PHIL2206 Philosophy of Mind
Week 7: Causal Role Functionalism

Causal role functionalism

  1. Last week we started looking at functionalism, according to which mental phenomena are roles.
  2. We looked, in particular, at machine functionalism, according to which mental phenomena are computational roles. More specifically: to have a mind is to be a Turing machine with a sufficiently complex machine table, and mental states are the internal states of the Turing machine.
  3. This week we are looking at causal role functionalism, according to which mental phenomena are causal roles. Pain, for example, is a certain causal role: the role of being caused by certain things, and causing certain things. To be a pain is to play this causal role, to have a pain is to have something which is playing this causal role.
  4. What is the pain causal role? Here is a starting suggestion, with one for the itch causal role as well:

A problem

  1. There are at least two problems for these suggestions:
  2. Let's set aside the first problem and assume that it can be fixed. To solve the second problem, we might try something like the following theory of what the pain role is:
  3. But now we get a new problem: in saying what the pain role is we have referred to the mental phenomena of being alert and being distressed (underlined), so we haven't said what the pain role is in purely non-mental terms.
  4. We might try getting rid of the reference to being alert and being distressed by plugging-in an account of what the alertness and distress roles are.
  5. But, like the pain role, no account of the alertness role will be correct unless it refers to mental phenomena, and so too for the distress role. So even when we plug these accounts in we are left with reference to mental phenomena.
  6. So it seems that we cannot give an account of what the pain and distress roles are in purely non-mental terms.

The Ramsey-Lewis method

  1. There is a method for specifying the pain role which promises to solve this problem.
  2. Suppose we have the following accounts of the pain, alertness and distress roles (each of which refers to mental phenomena, underlined):
  3. Now let's take these accounts and combine them into a single psychological theory, call it T:
  4. Now let's Ramseify T by forming its Ramsey sentence, TR (by existentially generalizing):
  5. Note that TR is logically weaker than T: T entails TR, but TR does not entail T. But TR is just as predictively powerful as T: TR and T make the same predictions about non-mental behaviour.
  6. Now let's give an account of what it is to be in pain, what it is to be alert, and what it is to be in distress:
  7. And now we have a causal role account of what it is to be in pain in purely non-mental terms. So too for being alert and being distressed.

    (But what about giving an account of the roles themselves: pain, alertness, distress?)

  8. Note that the Ramsey-Lewis method of giving an account of what it is to be in pain allows that pain can be multiply realized. For any kind of organism (e.g. Martians), as long as there are x, y, and z such that T(x,y,z) for that kind of organism, then organisms of that kind can be in pain. And it makes no requirement that x, y and z be physical (unless this is built-in to T itself, which it seems it wouldn't be if T is just about causal roles).
  9. So this seems like a promising method. We just need to make sure we start with the right psychological theory.

Choice of a psychological theory

  1. To get an account of all mental phenomena we need a theory that tells us about all mental phenomena.
  2. It needs to say enough about the causal role of each mental phenomenon for it to distinguish them. The following is obviously not enough:
  3. But if it says too much then it runs the risk of being false. And we don't want the theory to be false.
  4. Why not? Because if T is false then TR might also be false (note: it might be true, if we get lucky). And if TR is false then we get accounts of mental phenomena according to which they never occur.

    Why? Consider pain. For a to be in pain is for there to be x, y, z such that: T(x,y,z) and a is in x. But if TR is false then there are no x, y, z such that T(x,y,z), so there are no x, y, z such that T(x,y,z) and a is in x. So a is not in pain.

  5. There are two psychological theories that we might try starting with:
  6. There are pros and cons for each. We have a fairly stable commonsense psychology that has stood the test of time and seems to work very well, and it is a theory that we all seem to know pretty well (at least implicitly, just as we all know the rules of grammar). But when it gets put under scientific scrutiny it is sometimes found wanting (perhaps even massively false). And perhaps, when pushed, we are not really prepared to stand by our folk theory. But scientific psychology is young, its methodological foundations are still in dispute, and it is yet to produce a robust core of generally accepted laws and theories.
  7. Perhaps we don't have to choose. Perhaps we can distinguish folk-pain from scientific-pain and get an account of the first from our folk psychological theory, and an account of the second from our scientific psychological theory. But is this satisfying?
  8. Here is something else to consider: What if there is no such thing as the folk psychological theory, or the scientific psychological theory? What if there are two (or more) theories, each equally compatible with all observable non-mental behaviour? Which theory is the theory that we should use? They might, after all, yield different and incompatible accounts of mental phenomena.

Some problems for causal role functionalism

  1. Absent qualia. Suppose that John has something that is playing the pain causal role, but it does not feel any way to him: there is nothing it is like to have this thing, it lacks any qualia. According to causal role functionalism this thing is a pain, and John is in pain, because this thing is playing the right causal role. But is this the right result? Don't pains have to feel some way in order to count as pains?
  2. Inverted qualia. Suppose that John has something that is playing the pain causal role, but it feels to John the way orgasms feel to you and me. So there is something it is like for John to have this thing, it does have a quale. But has the 'wrong' quale. According to causal role functionalism this thing is a pain, and John is in pain, because this thing is playing the right causal role. But is this the right result? Don't pains have to feel a particular way in order to count as pains?
  3. Cross-wiring. Suppose that John has something, call it x, playing the pain role, and something, call it y, playing the itch role. So when x is activated John has a pain, and when y is activated John has an itch. Now suppose that we re-wire John (presumably his brain) so that y is now playing the pain role, and x is now playing the itch role (swap their inputs and outputs). According to causal role functionalism, when x is activated John now has an itch, and when y is activated John now has a pain. But is this the right result?
  4. A disjunctive property? Suppose that pain is a causal role, and that x and y can both play that role, and are the only things that can play that role. Then it seems that the property of being in pain is the disjunctive property of either having x or having y. But is there such a property? And if there is, is it the kind of property that serious scientific study should interested in?
  5. Causal powers. Do causal roles have any causal powers, or just the things that play those roles? If it is the latter, and if pain is a causal role, then why should be interested in pain at all - why not focus on whatever it is that plays the pain role?