Vous êtes sur la page 1sur 31

Social Studies

of Science
http://sss.sagepub.com/

Interests and Explanation in the Social Study of Science


Steve Woolgar
Social Studies of Science 1981 11: 365
DOI: 10.1177/030631278101100304
The online version of this article can be found at:
http://sss.sagepub.com/content/11/3/365

Published by:
http://www.sagepublications.com

Additional services and information for Social Studies of Science can be found at:
Email Alerts: http://sss.sagepub.com/cgi/alerts
Subscriptions: http://sss.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav

>> Version of Record - Aug 1, 1981


What is This?

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

ABSTRACT
Recent applications of the strong programme in the social study of science
involve (I) appeals for a naturalistic study of science and (II) the invocation of
interests as an explanatory resource. It is argued that the notion of naturalism
is insufficiently clear and that attempts to identify interests neglect important
features of scientific practice. Studies of the content of scientific knowledge
have proceeded at the expense of attention to the character of argument Itself. A
detailed examination of one example of the many recent case studies hrghlights a
series of explanatory strategies used to gloss the fundamental difficulties of
interests explanation. It is argued that rather than unreflectively attempting to
reveal interests it is more appropriate to turn our attention to the management
of explanatory strategies in the practice of scientific argument.

Interests and Explanation


in the Social Study of Science
Steve

Woolgar

One of the more intriguing aspects of the recent sociology of


science has been the emergence of the so-called strong programme
in the sociology of scientific knowledge. One major claim of the
strong programme is that the very content of science should be
amenable to social study. Not surprisingly, this kind of claim has
provoked strongest reaction where it is seen to violate wellestablished definitions of the legitimate focus of inquiry. Thus,
philosophers adhering to traditional tenets about the scope of their
investigation have been disturbed by the potential intrusion of
sociologists into their domain. At the same time, sociologists within
the Mertonian tradition have been puzzled by the daring extension
of sociological purview which is recommended. But quite apart
from resistance to potential transgression of established boundaries
of academic competence, one major reason for critical reservation
about the claims of the strong programme is uncertainty as to the
precise form of empirical work which could be said adequately to
follow the strong programme. What exactly does social study of the
content of science look like?
365-94

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

366
Those familiar with Barnes Scientific Knowledge and
Sociological Theory (SKST)2 and Bloors Knowledge and Social
Imagery (KSI)3 will recall the main arguments which can be sub-

sumed under the rubric of the strong programme. In addition to the


notion that the very content of scientific knowledge should be
susceptible to sociological analysis, it is argued that such analysis
should adopt a causal explanatory scheme; that analysis should be
impartial with respect to the perceived status (true or false, rational
or irrational, and so on) of the knowledge under study; that
analysis should be symmetrical in that the same types of cause
could explain both (perceivedly) true and false knowledge; and that
analysis of this kind could in principle apply equally to sociological
knowledge.4As presented in the opening chapter of Barnes more
recent Interests and the Growth of Knowledge (IGK),5 one main
thrust of these arguments can be summarized by the dictum that
passive, contemplative accounts of the character of knowledge
generation should be replaced by a more active sociological conception of the relation between knowledge producer and reality.6By
itself, however, the claim that the social is in some sense always involved in the production of knowledge (even scientific knowledge)
is surely no longer very remarkable. As Shapin puts it:
The mere assertion that scientific knowledge has to do with the social order or
that it is not autonomous is no longer interesting. We must now specify how,
7
precisely, to treat scientific culture as social product.~

that remains in the wake of the prostrong programme is how exactly should
sociologists proceed in constructing the influence of the social? Or,
in case this way of expressing it should unduly anticipate my argument, which specific aspects of the social are to be studied and how
are they to be incorporated into our understanding of the
mechanisms of knowledge generation?
The

important question

nouncements of the

Recently, these kinds of questions have begun to be answered.


Thus, in the few years since the publication of SKST and KSI, a

body of work has appeared which can reasonably be said to represent an attempt to extend and apply the initial programmatic formulations. This now makes it possible to discern a general trend
and to begin a tentative evaluation. The central feature of this particular body of work is the recommendation and use of interests as
an explanatory resource: change in (or involvement with) the con-

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

367
tent of scientific

knowledge, the relationship between the social and


knowledge products, is to be explained or understood in terms of
the social and/or cognitive interests of participants. I shall refer
to this usage as explanation in terms of an interests model. I suggest the time is now ripe for a close examination of the explanatory
schema employed in interests model explanations. In particular, we
need to ask whether there are any systematic weaknesses arising
from this particular interpretation of the general message of the
strong programme.
I should emphasize that the arguments I summarize under the
label of the strong programme are not per se the main focus of
discussion in this paper; a number of reviews and critiques are
available. Instead, I shall concentrate on recent extensions and applications embodied in the use of the interests model. In the writing
of those who recommend interests explanations there is a recurrent
appeal to the need for a naturalistic approach to the social study
of science. Indeed, the notion of naturalistic inquiry goes hand in
hand with the idea of explaining scientific action in terms of interests. This suggests that an examination of the use of the concept
of naturalism is prerequisite to an appreciation of some of the
deficiencies of interests explanations. Since the most readily
available exposition which incorporates both the ideas of
naturalism and interests is Barnes IGK, I shall begin my examination of these notions with special reference to their use by
Barnes in that volume. Subsequently, in the latter part of this
paper, I will turn my attention to empirical case studies employing
interests explanation in order both to illustrate and develop some of
the points made with reference to Barnes general discussion.

The Call for Naturalism


In the introduction to Natural Order (NO),9 naturalism is used to
characterize a recent change in the analytical perspective of social

studies of science: Perhaps the most significant change in the


history of science, and indeed in the study of science generally, over
the last decade is that it has become more relaxed and
naturalistic. 10 Unfortunately, the most striking feature of its recent
deployment in the science studies literature is that the concept of
naturalism is never explicated nor clearly defined.&dquo; Consequently, we are left with a variety of different impressions of the kind of

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

368
which is intended. Consider the following examples from
Barnes IGK. At one point, naturalistic seems to imply some
similarity to scientific method (... the forms of argument and explanations... are avowedly naturalistic, in the sense that the
natural sciences can be said to be naturalistic),2 notwithstanding
the obvious difficulties of presupposing the nature of the
phenomenon (that is, scientific method) which is supposed to be
one main object of study.3 Naturalistic also denotes a nonevaluative perspective, a descriptive rather than prescriptive approach, which can be contrasted with the concerns of philosophers
and epistemologists (The sociologist is concerned with the
naturalistic understanding of what people take to be knowledge
and not with the evaluative assessment of what deserves so to be
taken).4 It is also used to mean non-teleologica}l5 and, by contrast
with the grand speculative philosophical concerns of writers such as
Habermas, it is used to denote a preoccupation with low level
theoretical concerns. 16 At other places in the text naturalism
seems to denote coherence, straightforwardness (in the sense that
naturalistic analyses should not be concerned with evaluative questions) and untaintedness or lack of bias (as when it is said that the
orientations and conclusions of writers such as Smith and Ricardo
would have been naturalistic but for their modulation and amendment at the hands of ideological determination&dquo;). The undefined
and multifaceted usage of naturalism makes it difficult to discern
which particular aspect is being appealed to. More importantly, it is
unclear what precise form of explanatory account is envisaged. To
say, for example, that Historical materialism has been accepted
here only insofar as it has merits as an entirely naturalistic account
of mans activity and its historical development ,18 is to leave
unclear the specific features of historical materialist accounts which

thing

are

being applauded.9

In evaluating Barnes use of naturalism, it is instructive to draw


on the distinctions made by Matza in his attempt to develop a
naturalistic perspective in the sociological study of deviance.2
Matza notes the common use of naturalism to denote a philosophy
based on the methods and findings of natural sciences.2
Naturalism in this first sense is opposed to subjectivism, idealism
and phenomenology, and is largely equivalent to scientific
philosophy, experimental method, and a stress on the objective, external or observable features of phenomena.22 But in a second
sense, naturalism can denote an entirely different approach to

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

369

understanding. It is the philosophy that strives to remain true to the


nature of the phenomenon under study. (In terms of one of the
many slogans of latter-day interpretivists: Do not do violence to
the data.) It is against all forms of philosophical generalization
and in this sense is notable as an essentially anti-philosophical
philosophy. There is no commitment to any one single preferred
method for engaging and scrutinizing phenomena: observational
study can benefit from the inclusion of experience, introspection
and other methods usually associated with subjectivism. Matza
aligns himself unequivocally with this second version of
naturalism.23
Matzas twofold distinction makes it possible to discern a source
of confusion in Barnes usage. In very many respects, Barneys
naturalism is close to that of Matza. It is emphatically antiphilosophical in its desire to shun the concerns of philosophy and
epistemology. But at the same time it espouses an unexplicated version of scientific method in its own explanatory format. In addition, there is a very definite commitment to one particular method
for apprehending and explaining the nature of the phenomenon.

Naturalism and Interests

The confusion in Barnes use of naturalism is carried over into the


invocation of interests as a primary explanatory resource. Interests
of whatever kind are to be regarded as an explanatory resource
whose existence is taken for granted. It follows that interests not
only enjoy an unproblematic existence, to be drawn on at will by
the investigator, but that existence is essentially separate and
distinct from the scientific content they are said to explain. Given
the assumption of essential independence of phenomenon and explanatory resource, the stage is set for the deployment of a causaltype explanatory scheme. Interests can be used to explain
knowledge generation. This is not to suggest that the scope of the
game is restricted to simple causal determination, which is why I
referred to the scheme as causal-type. This means that interests can
be shown to influence rather than determine knowledge production, or that particular scientific episodes can be better understood
in the light of the particular interests of the involved parties, and so
on. The general strategy is to reveal interests as a kind of backcloth
of attendant circumstances, and to imply that this revelation

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

370
throws into better perspective the knowledge claim or event which
is at issue. In its weakest form the job can be managed merely by
juxtaposing, in the same report, the knowledge event and the
revealed circumstances, preferably with cautious caveats about the
difficulty of speculating on the precise causal mechanisms at work.
(But note how stressing the difficulty of specifying the nature of a
mechanism can do substantial rhetorical work in establishing its ex-

istence.)
In terms of my earlier comments, and to employ a contemporary
metaphor, Barnes version of naturalism is essentially two-tone.
For the phenomenon to be explained, Barnes quite rightly argues
to naturalism in the second of Matzas
But when it comes to the explanatory format, Barnes goes
for a neo-Durkheimian form of causal explanation which is
naturalistic in the first of Matzas usages. This means that the explanatory resource (interests) is emphatically not to be considered a
social resource. Interests are not to be treated as actively constructed assemblages of conventions or meaningful cultural
resources, to be understood and assessed in terms of their role in
activity, even though this is precisely the formula which Barnes
says has to be applied to all representations, pictorial or verbal,
realistic or abstract .24 For Barnes, knowledge products and scientific events of all kinds fall under the rubric of socially constructed
representations, but interests do not.
All this could be summarized by saying that the standard complaint made in connection with Mannheim could be turned back on
itself. To adopt Barnes own criticism of Mannheim, we could say
that Barnes knew and advanced many good arguments against the
contemplative account and in favour of the alternative he explicitly
advocated, but this did not suffice to orient his practical
approach.z5 The complaint against Mannheim is that he explicitly
stopped short of making natural science and mathematics the focus
of sociological study. The practical approach to which Barnes
alludes here had to do with Mannheims choice of focus. Barnes
thus criticizes Mannheim for failing to see the implication of his
own programmatic arguments; Mannheim is allegedly inconsistent
in practising a sociology of knowledge which assumes that (certain
kinds of) scientific knowledge are contemplatively available. But a
similar inconsistency is evident in Barnes own argument. For
Barnes practical approach itself involves the deployment of a body
of unexamined phenomena (interests) as sources of explanation in

for

perspective adhering

senses.

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

371

spite of their actively constructed

character. For the

practical pur-

poses at hand, interests themselves are assumed not to be amenable


to explanation; they are, as it were, contemplatively available to the

sociologist.
To anticipate one obvious objection, it could be said that some
aspect of explanation must always remain unexplicated. Indeed, it
could be argued, explanation of this kind simply cannot proceed
without assuming the pre-given status of one or other entity. Let
me return to this issue below and, for the time being, concede that
at least one factor in explanation must remain unexplicated. Even if
this is the case, it is not at all clear that there is any strong justification for the choice of interests, apart from the prevalence of this
concept in the traditional sociology of knowledge literature. Undoubtedly, Barnes arguments, especially as presented in IGK, are
both addressed to and influenced by the sociological concerns of
writers such as Marx, Lukacs and Habermas.26 Unfortunately this
orientation completely ignores evidence that the importance of
terms like interests should be approached from an entirely different
perspective.2 Scientific practice involves, crucially, the imputation
of social characteristics.28 In other words, scientists themselves can
be seen to be constantly engaged in monitoring, evaluating, attributing (in short, in accounting for) the potential presence or
absence of interests in the work and activities both of others and of
themselves. Interest-work is thus constitutive of scientific practice.
But it would not be unduly cautious to say that we have as yet little
appreciation of the way this kind of work is done by scientists. It is
at best inappropriate, therefore, to use interests as a resource at the
expense of investigating how they are accomplished. The construction and use of interests is an aspect of scientific activity which
demands treatment as a phenomenon in its own right.
In any case the notion that some aspect of explanation must remain unexplicated can only be offered as a solution given a commitment to causal-type explanation. If, for example, we undertake
an ethnographic approach to the study of scientific activity, it is
true that this will also involve the use of unexplicated resources. We
are necessarily unable simultaneously to make all features of the
scientists world the subject of our study. But since no causal explanation is required, less significance need be attached to the role
of these uninvestigated features. They remain available for further,
extended ethnographic study and our discussion need not be undermined by the suspicion that they have been constructed by the

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

372

analyst so as to perform his explanatory work. As we shall see later,


a close examination of the explanatory work involved in one example of interests explanation reveals a series of rhetorical and
argumentative strategies by which analysts attempt to avoid this
kind of suspicion. Before turning to a specific example of explanation in terms of interests, however, it is important to note one last
feature of Barnes general discussion.
Consider the following two excerpts from IGK:
... knowledge has the character of a resource, communally
achievement of whatever interests actors decide.29
Natural science,
which serve as

history, sociology,
resources

are

exploited

in the

(or potentially are) bodies of knowledge

to facilitate

prediction

and control in different

contexts. 30
Both these statements exhibit similar formats: the first part of the
statement characterizes the phenomenon to be explained
(knowledge) in terms consistent with the constructivist perspective
that is, knowledge is not just a pre-given, it is socially malleable
(one might say negotiable). In other words, knowledge is to be
understood as a resource. However, the second part of each statement indicates that actors treatment of this resource must be
understood as their treatment of a resource for a purpose. The initial part of these statements suggests that we attempt to get to grips
with the how of actors dealing in and with knowledge; the latter
part makes clear that this is just a means to an end - a second and
overarching analytical objective is that we understand why actors
do things in this way. Once again a basic asymmetry in the scheme
becomes apparent. The objects of the how domain (representations, arguments, knowledge) are to be treated naturalistically in
the sense that they are all to be viewed as actively constructed; by
contrast, the objects of the why domain are to be taken as pregiven, to be used as sources in a kind of causal explanation, and to
be invoked without making problematic their constructed
character.
As already noted, this distinction between how and why questions leaves unexamined the way in which interests are constructed,
invoked, and utilized. In addition, the unproblematically assumed
existence of interests has an important effect on the structure of ensuing explanation. If knowledge can always in principle be explained by recourse to the interests of the actors involved, then any
-

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

373
encountered in specifying the nature and extent of these
interests is merely technical. In other words, the problems
associated with attempts to specify the determinants of actions may
be severe but they are never sufficiently significant to threaten the
explanatory enterprise: the assumed existence of interests ensures
their eventual discovery or approximation, notwithstanding the
back-breaking sociological labour which might be said to be required. Of course, Barnes is aware of the difficulties. He notes that
the task of specifying concealed interests is particularly
troublesome. He speaks of the great technical difficulty involved
in identifying the role of concealed interests3 and concludes that
the difficulty of the task contrasts starkly with the confidence with
which sociologists and polemicists typically undertake it.32
Nonetheless, for Barnes, the problem remains a technical irritant
rather than the basis for any deep reappraisal of the nature of the
task. Relative to the goal of revealing the effect of interests, the
problem of identifying and attributing interests is portrayed as a
necessary evil: such technical difficulties should not obscure the
fact that the approach can be successfully applied in a considerable
number of instances.33
Thus far I have identified effects stemming from Barnes usage
of naturalism and from the selection of interests as an explanatory lynch-pin. The recourse to a causal-type explanatory format and the attempt to portray interests as the explanation of scientific action are both neglectful of the constructive activity involved
in scientific work. The proposed focus of investigation remains admirable : even the most esoteric detail of scientific knowledge
should be analyzed. But the specific form of proposed investigation
is very unsatisfactory. We are led to the heart of scientific practice
by arguments about focus, only to find that the phenomenon is
then to be subject to a sociological sledgehammer. It is not just that
the particular choice of explanans (interests) is at fault: I am not
simply advocating the replacement of interests by some alternative
explanatory factor. Instead, I am arguing that far greater reflective
attention be given to the explanatory form of interests explanations. In order further to explore the character of this explanatory
form, I now turn to an examination of some of the empirical case
studies which have adopted an approach similar to that recommended in IGK.

problem

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

374
Case Studies of Interests
There is

plethora of such

case studies. Thus we find, for exthat


the
debate
between the statisticians Yule and
ample, arguments
Pearson over measures of nominal association was informed by the
cognitive and social interests of these parties;34 that debates between biometricians and Mendelians must be understood in the
light of the different manipulative goals of the parties;35 that the
social interests of early nineteenth-century phrenologists affected
both the way in which character traits were said to be available
from the shape of a persons skull and the course of cerebral
anatomy itself;36 that the functional interests of the scientific community as a whole served to minimize the possibility of open
dispute about. the erroneous nature of Barklas discovery of J
phenomena;37 that the differing professional interests of
hereditarians and environmentalists shaped the substance of the
race-intelligence controversy;38 that interests gave rise to the ascendancy of charm over colour interpretations of particle
characteristics in theoretical high energy physics;39 that botanical
classifications are maintained, sustained and modified in the light
of social interests ;40 and so on. Similarly, some authors, although
not explicitly concerned with the revelation and use of interests,
argue that the context (whether social, historical or cultural) has a
marked effect on scientific work. 41 Of course, it is quite possible to
extend the present critique of the explanatory use of interests to the
use of context: here again is a notion mobilized by historians and
sociologists with little regard for its currency as a resource which is
constitutive of scientists argument; the concept is employed as unproblematically fixed (rather than constructed) with little reflective
attention to its employment. However, for reasons of space, I shall
confine my remarks to the use of interests.
This recent upsurge of explanations in terms of interests suggests
that we may soon reach a situation where, to borrow from the title
of Denis Wrongs famous paper (On the Oversocialized Conception of Man in Modern Sociology), we have an over-interested
conception of the scientist in modern sociology of science.42
Wrongs central complaint was that sociologists too readily assume
some processing of humans (socialization) whereby they passively
internalize norms and almost automatically respond to these internalizations in their subsequent actions. In the interests model approach we have a portrayal of scientists who act (that is, produce
knowledge) in response to their interests. For those who thought
now a

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

375
the Mertonian model had been radically displaced, the force of this
parallel is particularly striking. Granted, a major change has occurred in arguing that a focus of sociological analysis can be the very
content of scientific knowledge. But the form of analysis remains
essentially unchanged: instead of norms we now have interests.
Garfinkel has argued that the social sciences have an unfortunate
habit of making out actors to be cultural or judgemental dopes:43
that is, individuals behaviour is too readily explained by portraying it as both complying with and resulting from pre-established
alternatives of action provided by the available culture. As Garfinkel points out, this kind of portrayal neglects significant aspects
of individuals behaviour - for example, the work they do in anticipating possible rule violation, their management of the
relevance of alternative courses of action, their assessment of the
possible conditions and consequences of action, and the like. In
short, the portrayal of the individual as a judgemental dope ignores
an analysis of the individuals own language game.
To apply this to the present problem, we can see that we are dealing with two kinds of dope. On the one hand, we are all familiar
with complaints about philosophical idealizations of science.
Scientists have frequently committed themselves to print with the
view that there is no such thing as the scientific method, that
philosophers attach undue significance to the formal products of
research, that philosophers dwell selectively on instances of successful science, and so on. Recently, writers in the social study of
science have echoed these criticisms with enthusiasm, taking the inadequacy of these philosophical versions as the justification for
more sociologically-informed analysis. Indeed, as we shall see
below, some kind of preliminary invective against philosophical
characterization of science is a rhetorical sine qua non of nearly all
the interests model contributions. In short, the butt of the criticism
by sociologists is the philosophical portrayal of scientists as
rationality-dopes. That is, philosophers are held to portray scientific action as the actors response to existing knowledge and his extrapolation along some rational course. On the other hand, just as
philosophers often seem intent on portraying scientists as rationality dopes, it is now apparent that a growing body of authors are
intent on portraying scientists as interest-dopes. Once again, the
significance of this is that the social study of science, as exemplified
by interests model explanations, ignores the processes of construction whereby scientists themselves manage and attribute interests.

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

376
I should emphasize that I am not claiming any of these often very
careful case studies to be wrong: within their own terms they
display admirable consistency.44 In his review of NO, Neve also
seems to find the general approach fruitful from an historical point
of view, although he has some reservations about specific applications of the model in that volume, noting an unease which accompanies certain versions of the &dquo;interest&dquo; analysis,45 and referring
to the difficulty of the reductionist use of &dquo;interest&dquo; as a
methodology .46 In the light of the problems already raised in relation to Barnes argument, let us look in detail at an example of the
empirical application of the interests model. Apart from illustrating some of the argument already made about interests explanations, an examination of an empirical example also provides
an opportunity for inquiring into the social character of explanation in general. In other words, I shall take the critical examination
of empirical interests explanations as an occasion for beginning to
formulate key questions about the nature of explanatory techniques and strategies.

The

Example of the Yule-Pearson Debate

In what follows I will make special reference to MacKenzies


analysis of the debate between the statisticians Yule and Pearson in
the early years of the twentieth century. 41 I have not chosen
MacKenzies study because it provides the easiest opportunity for
picking holes; nor is that my objective. On the contrary, his is one
of the most careful and fascinating studies of the genre. I have
chosen it here because the extent of detail and care of presentation
make it especially useful as the basis for an examination of the

method employed.
MacKenzie begins with an account of the divergent views of Yule
and Pearson. As represented in their publications of 1900, Yule and
Pearson proposed alternative and competing measures of statistical
association between two nominal variables. Yules reasoning was
based, in part, on values which a measure should assume if the two
variables under study turned out to be independent (Q = 0), fully
positively associated (Q = + 1) and fully negatively associated
(Q = - 1). His proposed measure, Yules Q, is calculated by a formula comprising the four observed cell frequencies of a 2 by 2
table. Pearson reasoned that the observed frequencies should more

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

377

exactly be regarded as resulting from the conjunction of underlying


normal distributions. He proposed a measure (r) which reflected
this underlying distribution and went on to develop another
measure (Pearsons C) which could handle association between
nominal variables in tables larger than 2 by 2. Briefly, the substance
of the controversy between the two statisticians involved attacks by
Yule on Pearsons assumption about the nature of the underlying
distribution, and counterattacks by Pearson that Yules criticism
was ill-founded and that the Q measure was too simplistic.
MacKenzie argues that it is insufficient to account for this controversy simply by noting the objections exchanged between the
disputants. Instead, he suggests, it is preferable to take note of the
cognitive interests of the two parties. Pearson is said to have had a
deep-rooted commitment to the utility of statistical prediction; existing interval level theory of correlation and regression satisfied
this concern, and MacKenzie claims that it was therefore in Pearsons interest to develop nominal measures by analogy with the existing theory. Yule, on the other hand, is said to have exhibited a
much more pragmatic concern with the data itself, rather than with
any desire to maximize analogy with existing statistical theory.
Yules methods were thus structured by a cognitive interest in
prediction using nominal data as phenomena in their own rights
MacKenzie then argues that the cognitive interests of the two can
be related to their differing objectives in the development of
statistical theory and perhaps ultimately to differing social interests.49 A further deep-rooted commitment of Pearson is revealed : the overarching objective of his work turns out to be the
development of measures of hereditary relationship between successive generations and, in particular, the pursuit of some means of
testing association between non-measurable (nominal) variables
such as mental character and individual type. Yules proposed
measure did not facilitate direct comparison with existing interval
measures of association and hence did not serve Pearsons eugenic
interests. Yule had no particular interest in eugenics. MacKenzie
describes him as having a more diffuse goal orientation which led
to a looser and more empirical approach which embodied cognitive
interests of a more general nature.10 MacKenzie takes the argument further by locating the controversy in the context of the contemporary statistical community: Pearsons work, and the support
for his side of the debate, is found to be associated with membership of a small coherent school of statisticians working in

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

378

biometrics; whereas Yules affiliation was less specific. MacKenzie


finally suggests that the different attitudes of Pearson and Yule can
be further understood within two differing constellations of contemporary social interests: Pearson was an archetypical member of
the rising professional class whose growing importance was
associated with its espousal of ideologies which emphasized its difference from, and superiority to, the manual working class, and
pointed to the social value of professional knowledge and skill
as against the mere ownership of capital or land;5 Yules
background, by contrast, was that of the downwardly mobile traditional elite with generally conservative tastes and a dislike for
...

eugenics.
The central facets of MacKenzies argument are clear: firstly, the
focus of his analysis is the very nature of the alternative
mathematical measures proposed by Yule and Pearson; secondly,
differences in these measures are to be explained by various sets of
prevailing interests. For our immediate purposes, little significance
need be attached to the claimed distinction between cognitive and
social interests. Suffice it to note that the distinction does raise
issues requiring further consideration. In the first place, as has been
argued in some detail elsewhere, the distinction between what is to
count as cognitive rather than social (and vice versa) is an important feature of participants own discourse; the active achievement and management of the distinction is central to scientific
argument.52 Proponents of interests explanations overlook the importance of the social/cognitive distinction for scientists by attempting to provide definitive accounts of the type of interests at work.
Secondly, the particular distinction which is claimed to exist between cognitive and social interests can be read as indicating
that certain kinds of scientific action derive from the existing corpus of scientific knowledge while others do not. At least in some
forms, then, the claimed distinction between cognitive and
social interests (strangely reminiscent of the well-worn internalist/externalist dichotomy) is in danger of being heard as a plea
for the reintroduction of a classical philosophy of science.
As can be seen, even from a necessarily brief rendering of
MacKenzies argument, considerable constructive work has gone
into his revelation of the interests involved. Let us now begin to
unpack (or deconstruct) this work and, bearing in mind the
assumption that this is roughly typical of the case studies listed
above, identify some of its main explanatory features.

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

Explanatory Features of
Interests Explanation
J. What

seems

rational is

actually social

One strategy frequently employed in interests model explanations


in the sociology of science is to set up an initial springboard against
which the thrust of the argument is to be directed. Thus MacKenzie
opens his account with the following:
The esoteric knowledge to be found in the mathematical sciences is frequently
held to develop accordmg to its own laws, immune from social influence. The
purpose of this paper is to cast doubt on that assumption.53

This has the effect of supplying the reader with

an existing state of
affairs (a philosophical characterization of mathematical practice)
and suggesting that this state is somehow incorrect, wrong or otherwise requires re-examination. To contrast an allegedly incorrect
state of affairs with what is to follow is to claim strong grounds for
justification.54 Thus the interests model argument is claimed to be
an alternative to a deficient philosophical model - or, in terms of
the earlier discussion, a picture of scientists as interest-dopes is to
be substituted for their previous portrayal as rationality-dopes. For
present purposes the important point is that both are alternative
constructions. To replace one by the other misses at least two important questions: what counts as legitimate construction in practical argument; and what counts as adequate grounds for substitution of one construction for another? The central importance of
this initial justification can be assessed by discerning the effects on
the argument of its removal. Without the alternative philosophicalrational version for contrast, the claim for the social character of
science becomes rather mundane. This is evident from MacKenziess
own very candid concluding remarks:

In the absence of a great deal of further research, particularly on the


hypothetical constellation of interests suggested in the previous section, this conclusion must remain tentative. I hope, however, that this paper has shown that
hard sciences, such as the mathematical theory of statistics, should not be excluded a priori from analysis in terms of social interests.55

Elsewhere I have discussed the particular form of contrast being


worked with here in terms of the debate between rationalists and

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

380

the strong programme, and concluded that alignment to one or


other position is of little consequence for a fruitful understanding
of scientists activity.56 Instead we should recognize that the construction and attribution of everyday versions of both rationalist
and strong programme positions is a continual feature of interaction between scientists. It is the localized management of these
positions (which usually has both immediate and significant practical consequences) which demands our attention as analysts.
2.

Scientific action

is

expressive of concomitant interests

A second related strategy is crucial to the interests model. This is


the demonstration or revelation of interests behind instances of
scientific action to be explained. It is possible to represent this
strategy schematically as an algorithm without, I think, unduly
simplifying the explanatory forms
a. Identify an instance of scientific action.
b. Take action to be an indicator of an underlying desire on the
part of the actor.
c. Repeat steps a and b so as to generate a set of desires.
d. Equate this set of desires with actors concomitant interests.
e. Argue that action to be explained is consistent with concomitant interests.
In this general scheme, scientific action denotes a wide range of

activities corresponding to a number of different kinds of sources.


For example, action can refer to an entire dispute, to a specific
item of correspondence between scientists, or to a particular utterance made in the laboratory. In each case, the item is taken as an
indicator of some external or underlying desire. 58 In MacKenzie, for example, Yules formulation of the Q measure is taken as
indicator of his desire that it should attain certain values under certain conditions. Another action, Pearsons criticism of Yule, is
taken to indicate Pearsons desire to create a measure consistent
with existing statistical theory. Note also that the same instance of
action can be read as indicative of more than one desire, and conversely, that different actions can be interpreted as indicative of the
same desire. Thus in MacKenzies account, Pearsons development
of the C measure, just like his criticism of Yule, indicates his concern to maximize analogies with interval level theory.
Clearly, the desires generated in this manner do not stem solely
from observed action. Proponents of the interests model obviously

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

381

begin with some notion of the nature and variety of desires they will
identify. More importantly, the generation of various desires involves the specification of motives, intentions, underlying concerns
and so on in the light of previously revealed desires. Thus, the
specification of any particular desire is bound to be influenced by
previous interpretations of other actions. But my point is not
simply that the analysts choice of underlying desire is biased (it
would be difficult to imagine what an unbiased choice would look
like). Rather, it is clear that a very large (if not infinite) number of
alternative desires is in principle discernible; the selection and construction of a set of desires from the number potentially available
thus involves complex argumentative work.
If a large number of alternative desires is discernible, how in
practice are we to identify one particular desire as demonstrably
corresponding to the observed action? The answer must be that the
analyst has to specify what it would be rational for the actor to
desire in the prevailing circumstances: in other words, it is
necessary for the analyst to create some rational connection between action and desire. However this is done, it will involve some
judgement as to what is rational. In the example from MacKenzies
account, Pearsons criticism of Yules measure is taken to indicate
his concern to maximize analogies with other statistical measures,
in part because criticizing is reckoned to be a rational course of
action given this concern. It need hardly be pointed out, of course,
that other alternative actions follow rationally from this same
concern: in other words, there is no necessary correspondence between any specfic concern and any particular action. In addition, it
is fairly obvious that other alternative desires could be said to be
the rational antecedent of the same action. Pearsons criticism of
Yule could be said to follow rationally from his (Pearsons) desire
to be bloody-minded.59 This kind of alternative is either considered
and rejected or, more likely, is simply not considered, by virtue of
the influence on the analyst of his interpretation of the prevailing
circumstances in which the action took place. In a similar way, the
construction of a set of desires (step c) seems to involve judgements
as to the rational coherence of that set. Actions and desires will be
added to the set only if they do not conflict with other members of
the set. In general, the analyst will construct a two-way link between action and desire: it is the basis for the initial inference as to
the nature of the desire; at the same time, the desire is used to explain the occurrence of action.

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

382

perspective, this modus operandi could be said to be


The very phenomenon to be explained is the basis
circular.
patently
for the constructed explanation. But whether or not this kind of explanation actually is circular is of little concern here. Perhaps all
attempts to reveal socio-historical circumstances inevitably follow
some variant of this scheme. Nonetheless, the fact that a charge of
circularity could be made raises some intriguing issues. How is it
that in most instances of practical argument no significance is attached to its defeasibility on grounds of circularity? What are the
mechanisms by which parties to an argument either reveal or ignore
the presence of circularity? Of course, these and similar questions
are of considerable import in the context of furthering our
understanding of science. They suggest that sustained study of
scientific practice should pay much closer attention to the details of
argumentative rhetoric. For now, however, I shall confine my
remarks to the evasion of charges of circularity in sociological
usage of the interests model.
From

one

3. The identified interests


of the action they explain

are

independent

Garfinkel speaks of the documentary method to denote the process


by which underlying patterns are discovered on the basis of
observed appearances .6 He shows that neither the appearances
nor the underlying pattern is fixed and independent of the other.
Instead the sense and character of the appearance (in our present
case, the instance of scientific action) is modified as its relationship with the underlying pattern (in our case, the desires of the
actor) is constructed. Identification of an underlying pattern thus
involves a back-and-forth process whereby neither entity is ever independent of the other. The notion of reflexivity has also been
used to refer to the general property of discourse whereby the invocation of an ostensibly underlying pattern is inevitably part and
parcel of the scene which it purports to explain; the sense and
meaning of the scene depends on, and is inextricably tied to, the invocation of underlying pattern.
One of the best illustrations of reflexivity appears in Wieders
study of the convict code. 61 Indeed, there are some important
parallels between the present argument and that advanced by
Wieder. Wieder notes that the actions of the inmates of a halfway
house (in our case, the actions of scientists) could be sociologically

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

383
to the existence of a set of rules (interests).
he
shows
that
the specification of correspondence beHowever,
tween any particular action and the rule (interests) which it is said
to be following (expressive of) is in principle defeasible. In addition, rules (interests) often implied contradictory courses of action:
that is, a particular action could be explained as being governed
by one rule, even though it contravened another. The invocation of
the rule-governed character of action was reflexive in that it
simultaneously modified the nature of the action and the interpretation of the rule to which it was said to be related.62 Mulkay has
similarly pointed out that the interpretation of scientific action by
recourse to underlying rules (in particular, the norms of science) is
subject to difficulties of interpretation and contradiction.63 Most
importantly, both Wieder and Mulkay note that the issue of correspondence between action and underlying pattern (whether rules,
norms or, in this case, interests) is crucially a matter for the participants involved. The programmatic implication is that we should
develop an understanding of the practical management of correspondence between actions and underlying patterns, rather than
simply engage in unreflective attempts to advance such cor-

explained by appeal

respondences.
Garfinkel notes that the essential reflexivity of accounts is
uninteresting to their proponents.64 By this I take him to mean
that accounting practice depends for its success on the routine
denial of the intimate interdependence between underlying pattern and the scene which is being accounted. Translating this into
terms more apposite to the current discussion, it is necessary for
purposes of successful argument to background65 the constructive
interrelationship between the thing to be explained and the thing
which does the explaining. In other words, an important, if not
crucial, feature of argument in general, and a fortiori of scientific
argument in particular, is that independence be achieved between
explanans and explandum; in the present example, the desires of
the actors have to be made to seem independent of the actions from
which they are created. Only then is it possible to argue for the
prior existence of the desires and hence to imply that the actions
resulted from these pre-existing desires.
How is the independence of interests accomplished? We can shed
some light on this facet of interests model explanations by drawing
a parallel with a recent discussion of the process whereby facts are
constructed in neuroendocrinology.66 Typically, fact construction

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

384

comprises three main stages. Initially scientists deal in statements


of varying degrees of speculation. Secondly, however, as a stable
form of a particular statement is gradually established, a transformation occurs whereby the statement is no longer just a statement,
it is a statement about something. The collection of words acquires
a

currency in discourse which creates

referent in the real world.

The real

worldly object is thus created by virtue of the statement:


through a process of splitting, the statement becomes both separate
from and independent of the object which it is supposed to be
about. A third stage involves the inversion of this relationship:
the object becomes transformed into an entity which had existed
all along and, more importantly, the prior existence of the object is
now seen as having given rise to the statement about the object.
The statement, far from being regarded as the basis for the objects
construction, is now seen as a report which attempts to reflect
some of the actual character of the object.
In MacKenzies account, we are initially presented with the
published work of Yule and Pearson. As the account proceeds, we
find that the work of Yule and Pearson is the basis for MacKenzies
inference of underlying interests; the statistical publications are
made to speak to the existence of these interests. Thus the
arguments in these publications are presented not merely as
arguments, but as arguments-made-for-a-purpose. This allows the
arguments and the purposes (interests) to appear as two analytically
distinct entities. Finally, the interests which were initially constructed from the actions of Yule and Pearson (their published
arguments) are cited as the antecedents of these actions: The differing cognitive interests of Pearson and Yule led to their two positions being incommensurable. 67
This rhetorical transformation seems crucial to the success of an
interests model explanation. Without splitting, scientific action
and underlying desire are inextricably tied in the manner Garfinkel
points out. This is hardly the basis for adequate causal explanation.
In addition, inversion is necessary to avoid the charge that the explanans (underlying desires) is simply being created from the explanandum (the scientific action): this would be something akin to
assuming the answer in a mathematical proof. 68 Without inversion,
it would simply not do to take actions as indicative of underlying
desires and then subsequently cite these same desires as the guiding
hand behind scientific actions.

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

385

Of course, the application of the algorithm outlined above is


rarely as straightforward as I have suggested thus far. Sophisticated
applications (such as that by MacKenzie) make the suspicion of circularity or, to be more precise, the accomplished avoidance of
suspicion, much less immediately obvious. Three further features
of interests-model explanation, which I shall now briefly illustrate,
help to background or minimize the significance of possible cir-

cularity.
4. Multiple instances of action affirm
the independent existence of interests
that any constructed rational connection beprinciple defeasible: that is, it is always
candidate desire which underlies
to
an
alternative
possible propose
a specific instance of action. In the instance of Pearsons criticism
of Yule, for example, the desire to be bloody-minded is one such
suggested alternative to the desire to maximize analogies with existing statistical theory. One strategy for defending the choice of a
particular desire is the citation of other instances of action which
are said to support the initial interpretation as to the underlying
desire.69 Thus, in MacKenzies account, we are told about
Pearsons work in general that it was dominated by its reference to
an existing achievement of statistical theory, the interval level
theory of correlation and regression.10 The reader is asked to
assume that this is the outcome of the interpretation of several
other documents produced by Pearson. The fact that additional
constructive work is necessarily involved in the production of these
auxiliary interpretations is played down in several ways. Firstly,
the reader is not told how many other instances of scientific action
(in this case, examples of Pearsons work) have been consulted. Instead, these instances are characterized as a category, namely
Pearsons work, and the coherence of this category for its
relevance to the overall interpretation is assumed. Secondly, there
is no explicit reference to the construction of interpretations which
must have gone on. The facticity of the interpretation is enhanced
by minimizing or backgrounding the active involvement of the interpreter.&dquo; Sometimes (although not in the present example)
authors adopt a variant of the inversion strategy mentioned above
whereby the facts of the matter are presented as apparent to anyone
who cared to look. In a later part of the paper, for example,
I earlier

suggested

tween action and desire is in

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

386
MacKenzie refers to what does, I think, emerge from his [Yules]
letters.. :.72 Notice here that MacKenzie himself appears largely
incidental to the process whereby the interpretation of Yules letters
gets done. Fairly obviously, this effect results, in part, from his
avoidance of phrases like what I can construct from Yules letters.
Thirdly, it is implied that the multiple interpretations of different
instances of action have been done in isolation from each other. In
other words, the supposition is that each interpretation is independent of, and hence uninfluenced by, any other. As mentioned
already, it is impossible to make interpretations in complete isolation from other existing interpretations, expectations about what
should be the case, and so on: interpretation is emphatically not
observation-neutral. Nonetheless, attention to this difficulty of
principle is diverted by, for example, stressing the difference between the kinds of source on which multiple interpretations are
based. Thus, in the above case, the underlying personality of Yule
emerges not just from his letters, but also from comments on him
by those who knew him well and from occasional passages in his

writings .73 Fourthly, not only are the resultant set of interpretations created independently, they also turn out to be the same.
That is, each individual interpretation of action is presented as having given rise to the same underlying pattern even though there is
no mention of the criteria employed to determine similarity or difference.
A special case of the multiple interpretation strategy occurs when
particular emphasis is given to the separation in time of actions and
the desires which supposedly underly them. Their presentation as
separated in time enhances the independence of action and underlying desire, and this in turn paves the way for the adequacy of their
causal relationship. Thus the desire purportedly underlying Pearsons criticism of Yule is a desire which was there all along. The
desire (to maximize analogies with existing statistical theory) is proin this case, other work by
posed on the basis of prior action
Pearson. Obviously, the same difficulty of principle applies in proposing a connection between Pearsons prior work and the desire it
is said to reflect. But possible accusations of circularity are
backgrounded because the focus of explanation is the more recent
act of criticizing Yule. The action to be explained (the criticism) is
not directly the basis of the inferred desire.
-

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

387
5. The relationship between action and interests
is not actually causal

already mentioned, the aspiration to causality in the proposed


explanation is frequently played down. This is because the criterion
of independence of explanans and explanandum becomes much less
important if there are no strong claims for the causal character of
their relationship. To say, for example, that a discussion of interests helps to shed light on a particular series of scientific actions is significantly different from saying that these actions
resulted from or were caused by these interests. Typical in this
respect is the admirably cautious nature of MacKenzies remarks
As

about
examining possible connections between eugenically relevant research and social
interests .... Study of this situation can hopefully illuminate the choices [of individuals such as Pearson and Yule], even if it cannot provide a causal account
of them. 74

Nonetheless, the format of the explanatory work is substantially


the same, however cautious the claim finally made for the disclosed

relationship: the portrayal of scientific actions in the light of


prevalent interests is axiomatic to this style of explanation.
6. The interests specified

are

those

of a collectivity,

rather than of an individual


Another feature of sophisticated uses of the above algorithm hinges
on steps c and d and is almost exclusive to sociological explanation.
The significance of the constructed relationship between action and
desire is minimized by the claim that the (constructed) set of desires
need not actually apply to ~ny particular individual case. The
desires are worked up on the basis of a variety of observed actions
(usually on the part of individuals) and their internal coherence as a
set of desires is built into the construction. The set as a whole is
then attributed to a particular collectivity. The interests are not
necessarily the interests of an individual; rather they become the interests of a social group in which the action to be explained took
place. Although the construction of these interests is based on the
interpretation of individual actions, it is rhetorically necessary to

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

388
the putative cause as far as possible from the scene of individual action while nonetheless asserting its pervasive influence.
remove

So

between the needs of eugenic research and the cognitive inmanifested in the development of the theory of association by the
biometric school can be reasonably held to exist, irrespective of the particular
motives of individual members of the school. 75
a

relationship

terests

a companion study of the debate between Yule and


Norton
made explicit his reservations about the adoption
Pearson,
of this kind of sociological strategy:

In

concluding

we can see that many of Pearsons ideas appear to be enhancing the


of the group with whom he identified. Whether or not such harmonisation can be seen as explaining his espousal of these ideas is, it seems to me, a
question that brings us hard up against the philosophical difficulties inherent in
explaining an individuals thought in terms of the interests of a group to which
he has attached himself.76

In

short,

esteem

Conclusion
I began by noting that although the initial proposals of the strong
programme in the sociology of knowledge were unclear as to the
precise form of explanatory account which sociologists should
follow, a larger number of more recent contributions have pursued
interests model explanations. The objective of revealing the social
character of the content of scientific knowledge has been interpreted as a programme of specifying the interests which give rise
to scientists actions. In IGK in particular, we can identify a confusion between (a) the need to focus on knowledge content, and (b)
the adoption of an unexplicated causal-type format for explaining

this content in terms of interests.


It seems to me that the latter is a disappointing and, at best,
premature development from the former. Having admirably drawn
attention to the need to analyze scientific content, some sociologists
and historians have too readily jumped to the deployment of interests without yet understanding how such deployment is achieved, managed and sustained. This is particularly important in the
sociology of science where the practical character of argument must
surely be a major analytical concern. We should at least be committed to a high degree of methodological reflection, as a first step in

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

389

understanding the practical management of the


philosophical difficulty inherent in explanation.

many kinds of

In the second part of this paper, I examined an empirical example of interests explanation. I have shown that the construction of
interests on the basis of actions and their subsequent use in explaining actions entailed an explanatory schema in principle beset by
methodological difficulties. These difficulties were backgrounded,
minimized and otherwise made to seem inconsequential by the use
of various rhetorical and argumentative strategies. I suggest that
the kind of methodological reflection which I am advocating, and
which is tentatively illustrated by my examination of interests explanations, can have a positive pay-off for the social study of
science. Although the particular focus of my analysis here was an
example of the explanatory schema in recent contributions to the
social study of science, it would obviously be highly desirable to apply the same style of analysis to natural science. The central question is: what is the character of the constructive work involved in
scientific argument? More specifically, what counts as legitimate
avoidance of what might otherwise be regarded as insurmountable
philosophical difficulties? How are presentational devices used to
minimize the possibility of critical intervention by others? What
argumentative strategies enable scientists routinely to accomplish
and sustain the rationality of their interpretations in the face of
the ever-present possibility of better alternative interpretations?
At the very least, detailed attention to these kinds of questions may
help redress the current imbalance of overly enthusiastic sociologizing in the social study of science. More importantly, it is by this
kind of approach that we can most fruitfully proceed to realize the
potential of a programme which directs our attention to the very
content of scientific knowledge.

NOTES

I am grateful for the comments and suggestions made by Peter Halfpenny and
Malcolm Spector on their reading an earlier draft of this paper, and for the advice of
anonymous referees.

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

390
1. See, especially, S.B. Barnes, Scientific Knowledge and Sociological Theory
(SKST) (London: Routledge and Kegan Paul, 1974); D. Bloor, Wittgenstein and
Mannheim on the Sociology of Mathematics, Studies in the History and Philosophy
of Science, Vol. 4 (1973), 173-91; Bloor, Knowledge and Social Imagery (KSI) (London : Routledge and Kegan Paul, 1976).

2. Barnes, op. cit. note 1.


3. Bloor, op. cit. note 1.
4. Bloor uses reflexive to describe this last clause of the strong programme
(KSI, 5). I reserve the term for use in the sense proposed by Garfinkel: see H. Garfinkel, Studies in Ethnomethodology (Englewood Cliffs, NJ: Prentice-Hall, 1967),
7 ff. See also notes 61 and 62, below. For a discussion of reflexivity in the sense
Bloor uses the term, see B. Gruenberg, The Problem of Reflexivity in the Sociology
of Science, Philosophy of the Social Sciences, Vol. 8 (1978), 321-43.
5. Barry Barnes, Interests and the Growth of Knowledge (IGK) (London:
Routledge and Kegan Paul, 1978).
6. For present purposes, I leave aside the issue of reality and its relevance to
sociological arguments, except to note that Barnes is not arguing for simple
sociological relativism: he sees social context as one additional feature of explanations which show how knowledge derives from reality. See IGK, 2 and 10. For correspondence relating to this argument as presented in SKST, see H. Collins and G.
Cox, Recovering Relativity: Did Prophecy Fail?, Social Studies of Science, Vol. 6
(1976), 423-44; J. Law, Prophecy Failed (for the Actors)!: A Note on Recovering
Relativity, ibid., Vol. 7 (1977), 367-72; Collins and Cox, Relativity Revisited: Mrs.
Keech — A Suitable Case for Special Treatment?, ibid., 372-80.
7. S. Shapin, Homo Phrenologicus: Anthropological Perspectives on an
Historical Problem, in B. Barnes and S. Shapin (eds), Natural Order: Historical
Studies of Scientific Culture (London and Beverly Hills, Calif.: Sage, 1979), 42.
8. See, for example, the review of SKST
by S. Lukes, Social Studies of Science,
Vol. 5 (1975), 501-05; H. Meynell, On the Limits of the Sociology of Knowledge,
ibid., Vol. 7 (1977), 489-500; E. Millstone, A Framework for the Sociology of
Knowledge, ibid., Vol. 8 (1978), 111-25; and G. Freudenthal, How Strong is Dr
Bloors Strong Programme?, Studies in the History and Philosophy of Science,
Vol. 10 (1979), 67-83.
9. Barnes and Shapin (eds), op. cit. note 7.
10. B. Barnes and S. Shapin, Introduction, in ibid., 9.
11. Not that this is peculiar to science studies: cf., for example, N. Denzin, The
Logic of Naturalistic Inquiry, Social Forces, Vol. 50 (1971), 166-82.
12. Barnes, IGK, viii.
13. In his review of Bloors KSI, Johnston notes that a similar position is stated
with a breathtaking simplicity that belies the magnitude of the shift in common
belief involved: R. Johnston, British Journal for the History of Science, Vol. 11
(1978), 65-66.
14. Barnes, IGK, 1.
15. Ibid., 12 and 66.
16. Ibid., 13.
17. Ibid., 32.
18. Ibid., 86.
19. Michael Neve makes a similar point with respect to Natural Order, when he
says that it is not clear ... what "naturalism" means for the purposes of that

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

391
book: Neve, The Naturalization of Science, Social Studies of Science, Vol. 10
(1980), 375-91, at 386.
20. D. Matza, Becoming Deviant (Englewood Cliffs, NJ: Prentice-Hall, 1969),
Chapter 1.
21. Of course, my use of based on belies a complex and yet to be understood
process whereby a version of scientific method is abstracted from scientific practice
for purposes of philosophical discourse.
22. Cf. a dictionary definition of naturalism as an approach to psychology and
social science which assumes that human beings are essentially physico-chemical
systems, and can be studied in exactly the same way as the rest of the physical
world: A. Bullock and O. Stallybrass (eds), The Fontana Dictionary of Modern
Thought (London: Fontana, 1977), 411. Bhaskar similarly speaks of naturalism as
the thesis that there is (or can be) an essential unity of method between the natural
and social sciences: R. Bhaskar, On the Possibility of Social Scientific Knowledge
and the Limits of Naturalism, Journal for the Theory of Social Behaviour, Vol. 8
(1978), 2.
23. In an intriguing statement, Matza writes: The commitment of naturalism, as
I conceive it, is to phenomena and their nature, not to Science or any other system of
standards (op. cit. note 20, 3). At first sight, this formulation seems to provoke a
rather knotty problem as soon as we apply it to the study of science. What happens
if the phenomenon itself is science? But there is in fact no problem if we maintain
the distinction between science as a phenomenon, the character of which is managed
by and made available to participants, and Science as a pre-given system of standards which some (positivist?) researchers would wish to employ as the major
criterion of adequacy of their work.
24. Barnes, IGK, 9.
25. Ibid., 4.
26. IGK can be read as attempting to forge links between recent empirically based
work in the sociology of science and the more theoretical discussion in the sociology
of knowledge. Specifically, Barnes aim is to show that some of the traditional
preoccupations of the sociology of knowledge are of use in analyzing the content of
scientific knowledge. In this respect, the treatment of sociology of knowledge traditions in IGK is admirably more specific than we are generally used to. Another
discussion which relates sociology of knowledge concerns to the possibility of studying scientific content, and which includes a detailed appraisal of some recent empirical contributions in the sociology of science, is M.J. Mulkay, Science and the
Sociology of Knowledge (London: George Allen and Unwin, 1979).
27. It is worth noting that I am not simply recommending a micro as opposed to
macro perspective. It would be misleading to formulate my argument in terms of
this contrast because my main point is directed, not so much against the choice of
level of substantive analysis in interests model explanations, as against the unreflective form of explanation employed. As I hope will become clear, even those interests
explanations which try to explain the details of specific interactions between scientists suffer the same deficiencies of explanatory format which I am attempting to
elucidate.
28. B. Latour and S. Woolgar, Laboratory Life: The Social Construction of
Scientific Facts (Beverly Hills, Calif.: Sage, 1979), Chapter 4; Woolgar, Discovery:
Logic and Sequence in a Scientific Text, in K. Knorr, R. Krohn and R.D. Whitley
(eds), Sociology of the Sciences Yearbook, Vol. 4: The Social Process of Scientific

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

392
Investigation (Dordrecht: Reidel, 1980), 239-68. Further evidence of a similar kind is
available in some recent ethnographic studies of interaction between scientists in
laboratory settings: I am editing a collection of such studies for future publication as
a Special Issue of Social Studies of Science.
29. Barnes, IGK, 16.
30. Ibid.
31. Ibid., 37.
32. Ibid.
33. Ibid., 35.
34. D. MacKenzie, Statistical Theory and Social Interests: A Case Study, Social
Studies of Science, Vol. 8 (1978), 35-83; B. Barnes and MacKenzie, On the Role of
Interests in Scientific Change, in R. Wallis (ed.), On the Margins of Science: The
Social Construction of Rejected Knowledge (Keele, Staffs.: University of Keele,
Sociological Review Monograph No. 27, 1979), 49-66.
35. Barnes, IGK, 59-63; D. MacKenzie and B. Barnes, Scientific Judgement:
The Biometry-Mendelism Controversy, in Barnes and Shapin (eds), op. cit. note 7,
191-210.
36. S. Shapin, Phrenological Knowledge and the Social Structure of Early
Nineteenth-Century Edinburgh, Annals of Science, Vol. 32 (1975), 219-43; Shapin,
The Politics of Observation: Cerebral Anatomy and Social Interests in the Edinburgh Phrenology Disputes, in Wallis (ed.), op. cit. note 34, 139-78; Shapin,
Homo Phrenologicus, op. cit. note 7, 41-71.
37. B. Wynne, C.G. Barkla and the J Phenomenon: A Case Study in the Treatment of Deviance in Physics, Social Studies of Science, Vol. 6 (1976), 307-47.
38. J. Harwood, The Race-Intelligence Controversy: A Sociological Approach. I
Professional Factors, Social Studies of Science, Vol. 6 (1976), 369-94.
39. A. Pickering, The Role of Interests in High-Energy Physics: The Choice between Charm and Colour, in Knorr et al. (eds), op. cit. note 28, 107-38.
40. J. Dean, Controversy over Classification: A Case Study from the History of
Botany, in Barnes and Shapin (eds), op. cit. note 7, 211-30.
41. See, for example, B. Harvey, The Effects of Social Context on the Process of
Scientific Investigation: Experimental Tests of Quantum Mechanics, in Knorr et al.
(eds), op. cit. note 28, 139-63.
42. D. Wrong, On the Oversocialized Conception of Man in Modern Sociology,
American Sociological Review, Vol. 26 (1961), 183-93.
43. Garfinkel, op. cit. note 4, Chapter 2, 67 ff.
44. This is apart, of course, from the systematic inconsistency evident when these
studies include programmatic claims about the inadequacy of Mannheims sociology
of knowledge. See the discussion in the first part of this paper.
45. Neve, op. cit. note 19, 387.
46. Ibid., 390.
47. MacKenzie, Statistical Theory and Social Interests, op. cit. note 34. Since
completing this paper, MacKenzies work has become available as a book. I have
not been able to take this fuller version into account: see Donald A. MacKenzie,
Statistics in Britain 1865-1930 (Edinburgh: Edinburgh University Press, 1981).
48. MacKenzie, Statistical Theory..., op. cit. note 34, 52.
49. Ibid., 53.
50. Ibid., 60.
51. Ibid., 67.
—

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

393
52. Latour and Woolgar, op. cit. note 28, esp. Chapter 1.
53. MacKenzie, Statistical Theory ..., op. cit. note 34, 35.
54. This strategy enjoys general usage in sociological argument, most obviously
in the areas of deviance and studies of the media. Typically, a popular stereotype,
for example what-homosexuals-are-like, is used as the strawman in demonstrations
that this stereotype is false or inaccurate. False or inaccurate versions of the actual
state of affairs are then said to arise by virtue of the intrusion of social factors.
55. MacKenzie, Statistical Theory ..., op. cit. note 34, 71.
56. Woolgar, Discovery ..., op. cit. note 28.
57. I am grateful to Herminio Martins and Peter Halfpenny for pointing out that
this algorithm is a variant of the practical syllogism discussed by von Wright: see
G.H. von Wnght, Explanation and Understanding (London: Routledge and Kegan
Paul, 1971); also J. Mannien and R. Tuomela (eds), Essays on Explanation and

Understanding (Dordrecht: Reidel, 1976).


58. Desire is used throughout to stand synonymously for terms such as
motive, intention, purpose, concern, reason, and so on. All such items are
constituents of sets of interests.
59. Those who assume that such a possibility is ludicrous should recall that
similar interpretations are regularly accomplished and sustained by scientists in their
attempts to account for the actions of others, especially those of their competitors in
situations of controversy.
60. Garfinkel, op. cit. note 4, Chapter 3, esp. 94 ff.
61. D.L. Wieder, Telling the Code, in R. Turner (ed.), Ethnomethodology
(Harmondsworth, Middx.: Penguin, 1974), 144-72.
62. In his recent discussion of Wieders study, Leiter also notes the empirical consequences of reflexivity for the use of rules as causal elements by sociologists: see K.
Leiter, A Primer on Ethnomethodology (Oxford: Oxford University Press, 1980),
esp. 198-200.
63. M.J. Mulkay, Interpretation and the Use of Rules: The Case of the Norms
of Science, in T.F. Gieryn (ed.), Science and Social Structure: A Festschrift for
Robert K. Merton, Transactions of the New York Academy of Sciences, Series II,
Vol. 39 (1980), 111-25.
64. Garfinkel, op. cit. note 4, 7.
65. The term background is used throughout the remainder of the discussion in
place of the more cumbersome relegate to the background.
66. Latour and Woolgar, op. cit. note 28, 174 ff.
67. MacKenzie, Statistical Theory ..., op. cit. note 34, 52 (my emphasis).
68. Of course, this simile is not exact in that it incorrectly implies that
mathematical proofs which do not assume the answer are free from the reflexive
interconnection between explanans and explanandum. My argument here is that
such interconnections are symptomatic of all explanations, whether or not they are
subsequently said to be flawed. My thanks are due to Steven Yearley for raising this

point.
69. This of course reflects the common-sense notion that it is always better to
base an interpretation on several different indicators rather than to rely on one. The
sociologists version of this idea appears in some methods texts as an injunction to
engage in triangulation: see, for example, N.K. Denzin, The Research Act
(Chicago: Aldine, 1970), Chapter 12; H.W. Smith, Strategies of Social Research
(Englewood Cliffs, NJ: Prentice-Hall, 1975), Chapter 12. While advising in-

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

394
to triangulate, these texts are less forthcoming about ways of resolving
differences in interpretations which may result. For a preliminary discussion of
some practical problems associated with a multi-method perspective in the sociology
of science, see M.J. Mulkay, Methodology in the Sociology of Science: Some
Reflections on the Study of Radio Astronomy, Social Science Information, Vol. 13

vestigators

(1974), 107-19.
70. MacKenzie, Statistical Theory ...,

op. cit. note 34, 49.


71. This seems to be a recurrent theme in scientific literature. See for example, my
analysis of a Nobel Laureates account of his discovery: Woolgar, Discovery ... ,
op. cit. note 28.
72. MacKenzie, Statistical Theory ..., op. cit. note 34, 69.
73. Ibid.
74. Ibid., 66.
75. Ibid., 62.
76. B.J. Norton, Karl Pearson and Statistics: The Social Origins of Scientific Innovation, Social Studies of Science, Vol. 8 (1978), 30 (emphasis in original).

Lecturer in the Department of


Sociology at Brunel University and currently Visiting
Professor in the Department of Sociology at McGill
University. His current research includes an extended
investigation of the relationship between
ethnomethodology and the social study of science, as
well as the development of strategies for the participantobservation study of scientific practice. His publications
include (with Bruno Latour) Laboratory Life: The Social
Construction of Scientific Facts (Beverly Hills, Calif.:
Sage Publications, 1979). Authors Address: Department
of Sociology, McGill University, 855 Sherbrooke Street
West, Montreal, PQ, Canada H3A 2T7.

Steve

Woolgar is

Downloaded from sss.sagepub.com at Oxford University Libraries on August 13, 2014

Vous aimerez peut-être aussi