Thursday, 14 March 2013

Imponderable II: Free Will

I began the first Imponderable by bagging on Jim Flynn’s ideas about free will. I’m terribly sorry to give an unbalanced impression of Flynn, who is a first-rate political theorist, but I’m going to have to do it again. Some day I will do a political blog post which pays proper tribute to Flynn’s immense positive contributions to human understanding. For now, I’m afraid Flynn’s very clarity and force of expression make him the best starting-point for exposing the confusion in the traditional Western concept of free will.
The concept of free choice is perfectly coherent and easily stated. Free choice, to the extent that it is real, would be an uncaused cause. It is the opposite of what we call an epiphenomenon. A good example of the latter is the reflection of a tree in a pond: if you cut down the tree, the reflection disappears; but if you drop a rock on the reflection, the tree is unmoved. An epiphenomenon is all effect and no cause. If free choice exists, the present self has a genuine choice between (at least) two alternatives and creates a future that would not otherwise have existed. If we decide to pick up hitchhikers as an act of charity at a greater risk to our lives, the world will be different: more hitchhikers will get to their destinations quicker and some extra lives will be lost. Free choice breaks the flow of the world from past to future and thus the result is what philosophers call “metaphysical discontinuity”.
Jim Flynn, Where Have All the Liberals Gone? p. 265

Selves All the Way In

The word for someone who holds Flynn’s views on free will is libertarian, which is confusing, because there is also a political leaning called libertarianism, and this Flynn is most certainly not – he’s a passionate Social Democrat. Sam Harris, on the other hand, is a political libertarian but a metaphysical hard-determinist, which is to say he rejects free will outright. I mention Harris because he recently wrote a little book titled Free Will. It was available for free online for a short while, but I’m afraid I missed it, and the libraries in Dunedin haven’t got hold of a copy yet. So I can only quote directly from the bits that made it into the Amazon.com preview, and hope the related posts on Harris’s blog don’t miss anything important.
Free will is an illusion. Our wills are simply not of our own making. Thoughts and intentions emerge from background causes of which we are unaware and over which we exert no conscious control. We do not have the freedom we think we have.
Sam Harris, Free Will p. 5
We’ll see more from Harris later. The odd thing about (metaphysical) libertarians and hard-determinists is they both use the same definition of free will, which is the one given by Flynn above, invoking an “uncaused cause”. Now, that could mean one of just two things. I gave causation a quick going-over when I speculated about the nature of time, and basically it comes down to “Whenever A happens, B happens.” Of course anyone nerdy enough to be interested in this blog will long ago have had it drummed into them that correlation doesn’t prove causation, but that’s because A (whatever it might be) tends to go along with C and D and E and F as well, and it’s hard to tell, most of the time, which ones would stop B from working if you took them away. Therefore, the only sensible thing that “uncaused” could mean is that there is no A to be found for the B in question; nothing that could predict it beforehand. The other possible meaning, which I’ll get to shortly, is not sensible.
It should be clear that this kind of uncausedness is a lousy basis for free will, but some people have hung their hopes on it. An otherwise excellent book, which has helped demystify quantum physics for me to a great degree, identifies free will with indeterminacy:
This is an old problem in a new guise. In the heyday of classical physics it seemed that every physical event in the universe must be determined absolutely and predictably from previous events, because the rule of cause-and-effect was inviolable. But then everything in the universe must have been preordained from the outset, so that the absolute determinism of classical physics apparently negated the possibility of free will. No-one ever resolved that conundrum.
David Lindley, Where Does the Weirdness Go? p. 221
But subatomic particles, such as might behave quantum-mechanically, are only relevant to the activity of the brain insofar as they are the components of chemical reactions; and chemical reactions are no less deterministic inside human brains than they are in, say, laboratory flasks, or batteries, or tubs of soapy water, or anywhere else where we rely on them working deterministically en masse. More importantly, random chance is at least as bad for “free will” as determinism. If our actions are random, then nobody, not even we ourselves, can say why we do the things we do – there is no why. Flynn knows this perfectly well, and denies holding any such position.
Something can be indeterminate and not worthy of praise or blame. A random event like an electron jump cannot be praised or blamed because no present self exists faced with open alternatives and freely choosing between them. The same is true of a clock. The existence of a present self of this sort is the crux of the matter. You do not praise or blame a Mexican jumping bean simply because it is unpredictable.
Jim Flynn, Where Have All the Liberals Gone? p. 271
The “present self” is a problematic concept (to say the least!), as we’ll see below. Steven Pinker articulates the trouble with indeterminism with customary clarity:
But a random event does not fit the concept of free will any more than a lawful one does, and could not serve as the long-sought locus of moral responsibility. We would not find someone guilty if his finger pulled the trigger when it was mechanically connected to a roulette wheel; why should it be any different if the roulette wheel is inside his skull? The same problem arises for another unpredictable cause that has been suggested as the source of free will, chaos theory, in which, according to the cliché, a butterfly’s flutter can set off a cascade of events culminating in a hurricane.
Steven Pinker, How the Mind Works pp. 54–55
Anyone who doesn’t fit into the libertarian or hard-determinist camp is called a compatibilist, but since the only thing they share is their rejection of libertarianism and hard-determinism, I don’t know that we can meaningfully speak of “the compatibilists” as a school or movement. Pinker seems to be one.
Either we dispense with all morality as an unscientific superstition, or we find a way to reconcile causation (genetic or otherwise) with responsibility and free will. I doubt that our puzzlement will ever be completely assuaged, but we can surely reconcile them in part. Like many philosophers, I believe that science and ethics are two self-contained systems played out among the same entities in the world, just as poker and bridge are different games played with the same fifty-two-card deck. The science game treats people as material objects, and its rules are the physical processes that cause behaviour through natural selection and neurophysiology. The ethics game treats people as equivalent, sentient, rational, free-willed agents, and its rules are the calculus that assigns moral value to behaviour through the behaviour’s inherent nature or its consequences.
Free will is an idealization of human beings that makes the ethics game playable. Euclidean geometry requires idealizations like infinite straight lines and perfect circles, and its deductions are sound and useful even though the world does not really have infinite straight lines or perfect circles. The world is close enough that the theorems can be usefully applied. Similarly, ethical theory requires idealizations like free, sentient, rational, equivalent agents whose behaviour is uncaused, and its conclusions can be sound and useful even though the world, as seen by science, does not really have uncaused events. As long as there is no outright coercion or gross malfunction of reasoning, the world is close enough to the idealization of free will that moral theory can meaningfully be applied to it.
Steven Pinker, How the Mind Works p. 55
I don’t know that the card game analogy works very well. If there exist any two card games where the cards can make the same movements between the players and the table, the rules of neither ever being violated, and yet be working on different rules (so that, at the end, there would be two different winners depending on which game you were reading into the cards) – then somebody point them out to me, because I can’t imagine such a thing. And surely it only gets more unlikely when we move from the simplicity of a deck of cards to the hyperbole-defying complexity of the brain. Yet the ethics game works well enough, in the very same world that the science game discovers, that we can play it for a lifetime without collapsing in absurdity. Somehow, the reality in our brains must approach the idealization of a free, sentient, rational agent.
But the answer cannot be that there is a free, sentient, rational agent somewhere within the brain. Even if that were so – if the frontal cortex, say, turned out to act as an agent in its own right – that would not count as an explanation. And this is where Flynn goes wrong, and where the third alternative (besides determinism and “jumping-bean” indeterminacy) reveals itself for the nonsense it is. Last time, I asked you to imagine explaining the functioning of a car’s engine by positing a little car running on a treadmill under the bonnet. Flynn’s “present self” is just such a little car. If an entity can only be a free-willed, choosing agent by having a smaller free-willed choosing agent, its “present self”, inside it, then for the “present self” to be a free-willed choosing agent it must have its own little “present self” making choices inside it; and this must have a still smaller self, and so, literally, ad infinitum. It must be selves all the way in. If at any point the regress stops, then the innermost self must be either a deterministic system or a jumping bean. If it is, nevertheless, an agent with free will, then it is not true that a free-willed agent can’t be made up solely of deterministic and/or jumping-bean parts. If it isn’t an agent with free will, then on Flynn’s own logic neither are any of the onion-layers of larger selves around it, including the whole human being.
Actually, we know for a fact that the brain doesn’t go all the way to being an ideal free sentient agent, because of the work of a neuroscientist named Benjamin Libet. He attached EEG electrodes to his subjects’ heads and sat them in front of a display with a rapidly-circling dot. They were to perform a small, one-hand action (pressing a button, twitching their wrist, flicking a finger; Libet varied this part of the experiment) whenever they felt like it, and report where on the screen the dot was when they decided to do it. Their reports showed that they became conscious of the decision a ponderous 0.3 to 0.5 seconds after their brain actually initiated the movement. The “present self” turned out to be merely a commentator after the fact.
The philosopher Daniel C. Dennett discusses the implications in detail in Freedom Evolves. Anyone who follows this kind of stuff will know that name. Dennett is the most prominent of all the compatibilists, and no serious thinker on free will, consciousness, or the self can afford to ignore him. Those of you who’ve read Freedom Evolves and its predecessor Elbow Room, and since you’re here reading a blog about free will I’d imagine that’s quite a large percentage of you, will recognise his influence throughout the rest of this essay. According to Dennett, there are quite a number of hypotheses consistent with Libet’s data, all of them variations on the theme of “Consciousness takes time”. However,
Libet’s data do rule out one hypothesis, which might have been our favourite: Self-contained You, according to which all the brain’s chores are gathered into one compact location, where everything could happen at once in one place – vision, hearing, decision-making, simultaneity-judging... With everything so handy, the timing problem couldn’t arise: A person, a soul, could sit there and make free, responsible decisions and be simultaneously conscious of making them, and of everything else going on in consciousness at the time. But there is no such place in the brain.
Daniel C. Dennett, Freedom Evolves pp. 237–238 (emphasis and ellipsis original)
Flynn cites Dennett, but dismissively; apparently he does not see that his definition commits us to selves all the way in. I’m afraid he resorts to a quite unbelievable degree of special pleading:
[Dennett] rejects the reality of radical free will on the grounds that it cannot be reconciled with scientific explanation. But that does not justify salvaging science at the expense of free will. Why assume that we must reject the reality of free choice if that renders part of reality beyond scientific explanation? Why not assume the reverse: that we must recognise a limit on science if uncaused causes are part of reality. Dennett has a field day demolishing those who try to fit free choice into some niche created by scientific explanation (such as indeterminacy). They are mistaken to try. Science excites our admiration because of the wonderful explanations it has given us about the world thus far. But no-one has evidenced the hypothesis that all of reality is susceptible to scientific explanation.
Jim Flynn, Where Have All the Liberals Gone? p. 272
...any more than anyone has evidenced the hypothesis that all English-language texts are susceptible to being read by literate English-speakers. What modes of investigating the world, exactly, does Flynn exclude from “scientific explanation” here, and on what grounds? What are we to suppose is happening inside the brain, or out of it, when someone makes a decision? A flash of magical energy from nowhere? Mine, I have to say, has been remarkably boring. I’ve never once noticed even the slightest change to the temperature or mass of my head. Politically, a supernatural soul that can always choose (or be blamed for not choosing) to work hard and obey the law and not get pregnant out of wedlock, regardless of any stress, deprivation, or social disempowerment its owner experiences, would seem to sit better with Flynn’s opponents than with his own philosophy.
Even if you believe that every human being harbours an immortal soul, the problem of responsibility remains; I cannot take credit for the fact that I do not have the soul of a psychopath.
Sam Harris, Free Will p. 4
And even a supernatural soul would have to either respond predictably to inputs (and thus be deterministic), or not (and thus be indeterministic). For the traditional concept of free will, Flynn’s concept, to do the explanatory work it was intended to do, it must be the case that there is no correct theory of the origin of human actions. It is not merely mysterian in the sense of “We don’t know” or of “We can never know,” nor even really in the sense of “There is nothing to know”. It is directly contradictory in the sense of “Human action cannot happen either with or without a cause”.

To Do What I Want Any Old Time

The idea of having no control over our actions is scary, and for good reason. Imagine being locked in a cell, or tied to a chair, or surrounded by enemies with no escape route – such scenarios are the stuff of thrillers and horrors. Animals know this fear too, and shun enclosure. Natural selection gave us this instinct fundamentally because the universe doesn’t care about us. We are fragile. There are many more configurations of matter that harm us than benefit us, and nearly all possible pathways our existence could take will bump into them. The only things that do have our interests at heart are ourselves. If we are pinned to one spot by a rockfall the universe will not bring us water. If we are encircled by lions the universe will not turn their attacks aside. The freedom instinct is a survival necessity in a Godless world.
I’m afraid I have to emphasize this, because of one of the most common contexts in which we hear the phrase “free will”. Christians say that the world was once perfect, and as part of that perfection God gave us the gift of free will, which we misused by disobeying him and turning to sin. If you’ve come here looking for support for this doctrine, I can offer you nothing. If the world were ever perfect, if it once contained nothing that could harm us, then the ability to change it away from that perfection would have been a curse, not a gift.
Free will is often explained as follows: If we have it, then we “could have chosen to do something else”, instead of what we actually did do. That’s why the logic of determinism vs. indeterminism seems so damning. If our actions are deterministic, then we couldn’t have done anything else; and if they’re indeterministic, then we could have done something else, but we couldn’t have chosen it. We need to realize that this is not as straightforward as it seems.
A teenage boy, bored and anxious to prove his courage and agility, takes his skateboard up to the roof of an apartment building. Trundling around at low speed, he wobbles and stumbles off the board. Annoyed, he kicks it. It whizzes off to the other side of the roof, and there does not stall in the gutter as he had intended, but, to his horror, rattles over the edge and describes a neat parabolic arc towards the street below.
Now, any statement that some event “could have happened differently” must be contingent on an “if...” If the boy hadn’t kicked the skateboard, it wouldn’t have gone off the roof. But then, if the gutter had been deep enough to catch its wheels and stall it, it wouldn’t have off the roof. Either way, if something had been different from the way it was, the skateboard wouldn’t have fallen. In reality, the skateboard did fall. The statements “The boy did not kick the skateboard” and “The skateboard stalled in the gutter” are both false. I don’t see what it can mean, philosophically, to say that the first statement is a realer, more live, or more present falsehood than the second, simply because the first one describes a “choice” and the second the trajectory of a physical object.
I’ve said elsewhere that determinism is not the same thing as fatalism. In fact, it’s quite close to being the opposite. Granted, once the skateboard is over the edge, you might say it was “fated” to fall, in that nothing realistic can change the fact that it’s going to end up arriving at ground level with great force. But before it went over the edge, any number of things could have prevented that – rubble, weeds growing out of cracks, pipes left by builders, a less well-aimed kick. Some possible paths at that point would have taken the board over the edge, and as it turned out one did; other possible paths would not have. Because the skateboard’s path is determined by the moment-by-moment events unfolding, the end of its trajectory is not set in stone. A different set of moment-by-moment events would have had a different outcome.
There is a difference between a choice and a regular old physical event, but it’s not that the choice “really could have gone the other way” in some sense that the physical event couldn’t have. Unlike other events, a choice is aimed at some future outcome. The universe doesn’t care what happens to us, but we do, and we hope to adjust our circumstances to be better for us. An “intentional” or “deliberate” action – that is, a choice – is one that occurs as a result of a conscious calculation of the agent’s interests.
That, I suggest, is what we really want free will for. Conceptually, it’s more complex than you might expect. You can’t instruct a computer to “follow the best course of action”, because it doesn’t know what the best course of action is – not even if you spell out what outcome counts as “best”, because that still doesn’t tell it which actions will lead towards that outcome. How might an agent build a road-map towards its goals? I can think of at least six ways:
  1. Trial and error— Many people still have to resort to this when plugging a memory-stick into a USB port. Only one of the two possible orientations works, and it’s often hard to remember which one. To program this into a computer, you would need to write instructions on what counts as a “right” result, and what counts as an error. You’d also need to give it some way of figuring out exactly which choice it was that caused the error, and how far it needs to backtrack, and that, too, is more complicated than it sounds.
  2. Past trial and error— I don’t have trouble with my USB stick, because I’ve memorized the fact that the side with the brand name on is the top. Most simple interactions with technology depend on remembering past trials, especially those that depend on turning something one way or the other (taps, car keys, screws...) Given a computer that could do trial-and-error in the first place, adding a memory database wouldn’t be all that complicated, I suppose.
  3. Imagined trial and error— You lock your car in the garage because you can imagine what would happen if you didn’t. Often done with technological aids, such as pen and paper, when it gets too complex for the unaided brain, this is one of the major operations of what we call “reason”. I do not expect to see desktop computers doing this any time soon, but not because it can’t be executed algorithmically. On the contrary, I have myself written a Sudoku game which employs it in a small way. It has to “imagine” the set of possible answers when choosing which squares to reveal as clues because, being a puzzle, it mustn’t show the user what it’s doing. Usually, for a computer, the information space that constitutes its “imagination” is also its “real world” that the user is going to interact with. We who have to deal with the external world, where failure can be costly, eliminate the more disastrous options by running through the problem in the information space in our heads first.
  4. Instruction from a person who knows— either someone who has learned by one of the methods above, or someone who has been instructed by someone else who has learned by one of the methods above, or... lengthen the line indefinitely. This is how I learned to write computer programs, to speak French and te Reo Māori, and to eat with a knife and fork, though of course the other forms of trial and error came into it too. All computers might be said to do this, but of course the question then is making sure the instructor really knows what they claim to know.
  5. Instruction from a group of people whose collective experience amounts to trial and error— It’s said that the ancient Celts revered the oak and maintained sacred groves that were never to be felled. Oaks make an impractical orchard tree – they take decades to grow to maturity – but they also survive harsh weather that might kill crop plants, and acorns are nutritious if somewhat bitter to taste. Did the Celts study the life cycle of the oak tree scientifically over centuries before choosing it as their sacred tree? It’s unlikely. It’s equally unlikely that any one person could have discovered it by trial and error, because growing a new grove for a second trial would take up most of their lifetime. Assuming they really did revere the oak (that story may be nothing but a misinterpretation of the word druid), it’s more plausible that tribes who happened to preserve their oaks for any reason were well-placed, after freak storms, to conquer tribes who had cut theirs down and were left with no food. The trials and errors are distributed among many different subjects, but as long as they have some way to pass the custom on, knowledge accumulated in this way will turn out to be useful more often than chance, or a rational examination of the people’s own theory of why it works, would predict. I suspect that many of our most basic cultural practices have been wrung through this process.
  6. Inheritance of instinct from ancestors whose collective experience amounts to trial and error— This is the longest, slowest way to acquire knowledge, but also the one that requires the least cognitive apparatus. It’s even simpler than trial and error, because you don’t even need a capacity to tell right from wrong; those who get it wrong die without leaving offspring, and do not perpetuate their error to the next generation. The earliest version of my Sudoku program was intended to generate puzzles for print, not for playing in real time, and all it did was to generate eighty-one random numbers from 1 to 9, make a thousand copies with a few changes in each, pick the one with the fewest errors, and then go back to the thousand-copies step unless the number of errors was zero. It zipped along quite quickly until you got down to 16 errors, I think it was, at which point it slowed to a crawl; you ended up with a valid Sudoku grid about twenty minutes later. Natural selection seldom squares away those last few errors, but it’s exacting enough to have produced the cognitive systems the other five methods depend on, and they can cover the remaining ground.
All six operate in humans, and all six can be implemented on purely algorithmic systems. Deterministic or no, we can change our circumstances to suit our interests. I think I can hear some of you objecting that in a deterministic system it doesn’t count as a change because we were always going to do that anyway, so here is some more Dennett for you to read.
But in what sense was that baseball “going to” hit you smack in the face? You dodged it; you were caused to dodge it by the elaborate system evolution has built into you to respond to photons bouncing off incoming missiles on certain trajectories. It was “never really going to” hit you precisely because it caused your avoidance system to go into action...
The very idea of changing an outcome, common though it is, is incoherent – unless it means changing the anticipated outcome... The real outcome, the actual outcome, is whatever happens, and nothing can change that in a determined world – or in an undetermined world!
Daniel C. Dennett, Freedom Evolves p. 59
Sam Harris, who you remember uses much the same definition of “free will” as Flynn, does not think this counts. It’s not enough for actions to be aimed at serving the interests of the agent, or even to be effective at serving the interests of the agent. They must arise from the agent’s consciousness.
The popular conception of free will seems to rest on two assumptions: (1) that each of us could have behaved differently than we did in the past, and (2) that we are the conscious source of most of our thoughts and actions in the present... both of these assumptions are false.
Sam Harris, Free Will p. 6
In fact, our thoughts and actions do not arise in our consciousness. They pop up from who-knows-where; we observe them as they pass through. Our minds are complex systems, and they run themselves without any contribution from “us”.
The problem is not merely that free will makes no sense objectively (i.e. when our thoughts and actions are viewed from a third-person point of view); it makes no sense subjectively either. It is quite possible to notice this through introspection. In fact, I will now perform an experiment in free will for all to see: I will write anything I want for the rest of this book. Whatever I write will, of course, be something I choose to write. No one is compelling me to do this. No one has assigned me a topic or demanded that I use certain words. I can be ungrammatical if I pleased. And if I want to put a rabbit in this sentence, I am free to do so.
But paying attention to my stream of consciousness reveals that this notion of freedom does not reach very deep. Where did this rabbit come from? Why didn’t I put an elephant in that sentence? I do not know. I am free to change “rabbit” to “elephant”, of course. But if I did this, how could I explain it? It is impossible for me to know the cause of either choice. Either is compatible with my being compelled by the laws of nature or buffeted by the winds of chance; but neither looks, or feels, like freedom. Rabbit or elephant? Am I free to decide that “elephant” is the better word when I just do not feel that it is the better word? Am I free to change my mind? Of course not. It can only change me.
Sam Harris, Free Will pp. 64–65
I can’t help (heh) but remember Dennett’s remark, first made in Elbow Room and repeated several times in Freedom Evolves: “If you make yourself really small, you can externalize virtually anything.” I’m also reminded that, rationalistic as he is on morality and free will, Harris is a mysterian about consciousness. The causal chain leading to my actions may not originate, uncaused, from my brain, but it certainly passes through there. Various parts of my cortex critically assess the plans passing through them and make alterations. Not to pre-empt future Imponderables, but it appears that consciousness is a matter of different parts of the brain sharing information with each other. That being so, of course nothing can originate in consciousness, as it must have started in one part of the brain before being shared; but without the sharing, the critical assessment would not be possible, and my actions would be even more ill-judged than, in fact, they are.

You’re Not the Boss of Me Now

The second main thing that scares us, when we consider life without free will, is the idea of our behaviour being predictable or controllable. We want to control ourselves, we don’t want others to control us; and we know that if they could predict our actions perfectly they would be able to control us. But if our decisions come from a web of purely deterministic neuron-firing events in our brains, then surely in principle someone could plot out that web and predict what we’re going to do.
In principle? Perhaps. But that “in principle” is deceptive. Conway’s Game of Life is purely deterministic, governed by a very short list of extremely simple rules – far, far simpler than those of the brain. Now follow the link and see if you can predict its behaviour in anything remotely like real-time. Not easy, is it? Still, I suppose it’s all a matter of computer power. There’s a more fundamental difficulty for anyone setting out to do mind-control.
Two words sum up the obstacle. They are not “free will” or “human spirit” or “scientifically inexplicable” or “spooky, paranormal” or any word pair in that general area of the cloud of English word-associations. They are “recursive self-awareness”. Self-awareness means that a human mind can contemplate its own states, and their likely consequences, and weave them into its plans for future action; recursive means that self-awareness itself is among the things it can become aware of. To which I imagine your reaction is quite likely “What? That’s all?” The consequences are not immediately obvious. Anyone trying to predict exactly what your brain is going to do next, will need to account for your knowledge of their prediction. Having accounted for it, they then need to account for your knowledge of the accounting. Having accounted for that... you get the idea. The moment you figure out what they know about you, that knowledge becomes part of the chain of causes for your actions. Chesterton’s game of Cheat the Prophet can easily be adapted for day-by-day or minute-by-minute gameplay:
The human race, to which so many of my readers belong, has been playing at children’s games from the beginning, and will probably do it till the end, which is a nuisance for the few people who grow up. And one of the games to which it is most attached is called, “Keep tomorrow dark,” and which is also named (by the rustics in Shropshire, I have no doubt) “Cheat the Prophet.” The players listen very carefully and respectfully to all that the clever men have to say about what is to happen in the next generation. The players then wait until all the clever men are dead, and bury them nicely. They then go and do something else. That is all. For a race of simple tastes, however, it is great fun.
G. K. Chesterton, The Napoleon of Notting Hill p. 1
I suppose your would-be-controllers could hide themselves and pretend they didn’t know any such thing, but that’s where self-awareness comes in. Even if you don’t know where your ideas are coming from, you can know what they are (and know that you know what they are, and so on); you can act on your meta-(meta-meta-...)knowledge, thus forcing their predictive software into an infinite loop. To maintain the accuracy of their prediction, they would have to make sure that their knowledge of your future actions never impinged upon you in any form. So if that’s the kind of “free will” you’re after (it’s certainly the kind I care most about) then you may breathe easy – we have it.
As a bonus, recursive self-awareness illuminates another puzzle: what are decisions for? What do they add to our understanding of human action? It doesn’t look like a puzzle, until someone brings it up. Why decide to do things? Why not just do them? What does the “decision” add to the picture? We can certainly do things without deciding to do them – I don’t decide what to do with my foot every time I take a step, unless the surface I’m walking on is somehow hazardous. We decide when the action requires conscious deliberation. But again, why does there have to be a moment of decision in between the deliberation and the act? I suggest it’s to avoid creating chaotic loops with our own recursive self-awareness. As we go through the act, instead of contemplating our own thoughts about our thoughts about our thoughts to figure out what we’re doing next, and wandering in circles and falling on our faces, we remember our decision, stick to it, and get our daily tasks done. If the alternative sounds unlikely, try deciding what you will think for the next two minutes, instead of what you will do. See how long it takes before a random thought pops into your head. Since thought, unlike action, is consequence-free on the timescale of seconds, natural selection didn’t provide us with a decision-commitment subroutine for it.

Won’t Be Held Responsible

As we’ve already seen, our actions are aimed at our interests. That deflates a couple of objections commonly put forth in favour of the traditional concept of free will. “Why bother reasoning with anyone if they don’t have free will?” Because, logically, their actions can be altered by providing information relevant to the calculation of their interests. “What good is it to hold people responsible for what they do?” Because their actions can likewise be altered by adjusting their interests with rewards and punishments (social as well as official). But usefulness is one thing, and rightness is another.
I signed off Imponderable I with a set of unanswered questions, which boiled down to: How can anybody be held responsible for anything? Because my theory of morality is in a bit of a bind if they can’t. You might not blame someone who isn’t responsible, but you wouldn’t trust them either; and trust, you’ll recall, is in my view the whole basis of moral reasoning. Dennett, as usual, has an answer.
Aren’t we headed toward a 100 percent “medicalized” society in which nobody is responsible, and everybody is a victim of one unfortunate feature of their background or another (nature or nurture)? No, we are not, because there are forces – not mysterious metaphysical forces, but readily explainable social and political forces – that oppose this trend, and they are of the same sort, really, as the forces that prevent the driving age from rising to, say, thirty! People want to be held accountable. The benefits that accrue to one who is a citizen in good standing in a free society are so widely and deeply appreciated that there is always a potent presumption in favour of inclusion. Blame is the price we pay for credit, and we pay it gladly under most circumstances. We pay dearly, accepting punishment and public humiliation for a chance to get back in the game after we have been caught out in some transgression.
Daniel C. Dennett, Freedom Evolves p. 292
That’s certainly more sophisticated than either Harris or Flynn. It bugs me, though, that responsibility in this conception is apparently a social fiction. Some of my Marxist friends would say exactly that, and it’s true enough that people publicly banging on about responsibility are generally trying to offload some. But social trust is what I’m concerned about, and it’s not served by punishing the innocent to “make an example” of them. I suppose a good deal of the problem is removed if it’s the innocent “making an example” of themselves, but that isn’t, in practice, the case. People don’t actually hand themselves in after committing crimes and demand full punishment as befits a responsible agent in a free society. We behave as if that was what they were supposed to do, and in that gap between personal responsibility and social expectation lies a great deal of potential for abuse.
Let’s tackle the question from the easy end. When is somebody not “responsible” for something happening? Am I responsible for Pope Benedict resigning, or Hugo Chávez dying? Of course not, I didn’t cause either one. For Entity A to be “responsible” for Event X, it must be the case that Entity A caused Event X.
I breathe air, returning it to the atmosphere with a little more carbon dioxide than it had before I got hold of it. I cause the carbon dioxide; am I responsible for it? No, I’m not, because I cannot choose to stop breathing, or not for more than a minute or two at a time. For Entity A to be responsible for Event X, it must be the case that Entity A could have chosen not to cause Event X. And so we return to where we started. What does it mean to say that one could have chosen otherwise?
(Well, hypothetically I could “choose” not to breathe after all, as long as I was comfortable with dying. If I’m not responsible for the consequences of my breathing, then it must be because it’s somehow not reasonable to expect me to choose to kill myself. Exactly what constitutes “reasonable” is an interesting question, but one that can wait for another time.)
Remember the teenager’s skateboard, hurtling off the roof. If it hits someone and seriously injures them, we would hold him responsible, but less so than if he had lain in wait and intentionally flung it at them. What’s the difference? Recall, a deliberate action is one that follows a conscious calculation of the agent’s interests. A careless action might also be based on an assessment of the agent’s interests, but we would judge that the assessment process had been prematurely curtailed – in this case, the teenager did not consider the skateboard’s potential for falling off the roof before kicking it. We hold him responsible for his thoughtlessness because we judge that that came from a calculation of another of his interests, namely his self-image as a powerful, decisive person. The connection with responsibility should be obvious: logically, the actions most appropriately altered by adjusting the agent’s interests are those performed under conscious contemplation of said interests.
All the easy parts aside, we must now face the “could have done otherwise” problem head-on. There is something, after all, in Pinker’s separation of the science and ethics games. Responsibility is not part of the science game. We do not expect to find it in the laboratory at the cellular and brain-tissue scales studied by neuroscientists. It is a property that pertains to real-time social interaction between human beings, where (besides the fact that we don’t have electrodes and fMRI scanners to play with) the other players are responding to us just as fast as we are responding to them. If we were to plot out the trajectories of every neurotransmitter molecule, and every calcium and sodium ion, in one another’s brains, after the fact and with no interference from the person concerned, we might well find there was only one outcome that didn’t break a law of physics somewhere. But we don’t have that kind of access to anyone’s neurons, which puts that information off-limits in the ethics game. Even if we did have an iPhone app (or such-like) for following neural activity in real-time, we are still impinging mutually on one another’s lives, and that means our future choices are screened behind recursive self-awareness. The proper truth-test is not “Could you have chosen something else, given the same initial conditions down to the position of the last atom?” It is “Could you have chosen something else, having calculated your interests differently, given the same relevant conditions as best we can discern them in real-time?”
Harris argues that responsibility-based ideas about justice are dangerous at best:
False beliefs about human freedom skew our moral intuitions and anchor our system of criminal justice to a primitive ethic of retribution.
I don’t think he’s entirely wrong. The human instinct for revenge is grossly overpowered for its evolutionary purpose once a society has entrusted its deterrent functions to a neutral arbiter, and moral leaders through the centuries have been right to urge people to damp it down. But, as I argued in Imponderable I, I believe that we can best serve the cause of mutual co-operation by responding proportionately to breaches of trust, after carefully determining which (if any) party to the action was intentionally advancing its own interests at the expense of others.
However, this theory of will does not suit all theoretical moral purposes. As I said before, if you have come here looking for support for the Christian concept of free will, I have nothing to offer you. Recursive self-awareness would not screen our brains from an all-knowing God. From God’s point of view, we do not have free will; we do precisely what he always knew and (as our creator) intended we should do. This model of free will does not, therefore, provide a loophole out of the problem that a supposedly good, supposedly all-powerful God supposedly created a world with evil in it.
Some Christians explain the invisibility of God with reference to free will as well: he wanted us to love him “freely”, they say, but he’s so awesome that if we could see him we’d be compelled to do it. The same Christians will also say that the consequence of not loving God is an eternity in hell. But since free will is a matter of being able to act on one’s interests, withholding information critical to assessing those interests does not preserve it but derails it. And do you remember the scene in The Lord of the Rings when Gandalf was imprisoned on the roof of Orthanc? Or was he imprisoned? He was “free” to jump off at any time, after all. The dreadful consequences of doing the wrong thing (absent a friendly eagle) made this “freedom” a sick joke. Same with hell, I would say.

Who Am I?

Do we have free will? Not in the traditional sense, the Christian sense, the sense Flynn describes. But it seems perverse to insist on that, when we do have so many of the things we think we want free will for. We can pursue our own interests and avoid harm; our choices do have consequences; choosing something else would have different consequences – unless we insisted on God-like precision in specifying exactly what conditions satisfied the hypothetical “Starting from the same beginning situation...” That being so, we can formulate a meaningful truth-test for responsibility (which, yes, would crumble given God-like knowledge, but we have not been given God-like knowledge). Moral reasoning and moral sanctions can dissuade other human beings from harming us or others. But I have saved Harris’s subtlest counter-argument for last.
...am I responsible for the fact that it has never once occurred to me that I might like to be a surgeon? Who gets the blame for my lack of inspiration? And what if the desire to become a surgeon suddenly arises tomorrow and becomes so intense that I jettison my other professional goals and enroll in medical school? Would I – that is, the part of me that is actually experiencing my life – be the true cause of these developments?
Notice how he has to provide a definition for the word “I”. That is the crux here. Who is “I”? Several of my arguments in this essay have assumed (like Flynn with his “present self”) that the meaning of “I” is common knowledge and uncontroversial, but this is not the case. For instance, I’ve said “We can pursue our own interests and avoid harm”. But what are “our own” interests?
There is an organism called Toxoplasma gondii which lives parasitically in the brains of mice. It does not harm them directly, but it has a dramatic and lethal effect on their behaviour: an infected mouse, faced with a cat, will run towards it instead of fleeing. This is because the sexual phase of Toxoplasma’s life-cycle can only take place in the digestive tract of a cat. Toxoplasma infects humans as well. It doesn’t drive us to suicide, because we are seldom preyed on by cats; but it’s easy to imagine a similar organism inspiring, say, an irresistible desire to swim with great white sharks. Now, obviously if mice could speak, they would say that Toxoplasma caused them to act contrary to “their own” interests. But an infected mouse might retort that its interests had changed – that “cat-fearing me” (its past self) was some other mouse, no longer “the real me”, and that if it felt that blissful proximity to a cat was worth paying for with its life, that would be its own business. Don’t laugh too hard; some people with life-threatening behavioural disorders (such as anorexia or tobacco addiction) make exactly equivalent statements.
What’s my answer? Who is “I”? What is the self? Once again, I’m going to have to defer the question. The Self will be the subject of Imponderable III.

No comments:

Post a Comment