[ExI] The Problem of Mental Causation

Sun May 4 16:44:48 UTC 2025

On Sun, May 4, 2025, 11:30 AM William Flynn Wallace via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

> All processing of information starts in the unconscious (obvious if you
> follow the stimuli through the brain centers before it reaches the
> forebrain).  Thus, thoughts and actions come from the unconscious.
>

Another interpretation is that the computations involved in realizing a
conscious state may be distributed across time and space. So even if brain
activity precedes a conscious thought in time, it may still be part of the
conscious state.

Consider a function on a computer that multiplies two million digit
numbers. The conscious state may be viewed as the function evaluation
itself. Though the physical realization of this function by the computer's
processor is spread out over time through many individual operations
processing 64 bits at a time. An alien looking at this system might say all
the antecedent computation by the processor is unconscious, when really it
is just all part of one larger function.

Now if the function is conscious we would have the same picture. And we
might ask, why certain neural activity seems conscious while other neural
activity does not, but this is an illusion created by the fact that one
large computations can be spread out across spacetime (a large computations
never occurs at a single spacetime point).

Another interpretation is that unconscious thoughts are independently
conscious. The fact that they inhabit the same skull is not enough
justification to logically imply that the consciousness you identify
yourself would necessarily be aware of other conscious parts. We see
evidence of this in split brains, where two conscious minds inhabit the
same skull.

One can also consider something like the China brain thought experiment,
where individual conscious humans act as neurons in one larger brain. The
larger brain doesn't and can't be aware of the individual conscious parts
that compose it.

  The conscious gets a veto power, (if it is very quick - if it isn't, we
> do and say things we regret), but that's about all.  I assume this is
> inconsistent with how an AI operates.
>

I think there's actually a surprising overlap between how humans seem to
think and how these LLMs seem to think.

> Have you read "Thinking Fast and Slow"???  Why not?  bill w
>

I haven't but it sounds quite good!

Jason

>
> On Sun, May 4, 2025 at 8:40 AM Jason Resch via extropy-chat <
> extropy-chat at lists.extropy.org> wrote:
>
>> Stuart,
>>
>> I was hoping you would have something deep and insightful to add, you
>> don't disappoint!
>>
>> On Sat, May 3, 2025, 3:09 PM Stuart LaForge via extropy-chat <
>> extropy-chat at lists.extropy.org> wrote:
>>
>>> On 2025-04-30 10:17, Jason Resch via extropy-chat wrote:
>>> > One of the great puzzles when it comes to understanding consciousness
>>> > and its role in the universe is the question of how conscious thoughts
>>> > could have any causal power in a universe fully governed and
>>> > determined by particles blindly following forces of nature.
>>>
>>> Thinking and information processing, conscious or otherwise, has casual
>>> power through its information content. This is a direct application of
>>> the Laundauer's principle. Mental causation is exactly how Maxwell's
>>> Demon works. It uses it knowledge of the positions and momentum of all
>>> the individual particles of gas to create a temperature gradient.
>>> Maxwell's Demon seems to violate the 2nd law of thermodynamics by
>>> decreasing the entropy of the gas. But, this is not the case, because in
>>> the process of memorizing the positions and momenta of every particle in
>>> the gas and enabling it to increase the system's potential energy, the
>>> Demon increased the entropy or information content of its own brain or
>>> data storage. This could only have been done by erasing whatever
>>> information was there before and incurring some minimal energy cost
>>> given by the Landauer principle E >= k * T * ln2 with k being the
>>> Boltzmann constant and T being Kelven temperature.
>>>
>>
>> To be clear, are you equating the causal potency of information, with
>> it's necessary generation/storage always incurring a cost of increasing
>> entropy elsewhere? Or is this just one example of how information (or it's
>> processing) can have physical effects?
>>
>> The process that analyzes an approaching gas molecule, judging it's
>> temperature and trajectory, and ultimately deciding whether to open or
>> close the door could itself be viewed as a kind of primitively aware
>> (conscious) thing. It's discriminated high-level information state then
>> occupy a spot in the causal chain, without it, the door could not respond
>> intelligently to it's environment. And I would say the discriminated
>> high-level information state is its conscious state.
>>
>>
>>
>>> Basically the causal power of wanting ice cream is the energy cost it
>>> takes to forget you want ice cream either by distracting yourself or by
>>> getting yourself the ice cream.
>>>
>>
>> Would this mean a conscious mind running on a reversible computer (which
>> escapes Landauer's principle) could have (or allay) no desires?
>>
>>
>>> >
>>> > Some solve this problem by supposing our will must somehow
>>> > miraculously intervene in physics. Others solve this problem by
>>> > denying human will or agency, relegating consciousness to an
>>> > ineffectual, and inessential "epiphenomenon."
>>>
>>> Our will directly intervenes in physics, not miraculously, but through
>>> the mathematical connection between information, entropy, and energy. We
>>> are  giant Maxwell's demons made possible by smaller Maxwell's demons
>>> called cells, which are made possible by smaller Maxwell's demons called
>>> mitochondria, and so forth all the way down.
>>>
>>
>> I agree there are many levels of organization all of which can possess
>> their own upwards and downwards causal forces.
>>
>>
>>> > But I think a more nuanced view can show that consciousness can have
>>> > causal power in a universe fully determined by particles following
>>> > physical equations. Here is my attempt at describing such a view:
>>>
>>> >
>>> https://drive.google.com/file/d/1qCuCc9kvbw5KKHJ223l7MbbNhZkTamhA/view?usp=sharing
>>> > An answer to this question is relevant to whether uploaded minds, AI,
>>> > or robots can have will or agency, despite their behavior being fully
>>> > determined by low-level machine code.
>>>
>>> Yes, but while consciousness or agency are examples of downward
>>> causation or what you call strange loops, these causal loops are far
>>> more general of a phenomenon than consciousness or agency.
>>
>>
>> I agree with this. Consciousness is by no means unique in expressing
>> downwards causation. I give the example of a nerve cell ordering
>> neurotransmitter molecules around.
>>
>>
>> For example
>>> subatomic particles give rise to atoms in a standard upward causation,
>>> but atoms also give rise to subatomic particles through radioactive
>>> decay which is downward causation. Another example would be the surface
>>> tension of a water droplet ordering the water molecules into a perfect
>>> sphere. Strange loops are not magic, they are physics incorporating
>>> information theory.
>>>
>>
>> Sperry gave the example of a tire rolling down a hill. The tire is made
>> of molecules, but the rolling of the tire largely guides the motions of all
>> the atoms in the tire.
>>
>> I guess the question then becomes what kinds of information processing
>> activities are conscious ones.
>>
>> You make the comparison to erasing or overwriting information, but is any
>> process of recording information conscious? And what of processing
>> information without overwriting or erasing? Are such processes not
>> conscious? I think the dividing line for consciousness may be something
>> other than entropy increasing operations. I agree that information
>> processing, consciousness, and entropy are all closely related, but are
>> they equal?
>>
>> Jason
>> _______________________________________________
>> extropy-chat mailing list
>> extropy-chat at lists.extropy.org
>> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>>
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20250504/265a7f8b/attachment.htm>