Searle’s Chinese Room
The Chinese Room is a thought-experiment designed by John Searle. Its purpose is to refute strong AI and functionalism. Searle describes the following situation:
A monolingual Englishman sits in a room. In the door there are two slots. Pieces of paper are passed into the room through one slot. On each of these pieces of paper are symbols, perhaps ‘squiggle squiggle squiggle’.
The man has an enormous book containing a set of instructions, written in English, some blank pieces of paper, and a pencil. By comparing the “squiggles” passed into the room with this book, the man derives instructions telling him “to draw some new symbols, perhaps ‘squoggle squoggle squoggle’.” He draws the symbols on a blank piece of paper and then passes it through the second slot in the door.
Unknown to the man, the symbols are Chinese characters, the notes passed into the room are questions from a Chinese speaker, and the notes passed out of the room are answers to the questions. The instructions that the man is following are such that the answers to the questions that he derives are indistinguishable from those that a Chinese speaker would give.
This works against functionalism in the following way: Functionalism tells us that being in a mental state is to be in any state that plays the right functional role. So, to take a simplistic example, to be in pain is simply to be in a state that typically results from bodily damage and gives its subject a propensity to roll around on the floor, clutching a part of his body, screaming in as loud a voice as he is able. This doesn’t mean that pain invariably involves this sort of behaviour, just that subjects of pain are liable to act in this sort of way for the duration of the experience. Note that the particular state of the person in pain is unimportant here, any state which plays the causal role associated with pain is an instance of pain, and likewise for other types of state. The model simply involves a state resulting from the right kind of causes and itself playing the right kind of causal role.
It is this model of the mind that the strong AI program attempts to exploit. If a computer can be programmed to behave in the right way in a wide variety of situations, then it will at least appear intelligent. Its program will be reducible to a set of propensities to perform certain actions, given the right situation. These can be expressed as conditionals; if input (situation x), produce output (action y). On the functionalist view, if the right set of conditionals are true, and they are true because the computer is in a particular physical state, then those physical states will be playing the right causal role, and the computer instantiating the program will literally have a mental life. This is the first component of the position labelled “strong AI” by Searle.
The Chinese Room scenario is, Searle claims, analogous to this. The mental state in question is that of understanding Chinese, and the analysis of this capacity that Searle is attacking is that which equates it with being in any state which plays the right causal role, which produces the right output when given the right input, which makes a particular set of conditionals true. The man sits in the room with his book, to take the AI case, he is the computer. Along comes an input, ‘squiggle squiggle squiggle’, through the first slot in the door. He follows the instructions in the book, runs the program if you like, and produces the output ‘squoggle squoggle squoggle’. By stipulating that the man’s ‘squoggles’ are indistinguishable from the characters a Chinese speaker would draw, Searle intends to ensure that the set of input-to-output conditionals that are true is the right set, i.e. the set appropriate to the mental state of understanding Chinese. So the functionalist condition appears to have been met. Yet if you take the man out of the room, not only is he still a monolingual Englishman, he also has no idea that he has been part of a written conversation in Chinese. Where, asks Searle, is the understanding?
|
