You Can't Argue with a Zombie

by Jaron Lanier
(cached and reformatted by David Chess from this original)

It is impossible to awaken someone who is pretending to be asleep.
- Navajo proverb [1]

The Zombies Among Us

It turns out that it is possible to distinguish a zombie [2] from a person. A zombie has a different philosophy. That is the only difference. Therefore, zombies can only be detected if they happen to be philosophers. Dennett [3] is obviously a zombie.

Zombies and the rest of us do not have a symmetrical relationship. Unfortunately, it is only possible for non-zombies to observe the tell-tale sign of zombie-hood. To zombies, everyone looks the same.

Arguing with zombies is generally futile, of course. But I have found it to be a valuable experience for two reasons. First, I believe that zombies are having a significant indirect influence on cultural and political thought, and I wish to thwart them. And then, there's a lot to be learned from zombies; they are useful, at the very least, as conversation pieces.

Part One

Entering Zombie Culture: Playing the Intuition Pump Game[4]

Zombies are concerned for some of us. They believe we are zombies too, but that we are confused, so they produce literary devices to help us think like zombies. For example, Dennett proposes a thing called a zagnet, which is like a magnet in every way except for some ineffable difference. He then suggests that consciousness (the old kind[5]) in people is a superfluous hypothetical quality just like zagnet-hood, and that philosophy gets so much easier without it. It doesn't occur to him that a zagnet like me might write its own essay, but, then again, he's a zombie. He can only think about zagnets from the outside.

The zombie/zagnet dialog has surpassed any other in my experience as a source of passionate lack of communication. Only a zombie like Dennett could write a book called "Consciousness Explained" that doesn't address consciousness at all. Only zombies could respond with a book called "Dennett and His Critics" which doesn't criticize Dennett at all. What other topic could drive a fine physicist into the strange speculations of Roger Penrose? I feel as though I'm able to occasionally communicate across the zombie/zagnet gap, because I'm in the unusual position of being a zagnet and a computer scientist, a sadly rare combination. It's easy to see why computer scientists are so ready to embrace zombiehood. Wouldn't you love it if a major school of philosophy claimed that what you did was uniquely at the root and the apex of everything?

My Arguments with Zombies

I faced a class comprised mostly of zombies at Dartmouth a few years ago. Here is what I told them. I decided to push a tired old intuition pump just a little harder than usual to see if a few more drops of intuition would squirt out. Some of the zombies stirred as a result and I wondered if there might be a cure after all for zombie-hood.

I started with the usual sort of brain-replacement yarn. Your neurons are replaced one-by-one with silicon devices. That sort of thing. Young zombies-in-training assume that nothing fundamental will have changed if they are turned to silicon.

We then transferred our brains into software. Each neuron was now replaced by a software expression and they all connected together functionally in the same way as they did when they were mushy.

The zombies still felt at one with this proposed zombie-on-a-disk. It is worth pausing for a moment and noting that accepting one's ontological equivalence to some data on a disk does not necessarily banish the demons of vitalism. Zombies might still imagine their data interacting with biological humans (as we see in the Star Trek character "Data"). They might still turn to the natural world for confirmation, relying on that old ritual of vitalism, the Turing Test.

Harder core zombies are ready to leave all that behind and imagine living on a disk in which they only interact with other minds and environmental elements that also exist solely as software. It is here that we must ask a question that seems obvious to me, but seems to shock zombies: What makes this software exist? What makes the computer that it runs on exist?

There can be only one proper basis to judge the existence of computers and software. We should be able to confirm their existence empirically, using the same scientific method we use to study the rest of the natural world. As it turns out, we cannot do that, for reasons that I will make clear later in this paper. We are the only measure of the existence of computers. So the assertion that computers and software exist is a stealthy conveyor of rampant vitalism and mystical dualism.

How Far Can You Stretch a Zombie?

Back to my class of eager young zombies:

I asked them if it mattered to them what kind of computer their software selves would run on. No, they replied, it doesn't matter. All computers are considered to be equivalent by virtue of the Church-Turing Hypothesis[6]. If they and their classmates were implemented on a vacuum tube computer, or on a computer made of mechanically-linked Lego blocks, they would still feel the occasional rush of adrenaline as a desired mate strolled by, and the agony of a parental visit.

Zombies are obviously quite flexible. So let's build a truly unusual sort of computer for them to inhabit. We first record a bunch of data from the natural world, such as the trajectories in a meteor shower. We then rely on massive quantities of computation and luck to construct just the right computer that happens to read the meteor shower as a program, such that it happens to be equivalent to a zombie's brain.

In other words: When a natural phenomenon, like a meteor shower, is measured, it turns into a string of numbers. The program that runs a computer (the object code[7]) is also a string of numbers, so we have two similar items. The string of numbers that runs a particular computer has to perfectly follow the rules of that computer or the computer will crash. But if you can find the matching computer, any particular string of numbers can run as a program[8]. In fact, for any string of numbers, you can in theory find or construct many computers, each of which will run the same string of numbers as a different program. So one computer might read the meteor shower and end up doing your taxes as a result, while another might calculate racetrack odds from exactly the same "object code". If your brain is functionally equivalent to a computer program, there is no reason a meteor shower can't be that program, if you take the trouble to find the right computer to run it.

Does even the possibility[9] of this computer give the meteor shower consciousness, if only for a moment[10]?

Of course a zombie would probably object that this hypothetical computer isn't functional at all. Even if it interprets the meteor shower as having the functionality of a brain, that could only be true for a limited period of time. Certainly after a very short while Newton and Einstein would take over again and the brain would dissipate. Furthermore, there could be no functional relationship with anything outside of the meteor shower. Is the meteor shower not even a zombie?

If my meteor shower doesn't seem functional enough for you, widen your search. Just gather more data, and find an even harder-to-design computer that will interpret it as a mega-program that is equivalent to not only your brain, but also to your brain's surroundings, including other brains. Widen further to gain enough object code to implement a lifetime.

Zombies will sometimes remain unmoved by this idea because it might seem as though we still have to do the work of building a computer in order for the meteor shower to be considered as a program. Maybe something magical does happen when a program is read by a real computer. Well, if you're patient enough, you can even find a real computer hiding in a meteor shower.

Now we're going to conduct a slightly different kind of search; an even harder search than the previous ones. This time we are looking for a computer (which we'll call Ralph) such that, when you point it at our overworked meteor shower, it implements an emulation of itself (which we'll call Fred), as well all the brains and their environment.

Once we're done there'll be two computers looking at the same meteor shower; the one we've actually built (Ralph) and the emulation[11] of it (Fred). From an empirical point of view they are equally prime suspects for being the "real", functional computer, and they will be observable for the same period of time.

Time is only one aspect of the subjectivity of computers, but let's look at it a little more closely. A Macintosh functions more-or-less predictably during its useful life. Ralph and Fred last for less time, perhaps. But the Macintosh will seem to be a more legitimate computer to us in part because it matches our time frame better. We can borrow some fast moving trains from Einstein to illustrate this idea with another image. Suppose you visit a train yard where trains are moving at wildly different speeds. Some move so quickly that they blur into continuity. If you start moving very quickly to keep up with some of the fast trains, the stationary ones fall away into a blur. Only those trains moving at about the same speed as you look like trains. In the same way, a Macintosh user interface responds slowly enough for us to catch the images, and the machine generally keeps running for a few years, and this makes it sensible within a human time frame. Ralph and Fred might be coherent for only a fraction of a second, and thus seem less like computers to us, but if we could "move fast enough", the Macintosh would disappear and Ralph and Fred would come into focus.

Zombies sometimes object that a computer that someone built is different from one that is simply detected. I suppose they might believe in Ralph more than Fred. It's a good sign when zombies think this; they are on their way to being cured. This is a fine example of vitalism, and it warms my heart.

Let's suppose you actually did physically build Ralph, in order to bring into focus the brains living in the meteor shower, as well as a Ralph clone (Fred) looking at those brains. The meteor shower isn't affected if you turn off Ralph[12], so Ralph is a part of the system that is not functional. Since a program can't tell if it's being read, it should exist equally well even if the computer is only hypothetical, right?

I know I'm pushing the intuition pump very hard here, so let's let's leave Ralph and Fred behind and approach this last point from another angle.

Let's suppose you run a more normal program (not a meteor shower) that implements the functional equivalent of your brain, a bunch of other people's brains, and the surrounding environment, so that you and the rest of the brains can have lots of experiences together. (This is the condition in which my test zombies thought that nothing fundamental would have changed; they'd still experience themselves and each other as if they were flesh.) You save a digital record, on the same disk that holds the program, of everything that happens to all of you. Now the experiences "pre-exist" on the disk. Take the disk out of the computer. Is this free-floating disk version of you still having experiences? After all, the information is all there. Why is this information sanctified into some higher state of being by having a processor just look at it? After all, the experiences have already been recorded, so the processor can do no new computation. A much simpler process that just copied the disk would perform exactly the same function[13] as running your brain a second time.

The meteor shower arguments hint that computers might not be as objectively present as zombies like to assume, but the question needs further examination.

I claim that to a Martian, a Macintosh is the same sort of thing as a toaster or a rock. In order to perceive information, you have to put it in a cultural context, and that re-opens the can of worms that zombies have been trying to solder shut. Could "information" just be a shell game that hides the nut of old-style consciousness?

Can We Build a Computer Detection Machine?

Now, a zombie might object that there must be some method for the objective analysis of a proposed computer-object that would confirm its computer-hood. If you don't want to rely on a mushy brained human's intuition, you'll have no choice but to bring a confirmed computer on stage to analyze your proposed computer. (If you do rely on a mushy human, you've retreated to vitalism. ) Even if you somehow came up with a confirmed computer, you would have to have a rigorously objective answer to the question: What would make an object a non-computer?

If we want a computer detection test to be rigorous, we should assume that we are attempting to discover a truly alien computer, or that an alien is trying discover our computers. Our first problem is that if we were presented with an alien computer we might not be able to figure out where to stick our probes, but even worse, we might not have a clue that we should try.

Any purported computer we study can be fully understood as a non-computational complex phenomenon. Computer science is unnecessary to explain the behavior of computers. Computers are simply pieces of the physical universe obeying physical laws. Everything a specific, physical computer can be observed to do[14] can be understood without having to think of it as a computer. What makes a computer a computer is our way of thinking about its potential, not its observed actuality.

This is not a trivial point. Computers have been given the ontological kid glove treatment. Relativity is necessary to explain the observed universe, while computer science is not.[15]

If we designed a test that could detect an alien computer, then that test could also find computers and their programs wherever we chose to look[16] (even in a meteor shower), so long as we looked hard enough. This is not what you call a useful detector.

If computers are to definitely exist we should know that we could someday build an instrument to find them. Scientific instruments can lack accuracy, but they must be able to distinguish between phenomena. If there was no conceivable device that could distinguish heat from other phenomena like gravity, for example, heat would not be a useful concept, and science would pursue a parameter that could be measured. No one has thus far been able to define a notion like "complexity" sufficiently well that we could someday hope to build a complexity detector. Even if we got to that point, a computer detector would be unable (for the reasons stated above) to distinguish proposed computers from other examples of "ordered complexity" or "intricately bundled causality"[17].

What is amazing to me is that even zombies on the whole aren't quite willing to drop the idea of something special called "consciousness". It's just that they want zombie-think to own it. Such zombies start using the word "emergent". They say that computers differ from one another in that some of them have this consciousness thing "emerge". David Chalmers argues that all action in the universe is at least a little computational, and the right computation give rise to consciousness, so consciousness is everywhere, but in varying degrees. I like the hippie-ish egalitarianism of it; that even a thermometer gets to be a "little" conscious.

But where are the computers in Chalmers' universe? A computer-detector would not find a single computer in a test location because it would always find an infinite[18] number of them. As soon as we cordoned off a piece of this universe and called it a particular computer, we'd also have identified the infinite superset of universe pieces that contained it, as well as the possible subsets that were functionally equivalent (by some measure or other). The conclusion has to be that the choice of where to cordon is arbitrary.

Furthermore, and most importantly, if we are interested in computers that are similar but not identical to the one first cordoned, then we can always identify an even huger swarm of computers that are co-present. If computation gives rise to consciousness according to Chalmers' scheme, then each emerged consciousness would in fact carry with it an infinite swarm of consciousnesses that contained it, or almost contained it. All sense of discreteness and locality for consciousness is lost if it is only a byproduct of computation and computation is ubiquitous. Hypothesizing an infinite cloud of slightly different consciousnesses floating around each person seems like an ultimately severe violation of Occam's razor.

Other zombies might argue that computers are mathematical objects, so they don't need to be confirmed empirically. I would argue that any particular computer does need to be confirmed. Otherwise we are in a state in which all possible computers always exist, even more of them than would exist in Chalmer's universe.

There are other possible arguments[19], perhaps, but I think ultimately an honest zombie will have to accept that any particular observation of computation is a non-confirmable human interpretation of events that can be more objectively described in other ways.

When Zombies Get Weird

The sample argument above will not be unfamiliar to hard core zombies and their antagonists. What interests me most is the ultimate position that zombies arrive at when this argument is driven to its conclusion. After abolishing ontological distinctions based on human epistemological difficulties, zombies invent new ontologies for the benefit of computers. Inside every zombie is a weird new kind of dualist.

The new weird dualism can take a number of forms, distinguished by the choice of meaningless code words, such as "emergent" or "semantics". But the hallmark of zombie dualism is the belief in the independent, objective existence of information and computers.

I am certainly not trying to convince zombies that they exist in some special way, that they might have a sense of experience. By now I know better. What I would like them to consider, rather, is that they are granting to the process of computation not only a type of indisputable objective existence that it probably doesn't have, but also a magical ability to confer ontological properties onto yet other objects. I'd like zombies to consider that this purported ability is even more bizarre and insupportable than the phenomenological reportage of non-zombie experience.

Zombies believe in something called information, and also in the existence of objects called computers. Zombies are so quick to criticize the notion of old-style consciousness as being the worst sort of murky, sentimental dualism, but they themselves are zagnetizing the universe with these new ineffable concepts.[20]

If the universe were populated solely by zombies, there would be no computers. Computers cannot make each other exist, because they cannot even recognize each other.

There also would be no information. Information is another thing that only exists by virtue of experience. (My old catch-phrase: Information is Alienated Experience.) Zombies owe us zagnets a great debt for making their information exist.

Part Two:

The Zombie Effect

If I'm Not a Zombie, What Kind of Science Do I Like?

Zombies probably think that I am a mystical dualist of some stripe. I can accept that[21], but I don't act like a mystical dualist. I am enthused by progress in neuroscience. I don't foresee any brick wall that would prevent a scientific examination of the minutiae of thought and behavior. In fact, I'm thrilled to think about brains. I must appear to be a monstrous anti-zombie to the zombies; someone who claims to have ineffable subjective experience and yet acts just like them.

Let's imagine a society in the future in which neuroscience has gotten as good as, say, quantum electrodynamics is today, that is to say essentially complete within its framework. Would every educated person be a zombie? Would the consciousness debate still exist? Would it have any practical consequences?

This is an entertaining future to imagine. Suppose you could buy a brain self-examination unit where you could stimulate any firing pattern in any set of your own neurons at will. My working assumption is that I would experience a complete correlation between the objective state of my brain and the content of my experience. (I think this machine would be fun, when used gingerly. I want one![22]) And of course that means that inside every zagnet's brain would be seen some little gizmo comprising the thoughts of self-experience.

Even in this hard zombie future, any person who experiences him/herself as a non-zombie would find that the philosophical issues remain unchanged. The study of the brain might yield a perfect theory of what is experienced, but not of experience itself. To a zombie, of course, the philosophical problem didn't exist before and still will not.

So, if the consciousness problem has little consequence and will not yield to further physical study, why do zagnets like me care about it? I might ask the same question of some of the zombies.

Zombies delight in annoying us zagnets and seem to have an irrational need to wield rationality as a prank. Titles like "The Astonishing Hypothesis" and "Consciousness Explained" do not reflect the falsifiable aspects of their authors' work, but rather are intended to assert an authority over the reader's own intuition. Arrogance is always a bad strategy in science. In philosophy I suppose it's fine[23], but here we are dealing with a combination of the two and we should work out some practical etiquette.

Teaching Zombies Good Behavior

Zombies and zagnets often do not seem to speak the same language. They stumble over each other like an anthropologist's case study in cultural misunderstanding. Most zagnets in the arts and humanities cannot even imagine the hard zombie positions taken by writers like Dennett. Zombies often cannot even find a stated position in zagnet rhetoric. Marvin Minsky, a zombie I'm very fond of, once said to me that he didn't like non-science fiction literature, because nothing seemed to happen. One gets the same sense reading Dennett's appraisals of thinkers like Searle and Nagel. He doesn't just disagree, he doesn't even agree that anything's been said.

I can think of two consequences of the consciousness debate that matter to me currently. One is that it would be pleasant for non-zombies to have a philosophy that does not require that we ignore our own experience of existence. The other is that zombies have come up with a batch of metaphors that are radiating out in the world at large and are having an effect on politics and culture.

An overly dry metaphysics inevitably trickles down to a narrow reductionism in many practical instances, even though in theory it need not. An example is found in the design philosophy of computer systems. Convinced by zombies of the ontological equivalence of people and computers, a generation of software designers is asking users to shrink to the level of so-called "intelligent agents"24. For another example, we have the narrow application of Darwin, as he's been zombified by Dawkins and Dennett, to human affairs in Robert Wright's The Moral Animal, and even in a degenerate work like The Bell Curve. Then there is the strange abrogation of human agency in favor of algorithms that is found in some current political rhetoric. Newt Gingrich believes that it is counterproductive to try to do anything about problems, such as fund basic research or feed poor children, because only the algorithm of economic/technological evolution can achieve anything. The one exception he is willing to propose is to give computers to the poor.

If indeed we could be certain that the zombies were right, we could endure the bad metaphors that they are radiating, but as long as there is one zagnet around (count me in!) there is uncertainty. Should philosophers be held partly responsible for the damaged reflections of their ideas that are hammered into an approximate fit with the practical world by people who have to make practical choices? Absolutely. If not, then what kind of responsibility for anything can philosophers ever accept?

Philosophy will take on an even more heightened importance if an era of successful neuroscience comes into being. Definitions of personhood will be required at every turn to guide the design of our medicine, our information systems, and our laws. It is critical that this philosophy acknowledge the whole of the human experience, and not just that of the zombies.

It is true that if zombies have a weakness for excessive reductionism, zagnets are prone to succumb to superstitions which can be equally pernicious. This is what zombies seem to fear most, that zagnets will insist on believing in fantastic creatures like homunculi[25]. Some of us zagnets probably will, but not all of us. A zagnet can accept the state of the brain as being the same thing as the content of the mind and still wonder in awe about the nature of experience itself.

There is a tightrope to be walked, between reductionism and superstition. I propose that if we can keep on that tightrope in our philosophy, the metaphors we radiate will be more useful and beautiful.

Part Three:

What Might Philosophy Be Like After the Aliens Arrive?

It is easier to criticize than to construct a new opinion worthy of defense. But since we do seem to be epistemologically impoverished creatures prone to illusion, criticism is important. Nonetheless, one of the rewards of my persistent arguments with zombies has been a thumbnail sketch of a way to think about consciousness that might perhaps be satisfying to both zombies and zagnets.

I'll state it here. It's a sort of a cross between Plato and a radio dial.

Despite my demonstrated skepticism about the existence of computers, I am clearly able to recognize and use them. I am even using one to write this paper. Something curious is going on with computers, and I think it is exactly as curious as the existence of language.

Here's my thought: Consciousness is the choice of which abstractions we experience, out of an infinite number of ways of slicing the continuity of the universe. It's the feeling of existence that is the choice[26].

A zombie might object that it is your neurons and what they're up to that make this choice. My counter-argument has already been stated: Brains don't exist on their own any more than computers do. It is layers of abstraction, known sometimes as concepts, platonic forms, cultural context, or words, which make a brain, or the thought-processes in it, exist.

So consciousness is like a radio with a dial that might be marked "qualia" or "semantics", that selects from an infinity of equally available "layers of abstraction". Without the cosmic qualia dial, a brain, or a thought, is just another utterly arbitrary slice of the continuous causality that is the universe.

Levels of abstraction, like language and computation, are only singled out for existence because we experience them. Or rather, out of an infinity of equally valid layers of abstraction that potentially co-exist, experience illuminates a specific layer. Our brains and all their activity can be fully understood using only the ideas of physics (unless something really weird and mystical is going on). We don't need to be able to detect words or computers to explain every little thing that people and machines do. One brain, or many brains, can be understood as just another vector field[27].

Importantly, this does not mean that any possible layer of abstraction can be applied to any piece of the universe. If that were the case, then the universe itself wouldn't have a function in this philosophy. The universe in this theory is one that does have some particularity; an affinity for a particular infinity of possible slicings. Without consciousness it wouldn't be sliced, however; it would just be a continuity.

Even the neuronal activity that corresponds to our experience of abstractions like "abstraction" and "computation" can be completely understood without reference to those concepts28. We don't need chemisty, biology, or psychology to "understand" a thought scientifically and empirically, but we do need them to recognize a thought.

Nature doesn't have nouns, and indeed the more nouns we make use of in our science, the less complete and accurate our science becomes. Current physical theories like QED that conceive of the world as a near-continuity are the ones that are most successfully verified by experiment. They also have the greatest universality. Every time a noun is added, such as "atom", "molecule", "cell", or "organism", science becomes more convenient for us, but less generally applicable to the universe (as well as generally less accurate when tested empirically). The reason for this is that nouns, while necessary for science, are also arbitrary.

Words and Thoughts: Should Zagnets Discriminate?

Can I believe that thoughts are no more than patterns of activity in a brain and still be a Zagnet? The "qualia dial" formulation presented here is optimistic about the progress of neuroscience and doesn't assume that any veil will indefinitely continue to shroud the mechanisms of thought. But it is that very acceptance of the ontological equivalence of thoughts and ordinary objects that forces us to ask why we experience these particular thoughts and things and not some other slicing of the universe.

I once read a satirical piece in which aliens came to Earth and decided that cars were intelligent life forms that made use of disposable guidance systems called "people". The fable would seem to be, at this point, obsolete. It joins the "Chinese Room" and many other zagnet thought experiments in its anachronistic treatment of the human head as a black box. Modern zagnets will need a new collection of thought experiments that assume advanced neuroscience complete with instruments that render the inner workings of the brain both visible and comprehensible. If we push this framework far enough, it turns around to favor zagnets.

Let's join some alien scientists with superb measurement equipment that makes the human brain appear to be transparent, with all neuronal states exposed. Furthermore, our aliens have ultra-fast computers tied in to this equipment that perform generalized pattern correlation algorithms on everything that is measured.

Can these well equipped imaginary scientists detect human language? We think language is the most obvious interpretation of what our brains do together because it's the interpretation we experience. My claim is that if truly unbiased aliens were observing my neuronal patterns as I write about language, they would lack the reference point of my experience to choose from the infinity of interpretations available. Even watching my neurons as I write these sentences wouldn't give them a clue about which of the infinity of available layers of abstraction is the one I am experiencing. So, for instance, they wouldn't necessarily slice the world up into words and objects the way we do. They might think vertically and sort all phenomena they came across in strange alien/Platonic terms, according to connectedness and shape, for instance. In this case they might fail to distinguish what we'd call a real chair from a neuronal thought of a chair.

A thought of a chair and a real chair are both no more than temporary perturbations in the vector field. To an alien they might have more in common than most perturbations. Each might only be noticed (by an alien) in conjunction with the other, and they reflect each other in a way that might place them in the same Platonic class, if aliens happen to think that way. They are distributed in different ways in space and are stable for different amounts of time, but my aliens are pretty weird and these differences seem trivial to them. It's a toss up which they would "single out" first; an otherwise unconnected set of transient physical objects that we would call "chairs", or the distributed correlation across many transient brain-objects that constitutes the word "chair". And they might never detect either because neither are special.

In general, whenever discreteness enters a definition of a thing, objectivity is lost. The discreetly existing version of what a chair is, the everyday definition "chair", does not have objectivity. Neither do neurons.

The Zagnet Pitch

I've been framing this idea as I would in introducing it to zombies, because I expect them to be the toughest customers[29]. I'd introduce the concept to a zagnet in this way: Phenomenological experience has to have some specificity. This usually leads zagnets into trouble because they end up imagining a dualistic alternate plane of existence where the specificity is held. The "qualia dial" avoids that problem because it adds no content at all. All the particulars, including all the details about how a qualia choice "feels" are held in the physical world, particularly in neuron states. It is the dial, however, that has singled out a particular slicing of the universe, in which we experience the neurons in our particular brains, instead of their momentary coherence with a plate of pasta, for instance.

The qualia dial validates zagnets while still letting the universe exist independently. Zagnets frequently end up having to deny the existence of the objective universe in order to exist themselves. Sometimes, to get around this problem, zagnets propose that consciousness is a part of the natural world, just not the part that zombies are competent at observing. Taking this approach, zagnets can run but they can't hide. Eventually, some grandson of Dennett might be insulting Penrose-style zagnets with quantum measurement devices and Searle-style zagnets with group-mind detectors even though today we believe such devices to be impossible.

The qualia dial gives both subjectivity and objectivity their due.

Qualia Dial -- Conclusion

In brief: Abstractions are superfluous to the world but they are experienced by zagnets anyway (I can confirm this). Experience is the mooring of the floating, ephemeral world of words, brains, and computers.

I have found this philosophical framework to be helpful because it not only acknowledges experience but gives it a function. At the same time it doesn't propose that anything ineffable should prevent scientific inquiry about the brain and mind. And, it provides a constant motivation to consider alternate frameworks of thought, alternate layers of abstraction. It also avoids a Cartesian trap door with homunculi peering out, because it conceives of experience as an epistemological necessity that is mysterious, certainly, but that does not require any additional mechanism inside it. It's a clean kind of dualism that doesn't mystify the empirical world, ever. It doesn't grant excessive primacy to language. And it conforms to the experience of being alive.


Footnotes

[1] At least I'm told it's a Navajo proverb by John Perry Barlow.

[2] Zombies have no internal experience. They are unconscious, but give no obvious externally measurable evidence of that fact. Zombies have played a distinguished role as fodder in the rhetoric around the mind/body problem and consciousness research. There has been much debate about whether a true zombie could exist, or if internal subjective experience inevitably colors either outward behavior or events in the brain in some way.

[3] Daniel Dennett, author of Consciousness Explained and many other zombie manifestos.

[4] Dennett calls his thought experiments "intuition pumps".

[5] The meaning of the term consciousness has been subjected to a tug-of-war lately. It used to mean "subjective, ineffable experience", and now it might mean "a part of a program that models other parts and can exercise executive control". I like to use the word "experience" to refer to the subjective experience of experience, which is the thing that makes consciousness into a hard problem.

[6] The term "computer" can mean a number of things. For the purpose of this paper, "computer" will mean a practical object that can exist, such as a Macintosh. The Church-Turing Hypothesis suggests that there is a ceiling of capability above which no computer made of ordinary materials can rise. This ceiling is defined by an "ideal computer", called a Turing Machine, which is like an ordinary computer, but with infinite memory. The hypothesis is treated, in general, as a truth by computer scientists. Penrose and others are interested in quantum computation because it might result in a computer that does more than a Turing Machine can. The Church-Turing Hypothesis "trickles down" in most debates about consciousness to a similar assertion that as finite computers become very large, they should be treated increasingly as being functionally equivalent to each other. This is why the brain is seen by many as a large computer. If either the pure or trickled hypotheses turn out to be false, nothing in these arguments really changes; the bar is simply raised to a new level corresponding to the new ideal computer.

[7] "Object code" is the kind of program that a computer operates from, as opposed to source code, which is written by people. Source code has to be converted into object code before a computer can do anything. Each different kind of computer uses a different, incompatible type of object code. DNA is understood by some biologists as a type of object code. A computer without any object code to run is inert, as is a specimen of code for which the proper computer cannot be found. I am here suggesting that any arbitrary piece of nature might turn out to be object code for some possible computer.

[8] Two different pieces of object code can have exactly the same effect, and are said to be two different implementations of the same program. From a functionalist viewpoint, the two are identical. For example, if two versions of a program behave identically on Macintosh and Windows machines, there are then two different pieces of object code that are functionally equivalent. From a functionalist standpoint, there could be many different pieces of object code, running on the same or different computers, that could be equivalent to your brain. I am here suggesting a particular one, which happens to be a meteor shower.

[9] You could trivially construct a computer to treat any sufficiently large data set as object code to emulate any program simply by including a big "lookup table". A lookup table would simply map whatever data you found into the data you want (which in this case would be object code to run your brain). This feels like cheating. It seems as though it isn't magical enough. To get magical, we'd need to construct a computer that doesn't have any information built in to it that reflects advanced knowledge of your brain. Fine. Then you have to search through the space of possible computers (defined here as finite state machines) until you find the first one that works properly. Finding such a computer is similar to cracking a very, very large cryptographic code. It might not be practical, but it is theoretically possible. If zombies wish to dispute this, they'll have to join ranks with Penrose and seek a fabulous quantum element in the brain that would cause it to evade such a search.

[10] There could be many different computers that each interpret the same meteor shower data as a different brain, thus giving this arbitrary bit of nature a very rich inner life.

[11] An emulation is a program running on a computer that simulates the existence of another computer. For instance, Macintosh computers can run emulators that seem to be IBM-type machines "living inside" the physical Macintosh. A Macintosh could also certainly run a program that simulates a copy of itself. There is a potential problem of an infinite regress, of course, but that can be easily avoided in my example by having the emulation be incomplete; it will not include a further interior emulation of itself.

[12] It might be "effected" if you turned off your measuring instruments, but it will not be perturbed by the status of your computer that runs the data as a program. Even Schrodinger's cat wouldn't be affected by that computer.

[13] If you try this argument on zombies this is the point at which they suddenly renounce functionalism.

[14] I'm assuming that our Martian's instruments can record the internal states of the transistors in the computer's chips.

[15] Other fields of science like chemistry and biology are also not needed to explain the observed universe, but those frameworks of understanding are recognized to function only within limited parameters. No one would claim that chemistry alone can explain the Sun's source of energy, for instance. Because of its limited scope, chemistry isn't ontologically challenging while physics, being necessary, is. I believe that this is the reason so many physicists end up as zagnets.

[16] The secret to finding functioning computers (not just free-standing programs) wherever you look is in choosing pieces of the universe which are exerting influence on one another over time (this isn't hard). For instance, in my large meteor shower, all the meteors exert some gravitational pull on each other, so they are causally linked. If you searched hard enough, you could find a computer which read the relative motions of the meteors over a specific period of time as a record of the changing states of many minds, and the process of communication between them. In this case, an alien could not only find a computer program in a meteor shower, but a tangibly functional one.

Zombies sometimes object to this argument by saying that a "real" computer doesn't have to know what it will do in advance, while my constructed computers do. This is a remarkable argument, because it asserts that deterministic computers exercise a mystical kind of free will. Or Zombies might argue that computers are different from meteor showers because they have a special, practical relationship with their surroundings. These kinds of arguments are touching because they are a rehashing of the most sentimental old zagnet arguments for the specialness of people.

[17] It should be pointed out that many objects which aren't needed for an understanding of the universe can still be detected by instruments. An example would be chemicals (since chemistry is a theoretically unneeded, though immensely practical, layer of abstraction above physics). This suggests an ontological spectrum reflecting the dispensability of things. Chemicals could be said to "exist" more than computers do, and energy to exist even more, since it is both measurable and a less dispensable concept.

[18] I've decided to use the word "infinity" in the vernacular sense in parts of this essay to make it more accessible to non-technical readers. It should be taken to mean "unbounded".

[19] Zombies, changing the rules of the game, might suggest that we could detect a computer objectively by redefining it as the "best fitting" or "most efficient" finite state machine to explain the behavior of a selected piece of the universe (in this case the piece we call a computer). Accordingly, aliens would recognize our computers because our interpretation is in fact the best one available, and all who seek will arrive at approximately the same point of view. I think schemes such as this are really only hiding some step in which human guidance would be needed (not because humans have the best point of view, but because we don't). Obviously there is the matter of selecting a piece of the universe, which in itself might be the conveyor of "semantics". Even if the aliens can choose the right slice without human assistance, this problem is different from the meteor shower example above. In that problem, we searched for a particular computer, but in this one we'd have to find the best possible computer that included as much as possible of the meteor shower. In other words, we'd have to look at all of the computers hiding in nature instead of just finding one of them. This gives us an unbounded problem instead of a large finite one. It is just like hoping for an algorithm that you could feed a bunch of data into and then be rewarded with the best possible scientific theory to explain the data.

[20] 20 I was both bolstered and disappointed (I wanted to publish this idea first!) to learn that Searle has also argued that computation is not intrinsic to nature, in "The Rediscovery of the Mind". Searle's position is actually a little different from mine, in that he doesn't entirely dismiss the idea that some kind of computer could have an objective existence in the right context, and he doesn't view computers as being similar to other phantoms like language.

[21] And I would say they're even worse dualists. My dualism is cleanly defined by the existence of two different epistemological channels, the empirical and the subjective. Theirs is cloaked in weird fantasies of imaginary objects like information with undefined properties like "semantics".

[22] I am told by my friends who have experimented with psychedelics they have experienced this correlation, where every aspect of experience is radically altered by changes in the physical brain. What is notable to me is that experience itself continues during these radical "trips". This is, once again, why I choose to use the word "experience" instead of "consciousness". Consciousness is something which is said to exist in altered states, where experience is a thing, as I understand it, without state.

[23] As this essay demonstrates.

[24] See my essay "Agents of Alienation".

[25] Or Penrose's quantum computations.

[26] I'm not suggesting a "free-will" or conscious kind of choice. It is rather an implicit choice that has been made in the act of perception.

[27] Vector fields are the mathematical way of expressing the continuous aspect of the universe.

[28] Does subjectivity disappear if you're thorough enough? This is what some zombies believe. If enough well-instrumented alien scientists studied enough situations on Earth, would they eventually weed out, perhaps relying on an evolutionary process, all of the possible but more awkward interpretations of what's going on here? Would they eventually "parse" our world the way we do, into people with brains using words to refer to objects, because that interpretation is the easiest? This is similar to the idea addressed in footnote 19. I would argue that aliens who learn to think like us must have cheated and gotten a hint or two to find their way.

Some recent speculation concerned with "Complexity", coming from Stuart Kaufman, Brian Goodwin, and other, suggests that forms in the universe are limited to a far smaller variety than we might have thought, following the contours of a new class of mathematical objects, such as the "catastrophes". What I currently think is that even if this turns out to be right, it doesn't mean that the number of possible "layers of abstraction" would be similarly reduced. A limited variety of territories does not imply a limited number of maps. In fact, this thinking might be very compatible with the idea of the objective universe stated here ("an affinity for a particular infinity of possible slicings").

[29] Zombies will probably ask whether there is one dial per person, or one for the universe. I would reply that dials exist in "epistemological space" not physical space, so that question is not sensible.


Jaron's home page

The Problems of Consciousness Bibliography