Now for the big question: Are you still conscious after the process has been completed? Furthermore, because the computer is completely responsible for the dynamics of your brain, you can forgo the physical artificial neurons and let the neuron-control programs connect with each other through software alone. Does the computer then become conscious? If you believe in consciousness, is your consciousness now in the computer, or perhaps in the software? The same question can be asked about souls, if you believe in them.
Here's the second thought experiment, which I cooked up in the mid-1990s. It addresses the same question from the opposite angle. Instead of changing the program running on the computer, it changes the design of the computer. First, imagine a marvelous technology: an array of flying laser scanners that can measure the trajectories of all the hailstones in a storm. The scanners send all the trajectory information to your computer via a wireless link. What would anyone do with this data? Read on.
As luck would have it, there's a wonderfully geeky store in this thought experiment called the Ultimate Computer Store, which sells a great many designs of computers. In fact, every possible computer design that has fewer than some really large number of logic gates is kept in stock. You arrive at the Ultimate Computer Store with a program in hand. A salesperson gives you a shopping cart, and you start trying out your program on various computers as you wander the aisles. Once in a while you're lucky, and the program you brought from home will run for a reasonable period of time without crashing on a computer. When that happens, you drop the computer in the shopping cart.
For a program, you could even use the hailstorm data. Recall that a computer program is nothing but a list of numbers; there must be some computers in the Ultimate Computer Store that will run it! The strange thing is that each time you find a computer that runs the hailstorm data as a program, the program does something different. After a while, you end up with a few million word processors, some amazing video games, and some tax preparation software—all the same program as it happens to run on different computer designs. This takes time; in the real world the universe probably wouldn't support conditions for life long enough for you to make a purchase. But this is a thought experiment, so don't be picky.
The rest is easy. Once your shopping cart is filled with a lot of computers that run the hailstorm data, settle down in the store's café. Set up the computer from the first thought experiment, the one that's running a copy of your brain. Now go through all your computers and compare what each one does with what the computer from the first experiment does. Do this until you find a computer that runs the hailstorm data as a program equivalent to your brain.
How do you know when you've found a match? There are endless options. For mathematical reasons, you can never be absolutely sure of what a big program does or if it will crash, but if you found a way to be satisfied with the software neuron replacements in the first thought experiment, you have already chosen your method to approximately evaluate a big program. Or you could find a computer in your cart that interprets the motion of the hailstorm over an arbitrary period of time as equivalent to the activity of the brain program over a period of time. That way, the dynamics of the hailstorm are matched to the brain program beyond just one moment in time.
After you've done all this, is the hailstorm now conscious? Might it have a soul? My point is that you can neither reason nor design experiments to study core beliefs about the self or consciousness. There is no marker of consciousness but consciousness itself (if it exists), so if you try to define an experiment, you have to arbitrarily decide whether your object of study is a computer on your desk, some tissue in your head, or a hailstorm.
Dan Dennett and I have argued about this idea over the years. He takes the view that consciousness is a cognitive illusion and a nuisance; jettison consciousness and it's easier to design an airtight philosophy. What Dan sees as a road to clarity, though, seems to me to be taking the easy way out. He worries that ideas about consciousness and other matters of faith pave the way to weird and dangerous superstitions. While I share that concern, I take a more accepting view. Consciousness is a loose end, and it takes discipline to live with loose ends. Dispelling the idea of consciousness (or religion) doesn't necessarily dispel what Dan fears are divisive beliefs, like dreams of everlasting life. Some of my nonreligious computer science colleagues, like Ray Kurzweil, hope to cheat death by backing themselves up on a computer, along the lines described in the first thought experiment.
In the end, a sense of humor is essential in these matters. Dan Dennett still thinks I'm wrong and that consciousness is an illusion. I've accused him, with tongue in cheek, of being one of those unusual people who just happen not to have consciousness.
Previous Jaron's World columns: