2nd Instalment (continuation) - scroll down if you have read 1st
instalment
Trans. Kim Jones (extract only)
1.1 Mechanist Philosophies
1.1.1 Different types of Mechanism
I distinguish the following mechanist hypotheses:
BEHAVIOURIST MECHANISM
Some machines can behave as thinking beings (living, conscious etc.)
(BEH-MEC)
STRONG MECHANISM
Some machines can think (living beings, conscious beings, have a
private life etc.) (STR-MEC)
INDEXICAL MECHANISM
I am a machine (or - you are a machine, or again - human beings are
machines) (IND-MEC)
By replacing "machine" by "digital machine" one obtains the
corresponding digital theses.
The behaviourist digital mechanism BEH-DIG-MEC corresponds largely to
that of Turing in his 1950 article. In the same way, the strong
digital mechanism STR-DIG-MEC corresponds to what is called in the
literature the strong artificial intelligence thesis (strong AI).
In this work I am exclusively interested in indexical and digital
mechanism (IND-DIG-MEC or just IDM). "Digitality" necessitates
Church's Thesis, which is why the digital aspect is explained in its
turn in the second part. There, I will show how a procedure, due
essentially to Goedel, permits an indexical treatment of machines in
general.
Proposition:
IND-MEC => STR-MEC => BEH-MEC, and
BEH-MEC ≠> STR-MEC ≠> IND-MEC.
(with or without the hypothesis of digitality)
Reasoning: One admits that humans know how to think (conscious
beings, having private lives etc.) In this case IND-MEC entails STR-
MEC and STR-MEC entails BEH-MEC. That BEH-MEC does not entail STR-MEC
is supported by Weizenbaum (1976) (see also Gunderson {footnote 1}
1971). STR-MEC does not entail IND-MEC, since the fact that machines
are able to think does not entail that they alone are able to think.
It is conceivable that machines are able to think without we ourselves
being machines. Wang (1974) presents a similar reasoning.
Nevertheless, numerous philosophers make implicit use of an opposing
opinion: STR-MEC => IND-MEC, see for example Arsac 1987.
{Footnote 1: Gunderson 1971 criticises the Turing Test. The Turing
Test is a test for BEH-MEC. Simply put, a machine (hidden) passes the
test if it is able to pass itself off as a human being during a
"conversation" by means of a computer keyboard terminal.}
1.1.2 Mechanist Philosophy: Historical Summary
Contemporary digital mechanist philosophy is due in large measure to
Descartes and Hobbes {footnote 2} (see Rogow 1986, Bernhardt 1989).
Descartes wanted to distinguish Man from the animals. He argues that
animals, as much as Man's body (including the brain), are/is a
machine. He intended by this a finite assembly of material components
that unequivocally determines the behaviour of the whole. Descartes
surmises that the soul is not mechanical. In separating the soul from
the body in this way, and thus the mind from matter, he is the
originator of the dualist position, widely encompassed by the
philosophy of mind. One speaks of Cartesian Dualism.
Following are three arguments that Descartes presented in favour of
his distinction of Man from the animal-as-machine (note that this
distinction entails the negation of IND-MEC.)
{Footnote 2: One can detect some mechanist affirmations or questions
among (pre and post-Socratic, though not necessarily materialist)
philosophers, from Greek antiquity (cf Timaeus and Plato, see also
Odifreddi 1989). Among Chinese philosophers, for example Lao-Tzu, a
certain monk is admired for having passed off his "automated" servants
as flesh and blood beings. Among Hindu philosophers for example, in
the "Questions to the King Milinda", the human body is compared to the
chariot, and the human mind is compared to the different parts of the
chariot, similar to Hume's (1739) manner of tackling the problem of
identity with his boat. The temptation to set up artefacts in the
image of Man is also a component of several myths, (for ex. the Golem
in Jewish culture, see for ex. Breton 1990). It is no exaggeration to
maintain that the very idea of mechanism appears wherever and whenever
machines themselves are developed.}
1) Animals are not endowed with reason and cannot engage in linguistic
communication
This argument is losing credibility since language and reason seem
more accessible to today's machines than for example, emotion which is
communally allowed in the case of certain animals (see for ex. Lévy
1987). Here Descartes takes Aristotle's position which asserts that
Man is a "reasoning animal".
2) Machines are finite beings. A finite being cannot conceive of the
infinite. Now, I am able (said Descartes) to conceive of the infinite.
Thus I am not a machine.
This argument against IND-MEC brings into relief two fundamental
questions:
a) Can Man conceive of infinity?
b) Can a machine conceive of infinity?
Question a) differentiates Hobbes' point of view from Descartes'.
Hobbes concludes that he cannot in effect conceive of infinity.
3) A machine can only carry out particular tasks, as it turns out,
those tasks for which it was constructed. In effect, Descartes is
saying:
"Since, in the case that reason is a universal instrument that
participates in every sort of encounter, these organs need a certain
particular disposition for each and every action; from this comes the
idea that it is morally impossible that a machine might possess
sufficient diversity such that it might act in every living occurrence
in the same way that our reason assists our actions (Descartes,
"1953", page 165).
The idea of a universal machine had nevertheless crossed the mind of
Raymond Lulle (1302) whom Descartes studied. This same idea will
reappear with Leibnitz, culminating in the work of Turing, and this
will be explained in the second part.
La Mettrie will rehash Descartes' animal-as-machine for the purpose of
extending it to Man (La Mettrie 1748, see also Gunderson 1971).
In parallel with Descartes, Hobbes himself develops the mechanist
hypothesis (Rogow 1986). One can date Hobbes' motivation toward
mechanism from the time of his discovery of geometry. Having been
particularly impressed by the fact that he may have been convinced by
a *finite communication* based on logical geometrical reasoning,
Hobbes conceives of the mechanistic character of thought. He then
reasons that it should be possible to reduce thinking to addition and
subtraction. (see Webb 1980). He is thus very close to the
*functionalist* position in the philosophy of mind: that the additions
and multiplications might be realisable by a *telegraphic network* , a
*hydraulic system*, an *electromagnetic device* , or even *a
windmill*, a *catapult* or a *calculating device* (ordinateur), to
cite Searle's enumeration (Searle 1984). Thought is thereby reduced to
operations not necessarily equipment-dependent, and to the constituent
matter employed to realise these operations. La Mettrie, in his own
way argues in something like the same sense:
"Thus a Soul of mud, discovering in the twinkling of an eye the
relations and the consequences of an infinity of ideas difficult to
conceive, would be preferable evidently to an ignorant and stupid
Soul, which might be made of all the more precious Elements" (La
Mettrie 1748).
Similarly, Lafitte engages us on the subject of Babbage, precursor of
19th century information processing, to which we will return in the
second part:
"For Babbage, all machines being a composition of different organs
linked together in a complex manner, the important thing to fix is
less the very form of the organs than the sequencing of their
functions, which relates to organic linkages causing the ensemble to
function." (Lafitte 1930).
Differing with Descartes, Hobbes concludes that it is not possible
that Man - whom he considers as a finite being - might conceive of the
infinite. Hobbes' motivation, being finitist and indexical (human
thought is mechanisable) is therefore opposed to Descartes' animal-as-
machine and is, in this sense much closer to the contemporary
motivation in the direction of artificial intelligence. Soon I will
return to the relation existing between mechanism and functionalism.
1.1.3 What is a machine?
Given the familiar connotations of the word "machine" - locomotives,
electric kettles, automobiles, computers, microscopes, dish-washers,
sewing machines, rice-cookers, time-pieces - (the concept of)
mechanism may well seem grotesque.
Even if machines are considered to be artefacts of exclusively human
construction, in other words artificial, the concept of the machine is
difficult to define. Lafitte, in 1911 argues that just such a
definition can only be made in vain:
"To claim to be able to define the concept of a machine is to suppose
that the science of machines has come about, or that it might one day
come about in all it's perfection. Other than what amounts,
chimerically-speaking, to assigning limits to the development of
mechanical forms, it really is to suppose in the first place an entire
and complete knowledge of the character of every individual present
and future mechanism, followed by the perfection of a measuring
instrument capable of situating each into a definitive category
according to the ensemble of characteristics. But, this again
implicitly admits to a massive division of sorts, conforming entirely
to those contours that we can cleanly envisage, having no link
whatsoever with other bodies." (see also further on 2.3)
Similarly, La Mettrie, in "Man as Machine" writes:
"Man is a Machine composed in such a way that it is frankly impossible
to initially get a clear idea of it and consequently to arrive at a
definition"
What Hobbes and Descartes have in common is that a machine is a
locally finite being. Its global behaviour is determined by the
behaviour of its elementary constituents, these being finite in number
at each instant (call this the "digital aspect"). The number of
components can nonetheless grow according to the work performed by the
machine.
A philosophy called "Mechanical Philosophy" developed not only under
influence of Newton's works but also under those of Boyle (see
Broukère 1982), was more materialist and determinist (in the vague
contemporary sense) than finitist. Action at a distance which seems to
exist between material bodies in Newton's mechanics which in addition
worried Hobbes, seemed to exclude any finite component (see
Metaxopoulos 1986). Similarly, Searle - who I do not classify as a
mechanist philosopher - argues that a machine alone such as (according
to him) the human brain, is capable of thought, but he is unclear as
to the nature of the identifying factors. The difficulty inherent in
defining machines is reflected in the difficult task of circumscribing
mechanist philosophy.
Gandy (1980) isolated 4 principles capturing the intuitive idea of a
machine (digital or effectively digitisable):
1) Determining Principle: a machine may be described by the givens
within a hierachical structure S, equally by a function F such that
iteration F(S), F(FSS)), F(F(F(S))),....specifies the evolution of
that machine.
2) Limiting Hierachical Principle: a structure S is hierarchised via a
finite number of levels.
3) Principle of Unique Reassembly: a structure S can be disassembled
into limited parts and reconstituted in a unique way according to
those parts, following a guide or plan (which itself must be finite).
4) Locality Principle: the hierarchical structure of a machine admits
a topological description such that the state of a part of the machine
Fn(S) stemming from its evolution only depends on the state of parts
in the immediate vicinity in Fn-1(S) (no "action at a distance").
Gandy and Shepherdson (1988) give a much more precise formulation in
terms of groups deemed finite by heredity. Gandy demonstrates (and
Shepherdson generalises) the equivalence between this definition and
Turing's conceptual definition (with oracle), so that a generalised
version of Church's Thesis will permit (in the second part) that we
can do without Gandy's explicit principles. I have nonetheless laid
bare these principles informally since they fall back on Descartes'
and Hobbes' conceptions and illuminate the biological pretext for
(digital) mechanism, as well as certain doubts arising from chemistry.
1.1.4 Biological Incentives
These concern indexical mechanism directly since they stem from self-
observation, unlike incentives arising from the contemplation of
engineering work (see for ex. Lafitte 1930, Vèsale 1543, Ambroselli
1987).
Wanting to show that animals are machines, Descartes runs up against a
problem that he can never resolve.
1* Descartes'problem
How could an animal-as-machine be capable of reproduction? How could a
machine construct a self-same and identical machine? It would have to
contain a complete description or plan of itself, and this seems
impossible. {footnote 3}. The fact that cellular division is, broadly
speaking, understood to take place at the molecular level (Cairns,
Watson and Crick, (see Watson (1965, 1989) for detailed references)
constitutes a contemporary biological incentive for mechanism.
(Footnote 3: This argument is used from time to time, as in Cossa in
1955: "Were machines to find themselves - via an impossible form of
reasoning - endowed with reproductive power, these irregularities,
these defects - although tiny at the start and without functional
consequences - would undergo an amplification from generation to
generation such that the machine would rapidly cease to work at all."
2* Driesch's problem
Similarly, when Driesch conducted his first experiments in embryology,
cellular division seemed to him at this point to be a surprising
phenomenon which he would go on to use as a decisive argument in
favour of vitalism, arriving at a counter-mechanist {footnote 4}
conception of life {footnote 5}.
(Footnote 4: The difference between the non-mechanist and anti-
mechanist view is identical to that of the agnostic to the atheist.)
(Footnote 5: By way of reaction, Helmholtz proposes a mechanist theory
of perception - something we will in a certain sense rediscover - and
opts for an anti-vitalist pact ( de Broukère 1982)
By the same token, how might we explain the flexibility exhibited by
cells in the analogous phenomenon of cellular regeneration? Think for
a moment of the cloning of a frog (see Watson et al 1989 p. 835) or
yet, of the stupefying example of the planarian, small flatworm
roughly 1cm in length which lives in certain freshwater habitats (see
the drawing, following page) which qualifies as a kind of champion of
tissue regeneration among animals possessing a central nervous system
(Buchsbaum 1938, Buchsbaum et al. 1987). Phenomena of genetic
regulation with regard to mechanism are eloquent [elegant?=poss.
error:] Kim) (Jacob and Monod 1961, Thomas 1978, Thomas and van Ham
1974).
Here again is what Diderot said in his conversation with d'Alembert,
in confronting Cartesian mechanism and the development of the embryo:
"Do you see this egg? With this, one can upend every school of
theology and every temple on Earth. What is this egg? Before the germ
(of life) is introduced, no more than an insensate mass; and after
it's introduction, what is it then? An insensate mass, since the germ
itself is but an inert and coarse fluid. How might this mass progress
to another form of organisation, toward the sensation of feeling,
toward life itself? By warmth. What produces this warmth? Movement.
What will be the successive effects of movements? Instead of my
response, sit here and together we will follow these movements from
moment to moment. Starting with a point that oscillates, a thread that
extends and gathers colour, to the flesh which forms; a beak, tiny
wing-ends, eyes, feet appear; a yellow-tinged matter that divides and
which produces intestines; behold an animal. An animal that moves,
becomes agitated, sounds its voice; I hear its squawking through the
shell; it grows its downy coat; it sees. The weight of its head, which
bobs back and forth, unceasingly brings its beak against the inner
rampart of its prison; this now breaks; it leaves, it walks, it flies,
it registers irritation, it flees, it returns, it complains, it
suffers, it loves, it desires, it experiences joy; it possesses each
of your affects; all of your actions, it can perform them all. Can you
claim, with Descartes that it is no more than a purely imitative
machine? In that case, tiny children laugh at you with derision and
the philosophers' rejoinder is that if such is a machine, then you are
but another." {footnote 6}
(Footnote 6: We note here the essentially modernist mindset of Diderot
who places the animal on the same rung as the human, thus rejecting
Descartes' distinction. In general, with the notable exception of La
Mettrie, mechanism will face a poor reception. This brings to mind
Pascal's argument. This genre of "argument" is not all that far from
what Turing called "head in the sand objection" qualifying more as
"consolation" than refutation. (Turing 1950)
The contemporary biologist may surmise that - relative to the laws of
chemistry - the problem of biological reproduction is solved. The
discovery by biochemists and molecular geneticists of the plan or
description of the cell and the fashion by which this map is
chemically represented, decoded and executed within the organism
constitutes cause for the application of the Principle of Unique
Reassembly, the Determining Principle and the Limiting Hierachical
Principle (this last appearing already with classical genetics, see
Cuny 1969). In the same way, the older discovery of the importance of
particle exchanges with the surrounding environment or between
organisms - as happens during breathing, during digestion, during
conception, favours the application of the Locality Principle (Van
Helmont, Mendel, Lavoisier, Vesale - to cite the more well known ones;
see de Broukère 1982, Ambroselli et al 1987, Vesale 1543).
1.1.5 Doubts Arising from Chemistry
Watson has said "the cell obeys the laws of chemistry", and the
preceding incentives perhaps justify a belief in indexical mechanism
relative to those laws. If these laws prove themselves to be non-
mechanisable, mechanism will thereby find itself weakened, perhaps
even refuted but certainly relativised.
This suggestion is all the more well-founded in that the laws of
chemistry are captured by quantum mechanics. Despite its name
("mechanics" is here used in the Newtonian sense), philosophers and
theologians are attracted to QM and see in the factual descriptions
(up to here confirmed) of this theory an empirical justification of
the non-mechanist nature of the world and/or of consciousness.
{footnote 7}
(Footnote 7: Letovski 1987 takes up a (too?) rare encounter between
cognitivists open to computational approaches to consciousness and
neuroscientists open to the use of QM to resolve the brain/mind gap.)
Anti-mechanist arguments founded on QM are various. We shall briefly
examine several:
a) The oldest argument: QM provides evidence of an intrinsic
indeterminism in the world (or more precisely concerning the relations
between the observer and the world. Mechanism is determinist. Thus,
our relation to the world is not mechanist.
Those who use this argument are tempted to "explain" such a free
interpretation by means of this indeterminism. This argument has
already been refuted by Carnap or Mackay or Schroedinger. In addition,
I will show that mechanism *is not* determinist.
a) The most recent argument: QM makes possible very particular forms
of material, for example the quasi-crystals of Penrose and Schectman
(see Penrose 1989). Penrose suggests, though without any seeming
conviction, that the brain could be a *sort* of quasi-crystal.
Similarly, Margenau 1984 and Squires 1990 seek to utilise QM to
develop a dualist and non-mechanist theory of the mind (see also Stapp
1993).
The following arguments merit close and detailed examination since the
(indexical) mechanist hypothesis considerably clarifies them. To this
end, I will make use of a bare minimum of assumed quantum mechanical
knowledge to allow the reader to follow the argument.
Newton conceived of matter and light as constituted of particles
interacting with one another. Huygens was more prepared to reserve
this way of seeing things for matter alone. He develops a successful
wave theory of light which takes account of a number of luminous
phenomena. Einstein will provide evidence, in his work on the photo-
electric effect, of the corpuscular aspect of light, without
dethroning the wave theory in the process. He also arrives at the
quantum theory of light. De Broglie extends the wave-particle aspect
of light to matter. This permits the taking into account of the
behaviour of electrons in Bohr's description of atoms and signals the
birth of the quantum theory of matter.
(to be cont.)
K
Email:
kimjones.domain.name.hidden
Web:
http://web.mac.com/kmjcommp/Plenitude_Music
Phone:
(612) 9389 4239 or 0431 723 001
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Everything List" group.
To post to this group, send email to everything-list.domain.name.hidden
To unsubscribe from this group, send email to everything-list+unsubscribe.domain.name.hidden
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en
-~----------~----~----~----~------~----~------~--~---
Received on Fri Feb 06 2009 - 22:48:13 PST