3 ms·

Everyone in this thread keeps claiming that the Chinese Room argument is bad. Nobody says why.

dghf 9y ago
Dennett does a pretty good job (to my mind) of undermining it in Consciousness Explained: you can read part of the relevant section on Google Books [0]. Or this SMBC [1] summarises the argument pretty well, if somewhat sarcastically. [0] https://books.google.co.uk/books?redir_esc=y&id=d2P_QS6AwgoC&q=chinese+room#v=snippet&q=chinese%20room&f=false https://books.google.co.uk/books?redir_esc=y&id=d2P_QS6AwgoC... [1] https://www.smbc-comics.com/comic/the-chinese-room https://www.smbc-comics.com/comic/the-chinese-room
Analemma_ 9y ago
Searle does not prove that the distinction between "actually understanding Chinese symbols" and "simulating the ability to understand Chinese symbols" is a real one. Physicalists, including me, believe this is just a confusion generated from our inside view of our consciousness, and that "really understanding" is just something a computer program thinks it can do once it gets complex enough to be conscious. At the bottom level it's still just pushing symbols around via syntactic rules (in the case of the mind, natural law).
- noam87 9y ago
  > "really understanding" is just something a computer program thinks it can do once it gets complex enough to be conscious. At the same time, consciousness might not be a requisite of higher intelligence at all; it could merely have been evolutionarily advantageous early on in the development of complex brains because of our natural environment... it's hard to imagine an intelligent animal with no "me" program doing very well. But maybe a digital intelligence (one that did not evolve having to worry about feeding itself, acquiring rare resources, mating, communicating socially, etc.) would have no use for a central "me" program that "really experiences" things. Such a creature is kind of eerie to think about.
hackinthebochs 9y ago
Searle's argument depends on our intuition of the distinction between syntax and semantics, but this distinction seems very outdated to me. It's an easy intuition to have in 1980, but its hard to defend in 2018. It's true that syntax itself isn't enough for semantics, but syntax is a component of a system that recognizes semantics, i.e. low level rule based operations that in aggregate entail semantics. So to say that because computers operate based on syntactic structures means the computer can never understand semantics is just a mistake on many levels.
dvt 9y ago
Anyone that claims it's bad/outdated/unequivocally wrong probably doesn't have the requisite background to make such a claim, let alone any academic training in the field. The Chinese Room is still studied and debated vigorously. For example, Baggini thinks that Searle's Chinese Room is a knock-down argument against functionalism. Dennett, on the other hand, thinks it's bunk.
mocrath 9y ago
Because the CRA is based on computers, which are based on math, anyone can know why the CRA is inconclusive -- based on facts, not opinions. The theory of Turing machines (TMs), programs, and universal TMs explains objectively why in the CR there is a 2nd computation (which is instantiated via Searle's "universal" one). (E.g., see https://en.wikipedia.org/wiki/Universal_Turing_machine https://en.wikipedia.org/wiki/Universal_Turing_machine) Basically, the Searle-UTM has TWO inputs, (1) the program and (2) the Chinese. The Searle-UTM itself processes the 1st, which is in English(!!), which in turn instantiates the program's TM per the theory. The program's TM thereby processes the Chinese symbols, which is why Searle, as merely the program's agent, can (and must!) process them purely formally. The details are further explained and illustrated here: https://chineseroom.info https://chineseroom.info This also explains why the Systems Reply is the most popular type of CRA rebuttal: it backed by TM theory, even if many don't know the details. Note that although this refutes the CRA, it does not prove the Strong AI (computationalism) thesis, which remains open.

dghf 9y ago

Dennett does a pretty good job (to my mind) of undermining it in Consciousness Explained: you can read part of the relevant section on Google Books [0]. Or this SMBC [1] summarises the argument pretty well, if somewhat sarcastically. [0] https://books.google.co.uk/books?redir_esc=y&id=d2P_QS6AwgoC&q=chinese+room#v=snippet&q=chinese%20room&f=false https://books.google.co.uk/books?redir_esc=y&id=d2P_QS6AwgoC... [1] https://www.smbc-comics.com/comic/the-chinese-room https://www.smbc-comics.com/comic/the-chinese-room

Analemma_ 9y ago

Searle does not prove that the distinction between "actually understanding Chinese symbols" and "simulating the ability to understand Chinese symbols" is a real one. Physicalists, including me, believe this is just a confusion generated from our inside view of our consciousness, and that "really understanding" is just something a computer program thinks it can do once it gets complex enough to be conscious. At the bottom level it's still just pushing symbols around via syntactic rules (in the case of the mind, natural law).

noam87 9y ago
> "really understanding" is just something a computer program thinks it can do once it gets complex enough to be conscious. At the same time, consciousness might not be a requisite of higher intelligence at all; it could merely have been evolutionarily advantageous early on in the development of complex brains because of our natural environment... it's hard to imagine an intelligent animal with no "me" program doing very well. But maybe a digital intelligence (one that did not evolve having to worry about feeding itself, acquiring rare resources, mating, communicating socially, etc.) would have no use for a central "me" program that "really experiences" things. Such a creature is kind of eerie to think about.

hackinthebochs 9y ago

Searle's argument depends on our intuition of the distinction between syntax and semantics, but this distinction seems very outdated to me. It's an easy intuition to have in 1980, but its hard to defend in 2018. It's true that syntax itself isn't enough for semantics, but syntax is a component of a system that recognizes semantics, i.e. low level rule based operations that in aggregate entail semantics. So to say that because computers operate based on syntactic structures means the computer can never understand semantics is just a mistake on many levels.

dvt 9y ago

Anyone that claims it's bad/outdated/unequivocally wrong probably doesn't have the requisite background to make such a claim, let alone any academic training in the field. The Chinese Room is still studied and debated vigorously. For example, Baggini thinks that Searle's Chinese Room is a knock-down argument against functionalism. Dennett, on the other hand, thinks it's bunk.

mocrath 9y ago

Because the CRA is based on computers, which are based on math, anyone can know why the CRA is inconclusive -- based on facts, not opinions. The theory of Turing machines (TMs), programs, and universal TMs explains objectively why in the CR there is a 2nd computation (which is instantiated via Searle's "universal" one). (E.g., see https://en.wikipedia.org/wiki/Universal_Turing_machine https://en.wikipedia.org/wiki/Universal_Turing_machine) Basically, the Searle-UTM has TWO inputs, (1) the program and (2) the Chinese. The Searle-UTM itself processes the 1st, which is in English(!!), which in turn instantiates the program's TM per the theory. The program's TM thereby processes the Chinese symbols, which is why Searle, as merely the program's agent, can (and must!) process them purely formally. The details are further explained and illustrated here: https://chineseroom.info https://chineseroom.info This also explains why the Systems Reply is the most popular type of CRA rebuttal: it backed by TM theory, even if many don't know the details. Note that although this refutes the CRA, it does not prove the Strong AI (computationalism) thesis, which remains open.