Chinese room argument

« alone.

Such an ensemble he asserts is just the wrong sort of thing literally to understand or to have a mind.

Second, he tries to avoid the objection by altering his example and removing himself from the room to a wide open field and committing all the rule books to memory.

In such a case he would perform all the purely formal operations in his head, re-establishing his analogy with the whole computer.

He asserts he would still not understand a word of Chinese (though some of his critics claim otherwise). The robot reply concedes that the original Chinese room might not suffice for understanding Chinese.

Since all its behaviour is strictly verbal, it is able to make connections between words and words, but not between words and the world: it might give coherent answers to questions about boiled eggs, but not have any ability to recognize an egg.

The robot reply thus imagines a program installed within a robot that allows it to simulate the full range of human behaviour, verbal and non-verbal; and that , it is claimed, would count as understanding Chinese and having a mind. Searle, however, disagrees.

He notes that the robot's sensors would supply its internal processors only with more formal symbolic inputs.

Were he and his Chinese room installed within the robot with additional rule books specifying how to respond to those additional symbols, and were these responses to drive the robot's apt behaviour, he, Searle, would still not understand a single word of Chinese.

Nor, he argues, would the robot.

Again, the assumption that mentality requires consciousness seems in part to drive Searle's intuition.

Critics less sympathetic to that assumption and more inclined to analyse intentionality in terms of causal interactive relations have been predictably more persuaded by the robot reply, especially if it is combined with the systems reply (see Semantics, informational ). Other objections to the Chinese room have sought to discredit it as a source of unreliable and misleading intuitions. Some have focused on the enormous disparity between the complexity and speed that would be required in a computer able to pass a Turing test, which the man in the room could not match.

Others have stressed that strong AI need not be committed to the Turing test, and that the specific program by which behaviour is produced is crucial to whether something has a mind.

In reply, Searle denies that speed or style of program is of any importance; what matters, he argues, is that the operations he performs are merely formal.. »