BRS Quarterly Home

Recent Issues

May - August 2006 Contents

Cover / In This Issue

Society News

Waismann’s Philosophy

On Russell’s Naturalism

More on Russell and Quine

Orwell on Russell’s Power

Review, On Russell and Education

Review, The Salmon of Doubt

Traveler’s Diary

on russell’s naturalism

Graham Stevens

In an article published recently in this journal,1 Andrew Lugg contends that Quine’s naturalized epistemology was pre-empted in most important respects by Russell’s epistemological project from roughly 1912 onwards. Contrary to the (arguably) standard interpretation of Russell as a Cartesian foundationalist in epistemology, Lugg presents the following portrait of Russell the epistemologist:

He is an empiricist in the Quinean mode, one who takes the doctrine that there is nothing in the mind about the world not first in the senses to be a finding of science (as opposed to a result of pure inquiry prior to scientific research). His empiricism is integral to his naturalism and he intends his claims about the evidence of the senses and our knowledge of the external world to be understood as hypotheses open to criticism and improvement.2
I share Lugg’s conviction that the naturalistic elements of Russell’s philosophy are important. The subject is one deserving of further attention. In the following paper, I will offer a somewhat different slant on Russell’s naturalism to the one Lugg presents. Although I am in agreement with Lugg’s general theses that (1) Russell’s naturalism is an important element of his philosophy that has been overly neglected in studies of him, and (2) Russell’s naturalism is an important precursor to Quine’s, I will take issue with the details of his take on each thesis. With regard to (1) I will argue that naturalism of the Quinean variety cannot be accurately attributed to Russell in as neat and simple a fashion as Lugg does. One reason for this is that Russell cannot be accurately characterized as an empiricist, even if the characterization is a qualified one of an empiricist “in the Quinean mode”. With regard to (2) I will argue that Russell’s greatest influence on Quine’s naturalistic project did not stem from his epistemology but from his semantics. In criticizing Lugg’s (2), I will therefore simultaneously be defending my own interpretation of Russell’s naturalism as given in detail elsewhere.3 On that interpretation, Russell took the naturalistic turn when he looked to psychology to provide a new home for propositional content. Once located in empirical psychology, Russell then took the further natural step of seeking to explain content in purely causal terms. Russell’s greatest contribution to philosophical naturalism was his attempt to naturalize content via a causal theory of meaning. It is not, as Lugg claims, Russell’s empiricism that is integral to his naturalism; it is his psychologism. The point is important for two reasons: first, Russell’s attitude to empiricism was variable and rarely resulted in unconditional subscription;4 second, it means that Russell was only really engaged in a project that can be usefully labelled “naturalistic” after he abandoned the anti-psychologism that was central to his early philosophy.

There is, as Lugg notes, plenty of evidence against the picture of Russell as a naturalized epistemologist. Russell repeatedly talks about the importance of establishing certainty in philosophy and it seems that the quest for such certainty was the original motivation for his interest in philosophy and, more particularly, for his desire to establish the truth of logicism in mathematics.5 But these issues are only apparent obstacles to Lugg’s thesis. For one thing, he does not attribute any commitment to epistemological naturalism prior to 1912. (Lugg does not explicitly date the emergence of Russell’s naturalism but he does cite 1912’s Problems of Philosophy as evidence of it, so I will assume that he holds Russell’s naturalism to be an active component of his philosophy from then onwards.) For another, even had he done so, it would be feasible to assume that one must tell a different epistemological story with regard to mathematical knowledge to that told about empirical knowledge. Whatever problems Russell’s philosophy of mathematics might face when it comes to explaining how we access the logical truths that mathematical truths are taken to be, these problems may be safely kept in quarantine, leaving the rest of Russell’s epistemology uninfected by them.6 Empiricists (of which, it will be recalled, Lugg thinks Russell is one) have always had to make a special case for logical and mathematical knowledge. If the influence of Wittgenstein’s Tractatus on Russell was as great as some maintain, it may have convinced Russell, as it did the Vienna Circle, that mathematical knowledge is a special case because, being trivial knowledge of analytic truths, it scarcely counts as knowledge at all. It is knowledge of truths which are “all of the same nature as the ‘great truth’ that there are three feet in a yard”.7

Furthermore, there is evidence for Lugg’s claim that Russell was a naturalistic epistemologist. Aside from the Russell texts he cites, the portrait of Russell as an early proponent of naturalized epistemology fits well with Russell’s own characterisation of his philosophy as a “a gradual retreat from Pythagoras” (that is, from the view that mathematical objects and the truths about them are wholly independent of the minds that grasp them). If Russell once believed that secure foundations for knowledge could be uncovered prior to (and distinct from) the gathering of scientific knowledge, he appears to have rejected it by the time he parted company with Pythagoras.8 It is no easy task, though, to determine just when Russell really did turn his back on Pythagoreanism, as he preferred to call the doctrine that most philosophers of mathematics nowadays would not distinguish from Platonism. Some rough location of that point in Russell’s philosophy is surely required, however, if Lugg is to establish his claim that the Russell of The Problems of Philosophy, Theory of Knowledge, and Our Knowledge of the External World is seeking to naturalize epistemology. It is doubtful, to say the least, that the class of Quinean naturalists and the class of Pythagorean realists intersect.

Russell’s memory, I believe, had a tendency to both accelerate and overstate his retreat from Pythagoras in his later recollections of it. This has encouraged some commentators to see Russell as one who, on discovering the theory of descriptions in 1905, wielded Occam’s Razor with all the fury of a demented axe-murderer, slaughtering all but the most indispensable members of his ontology in a violent bloodbath that left reality as he envisioned it by the time Principia was completed, if slightly more populated than that envisioned by the nominalist, then nonetheless comparable in taste to the “desert landscapes” relished by Quine.9 This version of events, propagated in no small measure by Quine himself, has been severely challenged – arguably refuted – in recent years.10 As an account of Russell’s ontological development it is no more than a crude caricature. Russell’s retreat from Pythagoras was more complicated and drawn out than this. For one thing, the theory of descriptions played a somewhat different role in Russell’s philosophy than the one it played when absorbed into Quine’s. For Quine the theory of descriptions was a method of ontological pruning. For Russell it was something more: it was a method of logical construction.

It is not my intention to get drawn here into well-known debates about the ontological status of Russellian logical constructions. I do however want to point out that whatever the theory is employed in constructing, and whatever the ontological status of those constructions, the raw materials of construction are essential to the process. It is here that Russell’s epistemology famously infiltrates his logic, his semantics, and even his metaphysics: the raw materials from which logical constructions are constructed must be items with which the constructor has immediate acquaintance. The paradigm case, of course, is the case of definite descriptions. As sentences containing them contain no corresponding constituent when reparsed into their correct logical form, definite descriptions are “incomplete symbols” and their apparent referents are “logical constructions” the existence of which we need neither deny nor affirm.11 Note that the things we are going to have to be acquainted with in order to understand the propositions expressed by descriptive sentences according to Russell’s principle of acquaintance (“in every proposition that we can apprehend … all the constituents are really entities with which we have immediate acquaintance”)12 are going to have to be just the kinds of things that one would not expect to find obscuring the elegant view provided by any Quinean desert landscape: namely universals or attributes in intension. Since, on analysis, the descriptive sentence ‘G[the F]’ has the logical form

x((Fx & ∀y(Fyx = y)) & Gx),
acquaintance with the universals F and G is needed for its proper understanding.13 Now for the most part, these universals are not of the kind where the problem of explaining our epistemic access to them can be conveniently restricted to the philosophy of mathematics in the way outlined above. When invoking the epistemic relation of acquaintance to explain my understanding of ‘the present King of France is bald’, no presumed privileged access to an a priori realm of mathematical truths will be relevant. Bearing in mind, then, the centrality of the acquaintance relation to Russell’s epistemology, how is it to be explained as a constituent of a naturalistic epistemology?

According to Lugg (p. 16), Russell’s obsession with acquaintance, while alien to Quine, is not fundamentally at odds with his epistemological project. Lugg thinks that the two following quotations, the first from Russell, the second from Quine, are so similar that Quine’s remark contains “more than a slight echo” of the view expressed in Russell’s remark:

The meaning we attach to our words must be something with which we are acquainted. (Russell)14

All inculcation of meanings of words must rest ultimately on sensory evidence. (Quine)15

Contrary to what Lugg says, I do not think there is the slightest hint of an echo here. Quine might not have disapproved of the intrusion of an epistemological principle into a semantic doctrine such as we find here in Russell’s comment. After all, Quine thinks that once naturalized, “epistemology now becomes semantics”.16 But the semantic theory Russell’s epistemological principle is associating with is one that Quine holds to be very bad company. To say that the meanings we attach to our words are things we are acquainted with is to say that the meanings we attach to our words are things we attach to our words. This is just the semantic theory that Quine dismissed as “the myth of a museum”.17 The view that Quine is offering in the above quote is antithetical to such a semantic theory. The inculcation of meanings of words rests on sensory evidence for Quine, because of his commitment to a behaviouristic account of how languages are first ingested by their speakers. The semantic theory associated with Quine’s behaviourism does not assign to our words “something with which we are acquainted”. On Quine’s semantic theory, there is nothing more to the “meaning” of a word than the systematic contribution it makes to determining the conditions under which sentences containing it are true.18 The assignment of truth-values to observation sentences is then bestowed on them by the “tribunal of sense-experience” not as individuals but as holistically united portions of the “web of belief”. There is no place for Russell’s atomistic principle of acquaintance with the meanings of individual words in Quine’s landscape.

The principle of acquaintance, it seems, is a greater obstacle to Lugg’s thesis than he thinks. This is partly because the principle captures the complexity of Russell’s attitude towards empiricism. It is easy to mistake the principle as nothing more than an elaborate statement of empiricism. But it would be mistaken to see the principle this way because it devalues the principle. Russell’s principle is not a recycled relic of early modern philosophy; it is a truly insightful and original contribution to contemporary analytical philosophy. But its proper home is in the philosophy of language, not in epistemology. It places a restriction on what counts as understanding in order to gain a better insight into what the things that we understand are. (At the time the principle is first enunciated these things are Russellian propositions.) Obviously it is an epistemic remark, but it is intended to motivate a semantic theory. That semantic theory is hard to square with an epistemology that “simply falls into place as a chapter of psychology and hence of natural science”,19 as it is a semantic theory that relegates psychology to a position where it is unable to contribute anything to semantic matters.

Russell’s naturalism, in my view, emerges only after the rejection of his anti-psychologism. But this change of heart on Russell’s part was not the result of any epistemological considerations. Nor, for that matter, did it have much to do with his often self-advertised commitment to a scientific method in philosophy. Rather, as Russell made plain in later discussion of this development in his thought, the motivations again stemmed from reflection on semantics: “The problem of meaning is one which seems to me to have been unduly neglected by logicians; it was this problem which first led me, about twenty years ago, to abandon the anti-psychological opinions in which I had previously believed”.20

Having abandoned those opinions, Russell became persuaded that propositions, far from being mind-independent abstract objects, are mental occurrences of some kind. This is a dramatic alteration in Russell’s thought: the robust mind-independent reality of propositions was central to his and Moore’s rejection of Hegelian idealism and their development of analytical philosophy. In rejecting Russellian propositions, Russell was rejecting the very doctrine that most of us who are happy to be called “Russellians” subscribe to. Of course, Russellian propositions had been officially rejected for around a decade by the time Russell endorsed a psychological theory of propositional content. Throughout this period, however, Russell had seemingly nurtured the hope of replacing Russellian propositions with some alternative truth-bearers, such as the judgement-complexes of the multiple-relation theory, that would be compatible with his anti-psychologism. The psychologising of propositional content marks the moment when Russell conceded defeat for his semantic theory.21

Along with many others, I think that Russell was overly hasty in abandoning that semantic project and take it to be the most important of his many lasting contributions to philosophy.22 One philosopher who would certainly not have shared my view, however, is Quine. The psychologised theory of content, in contrast to its more famous Russellian predecessor, quickly took shape in Russell’s writings from 1919 onwards as a theory that is much more in tune with Quinean intuitions. Having located propositions within the domain of psychology, Russell embarked on an extensive attempt to “reconcile the materialistic tendency of psychology with the anti-materialistic tendency of physics”.23 I will not here enter into debate over the degree of success this project, carried out rather fitfully over several years and published in The Analysis of Mind and The Analysis of Matter, had. What is of interest to this discussion is the form that Russell’s psychological analysis of propositions took in that project. What emerges is a causal theory of meaning which substitutes for the principle of acquaintance a causal relation between a word and its meaning. In short, Russell offers an early naturalized semantic theory.

Very early in Russell’s philosophical career he wrote: “That all sound philosophy should begin with an analysis of propositions, is a truth too evident, perhaps, to demand proof”.24 By the time he was endorsing a causal theory of meaning, he clearly could not have held to this view anymore. For now philosophy is surely entitled to help itself to scientific theory in explaining propositional content: meaning is just an object of study for empirical psychology (or perhaps other branches of empirical science) and is not something that can be explained in advance of scientific findings. It is just another element of the causal order. No doubt Quine approved. No doubt he saw similarities with his proposed revamping of epistemology. But the key to Russell’s naturalism is to be found in his theory of meaning, not his theory of knowledge.

I have argued that Russell’s naturalism cannot be present quite so early in his work as Lugg alleges. More importantly, I have argued that this is because the catalyst for Russell’s naturalistic turn was his psychologising of propositional content in 1919. I do not doubt that a naturalistic approach to epistemology is present in Russell’s work after this time. But to present Russell’s epistemology rather than his account of propositional content as the source of his naturalism is to paint a distorted portrait of Russell’s philosophical development.


1 Andrew Lugg, ‘Russell as Precursor Quine’ Bertrand Russell Society Quarterly 128-129, 9-21.
2 Ibid., pp. 18-19.
3 Stevens, ‘Russell’s Re-Psychologising of the Proposition’, in The Russellian Origins of Analytical Philosophy, ch. 5.
4 See Anthony Grayling’s ‘Russell, Experience, and the Roots of Science’ for detailed discussion of Russell’s attitude towards empiricism and pp. 38-41 of Nicholas Griffin’s introduction to The Cambridge Companion to Bertrand Russell for an overview of the exegetical dispute regarding that attitude.
5 See, e.g., Ray Monk’s Bertrand Russell: The Spirit of Solitude.
6 This might be thought to be difficult due to Russell’s use of set-theoretic constructions in his analyses of the alleged denizens of the external world. E.g. physical objects are defined as series of classes of sense-data in Russell’s logical atomist period. But there is no need to appeal here to our knowledge of the raw logical materials out of which classes and series are constructed according to the doctrines of Principia Mathematica in order to explain our knowledge of objects. Rather objects “in themselves” (insofar as it is admissible to use such a locution at all) are constructed out of the immediately available empirical information we already do have (sense-data).
7 Russell, A History of Western Philosophy, p. 860. See pp. 54-58 of my ‘From Russell’s Paradox to the Theory of Judgement’ for discussion of Wittgenstein’s influence on Russell on this point.
8 See, e.g., My Philosophical Development, p. 17.
9 Quine, ‘On What There Is’, p. 4.
10 This applies not just to the immediate motivations behind the development of the theory of descriptions, but also to Russell’s general ontological development, including the ontological status of the theory of types. See my Russellian Origins of Analytical Philosophy (especially chapters 1-4) for a detailed discussion of these points, including an overview of the recent exegetical disputes surrounding them.
11 See, e.g., ‘The Philosophy of Logical Atomism’, p. 273.
12 ‘On Denoting’, p. 56.
13 I am deliberately giving Russell the benefit of the doubt by ignoring the questions of whether the existential quantifier, the conjunction and implication relations and even suitably ontologized variables must also be constituents with which we are acquainted in order to understand ‘G[the F]’ on his account as given in ‘On Denoting’. (Russell did not take the logical constants to be truth functions in 1905 but still maintained his view that they were relations. See chapters 1-3 of The Russellian Origins of Analytical Philosophy for arguments in support of this claim.)
14 The Problems of Philosophy, p. 58.
15 ‘Epistemology Naturalized’, p. 75.
16 Ibid., p. 89.
17 Quine, Ontological relativity and Other Essays, p. 27.
18 See Quine, Word and Object, ch. 2.
19 ‘Epistemology Naturalized’, p. 82.
20 ‘The Relevance of Psychology to Logic’, p. 362. It is also worth noting the explicitly semantic flavour of the title of the paper in which Russell first sets out his new commitment to psychologism: ‘On Propositions: What They Are and How They Mean’.
21 This is not to say that there weren’t benefits to be had from Russell’s psychologistic turn. See my ‘Russell’s Re-Psychologising of the Proposition’ for details.
22 I am not claiming that naturalism is incompatible with what we now call “Russellian” semantics. I am claiming that psychologism is.
23 The Analysis of Mind, p. 114.
24 The Philosophy of Leibniz, § 7.


Grayling, A., 2003, ‘Russell, Experience, and the Roots of Science’, in Griffin 2003.
Griffin, N., ed., 2003, The Cambridge Companion to Bertrand Russell, Cambridge University Press, Cambridge, UK.
Lugg, A., 2006, ‘Russell as a Precursor to Quine’, The Bertrand Russell Society Quarterly 128-129, 9-21.
Monk, R., 1996, Bertrand Russell: The Spirit of Solitude, Jonathan Cape, London.
Quine, W.V., 1948, ‘On What There Is’, in Quine 1961.
Quine, W.V., 1960, Word and Object, MIT Press, Cambridge, MA.
Quine, W.V., 1961, From A Logical Point of View, 2nd ed., Harvard University Press, Cambridge, MA.
Quine, W.V., 1968, ‘Epistemology Naturalized’, in Quine 1969.
Quine, W.V., 1969, Ontological Relativity and Other Essays, Columbia University Press, New York.
Russell, B., 1900, A Critical Exposition of the Philosophy of Leibniz, Allen and Unwin, London.
Russell, B., 1905, ‘On Denoting’, in Russell 1956.
Russell, B., 1912, The Problems of Philosophy, Oxford University Press, Oxford.
Russell, B., 1914, Our Knowledge of the External World, Routledge, London.
Russell, B., 1918, ‘The Philosophy of Logical Atomism’, in Russell 1956.
Russell, B., 1919, ‘On Propositions: What They Are and How They Mean’, in Russell 1956.
Russell, B., 1921, The Analysis of Mind, Allen and Unwin, London.
Russell, B., 1927, The Analysis of Matter, Allen and Unwin, London.
Russell, B., 1938, ‘The Relevance of Psychology to Logic’, Aristotelian Society Supplementary Volume 17: 42-53, reprinted in Russell 1996.
Russell, B., 1946, A History of Western Philosophy, Allen and Unwin, London.
Russell, B., 1956, Logic and Knowledge: Essays 1901-1950, ed. R.C. Marsh, Allen and Unwin, London.
Russell, B., 1959, My Philosophical Development, Allen and Unwin, London.
Russell, B., 1984, Theory of Knowledge: The 1913 Manuscript, ed. E.R. Eames, Allen and Unwin, London.
Russell, B., 1996, The Collected Papers of Bertrand Russell, vol. 10., Routledge, London.
Stevens, G., 2004, ‘From Russell’s Paradox to the Theory of Judgement: Wittgenstein and Russell on the Unity of the Proposition’, Theoria, 2004 (1): 28-61.
Stevens, G., 2005, The Russellian Origins of Analytical Philosophy, Routledge, London.
Stevens, G., 2006, ‘Russell’s Re-Psychologising of the Proposition’, Synthese 151(1): 99-124.

Lecturer of Philosophy
University of Manchester
Manchester M13 9PL
United Kingdom