NEIL W. TENNANT
tennant.9@osu.edu
If you email me, please use the header PHIL 101: [YOURNAME].
Professor
Department of Philosophy
Lecture summary © Neil Tennant, November 2004
Critical assessment of Functionalism (continued)
In the last lecture we covered the `phenomenalist' objections to functionalism. They all alleged that functionalism cannot be the whole story about subjective sensory experience. We also mentioned the Disentanglement Problem.
An even more radical line of objection to functionalism would maintain that it actually fails where it is thought to be at its strongest and best. Usually, critics focus on the phenomenal, agreeing (concessively) that functionalism might still be a workable model of what can be known about the mind from a third-person, psychological point of view. But what if we refuse to make even this concession?
One intriguing line of possible objection would be that there are not only sensory qualia; there are semantic qualia as well. Genuine communication, in language, is accompanied by a kind of subjective experience that can loosely be described as really knowing that you understand the words and grammar involved. When one hears or makes an assertion, there is a `subjectivity' to the experience; there is soemthing that it is like to be trying to represent the world as being this way rather than that. When you make an assertion, you know what it is you are trying to do; and you can also imagine what it would be like to be at the receiving end, and acquire (for the first time) the information that you are seeking to convey. This sort of subjective experience is very different indeed from exerience made up of sensory qualia; but it is genuinely subjective, for all that.
This semantic subjectivity would, however, arguably be absent from a mere `information-processing' device like a computer, even if it instantiated the program that the functionalist says underlies our linguistic competence. The suggestion here is that `semantic zombies' might be easier to imagine than `sensory zombies'. And zombies of any kind are bad news for the functionalist.
A related, though different, line of objection to functionalism comes from John Searle, and is called the `Chinese-Room Argument'. Searle attacks the functionalist's claim even to have characterized linguistic understanding of a `non-qualitative' kind, devoid of any `semantic qualia'.
The Chinese RoomThe functionalist, note, insists that all there is to linguistic understanding is this functionally describable ability to manipulate symbols (either sounds or written marks). If the functionalist's claim is true, then Searle's monoglot English-speaker inside the `Chinese Room' will be instantiating such a program, but—and this is Searle's main point—with absolutely no understanding of Chinese, qualitative or otherwise. Yet his output, in the form of answers written in Chinese in response to questions put to him in Chinese about the story told in Chinese, would convince any Chinese speaker that he understood Chinese! (He would, of course, have to work very fast when implementing the instructions from the manual he is given; but let us suppose he is somehow motivated to do that, and succeeds at it.)
Take a person who has absolutely no knowledge of, say, Chinese. Put him in a room with an in-window and an out-window. Feed him input in the form of stories written in Chinese ideography, followed by questions (again, in Chinese) intended to test his understanding of the stories. Let him be equipped with the functional program of the fully competent speaker/reader/writer of Chinese. This takes the form of a manual of instruction, written in English, for the manipulation of sequences of Chinese ideograms into other sequences of Chinese ideograms.
The conclusion, says Searle, is that linguistic understanding cannot be captured by any functionalist account. Whatever program is put forward as embodying one's competence in Chinese, it could be given to the monoglot English speaker in the room as a manual written in English, for the manipulation of Chinese characters. And he could successfully manipulate these characters in accordance with the program's instructions, all the while being blissfully unaware that his output was even interpretable as a response in any kind of language at all!
The functionalist's reply to Searle's Chinese-Room argument is known as the `Systems Response'. In my imagined version, it runs as follows.
You, Searle, are focusing on the wrong `locus of understanding' in your set-up of the Chinese room. To be sure, the man inside the room knows no Chinese. You get to legislate on that point!—it is, after all, your story. But you have misunderstood where we functionalists would locate the alleged linguistic understanding that is involved when Chinese speakers outside the room are impressed by the apparent fluency in Chinese of the `system' as a whole. Imagine that the room were inside the head of some artificial android, and that the inputs were not only the written Chinese characters of your story, but also perhaps other characters encoding sensory information from sensory transducers with which the system is equipped. And suppose further that the outputs were not only the written Chinese responses to questions, but also motor-system commands being issued by the Room (the system's Brain) to its locomotive apparatus. Call the big system Godzilla. Godzilla uses sensory information and written Chinese input to determine his course of actions, which include not only the emission of Chinese outputs but also locomotion, grasping of objects, consumption of fuels, etc. Suppose, moreover, that the streams of input and output could interlace, so that Godzilla could receive, say, a written instruction in Chinese to raise his arm, and he would do so! Or he could receive an invitation to a dinner party at the local gas station pump, and he would arrive there, tank-cap unscrewed and fuel-bladders panting. Godzilla as a whole really would understand (written) Chinese! You ask about the man inside Godzill'a head? He's just a sub-personal cog inside Godzilla's hardware, running a bit of his own sub-personal software in his own head. We call such bits of sub-routining software modules in computer science, and we call such partial bits of system-hardware homunculi. No one ever requires a homunculus to possess any properties that could only be emergent properties of the system as a whole. So, the ignorance of Chinese on the part of the homunculus (the man-in-the-room) within Godzilla's head is irrelevant; it's Godzilla whom we are claiming to understand Chinese, not the homunculus.