functionalism
"The Mind-Body Problem" (1981)
Jerry Fodor
In Heil, pp. 168-82
I. Introduction
A. writing in 1981, Fodor describes functionalism as a new philosophy of mind that is neither dualist nor materialist (168)
B. the functionalist philosophy reflects current work in AI, psychology, linguistics, etc.,
1. that is, in the "cognitive sciences"
2. all of these disciplines are concerned with information processing at an abstract level
C. functionalism tries to explain and clarify this method of abstraction
1. recognizes that it's possible for all sorts of systems, from humans to machines to spirits, to have mental states
2. psychology of a system depends not on what it's made of but how it's put together (169)
II. Functionalism's predecessors
A. Dualism
1. faces the problem of explaining how a non-physical substance can interact with physical substances
a. a committed dualist might argue that we don't even really know how physical substances interact (169)
b. however, Fodor argues, at least we meet with many examples of physical causation in experience. Non-physical causation may simply be an artifact of the dualist theory
2. also, dualism seems to be incompatible with the practice of research psychology: at least, there is no reason to expect experiments to work in a non-physical realm
B. radical behaviorism
1. here there is no problem of mind-body interaction: there are no mental causes
2. has always seemed paradoxical, since the idea of mental causation is deeply embedded in language and the way we understand one another
3. the strongest argument against behaviorism is that psychology did not turn out this way: it has discovered that mental states and processes are needed especially to explain human behavior (170)
C. nevertheless, many philosophers were initially attracted to behaviorism because it seemed to be the only alternative to dualism
1. by the early 1960s, however, they began to suspect that these were not the only choices
2. they sought a materialist philosophy of mind that allowed for mental causes
3. two alternatives were developed: (170)
a. logical behaviorism
b. central state identity theory (Fodor’s term for mind-brain identity)
D. logical behaviorism
1. a semantic thesis about the meaning of mental states
a. it analyzes the meaning of a mental state as a disposition to behave in a particular way
b. that is, every mental state is equivalent to an open-ended set of if-then statements
c. e.g., "Jones is hungry" means that "if food were available, Jones would eat," etc.
2. the advantage of logical behaviorism is that it provides a materialist account of mental causation
a. a mental cause is simply the disposition to act in a certain way
b. it is no more mysterious than a physical disposition (171)
1.) for example, we can analyze dispositions like "soluble in water" to mean "if it is placed in water, then it will dissolve"
2.) a materialist would not object to such physical dispositions, and hence, presumably, would accept behavioral dispositions as well
3. a limitation of logical behaviorism, however, is that the only form of mental causation that it allows is the manifestation of a disposition as behavior (171)
a. specifically, it does not allow for one mental state to cause another
b. this is peculiar, as even the physical sciences allow one physical event to cause another. In fact, this sort of causation is more basic than the manifestation of a disposition
c. human behavior seems to result from a combination of causes -- e.g. being hungry leads to eating if it's the sort of food one likes, one believes that it is fresh, or well-done, etc. (171-72)
4. hence, the logical behaviorist is really in no stronger position than the radical behaviorist (172)
a. although it allows for mental causation, it is all construed as if-then hypotheticals about behavior
b. it's just the semantic version of radical behaviorism -- deep down it really does not accept the existence of mental causes
E. Central State Identity Theory
1. Mental states are identical to neurophysiological states
2. allows for mental states to interact causally even when they don't issue in behavior
3. it says that our terms for mental states refer to neurophysiological states
a. unlike the behaviorist, who says that these terms either refer to nothing at all or to stimulus-response relations (172-73)
b. thus the identity theorist is able to take seriously the explanation of behavior through mental causes (173)
4. not a semantic thesis
a. it does not say that our terms for mental states mean the same thing as our terms for neurophysiological states
a. hence not open to the sorts of objections that one might raise to logical behaviorism, e.g., that when we say we have a toothache we don't mean that we're engaging in a certain sort of behavior
5. two sorts of identity theory
a. token: holds for particulars
b. type: holds for universals
6. on Fodor's reading,
a. token says only that all the mental states that happen to exist are states of the cns
b. type says all possible mental states are states of the cns
F. for Fodor, type physicalism is not plausible
1. imagine an alien with a totally different neurobiochemistry than ours, one based on silicon instead of carbon. Isn't it possible for the alien to feel pain?
a. for that matter, couldn't a machine with the right program have beliefs? (173) Or a spirit?
b. the psychology of a system seems to depend more on its software than its hardware
2. in sum, psychology seems to deal in abstract generalizations about information processing that cut across physical differences (q.v.)
G. hence a dilemma: (174)
1. the identity theory, but not behaviorism, seems to have got the mind-body interaction problem right
2. while it was logical behaviorism, not the identity theory, that understood the relational character of mental states
H. functionalism was able to solve this dilemma through the software/hardware distinction
III. Functionalism
A. Functionalism is the philosophy of mind that defines types of mental states in terms of their causal roles
1. that is, mental states are defined in terms of their causal relations with:
a. environmental or external effects on the body
b. other types of mental states
c. bodily behavior
2. it is the second that distinguishes it from logical behaviorism
3. it is neither a reductionist nor an eliminativist position
C. functionalism combines the advantages of logical behaviorism and the identity theory (175)
1. it is compatible with there being token-token identities between mental states and brain states
a. hence, it is compatible with the idea that mental causation is a kind of physical causation
b. hence, like the central state identity theory, it allows for a materialist solution to the problem of mind-body interaction
2. like logical behaviorism, it understands mental properties in a relational way
D. the machine analogy
1. the machine analogy illustrates two concepts central to functionalism:
a. that mental states are defined in relation to each other
b. that functional systems can be multiply realized
2. Fodor explains this by comparing a "behaviorist" with a "functionalist" Coke machine
a. the behaviorist machine: its single state S0 is defined only in terms of the input and output
b. the functionalist machine: the states of the machine S1 and S2 are defined not only in terms of inputs and outputs but in terms of each other
3. in both machines, there are no constraints regarding what they could be made of
a. that is, as long as it was capable of having states with the right sort of relations to inputs, outputs, and other states (175)
b. nothing in the machines program requires any specific hardware (176)
c. in the same way, nothing in our program requires neurons
d. as far as functionalism is concerned, both minds and coke machines could be made out of ectoplasm
4. n.b. he is not saying that a Coke machine has a mind
E. Some philosophers object to functionalism on the grounds that it is question-begging, circular, or trivial
1. that is, it allows the functionalist to explain a fact simply by postulating the existence of a thing that causes the fact to be explained
2. for example, it would be like explaining the opening of the valves in an engine in terms of the operation of a valve-opener, and then defining a valve-opener as something that opens valves
3. in psychology this sort of question-begging explanation typically involves the postulation of some "faculty" or "homunculus" that is responsible for the thing to be explained
a. e.g., visual perception is processed by the visual-information processor
b. the problem with such explanations is not so much that they're false as that they don't really say very much
F. the way to meet this objection is to allow the postulation of functions only where there is some mechanism that can carry out this function and it is known how it works (176, q.v.)
IV. Machine Functionalism
A. one way to meet this requirement is to say that the mental processes psychology postulates must be the sort that could be carried out by a Turing Machine (176, q.v.)
B. a Turing machine is a very simple, theoretical computer first characterized in 1936 by A. M. Turing
1. what is essential to being a Turing machine is that:
a. it is capable of being in any one of a finite number of program states at any particular time
b. it has a finite alphabet
2. for example, Turing himself conceived of this machine as supplied with a linear tape
a. this tape is divided into squares, in each of which is written a symbol taken from a finite alphabet
b. in the simplest version, there would be a "1" or a "0" written in each square
c. the machine would be able to read only one square at a time
d. based on both the (1) input and the (2) machine state (these two things together are often called the configuration of the machine), the Turing machine would either:
1.) erase the symbol and write a new one there (176-77)
2.) move one square to the left
3.) move one square to the right
3. program states in the Turing machine are defined in terms of their relations with inputs, outputs, operations, and other program states (177)
C. the Turing Machine version of functionalism captures both:
1. the relational character of mental states
2. the idea that the character of internal states is independent of their physical realization. The Turing machine could be made out of anything -- mechanical, electrical, even a roomful of people
D. Machine functionalism then avoids the circularity problem by restricting functionally defined mental states to those that can be expressed in terms of the program states of a Turing machine
1. in this way, we guarantee that there is some mechanism and our functional account is not empty (177, q.v.)
2. there are of course other sorts of computational devices, so the appeal to Turing machines is only a sufficient condition for mechanism
3. however, a Turing machine can carry out any well-defined operation on discrete symbols -- it's a universal machine
4. hence, if a mental process can be functionally defined as an operation on symbols, there is Turing machine that can carry out this function (177)
E. machine functionalism allows the psychological theorist to do away with the homunculi
1. that is, the theorist is not allowed to postulate any psychological functions that could not be carried out by a symbol-processing mechanism
2. of course, sometimes the theorist may postulate such a function or mechanism before he or she has figured out the program
V. Objections to Functionalism
A. too many different sorts of things can be given functional descriptions (178)
1. thermostats
2. mousetraps, etc.
B. what in particular makes functional state a mental state?
C. traditionally, philosophers of mind have held that mental states are distinguished by two things:
1. qualitative content
2. intentional content
D. Qualitative content
1. this is the problem of the subjective quality of our experiences that we have already seen in our discussion of materialism
2. it's also a problem for functionalism
a. which defines mental states in terms of their causes and effects (178)
b. it seems to be possible for two mental states to have the same causal relations yet differ in their qualitative aspects -- the inverted spectrum problem (178-79)
1.) there is some question whether an inverted spectrum is actually possible (179)
2.) nevertheless, it raises a serious threat: what it is that makes mental states conscious
3.) (however, it doesn't seem that anybody has a good explanation for qualia -- not even dualists)
E. functionalism has done much better with the problem of the intentional content of mental states
1. to say a mental state has intentional content is to say that it has semantic properties
2. that is, that things like beliefs are about something, can be true or false, etc.
VI. Functionalism and the representational theory of mind
A. the major achievement of the cognitive sciences is that they can account for the intentional content of mental states
B. however, it is not the functionalist approach by itself that allows us to account for these semantic properties (179)
1. we can provide a functional description of a mousetrap
2. but a mousetrap has no semantic properties
C. symbols
1. this is the important concept that allows us to account for meaning
2. like mental states, symbols have intentional content
D. Mental symbols (179)
1. so the trick to explaining content is to assume that there are such things as mental symbols or representations and that these things have meanings
a. then having a belief or thought about something can explained in terms of standing in a certain relationship to a symbol
b. and the meaning of the thought or belief derives from that of the symbol
2. similarly, the meanings of the words we speak or write derives from the meanings of our mental symbols (127a)
E. the assumption of mental symbols is compatible with the computer metaphor of the mind
1. a computer is also a symbol manipulator
2. in fact, the computer metaphor implies the existence of mental symbols, since there is no computation without symbols (180, q.v.)
F. the representational theory of the mind actually predates computers
1. it goes back to classical philosophers like Descartes, Hume, and Kant
2. for Hume, for example,
a. mental representations are called "ideas"
b. having a belief involves having an idea (180)
c. mental processes are causal processes, explained by the laws of the association of ideas
d. ideas are like pictures in the head
e. ideas get their meaning from the things which these pictures resemble
H. the contemporary representational theory differs from the traditional in several ways
1. it provides better accounts of mental processes than the laws of association
2. it does not hold that most representations are like pictures in the head
3. most importantly, it does not hold that ideas get their meaning from the things they resemble
I. Problem with the resemblance theory of meaning
1. a picture can resemble something in many different ways
2. the problem is then how do we pick out which of these ways is the one that is meant? (181)
3. for example, how does my mental picture of John mean "John is tall" when this picture includes many other characteristics of John?
J. the contemporary view is that mental representations derive their meaning not from resemblance but from their functional role
1. that is, from their causal relations to inputs, outputs, and other mental states
2. Hence, the belief that "John is tall" derives its meaning (181)
a. from its having resulted from seeing that John is tall
b. from causing us to say things like "John is tall"
c. from its allowing us to infer other beliefs, such as that if John is in the room, there is at least one tall person in here
K. the notion that the semantic properties of mental representations are fixed by their functional role is central to current work in the cognitive sciences (182)
1. thus this work has made mental representations respectable again
2. although the behaviorist may still dislike mental states, it's hard to argue with success
3. methodological principles historically give way to successful theories