Operationalism and the Perspectival Character of Scientific Knowledge
The Chinese Room thought experiment sums up Searle’s view about the limits of AI. Here it is in a very concise version:
Imagine a native English speaker [...] who knows no Chinese locked in a room full of boxes of Chinese symbols (a data base) together with a book of instructions for manipulating the symbols (the program).
Imagine that people outside the room send in other Chinese symbols which, unknown to the person in the room, are questions in Chinese (the input). And imagine that by following the instructions in the program the man in the room is able to pass out Chinese symbols that are correct answers to the questions (the output). The program enables the person in the room to pass the Turing test for understanding Chinese, but he does not understand a word of Chinese.[29]According to the prevailing interpretation, the main point of this thought experiment is that the purely formal, abstract, syntactical processes of an implemented computer program “could not by themselves be sufficient to guarantee the presence of mental content or semantic content of the sort that is essential to human cognition.”[30]
It is interesting to note that in 1967 Agazzi devised a thought experiment in some way similar to Searle’s. If we exclude “Leibniz’s Mill”,[31] Agazzi’s thought experiment is the first antecedent of Searle’s Chinese Room argument, anticipating for example Ned Block’s “Chinese Gym”.[32] Agazzi too argues against the Turing Test by pointing out that it is intrinsically impossible to determine intentionality through behaviour. He claims that a computer or a robot can imitate a human being only so far as syntax is concerned. Now, syntax is neither identical with nor sufficient for semantics; but minds have semantic contents and it is impossible to understand the meaning of a symbol without human intentionality:
it is quite reasonable to suppose that a person who does see may simulate the behaviour of a blind person (i.e.
there is no apparent behavioural experiment to discriminate a genuinely blind man from one who is not). On the other hand, if we accept for a moment that it would really be possible to equip a robot with such electronic devices as would allow it to simulate perfectly the behaviour of a seeing person (i.e. to avoid obstacles, to read, to recognize colours and figures, etc.), we may also imagine equipping a blind person (or simply somebody whose eyes were adequately sealed) with similar electronic devices, and teaching him/her to interpret the stimuli received through these electronic “sense organs” (stimuli that might be for example of a tactile character). In this condition our person would be able ex hypotesi to behave as if he/she were seeing, without seeing at all.[33]I think that Agazzi, at least to some extent and in certain respects, clearly anticipated Searle’s thought experiment. In my opinion, this anticipation is no accident, but depends on (at least) one important assumption shared by the two authors, namely the perspectival view of empirical knowledge.
Why, Agazzi asks, do we not say that a camera “saw” the object that is now on the photographic film? The difference is in “the way in which the image is present”:
when we feel compelled to state that a camera, differently from animals, does not “see” an object, we are referring to a “something more” which accompanies the pure recording of an image in the case of the animal, that is, to intentionality [...] The difference between recording an image on a photographic plate and seeing it does not consist in the fact that there is an image, but in the way in which the image is present; and it should be no wonder that such a way is not subject to further analysis by means of facts, for there could be a risk of falling into a regressum ad infinitum, as one might always formulate a question about the way the subsequent facts are given.[34]
Searle is not only clearly aware of the perspectival character of scientific knowledge; he also realizes that it is an intrinsic ingredient of human intentionality.
It is in this sense that intentionality is, in Searle’s words, “the feature of certain mental states by which they are directed at or about objects and states of affairs in the world”.[35] In particular, the Chinese Room thought experiment presupposes that intentionality has a necessary “aspectual” (that is, perspectival) character which would be very hard to ascribe to a computer or to a robot. Searle uses this claim against the thesis that intentionality is reducible to computability, when he argues that every intentional state has an “aspectual shape”, in the sense that it is directed at an object only “under an aspect”.[36]Given that they share the important thesis of the intentional-perspectival view of empirical knowledge, it is unlikely that it was a coincidence that in 1967 Agazzi relied on a thought experiment that was very similar to Searle’s in order to show the limits of AI. Indeed, the intentional-perspectival view of empirical knowledge is a strong reason for rejecting any strong reductionism. The fact that reality must be investigated from particular points of view which cast light on particular aspects of it, is sufficient to dissolve the reductionist illusion that there is a single privileged and fundamental level of description. Conversely, to assume a single privileged and fundamental level of description (which is the very heart of radical reductionism, whether idealistic or naturalistic) would be in stark contradiction with the abstractive, selective, partial, idealizational and theoretical character of any empirical investigation of reality, which requires thought-projects which are “subjective” or, more precisely, always guided by our purposes and values.
There is an obvious difference between Searle and Agazzi. Searle does not exclude the possibility in general of duplicating human intelligence but only the possibility of duplicating it by computational means. Agazzi, on the contrary, maintains that, from a scientific point of view, the question in general as to whether an automaton, besides being able to behave like a seeing entity, is or is not actually able to see, must remain open, even though “the onus probandi must be taken on by those who maintain this fact, for it is a matter of asserting “something more” than is necessarily implied by the robot’s behaviour.”[37]
However, this is a minor point, in the light of what we have been saying.
The most important difference between the two authors lies elsewhere. Agazzi’s solution of the problem of intentionality is consistent with the main tenets of his philosophy of science, and especially with the perspectival sense of objectivity. Searle’s position, on the other hand, is not internally consistent, since the perspec- tival (or “aspectual”) account which he rightly gives of the problem of intentionality is clearly incompatible with the main tenets of his philosophy.As is well known, Searle insists that intentionality is produced by the brain. Intentionality is for Searle a “biological phenomenon, and it is as likely to be as causally dependent on the specific biochemistry of its origins as lactation, photosynthesis or any other biological phenomena.”[38]
This claim, which sees the brain or some of its properties as an ultimate constituent of matter or reality, is inconsistent with the aspectual or perspectival character of intentionality. If an ultimate biochemical phenomenon existed independently of any particular point of view, if it existed in itself apart from our knowledge interests, it would be a kind of atomic, self-enclosed reality. But what we call the brain and its properties (including intentionality as a biological phenomenon) cannot exist or be understood apart from theoretical constructions of some type. The brain and its properties appear as biochemical realities only through the concepts, terms and technical apparatuses that define the viewpoint from which biochemistry investigates reality.
Searle’s claim that intentionality is produced by the brain is inconsistent with the perspectival character of scientific knowledge according to which nature can be known scientifically only from a potentially infinite number of perspectives or theoretical points of view. If a “part” of reality, such as a table, can be examined as a physical phenomenon as well as a chemical or a commercial one, it turns out to be a mechanical phenomenon only when it is considered from a point of view that takes into account only some of its properties such as force, mass and certain spatial and temporal relations.
If this is so, whenever physics and chemistry grasp certain law-like connections inherent to a phenomenon, it would be arbitrary to assume that the phenomenon is thereby fully explained, perhaps apart from a few minor details. It would also be arbitrary to assume that any other kind of explanation would necessarily clash with the one provided, for example, by biology. Recognition of this fact is sufficient to undermine all reductionist approaches, whether in terms of old-fashioned symbols and rules systems a la Turing, or in terms of distributed processing connectionist systems, or in terms of chemical- biological duplication a la Searle.On the contrary, consistently with this point of view, Agazzi writes that reductionism
always consists in giving to assertions, which are true in their own specific domain of scientific reference, a general scope which they do not possess. For instance, the sentence “the brain is a very powerful computer” is true, provided that it is meant that this is the case if the brain is considered from the point of view of processing speed and storage capacity, but it is clear that this sentence is not the whole truth about the brain, and it would therefore be incorrect to claim (as often happens) that the different brain functions can be reduced to algorithmic processes, even though they also undoubtedly have algorithmic aspects.[39]
To sum up, our brief comparison between Agazzi’s and Searle’s treatment of Turing’s test confirms that one of the most important aspects of Agazzi’s opera- tionalism lies in his attempt to connect intimately the main idea of operationalism with a theoretical-perspectival view of scientific knowledge.
In the rest of this paper I shall suggest some modifications which are needed in order to make Agazzi’s overall account more consistent.
4