Consciousness ~=~ self-referentiality' (was Re: Consciousness, New Thinking About

Matt Jones jonesmat at physiology.wisc.edu
Tue Jun 4 10:11:45 EST 2002


Jim Chinnis <jchinnis at alum.mit.edu> wrote in message news:<61dlfu4464l1vldm92n9cjgucvd0op62he at 4ax.com>...
> jonesmat at physiology.wisc.edu (Matt Jones) wrote in part:


> There is the possibility, as some have suggested, that consciousness is one of
> the primitives of the universe, like matter-energy. That's very appealing in
> some ways, and intellectually frustrating in others.

I agree. Appealing and frustrating. It seems impossible to consider
this idea only part-way. If consciousness is a primitive, then like
matter/energy it must pervade everything, by definition. This means
quarks are "made of" consciousness. At first blush, many people would
reject that as utter nonsense. But what's a quark? Essentially, it's a
particular solution to an equation. That is, you can't hold one up and
point to it and say, "Here's a quark!".  Instead, you do an experiment
in which you most definitely do not directly observe a quark, and then
you show that the result is consistent with the solution of an
equation that contains quarks as one of the mathematical elements. So
that's a bit like a quark being "made of" consciousness.


> Most neuroscientists probably subscribe to the different view that
> consciousness is an emergent property of certain processes. If that's true, it
> didn't exist back when there were only rocks.


Yeah, I know. But how many of those neuroscientists can provide a
precise (and non-trivial) definition of "emergent property"? Further.
how many can give a concrete example of an emergent property?

The usual definition is something like: "An emergent property of a
system is a property that is more than the sum of its parts." (This
was one of the ways that John Stewart Mill originally defined it). 
Another definition is that one cannot predict an emergent property
based on knowledge of the properties of the individual components
(also used by Mill). With all due respect to Mill, these definitions
easily lead to trivial results.

Consider a rock (again) made of a particular substance, calcium
carbonate, say. There's clearly no way to predict the shape of that
rock from the properties of calcium carbonate. In that sense, "shape"
would be an emergent property. On the other hand, one could measure
the size of calcium carbonate molecules, count them and show that the
shape of the rock is indeed due to the exact spatial sum of the
components. So applying these two (very common) definitions of
emergence, one must conclude that shape is simultaneously an emergent
property and a non-emergent property.

Like I said, the example is trivial, but I think it points out the
problem that these so-called "emergent properties" are not
well-defined.

People often talk about emergence in the context of neural network
behavior, for example, the doistributed representation of information
in a net is often called an emergent property. But nobody would deny
that the representation is 100% deterministic (even in the presence of
noise, because if you use the same random seed you get the same
representation every time). Furthermore, after the thing has
converged, you can go back and examine all the weight changes, and do
a sort of "reverse-prediction" of exactly why the representation took
the form it did.

Anyway, I think the notion of emergent properties is just a
hand-waving fancy word for "really complicated".



> I'll go one step further. I also question the sharp boundaries we draw
> between, say, one person and another or between a person and the air in their
> lungs. We form "objects" cognitively, but we know that those objects are not
> truly physically or functionally distinct. Hmmm..."I" have consciousness or
> *am* consciousness, but what about my foot?
> 
> Just to add to our common pool of ignorance. ;-)


Thanks! I was afraid that pool was running low. I'm always happy to
pitch in too.


Cheers,

Matt




More information about the Neur-sci mailing list