I’m puzzled by a feature of much of the scholarly discussion of the Epicurean swerve. So many of the discussants seem to be assuming that there must be a swerve corresponding to each human free action. I don’t see why. If indeterministic swerves occur, then every atomic motion – not just the times when the atom swerves, but also the times when it doesn’t – is going to be an instance of indeterministic motion. And I take it that it’s the causally undetermined nature of the atomic motions underlying our actions that’s crucial to Epicurus’s account, not their specifically being swerves. (And this seems to me to be true regardless of what stance one takes on the much-debated questions as to the precise nature of the relationship between human actions and underlying atomic motions and how the indeterminacy of the latter serves to guarantee the freedom of the former.)
I’m also having painful flashbacks to Leonard Peikoff’s grossly uncharitable and distorted discussion of the swerve, delivered in complete isolation from any tincture of familiarity with the scholarly literature on the subject.
Maybe the title of this post should have been “They Also Swerve Who Only Stand and Wait.”
I’m not sure I completely see your reasoning for why any swerves should affect every atomic motion—if that’s what you’re saying—but still I can suggest an answer to the question of why there should be a swerve for every “free” act.
Epicurus’s system, as I understand it, is one of reductionistic, mechanistic determinism, except for the occasional swerves. So, it runs like a clockwork unless a swerve specifically intervenes. Everything, such as a human soul, is therefore like one of those chess playing electronic computers you can get at Amazon for $50. (https://www.amazon.com/dp/B099S1339R/) How would a swerve affect such a machine? It seems like it would have to alter some specific operation, like the direction of a logic gate or changing a one to a zero. And once the change has happened, it is set, and operation goes forward deterministically once again. The change could have far-reaching effects or effects that take a long time to be manifested. For example, the swerve might alter a value stored in memory, and this alteration might affect every subsequent move the machine makes or it might take a long time, even years, to be called into play. But that wouldn’t change the predictability of the machine, given the particular swerve, once it occurred. Given that swerve, the machine is no longer “free.” If you want another free action, you need another swerve.
Is the thought that, if there is any uncaused (indeterministic) motion, then, since everything is causally relevant to everything, everything is infected with indeterminism? That may be, but I doubt that that is considered to be the important question. The important question surely is whether the future is open or closed—and specifically as this concerns human actions. If I were worried about whether I had free will, I would not be comforted to hear, “All your actions are predetermined as a result of the deterministic motions of the atoms that compose the universe, including your body, given the state they were in at the moment of your conception. But not to worry! Ten thousand years ago there was a single swerve, and that has infected all subsequent motions with indeterminism!” I should think that what Epicurus wants is for a person’s future actions to be indeterministic, given the state of things now. That means (if the above is correct) that we need a continuing supply of new swerves, and in particular a new swerve for every “free” action.
LikeLiked by 1 person
Well, here’s how I’m thinking about these matters:
Click to access TulaneFreewill5.pdf
LikeLiked by 1 person
Thanks for the paper, it was thought-provoking and interesting.
On the question of whether an Epicurean needs a new swerve for every free human action, I gather from the paper that your view is that the mere fact of indeterminism—an open future—is enough to establish free will or anyway to legitimate an assumption of free will. (This seems questionable, but I will stick to what you actually say in the paper.) The argument seems to be: the act of deliberation assumes “that it is not already settled, prior to our choice, what decision we shall make” (8). Since determinism is incompatible with that assumption, it must be rejected on pain of “pragmatic incoherence.” The bulk of the argument is addressed to rebutting a “compatibilist response” to this incoherence claim; namely, that determinism alone doesn’t render deliberation pragmatically incoherent, but only determinism in conjunction with knowledge of our specific choice. That is, you try to show, using an argument derived from Newcomb’s Problem, that it is not my knowing that I am foreordained to choose A, when I deliberate between options A and B, that renders my deliberation incoherent, but my knowledge merely that the outcome is settled in advance in any way at all. And having shown this, you conclude that “I must affirm my own libertarian freedom a priori” (12).
If this is right, then I don’t see how it addresses my comment about why new swerves are necessary. Namely, that since atomic motion is deterministic, every post-swerve motion is deterministic until a new swerve comes along. Thus, for instance, a second post-swerve decision will be deterministic with respect to any earlier post-swerve decision. And certainly, a single swerve 1000 years ago will not be enough to make my present-day decisions indeterministic relative to life today. From a snapshot of the universe taken at any instant post-swerve, the future is completely determined until there is another swerve.
However, I think there is a problem with the argument from Newcomb’s Problem to the conclusion that the future must be causally open, not merely epistemically open, for deliberation to be pragmatically coherent. I don’t know if you’re seeking comments on this paper or intending to revise it or just offering it as an account of your thinking—at first I thought is was new, but the document details say it was made ten years ago—so I’ll try to be brief. [Looking at how this comment turned out, clearly I failed!]
You say that the crucial feature of Newcomb’s Problem that generates the paradox is that the option—whether I open just one box or both boxes, or, equivalently, whether the predictor put the money in the opaque box or not—is already settled in advance of my action. It is already settled because the money is already in the opaque box or not, and the predictor already knows my choice. And your claim is that what matters is the mere fact of the option being already settled, in whichever way, not my ignorance of which way it has been settled.
To show this, you construct a version of the problem intended to be structurally identical to the original, but with causal determinism playing the role of predictor. Thus, you have an agent A deciding whether to ϕ and an antecedent condition C that causally determines his choice: if C, he will ϕ, otherwise not. (This is like the predictor knowing whether you will open only the opaque box in the original Newcomb problem.) Now, C being out of A’s control, the combination of C’s holding or not and A’s ϕ-ing or not produces four logical possibilities: (1) C, and A does not ϕ; (2) C, and A ϕs; (3) Not-C, and A does not ϕ; (4) C, and A ϕs. And let us suppose that A’s preference ordering for the four logical possibilities is the same as the order just presented. (This also matches the original Newcomb problem: best would be to get money from both the opaque and transparent boxes, next best is to get the money from the opaque box alone, third best is to get the money from the transparent box alone, and worst is to take only the opaque box and find no money in it.)
Next, you say there are two ways A might reason about the choice, each being rationally decisive yet leading to opposite conclusions. First, considering that whether C obtains is out of his control, A should not ϕ, since that is A’s first choice regardless of whether C obtains or not (i.e., cases (1) and (3)). But, second, considering that determinism is true and thus cases (1) and (4) are causally impossible, A should ϕ, since that is his preference among the two remaining cases (i.e., (2) and (3)). Since one line of decisive reasoning concludes that A should ϕ and the other concludes that A should not ϕ, this situation is pragmatically incoherent. And, since the root of the incoherence is the assumption of causal determinism (because that is the only assumption, given that A can have a preference ordering between outcomes), causal determinism must be (at least pragmatically) rejected.
But here’s the thing: The first line of reasoning, which concludes that A should not ϕ, does not respect the assumption of causal determinism. If determinism is true, then case (1) is impossible. This is indeed the ground for dismissing case (1) in the second line of reasoning, and it should apply in the first line also. In other words, the first line of reasoning is incorrect. The second line alone is correct and decisive, and so there is no incoherence.
The statement of the argument is a little odd, actually, regarding the matter of determinism. Condition C is said to be sufficient for bringing it about that A ϕs, and then it is remarked that “if causal determinism is true,” then C is also necessary for A to ϕ, as though there were some question about it. And yet the conclusion of incoherence is said to depend on determinism: “The assumption that it is causally determined whether or not I shall ϕ at t, … entails both that the rational thing for me to do is to ϕ at t and that the rational thing for me to do is to refrain from ϕ-ing at t” (12). Again, the incoherence in Newcomb’s Problem, which the present argument is supposed to mirror, is also said to depend exclusively its being settled in advance of my action whether the money will be in the opaque box and also whether I will choose only that box: “the presence of the money is sufficient (not causally sufficient, but sufficient nonetheless, as the Newcomb story goes) for my choosing the one-box option, and its absence is sufficient for my choosing the two-box option, so it is already settled which option I will choose; and this is the crucial feature that leads to the paradox” (10). But if this is so, then it is incorrect to treat impossible options, such as case (1) in the above argument or the case of finding money in both boxes in the Newcomb problem, as though they were possible. That means that the first line of reasoning in the above argument is incorrect, as is the two-box argument in Newcomb’s problem.
Another point is that we can’t run the first line of reasoning on one assumption (indeterminism) and the second on another assumption (determinism), and then conclude that the incompatibility of their decisions is paradoxical or incoherent. Rather, we should conclude that, on the assumption of indeterminism, A should not ϕ, and on the assumption of determinism, A should ϕ. This is not pragmatically incoherent and would not be an argument against regarding our choices as causally pre-determined.
As a matter of interest, what if we relaxed the strict determinism of the above argument? That is, what if condition C were merely sufficient for A’s ϕ-ing, but not necessary? Then case (1) should still be ruled out as impossible, since C is still causally sufficient for ϕ-ing, but the three remaining cases would be live possibilities. Also, the second line of reasoning would be incorrect, since it depends on cases (2) and (3) being the only live possibilities. We must choose between ϕ-ing or not ϕ-ing with all three of cases (2), (3), and (4) considered as live possibilities. And unfortunately, I can’t see any way in which ϕ-ing dominates not ϕ-ing or vice versa, given these possibilities. Without further information about the probability of C holding, the conditional probabilities between (not) C and (not) ϕ-ing, and A’s degree of preference for ϕ-ing or not ϕ-ing, I don’t think a rational decision is possible.
All of this maps very well onto Newcomb’s Problem, in my opinion. For the reason explained above, if the predictor is infallible, then the one-box option is correct and there is no paradox or incoherence. On the other hand, if the predictor is fallible, then the best option depends on one’s estimate of the accuracy of the predictor and the decision will have to be based on an expected value calculation. It would also be nice to know just what cues the predictor uses to make his predictions. Best of all would be find some public way of inducing an unbreakable commitment to open only the opaque box, say by having oneself hypnotized and imprinted with the suggestion to open just one box a day or with a horror of money in transparent boxes. In any event, I don’t think there’s anything paradoxical or incoherent about Newcomb’s problem. (Nor does its philosophical interest depend on there being such.)
There are other interesting elements of the paper, especially the discussion of “Objection 3,” whether free will is compatible with modern science, and “Objection 4,” concerning “Aristotle’s Fantasy” and Leibniz’s Law, but this comment is very long, so I’ll stop. I should say that I am quite sympathetic to what I take to be your basic views here. I am strongly inclined toward libertarian incompatibilism, and I think that the only real alternative is a mostly incompatibilist determinism. However, I am inclined to accept the point that determinism does not render deliberation pointless as long as we do not know which choice we have been fated to make. And I fear libertarian incompatibilism faces strong challenges, especially from modern physics (I am thinking mainly of the conservation of energy), which I don’t know how to solve.
Thanks for your comments! I’ll examine them and reply more fully in future, but for the moment I just want to reply to this:
“since atomic motion is deterministic, every post-swerve motion is deterministic until a new swerve comes along”
How so? Since it’s causally undetermined when a swerve will occur, EVERY atomic motion is indeterministic, whether it’s a swerve or not (since at every moment it’s unsettled whether the next moment will be swervy or non-swervy). So I’m puzzled as to what you mean.
“I’ll examine them and reply more fully in future”
Sure about that?
Yes, I see what you mean. That’s a nice point! And so this must be the reasoning behind your original question:
Since even motions that obey the deterministic laws might not, (because swerves are always physically possible, even when they don’t occur,) such motions are in that sense indeterministic.
So, if indeterministic means “not 100% predictable,” then all motions are indeterministic in a world where a swerve is physically possible for every motion. Of course, what I meant by deterministic in the quote above was not “100% predictable,” but that the motion is in accordance with the deterministic causal laws. I imagine it’s not uncommon for people to speak this way, using deterministic to refer to motions governed by deterministic laws and indeterministic to refer to motions that deviate from them. But this is at least imprecise. Maybe it would be better to say “law governed” versus “lawless” or some such.
This correction leads to another good point, which is that if we’re thinking of free will as resulting from the existence of random swerves, every decision that counts as free need not be the result of a swerve. Instead, there could be a “default” system of deterministic decision making that sometimes gets disrupted by a swerve. The default decision could be what one would do from one’s predominant desire, for example, and this could be a deterministic mechanism (in the naughty sense I just said is imprecise—it’s hard to avoid!). But a chance swerve could disrupt this process and allow some weaker desire to prevail. We would not have to say the decision was free only in the latter (swervy) case. The default decision could also be regarded as free, if a swerve’s not happening is construed as the default decision being allowed to go through.
However, this is all leaving aside what is for me the fundamental mistake of identifying free will with random action. “Free will” should be a will. For me, it means that the agent self-determines his actions. I do not see how this is accomplished by random swerves (or quantum indeterminacy, etc.). This is why I think that merely establishing that determinism is false is insufficient to demonstrate the existence of free will. What is needed for that is some account, however vague, of how there might be a means by which an agent indeterministically self-causes some of his actions.
On further reflection, I was too hasty in saying that the analysis of Newcomb’s Problem is basically the same as the one in the paper. The problem is that, whereas C is sufficient for ϕ-ing, the predictor putting the $1M in the opaque box is not sufficient for the agent to be a one-boxer. So, in Newcomb’s Problem, the two-box option dominates in the case where the predictor is fallible. Therefore, there seems little point in trying to estimate the predictor’s accuracy, since there’s almost no chance the predictor will put the $1M in the opaque box. The only real hope for the agent would seem to be to find some ironclad, publicly knowable (i.e., by the predictor) means of preventing himself from opening the transparent box. For instance, he might eat a pill that makes him temporarily perceive money in transparent boxes as poisonous spiders. Since realistic means of preventing himself from opening the transparent box will be hard to come by, it’s too bad for the agent, who now has very little chance of getting the $1M.
EPICURUS ON FREEDOM
“In this book, Tim O’Keefe reconstructs the theory of freedom of Epicurus. Epicurus’ theory has attracted much contemporary interest, but our attempts to understand it have been hampered by reading it anachronistically as discovery of the modern problem of free will and determinism. O’Keefe argues that the sort of freedom which Epicurus wanted to preserve is significantly different from the ‘free will’ which philosophers debate today, and that in its emphasis on rational action, has much closer affinities with Aristotle’s thought than with current preoccupations.”
Mi-Kyoung Lee wrote that this book is “a helpful and masterful guide through the complex philosophical issues and fragmentary pieces of evidence relevant to Epicurus, determinism and the swerve.”
Good. It remains, I’d say, that if one’s interest is in libertarianism, determinism, and compatiblism (and incompatibility of knowledge with determinism, as in Nathaniel Branden, early 1960’s) set within our own physics, biology, and neural science: the pictures and maneuvers in Epicurean texts have analogies in contemporary takes on the analogous contemporary cluster of issues.
Nozick had a quip somewhere about the supposedly physical tail wagging the supposedly metaphysical dog. So true. So often the setting out of the problem of determinism (contemporary) rests on a false picture of our physics. People run on as if back speculating with Laplace, as if all the thermo, statistical mechanics, and chaos (v. regular) in the classical regime never came to light. Nature in the classical regime is shot through with intersecting independent causal lines (as in my old paper “Volitional Synapses”), physical contingencies within life processes and in their environments is how engineered-looking systems, such as living ones, are possible. That some living molar brain processes of humans can get to deliberate creative engineering is hardly ruled out by modern classical physics, provided one does not substitute handwaving or the models of Democritus/Epicurus for the latter.
(Quantum physics is evidently irrelevant to the pertinent brain-state transitions because of the slow speeds at which the latter occur—at like 10-200 ms. Classical chaos in neuronal action on the other hand has been implicated in some perceptual discriminations, by the 1990’s—I haven’t kept up since then.)