The Future of NASA
Michael Griffin is gearin...
More Features
Looking to apply for a Discover Credit Card? Members/Subscribers Log In      
My Problem with Big Pharma
Has Newsweek Sold Out to Big Pharma?
Dark Side of Green, Continued
The Dark Side of Green
The Green Bandwagon
Green Book Award: Nominations Wanted
Wilson Wins “Green Book Award”
The End of Total War?
Does the Desire for Peace Cause War?
[ Full Blog Archives ]
[ Who is John Horgan? ]
[ What is Horganism? ]
Mind & Brain
Ancient Life
All Newsletters
Discover Magazine  Blog  Archives

« Saddam’s Execution and the End of War | Main | Free Will Free Fall »

Free Will Is NOT an Illusion

I choose to muse on free will this morning. The ultimate cause? I’m obsessed with free will, as any sentient creature should be. Proximate causes? First, one of my New Year’s resolutions was to stop drinking. Alcohol. One night and counting!

Second, I just read an essay on free will in today’s New York Times by one of my favorite writers, Dennis Overbye. Overbye confesses to being disturbed by a “bevy of experiments” which “suggest that the conscious mind is like a monkey riding a tiger of subconscious decisions and actions in progress, frantically making up stories about being in control.” I know the feeling. I fretted over some of these same experiments—by Benjamin Libet and Daniel Wegner, among others--in a Times essay published on New Year’s Eve, 2002.

But my faith in free will is stronger now. First, I reject the notion that psychological deliberations have no real impact on our actions. Should I marry this woman? Should I take this job? Should I have one last beer? To be sure, sometimes we deliberate insincerely, toward a foregone conclusion, or fail to act upon our resolution. But not always. Sometimes, psychology determines physics, chemistry and genetics, not the other way around.

Moreover, free will must exist, if some creatures have more of it than others. My daughter and son have more free will—more choices to consider and select from--than they did when they were infants. They also have more than our dog Merlin does. As a mature adult (hope my wife doesn't see this), I have more free will than Merlin and my kids do. I also have more than adults my age suffering from schizophrenia or obsessive-compulsive disorder--or severe, compulsive alcoholism.

I haven’t converted many people into free-will believers with this argument, but it works for me.


Tom Clark


It's of course true that conscious deliberation sometimes plays an important role in decision-making (although John Bargh points out its limitations at ). And of course you have more freedom of opportunity and control than your kids, your dog, the addicted and the insane. But the sort of free will that Overbye was denying, primarily, is the libertarian free will of not being fully subject to causation.

As far as science can tell, the neural processes that underlie (or constitute) consciousness are just as causally determined as any other process in nature at the macro level. And if they weren't, that wouldn't give you any sort of freedom or authorship worth wanting, only add a tad of randomness.

I hope we can agree that we're not causal exceptions to nature, but that this still leaves us with important sorts of freedom, *and* responsibility, see .


Tom Clark
Center for Naturalism

Hal K

What does it mean to say that there is no free will when we don't understand how the mind works? John Horgan has said that he thinks that one day the mind will be understood but that the explanation will be very difficult to understand, like quantum physics. I don't see how this helps his case for free will. If a system can be completely described mathematically then there is no free will within the system. (Here I am setting aside the issue of randomness.) My own view is that the mind will never be completely understood.


You can recognize more options than your children. Just as children can recognize and respond to more options than infants. This is not 'free will'; it is age and experience.

Your examples of schizophrenics and alcoholics serve to show that we are, in fact, highly dependent on an appropriate balance of chemicals in order be able to exhibit what most people recognize as 'free will'. What does *that* say for the definition of 'free will'?

Phil Oliver

The Overbye piece invokes William James, only to shunt him aside unceremoniously ("Get over it, Doctor James.") But for my money, James's is still the most engaging (I started to say "compelling"!) free will defense going. His resolute "first act of free will... to believe in free will" is question-begging, but more importantly it was for him life-starting (see the new Robert Richardson bio on this). Of course our psychological deliberations matter, because experience matters -- and if we directly experience anything at all, we directly experience our own effort and attention as life-changing events both personal and real.

Jason S.

I don't want to be rude, but isn't your argument for free will circular? You’re making the assumption that free will exists when you claim to be able to detect it in different quantities. It's like saying that leprechauns must exist because there are more of them in the woods of Ireland than at the bottom of the ocean, without telling us why we should believe that you’ve found leprechauns in the first place.


Come on, man, define your terms! The reason the "Free Will" question is hard is that nobody knows what it is. Apparently you think it has something to do with consciousness. Well I don't know what _that_ is either.


It's pretty obvious that if we at a small (atomic, or nuclear) size scale, the chaotic nature of the vacuum on such a scale (with particles appearing and disappearing at random and causing all kinds of complicated effects), everything would be completely chaotic and the only rules would be statistical averages.

Because we're on a larger scale, most of the chaos cancels out and free will (even determinism) looks real.

However, if you are honest, you must admit that there is very little free will possible in the first place.

People are all very similar to one another genetically. They all live in the same general environment, the same planet. They all have access to more or less the same news and technical information. Any radical ideas are automatically first dismissed by the majority as being crackpot, eccentric, crazy, and so on.

The greatest example I know is the space rocket. The blueprint of the V2 rocket was found to be virtually identical to Goddard's liquid fuel rocket, just bigger and with a ton of high explosive in the nose.

When one of the captured German rocket scientists (Walter Dornberger I think, maybe Wernher von Braun) was asked to explain the similarity, he replied: "it is not a coincidence, there is just one way to design a rocket!"

This is the problem for determining whether we have free will or not. There are just not endless possibilities out there. If you want to have freedom, you are limited by the way nature is. Increasing the freedom of the designer comes at the cost of reducing efficiency.

There are few ways to do things effectively, so there is little choice.

Most people choose (efficiently) to use the jungle tracks already beaten by others, rather than hack their own path through the undergrowth and obstructions.

The cost of free will can be very high. Really, you need to admit that in many cases there isn't free will. The choices a person has is constrained by the need to earn money to live and to find a partner, or to make a work project work with little risk, on a limited budget, within a limited time. There's very little free will involved in life.

Mike Cook

Of course, if we live in a universe where random numbers are never really random and chance and probability themselves are illusions of consciousness, then free will is a non-starter.

The teaser on a Discover magazine story this month informs us that a single type of RNA may be responsible for our human brains ending up a lot more complex than ape-brains. Presumably, evolution had something to do with this and evolution ultimately depends on mutations, which are hard to get because gamma rays have to penetrate deeply enough through the atmosphere and then through flesh to get to our sex cells.

If this mechanism is really important, we should probably require that astronauts and climbers of Mt. Everest continue to produce children after their high-altitude exploits, as we may need the mutations. In fact, maybe some type of primordial urge drives us to thin air, or no air. Evolution on the bottom of the ocean is really kind of stuck. The mutations would have to happen near the surface then accidentally filter down to the depths.


"Of course, if we live in a universe where random numbers are never really random and chance and probability themselves are illusions of consciousness, then free will is a non-starter."

And what if there are real random numbers in the universe? How would that help? Surely whatever free will means it doesn't mean doing something probabilistically?



I disagree. Probability remains a problem.

A guy tosses an unbiased coin four times, getting - against unbelievable odds - heads each time.

Then another person comes into the room, just as he is about to toss again.

For the first guy, that coin is more likely to be tails because he knows that the run of heads is likely to end sooner or later, and any run of luck one way will on the average be cancelled out eventually (or it would not be random).

To the new guy, who doesn't know the history, the coin is just as likely to give heads as tails.

So even if things are completely random, you can still sometimes make use of random patterns to make useful (although approximate) predictions about what is more likely to happen next.

Sometimes you hear people who don't understand physics making claims that "according to probability theory, all the air in a room could segregate in one corner, although that would be unlikely."

In fact, that is physically impossible, because there are physical constraints on such natural randomness. The mean free path of air molecules is very short, and this sets a physical range on the amount of clumpiness that can ever develop. The fluctuations in density can't become big without work being done, or the laws of energy conservation and thermodynamics will be violated. It's just not a case of probability allowing anything to occur with some likelihood. You can't ever in principle find the electron randomly located 10 metres from the proton in a hydrogen atom. That's not possible physically. The Schroedinger equation predicts otherwise but we have physical evidence that it is incomplete; it is a statistical approximation and isn't real at the limits.

You would have to supply enough energy to the electron for it to escape from the ground state, for it to end up so far from the proton. The Heisenberg uncertainty principle allows any amount of energy to be borrowed, but the time it can be borrowed for is inversely proportional to the energy. So to get enough energy for the electron to escape from from the proton, the time the energy is available for will be inadequate to allow the electron to travel far.

Hence, there are physical limitations on probability.

If everything was governed by probability, there would be no free will. Everything would occur at random. But fortunately, physical laws set a definite constrain on the amount of interference due to randomness. First, the small size of Planck's constant means that chaos is limited to small scales (the waves around moving particles are small). Second, the conservation principles and limits like the speed of light, keep chaos at bay on large scales.

There are probably no purely random numbers in nature. There are always constraints and skewed statistical distributions. Anyone who does statistics knows that in reality, the Gaussian or normal distribution (bell curve for Americans) is skewed. The probability theory and the random number generators are the approximation to nature, not vice-versa. Don't get confused about this. Maths is an abstract tool. You don't really find totally perfect circles in nature.


NC: I don't think I understand your point about a run of heads influencing future coin tosses. No matter how many heads you've thrown in a row, the probability that a toss will be tails is 0.5. It is sensible to expect a run of heads to end, but each toss is independent, no?

Hal K

If you read carefully it seems that nc is arguing the opposite of what Mike Cook is saying. Mike Cook is saying that randomness is essential for free will, and nc is saying that randomness actually diminishes free will. I agree with csrster that we are talking about things that cannot be precisely defined in a scientific sense. A precise mathematical description (probabilistic or otherwise) of a system in which free will exists is impossible. This tells us that either free will is an illusion or that ultimately there can be no precise mathematical description (probabilistic or otherwise) of reality.

Hal K

I agree with Comstock. If anything the guy who just got heads four times might think the coin was biased and that heads was more likely (unless of course he didn't understand probability theory).



Bayesian probability theory allows observers to update chances based on actual observations.

If you don't know whether a coin is biased, you'd start assuming it was unbiased but as a long run of heads developed, you'd increase your estimate of the probability of heads above 0.5 to take account of the incoming data.

If however you are CERTAIN the coin is unbiased and yet by chance get a long run of heads, then you can use logic to estimate that the probability of continuing heads will be less than 0.5.

The likelihood of 5 heads in a row is ~3%. The likelihood of 4 heads in a row is ~6%. So the chance of a block of heads becoming ever bigger becomes smaller and smaller. You can use this logic to overide probability law rules.

You are right than the coin has no memory and the likelihood of heads remains 0.5 on each throw, IF seen individually.

But probability laws describe only ignorance, not knowledge. If you have any knowledge of a non-random run of results, that influences the final result!

The bigger the run of regularity, the greater is the CUMULATIVE chance it will end on the next throw.

If you insist on looking at each result piecemeal, then the chance is 0.5 each time.

As soon as true coincidental "regularity" appears in random data, you can kiss the probability definition goodbye.

The OBSERVER's existence influences probability.

Suppose I toss a coin and know it to be heads, but don't tell you. You estimate it is 50% heads. All you are doing is quantifying your ignorance. Probabilities don't exist in nature without observers, even in quantum theory:

"Thus, the first stage of the measurement process is caused by the physical interaction of the system with the device; the second stage, the reading, involves the conciousness of the observer, as remarked by Wigner. This does not mean that physics is subjective, any more than it does in classical probability. Indeed, once X has been singled out for measurement, and the measurement has been done, the theory reduces to classical probability in its techniques and interpretation." -


Even if you insist that probability is NOT changed by experience, my argument about block results is valid.

Suppose, before I start tossing a coin, I calculate that the chance of a run of 5 heads is 3%.

I would argue that the instant I may find out whether I have a run of 5 heads is just after the last toss, when the coin lasts the final time.

If you insist that probabilities should DEFINITELY NOT be updated each time data comes in, then you are stuck with the initial 3% prediction right up to the final toss!! Hence, at the moment of the fifth toss, the chance of heads is 3% and the chance of tails is 97%.

Hal K

The conditional probability that the last toss is heads given that the previous four tosses were heads is still 50%. The probability of five heads in a row starts at 3% and then increases by a factor of 2 as each successive toss turns out to be heads.


Hal, if you increase the probability based on information coming in, then you're breaking the "rule" which claims the probability is independent of previous results.

(1) If you agree to incorporate information on previous toss results into your prediction, you should be reducing the probability of heads next time, as it becomes less and less probable to keep getting the same thing.

(2) If you deny the inclusion of information on previous tosses into the prediction, you must stick to the original prediction, which is 3% fo 5 heads in a row.


You might want to check out Bayesian probability. Conditional probability is confusing you. It segregates past events and future events as separate blocks of data. In reality the only difference is a lack of information.

Hence if I toss a coin and see the result but you don't, we disagree on the probability of it being heads. I know with 100% certainty what it is, but you don't.

It doesn't matter when the event occurred. Your uncertainty disappears when you get the definite information.

This is the whole issue with entanglement and "wavefunction collapse", Schroedinger's cat, and so on.

It's a complete lack of physical understanding of what probability is. Probability is relative to the observer, depending on what information he has. It isn't real. There is no evidence the electron is in a superposition of spin states until measured, or that radioactive material only definitely decays if you turn on a geiger counter to detect it. That's junk philosophy, like 10/11 dimensional M-theory.

Hal K

nc: Your example about the guy tossing the coin is wrong as stated. If you toss an unbiased coin four times and it comes up heads each time then the probability is still 50-50 at the fifth toss. The person who was there for the first four tosses doesn't have any more information about what will happen on the fifth toss than the person who just walked in. You are just wrong.

I was not breaking any rule about independent trials in my previous comment. I was speaking of the increasing probability of five heads in the case that heads happen to turn up in the first however many tosses. The result for the first few tosses is not independent of the probability of five heads and cannot be considered "previous" to the five heads outcome since they are included in it.

After this thread I am starting to have doubts about you, nc. I am not going to say any more about this.

James McWilliams

Damn... My head hurts after reading all that.

So, what are "free will is an illusion" supporters saying exactly? That everything we do is already layed out for us? Some kind of destiny thing? You cannot escape your destiny, you must face Darth Vader again! :)


Hal K,

You just need to study probability more. A person who knows the result of the toss doesn't have any uncertainty, but the person who doesn't know it does have uncertainty.

If you have a set of 5 trials and your rules forbid any change in the probability after each trial, then the probability remains fixed at 3% for 5 heads in a row, because you refuse to update.

Mike Cook

I have been plunging into Andrei Kirilyuk's papers on dynamical machine consciousness (DMC) to see, among other things, if they speak to the relationship between consciousness and mathematical concepts.

My understanding of DMC at the moment suggests a godlike presence created by any entities with sufficient degrees of freedom in a milieu where they have sufficient chances of interacting through the process of their interactions alone. DMC seems to say that the patterns of interactions inevitably and sufficiently conjure up group behavior that is considerably more than a mathematical abstraction. DMC also seems to say that true randomness and other math principles will thereafter govern the behavior of these material objects. This is a very unusual claim, because when I count a dozen eggs and then multiply 12 by the asking price per egg and a 75% discount, the math involved does not suddenly become a real presence that tells the eggs whether to jump out of the box or not.

But in DMC that does happen. We expect that at the grainy quantum level such things will happen, but DMC arguments reach up into higher realms and suggests that evolutionary mutations, for only one example, might be driven by macro-quantum mathematical principles. (Neuron interaction giving rise to consciousness would be another example.)

I think I like DMC, because the idea of random gamma rays zipping through DNA and just by accident creating a good mutation that can actually be translated into a useful protein that then confers survival advantages seems to me to be about as likely as a person setting out to obtain a particular evolution of Microsoft products by driving up to the Microsoft campus in Redmond and sniping three engineers out in the parking lot every six months until the wished-for changes appear, on the theory that if the old personnel aren't giving you what you want, the new hires will eventually. Blind chemical accidents in place of radiation snipers are the same scenario.

DMC suggests that the cosmic rays can zip through a bundle of DNA almost anywhere and the bundle will react to the stimulus holistically. In fact, the reaction might be to trigger the production of a volutionary "potentiality" that has been hiding, awaiting for the right stimulus from the environment, or even the right pattern of stimuli.

Where would such information hide? Well, there is a lot of "junk" DNA that seems to do nothing but jump up at the odd time to help in a timing or a revolutionary production process. I am intrigued by a 10 or 11 dimensional universe, because objects that seem to be distantly separated in our conventional dimensions might actually be the same point exactly in the curled up dimensions.

Mostly I like DMC a lot because it imagines an abstract mathematical presence that really is quite supernatural. Plus, how can we not look at the matter which makes up our physical bodies and not be impressed with the fact that hidden somewhere in that matter itself is a pre-planned way to deal with almost any environmental crisis that comes along?

DMC as Dr. Kirilyuk presents it does seem to operate on evolution in an intelligent design kind of fashion. There may be some "testing" of new components, but basically matter when arrayed in sufficiently complex patterns knows what to do already when perturbed, about almost anything.

The other noteworthy thing about DMC is that practitioners may put in by hand a number like one hundred billion and proclaim that is the number of neurons in the brain and the number of neuronal interactions is therefore such-and-such a factor higher, but you can't really partition off a part of reality like an individual brain and consider it in isolation. Here on the Internet all our brains are wired together, continuing a group-think interaction venue that has been in existence since before Guttenberg invented printing. Holistic really means holistic in a dynamical machine consciousness universe, but I for one will bet my last hard-earned dollar that Kirilyuk et al's philosophy in this regard is the major scientific insight of our times, far and away so.

The comments to this entry are closed.

Wishful Seeing
Shiny Happy People
20 Things You Didn't Know About... Sleep
Can New Neurons Teach an Old Mouse?
The Woman Who Never Forgets
Why We Get Diseases Other Primates Don't
Vital Signs: Trouble in the Nursery
Natural Selections: The Potential Pandemic You've Never Heard Of
20 Things You Didn't Know About... Death
Natural Selections: The Potential Pandemic You've Never Heard Of
Recently Covered in Discover: The Man Who Finds Planets
Sky Lights: The Dark Side of the Universe
20 Things You Didn't Know About... Meteors
Sky Lights: The Dark Side of the Universe
Islam Hits International Space Station
Neighborhood Watch Goes High Tech
Going Atomic... Again
Jaron's World: The Murder of Mystery
How to Make Anything Look Like a Toy, Round II
Raw Data: The Rigorous Study of the Ancient Mariners
Will We Ever Clone a Caveman?
This Month's Ask Discover
How Life Got a Leg Up
Mammals Stake Their Place in Jurassic Park
You Say "Ook Ook," I Say "Aak Aak"
Guilt-Free Gossip for Greens
A Greener Faith
Whatever Happened To... the Exxon Valdez?
Life After Oil
The Next Katrina
  Full access to all site content requires registration as a magazine subscriber.
© 2005 Discover Media LLC. All rights reserved. Reproduction in whole or in part without permission is prohibited.
Privacy Policy / Your California Privacy Rights | Terms and Conditions | Educator's Guide | Subscribe Online Today | Online Media Kit
Customer Care | Contact Us