Straws, Camels, and continuous discretion

O'Regan, Emlyn (Emlyn.ORegan@actew.com.au)
Wed, 23 Jun 1999 15:53:08 +1000

>From Harvey:

> Take the example of the straw that broke the camel's back. You can keep
> loading a camel up with straw. You can add one straw at a time. I can't
> predict when the weight will be too much for the camel. But you can't put
> an infinite amount of weight on top of the camel before it collapses.
>
Point absolutely taken, and this reminds me of some problems that I had with Chalmers' arguments about dancing qualia (slowing swapping components in a structurally identical brain, which disagree about blue and red).

A similar example would be that you could take a brain, and remove a neuron at a time. Does removing one neuron kill the brain/extinguish consciousness? No. Then a brain with no neurons, by induction, is alive+conscious, if the original was. Silly stuff.

I think that the problem lies in the discrete nature of the arguments - consciousness is on or off. I would argue that there is a continuum, between conscious and not conscious, which is more believable. Then the straws & camels problem is dealt with, and a brain with no neurons can finally be pronounced dead.

Also from Harvey:

	The only acceptable upload I have heard involves replacing my
	biological neurons one at a time with mechanical ones.  Eventually,
I would
	be in the mechanical robot and would feel it was me.

Or maybe "you" would slowly, inexorably slip away during the process, being replaced by a new copy of "you" (by "you" I mean the conscious you, the bit that means it when it says "I"). I am not convinced by this gradual replacement argument that the conscious continuation of self would be achieved. I am sure that the final self would report "Yes, it's me, it worked", but would the old version of your self have experienced a devestating demise? In tumbolia, no-one can here you scream...

---

I feel intuitively that causality is actually the basis of consciousness,
that mere patterns are not conscious. The problem is that it seems to be
very difficult to define the difference between a dynamic (causal) system
and a ruddy great lookup table. Where is causality?

At its basis, isn't it about fundamental particles/waves/energy/other messy
tiny stuff interacting? When one neuron fires and that sends a signal to the
next neuron, there are actually a bunch of subatomic things happening, one
sparking the next sparking the next.

Except as you get closer, the causal quality of these interactions retreats
before you.

Even if you can chain together causality at that level, and build up from
there to show it at the level of cells/silicon/email, the isomorphisms are
incredibly different and complex, between a chain of causality in a neuronal
brain and one in a nanorobot copy of that brain, even more so in a software
sim of the same brain. If you are prepared to go that far with isomorphisms,
you'll probably find more isomorphisms between "causal" and "non-causal"
intelligences which are no more complex. Is there a decent distinction?

My apologies if this stuff has been covered before, but I think it is more
interesting than g*ns...

Emlyn (not the copy)