Concept Cagematch: Zombies vs Consciousness

The concept of Zombie is fundamental to understanding consciousness. By comparing Zombies to what we know of physics, can we discover anything about consciousness?

If Zombies are possible, then consciousness is purely ineffable, where even the incidence of consciousness is defined without reference to physics. There are two reasons Zombies may be impossible; either consciousness inevitably arises, or consciousness is a fundamental substance. The former, however, reduces to the latter, and in both cases everything is conscious. If everything is conscious, then the other-minds problem prevents consciousness from being accessible to experiment, resulting in essentially the same situation as with possible-Zombies, and we have an unavoidable contradiction.

I would like to clarify what exactly I mean by ‘consciousness,’ because as we can easily see, there’s a lot of potential confusion on what exactly many discussions of consciousness are discussing. There is a feature of my experience that I experience but at present cannot be explained, specifically the existence of that experience itself. When my experience exists I say, “I am conscious.” We can see that the language used to describe it is a bit weird, too, which is probably the source of many mistakes.

This is a problem primarily because consciousness does not appear to do anything. All the things commonly considered the functions of consciousness have been shown to be unnecessary, can be done with a purely mechanistic computer, or both. To attempt to probe this mystery, I analyze the inevitable logical consequences of the concept, “Philosophical Zombie.”

First I will assume that constructing a Zombie is possible.

Zombie World
A Zombie, strictly defined, is a permanently unconscious human being that can perfectly fool observers into believing that the Zombie is conscious.

This means there is no physical test - no physical interaction - that is different between the human and the Zombie, no conversation, creation, or surgery you can do to uncover which is which. This means that, physically speaking, consciousness does not exist. Thus, consciousness must be pure magic - it exists, can discriminate on where and when to occur, and yet will never affect the world. Much like unicorns, leprechauns, and limyin laparlaxian trumpets, which I just made up, never affect the world. (If this is the case, would the other-minds problem be joined by an other-unicorns problem?)

If Zombies are possible, consciousness is purely nonphysical.

Presumably, then, Zombies are impossible?

No Zombies Allowed; Physically and Mentally Incoherent
I will first assume that consciousness isn’t a fundamental physical entity, but arises through some interaction in the brain.

If we construct a pseudo-Zombie, a human specifically missing all components that lead to consciousness, there is some physical thing that the construct cannot do, specifically the physical function of that missing component. For sake of argument, the pseudo-Zombie cannot paint original paintings. By assumption, in this scenario the pseudo-Zombie is indeed unconscious.

Yet, this leads to a contradiction.

For example, while it is true that we can make new images with machines, most of these images are simply random noise from a random number generator. They are not true paintings, they do not have order and cannot be assigned a purpose; the artist clearly didn’t have anything in mind. However, we are closing in on a solution in the form of cellular automata. These simple programs can create breathtakingly complex and beautiful new images. Some can even create entire games. Yet, to assume that these simple scripts are conscious would be to imply that every script is conscious, which would mean that the program that runs my keyboard is conscious. From here it is a small step to prove that everything in conscious, contradicting the assumption that the Zombie is unconscious.

Now, this is only one example. It could be argued that it is deficient in generality. Therefore, I will now generalize it.

First, let me do a small proof of physicalism.

We can see that the numbers we use to count sheep inevitably imply the mathematics that we use to describe the quantum world. This kind of long-range consistency is a feature of consistent mathematical systems, and indeed apart from the rift between General Relativity and Quantum Field Theory, physics is completely unified. Next, quantum particles are affected by gravity; their math does not actually conflict in the real world. The rift is a human misunderstanding, which we will eventually fix.

Another feature of consistent mathematical systems is that every part either necessarily follows from or is implied by every other part. (Except the axioms, as per Godel.) We can see this in action with the success of theoretical physics, especially in Einstein's work.

From this idea we would expect that, for example, one could learn logic applicable to the physical world through studying mathematics.

One of the things that would happen if a non-mathematical object interacted with a mathematical one is that, unless the non-math object temporarily followed a consistent mathematical law, inconsistencies would pop up for the mathematical object, such as divide by zero errors. I don't know exactly what the physical manifestation of a 1/0 error would be, but 'oblivion shockwave' is not hyperbole.

We are physical. One part of physics is mathematical, therefore, every part of physics is mathematical, because we still exist.

Therefore, a Zombie under these assumptions would be every equation that describes a human but one, and the human would be all those plus the equation for consciousness.

Let's use the wave equation for the sake of argument. Again I'm using a specific equation for clarity, but my arguments will apply equally to all possible equations.

Usually, the wave equation is used to discover waves. Given any initial set of equations, if you can derive the wave equation, the equations describe a physical wave. This applies to light, water, air, strings, quantum mechanics, and many other situations. They all start differently, but end up the same. I shall assume that this equation is also conscious. (This would mean that all these things are also conscious, but specifically anything that cannot be worked into a wave equation is not conscious. I’m going to neglect the obvious conclusion that since everything has de Broglie wavelength, everything is conscious.)

Humans have brain waves, and waves are conscious, so humans are conscious.


In English, "The second time derivative of a function (u) is equal to the second space derivative of the function times a constant." (The constant is the wave velocity squared.)

So, which part differentiates this equation from the unconscious equations?

It can't be the constant, the variable, the power operator or the derivative operators, because these are found in an infinite variety of equations. If it is any particular element of the equation, every equation is conscious as an immediate consequence.

Can it be the function, u, itself? The solution to this equation is the sine function. The problem with this is that the sine function can be generalized. For example, a Fourier Transform can turn an arbitrary image into a combination of sine functions. (Your eye performs this exact transform in reverse, turning waveforms into images.) Or, simply set the sine function to a constant. Sin(x,t)=sine(π/2) is a valid solution, and Sin(x,t)=1. (The derivative is 1 is zero, and so the equation becomes 0=(c²)(0). Technically this is a wave of infinite wavelength and zero amplitude.) If you accepted sine as the 'conscious' function then you can simply re-derive everything from it, proving that every equation is conscious.

What I’ve done is I’ve simply set the equation equal to a single value that the equation takes at a single point, which can be applied to any equation. You could demand that sine(x) be at least first order in x. I would merely reply “What about sine(x)=sine(x^0.9)?” and bargain down from there. All cutoffs are arbitrary.

Does it also generalize to matrices and other types of functions with more than one output? Since each element is described by a simpler function: yes.

So, can it be the whole arrangement, where no element in particular is conscious, but put all together into the proper incantation, the whole is conscious?

There are two problems with this. First, every brain is different. If only the exact arrangement is conscious, then at most one person is conscious at any given time. This is also preposterous because even the wave equation itself is an ideal; very few actual physical situations correspond exactly to the equation.

Second, if we allow some leeway, how do we know where to stop? If we add new elements;+3

When does it stop being conscious? There is no qualitative difference between adding small correcting functions and adding 10²⁰, completely swamping the original function. Thus the cutoff is again arbitrary, and every equation is conscious. For instance, add a small correcting function multiplying everything by 0.9, and then add the equation for a line, y=mx+b. Then, jack the multiplying factor to 10^-20. The equation will be completely dominated by the equation for a line, but there will have been no non-arbitrary cutoff. Thus, every equation...

I will later do a case study of this analysis on an actual theory, Integrated Information Theory.

(Eventually I hope to perfect the actual equation for consciousness, which is special, but that is a separate essay.)

The inevitable conclusion is that, if Zombies are in fact logically incoherent, every equation is conscious, and we've redefined consciousness to be basically a synonym for 'existence.' Strong emergence is also and as a result, logically incoherent.

Emergent concepts are not necessary to explain the universe, and do not exist in a strict sense. Flocks don't exist, just the birds. But the birds don't exist, just their molecules, and so on. If the flock cannot be explained in terms of the particles that make it up, it cannot be explained at all and will not occur. Similarly, if consciousness is emergent, it is not necessary to explain a human, and does not exist is the strict sense, and thus isn't physical.

So, if Zombies are possible, then consciousness isn't physical. If Zombies are impossible because consciousness arises through emergence, then everything is conscious, because of the consistent nature of mathematics.

There's a second way of looking at this, which I mention because the mathematical treatment doesn't adequately illustrate the flaws of 'consciousness as epiphenomenon,' because you can look at the math two ways; in one, consciousness actually does stuff, like paint pictures, and in the second, the process that gives rise to painting pictures also gives rise to consciousness. The consequences of the second viewpoint do not follow directly.

To make this absolutely clear, I'm going to restate the definition twice.

"Zombies are impossible because there is some event that cannot occur without consciousness, such as painting original pictures."

"Zombies are impossible because there is some process, such as painting pictures, which cannot occur without also giving rise to consciousness."

Physically speaking these are almost identical; the process of painting is simultaneous to the process of consciousness; if consciousness ends so does the painting. This means we can never properly test for consciousness. In the first case, whatever it is that the Zombie cannot do, can also end in non-Zombies without consciousness ending, meaning consciousness is ineffable as is the case if Zombies are possible. Whatever arbitrary test we can concoct, it is a test of the effects of consciousness, not consciousness itself. In the second, consciousness has no effects per se; it is epiphenomenon, again ineffable.

None of the parts of the proposed Zombie are conscious in isolation; once you divide it up into modules (if that's not enough, fundamental particles) they stop being conscious. This is not a physical division, but rather an abstract one. The states and evolution of the individual particles will not change in this division, their interactions will be conserved virtually. If consciousness is not a fundamental physical thing, then these components cannot be considered conscious. This is analogous to analyzing a single term of the equation that describes the whole. Also analogous would be to consider a single brush stroke of the painting.

No emergent phenomenon can do things that cannot be explained in terms of pre-existing degrees of freedom in the underlying particles. Since our underlying particles do not have consciousness, it is not necessary to invoke consciousness to explain the whole; the painting cannot be anything but the sum of the brush strokes.

(Please don’t confuse the experience of the painting with the painting itself. By adding a conscious mind to a painting, the logical situation becomes very different.)

In fact, we know that every external action a human can take can be mimicked by a machine in isolation. There is no known function that only consciousness can perform. There is no test, not interaction.

Presumably, then, consciousness really is a fundamental physical concept?

Consciousness As Physical Axiom
Either consciousness is made of fundamental physical ‘stuff’ like leptons are, or math really is conscious, and physical things are fundamentally conscious in some way, analogous to spin, which is a property of every particle.

Obviously there must be some mechanism to divide consciousnesses up into discrete packages, unlike the various charge fields which permeate all space, and there's a good candidate, which is the Information Integration Theory or IIT that I mentioned earlier. In this, we measure the information content of an arbitrary set of elements, and compute how integrated the information is, to find the integrated information Φ. A coherent consciousness then is a set of conscious elements that have a maxima of integrated information. A brain, over a timescale of 0.2-3.0 seconds has a high Φ. Several brains in nearby human skulls has even more information but considered together have low integration because the brains are not strongly linked, and thus actually have a lower Φ.

This also gives a putative test of consciousness; the level of Φ. However, we run into the other-minds problem. The only way to assure that Φ corresponds to consciousness is to test for consciousness and see if it correlates to Φ. My brain hurts even writing that contradiction down.

It doesn’t matter which kind of fundamental thing consciousness is. It is subject to the other-minds problem. Which means that it’s impossible to test for. Which means it’s ineffable, exactly as in the Zombies-Possible regime.

And we have a contradiction. (A solution to the other-minds problem would obviously reverse this proof.)

So what are the remaining assumptions?

Consciousness exists.

Consciousness is physical.

There’s only one left to eliminate.

Dualism
Zombies are possible, but humans out-compete them anyway. That is, consciousness exists, is an event with causes and effects, but is not physical.

We build a Zombie. When the Zombie and the human are asleep, there is no test we can run to physically distinguish the two; they are physically identical. We conclude that there is no action the human can take that the Zombie cannot.

However, when we wake the Zombie and the human up and have them compete, perhaps in a painting contest or for mating privileges, the human wins. Consciousness is doing something for the human, but it is doing it nonphysically.

This immediately makes sense; if consciousness has effects, it must be an event, which means it must have causes, which correspond to energy costs. It will therefore be selected against by natural selection, unless there is a compensating upside. Therefore we can fairly safely assume that consciousness gives some benefit to evolutionary fitness, especially considering the huge amount of consciousness a human possesses. However, as explained above in the Zombie-impossible scenario, there is no physical benefit it can give; every possible function, described by any possible equation, can be reliably deployed by a machine. Therefore, consciousness must serve some nonphysical function. (Unless you want to accept that everything has full-on human consciousness, including undiscovered abstract mathematics.)

I suspect that the human will be acting on nonphysical information, which is unavailable to the Zombie. However, since we cannot define consciousness well, I have focused on what it isn't, instead of what it is.

However, as I mentioned in my small proof of physicalism, if the nonphysical consciousness does not interact consistently with physics, it will destroy the universe. Thus, while it would appear that consciousness is not physical, it must have some mathematically consistent interface with physics.

Specifically, there should be a hole in physical causality, which appears in the brain, that is plugged by consciousness. That, however, is the other article.

(Technically speaking, Zombies are still impossible. There is a specific component that uploads and downloads physical information to conscious information. You can just test for it. But! Direct testing is impossible, just as it is impossible to directly test quarks. Quarks are simply the most likely explanation, just as, for my component, consciousness is the most likely explanation.)