Dennett begins by claiming that the free will problem is sort of a faux-problem. According to Dennett, it is really a whole host of smaller, sometimes unrelated problems that are brought together and unified by one name -- "the free will problem". Dennett's take on the faux-nature of the free will problem, however, is not his aim here. Rather, he points out on pg 287 that
"Since the case for determinism is persuasive and since we all want to believe that we have free will, compatibilism is the strategic favorite, but we must admit that no compatibilism free of problems while full of the traditional flavors of responsibility has yet been devised".
What Dennett is saying is this:
Compatibilism is his preferred strategy, though it is recognized that the theory has problems.
Since it has problems, why should someone pick it? Well, one would pick this strategy if and only if it was the best theory available. So compatibilism can still "have some problems" and be the solution we should pick, if it can be shown that the alternatives are worse off.
What are the two alternatives? Hard determinism and Libertarianism (note here that Dennett does not acknowledge that what he calls "libertarianism" has two routes -- 'indeterminism' and 'agent-causation'. According to the first, random indeterminacy is required for free will, according to the second, it is immanent causation -- as per Chisholm or Aristotle -- that is required. In this article Dennett associates "libertarianism" with "indeterminism").
Dennett starts by suggesting that what we want to do is to lay out a case for free will. So that means that Hard Determinism as an alternative is ruled out immediately, since it denies free will entirely. So what is left is Libertarianism.
So we know that since Dennett prefers Compatibilism, his general strategy is going to be to try to show that Libertarianism is worse off than Compatibilism as a theory.
II. The Challenge of David Wiggins
Wiggins is a proponent of libertarianism. Now we all know the standard argument for libertarianism. It is:
P1. All events are either causally determined or random.
P2. If an event is the result of free will, it cannot be the result of causal determination.
C1. Thus, if an event is the result of free will, it must be a random event. (from P1 and P2)
Then, as Wiggins notes, we wind up with this quandry (taken as the argument against libertarianism):
P3. If an event is random, then the event is meaningless (no one had any control over the production of that event).
P4. If an event is meaningless (no one had any control over the production of an event), then no one is responsible for that event.
P5. If no one is responsible for the event, then the event was not the result of an act of 'free will'.
C2. Thus, if an event is random, no one is responsible for it (from P3 and P4)
C3. Thus, if the event is random, then it is not the result of an act of free will. (from P3 and P5).
Clearly, both arguments are valid. So if the premises are true, the conclusion(s) cannot be false. As such Wiggins -- as a libertarian -- must deny the truth of some one or more of the premises. Which one?
Dennett points it out -- Wiggins denies the truth of P1, the premise that really gets the whole argument moving. The reason he denies P1 is because he believes that it is possible to have an event that is not causally determined, but yet intelligible. So there must be a kind of non-determined event which is not simply "random".
Essentially this is called claiming that there is a "false dilemma". A false dilemma says: "there are only two options, P and Q. Since Q not preferable, only P is left over as an option". The dilemma would be a "false one" if there were actually other alternatives other than P and Q (in which case rejecting Q would not result in having to pick P).
This is Wiggins' claim. P1 is a false dilemma -- it states that the only two options are "causally determined action" and "random action". So if the free will advocate denies that free will and causally determined action are compatible, the only thing left is random action. But it is acknowleged that random action is meaningless. As such, it cannot ground free will. Thus the only way to ground free will is in causal determined action (compatibilism).
Wiggins argues that the dilemma is false because there is a third option; some events are not causally determined and not meaningless. If so, then the failure of random meaningless action to ground free will would not lead us immediately to compatibilism.
Dennett counterargues that in his article Wiggins raises this possibility but does not give any details. How is the third option possible? What would it look like?
In what follows Dennett tries to make a case for non-causally determined but intelligible action. If he can make such a case, he will have "given libertarians what they say they really want".
III. Making Non-Determined Action Intelligible
As Dennett sees it, the libertarian actually needs two things to get his/her argument working.
(1) Quantum indeterminacy must actually cause macroscopic indeterminacy in some cases
(2) Some of these non-determined macroscopic events must be intelligible
For the sake of the argument (for now anyway), Dennett grants that (1) is true. So the task is to make a compelling case for (2).
(a) First Pass -- External Indeterminacy: The Answer-Box
1. Physics vs Intentionality
Dennett, following Wiggins, introduces two ways in which an action can be 'meaningful':
For an event to be meaningful with respect to (a), it would have to be "predictable". The problem here is obvious -- if there is an indeterminate event in the physical sequence, then the whole sequence could not be predictable. This is usually what it meant when philosophers argue that libertarianism reduces to meaninglessness -- they equate the event simply with the physics. Obviously no event that is not deterministic can be intelligible in this way, by definition.
It is this connection, however (between libertarianism and intelligibility at the level of physics) that Wiggins is opposing.
On the other hand, for an event to be meaningful with respect to (b), it simply needs to be "rational" or "understandable", or even generally "predictable". This is what Dennett says on 289. He says:
"Surely if human actions were predictable in the low-level terms of practical deliberation, they would be intelligible in those terms".
What does it mean to say "from the intentional stance" or "from the intentional point of view"?
Take the example of a person going to get a coke from the vending machine. There are two different ways to describe this act. One, I could give it the kind of description a physicist would give. I would talk about how atoms are coming together and bumping against other ones, or I could give a more macro description and talk about muscles and bones moving to and fro. Or, I could just say "He is going to the coke machine because he was thirsty". This is an intentional description. It makes reference to the subject's reasons for the behavior. To give an intentional description is just to make sense of what happened on the level of beliefs, desires, reasons. To make sense of a subject's behavior on a physical level is to explain physically what happened, with no reference to beliefs, desires, reasons, etc.
With this in mind, Dennett takes his mission to be (as stated on pg 289):
"I propose first to demonstrate that there is a way in which human behavior could be strictly undetermined from the physicist's point of view while at the same time accurately predictable from the intentionalist's level."
So Dennett thinks that what the libertarian wants is intelligible action that is not entirely causally determined. His strategy is to see how he can provide that by claiming that while the some acts would not be intelligible from the physical stance, they could be intelligible from the intentional stance.
2. The Answer Box
The answer box is meant to record a person's answers to simple questions. Here is the construction of the box:
1. It has two buttons, a YES and a NO button
2. It has two foot petals, a YES petal and a NO petal
3. It has a display screen divided in half, one one side reading "USE THE BUTTONS" and on the other "USE THE PETALS"
4. A truly random generator is hooked up to the display, so that it is undetermined which side will light up at any given time.
Now we draw up a list of questions, like "Is Texas bigger than Rhode Island?" and "Can fish swim?"
We sit the subject at the box. Before each question, the random generator lights up one side of the display screen indicating how the subject should record his/her answer.
Now, assuming that the subject him/herself is a deterministic system,
(a) Can the physicalist predict the subject's behavior?
No, because the indeterminacy affecting the box will make the subject's behavior non-predictable. The best the physicist can do is say "IF the petals light comes on, then..."
(b) Can the intentionalist predict the subject's behavior?
Yes. The intentionalist will say "the subject will answer 'yes' to questions 1, 3, 5, and 'no' to questions 2, 4, and 6".
However, Dennett says that the Libertarian will not be interested in this "help" that he has provided. The reason is that indeterminacy is outside the agent. The libertarian wants decisions to be free and not causally determined. In this case the decisions are entirely determined. So Dennett hasn't yet given them what they want.
(b) Second Pass -- The Campbell Soup Cans
First Dennett poses a question: "What exactly are we interested in predicting when we make an intentional prediction?"
Say I predict that given a certain person's beliefs, desires, etc., that he will go and buy a Coke from the vending machine. Should the prediction be more precise? Does it need to refer to the way in which the person will walk, or the way in which he/she puts the quarters in the machine, or which Coke button is pushed?
Assume that the subject goes into the supermarket to buy some Campbell's Tomato soup. The subject arrives in the right aisle, and looks up to find 1,000 identical Campbell's Tomato soup cans. Each of the cans is qualitatively identical to the next one, and each can is equidistant from the subject's hands.
Now what? What further reason will suffice to pick one can over another? Dennett points out that no such reason exists -- each of the cans is identical to the next, so there is no good reason to pick any one of them over some other.
Now, Dennett says, we can install indeterminacy inside the agent in a way which preserves (a) inability to predict at the physical level and (b) intelligibility at the intentional level. What we do is install it such that when we are in a "Buridan's Ass" situation (when we are presented with two options each of which are essentially identical to the other) indeterminacy kicks in to determine the direction.
So as a result of indeterminacy I pick can 876, say.
According to Dennett, such a move will in fact maintain unpredictability at the physical level. The physicalist could not entirely predict what the subject will do. However, on the intentional level, the action makes perfect sense. The subject wanted a can of soup (that's what their beliefs and desires indicated) and so they grabbed on. It is irrelevant "which" can is picked -- all that is important is that the subject pick some can or other.
Will the libertarian be happy? No. Dennett says that they should be happy at least with respect to the fact that indeterminacy can be placed inside the agent and still get you intelligible action. But they won't be happy because the indeterminacy seems to be in the wrong place.
As Dennett says: what difference does it make to me to know that how my arm with the ax comes down to cleave the head of my victim is undetermined, but that the decision to murder the victim was determined?
So the libertarian will hold out for yet another solution.
(c) Pass Three -- Intelligent Selection
What the Libertarian really wants, Dennett thinks, is what he calls "intelligent selection". Here what we would have is the following:
(1) an (at least partly) random or arbitrary production of factors (beliefs, desires, etc)
(2) an intelligent meaningful selection or choice based on what was at least partially randomly laid out to choose from
Dennett (of course) has the answer.
Most of our decisions, Dennett suggests, are time-pressed. In other words, we really don't have forever to think and critically analyze all of our beliefs and desires relevant to the given decision that needs to be made. Instead, we "satisfice". In other words, we "settle" for a way to balance the time pressure with the desire to think through and analyze what is important to the situation. So we basically "do the best we can with what we've got and how long we've got to do it".
Dennett thinks we can install indeterminacy right here to give the Libertarian what he wants.
Let's assume that when subject X is trying to decide what to do in a given situation, X "satisfices". So X does the best he can with what occurs to him at the time. Now some of the beliefs that occur to X will be deterministically generated, and some will be indeterministically generated. So let's look at the example of Jones:
Jones is finishing his dissertation and must decide whether to take the job in Swarthmore or the job at the University of Chicago.
1. Jones "satisfices".
2. So although there are many beliefs that Jones has that could be relevant to making this decision, only A, B, C, D, E and F occur to Jones.
3. A, B, C are deterministically generated
4. D, E, F are indeterministically generated
Based on A - F, Jones takes the job at the U of Chicago. Right after the call, Jones realizes consideration G. If Jones had thought of G, she would have taken the job at Swarthmore.
What is important here?
(a) The behavior of subject X is physically unpredictable.
(b) The behavior of subject X is even intentionally unpredictable (unless one issues a conditional prediction).
(c) The behavior of subject X is intelligible.
Here the libertarian gets just what they want! The agent's actions are never totally predictable, intentionally or physically. But yet the behavior is intelligible.