Pergamon Stud. Hist. Phil. Sci., Vol. 29, No. 2, pp.

239–271, 1998
 1998 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
0039–3681/98 $19.00+0.00

On the Sociology of Scientific Knowledge

and its Philosophical Agenda

Michael Friedman*

For a philosopher seriously interested in the history and historiography of science,

perhaps the most interesting—if also the most troubling—theoretical development
of the last twenty years has been the sociology of scientific knowledge (SSK)
articulated at Edinburgh by Barry Barnes and David Bloor and at Bath by H. M.
Collins.1 On the one hand, this program has inspired innovative work in the history
of science that has applied methods of social history to key events in the evolution
of modern science so as to yield extraordinarily rich delineations of the wider
social, cultural, and political context of these events.2 On the other hand, however,
SSK, both theoretical and applied, has been framed by an explicitly philosophical
agenda—an agenda that aims to reject the traditional philosophical ideal of univer-
sal standards of rationality, objectivity, and truth (which ideal has, of course, been
traditionally taken to be paradigmatically exemplified in modern science itself) in
favor of a relativistic conception of scientific rationality, objectivity, and truth that
grounds these concepts, in the end, in local and particular social and cultural cir-

* Department of History and Philosophy of Science, Indiana University, Bloomington, IN 47405,

U.S.A.
Received 3 March 1997; in revised form 17 May 1997
1
B. Barnes, Scientific Knowledge and Sociological Theory (London: Routledge, 1974), T. S. Kuhn
and Social Science (London: Macmillan, 1982); D. Bloor, Knowledge and Social Imagery (London:
Routledge, 1976; second edition, Chicago: University of Chicago Press, 1991), Wittgenstein: A Social
Theory of Knowledge (New York: Columbia University Press, 1983); B. Barnes and D. Bloor, ‘Relativ-
ism, Rationalism and the Sociology of Knowledge’, in M. Hollis and S. Lukes, eds, Rationality and
Relativism (Oxford: Blackwell, 1982); H. M. Collins, Changing Order (London: Sage, 1985; second
edition, Chicago: University of Chicago Press, 1992).
2
Paradigmatic examples of this kind of social history are S. Shapin and S. Schaffer, Leviathan and
the Air-Pump (Princeton: Princeton University Press, 1985), and S. Shapin, A Social History of Truth
(Chicago: University of Chicago Press, 1994). Shapin provides a valuable survey (as of 1982) of histori-
cal work in this tradition in ‘History of Science and its Sociological Reconstructions’, History of Science
20 (1982), 157–211.

PII: S0039-3681(97)00021-6
239
240 Studies in History and Philosophy of Science

cumstances. According to what I will call the philosophical agenda of SSK, that
is, all there ultimately is to the notions of rationality, objectivity, and truth are
local socio-cultural norms conventionally adopted and enforced by particular socio-
cultural groups.
Barnes puts the idea this way:
Science is not a set of universal standards, sustaining true descriptions and valid
inferences in different specific cultural contexts; authority and control in science do
not operate simply to guarantee an unimpeded interaction between ‘reason’ and
experience. Scientific standards themselves are part of a specific form of culture;
authority and control are essential to maintain a sense of the reasonableness of that
specific form. Thus... science should be amenable to sociological study in fundamen-
tally the same way as any other form of knowledge or culture.3
In a well-known article explicitly devoted to these philosophical questions, Barnes
and Bloor explain that
[the relativist] accepts that none of the justifications of his preferences can be formu-
lated in absolute or context-independent terms. In the last analysis, he acknowledges
that his justifications will stop at some principle or alleged matter of fact that has
only local credibility$ For the relativist there is no sense attached to the idea that
some standards or beliefs are really rational as distinct from merely locally accepted
as such. Because he thinks that there are no context-free or super-cultural norms of
rationality he does not see rationally and irrationally held beliefs as making up two
distinct and qualitatively different classes of things.4
Collins, for his part, dubs his point of view ‘the Empirical Programme of Relativ-
ism’ and devotes the initial chapter of his major work to considering themes from
‘philosophical skepticism’ which point in an epistemologically relativistic direc-
tion.5 And it is not too much to say, I believe, that this relativistic and anti-tra-
ditional philosophical agenda is responsible for a large part of the intellectual
excitement that has surrounded SSK in both its theoretical and applied versions.
Moreover, this philosophical agenda of SSK, in both its theoretical and applied
versions, is explicitly traced to the work of one of the giants of twentieth century
philosophy, namely, Ludwig Wittgenstein. In particular, the concepts of ‘language-
game’ and ‘form of life’, which are central to Wittgenstein’s Philosophical Investi-
gations, are here interpreted as referring to particular socio-linguistic activities and
practices associated with particular socio-cultural groups—where the practices in
question are regulated by socio-cultural norms conventionally adopted by the rel-
evant groups. Wittgenstein’s insistence on the need for renouncing traditional philo-
sophy in favor of the careful description of particular ‘language-games’ expressing
particular ‘forms of life’ is then read as the call for an empirical sociological inves-

3
T. S. Kuhn, op. cit., note 1, p. 10.
4
‘Relativism’, op. cit., note 1, pp. 27–28.
5
Changing Order, op. cit., note 1, Chapter 1.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 241

tigation of the way in which the traditional categories of knowledge, objectivity,

and truth are socially constituted and determined by the norms, needs, and interests
of particular socio-cultural groups. This idea is of course most explicit in the work
of Bloor, who devotes an entire book to just such a sociological (and relativistic)
reading of Wittgenstein. And Bloor’s conclusion is nothing less than that the
sociology of knowledge should be the replacement for traditional philosophy.6 For
Bloor, the traditional problems of philosophy—which concern, in particular, ‘the
nature of rationality, objectivity, logical necessity and truth’7—can after all be
solved, and solved by empirical sociological means.8 That traditional philosophers
will find this ambition especially troubling goes without saying.
This paper is an exploration of the philosophical agenda of SSK. I will argue
that there are deep and significant tensions between the idea that SSK is an empiri-
cal scientific discipline, on the one hand, and its claim to solve the traditional
problems of philosophy (and, indeed, to replace the philosophical tradition), on the
other. I will approach these tensions by exploring the ways in which the aims,
methods, and arguments of SSK contrast with both those of the philosophical tra-
dition and those of Wittgenstein himself. My own aim, however, is not to indulge
in disciplinary boundary drawing. I wish rather to extend an invitation to philo-
sophically minded practitioners of SSK to engage more fully and systematically
with the philosophical context and tradition that frames their philosophical agenda.
For I believe, more specifically, that SSK has acted precipitously in actively seeking
to embrace and to advocate one particular side in a traditional philosophical debate.
Although there are indeed important reasons why disputes over philosophical rela-
tivism have arisen in precisely this intellectual context, I would urge philosophi-
cally minded historians of science to resist the temptation to become partisans in
these disputes and instead to treat both philosophy and science in an historically
and contextually sensitive manner. We should seek, in particular, to transcend the
starkly asymmetrical treatment of science and philosophy originating with Thomas
Kuhn’s ground breaking work on scientific revolutions.9 This work, as is well
known, uses sensitive attention to the history of science to argue for a new philo-
sophical picture of the nature of science. At the same time, however, the history
of philosophy receives no sensitive attention at all, but is rather portrayed stereotyp-
ically as a battle between a new, progressive view and ‘the’ tradition—which latter

6
Invoking Wittgenstein’s own characterization of his work as ‘one of the heirs to the subject which
used to be called “philosophy”’, Bloor continues (Wittgenstein, op. cit., note 1, p. 183): ‘My whole
thesis could be summed up as the claim to have revealed the true identity of these heirs: they belong
to the family of activities called the sociology of knowledge.’
7
Knowledge (second edition), op. cit., note 1, p. 83.
8
Collins, too, indicates that his program is to solve the epistemological problems traditionally raised
by philosophy: Changing Order, footnote 16 to Chapter 1.
9
T. Kuhn, The Structure of Scientific Revolutions (Chicago: University of Chicago Press, 1962).
242 Studies in History and Philosophy of Science

is treated as an undifferentiated and usually entirely unspecified monolith. Of

course it is perfectly normal for new intellectual movements to treat the history of
philosophy in this way. My point here is simply that it is now time for philosophi-
cally minded social historians of science to move beyond this stage.10

1. SSK and Traditional Philosophy

SSK is intended to be a naturalistic, empirical scientific enterprise. Indeed, it is
by emphasizing precisely this that we rebut the common idea that SSK is intended
as some kind of rejection or ‘denigration’ of science:
[T]he sociology of knowledge is emphatically not, as its critics often mistakenly
believe, itself a denigration of science; on the contrary, it is in many ways modelled
upon scientific investigation, and any claims to credibility it comes to have must be
closely related to those of science itself. The sociology of knowledge is a matter-of-
fact, empirical field of study which happens to include, among its subject-matter, the
knowledge and culture of science.11

Accordingly, the naturalistic and purely descriptive aims of SSK are explicitly
contrasted with the normative and prescriptive—‘moralizing’—aims of traditional
philosophy.12 And, in this sense, SSK eschews all temptation to evaluate histori-
cally given scientific practices from the point of view of a ‘universal reason’ and
confines itself rather to describing these practices as they actually exist:
Sociology is a subject with a naturalistic, rather than a prescriptive or normative
orientation; it simply tries to understand the convictions and the concepts of different
cultures as empirical phenomena. External evaluation of the convictions and concepts
is irrelevant to this naturalistic concern; all that matters is why they were actually sus-
tained.13
We thereby explain the widespread distaste for the relativistic conclusions of SSK
within the academic community:
A plausible hypothesis is that relativism is disliked because so many academics see
it as a damper on their moralizing. A dualist idiom, with its demarcations, contrasts,
rankings and evaluations is easily adapted to the tasks of political propaganda or self-
congratulatory polemic. This is the enterprise that relativists threaten, not science$

10
Historians of science in this tradition can perhaps take encouragement from the circumstance that
historians of philosophy are themselves paying increasing attention to social and other contextual con-
siderations. In addition to the works cited in note 69 below (which pay particular attention to the
¨
concomitant history of science), I would here cite K. Kohnke, Entstehung und Aufstieg des Neukantian-
ismus (Frankfurt: Suhrkamp, 1986); (partially) translated by R. Hollingdale as The Rise of Neo-Kantian-
ism (Cambridge: Cambridge University Press, 1991), and F. Beiser, The Fate of Reason: German Philo-
sophy from Kant to Fichte (Cambridge, MA: Harvard University Press, 1967).
11
Barnes, T. S. Kuhn, p. xi.
12
See Barnes, op. cit., pp. 58–63.
13
Op. cit., p. 5.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 243

If relativism has any appeal at all, it will be to those who wish to engage in that
eccentric activity called ‘disinterested research’.14
Paradoxically, then, it is the sober and empirically based relativism of SSK rather
than the traditional (philosophical) ‘Cult of Rationalism’ that best represents the
traditional values of disinterested and empirical scientific research.
Moreover, the practitioners of SSK are unanimous that sociological explanations
of ‘knowledge’ and ‘truth’ do not employ these terms as they are used within
traditional philosophy. Whereas in traditional philosophy knowledge is understood
to be justified true belief, so that, in particular, we have a sharp distinction between
what is taken to be and what actually is knowledge, for SSK ‘knowledge’ refers
rather to ‘any collectively accepted system of beliefs’15 or to ‘whatever people take
to be knowledge$ those beliefs which people confidently hold and live by’.16
Similarly, whereas nothing is more basic to traditional philosophy than the ‘distinc-
tion between what is “true” and what is merely taken to be so’, in SSK this distinc-
tion is definitively abandoned: truth is simply identified with ‘the body of locally
credible knowledge’.17 These terminological divergences clearly reflect the funda-
mental divergence in aims noted above. The traditional notions of ‘knowledge’ and
‘truth’ are normative rather than descriptive: ‘truth’ means what ought to be
believed, ‘knowledge’ means what is rationally or justifiably accepted. In rejecting
the normative aims of traditional philosophy in favor of an explicitly non-evaluative
and purely empirical point of view, SSK must also eschew the use of all traditional
normative notions. Rather than articulating the structure of what ought to be
believed, SSK simply describes and explains what is in fact believed.
In this way we straightforwardly obtain the celebrated ‘symmetry’, ‘impartiality’,
or ‘equivalence’ postulate:
Our equivalence postulate is that all beliefs are on a par with one another with respect
to the causes of their credibility. It is not that all beliefs are equally true or equally
false, but that regardless of truth and falsity the fact of their credibility is to be seen
as equally problematic. The position we shall defend is that the incidence of all beliefs
without exception calls for empirical investigation and must be accounted for by
finding the specific, local causes of credibility. This means that regardless of whether
the sociologist evaluates a belief as true or rational, or as false and irrational, he must
search for the causes of its credibility.... [The practitioners of SSK] simply investigate
the contingent determinants of belief and reasoning without regard to whether the
beliefs are true or the inferences rational. They exhibit the same degree and kind of
curiosity in both cases.18
As we have seen, SSK is an empirical and purely descriptive discipline, which,
for this very reason, must leave the traditional normative or evaluative uses of
‘knowledge’, ‘truth’, and ‘rationality’ wholly out of account. The aim is to explain,

14
Barnes and Bloor, ‘Relativism’, p. 47, footnote 44.
15
Op. cit., p. 22, note 5.
16
Bloor, Knowledge (second edition), p. 5.
17
See especially Shapin, Social History, op. cit., note 2, pp. 3–8.
18
Barnes and Bloor, ‘Relativism’, p. 23.
244 Studies in History and Philosophy of Science

not why beliefs are rationally or correctly accepted, but simply why beliefs are in
fact accepted. In thus explaining why beliefs are in fact accepted—how local con-
sensus is in fact achieved—we clearly have no option but, in all cases, regardless
of the normative status of the beliefs in question, to appeal to purely naturalistic
and entirely empirical contingent causal factors. And, since what is in fact
believed—what is in fact locally credible—obviously varies from context to context
and from group to group, a naturalistic and empirically based relativism is the
inevitable result.
At this point, however, one might very well wonder why SSK represents itself
as in conflict or competition with traditional philosophy. Why do we not simply
acknowledge the fundamental divergence in aims and methods and leave it at that?
Why, in particular, should the enterprise of empirically and naturalistically describ-
ing how beliefs become locally credible as a matter of fact compete or stand in
conflict with the enterprise of articulating the non-empirical and prescriptive struc-
ture in virtue of which beliefs ought to be accepted as a matter of norm? The
answer, of course, is that defenders of SSK represent themselves as explicitly
rejecting the aims and methods of traditional philosophy—not simply as leaving
them out of account. They feel compelled, that is, explicitly to deny the philosophi-
cal theses underlying the traditional normative enterprise: for example, ‘science is
not a set of universal standards, sustaining true descriptions and valid inferences
in different specific cultural contexts’; ‘there is no sense attached to the idea that
some standards or beliefs are really rational as distinct from merely locally accepted
as such’; ‘there are no context-free or super-cultural norms of rationality’; and so
on. Moreover, it is precisely by insisting on such negative philosophical con-
clusions that defenders of SSK adopt an explicitly philosophical agenda which
itself goes beyond the bounds of purely descriptive empirical research. Does the
practicing empirical ethnologist really need to be concerned whether there are—or
are not—‘super-cultural norms of rationality’ in the sense envisioned in traditional
philosophy? Is the assertion that there are no—absolutely no—such norms really
supposed to be an empirical generalization supported solely by currently available
empirical evidence?
Lying behind this rejection of traditional philosophical discussions of science is
the conviction that such accounts straightforwardly compete with the sociological
explanations offered by SSK. If there were ‘super-cultural’ norms of rational argu-
ment and evidence, so the argument goes, then scientific theories would be determ-
ined one way or another by reality, experience, and reason. There would be no
room left, as it were, for sociological, historically contingent explanations of the
content of scientific knowledge. Steven Shapin puts the idea this way:

If scientific representations were simply determined by the nature of reality, then no

sociological accounts of the production and evaluation of scientific knowledge could
be offered$ It would be pointless to argue against the kind of naive realism and
positivism which has few, if any, philosophic proponents at present. The underdeterm-
On the Sociology of Scientific Knowledge and its Philosophical Agenda 245

ination of scientific accounts of reality and the ‘theory-laden’ nature of fact-statements

are both quite widely accepted. Nevertheless, the way forward from these basic sensi-
bilities to a full-blown sociology of scientific knowledge is by no means generally
recognized. Even so, this is the best way to proceed: the sociology of knowledge is
built upon an appreciation of the contingent circumstances affecting the production
and evaluation of scientific accounts.19
It is striking, then, that Shapin’s survey of empirical work in the application of
SSK to the history of science begins with explicitly philosophical considerations—
just as it is similarly striking, as noted above, that Collins begins his major work
with a chapter on ‘philosophical skepticism’. In both cases it is thought that we
need such skeptical considerations precisely to undercut the competing explanations
of scientific knowledge developed by traditional philosophical ‘realism’ and
‘rationalism’. We need, in Bloor’s words, ‘to stop the intrusion of a non-naturalistic
notion of reason into the causal story’.20
But this line of thought rests on a misunderstanding. All that is necessary to
stop such an ‘intrusion’ of reason is mere abstinence from normative or prescriptive
considerations. We can simply describe the wealth of beliefs, arguments,
deliberations, and negotiations that are actually at work in scientific practice, as
Bloor says above, ‘without regard to whether the beliefs are true or the inferences
rational’. In this way, we can seek to explain why scientific beliefs are in fact
accepted without considering whether they are, at the same time, rationally or
justifiably accepted. And, in such a descriptive, purely naturalistic enterprise, there
is precisely enough room for sociological explanations of why certain scientific
beliefs are in fact accepted as the empirical material permits. Whether or not philos-
ophers succeed in fashioning a normative or prescriptive lens through which to
view these very same beliefs, arguments, deliberations, and so on, is entirely irrel-
evant to the prospects for empirical sociology. In this sense, there is simply no
possibility of conflict or competition between ‘non-naturalistic’, philosophical
investigations of reason, on the one hand, and descriptive, empirical sociology of
scientific knowledge, on the other.21

2. The Intellectual Context of Philosophical Relativism

We can deepen our appreciation of this last point, and, at the same time, begin
to understand why, nonetheless, there is indeed an intimate relation between recent

19
Shapin, ‘Sociological Reconstructions’, op. cit., note 2, p. 159.
20
Knowledge (second edition), p. 177.
21
Consider, for example, the philosophical thesis of the underdetermination of theory by evidence,
nowadays known as the Duhem-Quine thesis—discussed by Shapin in ‘Sociological Reconstructions’,
footnote 9, and often figuring centrally in discussions of SSK. This thesis challenges the view that
scientific theory is rationally or justifiably determined by evidence by invoking the mere logical possi-
bility of modifying background or auxiliary hypotheses in the case of a supposed negative crucial
experiment. The unavoidable historical contingencies governing how scientists in fact move from evi-
dence to theory in actual scientific practice are entirely independent of whether or not this philosophical
thesis is correct.
246 Studies in History and Philosophy of Science

interest in the sociology of scientific knowledge and philosophical relativism, by

briefly sketching a chapter in the history of scientific philosophy. This chapter
begins with the philosophy of Newtonian science developed in the late eighteenth
century by Immanuel Kant—who, we should add, was the first philosopher to make
‘scientific objectivity’ a fundamental philosophical theme. At the same time, as we
shall see, Kant was also the first important thinker of the modern period to insist
on a sharp separation between philosophy and empirical natural science.
Kant undertook the task of explaining how the best scientific knowledge of his
day—the mathematical physics of Newton’s Principia—is possible. And, on this
basis, he hoped thereby to reformulate the very meaning and function of philosophy
or ‘metaphysics’ as well. He hoped to answer the question ‘How is metaphysics
as a science possible?’ by answering the questions ‘How is pure mathematics poss-
ible?’ and ‘How is pure natural science possible?’22 In this connection, Kant inter-
prets the Newtonian achievement as indicating that the fundamental concepts of
space, time, and motion, along with those of material substance, force, and interac-
tion, have no application whatever to ‘things-in-themselves’ existing wholly inde-
pendently of the human mind. On the contrary, they are rather expressions of forms,
categories, and constructions of our own, on the basis of which alone nature—as
the object of truly objective and determinate scientific knowledge—is first possible.
Rather than describing an ultimate ‘noumenal’ reality of ‘things-in-themselves’,
metaphysics, in Kant’s hands, has now the task of articulating the conditions of
the possibility of experience, that is, the possibility of the truly objective and deter-
minate experience of nature exemplified, paradigmatically, in Newtonian natural
science.
The crucial question, however, is what do ‘objectivity’, ‘knowledge’, and ‘truth’
now mean here? For Kant, these concepts emphatically do not refer to a relation
of ‘mirroring’ or correspondence between the human mind on one side and a mind-
independent reality on the other. Kant’s insistence that we have no knowledge
whatsoever of ‘things-in-themselves’ is explicitly intended to block such a ‘realist’
interpretation once and for all. For Kant, ‘objectivity’ is now interpreted rather as
necessary intersubjective validity: as necessary validity for all times and for all
human knowers. The point is that pure mathematical knowledge is taken, above all,
to be exemplary of such necessary intersubjective validity, and what the Newtonian
achievements has shown, for Kant, is how such mathematical intersubjective val-

22
This strategy is most explicit in the Prolegomena to Any Future Metaphysics (1783). The metaphys-
ical tradition Kant is attempting to transform is that of Leibniz and Christian Wolff, wherein natural
science is assigned the role of describing spatio-temporal ‘phenomena’ while metaphysics describes the
underlying ‘noumenal’ reality of ultimate simple substances or monads. For discussion see my Kant
and the Exact Sciences (Cambridge, MA: Harvard University Press, 1992), and also G. Buchdahl,
Metaphysics and the Philosophy of Science (Oxford: Blackwell, 1969).
On the Sociology of Scientific Knowledge and its Philosophical Agenda 247

idity can be determinately injected or transplanted, as it were, into our experience

of sensibly given nature. In this way, empirical natural science becomes capable
of the same kind of intersubjective validity as is pure mathematics, and, for the
first time, it thereby becomes truly objective knowledge. And it is in this way, too,
that natural scientific knowledge—the kind of knowledge represented by modern
empirical natural science—becomes paradigmatic of the successful and objective
use of ‘universal’ human reason.
Now none of this is intended, by Kant, as a causal explanation of how natural
science actually evolves and develops as an historically given empirical phenom-
enon. On the contrary, Kant explicitly distinguishes between the ‘transcendental’
task of philosophy—the enterprise of articulating the necessary conditions of truly
objective scientific knowledge—and the ‘empirical’ task of the various scientific
disciplines. Philosophy, unlike the sciences themselves, is not concerned with
empirical objects and events occurring in space and time; it is not concerned with
what Kant calls ‘phenomena’ or ‘appearances’. The task of philosophy is rather to
articulate the necessary spatio-temporal background structure or framework within
which alone phenomena—as objects of empirical knowledge—are first possible.
In particular, the ‘transcendental’, philosophical enterprise has the task of
explaining the possibility (as opposed to the actuality) of scientific knowledge by
depicting the necessary framework of fundamentally mathematical presuppositions
within which all future empirical science must necessarily proceed. For all future
scientific knowledge, for Kant, must emerge as an extension and further develop-
ment of the Newtonian paradigm into more and more regions of empirically given
nature. This does not mean, however, that we know such an extension will in fact
occur. Rather, the idea of such continual progress is what Kant calls a regulative
ideal. It is something that we can neither achieve in fact (in real time, as it were)
nor confidently expect in the future. We can only hold it up before us as a goal
or norm towards which we strive to make our actual scientific knowledge converge.
And it is in this way that the Kantian conception of scientific objectivity is genu-
inely normative or prescriptive, without any hint at all of confusing the normative
with the descriptive.
Nevertheless, despite the purely normative character of this regulative ideal, the
actual further development of science created overwhelming problems for Kant’s
conception of scientific objectivity. For, as we have seen, the necessary framework
for all further scientific development, on Kant’s view, is modelled on the spatio-
temporal structure of Newtonian mechanics. Its mathematical expression, accord-
ingly, is through the laws of Euclidean geometry, on the one hand, and Newton’s
three laws of motion, on the other. Yet the great revolutions in late nineteenth and
early twentieth century mathematics and mathematical physics have shown that
precisely these laws must be given up if we are to progress beyond Newton. In
particular, the spatio-temporal framework of Einsteinian relativity theory employs
both a non-Euclidean geometry and a fundamentally non-Newtonian conception of
248 Studies in History and Philosophy of Science

how space and time are essentially interrelated through the concept of motion:
what we now call relativistic space–time has a fundamentally different underlying
mathematical structure from that of Newtonian space–time.23 And this means that
we now know that Kant’s conception of scientific objectivity is impossible in prin-
ciple, even conceived purely as a regulative ideal.
How, then, did scientifically minded philosophers respond to this new situation?
It is interesting to note that those philosophers who were most concerned to argue
that Kant’s particular conception of scientific objectivity is now untenable in the
light of Einstein’s work did not reject the Kantian project of articulating the a
priori or ‘transcendental’ presuppositions of scientific objectivity tout court. On
the contrary, these philosophers—who soon became known as the logical positivists
or logical empiricists—continued to hold that scientific objectivity is only possible
in virtue of an a priori mathematical framework that must first be injected or
transplanted into sensible nature before any properly empirical science of nature
is then possible. Kant’s mistake, however, was in thinking that he could specify
the precise content of such a mathematical framework entirely independent of, and
antecedent to, the actual future development of science—that he could specify the
particular structure of what we now call Newtonian space–time as eternally valid
once and for all. Relativity theory, in particular, has decisively refuted this idea;
for we now use, as we have seen, an entirely different mathematical structure as
our underlying framework for mathematical physics. Nevertheless, this new frame-
work, in the new context of Einsteinian relativisitic physics, has the same status,
relative to the new physical situation, that the old framework had in the context
of Newtonian physics. In both cases the underlying spatio-temporal structure rep-
resents a form or construction of our own that we must first inject into nature
before any properly empirical study of nature is then possible. Yet we now see
that such an underlying spatio-temporal mathematical framework is not, as Kant
thought, the expression of eternal and universally valid laws of thought. Following
the deep mathematical and philosophical study of non-Euclidean geometries under-
´
taken by Henri Poincare, we now see that such a priori constructions of our own
are better described as free conventions.24
The logical positivists then busied themselves in attempting to generalize and
apply these new ideas so as to develop a new philosophical conception of scientific
objectivity. Their first efforts involved a sharp distinction between the pure math-
ematics and logic implicated in the articulation of any mathematical framework

23
For an intellectual history of these developments see R. Torretti, Relativity and Geometry (Oxford:
Pergamon, 1983).
24
On these developments see my ‘Geometry, Convention, and the Relativized A Priori’, in W. Salmon
and G. Wolters, eds, Logic, Language, and the Structure of Scientific Theories (Pittsburgh: University
of Pittsburgh Press, 1994). For a general history of scientific philosophy from Kant through the logical
positivists see J. A. Coffa, The Semantic Tradition from Kant to Carnap (Cambridge: Cambridge Uni-
´
versity Press, 1991). For a history of geometry and its philosophy through Poincare’s ‘conventionalism’
´
see R. Torretti, Philosophy of Geometry from Riemann to Poincare (Dordrecht: Reidel, 1978).
On the Sociology of Scientific Knowledge and its Philosophical Agenda 249

for physics whatsoever and the particular applied mathematical representations of

nature employed by the various particular such frameworks. Following the first
published work by Ludwig Wittgenstein (who, at the time, was closely associated
with the logical positivists), his Tractatus Logico-Philosophicus (1921), the positiv-
ists argued that pure mathematics and logic (which, in accordance with the work of
Gottlob Frege, and with Bertrand Russell’s and Alfred North Whitehead’s Principia
Mathematica, were thought to be fundamentally identical) in fact possess the neces-
sary and absolutely universal validity Kant had attributed to all scientific knowl-
edge. For pure mathematics and logic express the absolutely necessary structure
of any system of linguistic representation whatsoever. Yet the further development
of mathematical logic during the 1920s suggested that this idea, too, must be called
into question. For it now appeared that a variety of different logico-mathematical
calculi, some essentially diverging from the ‘classical’ mathematical logic of Prin-
cipia Mathematica, were equally possible and legitimate. In particular, no over-
arching notion of universal logical ‘correctness’ appeared capable of ruling out the
new ‘intuitionistic’ mathematical logic developed by L. E. J. Brouwer and Arend
Heyting on the basis of a rejection of the ‘classical’ logical law of excluded middle.
In this respect, the situation in mathematical logic appeared precisely analogous
to that in geometry: in neither case did the idea of a single, uniquely ‘correct’
system appear to make sense.25
So, in 1934, Rudolf Carnap, the leading representative of logical positivism, bit
the bullet and took the decisive step of asserting that there is no over-arching notion
of universal validity or ‘correctness’ independent of the particular, and diverse,
rules of the particular, equally possible and legitimate, formally specifiable calculi.
For this reason, the very notions of ‘rationality’, ‘objectivity’, and ‘correctness’
must be relativized to the choice of one or another formal language or ‘linguistic
framework’. The particular logical rules of a given linguistic framework define
what counts as ‘correct’ and therefore rational within that given framework, and
it follows that no over-arching notion of rationality or ‘correctness’ can govern the
´
choice between different such frameworks. This latter choice, as Poincare had first
emphasized in the particular case of Euclidean and non-Euclidean geometries, can
only be a free stipulation or convention governed by pragmatic—as opposed to
rational—criteria. The result is Carnap’s celebrated ‘Principle of Tolerance’:
In logic there is no morality. Anyone may construct his logic—i.e., his form of langu-
age—as he wishes. But, if he wishes to discuss [matters] with us, he must clearly
specify how he wishes to construct it and give syntactical determinations instead of
philosophical considerations.26

25
All of these developments (including, in particular, the relationship between Wittgenstein and the
logical positivists) are discussed in detail in Coffa, op. cit., note 24.
26
R. Carnap, Logische Syntax der Sprache (Wien: Springer, 1934), §17; translated by A. Smeaton
as The Logical Syntax of Language (London: Kegan Paul, 1937).
250 Studies in History and Philosophy of Science

In Carnap’s later work the same idea is formulated as a distinction between internal
and external questions. Internal questions arise within the context of a particular,
already chosen and agreed upon linguistic framework. As such, these questions
are rationally and objectively answerable relative to the given logical rules of the
framework in question—rules which alone can give content to the notions of
‘rationality’, ‘validity’, and ‘truth’ in the first place. External questions, by contrast,
concern the choice between different such linguistic frameworks. These questions
cannot be rationally or objectively answerable, because the over-arching rules that
could define and characterize these notions are, in this case, necessarily missing.
External questions are therefore entirely a matter of conventional choice based, in
the end, on purely pragmatic criteria of suitability or adaptedness for one or another
given purpose.27
At this point, one might very well experience a strong sense of de´ja` vu. For this
Carnapian distinction between internal and external questions is closely analogous
to the central Kuhnian distinction between normal and revolutionary science. Thus,
the ‘puzzle-solving’ activities of normal science proceed against the background
of a generally accepted and agreed upon paradigm that defines, relatively unprob-
lematically, what could count as either a ‘correct’ or an ‘incorrect’ solution. In
revolutionary situations, by contrast, the very background framework which alone
can define such ‘correctness’ is itself at issue. And this is why revolutionary
science, for Kuhn, poses a particularly acute challenge to what he takes to be the
traditional philosophical conception of the rationality of science. For it is precisely
in this case—where we are faced with the choice between competing scientific
paradigms—that over-arching standards of rationality and validity are entirely miss-
ing. Hence, the traditional ideal of scientific rationality must here evidently give
way to non-rational factors in explaining the emerging new consensus. It is doubly
ironic, then, that this Kuhnian collapse of the traditional notion of universal stan-
dards of rationality was already clearly prefigured in the development of the philo-
sophical tradition itself, as this tradition evolved from the ‘transcendental’ univer-
salism of Kant to the explicitly relativized notion of rationality characteristic of
Rudolf Carnap’s work.28
And it is at this point, too, that we can discern the true relationship between the
explicitly philosophical agenda of SSK, on the one hand, and the philosophical
tradition it is intended to replace, on the other. It is not that the philosophical
tradition sets up a competing model for causally explaining the actual historical

27
See R. Carnap, ‘Empiricism, Semantics, and Ontology’, Revue Internationale de Philosophie 4
(1950), 20–40; reprinted in his Meaning and Necessity (second edition; Chicago: University of Chicago
Press, 1956).
28
The Structure of Scientific Revolutions was first published in the International Encyclopedia of
Unified Science edited by Carnap and Charles Morris. Carnap wrote to Kuhn in his editorial capacity
and comments enthusiastically about Kuhn’s work in two letters of 12 April 1960 and 28 April 1962.
(These two letters are reproduced, and discussed, in G. Reisch, ‘Did Kuhn Kill Logical Empiricism?’,
Philosophy of Science 58 (1991), 264–277.) Kuhn, towards the end of his career, regretted the fact that
he had interpreted Carnap’s letters as expressions of ‘mere politeness’ and acknowledged the point that
his philosophical conception is akin to ‘Kant’s a priori when the latter is taken in [a] relativized sense’—
see T. Kuhn, ‘Afterwords’, in P. Horwich, ed., World Changes: Thomas Kuhn and the Nature of Science
(Cambridge, MA: MIT Press, 1993), pp. 313, 331.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 251

evolution of science. On the contrary, the philosophical tradition is concerned

throughout with purely normative ideals of scientific rationality that are intended
to capture the underlying structure of what is taken to be exemplary in contempor-
ary science, so as to hold this up as a model for objectivity and rationality as such.
Yet, as we have seen, as contemporary science develops and changes, philosophers
have been forced to change fundamentally the resulting ideals of scientific ration-
ality as well. And it is in this way that the philosophical tradition is in fact essen-
tially intertwined with the actual historical evolution of science. Finally, as we
have also seen, this same process of mutual scientific and philosophical evolution
has, in the end, resulted in a thoroughly relativized and conventionalistic conception
of scientific rationality—a conception that in turn prepares the way for the philo-
sophical agenda of SSK. All we need do is turn our attention—as Carnap himself
does not—to the social factors actually responsible for the consensual acceptance
of the conventions underlying one or another Carnapian linguistic framework or
Kuhnian paradigm.29 Carnap plus Kuhn equals the philosophical agenda of SSK.

3. SSK and Wittgenstein

We already know, however, that the practitioners of SSK do not as a matter of
fact appeal to Carnap in articulating and defending their philosophical agenda.
Their main philosophical hero is rather Ludwig Wittgenstein, and, in particular,
the Wittgenstein of the so-called later philosophy—represented, above all, by the
Philosophical Investigations (on which Wittgenstein worked intensively throughout
the 1930s and 40s and which was finally published posthumously in 1953). And
this appeal to the later philosophy of Wittgenstein is, at least at first sight, perfectly
intelligible, since this philosophy undoubtedly takes as its basis and starting point
the fundamentally social and practical character of all human thinking and knowing.
In sharp contrast to both his own earlier philosophy as expressed in the Tractatus
and Carnap’s Logical Syntax of Language—both of which discern the philosophi-
cally interesting ‘essence’ of language in its purely formal, purely syntactic proper-
ties—Wittgenstein here introduces the concept of ‘language-game’ precisely ‘to
emphasize that the speaking of language is part of an activity, or of a form of life’
(PI, §23).30 Language as such, and therefore all human thinking and knowing, only

29
A version of this understanding of the sociological relevance of Kuhn’s work is completely explicit
in Barnes, T. S. Kuhn. Carnap himself was by no means unsympathetic to investigation of such social
factors. On the contrary, in unpublished notes in connection with his correspondence with Kuhn (note
28 above) Carnap compares the choice of scientific framework to a Darwinian process of selection ‘on
the basis of preference in the community of scientists, whereby all kinds of sociological, cultural, etc.,
factors are involved’. (This note is reproduced in J. Earman, ‘Carnap, Kuhn, and the Philosophy of
Scientific Methodology’, in Horwich, ed., op. cit., note 28.)
30
I shall cite Wittgenstein’s works as follows: (T) Tractatus Logico-Philosophicus, trans. C. K. Ogden
(London: Routledge, 1922); (BB) The Blue and Brown Books (Oxford: Blackwell, 1958); (PI) Philo-
sophical Investigations, ed. G. E. M. Anscombe (Oxford: Blackwell, 1953); (RFM) Remarks on the
Foundations of Mathematics, ed. G. H. von Wright, R. Rhees, and G. E. M. Anscombe (Oxford:
Blackwell, 1956); (OC) On Certainty, ed. G. E. M. Anscombe and G. H. von Wright (Oxford:
Blackwell, 1969).
252 Studies in History and Philosophy of Science

makes sense in the context of essentially social and practical ‘streams of life’.
Indeed, Kuhn’s decisive emphasis on the importance of scientific communities is
itself explicitly modelled on these later Wittgensteinian ideas.31
On second sight, however, there are also very significant philosophical disparities
and divergences between Wittgenstein and the philosophical agenda of SSK. In
the first place, as we have seen, SSK is itself intended to be an empirical scientific
discipline. It is intended to describe and explain the naturally occurring phenom-
enon of human scientific knowledge in the same way, and by the same methods,
that science describes and explains any other natural phenomenon. But Wittgenstein
himself is adamant, both early and late, that what he calls philosophy is entirely
distinct from natural science. This theme begins in the Tractatus:
Philosophy is not one of the natural sciences.
(The word ‘philosophy’ must mean something that stands above or below the natural
sciences, not beside them.) (T, 4.111)
The theme is continued, even more insistently, in the ‘Blue Book’ (unpublished
notes dictated in 1933–34 which are considered to initiate the period of the Philo-
sophical Investigations):
Our craving for generality has another main source: our preoccupation with the
method of science. I mean the method of reducing the explanation of natural phenom-
ena to the smallest possible number of primitive natural laws; and, in mathematics,
of unifying the treatment of different topics by using a generalization. Philosophers
constantly see the method of science before their eyes, and are irresistibly tempted
to ask and answer questions in the way science does. This tendency is the real source
of metaphysics, and leads the philosopher into complete darkness. I want to say here
that it can never be our job to reduce anything to anything, or to explain anything.
Philosophy really is ‘purely descriptive’. (BB, p. 18)
And, in the Philosophical Investigations itself, when looking back critically at the
earlier project of the Tractatus, the very same point stands out perfectly explicitly:
It was true that our considerations could not be scientific ones$ And we may not
set up any kind of theory. There must not be anything hypothetical in our consider-
ations. We must do away with all explanation, and description alone must take its
place. And this description receives its light, i.e., its purpose, from the philosophical
problems. These are certainly not empirical problems. They are solved, rather, through
an insight into the working of our language—and in such a way, in fact, that this is
recognized: contrary to an urge to misunderstand it. The problems are solved, not by
bringing to bear new experience, but by arranging what has long been known. Philo-
sophy is the battle against the bewitchment of our understanding by means of our
language. (PI, §109)
Wittgenstein’s interest in carefully and attentively describing the workings of vari-

31
See Kuhn, Structure, op. cit., note 9, pp. 44–46. Kuhn’s assimilation of Wittgenstein was probably
mediated by his association with the philosopher Stanley Cavell, who is cited in the Preface.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 253

ous particular language-games is in no way preparatory for the kind of general

socio-cultural ‘systematic theory of language-games’ envisioned by Bloor.32
By the same token, Wittgenstein shows no interest in articulating generalizations
about human thought and language—either such straightforward empirical gen-
eralizations as might be found in actual ethnological practice or in the kind of bold
and exciting philosophical generalities about human knowledge considered earlier
(‘there are no context-free or super-cultural norms of rationality’, and so on). On
the contrary, as stressed above, philosophy stays always on the level of particulars
so as solely to reveal perspicuously what we all already know:
Philosophy merely presents everything, and neither explains nor infers anything.—
Since everything lies open, there is also nothing to explain. For what might be hidden
does not interest us.
One could also call ‘philosophy’ what is possible before all new discoveries and
inventions.

The work of philosophy is an assembling of reminders for a definite purpose.

If one wanted to set up theses in philosophy there could never be any discussion
about them, because everyone would agree with them. (PI, §§126–128)
Thus, whatever may be the ‘definite purpose’ of Wittgenstein’s ‘arranging’ (§109),
‘assembling’ (§127), and ‘ordering’ (§92) of various particular ‘language-games’
(§23) so as to produce a ‘surveyable presentation’ of ‘the use of our words’ (§122),
this purpose is certainly not to marshal evidential support, by instances, as it were,
for any kind of generalized picture of language—whether empirical or philosophi-
cal.
In the second place, the particular language-games that Wittgenstein actually
considers are, as often as not, imaginary uses of language rather than real ones.
From the famous example of the primitive builders that opens the Investigations
(§2), to the ‘pupil’ who ‘understands’ simple arithmetical examples differently than
we do (§§144 ff.), or the strange ‘tribe’ or ‘people [Volk]’ whose forms of life are
essentially different from our own (e.g., §§200, 206–207), Wittgenstein is con-
stantly presenting us with entirely imaginary examples which are ‘invented’ (§122)
as ‘objects of comparison that are supposed to cast light on the relations of our
language through similarities and dissimilarities’ (§130). And it is for precisely
this reason that, although Wittgenstein’s purely descriptive enterprise is certainly
intended to point to the very general facts of human (and non-human) nature in
which our language-games are necessarily embedded, this enterprise is neither natu-
ral science nor natural history:
If the formation of concepts can be explained on the basis of facts of nature, should
we not be interested, instead of in grammar, in that in nature which lies at its basis?—
We are certainly interested also in the correspondence of concepts with very general
facts of nature. (Those that mostly do not strike us because of their generality.) But
our interest does not extend back to the possible causes of the formation of concepts.

32
See Wittgenstein, Chapter 7.
254 Studies in History and Philosophy of Science

We are not engaged in natural science, and not even in natural history—since we can
also surely provide fictitious natural history for our purpose. (PI, II, xii)
The importance and prominence of entirely imaginary language-games in Wittgen-
stein’s work very clearly brings out its fundamentally non-empirical character.
A third, closely related point is that Wittgenstein shows very little interest in
the kind of historical and cross-cultural variation in human linguistic and cultural
practices that is the basis and starting point for the empirically oriented enterprise
of SSK. He shows very little interest, that is, in either the kind of socio-cognitive
changes studied in the history of science or in comparative cultural ethnology.
Wittgenstein is interested, to be sure, in alternatives to ‘our’ ordinary practices;
and he is interested, in particular, in showing that there is therefore no absolute
necessity somehow inherent in these practices.33 Yet the deviation from ‘our’ ordi-
nary practices manifest in such Wittgensteinian alternatives—which are typically,
as emphasized above, purely imaginary alternatives—is much more radical than in
the alternative socio-cultural practices studied in the history of science and eth-
nology. An imaginary ‘tribe’ of people who consistently ‘misunderstand’ the most
elementary possible arithmetical rules, for example, is clearly quite another kettle
of fish from the alternative socio-cognitive communities encountered in actual his-
tory and ethnology. Indeed, it is typical of Wittgenstein’s alternative ‘tribes’, unlike
the alien communities we study in actual history and ethnology, that, try as we
might, we literally cannot, in the end, understand them.34
Finally, there is no trace of socio-cultural relativism in Wittgenstein. Although
the ‘we’ that constitutes Wittgenstein’s central object of concern is normally left
entirely unspecified, it is also quite frequently used as synonymous with ‘humanity’
or the totality of ‘human beings’.35 The contingent ‘agreement in form of life’
(§241) lying at the basis of human thought and knowledge appears to have more
to do with ‘the common human way of acting’ (§206) than with the socio-cultural
conventions of one or another particular socio-cultural group.36 And it appears to
be for precisely this reason, in fact, that Wittgenstein focusses on purely imaginary
alternatives to ‘our’ customary forms of life—alternatives which are so extreme
that ‘we’ human beings cannot, in the end, understand them. Accordingly, when
Wittgenstein speaks of ‘language-games’ and ‘forms of life’ in the plural, he is
typically calling attention to the manifold uses of language and linguistic activities
found within ‘our’ single cultural-linguistic community. Indeed, Wittgenstein is
doing precisely this in the very passage where he first introduces these expressions:

33
See the continuation of PI, II, xii: ‘If someone believes that certain concepts are absolutely the
correct ones, that whoever might have different ones would simply not comprehend what we do, then
let him imagine certain very general natural facts to be different from what we are used to, and the
formation of concepts different from the usual ones will become understandable to him.’
34
See, e.g., PI, §207, and II, xi, p. 223. This point is emphasized in an insightful discussion by Barry
Stroud, ‘Wittgenstein on Logical Necessity’, reprinted in G. Pitcher, ed., Wittgenstein (New York:
Doubleday, 1966).
35
See, e.g., PI, §§204, 206, 241, 242.
36
Compare OC, §156: ‘In order to make a mistake, a human being must already judge in conformity
with humanity.’
On the Sociology of Scientific Knowledge and its Philosophical Agenda 255

But how many kinds of sentences are there? Perhaps assertion, question, and com-
mand?—There are innumerable such kinds: innumerable different kinds of uses of
everything that we call ‘signs’, ‘words’, ‘sentences’. And this multiplicity is nothing
fixed, given once and for all; rather, new types of language—new language-games,
as we might say—come into being and others become obsolete and are forgotten.
(The changes in mathematics can give us an approximate picture of this.)
The word ‘language-game’ is here intended to emphasize that the speaking of langu-
age is part of an activity, or of a form of life.
Review the multiplicity of language-games in these examples and others:
Giving orders and acting in accordance with them—
Describing an object by its appearance, or on the basis of measurements—
Constructing an object in accordance with a description (a drawing)—
Reporting an event—
Speculating about the event—
Setting up and testing an hypothesis—
Presenting the results of an experiment by means of tables and diagrams—
Inventing a story; and reading it—
Play-acting—
Singing rounds—
Guessing riddles—
Making a joke; telling it—
Solving an applied calculation—
Translating from one language into another—
Requesting, thanking, cursing, greeting, praying—
—It is interesting to compare the multiplicity of linguistic tools and their uses, the
multiplicity of kinds of words and sentences, with what logicians have had to say
about the construction of language. (Including the author of the Tractatus Logico-
Philosophicus.) (PI, §23)
Here Wittgenstein is in no way considering conflicting or competing socio-cogni-
tive communities as appear in the history of science or alternative socio-cultural
communities as studied in comparative ethnology. Wittgensteinian ‘forms of life’
are, in this respect, something entirely distinct from the alternative ‘forms of life’
considered in SSK.
I do not mean to suggest, however, that practitioners and defenders of SSK are
simply ignorant of these fundamental divergences between their agenda, on the
one hand, and Wittgenstein’s actual philosophical conception, on the other. On the
contrary, Bloor himself forthrightly calls attention to the sharp contrast between
Wittgenstein’s explicitly anti-scientific and anti-theoretical characterization of his
enterprise and Bloor’s own frankly empirical and theoretical ambitions at the very
beginning of his discussion. Yet, in his eagerness to get on with the business of
developing an empirically based ‘systematic theory of language-games’, Bloor
merely brushes this divergence aside: what he wants to do is ‘replace a fictitious
natural history by a real natural history, and an imaginary ethnography by a real
ethnography’.37 My point, in emphasizing the significance of these divergences

37
Wittgenstein, p. 5. In Chapter 8 Bloor seeks to account for Wittgenstein’s ‘unfortunate’ anti-scien-
tific and anti-theoretical attitude as a product of the ‘pessimistic’ Lebensphilosophie popular in the
1920s. I here attempt to show, on the contrary, that Wittgenstein’s anti-scientific and anti-theoretical
256 Studies in History and Philosophy of Science

nonetheless, is not at all to dispute the possibility or importance of the kind of

empirically based ethnology of science advocated and articulated in SSK. Nor does
it follow, of course, if practitioners of SSK in fact diverge significantly from Wittg-
enstein in their philosophizing, that Wittgenstein is ‘right’ and they are ‘wrong’.
Indeed, it is not at all my purpose here to argue that practitioners of SSK are
(philosophically) ‘wrong’. I wish rather to argue that, if one insists on embedding
the empirical enterprise of SSK within an explicitly philosophical agenda, then one
might very well profit from attending more carefully and systematically to the
particular philosophical conception of that historically given philosopher one takes
as one’s philosophical model.38

4. ‘The Hardness of the Logical Must’

In attempting to understand the philosophical conception underlying Wittgen-
stein’s later philosophy, it is very helpful to begin with his earlier conception
expressed in the Tractatus Logico-Philosophicus. Indeed, Wittgenstein himself says
as much in the Preface he prepared for the Philosophical Investigations: ‘[My new
thoughts] can be seen in the proper light only in contrast with and against the
background of my old way of thinking’ (PI, p. x). And the Tractatus, as indicated
above, attempts to present the philosophical foundations of the new mathematical
logic recently articulated by Gottlob Frege and Bertrand Russell—which logic,
according to the Tractatus, depicts the universal and absolutely necessary frame-
work within which alone any linguistic representation whatsoever is possible.39 It
is in virtue of this logical framework, and this framework alone, that it is possible
at all to represent the world:
Logical propositions describe the framework of the world, or rather they present it.
They do not ‘treat’ of anything. They presuppose that names have meaning and
elementary propositions have sense; and this is their connection with the world. It is
clear that it must indicate something about the world that certain connections of sym-
bols—which essentially have a determinate character—are tautologies. Herein lies the
decisive point. We said that some things are optional in the symbols we use and some
things are not. In logic only the latter is expressed. This means, however, that in logic
we do not express what we wish with the help of signs; rather, in logic the nature of
the absolutely necessary sign itself gives testimony: if we know the logical syntax of
any sign language whatsoever, then all the propositions of logic are already given.
(T, 6.124)

attitude towards philosophy, in particular, is an integral part of his philosophical conception that cannot
be so easily brushed aside.
38
In §5 below I attempt to illustrate how the recent theoretical debates surrounding SSK have become
bogged down in tortuous metaphysics precisely by paying insufficient attention to Wittgenstein’s own
philosophical conception.
39
As Peter Winch has emphasized to me, in particular, this is not to say that Wittgenstein does not
have serious disagreements with Frege and Russell about the nature and form of this new logic. On
the contrary, he emphatically rejects their attempts to incorporate something like what we now call set-
theory into the foundations of mathematics and logic: see especially T, 6.031.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 257

And it is in this way, too, that logic obtains a uniquely privileged status: ‘Logic
is not a doctrine, but a mirror-image of the world; logic is transcendental.’(6.13)
Thus, Wittgenstein’s conception of logic in the Tractatus is a clear descendant of
the Kantian conception of universal reason—a reason that yields the absolutely
necessary presuppositions or conditions of possibility for any thinking about
reality whatsoever.
There is one important respect, however, in which the conception of the Trac-
tatus dramatically diverges from Kant. Central to the Tractatus is a sharp distinc-
tion between what can be said [gesagt] on the one side and what cannot be said
but rather can only be shown [gezeigt] on the other. And this means, in particular,
that the philosophical conception of logic just reviewed cannot, strictly speaking,
itself be formulated or articulated. This philosophical conception can itself only be
shown—it is shown in and through the logical relationships among propositions:
Propositions cannot present logical form; this mirrors itself in them.
That which mirrors itself in language, language cannot present.
That which expresses itself in language, we cannot express by means of language.
Propositions show the logical form of reality.
They exhibit it. (4.121)

What can be shown cannot be said. (4.1212)

This is why, at the end of the Tractatus, Wittgenstein paradoxically characterizes
his own philosophical reflections as ‘non-sensical’, as a ‘ladder’ that one must
‘throw away after one has climbed it’ (6.54). Unlike Kant, then, Wittgenstein can
find no place at all for a ‘transcendental’ philosophical theory of universal reason.
The reality, as it were, of this universal reason can only be shown—shown in the
properly understood logical workings of our language.
In this respect, moreover, there is also a sharp divergence between Wittgenstein’s
conception of logic, in the Tractatus, and the related view later articulated by
Carnap—which view, as Carnap himself insists, is explicitly inspired by the Trac-
tatus. Both Wittgenstein and Carnap hold that logic is based solely on the purely
formal or syntactic combinatorial possibilities of symbolic language—and is there-
fore, in particular, entirely independent of the meanings of the signs in question
(compare T, 3.33). It is for this reason, for both Wittgenstein and Carnap, that
logic is thereby entirely independent, in addition, of all particular empirical facts
holding in the actual world. For Carnap, however, there is a theory or discipline
whose subject matter is the logical syntax of language, and this theory is precisely
the new discipline of metamathematics recently developed by David Hilbert. Here,
in the meta-language, we can indeed explicitly describe the formal logical syntax
of the object-language; and, on this basis, Carnap himself rejects Wittgenstein’s
conception of the inexpressibility of logical syntax.40 For Wittgenstein, by contrast,
there is no point of view outside of logic and language—outside, that is, of the
single, absolutely universal logical language—from which one could possibly

40
See, e.g., Logical Syntax, op. cit., note 26, §8. For discussion see my ‘Carnap and Wittgenstein’s
Tractatus’, in W. Tait, ed., Early Analytic Philosophy (La Salle, IL: Open Court, 1997).
258 Studies in History and Philosophy of Science

describe or articulate what he calls logical syntax (4.12). Hilbertian meta-languages,

from Wittgenstein’s own point of view, are therefore entirely irrelevant.41
There is a close connection, then, between Carnap’s rejection of Wittgenstein’s
conception of the inexpressibility of logical syntax, on the one hand, and Carnap’s
logico-linguistic pluralism, on the other. It is precisely because there are many,
logically distinct formal languages that Carnap has room for the distinction between
object-language and meta-language—and hence room for an explicit formal theory
of logical syntax. During the late 1920s and early 1930s, the period when Wittg-
enstein was most closely associated with Carnap and the other logical positivists,
Wittgenstein, too, came to reject his earlier commitment to the uniquely privileged
status of the ‘classical’ mathematical logic of Frege and Russell. Indeed, both
Wittgenstein and Carnap were greatly impressed by the development of
‘intuitionism’ by the mathematician L. E. J. Brouwer, which showed, in particular,
that alternatives to the ‘classical’ logical framework are in fact quite possible.
Moreover, during this same period Wittgenstein developed a conception of ‘gram-
matical rules’ that appeared to be quite similar in many ways to Carnap’s concep-
tion of the ‘logical rules’ definitive of a given linguistic framework.42 Nevertheless,
Wittgenstein did not, in the end, go the route of Carnap’s Logical Syntax of Langu-
age. In particular, he never adopted the conception of the arbitrariness of logical
grammar or logical syntax expressed in the Principle of Tolerance, and he never
followed Carnap in embracing the possibility of an explicit and articulated philo-
sophical theory of language. Wittgenstein instead pursued the entirely different
path that eventually issued in the Philosophical Investigations.
In the Philosophical Investigations, as we know, Wittgenstein abandons the logi-
cal absolutism of the Tractatus in favor of a kind of linguistic pluralism—a plural-
ism embracing an indefinite variety of language-games and forms of life. Yet this
is not, as in Carnap, a pluralism of alternative logico-mathematical frameworks,
each articulating its own particular version of logico-mathematical validity.
Wittgenstein’s linguistic pluralism rather involves the even more philosophically
radical idea that logico-mathematical thinking as such is in no way foundational
for the workings of language as a whole. In particular, one should no longer take
the fact-stating, assertive or descriptive use of language—the use for which the
concepts of ‘true’ and ‘false’ together with the logical principles that essentially
govern these concepts encapsulate the rules of the language-game—as the funda-
mental use of language to which all other uses must somehow be reduced. On the
contrary, language in the end rests simply on the indefinite multiplicity of human
forms of life, and the fact-stating, assertive or descriptive use of language is merely
one part of this immense multiplicity (PI, §23). Nevertheless, Wittgenstein does
not in any way lose interest in the particular case of logico-mathematical thinking.
On the contrary, this particular human linguistic phenomenon continues to play a

41
The importance of this rejection of the project of metamathematics for Wittgenstein’s later philo-
sophy as well is stressed in Paul Feyerbend, ‘Wittgenstein’s Philosophical Investigations’, reprinted in
G. Pitcher, ed., op. cit., note 34.
42
The overlap between Carnap’s views and Wittgenstein’s at this time was in fact so extensive that
Wittgenstein accused Carnap of plagiarism. See Coffa, op. cit., Chapters 13–14, for a discussion of this
episode by a Carnapian partisan.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 259

very central role indeed in his own philosophical reflections in the Philosophical
Investigations—and, even more, in the unpublished notes he originally intended to
integrate into the Investigations that are now collected together as the Remarks on
the Foundations of Mathematics. And Wittgenstein continues to be obsessed, just
as he was in the Tractatus, with the question of the ultimate source or ground of
logico-mathematical necessity: with the question of ‘[t]he hardness of the logical
must’ (RFM, I, §121).
In sharp contrast to the Tractatus, however, it is clear that logico-mathematical
necessity, in the period of the Investigations and the Remarks, now has no ultimate
source or ground—no philosophical foundation or explanation of the kind the Trac-
tatus had attempted to depict. The real problem, in philosophy, is now precisely to
resist the characteristically philosophical demand for such an ultimate justification:
The danger here, I believe, is to give a justification of our procedure, where there is
no justification and we should simply say: that’s how we do it. (RFM, II, §74)
[I]it is a peculiar procedure: that I go through the proof and then accept its result.—
I mean: that is simply how we do it. That is customary among us, or a fact of our
natural history. (I, §63)43
In particular, we should reject the attempt of the Tractatus to trace the source of
logical necessity to ‘the nature of the absolutely necessary sign’ which, entirely
independently of what we wish to do with it, ‘itself gives testimony’ (T, 6.124).
We should reject the whole idea of such an ‘absolutely necessary sign’—which is
made, as it were, of ‘the purest crystal’ (PI, §97) and which, ‘hidden in the medium
of the understanding’, (PI, §102) determines in advance how it is to be understood.
No sign—not even the very simplest combinatorial sign configuration used in
elementary logic and arithmetic—can have this latter property. For all signs, in the
end, only unfold their meanings in and through their concrete applications at the
hands of actual human beings. In the end, it must after all be we who determine
the meanings of logico-mathematical rules by the way in which we all agree in
how to go on in continually applying these rules.44
Does this mean that Wittgenstein now intends to advocate a naturalistic, empiri-
cally oriented theory of logico-mathematical necessity that locates the source of
this necessity in human agreement and convention—a sociological theory of logico-
mathematical necessity of the kind explicitly championed by David Bloor? Does
Wittgenstein, that is, intend a naturalistic and reductionist account of this phenom-
enon according to which all there is to logico-mathematical necessity is socio-
cultural convention? Nothing, I think, could be further from the truth.45 For, not

43
Compare PI, §217, in the context of the celebrated ‘rule-following’ argument.
44
Compare RFM, I, §118: ‘it is we who are inexorable in applying these laws’.
45
Shapin and Schaffer, as the motto for Chapter IV of Leviathan, op. cit., note 2 (p. 110), cite a
remark from RFM, I, §116, where Wittgenstein says: ‘Yet one can nevertheless say that the laws of
inference compel us; in the sense, namely, that other laws in human society do so.’ The implication,
it seems, is that here Wittgenstein is advocating a general theory of logico-mathematical necessity that
equates it with societal laws and conventions (an implication which is fully explicit in Bloor, Witt-
genstein, pp. 121–122). This kind of reading perfectly illustrates the dangers of overly generalizing
approaches to Wittgenstein. For the remark in question is tied to a very specific example—the example
of an official clerk (apparently a military clerk) who follows instructions to assign people in accordance
with height to various ‘divisions’ (I, §17). This clerk, in the words of I, §116, ‘would be punished if
260 Studies in History and Philosophy of Science

only does Wittgenstein explicitly reject all attempts to provide theoretical accounts
or explanations for the phenomena he considers—whether these explanations be
philosophical or empirical—but he also insists that his investigation is a purely
‘grammatical’ one (PI, §90), which, just as in the Tractatus, is concerned precisely
with the ‘essence’ of our language:
[I]f we, too, are striving in our investigations to understand the essence of language—
its function, its construction—it is not this that these questions [of the Tractatus] have
in view. For they see in the essence, not something that already lies open and becomes
surveyable [u¨bersichtlich] by ordering, but rather something that lies under the sur-
face—something that lies within, that we see if we penetrate [durchschauen] the thing,
and which an analysis is supposed to dig out. (PI, §92)

Since, in Wittgenstein’s new conception, ‘essence is expressed in grammar’ (§371)

and ‘grammar says what kind of object something is’ (§373), we see that the chief
difference, in this connection, between the new conception and the old can be
expressed as follows: whereas, on the old conception, the essential—and therefore
non-empirical—features of language are uncovered when we penetrate [durchs-
chauen] to the ‘crystalline purity’ of logical form underlying our ordinary uses of
language via analysis, these same features, on the new conception, are rather laid
¨
open to view when we survey [ubersehen] a number of rightly chosen particular
examples of such ordinary uses—whether real or imaginary—to reveal their ‘gram-
mar’ (compare §122). In this way: ‘The prejudice of crystal purity can only be
removed by rotating our entire examination. (One could say: the examination must
be rotated, but around the fixed axis of our real need.)’ (§108, and see the immedi-
ately following §109).
When we perform this ‘rotation’, moreover, we see that logico-mathematical
thinking has by no means lost its central, and essentially non-empirical position in
our thinking and acting in general. For, as a matter of fact, logico-mathematical
thinking, characterized precisely by the peculiar ‘hardness of the logical must’,
plays a central and very particular role—both theoretical and practical—in our
actual human life:
[The logico-mathematical must] is an expression of an attitude towards the technique
of calculating that shows itself everywhere in our life. The emphasis of the must
corresponds only to the inexorableness of this attitude—both towards the technique
of calculating and towards innumerable similar techniques.
The mathematical must is only another expression of the fact that mathematics
forms concepts.

he inferred differently’. But consider an only slightly more complicated example: a clerk who must
give correct change for purchases. Such a clerk is certainly legally liable for making false change—
but only if he is knowingly and intentionally cheating. In such a case the clerk is emphatically not
being punished for making mistakes in arithmetic—indeed, if this is the problem he is not legally liable.
Here the connection between ‘laws of inference’ and ‘laws in human society’ is already considerably
more complex and indirect than in the first example.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 261

And concepts serve for conceptualizing. They correspond to a definite way of dealing
with situations.
Mathematics forms a network of norms. (RFM, V, §46)46
Thus, although logico-mathematical thinking simply rests, in the end, on the very
general facts of human (and non-human) nature in virtue of which we all agree in
thus going on, in spite of this fact—or better, precisely because of it—such thinking
is in no way arbitrary or optional: ‘thinking and inferring (like counting) is of
course not bounded for us by an optional definition, but rather by the natural limits
corresponding to the body of that which we can call the role of thinking and
inferring in our life’ (RFM, I, §116). As a matter of grammar, not a matter of
experience, then: ‘The propositions of logic are “laws of thought”, because they
express the essence of human thinking—or, more correctly: because they express,
or show, the essence, the technique, of thinking. They show what thinking is, and
also types of thinking.’ (RFM, I, §133, and compare §131).
We have now reached the heart of the matter, I believe, for it is at precisely
this point that three fundamental themes of Wittgenstein’s later philosophizing
become clear. First, there continue to be non-empirical, essentially normative
elements in our thinking; and logic and mathematics continue to be paradigmatic
of such elements. Second, such normative features of linguistic practice are the
particular province of a peculiarly philosophical investigation—an investigation
that is itself non-empirical or ‘grammatical’.47 And third, this peculiarly philosophi-
cal investigation nonetheless issues in no special body of doctrine or theory. On
the contrary, it can only proceed by a piecemeal examination of various particular
language games that ‘shows’ or ‘expresses’ the grammatical features in question.
All of these themes—with the exception, of course, of the emphasis on piecemeal
examination of various particular language-games—are, as we have seen, also cen-
tral to the Tractatus. The fundamental difference, as we have also seen, is that the
Tractatus points to a philosophical explanation of the normativity of logic in terms
of the absolute necessity and ‘crystalline purity’ of underlying logical form,
whereas the Investigations points rather to the ineluctable connection of such norm-
ativity to the ultimately contingent and ‘mundane’ facts of practical social life.
Yet it is of crucial importance in assessing the philosophical agenda of SSK, I
believe, that the Investigations only gestures towards these ‘mundane’ facts of
practical social life. At precisely this point, Wittgenstein himself self-consciously
and deliberately steps back from embracing any kind of naturalistic theory claiming
to ‘explain’ the normativity in question. And he will have absolutely nothing to
do, in particular, with any kind of reductionist account according to which the

46
For the essentially non-empirical—‘grammatical’—character of ‘concept formation’ in general and
mathematics in particular see, e.g., RFM, I, §§128, 155; II, §§71–75; III, §29; V, §14–18. And compare
PI, II, xi (pp. 224–226), xii. In the passages from PI, II, xi, Wittgenstein emphasizes the constitutive
¨
importance of the (by him unexplained) fact that there is general or complete agreement [volle Uberein-
stimmung] in mathematics. In is in this context, especially, that one should understand the famous
remarks about the relationship between ‘human agreement’ and truth at PI, §§240–242.
47
The importance of Wittgenstein’s non-empirical notion of ‘grammar’ is emphasized by Stanley
Cavell, ‘The Availability of Wittgenstein’s Later Philosophy’, in G. Pitcher, ed., op. cit. In particular,
Cavell compares Wittgenstein’s distinction between ‘grammatical’ and ‘empirical’ propositions with
Kant’s ‘transcendental’/‘empirical’ distinction.
262 Studies in History and Philosophy of Science

normativity of logico-mathematical thinking is nothing but a matter of social con-

vention: as emphasized above, there is no trace of socio-cultural relativism in
Wittgenstein’s own philosophizing. In sharp contrast to all such tendencies, Wittg-
enstein himself is rather centrally concerned with what he terms the ‘dignity’ of
logic and mathematics:
What relation has [the mathematical proposition] to these empirical propositions
[describing how calculations in fact turn out]? The mathematical proposition has the
dignity of a rule.
This much is true in the thesis that mathematics is logic: it moves within the rules
of our language. And this gives it its special solidity, its separate and unassailable
position. (RFM, I, §164)
Moreover, Wittgenstein explicitly considers the relationship between the ‘dignity’
of logico-mathematical rules and a possible ethnological description of them:
But can one not then replace the rule by an empirical proposition that says that meas-
uring instruments are made in such and such a way, that people make use of them
in this way? One might give an ethnological portrayal of this human institution.
Now it is obvious that this ethnological portrayal could take over the function of
the rule.
$
Whoever knows a mathematical proposition is not yet supposed to know anything.
I.e., the mathematical proposition is only supposed to supply the framework for a
description. (RFM, V, §2)

A rule qua rule is detached, it stands there, so to speak, tyrannically—even though

what gives it importance are the facts of everyday experience.
What I have to do is something like this: describe the office of a king—whereby I
may not make the mistake of explaining the kingly dignity by the king’s usefulness;
and yet I may leave neither usefulness nor dignity out of account. (V, §3)
Even though the ethnological portrayal ‘could take over the function of the rule’,
it is clear that the point of these passages is precisely to distinguish the two. An
ethnological description functions as a rule, for example, when we correct the
linguistic usage of a foreigner by the phrase, ‘we don’t say that’, which here means
‘you shouldn’t say that’.48 Used in this way, however, the ethnological portrayal
is precisely not a naturalistic, purely empirical description of our practice; for the
latter would necessarily leave out the ‘dignity’ of the rules governing our practice—
their actual normative power.49
We can now appreciate the full force of Wittgenstein’s strategy of attempting
to describe ‘the essence of language—its function, its construction’ from within
the very norms and practices he is describing. We can now see, in particular, why
Wittgenstein attempts to ‘show’ or ‘exhibit’ the essential features of our socio-
linguistic practice by adducing particular examples of ordinary usage rather than

48
Compare RFM, II, §65, for the example of a judge attempting to use a ‘handbook of anthropology’
as a statute book.
49 ¨ ¨
The German word Wurde, unlike the English ‘dignity’, has the connotation of ‘worthiness [wurdig-
¨
keit]’. In this sense, a rule that has Wurde is ‘worthy’ of being followed.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 263

engaging in ethnological theorizing. For the point of view of the ethnological theor-
ist is explicitly outside the cultural system being considered. The ethnological the-
orist, unlike the members of the cultural system themselves, is in no way bound
by or committed to the norms of the system in question—for the ethnologist, these
norms certainly do not have ‘the dignity of a rule’. Indeed, precisely in so far as
the ethnologist explicitly conceives the cultural system in question as simply one
possible system among many diverging alternatives, the normativity of the system,
from this point of view, must inevitably be called into question: its natural and
‘internal’—taken-for-granted—character is thereby necessarily dissolved. But
Wittgenstein, as we have seen, does not conceive his own investigation as belong-
ing to comparative ethnology in this way. On the contrary, Wittgenstein takes
himself to be describing features of our socio-linguistic practice that lie so deep
and are so pervasive, as it were, that there is, as a matter of fact, no point of view
outside them—no such point of view that we human beings can, as a matter of
fact, understand. Wittgenstein then attempts to depict the true character of these
most fundamental features of our practice—features through which we see what
thinking and inferring amount to for us, for example—from within, and thus in a
way that embraces, rather than dissolves, their normativity.
It is this that gives Wittgenstein’s own investigation its deeply philosophical
quality and continues, despite all his philosophical radicalism, to bind him to the
philosophical tradition. For, in considering the most fundamental features of our
thinking—those that define what ‘objectivity’, ‘rationality’, and ‘truth’ amount to
for us—the philosophical tradition has attempted to attain a reflective position
wherein these notion lose their ‘internal’ or taken-for-granted character, but are
nevertheless, from this same reflective position, shown to be ultimately grounded
or justified. In this way, the philosophical tradition has attempted to take reflective
responsibility, as it were, for the normativity of our most fundamental cognitive
categories. Yet Wittgenstein has reached a point where the traditional philosophical
enterprise of depicting such an ultimate explanation or justification appears hope-
less. His project, accordingly, is to wean us once and for all from the need for
such a justification while, at the same time, showing us how we can continue to take
reflective responsibility for our most fundamental human practices nonetheless. We
can see that the normativity of these practices rests, in the end, on nothing more
nor less than certain very basic but ‘mundane’ facts of practical social life—yet
we can do this in such a way that our commitment to, and responsibility for, the
normativity of these practices is in no way endangered or diminished.50
It is of particular interest, then, that proponents of the philosophical agenda of
SSK also emphasize the new perspective on reflective responsibility made possible
by a sociological conception in which objectivity and rationality are seen to be our
own constructions. Bloor, at the beginning of his book on Wittgenstein, puts the
point this way:

50
The themes of commitment and responsibility are emphasized by Cavell, op. cit., in a closely
related context.
264 Studies in History and Philosophy of Science

If what [Wittgenstein] says is true, or anywhere near the truth, the great categories
of objectivity and rationality can never look the same again. Think how often our
polemical appeals to these two things depend on portraying them as forms of external
compulsion. A social theory of knowledge changes all this. Objectivity and rationality
must be things that we forge for ourselves as we construct a form of collective life$
The things we had seen ourselves as answerable to, we are now answerable for. So the
body of work that we are about to examine redraws the boundaries of responsibility; it
is a subtle attempt to alter our cultural self-consciousness.51
And the often cited penultimate sentence of Leviathan and the Air-Pump explains
that ‘[a]s we come to recognize the conventional and artifactual status of our forms
of knowing, we put ourselves in a position to realize that it is ourselves and not
reality that is responsible for what we know’.52 But to my ear these professions
of responsibility ring hollow. One precisely does not express commitment and take
responsibility by adopting the purely naturalistic and deliberately non-evaluative
point of view of the empirical ethnologist towards one’s own intellectual standards.
And still less does one do so if, in a philosophical vein, one then proceeds to make
sweeping negative claims of the form ‘science is not a set of universal standards’
and ‘there is no sense attached to the idea that some standards or beliefs are really
rational as distinct from merely locally accepted as such’. For the normativity of
one’s own standards is now explicitly reduced, from this point of view, to the
status of an otherwise arbitrary ‘preference’ for the practices of one’s own parti-
cular social group;53 and the actual normative force or binding power of these
standards is thereby inevitably dissolved. Wittgenstein’s intense philosophical
struggles with the ultimately social character of all human thinking and acting—
and with the question of the normativity of our standards in the light of this ulti-
mately social character—show us, in the end, that genuine reflective responsibility
cannot be so easily attained.54

5. Entanglements of ‘Reflexivity’
I have tried to make it clear why Wittgenstein himself does not follow the prac-
tice of SSK in linking his philosophical exploration of the ultimately social charac-
ter of all human thought to an empirical, naturalistic sociology of knowledge. An
empirical, purely naturalistic treatment of our scientific and other intellectual prac-

51
Wittgenstein, p. 3.
52
Shapin and Schaffer, Leviathan, p. 344.
53
See Barnes and Bloor, ‘Relativism’, pp. 26–27, 29–30. The idea is perhaps most explicit on p. 27,
where Barnes and Bloor consider two primitive ‘tribes’ with different beliefs and standards: ‘Faced
with a choice between the beliefs of his own tribe and those of the other, each individual would typically
prefer those of his own culture. He would have available to him a number of locally acceptable standards
to use in order to assess beliefs and justify his preferences. What a relativist says about himself is just
what he would say about the tribesman$ When confronted with an alien culture he, too, will probably
prefer his own familiar and accepted beliefs and his local culture will furnish norms and standards
which can be used to justify such preferences if it becomes necessary to do so.’ (There follows the
passage denying the existence of ‘super-cultural norms of rationality’ with which we began.)
54
We are here faced, of course, with the problem of ‘reflexivity’, which has been much discussed
in the more recent theoretical literature on SSK. In the following section I briefly consider this literature.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 265

tices occupies a detached or external standpoint relative to these practices. In parti-

cular, the norms and standards that govern these practices from the inside, as it
were, now lose their normal, taken-for-granted character. Indeed, when we now
ask ourselves, from the outside, why precisely these norms and standards should
govern our behavior, they entirely lose, from this point of view, their customary
normative force. Traditional philosophical efforts at depicting the ultimate sources
or grounds of our knowledge, by contrast, have attempted to occupy an anal-
ogous—but non-empirical—external standpoint from which one could underwrite
our customarily taken-for-granted standards by depicting a ‘transcendental’ ration-
ality behind our ordinary one. The unique contribution of Wittgenstein’s genius
was to create a new style and method of philosophizing, wherein we acknowledge
both the failure of such philosophical attempts and the ultimately social character
of all human thought, while continuing to maintain—and, indeed, to take reflective
responsibility for—the normativity of our most fundamental norms and standards
without which we could not think at all. And it is therefore absolutely essential,
as I hope to have made clear above, that what Wittgenstein has created is a new
style of philosophy held self-consciously and deliberately distinct from all empirical
social science.
It is striking, then, that extended theoretical discussions revolving around pre-
cisely these issues have emerged in the recent literature on sociological approaches
to science inspired by SSK. What we find, in fact, is the emergence of approaches
to the social study of science that have been driven by precisely the tensions
between philosophy and social science we have been considering to articulate new
methods explicitly intended to overcome these tensions. The discussions in question
revolve around the problem of ‘reflexivity’—the problem, that is, of applying skept-
ical and relativistic conclusions about science in general to the particular case of
the social study of scientific knowledge itself. And they have led to new attempts
to interpret and apply the thought of Wittgenstein, on the one hand, and to efforts
at treating social reality entirely on a par (‘symmetrically’) with natural reality, on
the other. In both cases, perhaps not too surprisingly, the newer approaches have
faced vigorous opposition from the ‘classical’ SSK first articulated at Edinburgh
and Bath.
The most detailed discussion of the proper interpretation of Wittgenstein, in this
context, occurs in a debate between Michael Lynch, who opposes the theoretical
and ‘analytical’ sociology of knowledge developed by the Edinburgh school on
behalf of the more particularistic and descriptive methods of ethnomethodology,
and David Bloor.55 Unlike Bloor, Lynch takes Wittgenstein’s anti-theoretical,

55
For this debate see Lynch, ‘Extending Wittgenstein: the Pivotal Move from Epistemology to the
Sociology of Science’, Bloor’s reply, ‘Left and Right Wittgensteinians’, and Lynch’s rejoinder, ‘From
the “Will to Theory” to the Discursive Collage’, all in A. Pickering, ed., Science as Practice and Culture
(Chicago: University of Chicago Press, 1992). I am indebted to both Thomas Gieryn and Steven Shapin
for urging me to consider this encounter.
266 Studies in History and Philosophy of Science

purely descriptive ambitions very seriously indeed. And, again unlike Bloor, Lynch
pays careful attention to the anti-skeptical dimension of Wittgenstein’s treatment
of rule following—the extent to which agreed upon intellectual rules, like those of
logic and mathematics, can and must have genuine normative force from within
the standpoint of the practice in which they are embedded. Accordingly, Lynch
reads Wittgenstein as explicitly warning us against the kind of explanatory social
science recommended by Bloor, in which we try to reduce intellectual norms to
social conventions:
As I understand Wittgenstein’s discussion of rules, he specifically warns against such
an explanation, but not because of any anticausal irrationalism. He insists that there
is no better site for the explication of a rule’s sense, relevance, and agreed-to use
than the ensemble of expressions and techniques that make up the practice in which
it is embedded. He does not offer an alternative form of causal explanation, since a
practice is not a center of agency or a causal factor. Rather than trying to explain a
practice in terms of underlying dispositions, abstract norms, or interests, a task for
sociology would be to describe the ensemble of actions that constitute the practice.
This is precisely what ethnomethodology seeks to do.56
For Lynch, then, intellectual practices have their own ‘internal’ authority and integ-
rity, which is not to be reduced to or explained in terms of ‘external’ social factors
such as conventions, institutions, or interests.
There is no doubt, I believe, that Lynch’s Wittgenstein is both closer to the
actual Wittgenstein and more sophisticated philosophically than Bloor’s. Neverthe-
less, Lynch, too, recommends that we diverge from Wittgenstein’s actual philo-
sophical practice by ‘replac[ing] a fictitious natural history by a real natural history,
and an imaginary ethnography by a real ethnography’. For ethnomethodology,
according to Lynch, is ‘an empirical extension of Wittgenstein’ in which we pro-
duce faithful empirical case studies of actual instances of scientific practice—for
example, by analyzing tape-recorded or videotaped behaviors and interactions of
scientists in the laboratory.57 Yet, at the same time, Lynch takes such an ethnome-
thodological perspective to be incompatible with philosophical ‘realism’ and ‘Pla-
tonism’ (which, I imagine, is the same as what others call ‘rationalism’), and he
considers this new enterprise to be an extension and development of ‘epistemology’
which represents, presumably, a distinctive philosophical stance.58 Here Lynch con-
tinues to miss the central Wittgensteinian distinction between ‘empirical’ and
‘grammatical’ investigations—which Wittgenstein himself marks, as stressed
above, precisely by appealing to imaginary rather than to real examples. Lynch
thereby misses the Wittgensteinian connection, also emphasized above, between
‘grammatical’ explorations of the ‘essential’ features of our language and the exhi-
bition, from within our language, of the normativity of such features. Although
Lynch’s ethnomethodology indeed eschews sociological theorizing, it continues to

56
Lynch, ‘From the “Will to Theory”’, in Pickering, ed., op. cit., note 55, p. 290.
57
See Lynch, ‘Extending Wittgenstein’, op. cit., pp. 256–259.
58
Ibid., pp. 247, 251, 257–259.
On the Sociology of Scientific Knowledge and its Philosophical Agenda 267

be an empirical, purely naturalistic enterprise. As such, it can neither display the

normativity of our practices from within nor engage the philosophical tradition.59
An even clearer illustration of the tensions between philosophical and empirical
investigation here, however, is provided by a second recent debate responding more
directly to the problem of ‘reflexivity’. This debate involves Bruno Latour and his
associates, on the one side, and H. M. Collins and his associates, on the other.60
Latour began the debate by complaining that the idea of a ‘social construction’ of
scientific facts is groundlessly asymmetrical in its treatment of the facts of natural
science which are the primary objects of SSK and the facts invoked by SSK itself
(facts about social conventions, institutions, and interests) in order to explain the
‘construction’ of the former. For, if SSK is itself a scientific enterprise, then the
facts to which it appeals should have precisely the same status as all other scientific
facts. In this sense, we should no longer speak of an asymmetrical construction of
natural reality out of social reality, but we should rather seek for an entirely sym-
metrical construction of both natural reality and social reality simultaneously.61
Collins, for his part, objects that a fully symmetrical treatment of natural and social
reality simply repeats the ‘realist’ and ‘rationalist’ errors of traditional philosophi-
cal and historical approaches to science, and, more generally, that it gives too much
weight to the natural sciences in relation to the social sciences.62
How, then, does Collins deal with the problem of ‘reflexivity’? How does he
rebut the charge that SSK is groundlessly asymmetrical in refusing to apply the
skeptical and relativist philosophy it brings to the productions of the scientists
under study also to its own productions qua social science? Here Collins invokes
the concept of ‘meta-alternation’, inspired by the sociologist Peter Berger’s concep-
tion of the ‘alternation’ between cultural worlds that good sociology and anthro-

59
In this connection, it is also worth emphasizing that Wittgenstein’s own attitude towards the tra-
ditional debate between ‘realism’ and ‘constructivism’ (that is, ‘idealism’) is definitely not one of philo-
sophical partisanship: see especially PI, §402.
60
For this debate see Collins and S. Yearley, ‘Epistemological Chicken’, S. Woolgar, ‘Some Remarks
About Positionism’, M. Callon and Latour, ‘Don’t Throw the Baby Out with the Bath School!’, and
Collins and Yearley, ‘Journey into Space’, all in A. Pickering, ed., op. cit. See also the Afterword to
the second edition of Collins, Changing Order.
61
This idea, as is well known, is explained in the Preface to the second edition (Princeton: Princeton
University Press, 1986) of Latour and Woolgar, Laboratory Life, where the subtitle is changed from
that of the first edition (London: Sage, 1970), The Social Construction of Scientific Facts, to The Con-
struction of Scientific Facts.
62
Although there is not space adequately to consider this issue here, the question of the relationship
between the natural sciences and the social sciences traces back, ultimately, to the debates over the
relationship between the Naturwissenschaften and the Geisteswissenschaften that arose within the late
nineteenth and early twentieth-century neo-Kantian tradition. Within the tradition of Husserlian phenom-
enology it then emerged in an especially stark form in an encounter between Husserl and Wilhelm
Dilthey in 1911. And this encounter was crucial in moving Martin Heidegger decisively to transform
Husserlian phenomenology—in explicit opposition to Husserl’s own scientific ‘universalism’—in the
direction of ‘historicism’ and ‘relativism’. This same issue, finally, forms part of the essential backdrop
to the twentieth-century divergence between ‘analytic’ and ‘continental’ philosophical traditions—and
thus, by implication, to the current problematic of ‘post-modernism’. For discussion see my ‘Overcom-
ing Metaphysics: Carnap and Heidegger’, in R. Giere and A. Richardson, eds, Origins of Logical Empiri-
cism (Minneapolis: University of Minnesota Press, 1996).
268 Studies in History and Philosophy of Science

pology requires.63 Consistency does indeed require that we apply skeptical and
relativist philosophy to SSK itself, just as much as SSK applies this philosophy to
the productions of natural science which are its primary object of study. Neverthe-
less, we cannot do both simultaneously, as it were, for, as scientists, practitioners
of SSK need to take a ‘natural attitude’ of ‘social realism’ towards the explanatory
factors they invoke (conventions, institutions, interests, and the like), just as much
as natural scientists need to take a parallel attitude of ‘natural realism’ towards the
explanatory factors invoked in natural science (electrons, quarks, gravitational
waves, and the like). We therefore need to perform a ‘meta-alternation’ between
the two standpoints whereby they are hermetically ‘compartmentalized’ from inter-
fering with one another:
[SSK] reexamines the nature of science while at the same time doing science. But
for the two activities not to interfere with one another one needs to put them in
separate compartments. This is the crucial reflexive insight.... If, then, I want to make
some new scientific objects—in this case some objects belonging to the social
sciences and having to do with the institutions of natural science—I will have to
make absolutely certain that I ignore the social origins of these objects within my
own practice as a (social) scientist and that my audience is encouraged to remain
equally ignorant. Science—the study of an apparently external world—is constituted
by not doing the sort of thing that the sociology of scientific knowledge does to
science; the point cannot be made too strongly. Sociologists of scientific knowledge
who want to find (or help construct) new objects in the world must compartmentalise;
they must not apply their methods to themselves.64
The tension between the internal norms and standards governing the practice of
science, on the one hand, and the distanced, external conclusions of skeptical and
relativist philosophy, on the other, could not be more sharply exhibited.
One interpretation of Collins’s ‘meta-alternation’ is congenial to the general
point of view I have been urging here. What I have been calling the philosophical
agenda of SSK is really not a part of the social scientific and historical practice
of SSK. From the point of view of this practice it appears rather as a fundamentally
extraneous addition which is in fact quite superfluous to the genuine empirical
achievements of SSK. (Indeed, from this point of view, it now appears that the
skeptical and relativistic philosophical agenda of SSK is positively harmful to its
empirical practice.) But this cannot be what Collins himself intends. His view is
rather that what I call the philosophical agenda of SSK is part and parcel of SSK
itself. By ‘reexamin[ing] the nature of science while at the same time doing
science’, SSK operates against the background of a presupposition in favor of
skeptical and relativistic philosophy—a presupposition which is then later con-
firmed by the empirical success of SSK.65 It is essential to Collins’s own view, in

63
See P. Berger, Invitation to Sociology (New York: Doubleday, 1963). This is the same Berger, of
course, who later, together with Thomas Luckmann, produced the classic text, The Social Construction
of Reality (London: Allen Lane, 1967).
64
Changing Order, Afterword to second edition, p. 188.
65
See op. cit., p. 185: ‘[W]hile sociology of scientific knowledge does not prove relativism it does
lead inexorably in that direction. This is because the more successful an analysis based on certain
presuppositions the more those presuppositions look right$ Just as the empirical success of descriptions
of the world based on Euclidean geometry encourages us to think that parallel lines never meet, it is
On the Sociology of Scientific Knowledge and its Philosophical Agenda 269

other words, in harmony with the general manner of proceeding within the literature
of SSK, that we do not make the crucial distinction between philosophical and
empirical investigation on which I have been insisting. How, then, can Collins
himself possibly ‘compartmentalize’ the two standpoints? How can he insulate the
empirical scientific practice of SSK from the corrosive effects of skeptical and
relativistic philosophy?
We can clarify and deepen our appreciation of this problem by glancing back
briefly at the contrasting philosophical practice of David Hume, undoubtedly the
greatest modern proponent of philosophical skepticism and naturalism. For Hume
is careful explicitly to segregate the skeptical conclusions about human reason he
reaches as a philosopher, alone in his ‘chamber’ in a condition of ‘forelorn soli-
tude’, from the ‘natural’ attitude exhibited in science and everyday life when he
emerges, whereupon his previous skeptical reflections appear ‘cold, and strain’d,
and ridiculous’. These skeptical reflections, undertaken from the philosophical
standpoint, indeed provide an effective counterweight to the pretensions of philo-
sophical rationalism. But the practicing scientist, immersed in the world and thus
in the ‘natural’ attitude, has no real use for either of these philosophical convic-
tions.66 And an analogous situation obtains in the relativism of linguistic frame-
works articulated by Rudolf Carnap. This relativism is addressed entirely to philos-
ophers, and it is intended, in particular, to free philosophy from fruitless and
interminable disputes by leading philosophers to reconceive their problems as
involving purely conventional, purely pragmatic choices of one or another form of
linguistic framework for the total language of science. In this way, most
importantly, we see that philosophy—as opposed to science—is involved with no
genuine theoretical questions whatsoever.67 For both Hume and Carnap, therefore,
we insulate the practice of science from the corrosive effects of skeptical and rela-
tivistic philosophy precisely by sharply segregating philosophy, as a discipline,
from the scientific enterprise. Here, as we have seen, they are also entirely in
agreement with Wittgenstein.

6. Conclusion
The theoretical debates in the recent literature on sociological approaches to
science appear, to my mind, to have become bogged down in tortuous metaphysics.
I fear that the smoothly running gears of empirical social history are in danger of

the fruitfulness of the sociological case studies that leads us to reevaluate the nature of science.’ For
Collins, then, philosophical relativism is as much an essential part of the empirical practice of SSK as
Euclidean geometry is of classical physics. (I hate to be pedantic, but it is tiresome to see the same
mistake so often repeated: ‘not meeting’ is the definition of ‘parallel’—the Euclidean parallel postulate
then asserts that given any line and any point not on that line in a plane, there exists one and only one
line through this point in the plane that is parallel to the original line.)
66
See A Treatise of Human Nature, Book I, Section VII. Here I am especially indebted to an unpub-
lished paper by Graciela De Pierris, ‘Causality as a Philosophical Relation in Hume’, a version of which
was presented at the Twenty-Third Hume Society Conference, University of Nottingham, July 15–
16, 1996.
67
See my ‘Tolerance and Analyticity in Carnap’s Philosophy of Mathematics’, in J. Floyd and S.
Shieh, eds., Future Pasts (Cambridge, MA: Harvard University Press, forthcoming).
270 Studies in History and Philosophy of Science

being ground to a halt by too precipitous philosophical partisanship. Nevertheless,

as we have seen, there are good historical reasons behind the contemporary fasci-
nation with skeptical and relativistic philosophy—reasons arising from the mutual
interaction between the historical evolution of scientific philosophy and the histori-
cal evolution of the sciences themselves. So I am not suggesting that social his-
torians of science simply give up on their fascination with philosophy.
My suggestion is rather that philosophically minded social historians of science
apply their historical and contextual skills to the evolution of philosophy as well—
that they treat philosophy in precisely as historically and contextually sensitive a
manner as they treat the scientific disciplines. Indeed, we already have works by
socially sensitive historians of science that do precisely this—at least with respect
to aspects of the older philosophical tradition. One important example is just the
seminal Leviathan and the Air-Pump itself, for one of the main innovations of this
study is the way in which it traces the unexpected involvement of Hobbes’s political
philosophy in seventeenth-century debates over the emergence of experimental
science.68 A second example is the work of Peter Dear on the early modern period,
in which philosophical themes in such thinkers as Descartes and Mersenne are
subtly woven together with the parallel evolution of the new science.69 A third
example is recent work of Joan Richards on the reception of non-Euclidean
geometry in Victorian England, where themes from the philosophy of geometry—
culminating in the early philosophy of geometry of Bertrand Russell—are illumi-
natingly integrated with social and intellectual history of mathematics.70 And,
finally, although I am by no means a social historian, the present discussion of the
intellectual context and development of Wittgenstein’s philosophizing may perhaps
also serve as an example, which could, in particular, point those centrally concerned
with our contemporary relativistic philosophical predicament in a more fruitful and
promising direction.

Acknowledgements—Earlier versions of this paper were presented at the University of California at San
Diego, the University of Western Ontario, Tel Aviv University, the University of St. Andrews, and the

68
Shapin and Schaffer, op. cit., especially Chapters III and IV.
69
See P. Dear, Mersenne and the Learning of the Schools (Ithaca: Cornell University Press, 1988)
and Discipline and Experience: The Mathematical Way in the Scientific Revolution (Chicago: University
of Chicago Press, 1995). I might also mention D. Bertoloni Meli, Equivalence and Priority: Newton
versus Leibniz (Oxford: Clarendon, 1993), where philosophical divergences between Newton and Leib-
niz are interwoven with mathematical, physical, and methodological ones. Happily, historians of early
modern philosophy are now paying increasing attention to such interweaving as well. See, for example,
Daniel Garber, Descartes’ Metaphysical Physics (Chicago: University of Chicago Press, 1992); Douglas
Jesseph, Berkeley’s Philosophy of Mathematics (Chicago: University of Chicago Press, 1993); Lynn
Joy, Gassendi the Atomist (Cambridge: Cambridge University Press, 1987); Paolo Mancosu, Philosophy
of Mathematics and Mathematical Practice in the Seventeenth Century (Oxford: Oxford University
Press, 1996); Catherine Wilson, Leibniz’s Metaphysics (Princeton: Princeton University Press, 1989)
and The Invisible World: Early Modern Philosophy and the Invention of the Microscope (Princeton:
Princeton University Press, 1995).
70
See J. Richards, Mathematical Visions: The Pursuit of Geometry in Victorian England (Boston:
Academic Press, 1988).
On the Sociology of Scientific Knowledge and its Philosophical Agenda 271

University of Chicago. I am indebted for discussions and advice to Gillian Barker, Frederick Beiser,
D. Bertoloni Meli, Mario Biagioli, Graciela De Pierris, Michael Dickson, Thomas Gieryn, Nicholas
Jardine, Philip Kitcher, Steven Shapin, Richard Westfall, Peter Winch, Eric Winsberg, and an anony-
mous referee for Studies in History and Philosophy of Science. And I am especially indebted to Bertoloni
Meli, in particular, for carefully reading and commenting on several different drafts.