Have any observational, i.e., empirical, tests been proposed for determining whether an entity has free will? That is, given an arbitrary entity, how would one test whether it has free will. Asking does not count as a valid test.

Or to put it another way, suppose you as a scientist want to establish a test that could be applied to an arbitrary entity to determine whether that entity has free will. What would such a test look like?

  • 4
    You're going to need to give a definition for "free will" before you can really ask whether anyone has thought of a way to test it.
    – virmaior
    Jun 20, 2017 at 4:55
  • 1
    That's part of the question. As you say, in order to test for it, one must have some sense what one is testing for. If any tests have been proposed, they implicitly or explicitly define what free will means. So this is one way of asking for a definition of free will without opening that question up to what would likely be a not-very-useful discussion. In effect, this is asking for any proposed operational definitions of free will.
    – RussAbbott
    Jun 20, 2017 at 5:34
  • 2
    See Do We Have Free Will?: "The will has also recently become a target of empirical study in neuroscience and cognitive psychology. Benjamin Libet conducted experiments designed to determine the timing of conscious willings or decisions to act in relation to brain activity associated with the physical initiation of behavior. Interpretation of the results is highly controversial. [...] 1/2 Jun 20, 2017 at 6:49
  • 1
    ... "Libet himself concludes that the studies provide strong evidence that actions are already underway shortly before the agent wills to do it." See also Benjamin Libet 2/2 Jun 20, 2017 at 6:50
  • 2
    There are no empirical tests for freewill. No such test is possible. Libet did some interesting experiments but they proved nothing.conclusively. To try to test freewill empirically is to make a major category-error. We're better off studying metaphysics, where freewill can be tested as a logical idea, or studying our own consciousness. This is where empiricism runs out of steam. .
    – user20253
    Sep 19, 2017 at 11:38

7 Answers 7


In 2006 Conway and Kochen proposed The Free Will Theorem and in 2009 offered an emended variant, details, ref and links are to be found in the wikipedia article. Ever Since there has been a lively debate - a few hundred citation of the seminal papers are given by Scholargoogle. A few days ago Kochen posted a new paper , and a year ago in St.Hist.Philosophy of Modern Physics appeared a paper On the notion of Free will in the Free Will Theorem by Landsman .

In 2009 C&K wrote:

if indeed we humans have free will, then elementary particles already have their own small share of this valuable commodity.

This has been widely discussed but an excellent place where their meanings have been unpacked is The Free Will Function Free will from the perspective of a particle physicist by Sabine Hossenfelder. She notes (p6) that the converse is

if elementary particles do not have free will then experimenters have no free will either

Besides Hossenfelder has taken care to offer human-independent definitions e.g. stating

An agent in possession of free will is able to perform an action that does not inevitably follow from all in principle available information at any time preceding the action.

Actually the whole topic (FW Theorems) derives from Bell's seminal ideas about quantum mechanics and experimental work connected to them could hint what relevant tests would look like.

Another approach with experimentals proposals, mixing QM and neurology (like Libet's), has been reacently published by Nayakar C.S.M., et al (2014) Consciousness, Libertarian Free Will and Quantum Randomness. In: Menon S., Sinha A., Sreekantan B. (eds) Interdisciplinary Perspectives on Consciousness and the Self. Springer, New Delhi.

  • Actually addressing the question in it's own terms! Top marks. I am intrigued Hossenfelder has worked on this, her ideas on the dubiousness if beauty as a guide in physics is controversial and interesting. I hope to get a copy of the paper when libraries reopen
    – CriglCragl
    Apr 5, 2020 at 20:31
  • This theorem has nothing to do with free will as normally understood by philosophers. If quantum noise is truly random, and you program a computer to read a bit from /dev/random and output either "steak" or "fish", that's free will by their definition. The theorem doesn't imply the existence of an empirical test for true randomness, so it doesn't give you a way to test for even this weak and uninteresting notion of free will. Also, Sabine Hossenfelder's opinion on this particular subject is not representative, since she believes in superdeterminism, which hardly anyone else does.
    – benrg
    Mar 30, 2023 at 21:19

There cannot be any successful experiments because free will cannot be proven. In any case, whatever the results of any experiment would be, it could be interpreted as determined.

The complete history of the universe including every "quantum jump" in atoms, nuclei and particles could be like the determined frames of a movie. I do not say that I believe in that possibility but at the end of times God could lean back satisfied that everything has happend as he had scheduled it. (Of course it must have been a boring movie for him.)

  • 2
    By this logic there could never be successful experiments proving relativity since with enough ingenuity one can fit all available data into absolute space with Lorentz's ether. For that matter, it can not even be proven that we are not plugged in into a Matrix and fed hallucinations. The standard of success in science is not to be beyond idle philosophical skepticism. Which means that there may well be experiments interpreted as either proving or disproving "free will" scientifically, one certainly can not decide this a priori from an armchair.
    – Conifold
    Sep 18, 2017 at 20:57
  • @Conifold: This is not logic but simply a point of view that cannot be falsified. The experiments to check for free will do not attempt and cannot attempt to disprove this view but are designed under the premise that the world is not deterministic but ruled by the laws of physics which include quantum theory and its indeterminism. I only wished to point to the fact that a complete determinism of all and every cannot be excluded.
    – Hilbert7
    Sep 19, 2017 at 13:36
  • On "did not" you are mistaken, that is exactly what Libet and other neuroscientists are attempting, as for "can not" neither you nor I can possibly know. The experiments are not designed under either deterministic or indeterministic premises, they suspend judgment on the matter and observe what happens. It is true of course that we can not "prove" that matter is made of atoms, or that desks and chairs do not disappear behind our backs, but beyond reasonable doubt need not be beyond skeptical doubt.
    – Conifold
    Sep 19, 2017 at 17:57
  • 1
    There is no empirical test for freewill. There is no such test for consciousness. This can be calculated while sitting in an armchair. I'd call it common sense, but whatever it is it's not difficult to work out. . .
    – user20253
    Oct 5, 2017 at 14:31
  • @Conifold There are plenty of tests that distinguish between relativity and Newtonian mechanics. Can you come up with one that would distinguish between determinism and free will? Sep 26, 2018 at 14:55

Every time you make a conscious decision, you generate an observation that is consistent with the concept of free will. If you decide to vote for this party or that (or not to vote at all), that is an observation consistent with he concept of free will; if you decide to have that piece of cheesecake (or skip it, because you need to take off a few pounds), that is an observation consistent with the concept of free will. There are people who have an extremely difficult time making certain kinds of decisions: e.g., addicts of various sorts, people with obsessive compulsive disorder or impulse control disorders, etc. But the self-evident fact that people do overcome these conditions, and often struggle with them even if they cannot overcome them: that too is an observation consistent with the concept of free will.

This doesn't imply subjectivity (at least no more subjectivity than any other empirical observation). If I ask you what decision you made or are going to make, you will give me an answer, and I can observe the outcome; those are external observations consistent with the concept of free will.

Now, any of these observations may someday be proven to be deterministic. Science would have to advance to a point where we can make detailed, real-time analyses of the human brain in action, to show that the brain mechanisms have no points at which 'free will' might have operated. But we are a long, long, long way from that. It's unwise (though often entertaining) to speculate on what future science might show us, because future science has a way of surprising us. Note the total absence of those flying cars that were such a staple of mid-twentieth century science fiction, or some of the sillier ideas spun out by H.G Wells...

So what he have on one side is a theory that free will does not exist (an extension of the Newtonian 'clockwork universe' concept applied to humanity), and a large body of observations which seemingly contradict this theory and support the theory that free will does exist. There is a strong parallel here to the creationism/evolution kerfuffle. We have no clear and direct evidence of evolution in action, but lots of observations that fit within the picture that evolution paints. Mostly we consider it common sense to accept evolution, but many people still cling to that evidence-free presumptions of creationism. Yet somehow the people who are most ardent about embracing the observations backing the theory of evolution are most likely to refuse to consider the host of observations that support the theory of free will. It is a peculiar inconsistency that has more to do with the whole-cloth rejection of religious ideation than with any real appreciation for the scientific method.

But I digress...


I'll summon all my moral courage and provide this controversial answer: the test that rules out free will is whether we can predict the behavior of the subject. If we cannot, then free will may be present. We can strengthen this conclusion inductively to the point of practical certainty, as is always the case with induction.

(Per my kind first commenter's recommendation below, I'm adding some references in parentheses. It will take a bit to complete these edits, please be patient!)

To explain, please permit me to start from basics. We disprove assertions using the contrapositive, we prove assertions, always provisionally, using induction. One of the surest conclusions of science, inductively demonstrated a thousand times a day across the globe and through the heavens, is that natural processes are predictable and amenable to representation by "physical law". We can refine those laws to predict outcomes of natural processes to 1 part in 1012 or better. (Hoyle's C12 resonance prediction based on physical processes was good to about 1 part in 1012; multiple physical processes yield QED's fine-structure alpha values consistent to about 1:109, and we have G values across many physical processes consistent to about 1:103, the black sheep in the fold [Nature (2014) doi:10.1038/nature13433]. See also A physicists guide to skepticism, p. 102ff.) Never in any of these predictions do the phenomena ever require us to invent the concept of "free will" to understand them. Determinism reigns. Predictions are not only possible, they typically are of amazing precision and ongoing validity. We send probes to Saturn on the strength of these laws.

Under naturalism's view we human beings are no more nor less than natural phenomena. (see esp. section 1.3 in the linked page.) Yet somehow we have been required to invent the concept of "free will" to explain the behavior of living things. The clear reason for this is that living things behave unpredictably. In the case of human beings, the behavior is so unpredictable that we struggle to maintain civil society and even risk self-annihilation. (While there are many references that represent habits as predictable behavior, I find none so far that accurately represent the validity and unpredictability of arbitrary choice by humans. So for a reference accessible to anyone, I recommend you simply predict what your friend or spouse will say first at any given point in time. If you're not successful 100% of the time, just go find someone else, or some computer algorithm, that can do so with 100% success. If you can't find such a person or thing, this should clearly show that we humans are routinely capable of unpredictable behavior in utter contrast to the predictability of nonliving phenomena. You may distinguish between practical, positivist unpredictability and unpredictability in principle as I discuss below, but Occam's Razor will not be your friend, I think.) So the simple test for free will can be, "Construct a predictive law that works with a comparable degree of success as the laws of physics to predict the behavior of the subject in question." If this cannot be done, then the factor we call "free will" could be in operation.

The mainstream response to this point is typically to vigorously restate the assertion that we living things are simply natural objects governed completely by deterministic physical law, and that is that. The only reason we can't predict the precise behavior of living things, opponents will say, is that living things are the most complicated natural structures in existence, which means the problem is just too hard. But someday we will be able to make such predictions.

I would argue in reply that a truly scientific approach to this question means that in the complete lack of any evidence that such predictions can be made (that is, no one can ever predict what another person will do next in the way one can predict where Saturn will be next) we will refuse to make unsupported assumptions. If you cannot predict precisely what I and others will do next, you can tell me all day long that my behavior is predictable with the utmost precision because all of me is completely determined by natural law, but I have no reason to believe you from a scientific point of view. You may try to convince me using logical inference in the absence of any supporting data for the premise, but we normally consider arguments like that to be of the weakest kind.

So the test would be, "Is the subject's behavior predictable to a degree which matches other nonliving physical phenomena?" If so, then free will is not a factor. If this prediction is not achievable (and Libet's work does not achieve it, by the way) then free will must be considered. Moving from this to a firm conclusion that free will exists in the world then becomes a matter of inductive proof.

Please reply kindly.

  • People normally don't reply unless they have a question. It is not a forum, but they do vote. You could strengthen your answer if you provide sources to help back up your position. These would give a reader a place to go for more information. Welcome to this SE! Sep 23, 2018 at 12:24
  • I strongly agree with this approach. I see identity and character as heuristics that make predictions of others tractable. When prediction of others becomes intractable using realistically available resources, identity and character become irreducible, causal agency enters the explanatory narrative as much as particles and fields. Prediction has to be the fundamental arbiter, and not just in theory, in practice, until we can do better.
    – CriglCragl
    Apr 5, 2020 at 20:36

Honestly, the biggest problem with even approaching such debate with an attempt to experimentally strengthen one side of the debate (note that I don't even say "prove") is that there are so, so many views on the spectrum of debate, so much that declaring strengthening one view would immediately strengthen a similar view that supports the opposite side.

A second problem, immediately related to the first one, is that no matter how you attempt to test such experiment, you'll end up with a serious amount of premises that won't be acceptable by the majority of the views outside of your test subject. This will result in an easily refutable experiment.

On top of all that, these two problems come with the very controversial assumption that free will can even be tested, directly or indirectly, as Heinrich's answer claim to the contrary.

  • Don't they broadly fall into libertarian free will, compatibilism, and no free will? It seems you don't address the question. Does your answer suggest free will is incoherent or meaningless?
    – CriglCragl
    Apr 6, 2020 at 9:19
  • @CriglCragl those are the 3 big "camps", so to speak, yes. But at least from what I've seen, almost every experiment could have been (and have been) reinterpreted into a different camp very easily because each camp is so wide and the line between them so thin. What I suggest isn't that free will itself is meaningless, rather that using experiments to support it, is. Apr 6, 2020 at 11:07
  • Ie. Unfalsifiable, so meaningless in science, not in truth a hypothesis, but metaphysical. I don't think that's true. Compare to the interpretations of quantum mechanics: you can make the same case, no experiment done or proposed. But that does not mean there cannot be such experiments, and the expectation is there will be, is what makes the measurement problem part of science, even though the experiments have not been clearly articulated.
    – CriglCragl
    Apr 7, 2020 at 13:40

I think that it's easy to do, as long as you can empirically identify the person. If empiricism can see the person, then empiricism could learn whether that person is guilty or honorable. And a person could only be guilty or honorable if the actions of their will had been free.

But persons are invisible to an empirical study, owing to difficulties such as the mind/body problem and the soul. So for this reason, I suspect that it will be impossible to measure guilt or honor.

  • Couldn't you make an arbitrarily accurate simulation, from observations, then interrogate that as a proxy? Isn't that effectively, what we do for each other? There are more fundamental problems to reconciling free will with physicalist-materialism surely
    – CriglCragl
    Apr 7, 2020 at 15:29

If A implies B, then not B implies not A.

If the human will transcends physical law, then alcohol does not impair speech; alcohol impairs speech, therefore, the human will does not transcend physical law.

The apostle of free will maintains that a man can by will avoid getting drunk, but he does not maintain that when drunk a man can say "British Constitution" as clearly as as if he were sober. --Russell Bertrand. Why I Am Not a Christian.

As a matter of fact, even the US military does not believe in free will in practice. Nowadays alcoholics are ordered to visit psychiatrists in addition to extra duty; an under-performing soldier is checked for medical conditions such as dehydration and heat stroke before being disciplined for lacking fortitude. PTSD is widely recognized as a medical condition instead of cowardice.

Sleep requirements before flight duty period are taken for granted everywhere; it will be absurd if someone asks a pilot to use his willpower to get over his fatigue.

Noise wrecks mental health and productivity. The US military acknowledges this fact and issues ear plugs in noisy work environments and enforce very strict noise discipline in living quarters. Yet, in today's software industry, there are managers who firmly believe that one can use yoga to tough out noise; they will tell you to practice mental yoga when the noise becomes unbearable.

  • 1
    -1 I don't think this is true: "If the human will transcends physical law, then alcohol does not impair speech" I don't think free will must be without any constraints whatsoever. All one needs is one event which is free to counter determinism which requires that all events not be free. Jan 8, 2018 at 1:40

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .