Friday, July 11, 2014

Consciousness Machine: Refined

Sometimes I call it a phrencell, a little room of the mind. (Mixing Greek and Latin because trolling.)

I'm not certain I'm correct. I'm merely certain that I haven't been rebutted, yet, but the point isn't necessarily to be right. The point is to show that the box has an outside to go look around; the point is to show what kind of thing Decartes' pineal gland will look like in physical implementation. For example, even if I'm wrong, it's clear enough that physics doesn't have to be causally closed, and that dualism is compatible with naturalism.

One example of the machine is this:

Take an FPGA. Hook its output into its configuration register, so when it runs it re-programs itself. To its input, hook a true-random source, so it re-programs itself randomly. Ideally the FPGA will have infinite gates so it can, in principle, be programmed with any program and thus emit arbitrary bitstreams.

I imagine a pentagon of five anodes and an electron in a superposition of being about to be absorbed by all of them. If we're feeling mystic (I am) the pentagon is an oracle and the FPGA is its interpreter.  It probably helps but probably isn't necessary to let the FPGA bias the oracle, for example tuning the voltages on the anodes.

I assert that the probability of the FPGA at any present time is in a state of both 100% and 0% probability. The previous input had a 0% probability, all past states had a 0% probability, but of course it's right there. Look at it. The present has a 100% probability.

Future evolution has a well-defined probability. For example, with the pentagon, the FPGA will have five possible evolutions with average 20% probability. We can just look at it and compute the probability, entirely straightforward. Except that when one of those states becomes the present, it too will have a 0% probability, because it evolved from a state with 0% probability.

The reason it has a 0% probability is because at infinite time, the (ideal) FPGA has an infinite number of possible states, which means at most infinitesimal probability for any one state, which is physically identical to zero. Which is to say it can't happen, it's impossible. Normally we can ignore things at infinity for physical purposes; parallel lines cross at infinity, so what. All particles decay by infinity - that's just physics speak for 'never.' No observer will ever observe the decay of certain particles. (Neutrons and electrons?)

However, the FPGA is, at any time, in a state bit-for-bit identical to an impossible state, because states are degenerate in time. As long as the configuration register is properly set up, the FPGA is allowed to have constant states, cyclical states, random states or chaotic states. This means the FPGA cannot know if it reached its current state by passing through infinity or not. (E.g: a constant state that might have not been constant, but happened not to be.) Which means we, as the observers, cannot know whether it passed through infinity or not. Which means everything that is true at infinity must be true now, at the present. Which means its past and current states have a probability indistinguishable from zero.

But it's still right there. Paradox.

My solution to the paradox is to say that the FPGA's future states in fact never had a probability. Quantum events look like probability because they're a special case of non-probability. Namely, if the canvas of possibilities is finite, then no matter what you do you'll converge on some probability distribution. (This is why consciousness' effects on physics are about information - bitstreams have few inherent limitations.)

Specifically, quantum objects have a quantum of consciousness and free will (with respect to physics). As it turns out, if you have 6*10^24 persons with no memory and only two options, the variations in personality and whim will cancel out and converge on some well-defined probability. If you have 6*10^24 randomly selected persons, there's guaranteed to be a representative sample of all personalities, which means you're measuring the average personality, and will do so reliably.

However, if you hook only one personality up to a randomness amplifier, none of this holds, and the true non-deterministic, non-stochastic nature of the beast emerges. (Specifically, the phrencell deliberately breaks the independence clause of independent random events.) This is why the interpreter may need to be able to bias the oracle. The oracle needs to be able to feel the circuit or it won't be able to tell what it's doing and thus shape the interpreter to its will.

No comments: