The Chinese Room experiment places Searle inside a room, in which he receives inputs in Chinese and instructions on how to process those inputs in English furthermore Searle has no previous knowledge of Chinese or its grammatical structures. Suppose that Searle and the programmers who are giving him instructions get so good at their respective tasks that Searle’s Chinese outputs become indistinguishable from those of a native Chinese speaker.fn-searle
Searle’s argument goes as follows:
- In the Chinese Room Experiment Searle uses the formal properties of symbols to produce output symbols, but this does not generate understanding.
- There is nothing more that a computer program has than what Searle has inside the room.
- So a computer program doesn’t understand what the symbols mean and therefore does not have a mind.
One would intuitively conclude that even though Searle can respond to Chinese inputs, he has no idea of what those inputs are about or the meaning of his response. This contradicts the Strong AI view that a machine can be said to literally understand the content it is manipulating, rather according to Searle it is only manipulating symbols. As a human can carry out any computer function without being said to understand the content, the machine can therefore not understand and is not a mind.
We’re all symbol manipulators!
The first objection to Searle’s Argument is from Robert Abelson, who’s response is titled “Searle’s argument is just a set of Chinese symbols”fn-abelson, I have to agree with Abelson, we express the vast majority of our thoughts to the world through the manipulation of formal symbols, both in Speech and Writing. For the most part we accept the symbolic and formal definitions of objects in our world and interact with them in a symbolic manner. Take people with a fundamental understanding of Mathematics, such people understand that the symbols “1 + 1” equals the symbol “2”, but very few people know why this is the case, and even fewer are able to prove that it is so. These people are still said to understand basic mathematics, but if Searle was correct they would be said to not really understand, but are rather playing a game of symbol manipulation. As Abelson puts it “We might be humble and give the computer the benefit of the doubt when it performs as well as we do.”
Additionally this does not mean that the necessarily machine understands as we do, but rather that there are situations in which our understanding and a machines symbol manipulation are carried out through the same formal process. And for that reason we ought to give them the benefit of the doubt, and to not assume that there is a significant difference if the same process is occurring.
In order for a mind to work, either formally or in a manner that ours does, logical connectives are essential. In fact it is this tipping stone that determines what a mind is. Searle states that intentionality is the key factor and that machines only have intentionality in virtue of their use case. (Searle, 1980, p. 419) However we must examine the root of understanding in order to determine that what it means to be a mind resides in a much more fundamental process. Take for example Lewis Carroll’s “What the Tortoise Said to Achilles”fn-carrol, in which a tortoise presents the following argument:
A: “Things that are equal to the same are equal to each other”
B: “The two sides of this triangle are things that are equal to the same”
Therefore Z: “The two sides of this triangle are equal to each other”
The tortoise then asks Achilles if the conclusion logically follows from the premises, Achilles agrees. The tortoise then asks if a reader could reject the premises, Achilles says yes, such a reader could exist. The tortoise asks Achilles if a reader might accept both the premises and but reject the conclusion. Achilles says such a reader might exist, and the tortoise asks Achilles to prove the argument valid to such a reader. Achilles then adds the premise,
C: “If A and B are true, then Z must be true”.
The tortoise agrees to this new premise but still rejects the conclusion. Such a reader could never be convinced of the argument, as Achilles only recourse is to continue to add premises, which will have no effect.
This ability to make logical leaps is key, as a system can just as easily reject all aspects of an argument or system altogether. Intelligence comes for a unique ability to reason about inputs and logically determine the proper output. This ability to make logical connectives is essential in a minds operation, and furthermore this characteristic is integrated at the physical level (in the form of logic gates), so it can be said that computers physically understand the logical connectives just as we physically understand our senses on a physical level. This also divorces content from the Boolean aspects of that content, it does not matter if the subject matter is well know, but purely the ability to reason about it. This is a much better way to determine a mind, as opposed to Searle’s intentionality, which is unknown if we even have original intentionality, and this question is inherently unresolvable. A logical connective criteria therefore makes much more logical sense, and also is something that can be worked towards, which unfortunately original intentionality cannot.
Some Possible Objections
In response to the first objection mentioned to Searle’s Chinese Room, one might argue that it does not matter than machines and humans both sometimes operate in formal manners, as Humans have a possibility of going further, into the realm of true knowledge as opposed to only looking at formal symbols.
To rephrase this view, if a human only operates formally, but the potential of understanding exists then it is a mind. In order to know that potential exists, the mind must actualize on that potential, and therefore a mind is any thing that shows true understanding in any domain, and the amount of understanding or range of understanding is irrelevant. Therefore Logical connectives fit this description as outlined above, and allows for true understanding of one domain, which satisfied the potential of true understanding in the nature of the machine itself.
Just because logical connectives are more fundamental does not mean that they are necessarily the tipping point in determining a mind.
This objective misses the point with logical connectives. It is impossible for a system to understand content without understanding a logical connective. In order to understand a systems place in the world, one must understand if I push the block, then the block will move. Or that the sky is blue, and the sky is above me. Without such abilities the world is a stream of inputs that make no sense, and cannot be structured to uncover any truth in the world. In order to at a very basic level, make sense of inputs and develop core logical connectives are needed. Luckily they are the one things computers can be said to truly understand in virtue their physical nature.