Re: Determinism

From: Eugen Leitl <>
Date: Fri, 16 Jan 2004 20:46:57 +0100

On Thu, Jan 15, 2004 at 10:25:06PM -0500, Doug Porpora wrote:

> Okay, well here you have just asserted that human beings can be
> defined as the sum of discrete quantum states and the like and that
> thoughts are therefore not infinite.

You make it sound like a fringe statement, the reverse is in fact the case.
Bekenstein's limit (and subsequent works based on that) is mainstream

In practice these limits are ridiculously high: physical systems like people
are physical systems at 300 K, not singularties, nor balls of quark-gluon

If you look at how tissue computes (ignoring more lunatic claims as QC), it's
even way below chemical bond dissotiation limit -- wet 300 K systems with
some 10 W/l energy density, for Pete's sake, and most of it is used for
homeostasis/autopoiesis, so it's unavailable for information processing.

I'd be really thankful if people used observed event rates and metabolic
rates, as well as input from (computational, and otherwise) neuroscience, as
well as robotics & Co when discussing cognition and biological infoprocessing
in general.

It would take a lot of slack and jaw out of the discussion.
> If what you suggest is true, then your conclusion would follow. I
> claimed, though, that the reductionist thesis is in trouble, and you
> asked why. There is a huge literature on emergentism and
> reductionism, but let me just stick with the so-called mind-body
> issue that Hal also alluded to.
> There have been two main reductionist strategies to deal with mental
> states, and they both -- to say the least -- have stalled. The two

I'm pretty sure it's in the eye of the observer. I would agree that
philosophy has been stalling for the last two kiloyears, when attempting to
explain cognition. Its bastard child science, not quite so.

> strategies are:
> 1. Eliminative materialism
> 2. Identity theory
> Eliminative materialism argued that human behavior could be explained
> scientifically without reference to the mental states of folk
> psychology. S-R behaviorism -- as in Skinner -- was the great effort

It's a complex enterprise; both requiring bottom-up and top-down. It appears
a bit premature to eliminate any approach for the benefit of another.

> here, and it is now largely judged a failure. We seem to need mental
> states to explain human -- and even lower animal -- behavior.

No offence, but I seem to detect a strong whiff of vis vitalis in the way you
use 'mental states'.

All biological systems process information, from the lowly procaryonte to
the primate. Chemotaxis as biased randomwalk through frequency of flagellar
inversions in concentration gradients is not magically different from
saltatoric spike propagation in myelinated neurons of mammals. Both are
relatively simple events, describable at molecular scale. Of course, there's
a lot of emergence going on, if you have some 10^12 components.

Mental states are just high-order emergent stuff. The only magic they have is
when you're introspecting, or trigger similiar representation systems when
emphatising during observation.
> So then there is the identity theory, the attempt to show that each
> mental state is identical to some (or finitely many) physical states.

Isomorphic computation doesn't exist? I guess this means the end of emulator
industry. Somebody better tell them, before their stocks plummet.

> Well, this has not panned out either. At worst, we may be in for
> some many-to-many correspondence between mental states and physical
> states, which spells doom for identity theory and reductionism.

I must say I can't follow you conclusion here. You're even not bothering to
dismiss the strawman very convincingly.
> I probably have not said enough to convince anyone here. This is a
> big issue, and much more could be said. I am just trying to
> summarize the current status of the mind-body debate. At the very

I'm not sure there is much difference between the "mind-body" debate and the
phlogiston, and the vis vitalis debate. You will remember, all of them were
associated with much gusto, and acrimony. Where are they now?

> least, the reductionist argument has been stalemated -- and there are
> good reasons, having to do with the role of language, for thinking it
> is false.
> Norman suggests the upshot for the Tegmark thesis:
> >"Thoughts" are therefore NOT
> >infinite because they can be conceptually defined in terms of particles and
> quantum states, and there are not an infinite number of these permutations.
> Well, only if reductionism succeeds. If reductionism fails, then,

Science is based on reductionism. The only way it can handle holism is
looking at emergent behaviour of pieces it has before dissected -- the old
Frankenstein thing. This is a surprisingly powerful mechanism, given modern
tools like full inspectability at molecular scale, and modeling. (Who'd
thunk, these computers are good for anything, after all).

> unlike universes, which, on my reading of Tegmark, are discrete and

Is the population of universes in the metaverse infinite, or just very, very
large? How can we tell the difference?

> countable, thoughts are not only infinite but uncountably infinite.

An 18th century poet could have said that.

> In that case, thoughts -- and persons -- comprise an even larger
> infinity than universes. And -- although this is another argument --

Apart from the fact that you're not providing any evidence for that,
have you ever considered the number of possible states in a system as complex
as the insides of a human noggin? If you look at how much trouble crypto
people are having when bruteforcing meek DES, there's a lot of them bits in
the space between our ears.

> at least a part of the universe would not behave deterministically.

How can you tell the difference between random, and (deterministic)
pseudorandom? The experimental answer: you can't, even for trivial sized

The question of free will and determinism is undecidedable for any being but
the omnipotent, omniscient (and even transcendent, given that you have to
violate the known laws of physics).
> If you tend to resist what I am suggesting, consider three things:
> 1. How do you even individuate thoughts so as to count them or
> correlate them with physical states? Is the belief that Mark Twain

Why would anyone be so foolish do that? It takes a lot of trouble to even
model single cells, save trivial-sized critters like a nematode.

> wrote Huckleberry Finn the same as or different from the belief that
> Samuel Clemens wrote Huckleberry Finn? Would that be one physical
> state you would seek to correlate with it or two? There are lots of
> well-discussed conceptual problems here.

Does a GNUchess system running on SPARC, MIPS or x86 care much on which
platform it runs? It still plays chess.
> 2. The mind-brain relation has sometimes been compared to the
> relation between software and hardware in computers. A certain

With very good reasons, because it's a good model.

> software function might be endlessly realizable by different physical
> (hardware) configurations in different computers. Similarly, I

See, you're agreeing.

> suppose, the same hardware configuration might realize different
> software functions in different computers. The analogy might break
> down, but this is the idea.

Layers are a human design artifact, evolved systems tend to not have very
well defined modularity.
> 3. The denial of reductionism does not necessarily entail belief in
> what is called "a ghost in the machine," i.e., a soul or other
> mystical something. The denial of reductionism may instead imply
> that not only is there no ghost, there also is no machine (i.e., we
> don't behave in machine-like ways). (This is a point made by Searle.)

Humans are always making analogies to well-known artifacts. Animals are
complex, so are some (very few, very large ones) machines. Searle is locking
in those aspects which the metaphor authors explicitly wanted to omit.

Machines are not at all like animals. (In many aspects, yet).
> John, I am not sure I understand everything you said. One thing I
> would say along lines I think you suggest: Determinism suggests a
> closed system. If you don't have a closed system, you don't get

What is a closed system? It depends on the scope. The universe is a closed
system, arguably. Yet, there's a lot of state in there.

> deterministic predictiveness. Human thought is both holistic and

You will observe that PRNGs are very deterministic, and utterly
unpredictable. They are explicitly made to be anything but unlinear, but you
of course can make a PRNG take input from the proverbial bicycle toppling in
Peking. In fact, even many current computers take pains to tap physical
noise, to effectively evade the known-inner-state attack.

> unclosable. Those features do not preclude mental causality, but
> they do preclude deterministic, causal laws.

The dichotomy between PRNG and RNG (determinism, and free will) only makes
sense if you're God.

For us meek end users, the difference is undetectable.
> Well, I hope I have not bored you all, but I do think that there are
> considerations from the social sciences that bear on -- and possibly
> challenge -- Tegmark's thesis.

-- Eugen* Leitl <a href="">leitl</a>
ICBM: 48.07078, 11.61144
8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE

Received on Fri Jan 16 2004 - 14:48:40 PST

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:09 PST