THE CATHOLIC UNIVERSITY OF AMERICA

 

JOHN SEARLE’S CHINESE ROOM ARGUMENT

by

Nathan D. March

 

Rev. David D. Thayer, S.S., Ph.D.

PHIL 315 Philosophy of Language

04 December 2002

 

Introduction

 

John Searle, in his paper “Minds, Brains, and Programs” argues against the two strong claims of artificial intelligence (AI) proposed by researchers such as A.M. Turing and Roger Schank.  The first claim suggests that properly programmed computers can understand as human beings understand.  The second claim alleges computer programs themselves can explain the human activity of understanding.  Searle rejects these claims.  His thesis is that machines cannot be said to understand if their operation is defined as an instantiation of a computer program because “no purely formal model will ever be by itself sufficient for intentionality.”[1]

 

The Claims of Artificial Intelligence

Turing’s Imitation Machine

 

  A.M. Turing in his article “Computing Machinery and Intelligence,” considers the question, “Can machines think?”[2] He proposes a scientific solution to the problem by restating it in terms of an “imitation game” of question-and-answer.[3]  Turing defines the game as a human interrogator questioning, in separate rooms, both a human computer and a digital computer.[4]  He considers the best strategy for the computer is to “provide answers that would naturally be given by a man.”[5] Turing envisions human beings as machines who follow fixed rules with no freedom or authority to deviate from them.[6]  For the experiment, the human is required to answer according to rules supplied in a book.[7]  It seems plausible to him that from the perspective of the human interrogator the computer simulation could perfectly correlate with the human computer.  Turing believes this perfect correspondence, or correctness, validates the claim that the computer thinks.

 

Robert Schank’s Machine

 

Following Turing’s work, Robert Schank considered the human ability to understand stories and answer questions about them, even regarding information not explicitly stated in the story.[8]  Schank’s machine can answer this way because it has a “representation”, or database, of the sort of information that human beings would have about the objects referenced in the stories.[9]  He concludes, a computer simulation that can “print out answers of the sort that we would expect human beings to give if told similar stories,” can be said to literally understand the story.[10]  He claims what the computer and its program do explains the human ability to understand stories and answer questions about them.[11] 

 

John R. Searle’s Argument

 

Searle does not believe these formalist accounts are able to explain the human mind’s ability to ascribe meaning or intentionality to formal symbols.[12]  However, he does not reject the claim that machines can think or understand; for Searle, human beings are thinking machines or even digital computers.[13]  Rather, his argument is that no machine can think, mean, or understand solely in virtue of its instantiating a computer program.[14]

 

The Chinese Room Experiment

 

He begins his argument by suggesting, “A way to test any theory of mind is to ask oneself what it would be like if one’s own mind actually worked on the principles that the theory says all minds work on.”[15]  He employs an ingenious experiment based on Schank’s example, but indicates it will work for any Turing machine.[16]

The experiment places Searle in a locked room with two stacks of Chinese writing and a book of rules written in English. From outside the room an interrogator submits a third batch of Chinese writing to him, along with a set of instructions written in English.  The instructions allow him to correlate elements of the third batch of Chinese writing with the first two batches according to the book of rules, and instruct him how to return Chinese symbols in response.

Searle can understand English since it is his native language.  With respect to the Chinese language he states, “I know no Chinese either written or spoken, and that I’m not even confident that I could recognize Chinese writing as Chinese writing distinct from say, Japanese writing or meaningless squiggles.”[17] Since he does not know Chinese, Searle does not recognize the first batch of Chinese writing as “a script”, the second batch as “a story”, and the third set as “a question.”  The English instructions serve as “a program” and allow him to generate Chinese symbols as “an answer.”  In this way, Searle is simply behaving like a computer and performing computational operations on formally specified elements. He is an instantiation of a computer program.[18]

Searle supposes that given time he could become proficient at manipulating Chinese symbols in this manner such that, from the point of view of the interrogator, his answers to the questions would be indistinguishable from those of native Chinese speakers.[19] This perfect correspondence was the condition for the possibility of understanding for Turing.

 

The Chinese Room and the First Claim

 

The first AI claim states that properly programmed computers can understand as human beings understand.  Through the Chinese Room Seale responds, “it seems to me obvious in the example that I do not understand a word of the Chinese stories…I can have any formal program you like, but I still understand nothing.”[20] In the same way, Schank’s computer understands nothing, “since in the Chinese case the computer is me; and in cases where the computer is not me, the computer has nothing more than I have in the case where I understand nothing.”[21]

 

The Chinese Room and the Second Claim

 

The second AI claim suggests computer programs themselves can explain the human activity of understanding.  Searle responds, “we can see that the computer and its program do not provide sufficient conditions of understanding, since the computer and the program are functioning and there is no understanding.”[22]  What the Chinese Room experiment suggests is that as long as the program is defined in terms of computational operations on purely formally-defined elements they by themselves have no interesting connection with understanding since a human will be able to follow the formal principles without understanding anything.[23]

 

The Problem of Artificial Intelligence

 

Having countered the two claims of Artificial Intelligence, Searle asks the question, “What is it, then, that I have in the case of the English sentences which I do not have in the case of the Chinese sentences?”[24]  He answers that on the one hand, he knows what the symbols mean, and on the other, he has not a clue.  For Searle, the fundamental question of AI becomes, “In what does understanding consist, and why couldn’t we give it to a machine?”

 

Intentionality

 

Searle understands understanding as a cognitive state conditioned by an intentional phenomenon.[25]  This intentionality allows for the possibility of ascribing meaning to symbols.  He defines intentionality as, “that feature of certain mental states by which they are directed at or are about objects and states of affairs in the world.”[26]  Searle asserts that he understands, not because he is an instantiation of a computer program, but because he is, “a certain sort of organism with a certain biological structure, and this structure under certain conditions is causally capable of producing perception, action, understanding, learning, and other intentional phenomena.”[27] 

The fact computers are a machine does not deny the possibility they might think.  Searle simply claims, as demonstrated in the Chinese Room experiment, the formal properties instantiated in a computer program lack the causal power to produce intentionality.  Searle suggests AI’s misplacement of intentionality is the result of the misunderstanding of analogies proponents of AI use when discussing the subject.

 

The Problem of Analogy

Analogy of Intentionality

 

Searle perceives AI’s emphasis on the instantiation of a formal program as the result of the misapprehension of intentionality.  He states, “We often attribute ‘understanding’ and other cognitive predicates by analogy.”[28]  Further, he suggests human beings metaphorically extend human intentionality to their artifacts and tools as an extension of their purposes.[29]  Thus, whatever intentionality computers appear to have is exclusively in the minds of those who program them.[30] Therefore, the claim made by AI that instantiating a program is a sufficient condition for intentionality is ungrounded.

The aim of the Chinese Room example was to show that the attributions of intentionality have nothing to do with formal programs.  They are based on the assumption that if the computer looks and behaves sufficiently like a human it must have similar mental states.[31]  The experiment illustrates that, if something with intentionality is placed into the system and formally programmed, the formal program produces no additional intentionality and adds nothing to the intentional being’s ability to understand.[32]

 

Analogy of Mind-Program

 

Part of the confusion arises from the mind-program analogy: “Mind is to brain as program is to hardware.”[33]  Searle implies such a conviction stems from a “residual behaviorism or operationalism that acknowledges similar input and output patterns between computers and humans and therefore postulates similar mental states.”[34]  Such a position is consistent with modern technological objective thinking that defines truth in terms of correctness and duplication.  The Chinese Room experiment questions the position by illustrating an example of a system that has input and output capabilities, which duplicate those of a native Chinese speaker, and does not understand Chinese.[35]

The mind-program analogy, and the dualistic assumption that the mind is independent of the brain, allows for the AI claim that programs, independent of their realization, provide the causal capacity to produce intentionality. The entire AI endeavor, to emulate and explain the mental by designing programs that are independent of any realization, necessitates the mind’s empirical independence from the brain.[36]  The Chinese Room example specifically denies this by illustrating that “no purely formal model, independent of it realization, will ever be by itself sufficient for intentionality.”[37]

 

Analogy of Information Processing

 

According to the mind-program analogy, computers perform information processing and the human brain with its mind must do the same.[38]  Proponents of AI claim that if a computer is properly programmed the information processing would be identical to human information processing and therefore the computer would be said to understand as a human understands.  Searle suggests this argument is supported by an ambiguous notion of the term.[39]  He notes that computers do not do information processing in the same sense that people process information.  Unlike humans, computers manipulate formal symbols with no associated meaning. “Thus if you type into the computer, ‘2 plus 2 equals?’ it will answer ‘4’ but it has no idea that ‘4’ means 4, or that it means anything at all.”[40] Searle suggests, “If AI workers totally repudiated behaviorism and operationalism, much of the confusion between simulation and duplication would be eliminated.”[41]

 

Conclusion

 

John Searle argues against the two strong claims of artificial intelligence: that properly programmed computers can understand as human beings understand and computer programs themselves can explain the human activity of understanding.  Using the Chinese Room experiment he demonstrates the inability of a formal instantiation to result in understanding. Searle defines human understanding as a biological phenomenon resulting from the brain’s causal capacity to produce intentionality.[42] His prejudice towards cognitive science and the philosophy of the mind limits his exploration of artificial intelligence with respect to the philosophy of language. Only secondarily, does he consider how misinterpretation and ambiguous analogy provide the impetus for artificial intelligence.