|Understanding Quine in Terms of Aufbauian Reductionism:
Another Look at Naturalized Epistemology
Oneonta, NY USA
True to at least Burton Dreben's word, lifelong friend and student of Quine, it is simply intellectually and historically irresponsible to attempt to understand Quine’s work apart from Carnap’s.1 In fact, it could even be said that Quine's dedication of his first major epistemological undertaking to Carnap as his "teacher and friend" (Word and Object (1960)) was a far stretch more than a polite gesture; it might even be ventured that without Carnap, there might not have been a Quine. Or at the very least, Quine as we know him. In fact, as if to underline this very point, Quine writes in his short essay "Homage to Rudolph Carnap" (1970) that: "Carnap was my greatest teacher … I was his disciple for six years. In later years his views went on evolving and so did mine in divergent ways. But even where we disagreed he was still setting the theme; the line of my thought was largely determined by problems that I thought his position presented." 2 And these problems, at their most general level, revolved around what Russell once gave the title of a book: Our Knowledge of the External World (1914). For the fundamental questions that initially drove Russell, then Carnap, and eventually Quine, were simple, although admittedly perplexing: What does our knowledge3 of the external world consist of and how does one acquire it? In what sense is it certain, and/or justifiable, if at all?
However, in this essay I certainly cannot explain how Carnap influenced Quine in regard to these matters in any degree of comprehensive detail—that would take a book. So instead, I focus on explicating a small portion—particularly, a third—of what I claim is the central tension between these two philosophers, namely, one of Quine’s reactions to the three fundamental epistemological circles he believed to be present in Carnap’s work, or what I think we may characterize as three contemporary variants of Meno’s Paradox.4 More specifically, I focus here on what we may identify as the first circle:  The “rational reconstruction” (LAW 158) of knowledge in the Logische Aufbau der Welt, where knowledge seems to paradoxically emerge from knowledge, where the latter, more primary mode of knowledge consists of “elementary experiences,” the relation R of “remembering as similar” and Russell and Whitehead’s theory of relations. Meanwhile, if only to frame our discussion in terms of Quine’s more comprehensive reaction to Carnap, realize that the other two circles are:  “The linguistic doctrine of logical truth,”5 or in other words, Carnap’s conventional doctrine of logical truth, as spelled out in the Logical Syntax of Language. For according to Quine, as well as any of those who wrestled with the “logocentric predicament,” 6 on this account, we seem to have to already know logic to acquire logic; in particular, in order to accept a logical inference as logically valid, one must presuppose the validity of that very inference.7  Carnap’s brand of analyticity, which is not be confused with the linguistic doctrine of logical truth. For as Quine points out in a number of papers and letters to Carnap,8 and most famously in “The Two Dogmas of Empiricism” (1950), it seems that to define analyticity, one must already have some definition of analyticity in mind, and so, correspondingly, to know what ‘analyticity’ means in terms of grasping it by way of its definition, we must already know what ‘analyticity’ means.
However, as noted above, as interesting as it would be to examine circles  and  in more depth, time and space simply do not permit. As a result, I focus on just circle  here, namely, the paradox of reductionism. To do so, I have divided this paper into five sections: In §1, I sketch the general structure of the reductionist project as it is presented in the Aufbau. In §2, after canvassing the distinction between what Quine characterizes as “radical” reductionism (namely, Aufbauian reductionism; FLPV; TDE 39) v. “attenuated” reductionism (namely, the idea that the truth value of a sentence can be determined purely by empirical means; FLPV; TDE 41), I explain why Quine thought the Aufbau was flawed. In particular, I begin by examining his criticism that it smacked of “mentalistic monism” (FSS 15). Following, I show that in virtue of this critique, Quine puts his finger on—however implicitly—the epistemological circularity of the “radical” reductionism of the Aufbau, which, I claim, reminds us of the second horn of Meno’s Paradox.
However, in §3, I point out that this breed of epistemological circularlity is not be confused with what we may characterize as the “naturalistic circle—“ where it is alleged that using the scientific method to examine science is problematic—in two fundamental respects, where the first problem was brought into focus by Hume,9 and the second, by “physicalism,” particularly, Quine’s notion of physicalism. These circles are, respectively: a.) Science may only yield fallible results and so, if the claim “We must use science to examine science” is a scientific claim, it is a fallible claim. As a result, the empiricist, or in other words, the naturalized philosopher, appears to be undermined from within. b.) The epistemological foundations of science are, according to contemporary scientific research itself, overwhelmingly sparse, consisting of unrelated sense data manifest by certain sets of nerves firing. Consequently, if it is shown that scientific theory is equivalent to such scanty data as a result of reducing science to it, then our scientific claims seem to be equally scanty; they amount to nothing more than intermittent and unrelated sets of nerves firing.
Subsequently, in §§4-5 I show that Quine’s repudiation of “radical” reductionism, in light of its paradoxical nature (in a respect reminiscent of the second horn of Meno’s Paradox) could only have motivated Quine to adopt naturalism for reasons that appear to be independent of his pragmatic concerns, simply because it is not reasonable (namely, it is paradoxical)10 to adopt a Carnapian phenomenalistic/mentalistic approach to epistemology. Armed with what could only be his invigorated faith in the naturalistic method, he was then, as I see it, equipped to break the physicalistic version of the naturalistic circle, a repudiation that, I show, entails his rejection of “attenuated” reductionism and concomitantly, his rejection of “analyticity” if not “certainty” altogether. As such, the Humean version of the naturalistic circle could simply be dismissed. Meanwhile, the practicality of an admitably fallible science could be unashamedly embraced, although not just for the sake of its practicality—as Quine himself seems to misleadingly indicate throughout his work—but instead, as just noted, to avoid the seemingly Platonic paradox of Aufbauian reductionism.
§1 The Aufbau: a General Overview
As noted above, and as Quine rightly sees it throughout his work and his letters,11 the early Carnap was the champion of the great Russellian project in Our Knowledge of the External World, which was, according to Quine, the explication of the "construction" of the external world from bits of sense data. (TT; ROD 83) In Carnap's case in the Aufbau, this consisted of “attempt[ing] to apply the theory of relations to the task of analyzing reality.” (LAW 7) That is, as noted earlier, for his method of construction, Carnap employs the theory of relations as it was laid out by Russell and Whitehead in the Principia. Broadly speaking, this means that Carnap attempted to show that all concepts/objects12 may be understood as logically "reducible," or translatable13 to the primary relation “remembering as similar” (or what we may also refer to as R in this paper) and certain unanalyzable “elementary experiences” (or what we may also refer to as E in this paper) (see LAW Chapter C).14 In this respect, Carnap hoped to "rationally reconstruct" the concepts of all fields of knowledge (including science) by showing that they may be translated into the strictly formal world of a "constructional language." As such, this constructed language served as a model for how all fields of knowledge may be redefined, or in other words, logically reduced to R and elementary experiences.15
In a bit more detail—although it is not relevant to indulge ourselves in all the Aufbau’s technical machinery here—as far as reducing concept/objects to other concept/objects goes, and conversely, constructing concepts/objects out of other concepts/objects, Carnap explains that:16
if an object a is reducible to objects b, c, then all statements about a can be transformed into statements about b, c. To reduce a to b, or to construct a out of b, c, means to produce a general rule that indicates for each individual case how a statement about a must be transformed in order to yield a statement about b, c. This rule of translation we call a constructional rule or constructional definition.” (LAW, 6)
That is, "construction rules" are the rules that allow us to change any statement about a into given statements about b and c, provided that the concept/object a is "reducible to," namely, may be redefined as the concept/objects b, c. For instance, Carnap points out further on in the Aufbau that we may "reduce" the concept/object of a prime number to two other concept/objects, a natural number and division. As a result, a definition that initially concerned just prime numbers (particularly, a, with b and c absent), namely, "x is a prime number" (LAW 16) may be restated to concern only b & c: "x is a natural number whose only divisions are 1 and x itself.” (LAW 16) Construction proceeds exactly opposite, more specifically, if there is a statement about b and c and it can be translated into a statement about a, b, and c (where b and c may be absent), then the concept/object a is constructed. As a result, the construction of a concept/object is actually a new definition of a concept/object via other concept/objects. Or as Carnap puts it (where here, he uses the term ‘statement’ rather than ‘definition’): “An object (or concept) is said to be reducible to one or more other objects if all statements about it can be transformed into statements about those other objects ... [thus] if a is reducible to b, and b to c, then a is reducible to c. Thus, reducibility is transitive.” (LAW 6) In short then, this constructing/reduction process is what Carnap also refers to as “analysis” which, as such, is “analytic” in nature; as a result, any statements about concepts/object a—where a is reducible to other concepts/objects b and c—are interchangeable with statements about b and c.
Ultimately, as Carnap explains in Part III, Chapter C of the Aufbau, all basic concepts/objects may then, by a method he calls “quasi-analysis,” (LAW §§69-74) be reduced to a network of “basic relations,” (LAW 98) where, as noted earlier, the relation that is logically primary is R, or in other words, “remembering as similar” and the components that these relations obtain of consist of unanalyzable “elementary experiences.” However, “analyzing” “unanalyzable wholes” (namely, “elementary experiences”) appears to be somewhat problematic. As a result, Carnap explains that “quasi-analysis” is actually “a synthesis which wears the garb of an analysis.” (LAW 121; emphasis added). That is, as just noted, although the construction/reduction process of translating concepts/objects into other concepts/objects in the Aufbau is strictly analytic, the method of constructing-from/reducing-to concepts/objects by way of “quasi-analysis” is synthetic. But what is this “synthesis?” In particular, does it mean that statements about say, a lower-level concept/object c are not interchangeable with statements about E and R? No. To see why this is the case, first note the following general explanation Carnap gives of quasi-analysis: “We overcome the difficulty which results from the fact that elementary experiences are unanalyzable by introducing a constructional procedure which, even though synthetic, leads from any basic elements to objects which can serve as formal substitutents for the constituents of the basic elements.” (LAW 110; emphasis added) That is, quasi-analysis is simply “quasi” because it attributes a certain formal structure to what is, in principle, unstructured in virtue of being unanalyzable; namely, it provides a structure that acts a logical, analyzable proxy for elementary experiences, which as such, serves as the analyzable foundation for all fields of knowledge. Carnap makes this clear when he continues in the immediately following passage: “We call [these objects constructed by quasi-analysis] formal substitutents, because all assertions which hold for the constitutents hold, in analogous form, also for them. We call this procedure quasi analysis.” (LAW 110) That is, and crucial to note, a reduction to the objects created by quasi-analysis, is, analogously, a reduction to the elementary experiences and R, a point that is behind Carnap’s remark that “the subjective origin of all knowledge claims lies in the contents of experiences and their connections” (LAW 7; emphasis added). For, Carnap writes on p. 111, “It is of importance whenever we are concerned with unanalyzable units of any kind, that is, with objects which, in their immediate given-ness, do not exhibit any constituents or properties or aspects. These objects are given, as it were, only synthetically; nevertheless, as a result of our procedure, we can ascribe various characteristics to them” (LAW 111). In short then, for our purposes, this means that according to Carnap, all knowledge claims are first reducible to the products of quasi analysis (“basic relations” (§§75-83) which obtain of elementary experiences), and then second, analogously, by way of quasi-analysis, to the elementary experiences themselves, and R.
So in short, with all additional technical detail set aside, Quine writes in "Russell's Ontological Development" (1966) that: "[In the Aufbau,] Carnap achieved remarkable feats of construction, starting with sense data and building explicitly, with full Principia techniques and Principia ingenuity, toward the external world.” (TT; ROD 84) In still other words, as Quine indicates here and elsewhere, in the Aufbau, Carnap had taken the Russellian epistemological concerns to subtler, if not more daring heights; according to Quine, epistemology had, thanks to Carnap, taken a clear and decisive logical turn. And initially, this seemed to mean a turn for the better.17
§2 Quine’s Reaction to “Radical” Reductionism: A Sensitivity to Circularity
§2.1 “Attenuated” v. “Radical” Reductionism
As is well-known, Quine rejects the idea that any sentence claim can, both in principle and in practice, be confirmed or denied on the basis of just experience; this rejection is behind his attack on the “second” dogma of empiricism, namely his distaste for allegedly “synthetic” claims. In still other words, Quine’s renouncement of the synthetic amounts to a repudiation of what he refers to a “subtler and more tenuous form” (FLPV; TDE 40) of reductionism, or what we can refer to, after Quine, as “attenuated” (FLPV; TDE 41; emphasis added) reductionism. Meanwhile, it is also well-known that Quine rejects what he construes as Carnap’s more “radical” (FLPV; TDE 39; emphasis added) form of reductionism, where, as explained in §1 above, it is alleged that all statements may be reduced, or in other words, translated into sense data and ordered by at least one fundamental relation (which, as we saw in Carnap’s case, consisted of, respectively, E and R). Or as Quine puts it: “Radical reductionism, conceived now with statements as units, set itself the task of specifying a sense-datum language and showing how to translate the rest of significant discourse, statement by statement, into it. Carnap embarked on this project in the Aufbau” (FLPV; TDE 39; emphasis added).18
In short then, according to Quine, the distinction between attenuated reductionism v. radical reductionism is, respectively:  The truth value of certain sentences (namely, “synthetic” ones) may be established solely by appealing to empirical evidence and  All knowledge claims may be reduced, or in other words, translated into empirical experiences, where such empirical experiences are ordered with a select amount of fundamental relations (e.g. Carnap’s “remembering as similar”).
§2.2 The Paradox of Radical Reductionism
As briefly noted above, it appears that, on the face of it, Quine eschewed the notion of “synthetic” claims—and so, the theory of “attenuated” reductionism—in light of his holism, where, according to this theory, no sentence stands or falls on its own, regardless of what the empirical evidence tells us. For instead, the Quinean story goes, whether or not we reject or accept a given sentence also depends on its relationship to the rest of the theory at hand; as a result: “statements about the external world face the tribunal of sense experience not individually, but only as a corporate body.” (FLPV; TDE 41)19
But as noted in my opening remarks to this paper, I think that the reasons behind Quine’s rejection of attenuated reductionism are a bit more complex than this, although Quine never directly says as much. In particular, as already suggested above, it may be shown that Quine’s repudiation of “radical” reductionism ultimately entails his rejection of “attenuated” reductionism by way of a rejection of what I refer to as the physicalist version of the naturalistic circle.
To see why this appears to have been the case, we must first understand how and why Quine rejected the paradoxical “radical” reductionism manifest in the Aufbau. To begin, note a particularly revealing line from Quine’s 1993 paper “In Praise of Observation Sentences:” "The lively objection to [protocol sentences], as vehicles of evidence for our knowledge of the external world, is that they already assume such knowledge … But the answer is that they need not.” (IPOS 108; emphasis added). In other words, here, Quine is defending the idea that his notion of an “observation sentence—” a term that he uses interchangeably in this paper (IPOS 108) with the term “protocol sentence—” is not, in the course of the human being’s learning process, initially “theory-laden” (IPOS, 110). Rather, our initial use of language is purely reflexive; a product of our being habituated to say a given sentence S when we experience a given range of neural intake M (IPOS, 109). Only later, through a process that is not relevant for us to examine here, do human beings acquire theory, and relatedly, knowledge. But what we must ask is: Why does Quine balk at the idea that our observation sentences somehow initially “assume” knowledge such that, say, when we first learn to use the word “Mama” correctly, that we allegedly know who and what Mama is when we say it?
Two reasons, where the first, and most well-known is: According to Quine, science tells us that our initial input is remarkable “meager” (FSS, 16), consisting of “the mere impacts of rays and particles on our surfaces and a few odds and ends such as the strain as walking uphill” (FSS 16). In this respect, our initial input consists of unrelated bits and pieces of sense data. As a result, we are not in any respect “aware” of such nerves firing (see at least FSS, 17-18, RR, 2-4). Instead, such input must be understood as our body’s initial and unconscious state of reacting to the world—not only when we are infants and first experience the world, but also in terms of our “outermost” interface with the world, as it assails us throughout our entire lives. Thus, “knowing—“ where, according to Quine, knowing consists of at least some kind of psychological ability to evaluate data such that we may say the right thing at the right time, and as a result, at the very least, be “aware” of what is being experienced and what to say20—is simply not an option at this stage of our data acquisition. Rather, according to Quine, our initial, most outermost interface with the world occurs at what he characterizes as the the purely “receptual” (FSS 17) level of data acquisition; accordingly, “awareness,” and so, the possibility of knowledge occurs only at the “perceptual” (FSS 17) stage of data acquisition.21
Second, although Quine never explicitly says as much, it seems that assuming knowledge in our initial experience of the world incurs the second horn of Meno’s Paradox, namely, “[A man] would not seek what he knows, for since he knows it, there is no need of inquiry.” (M 80 d-e). To see why this is the case, we must briefly re-examine what is going on behind the scenes in Carnap’s Aufbau. In particular, realize that if any statements about a concept/object a are, in fact “translat[able” (FLPV; TDE; 39) or in other words, are logically equivalent to some set of elementary experiences E and the relation of R, it seems that in virtue of just experiencing E and knowing R and the theory of relations, we must simultaneously know a and any statements about a as well—at least implicitly—much in the same way that if we know A B and we know our equivalence rules, then we know ~A v B. For again, if it is the case that all concepts/objects may be reduced (namely, are logically equivalent to elementary experiences and R), then it seems that all our possible knowledge of the external world (including scientific theory) “lies in” (LAW 7) our “elementary experiences” and R, if only in the respect that “lies in” means logically equivalent to.22 For as noted above, keep in mind that the Aufbau’s reductionist program was modeled after the logistic reduction carried out in the Principia (although the latter engaged in a fatally flawed reduction, as pointed out by Gödel (1931)). In fact, Quine writes in this regard in From Stimulus to Science:
The conclusion [Whitehead and Russell] drew was that mathematics is translatable into pure logic (FSS 9)…[The] total translatability [of mathematics and their basic laws and interrelations] into just elementary logic and a single familiar two-place predicate, membership, is of itself a philosophical sensation (FSS 9-10; emphases added)…Russell adumbrated [the idea of this kind of construction] in Our Knowledge of the External World, and a dozen years later, Rudolf Carnap was undertaking to carry it out. Carnap’s effort found expression in Der Logische Aufbau der Welt (1928). (FSS, 10; first two emphases added)
However as far as I can tell, Quine did not overtly reject the Aufbau because, from at least a strictly logical point of view, it seems that to show that all knowledge that is possibly derivative of a given elementary experience E, the relation R and the theory of relations, is as such, “assumed” in these primary elements, through, quite simply, the “total [logical] translatability” (FSS 10) of knowledge into these primary elements. Rather, as noted in my introductory remarks, he accuses Carnap of embracing a “mentalistic monism” in the Aufbau because, Quine asserts, according to Carnap, “elementary experiences” were psychological “global units” (FSS 10) that as such, consisted of “the individual’s total experience at that moment” (FSS 10).23 As a result, Quine explains, these psychologically experienced wholes were, according to gestalt psychologists—as well as according to Carnap and a number of his philosophically-inclined contemporaries—24 units that one is necessarily “aware of” (RR 1-4) when they are being experienced. And in the respect that one would be “aware” of such units, one would, to at least some inchoate degree, know them. In short then, as Quine sees it, according to this psycho-epistemological account of the world, “awareness of” was roughly equivalent to “knowledge of”—such “wholes” were alleged to be the psychological foundations of knowledge (RR 1-4).25
With this in mind, realize that on the face of it, although it does not appear that Quine calls the Aufbau a “sort of fiction” (IPOS, 116) and “make believe” (OR; EN 75) because it invokes the second horn of Meno’s Paradox, for all intents and purposes, this must be understood as his deeper objection; overt or not: Carnap’s “mentalism,” or what Quine also refers to as his “phenomenomalism” (FSS 15-16) is not just unconvincing because it incurs vague and suspicious mental entities (namely, “elementary experiences”), but worse still, as just noted, these entities allegedly admit of immediate awareness, and thus, knowledge of the world. And not just rudimentary knowledge, but, it seems, knowledge that is logically equivalent to all the theories and knowledge claims possibly derivative of a given elementary experience or experiences. Note in fact, where Quine does somewhat obliquely admit as much in “Epistemology Naturalized” (1968), paying particular attention to the idea that it would be nice, Quine thinks, if we could show that all of science is translatable to “logic, observation terms and set theory” (OR; EN, 76), and as such, “[show] that everything done with the one apparatus could in principle be done with other” (OR; EN, 76). But quite, frankly, he tells us, this is impossible. In other words, Quine did indeed think that the Aufbauian project was paradoxical, although not quite in Meno’s respect that we would be unmotivated to learn what we already know, but in the respect that it is simply not possible that in virtue of just “elementary experiences” and R we simultaneously know all of what we eventually come to learn, particularly, scientific theory, or as Quine puts it: “certainly we did not grow up learning definitions of a physicalist language in terms of a prior language of set theory, logic and observation” (OR; EN, 76; emphasis added).26 In fact, Quine had already put his finger on what seems to be the sheer impossibility of assuming such a cache of sophisticated logical ability some eighteen years earlier when he wrote in “The Two Dogmas of Empiricism:”
The language which Carnap adopted as his starting point [in the Aufbau] was not a sense-datum language in the narrowest conceivable sense, for included also the notations of logic, up through higher set theory. In effect it included the whole of language of pure mathematics. The ontology implicit in it (that is, the range of values of its variables) embraced not only sensing events but classes, classes of classes, and so on. Empiricists there are who would boggle at such prodigality. (FLPV; TDE 39; emphasis added)
Who, then, was “boggling?” We might, given what we have seen above, conclude that it was Quine who found himself gasping at such epistemological “prodigality.” In fact, in so many words, Quine repeats this point in his last book, From Stimulus to Science, when he writes: “we are given a canon or procedure [in the Aufbau], and a brilliant one, but not one that makes the theory of the external world translatable into the language of sense experience. That is too much to ask.” (FSS 13; emphasis added)
So in short, according to Quine, the “radically” reductionist program that Carnap set up for himself in the Aufbau, however technically deft, was not only plagued with mentalistic preconceptions—namely, “awareness” at the initial stages of knowledge acquisition—it was a hopelessly impossible project—we are not, thanks to just “set theory, logic and observation” (OR; EN 76) initially privy to our knowledge of the external world. So, to be perfectly clear, in this respect, Quine’s rejection of the “radical” epistemological reductionism invoked in the Aufbau should be understood in the context of an implicit reaction to a variant of Meno’s Paradox,27 particularly, the idea that to know what we do (particularly, scientific theory), we must somehow, already know it, or at the very least, it must be immediately accessible to us by way of an “awareness” of our “elementary experiences, ” R, and our grasp of logic.28
§3 The Naturalistic Circle
With Quine’s rejection of “radical” reductionism in mind, where, as just shown, this rejection seems to turn on a somewhat implicit rejection of a form of Meno’s Paradox, we must now take the two versions of what I characterize as the “naturalistic circle” into account. For as noted in the introduction of this paper. I argue that Quine’s rejection of what we may, after Quine, characterize as Carnap’s “phenomenalistic” radical reductionism—in light of its paradoxical nature—could only have reinforced his faith in naturalism. In turn, Quine rejects what we may characterize as—after Quine—“physicalistic” radical reductionism and as such, breaks what I call the physicalist version of the naturalistic circle, which, it may be shown, entails his rejection of “attenuated” reductionism. Concomitttanly, because “the two dogmas are indeed, at root identical” (FLPV; TDE 41), a rejection of the “synthetic” amounted to a rejection of the “analytic” and thus, the need for “certainty” altogether. As a result, as I see it, Quine could simply dismiss the Humean version of the naturalistic circle while the scientific method, despite its fallibility, could be embraced.
§3.1 The Naturalistic Circle: Hume’s Version
According to at least Hume, the scientific method—29 where that method consists of, simply put, gathering our information about the world by means of our senses and/or various hypotheses compiled from the information given to us by the senses—is a fallible method in the respect that its subject matter consists of “matters of facts.” (T 1.3) This is the case because according to Hume, knowledge claims that are based on matters of fact are based on the relation of cause and effect. (T 1.3) However, no causal relation is, according to Hume, necessary, but instead, is a product of imagining certain constantly conjoined events as apprehended through any and/or all of our five senses, and as such, any causal relation can always be imagined otherwise without creating a contradiction. As a result, no matter of fact is necessarily true. Accordingly, because the scientific method is constructed from consideration of matters of fact, no result it yields is necessarily true (see T 1.3). Consequently, by appealing to the scientific method to show that we must use the scientific method to philosophically examine the world, where a component of that world is the scientific method itself, one makes a claim that is not necessarily true; no claim derivative of the scientific method is necessarily true, even if it’s the claim: “We must use science to examine science.” As a result, it is simply not certain that we should be doing naturalized philosophy at all; this is what we may identify as Hume’s version of the “naturalistic circle;” a circle that Hume intermittently torments himself with, particularly in the Treatise.30
It is no surprise then, that Quine, latter-day Humean empiricist that he was, puts his finger on this circle approximately two hundred years later in Roots of Reference, although here, Quine locates the circle in a historical venue that preceded Hume by thousands of years:
Ancient skepticism, in its more primitive way, likewise challenged science from within. The skeptics cited familiar illusion to show the fallibility of the senses; but this concept of illusion rested on natural science, since the quality of illusion consisted simply in deviation from external scientific reality (RR, 2-3)
In other words, Quine’s point is: Ancient skepticism challenged science from a “scientific point of view” because the concept of an illusion is itself a scientific concept; an “illusion” is, by definition, a “deviation” from an empirically confirmable fact. As a result, the idea that the senses are fallible—and thus susceptible to illusion—is itself a claim that is derived from the senses, which means that the claim “The senses are fallible” seems to be fallible itself, and thus, just as susceptible to illusion. As such, just as Hume would complain thousands of years later, it seems that science may be “challenged … from within” (RR, 2).31
How then, can an empiricist, a naturalistic philosopher, avoid this? For instance, should one, like Husserl—in partial response to Hume’s version of the naturalistic circle32—appeal to a “first science,” or in other words, to what Husserl called the “epoché?” For on this account, one might, it seems, properly justify his or her decision to use science to talk about the world. For in the case of Husserlian phenomenology, the decision to use science would not be a product of the insights gained from empirical (and thus fallible) experience, but instead, would be a product of insights gained from the epoché, or the “first science,” which, according to Husserlian phenomenologists, yields necessary truths (see at least I 1, lines 59-60). However, as is well-known, Quine was loath to appeal to anything like a “first science,” that as such, is alleged to stand on a foundation built from anything other than empirical observation.33
§ 3.2 The Naturalistic Circle; The Physicalist Version
With the general structure of the Humean version of the naturalistic circle in mind, let’s now take a look at what I characterize as the physicalist version, which was brought squarely into focus by Quine. As noted earlier, according to him, sense data, which he prefers to construe as “neural input,” is fragmented, which as such—at least at the level that Quine identifies as reception—consists of experiences that we are in no way “aware” of and so, as explained above, could not possibly admit of knowledge, even knowledge in Quine’s behavioristic sense of the word.34 Rather, as also explained earlier, according to Quine, “awareness” and so, any possibility for knowledge, only emerges at the level of perceptual similarity. In this general respect, Quine’s epistemology is fearlessly “physicalistic—“ it paints a picture where the primary source of all our knowledge consists of nerves firing, where “awareness” of such events is decisively absent. Recall that this way of looking at data acquisition is opposed to Carnap’s “phenomenalism,” where, as noted, the source of all our knowledge consists of mentalistic entities that are related (at least by R) and we are aware of.35
But as noted in my introductory remarks, adopting the physicalist stance appears to incur another version of the naturalistic circle, which unfolds, quite simply, as follows: If one assumes physicalism and attempts to reduce, or in other words, claims to translate knowledge, particularly, knowledge of scientific theory, to nerve inputs, where such input does not admit of knowledge, then it seems that such knowledge is effectively equated to something that is not knowledge, namely: “smells, noises, feels, flashes, patches of color and the like” (RR 1). As a result, it simply follows that if we translate science, say, the scientific claim X “All our scientific theory may be reduced to physical input” to physical input, then all of science, e.g. in this particular case, X, is equivalent to nonsense; that is, mere “impacts on our sensory surfaces” (POT 1). This then, is the physicalist version of the naturalistic circle, which Quine discusses tirelessly throughout the body of his work.36
§4 A Summary of What is at Stake
To best organize the three main concerns that we have been dealing with, namely, Meno’s Paradox as it is manifest in the “radical” reductionism of the Aufbau, Hume’s version of the naturalistic circle and the physicalist version of the naturalistic circle (setting aside any discussion of “attenuated” reductionism for the moment), realize that Quine was faced with the following epistemological mess: If we “radically” reduce or translate knowledge to sense data then it seems that:
a.) We must assume such knowledge in the sense data by way of the Gestalt psychologists’ “wholes,” or what Carnap preferred to call “elementary experiences;” this constitutes what we might call, after Quine, “phenomenalistic” radical reductionism. Yet as noted, according to Quine, with all suspect mentalistic overtures aside, doing as much is simply impossible. In other words, the Aufbauian project reminds us of the second horn of Meno’s Paradox, namely, the idea that we would not seek knowledge if we already knew such knowledge, where in this case, Quine translates the problem of a lack of motivation to seek what we already know into bleak impossibility: it’s just not the case that in virtue of our “elementary experiences” and knowledge of R and all of set theory that we know all possible knowledge claims logically derivative of E and R; this would simply “be too much to ask” (FSS 13)
b.) Regardless if we assume that knowledge is or is not present in the sense data (e.g. in terms of “elementary experiences”), empiricists widely accept the fact that empirical, and thus scientific claims, are fallible—as was made particularly clear in Hume’s philosophy. Thus, if the claim “We must use science to examine science” is a scientific claim, then it is a fallible claim; this is Hume’s version of the naturalistic circle, and as such, “challeng[es] science from within.” (RR 2)
c.) Moreover, if—unlike the Aufbau Carnap—we conclude that our sense input does not admit of knowledge, it seems that if we reduce, or in other words, translate scientific theory and/or knowledge into sense data conceived of from a physicalistic point of view (and thus, engage in what we might call, after Quine, “physicalistic” radical reductionism), then knowledge, particularly knowledge of scientific theory, equates to nerves firing, and thus, it seems, to nonsense. This is the physicalist version of the naturalistic circle.
§5 The Solution: Naturalism Embraced; “Radical” and “Attenuated” Reductionism Rejected
Quine’s three-fold solution to this mess is, as I see it, quite simple although many aspects of it have been much contested.37
 We cannot, according to Quine, assume knowledge in our initial input; in other words, Quine must flat-out reject Carnap’s “phenomenalistic” radical reductionism as it is manifest in the Aufbau. For as noted above, if we did not, one simply assumes too much, causing us in fact, to “boggle at [the] prodigality” (FLPV; TDE 39) inherent in such an endeavor. Or in still other words, this means that Quine was aware—however implicitly—that Carnap’s Aufbau invoked a paradox, reminiscent of the idea that one “would not seek what he knows, for since he knows, there is no need of inquiry” (M 80 d-e). Meanwhile, as Quine sees it, contemporary scientific research shows that our initial input (“reception”) does not admit of knowledge; namely, science seems to favor the physicalistic approach. In other words, ironically enough,38 it seems clear enough that a simple Platonic Paradox appears to have justified Quine’s endorsement of contemporary scientific research, but quite independently of Quine’s pragmatic reasons for embracing naturalism. Put still another way, the paradox of “radical” reductionsim may have made it quite clear to Quine that it is just unreasonable to assume knowledge in our initial interface with the world; or in still other words, it is simply unreasonable to take what Quine characterizes as a “phenomenalistic” approach. So instead, we must adopt what contemporary science seems to favor, namely, the “receptive” model of data acquisition. As a result, problem a.) noted above is avoided; our initial input is not only bereft of knowledge (even as adults), we are, according to contemporary scientific research, born knowing virtually nothing.39
 However, as noted above, if one translates knowledge claims into what science tells us constitutes our outermost interface with the world (namely, fragmented sense data (“reception”)) one effectively translates knowledge into nonsense; recall that this is the physicalized version of the naturalistic circle explained above. To specifically avoid this predicament, Quine must claim that at best, knowledge claims are “evidence[d]” (See at least TT; EC, 24; emphasis added) by stimulus, where, crucial to note, the notion “evidenced by” is not, equivalent to “equivalent to.” In other words, by "evidence," Quine is referring to the significant but not comprehensive influence that sense data has on a sentence. So instead, according to Quine, and as noted above, to grasp the given sentence’s entire meaning, and likewise its truth value, and thus to properly know it, it must be understood in terms of the larger context of the theory it is embedded in. As a result, and crucial to note, this means that a rejection of “radical” reductionism (in terms of both a phenomenalistic and physicalistic approach) entails a rejection of “attenuated” reductionism. For if a knowledge claim is not reducible to, and thus, is not equivalent to sense data, but instead, its meaning, and likewise, its truth value can only be obtained upon considering it in terms of the theory it is embedded in, 40 it simply follows that there are no knowledge claims whose truth values may be obtained solely in virtue of empirical confirmation or disconfirmation. In short then, problem c.) noted above is avoided; knowledge, knowledge of scientific theory in particular, is simply not equivalent to nonsense. And as an added bonus, the “synthetic” dogma (namely, “attenuated” reductionism) is revealed for what it is—a dogma.
 With the rejection of “attenuated” reductionism—which, as we just saw, is entailed by Quine’s rejection of physicalistic “radical” reductionism—the expectation for “certainty” is concomitantly dropped, for “the two dogmas are, indeed, at root identical” (FLPV; TDE 41).41 In other words, according to Quine, just as no claim may be proven true or false purely in virtue of sense data, no claim may be proven true or false purely in virtue of its meaning, where the latter (“analytic”) claims, it is alleged, admit of “certainty” and the former (“synthetic”) claims do not. Rather, no claim, according to Quine, is “certain,” or in other words, “necessary” (see also, for instance, WOP; NT).42 As a result, and crucial to note, it simply does not matter if the scientific method, or our choice to employ it, is fallible. What does matter, according to Quine is that science works, “[the scientific, empirical method] is the best [method] we know” (WO 4); as such, it allows us to make certain predictions about the world, which include predictions and hypotheses about scientific theory. According to this line of thought then, we might say that allegedly certain Husserlian “first sciences” do not allow us to fly to the moon or cure cancer, but regular old “fallible” science surely does; in this respect, science “matters.” And so, problem b.) noted above is avoided; the Humean version of the naturalistic circle is simply irrelevant; it is a manifestation of misplaced expecations.
So in conclusion, when faced with the question of: Where does our knowledge of the external world come from—a question that Socrates wrestled with thousands of years before Russell, Carnap and Quine did—Quine was forced to turn to science; in fact, it is no wonder that Quine’s very last book, From Stimulus to Science, was devoted to “physicalizing” the Aufbau. For, the story goes, when we see that reducing knowledge to knowledge (e.g. “elementary experiences” R and all of set theory) is absurd (namely, when we see that phenomenalistic radical reductionism is absurd), we must turn to the more reasonable scientific account of knowledge acquisition, namely, the idea that we are born knowing virtually nothing. Yet this does not mean that we should equate such an initial lack of knowledge (nerves firing, etc.) with what we do know; that is, we should not engage in physicalistic radical reductionism either, and as such, the physicalized version of the naturalistic circle is avoided. Rather, we must realize that although all our knowledge claims are supported by empirical evidence, they may not be translated into them, solving the problem of “if our science were true, how could we know it?” (RR 2). Concomitantly, this means that no knowledge claim may be proven true or false purely in virtue of empirical evidence, namely, the synthetic dogma is revealed as a dogma; that is, “attenuated” reductionism may be rejected. Simultaneously—because the two dogmas are, at root, identical—this also means that no knowledge claim admits of absolute certainty; inspiring us to lower our somewhat childish philosophical expectations for science. As a result, we may simply sidestep Hume’s torment (namely, Hume’s version of the naturalistic circle), and instead, exercise science for all that it’s worth.