Intensionality and Intentionality

Since Brentano, and later with Chisholm, there is a deep and privileged relation between intentionality and propositional attitudes like “believing”, “knowing”, etc., which involves the relation between intentionality (with a ‘t’) and inten­sionality (with an ‘s’).

Searle is well known among the authors who most insisted on the original char­acter of intentionality and on its role in the foundation of meaning, and his argu­ment of the Chinese Room has had a certain role in the disillusion of the perspectives of strong artificial intelligence. The Chinese room experiment suggests us imagining a English speaker inside a room where he receives from outside papers written in Chinese characters. The man has a system of rules of transforma­tion (into English) according to which the Chinese sentences are mapped into other sentences (to questions he will produce answers, and so on) in Chinese characters, although nobody explains him the meaning of the Chinese characters.

Searle’s famous paper has been published in 1980, and its main point is that intentionality is the characteristic feature that makes it impossible to say of a machine that it “thinks” or “understands” as we say it of a human being. Several years before the publication of Searle’s paper, Agazzi had presented a similar argument in an Italian journal (Agazzi 1967). He first develops a kind of “intentional stance” as Dennett (1987) developed later. He claims that com­plex machines impose us to use an intentional language; in fact, if we want to explain a robot’s performance on the ground of our knowledge of human activi­ties, it would be “not only possible, but also necessary to illustrate the prop­erties of machines in terms of human psychological predicates” (Agazzi 1967: 14). The use of anthropomorphic predicates or of an intentional language (“the computer has not enough memory...”, “it has still to learn how to make the program run...”) are therefore acceptable in a context of advanced robotics. But the use of anthropomorphic predicates cannot make us forgetting the fun­damental difference between the human sight and the recording of visual data from a sophisticated machine. What is a “plus” in human action “is denoted in philosophy with the term of intentionality, with which we refer to the fact that in the knowledge activity of a living being there is a kind of participation or iden­tification of the subject in front of the objects Therefore, we are brought to think that we cannot properly say that a machine sees an object, “the plus that accompanies the recording of the image in the living being is intentionality or its consciousness” 1967: 16).

Summarizing with a slogan we might say that the possibility to receive infor­mation is the common feature between a machine and a living organism, while the intentional activity is what differentiates them. Intentionality would become then, to take a term from the Theory of Evolution, the “missing link” of artificial intelli­gence that makes machines incapable to have that specific quality of human sub­jectivity consisting in thinking (see Agazzi 1991). Intentionality however is not translatable into operations of behaviours^[148] and its characterization has raised many debates. Searle considers intentionality as a characteristic feature of the brain; is this the best definition of intentionality? Or does it take for granted what one should argue for (that is, that intentionality is typical of living beings)? Agazzi looks for a different kind of definition: as “symptomatic characterization”: the most typical symptom of intentionality is the involvement of the living organism in its entirety (and not of subparts of it) in answering the signals that it perceives. We might also imagine to reconstruct a brain and all the details of a nervous sys­tem, but this would not amount to have intentionality. Only the activity of the liv­ing being in the environment may produce this peculiar feature, and the brain is just a part of it, a necessary but not sufficient part of it.

The definition of the original intentionality is what differentiates Agazzi from Searle (1980); for other aspects, Agazzi’s argument is very similar to the one pre­sented by Searle, although Agazzi arrives at a similar conclusion in a different way, through an analysis of the limits of formalisms, and in particular of Godel theorem (of which Agazzi made the first translation in a language different from German). We can assert the validity of Godel’s formally indecidable proposition (neither provable nor refutable in the formal system) only by metatheoretical con­siderations that are grounded on the meaning of the Godelian formula (and on the fact that it is satisfied by every interpretation of the symbols in the universe of natural numbers).

this further step (intentionality) seems to be in a structural way beyond the possibility of a machine, that is always in a condition very similar to the one of a man to whom somebody teaches the grammar and the syntax of a language of which he knows only the alphabet, without having been communicated the meaning of words; he would be able to build cor­rect sentences, or to recognize the correctness of some and incorrectness of others... but he could not distinguish the true from the false propositions (Agazzi 1967: 23)

Although not framed in the form of a Chinese room, Agazzi’s thought experiment is devised to make us imagine somebody endowed with syntax but not with semantics, exactly as Searle’s English speaker in the Chinese room. Is this the end of the matter? Difficult to say, and the discussion on the Turing test against which Agazzi and Searle fight, leaves open other possibilities, and Agazzi himself partic­ipated to the new discussion on the topic in a recent anthology.^[149] But this would bring us too far.

What seems to me clear enough from this short discussion of some of Agazzi’s ideas on meaning is twofold: on the one hand the particular consonance of Agazzi’s views with some of the most striking solutions to the main problems of contemporary philosophy of language and mind, and on the other hand the atten­tion he pays to some particular features, which makes his contribution always partly different from the standard solutions, suggesting either an update or an alternative to them.