|
7. Four Challenges to Functionalism |
||
|
Consider
3 things that are functionallly like us but psychologically unlike us. As
such they constitute challenges to F-ism. 1.
The 2.
The Chinese Room 3.
Blockhead 4.
Zombie For
each consider: a.
Is the thing really psychologically unlike us? b.
Does it contradict F-ism, or just some kinds of F-ism? The
Suppose
each Chinese person is given the running instructions for a neuron of your
brain. The population of Let
this China Brain (CB) be fed inputs from an exact android simulacrum of
yourself, and the outputs go to this android’s limbs, etc. The android
behaves just as you do, but all its brainwork is outsourced to Does
CB have M states like you? No?
Then F-ism must be false. Denying the intuition Why
trust that intuition? It results from our being impressed by irrelevant
differences such as sheer size or speed. Consciousness Note
that we’re not necessarily concerned with consciousness here. We may
only be talking about M states such as belief and desire. This seems
easier to attribute to CB, since its actions are thus explicable. Connection to the environment Suppose
the android is omitted. Then the functions are not connected to the
environment and the functions are purely abstract. This is like the case
that we said was in danger of excessive liberalism – mere
number-crunching can’t produce M states. The
Chinese Room Suppose there was a room in which there was a
person who knows only English. The room has a slot in the door through
which from time to time there comes a piece of paper with strange
squiggles drawn upon it. In the room there is also a large book that has a
set of instructions in English about what to do if someone sends you a
scrap of paper with squiggles on it. The person follows those
instructions. It turns out that what is happening when the person is doing
this is that he’s participating in a conversation conducted in Chinese.
Does the system of the person
plus all the other paraphernalia understand Chinese? The
example embellished Searle says NO, the person in the room is simply
manipulating symbols; and yet the CR is passing the Turing Test for
thinking. (Note that we can elaborate the instruction book as required to
produce memory, flexibility, etc) For understanding we need semantic capabilities that the CR doesn’t have. It is a syntactic
engine only. A further embellishment Give the room causal connections to the world.
Connect the room to an android as we did for the CB, and let the inputs be
environmental data and the outputs be actions. A system that does not understand Chinese Now the intuition is that the CR plus
the android body understands Chinese. Searle would say: let the man memorise the
instructions. He now is the
CR+body. Does he understand? Searle says No, but isn’t this pretty
similar to multiple personalities? We should say Yes. A computer analogy We should think of this as being like a Mac
emulating an MS Windows PC. The Mac doesn’t really know what it’s
producing in the way that the PC would, but it does the same thing
nevertheless. What we have now is a system that our intuitions
tell us might understand Chinese (even if the person in the CR never
does.) Blockhead Input-output functionalism We
earlier saw that EF-ism that is too restrictive of internal organization
is possibly chauvinistic. Can we make it so that the only constraint on
internals is that they are responsible for there being the right
relationship between inputs and outputs? This we might call input-output
or stimulus-response functionalism.
It is like Behaviourism, but makes M states internal causal states. In
fact Ned Block’s Blockhead example shows this is not a possibility. Good chess versus being good at chess Copycat chess Suppose
that chess grandmasters create a look-up tree that gives the best moves in
response to all possible moves by a chess opponent. Anyone with such a
tree could play chess at grandmaster level without knowing anything about
chess. The game of life Look-up trees for life Treat
life as a game. For every person, at any time there are finitely many
(effectively distinguishable) inputs and finitely many (effectively
distinguishable) responses. Therefore, in principle for any person one can
make a look-up tree for all occasions that will perfectly describe their
behaviour. Do it for Jones, say. Blockhead Now
create Jones’s blockhead twin as a Jones-bot whose actions are
determined by the Jones look-up tree written to a chip in its head. The
input-output relations for Jones and the blockhead-Jones are identical,
and the relations are the result of the internal states of the
blockhead-Jones. However, we’re sure that BJ doesn’t think or have
real M states. So I-O F-ism can’t be true. Blockhead’s challenge to us all Note
that the example demonstrates that I-O F-ism is wrong even about
intelligence. We are sure a blockhead isn’t intelligent. Since
intelligence is, we often think, all about getting right answers to
problems, why do we think blockhead isn’t intelligent? Some wrong turns It’s
not because everything is written in advance. The same might be true of us if determinism is true. It’s
not because of the practical/nomological/epistemological/… impossibility
of the look-up tree. The conceptual
possibility of the tree is all that is required. Our
intuitions don’t count in such impossible cases.
But is is so like similar more possible cases – like someone who uses a
chess book to play. Why
Blockhead is not a thinker Causal
connections are important in many things. We see X because our perceptual
states of X are properly causally connected to X; a person at t1
is the same person at t2 if there are the right causal
connections between the two; etc. Rationality and causal history Rationality
depends upon beliefs developing in the right way from previous beliefs and
perceptions. Ditto intelligence. Part
of being a belief is that it tends to develop in certain ways. X can only
be a belief that ‘if A then B’ if when one has beliefs X and A that
will tend to cause the belief B. Blockhead’s causal peculiarity Look-up
tree devices don’t have the right causal relations. They use static
trees in which at any one time there is just one active node. But note
that the set of nodes at depth d1 does not generate the
‘later’ nodes at d2. This is not how we think of thinking.
For us to be thinking, our M states at t1 must
generate the M states at t2. Common-sense
functionalism and Blockhead Note
that Blockhead doesn’t affect C-S F-ism since it already held that the
right causal relations between M states were required. However, Blockhead is
an objection to I-O F-ism. [Indeed
it’s hard to see any intuitions blocking C-S F-ism, since that is
supposed to be the sum of all our intuitions.] The Zombie Objection Zombies invade the physicalist paradise The
Zombie Objection is made against any form of physicalism. A zombie
is a creature physically just like us, and identical in its behaviours,
but which has no inner ‘feels.’ It isn’t like anything to be a
zombie. So there’s something non-physical that a zombie lacks that is
necessary for our feels. So physicalism is false. Expand
the argument: P1
We can conceive of a minimal physical duplicate of this world where
people are zombies. P2
Conceivability suggests possibility P3
We aren’t zombies C4
Zombies are possible (1+2) C5
There is a minimum physical duplicate of this world mentally
different from it. (3+4) C
Physicalism is false (5+definition of Ph.) This
is a valid argument, so a physicalist must deny a premise. Analytic
F-ism denies P1. It’s a matter of the meaning of the words that you have
M states when the right functions are instantiated; and here they are
instantiated physically. There is no conceiving of the comceptally (a
priori) impossible. Analytic functionalism and ideal conceivability But
we think we can conceive of
zombies – so is A F-ism false? Most
think that if zombies are impossible this is a metaphysical fact not a
semantic one. Perhaps
our intuition of conceivability is incorrect. Compare getting a math
problem wrong. But what about the case of ideal conceivability, when all the facts are presented to a rational
mind? Then we can’t get the math problem wrong. Well,
maybe we are only concerned with unideal
conceptions in the zombie case.
What work does the additional clarity and rationality actually do
for our intuitions? Perhaps we should take strong intuitions of
possibility as indicating that the A F-ist theory of M state terms is
wrong. Empirical functionalism and zombies For
E F-ists it seems to be possible to deny P2. It may be claimed that
zombies are conceivable, but impossible, because we can discover a posteriori that qualia really are physical. This
argument doesn’t work for one kind of E-Fism which is coherent, and
those E-Fisms for which it does work don’t appear to be coherent. Reference-fixing Consider
some forms of E F-ism in which folk-roles, say, pick out M states, and we
then rigidify on the internal features of those M states. If we reference
fix on brains and find neural feature X playing the qualia role, then if
we rigidify we find that qualia are necessarily
neural feature X. So
zombies, though conceivable, are a
posteriori impossible: anything with NF X has qualia, and zombies have
NF X, so zombies have qualia, so zombies aren’t zombies. OTOH. Suppose we allow that ‘qualia’ refers to
what plays the folk roles that we used to reference-fix – i.e. NF X.
Still, we knew enough to reference-fix before we knew any neuroscience.
Let ‘qualic’ mean having the roles played. We knew we were qualic and
that qualia were the role players (but we didn’t know what qualia were.) It
is conceivable that there are possible physical duplicates that are not
qualic. These are R-zombies. But
if being qualic is just having roles played, then nothing can be like us
and fail to be qualic. So
there can’t be R-zombies. Finally, There are E F-isms that deny that we should
use folk roles. However; whatever they think indicates a qualitative
nature will also be vulnerable to the zombie. A modification of analytic functionalism The
A F-ist’s position should be that if
Dualism is false then A F-ism is
true. Then
the conceivability of zombies is partly the conceivability of Dualism.
But if A F-ism is a priori
true, then zombies should still be ideally inconceivable. Instead
let the a priori truth be that: if
Dualism is false then A F-ism is true. So
the analytic truth we grasp is this: If there are dualistic states then
In @ qualia are D states, and in all PW all and only qualia are D
states
Else In @ qualia are the states that play the roles here,
and in other PW they are the states that play the roles there If
so then if D is false then zombies are impossible, but we can’t know
that a priori. So the zombie intuition conflates two things; the possibility of dualism and the possibility of zombies.
|