Free Will and Randomness (Column 645)
This past weekend I took part in the second conference on free will (the first was two years ago; you can see the recordings of the talks given there here. The talks from the second conference are at this link). The conference included researchers from all fields relevant to the topic: philosophers, neurologists, psychologists, psychiatrists, mathematicians, computer scientists, biologists, and brain researchers. I was glad to be invited, even though I had to make a few halakhic compromises (the considerate accommodation by the organizers and participants helped greatly), because I saw great importance in my participation there, for two main reasons: (a) In my impression, in recent years a materialist–determinist consensus has been forming in the scientific world that I consider dangerous, so it was important to me to be there and present a reasoned dualist–libertarian position. (b) I have written more than once that even professionals tend to err in defining the concepts and in the philosophical analysis of scientific findings, and thus arrive at mistaken conclusions regarding the existence or non-existence of free will. It was important to me to lay out the conceptual framework for the discussion, from which one can present the libertarian position and its implications more clearly and clear away a variety of straw men.
The conference was fascinating, and indeed my forecasts did not fail. In my view, the two points were very necessary, and I hope and believe I made my contribution there. My talk outlined a conceptual framework for discussing free will, and in effect summarized my book The Sciences of Freedom. Here I wish to focus on a point that came up again and again and to which I repeatedly called attention in the various panels held after each session: the relation between free choice and randomness, and from this the rejection of the possibility of rooting free will in scientific ground. I will also take this opportunity to comment on a few more general methodological points.
Introduction: The Aim of the Discussion
I will start at the end and state my position. I am a libertarian—that is, I believe we have free will. As I have already noted, many philosophers and scientists today are determinists, but even those who are not usually try to root free will in scientific soil. Some appeal to chaos theory, but as I explained in my book there is nothing random there. Others appeal to quantum theory, but as I argued in my book that is not possible either, since randomness is not choice. Others appeal to emergentism, but that is not really an option distinct from the previous ones (see, for example, Column 593. This column is in a certain sense a complement to that one).
The upshot is that if you are a materialist–physicalist, i.e., you believe the world is all physics and nothing else, you cannot be a libertarian. You must be a determinist. This is not an a priori conceptual conclusion. Conceptually, one could be both a materialist and a libertarian without contradiction—but only if physics contained “gaps” that allow this (chaos, quantum, etc.). As a matter of fact, as I understand it, there are no such gaps within physics, and therefore this option is ruled out a posteriori. Hence I argue that libertarianism requires interactionist dualism, namely the belief that there is in us a spiritual dimension beyond matter, and that this dimension is in two-way interaction with matter. Each affects the other. Therefore my struggle at the conference, and generally, is conducted on two fronts: one against determinism, and the other against physicalist libertarianism.
Incidentally, I will add that one of the speakers at the conference, Prof. Gil Kalai, pointed to a phenomenon I did not know and which, according to him, indeed opens a door to physicalist libertarianism. I am currently in discussions with him about this. If there are developments I will try to update.
Peter van Inwagen’s Argument
Peter van Inwagen (who, as I learned at the conference, is himself a libertarian)[1] offered several arguments for determinism. The best known appears in almost every discussion of the subject, and even when not mentioned explicitly it is usually in the background. It is a dilemma-type argument (in Talmudic parlance, “mima nafshakh”—whichever way you look at it): for any event that occurs, either there is a cause or there is not. The Law of the Excluded Middle states there is no third option. But if the event has a cause—then it is not the product of free will but a deterministic event. And if it has no cause—then it is a random, brute event, and again it is not the product of free will. Since there is no third possibility, it follows that, whichever way you look at it, there is and can be no event produced by free will.
Schematically, one can present the argument as follows:
According to the Law of the Excluded Middle, the orange line is not possible, and therefore there is no choice.
A Look at Dilemma Arguments: The Sorites Paradox
Dilemma arguments are very common in philosophical, mathematical, and other discussions. If some conclusion is true whether we assume X or we assume “not-X,” then it is necessarily true. Formally it looks like this:
(A → B ; ~A → B) ► B
Seemingly this is a necessary argument and one cannot dispute its conclusion. That is the charm of dilemma arguments and of logical arguments in general, and thus, apparently, Peter van Inwagen has settled the ancient dispute about free will and determinism. Precisely for this reason, I must preface by sharpening the need to beware of being misled by such arguments (see, for example, the beginning of Column 634 on the pitfalls that can arise from logical formalization).
Consider, for example, the following argument: There is no point in giving exams, since a diligent student has no need of them (he will study anyway, even without an exam), and for a lazy student they won’t help (he won’t study even if an exam is held), and therefore exams are pointless. This is a typical dilemma argument. What is the problem here? I have often explained that there is an implicit dichotomous assumption that all students are divided into two categories that constitute a ‘complete partition’ (i.e., they do not overlap—they are disjoint—and together exhaust the entire space—there is nothing outside them). But in many cases this assumption is false, since diligence and laziness are not binary notions. There are students who are not wholly diligent nor wholly lazy. They possess some degree of diligence, and for them an exam can indeed help. Such a student would not have studied without an exam, and the exam will prod him to study. This actually reflects what is called in philosophy the Sorites Paradox (see, for example, Column 110).
However, in our case the situation looks different. Either there is a cause or there isn’t. There does not seem to be a third option—at least according to the usual definitions of “cause”: a sufficient condition that produces the effect. On this definition, if it produces it only partially (e.g., further conditions are needed—that is, it is a necessary but not sufficient condition, and certainly if it is neither necessary nor sufficient) then it is not a cause. If so, van Inwagen’s dilemma argument looks crushing, and it is no wonder that it stars in almost every discussion of free will. And yet, I will argue below that it is not as decisive as it seems.
Logical Arguments and Begging the Question
Begging the question is an argument that assumes its conclusion. It is commonly thought that begging the question is a fallacy—for what is the point of basing the conclusion on itself?! But I have repeatedly argued that a valid logical argument always begs the question (see here and especially the thread here). In fact, I showed that validity of an argument means that it does beg the question. In brief: the validity of an argument means its conclusion follows necessarily from its premises, i.e., whoever accepts the premises cannot reject the conclusion. Why is that so? Because the conclusion is already embedded in the premises. For example, from the two premises “All humans are mortal” and “Socrates is a human,” the conclusion “Socrates is mortal” necessarily follows. But the premise “All humans are mortal” is actually a shortened formulation of many particular claims: Jacob is mortal, Moses is mortal, Muhammad is mortal, Yocheved is mortal, and so on. Among them is also the assumption that Socrates is mortal, since according to the second premise he is also human and therefore necessarily appears on that distinguished list. If so, the conclusion of the argument is already contained in its premises, and therefore it is a valid argument. That is why, if one adopts the premises, one has no choice but to adopt the conclusion as well. The meaning is that a valid logical argument never adds information beyond what is already in the premises. It merely exposes information latent within them.
In simple arguments like the one about Socrates, this is easy to see. But I have often explained that this holds for every valid logical argument. Sometimes it is quite complicated to see, but keep this rule of thumb: if the argument is valid, it always begs the question. This matters because it guides you in how to attack an argument that leads to a conclusion you do not accept. It is always worth looking for that conclusion somewhere within the premises. I assure you that if you look carefully, you will always find it there.
Example: The Principle of the Identity of Indiscernibles
A good example is Leibniz’s “Principle of the Identity of Indiscernibles” (see Column 519). His claim is that any two objects that have the same set of properties are not two but one and the same object. If they are indiscernible (in their properties), then they are identical (ontically). He offers a proof of this principle via reductio: assume the opposite, namely that there are two distinct objects (non-identical, two and not one), A and B, that have the same set of properties. But in that case, object A has the property of “not being B,” which B of course does not have. That is, if they are two, the premise that they have exactly the same set of properties is contradicted, and therefore, if they do have the same set of properties, it is clear we cannot say they are two. Q.E.D. Seemingly this is a crushing proof, since it is clearly a valid logical argument.
But if it is a valid logical argument, then it necessarily begs the question. To see this, let us return to the philosophical dispute Leibniz sought to settle. What is the position he is trying to attack? That there are two objects with the same set of properties, and yet they are two and not the very same object. This position apparently assumes that “being not-A” is not a property of B (otherwise they would not have the same set of properties, as Leibniz rightly claims). The conclusion is that, according to this position Leibniz attacks, the properties of an object are the collection of its attributes, but its very identity (its individuation—that it is itself and not another) is not a property. This is a statement about the thing as such (the noumenon) and not about its properties (the phenomenon).
Along similar lines, I once wrote here about attempts to locate where the “I” sits on the psychoanalytic map of the person (this also came up at the conference). I have always thought the question is based on a category mistake, since the “I” is not one of the person’s psychological functions. The “I” is the entity that has those functions. It is the object to which those functions belong (the one who wants, feels, thinks, remembers, and so on). Memory, intellect, emotion, etc., are all functions of the I, but one cannot find the I itself on the map of functions. It is like asking where we can locate the table’s “this-ness” within the object called a table. We can locate its leg there, and even the tabletop, but not the table’s very “table-ness.” The table is the bearer of the leg or the plank. They are parts of it. Likewise, the object’s haecceity is not a property of it but the bearer of properties. Miki Avraham is the person for whom such-and-such properties are properties. His being “Miki Avraham” and not, say, Yosef Cohen is not one of his properties (part of the phenomenon, in Kantian terminology). It is a statement about him as such (the noumenon).
You understand that if we assume this premise (which Leibniz is trying to attack), Leibniz’s argument collapses. His argument assumes that “not being A” is a property of object B, but that is precisely the disputed claim. He assumes what he seeks to prove, which is begging the question. He may of course be right, but there is no argument here that can persuade adherents of the opposing view. The dispute remains. That is the nature of logic. A logical argument is based on premises. When the conclusion is part of the premises (and that is always the case in a valid logical argument), anyone who disagrees with them will of course not be persuaded by the argument, since a logical argument cannot persuade someone who rejects its premises.
It is important to distinguish between the two cases I discussed. In the Socrates case, the question-begging is obvious and simple, and therefore the argument, while valid, is worthless. The begging of the question is blatant and self-evident, and there is no point in offering such an argument against someone who thinks Socrates is not mortal. Where, nevertheless, is it possible and appropriate to use logical arguments? Where the question-begging is more subtle, as with Leibniz. There, sometimes it may turn out that your interlocutor will indeed be persuaded by the argument. How can that be if it begs the question and the conclusion is not accepted by him, at least initially? In such a case he realizes that what he thought was a correct claim is actually wrong by his own lights. Suddenly he understands that, in his view, the fact that object A is not B is a property, and therefore Leibniz is right. At the outset he did not realize that his conclusion depends on that premise, and Leibniz’s argument clarified the matter for him.
For example, a proof in geometry is a valid logical argument, and therefore it begs the question. But clearly there is value in teaching someone geometry. Even one who knows and understands the axioms well will not necessarily know that the sum of the angles in a triangle is 180°. He does not see that this is contained in the axioms, since the route to see it is not at all simple. In such a case, if he thinks the sum is different (or that there is no constant angle sum for different triangles), the proof will persuade him that he is mistaken. Here the relation between the premises and the conclusion is more complex than in the Socrates case, and in such situations valid logical arguments have value.
Let us now return to our subject. We now understand that once we are faced with a valid logical argument, it necessarily begs the question. If so, we must find where van Inwagen assumes determinism in the premises of his argument.
Back to van Inwagen
To do so we must examine the libertarian claim he attacks (exactly as we did with Leibniz’s argument). We will ask ourselves what it assumes regarding the relation between free will and cause or randomness. The determinist claims there is no free choice, and therefore, for him, there is either causation or sheer randomness. If there is a cause, that is determinism; if there is not, that is randomness.
But the libertarian disagrees precisely at this point. He claims there is a third mechanism, which is neither causation nor indeterminism (sheer randomness). If so, van Inwagen’s dilemma (“whichever way you look at it”) collapses. Where exactly is the flaw? After all, the Law of the Excluded Middle is a logical axiom that cannot be denied, and the libertarian cannot reject it either. Is there an additional assumption in van Inwagen on which one can forgo? Indeed there is. Van Inwagen assumes that if there is no cause, then necessarily the matter is indeterminism. But note that on the libertarian’s map there are two states that fit “no cause”: indeterminism (sheer randomness) and choice. Choice, too, has no cause. If so, there is a hidden premise in van Inwagen’s claim: that if there is no cause, then necessarily we have indeterminism. It is precisely this premise that the libertarian disputes, and van Inwagen’s argument begs the question: its conclusion is a hidden premise in the argument, beyond the logical Law of the Excluded Middle, and therefore his argument begs the question. It may be true, but it obviously loses its value as a tool for persuading the libertarian. The determinist himself can still hold his position consistently.
[In brackets, note that in this argument the determinist is willing to accept randomness (though that is not a deterministic mechanism). He rejects only free will. He can of course continue and, after van Inwagen’s argument, offer an additional argument: van Inwagen’s argument claims there is no category of free will because there is no logical place for it on the map (only indeterminism or determinism), but afterward one can also reject randomness by appeal to the principle of causality. That is, the determinist raises this argument against the libertarian, but he need not stop there.]
Sharpening the Libertarian Position
In the terms discussed above, one can say that such an argument has value. The connection between the premises and the conclusion is complex and non-trivial, so a person can hear this argument and be persuaded that his libertarian premise is mistaken. And indeed, to my regret, there are a fair number of people who have been persuaded by it and think there is no logical possibility for a mechanism of free will. This happens when someone is convinced that whenever something occurs without a cause it is an indeterministic state, and thus realizes that the notion of “free will” he has been using is in fact empty. He discovers that until now he has been living in an illusion.
But the fact that this argument has value does not mean everyone must be persuaded by it. In my case, for example, it only helped me further sharpen my libertarian position. I understood from this what many others miss (I saw this throughout the conference): that “choice” does not mean merely the absence of a cause, full stop. While an act of choosing has no cause, there must be something else that distinguishes choice from randomness. What can that be? Here I draw on van Inwagen’s argument precisely to sharpen the libertarian position.
A brute act is an act done without a cause and without a purpose. It just happens, that’s all. An act with a cause is an act that is brought about by some cause that determines it. Free will is not produced by a cause, but it is not brute. It is the result of deliberation and decision usually oriented toward the future. When I choose between conflicting values, it is the result of deliberation, not of a lottery. Deliberation can determine, for example, how the world will look better. I think about what I should do to realize some value or achieve some outcome. This is a mechanism oriented toward the future, a teleological action (purposeful). In the libertarian picture, free will consists in actions that move toward a future one wishes to realize and not by virtue of some cause in the past (as in determinism). It is done “in order to” achieve something, not “because of” something.
If so, libertarianism holds that there are three mechanisms or types of relations between circumstances and an event that occurs within them:
- When the circumstances determine the event—this is determinism. The event occurs by virtue of a cause.
- When the circumstances are unrelated to the event that occurs within them—this is indeterminism. The event occurs randomly (independently of the circumstances).
- When the circumstances do not determine the events, yet they do not occur randomly but as a result of deliberation oriented toward future goals (reaching them from the existing circumstances)—this is free choice.
In short: when an event has a cause—it is deterministic. When it has a purpose and no cause—it is a choice. And when it has neither purpose nor cause—it is indeterministic.
From here you can see why van Inwagen’s argument does not persuade me to be a determinist, but it certainly helped me sharpen the meaning of my libertarianism and situate it relative to determinism and indeterminism. Its flaw was not in the Law of the Excluded Middle as in the exam example—that is, not because there is a third option in addition to the two presented—but, on the contrary, because the “third” option here (choice) is not third but part of the second (it is another kind of cause-less event, beyond indeterminism).
Note: Experience and Intuition
Of the three mechanisms, the determinist treats causation as self-evident. That is certainly a mechanism that exists in his view. Moreover, he assumes the principle of causality that states that for every (!) event there must be a cause. By contrast, randomness may or may not exist, as I remarked above (and see also below on quantum theory). But free choice—this he is not prepared to accept at all. Note that we have just seen that this “mechanism” is well-defined and conceptually distinct from determinism and indeterminism. Yet the determinist is still unwilling to accept its existence.
Now is the time to ask: why? Why is a causal mechanism understandable and acceptable to him—and perhaps also indeterminism—but free choice is not? We should recall what we learned from David Hume, that the causal relation is not the result of observation. The causal relation is not learned from experience but is an a priori insight of ours (see, for example, Column 586 and many others). As for randomness, no need to belabor the point. One need only observe the embarrassment of philosophers and physicists in the face of quantum theory to understand how unwilling people are to accept the existence of brute, uncaused events. Paradoxically, free will is the mechanism most familiar to us from immediate experience. We experience it every day, again and again, within ourselves when we make decisions. And yet the determinist claims specifically about this mechanism—the most familiar to us—that it is an illusion, while he is ready to accept the other two more readily, despite their lacking any observational basis that would ground and clarify their existence. Of course, in light of the argument I have raised here, doubts arise with regard to any pair of events we link in cause-and-effect terms, but the assumption that everything must have a cause is certainly undermined. It is undermined even more in light of quantum theory (which indeed demonstrates randomness, not choice). But this does not prevent the determinist from treating libertarianism specifically as mysticism.
There is another important point. Many have argued against me that I speak of a libertarian mechanism but do not explain it. If there is no cause here, what, then, makes the event occur? Again and again I am asked: Yes, but why does this event occur? What produced it if it has no cause? First, this question also arises regarding random, brute events (indeterminism). Quantum theory teaches us that such phenomena can exist. Beyond that, the question is categorically mistaken. The questioner expects me, as a libertarian, to present, in my answer, a cause by virtue of which an event of free will occurs. But my entire claim is that there is no such cause, since this event is the result of free will. That is precisely the point of our dispute. What, then, is the point in asking me what cause produced this event? The search for explanations is in fact a search for causes—but that is, again, begging the question. The determinist presupposes determinism instead of proving it.
Incidentally, to look for an explanation of the mechanism of free will on the basis of a causal mechanism is like using a Hebrew–English dictionary. You expect me to explain to you a mechanism we both know best and most directly and immediately (like a Hebrew word) by means of another mechanism we have never encountered observationally and for which we have no empirical basis (a collection of words in English). For some reason, determinists accept the conceptual clarity and existence of deterministic mechanisms they have never observed, and deny the clarity and very existence of “mechanisms” of free will that are familiar to us in the most intimate way. And this is considered the rational, scientific approach. Very strange.
What Does All This Say About Randomness?
Suppose we see that some event cannot be fully predicted on the basis of the circumstances, as with the toss of a coin or a die. In such cases I can present a distribution of probabilities for the outcome, but still cannot predict clearly what exactly will happen. The same holds in quantum theory. In a quantum state one cannot know, on the basis of present circumstances, what will happen next. There is a distribution of probabilities concerning what will happen, but the outcome is not dictated by present circumstances.
Many think that if we show there are events that lack a cause in the deterministic sense, we have thereby shown free will. If the outcome cannot be predicted, that is freedom. At the conference I heard more than once people pointing to statistical events whose outcomes cannot be predicted as a basis for free will within science. Thus, some see chaos as a scientific basis for free will. But that is a mistake. As we have seen, randomness is indeed an occurrence without a cause, but it is not choice.
Chaos: On Predictability and Free Choice
Chaos is a wholly deterministic process; it is just that we find it difficult to predict its outcome because of computational complexity, and therefore we employ probabilities. A simple example is the coin toss. In fact, it is entirely deterministic. There is a physical object (the coin) on which ordinary forces act (gravity, friction, the push from the tossing hand), and therefore, in principle, one could calculate where and on which side it will land. These are Newtonian laws of basic mechanics. However, that calculation is very complex and highly sensitive to the way the coin is tossed, and therefore we habitually use probabilities. But there is no genuine freedom here. Everything is deterministic. It is therefore an error to hang our free will on this. The fact that we know of chaotic processes within physics does not constitute an explanation and does not allow physics to encompass free will.
More generally, the fact that we cannot predict some outcome does not mean it occurred by virtue of free will, or even randomly (and as noted, those are not the same). True, if something happens by free will, you cannot predict it—but it is not true that if you cannot predict something, it expresses freedom. As we will now see, the same holds for genuine randomness, not only for chaos which, as noted, is a deterministic process.
Quantum Theory: Randomness and Free Choice
We must understand that all the “random” events we know in our lives are not truly random, but something like the die. These are events whose outcomes are difficult to compute or predict because of computational complications. In our ordinary world there is no genuine randomness. I dealt with this type of events and their meaning for the question of determinism in the previous section. But in quantum theory the situation is more complex. There the process is indeed non-deterministic. In quantum theory one cannot predict the outcome not because the calculation is complicated but because, in principle, it cannot be predicted. It is genuinely random (not merely apparently so, as with chaos). The meaning is that in quantum theory the outcome is not determined unequivocally by the present circumstances. There we already have genuine randomness (on the common interpretations), ontic and not merely epistemic. Does this allow us to speak of free will?
In my view, absolutely not, for several reasons. First, quantum effects occur on very small scales, and they dissipate when we reach the scales of everyday life. There are no quantum effects in bodies larger than a micron (one-thousandth of a millimeter), and even that only at very low temperatures. This certainly does not occur at ordinary temperatures (“room temperature,” as physicists say). More importantly, the second point: randomness is not choice. Even if quantum theory could describe what happens in a living body or in its brain and there were no scale problems, the “choice” of particles among possibilities in quantum theory is a lottery, not a choice. We have seen above that randomness (indeterminism) is not choice.
To sharpen this further, consider the two-slit experiment. A particle is fired toward a screen with two slits. If we do not place a detector by one of the slits, then this single particle will pass through both slits. This is a state of superposition that occurs only when we do not measure what happens. But if we place a detector by slit A, the detector will tell us whether it passed through that slit (if the detector fires) or through the other (if the detector remains silent). Thus, by the detector we measure which slit the particle passed through. Quantum theory says that in this case there is no superposition, i.e., the particle “chooses” just one slit and passes through it (this is the famous “collapse” of the wave function). In such a case, the probability that the particle will pass through slit A or B is determined by the particle’s quantum wave function, which is a solution to the Schrödinger equation in the given state.
Now think of this particle as a person. If the person chooses to pass through slit A or B, the probability of passing through either is not determined by a quantum wave function of the person (a solution to Schrödinger’s equation for him), but by his values and reasons. This will not necessarily match the distribution dictated by quantum theory. Remember we are assuming free will here, i.e., the person can freely choose, by his various considerations, through which slit to pass. There is no reason to assume that the resulting distribution will obey the Schrödinger equation that sets the quantum distribution. In other words, free will means the outcome is not determined even probabilistically.[2] Moreover, even if we treat placing the detector as the person’s act of choosing—i.e., placing the detector at slit A is the choice of which slit the particle will pass through—behold your human free choice. But placing the detector does not determine which slit the particle will pass through. It only ensures that it will not be in a superposition. If we place the detector at slit A, the particle can still pass through A or B “as it wishes.” Therefore, placing the detector cannot be treated as the person’s choice (and besides, the person’s choice to place the detector at slit A is itself an event that can be analyzed in the same way). If so, quantum theory cannot explain human free choice either. We repeatedly see that randomness is not choice. In the terminology of my series of columns 126–131, this is freedom, not liberty.
Conclusion: Libertarianism Is Incompatible with Physicalism
The conclusion is that indeed, when a person acts by free will, the outcome cannot be predicted. But if there is an event whose outcome cannot be predicted, there is no necessity to infer that it is an act of choice. It could be a deterministic event that is hard to predict (like chaos) or a brute indeterministic event whose outcome is determined by a lottery according to a given distribution (as in quantum theory). Contrary to the hidden assumption in van Inwagen’s argument, randomness is not choice.
The conclusion is that we cannot root our free will in the field of physics. If the world is only physics—then it is necessarily deterministic. Hence, if someone holds that a person has choice, we are compelled to say there is something beyond the laws of physics at work in him. In my view, it is plausible that this is not merely a matter of additional laws. Physics works very well, and there is no reason to think there are more physical laws we have not discovered. It is more plausible that there is a different kind of entity—spiritual entities (soul, spirit, psyche)—that are not subject to the laws of physics. Our choosing takes place within the will, i.e., in the spirit, and only afterward it passes into the physical plane and produces physical results.
This means that an event of free will occurs in two successive stages: in the first stage, some desire arises in us. This has no cause but does have a purpose (it is the result of deliberation). Here the laws of physics do not apply because it is a mental–psychic process, but the principle of causality is violated. After that, this desire initiates a chain of events that lead to a physical action. These do have a cause—the desire—but not a physical cause. For example, I decide to punch someone. Now an electrical current is formed in the brain that ultimately arrives as an electrical instruction to my right hand to send a fist to that poor fellow’s face. It ends with the delivery of a physical punch (unless he responds, in which case the ending is less attractive for me). If so, the first stage violates the principle of causality even if not the laws of physics (it departs from them, of course, because it is a non-physical event). The second stage has a cause—the desire—but not a physical cause. Therefore it departs from the laws of physics. The first electron in this process begins to move without a physical force acting upon it.
This course is depicted in the following figure:
One must understand that this is the necessary meaning of a libertarian picture. If we cannot root free will in physics, then necessarily, if we adopt a libertarian view and oppose determinism, at some stage there must be a departure from the laws of physics.
How Do We Decide?
Note that thus far I have said nothing for or against determinism. I merely set the two conceptions side by side. I assume that from my tone it is easy to see where my opinion leans (and I also said so at the beginning of the column), but all you have seen so far is a sketch of the two pictures side by side. In my book The Sciences of Freedom I explained why, in my view, there is currently no scientific way to examine this question, and the main points were presented here in the column. There I proposed several thought experiments that may nevertheless help us form a position. Here I will briefly present two possible ways of resolution.
Resolution: Thought Experiments and “Buridan’s Man”
One of the thought experiments I proposed there is what I called “Buridan’s Man” (see also a discussion of it here). Jean Buridan was a French scholar in the 15th century. He wanted to illustrate the meaning of rational decision through the following case. Imagine a donkey standing between two mangers on either side of it. The distances are equal, the contents identical, there is nothing else in the universe, and even our donkey is point-like. That is, there is perfect symmetry regarding the state of the world. What will happen to such a donkey? Buridan’s claim was that it would die of hunger. He explains that people usually think a rational decision is to act when you have a good reason to do so. But here our donkey has no reason to go to the right manger rather than the left, and therefore a “rational” donkey would die of hunger. A human being, by contrast, is truly a rational creature (and not merely “rational,” or rationalistic). Therefore, even though there is no clear reason to go right or left, he will surely choose one of the directions at random and go eat—without a reason and without a rational ground. Buridan wanted to illustrate that at times rationality means acting even when there is no specific reason for one’s choice.
I now continue this line of thought and say that in the experiment of Buridan’s Man—i.e., a point-like person standing in a perfectly symmetric situation between two tables laden with good things—according to physicalist–determinists he would die of hunger. The reason is that if we are dealing with a physical system, then the symmetry of the problem constrains the symmetry of the solution. The equation treats a perfectly symmetric situation between right and left, and therefore the solution (i.e., what the person will do) must also have that symmetry. That means the person cannot go to one of the two tables at his sides. No solution of a deterministic equation or computation can yield such a result. In the physicalist picture of the world, a person in such a situation will die of hunger just like a donkey. His exalted intellect will not help him, because his body is a physical entity whose motion must conform to the laws of physics.
This is, of course, a thought experiment. We cannot perform it in practice, and we can only guess what would happen there. What does this mean for the determinist? One of two things: either he decides to remain in his determinist position and then must agree that a person in such a case would die of hunger; or he must forgo his position and understand that he is not a determinist. If the person decides to go right and eat in order not to die, this means that decision does not reflect a physical computation but a free decision (arising from deliberation—therefore it is not randomness). Now every determinist can think for himself which of the two options he chooses. Note that this thought experiment is nothing but a logical argument in a complex situation (the relation between the premises and the conclusion is non-trivial). As we have seen, such an argument gives the determinist two possible exits: either remain in his position and pay the intuitive price, or realize he was mistaken, forgo his premises and position, and acknowledge that he is a libertarian (only he did not realize it until now).
A General Way to Resolve Between Intuitions: “Perforated” Determinism
What stands here, face to face, is the principle of causality (the claim that everything has a cause), on the one hand, and the sense of freedom of will that each of us has regarding himself, on the other. I remind you that the principle of causality has no empirical source. It is an a priori intuition of ours. The sense of freedom is likewise grounded in immediate experience, and we have already seen that this is no worse a source—and perhaps a better one—than the sources we have for random or deterministic events. There is therefore no a priori priority to either of these two intuitions. If anything, it seems that the libertarian option is preferable.
What do we do when faced with two conflicting intuitions? My claim is that we should apply here the legal principle known as lex specialis (the preference for the specific). To understand this better, take the following halakhic example. The Torah contains a general prohibition on murder, and a specific obligation to put Sabbath desecrators to death. These are, of course, two conflicting directives. Now a Sabbath desecrator comes before us. What should we do? If we prefer the general principle—the prohibition of murder—then of course we will not kill him, but then the specific principle loses all content. The verse instructing us to execute Sabbath desecrators is emptied of content. By contrast, if we prefer the specific—i.e., we execute the Sabbath desecrator—the general prohibition still retains substance. There is a general prohibition on murdering human beings—except in the case of one who desecrated the Sabbath. Therefore, this solution is preferable to following the more general principle. The specific principle perforates the general one but leaves it intact.
So too in our case. If we prefer the principle of causality, then the experience/intuition of free will is emptied of content. We have given it up entirely. By contrast, if we prefer it—i.e., we continue to assume we have free will—the principle of causality will still stand regarding every other event and situation (apart from acts of human choice). That initial electron indeed moves without a cause (if we compress the two stages from the figure above), but from then on everything operates according to the principle of causality and the laws of physics. One can call such a view “perforated determinism.” As a rule, physics is deterministic (or random, in quantum theory), but there are very rare “holes” in which this physicalism breaks, the will intervenes, enters through them, and affects physics. The libertarian resolution is reasonable and called for in such a case.
[1] Just like the neurologist Benjamin Libet who, despite his well-known experiment from the late 1970s that many use to prove determinism, was himself a libertarian.
[2] It is precisely on this point that Prof. Gil Kalai offers his novelty. He claims there is an additional uncertainty in quantum theory beyond the quantum distribution described by the Schrödinger equation (he calls it “noise”), and perhaps that is where free will can be inserted. As noted, I will update if I have developments.