Functionalism: A first problem

So far what I have said about functionalism is very abstract. If we are to make it plausible, we will need a more concrete case to consider.

Beliefs, for a functionalist, are characterized as states that are caused by sensations and perceptions of the appropriate kind, and that can cause other beliefs, and that interact with desires to pro­duce action. Thus, for example, seeing a gray sky causes me to believe that the sky is gray, which may lead me to believe that it will rain, which may lead me to take my umbrella, because I desire not to get wet. Here the input is sensation and perception and the out­put is action; the internal states that mediate between the two are beliefs and desires.

There is an immediate and obvious problem for anyone who wants to say what beliefs are in a theory of this kind. Remember that a functionalist says what an internal state of the system is by describ­ing its functional role: by saying how it functions in mediating between input and output in interaction with other internal states. Suppose we try to do this for some particular belief—say, the belief that the sky is gray. You might think you can say fairly precisely what would cause this belief. Looking up, eyes open, fully conscious, at a gray sky ought to do it. But the trouble is that this is really neither a necessary nor a sufficient condition for acquiring the belief. It isn't necessary, because you can acquire the belief in lots of other ways: looking at the sky's reflection in a pond, for example, or listening to a weather forecaster. It isn't sufficient, because, in suitably weird circumstances, you might reasonably believe that the sky wasn't gray when it looked gray. (Suppose, for example, I told you I had inserted gray contact lenses in your eye while you were asleep; suppose you believed me.

A similar problem arises with output, though here the issue is even more complex. For what you do on the basis of the belief that the sky is gray depends not only on what other beliefs you have—for example, do you believe that gray skies “mean” rain?—but also on what desires you have—for example, do you want to avoid getting wet? So whereas for a heat sensor in a thermostat the effect of input doesn't depend on an indefinitely large number of other internal states, in the case of belief in a mind it does.

In finding a way to handle this increased complexity, the analogy with the computer is helpful. For, in this respect, computers are more like minds than like thermostats. The results of inputting a number to a computer depend also on a complex array of internal states. If I put in a “=” to an adding program after putting in “2” fol­lowed by “+” followed by “2”, then the output will be “4”; but if I put in the same sign, “=”, after putting in “4” followed by “+” fol­lowed by “2”, then the output will be “6”. Yet we can still give a func­tional role to each internal state of the system: we can do it by say­ing, for example, that when the adding program is in the functional state of having a “2” stored, entering “+” followed by any numeral, “n”, followed by “=” will result in outputting the numeral “n + 2”. The general strategy is this: we must specify the functional role of a state, A, by saying what will happen, for any input, if the computer is in state A, but in a way that depends on what the other internal states are.

So for a functionalist account of the belief that the sky is gray, we can say, at the level of input, that it will be caused by looking at gray skies, provided you don't believe that there's some reason why the sky should look gray when it isn't; and that it will also be caused by acquiring any other belief that you think is evidence that the sky is gray.

It may look as though we have still not solved the problem we started out with. For this definition of the belief that the sky is gray still seems to define it in terms of other states of belief and desire, and these other states are ones we want to give functionalist defini­tions also. So, you might ask, isn't this sort of definition going to be circular? We are going to define the belief that the sky is gray partly in terms of what it will lead you to do if you believe that gray skies mean rain; but aren't we going to have to define the belief that gray skies mean rain partly in terms of what it will lead you to do when you believe the skies are gray?

This is a genuine problem if you want to use functionalist defini­tions, but there is a procedure that allows us to solve it in a way that avoids this circularity. Applying it in the case of beliefs is extremely complex, so it will help, once more, to start with a simpler case.

1.7