Philosophy of Mind

1
Philosophy of Mind

• Lecture Seven Functionalism

2
Previously

• Weve examined Type-Type Identity Theory (TTID).
• Weve seen some of its alleged problems it
  cannot account for the multiple realisability of
  mental states.
• So we make a move to Token-Token Identity Theory
  (ttID).

3
This Week

• We will look at one of the more dominant
  philosophical theories of the mind
  functionalism.
• Functionalism is (allegedly) a version of ttID.
• We will introduce the theory, and examine just
  one of its problems.
• Next week, we will look at a problem for ttID in
  general.

4
Functionalism

• Functionalism comes about during the late 50s
  and early 60s.
• The early papers on it often made reference to
  computers, and the way they worked.
• Indeed, they were closely aligned to Turing
  Machines.

5
Computers and Turing Machines

• Think back to last lecture.
• I gave the example of two objects which had
  radically different workings and properties
  carrying out the same process/task.
• Namely, a contemporary PC and an old 80s ZX
  Spectrum.

6
Computers and Turing Machines

• The two computers are radically different.
• Their microchips and circuit boards are obviously
  very different in terms of complexity and the way
  in which they work.
• But they both can still carry out certain tasks.
• For instance, calculating what the square root of
  1,241,928,081 is.
• Or playing a ZX Spectrum computer game.

7
Computers and Turing Machines

• Both can carry out the function of following
  certain instructions and producing certain
  outputs (e.g. producing the number 35,241 or
  certain images of a computer game)
• So even though they are physically different,
  they can both carry out certain functions the
  same.
• We can see this feature more generally with
  Turing Machines.

8
Computers and Turing Machines

• Turing was writing before the invention of
  complex computers.
• He was interested in what it was for something to
  be computable.
• So certain things were computable, and others
  were not.
• You could build a machine to compute certain
  sequences.
• For instance, a machine to compute the square
  root of 1,241,928,081.

9
Computers and Turing Machines

• Of course such a machine would be pretty limited.
  Instead we build computers that can calculate all
  kinds of things.
• Example Calculators.
• But even they are limited. A calculator cannot
  compute everything.
• Turing managed to demonstrate that one could
  create a machine that, theoretically, could
  compute any computable sequence whatsoever.
• A universal Turing machine.
• (there was lots of complex maths involved)

10
Computers and Turing Machines

• So when these early philosophers are talking
  about Turing machines, all that is important is
  that they are thinking of machines that can
  simulate other machines.
• Like the PC and the ZX Spectrum.
• Getting to grips with the die hard maths and
  logic isnt essential (but may be useful, as
  dictated by where you take your essay)

11
Functionalism

• What the functionalist takes from all this is
  that whats important is not how the task is
  carried out but the function that is carried out.
• They think the mental states are a bit like that.
• What matters is the functional nature of mental
  states, not what implements them.

12
Functionalism

• So whats important to being in pain is the
  carrying out of a certain function.
• When certain inputs are received (e.g. those
  received when your hand touches a hot stove)
• certain outputs are produced (e.g. removing your
  hand and seeking medical attention)

13
Functionalism

• It doesnt matter how these functions are
  implemented.
• Example Humans and C-fibres
• Example Martians and feet cavities
• Example Angels and ectoplasm
• That functional role can be realized in many
  different ways.

14
Functionalism as ttID

• There are actually different brands of
  functionalism.
• Here were interested in their division into two
  types realizer functionalism and role
  functionalism.
• Youve already seen an example of realizer
  functionalism when we looked at Lewis.
• There pain was whatever property realized a
  certain function. So as the function was realized
  by C-fibres firing, thats what pain in humans is.

15
Functionalism as ttID

• Realizer functionalism, whilst important, plays
  second fiddle to role functionalism.
• Role functionalists identify the mental states
  not with the thing that realizes the role, but
  with the role itself.
• Hence the names!
• Clearly realizer functionalists are TTID (as
  discussed last week).
• Role functionalists claim to be ttID.

16
Functionalism as ttID

• So for role functionalists, pain is the role
  type, not the brain type.
• So whilst (angels aside) every mental token is
  also a token of a physical type (i.e. ttID is
  true)
• whats important is that they are tokens of a
  (non-physical) role type (i.e. TTID is false).

17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
What is a function?

• So thats one dimension along which
  functionalists disagree.
• There are other dimensions.
• The next one to concentrate on is what counts as
  a function?
• The theories disagree over what counts as a
  function, and which functions mental states are
  to be identified with.

24
What is a function?

• A functions are defined by what they take as
  inputs, what they take as outputs and how they
  relate to other functional states.
• Take a simple function a coke machine.
• Imagine the coke costs 1, and machine takes 50p
  and 1.
• So it has two inputs (50p and 1).
• It has two outputs (your change in 50p, and cans
  of coke).
• The outputs depend upon what state the machine is
  in.
• What it does when you stick in 50p depends upon
  what state the machine is in.

25
What is a function?

• Imagine the machine starts in state one.
• Stick in 1, and the machine emits coke, and
  stays in state one.
• Stick in 50p and the machine shifts to state two.
• In state two, if it receives 50p, it emits coke
  and switches back to state one.
• If it receives 1, it emits coke and 50p. And it
  also switches back to state one.

26
What is a function?

• We can build up a simple machine table to express
  this function.

27
What is a function?

• We can build up a simple machine table to express
  this function.

28
What is a function?

• We can build up a simple machine table to express
  this function.

29
What is a function?

• We can build up a simple machine table to express
  this function.

30
What is a function?

• We can build up a simple machine table to express
  this function.

31
What is a function?

• We can build up a simple machine table to express
  this function.

32
What is a function?

• We can build up a simple machine table to express
  this function.
• This is a very simple machine table. And a very
  simple function.

33
What is a function?

• Note that the functions can depend upon what
  other functions are doing.
• So a machine might have a tampering detector
  function to prevent it being fooled by fake
  coins.
• All those outputs would only occur if the
  tampering detector function produced outputs
  saying it wasnt being tampered with.

34
What is a function?

• So the functions depend upon inputs, outputs and
  other functional states.
• Whilst these functions are simplistic, and the
  brain carries out far more complex functions, the
  idea is basically the same for your brain.
• Its that last bit that makes functionalism
  different from behaviourism.

35
What is a function?

• So the functions depend upon inputs, outputs and
  other functional states.
• Whilst these functions are simplistic, and the
  brain carries out far more complex functions, the
  idea is basically the same for your brain.
• Its that last bit that makes functionalism
  different from behaviourism.

36
What is a function?

• Beh. tried to eliminate mental states.
• So x is in pain iff where the RHS lacked any
  mention of mental states.
• Functionalism does have in common this close
  attention to what things do, how they react i.e.
  their outputs.
• But it doesnt try to eliminate mental states at
  all.
• Functions often depend upon what other functions
  are doing.
• So if mental states are functional states, that
  state will depend upon what other mental states
  are doing.
• So x is in pain iff where the RHS may well
  include reference to other functional/mental
  states.
• No elimination!

37
Levels of Abstraction

• With the idea of a function in place, we can
  start to see some of the problems functionalism
  suffers from.
• Functions can be seen at different levels of
  abstraction.
• Imagine we filled the coke machine with Lilt.
• Is it now carrying out the same function?

38
Levels of Abstraction

• In a sense, it isnt.
• Its outputs are now cans of Lilt, rather than
  cans of coke.
• So the outputs have changed, and if a function is
  defined in terms of its outputs, then it is now a
  different function.

39
Levels of Abstraction

• So whereas we previously had
• We now have

40
Levels of Abstraction

• But in a sense the function hasnt changed.
• The coke machine and Lilt machine are still
  carrying out exactly the same functions if we
  take the outputs to be a bit more abstract.
• So we end up with this kind of machine table

41
Levels of Abstraction

• So it doesnt matter what its emitting (coke
  cans Lilt cans mars bars fifteen cubic metres
  of deadly nerve toxin etc.) all such machines
  have, at that level of abstraction, the same
  function.
• And of course we can make the inputs more
  abstract as well.
• So it doesnt really matter what goes in (pence
  and pounds cents and dollars whatever)

42
Levels of Abstraction

• Whats this got to do with the mind?
• Well, the level of abstraction we use for the
  functions we identify with mental states causes
  great difficulty for functionalism.
• Either it ends up being chauvinistic (excluding
  entities that should have the mental state from
  having it) or liberalistic (including entities
  that shouldnt have the mental state as having
  it).

43
Chauvinism

• Lets see that in action.
• Lets say that you thought being in pain was a
  certain function, f.
• And f took as inputs things like certain nerve
  signals travelling along C-fibres.
• And f took as outputs things like removing hands
  from stoves, and screaming, and shaking somewhat.

44
Chauvinism

• But this means that creatures without C-fibres
  wouldnt have those inputs.
• The Martian has quite different inputs.
• Nor do creatures that respond to pain in a
  different way have those outputs.
• The Martian may not cry out, but do something
  quite different.
• So if pain was that functional state, Martians
  cant be in pain which is clearly not what we
  wanted to say.

45
Chauvinism

• So the move is to talk about a more abstract
  function.
• One that both Martians and humans (and angels
  etc.) can realize.
• In the same way that we move to a more abstract
  machine table for our Lilt/coke/nerve gas machine
  above, we can move to a more abstract function
  for being in pain.

46
Chauvinism

• So just as it didnt matter what the input or
  output was for the machine, it doesnt really
  matter what the input or output is for beings in
  pain.
• Its just whatever plays the correct role in a
  given functional state.

47
(No Transcript)
48
Liberalism

• The Martian and the human can both realize that
  more abstract functional state.
• So if we take this version of functionalism, we
  are no longer chauvinistic.
• Alas, it looks like we will be liberalistic.
• Thats a really abstract functional state.
• All kinds of things will end up being in that
  abstract functional state.
• Things that intuitively arent in pain.

49
(No Transcript)
50
Liberalism

• Clearly the gas companies computer isnt in pain
  even though it is carrying out exactly the same
  abstract function we detailed.
• Of course, what weve been doing so far is highly
  simplified.
• The functional state that pain will be identified
  with is going to be vastly more complex than the
  example Ive been using.
• The gas company computer realizes a functional
  property, but not a functional property as
  complex as what pain will be.
• I wanted you to get an idea of the nature of the
  problem first.

51
Liberalism

• There are various thought experiments that are
  meant to be examples of things that could
  instantiate these more complex functional states.
• So imagine we detailed, at that level of
  abstraction, the highly complex function that is
  to be identified with pain.

52
Liberalism

• Now imagine that we mapped your brain.
• For every neuron, we got a person from China
  (chosen for its population) to play the role of
  that neuron.
• There are only so many ways that a neuron can
  react, taking inputs from other neurons, and
  sending outputs to further neurons.
• We teach each Chinese person how to replicate
  these functions.
• They represent the inputs/outputs by scrawling
  appropriate words on a small blackboard.

53
Liberalism

• So where your neuron receives a charge of type
  one from the appropriate neuron and emits a
  charge of type two
• Our Chinese guy will, upon seeing that the
  appropriate guys board has a certain sign, will
  respond by scribbling a certain sign on his
  board.
• There are only so many types of inputs and
  outputs from the neuron, so we only have to teach
  a small number of these signs to each Chinese guy.

54
Liberalism

• The Chinese nation will now be able to carry out
  exactly the same function as your brain does.
• Obviously itll do it slower, but what does that
  matter?
• Example Asking China if it would like a cup of
  tea.
• Clearly, one wouldnt think the Chinese nation
  was thinking!
• But it does have the same abstract function!

55
Liberalism

• Similar examples are found elsewhere
• Searles Chinese Room.
• The Homunculi Headed Robot.
• The Bolivian Economy.
• So it looks like there will be examples of
  things instantiating the relevant function, and
  therefore being in the appropriate mental state,
  when (intuitively) they are not.

56
Liberalism

• So one difficulty for functionalism is exactly
  how it goes about defining function without
  committing us to ridiculous claims about what
  thinks and what doesnt.
• So say the critics, there has to be something
  else to the mind other than the functions it
  carries out.

57
Recap

• Weve looked at functionalism.
• There is a version (realizer functionalism) which
  is Type-Type.
• There is a version (role functionalism) which is
  Token-Token.
• So we have an example of ttID.
• Weve then turned to a general criticism of
  functionalism.
• What is a function?
• Can it be defined in such a way to exclude
  ridiculous things from thinking, whilst still
  allowing us to account for multiple realisability?