Building Models - Part One
Introduction
For Quine, the relationship between sensory data and theory is indeterminate in the sense that our theories are under-determined by the evidence. According to Quine, one could construct many equivalent theoretical models which would be consistent with the available evidence, and, as such, the data does not lend any more credibility to one theory over another.
In fact, Quine took the argument one step further and said that the relationship between language and the world is indeterminate as well. Quine maintained there is an aura of inscrutability surrounding the way language attempts to make reference to the world such that one couldn't be sure in many, if not most, instances precisely what was being referred to in the context of any given word usage.
In effect, Quine believed there is, potentially, an indefinite number of ways of conceptually slicing up the world. As a result, one could not be certain that a given reference was slicing up the world in one way rather than another.
Quine augmented the ‘ inscrutability of reference’ idea with his thesis on radical translation. In essence, the theory of radical translation held that one could not translate, with any sense of conviction, from one language to another, remote, radically different language, such that the translation would be capable of generating a meaning expression in the former language that was truly equivalent with the expression in the language being translated.
One of Quine's famous examples, in this respect, concerned the expression "gavagai" which was uttered in contexts when a rabbit was present. According to Quine, one would have no non-arbitrary and objective means of ascertaining whether the proper translation of that expression should be, for example, "rabbit", "rabbit stages", "undetected rabbit parts", or any number of other possibilities that might concern contexts in which a rabbit was present.
Once again, because, for Quine, there are an indefinite number of ways of conceptually slicing up the world into different categories and so on, with different focal points and emphases, one could not be certain, in any given instance of radical translation, that the method of translation would be able to isolate the precise manner of slicing up the world that had been behind the use of a given expression such as "gavagai".
The underlying issues with which Quine was concerned in the problems of the inscrutability of reference, as well as the problems of radical translation, emerged, as well, in Quine's position concerning the relativity of theoretical and language networks. Indeed, in a sense, the relativity of theory and language were merely flip sides of the same set of issues.
However, in discussing the relativity of such networks, Quine was not maintaining just that we are trapped by the boundaries of our language and theories in our attempts to understand or contact reality in some extra-linguistic or extra-theoretical manner. He also was contending that our theories and talk about the world only made sense relative to some arbitrarily selected background language which had to be accepted as the base line from which we linguistically and theoretically approached our dealings with the world.
In other words, in exploring the character of the sort of ontology to which any given theory commits one, that theory makes sense: a) only when placed in the context of a background language which is arbitrary; and, b) only when that theory is translated into the background language according to some equally arbitrary set of rules for translating theoretical language into backgrou
nd language. As such, Quine believed it is meaningless to talk in terms of inter-theoretical dialogue or exchanges.
Instead, Quine believed one only can make epistemological progress within the confines of one's own theoretical perspective by being seriously committed to the particular brand of 'aggregate science' and the associated scientific method that is generated through such a theoretical perspective. Moreover, according to Quine, the scientific method which is associ
ated with the aggregate science made possible by our theoretical perspective constitutes the last arbiter of truth.
In one way or another, Word and Object, taken as a whole, deals in considerable detail with all of the foregoing issues. It discusses problems of radical translation, the inscrutability of reference, the relationship of language and theory, and especially the problem of translation and background language. Indeed, much of Word and Object can be understood as an effort to spell out some of the fundamental characteristics in relation to the sort of background language that Quine believes stands behind theory and into which the latter ultimately must be translated in order to be given a frame of reference in which such theory makes sense.
Chapter One of Word and Object represents something of a portrait in miniature of the issues that he discusses and develops, in detail, throughout the remainder of his book. As such, Chapter One represents a very good opportunity to explore, albeit on a limited scale, some of the issues and problems with which Quine is concerned in Word and Object as a whole.
If one were to reduce, further, the various elements of Quine's position in Chapter One down to their essential features, there are two fundamental themes that stand out for consideration. First, Quine is interested in developing at least the outline of a model for the learning or acquisition of language. In this respect, he places heavy emphasis upon a somewhat behavioristic approach to language learning that relies on the notion of 'contextual conditioning'.
Secondly, Quine is interested, equally, in establishing the rudiments of a model concerning the meaning, significance and construction of theoretical networks within the context of aggregate science. For Quine, this sort of model is caught up in, among other things, notions such as 'simplicity', 'positing', 'theory-building process' and a scientific method that is "unsupported by ulterior controls".
In point of fact, however, and as intimated above, as far as Quine is concerned, one really cannot separate issues of language learning from issues of theory development. Indeed, the former stands as the deep and immediate background against which and through which theoretical discussion and progress takes place. In this sense, Quine's theory of language acquisition is a sort of prototype for the development and learning of models in general - theoretical or otherwise.
In the present essay, I intend to argue against both of the aforementioned themes of Quine's opening chapter of Word and Object. More specifically, I propose to analyze, critically, Quine's theory of language learning as it is presented in Chapter One of his book. I maintain his idea of contextual conditioning (together with related notions such as induction, analogy, abstraction, chain stimulations and the inter-animation of sentences) is a flawed and inadequate representation of language learning. Furthermore, I intend to take issue with Quine's position on the character of the theory building process and how that process relates to such issues as simplicity, hypothetical positing, the role of scientific methodology, theoretical relativity, and the possibility of meaningful inter-theoretical exchanges.
The structures and structuring process of understanding do not give expression to just themes of identifying reference, meaning, and the capacity to differentiate between belief and knowledge. Understanding also gives expression to the way in which all of the foregoing elements are woven together into models or representations that identify, signify (i.e., mean) and involve distinctions between belief and knowledge. In this respect, the dimension of understanding concerned with model building encompasses those aspects of the phenomenology of the experiential field which are preoccupied with the character of the processes by which models are acquired, generated and developed (or discarded) over time.
Quine's perspective in Chapter One of Word and Object constitutes a heuristically valuable means of exploring the foregoing issues since they are precisely the focus of his concerns as well. By investigating Quine's position on such matters, one has a very good opportunity to reflect upon the character of the structuring process of understanding as it encounters and inquires into one of its own dimensions - namely, the nature of one’s own model/theory building process as it engages someone else’s (i.e., Quine’s) ideas concerning the model/theory building process.
The Context of Language Acquisition
On the opening pages of Word and Object, Quine claims:
"Each of us learns his language from other people, through the observable mouthing of words under conspicuously inter-subjective circumstances. 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. Talk of subjective sense qualities comes mainly as a derivative idiom. When one tries to describe a particular sensory quality, he typically resorts to reference to public things - describing a color as orange or heliotrope, a smell as like that of rotten eggs....
"Impressed with the fact that we know external things only mediately through our senses, philosophers from Berkeley onward have undertaken to strip away the physicalistic conjectures and bare the sense data. Yet even as we try to recapture the data in all their innocence of interpretation, we find ourselves depending upon sidelong glances into natural science." (pp.1-2)
To be sure, one of the most fundamental inputs of language learning for an individual does consist of the data acquired by observing, in a variety of circumstances, the linguistic behavior of those around that individual. However, the fact is, being exposed to this kind of information is not, in and of itself, enough. If it were, then, presumably, all domesticated animals also would learn language in precisely the same way. Thus, one would expect, say, kittens and puppies to pick up a language "through the observable mouthing of words under conspicuously inter-subjective circumstances".
There is some evidence that, for example, dogs and chimpanzees can acquire a certain amount of understanding with respect to a fairly large range of words, symbols and phrases. Nevertheless, for whatever reason, there is a tremendous difference in the scope, depth and flexibility of language competency which occurs in human beings from that which seems to occur (if one actually can refer to it in terms of language competency) in dogs and chimpanzees.
This difference suggests that the claim, "each of us learns his language from other people", is only partially true. What is necessary, as well, is some internal, subjective means capable of: a) directing intellectual focus to the aspects of the behavior which are linguistically relevant (as opposed to those features of behavior which are linguistically irrelevant); b) characterizing the nature of the behavior in question; c) inferentially linking the characterized structure of linguistic behavior to other facets of the experiential field in such a way that one can differentiate appropriate from inappropriate experiential contexts in relation to the referential dimension of the given linguistic behavior; d) narrowing down the appropriate (that is, language-behavior relevant) experiential context sufficiently to isolate the specific criterial features that are central to establishing the precise combination of features which constitute the focus of the identifying reference of the given language behavior; and, e) assigning a framework of significance or a hermeneutic to the aspect(s) of the current experiential field being identifyingly referred to such that the character of one's understanding of the reference one has identified reflects the character of the reference which the speaker intended to identify by means of his or her (i.e., the speaker's) linguistic behavior.
There is, of course, a great deal of disagreement over what stands behind capacities a) through e) (as well as other language-related capacities), and, thereby, makes those capacities possible. Nevertheless, anyone attempting to construct a plausible theory of language acquisition which ignored the considerable input of effort and ability that is required by an individual to learn a language, and which is above and beyond the data the individual is able to pick up by observing the behavior of others, would, presumably, encounter many problems on the road to trying to generate a plausible theory of language acquisition.
In the foregoing quote, Quine says, "Linguistically, and hence conceptually, the things in sharpest focus are the things that are public enough to be talked of publicly." In saying this, Quine appears to be saying that language and conceptualization are, for all intents and purposes, one and the same thing, or that conceptualization is a function of, and follows from, language acquisition.
Even if one were to agree with Quine that those facets of experience which are most public, most inter-subjectively conspicuous, most frequently encountered, as well as being most readily likely to be registered through the senses, are the facets of experience for which words will be learned most quickly, one is not compelled to conclude that what is "in sharpest focus" for purposes of initial language learning will be precisely the same sorts of things which will be "in sharpest focus" conceptually. Indeed, for many months prior to the onset of even the first spoken instance of co-ordinated identifying reference, there are many aspects of an infant's experiential field which are being conceptualized in various ways that are not tied exclusively to those "things" which might be considered by adults to be most publicly conspicuous and sensorily accessible.
Sensations of hunger, thirst, tiredness, waking, being startled, fear, pain (e.g., from diaper rash, colic, excessive exposure to heat or cold), pleasure of tactile and proprioceptive stimulation, play, dreaming, various felt emotional responses of anger, likes and dislikes, frustration (even if none of these emotions are identified as such by the infant), and so on, have been going on for a long time before spoken language ever makes an appearance. In this sense, the phenomenology of subjective experience, in all its diversity, marks the most conspicuous, the most readily accessible, the most commonly encountered aspect of existence for the child.
Consequently, within the pre-linguistic child, some form of experiential co-ordinate grid system already has begun to be laid down, in however primitive a fashion. This primitive, experiential grid system will help form the pre-understandings through which the language behavior of others will be engaged by a child.
In other words, these early conceptualizations, of one sort or another, constitute the backdrop against which subsequent experience - including that of language - is to be considered. Naturally, in order to communicate with someone else, a common experiential ground must be found with respect to which one can begin to establish points of identifying reference about which some sort of interaction- including, possible, further discussion- can take place.
Furthermore, the "things" which are public in some physical manner that renders them easily accessible (experientially) to human sensory capabilities do tend to constitute the things that are most conspicuous in Quine's sense of being high-profile candidates for being talked about inter-subjectively, and, thereby, becoming the focus of a child's early word learning experience. Nonetheless, language is only one dimension about which the individual conceptualizes.
Very likely, whatever words are learned will be placed in an already existing, albeit limited, conceptual geometry which represents the way that experience has, up until the time of learning words, been individualized, particularized and inferentially related in terms of the hermeneutical co-ordinates which structure a given child's world. After all, if the words that are learned refer to "things", "objects" and "items" which are conspicuous enough to be accessible to inter-subjective processes of identifying reference, then, they represent features which very likely have been encountered experientially by the child prior to the learning of the appropriate words. The word-label merely represents one more component to fit into the co-ordinate system which has been developed during the course of experiential encounters with the physical environment prior to the learning of the word.
A child may resort "to reference to public things" in order "to describe a particular sensory quality" of a subjective nature. Yet, this only is because the nature of communication is such that one has to constantly work at maintaining some sort of common ground through which the hearer can have a point of identifying reference to hold onto as he or she (the hearer) tries to place what is said within an aspect of his or her (the hearer's) conceptual geometry in order to make some sort of sense out of what is said.
Thus, in contrast to Quine's position, "talk of subjective sense qualities" may not come "as a derivative idiom" in relation to talk of public things. Rather, "subjective sense qualities" are experientially prior "to talk of public things".
As a result, the latter may be treated as merely an extension of the former. Or, public things may be talked about in terms of (i.e., from the perspective of) the experience one has had with those "public things" and in terms of the conceptualizations which have been built up over time in relation to one's experiential interaction with these "public things".
On the basis of the discussion of the last several pages, one is not forced to maintain "that we know external things only mediately through our senses" as Quine seems to indicate in the quote noted earlier. Our senses do mediate between us and external things. However, one could say, equally well, that our conceptual geometries mediate between us and external things.
We do not always leave sensory data in an unprocessed, raw form which is a purely physiological/biochemical framing of such experience. We also tend to individuate sensory data, particularize it, interpret it and assign various shadings of valuation or significance to the incoming data.
This manner of characterizing our experience forms the basis for the conceptual co-ordinate grid system which is constructed over time and forms the structural character or logic of the conceptual geometry by which, and through which, one interacts with, and interprets, both subsequent and past experience. Just as conceptualization need not be restricted to the language we learn from others, so, too, our epistemological interaction with the external world need not be restricted to purely sensory events. Data must be taken and worked into a conceptual model capable of reflecting the character of that to which the sensory data in question gives expression.
Moreover, this model often is capable of assigning to data, or discovering in it, a form of significance or meaning which is consistent with such a characterization. The conceptual characterization, in turn, must be reconcilable with the character of the sensory data that is available, so that the former reflects, in some sense, various facets or dimensions of the latter.
Consequently, the epistemic character of the task facing an individual need not be one which requires the individual "to strip away the physicalistic conjectures and bare the sense data" ala Berkeley et al. On the other hand, the epistemic character of the task before us need not be one in which "we find ourselves depending upon sidelong glances into natural science" (with its physicalistic overtones) as Quine seems to be suggesting in the previously cited quote from Word and Object.
In point of fact, the experiential or phenomenological field in which each of us find ourselves embedded has a variety of characteristics. We particularize and individuate these features according to an existential dialectic which plays off sensory features against hermeneutic features. This dialectic generates epistemic structures which vary, to some extent, from individual to individual, as well as display similarities, to some extent, among different individuals.
Even if we were able to isolate, clearly, the nature of the contribution of un-interpreted sense data, we still are faced with the problems of determining the following. What do such 'base sense data' mean? To what degree does this data distort or accurately reflect the character of that which helps give rise to a certain sensory experience of given character?
Furthermore, let us assume one were to grant there are physical, external objects in the world. Nonetheless, one still is faced with the problems of trying to determine what the character of the relationship is between the nature of one's experiential field and those 'physical, external objects' which are assumed to exist.
One also is confronted with questions of the following sort. What exactly is entailed by the existence of such objects? What does it mean to describe something as "physical"? Are physical objects necessarily material objects? If they aren't, just what is the metaphysical status of the physical character of such "objects"? That is, what manner of objects are physical ones, if by "physical" we do not mean material in nature (whatever this means)?
The foregoing problems are complicated further when one is faced with trying to figure out precisely how language fits into all of this. In other words, is language a distorting or a reflective medium with respect to the way it relates the conceptual aspects of the experiential field to other aspects (e.g., so-called sensory data) of that field? Or, what is the distorting or reflective character of the way language relates the experiential field to that which makes, or is thought to make, such a field possible and gives it the diversified character it has?
In Search of Bare Sense Data
While criticizing those philosophers like Berkeley who are in search of bare sense data, Quine states:
"Aware of the points thus far set forth, our philosopher may still try, in a spirit of rational reconstruction, to abstract out a pure stream of sense experience and then, depict physical doctrine as a means of systematizing the regularities discernible in the stream.... Talk of ordi-nary physical things he would then see as, in principle, a device for simplifying that disorderly account of the passing show.
"But this is a misleading way of depicting matters, even when the idea of a sense-datum "language" is counted frankly as a metaphor. For the trouble is that immediate experience simply will not, of itself, cohere as an autonomous domain. References to physical things are largely what hold it together. These references are not just inessential vestiges of the initially inter-subjective character of language, capable of being weeded out by devising an artificially subjective language for sense data. Rather they give us our main continuing access to past sense data themselves; for past sense data are mostly gone for good except as commemorated in physical posits. All we would have apart from posits and speculation are present sense data and present memories of past ones; and a memory trace of a sense datum is too meager an affair to do much good. Actual memories mostly are traces not of past sensations but of past conceptualization or verbalization." (pp. 2-3)
One might agree with Quine that a position which viewed "talk of ordinary physical things ... as, in principle, a device for simplifying that disorderly account of the passing show" could be "a misleading way of depicting matters". However, the agreement would not have to be predicated on Quine's belief that "immediate experience simply will not, of itself, cohere as an autonomous domain".
Part of the problem here is not knowing what the parameters are of Quine's use of the term "autonomous domain" with respect to immediate experience. By this phrase, Quine could be raising an implicit question for those philosophers against whom he is arguing.
This question concerns the problem of explaining why immediate experience has the character it does and why one cannot assume immediate experience coheres into a given structural form, in and of itself, as an autonomous realm of internal events independent of the rest of reality. If this is the case, then, one could acknowledge the importance and relevance of the question so raised.
In effect, the question is: if immediate experience does cohere, in and of itself, then, what is the character of the autonomous domain which causes immediate experience to cohere in the various ways it does? Quine contends that one cannot come up with a plausible explanation for the reasons why immediate experience has the character it does unless one makes reference to physical things.
Unfortunately, this sort of response can be as misleading as the kind of position advocated by those Quine is criticizing. In other words, if the reason why part of the immediate experience has the character it has is because physical things have the character they do, then, only by referring to the character of these physical things will the observed character of immediate experience make sense.
On the other hand, part of the reason why immediate experience has the character it does might be because the internal phenomenological laws, principles or rules of coherence which give expression to an individual's immediate experience have a certain autonomous character that is a function of the neurophysiological and/or biochemical and/or mental structure of the individual undergoing the immediate experience. In this respect, the character of a given sensory experience is as much a function of the structural and process logic of the capabilities inherent in an individual's sensory/conceptual equipment (both in terms of limitations as well as in terms of access ranges) as it is a function of the character of the physical thing being encountered.
Of course, these capabilities may need to be activated by 'something' (external or internal) in order to register an experience which will show up phenomenologically as an experience with determinate characteristics of a certain nature. Nevertheless, the structure of the logic by which they operate is, in a sense, autonomous and independent of physical things being the way they are.
Some organisms respond to a certain range of sound waves. Others do not respond to the same range. Some organisms respond to differences in magnetic field properties. Other organisms do not respond to such field differences. Some organisms are capable of living in physical environments which prove to be deleterious, if not fatal, to other kinds of organisms.
To use some computer terminology, there seems to be a certain amount of interfacing which goes on between organism and environment. During this interfacing, each side brings something in the way of autonomous domains of principles, structural logic, and so on, to the ontological points that mark the areas or instances of interfacing engagement. From the organism's point of view, out of these interfacing-engagements arises an experiential field of a given character.
Assuming the foregoing is true, then, if one is being misleading by saying plysical things are "a device for simplifying that disorderly account of the passing show", one also is being misleading when one says "references to physical things are largely what hold it [i.e., immediate experience] together". This is the case becuase there may be principles of "coherence-logic" on both sides which constitute the basis for organisms and physical things having the apparent character for which their ontological expression seems to provide evidence. Therefore, while one might not be quite correct if one were to say that immediate experience coheres in and of itself, one might not be quite incorrect if one were to make such an assertion, provided it is qualified in a defensible manner.
Consider another aspect of Quine's position. Let us assume one were to agree with Quine that references to physical things "are not just inessential vestiges of the initially intersubjective character of language".
This agreement does not commit one, automatically, to maintaining that physical things "give us our main continuing access to past sense data themselves; for past sense data are mostly gone for good except as commemorated in physical posits". That Quine should claim past sense data are commemorated only in the form of "physical posits" is strange.
After all, just a few sentences later, he stipulates: "memories are traces not of past sensations but of past conceptualization or verbalization." Is he equating "physical posits" with past conceptualizations and verbalizations? If so, in what way are past conceptualizations and verbalizations physical posits? And, what is the character of a "physical posit"?
Is he advancing some sort of mind/brain identity theory in which all conceptualizations and verbalizations reduce down to the functioning of a physical thing - namely, the brain? Even if he is advancing such a thesis, why should one not suppose that memories are indeed traces of past sensations which have been encoded in a stored form whose character accurately reflects the character of the original sensation experience? Certainly, Quine has not put forth any arguments, yet, to demonstrate why memory traces cannot reflect, accurately, the character of past sensations - as sensations - and not as some conceptualization or verbalization.
Conceivably, one’s memory of a past sensation might be a function of some conceptualization or interpretation of a certain experience. As a result, what one remembers is not an accurate encoding of the past sensation, but is, instead, one's interpretation of that sensation. However, such a possibility represents only one alternative.
For example, what one remembers might not be a conceptualization or interpretation of past experience. It might be an accurate reflection of, or even a re-invoking of, the full phenomenological nature of the past experience. This sort of possibility is something which is suggested by the findings of psychologists who have worked in the area of eidetic memory images.
One might conceive of an innumerable set of combinations of the possibilities stated at the be-ginning of this paragraph. Presumably, one of the jobs of epistemology is to try to sort out which of the foregoing alternatives might be most plausible in any given instance of memory. In any event, one, simply, cannot assume, as Quine seems to be doing, that the memory trace is a "physical posit" (whatever this means) or that it is only a conceptualization or verbalization (whatever this means).
|
