The Generalized Anti-Zombie Principle

“Each prob­lem that I solved be­came a rule which served af­ter­wards to solve other prob­lems.”
—Rene Descartes, Dis­cours de la Methode

Zom­bies” are pu­ta­tively be­ings that are atom-by-atom iden­ti­cal to us, gov­erned by all the same third-party-visi­ble phys­i­cal laws, ex­cept that they are not con­scious.

Though the philos­o­phy is com­pli­cated, the core ar­gu­ment against zom­bies is sim­ple: When you fo­cus your in­ward aware­ness on your in­ward aware­ness, soon af­ter your in­ter­nal nar­ra­tive (the lit­tle voice in­side your head that speaks your thoughts) says “I am aware of be­ing aware”, and then you say it out loud, and then you type it into a com­puter key­board, and cre­ate a third-party visi­ble blog post.

Con­scious­ness, what­ever it may be—a sub­stance, a pro­cess, a name for a con­fu­sion—is not epiphe­nom­e­nal; your mind can catch the in­ner listener in the act of listen­ing, and say so out loud. The fact that I have typed this para­graph would at least seem to re­fute the idea that con­scious­ness has no ex­per­i­men­tally de­tectable con­se­quences.

I hate to say “So now let’s ac­cept this and move on,” over such a philo­soph­i­cally con­tro­ver­sial ques­tion, but it seems like a con­sid­er­able ma­jor­ity of Over­com­ing Bias com­menters do ac­cept this. And there are other con­clu­sions you can only get to af­ter you ac­cept that you can­not sub­tract con­scious­ness and leave the uni­verse look­ing ex­actly the same. So now let’s ac­cept this and move on.

The form of the Anti-Zom­bie Ar­gu­ment seems like it should gen­er­al­ize, be­com­ing an Anti-Zom­bie Prin­ci­ple. But what is the proper gen­er­al­iza­tion?

Let’s say, for ex­am­ple, that some­one says: “I have a switch in my hand, which does not af­fect your brain in any way; and iff this switch is flipped, you will cease to be con­scious.” Does the Anti-Zom­bie Prin­ci­ple rule this out as well, with the same struc­ture of ar­gu­ment?

It ap­pears to me that in the case above, the an­swer is yes. In par­tic­u­lar, you can say: “Even af­ter your switch is flipped, I will still talk about con­scious­ness for ex­actly the same rea­sons I did be­fore. If I am con­scious right now, I will still be con­scious af­ter you flip the switch.”

Philoso­phers may ob­ject, “But now you’re equat­ing con­scious­ness with talk­ing about con­scious­ness! What about the Zom­bie Master, the chat­bot that re­gur­gi­tates a remixed cor­pus of am­a­teur hu­man dis­course on con­scious­ness?”

But I did not equate “con­scious­ness” with ver­bal be­hav­ior. The core premise is that, among other things, the true refer­ent of “con­scious­ness” is also the cause in hu­mans of talk­ing about in­ner listen­ers.

As I ar­gued (at some length) in the se­quence on words, what you want in defin­ing a word is not always a perfect Aris­totelian nec­es­sary-and-suffi­cient defi­ni­tion; some­times you just want a trea­sure map that leads you to the ex­ten­sional refer­ent. So “that which does in fact make me talk about an un­speak­able aware­ness” is not a nec­es­sary-and-suffi­cient defi­ni­tion. But if what does in fact cause me to dis­course about an un­speak­able aware­ness, is not “con­scious­ness”, then...

...then the dis­course gets pretty fu­tile. That is not a knock­down ar­gu­ment against zom­bies—an em­piri­cal ques­tion can’t be set­tled by mere difficul­ties of dis­course. But if you try to defy the Anti-Zom­bie Prin­ci­ple, you will have prob­lems with the mean­ing of your dis­course, not just its plau­si­bil­ity.

Could we define the word “con­scious­ness” to mean “what­ever ac­tu­ally makes hu­mans talk about ‘con­scious­ness’”? This would have the pow­er­ful ad­van­tage of guaran­tee­ing that there is at least one real fact named by the word “con­scious­ness”. Even if our be­lief in con­scious­ness is a con­fu­sion, “con­scious­ness” would name the cog­ni­tive ar­chi­tec­ture that gen­er­ated the con­fu­sion. But to es­tab­lish a defi­ni­tion is only to promise to use a word con­sis­tently; it doesn’t set­tle any em­piri­cal ques­tions, such as whether our in­ner aware­ness makes us talk about our in­ner aware­ness.

Let’s re­turn to the Off-Switch.

If we al­low that the Anti-Zom­bie Ar­gu­ment ap­plies against the Off-Switch, then the Gen­er­al­ized Anti-Zom­bie Prin­ci­ple does not say only, “Any change that is not in-prin­ci­ple ex­per­i­men­tally de­tectable (IPED) can­not re­move your con­scious­ness.” The switch’s flip­ping is ex­per­i­men­tally de­tectable, but it still seems highly un­likely to re­move your con­scious­ness.

Per­haps the Anti-Zom­bie Prin­ci­ple says, “Any change that does not af­fect you in any IPED way can­not re­move your con­scious­ness”?

But is it a rea­son­able stipu­la­tion to say that flip­ping the switch does not af­fect you in any IPED way? All the par­ti­cles in the switch are in­ter­act­ing with the par­ti­cles com­pos­ing your body and brain. There are grav­i­ta­tional effects—tiny, but real and IPED. The grav­i­ta­tional pull from a one-gram switch ten me­ters away is around 6 * 10-16 m/​s2. That’s around half a neu­tron di­ame­ter per sec­ond per sec­ond, far be­low ther­mal noise, but way above the Planck level.

We could flip the switch light-years away, in which case the flip would have no im­me­di­ate causal effect on you (what­ever “im­me­di­ate” means in this case) (if the Stan­dard Model of physics is cor­rect).

But it doesn’t seem like we should have to al­ter the thought ex­per­i­ment in this fash­ion. It seems that, if a dis­con­nected switch is flipped on the other side of a room, you should not ex­pect your in­ner listener to go out like a light, be­cause the switch “ob­vi­ously doesn’t change” that which is the true cause of your talk­ing about an in­ner listener. What­ever you re­ally are, you don’t ex­pect the switch to mess with it.

This is a large step.

If you deny that it is a rea­son­able step, you had bet­ter never go near a switch again. But still, it’s a large step.

The key idea of re­duc­tion­ism is that our maps of the uni­verse are multi-level to save on com­put­ing power, but physics seems to be strictly sin­gle-level. All our dis­course about the uni­verse takes place us­ing refer­ences far above the level of fun­da­men­tal par­ti­cles.

The switch’s flip does change the fun­da­men­tal par­ti­cles of your body and brain. It nudges them by whole neu­tron di­ame­ters away from where they would have oth­er­wise been.

In or­di­nary life, we gloss a change this small by say­ing that the switch “doesn’t af­fect you”. But it does af­fect you. It changes ev­ery­thing by whole neu­tron di­ame­ters! What could pos­si­bly be re­main­ing the same? Only the de­scrip­tion that you would give of the higher lev­els of or­ga­ni­za­tion—the cells, the pro­teins, the spikes trav­el­ing along a neu­ral axon. As the map is far less de­tailed than the ter­ri­tory, it must map many differ­ent states to the same de­scrip­tion.

Any rea­son­able sort of hu­man­ish de­scrip­tion of the brain that talks about neu­rons and ac­tivity pat­terns (or even the con­for­ma­tions of in­di­vi­d­ual micro­tubules mak­ing up ax­ons and den­drites) won’t change when you flip a switch on the other side of the room. Nu­clei are larger than neu­trons, atoms are larger than nu­clei, and by the time you get up to talk­ing about the molec­u­lar level, that tiny lit­tle grav­i­ta­tional force has van­ished from the list of things you bother to track.

But if you add up enough tiny lit­tle grav­i­ta­tional pulls, they will even­tu­ally yank you across the room and tear you apart by tidal forces, so clearly a small effect is not “no effect at all”.

Maybe the tidal force from that tiny lit­tle pull, by an amaz­ing co­in­ci­dence, pulls a sin­gle ex­tra cal­cium ion just a tiny bit closer to an ion chan­nel, caus­ing it to be pul­led in just a tiny bit sooner, mak­ing a sin­gle neu­ron fire in­finites­i­mally sooner than it would oth­er­wise have done, a differ­ence which am­plifies chaot­i­cally, fi­nally mak­ing a whole neu­ral spike oc­cur that oth­er­wise wouldn’t have oc­curred, send­ing you off on a differ­ent train of thought, that trig­gers an epilep­tic fit, that kills you, caus­ing you to cease to be con­scious...

If you add up a lot of tiny quan­ti­ta­tive effects, you get a big quan­ti­ta­tive effect—big enough to mess with any­thing you care to name. And so claiming that the switch has liter­ally zero effect on the things you care about, is tak­ing it too far.

But with just one switch, the force ex­erted is vastly less than ther­mal un­cer­tain­ties, never mind quan­tum un­cer­tain­ties. If you don’t ex­pect your con­scious­ness to flicker in and out of ex­is­tence as the re­sult of ther­mal jig­gling, then you cer­tainly shouldn’t ex­pect to go out like a light when some­one sneezes a kilo­me­ter away.

The alert Bayesian will note that I have just made an ar­gu­ment about ex­pec­ta­tions, states of knowl­edge, jus­tified be­liefs about what can and can’t switch off your con­scious­ness.

This doesn’t nec­es­sar­ily de­stroy the Anti-Zom­bie Ar­gu­ment. Prob­a­bil­ities are not cer­tain­ties, but the laws of prob­a­bil­ity are the­o­rems; if ra­tio­nal­ity says you can’t be­lieve some­thing on your cur­rent in­for­ma­tion, then that is a law, not a sug­ges­tion.

Still, this ver­sion of the Anti-Zom­bie Ar­gu­ment is weaker. It doesn’t have the nice, clean, ab­solutely clear-cut sta­tus of, “You can’t pos­si­bly elimi­nate con­scious­ness while leav­ing all the atoms in ex­actly the same place.” (Or for “all the atoms” sub­sti­tute “all causes with in-prin­ci­ple ex­per­i­men­tally de­tectable effects”, and “same wave­func­tion” for “same place”, etc.)

But the new ver­sion of the Anti-Zom­bie Ar­gu­ment still car­ries. You can say, “I don’t know what con­scious­ness re­ally is, and I sus­pect I may be fun­da­men­tally con­fused about the ques­tion. But if the word refers to any­thing at all, it refers to some­thing that is, among other things, the cause of my talk­ing about con­scious­ness. Now, I don’t know why I talk about con­scious­ness. But it hap­pens in­side my skull, and I ex­pect it has some­thing to do with neu­rons firing. Or maybe, if I re­ally un­der­stood con­scious­ness, I would have to talk about an even more fun­da­men­tal level than that, like micro­tubules, or neu­ro­trans­mit­ters diffus­ing across a synap­tic chan­nel. But still, that switch you just flipped has an effect on my neu­ro­trans­mit­ters and micro­tubules that’s much, much less than ther­mal noise at 310 Kelvin. So what­ever the true cause of my talk­ing about con­scious­ness may be, I don’t ex­pect it to be hugely af­fected by the grav­i­ta­tional pull from that switch. Maybe it’s just a tiny lit­tle in­finites­i­mal bit af­fected? But it’s cer­tainly not go­ing to go out like a light. I ex­pect to go on talk­ing about con­scious­ness in al­most ex­actly the same way af­ter­ward, for al­most ex­actly the same rea­sons.”

This ap­pli­ca­tion of the Anti-Zom­bie Prin­ci­ple is weaker. But it’s also much more gen­eral. And, in terms of sheer com­mon sense, cor­rect.

The re­duc­tion­ist and the sub­stance du­al­ist ac­tu­ally have two differ­ent ver­sions of the above state­ment. The re­duc­tion­ist fur­ther­more says, “What­ever makes me talk about con­scious­ness, it seems likely that the im­por­tant parts take place on a much higher func­tional level than atomic nu­clei. Some­one who un­der­stood con­scious­ness could ab­stract away from in­di­vi­d­ual neu­rons firing, and talk about high-level cog­ni­tive ar­chi­tec­tures, and still de­scribe how my mind pro­duces thoughts like ‘I think there­fore I am’. So nudg­ing things around by the di­ame­ter of a nu­cleon, shouldn’t af­fect my con­scious­ness (ex­cept maybe with very small prob­a­bil­ity, or by a very tiny amount, or not un­til af­ter a sig­nifi­cant de­lay).”

The sub­stance du­al­ist fur­ther­more says, “What­ever makes me talk about con­scious­ness, it’s got to be some­thing be­yond the com­pu­ta­tional physics we know, which means that it might very well in­volve quan­tum effects. But still, my con­scious­ness doesn’t flicker on and off when­ever some­one sneezes a kilo­me­ter away. If it did, I would no­tice. It would be like skip­ping a few sec­onds, or com­ing out of a gen­eral anes­thetic, or some­times say­ing, “I don’t think there­fore I’m not.” So since it’s a phys­i­cal fact that ther­mal vibra­tions don’t dis­turb the stuff of my aware­ness, I don’t ex­pect flip­ping the switch to dis­turb it ei­ther.”

Either way, you shouldn’t ex­pect your sense of aware­ness to van­ish when some­one says the word “Abra­cadabra”, even if that does have some in­finites­i­mal phys­i­cal effect on your brain—

But hold on! If you hear some­one say the word “Abra­cadabra”, that has a very no­tice­able effect on your brain—so large, even your brain can no­tice it. It may al­ter your in­ter­nal nar­ra­tive; you may think, “Why did that per­son just say ‘Abra­cadabra’?”

Well, but still you ex­pect to go on talk­ing about con­scious­ness in al­most ex­actly the same way af­ter­ward, for al­most ex­actly the same rea­sons.

And again, it’s not that “con­scious­ness” is be­ing equated to “that which makes you talk about con­scious­ness”. It’s just that con­scious­ness, among other things, makes you talk about con­scious­ness. So any­thing that makes your con­scious­ness go out like a light, should make you stop talk­ing about con­scious­ness.

If we do some­thing to you, where you don’t see how it could pos­si­bly change your in­ter­nal nar­ra­tive—the lit­tle voice in your head that some­times says things like “I think there­fore I am”, whose words you can choose to say aloud—then it shouldn’t make you cease to be con­scious.

And this is true even if the in­ter­nal nar­ra­tive is just “pretty much the same”, and the causes of it are also pretty much the same; among the causes that are pretty much the same, is what­ever you mean by “con­scious­ness”.

If you’re won­der­ing where all this is go­ing, and why it’s im­por­tant to go to such tremen­dous lengths to pon­der such an ob­vi­ous-seem­ing Gen­er­al­ized Anti-Zom­bie Prin­ci­ple, then con­sider the fol­low­ing de­bate:

Albert: “Sup­pose I re­placed all the neu­rons in your head with tiny robotic ar­tifi­cial neu­rons that had the same con­nec­tions, the same lo­cal in­put-out­put be­hav­ior, and analo­gous in­ter­nal state and learn­ing rules.”

Ber­nice: “That’s kil­ling me! There wouldn’t be a con­scious be­ing there any­more.”

Charles: “Well, there’d still be a con­scious be­ing there, but it wouldn’t be me.

Sir Roger Pen­rose: “The thought ex­per­i­ment you pro­pose is im­pos­si­ble. You can’t du­pli­cate the be­hav­ior of neu­rons with­out tap­ping into quan­tum grav­ity. That said, there’s not much point in me tak­ing fur­ther part in this con­ver­sa­tion.” (Wan­ders away.)

Albert: “Sup­pose that the re­place­ment is car­ried out one neu­ron at a time, and the swap oc­curs so fast that it doesn’t make any differ­ence to global pro­cess­ing.”

Ber­nice: “How could that pos­si­bly be the case?”

Albert: “The lit­tle robot swims up to the neu­ron, sur­rounds it, scans it, learns to du­pli­cate it, and then sud­denly takes over the be­hav­ior, be­tween one spike and the next. In fact, the imi­ta­tion is so good, that your out­ward be­hav­ior is just the same as it would be if the brain were left undis­turbed. Maybe not ex­actly the same, but the causal im­pact is much less than ther­mal noise at 310 Kelvin.”

Charles: “So what?”

Albert: “So don’t your be­liefs vi­o­late the Gen­er­al­ized Anti-Zom­bie Prin­ci­ple? What­ever just hap­pened, it didn’t change your in­ter­nal nar­ra­tive! You’ll go around talk­ing about con­scious­ness for ex­actly the same rea­son as be­fore.”

Ber­nice: “Those lit­tle robots are a Zom­bie Master. They’ll make me talk about con­scious­ness even though I’m not con­scious. The Zom­bie World is pos­si­ble if you al­low there to be an added, ex­tra, ex­per­i­men­tally de­tectable Zom­bie Master—which those robots are.”

Charles: “Oh, that’s not right, Ber­nice. The lit­tle robots aren’t plot­ting how to fake con­scious­ness, or pro­cess­ing a cor­pus of text from hu­man am­a­teurs. They’re do­ing the same thing neu­rons do, just in sili­con in­stead of car­bon.”

Albert: “Wait, didn’t you just agree with me?”

Charles: “I never said the new per­son wouldn’t be con­scious. I said it wouldn’t be me.

Albert: “Well, ob­vi­ously the Anti-Zom­bie Prin­ci­ple gen­er­al­izes to say that this op­er­a­tion hasn’t dis­turbed the true cause of your talk­ing about this me thing.”

Charles: “Uh-uh! Your op­er­a­tion cer­tainly did dis­turb the true cause of my talk­ing about con­scious­ness. It sub­sti­tuted a differ­ent cause in its place, the robots. Now, just be­cause that new cause also hap­pens to be con­scious—talks about con­scious­ness for the same gen­er­al­ized rea­son—doesn’t mean it’s the same cause that was origi­nally there.”

Albert: “But I wouldn’t even have to tell you about the robot op­er­a­tion. You wouldn’t no­tice. If you think, go­ing on in­tro­spec­tive ev­i­dence, that you are in an im­por­tant sense “the same per­son” that you were five min­utes ago, and I do some­thing to you that doesn’t change the in­tro­spec­tive ev­i­dence available to you, then your con­clu­sion that you are the same per­son that you were five min­utes ago should be equally jus­tified. Doesn’t the Gen­er­al­ized Anti-Zom­bie Prin­ci­ple say that if I do some­thing to you that al­ters your con­scious­ness, let alone makes you a com­pletely differ­ent per­son, then you ought to no­tice some­how?”

Ber­nice: “Not if you re­place me with a Zom­bie Master. Then there’s no one there to no­tice.”

Charles: “In­tro­spec­tion isn’t perfect. Lots of stuff goes on in­side my brain that I don’t no­tice.”

Albert: “You’re pos­tu­lat­ing epiphe­nom­e­nal facts about con­scious­ness and iden­tity!”

Ber­nice: “No I’m not! I can ex­per­i­men­tally de­tect the differ­ence be­tween neu­rons and robots.”

Charles: “No I’m not! I can ex­per­i­men­tally de­tect the mo­ment when the old me is re­placed by a new per­son.”

Albert: “Yeah, and I can de­tect the switch flip­ping! You’re de­tect­ing some­thing that doesn’t make a no­tice­able differ­ence to the true cause of your talk about con­scious­ness and per­sonal iden­tity. And the proof is, you’ll talk just the same way af­ter­ward.”

Ber­nice: “That’s be­cause of your robotic Zom­bie Master!”

Charles: “Just be­cause two peo­ple talk about ‘per­sonal iden­tity’ for similar rea­sons doesn’t make them the same per­son.”

I think the Gen­er­al­ized Anti-Zom­bie Prin­ci­ple sup­ports Albert’s po­si­tion, but the rea­sons shall have to wait for fu­ture posts. I need other pre­req­ui­sites, and be­sides, this post is already too long.

But you see the im­por­tance of the ques­tion, “How far can you gen­er­al­ize the Anti-Zom­bie Ar­gu­ment and have it still be valid?”

The makeup of fu­ture galac­tic civ­i­liza­tions may be de­ter­mined by the an­swer...