Bertrand Russell's Naturalistic Epistemology

Author(s): Richard F. Kitchener

Source: Philosophy , Jan., 2007, Vol. 82, No. 319 (Jan., 2007), pp. 115-146
Published by: Cambridge University Press on behalf of Royal Institute of Philosophy

Stable URL: https://www.jstor.org/stable/4619710

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/4619710?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

and Cambridge University Press are collaborating with JSTOR to digitize, preserve and extend
access to Philosophy

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic
Epistemology'
RICHARD F. KITCHENER

Introduction

According to many people, a revolution in philosophy occu

around the turn of the 20th century.2 Some have labelled
approach 'analytic' philosophy, whereas others refer to it as
'linguistic philosophy'. But however we refer to it, something
radical seems to have happened as a result of the views of Frege,
Russell, Moore, Wittgenstein, Carnap, et al. Some ascribe this
revolutionary change to the work of Frege; others see it first
exemplified in the new approach of Russell and Moore; while still
others attribute this change to Wittgenstein. But however we
characterize this new movement, most individuals take Russell to
be a major figure in this analytic movement.
This has resulted in a systematic stress placed on Russell's
early--1904-1913-philosophical work-his logic, philosophy of
logic, theory of denoting and proper names, philosophy of
mathematics, logical atomism, etc.3 As several individuals have

1 An earlier version of this paper was presented at a departmental

colloquium at Colorado State University. I am grateful to the audience for
all of their helpful comments.
2 A. J. Ayer, et al (eds.), The Revolution in Philosophy (London: St.
Martin's, 1960).
3 A. D. Irving & G. A. Wedeking, G. A. (eds.). Russell and Analytic
Philosophy (Toronto: University of Toronto Press, 1993); E. D. Klemke,
(ed.) Essays on Bertrand Russell. (Urbana, IL: University of Illinois Press.,
1970); R. Monk, R. & A. Palmer, A. (eds.)Bertrand Russell and the Origins
of Analytic Philosophy (Bristol: Thoemmes Press, 1966); D. Pears,
Bertrand Russell and the British Tradition in Philosophy (New York:
Random House, 1967); R. M. Sainsbury, Russell (London: Routledge &
Kegan Paul, 1979); C. W. Savage & C, Anderson (eds.), Minnesota Studies
in the Philosophy of Science. Vol. XII: Rereading Russell: Essays in Russell's
Metaphysics and Epistemology (Minneapolis: University of Minnesota
Press, 1989).

doi: 10.1017/S0031819107319050 @2007 The Royal Institute of Philosophy

Philosophy 82 2007 115

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

noted, his later (post-1918) work has not received the atten-
tion has earlier work has received.4
Although I have no wish to denigrate this early period in
Russell's thought, I believe this account misrepresents a central
component of Russell's philosophical interests. For, as he put it,
throughout his philosophical development, 'There is only one
constant preoccupation: I have throughout been anxious to discover
how much we can be said to know and with what degree of
certainty or doubtfulness'.5 Such a concern sounds more like a
passion for epistemology and philosophy of science than for
philosophy of language; and yet there have been fewer discussions
of Russell's epistemology than his linguistic philosophy. What
discussions of Russell's epistemology we do have tend to interpret
it in ways that I believe are mistaken or seriously misrepresent it,
e.g., according to the standard reading of Russell's epistemology,6
Russell was engaged in a more or less traditional epistemological
project of analyzing our concept of knowledge, providing the
foundations of our knowledge, etc. Such a traditional project has
usually rested on the traditional view about meta-epistemology,
about how one should do epistemology: epistemology is first
philosophy. Although Russell's philosophy of science has also been
neglected, a similar account would seem to apply to it.
In this paper I would like to present a different interpretation of
Russell's epistemological program,' one in which he is pursuing a
project that has come to be called naturalistic epistemology (NE).
What I am claiming is that there can be found running throughout
much of Russell's work (especially after 1919, when he took a
psychologistic turn), an account of the nature of knowledge and
epistemology that is naturalistic in spirit, in fact, psychologistic in

4 Although many individuals would place this period as 1904-1918,

there are reasons for not including the years of 'Logical Atomism' in this
period, reasons having to do with Russell's transition from his earlier
theories of judgement to his later psychological theory.
s B. Russell, My Philosophical Development (New York: Simon &
Schuster, 1959), 11. See also 217, 219.
6 R. M. Chisholm, 'Russell on the Foundations of Empirical
Knowledge' , The Philosophy of Bertrand Russell, P. A. Schilpp (ed.) (New
York: Tudor, 1944).
7 In the remainder of this paper, I combine epistemology and
philosophy of science under the rubric 'epistemology'. According to the
canon, philosophy of science just is epistemology-the epistemology of
science. Whether it is to be conceived as a branch of epistemology, applied
epistemology, or the successor to traditional epistemology I leave open.

116

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

spirit, an account that anticipated several of the features to be

found in Quine's later and canonical program of NE.8 I am not
denying, let me stress, that there can be found a traditional account
of epistemology in Russell; clearly there can be. However, this
account does not play the role in Russell's epistemology that many
traditional philosophers believe it plays. If I am correct that Russell
had a NE and was committed to a philosophical naturalism,9 then
not only must we reinterpret Russell's philosophical views, we must
also reinterpret out understanding of the history of analytic
philosophy.
First, I briefly sketch Russell's views about the nature of
philosophy, indicating their potential for a naturalistic interpreta-
tion. Next, in order to get clearer about Russell's conception of
analytic philosophy, I briefly sketch the traditional method of
analysis and synthesis. Then I discuss Russell's own particular
interpretation of this method. Next, I present Russell's version of
NE, showing how it follows from his conception of the nature of
philosophy and the method of analysis. Finally, I discuss Russell's
particular interpretation of this psychological epistemology, leading
him to construct a behavioristic epistemology importantly similar
to that of Quine's. Implications for our understanding of the
history of analytic philosophy are then discussed.

I. The Nature of Philosophy

Throughout his long philosophical career, Russell characterized the

nature of philosophy in a variety of ways. But running throughout
all of his accounts is a conception of philosophy squarely at odds
with the received view concerning the nature of philosophy.

8 W. V. Quine, 'Epistemology naturalized', Ontological Relativity and

other Essays (New York: Columbia University Press, 1969).
9 Although several other individuals, e.g., T. Baldwin, 'Critical
Introduction', Bertrand Russell: An Inquiry into Meaning and Truth
(London: Routledge, 1995); E. R. Eames, 'Introduction', Bertrand Russell:
Theory of Knowledge: The 1913 Manuscript (London: Routledge, 1992);
W. V. Quine, 'Russell's Ontological Development', Bertrand Russell: A
Collection of Critical Essays (Garden City, NY: Doubleday, 1972);
Sainsbury, op. cit., J. G. Slater, 'Introduction', Bertrand Russell: An
Outline of Philosophy (London: Routledge, 1995), have noted this
naturalistic element in Russell, no one, as far as I know, has provided a
systematic account of it.

117

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

According to this traditional view, philosophy provides the

conceptual and theoretical foundations for every intellectual
endeavour including that of science. Philosophy's task may be
described in different ways, but (according to one common
conception) the task of philosophy is to answer scepticism by
securing something like absolutely certain foundations and hence
providing the rationale for all other epistemic pursuits. Science,
ethics and religion are thus logically and conceptually dependent on
philosophy.
Now, throughout his long career, Russell was fairly consistently
committed to a certain version of scientism-towards looking to
science for answers to questions about the nature of the world, the
nature of the mind, and the nature of knowledge (a view we can call
metaphysical scientism). Russell was not opposed to metaphysical
speculations about reality.
More than this, however, Russell believed that it is only by means
of following the scientific method that one can have knowledge,
that the only way of knowing is the scientific way (methodological
scientism). 'I believe', Russell said, 'that there is one method of
acquiring knowledge, the method of science; and that all specially
"philosophical" methods serve only the purpose of concealing
ignorance'.10 There was for Russell no possibility of philosophy as
a 'first philosophy' since there is no philosophical standpoint higher
than science from which to make epistemic pronouncements. Early
in his career, for example, he said: 'There is not any superfine
brand of knowledge, obtainable by the philosopher, which can give
us a standpoint from which to criticize the whole of the knowledge
of daily life'."1 In one of his last works, he echoed these same
remarks: those who advocate logical analysis, Russell said, 'refuse to
believe that there is some "higher" way of knowing, by which we
can discover truths hidden from science and the intellect'.12 In
particular, there is no such thing as philosophical truth (as dis
from scientific truth); there are no philosophical questions tha

10 B. Russell, 'The Meaning of "Meaning" ', The Collected Pape

Bertrand Russell. Vol. 9: Essays on Language, Mind and Matter: 191
J. G. Slater (ed.) (London: Unwin Hyman, 1988; original work publi
1920), 91.
1 Russell, B. Our Knowledge of the External World as a Field for
Scientific Method in Philosophy (London: George Allen & Unwin, 1993;
original work published 1914), 73.
12 B Russell, A History of Western Philosophy (New York: Simon &
Schuster, 1945), 835.

118

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

not scientific questions; there is no philosophical method different

from the scientific method.13 How then can we demarcate
philosophy from science?14
Although philosophy does not differ in kind from scienc
differ in degree or scope: philosophy can be seen to be a
'generalized science'; for, unlike the regular formal an
sciences, which are concerned with detailed and specialized
questions about restricted domains, philosophy is interested in the
most general set of questions about the nature of things and how
we know them. Philosophy aims at a general, theoretical
understanding of the universe and our place in it.15 Philosophy
does this not by virtue of any special philosophical method but
rather (he sometimes suggested) by synthesizing the results of
science into one overall worldview. Philosophy, therefore, is
generalized science.16 In other places, he suggested that the task of
philosophy might be to suggest hypotheses that could then be
empirically validated.

13 B. Russell, 'Logical Atomism', Logic and Knowledge: Essays

1901-1950, R. C. Marsh (ed.) (London: Routledge, 1994; original work
published 1924), 325; B. Russell, An Outline of Philosophy, Rev. ed.
(London: Routledge, 1995; original work published 1927), 239; My
Philosophical Development, 250.
14 In his characteristic style, Russell commented on the relationship
between philosophy and science: The only difference between science and
philosophy, Russell says ('The Philosophy of Logical Atomism', Logic and
Knowledge: Essays, R. C. Marsh (ed.) [London: Routledge, 1994;
originally published 1918] is 'that science is what you more or less know
and philosophy is what you do not know. Philosophy is that part of science
which at present people choose to have opinions about, but which they
have no knowledge about. Therefore every advance in knowledge robs
philosophy of some problems which formerly it had ...' ( 281). Elsewhere
(An Outline of Philosophy), he said that all we can say about a definition of
philosophy 'is that there are certain problems, which certain people find
interesting, and which do not, at least at present, belong to any of the
special sciences' (1).
15 Our Knowledge of the External World, 28, 36; B. Russell, Mysticism
and Logic (New York: Doubleday, 1957; original work published 1917),
105; An Outline of Philosophy, 1-2; My Philosophical Development, 217,
230.
16 Our Knowledge of the External World, 28, 94; 'Philosophy in the
Twentieth Century', The Collected Papers of Bertrand Russell. Vol. 9:
Essays on Language, Mind and Matter: 1919-26, J. G. Slater (ed.)
(London: Unwin Hyman, 1988; original work published 1924), 341; My
Philosophical Development, 108.

119

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

Although Russell's conception of the nature and demarcat

philosophy varied over time in terms of its details, Russell
consistently to have endorsed something like a naturalistic
interpretation. Indeed, he saw such a conception of philosophy to
be the prevalent tendency in 20th century philosophy in general:
The first characteristic of the new philosophy is that it abandons
the claim to a special philosophic method or a peculiar brand of
knowledge is to be obtained by its means. It regards philosophy
as essentially one with science, differing from the special sciences
merely by the generality of its problems, and by the fact that it is
concerned with the formation of hypotheses where empirical
evidence is still lacking. It conceives that all knowledge is
scientific knowledge, to be ascertained and proved by the
methods of science ... It regards knowledge as a natural fact like
another, with no mystic significance and no cosmic importance.17
That Russell could have ascribed such views to individuals such as
Frege and Husserl indicates, at the very least, that his conception o
philosophy is radically at odds with how most contemporary
individuals would characterize such philosophers.1s
Now, on this conception, it would be difficult to construe
philosophy in any of the traditional ways, e.g., as transcendental,
issuing in analytic propositions, as employing a priori methods,
being normative, etc.19 Russell's conception of philosophy is thu
very close to Quine's view that philosophy should be thoroughly
naturalized. This view underlies Russell's suggestion that philoso-
phy should become scientific: it should adopt the logico-analytic
method and proceed the way the sciences do-by following the
scientific method. As Russell conceived it, the scientific method
and philosophical method are not different because both of these
are instances of the method of analysis.

17 'Philosophy in the Twentieth Century', 460-1.

s18 Another aspect of Russell's conception of philosophy is his
ambivalence towards the theoretical vs. the practical nature of philosophy.
He had a kind of love-hate attitude towards both his technical 'analytic'
philosophy and his many popular works on politics, sex, marriage, etc. I
believe most of the history of 20th century analytic philosophy attitude
towards applied philosophy can be traced back to Russell's attitude.
19 At the same time, it should also be pointed out that Russell does
sometimes claims that philosophy is an a priori discipline (Mysticism and
Logic, 105; Our Knowledge of the External World, 190). These passages
must be balanced against other passages where Russell denies there is any
a priori knowledge.

120

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

II. The Method of Analysis/Synthesis

Philosophers typically classify Russell not only as an analytic

philosopher but as one of the founders of analytic philosophy and
for good reasons, for one thing he always insisted upon was that his
method of philosophizing (and the one others should follow) was
the method of analysis. This method of analysis is a version of the
older version of the method of analysis/synthesis (MAS).
It is no accident that one of the founders of analytic philosophy
would have been trained in mathematics, for one of the most
ancient methods of mathematics (especially geometry) practised
since ancient Greek times and continuing throughout the history of
mathematics up until the 20th century was MAS. This method, in
turn, was also part of a long-standing philosophical tradition of
analysis found throughout many of Plato's works: Phaedo (100Aff,
107B), Phaedrus (265d-277c), Meno (86e-87b), Sophist, Parmenides,
and especially the Divided Line Analogy in the Republic (510Bff). It
also played a central part in Aristotle's philosophy: Nichomachean
Ethics (1112b1 5-24), On Sophistical Refutations (175a17-28), Meta-
physics (1051a2lff), the Meterology (376a3ff), his views concerning
definition, division, and classification in the Topics, and his views of
scientific method in the Posterior Analytics (78a6-13; 94a28-35), In
Greek mathematics, it is to be found explicitly addressed in the
writings of Euclid (Elements, the Data), Archimedes (On the Sphere
and Cylinder), Apollonius (On Conic Sections), Pappus (the
Collection), and Diophantus (the Arithmetica). It can also be found
in the methodological views of Galen (the Tegni) and throughout
the Middle Ages (Aquinas, Zabarella, Grosseteste, Peter Albano,
Averroes). In modern times, one can find this method employed
and discussed both by philosophers, mathematicians, and natural
scientists. For example, it plays a central part in the philosophical
methodologies of Hobbes, Descartes, Locke, Leibniz, and Kant as
well as in the works of mathematicians such as Vieta, Descartes,
and Bernoulli; natural scientists such as Galileo, Boyle, and
Newton explicitly employ it in their new experimental philoso-
phy.20 It is perhaps not an exaggeration to claim that it was the

20 A. C. Crombie, Robert Grosseteste and the Origins of Experimental

Science: 1100-1700 (Oxford: Oxford University Press, 1953); N. W.
Gilbert, Renaissance Concept of Method (Columbia: Columbia University
Press, 1960); J. H. Randall, The School of Padua and the Emergence of
Modern Science (Padova: Editrice Antenoire, 1961); M. Otte, & M. Panza,
M. (eds.)Analysis and Synthesis in Mathematics: History and Philosophy

121

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

methodological secret underlying the scientific revolution. In any

case, the method of analysis 'has served as a conceptual model for
some of the most important ideas in the history of philosophy,
including the history of the methodology and philosophy of
science'.21 Those individuals employing the metaphor of the arch
of knowledge22 are explicitly indebted to it.
The exact nature of MAS, however, is a question scholars have
not agreed upon. This is perhaps understandable considering it has
had a history of conceptual change and development spanning 2500
years. Although Euclid was an explicit advocate of the MAS
(Elements, Book III; Data), the most influential and detailed
remarks about this geometrical method are to be found in Pappus of
Alexandria:

Analysis, then, takes that which is sought as if it were admitted

and passes from it through its successive consequences to
something which is admitted as the result of synthesis: for in
analysis we assume that which is sought as if it were already
done, and we inquire what it is from which this results, and again
what is the antecedent cause of the latter, and so on, until by so
retracing our steps we come upon something already known or
belonging to the class of first principles, and such a method we
call analysis as being solution backwards .
But in synthesis, reversing the process, we take as already done
that which was last arrived at in the analysis and, by arranging in
their natural order as consequences what before were anteced-
ents, and successively connecting them one with another, we
arrive finally at the construction of what was sought; and this we
call synthesis.
Now analysis is of two kinds, the one directed to searching for
the truth and called theoretical, the other directed to finding what
we are told to find and called problematical. (1) In the theoretical
kind we assume what is sought as if it were existent and true,
after which we pass through its successive consequences, as if
they too were true and established by virtue of our hypothesis, to

(Dordrecht: Kluwer, 1997); R. E Kitchener, Holism: An Analytic

Approach. Unpublished manuscript. (1995).
21 J. Hintikka, & U. Remes, The Method of Analysis: Its Geometrical
Origin and its General Significance (Dordrecht: D. Reidel, 1974), 1.
22 D. Oldroyd, The Arch of Knowledge (New York/London: Meth-
uen.1986).

122

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

something admitted: then (a), if that something admitted is true,

that which is sought will also be true and the proof will
correspond in the reverse order to the analysis, but (b), if we
come upon something admittedly false, that which is sought will
also be false. (2) In the problematical kind we assume that which
is propounded as if it were known, after which we pass through
its successive consequences, taking them as true, up to something
admitted: if then (a) what is admitted is possible and obtainable,
that is, what mathematicians call given, what was originally
proposed will also be possible, and the proof will again
correspond in the reverse order to the analysis, but if (b) we
come upon something admittedly impossible, the problem will
also be impossible.23
One of the most puzzling features of the above remarks is the
problem of the direction of analysis: does analysis proceed
downward-from an axiom or set of axioms to a theorem in a
deductive fashion-or does it proceed upwards-from a theor
something from which the theorem deductively follows? I
first, then both analysis and synthesis would involve dedu
processes, but if the second, then analysis would involve a
non-deductive inference to a suitable proposition, something like an
abductive inference or inference to the best explanation.
Although the downwards interpretation, in which both analysis
and synthesis are deductive, has had its defenders, I favour the
upwards interpretation, which has had its own champions. On this
interpretation, the direction is to proceed inductively from assumed
datum to (known) principle and then deductively from principle to
new results. Suppose you have a proposition, p, which you take to
be true (a datum), but which you want to show to be true by
proving it. In order to do this, you need a principle or set of
principles, q, which provide the basis of your proof. The step of
analysis consists of an 'upstream' movement or inference (bottom-
up) from your p to those principles q, whereas the step of synthesis
is a 'downstream' movement (top-down) consisting of an inference
or movement from q to p (See Figure 1).
The step of synthesis is one with which we are very familiar: it is
deduction, the deduction of a theorem (the datum) from a set of
axioms (the principles). But throughout the centuries, analytic
methodologists expanded this step of synthesis to include other

23 T. L. Heath, A History of Greek Mathematics (2 Vols.) (New York:

Dover, 1981; original work published 1921), 400-401.

123

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

Principle q

Analysis Synthesis
(induction) (deduction)

Datum p

Figure 1. The Method of Analysis/Synthesis

kinds of relationships. In a causal context, for example, the

downward movement of synthesis is a movement from cause to
effect, and in an ontological context, it is the process by which the
basic elements (e.g., atoms) give rise to or compose a larger whole.
Hence, synthesis is an operation of 'adding to' or 'putting
together', thus Kant's notion of a 'synthetic judgment'. In all of
these cases, one begins with the axiom, cause, or set of atoms, and
proceeds to their consequences, effects, or whole. But how do we
arrive at these original axioms, causes, or atoms? How do we
discover or infer them? This occurs by analysis.
Analysis is the inverse of these synthetic operations: it is not a
deductive process, but an inductive one-inducing a set of axioms
from a mathematical or logical datum, inferring a cause from its
effects, decomposing a whole into its parts.
Although the downward part of the analysis-synthesis method is
well-known and represents a well-worn mode of thought, the
upward step of analysis is less familiar. How do we perform an
inductive inference and arrive at a geometrical axiom? How do we
infer the cause of something? How do we decompose an entity into
its parts? How, in short, do we come up with these initial
hypotheses? Aristotle invoked a process of intuition-intuitive
induction (epagoge)-and lodged it in nous. Galileo and Newton
also believed the mind had the ability to analyze a phenomenon
(e.g., motion) and to arrive at its basic constituents or variables
(e.g., velocity, distance, time). Others have argued that the basic
pattern of inference here should be called abduction or inference to
the best explanation. According to Peirce,24 for example, an
abductive inference has the following form:

24 C. S. Peirce, 'Pragmatism and abduction', Collected Papers of

Charles Sanders Peirce. Vol. IV Pragmatism and Pragmaticism, C.
Hartshorne & P. Weiss (eds.) (Cambridge, MA: Harvard University Press.
1934).

124

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

The surprising fact, C, is observed

But if H were true, C would be as a matter of course
Hence, there is reason to suspect that A is true.
This has the logical form:
C
If H then C
Hence, A
According to Harman,25 this type of inference is what he calls a
inference to the best explanation (IBE). Here, 'one infers, from th
fact that a certain hypothesis would explain the evidence, to the
truth of the hypothesis'.26 But, if these are several hypotheses th
would explain this evidence, one rejects all other hypotheses and
infers that the remaining one is the best. According to Harman,
hypothesis is judged to be better than the others because it satisfie
certain criteria: simplicity, plausibility, explanatory power, degre
of ad hocness, etc.
In general, there will be several hypotheses, which might explain
the evidence, so one must be able to reject all such alternative
hypotheses before one is warranted in making the inference.
Thus one infers, from the premise that a given hypothesis woul
provide a 'better' explanation for the evidence than would any
other hypothesis, to the conclusion that the given hypothesis is
true. 27

The expanded IBE argument might, therefore, be construed in the

following way:
We observe some phenomenon C
But if H1 were true, C would be expected
If H2 were true, C would be expected
If H3 were true, C would be expected
H1 is simpler, more plausible, explains more, and is less ad hoc
than H2, H3, etc.
Therefore, H1 is likely
Although there is some confusion as to whether such an inference is
deductive or inductive, I think we must construe it as an inductive
inference, an inference from data to hypothesis; as such it

25 G Harman, The inference to the best explanation, Philosophical

Review, 74 (1965).
26 Ibid, 89.
27 Ibid.

125

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard F. Kitchener

constitutes a crucial step in the method of analysis (or the m

of analysis/synthesis). Furthermore, I will suggest, this
constitutes a large part of what Russell conceived the scien
method to be. Since his new 'scientific philosophy' was to e
the scientific method, this meant that the new philosophical
method was the MAS.

III. Russell's Analytic Methodology

In an early 1907 paper entitled 'The Regressive Method of

Discovering the Premises of Mathematics',28 Russell begins with
the statement

My object in this paper is to explain in what sense a

comparatively obscure and difficult proposition may be said t
be a premise for a comparatively obvious proposition, to conside
how premises in this sense may be discovered, and to emphasi
the close analogy between the methods of pure mathematics an
the methods of the sciences of observation.29

One can look at this method either in its upward direction

(resolution) or its downward direction-what others have called
composition. The upward direction concerns what Russell calls the
order of knowledge, whereas the downward direction concerns the
order of exposition (sometimes called the logical order).
In mathematics and the natural sciences, we begin with a certain
fact or datum, which Russell calls the empirical premise. This
epistemic premise has a certain degree of certainty or intuitive
obviousness and is the basis of our knowledge. In mathematics as
well as the empirical sciences, the initial propositions believed-the
empirical premises-must be 'intrinsically obvious': 'In the natural
sciences, the obviousness is that of the senses, while in pure
mathematics it is an a priori obviousness, such as the law of
contradiction.'30
We can ask either what consequences follow from this empirical
premise or from whence it comes; in the latter case this will be
another proposition Russell calls the logical premise. This 'is some

28 B. Russell, 'The Regressive Method of Discovering the Premises of

Mathematics', Essays in Analysis, D. Lackey (ed.) (New York: Braziller,
1973).
29 Ibid, 272.
30 Ibid, 279.

126

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

logically simpler proposition or propositions from which, by a valid

deduction, the proposition in question can be obtained'.31 Hence,
the empirical proposition is deduced from the logical premise,
whereas the relation between the empirical premise and the logical
premise is an inductive one. For example, suppose we have the
premise 2 + 2 = 4. This generalization was no doubt, Russell says,
actually obtained from other, more particular propositions, such as
2 sheep + 2 sheep = 4 sheep, 2 trout + 2 trout = 4 trout, etc.; here
we have an inference in the upward direction, the order of
knowledge, and, in this example, the inference is enumerative
induction.
These particular propositions about sheep and trout constitute
the empirical premises underlying our inductive inference. But the
empirical premises are deducible from the logical principle 2 + 2 =
4; this constitutes the order of exposition, which is deductive.
Hence, although we knew 2 sheep + 2 sheep = 4 sheep before we
knew 2 + 2 = 4,32 when we know the latter, we know 'the reason
why we believe the given proposition.' (see Figure 2).

Logical Premise
P

Epistemic Order Logical Order

(Inductive) (Deductive)

q
Empirical Premise

Figure 2. The Epistemic Order vs. the Logical Order

Russell is an inductivist (or better: inductive-deductivist) in his

philosophy of science, but, more surprisingly, he is an inductivist
also in his philosophy of mathematics. According to Russell,
inductive inferences occur in the analytic stage of mathematical
analysis:

31 Ibid, 272-273.
32 In one place ('The Philosophy of Logical Atomism', 180), Russell
claims that such an inductive inference should be seen as a causal,
psychological relation.

127

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

... the inferring of premises from consequences is the esse

induction; thus the method in investigating the princi
mathematics is really an inductive method, and is substan
the same as the method of discovering general laws in any
science.33
Russell is careful to point out, however, that not all inductive
inferences are cases of enumerative induction, since they also include
what has been called inferences to the best explanation. When we
infer a logical premise from an empirical premise, the logical
premise is complex, whereas the empirical premise is simple. In
such cases, Russell seems to be employing something like inference
to the best explanation (IBE) model.
But, more than this, induction enters at the synthetic stage.
Suppose q is our empirical premise and p is our logical premise, and
suppose p deductively implies q.
But p is only believed on account of q. Thus we require a greater
or less probability that q implies p, or, what comes to the same
thing, that not-p implies not-q. If we can prove that not-p implies
not-q, i.e., that p is the only hypothesis consistent with the facts
that settles the question. But usually what we do is to test as
many alternative hypotheses as we can think of. If they all fail,
that makes it probable, more or less, that any hypotheses other
than p will fail.34

Russell concludes, therefore, that mathematical knowledge, as

contrasted to mathematical proof, is inductive in nature and hence
probable. Hence, mathematical knowledge is not different, in
principle, from the knowledge obtained in the natural sciences. In a
reply to Poincar6, Russell makes this surprising claim:
The method of logistic is fundamentally the same as that of
every other science. There is the same fallibility, the same
uncertainty, the same mixture of induction and deduction, and
the same necessity of appealing, in confirmation of principles, to
the diffused agreement of calculated results with observation ...
In all of this, logistic is exactly on a level with (say) astronomy,
except that, in astronomy, verification is effected not by intuition
but by the senses. The 'primitive propositions' with which the
deductions of logistic begin should, if possible, be evident to
intuition; but that is not indispensable , nor is it, in any case, the

33 Ibid, 274.
34 Ibid, 274-275.

128

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

whole reason for their acceptance. This reason is inductive,

namely, that among their known consequences (including
themselves), many appear to intuition to be true, none appear to
intuition to be false, and those that appear to intuition to be true
are not, so far as can be seen, deducible from any system of
indemonstrable propositions inconsistent with the system in
question.35
For Russell, therefore, mathematical and logical knowledge is
(partly at least) inductive in nature36 and the formal sciences are not
different in kind from the natural sciences. This is because all the
sciences (and philosophy) employ the MAS.
The various sciences are distinguished by their subject matter,
but as regards method, they seem to differ only in the
proportions between the three parts of which every science
consists, namely (1) the registration of 'facts', which is what I
have called empirical premises; (2) the inductive discovery of
hypotheses, or logical premises, to fit the facts; (3) the deduction
of new propositions from the facts and hypotheses.37
It is not unreasonable to conclude, therefore, that, for Russell,
mathematics and logic are not different in degree from the natural
sciences but (at most) in degree. Such a view seems indistinguish-
able from that of Quine's.

Scientific Philosophy

Russell argued that there is an underlying similarity in the method

employed by the various sciences-the method (roughly) of
analysis followed by synthesis, which is essentially equivalent to the
scientific method. There is, therefore, a single thing called the

35 B. Russell, (1906/1973). 'On "insolubilia" and their solution by

symbolic logic', Essays in Analysis D. Lackey (ed.) (New York: George
Braziller, 1973; original work published 1906), 194. Russell points out that
intuition is not infallible and certain.
36 He echoes this view later in this life: 'The reason for accepting an
axiom, as for accepting any other proposition, is always inductive, namel
that many propositions which are nearly indubitable can be deduced from
it, and that no equally plausible way is known by which these proposition
could be true if the axiom were false, and nothing which is probably fals
can be deduced from it' (My Philosophical Development, 121).
37 'The Regressive Method', 282.

129

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard F. Kitchener

scientific method and it is employed in all of the sciences. H

does philosophy fit into this general account? How, if at all,
Russell demarcate philosophy from the formal and factual scie
If there is no sharp distinction between the methods of the f
and the factual sciences, and if philosophy pursues the meth
analysis/synthesis, philosophy is (or must become) indisting
able from science. This comes out clearly in Russell's work of
e.g., in his article on 'Scientific Method in Philosophy',38 and
major epistemological work of this period-Our Knowledge of
External World, whose subtitle is: as a field for scientific meth
philosophy. Philosophy, properly conceived, engages in analy
such it ideally is and can become scientific by using the scie
method. The result would be what he called a scientific
philosophy-in particular, a scientific epistemology-a new kind of
philosophy that practiced the scientific method in its philosophiz-
ing. This new method was termed various things, e.g., the
logico-analytic method, the method of construction, etc.-but how
Russell imagined it to be an instance of the scientific method and
what he takes the scientific method to be remain to be decided. If I
am correct about his views of scientific method being roughly
equivalent to the MAS, then we have an answer to our question:
philosophy is to become scientific by pursuing the MAS. But the
MAS has two directions-upwards and downwards-with the
deductive sciences pursuing the downward path of synthesis and
the inductive sciences pursing the upward path of analysis. How
does philosophy (epistemology) fit into this scheme? Is philosophy
to be assigned the upward path, the downward path, or a
combination of the two? If the first, then it would be equivalent to
the deductive sciences and philosophy would be assimilated to
logic; if the second, it would be an inductive science no different in
principle from physics and psychology.39
On this question Russell is far from clear. On the one hand, he
sometimes suggests that philosophy is equivalent to logic. In his
1914 work, for example, chapter 2 is entitled 'Logic as the essence

38 Mysticism and Logic.

39 D. J. O'Connor ('Bertrand Russell', A Critical History of Western
Philosophy, D. J. O'Connor (ed.) [New York: Free Press, 1964]) suggests
that, by the scientific method Russell means the method of the formal
sciences, namely, the methods of logic and mathematics. If this means the
method of deduction, then I believe such a view is incorrect (especially if
one sharply distinguishes the formal from the factual sciences).

130

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

of philosophy', a point he makes elsewhere,40 and a point not

wasted on later thinkers such as Carnap, who advocated the same
thing (e.g., philosophy is just logical syntax). But, as we have seen,
what he means by this and how he conceives logic remain opaque.
Philosophy might be equivalent to formal deductive logic, engaging
in the downward step of synthesis. But Russell does not restrict
mathematics and logic to pure deduction. On the other hand,
philosophy might be identified with logic in a broader sense-
deductive and inductive logic; on this account, philosophy is both
analysis and synthesis. Sometimes,41 this seems to be what Russell
has in mind, namely, philosophy is logical analysis followed by
logical synthesis. Finally, Russell sometimes distinguishes philoso-
phy from (deductive) logic and, following a Kantian tradition, sees
philosophy's task as that of analysis (which Russell would describe
as being inductive), whereas the task of mathematics is synthesis
(deduction). This comes out clearly in Our Knowledge of the
External World, where Russell contrasts philosophy and mathemat-
ics:

Mathematics and philosophy differ ... in their manner of tre

the general properties in which all possible worlds agree; fo
while mathematics, starting from comparatively simple pro
tions, seeks to build more and more complex results by dedu
synthesis, philosophy, starting from data which are common
knowledge, seeks to purify and generalize them into the simplest
statements of abstract form that can be obtained from them by
logical analysis.42
Philosophy thus engages in analysis, mathematics in synthesis.
Similar questions can be raised concerning the nature of
epistemology and its relation to logic and psychology: How is
epistemology different from psychology and logic? On the one hand
it is tempting to equate epistemology with logic-deductive logic;
however, this is a move that Russell does not make and in fact
guards against making: epistemology, he often says, is not the same
as logic (although it is related to logic). Should one, then, equate
epistemology with psychology-with the more proper step of
analysis? If so, then epistemology would be reducible to
psychology-a version of psychologism. This is a move Russell also
resists:

40 The Principles of Mathematics, 2nd ed. (New York: W. W

1938), 42.
41 'The Philosophy of Logical Atomism', 34.
42 Our Knowledge of the External World, 190.

131

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

The analysis of experience, the distinction between sensation,

imagination, memory, attention, etc., the nature of belief or
judgment, in short all the analytic portions of the subject, in so
far as it does not introduce the distinction between truth and
falsehood, must, I think, be regarded as strictly part of
psychology. On the other hand, the distinction between truth and
falsehood, which is plainly relevant to the theory of knowledge,
would seem belong to logic.43
In his early period, Russell accepted Brentano and Meinong's
act-content-object distinction involving mental states. The content
of thought is propositional and belongs to the logic of propositions,
which (in the present case) is the epistemological order. This, he
says, includes the consideration of truth and related notions such as
verification. But the act of thinking belongs to psychology. Hence, a
full account of knowledge requires both psychology and logic.
Epistemology, therefore, is just psychology plus logic; psychology
engages in analysis, whereas logic engages in synthesis. It thus
appears that epistemology is a hybrid discipline, involving both
empirical science and logic. But Russell proceeds to argue that even
the logical part of epistemology-the search for epistemological
order-can not be separated completely from psychology.44 It thus
appears that epistemology is not an autonomous, philosophical
discipline. But this raises the question of the very status of
epistemology: is it a philosophical discipline or is it to become a
science such as psychology? Since it does not appear to be
autonomous, the possibility that epistemology can be naturalized
must be explored.

IV. Russell's Naturalistic Epistemology

Throughout Russell's many writings on epistemology, two different

conceptions of epistemology can be found.45 On the one hand,

43 B. Russell, Theory of Knowledge: The 1913 Manuscript, E.R.

Eames (ed.) (London/NY: Routledge, 1984). 46.
44 Ibid, 40.
45 B. Russell, The Problems of Philosophy (Amherst, NY: Prometheus.
1984; original work published 1912); 'The Philosophy of Logical
Atomism'; An Outline of Philosophy; B. Russell, An Inquiry into Meaning
and Truth (London: Routledge. 1995; original work published 1940); B.
Russell, Human Knowledge: Its Scope and Limits (London: Routledge,
1992; original work published 1948).

132

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

there is a more s traditional Cartesian account of the nature and

task of epistemology: according to this conception, the primary
of the epistemologist is to answer the sceptic and to show tha
knowledge is possible by showing that we do have certain,
indubitable knowledge. From these secure foundations, one then
shows how it is possible to derive the rest of one's knowledge. Until
one secures these certain foundations, however, no other kind of
knowledge can be assumed to exist. Russell often wrote about
epistemology in this way-as a concern for the epistemological
ordering of one's epistemic states; as a result some individuals46
(e.g., Chisholm, 1944) take this to be the only kind of epistemology
to be found in Russell and hence that Russell is a traditional
epistemologist.
Alongside this traditional concern of epistemology, ho
there is another, quite different conception of epistemolo
found in Russell-a NE. The history of this other concep
epistemology has not been charted. It started off, I would s
as a kind of traditional Kantian epistemological project-of
how it is possible for us to have the knowledge we do have
conception can be found, for example, in Russell's An Essay
Foundations of Geometry.47 This original kind of transcen
question was gradually turned into a different kind of epi
logical question, however, for the question about what m
operations and faculties are necessary in order to have the
knowledge we do have was turned into an empirical psychological
question.48 The result was thus a kind of descriptive or naturalistic
epistemology founded upon the sciences of physics, physiology, and
especially psychology. It is this second conception of
epistemology-an inchoate NE-that has largely gone unnoticed i
Russell.
The Cartesian conception of epistemology is the more traditiona
view, the notion that the epistemologist must confront and answe
epistemological scepticism. This is driven by the doubt one ca

46 Chisholm, op. cit.

47 B. Russell, An Essay on the Foundations of Geometry (London
Routledge & Kegan Paul, 1899).
48 Here, the important work seems to be G. E. Moore's critique of the
Kantian program (G. E. Moore, 'Review of Russell, An Essay on the
Foundations of Geometry', Mind, 8 [1899]), which included his critique o
Russell's distinction between the transcendental and the psychological.
Moore's criticism was that 'the Kantian fallacy' (as he called it) consiste
of confusing what is known with what is true, where the former belongs to
psychology.

133

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

have concerning a proposition. An epistemological answer can only

come from the certainty attaching to immediate experience; hence,
a Cartesian-type epistemology is based upon a first-person
perspective sometimes called internalism. The epistemological task
is to show how one can move from this initial stage of subjective
immediacy to our common sense and scientific knowledge.
In addition to this Cartesian conception, or what Chisholm49
calls methodism, there is another epistemological approach, one that
is much more in accord with a Kantian conception (Chisholm calls
this particularism). Here one begins with items that are clear-cut
examples of knowledge. Ignoring the initial challenge of the
sceptic, one's epistemic task is thus to show how such knowledge is
possible. Epistemologists sometimes call this externalism, since it is
concerned with what is known by other people. This epistemologi-
cal strand emerged very early in Russell and continued to press
upon his attention.
Russell's initial foray into epistemology occurred in 1913 in
which he began writing 'the big book' on epistemology; the result
was Our Knowledge of the External World and his Epistemology
manuscript of 1913.50 After a brief interval, Russell reports he
returned to epistemology in 1918. 'I found my thoughts turning to
theory of knowledge and to those parts of psychology and of
linguistics which seemed relevant to that subject. This was a more
or less permanent change in my philosophical interests.'51 This
epistemological change involved Russell turning towards psychol-
ogy (in particular behaviorism) for assistance in answering
epistemological questions.

Russell's Psychological Turn

This change in Russell's epistemological program, his turning

towards psychology, occurred as a result of the felt inadequacy of
his earlier theories of judgment. Russell's initial theory of
judgment and propositions-his 'binary theory'52-soon gave way

49 R. M. Chisholm, 'The Problem of the Criterion', The Foundations

of Knowing (Minneapolis: University of Minnesota Press, 1982).
50 Theory of Knowledge: the 1913 Manuscript.
51 My Philosophical Development, 128.
52 The Principles of Mathematics.

134

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

to his 'multiple relation' theory of judgment of 190653 which was

maintained (with subsequent revisions) in his Problems of Philoso-
phy, his 'big book on epistemology' of 1913, his Our Knowledge of
the External World work of 1914, and his lectures on logical
atomism of 1918. A central issue here concerned the question of
'the unity of the proposition'-how to explain the fact that a
judgment (or proposition), e.g., the cat is on the mat, is not just a
concatenation of terms-cat, on, mat-but has a certain kind of
unity to it. The options seems to be three-lodge this unity in the
objective realm, in the linguistic realm, or the psychological realm.
Russell (initially) took the first course and then the third,
suggesting that, in some way, the unity of the proposition or
judgment was due to an action of the mind. The unity of the
proposition thus turned on the question of the nature of this
mental operation-the act of believing-and this, Russell thought,
was a psychological question.
Wittgenstein had raised devastating objections to Russell's
multiple relation theory of judgment in personal conversations with
him in 1913, resulting in Russell's abandoning his 'big book' on
epistemology (but still retaining it in Our Knowledge of the External
World). Russell, perplexed by the problem (as he put it) of 'facts,
propositions, and judgments' during the war years, thus turned to
psychology for a more adequate theory of belief. In a letter to his
brother (July 8, 1918), Russell says: it was for the sake of 'facts,
judgments and propositions'
that I wanted to study behaviorism, because the first problem is
to have a tenable theory of judgment ... All the psychology that I
have been reading and meaning to read was for the sake of logic;
but I have reached a point in logic where I need theories of (a)
judgment (b) symbolism, both of which are psychological
problems.54

As a matter of fact, Russell seems to have rather consistently

maintained that all problems of meaning are psychological. In a
later work he said : 'I think that the notion of meaning is always
more or less psychological, and that it is not possible to get a pure

53 B. Russell, Philosophical Essays (London: Routledge, 1994; original

work published 1910).
54 B. Russell, 'Manuscript notes [1918]', The Collected Papers of
Bertrand Russell. Vol. 8. The Philosophy of Logical Atomism and other
Essays, 1913-1919, J. G. Slater (ed.) (London: George Allen & Unwin.
1986), 249.

135

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard F. Kitchener

logical theory of meaning, nor therefore of symbolism.'ss He

always to have believed this. It was psychology, therefore, t
Russell thought could provide an answer to the question, Wh
the nature of belief? Russell initially thought that behaviori
might be able to answer this question, but his hopes were da
nevertheless, he continued to maintain it was a psycholog
question.
As Russell immersed himself in psychology, especially during
the year 1918 while he was in prison, he struck out on a naturalistic
path. At first, this NE was tacit. In his Lectures on Logical
Atomism of 1918, he first entertained the possibility of a
behavioristic analysis of 'belief' ascribing such a view to James and
Dewey.56 In his article, 'On Propositions',57 he is even more
favourably inclined towards a behavioristic analysis of 'belief' and
'desire', this time explicitly introducing the views of the behaviorist
John Watson. It is here that Russell first suggested the possibility
of a naturalistic (psychologistic) theory of knowledge.8s Such an
approach then became explicit in Russell's The Analysis of Mind59
and An Outline of Philosophy, where his infatuation or flirtation
with behaviorism is most evident.60 Although these two works
contain Russell's most self-conscious program of NE, it is still
present in An Inquiry into Meaning and Truth and Human
Knowledge. Hence, until the very end of his life, Russell entertained
the possibility of a NE (although not a purely behavioristic one).

A Behaviouristic Epistemology

The basic tenet of Russell's NE is that knowing is a natural state in

the world, basically a relation between an organism and its
environment. 'The world, as presented by science, contains a
phenomenon called "knowing", and theory of knowledge ... has to
consider what sort of phenomenon this is'.61 Knowledge,
considered to be a naturalistic phenomenon, observable by the
ss 'The Philosophy of Logical Atomism', 186.
56 Ibid., 219.
57 B. Russell, 'On propositions', Logic and Knowledge: Essays, R. C.
Marsh (ed.) (London: Routledge, 1950; original work published 1919).
58 Ibid., 307.
59 B. Russell, The Analysis of Mind. (London: Allen & Unwin, 1921).
60 See R.E Kitchener, 'Bertrand Russell's Flirtation with Behavior-
ism', Behavior and Philosophy, 32 (2004).
61 An Inquiry into Meaning and Truth, 12-13.

136

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

senses, and studied like any other natural phenomenon by science,

in this case, psychology, would involve taking up a third-person
point of view, that of the epistemologist, and observing the
epistemic subject as it engages in various kinds of epistemic
activities resulting from various kinds of epistemic states.
'Knowledge, traditionally, has been viewed from within, as
something which we observe in ourselves rather than as something
which we see others displaying'.62 Instead, he suggested, know-
ledge could also be viewed from an external point of view-'man
from without'63 although this would also include animals besides
man. NE was thus (initially) committed to an externalist
perspective.
For a behaviourist, the activity of knowing must consist of
observable behaviour of the organism in an environment. When we
do observe such behaviour, we seem to be committed to a belief in
a causal sequence of events running from the remote environment
to the proximal environment to hidden processes in the nervous
system to proximal responses and finally to distal responses. It is
this complex causal sequence that must be the locus of knowledge;
hence Russell is firmly committed to a causal theory of knowledge.
As Russell developed his NE, he also come to endorse a causal
theory of belief, a causal theory of justification, a causal theory of
truth, and a causal theory of meaning.
Russell gave several behaviouristic accounts of knowledge, but
all of them included the notion that knowledge is a certain kind of
response (e.g., a verbal answer) to a stimulus (e.g., a question). We
can observe such behaviour in other humans and in animals, the
best example of which involves 'knowing-how' to do a certain kind
of task and hence the ability to reach a certain kind of goal. Russell
calls this animal knowledge and distinguishes it from the mechanical
response an instrument, such as a thermostat, makes to an external
stimulus.
Could a machine, a thermostat, be said to 'know' on the basis of
its fixed and mechanical response to a stimulus? One might, Russell
concedes, call this a kind of knowledge-a very low-level kind of
knowledge. Hence, we could say a thermostat knows the
temperature, a compass knows the direction, or a barometer knows
the weather. Some epistemologists have been impressed with this

62 An Outline of Philosophy, 14.

63 The Analysis of Mind, 28-29; An Outline of Philosophy, 14, chapter
8; An Inquiry into Meaning and Truth, 12-13.

137

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

kind of machine knowledge. David Armstrong64 for exa

advocates such a 'thermometer model of knowledge' in his
version of NE, and Ernest Sosa65 calls this servo-mechanic
knowledge. Although Russell is also willing to concede that there is
such knowledge possessed by instruments, he does not believe such
examples are adequate as an account of other, higher forms of
knowledge, such knowledge as is possessed by animals. Animals,
unlike machines, manifest behavioral properties that thermometers
don't, namely, they learn. Hence, there appears to be what can be
called animal knowledge, the kind of knowledge animals demon-
strate in their improvement in behaviour over time as they acquire
and perfect a skill or ability. A naturalistic account of knowledge,
therefore, must secure a place for such improvement in response
over time as a result of environmental input, and this will
necessitate a stress on the plasticity and adaptiveness of behaviour.
To incorporate these aspects into a behaviouristic epistemology,
one would also have to carve out a behaviouristic account of desire
and purpose.
Russell briefly addressed these issues in The Analysis of Mind
where he argued that for something to be a case of animal
knowledge, it must possess the characteristics some individuals
would call 'persistence until' a goal is reached, where such
persistence is manifested in behavioural variability (plasticity) that
is correlated with goal attainment. Such behaviour is goal-directed
and shows purpose.66 Such an account shows a strong family
resemblance to the views of the neo-realists Edwin Holt and Ralph
Perry, whose early works Russell read.67
Animal knowledge, e.g., an animal knowing how to obtain a goal,
is a kind of knowledge higher than servo-mechanic knowledge. A
comprehensive epistemology, therefore, must account for both
types. Of course, the kind of knowledge most epistemologists have
been concerned with is propositional knowledge, knowing-that
something is the case. Such knowledge seems to involve beliefs,

64 D. Armstrong, Belief Truth and Knowledge (Cambridge: Cam-

bridge University Press. 1973).
65 E. Sosa, Knowledge in Perspective: Selected Essays in Epistemology
(Cambridge: Cambridge University Press, 1991).
66 An Analysis of Mind, 63.
67 E. B. Holt, The Freudian Wish and its Place in Ethics (New York:
Henry Holt, 1915); R. B. Perry, 'Purpose as Tendency and Adaptation',
Psychological Review, 26 (1917).

138

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

which are what Russell called propositional attitudes; and such

beliefs must not only be true, they must fulfill an evidence
criterion.

Russell's account of belief is a complex and changing story b

beginnings about 1918, he abandoned his earlier model of a
belief-state as involving an act-content distinction and began
constructing a new theory of belief, a naturalistic theory of belief.
Russell had always been reluctant to accept the existence of
propositions as separate entities in the world. Instead, he was
inclined to apply Occam's razor and to treat them as logical fictions
(along with numbers and classes).
In his new theory of belief, the old self-act-content-objective
schema was abandoned, inspired by neutral monism, and the
ontological status of at least several of these entities was
questioned. Instead, he analyzed a belief state into his neutral
monistic elements-sensations and images-at least for the old self,
act, and content parts. A belief state now consisted of images plus
feelings (sensations), with images now taking over the role that
propositions once played, and feelings accounting for the attitudi-
nal part of propositional attitudes, where the feeling is that of
expectation (as opposed, say, to desire). Hence, when S believes
that the cat is on the mat, this is to be analyzed as (1) S has an
image of a cat on mat, and (2) S has a feeling of expectation, of
expecting the cat to be on the mat. The image has one kind of
representational function, the lion's share as it were, and the
expectation has a different one, a more behavioristic one.

Russell's Causal Theory of Knowledge

In adopting the older empiricist account of images as the vehicle of

meaning, in which images are copies of impressions, Russell took
over much of their intellectual framework. An external situation,
say, A, Russell suggested, can cause an image of it, say, B in which
case B is said to be the prototype of A, which is the meaning of B.
More precisely: A is an image of B if and only if B is the prototype
of A. This in turn is to be analyzed causally:
B is the prototype of A iff

(1) There is a resemblance (of structure) between B and A

(2) B causes A
(3) A and B have common effects.

139

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

The key component in all three clauses is that of causality. By

means of this notion, Russell proposes to provide a naturalistic
account of the truth and evidence.
In order for S to know that p, it is necessary that S have adequate
evidence for p, that S's belief that p be rational, that S's belief that
p be verified or confirmed, etc. Opinions differ on the precise way
of formulating this requirement of justification. Russell's describes
it in various ways, including that of verification. How, then, can one
naturalize this condition? The answer is by means of a causal
theory of evidence or verification.
Russell strongly opposed several of the views of the logical
positivists regarding the much-debated issue of protocol sentences
and their relation to experience. Rejecting their popular coherence
theories of verification, Russell suggested there must be a causal tie
between a protocol sentence (or better the acceptance of a protocol
sentence) and the experience in question. But more than this, since
several of the positivists also accepted a causal condition, Russell
offered a causal interpretation of verification itself, in fact, a
psychological interpretation of verification rooted in the feeling of
expectedness that is one crucial part of his account of belief.68 In
his article 'On Verification', Russell sets out this psychological
account as involving a statement, an expectation, a fact, and the
resulting feeling or emotion:
Verification of a statement about the future, according to what
we have said, occurs in two stages. First, the statement (if
believed) produces a certain bodily and mental condition which I
call expectation; then a certain occurrence or set of occurrences
transforms the expectation into surprise or into the feeling
expressed by 'ah yes!' In the latter case the statement is verified,
in the former it is refuted.69

The verification, falsification, confirmation, etc. of a belief thus

involves a causal account in which, first, a statement or assertion
causes an expectation, which is conceived to be a psychological
'set', and, second, facts cause certain emotions in the subject. This

68 The Analysis of Mind, 269; An Outline of Philosophy, 207; An

Inquiry Concerning Meaning and Truth, 80.
69 'On Verification', 18.

140

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

account, Russell says, is 'emotional rather than intellectual'.

Something like this can be found running throughout several of his
works.70
Even when it came to the notion of truth Russell did not shy
away from proposing what may be called a causal theory of truth. A
belief or cognition is analyzed as an image plus feeling with the
image component constituting the semiotic or representational part
of belief. An image represents the original situation (or impression)
causing it, which is the prototype of the corresponding image.
Hence, one can say that a fact F is the prototype of the belief image
B just in case there is a resemblance (of structure) between F and B,
F causes B, and F and B have common effects. In such a case,
Russell says, we can say B is true. 'A sentence of the form 'this is A'
is called 'true', Russell says, 'when it is caused by what "A"
means',71 or in less sociological terms, ' "this is red" is "true" if it is
caused by something red'.72
Such a causal theory of truth is present in several of Russell's
works.73 For example in An Outline of Philosophy, Russell says:
a form of words is true if a person who knows the language is led
to that form of words when he finds himself in an environment
which contains features that are the meanings of those words,
and these features produced reactions in him sufficiently strong
for him to use words which mean them ... The environment
causes words, and words directly caused by the environ
they are statements) are 'true'.74
It thus appears that Russell found no incompatibility in b
in both such a causal theory of truth and, at least in som
later writings, in a correspondence theory of truth no doubt
because he believed that, in some way, the correspondence relation
could be explicated in naturalistic (causal) terms. How this was
possible he never says and, indeed, the problems surrounding a
causal explication of correspondence seem formidable.75

70 The Analysis of Mind, 269; An Outline of Philosophy, 207; An

Inquiry concerning Meaning and Truth, 80; Human Knowledge, 142.
71 Human Knowledge, 134
72 Ibid., 141.
73 An Inquiry concerning Meaning and Truth, 244; Human Knowledge,
133-4.
74 210.
75 An account of this would necessitate a discussion of Russell's
theory of causality and his theory of 'structuralism', both of which are
beyond the bounds of the present essay.

141

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard F. Kitchener

It is to causality, therefore, than Russell looked to provide

naturalistic account of knowledge; indeed, as we have seen, h
it to explain or analyze servo-mechanistic knowledge and ani
knowledge. Russell had initially thought that behaviorism co
account for such animal knowledge but he soon gave up this
assumption largely because Watson had denied the existence of
mental images and Russell crucially depended upon them in his
later theory of belief. A purely behavioristic theory of knowledge
he thus declared to be ultimately incomplete, requiring supplemen-
tation by an account of our inner life, our introspective awareness.
It would thus appear that Russell believed an externalist account of
knowledge would have to be supplemented with an internalist
account. A naturalistic account of such an externalist epistemologi-
cal perspective, although encountering well-known philosophical
objections, would seem to be a viable option. But can one provide a
naturalistic account of an internalist epistemology? Russell did not
address this address, an answer to which is difficult to ascertain
largely because of the unclear status of such an internalist
perspective at least within Russell's program. Nevertheless, there is
no evidence that Russell believed such an internalist account was
incompatible with naturalism (even if it might be incompat
with materialism). In fact, there is considerable evidence of an
indirect support indicating that that he did believe that introsp
tion (first-person awareness of things like images) was someth
that could be handled in a purely naturalistic way. For if one
adopts, as Russell did, a version of central state theory, images are
to be equated with neural states and hence located in the brain.
Furthermore, there does not seem to be anything in contemporary
cognitive psychology that would argue against such a naturalizing
of introspection. So, even if behaviorism was not a fully complete
account of knowledge, this did not mean that Russell gave up (or
had to give up) a naturalistic, psychological account of know-
ledge.76
I have tried to make a case that Russell had a inchoate version of
NE. True, it was sketchy, fragmentary, and undeveloped, but it wa

76 I have given reasons (above) for thinking that Russell gav

philosophy no special kind of knowledge or wisdom over and above that
science, nor did he believe there was any special kind of epistemologic
method of a radically different kind from that of the scientific metho
This raises the question of whether Russell believed in the analytic-
synthetic, the a priori-a posteriori, the necessary-contingent, and the
normative-empirical distinctions, and if he did, what kinds of accounts he
gave of them. Unfortunately, I have no space to discuss these important

142

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

nevertheless, present alongside his other, more 'Cartesian' account

of epistemology with Russell's emphasis switching from one to the
other. Russell seems to have thought that any adequate theory of
knowledge had to take into account both types of knowledge. He
also seems to have thought that knowledge might be completely
accounted for in naturalistic terms.
One way of viewing the relation between these two types of
knowledge (and the corresponding two types of epistemology) is to
imagine that what we can call characteristically human knowledge
occurs at the highest epistemic level but is based upon lower levels
of animal knowledge, and animal knowledge, in turn, having its
roots in what Sosa calls servo-mechanism knowledge. As we have seen
Russell allowed for these three different kinds of knowledge and so
does Sosa.
Such a three-tiered account matches (roughly) what we kn
about the evolution of knowledge in the species and the
development of knowledge in the person. Most epistemologists
have been concerned with characteristically human knowledge-the
type of knowledge possessed by mature, normal, language-using
adults who can introspect and who have certain reflective abilities.
Adult humans can introspect and reflect upon their epistemic
states, subjecting the grounds and reasons for their beliefs to critical
scrutiny. Such an ability can be called a meta-epistemic or
meta-cognitive ability.77 This is the first person, internalist
perspective of Descartes, normally distinguished from the third
person, externalist perspective of naturalistic epistemology. But, as
I have already suggested in our earlier discussion of Russell, why
should we view internalism as a version of non-naturalism? At least
four reasons might be given: (1) introspection is a non-naturalist
procedure, (2) indubitability is a non-naturalistic state of certainty
(3) internalism is wedded to a rationalistic account of a priori
knowledge, epistemology is an a priori endeavour, and rationalism
is non-naturalistic, and (4) the normativity of epistemology is
incompatible with naturalism. I believe all four of these claims are
mistaken and based upon conceptual misunderstandings. If so,
then there is no reason for supposing that even the traditional
conception of epistemology-what I have called the Cartesian

questions and, as far as I can tell, there are no systematic discussions of

these issues. So, I must leave them for another occasion (or for others to
discuss).
77 K. Lehrer, Metamind.(Oxford: Clarendon Press, 1990).

143

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard F. Kitchener

conception-should be construed in a non-naturalistic way. R

may have done this but there was no reason for him to do
both conceptions of epistemology-Cartesian internalism and
behavioristic externalism-can be interpreted in a naturalistic way.
If we adopt something like this three-level approach to
knowledge that Russell and others such as Sosa suggest, knowledge
can be said to develop from servo-mechanism knowledge to animal
knowledge to human knowledge in a stage-like manner. Cognitive
developmental psychologists and developmental epistemologists
have studied and analyzed this developmental process (although
they have used somewhat different terms). But it does seem to be
clear, for example, that meta-cognition, this meta-epistemic faculty
underlying human knowledge, develops from childhood clear into
adulthood. But it is also clear that it is based upon earlier forms of
epistemic cognition and that these forms may be like animal
knowledge.
If there is such a developmental process, presumably it would
require a developmental explanation.78 But such a developmental
explanation is (arguably) a naturalistic one-an explanation
involving natural developmental processes-and, at the same time,
a normative one. The normative dimension is one of progress and
improvement in knowledge by virtue of the fact that the
subsequent stage of knowledge is epistemically better than the
preceding one. But, at the same time, any epistemic norms involved
must be supervenient on naturalistic facts. Russell was notoriously
sceptical about the normative dimension of life but, if I am right,
he should not have been. Russell's NE can, suitably expanded by
current scientific knowledge, be seen to be a plausible research
program, one to be analyzed, subjected to scrutiny, and then
suitably modified. This is precisely how Russell saw the task of the
epistemologist-to propose tentative hypotheses and then to
critically evaluate them in keeping with the best scientific evidence
available. This is what Russell himself did and it is something that
other epistemologists should be doing. This would be perfectly in
keeping with the kind of NE Russell was advocating.

78 R. E Kitchener, 'Developmental explanations', Review of Meta-

physics, 36, (1983).

144

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
 Bertrand Russell's Naturalistic Epistemology

V. Revisiting Analytic Philosophy

If I am correct that, at least during one long period of Russell's

thought, he advocated what later thinkers would call a NE (and
naturalistic philosophy of science), then we must re-evaluate our
understanding and interpretation of the history of 20th century
analytic philosophy. For, according to the standard view, the
revolution produced by 'the analytic turn' in philosophy was
committed to a conception of philosophy radically different from
those prevalent in the 19th century. This new philosophy was
different from its predecessors in securing a place for an
autonomous philosophy, one radically different from those activi-
ties pursued in the empirical sciences. Although there were
different ways individuals characterized the distinctiveness of
philosophy, e.g., as taking 'the linguistic turn', such an analytic
philosophy was conceived to be radically different from the
methods and results of the natural sciences. In short, philosophy
could not be naturalized (contrary to what earlier, 19th century
philosophers had suggested).
The key individuals here were (initially) Frege, Moore, Russell,
and Wittgenstein. Russell's contribution to this analytic revolution
(and hence to the non-naturalizing of philosophy) has been the
focus of many works devoted to the philosophy of Russell. But if I
am correct, there is another side of Russell that has been
systematically ignored or slighted-the naturalistic side of Russell.
If this is correct, then the question that arises is how one should
conceive the relation between analytic philosophy and naturalistic
philosophy. Most individuals have felt that analysis as a method of
philosophy was some kind of non-naturalistic method, but this
picture needs to be challenged, I think, if indeed the method of
analysis plays the role I suggest it does in Russell's philosophy. Of
course, many of Russell's remarks about analysis and the method o
analysis support the more traditional interpretation, but many
other remarks of his support a quite different interpretation. Wha
this means, I think, is that the notion of 'analysis' needs to be
re-examined in the light of the naturalistic revolution of
philosophy.
With the hindsight of such a revolution, ushered in by Quine's
initial foray, a re-examination of the history of recent philosophy
seems to be called for. This has already occurred with respect to
some members of the Vienna Circle and logical positivism, e.g.,
Neurath. Other individuals have also suggested a naturalistic
interpretation of some of the work of Wittgenstein. If these
145

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms
Richard E Kitchener

individuals, then who next? G. E. Moore? Rudolph Carnap? As

strange as it may sound, this is precisely what I am suggesting
needs to be done. My hunch is that there are prominent naturalistic
strains in Moore and Moore's theory of analysis. Likewise, a
substantial part of the work of Rudolph Carnap lends itself also to
a naturalistic interpretation-something individuals are now
beginning to suggest.
All of this raises the question of the nature of philosophy, of
philosophical method, of demarcating philosophy from science. Is
there a way of conceptualizing how philosophy is different from
science? Is there just a difference of degree or of kind? Is
philosophy destined to wither away on the vine, to be replaced by a
future science? Needless to say, most philosophers would dig in
their heels when it comes to defending the autonomy of philosophy.
But, as I have suggested, such individuals would not derive much
sustenance from one of their heroes-Russell. He may have been
an analytic philosopher but he was also a naturalistic philosopher,
and naturalistic philosophies such as Russell's do not provide the
kind of support needed to undergird philosophy as the queen of the
sciences.

Colorado State Unive

146

This content downloaded from

143.106.203.164 on Sun, 14 Apr 2024 04:43:01 +00:00
All use subject to https://about.jstor.org/terms