Peter Watts and p-zombies

John Ohno on 2015-08-20

I was surprised, upon listening to a two part interview with Peter Watts, to find him tentatively supporting Chalmer’s positions on qualia and the hard problem. Part of the reason is that Watts is a(n ex-) scientist with a background in biology and neuroscience, and also both very intelligent and spectacularly good at not avoiding unpleasant trains of thought. The other reason I was surprised is that I read Blindsight, and interpreted it as an amazingly good takedown of the Chalmers philosophical zombie idea along the same lines as Dennett’s.

This essay will contain spoilers for Blindsight, probably. Also, spoilers for the epistemology of Chalmers and Dennett. If you don’t like to learn things in orders not officially sanctioned by the establishment, I recommend you at least read Blindsight — it’s a great read, and Watts has been nice enough to put it online for free.

Chalmers presents the idea of consciousness as indicated by qualia — a representation of the subjective feeling of the outside world. His position, in my understanding, is that subjective feeling is a more difficult thing to model than other properties of the world. While I’m not sure about Chalmers himself, other people have used this idea that qualia is a “hard problem” as an excuse for reintroducing cartesian dualism into the world of epistemology — by claiming that qualia is so difficult to model that not even straight-up neurons can model it, and thus we need to bring in quantum nanotubules or some other structure as a stand-in for the soul.

A lot of people have been suspicious of the idea of qualia. After all, isn’t a representation a representation? Isn’t a subjective representation just a second-order representation? I agree with Dennett when he argues that it’s an unnecessary complication, with no evidence for it. I would furthermore argue that it’s a matter of preferring a mysterious answer to a mysterious question: complex behavior can be difficult to predict not because it’s irreducible — not because each piece is complex — but because lots of simple pieces combine in a complex way, but there’s a general tendency among people to try to keep emotional parity with explanations (mysterious things need to be explained in a way that retains the mystery or else you’ve lost the mystery; negative events can’t be explained as an interaction between purely positive intentions, or else where did the negative essence come from?) but ultimately reality doesn’t deal in emotional valences and so feelings of mystery do not need to be conserved.

Chalmers came up with a fascinating thought experiment in order to “prove” the existence of qualia. He suggested the idea of a ‘philosophical zombie’: a person indistinguishable from a regular person, but without qualia. Because qualia cannot be tested for, this person would be completely indistinguishable from a regular person.

Somehow, a lot of otherwise intelligent people thought that this was a good argument. I can’t see the invisible dragon in my garage, and therefore it must exist.

In Blindsight, Watts plays with a few variations on the philosophical zombie idea. He puts forth the idea of vampires being said to lack qualia — along with other cognitive anomalies that are of benefit to a humanoid with a very different position in the food chain. Certain optical illusions and cognitive biases don’t work on them. They have some differences in social behavior. They are largely lacking in empathy, without having the problems with impulse control that tend to be comorbid with lack of empathy in human sociopaths. A vampire, along with a split-brain patient, a personality collective, a person with extreme sensory modifications, and some other various neurodivergents take a space trip to meet a colony of intelligent starfish/squid-like aliens that are determined to have no qualia either and no sense of identity.

But, the ideas about qualia don’t line up here. I assumed it was on purpose.

Rather than ‘qualia’, each of these neurodivergent characters has some facility or attribute missing or strongly modified that is very clearly defined and very clearly not the same as qualia. And furthermore, each of these characters has very different behaviors based on their divergence from the norm. (This is along the same lines as the Rifters trilogy, particularly Starfish — we’re basically talking about circumstances where people who are psychologically and neurologically maladapted to normal life in a normal society end up being very well adapted to a fundamentally different environment.)

In other words, it’s a strong argument against philosophical zombies.

In the end of Blindsight, our protagonist gets back within radio range of Earth and can tell it’s been taken over by the vampires. Because Earth had stopped broadcasting music and entertainment, in favor of utilitarian communications. The vampires aren’t philosophical zombies, because they can be distinguished from humans. Because the particular kinds of things that they don’t experience lead them to live in a more utilitarian manner.

Indeed, no novel could deal with philosophical zombies. Because, by definition, philosophical zombies could not be distinguished from normal people. A novel about philosophical zombies could not be distinguished from a novel with no philosophical zombies in it.

Now, the argument for qualia is that, while human beings can experience something through their senses (like the color green), that experience cannot be identified in the brain itself. There is no neuron for ‘green’, and even if there was, the neuron itself wouldn’t be ‘green’ or contain the concept of ‘green’.

This argument has a handful of big flaws, some of which have been dissected elsewhere, so I’m going to dispatch it as efficiently as possible. First off, while some things do seem to have dedicated neurons (this is the ‘Grandmother Neuron’ model), most things don’t — however, this is not terribly unusual; we are very accustomed to another system for modeling the world where some configurations of state have single symbols and others have sets of meaningfully interconnected symbols: language. The word ‘green’ is not necessarily green — in fact, it might be red — and does not contain the concept of green, but instead gains its meaning from its relationship to other things. Ultimately, we can say that it gains real meaning by being in a relationship with other symbols in a manner that represents some configuration of the outside world as perceived through some people’s sensory apparatus, and gains utility insomuch as it allows us to communicate and make predictions. However, we can have syntactically meaningful configurations of symbols that could not have any semantic meaning — the colorless green ideas sleep furiously — or syntactically and semantically meaningful configurations of symbols that could not represent our universe — maxwell’s demon mounted the pink unicorn’s dragon-skin saddle and rode off at six times the speed of light in order to find some anti-entropic material and transmogrify it into orgone. Since language does this, there’s no reason for the brain to be incapable of it; since the brain makes language, the brain must be capable of doing it. It’s also not mysterious — even toy languages with heavily simplified grammars designed for computers to manipulate can do thing kind of thing (think RDF, or PROLOG).

As someone who has a background in biology and neurology, who works with words and language professionally, and who thinks deeply and clearly about most things, I would expect Watts to make these same judgments. If he has a counterargument in favor of qualia, I’d like to hear it. But, my general position is that to the extent that something that behaves similar to qualia exists, it is symbol manipulation, and to the degree that something like consciousness exists, it is something like self-simulation.

(Originally posted here)