In the building where I usually work, there are four doors on the ground floor.
The main front door, that I usually exit during the day; a back door that I usually enter and leave at the ends of the day; and a left and right side door.
The door to the left leads to a parking lot, with no sidewalk right where the cars turn into the lot entrance, but it is closer to the post office and coffee; the door to the right leads to a nice sidewalk, but it is set back from the main road, and closer to the downtown restaurants.
Two summers ago, the front door was blocked for many weeks by road construction. Inevitably I would forget this until I came out of the elevators and would have to make a choice, turn right or left?
Could you predict which way I turn?
As fans of old science fiction, and "Sliding Doors", know, this is a potentially critical decision.
One way, a late delivery truck might crush me coming around the corner at the wrong time, ending it all! The other way I might bump into an old colleague I hadn't seen for months, triggering a new collaboration leading to Nobel Prizes all 'round and significant betterment of mankind!
So... which way?
Now, in practise, I'm more likely going to grab coffee, and in spite of the less pleasant and safe environment would go left maybe 60% of the time... but I would never know until I actually did it, which way I'd turn.
Could my behaviour be predicted with better than 50% accuracy?
Well, classically it could. Given the initial finite state condition of my mind, and all classical inputs to my brain - a cumbersome but clearly feasible task - a finite state replica of my brain could be constructed, and hence my exact future behaviour deterministically calculated. Just engineering.
And, of course, the root cause of the problem of "free will" in classical science.
But! The Real World is not classical, it is quantum mechanical!
This does make a difference: first of all, my actual state is quantum contingent - as an illustration, consider the possibility of a radionuclide (C-14 or K-40 or actinide) in the testicles of a male.
Radioactive decay is random, truly random near as we can tell, in both time and direction, and for some decays in energy. So the survival of the particular combination of DNA in the actual sperm cell that lead to me is completely contingent on unpredictable quantum mechanical time evolution.
It is hard to get less predictable than that.
But, for free will, we don't just need freedom on some arbitary future time slice from arbitary past initial conditions, that is necessary but not sufficient. An actual evil genius with arbitary computing power could always continually update initial conditions and measure my state of mind a new, to update the predictability. So it should be possible to approach arbitary predictability with enough freedom to measure?
Not quite: there are two problems - one is that measuring disturbs the system - at some point measuring all inputs and states of my brain becomes equivalent to controlling them, which is interesting but misses the point of whether I have free will in the wild, as it were.
But, the real inputs to my brain are also quantum mechanical - the eyes for example are quite good quantum detectors, so what I see, ultimately, is sensitive to the individual photons entering my eye (ok, summer in Pennsylvania is more in the classical limit, but it is the principle dammit), and photon emission has some of the same quantum uncertainties radioactive decay does.
So my input function are intrinsically somewhat uncertain. That limits even instantaneous predictability in principle, if not in practise, in most cases.
What of the mind? Well, the brain is also quantum - and I don't mean to invoke any mysterious coherences in cell organelles etc, the normal brain suffices.
For example, the ionization potential of nerve junctions involves small amounts of charge, and in particular for marginal decision making the go/no-go firing of neurons is sensitive to small changes in potential - such as maybe the ionization trail of a particular cosmic ray?
So whether I turn left or right may depend on a proton from a burp in BL Lac a few hundred million years ago hitting the atmosphere just so as to leave a few extra electron at a critical sub-micron spot in my frontal lobe.
That is bothersome. But useful. Even if it is only the illusion of free will obtained by random cosmic fluctuations I will take it over a certainty of deterministic doom.
However good the Evil Alien Space Bats are at monitoring us, we can still randomly turn the wrong way triggering all sorts of unforeseen consequences. Possibly only in an alternative universe if the "many world" interpretation is right, but that is still something.
BUT, I am not done yet...
all this assumes the Evil ASBs are secretly monitoring us, and predicting our actions, laughing at our foolish notion of free will.
What if we knew someone had constructed a best possible model of our mind and was trying to predict our actions?
How would that affect our actions?
We know, they know, we know, they know... you get the picture.
(of course if they are really smart they will predict both doors and run out the back themselves)
Seriously - there are two possible outcomes - either my mind and their model converges, I collapse in shame and go left, and get run over by a truck.
Or, there is no convergence, just an infinite sequence of +1/-1 states that doesn't converge. Now in practise I must have my coffee and will break the cycle after a finite number of iterations, but is the parity of iterations predictable even in principle? I can be quite ornery, and have no obligation to stop after only 17 iterations, so maybe, if I know to be, I can be truly unpredictable.
And isn't that the essence of free will?
That is, at least a definition of free will: not whether we are predictable in general, but whether we are still predictable if we know that someone is trying to predict our actions.
I contend that for at least a finite and fairly large class of possible binary actions, our predictability becomes no better than 50:50 if part of the input is the knowledge that someone is trying to predict our actions.
- Log in to post comments
And isn't that the essence of free will?
Not as I see it. The essence of free will is that we are fundamentally distinct and discrete entities and hence sovereign rather than the realized outcome of a dynamic process i.e. workings of the brain or whatever. In other words, if the mind subserves the self, then there's no free will; if the self creates the mind (conscious and sub-conscious), then there is free will.
The mainstream dominant metaphysic of physicalism promotes the former view i.e. the mind and self are illusions and their true etiology is as the epiphenomena* of the purely physical brain. There's no free will because there is no thing in actual existence that can possess the free will, just like there's no such thing as a chair, ontologically speaking. It's a dynamic but stable aggregate of a multitude of atoms which themselves are composite phenomena. It's only a feature of the mind that it categorizes perceptions and bundles them as "objects". But if objects don't exist, then there's no such thing as an attribute of th object, except as assigned within our cognition.
In the end, if the self does not really exist, neither does free will.
*Of course, there are functionalist views of the mind as well.
if randomness enhanced our free will we'd carry dice in our pockets
Does predictability alone eliminate the possibility of free will? It seems that if humans were truly rational, many actions could be predicted. Wouldn't we always choose the course of action that resulted in the best outcome for ourselves or for other individuals whose welfare we care about? Of course a lot of considerations could go into a decision about whether to turn left or right. Is coffee good for you? Do you need an additional caffeine stiumulant? Should you be saving your pennies for your childrens' education (or birthday presents)? Should you buy a coffee maker for your office?
I'm not sure how rationality enters into the selection of a favorite color.
By that logic there is nothing with more free will than a free neutron waiting to decay. I really doubt that is what most people mean by 'free will'.
In the philosophic sense, true randomness is just as deterministic as the 'clockwork universe' of Newton. Neither provides any room for 'ghost in the machine' that advocates of free will want to exist.
There remains absolutely no evidence beyond people's 'gut feelings' that free will exists in any form what-so-ever and substantial evidence that our perception of being 'in control' is simply a 'just so' story told by our brain to ourselves quite literally after the fact.
Of course you have no free will. Every morning you showed up at the front door, forgetting the fact that it was closed due to construction.
People's behaviour is a set of patterns and habits. Breaking out of those is hard, but fortunately rare.
Now, I am going to go back to writing a proposal, just like I did this time last year, or last month, or last week....
Yes, free will is a folk psychology concept in theory of the mind, and seems contrafactual to neuroscience.
On the list of physical problems in describing actions of free agents are besides finegrained (quantum) and coarsegrained (classical) contingency also complexity (emergence) and unpredictability (chaos). Plenty of freedom to choose from. :-)
I wonder if the brain is really so sensitive to individual neurons firing... surely this would make your overall mental state extremely unstable? Even if there's a filtering process that prevents single-neuron events from triggering activity at a level where you can reflect on them, presumably you'd find yourself feeling all sorts of inexplicable urges and having all sorts of inexplicable thoughts all the time if those events led anywhere at a subconscious level.
Oh, wait...
I knew you were going to say that!
So, does nobody read the "The Dice Man" by Luke Rhinehart any more?
It is perfectly possible to be predicatble and still have free will - if I have free will I may choose to be predicatble.
Similarly, randomness is not free will, we may be deterministic but with a non-volitional random component.
But, IF we know that our actions are being predicted, and in all cases our actions remain predictable, then it doesn't matter whether we have free will or not, we are effectively acting deterministically.
Conversely IF we are purely acting deterministically, then it is possible to predict our actions completely, given enough (and recent enough) information.
Classical chaos does not help, it just limits the time interval over which predictability is robust, and then the information about the state of our mind must be updated or prediction fails.
In this picture "emergent behaviour" is irrelevant, it is just a failure to encompass all the information, and replacing some information by ensemble averages instead of the microphysical info.
From a functional perspective, you may think you have free will but demonstrably not have it, until proven contrary; conversely an agent may think they do NOT have free will but demonstrably act as if they do.
In practise, quantum effects will at least give the illusion of a free will in that our actions are inherently unpredictable, particularly if we know there is an attempt to predict them.
Whether that actually constitutes "free will" is a question that is either semantic or metaphysical, but not really scientific.
The definitive text on this is the Foundation universe by Isaac Asimov. Asimov, right up into college, wasn't sure if he'd go into History or Biochemistry. Foundation novels were originally a galactic expansion of "The History Of The Decline And Fall Of The Roman Empire" by Edward Gibbon. The latter was itself a kind of trilogy:
Preface
I/ The Decline And Fall In The West ( Chapters 1 - 38)
II/ General Observations Of The Fall In The West
III/ The Decline And Fall In The East ( Chapters 39 - 71)
A Vindication Of Some Passages In Chapters 15 and 16.
When Asimov found a kind of analytic continuity between the Robot, Empire, and Foundation novels he deepend his exploration of chance versus destiny in human affairs. Finally, almost at the end of his life, he applied Chaos Theory to radically reject the foundations of Psychohistory itself.
I really feel that this is the deepest philosophical study of "Free Will" and its paradoxes in human history. Asimov knew the History (and published considerable nonfiction about it), knew the Science (as the greatest Science Writer of the 20th century per his agreement with Arthur C. Clarke), and knew how to present it all in a compelling narrative.
It seems you have defined free will as absence of determinism, while I was discussing descriptions of agents since free will to me seems to be about observing agent behavior and the possibility to distinguish non-predictability.
What we feel ("I may choose to be predicatble") would be about the folk psychology concept, I think.
Well, absence of determinism is necessary, but not sufficient for free will.
So I think we have to start there.
It is still possible to be non-deterministic but not to have free will, merely to be subject
to stochastic processes.
One could, and probably ought to, take into account self-reporting of experiential free will, while remembering that the self is an unreliable narrator.
I think logically, a key issue in determining whether there is anything functionally resembling free will, is the examination of the iterated free will problem - namely whether we can choose to be contrary knowing we are predicted not to be.
Anecdotally, it would seem that this is possible, although taking it beyond gedankenexperiment would seem problematic.
Which implies at least functional free will in principle. We can choose to do stupid unpredicatable things if we know we're predicted not to do so, and what could be freer than that - just ask any two year old, or teenager...
Stein,
I am not sure if I can agree with your statement:
"Given the initial finite state condition of my mind, and all classical inputs to my brain - a cumbersome but clearly feasible task - a finite state replica of my brain could be constructed"
In order to predict your brain activity one would have to decouple it from the environment. I am not sure this is possible (without killing you) in principle and certainly not in everyday situations.
Also, once your brain is sufficiently decoupled from the environment (or perhaps just simulated, but how would we measure the initial state without destroying the original brain?) would it be still "your" brain as it is now ?
Classically measuring the state of the brain is not a problem, since it can in principle be
done with arbitarily delicacy, and the same is the case with the inputs, they can be tapped with infinitesimal perturbation.
Quantum mechanically this is not the case, as I noted, so the initial condition can not be cloned without destroying it. Which is a problem for prediction of course.
The brain state could in principle be probed to the quantum non-demolition limit, which is quite sensitive, but not perfect.
> Classically measuring the state of the brain is not a problem
I am not so sure. Every molecule in your brain interacts via gravity with the whole planet. There is no way to shield this interaction. Since your brain consists of many molecules you would have to measure and then predict their positions and velocities with incredible high precision to forecast just a few milliseconds. (In order to predict the microstate of (ideal) gas in a canister beyond a few collisions one would need to measure positions with accuracy well better than the Planck length - if I remember correctly.)
This means, you probably cannot neglect the interaction of the molecules in your brain with the rest of the planet if you want to predict its microstate.
In order to predict the state of the Earth you need to consider the whole solar system of course ...
... and the system Steinn's brain - earth - solar system necessarily includes the computer and other machinery to determine and predict the state of Steinn's brain.
Since this prediction system cannot know its own microstate,
but couples via gravity to your brain, it may be impossible *in principle* to dtermine and predict your microstate.
In this case 'free will' (as unpredictability) could perhaps coexist with deterministic Newtonian mechanics.
PS: Of course one should not take this too far, since classical mechanics is inherently unstable (as we know since Bohr).
Yes, you definitely assume that any objective definition of free will must be based on determinism.
I'm not so sure that this is either practical or close to what it could mean. On the practical side, how do we observe it? A requirement for reported introspection would preclude animals from having free will, which is an artificial and non-biological demarcation.
On the definitional side, it is suspiciously like the folk-psychology theory of mind concept, which are entirely abstract concepts modeling free agents. But I'm also convinced that an in principle deterministic agent like the innards of the brain or an AI would think he possessed free will, correct or not.
So I would like to cut to the chase and propose that observing the behavior of agents can yield an objective definition. There are problems here too. If we limit our observations to the ability to choose, processes like evolution have free will. (From natural selection.) So the agent must be embodied. (Which hopefully lets AI's be included.)
To rule out too simple systems like dice or toy cars with swerve function, we would also require generality in behavior. (But I know children who are completely convinced that these things have free will. :-)
My choice for definition, which I predict no philosopher will be satisfied with, is that when we observe embodied general agents make unpredictable choices they are exercising free will.
This is coincidentally (or not) also what the theory of mind would conclude as default, unless it is constrained to only accept humans.
With his permission, I am posting some comments by Forrest Bishop, and my reply.
Steinn: "...and hence my exact future behaviour deterministically
calculated. Just engineering."
Forrest: Not so. The measurement problem exists in a classical system as well as in QM. The finite-state-machine emulator cannot be constructed in the real world. Clearly not feasible, and this is another of those "reams of rubbish subjects", pursued by the transhumanism types as well as by the QM-interpretation cottage industry.
Steinn: "...It is hard to get less predictable than that."
Forrest: QM is only one of the ways of interpreting (theory) the apparently randomness of radioactive decay (reality). E.g., a stochastic interpretation ogf QM goes right back to the classical meaurement problem above. There also may be other factors in radioactive decay rates that are completely missing from herdstreamthink. There is some evidence that these are variable and subject to outside influence- this of course would ruin the imputed timeline of the Earth's history among other things.
Jonathan: "... Finally, almost at the end of his life, he [Asimov] applied Chaos Theory to radically reject the foundations of Psychohistory itself."
Forrest: "He also applied sociobiolgy or evolutionary psychology or whatever the PC buzzwords were back then to his later Foundation works, in a rather simplistic manner. Chaos theory is important, but chaotic behavior does not at all remove the prospect of predictability of the psychodynamic structure. cf
www.usurpthis.iwarp.com , bottom of page, for a few terse examples of what makes for predictability of the System. The mathematization (e.g. econometrics) described by Asimov is what is not possible, for reasons explained in great detail by Ludwig von Mises- www.mises.org . This is not the same as saying that Psychohistory is not science.
Jonathan: "Asimov... knew the Science (as the greatest Science Writer of the 20th century per his agreement with Arthur C. Clarke), and knew how to present it all in a compelling narrative."
Forrest: "The above is hyperbole, no further comment."
Jonathan: "My concluding praise for Asimov is not intended as 'hyperbole' but rather (and maybe I should have been explicit) a famous 'compact' between Asimov and Sir Arthur C. Clarke made in a taxicab on (I think) Madison Avenue, New York City. Each agreed to tell any interviewer:
(1) Clarke is the greatest Science Fiction Writer in the World;
(2) Asimov is the greatest Science Writer in the World.
Of course, they were simultaneously hyperbolic and serious. Not quite like Spain and Portugal dividing the New World by drawing a line on a map, but strangely parallel.