First, Eliezer seems to imply here that brains only work (to the extent that they can have beliefs capable of being acted on) when they work digitally, with at least some neurons having definite on or off states.
I assume you mean this section:
Your world does not split into exactly two new subprocesses on the exact occasion when you see “ABSORBED” or “TRANSMITTED” on the LCD screen of a photon sensor. We are constantly being superposed and decohered, all the time, sometimes along continuous dimensions—though brains are digital and involve whole neurons firing, and fire/not-fire would be an extremely decoherent state even of a single neuron… There would seem to be room for something unexpected to account for the Born statistics—a better understanding of the anthropic weight of observers, or a better understanding of the brain’s superpositions—without new fundamentals.
He’s not exactly saying that brains only work digitally—they don’t; neuron activation isn’t only about electrical impulses—he’s just talking about one particular process that happens in the brain. At least, as far as I can tell.
They certainly don’t work only digitally, but the suggestion seems to be that for most brain states at the level of “belief” it is required that at least some neurons have definite states, if only in the sense of “neuron A is firing at some definite analog value.”
I assume you mean this section:
He’s not exactly saying that brains only work digitally—they don’t; neuron activation isn’t only about electrical impulses—he’s just talking about one particular process that happens in the brain. At least, as far as I can tell.
They certainly don’t work only digitally, but the suggestion seems to be that for most brain states at the level of “belief” it is required that at least some neurons have definite states, if only in the sense of “neuron A is firing at some definite analog value.”