Madralin, 10-19 August 1999
Meaning and Experience
In this paper I will concentrate on one aspect of meaning: its relation to perception. The paper has four parts:
(1) I will first say something about Quine's view on the public nature of meaning and its consequences for the relation between meaning and the mind.
(2) Next, I will discuss an important problem in Quine: the connection between perception and meaning.
(3) Thereafter I will discuss two alternative solutions proposed by Donald Davidson: first his maxim of maximizing agreement, secondly his idea of triangulation.
(4) Finally, I will set forth my own view. My view leads to a circle, but I will argue that this is a good circle. I will end my paper with some remarks on the consequences of this view for our conception of consciousness.
Philosophers and linguists have always said that language is a social institution. They have, however, immediately forgotten this and have adopted notions of meaning that are not publicly accessible and where it remains unclear how such entities are grasped by us.
Quine seems to have been the first to take the public nature of language seriously and explore its consequences for meaning and communication. He begins with a situation where two people, each with their own language and view of the world attempt to communicate. They have no previous translation manual to fall back on, no grammar and dictionary, but must carry out "radical translation," where they try to establish a grammar and a dictionary that they test out by observing one another's behavior.
Quine specifies two constraints that translation manuals have to satisfy. First, a condition on observation sentences, that is sentences which the other person assents to or dissents from only in certain observational circumstances. Such sentences should be translated into sentences that we assent to or dissent from in similar circumstances. Secondly, a principle of charity. Sentences which the other person accepts should not be translated into sentences which we regard as absurd, and sentences which the other person dissents from, should not be translated into sentences that we regard as banal.
As Quine points out, several different translation manuals can satisfy these constraints. Given that these two constraints are all the evidence there is for correct translation, Quine concludes that translation is indeterminate; there are several translation manuals between two languages, and they are all correct.
Quine's conclusion depends upon two premisses: First a premiss that is fairly generally accepted: the Duhem-Poincar -Quine thesis of under-determination of physical theory by evidence. In the natural sciences we regard only one of these theories as true, the others as false. In order to get indeterminacy, we need in addition a premiss concerning meaning. There are here at least three candidates. First, two that are explicitly stated by Quine: first a verificationist theory of meaning la Peirce, and secondly physicalism: an ontological thesis to the effect that while there are physical states, there are no such things as meanings. There is, however, also a third candidate that will do the job and which seems to me to be more plausible and go more to the heart of Quine's insights into the public nature of language. This is what I have called the MMM thesis, the thesis that meaning is man-made: Meaning is a product of human interaction and is based on the same behavioral evidence as we use to find out what a person means. That is, where the evidence leaves off, meaning leaves off. There is no realm of meaning which like physical nature was there before we started to interact and is waiting to be discovered. Where the behavioral evidence for translation leaves off, there is nothing more which different translations could be right or wrong about.
This has radical consequences for our view on the mind and the relation between mind and language. We will turn to these in the fourth part of this paper. Let us now consider Quine's two constraints on translation.
I find the second of Quine's constraints, the principle of charity, well justified. It reflects an old and well-established hermeneutic principle and Quine supports it with good arguments. The first constraint, however, the observation constraint, I find very problematic. Not because observations are irrelevant to understanding and translation -- their relevance will be a main theme of this paper -- but because Quine defines observations in terms of the behaviorist notions of stimulus and response.
My problem is not the usual objections against behaviorism, such as those of Chomsky against Skinner. Chomsky's objections are pretty irrelevant against Quine's more discerning behaviorism. My problem is that I find that Quine through his focus on stimulus and response has forsaken the public nature of language. Stimuli can be empirically studied, but they are not publicly accessible. And according to Quine's fundamental insight the emergence and development of language, the learning of language and the use of language in communication must all be founded on publicly available evidence. In my daily life, where I learn and use language, I cannot observe the sensory stimuli of others. And I have never observed my own. How can I then compare the stimuli of others with those of my own, as Quine requires?
During the years 1961-64, when Quine and I were colleagues, we discussed these problems, and in 1974, when we were together for a year in Oxford, I gave a public lecture, "Meaning and Experience," where I expressed my objections.
Quine had his reasons for concentrating on stimuli. First, stimuli can be described in a fairly precise way. Secondly, and that was most important for Quine, we cannot determine through observation which objects other people perceive; what others perceive is dependent upon how they conceive of the world and structure it, and that is just what we are trying to find out. When we study communication and understanding, we should not uncritically assume that the other shares our conception of the world and our ontology. If we do, we will not discover how we understand other people, and we will not notice the important phenomena of indeterminacy of translation and of reference.
3. The early Davidson. "Maximize agreement" 
Inspired by Quine, Davidson has developed a theory of "radical interpretation." In order to compare this theory with Quine's theory of translation, let us transform Davidson's theory into a theory of the conditions a correlation between two languages must satisfy in order to be a translation. We see then that Davidson differs from Quine on the following two points:
(1) Davidson replaces Quine's systematization via grammar with a systematization by means of Tarski's theory of truth. This change reflects the fact that the systematization concerns semantics: one wants to see how the semantic features of complex expressions depend upon the semantic features if their component expressions. More accurately: given that one knows, through behavioral evidence, which sentences a person assents to -- that is, regards as true -- and which sentences he dissents from -- regards as false -- we try to divide these sentences into parts, that is words, and to find extensions and references for these words that make most of the sentences that the person assents to true and most of the ones he dissents from false.
This proposal by Davidson could be looked upon as applying Tarski upside down. While Tarski assumed that we know the extensions and references of the smallest components and built up from there, Davidson starts with the truth and falsehood of sentences and tries to determine the parts and their semantic features from there.
I regard this first proposal of Davidson's as an improvement upon Quine. And Quine has accepted it.
Davidson's second proposal is to replace Quine's two constraints on translation that I outlined above with one single constraint, a sweeping principle of charity that he expresses as a maxim: maximize agreement. That is, try to correlate the two languages in such a way that the sentences to which the other person assents are correlated with sentences to which we assent, and sentences from which he dissents are correlated with sentences from which we dissent.
This simple constraint was the only condition Davidson put on translation in his early writings. He had recognized Quine's problems in connection with reception, and he formulates his constraint without any appeal to perception. In Davidson's early writings there is no mention of perception as one of the factors one has to take into account when one interprets somebody else.
It would certainly simplify matters if perception did not have to enter the picture. However, I found Davidson's thesis problematic, and in 1973 I discussed with Davidson the indispensability of perception for interpretation. I used the following example of a rabbit behind a tree, and was curious as to how he would handle it:
I am together with a person who speaks a language which I do not know, but would like to learn. He frequently uses the phrase 'Gavagai' and I have formed an hypothesis that it has to do with rabbits. While we are in a forest and I note a rabbit I try out the phrase 'Gavagai'. However, my friend dissents. According to Davidson's thesis of maximizing agreement this would be a reason against my hypothesis that 'Gavagai' should be translated by 'Rabbit'. If I now discover that there is a big tree between my friend and the rabbit, I immediately have an explanation for our disagreement: I take it for granted that my friend, like me, is not able to see through trees and that he therefore does not think that there is a rabbit there. I even take my friend's dissent as confirming my hypothesis, I do not expect him to believe that there is a rabbit there.
The thesis of maximizing agreement hence has to be modified into "Maximize agreement where you expect to find agreement." Here both of Quine constraints on translation comes in, the observational and the principle of charity. Interpretation recapitulates epistemology, and Quine's two principles reflect the two main ingredients in epistemology: perception and reason.
The rabbit-behind-the-tree example illustrates how the perceptual situation which we assume the other to be in may be decisive for the beliefs we ascribe to him and thereby for how we interpret and translate what he says. When Davidson was confronted with this example, he agreed that perception is important for translation and interpretation. In his later writings he gives prominence to this.
How has then Davidson incorporated perception in his theory of meaning? He does it by help of a process that he calls 'triangulation'. 'Triangulation' was a word Donald Davidson introduced in 1982, in the second to the last paragraph of "Rational Animals," for the process of language learning. From 1989 on he has developed the idea further in a number of articles. He depicts this process as a triangle where one line goes between the learner and the object, one between the teacher and the object, and one between the learner and the teacher:
This triangulation process was exactly what Quine argued for in "Ontological Relativity" from 1969, where the triangle is spelled out in detail:
IP5,5 The learner has now not only to learn the word phonetically, by hearing it from another speaker; he also has to see the object; and in addition to this, in order to capture the relevance of the object to the word, he has to see that the speaker also sees the object. IP 
From the very beginning of Word and Object (1960) Quine stresses the importance for language learning of publicly observable uttering of words in the presence of things and events that we observe and take our fellow users of language to observe. In the first sentence of the preface to the book Quine stated his view in a nutshell:
IP5,5 Language is a social art. In acquiring it we have to depend entirely on intersubjectively available cues as to what to say and when. Hence there is no justification for collating linguistic meanings, unless in terms of men's dispositions to respond overtly to socially observable stimulations. (Word and Object, p. ix) IP On the first page of the main text of the book he reiterates: IP5,5 Each of us learns his language from other people, through the observable mouthing of words under conspicuously intersubjective circumstances. (Word and Object, p. 1) IP
He elaborates on this in the next sentence:
IP5,5 Linguistically, and hence conceptually, the things in sharpest focus are the things that are public enough to be talked of publicly, common and conspicuous enough to be talked of often, and near enough to sense to be quickly identified and learned by name; it is to these that words apply first and foremost. IP Quine saw this clearly from the very beginning of Word and Object. It is a pity that he so quickly turned to the stimuli.
How does Davidson solve the problems of perception? He turns to a causal theory of perception and says that the object, event or situation an expression relates to is the last common cause in the two infinite causal chains that lead to the sense organs of the teacher and learner in the learning situation.
I have great problems with this solution. Generally, I feel unease when adherents of causal theories of perception or of reference speak as if causality can individuate objects. Thus, for example, when Gareth Evans developed his causal theory of reference, I was in Oxford and I gave a talk on reference with Evans as commentator. I took the opportunity to include in my talk some objections against his view that a name refers to the object which is the main cause of the information that I connect with the name. Alfred Ayer, who was in the audience, said in the discussion that if Evans theory were true, the reference of most of Ayer's names would be his nurse. She was the main source of the information he connected with these names.
The problems become even worse when one speaks of causal chains, as does Davidson. The events with which I am familiar have a multitude of causes. One should rather talk about causal trees. One then sees immediately that the notion of the last common cause makes little sense. Which of the many junctions between our two causal chains is the last one?
I contend that in order to talk about objects one needs complex mental structures, such as Husserl's noemata, which are closely connected with linguistic meaning. Perception has propositional content and can thereby serve as evidence for judgments. Perception and language are jointly dependent upon intersubjective adaptation. Husserl studied this adaptation and concluded that "even what is straightforwardly perceptual is communal."
Since 1986 Quine has relinquished his "proximal" approach to translation. He has seen that the stimuli do not play a central role, and he has now a "distal" theory, a theory of triangulation, such as he originally conceived of it at the beginning of Word and Object. Quine has not yet written much about this new theory. Already in his Paul Carus lectures in 1973 he observed the social nature of perception:
IP5,5 Perception being such a private business, I find it ironical that the best evidence of what counts as perceptual should be social conformity. I shall not pause over the lesson, but there is surely one there. IP 
In his later writings Quine stresses the importance of reification. Thus he writes in a manuscript from 1990 that he has permitted me to quote: IP5,5 Our sophisticated concept of recurrent objects, qualitatively indistinguishable but nevertheless distinct, involves our elaborate schematism of intersecting trajectories in three-dimensional space, out of sight, trajectories traversed with the elapse of time. IP 
Reification is a process which is fundamental for our theories and at the same time depends upon them. It is in my opinion also crucial to language learning and communication. When we try to understand another person, we have to make assumptions concerning which objects he perceives and which properties he takes them to have, and thereby concerning his theories and the structure of what he perceives, his expectations, or his noema, to speak with Husserl. As the process of understanding progresses, these assumptions may be modified in the light of publicly accessible evidence, the way Neurath modified his ship. Our understanding always remains tentative. There is no dry dock, where we can build up our understanding from a firm, non-intensional basis, such as stimuli or causality. We are hence moving in a circle, we use assumptions concerning perception to understand language, and we use our tentative understanding of language to improve our assumptions concerning perception. However, this is no vicious circle. We are just extending Neurath's boat simile from science to translation and interpretation.
This important change in Quine's view, from a proximal to a distal theory of language learning, does not eliminate indeterminacy of translation. On the contrary, the indeterminacy becomes greater. We must from the very beginning make assumptions about which objects a person perceives, and these assumptions are far more under-determined than assumptions about which stimuli he receives. As I will argue below, we have here partly to do with under-determination, where there is something to be right or wrong about, similar to what we have in the natural sciences, and partly we have indeterminacy.
Independently of this, the indeterminacy of translation is, on the other hand, also smaller than it might seem from Word and Object. I think that a large number of human activities, practices and customs play a role on communication and therefore -- according to the MMM thesis -- contribute to establishing the meaning and reference of linguistic expressions. All this evidence must be brought into the study of meaning as constraints on translation and interpretation, not just assent and dissent.
We are born with a lot of dispositions and abilities to register certain features of the world and extrapolate them inductively. These abilities are decisive for perception and action and also for the learning of languages. Through the learning of languages these abilities become more developed and refined. This in turn facilitates further language learning.
When we start learning a language, we associate linguistic expressions with our different anticipations and other dispositions on the basis of publicly accessible evidence. In the case of perception we have many expectations and tacit assumptions that can be confirmed or disconfirmed. There is therefore here much under-determination, but little indeterminacy. However, as soon as we extrapolate from the perceptual realm into more theoretical domains, the interplay between theory and meaning becomes more pervasive. In theoretical areas indeterminacy of translation and of reference thereby becomes more prominent. In these areas it is important that we not regard meaning as something that first exists in our mind and then gets expressed through language. There are no proto-meanings in our mind, as Fodor and many others maintain. There are intimate and interesting connections between mind and meaning. But we get a wrong picture of these connections if we fail to take seriously the public nature of language.
3 For more on this, see my "Indeterminacy and Mental States," in Robert Barrett and Roger Gibson (eds.), Perspectives on Quine. Oxford: Blackwell, 1990, pp. 98-109. Further my "In What Sense is Language Public?" in Paolo Leonardi and Marco Santambrogio, eds., On Quine. Cambridge: Cambridge University Press, 1995, pp. 53-67. [back]
5 Donald Davidson, "Rational Animals." Dialectica 36 (1982), pp. 317-27. Reprinted in E. LePore and B.McLaughlin, eds., Actions and Events: Perspectives on the Philosophy of Donald Davidson. Oxford: Blackwell, 1985, pp. 473-80, and several other anthologies. [back]
6 "The Conditions of Thought." Le Cahier du Coll ge International de Philosophie. Paris: ditions Osiris, 1989, pp. 165-171. "Meaning, Truth and Evidence." In R. Barrett and R. Gibson, eds., Perspectives on Quine. Oxford: Blackwell, 1990, pp. 68-79. "Epistemology Externalized." Dialectica 45 (1991), pp. 191-202 (originally in Spanish, 1990). "Three Varieties of Knowledge." In A. Phillips Griffiths, ed., A.J. Ayer: Memorial Essays: Royal Institute of Philosophy Supplement, Vol. 30. Cambridge: Cambridge University Press, 1991, pp. 153-166. "The Second Person." Midwest Studies in Philosophy 17 (1992), pp. 255-267. "Locating Literary Language." In R.W. Dasenbrock, ed., Literary Theory After Davidson. University Park, PA: Pennsylvania University Press, 1993, pp. 295-308. [back]
10 Quine "From Stimulus to Science," lecture at Lehigh University Oct. 15, 1990, page 21 of the manuscript. See also Quine's book From Stimulus to Science, Cambridge, Mass.: Harvard University Press, 1995, pp. 35-40. Similar observations are found also in other of Quine's later writings, for example in his "Reactions," in Paolo Leonardi and Marco Santambrogio, eds., On Quine, Cambridge: Cambridge University Press, 1995, p. 350. [back]