Mark Balaguer - Free WIll as an Open Scientific Problem

In June, Neil Levy posted to the Garden a draft chapter of Mark Balaguer's new book "Free Will as an Open Scientific Problem," MIT Press (December 2009).

I bought a copy of Mark's book and read it closely. It has a new introductory chapter, but chapters 2 through 4 are expanded, and in many places altered, versions of papers that Mark previously published in the Southern Journal of Philosophy, Noûs, and Synthese.

Gardeners had a chance to read and comment on chapter 2 - "Why the Compatibilism Issue and the Conceptual Analysis Issue Are Metaphysically Irrelevant." Balaguer maintained that the question whether free will exists  - or of what kinds of free will exixt - was independent of the question of what free will is, i.e., whether it's Humean freedom, Frankfurtian freedom, or Libertarian freedom, for example.

In his Response to Balaguer, Levy criticized that view. He agreed that it has been a common mistake to think that we could do metaphysics just by doing conceptual analysis. But it is more than mere semantics to insist on a working definition of free will, which seems to Levy necessary to make progress on whether that free will exists or not.

Levy argued that conceptual analysis without an investigation of the real causal structure of the world is just a kind of folk psychology. But metaphysics without conceptual analysis is blind to the world it would investigate.

Balaguer urged us to wait for his chapter 3 to understand his new reformulation of the problem of free will, and his own libertarian view, which combines indeterminism and non-randomness.

He promised to explain how indeterminacy could actually generate the non-randomness, or procure it, or enhance it, or increase it, or something along those lines.

In the arguments about whether indeterminism diminishes or increases agent control, Balaguer, like Robert Kane, is an optimist.

In his Noûs article (A Coherent, Naturalistic, and Plausible Formulation of Libertarian Free Will (Noûs 38:3 (2004) 379–406), Balaguer began with a variation of the standard argument against free will when he says:

Thus, if our decisions are appropriately non-random, then they couldn’t possibly be undetermined. Therefore, libertarianism is simply incoherent: it is not possible for a decision to be undetermined and appropriately non-random at the same time. (p.379)

He then develops and extends Robert Kane's idea of "torn decisions."

These are decisions that require significant "effort" (C.A.Campbell) or what Kane called "self-forming actions" (SFAs).

Balaguer's model and Kane's model are "restrictive," a term coined by John Martin Fischer to describe Peter van Inwagen's claim that only a tiny fraction of our decisions and actions could be free actions.

For van Inwagen, free will is restricted to those which have closely balanced alternatives (the ancient problem of the liberty of indifference).

For Kane, it is those important decisions that provide us with what Kane calls ultimate responsibility or UR. They are those moments in which are character is formed. Later decisions made consistent with our character and values can then be traced back to these "self-forming actions."

Balaguer claims we that we may make many torn decisions a day than Kane admits.

But Kane argues that SFAs of the several kinds that he distinguishes (some need not be not moral decisions) are far more numerous in everyday life than we suspect and in his view occur regularly on a daily basis. So Kane is much closer to Balaguer than to van Inwagen.

Balaguer's decisions are still restricted to cases where reasons for the alternatives are closely balanced. The ancients called freedom in such cases liberum arbitrium indifferentiae. To prove that only humans had such a freedom, they denied it to animals in the classic example of Buridan's Ass.

Balaguer says,

A torn decision is a decision in which the person in question (a) has reasons for two or more options and feels torn as to which set of reasons is strongest, i.e., has no conscious belief as to which option is best, given her reasons; and (b) decides without resolving this conflict —i.e., the person has the experience of "just choosing". (p.382)

He defines L-freedom, in the case of these torn decisions, in terms of "appropriate non-randomness" and authorship and control as follows:

L-freedom is defined as the ability to make decisions that are simultaneously (a) undetermined and (b) appropriately non-random. Much needs to be said about what appropriate non-randomness amounts to, but for now, let me just say that the central requirement that a decision needs to satisfy in order to count as appropriately non-random is that of having been authored and controlled by the person in question; i.e., it has to have been her decision, and she has to have controlled which option was chosen. (p.382)

Balaguer knows that he is open to the criticism that he makes indeterminacy the direct cause of actions. He quotes Daniel Dennett that

"It would be insane to hope that after ... deliberation had terminated with an assessment of the best available course of action, indeterminism would then intervene to flip a coin before action (1978, p.51)"

Balaguer's basic and original claim is that a decision that is in not causally influenced at the moment of choice by any prior or external events must have been made by the agent alone because "nothing made him do it." When we combine this lack of external causation with conscious intentional purposefulness, we seem to get authorship and control, he says.

He confronts two objections.

Objection 1) If a decision is undetermined, then it isn't determined by the agent's reasons or character, which seems like less control and authorship.

Objection 2) The "luck" objection.

So what is new in the book, and especially in chapter 3, "Libertarianism Reduces to a Kind of Indeterminacy?"

Mark promised to Gardeners he would explain his “appropriate non-randomness.”

On page 10 of his new introductions he adds a new clause b)

Libertarianism is the view that human beings are L-free, where a person is L-free if and only if she makes at least some decisions that are such that (a) they are both undetermined and appropriately nonrandom,  and (b) the indeterminacy is relevant to the appropriate nonrandomness in the sense
that it generates the nonrandomness, or procures it, or enhances it, or increases it, or something along these lines.

In the Noûs paper, Balaguer argued that undetermined decisions simply are appropriately non-random. In the book, he argues that the indeterminacy actually generates the non-randomness. This relevancy is difficult to understand (for me at least) and I missed it on first reading.

In chapter 3, Balaguer expands his idea of an appropriately nonrandom decision, defining it as a decision that is authored and controlled by the agent.

Balaguer notes that from many alternative possibilities an agent might narrow down the choices for prior determining reasons, but that might leave some of the possibilities “tied-for-best.”  It is choosing among these “tied-for-best” options that is “wholly undetermined at the moment of choice.”

He gives a specific example as choosing to have dessert after dinner for a prior reason, but leaving the particular dessert undetermined. 

Balaguer then (p.92) confronts two objections. 1) the "luck" or “chance” objection.  and 2) if a decision is undetermined, then it isn't determined by the agent's reasons or character, which seems like less control and authorship.

Contrary to most commentators (van Inwagen, Fischer, Mele, Clarke, G. Strawson, et al.), Balaguer maintains that even if the agent's decision were to be randomly distributed (if the world could be rewound and the same circumstances were replayed - following van Inwagen's thought experiment  - and the results were random), this would simply show that it is the agent who makes the choices (p.92-94). If there were a pattern in the decisions, it would imply a hidden cause.  For Balaguer, randomness is evidence that no prior cause was involved! So we ourselves must provide the cause.

He then denies the familiar idea that “the looser the connection between the agent’s reasons on the one hand and her decision on the other, the less authorship and control she has over the decision” (p.95).

Balaguer's basic idea seems to be that the agent had good reasons for winnowing down her alternatives to the "tied-for-best" options, so whichever of these she chooses, it can be considered a conscious, intentional and purposeful choice.

He then (p.96) offers two theses why "Wholly Undetermined Torn Decisions" (he calls this TDW-Indeterminism) procure, increase, and enhance authorship and control.

Thesis 1) Such decisions, he says “provide as much authority and control over them as we could possibly have.”

Thesis 2) if the decision were in any way causally determined , he says we would “have less authorship and/or control.” 

(Does he include reasons, character and values here, as Bob Kane does with his self-forming actions (SFAs)?

These two theses, he claims, entail that TDW-Indeterminism is freedom-enhancing, i.e., procuring or increasing authorship and/or control.

Since any less indeterminism (e.g., prior reasons) would reduce control, he argues that more indeterminism must increase it. He makes this synonymous with generating and procuring  freedom. (Does this make sense?)

One interesting new item is that Balaguer says (p.123) that readers might have assumed from arguments in the book thus far that L-free decisions must be wholly undetermined and appropriately non-random (he calls them "type-1" decisions).

But, he says (p.123), "I never said that." He only claimed that if decisions are undetermined and appropriately non-random, then they are L-free.

So a major Balaguer novelty is his argument that even actions that are causally determined by reasons ("type-2" in the article, "type-5" in the book) can now be regarded as "L-free" (libertarian-free). The argument is that if he has established that agents are capable of "type-1" (undetermined and appropriately non-random) decisions, even decisions that are "determined" by reasons could now be regarded as L-free.

(It's not clear to me how the prior reasons that previously reduced authorship and control are now brought out as L-free themselves.)

Balaguer notes that he again differs from Robert Kane, who says that if our reasons and motives even partially cause our decisions, then they are not free, unless the reasons in question were caused by prior undetermined L-free choices (SFAs). On the contrary, Balaguer says that if an agent is L-free, and makes many undetermined L-free decisions every day, then her decisions that are caused by her reasons can also be called L-free.

In his Noûs article, Balaguer made the common mistake among compatibilist philosophers of claiming that R. E. Hobart's classic 1934 article asserted that determinism was required for free will. Like many others including Philippa Foot in 1957, Balaguer misquoted Hobart's title - "Free Will as Involving Determinism."

I think that he also misunderstands Hobart, when he says (p.6), "most compatibilists endorsed determinism. Some of these philosophers (Hobart for sure and arguably Hobbes and Hume as well) also held that freedom requires determinism."

Hobart's correct title is "Free Will as Involving Determination," which he defines only as the idea that reasons, character, values, etc are determining factors in our free decisions. Neither he nor Phillipa Foot argue for determinism in the sense of predeterminism. Both think that chance is real

Balaguer (or his fact-checking editors) corrected Hobart's title in the book's bibliography. And in new material in the book, Balaguer argued, as had Hobart and Foot, against predeterminism, which is of course the real problem that is solved by some indeterminism in the world.

Hobbes and Hume did think they had reconciled free will with determinism, The hard problem is to reconcile free will with indeterminism. Balaguer attempts to do this with his claim that indeterminacy in the torn decision enhances "appropriate non-randomness."

In his Noûs article, Balaguer had distinguished only type 1 and type 2 decisions. Type 1 are "a) undetermined at the moment of choice and b) appropriately non-random." Type 2 are "determined by the agent's reasons for choosing." In his book (p.122-3), Balaguer distinguishes several more types of decision, including "Buridan's-ass decisions" that are wholly undetermined, "Maybe decisions" that are spontaneous and "not teleologically rational," and decisions in which the agent is "leaning toward one or more options."

With ideas that resemble Peter van Inwagen's Consequence Argument and Ted Honderich's view of  determinism as a "black thing", Balaguer says

"The idea that determinism might be true can seem disturbing and depressing to us, and the reason, I think, is that it can seem to follow from determinism that we are something like puppets. If it was already determined before any of us were born that our lives would take the exact course that they in fact take, then it can seem that we don't have free will in any interesting or worthwhile sense."

In Balaguer's final chapter, "No Good Arguments for or against Determinism," he argues that there is no good evidence from empirical science in favor of determinism (quantum mechanics clearly denies it).  He also notes that logical arguments alone cannot establish an empirical truth and therefore indeterminism is an open question. And since he argued in chapter 3 that the right kind of indeterminism would generate L-freedom, he concludes that free will is indeed the "open scientific question" of his title.

Please see more extensive remarks on Balaguer's work in his web page on Information Philosopher.

http://www.informationphilosopher.com/solutions/philosophers/balaguer

I want to thank Mark Balaguer and Robert Kane for critical comments on earlier versions of this review.

Comments

[Bob Doyle]

Bob Kane has asked me to post the following note to me clarifying his views on free will relative to those of Mark Balaguer and Peter van Inwagen. No doubt I have misstated some elements of Kane's position and/or Balaguer's or both. I hope that Balaguer will also clarify his views for the Garden.
_____________
Dear Bob,

In your recently posted review of Balaguer's Free Will as a Scientific Problem, you have the following paragraph:

Balaguer notes that he again differs from Robert Kane, who says that if our reasons and motives even partially cause our decisions, then they are not free, unless the reasons in question were caused by prior undetermined L-free choices (SFAs). On the contrary, Balaguer says that if an agent is L-free, and makes many undetermined L-free decisions every day, then her decisions that are caused by her reasons can also be called L-free

If this is an accurate account of what Balaguer says, it is misleading about my view in a number of ways that I would like to clarify for potential readers of his book or your review. (I should add that I agree with much that Balaguer says in his book, but not all.) How the paragraph is misleading is made clear by the following comments which I have expressed in a number of different ways in several papers concerning three kinds of "free acts" which I regularly distinguish:

(1) acts done voluntarily, on purpose and for reasons that are not coerced, compelled or otherwise constrained or subject to control by other agents.

(2) acts [free in sense 1 that are also] done “of our own free will” in the sense of a will that we are ultimately responsible (UR) for forming.

(3) “self-forming” acts (SFAs) or “will-setting” acts by which we form the will from which we act in sense 2.

Acts of type 1, as I understand them, are compatible with determinism. One can act freely, in the sense of voluntarily, on purpose and for reasons, without being coerced, compelled or otherwise constrained or controlled, even if determinism should be true and even if one’s act is determined. Type 1 freedom is thus a compatibilist freedom. (Aristotle generally called type 1 acts “voluntary” by which he meant acts that were done on purpose or willingly without being coerced or compelled.)

Free acts of types 2 and 3 by contrast, as I understand them, are incompatibilist or libertarian free acts. They could not exist in a determined world. But only acts of type 3 have to be undetermined. Acts done “of our own free will” of type 2, on my view, may be determined (though they need not be) and may even be such that the agents could not have done otherwise. In what sense then are free acts of type 2 incompatibilist or libertarian free acts? The answer is that while acts of type 2 may themselves be determined, they could not exist in a determined world and hence their existence is incompatible with determinism because they presuppose other acts (of type 3) that are not determined. Often in everyday life we act of our own free will (type 2 free acts) in the sense of a will already formed. Our characters, motives and intentions are such that, we could not have done anything else then and there voluntarily and rationally. But on such occasions, the will (i.e., character, motives and purposes) from which we act is “our own free will,” to the extent that we had a role in forming it by earlier acts of type 3 that were not determined and with respect to which we could have voluntarily and rationally done otherwise.

It is important to recognize that all three of the above acts, including type 1 acts, are legitimate kinds of freedom. The word “freedom” does not have a single meaning (no surprise there for such a much used term). And, though I am a libertarian about free will, I have always conceded that type 1 freedom of the compatibilist kind is a significant kind of freedom. I have merely insisted that there is an “additional freedom worth caring about that is not compatible with determinism,” and it is “what was traditionally called ‘free will.’” Freedoms of all three types are thus significant freedoms, as I see it. The difference is that freedom of type 1 is freedom of action, while freedom of types 2 and 3 is freedom of will.

In addition, the three are related. Type 3 acts (“self-forming acts” or SFAs, as I call them) are also free in sense 2 (they are ultimately responsible acts of free will, albeit of a special kind). And acts of types 2 and 3 (acts of free will) are also free acts of type 1 (they must be voluntary, uncoerced, non-compelled, etc.). So freedom of will (of types 2 and 3) is a kind of freedom of action (of type 1), albeit a special kind.

I also think free acts of all three kinds are common in everyday life (including type 3 SFAs). That is why I reject the term “restrictivism” for my view. The term “restrictivism” was first used by Fischer to describe a view put forward by van Inwagen in the late 1980s according to which only acts of type 3 were really “free” acts, though determined acts of type 2 could be morally responsible acts. I was the commentator on an earlier version of van Inwagen’s 1989 paper “When is the Will Free?” delivered at an APA meeting. At that session, I agreed with van Inwagen that a distinction between (what are here called) type 2 and type 3 acts was important for understanding libertarian free will. But I objected, first, that acts of type 2 could also be called libertarian “free” acts (acts of free will) as well as responsible acts and, second, that type three 3 acts were far more common in everyday life than his paper implied."

[Mark Balaguer]

Bob,

Thanks for taking the time to write up this piece on my book. I’d like to correct a few points where I think you get my view wrong…

1. Near the beginning, you say that my view is “restrictive” in the sense that it entails that only a tiny fraction of our decisions and actions are free. I don’t think that’s right. First of all, I claim that we make many torn decisions every day and that if these are undetermined in the right way (wholly undetermined, or TDW-undetermined), then they’re L-free. Second, I claim that decisions that are fully determined by reasons for actions can be free. So my view is in fact not restrictive.

2. There is an important difference between what I call torn decisions and decisions involving a liberty of indifference, or what might be called Buridan’s ass decisions. In the latter, the reasons for choosing the various tied-for-best options are the same reasons. E.g., if two elevator cars arrive at the same time, you have a reason to get into both, but it’s the same reason. In a torn decision, on the other hand, the reasons for the various tied-for-best options are different, and this is what leads to feeling torn. In a Buridan’s ass decision, we don’t care which option we choose, but in a torn decision we do. E.g., if you have a great job offer in a city you hate, you have completely different reasons for picking two different options, and you might feel utterly torn about what to do. But you would most certainly care about the choice. But, again, in Buridan’s ass decisions, we don’t care which option we choose because the various tied-for-best options are essentially equivalent to us. This is why these decisions are associated with the term ‘liberty of indifference’. And this is why, in my view, torn decisions are more interesting and important than Buridan’s ass decisions.

3. At one point you say that “Balaguer knows that he is open to the criticism that he makes indeterminacy the direct cause of actions.” I’m not sure what this means. An action can be determined or undetermined, but presumably it doesn’t cause determinacy or indeterminacy. In fact, if it caused its own indeterminacy, that would seem to involve something very strange--backward causation, or something like that.

4. At one point, you say that I define an appropriately nonrandom decision as a decision that is authored and controlled by the agent. But that’s not really right. I say that authorship and control are the most important components, but there other components as well, e.g., rationality and some plurality conditions (in particular, plural authorship, control, and rationality).

5. You say that on my view, choosing among tied-for-best options is wholly undetermined at the moment of choice. But I don’t commit to that claim. What I say is that this is what’s needed for torn decisions to be fully L-free. But on my view, it’s an open question whether any of our torn decisions really are wholly undetermined in this way.

6. Regarding the rollback argument: My claim is not that randomness is evidence that no prior cause was involved. Rather, my claim is as follows: If Betty is torn between A and B, and if we replay her decision 100 times and she chooses A 50 times and B 50 times, then that would be evidence that nothing external to Betty’s conscious reasons and thought came into the picture and causally affected her choice. The thesis that no external cause comes into the picture predicts a 50-50 distribution, so if that’s what we found, that would be a friendly result to libertarianism. If, on the other hand, Betty chose A 100% of the time, that would be evidence that something external to her conscious reasons and thought was involved in the causal story.

7. You say at one point that I say that “any less indeterminism (e.g., prior reasons) would reduce control”. This isn’t quite right. If prior conscious reasons entered the picture and swayed Betty to choose A over B, that wouldn’t necessarily reduce control. My claim is that if Betty is making a torn decision, then external causal influences would diminish control. But if she’s making a torn decisions, then (by definition) her conscious reasons are neutral between A and B. So if prior conscious reasons push her to A, that’s simply not a torn decision. On my view, what diminishes control are causal influences that are external to the agent’s conscious reasons and thought.

8. With respect to the paragraph about Kane and the idea that partial causation by reasons destroys freedom: I think Kane and I agree that this is wrong. He can correct me if I’m wrong, but I think we would both say that (a) in torn decisions, our reasons do partially cause our choices because they are causally relevant to which of our options are tied-for-best; and (b) choices that are fully determined by reasons can still be free.

9. In the last paragraph about Hobart, you say that I “argued…against predeterminism.” That’s not really right. I argue that there are no good arguments for determinism. But I also think that there are no good arguments for indeterminism, so I don’t argue against determinism.

Relatedly, you say a few paragraphs later that “quantum mechanics clearly denies” determinism. It’s not clear to me if you meant to attribute that claim to me or if you were asserting it yourself. In any event, I’d like to emphasize that I do not think that quantum mechanics (QM) provides us with any good reason to reject determinism. QM has probabilistic laws. But it’s possible that there are deterministic mechanisms underlying those laws. Bohm has given a theory that involves this kind of determinism. But the problem is that, as of right now, there is no good evidence for or against Bohm’s theory. And more generally, there is no good reason for endorsing either a deterministic or an indeterministic interpretation of QM. Philosophers often go around saying that there is good reason to favor an indeterministic interpretation, but I think this is pretty clearly false. I talk about this a bit in the book, in section 4.4.1.

10. Finally, and this is more amusing than serious, I didn’t mean to name non-teleologically-rational decisions “Maybe decisions”. The purpose of the ‘maybe’ was to say that we might make decisions like this but we might not.

[Lyndon]

Mark, quick question, if you don't mind. On number two above, can we intuit that the mechanism for choosing between "torn decisions" is the same/similar as for "Buridan Ass Choices"? And, if the same, is what makes "torn decisions" more interesting only the outcome, the fork, that they play in our lives and not necessarily any controlling power that we have within such decisions?

Also, I assume, under your analysis, our caring more about a "torn decision" can not, does not, lead to bettering our ability to choose between the options, given the definition of a "torn decision."

Given this, although an interesting fact about decision making, I do not see how it is going to help in the free will issue.

Thanks,
Lyndon

[Mark Balaguer]

Lyndon,
Sorry it took me a few days to respond. I didn't see this until today. Anyhow, I guess I want to say that it's an empirical question whether the mechanisms of choice are the same in the two cases. It may be that because we care about torn decisions, this leads to a different sort of process. But I think the two cases have this much in common: In both cases, I think it can be argued that if they're undetermined in the right way, then we author and control them, and they are libertarian free. But I think that in Buridan's ass decisions, we don't really care whether we control them, because we don't care which option is chosen. Finally, as for why torn decisions are more interesting, I think this is just because we care about these decisions, and we don't care about Buridan's ass decisions. And the reason we care about the former but not the latter is that in torn decisions, the tied-for-best options and reasons are substantively different.

[Lyndon Page]

"In both cases, I think it can be argued that if they're undetermined in the right way, then we author and control them, and they are libertarian free."

I am still struggling with why the mechanism of control we use in a Buridan Ass choice would be beneficial to us. The ability to blindly/randomly choose between Door 1 or Door 2, given no indicator of what is behind them, does not seem to be a property of Humanness with any significant importance, just as this ability is not that big of a deal to a non-Buridan Ass, besides allowing him not to starve.

Perhaps what I am getting at, is that the mechanism of choice, of authorship, and control in these cases does not provide a significant and meaningful account of libertarian free will. Our choosing one choice in a "torn" decision is not because of the reasoning behind either choice, but is chalked up to the "flip of the coin" between two equal choices with equal reasons to choose either one. Even if undetermined, the choice between the two seems to be of little merit, although the difference in outcome could be great. Our authoring the outcome of such a decision, choosing one over the other, does not rest in the reasoning (good or bad) that we used to support the decision. It was only luck that we made the one choice and not the other. Why would we hinge authorship of our character, of our life, on such a mechanism (and then claim it is of upmost importance)?

[Mark Balaguer]

Hi Lyndon. A couple of things. First, I want to resist running together Buridan’s ass decisions and torn decisions. I think they’re very different in connection with some of the things you mention, e.g., importance and significance. We usually don’t care at all which option is chosen in a Buridan’s ass decision, and so I wouldn’t deny any of your claims about the significance of the issue there. I do think that it might turn out that we have libertarian freedom in connection with Buridan’s ass decisions, but I don’t think it MATTERS whether we do. Our Buridan’s ass decisions are L-free iff (i) they’re undetermined and (ii) we author them, control them, etc., and (iii) the indeterminacy generates (or procures or some such thing) the authorship, control, etc. This MIGHT all be true, and so it might be that our Buridan’s ass choices are L-free. But I don’t particularly care if it’s true, and I don’t think it matters, for the simple reason that in Buridan’s ass decisions, it doesn’t matter which option we choose.

Torn decisions are different. Since the options can be significantly different (e.g., stay in Mayberry or move to New York), we care which option is chosen. And for whatever reason, we WANT to author and control our choices. You seem to think that the reason we want L-freedom is that we want it to be the case that our REASONING determines our choices, so that our choices will have MERIT, or VALUE, or something like that. But I don’t think that’s the central issue. I think the reason we want L-freedom is that we want it to be the case that we author and control our choices. I don’t think it has much to do with our reasoning, or even with merit or value or anything like that. I think it’s about authorship and control. We just don’t want to be puppets. Choosing randomly (among my reasons-based tied-for-best options) is OK, so long as it’s ME who’s choosing randomly. This isn’t to say that we don’t value good reasoning. It’s just that (for me, anyway) the worry about L-freedom isn’t primarily about our reasoning. It’s (primarily) about authorship and control.

Now, of course, there are other worries about free will that are related to reasons. E.g., we want to be reasons responsive. But I think the worry about L-freedom is just a different worry.