Robin Durie - The Mathematical Basis of Bergson' Philosophy
Robin Durie - The Mathematical Basis of Bergson' Philosophy
Robin Durie - The Mathematical Basis of Bergson' Philosophy
Journal of the
British Society for
Phenomenology
Publication details, including
instructions for authors and
subscription information:
http://www.tandfonline.com/loi/
rbsp20
The Mathematical
Basis of Bergson's
Philosophy
a
Robin Durie
a
University of Exeter
Published online: 21 Oct 2014.
Journal of the British Society for Phenomenology, Vol. 35, No. I, January 2004
The original impetus for the divergence in the trajectories of analytic and
continental philosophy during the 20th century has, in recent years, become
a subject of scrutiny for both traditions. Most writers are agreed that the two
traditions share a more or less common point of departure, represented by
the shared concerns of Husserl and Frege at the start of the century. After
this, it is argued that Husserl's transcendental tum in Ideas (1913) marks the
moment at which the continental tradition begins to veer away from the
analytic tradition. Subsequently, a fundamental commitment to interiority,
and an affirmation of the irreducibility of subjectivity, are seen to be
hallmarks of the continental tradition. In contradistinction, the analytic
tradition has striven for objectivity. The methodological model to which it
has aspired in seeking to attain this objective has been scientific, and
specifically, mathematical, reflected in the central role accorded to logic by
the analytic tradition. From this perspective, it could therefore be argued
that the two traditions have been developing in opposition a twofold legacy
of Cartesianism - on the one hand, the analytic tradition seeking objectivity
by means of a contemporary reinterpretation of the more geometrico
delineated in the 'Reply to the Second Set of Objections', while, on the
other hand, the continental tradition re-enacts the subjective turn of
Descartes' Meditations.
Arguments of this nature highlight the fact that much of the suspicion of,
and indeed antipathy towards, the style of continental philosophy stem from
a conviction that continental philosophy fails to engage with the
mathematical or logical grounds of an adequate philosophical methodology.
It is just such a conviction that underpins the single text which did as much
as any other to provoke analytic philosophy's disenchantment with the
practices of its continental European sibling, namely, Russell's address to
The Heretics at Cambridge, on the evening of 11th March, 1912. This lecture
was subsequently published simultaneously by Open Court as 'The
Philosophy of Bergson' in volume 22 of The Monist (July, 1912), and as a
separate pamphlet bearing the same title. A revised version of the paper was
then included in Russell's History of Western Philosophy (1945). 1
54
But what of actual collections, such as the collections of the months of the
year, the signs of the zodiac, and the apostles, to all of which the number 12
can be applied - do these provide sufficient grounds for supporting
'Bergson's contention that every plurality of separate units involves space?' 7
Russell thinks not, and refers to some of Bergson's own examples, such as
that of hearing a clock strike 12. Far from depending upon our 'rang[ing] the
strokes of a clock in an imaginary space,' Russell claims, most people 'count
them without any spatial auxiliary.' This being the case, it demonstrates that
spatialisation, and hence separateness, is not necessary for conceiving
collections. Russell therefore concludes that Bergson's recourse to number,
and hence mathematics, is wholly spurious, commenting that 'as regards
mathematics, he has deliberately preferred traditional errors in interpretation
to the more modem views which have prevailed among mathematicians for
the last half century.' 8
There is, then, from Russell's point of view, a way of doing philosophy
which trades in problems that emerge as a consequence of philosophers'
adherence to assumptions which themselves follow as a consequence of
these philosophers' failure to understand and adopt the findings of
contemporary mathematics. This very failure, we could then argue, is
perpetuated by the tradition of continental philosophy, whereas the analytic
tradition begins to describe its own distinctive trajectory from the moment
that it overcomes this failure.
Whether or not we might wish to give credence to such an account, it
nevertheless behoves us to reflect on the legitimacy of Russell's portrayal of
Bergson. What I now aim to show is that, in fact, Bergson's position is far
more nuanced than Russell allows, and indeed, that it is so precisely because
of his keen awareness of certain theoretical developments in the
'mathematics of the last half century.' In fact, this should not come as a
surprise, since, in 1877, Bergson won first prize in mathematics for the
Concours General, and, with the publication the following year of his 'plane
solution of Pascal' in Nouvelles Annates de Mathematique, it was assumed
that his academic future lay in geometry - indeed, it was expected that he
would enter the Ecole Normale to study mathematics. 9
The first clue we come across suggesting that all is not quite as Russell
makes out is to be found in the long footnote that Bergson appends to the
chapter heading which is itself immediately followed in the body of the text
by the definition of number cited by Russell. In this footnote, Bergson
admonishes a certain F. Pillon for failing to 'distinguish between time as
quality and time as quantity, between the multiplicity of juxtaposition and
that of interpenetration [penetration mutuelle].' Bergson then writes that 'it
is the chief aim of the present chapter to establish ... this vital distinction.'
He goes on to conclude the footnote by observing that 'the verb "to
56
58
60
Why then does Bergson seek to argue in this way, and why should we
continue to insist on the connection to Riemann? In order to answer these
questions, it is necessary to ascertain in what the essence of quantity
consists. What is at stake in the more or the less with regard to quantity is to
be discerned in that fact that, in order to speak in this way of the more or
less, the initially given quantity must be able to increase or diminish. But if a
quantity x increases, and we subsequently wish to compare the initial
quantity x with the final quantity, to determine which is the greater, then we
must be able to divide, or separate out, the initial quantity x from the final
quantity. [TFW 3/(E 6] Furthermore, if the comparison is to be worthwhile,
then during this process of division, the quantity x must not change. This
seems to me to be the true import of Bergson's argument with regard to
separation - namely, that it is not spatial separateness that is significant,
which is what Russell wishes to claim, but rather, the capacity for
separability, that is definitive of quantity. From this perspective, therefore,
we can begin to argue that, contrary to Russell's claim, Bergson is not
advancing a thesis about number as such, but rather about the conditions of
possibility for counting (where 'countability' is in tum to be understood as
condition for either measuring or assimilation to a magnitude). Thus, the
condition of possibility for counting is that the elements of a multiplicity be
'denumerable' - and, correlatively, if the elements are non-denumerable,
then they cannot be counted. In turn, the condition of denumerability is
precisely that the elements of the multiplicity be distinct, or discontinuous.
Now, the distinctness of elements in a multiplicity does not, as such, require
that they be laid out in space, as Russell would have Bergson argue. Rather,
distinctness, or discontinuity, constitutes the condition for separability. As a
consequence, Bergson should be understood as arguing that the property of
distinctness which characterises the elements of a discrete multiplicity is also
a property that characterises space.
Thus, Bergson will argue that intensities are not quantifiable precisely
because they do not possess this capacity for divisibility or separability.
Now, if our preceding argument is correct, then this way of interpreting
Bergson has a more fundamental significance. For it is tempting to read
Bergson as propounding a series of more or less metaphysical dualisms such as duration/space, quality/quantity, intensity/extensity,
continuous/discrete - between which he then seeks, as we have seen, to
establish an equivalence. The criticism which is then levelled against
Bergson is that none of these concepts is rigorously defined; rather,
Bergson's style of writing consists in a tendency to conflate concepts at
crucial junctures, explaining one in terms of the other. The problem would
then be that at no point does Bergson determine the justification for
establishing the equivalence between the concepts, with the consequence that
62
64
number is a collection of units; we must add that these units are identical
with one another. .. [or if] they are all different from one another ... we agree
in that case to neglect their individual differences and to take into account
only what they have in common.' [TFW 76/CE 52] This shared identity of the
units within a collection corresponds to the common property that is shared
by members belonging to a set. What, then, is of fundamental importance is
that, in a process of division, made possible by the discontinuity between the
elements, the identity of the elements so divided is maintained. The elements
do not suffer any change in themselves as they undergo the process of
division. This is precisely because the process of division consists in nothing
other than a separation of parts from one another, a separability the condition
of which is, as we have seen, founded in the fact that the multiplicity is
discontinuous.
On the basis of this argument, we can begin to determine in what nondivisibility will consist. Let us imagine, for the sake of argument, dividing a
unit which is conceived as such ultimately rather than provisionally. Since,
from this perspective, it does not consist in separable parts, what would be
the effect of such a division? It would, of necessity, change the nature of the
unit essentially. It would create something other than the original unit,
something that would, as a consequence, be unable to continue fulfilling its
role as a unit with regard to the constitution of the number. This is precisely
what Bergson has in mind, I propose, when he talks in terms of a 'change in
nature,' or a 'difference in kind.'
The principle, therefore, that underpins the various dualisms between
which Bergson seeks to establish an equivalence, and which, furthermore,
determines the precise nature of his recourse to the Riemannian theory of
multiplicities, consists precisely in the difference between the two senses of
'to distinguish' to which Bergson, as we noted above, draws attention in the
crucial footnote which supports the opening to Chapter 2 of Time and Free
Will. [TFW 75-6/CE 51-2] The defining characteristic pertaining to the
Riemannian discrete multiplicity, the elements of which are discontinuous, is
that distinguishing the elements is a process that leaves these elements
unchanged. The defining characteristic pertaining to the Riemannian
continuous multiplicity, the elements of which are continuous, is that
distinguishing the elements is a process that changes the elements. Thus,
more properly called, this latter sense of distinction consists not in
quantitative division, but rather in 'qualitative differentiation.' [emphasis
added; TFW 95/CE 64] The difference between the two Riemannian
multiplicities as they are taken up by, and function for, Bergson, therefore,
consists in the difference between the nature of difference which determines
each multiplicity, division on the one hand, differentiation on the other. As
Bergson writes: 'the multiplicity of conscious states, regarded in its original
65
purity, is not at all like the discrete multiplicity which goes to form number.
In such a case there is, as we said, a qualitative multiplicity. In short, we
must admit two kinds of multiplicity, two possible senses of the word
'distinguish,' two conceptions, the one qualitative and the other quantitative,
of the difference between same and other.' [TFW 121/CE 80-1] And, because
division leads to no change in the elements of the multiplicity so divided,
Bergson characterises Riemann's discrete multiplicity as the homogeneous
multiplicity. On the other hand, since differentiation leads to a change in the
multiplicity so differentiated, Bergson characterises the Riemannian
continuous multiplicity as heterogeneous multiplicity. 2 Furthermore, each
multiplicity determines the nature of the reality to which they pertain - thus,
for Bergson, the nature of space is homogeneous, whereas the nature of
duration is heterogeneous: 'What we must say is that we have to do with two
different kinds of reality, the one heterogeneous, that of sensible qualities,
the other homogeneous, namely space.' [TFW 97 ICE 66]
Conclusion
The preceding argument reveals the extent to which Russell's criticisms
of Bergson are unfounded. Equally, however, it also reveals the extent to
which the Russellian diagnosis, that a healthy philosophy consists in
renouncing metaphysics in favour of mathematics, is misguided. What is
most striking about Bergson's recourse to Riemann is the way in which
Riemann's reflections on the fundamental axioms of geometry are given an
ontological significance in Bergsonism. It is from such a perspective that one
should understand Bergson's objective, for instance in his work on Einstein's
theory of relativity, of providing a metaphysics to complement science. At
the same time, it is just such a perspective that reveals the poverty of
Russell's formulation of the relation between mathematics and philosophy as
one of mutual exclusivity.
University of Exeter
References
1. Our proposal regarding the mathematical basis of the divergence between analytic and
continental philosophy is inadvertently borne out by Bergson himself when, in an
interview with Jacques Chevalier, he attributes to Russell the following observation: 'For
Bergson, evolution has culminated on the one hand in intelligence, which reaches its
most complete development in the mathematician and, on the other hand, in instinct,
which reaches its peak in bees, ants, and Bergson.' Quoted in The Collected Papers of
Bertrand Russell. Volume 6: Logical and Philosophical Papers 1909-13, edited by John
G. Slater (London: Routledge, 1992), 319. All further citations of Russell will be from
this volume.
2. Russell (1992), 322.
3. Russell (1992), 323; referring to Bergson, Creative Evolution, translated by Arthur
Mitchell (New York: University Press of America, 1984), 210-3; CEuvres, edited by Andre
Robinet (Paris: Presses Universitaires de France, 1959), 674-6.
66
4. Russell (1992), 328; referring to Bergson, Time and Free Will, translated by F.L. Pogson
(London: Macmillan, 1910), 78-9; (Euvres, 53-4. Hereafter cited in the body of the text as
TFW/(E, followed by the relevant page numbers.
5. Russell (1992), 328.
6. Russell (1992), 329.
7. Russell (1992), 329.
8. Russell (1992), 330-1. Following this passage, Russell proceeds to attack Bergson's
refutation of Zeno' s paradoxes, paradoxes which purport to demonstrate that change is
impossible.
9. Cf. the 'Chronology of Life and Works' in Bergson, Key Writings, edited by Keith Ansell
Pearson & John Mullarkey (London: Continuum, 2002), viii.
10. For the record, Russell's citation of this definition necessarily overlooks this identity,
following, as it does, Pogson's authorised translation of 'l'un et du multiple' as 'the one
and the many.' Russell (1992), 328.
II. Gilles Deleuze, Bergsonism, translated by Hugh Tomlinson & Barbara Habberjam (New
York: Zone Books, 1991), 39; Le bergsonisme (Paris: Presses Universitaires de France,
1966), 31.
12. Deleuze, Bergsonism, 'Afterword' to the English translation, 117.
13. All citations from: Bernhard Riemann, 'On the Hypotheses which lie at the Foundations of
Geometry', translated by W.K. Clifford, Nature, vol. 8, no. 183 (1873), 14; Gesammelte
mathematische Werke und wissenschaftlicher Nachlass (Leipzig: Teubner, 1876), 254.
14. Riemann (1873), 14/(1876) 254.
15. Riemann (1873), 14-15/(1876) 254-5.
16. Riemann (1873), 17/{1876) 257. The consequence of this argument, returning to the issue
of the 'simplest matters of fact from which the measure-relations of space may be
determined,' matters of fact would form the hypotheses for the geometrical system of
Euclid, is that 'the reality which underlies space [das Raume zu Grunde liegende
Wirkliche] must form a discrete multiplicity, or we must seek the ground [Grund] of its
metric relations outside it, in binding forces [bindene Kraften] which act upon it.'
17. Russell (1992), 328.
18. This condition goes to the heart of Cantor's use of set theory in his remarkable work in the
second half of the 19th century- in particular, given the significance of the role played by
infinite sets in Cantor's work the question which arises is the extent to which an infinite set
can indeed be 'grasped as a whole.'
19. This indivisibility arises as a consequence of their being conceived as 'ultimate' units,
whereas the divisibility would arise as a consequence of their being conceived as
'provisional' units. [TFW 81/(E 55]
20. Thus Bergson writes: 'In a word, pure duration might well be nothing but a succession of
qualitative changes, which melt into and permeate one another, without precise outlines,
without any tendency to extemalise themselves in relation to one another, without any
affiliation with number: it would be pure heterogeneity.' [TFW 104/(E 70]
67