The Brain Scans Have Not Found the Decision Before the Decider
(●)The laboratories have watched the brain at work without catching the decision before the decider.
+A nurse coming off a night shift reads a headline on her phone at the bus stop: scientists have found that your brain makes your decisions before you know it. She has spent the whole night deciding things. Just after three she decided to wake the doctor for the man in bed twelve, whose blood pressure was slipping, though she knew the doctor would be short with her. At five she decided how to tell a daughter that her father would not see the morning, and which words to leave out. She weighed each of those. If the headline is right, the weighing was a show put on for her benefit, and the real decisions were made somewhere underneath it, a moment before she arrived.
The headline is not a reporter's invention. The writer Sam Harris puts it plainly, in a short book from 2012: "Free will is an illusion." And he says why: "Some moments before you are aware of what you will do next ... your brain has already determined what you will do."
Everything turns on one word. The headline is really a claim about who, or what, settles the nurse's choices. To settle a choice is to be what makes it go the way it goes, which is more than making its outcome certain. Knowledge and guarantees settle nothing. A colleague who knew for certain that the nurse would wake the doctor would not, by knowing it, have made her do it. She would still have had to go and wake him herself, and put up with his temper. That a thing is certain to happen, or known in advance, is one fact. What makes it happen is another.
In those terms, a choice is free when nothing settles it but the one who makes it, for reasons of his own. He is then its source, the one it comes from. Defining freedom this way is called the source view. If nothing but the nurse herself settled whether to wake the doctor, and she did so for her own reason (the man's slipping blood pressure), then she was the source of that choice, and it was free.
The headline says the sciences have found something else doing the settling. It might be the past together with the laws of nature. It might be the brain, a moment before the person. Or the feeling of choosing, her sense that she was the one deciding, might be a trick the mind plays after the fact.
The first of these has a name. A choice is determined when it is settled beforehand by prior conditions (the past together with the laws of nature, or else a decree of God) that are enough by themselves to make it go the way it goes. Think of a row of dominoes, where the first push and the spacing of the row are enough by themselves to bring the last one down. The view that every event is determined is called determinism. Those who hold that we really are free, and that a free choice is not determined, are called libertarians, a name that has nothing to do with politics.
The argument is a cumulative case: several independent lines of evidence brought to bear on one question. A chain is only as strong as its weakest link, but a rope still holds if one of its strands frays. Here the question is whether science has caught something other than the person doing the settling. The case against free will arrives by four routes:
- the physics of the very small (P1);
- the recordings of brains just before a movement (P2);
- the experiments on the feeling of choosing (P3);
- an argument from the statistical laws of physics (P4).
Each strand examines one route on its own evidence, and none of them leans on another. Think of a defense lawyer facing a prosecution built from a fingerprint, a receipt, a witness and a motive. She answers each piece on its own terms, and each answer is a strand of her case. Suppose one of her answers fails and the receipt stands up: her client really did buy a coffee two streets away that night. The jury can still return "not proven," so long as the receipt does not by itself put him at the scene.
So it is here. Each route against free will, even taken at its strongest, still stops short of the nurse's decision at three in the morning. That is why the case survives the loss of any one strand (C).
Nothing the laboratories have found shows that anything other than the person settles his choices. Science has not shown that we are free either, and nobody should claim it has. Call the result a clear field. A farmer who walks his field in spring and finds nothing that will stop the plow has a clear field, though he has not yet grown a thing in it. In the same way, a clear field is a finding about the evidence, not yet a verdict that we are free. The reasons for believing we are free are a separate case, and they do not come from a brain scan.
Who holds what, on the scientific case against free will:
- Support: Benjamin Libet (1985) concluded that the brain begins a spontaneous act before the person is aware of intending it, though he held that conscious will could still stop the act. Daniel Wegner (2002) called the experience of conscious will an illusion. Sam Harris (2012) gave the popular form. Derk Pereboom (2001, 2014) argues from the statistical laws of physics.
- Oppose: Alfred Mele (2009), Eddy Nahmias (2014) and Adina Roskies (2014) on the brain and psychology experiments; Aaron Schurger and his colleagues (2012, 2021), Uri Maoz and his colleagues (2019) and Liad Mudrik and her colleagues (2022) from inside neuroscience; Randolph Clarke (2010), Timothy O'Connor (2009) and Jason Runyan (2018) against Pereboom.
- Between: Stephen Kearns (2015) recommends agnosticism about free will: believing neither that we are free nor that we are not. Manuel Vargas (2013) recommends a revised idea of it.
(P1)Physics has not shown that every event is fixed in advance.
+Determinism would settle every choice before it was made, and for two centuries many people took it to be the verdict of physics. It is not the verdict of physics now.
(1) The clockwork universe was a picture drawn from Newton.
- Isaac Newton's laws of motion let astronomers calculate where the planets would be centuries ahead. The success was so great that the whole universe came to be pictured as a clock.
- In 1814 the French mathematician Pierre-Simon Laplace imagined an intellect that knew, at one instant, every force in nature and the position of every particle. For such an intellect, he wrote, nothing would be uncertain, and the future would lie open before it as plainly as the past.
- Think of a billiard table. An expert who knew the exact position and speed of every ball could call every collision before the cue struck. The clockwork picture said the whole world is a billiard table, and people are some of the balls.
(2) Quantum mechanics made the clockwork doubtful.
- Quantum mechanics is the physics of the very small: atoms, electrons, light. On its standard textbook reading, many of its laws fix only the odds of what will happen, not what will happen.
- Take radium. Half of any lump of it will turn into other elements in about 1,600 years (roughly the time since Augustine was writing), and the law says so precisely. Yet on the textbook reading, nothing at all fixes which atom goes first.
- The standard encyclopedia survey of the free will debate, a reference philosophers themselves rely on, says the quantum revolution "has made that 'clockwork universe' image at least doubtful at the level of basic physics."
(3) The rival interpretations leave the question open both ways.
- An interpretation of quantum mechanics is a rival reading of what its equations say about the world. The equations make the same predictions on every reading, but the readings disagree about what lies behind them.
- Some readings are indeterministic, leaving real chance at the bottom of things. Others are fully deterministic: one worked out in 1952, for instance, has every particle moving along an exact path, guided by a wave. In the survey's words, "there are competing indeterministic and deterministic interpretations," and no experiment has yet told them apart.
- Think of two detectives who agree on every clue and disagree about what happened. The clues alone do not decide between them, and neither can honestly claim the case is closed.
(4) Whether small-scale chance washes out at our scale is being tested.
- A single coin toss is unpredictable, but a million tosses come out very close to half heads. It was long assumed that brains are like the million coins, so that whatever chance there is among atoms averages away in people.
- The survey reports that this idea, "once common, is now being challenged empirically, even at the level of basic biology."
- It adds that the social, biological and medical sciences "are rife with merely statistical generalizations": rules that say how often something happens without saying what will happen each time. Whether those rules reflect real chance, or only what we do not yet know, is part of what is still open.
- Nobody yet knows how the question will come out, and that is the point. It is a question still being asked, not an answer already in hand.
(5) Physics leaves determinism unproven, not disproven.
- The survey's own summary is that "current science does not decisively support the idea that everything we do is pre-determined by the past, and ultimately by the distant past, wholly out of our control."
- None of this puts freedom in quantum chance. A choice left to chance would be a choice that nothing settles, and that is not freedom either. One defender of free will puts the point exactly: we are held accountable for our choices "because they are not the result of random brain events but are undertaken for reasons which we weigh and act upon."
- Even a deterministic physics would need one more step before it reached the nurse. The survey observes that the arguments from physics rely on "assumptions concerning the way physical phenomena fix psychological phenomena," that is, on the claim that the physics of particles fixes what persons do. That further step is widely assumed, but it is a claim of philosophy, not a finding of physics.
Therefore, physics has not shown that every event, or every choice, is fixed in advance.
(P2)No brain recording has shown a decision made before the person made it.
+The recordings are real, and they are the heart of the popular case. What they show, though, is less than the headline says.
(1) A slow build-up of brain activity comes before a chosen movement.
- In the 1980s a physiologist asked volunteers to flex a wrist whenever they felt like it, and to note where a spot on a fast clock stood when they first felt the urge. Electrodes on the scalp picked up a slow rise in electrical activity, which scientists call the readiness potential.
- The rise began about 550 milliseconds before the movement (a millisecond is a thousandth of a second, so this is a little over half a second). The volunteers reported the urge about 200 milliseconds before it. So the build-up came first by some 350 to 400 milliseconds, a little over a third of a second.
- Later studies went further. A 2008 brain-scanning study predicted which of two buttons people would press up to ten seconds before they reported deciding, with an accuracy of about 60 percent where guessing would score 50. A 2011 study, recording single brain cells, predicted the coming decision to move a finger 700 milliseconds before awareness, with better than 80 percent accuracy.
- None of this is bad science. The basic finding has been repeated many times, and an answer that doubts the data has not read the papers.
(2) What the build-up means is now uncertain.
- A team of neuroscientists proposed in 2012 that the build-up is ordinary background noise in the brain. When a task says "move whenever you like," the movement tends to happen when the noise drifts high enough to cross a threshold, a trigger level. In their words, "the precise moment at which the decision threshold is crossed leading to movement is largely determined by spontaneous subthreshold fluctuations in neuronal activity."
- Averaging creates the rising curve. Line up many trials (the separate runs of the experiment) by the moment of movement and average backwards from it, and a build-up appears in the average even if, on each run, the activity was only drifting until it happened to cross the line.
- Think of photographs of the sea, lined up by the moment a wave broke over a rock and then averaged. You would see a swell rising before every break, though the sea was not planning anything.
- The model has not replaced the older reading by general agreement, and the argument is still running. But the survey states where things stand: "After extensive subsequent study, neuroscientists are uncertain what it signifies."
(3) The experiments studied choices with nothing at stake.
- Flexing a wrist whenever you like, or pressing left rather than right, is arbitrary: no reason favors one moment or one button over another. The only real decision in the room was made at the start, when the volunteer agreed to do the task.
- A 2019 study compared the two kinds of choice directly. Sometimes the volunteer's choice decided which of two charities received a real donation of $1,000, and sometimes the choice had no consequences at all. The build-up appeared for the arbitrary choices and was "strikingly absent" for the deliberate ones, the choices made for reasons.
- This is one study, and it tracked one marker, a single measurable sign of brain activity. What it shows is that the build-up the famous argument rested on was missing where the choice mattered, not that such choices have no brain activity behind them.
- Think of the nurse again. Deciding whether to wake the doctor for the man in bed twelve is nothing like flexing a wrist whenever the mood strikes. Yet the one kind of choice nobody ever thought was deliberated is the kind the famous experiments measured.
(4) Predicting a choice is not settling it.
- A prediction, even a perfect one, is a piece of knowledge, and knowledge settles nothing. The forecaster makes no one do anything.
- What matters is what the forecast is built on. A forecast drawn from earlier brain activity is evidence on one question: whether that activity made the choice go the way it went.
- A forecast that was right nearly every time, made before any weighing began, about hard choices that mattered, would be strong evidence that it did. Easy cases prove less: anyone who knows a good nurse can forecast that she will answer a patient's call bell.
(5) The forecasts so far are modest.
- The forecasts in hand are nothing like that near-perfect forecast. A prediction that is right 60 percent of the time is wrong four times in ten. It can reflect a leaning, a slight readiness one way, and a leaning inclines a person without deciding for him.
- A leaning shows up in the brain because a person acts through a brain, and a brain that is working will show activity before a movement. A typist's fingers move by muscle and nerve every time she types, and nobody concludes from a recording of her nerves that she is not the one typing.
- Think of a friend who guesses your lunch order right six times in ten. He knows you a little. He has not ordered for you.
(6) The scientist who began it did not think he had abolished free will.
- The physiologist who ran the first experiments concluded that "cerebral initiation of a spontaneous voluntary act begins unconsciously," but also that "the final decision to act could still be consciously controlled during the 150 ms or so remaining." In plain words, the brain starts the act before the person is aware of it, and the person could still control whether it goes ahead in the last seventh of a second or so.
- In his reading, the role of conscious will "would be not to initiate a specific voluntary act but rather to select and control volitional outcome." Whatever one makes of that, the man at the center of the story did not draw the conclusion the headline draws.
Therefore, no brain recording has shown a decision made before the person made it.
(P3)No experiment has shown that the experience of choosing misleads us in the ordinary case.
+Psychologists have shown that the feeling of having chosen can be produced where no choosing happened, and missing where a person plainly acted with purpose. Those cases are real, but they are special cases.
(1) The feeling of willing can be fooled.
- In one well-known study, two people moved a computer cursor together over a screen of pictures, using a board like a Ouija board. One of them was secretly working for the experimenters and sometimes forced the cursor to stop on a picture. When the real volunteer had just heard the picture's name through headphones, he tended to feel he had stopped there on purpose.
- The researchers summed it up: "people can arrive at the mistaken belief that they have intentionally caused an action that in fact they were forced to perform when they are simply led to think about the action just before its occurrence."
- In another study, people chose the more attractive of two faces, and the experimenters secretly handed them the other one. Many did not notice, and went on to give reasons for the choice they had not made.
- The feeling can also go missing. Some patients with certain neurological disorders act with evident purpose while sincerely denying that they are the ones acting.
- One psychologist who gathered these cases concluded that the experience of conscious will is "an illusion."
(2) A detector that sometimes misfires still detects.
- Think of a smoke alarm that goes off when the toast burns. There is no fire, but there is smoke, so the false alarm is evidence that the alarm works. It is no evidence that there is never a fire.
- Our eyes are fooled by optical illusions, such as two lines of equal length that look unequal because of the arrowheads on their ends. Nobody concludes that we never see anything.
- The laboratory illusions of will were produced under conditions built to fool: a hidden partner, words piped into headphones, a photograph swapped by sleight of hand. The rest come from rare disorders of the brain, or from practices like the Ouija board, where expectation does the fooling. The ordinary case has none of those.
(3) The illusions work by hijacking cues that ordinarily track the truth.
- On the researchers' own account, we feel we willed an act when a thought about it comes just before it, fits it, and has no rival: "The thought must occur before the action, be consistent with the action, and not be accompanied by other causes."
- Those are the very signs by which we ordinarily tell what caused what. In ordinary life, a thought that comes just before an act, matches it, and has no other cause in view usually is its cause. That is why fooling the sense takes special conditions: a hidden partner, a rare disorder, or a practice built on expectation.
- Think of a detective's rule of thumb: the man seen in the room just before the vase broke, holding a hammer, with nobody else around, is usually the one who broke it. A frame-up exploits the rule. It does not show the rule is worthless.
(4) Science itself relies on people doing what they set out to do.
- Agency is the power to act, and it is effective when what a person intends makes a difference to what happens. The survey makes the point directly: "effective agency is presupposed by all scientific inquiry and so cannot rationally be doubted."
- The experimenters planned the studies, chose the words, and decided when to force the stop. Every step of the argument that choosing does nothing was chosen by someone.
- Think of a scientist who designs an experiment to show that intentions never make a difference. She has to intend to design it, and her results depend on her intentions having made one.
- The survey is careful about how far this reaches. It shows that our intentions make a difference, which is exactly what the illusion argument denies. Whether they are also undetermined takes a further argument.
Therefore, no experiment has shown that the experience of choosing misleads us in the ordinary case: the special cases show a working sense that can be fooled, not one that never works.
(P4)No statistical law of physics has been shown to rule out a person causing his choices.
+The sharpest scientific argument comes from a philosopher rather than a laboratory: if persons caused their choices, those choices could match the statistics of physics only by coincidence. Nobody has shown that the argument works, and there are good reasons to doubt it.
(1) The argument turns on laws that fix only the odds.
- A statistical law fixes how often something happens without fixing any single case, like the law that half of a lump of radium decays in about 1,600 years.
- Suppose the brain events that make up a certain action have, by such a law, a 32 percent chance of occurring. Over many cases the action would then have to happen about 32 percent of the time.
- Now suppose the person himself causes his choices, which is called agent causation: the cause of the choice is the person, not merely his desires or his brain. On this view it was the nurse herself, and not merely her worry or her brain, that caused her choice to wake the doctor.
- Then, the argument says, it would be a wild coincidence if a person's choices kept landing on exactly the percentages the physics predicts. And if they strayed from those percentages, the laws of physics would be broken in his brain.
(2) Nothing in freedom calls for breaking the odds.
- The argument assumes that free choices should be expected to depart from the statistics. Why should they? A free choice is one that nothing settles but the chooser, and the laws fix only how often, not which case goes which way.
- One reply presses exactly this question. It finds nothing in freedom that should lead anyone to expect a free agent to defy the physical statistics.
- Think of an election poll. Pollsters can say within two points how a million people will vote, and each of those people still decides how to vote. Nobody calls it a wild coincidence that the votes match the poll.
(3) The power to choose may itself follow laws that fix only the odds.
- The poll's numbers come from the voters' own leanings, of course. A second reply argues that the odds in physics may be like that too.
- On this reply, a person's power to choose could itself come with odds attached. Given his reasons, he is more likely to choose one way than another, and the physical statistics are the footprint of that.
- Odds of that kind do not settle any single choice. They say how often, and the person settles each time which way he goes.
- Think of a choir. Its harmony belongs to the whole choir and not to any single singer, and it follows rules of its own. A power that belongs to the whole person could follow laws of its own in the same way, including laws that fix only the odds.
(4) The argument rests on untested assumptions about the odds.
- The odds it needs for the brain are inferred, not measured. Physics finds statistical laws for isolated, small-scale events, and the argument reasons that since people are wholly made of the things those laws govern, the same laws govern every physical part of a human action. Nobody has measured the odds for the brain events that make up a real choice.
- A third reply finds a further assumption: that the odds for a given kind of behavior hold steady over time, when the odds of what people do shift over a life as they learn and form habits.
- Think of a man trying to quit smoking. The odds that he lights up after dinner fall month by month as he keeps refusing. Odds that move with what he keeps choosing make a poor yardstick for catching his choices in a coincidence.
(5) The replies are contested.
- The argument's author and others dispute each of these replies, and the dispute is live.
- Even so, look at where the dispute stands. The argument alleges a coincidence. The replies ask why anyone should expect a mismatch, and they offer an honest account of the match.
- Until that account is shown to fail, the allegation is a suspicion rather than a finding. Reducing it to a suspicion is less than proving that a person causes his choices. But it is all a clear field needs, since a clear field asks only that no finding stand in the way.
- Think of an auditor who suspects the books were cooked because every month the figures land exactly on the forecast. The suspicion is reasonable. It lapses once someone shows how honest books would land there too, unless the auditor can show that they could not.
Therefore, no statistical law of physics has been shown to rule out a person causing his choices.
(C)Therefore, science has not shown that anything other than the person settles his choices.
+Nothing the laboratories have found shows that the past, or the brain a moment early, settles our choices in our place, or that our sense of choosing is a trick. The field is clear. But a clear field is a finding about the evidence, not yet a verdict that we are free.
(1) The four strands are independent.
- They rest on different evidence: the equations of physics (P1), recordings of the brain (P2), experiments on the feeling of will (P3), and an argument about statistical laws (P4).
- Two of them concern physics, but they cover different ground. One asks whether the laws fix everything (P1). The other asks whether laws that fix only the odds count against a person causing his choices (P4). Between them they cover both ways physics could turn out, and a reader can doubt either answer and keep the other.
- Think of the defense lawyer again. She answers the fingerprint one way and the receipt another. If her answer to the fingerprint falls through, her answer to the receipt is exactly as good as it was, because neither ever depended on the other.
(2) The strands are not equally strong.
- The strongest strand is the brain recordings (P2). The researchers' own field now doubts the headline reading of the build-up, and in the one study that compared the two kinds of choice, the build-up was absent when the choice mattered.
- The feeling-of-will strand (P3) is nearly as strong. The illusions in the literature come from rigged conditions, rare disorders, or practices like the Ouija board, and none of them shows the ordinary case failing. Besides, science itself presupposes that our intentions make a difference.
- The statistical strand (P4) is the most contested. It brings no new finding, only replies to an argument, and serious people dispute them.
- The physics strand (P1) carries the least weight, because the interpretations of quantum mechanics leave the question open in both directions. It shows only that physics has not delivered determinism, not that physics has delivered its opposite.
(3) The case survives the loss of any one strand.
- Lose the physics strand. That would take a deterministic reading of quantum mechanics winning, or chance washing out at our scale after all (P1). It would put determinism, or something close to it, back on the table for the matter in our brains.
• Determinism would still have to be carried from particles to persons, by the assumption that the physics of particles fixes what persons do (P1). That assumption is widely held, but it is philosophy rather than a finding, and the laboratory work on persons has not supplied it (P2, P3).
• The case would then lean on a contested claim: that nobody has shown the physical world to be closed to a person's causing anything in it (D7).
- Lose the brain recordings, and suppose the build-up really is an unconscious start to a flick of the wrist. That would be a finding about arbitrary flicks, not about the nurse's decision at three in the morning. And even for the flick, awareness arriving late would show that the person noticed late, not that something other than the person began it.
- Lose the feeling-of-will strand, and suppose an experiment showed the sense of choosing misleading us even in ordinary conditions. That would weaken one kind of evidence for freedom, which belongs to the separate positive case. It would not name anything else as the settler, because a sense that misreports what happened does not show who did it.
- Lose the statistical strand, and suppose the wild-coincidence argument succeeded. It would count against agent causation. The libertarian would then fall back on the event-causal account, on which a choice is caused by the person's desires and efforts without being fixed by them. He would fall back, too, on the replies its defenders give to the charge that on their account nobody settles the choice.
• Determinism would gain nothing, since the argument concerns laws that fix only the odds. The argument would still not show that the past, or the brain before the person, settles our choices.
(4) A laboratory cannot see a person settling a choice.
- Nor should anyone expect it to. Agent causation is not a further particle or an extra signal. It is the person, acting through his brain, and a working brain will show activity whenever a person acts.
- Think of a page of a novel under a microscope. Every word is there in the ink, and no amount of studying the ink, however fine the lens, will show you the story.
- What a laboratory can test is the record a person's settling would leave: weighing that makes a difference, choices that answer to reasons, a sense of choosing that tracks what happened. Those are fair tests, and nothing in them has come out against a person who settles his choices (P2, P3, D11).
- Scripture, for its part, pictures a human being as a living creature, formed from the dust and given life by God's breath: "the Lord God formed the man of dust from the ground and breathed into his nostrils the breath of life, and the man became a living creature" (Genesis 2:7). The psalmist looked at how he was made and answered with praise, not suspicion: "I praise you, for I am fearfully and wonderfully made" (Psalm 139:14).
(5) The prediction of a piano-key has not come true.
- In 1864 Dostoevsky's underground man mocked the confident promise of his day that "science itself will teach man ... that he never has really had any caprice or will of his own, and that he himself is something of the nature of a piano-key or the stop of an organ."
- A century and a half of real and impressive science later, it has not taught that a person is a piano-key. It has shown how much of our lives runs below awareness, how the brain prepares a movement, and how the sense of doing can be fooled. It has not found anything other than the person settling his choices.
Whether there is positive reason to believe we are free is a separate question, and its evidence is not a brain scan.
See also:
Philosophy / Free Will: We Have Good Reason to Believe We Are FreePhilosophy / Free Will: The Agent, Not Merely the Agent's States, Settles the Choice
On the brain recordings:
Fried, Itzhak, Roy Mukamel, and Gabriel Kreiman. "Internally Generated Preactivation of Single Neurons in Human Medial Frontal Cortex Predicts Volition." Neuron 69 (2011): 548-562.
Libet, Benjamin. "Do We Have Free Will?" In The Oxford Handbook of Free Will, edited by Robert Kane, 551-564. New York: Oxford University Press, 2002.
Libet, Benjamin. "Unconscious Cerebral Initiative and the Role of Conscious Will in Voluntary Action." Behavioral and Brain Sciences 8 (1985): 529-566.
Maoz, Uri, Gideon Yaffe, Christof Koch, and Liad Mudrik. "Neural Precursors of Decisions That Matter: An ERP Study of Deliberate and Arbitrary Choice." eLife 8 (2019): e39787.
Mele, Alfred R. Effective Intentions: The Power of Conscious Will. Oxford: Oxford University Press, 2009.
Schurger, Aaron, Pengbo Hu, Joanna Pak, and Adina L. Roskies. "What Is the Readiness Potential?" Trends in Cognitive Sciences 25 (2021): 558-570.
Schurger, Aaron, Jacobo D. Sitt, and Stanislas Dehaene. "An Accumulator Model for Spontaneous Neural Activity Prior to Self-Initiated Movement." Proceedings of the National Academy of Sciences 109 (2012): E2904-E2913.
Soon, Chun Siong, Marcel Brass, Hans-Jochen Heinze, and John-Dylan Haynes. "Unconscious Determinants of Free Decisions in the Human Brain." Nature Neuroscience 11 (2008): 543-545.
On the experience of will:
Johansson, Petter, Lars Hall, Sverker Sikström, and Andreas Olsson. "Failure to Detect Mismatches Between Intention and Outcome in a Simple Decision Task." Science 310 (2005): 116-119.
Mudrik, Liad, Inbal Gur Arie, Yoni Amir, Yarden Shir, Pamela Hieronymi, Uri Maoz, Timothy O'Connor, Aaron Schurger, Manuel Vargas, Tillmann Vierkant, Walter Sinnott-Armstrong, and Adina Roskies. "Free Will Without Consciousness?" Trends in Cognitive Sciences 26 (2022): 555-566.
Nahmias, Eddy. "Is Free Will an Illusion? Confronting Challenges from the Modern Mind Sciences." In Moral Psychology, vol. 4, Free Will and Moral Responsibility, edited by Walter Sinnott-Armstrong, 1-25. Cambridge, MA: MIT Press, 2014.
Wegner, Daniel M. The Illusion of Conscious Will. Cambridge, MA: MIT Press, 2002.
Wegner, Daniel M., and Thalia Wheatley. "Apparent Mental Causation: Sources of the Experience of Will." American Psychologist 54 (1999): 480-492.
On physics and the statistical laws:
Balaguer, Mark. Free Will as an Open Scientific Problem. Cambridge, MA: MIT Press, 2010.
Bohm, David. "A Suggested Interpretation of the Quantum Theory in Terms of 'Hidden' Variables, I and II." Physical Review 85 (1952): 166-193.
Clarke, Randolph. "Are We Free to Obey the Laws?" American Philosophical Quarterly 47 (2010): 389-401.
Ellis, George. How Can Physics Underlie the Mind? Top-Down Causation in the Human Context. Berlin: Springer, 2016.
O'Connor, Timothy. "Agent-Causal Power." In Dispositions and Causes, edited by Toby Handfield, 189-214. Oxford: Clarendon Press, 2009.
O'Connor, Timothy, and Christopher Franklin. "Free Will." In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, 2022.
Pereboom, Derk. Free Will, Agency, and Meaning in Life. Oxford: Oxford University Press, 2014.
Pereboom, Derk. Living Without Free Will. Cambridge: Cambridge University Press, 2001.
Roskies, Adina. "Can Neuroscience Resolve Issues about Free Will?" In Moral Psychology, vol. 4, Free Will and Moral Responsibility, edited by Walter Sinnott-Armstrong, 103-126. Cambridge, MA: MIT Press, 2014.
Runyan, Jason. "Agent-Causal Libertarianism, Statistical Neural Laws and Wild Coincidences." Synthese 195 (2018): 4563-4580.
On the scientific stance:
Kearns, Stephen. "Free Will Agnosticism." Noûs 49 (2015): 235-252.
Vargas, Manuel. Building Better Beings: A Theory of Moral Responsibility. New York: Oxford University Press, 2013.
On belief in free will as basic:
O'Connor, Timothy. "How Do We Know That We Are Free?" European Journal of Analytic Philosophy 15 (2019): 79-98.
Reid, Thomas. Essays on the Active Powers of the Human Mind. Edited by Baruch Brody. Cambridge, MA: MIT Press, 1969. First published 1788.
Swinburne, Richard. Mind, Brain, and Free Will. Oxford: Oxford University Press, 2013.
On freedom defined by its source:
Craig, William Lane. "Free Will." Reasonable Faith, Question of the Week #320, June 3, 2013.
Critic sources engaged:
Harris, Sam. Free Will. New York: Free Press, 2012.
Historical sources:
Dostoevsky, Fyodor. Notes from Underground. Translated by Constance Garnett. 1918. Project Gutenberg eBook #600.
Laplace, Pierre-Simon. A Philosophical Essay on Probabilities. Translated by Frederick Wilson Truscott and Frederick Lincoln Emory. New York: John Wiley and Sons, 1902.
Biblical citations:
Genesis 2:7; Psalm 139:14.
+The predictions keep improving: brain scans have forecast a choice ten seconds ahead, and single brain cells a decision to move with better than 80 percent accuracy. Whatever the brain shows beforehand, the brain decided.
1. The predictions are real and worth taking seriously.
- The 2008 scanning study asked people to press a left or a right button whenever they liked. It found patterns of activity that forecast the choice up to ten seconds before they reported deciding.
- The 2011 study recorded 1,019 single brain cells in twelve people making self-timed finger movements. A group of 256 of those cells forecast the coming decision to move 700 milliseconds before the person was aware of it, with better than 80 percent accuracy.
Those are careful studies, and a reply that waves them away has not understood them.
2. What was predicted was a pick with nothing at stake.
Left button or right; this moment or the next. These are arbitrary choices (P2), where no reason favors one option over the other. When a later study let the choice decide where real money went, the slow build-up that had come before the arbitrary choices was absent before the deliberate ones.
A forecast of which way a person will pick when nothing hangs on it tells us little about how he decides when something does.
3. A forecast is not a settling.
Knowledge and guarantees settle nothing, so the forecaster makes nobody do anything. The real question is what the forecast is built on, because when brain activity allows a forecast, that is evidence about what the activity does.
As evidence, these forecasts are modest. The 60 percent one is wrong four times in ten, and even the one right better than 80 percent of the time is wrong nearly one time in five. They show the brain leaning before the person acts. A leaning inclines; it does not decide. A forecast right nearly every time, made before any weighing began, about a hard choice that mattered, would be another matter (P2).
Think of dark clouds in the morning. They make rain likelier, and on four days in ten they blow over by noon. Nobody says the morning clouds decided the afternoon.
4. A person prepares through his brain.
A person who is about to act has a brain that is getting ready to act, and it would be strange if it showed nothing beforehand. The question is never whether the brain is busy before a choice. It is whether something other than the person makes the choice go the way it goes, and a forecast of a pick with nothing at stake does not show that.
Think of a sprinter in the blocks. Her muscles tense before the gun, and nobody says her muscles ran the race.
See also:
Philosophy / Free Will: What Free Will Is, and Why It Matters
+If the build-up is random noise crossing a threshold, then the noise decided. The 2012 model does not rescue freedom; it makes the movement a matter of chance.
1. For the flick of a wrist, that may be exactly right.
The volunteers were told to move whenever they felt like it, and to let the moment come. For a task like that, letting the moment fall where it falls is a sensible way to obey the instruction. The volunteer decided to do the task; the precise millisecond was left to whatever happened to tip it.
Think of a child told to pick any card from a fanned deck. He does not deliberate over the fourth card from the left. He decided to pick one, and let his hand fall.
2. Deliberate choices did not show the build-up.
The 2019 study found the build-up before arbitrary choices and "strikingly absent" before deliberate ones. Its authors concluded that the build-up fits a picture of noisy, random fluctuations that drive the arbitrary decisions and not the deliberate ones.
So, on the best evidence so far, the noise that may time a spontaneous flick is not what drives a weighed decision. Where reasons are weighed, something else is going on.
3. Chance is not freedom.
A choice left to chance would be a choice that nothing settles, and that is not a free choice either. The source view never placed freedom in noise, or in the timing of a flick with nothing at stake. It places freedom in the person settling what he does, for reasons of his own.
Think of a roulette wheel. When the ball drops into red, nobody says the wheel chose red, and nobody asks it why.
4. A weighed choice has a brain story too.
Nobody should conclude that deliberate choices happen without the brain. The 2019 study found one marker missing before the choices that mattered, not a brain at rest (P2). Weighing reasons is something a person does with a working brain, and a recording would show that brain busy the whole time.
Think of the nurse at three in the morning, reading the chart. Her brain was busy all the while she weighed whether to wake the doctor. That is what weighing looks like from the outside.
So is that busy brain the person weighing her reasons, or something settling the choice in her place? That is the question whether the agent herself makes her choices, or only her states do: her desires, her beliefs, the events in her brain. The recordings so far do not decide it either way.
See also:
Philosophy / Free Will: An Undetermined Choice Is Not a Lucky One, and Responsibility Does Not Require Self-CreationPhilosophy / Free Will: The Agent, Not Merely the Agent's States, Settles the Choice
+You cannot choose your next thought. As Sam Harris argues, thoughts and intentions simply appear in awareness from causes you never see, so the sense of authoring them is an illusion.
1. Harris's own words are worth hearing.
"Thoughts and intentions emerge from background causes of which we are unaware and over which we exert no conscious control." The claim is not only about laboratories. It asks you to look inward and notice that you never see a thought coming until it has arrived, and that even a decision to take a thought up arrives the same way.
2. Some thoughts do simply arrive.
The smell of bread brings back a grandmother's kitchen. A worry surfaces in the shower. Nobody chose those, and the source view does not say anyone did.
Other thoughts are taken up. You decide to work out the tip on a restaurant bill, and you do the sum. The difference between a thought that happens to you and one you take up is part of the ordinary experience of choosing, which no experiment has shown to mislead us (P3). Harris's report that even the taking up merely arrives is his reading of his own experience, not a finding of any laboratory.
3. Freedom never required choosing where every thought comes from.
To choose each thought before having it, you would need an earlier thought about it, and an earlier one about that, without end. No one could meet that demand, and nothing in freedom asks it. What freedom asks is that the person settles what he does with what arrives: which worry to act on, which reason to follow.
Grant, too, that the reasons that occur to a person come from causes he never sees. It does not follow that those causes settle which reason he acts on. The objection's premise is about where thoughts come from, and its conclusion is about who settles what is done with them.
Think of a cook. She did not grow the vegetables, and nobody concludes that the farmer made the soup.
4. The argument asks to be weighed.
Harris asks his readers to weigh his evidence and change their minds. That request makes sense only if weighing makes a difference to what we conclude. Whether our reasoning is ours, and what follows if it is not, is part of the positive case for believing we are free.
See also:
Philosophy / Free Will: We Have Good Reason to Believe We Are Free
+The feeling of will is always a construction. Daniel Wegner argued that the mind infers its agency from cues while unseen processes produce both the thought and the act, so the feeling is a guess about the cause, never a sight of it.
1. The mechanism Wegner described is real.
And he described it well. The feeling that we willed an act arises, on his account, when a thought comes before the act, fits it, and has no rival cause in view. The rigged studies (P3) produced the feeling by supplying those signs falsely.
2. A constructed experience can still be accurate.
Vision is constructed too. The brain builds the scene you see from light on the eye, filling gaps and correcting for shadow, and it can be fooled. Nobody concludes that sight never shows us the world. What matters is whether the construction tracks what is there, and in ordinary conditions vision does.
The same holds for the feeling of will. Its cues are the ordinary signs of causation, and in ordinary life they usually point to the real cause (P3). That our intentions make a difference is presupposed by every experiment, Wegner's included. The standard survey of the debate observes that some of the more extreme critics of conscious will, Wegner among them, overlook it.
3. The hidden common cause is a hypothesis about the ordinary case.
A common cause is one cause behind two effects, neither of which causes the other, the way a cold brings on both the sneeze and the sore throat. On Wegner's picture, unseen processes are the hidden common cause of the thought and the act.
The experiments show that the cues can produce the feeling when the thought did not cause the act. They do not show that when the nurse thought about waking the doctor and then went to wake him, her thought did nothing. That is a claim about the ordinary case, and the rigged cases were built by taking away exactly what the ordinary case has.
An argument proves too much when, if it worked, it would also prove things nobody believes. Reasoning from "can be produced falsely" to "is never accurate" does exactly that. Every sense can be fooled under the right conditions, so the same step would show that no sense is ever accurate. A mechanism that fails under rigged conditions is shown to be fallible, not worthless.
Think of a magician who forces a card on you while you feel you picked it freely. The trick shows that the feeling of picking a card can be produced. It does not show that you never pick a card yourself.
4. Mistaken reasons after the fact do not show that nobody chose.
In the swapped-faces study, people gave reasons for a choice they had not made. That shows that our accounts of why we chose can be rebuilt afterwards, and sometimes wrongly. It does not show that the original choice was not theirs: each had picked, by his own lights, the face he found more attractive, and the swap came afterwards. A man who forgets why he bought a coat still bought it.
See also:
Philosophy / Free Will: The Agent, Not Merely the Agent's States, Settles the Choice
+Unconscious influences are everywhere: the order of items on a menu, the music playing in a shop, a word glimpsed a moment before. If our choices bend to forces we never notice, they are not really ours.
1. The influences are real and significant.
Psychology has documented many ways that choices are nudged by things people do not notice. A team of neuroscientists and philosophers reviewing that work in 2022 drew a careful conclusion: "while unconscious influences on decision-making do not threaten the existence of free will in general, they provide important information about limitations on freedom in specific circumstances."
2. An influence inclines without settling.
A shopper who buys the chocolate at the checkout because it sits at eye level has been nudged. He has not been forced, and on another day, with another resolve, he walks past it. The source view never said choices are made in a vacuum. Reasons, moods and surroundings all incline a person, and the person settles what he does.
3. Pressure can lessen blame without taking the choice away.
Think of the nurse at five in the morning, after ten hours on her feet. She still settles how she speaks to the daughter, but tiredness narrows what she notices and makes her reasons harder to hold on to. Fear and exhaustion work that way on everyone, and we rightly judge a person under them more gently.
Findings about unconscious influence are valuable precisely because they show where choosing is hardest. That is where a person should take more care, and where others should be slow to blame.
+Damage the brain and you change the person. A tumor, a stroke or a drug can turn a gentle man cruel, so the person just is his brain, and whatever settles the brain settles him.
1. Nobody disputes that a person depends on his brain.
A person acts through a brain, and injury to it can limit or distort what he does. A tumor pressing on the front of the brain can change a man's character.
Think of a man after a stroke who reaches for words and cannot find them. The damage is real and it limits him, and his family is right to go on treating him as the man who is reaching.
2. The law already treats such cases the way the source view predicts.
When a tumor drives a man to act out of character, courts and ordinary people excuse him, or blame him less. That is exactly what the source view says: what excuses a person for a deed is something other than him settling it. The cases where the disease takes over are the cases where we stop holding the person fully responsible. That fits the conviction that in the ordinary case the person himself is doing the settling.
3. Dependence is not identity.
The objection moves from "the person depends on his brain" to "the person is nothing but his brain." The first is a finding. The second is a philosophical claim, and no scan or injury can establish it. That is because both views (that the person is his brain, and that he acts through it) predict that damage to the brain will change how a person acts.
Think of a house and its foundation. Crack the foundation and the whole house shifts, yet nobody concludes that the house is nothing but its foundation.
4. Even identity would not deliver the conclusion.
Grant, for the sake of argument, that the person is his brain. Then his brain weighing a choice and settling it, with nothing earlier settling it, would simply be the person settling it. The objection needs one more step: that whatever happens in a brain is settled beforehand by earlier events. That step is determinism, and physics has not delivered it (P1).
Think of the nurse. If she is her brain, then her brain weighing the chart at three in the morning is the nurse weighing it.
Christians disagree about how a person is related to his body, some holding that a person has a soul distinct from the body and others that a person is wholly a bodily creature. On either view the objection needs that same further step, and physics has not supplied it.
+A person cannot push atoms. Every physical event has a physical cause sufficient to produce it, and energy is always conserved, so a person who is not simply his brain can make no difference to his body.
1. This principle is not a finding of physics.
The claim that every physical event has a sufficient physical cause, one that is enough by itself to bring the event about, is called causal closure. On the textbook reading of quantum mechanics it is false as stated, since the decay of a radium atom has no sufficient cause of any kind (P1). Its careful defenders therefore state it so that it allows for chance: every physical event has its odds fixed by earlier physical events alone.
Stated either way, it is a principle inferred from the history of science, not a result any laboratory has produced. Biologists once looked for a special life-force in living things and found chemistry instead, and the principle generalizes from successes like that. In its careful form it is close kin to the premise of the wild-coincidence argument, which has not been shown to hold for a brain (P4).
2. Settling is not pushing.
The objection pictures a person's acting as one more push, a shove with energy of its own, added to the pushes already there. Agent causation needs no such push. It says that the person, a living whole with powers, is a cause of what his body does. Where the laws allow more than one outcome, each of them keeping the energy books balanced, settling which one happens adds nothing to the books. That holds whatever a person turns out to be made of.
Think of a ball balanced on a knife edge. Whichever side it rolls down, it releases the same energy; what differs is only the direction.
3. Conservation of energy does not decide the question.
The conservation laws say that an isolated system, one that nothing outside it acts on, keeps its total energy. They do not say which systems are isolated. Whether a brain in which a person is at work behaves as an isolated system is the very question in dispute, and it cannot be answered by assuming it.
Think of a sealed thermos. It keeps the coffee hot only while nothing gets in or out, and the physics of the coffee cannot tell you whether the lid is on.
See also:
Philosophy / Free Will: The Agent, Not Merely the Agent's States, Settles the Choice
+Quantum chance gives only randomness. If physics is indeterministic, then whatever the past does not fix is fixed by nothing, and a random choice is no more free than a determined one.
1. The objection is right about chance.
A choice left to chance would be a choice that nothing settles, and that is not freedom. Nothing in (P1) puts freedom in quantum chance: it shows only that physics has not delivered determinism.
2. Determined and random are not the only options.
The objection offers two: either the past fixes the choice, or nothing does. The source view names a third: the person himself settles it, for reasons of his own. Defenders of undetermined choice often put the point this way: control is not exercised before a choice, by earlier states that fix it, but at the time of the choice, in the very act of making it. We are held accountable for our choices, in one defender's words, "because they are not the result of random brain events but are undertaken for reasons which we weigh and act upon."
Think of a lottery machine and a woman choosing a birthday present for her mother. Neither outcome was fixed a minute before. Only one of them was somebody's doing, and she can tell you why she chose it.
See also:
Philosophy / Free Will: An Undetermined Choice Is Not a Lucky One, and Responsibility Does Not Require Self-Creation
+Derk Pereboom's wild coincidences stand. If persons caused their choices, those choices could match the odds physics sets for the brain only by a coincidence beyond belief, and if they strayed from those odds, physics would be broken.
1. The objection in its author's words.
"The proposal that agent-caused free choices do not diverge from what the statistical laws predict for the physical components of our actions would run so sharply counter to what we would expect as to make it incredible."
In plain words, the argument is a dilemma, one that offers two options and finds trouble in both. If persons caused their choices, then either those choices land, over time, exactly on the odds physics gives for the brain events involved, which would be beyond belief, or they stray from those odds. Straying would mean the laws of physics fail in the brain, and nothing shows that they do.
This is a serious argument by a serious philosopher, and the replies to it (P4) are contested.
2. Matching the odds is not fixing the cases.
A popular form of the objection adds that if the choices must match the statistics anyway, the person adds nothing. But the statistics fix how often an action happens over many cases. They do not fix which case goes which way. A person who settles each case is adding exactly what the laws leave open, and that is not nothing: it is the choice itself.
Think of the election poll again. The poll can say that 52 percent will vote one way. It cannot say how your neighbor will vote, and your neighbor's vote is not idle because the total came out as forecast.
3. The match need not be a coincidence.
If a person's power to choose comes with odds of its own, given his reasons, then his choices matching the statistics is what we should expect (P4). And if the odds of what people do shift as they learn and form habits, as a smoker's odds of lighting up after dinner fall month by month while he keeps refusing, the picture of fixed odds waiting to be matched does not fit a human life.
4. Nobody has measured a match or a mismatch.
The second option assumes that the odds for the brain are known. They are inferred: physics finds statistical laws for isolated, small-scale events, and the argument extends them to every physical part of a human action (P4). Whether laws found for such events hold unchanged in a living brain, where a person is at work, is part of the question, and nobody has measured either a match or a mismatch there.
Think of a tide table worked out for the open coast. It serves the coast well, and no harbor master would trust it for a long, narrow bay until someone had checked, since the shape of a bay can change its tides.
+The honest scientific stance is agnosticism. The laboratory cannot see agent causation, and a believer who needs free will for his theology should not fill the gap with a metaphysics the lab cannot see.
1. It is no surprise that the laboratory cannot see it.Metaphysics is the study of what is real at the deepest level: what a cause is, whether a person is more than his parts. Agent causation is a claim of that kind, not a further particle, so no detector will find it directly. The same is true of determinism: no scan shows that the past settles a choice. The claim that "your brain has already determined what you will do" goes as far beyond the data as the claim it opposes. What a laboratory can test, for either view, is the record it predicts (D11).
2. Agnosticism about the laboratory's verdict is right.
Stephen Kearns argues that the evidence leaves us uncertain about free will, and recommends agnosticism: believing neither that we are free nor that we are not. Manuel Vargas argues that the ordinary idea of free will should be revised to fit what we know. Both take the science seriously, and so should everyone.
Science has not shown that anything other than the person settles his choices (C). That is agnosticism about what the laboratory has found, and it is exactly what the evidence supports.
3. Belief in freedom did not come from theology or from a scan.
The belief is not a theologian's invention. The standard survey of the debate observes that belief in free will seems to be instinctive for most people, built into ordinary experience, and central to how we understand ourselves.
Philosophers from the eighteenth century to our own have held that the belief is itself basic: reasonable without further argument, the way we trust memory and eyesight. Whether those marks (instinctive, built into experience, central to how we understand ourselves) are enough to make it so is debated, and the case for freedom does not need them to be. What is basic is the experience of deciding, which everyone has, and freedom is reasoned to from it as the best explanation of it. So agnosticism is not the neutral ground the objection takes it for, since setting aside the best explanation of an experience takes a better explanation, or a defeater (a good reason to give it up), and science has supplied neither (C, D12).
The survey makes a further point, that "effective agency is presupposed by all scientific inquiry." That point adds that our intentions make a difference, though not by itself that they are undetermined (P3). The rest of the case for believing we are free draws on the experience of choosing and on the conviction that people deserve praise and blame. It is a separate case, and its evidence is not a scan.
4. Waiting for neuroscience to find the decision is a promise, not a finding.
"Give it time" may be said of any open question. So far the trend has run the other way: the famous reading of the most famous result is now doubted by the field that produced it, and in the one study that compared the two kinds of choice, the build-up was absent when the choice mattered.
See also:
Philosophy / Free Will: We Have Good Reason to Believe We Are Free
+The source view cannot lose. If no laboratory could ever see a person settling a choice, then "science has not shown otherwise" is guaranteed in advance, and a claim no finding could touch earns nothing by surviving the findings.
1. The principle behind the objection is sound.
A claim is falsifiable when some possible observation would show it false. "Every swan is white" is falsifiable: one black swan would sink it. A view that every possible finding would leave standing gains nothing from surviving the findings, since it was never at risk. If the source view were like that, a clear field (C) would earn it nothing.
2. The source view makes predictions that could fail.
A person who settles his choices, for reasons of his own, would leave a record, and each of these findings would count against him:
- hard choices that matter, made for reasons, forecast from brain activity before the weighing began, and forecast nearly every time;
- a person's weighing shown to make no difference, so that he does the same thing whatever reasons he considers;
- the sense of having chosen shown to fail in ordinary conditions, and not only in rigged ones.
Think of the nurse. If a scan at midnight had told you, nearly every time and before she had seen the chart, whether she would wake the doctor at three, her weighing would have been shown idle. The experiments of (P2) and (P3) were fair attempts at tests like these, which is why they deserved a hearing, and they fell short.
3. A laboratory tests the record, not the chooser himself.
No laboratory can put a person's settling under a microscope (C). What it can do is check the record his settling would leave, which is how science deals with many causes it cannot see directly.
Think of a chess game watched through its moves alone. You never see the player's plan, but you can test whether there is one: a player with a plan makes moves that hang together toward an end, and a machine rolling dice does not.
+Divide and conquer. Each finding looks weak alone, but together they point one way: the brain prepares before we are aware, the feeling of will can be manufactured, unnoticed influences steer our choices, and physics leaves no room for an extra cause.
1. A cumulative case is fair on both sides.
The objection builds a cumulative case, the same kind of argument as (P1) through (P4), and it is entitled to one. Clues that are weak alone can be strong together, as any jury knows. So the question is not whether each finding is weak alone. It is whether they add up.
2. Evidence adds up only when it favors one view over the other.
A clue counts toward a suspect only if it is more to be expected if he did it than if he did not. A clue equally expected either way counts for nothing, and ten such clues still count for nothing.
Think of a detective who notes that the suspect has two hands, breathes air and was in the city that day. Each is true of the guilty man, and none separates him from anyone else in the city.
Apply that test to the strands:
- brain activity before an act is expected on both views, since a person acts through a brain (P2);
- a sense of willing that rigged conditions can fool is expected on both, since every working sense can be fooled (P3);
- odds in physics are expected on both, unless the wild-coincidence argument works (P4), and physics has not shown that there is no room for a person's causes (P1, D7).
3. Three of the strands share one assumption.
The brain-before-awareness strand, the manufactured-feeling strand and the unnoticed-influence strand all assume that a person acts only where his conscious awareness starts the act, so that whatever begins below awareness is not his doing. Strands that rest on one assumption are one strand, not three: if the assumption fails, they fail together.
And the assumption is doubtful. A pianist plays a run of notes faster than she can think about any one of them, and nobody concludes that the music was played by someone else.
4. What survives shows choosing under pressure.
One strand does point somewhere. Unnoticed influences really do incline people, and the careful conclusion drawn by the neuroscientists and philosophers who reviewed them is that they reveal "limitations on freedom in specific circumstances," not that free will is absent (D5). Findings that mostly count for nothing, plus one that shows real pressure on choosing, add up to a person who chooses under pressure, like the shopper at the checkout. Ordinary life already teaches that much.
See also:
Philosophy / Free Will: What Free Will Is, and Why It Matters