On the other hand, if we could get ahold of a human with a perfect memory, we could easily establish (at least on the basis of orders of magnitude) how much con

On the other hand, if we could get ahold of a human with a perfect memory, we could easily establish (at least on the basis of orders of magnitude) how much concious-level information can be stored in the brain. This could grant tremendous insight into the functioning of our brains, as it would also give us a good idea of the "size" (in terms of neural connections, for example) of memories, thoughts and so forth as they are stored. We already have information-theoretic bounds on this sort of thing based on a physical understanding of the number of neurons and synapses in a person's skull, but they're very loose. It's like trying to estimate the storage available in a chip from the number of transistors on the die. What kind of RAM does it contain? How much is support circuitry? Does it contain RAM at all, or is it something more like a dataflow CPU? If we can fit a high-entropy sequence of N distinct digits into a brain composed of M neurons we might presume that storing each digit involves M/N neurons. If N is close to or greater than M, we might instead guess that data must be stored in-flight, which leads us down different avenues of investigation. But this is all really a thought experiment, anyway...