The Outward Face

Updates Blog

Drawings Gallery

Writings & Rants

Lyrics & Poetry

Netherworldly
Horrors!

Chat & Contact

Forum of Paroxysms

Ateisma í
Føroyum

Consideration of Critiques

A vast number of varying critiques have been issued to challenge the Chinese Room argument. Some nothing more than mere recycled versions of old already refuted arguments and other genuinely original. Searle himself dealt with those he considered to be the most common at the time in his original article[1] and addressed some of the late-comers in exchanges and the like as they cropped up. As a serious mention of each and every one of these critiques would be far too lengthy and purposeless to include in this investigation, I shall choose to consider a selected few on no other criteria than my own personal estimation of relevance to this treatise and its underlying problem.

The Robot Reply considers the proposition of putting the computer in question inside a robot. The computer would then not merely be receiving formal symbol input and spewing out formal symbol output, but in fact, also be operating the robot to such an extent that it were programmed to observe the world with its eyes, taking the form of cameras attached to it, handle any physical object with its robot arms, hear through microphones, and even have a sense of smell through air-particle measuring devices etc. In short the robot could be programmed to do anything you would like and to sense the world through its gadgetry and interact with it accordingly.

The argument then poses the question of whether this particular robot could not be said to have understanding and states of cognition. Searle however is not so easily suaded. Because while The Robot Reply recognises that formal symbol manipulation is not a sole prerequisite of cognition, it fails to realise that adding such causal relations to the outside world in form of motor capacity and sensory perception is no addition, neither to understanding nor to intentionality, on part of the actual program governing these various affairs.

As an illustration Searle asks us yet again to consider the man inside the Chinese Room. The only alteration would be that some of the symbols the man now receives stem from the robot’s various apparatus and some of the outgoing symbols are encrypted commands that govern - for instance - the robot’s locomotion. From the man’s point of view there is no change in his situation. He is still as oblivious to what is actually going on outside of the room as ever and can therefore not be said to have any understanding. The man is, as always, merely manipulating the syntax of formal symbols without any causal semantics.

The Presumption Reply is a critique appearing in the article “Artificial Intelligence and the Chinese Room: An Exchange” by Elhanan Motzkin[2] in which he points out that our only knowledge of the man inside the Chinese Room not understanding Chinese stems from the fact that we assume this in advance. And while it is perfectly reasonable that the man might not understand Chinese, it is disastrous for any argument to presume what it is supposed to prove to begin with.

Technically it is an informal logical fallacy called ‘begging the question’, ‘circular reasoning’, or ‘petitio principii.’[3] Searle seems to take this objection lightly however, because it simply does not refute the fact that if the man did not understand Chinese when he first entered the room, an instruction manual for manipulating symbols solely by virtue of their shapes would scarcely suffice in teaching the man to understand it. One might want to inject that Searle has done nothing to prove that the man does not, but that would just be a bad case of spouting Negative Proof, and we shall not adhere to such silliness.

Curiously though Searle seems to demand an almost equally impossible proof of consciousness in addressing proponents of The Systems Reply when saying:

‘(…) the systems reply simply begs the question by insisting without argument that the system must understand Chinese.’ [4]

That is precisely why Turing argued that if a machine could indeed pass his test, how could we ever justify saying that it is not thinking?[5]  Searle on the other hand thinks that the crux of the question is:

‘Under what conditions in fact does a system have understanding regardless of how, or if at all, observers outside the system can tell?’[6]

But personally I find it quite dubious if Searle really answers this question satisfactorily himself.

The Zombies Reply, as seen in the technical report titled “Zombies in Searle's Chinese Room (…)” written by Dr. Louis J. Cutrona, Jr.,[7] posits that as a consequence of John R. Searle’s theory one could construct

‘Gedanken audioanimatronic mechanisms that are indistinguishable from real live people.’[8]

This would constitute a curious proof that such a thing, as a Zombie in the philosophical sense, i.e. a human being identical to its peers save for its absence of consciousness, would be an existing phenomenon (or at least a definitely possible phenomenon). Cutrona goes on to argue that the foregoing in turn leads to solipsism, because we would have no criterion on which to base our supposition that minds even exist aside from our own. Cutrona describes Searle’s wriggling out of the dreaded solipsistic trap from whence there is no return thusly:

‘Searle sees the danger, and tries to avoid it with the more-or-less standard philosophical meta-argument adduced to fight the temptations of solipsism, to wit: we obtain support for the surmise that others are conscious from the knowledge that they have a similar biological makeup to our own. But when Searle makes the argument in this context he uses it not to invalidate or avoid solipsism per se, but only to provide a carve-out, an exception that saves him from having to apply his arguments to the question of the possible existence of other human minds, but allows him to apply those arguments to the question of the possible existence of computer minds. It’s cheating.’[9]

To the best of my knowledge Searle has yet to issue any response to this particular piece of criticism, but a scrutinous comparison of how Searle has replied to other critiques might yield plausible approximations of his stance. In particular The Robot Reply, The Systems Reply and The Other Minds Reply might be somewhat applicable. In general there seems to be a problem with ambiguity in Searle’s writings and Searle’s insistence that:

‘(…) only something that has the same causal powers as brains can have intentionality.’ [10]

Does not really lend to any additional enlightenment of just what would cause these semi-mysterious causal powers to begin with. Just what it is the brain is supposed to possess that makes it so impossible to duplicate on a formally symbolic basis.

Were I less benevolently inclined I would be tempted to attribute a latent dualism to Searle. As of now, though, I have to be content that possibly Searle did not elaborate because that would be another discussion altogether and he thought it sufficient to simply disprove strong AI and The Turing Test without providing a replacement hypothesis of his own. I had hoped to be able to respond to Cutrona’s criticisms, as I believe Searle would have, but I am unable to deduce anything from his writings that Cutrona has not already refuted – or at least has attempted to refute - in his report.

The Brain Simulator Reply[11] is somewhat reminiscent of The Robot Reply in that it accepts the premise that formal symbol manipulation is not sufficient on its own to ensure cognitive states. The Brain Simulator Reply thereby proposes that we program a simulation that does not merely represent knowledge of understanding Chinese but perfectly simulates

‘(…) the actual sequence of neuron firings at the synapses of the brain of a native Chinese speaker when he understands stories in Chinese and gives answers to them.’ [12]

The question then is, what would be the difference between this program and the program that a Chinese-understanding human brain is running? Searle posits that this particular example is best likened to the aforementioned man in The Chinese Room operating water pipes with valves connecting them, where each water connection corresponds to a synapse in the brain. Thus simulating the correct sequential neural-firing by the man opening and closing the faucets designated in the English instruction manual. This man would thereby produce the same reply as that of the brain, which this construction was modelled by, while still maintaining ignorant bliss. Furthermore Searle reminds us that the conjunction of the water pipes and the man in question could not be said to understand either, (even if the man alone could not) since in principle the man could memorise the entire network of water pipes, valves, and faucets and thereby perform the same function simply by calculating the correct water-path in his mind.[13]