From Stimulus to Science. – book reviews
by W. V. Quine. Cambridge, Mass.: Harvard University Press, 1995. Pp. vi + 114.H/b. 15.50 [pounds sterling].
W.V. Quine’s focus on the principle of economy in science is long-standing and well-known. This slim volume finds Quine applying the principle to the exposition of his own philosophy. For here (if we do not count an Appendix on predicate functors), we have an integrated account of his views weighing in at just under one hundred pages. This is no mean feat, especially if one appreciates that this tour around one of the great philosophical mansions of the twentieth century takes in at least the following riches: a Quine’s eye view of the history of philosophy, from Thales to the present; a naturalistic account of the development of science and the learning of language; the nature of reification and speculations regarding its origins; quantificational logic and a predicate functor approach; the intersubjectivity of science and its roots; the status of mathematics and its “glorious reduction” (p. 41) to set theory; a precise account of denotation and truth; the argument from proxy functions to the indeterminacy of reference; the indeterminacy of translation; and a physicalist treatment of the mental, including the propositional attitudes.
Ideally, such economical treatments of complex and wide-ranging matters should negotiate both the Scylla of the stumped student and the Charybdis of the bored cognoscente. Quine sails clear of the first, placing his lucid and entertainingly precious style in the service of a tightly organized and well-developed exposition. Newcomers to Quine should benefit. This very success might encourage the thought that, whatever its virtues for the novice, the book will likely founder on the second danger. And it is quite true, and should occasion little wonder, that experienced readers will find most of Quine’s views (some of them now dating back six decades) largely unchanged. As Hume said a propos something else, surprises in these matters are not to be expected.
But in fact, against this background of great stability, an important shift has indeed taken place. It began with the publication of this book’s immediate ancestor (Pursuit of Truth (revised edition), Cambridge, Mass.: Harvard University Press, 1992), when Quine, after years of being bothered by the lack of a fully satisfactory account of the observation sentence, cut himself loose from a feature that had characterized his analysis since Word and Object (Cambridge, Mass.: MIT Press, 1960).
Before describing the nature of this shift, Quine’s reasons for it and its consequences, I would like to counter any first impressions that this might concern merely a small corner of his philosophy. It does not: the observation sentence is at the center of Quine’s views about how science comes to have the empirical content it does and about how language imbibes whatever meaning it has. Loosely put, an observation sentence is one such that a speaker’s response to it is closely tied to her immediate experience and, furthermore, agrees with that of others in her community upon their undergoing a similar experience. Because scientific theories issue in observation sentences, the first feature anchors science to experience, while the second secures its objectivity, or at least its intersubjectivity. Likewise, according to Quine, observation sentences make language learning possible. Their direct link to experience enables them to be learned without any prior language; and the shareability of the experiences that elicit them (or that elicit the same responses to them) makes possible the acquisition of language from others, as well as communication with them.
Quine, chary of the unreconstructed notion of experience, set out early on to rehabilitate it. He replaced it with what he now calls global stimulus: basically, the collection of all surface sensory receptors of the individual triggered at some moment. In terms of this, he defined a notion of a sentence’s stimulus meaning for a speaker: loosely, the class of global stimuli that would prompt the speaker’s assent when queried with the sentence. (For the full story, see Word and Object, [sections] 9, [sections] 10.) Translation was then viewed as constrained by (and, notoriously, only by) facts about stimulus meaning. These facts were the ones to which translation strove to be faithful. The goal of the translator, even if he did not actually conceive of his practice in this way, was to match up the stimulus meanings different speakers associated with the sentences of their language.
This account, operative for some thirty years, has now been abandoned. The sticking point for Quine is that global stimuli cannot be shared: the neurophysiologies of individuals vary so greatly that one cannot even plausibly speak of a homology of surface receptors. At the level of speaker neurophysiology, privacy reigns. (See Pursuit of Truth, [sections] 16, p. 20.) Thus, while one can continue to talk of the stimulus meaning of a sentence for a speaker, it remains unique to that speaker. Translation cannot consist in pairing intersubjective stimulus meanings across speakers, for these simply do not exist.
Quine has said that he had been reluctant to abandon a view of translation premised on intersubjective stimulus meanings because he could not otherwise see how to account for similarity of responses across speakers. In particular, Quine worried, we need to explain why, if two external events lead to perceptually similar, but private, global stimuli in one speaker, they would also lead to similar stimuli in another. (For a discussion of perceptual similarity, see pp. 17ff.) Natural selection, he now argues, provides the answer, for it induces a “preestablished harmony of perceptual similarity standards” (p. 21). Although the global stimuli of different individuals are not directly comparable, they are linked by virtue of being respectively located in two similarity spaces that are, thanks to natural selection, in mutual harmony.
There is, however, another reason, unacknowledged and less easily accommodated, for Quine to be reluctant to part with his old view. To appreciate it, one must first note that whatever it is that any new account will deem shared by, say, “There goes a rabbit” and “Gavagai”, the preservation of which licenses the translation of one by the other, it will not be characterizable in terms of experience: for experience, as Quine now explicitly presents it, is described only as unshareable global stimuli. What can be characterized at the level of experience are, for instance, the regularities holding between global stimulus and response that reflect my use of “There goes a rabbit”, and these are not the same as, or even comparable to, those regularities that likewise reflect the native’s use of “Gavagai”; certainly, if there is correct translation, then it cannot consist in identifying such regularities across speakers. But a sentence’s meaning is what is preserved by correct translation. It follows that if there are any determinate meanings at all, then the analysis of a sentence’s meaning will make no reference to experience qua neural stimulus. (Relatedly, we might put the matter this way: a sentence qualifies as observational in part because its translation is by and large determinate. Whatever under writes this largely determinate translation is not articulatable at the level of experience, construed as private global stimuli. In other words, what makes a sentence observational is not expressible in terms of experience.)
This conclusion, however, is in conflict with a substantive empiricism. For “two cardinal tenets of empiricism” remain “unassailable”, according to Quine: “One is that whatever evidence there is for science is sensory evidence. The other […] is that all inculcation of meanings of words must rest ultimately on sensory evidence” (“Epistemology Naturalized”, in Ontological Relativity and Other [Essays, New York, NY: Columbia University Press, 1969, p. 75). Yet, on Quine’s current conception, sensory evidence can play no crucial role in characterizing the shared meanings of language. Treated as global stimulation, it “is no part of the intersubjective business of semantics” (“In Praise of Observation Sentences”, The Journal of Philosophy, 4, 3, 1993, p. 114). Likewise, if evidence for science is shareable–and surely it is, for nothing would be accepted as scientific evidence unless it were intersubjectively available–Quine’s present view falls foul also of the first tenet of empiricism: for sensory evidence, construed as global stimuli, are private unto each speaker.
In sum, it appears that one cannot simultaneously (i) acknowledge that global stimuli are unshareable, (ii) identify experience with them, and (iii) affirm an empiricism that insists on a critical role for experience in the transmission of shared meaning and justification.
If (i) is now unavoidable and (iii) “unassailable”, then it seems that there must be some flexibility with regard to (ii). To see how this might be manifested, let us consider Quine’s account of observation sentences, now that he has abandoned shareable stimulus meanings. His present line is to say that a sentence is observational just in case it is, for each speaker, “keyed to a range of perceptually fairly similar global stimuli” and “all members of the language community are disposed to agree on the truth or falsity of such a sentence on the spot, if they have normal perception and are witnesses to the occasion” (p. 22; see also p. 44). What is it, then, for “Gavagai” to be correctly translated as “There goes a rabbit”? It is for both the native and me to be disposed to assent to, or dissent from, those respective sentences when we witness the same occasion.
But what exactly does this account of observation sentences consist in? Precisely which facts would it have their translation be beholden to? One answer is this: to those regularities that obtain between a speaker’s response and causally efficacious events in the environment. We have seen that, given (ii), this would leave no substantive role for experience to play in either the analysis of scientific evidence or linguistic meaning. But if we were to hold instead that such external events count as our “sensory evidence”, then Quine’s position could again accommodate the “two cardinal tenets of empiricism”: on this analysis, experience would once more be shareable as well as appealed to in stating the constraints on translation. Quine has insisted, however, that such events do not, for him, pass for experience. (See, for example, p. 18; Pursuit of Truth, p. 41.) This reconciling option, then, is not available.
There is another way, though, of interpreting Quine’s suggestion that a sentence is observational in so far as agreement in response obtains amongst speakers when they witness the same occasion. For we might also understand this to mean that speakers will agree when they perceive the same state of affairs, that is, when they are in the same perceptual state. Such states, in so far as they figure in the facts to which translation must do justice, are described using the mentalistic idiom: for example, perceiving that a rabbit is in the field. Given anomalous monism, which Quine embraces (see p. 87), each mentalistically characterized perceptual state is identical to some neurophysiological state, but still the latter can be picked out by a mentalistic predicate that resists reduction to non-mentalistic ones. Given this interpretation, if we now identify experience with these perceptual states, then we can render Quine’s account of the observation sentence compatible with his empiricism.
But is this position not very like (i)-(iii), except that instead of identifying experience with global stimuli, we have identified it with equally unshareable states of a speaker’s neurophysiology? If the combination of views was deemed incompatible before, then why not this one now? The explanation resides in the fact that the regularities that ground translation are linguistic in nature and so individuated according to the language we use to describe them. Quine’s official analysis makes available no means to refer to experience beyond those provided by the vocabulary of neurophysiology, and, in such terms, one cannot capture the regularities that underpin translation. On the present proposal, however, experience is characterized in mental terms and as such is shareable (even though the realizing neurophysiological states are not) and so capable of figuring in descriptions of what correct translation preserves. We can, after all, render (i)-(iii) (or a position very like it) consistent, if we are prepared to rely on the mentalistic idiom to describe experience.
This proposal fits well with one way of interpreting Quine’s current conception of the facts that are relevant to semantics. Let us first consider the situation of radical translation, which is designed to reveal the “conditions of sameness of holophrastic meaning in starkest purity” (p. 79). How does the radical translator proceed? To what does he attend? “A native utters something–`Gavagai’, of course–just when something salient has occurred, perhaps the scurrying of a rabbit” (ibid). What is salient for the native is determined by the translator through empathic projection into the native’s perspective. Empathy, which of course yields judgments about, not global stimulation, but a speaker’s perceptual state, is prominent in Quine’s account of radical translation. “Practical psychology”, he writes, “is what sustains our radical translator all along the way, and the method of his psychology is empathy” (Pursuit of Truth, pp. 46-7). Empathy is what helps the translator determine how the native’s response varies with his (mentalistically characterized) perceptual state, that is, to determine the facts that constrain translation. Empathy’s prominent role in radical translation reflects Quine’s judgment that, in communication and the learning of language, discernment of another’s perceptual state is paramount:
Empathy figures also in the child’s acquisition of his first
observation sentences. He does not just hear the sentence, see the
reported object or event, and then associate the two. […] In his as
yet inarticulate way he perceives that the speaker perceives the
object or event. (p. 89)
It might be objected that this cannot be Quine’s view for he would have translation constrained only by public facts. “There is nothing in linguistic meaning”, Quine has insisted, “beyond what is to be gleaned in overt behavior in observable circumstances” (Pursuit of Truth, p. 38). And what could be less observable than the mental lives of speakers, which, according to the above proposal, must be made reference to when articulating those facts that give translation whatever substance it has?
As the quotation but last makes plain, however, Quine does take perceptual states to be among the “observable circumstances” of a speaker’s use of language. That a speaker perceives some state of affairs is itself a perceptible state of affairs. Indeed, we find Quine affirming that, to some degree, “Perception of another’s unspoken thought […] is older than language” (p. 89; see also Pursuit of Truth, [sections] 24).
It might also be countered that none of the above suggests that Quine does, if only implicitly, take facts described using the mentalistic idiom to function as constraints on translation. This claim exhibits, the objection continues, a confusion between how a translator proceeds and what one’s analysis tells us he is really doing. The analysis, it might be objected, is free to make use of ideas with which the translator is completely unfamiliar, and the translator might well employ notions that will not survive a careful examination of what is actually going on in the course of translation.
Although a full discussion is impossible here, one might wonder whether the objection is in some tension with Quine’s apparent insistence that nothing is relevant to translation, to determining the facts of meaning, except in so far as it is attended to by speakers in the normal course of communication. “What is utterly factual”, Quine writes, “is just the fluency of conversation and the effectiveness of negotiation” (Pursuit of Truth, p. 43; see also From Stimulus to Science, p. 82). But “smooth conversation” (Pursuit of Truth, p. 47) surely involves responses on the part of one’s conversational partner that make sense in the light of, at least, what one judges her to be perceiving (as opposed, say, to responses that one can see she is caused to make by some distant event or, more proximally, by some global stimulus). Indeed, one might well ask how meaning could simultaneously be thoroughly public and yet at the same time grounded in causal, perhaps neurophysiology-involving, facts about which the ordinary speaker has no knowledge whatsoever.
All this, as Quine has said about his now abandoned thesis of the homology of surface receptors, “surely ought not to matter” (“Propositional Objects”, in Ontological Relativity and Other Essays, p. 157). I cannot pursue here the correctness of the suggestion that we can find peeking from under the sheets of Quine’s recent writing an appeal to mentalistically characterized perceptual states in his analysis of the constraints on translation. I think we can and furthermore, for the reasons sketched above, that we had better if he is to sustain a substantive empiricism. Such an appeal will have consequences, however. For one thing, it will become very difficult to see how the indeterminacy of translation can ultimately be defended. (For further details, see my “Quine and Observation”, 1996.)
COPYRIGHT 1997 Oxford University Press
COPYRIGHT 2004 Gale Group