Why the Chinese Room argument may work—but doesn't matter

Moreover, the presumptions here really are only those of computationalism—we already know that any computation can be performed by a wide variety of hardware, and indeed that any finite computation can be reduced to lookup tables (if A, do B and goto C). So doesn't it follow that if understanding Chinese is just computation, then Searle with his lookup tables must understand Chinese?

Scale arguments seem to miss the point. Unless Churchland can explain why computing in sub-second time makes consciousness but computing over several billion years doesn't (in the same way that indeed moving a magnet 10 times per second makes weak radio waves and moving it 1 billion times per second makes light), her "luminous room" counterargument can't work.

Yet it's important to understand what role scale does play here. Searle's argument cannot be used to undermine the Turing Test. This is because any entity which can pass a Turing Test must pass that test in real time—at the same rate of computation and response that a human being would. Given that the lookup tables for all possible 30-minute human interactions, written in 12-point font, would take up more space than the observable universe and require up to 90 billion years between steps, Searle's room of file cabinets can't possibly pass a real Turing Test. At best it could pass a highly-structured form of the Turing Test in which all the questions were known in advance—something I could make a Python script to do. And if we were to encode the full set of lookup rules more efficiently, say on a neural-network architecture of billions of microchips, saying the resulting entity—the resulting AI—wouldn't be conscious simply begs the question. Even with the Chinese Room notwithstanding, it's fair to say that any entity which passes a real Turing Test is conscious. (In fact, most people when asked agree that fictional AIs like Data from Star Trek: TNG or Romi from Andromeda would be, if real, conscious; I think this is precisely because these AIs pass a real-time Turing Test with flying colors.)

Inversely, this also explains why Dennett's evolutionary argument against zombies doesn't work. It could well be that zombies are possible, they're just so inefficient that they would never evolve. In his argument, Dennett presumes that zombies are "a bit simpler"—but in fact there is every reason to think that they would be vastly more complicated. A conscious human brain can do things in seconds which would take lookup tables billions of years. It could be that we have a choice between making real consciousness and simulating it—and it's easier to make it for real. I'm not committed to the idea that zombies are possible; in fact I don't think they are—but you can't prove this with evolutionary arguments. Evolutionary arguments merely prove that zombies are, if possible, inefficient.

As for the Turing Test, it's important to note that we aren't saying that anything which can't pass isn't conscious; the Turing Test was always meant to be a test of sufficiency for conciousness, not necessity; even Turing would have agreed that dogs and people with severe aphasia are conscious despite being unable to pass any kind of verbal test. And because of this I do think that Searle's argument does something to undermine computationalism, since it at the very least seems very counter-intuitive to say that the 90 billion light-year file cabinet system is conscious, but just a very slow, inefficient sort of consciousness that isn't smart enough to pass a Turing Test. I'm not sure whether that's true or false, but I agree it's very counter-intuitive.

So, I ask Searle: What else would be needed? What are these "causal powers" that neurons have which allow them to make consciousness?

Searle never answers this question, probably because he knows how trivial it makes his argument. Whatever the "causal powers" are, they are either physical or non-physical.

If they are non-physical, then despite all his claims to the contrary Searle is committed to substance dualism—and all the problems that substance dualism entails. This cannot be seriously considered credible in an age of neuropsychology; the brain and mind are too closely connected for us to posit any kind of non-physical "soul".

Otherwise, the "causal powers" must be physical. If they are physical, then we could create an artificial system capable of replicating them. Searle is right to say that a computer simulation of rain isn't rain—but surely we can't say that a machine which pours drops of water from the sky isn't making rain? Now, it could be that the "causal powers" are so closely tied to brain chemistry that we will end up synthesizing serotonin and making synthetic neurons that are actually alive. The resulting artificial brain would be indistinguishable from a natural brain—and hence our AI would be more like an artificial human. But how plausible is this really? Other biological functions are not so closely tied to their chemistry. An entity which intakes and outputs air is respirating, whether it has lungs and a diaphragm or an electronic pump. An entity which takes in light and carbon dioxide and makes sugar is photosynthesizing, regardless of whether it uses chlorophyll or photovoltaics. An entity which copies itself is reproducing, whether it uses sperm and ova or an electronic blueprint and construction apparatus. Biology happens to work well under carbon chemsitry—but carbon chemistry is not necessary for biological function.

Indeed, even within the human brain there are differences in neural structure and chemical composition—some neurons are serotonergic, others dopaminergic, still others GABAergic, and so on. Different animals have different chemicals in their brains—human serotonin receptors aren't the same as dog serotonin receptors. If nothing else, we ought to be able to construct an AI brain that has significant differences in chemical composition from any animal brain, even if for some reason it must in fact be pink and squishy. But is it really such a stretch to think that it need not be pink and squishy, that a network of silicon chips could work just as well?

Perhaps there is something more to consciousness than mere computation; but whatever else is required, that is surely something we could replicate in an artificial entity. Searle's "strong AI" is a straw man—artificial intelligence of even the Data and Romi sort remains possible under any physical account of consciousness.

Utilitarianism is not cold

This is a common intuition, and I don't think it is completely without foundation; but I am convinced that it is ultimately mistaken. While it might seem unconscionable to allow (e.g.) rape, murder, genocide in order to achieve some greater good, ultimately this revulsion comes from a failure of imagination regarding just how great that greater good would need to be. To justify a genocide, it would not be enough that the world would be better off without the targeted group; rather, it would have to be that without the genocide a far more terrible outcome would have occurred, like the total destruction of human civilization. Further, there must have been no other way to prevent such a terrible outcome---otherwise we ought to take that other path.

There is a kind of utilitarianism that is cold; the most naive Benthamite act-utilitarianism seems quite cold. If all we are measuring is subjective pleasure versus subjective pain, we ignore not only aesthetics and honesty, but indeed autonomy and death! To a Benthamite utilitarian, a painless death is morally neutral, and there is nothing wrong with benevolent slavery. Indeed, under such a philosophy something like the Matrix---a false but pleasant reality imposed upon us through delusion---is not only permissible, but good; it may even be obligatory. This seems quite cold indeed, but it is dangerously close to a straw man; no serious modern utilitarian would defend these notions. Indeed, nearly all utilitarians today agree with John Stuart Mill, and adopt a system of consequences that includes not only pleasure and pain simpliciter, but also the "pleasure" of aesthetics, the "pain" of fearing death, the "pleasure" of autonomy, the "pain" of dishonesty. One could argue that these experiences aren't really pleasure and pain, but in fact something broader, like attractive and aversive emotions generally (in fact, I would argue precisely this); but the point remains that serious utilitarianism today considers this broad range of experiences and not a simplistic calculation of subjective pleasure versus subjective pain. If the sense of coldness came from Benthamite utilitarianism, basically the argument is one of guilt by association.

Moreover, it seems to me that the other moral theories on offer are not only just as cold---they are in fact significantly colder. They value the wrong things, and turn a blind eye to suffering when preventing it would require us to sacrifice our moral righteousness.

In the first case, virtue theory values something like "human flourishing" or a "completeness of character" above all else. As such, a virtue theorist is committed to saying that if I must allow a million people to suffer and die in order to preserve my completeness of character, I not only may do so, in fact I ought to do so---it would be wrong of me to sacrifice my character to save that million people. What could be colder than that?

In the second case, deontology demands that we obey moral rules under all circumstances, without exception or hesitation. A deontologist is committed to saying that we must never break the rules---never kill an innocent person, or in Kant's infamous case, never even tell a lie---regardless of how many people must suffer as a result. Pacifism is a particularly extreme deontology in which the rule is literally "thou shalt not kill"; killing human beings is absolutely forbidden regardless of the circumstances. It's not hard to see how a nation of pacifists would be rapidly invaded and exploited by foreign powers.In essence, deontology values clean hands over all else: Regardless of whom it hurts, I must never dirty my hands with evil actions. This doesn't seem as cold as virtue theory, but it still seems quite cold!

On the other hand, one could devise a virtue theory or deontology that didn't have such terrible consequences; the most obvious way would be to define the virtues and rules such that they are based upon the predicted consequences. But once we've done this, aren't we really talking about utilitarianism anyway? If "virtue" is that which will make one likely to do good, and the "rules" are heuristics that are likely to produce good outcomes---that's utilitarianism! It's a nuanced Millian sort of utilitarianism, but that's what utilitarians have been advocating anyway.

Virulent Senioritis

It could just be luck: The infection and subsequent recovery triggered my depression, which triggered my migraines; the combination of the three and the resulting loss of attendance and focus has damaged my academic success. I have ambitious research plans that are difficult to achieve, and I reasonably expected better support in my research than I actually received; thereby I became discouraged.

But it could also be subconscious or semi-conscious intentions on my part. I stand before a crossroads in my life, with many options laid out before me. I could go on to graduate school; I could join the Peace Corps; I could seek a teaching fellowship; I could combine these things through Peace Corps Master's International; I could take a year off to finish my book; and there are thousands of other options I might sometimes be tempted to consider but would generally consider less viable.

And in the face of all these options, I am afraid, or anxious,or some similar aversive emotion. I fear that I will make the wrong choice, or that I won't be given the opportunity I deserve. I feel as a tiny kayak on a fast and forking river, trying desperately to choose my course as I am swept ahead. I feel that I need time to consider these options, time to reflect upon my true goals and aspirations, time to plan my future. And yet I am not being given this time; the hours and days and months carry me relentlessly closer to graduation.

"Life is what happens when you're busy making other plans" said John Lennon, and a month later he'd been murdered. I am continually reminded of this thought, because it is so troubling to me. I don't want my life to be so random, so far beyond my control.

I want time to think about what I want to do in my life; I want time to develop deeper theories of the Hard Problem; I want time to write my books. I want to shunt a log across the river so that I can plan my course.

I'm not sure this is a reasonable desire; life goes on regardless of what we want. Yet at the same time, I see so many people around me with no plans, no dreams, simply carried forth into lives in which they will ultimately be unhappy and unfulfilled. Worse, even if they are happy, they are unlikely to achieve great good in the world; some might be satisfied with helping only those near to them, but I feel a deep drive to achieve something great, something lasting. In order to do that I need to seriously reflect upon the best path for myself.

Perhaps I should take the next semester off? That's the closest I can see to shunting the river. I could spend the winter of 2010 reflecting, then come back in 2010-2011 to finish my degree. This feels unappealing to me... I think this is primarily because it seems "weak" or like "surrender". I feel as if I am giving in to my anxiety—or even to my depression. I am "losing" some sort of "battle".

Alternatively, with significant effort I could psych myself back to full potential, finish my degree in time. This latter plan raises additional questions: Should I cancel my honors thesis? What should I do after graduation—and shouldn't I be applying to these things now?

Increasingly I'm thinking time off is warranted. I'd need to make sure that I use the time—that it isn't just wasted—but I think I could do that. The sense of "weakness" or "surrender" increasingly seems a poor reason for choosing a particular direction in my life. Indeed, I feel a sense of relief simply from proposing the option: Even if I ultimately decide not to do so, I have found a way to shunt the river. Will my parents be disappointed? Will others think less of me? Possibly—though not necessarily. And even if I make the wrong choice, there will be time to change that decision—many people change careers well into middle age. I could stay on this path now and change later.

Yet this is my life, and my life's direction that I am trying to decide. I want to do it right the first time. My autonomy here is not negotiable. In the absence of some foolish fantasy of divine destiny, autonomy in planning our lives is essential to fulfillment. Achieving what I want means knowing what I want and planning appropriately. I should not sacrifice that autonomy for the sake of some mild and transient disapproval.

CFI hosted a speaker tonight.

A truly excellent analysis of theology