Edge‘s big question for 2017 is: “What scientific term or concept should be more widely known?” The compilation of answers (205 in all) makes for fascinating reading. For his part, Professor Jerry Coyne has nominated physical determinism as “a concept that everyone should understand and appreciate.” Unfortunately, Coyne’s defense of this concept leaves a lot to be desired. As I’ll argue below, even if you reject interactionist dualism (as most scientists do), you can still believe in libertarian free will.
Professor Coyne begins by mis-defining determinism as the notion that “all matter and energy in the universe, including what’s in our brain, obey the laws of physics.” I know of no philosopher who defines determinism in this way. The Stanford Encyclopedia of Philosophy, for instance, roughly defines causal determinism as “the idea that every event is necessitated by antecedent events and conditions together with the laws of nature.” Coyne says nothing about antecedent conditions, and fails to even mention the notion of necessitation.
To illustrate what’s wrong with Coyne’s definition, I’ll use an analogy which is often cited by philosophers: the game of chess. All the pieces obey the rules (or laws) of the game, but those laws don’t tell the player where to move the pieces. Even if the pieces were capable of moving themselves without the help of an outside agent, nothing in the rules of the game would determine the moves that followed. That’s because the rules of chess merely constrain the set of moves which are allowed, without determining the movement of any of the pieces. What Coyne needs to show is that the laws of physics are more than mere constraints, and that for any given collection of molecules, they narrow down the set of possible outcomes to just one, and no more.
At a given moment, all living creatures, including ourselves, are constrained by their genes and environment to behave in only one way — and could not have behaved differently. We feel like we make choices, but we don’t. In that sense, “dualistic” free will is an illusion.
Here, again, Coyne is making an unwarranted leap of logic. I would happily grant that I am constrained by my genes and environment. What Coyne needs to show, however, is that I am constrained to behave in only one way.
Coyne seems to believe that a basic knowledge of physics is all you need to establish the truth of determinism:
This must be true from the first principles of physics. Our brain, after all, is simply a collection of molecules that follow the laws of physics; it’s simply a computer made of meat. That in turn means that given the brain’s constitution and inputs, its output — our thoughts, behaviors and “choices” — must obey those laws. There’s no way we can step outside our mind to tinker with those outputs. And even molecular quantum effects, which probably don’t even affect our acts, can’t possibly give us conscious control over our behavior.
Coyne’s appeal to physics is a misguided one. The claim that physics rules out free will is often heard. I find it curious, however, that proponents of determinism seldom tell us exactly which laws of physics imply the truth of determinism. The law of the conservation of mass-energy certainly doesn’t; and neither does the law of the conservation of momentum. Newtonian mechanics is popularly believed to imply determinism, but this belief was exploded over two decades ago by John Earman (A Primer on Determinism, 1986, Dordrecht: Reidel, chapter III). Sometimes the Principle of Least Action is said to imply determinism. But since the principle does not apply to a system which is non-holonomic (i.e. a system whose geometrical constraints involve not only the coordinates but also the velocities), and usually doesn’t apply if the system is dissipative (i.e. if it includes frictional forces), then I fail to see how it could be realistically applied to the human brain – which we have absolutely no reason to regard as holonomic, and which is subject to frictional forces (both wet and dry). So much for Coyne’s theatrical appeals to “the first principles of physics.” However, there is a more fundamental problem with his case against free will.
Let us grant (for the sake of argument) Professor Coyne’s materialist premise that it is my brain that thinks and chooses, rather than:
(i) a Cartesian soul interacting with my body, or
(ii) an embodied person (me) performing a non-bodily action (which is what Aristotelian-Thomists believe happens when we make a rational decision), or even
(iii) a person (me) performing an action (choosing) which has some non-physical properties, in addition to its physical properties (as property dualists maintain).
Even if we accept that thoughts and decisions are actions performed by our brains, and that these actions can be fully described in physical terms, the truth of determinism does not follow. To rule out the possibility of free will, we need to import an additional premise: that any movement of a body can always be fully explained in terms of the interactions between its physical constituents (i.e. the particles of which its is composed) and the other bodies in their surroundings. This premise might sound trivially obvious, but what it denies is that bodies have any holistic physical properties which affect the way in which their constituent particles move. When you put it that way, it’s not at all obvious.
Holism and the failure of supervenience
Suppose now that brains possess holistic properties (which are physical but not attached to any particular part of the brain), and that thinking and choosing are simply holistic properties of conscious, normal human brains. Then since physics makes no reference to holistic properties in its mathematical descriptions of bodily motion, it follows that physics will be unable to fully account for the movement of any body possessing these holistic properties – which in turn means that physics can provide us with no guarantee that a set of inputs acting on the brain when its constituent particles are configured in physical state S will always generate the same output, since the brain’s holistic properties may turn out to vary, even when its constituent parts (i.e. its neurons) are configured in the same state.
The notion which I am proposing here, that a body whose physical constituents have exactly the same properties on two successive occasions may yet turn out to have holistic properties which are quite different on both occasions may strike some readers as outrageous, as it runs afoul of the widely accepted principle of supervenience, which the philosopher Donald Davidson pithily expressed in a now-famous quote:
“[M]ental characteristics are in some sense dependent, or supervenient, on physical characteristics. Such supervenience might be taken to mean that there cannot be two events alike in all physical respects but differing in some mental respect, or that an object cannot alter in some mental respect without altering in some physical respect.”
To someone who accepts the principle of supervenience, the contrary notion, which I am defending here, might seem to smack of occultism. In reality, it is nothing of the sort. For if holistic properties of bodies are not confined to a particular point in space, it should hardly be surprising that these same properties are not confined to a particular point in time, either – which means that we cannot infer what values these properties take, merely by examining the behavior of a body, here and now. The problem with the principle of supervenience is that it rests upon a “snapshot” view of bodies, which is biologically implausible.
An advocate of supervenience might reply that the principle can be reformulated to accommodate the passage of time: “there cannot be two events which are alike in all physical respects over an extended period of time, but differing in some mental respect.” But what period of time are we talking about here? One second? One minute? One hour? Any answer we give will be an arbitrary one. And in the case of human beings, we can certainly envisage an individual performing the same behavior on two separate occasions, but having entirely different reasons and intentions for doing so, on each occasion. You may tell me that subtle differences would still show up in their brains, on the two occasions, which a skilled neurologist could detect. That may well be so, but I see no reason why it must – even when the behavior being repeated occurs over a very, very short interval of time.
Perhaps the supervenience proponent is envisaging two (living, conscious) bodies, whose physical states are identical over their entire lifetimes, and asserting that it is impossible for there to be any mental differences between these bodies. But in that case, the assertion is simply wrong. For it will never be the case that the two bodies are receiving identical input from the outside world. The mere fact that they are situated in different locations prevents that. Hence their physical states will never be the same – which renders moot the question of whether their mental states will be the same. Thus the plausibility of supervenience turns out to rest on an absurd counterfactual.
But even if the claim that two living bodies whose physical states are identical over their lifetimes will have the same mental states were true, we could not use it to infer that if a particular living body is in the same physical state S over an extended period of time, on two successive occasions, its mental states are the same on both occasions. (And if we allow that a body’s mental states can affect its behavior, it follows that we cannot be sure that it will behave in the same way on both occasions.)
We have seen that attempts to argue for the principle of supervenience end in ignominious failure. At this point, a defender of determinism may argue that the idea that the mental states of bodies are holistic is wildly implausible, and not worthy of serious consideration – but I would ask: why? Attempts to localize semantic propositions in the brain have, to date, been wholly unsuccessful (although scientists have had limited success in localizing words that are stored in the brain). That being the case, we cannot speak of decisions based on propositional reasoning as having a location in the brain, either. And even if some decision-making region of the brain could be identified – the frontal lobe is often suggested, although other areas appear to be involved as well – it still would not follow that decision-making is an event which could be reduced to the neuronal level. In other words, the question is not whether the holism of the mental is true, but to what extent it is true.
Coyne’s dismissal of randomness as irrelevant to the question of free will
At one point in his article, Coyne briefly alludes to the possibility that at the quantum level, events might not be determined, after all – but he dismisses this possibility, on the grounds that it has no relevance to our decision-making: “molecular quantum effects, which probably don’t even affect our acts, can’t possibly give us conscious control over our behavior.” If Coyne is asserting that quantum randomness in no way implies free will, then I agree; but if he is claiming that randomness is incompatible with free will, then he is profoundly mistaken. Consider the following two rows of digits:
1 0 0 0 1 1 1 1 0 0 0 1 0 1 0 0 1 1
0 0 1 0 0 0 0 1 1 0 1 1 0 1 1 1 0 1
The above two rows of digits were created by a random number generator. The digits in some of these columns add up to 0; some add up to 1; and some add up to 2.
Now suppose that I impose the non-random macro requirement: keep the columns whose sum equals 1, and discard the rest. I now have:
1 0 1 1 1 0 0 0 0 0 1
0 1 0 0 0 1 1 0 1 1 0
Each row is still random (at the micro level), but I have now imposed a non-random macro-level constraint on the system as a whole (at the macro level). That, I would suggest, what happens when I make a choice. Choices are holistic events in the brain, which constrain not only the spatial pattern but also the temporal pattern of neuronal firings in the brain, leaving them random at the micro-level, but imposing a distinctive pattern at the macro-level, which varies with the choice being made. A choice, I would suggest, is not typically made at a point in time; rather, it usually occurs over a segment of time – for instance, the length of time taken to perform a deliberate bodily movement (which may be seconds, minutes or even hours in some cases).
Finally, it may be asked how the brain could possess the ability to “discard” (or veto) an ensemble of micro-level states in the brain which does not correspond to the desired macro-level pattern – just as I did when I kept the columns of digits whose sum was 1, and eliminated the rest. The scientific answer is that a combination of feedback and feed-forward processes is known to regulate our voluntary movements, and that the brain continually makes minor adjustments to the motor impulses associated with these movements, even as it executes them. Exactly how it aggregates signals in different regions and rules out the ones it doesn’t want is a question I’ll leave to scientists who study the brain. All that concerns us here is that the aggregation of signals is known to occur in the brain, in connection with motor tasks, so the question of mechanism is a purely academic one.
We are beginning to see how a combination of holistic macro-level properties and micro-level randomness makes it possible for human brains to manifest the elusive phenomenon that we call free will. All that remains is to dispose Professor Coyne’s objection that experimental data have already ruled out free will.
Has free will been experimentally disproven?
In his latest article on determinism, Professor Coyne cites a range of experimental findings which appear to militate against the idea of free will:
Physical determinism of behavior is also supported by experiments that trick people into thinking they’re exercising choice when they’re really being manipulated. Brain stimulation, for instance, can produce involuntary movements, like arm-waving, that patients claim are really willed gestures. Or we can feel we’re not being agents when we are, as with Ouija boards. Further, one can use fMRI brain scans to predict, with substantial accuracy, people’s binary decisions up to ten seconds before they’re even conscious of having made them.
Coyne’s first two objections merely establish that my belief that I freely chose to execute a bodily movement is neither infallible nor incorrigible (and why should it be?). What they do not establish is that my bodily movements are never the result of my free choices – which is an entirely different question. I would suggest that the real reason why people tend to claim that involuntary arm movements caused by stimulating the brain are really willed gestures is that they have an overwhelming psychological need to maintain internal consistency, when giving a narrative of their actions and choices, as brain scientist Nicholas Humphrey points out in his Edge article on Referential Opacity. And the movement of Ouija boards is explained by the mechanism of the ideomotor effect, where the movement of the board is caused by subconscious signals from the brain, which cause a person’s hands and arms to move the board in a certain way. It is misleading of Coyne to cite signals that we are not even conscious of making as an example of voluntary agency.
What about Coyne’s claim that fMRI brain scans can predict people’s decisions up to ten seconds before they’re aware of having made them? A video of the famous “No free will” experiment by John-Dylan Haynes (Professor at the Bernstein Center for Computational Neuroscience in Berlin), has previously been cited by Coyne as experimental evidence debunking the notion of free will. According to the video, an outside observer, monitoring my brain, can tell which of two buttons I’m going to push, several seconds before I consciously decide to do so. I have discussed this experiment in a 2012 article, titled, Is free will dead? It turns out that there are several problems with Coyne’s claim that it refutes the idea of free will.
First, as Coyne acknowledges in a 2011 post on his Web site, Why Evolution Is True, entitled, The no-free-will experiment, avec video, “the ‘predictability’ of the results is not perfect: it seems to be around 60%, better than random prediction but nevertheless statistically significant.” Sorry, but I don’t think that’s very impressive.
Second, the experiment was deliberately designed to exclude the possibility of reflection. In the experiment, as narrator Marcus du Sautoy (Professor of Mathematics at the University of Oxford) puts it, “I have to randomly decide, and then immediately press, one of these left or right buttons.” Now, most people would say that a reflective weighing up of options is integral to the notion of free will.
Third, the experiment relates to an artificial choice which is stripped of several key features which normally characterize our free choices:
(a) it’s completely arbitrary. It doesn’t matter which button the subject decides to press. Typically, our choices are about things that really do matter – e.g. who the next President of the United States will be. And we make choices for a reason, rather than for no reason at all.
(b) it’s binary: left or right. In real life, however, we usually choose between multiple options – often, between an indefinitely large number of options – for example, when we ask ourselves, “What career shall I pursue after I graduate?”
(c) it’s zero-dimensional. Normally, when we make choices, there are multiple axes along which we can evaluate the desirability of the various options. In the experiment described above, there were no axes along which we could weigh up the desirability of the two options (left or right button), as there was literally nothing to compare.
(d) it’s impersonal. We are social animals, and most of our choices relate to other people – e.g. “Whom shall I marry?” Pressing a button, on the other hand, is a solitary act.
(e) it contains no reference to second-order mental states. Typically, when we choose, we give careful consideration to what other people will think of our choice, and how they’ll feel about it. To entertain these thoughts, we have to be capable of second-order mental states: thoughts about other people’s thoughts.
(f) it’s future-blind. The choice of whether to press the left button or the right button is a here-and-now choice, with no reference to future consequences.
(g) it has no feedback mechanism. Not only do choices typically have consequences, but the results of our choices are usually communicated back to us in a way that influences our future behavior. In the button-pressing experiment, nothing is learned by the subject.
Fourth, the experiment described by Coyne made no attempt to evaluate Benjamin Libet’s hypothesis of “free won’t”: “while we may not be able to choose our actions, we can choose to veto our actions.” What happens if the subject is permitted to decide in advance which button they will press, but is also free to change their mind at the last minute? Can a trained outside observer, who is monitoring an MRI scanner, pick up this sudden change of mind on the subject’s part? Coyne does not tell us.
A fifth criticism that can be made of Haynes’ experiment is that the time scale involved makes it meaningless to speak of free will or its absence, just as it would be meaningless to ask what color a hydrogen atom is. Typically, our free choices are preceded by an extended period of deliberation, followed by the brain’s preparation for the execution of a bodily movement, followed by activation of specialized areas of the brain which are responsible for the contraction of specific muscles in the body. It could therefore be argued that freedom is a property which does not attach to the decision to act here and now, but to the entire process leading up to the decision.
So, what can the predictability (60% of the time) of an arbitrary, binary, impersonal choice, which involves no weighing up, no worries about what other people might think, no thought of the future and no feedback, possibly tell us about the existence of free will in human beings? Absolutely nothing.
Adina Roskies, a neuroscientist and philosopher who works on free will at Dartmouth College in Hanover, New Hampshire, wasn’t too impressed with Professor Haynes’ experiment either. “All it suggests,” she declared, “is that there are some physical factors that influence decision-making”, which shouldn’t be surprising.
The arguments I have discussed above are the only arguments against free will which Professor Coyne puts forward in his latest article. It appears to me that Coyne’s case against free will is based on the flimsiest of scientific and philosophical foundations: he fails to adequately define the notion of determinism; his theoretical arguments against free will rest on a misunderstanding of physics and illicitly assume the truth of reductionism and the falsity of holism; he overlooks the possibility that randomness in Nature, while not implying the existence of free will, may help make it possible to realize; and finally, the experimental data which he cites to discredit the notion of free will fail to do so. I conclude that the possibility of free will remains open, and that it can be defended without appealing to dualism.
What do readers think?