Dennett being misled by the complexity of the real world and the complexity of real-world agents, like humans. My vague impression is that Dennett firmly believes that complexity automatically leads to (increased) freedom. For instance, in Chapter Five, he describes simple life forms. Clams have some kind of simple on-and off- “switches” that respond to bumps on their shells by closing them:
I hope this not some kind of shell game where switches are being switched for freedom and complexity for explanation. I do not believe that complexity creates more freedom. I could build a simple hut or the Pentagon building, with an enormous difference in complexity. However, that does not mean that the Pentagon is dizzyingly more free that Uncle Tom’s hut.
I do not believe switches would make Uncle Tom’s freer, either. Imagine that I fill the switchless hut with tons of light switches, linked in parallel and series, in arrays that combine both sorts of links, forming larger networks. Is Uncle Tom’s hut freer than before? I do not see how it could be.
I don’t know what Dennett would reply. Maybe he would say that soulless huts are different, “in an important way”, from alive and kicking organisms. But I don’t believe complexity does any explanatory work in the organic world either. Take the DNA for instance, that is enormously more complex than simple organic molecules. I cannot imagine the DNA being freer in any way.
Dennett appeases us every two pages that we humans are enormously complex and some readers, in a moment when the light switch is dimmed, might inadvertently come to believe that complexity might lead to more freedom. At the same time, the reader might fall into another trap, which resembles the argument from ignorance fallacy. As we cannot understand such complex systems ourselves, the reader might believe, we do not know whether we are free, and Dennett might therefore be right. I hope we can avoid at least that.
Avoidance: if not shelly, than fishy?
In this section, I suggest that there is no real avoidance, that the concept itself is incoherent while being a determinist, and that we should learn to accept that.
Anticipating another objection about the Life World example, Dennett lets a toothless Conrad, the voice of the ombudsman, who seemed to me to be made of straw, declaim:
And Dennett corrects him immediately:
I have to admit that Dennett might be onto something. There is something fishy about the common use of words like “inevitable” or “avoidance” in a deterministic world. And the whole idea of changing an outcome is probably incoherent. But why not just admit that those concepts are incoherent and close the discussion? I would like to illustrate this with an example.
Imagine that there are two different asteroids, each of them 100 kilometers in diameter. From a great distance, both appear to be headed right for Earth. One would pass by at a safe distance. The other one, if nothing were to happen to stop it, would crash into Earth, destroying it. Based on our observations, we believe both asteroids would destroy the Earth. So we launch some kind of nuclear device that will explode both of them at a safe distance.
Is there any relevant difference between those two asteroids, speaking in terms of avoidability? Take first the one as an example, the one that would pass by. Does it make sense to talk about it in terms of avoidability? Is it avoidable? No, I do not think this kind of talk would make much sense. Now let’s talk about the other one, the one that would crash into Earth if nothing happened to stop it. Is this one more avoidable than the first one? I do not believe so. That is because we do not act on a so-called anticipated outcome. We act on our beliefs. There is nothing about either of the two asteroids that would make them avoidable or inevitable, for that matter. Neither of them has any extra attribute or quality, property or whatever else that would make it distinct from the other, just on the sole virtue of that property, a property that we might call avoidability.
The second asteroid looks much more like the first one: it has never crashed into Earth. It was never going to crash into the Earth because we can only say that about the things that actually crash into the earth. Exploding it does not do more than any collision between molecules, or between some bigger things in the Universe. There is a vast amount of collisions and forces in the Universe that change the movement of the things. I would not name all these changes as avoidance of what has not happened. The mighty nuclear explosion, it seems to me, is not different from that.
Therefore, it seems to me that all those concepts, of “inevitability,” “avoidance” and “changing an outcome” have something incoherent about them, if we want to think about the universe in a deterministic way. And I do not see any problem with that. There are more incoherent concepts, unimaginable concepts, paradoxes and such in human thought. Our thought world, the folk psychology, differs in many regards from the real world. Modern physics reminds us of that every day. Therefore, Dennett is posing an improper question when he asks Conrad: “But why do you insist that determined avoidance isn’t real avoidance?” I convey that, if nothing else, using the term “determined avoidance” is improper.
I don’t see any reason why Dennett would want to believe that avoidance exists at any cost, only because we have to have an extra concept that makes sense in our toolbox. A more natural step in the argument would have been for Dennett to simply declare the “avoidance” concept incoherent. Period.
Even if we adopt a design stance or an intentional stance, we cannot say that determinism does not entail inevitability. Dennett should have better concluded that the common use of terms like “avoidance” and “evitability” are somehow incoherent. A naturalistic Nature does not care what we think anyway.
Dennett, Daniel Clement. Freedom Evolves. (New York: Viking, 2003).
 Interview in Reason magazine.
All the other quotes are referring to the book [Dennett, 2003]