Christof Koch: Could the Internet Learn to Feel?
In a new interview by Steve Paulson, neuroscientist Christof Koch, a self-described “romantic reductionist,” discusses the science of consciousness, including the notion that all matter is, to varying degrees, sentient.
A few quotes from Koch follow.
In answer to the question “what makes consciousness such a difficult problem”:
“You actually see a world. How does this picture get into your head? That’s the mystery. And because it has both an exterior, third-person perspective as well as an interior, first-person perspective, it’s unique among all the phenomena in the universe. This means it’s a little bit more difficult to attack using a scientific point of view. ”
“The average human brain has a hundred billion neurons and synapses on the order of a hundred trillion or so. But it’s not just sheer numbers. It’s the incredibly complex and specific ways in which these things are wired up. That’s what makes it different from a gigantic sand dune, which might have a billion particles of sand, or from a galaxy. Our Milky Way, for example, contains a hundred billion suns, but the way these suns interact is very simple compared to the way neurons interact with each other.”
“Unless you believe in some magic substance attached to our brain that exudes consciousness, which certainly no scientist believes, then what matters is not the stuff the brain is made of, but the relationship of that stuff to each other. It’s the fact that you have these neurons and they interact in very complicated ways. In principle, if you could replicate that interaction, let’s say in silicon on a computer, you would get the same phenomena, including consciousness.“
Here I would say Koch goes out on quite a limb. All of the examples we have of conscious beings are made of essentially the same stuff: “neurons and bones and muscles,” in Koch’s words. He may very well be right that consciousness could arise out of silicon components wired together in the right way. Then again, maybe not.
Here a thought experiment may be relevant. See the XKCD cartoon “A Bunch of Rocks” in which a character with an infinite expanse of sand, an infinite supply of rocks, and infinite time to work in, uses the rocks as a Turing machine to build a complete simulation of the entire universe at the quantum level. Now, let’s forget about simulating the whole universe and imagine simulating “just” the operation of a single human brain. Could consciousness arise out of such a “brain?” After all, what you’ve got is really just a bunch of rocks lying in the sand, being arranged and rearranged according to some pattern. Does this collection of rocks begin to feel? Have experience?
It seems very far-fetched. Yet this bunch of rocks is formally equivalent to a computer. Why is it so much easier for us to imagine the computer becoming conscious?
Maybe it’s because the computer is wired up to devices like screens and speakers so it can visibly and audibly respond to us in ways that seem to us sort of person-like. We can ask it a question and get an answer in a form more or less like another person would give us. But this similarity to a person is an illusion – we just build computers that way so they’re easier for us to interact with. Inside, where the actual computation, data storage and retrieval, etc. is going on, it’s just electronic bits being turned on and off and shuffled around. They might just as well be a bunch of rocks being moved around in the sand.
Here’s the thing: In our rocks-in-the-sand simulation of a human brain, at any given moment, the rocks are in a configuration that represents some particular state of a brain. But this representation of a brain state is only one possible interpretation of that arrangement. The same configuration could equally well represent an infinite number of other things. What the arrangement actually “means” depends entirely on what the rock arranger (i.e. programmer) decided it should mean. It has no inherent meaning in and of itself.
Similarly, the state of a digital computer’s memory at any moment has meaning only according to how some program is designed to interpret it. The same pattern of bits could represent an image, some music, cells in a spreadsheet, or written text. It depends entirely on how the programmer decides to encode information for a particular purpose.
In our brains, on the other hand, specific states DO have inherent meaning. And that meaning is what constitutes our thoughts, our sensory experience, our emotions, our memories, the unconscious parts of our mind that influence our behavior, everything.
The difference is huge.