Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

The map of p-zombies

6 Post author: turchin 30 July 2016 09:12AM
No real p-zombies exist in any probable way, but a lot of ideas about them have been suggested. This map is the map of ideas. It may be fun or may be useful.

The most useful application of p-zombies research is to determine whether we could loose something important during uploading.

We have to solve the problem of consciousness before we will be uploaded. It will be the most stupid end of the world: everybody is alive and happy but everybody is p-zombie. 

Most ideas here are from Stanford Encyclopedia of Philosophy, Lesswrong wiki, Rational wiki, recent post of EY and from works of Chalmers and Dennett. Some ideas are mine. 

The pdf is here.


Comments (14)

Comment author: kokotajlod 04 August 2016 08:14:27PM 0 points [-]

I disagree with your characterization of 0. You say that it is incompatible with physicalism, but that seems false. Indeed it seems to be a very mainstream physicalist view to say "I am a physical object--my brain. So a copy of me would have the same experiences, but it would not be me."

Comment author: turchin 05 August 2016 12:34:50AM 0 points [-]

Objection accepted.

Comment author: mwengler 12 August 2016 03:00:21PM 1 point [-]

1) I do not understand why our experience of identical twins does not play into most discussions of my copy being "the same person as me." We know that twins do not share the same consciousness (unless Occam's razor is wrong and they are all lying.) We know from that that if we made a copy without destroying the original that the copy and the original would not share a consciousness. So why isn't at least the possibiliy (I would estimate overwhelming likelihood) that a copy is a different consciousness than the original, and that destroying the original kills one consciousness while making a copy creates a different consciousness, and that these are separate processes?

2) Does philosophy talk somewhere about what I would call "outer" and "inner" worlds? I know I'm conscious because I participate in my inner world. I figure by Occam's razor that you are conscious, but I don't have direct experience of your consciousness, because I can only see you in my outer world. We don't talk anywhere near as much about "inner" world because we don't share that with others, while our "outer" experiences are shared, and we have evolved a host of techniques including language and science for processing "outer" experiences. But "inner" experiences don't benefit from language and science because they are, so far, locked away inside us, not social phenomenon. Because I think the idea that our copy is a continuation of our own consciousness is a mistake we can make if we don't realize there is an "inner" experience quite distinct from our "outer" experiences. So sure, my copy thinks he is continuously conscious, and therefore may think my consciousness has jumped into him, but that is because to my copy, I am part of his "outer" world. But if a non-destructive copy of me was made, I think it is obvious from what we know about twins that despite my copies eloquence at explaining his continuity from me, that I, the original, would resist being killed as superflous. Yes in everybody else's outer world, where consciousness of others is indirectly inferred, they can't tell that my copy is not a continuation of my consciousness in separate matter. But in my inner world, it seems pretty clear that I, (the original) would know.

Comment author: turchin 12 August 2016 04:56:09PM 0 points [-]

I am going soon to publish Identity map which will sum up my research of the identity problem.

1) If twins are really similar, there will be indexical question for them. Each of them will not know if he is twin 1 or twin 2. So in practical situations they should think that any event which affect one of twins has 50 per cent situation to happen with them. So consciousness will not jump from one twin to another. It is already shared in this example.

2) Philosophers are well aware of the problem and it is called "other minds problem" and "hard problem" of consciousness. the main problem here is that if we adopt idea that consciousness could be different without any physical difference between the copies, we adopt the idea of p-zombies and reject physicalism that is modern version of materialism. It almost the same as to say that immaterial soul exist. It is very strong statement.

Comment author: g_pepper 12 August 2016 05:39:18PM 0 points [-]

any event which affect one of twins has 50 per cent situation to happen with them

I don't understand what that means. For example, if I were to punch one of a pair of identical twins, the twin that I punched would feel the pain, and the twin that I did not punch would not feel any pain. In this regard, the two twins would have no more of a shared consciousness than would two strangers.

I suspect that if someone were non-destructively copied, the situation would be similar, as mwengler suggests. The two would have a common set of memories up until the point that the copy was made, but from that point on they would have no common or shared consciousness; they would be two distinct people (albeit with an uncanny resemblance to each other).

Comment author: turchin 12 August 2016 06:52:28PM *  0 points [-]

It means the following in your example:

Suppose I am one of twins, but I don't know which twin I am. But I know that twin 1 will be hit in face in next 1 minute. In this case I am in the situation of so called indexical uncertainty and my best guess is that I have 50 per cent probability of being twin 1 and thus 50 per cent probability being hit in the future. After I will be hit, I will know with 100 per cent probability that I am twin 1.

In general there is problem with the idea of copies, as it mixes several different ideas and it results into paradoxes. The idea of copies should be broken on several ideas:

1) "past copies" - that is the copies of me which was made before now moment (twins are best example of them),

2) "mirror copies" which are exact my copies now

3) "future copies" - which will be made from me in future.

Comment author: g_pepper 12 August 2016 07:33:59PM *  0 points [-]

Thanks for the clarification. However when you say:

Suppose I am one of twins, but I don't know which twin I am. But I know that twin 1 will be hit in face in next 1 minute. In this case I am in the situation of so called indexical uncertainty and my best guess is that I have 50 per cent probability of being twin 1 and thus 50 per cent probability being hit in the future. After I will be hit, I will know with 100 per cent probability that I am twin 1.

This is really not unique to twins. Suppose I choose two strangers and tell them that I will flip a coin and punch one or the other of them in the face in one minute, depending on the outcome of the coin toss. Assuming the strangers stick around, they will be in the same position as the two twins in your example. So, I don't really see how this tells us anything interesting about identity, twins or copies.

Also, considering the first two of your three cases of copies, i.e. past copies and mirror copies, it seems to me that if I make a nondestructive copy of a person, the two will be mirror copies only for an instant; once the copy process completes and the two "wake up" (presuming that copying is done from an unconscious state), the two copies will immediately begin diverging based on different stimuli, etc., and will therefore become instances of case 1 (past copies). So, from a practical standpoint, there can be no instances of case 2 for any meaningful duration of time (that is to say, for enough time to interact with the copies).

Comment author: turchin 12 August 2016 07:58:12PM 0 points [-]

I would like to clarify my position: Identity is complex social adaptation and it is directly connected with hard problem of consciousness. So we can't finally solve any identity paradox on our current level of knowledge.

Indexical uncertainty in case of twins is a trick which may be used to skip identity problem. It doesn't prove that twins are identical. It just makes it not important.

So it doesn't prove that twins are the same. It may work for very different people as long as everyone don't know who is who. But it could be reasonable guide to make decisions in the situations where many my copies exist (including uploading, quantum multiverse statistic etc)

But it is not the only principle. Another one is "conservative approach" - that is we should try to preserve as much identity as possible as we don't know what is identity.

Comment author: g_pepper 12 August 2016 09:12:24PM *  0 points [-]

Identity is complex social adaptation and it is directly connected with hard problem of consciousness.

I agree that identity is directly connected with the hard problem of consciousness. That identity is a social adaptation seems plausible (to me) but not certain.

So we can't finally solve any identity paradox on our current level of knowledge.

It seems to me that, per mwengler's observation, we already have past copies (your category number 1); identical twins are past copies that branched shortly after conception. Past copies, it seems to me, do not share a common identity and are distinct people with distinct conscious experiences. I'm not sure that I see any identity paradoxes involving past copies.

Current, mirror copies (your category number 2), cannot exist in a conscious state for any meaningful amount of time (except perhaps as EMs where the hosting environment ensures that they have identical stimuli, are kept in sync from a simulation standpoint, etc.), so mirror copies can be ignored, it seems to me.

Future copies (your category number 3) do seem to have some paradoxes (or at least they are unclear to me). Specifically,

  1. if I know that I am going to be non-destructively copied in five minutes, should I care more about one future copy than I do the other? I suspect that I should not.

  2. If I know that I am going to be copied but the original will be destroyed in the process, should this concern me? It seems like, per #1, it should not. But, somehow, I don't think that I'd be very eager to go through with a destructive copy process.

  3. How can it be that the identity of a person and his/her future copy is the same (which seems plausible), but two past copies have distinct identities? It seems like personal identity should be transitive.

So it seems to me that future copies are paradoxical. And of course future copies will matter if/when uploading becomes possible, so we will eventually need to resolve (or accept) the paradoxes.

Comment author: turchin 12 August 2016 09:41:01PM 1 point [-]

I agree that only future copies are paradoxical. And most paradoxes are about are decisions. So we could try to solve them as decision theory problems.

If identity is something real, it should have properties, like transitivness, and copy paradoxes should have definite solutions.

But is identity is only illusion or social agreement we could redefine it according our needs.

One of the possible approaches to identity problem is to ask why identity exist at all? What are its roots?

One of its roots is in evolutionary psychology. In early human societies each person should be able to know what will be consequences of his actions (pain, paternity, meal) and who knows what in his tribe. It required very strong feeling of personal identity and ability to distinguish it. It gave rise to names and ability to distinguish individuality of other tribe members as well as idea that they are the same tribe members everyday. (Some people change their identity during initiation rituals by the way). But for different living beings identity could evolve differently - for viruses it include all viruses with the same code for example.

It resulted that our culture and our brain have very strong built-in instruments to distinguish identity - of me and my family members. Unfortunately, than we come to the age of uploading, these old built-in structures start to protest our new ways of breeding and survival. And here appear most paradoxes.

But this doesn't solve identity problem, as evolution may be able to harness some kind of natural process, which provide physical basis for identity, like qualia, causality, continuity, you name it. It had harnessed many other physical effects before for its goals, like light, sound, chemistry etc.

(Another solution to identity origins is that God created souls as basis of identity and controls their identity every moment, but we could not take such idea seriously in rationalist forum).

Comment author: mwengler 11 September 2016 11:00:46PM *  0 points [-]

if we adopt idea that consciousness could be different without any physical difference between the copies, we adopt the idea of p-zombies and reject physicalism that is modern version of materialism. It almost the same as to say that immaterial soul exist. It is very strong statement.

Not relevant to the problem. If you create a copy of me, the copy is not identical, if for no other reason than it occupies a different location than I do. I agree that if it occupied the same location that I do, atom for atom and quark for quark, that could lead to the concern you express. But copies cannot occupy the same location, and so there is no problem having the copy to the left be one consciousness while the original to the right is a different consciousness.

The strongest claim I might accept would be that both the original and the copy have "valid" claims to be the continuation of the pre-copying single consciousness that was me back then. But no matter how you slice it, killing the original when you make the copy is still destroying a separate consciousness, even if the remaining consciousness thinks it is the only continuation of the pre-copy consciousness.

Comment author: turchin 13 September 2016 08:22:51AM 0 points [-]

If we adopt criteria of identity based of same location, next-moment-of-me will be not me, as I move, Earth move etc. It results in situation where I will die every millisecond.

We could also imagine situation where Omega put my copy in my location, but instantly moves me to the next room.

Comment author: mwengler 11 January 2017 03:52:43PM 0 points [-]

Clearly we can differentiate between different-location-same-time and different-location-different-time. Two things in different-location-same-time are different things. Two things different-location-different-time may be same thing or may be different thing depending on the path through time. Your mathematical style of abstraction in thinking about identity will only be useful at explaining the real world if it is matched to real world processes, and does not ignore important real world insights.

Comment author: turchin 11 January 2017 05:24:07PM 0 points [-]

In fact my goal was to get rid of mathematical definitions of identity and move to instrumental definition. That is the definition of identity depends of the goal I am solving and identity itself is not a thing that actually exists but a supportive term. That is why we have problems this it.

Basically, there is two types of question where we need notion of "identity":

1) What I will experience in the next moment? (first-eye view) 2) Is it the same person? (third-eye view)