549

*Received July 2002.

†Send requests for reprints to the author, Department of Philosophy, Columbia Uni-
versity, New York, NY 10027; email: psk16@columbia.edu.

Philosophy of Science, 69 (December 2002) pp. 549–559. 0031-8248/2002/6904-0001$10.00
Copyright 2002 by the Philosophy of Science Association. All rights reserved.

The Third Way: Reflections on Helen
Longino’s The Fate of Knowledge*

Philip Kitcher†
Columbia University

Helen Longino’s The Fate of Knowledge offers a positive view of how to
integrate social considerations into epistemology, as well as a diagnosis of
the failures of previous efforts. I suggest that Longino’s positive account
is interesting and fruitful, although her favorite version of pluralism seems
unclear and unmotivated. I also argue that her attempt to distinguish her
own views from earlier ventures in social epistemology suffers from ob-
session with an allegedly pervasive Rational-Social dichotomy. This defect
doesn’t interfere with the insights of her positive proposals.

1. The R*D Menace. A specter is haunting Science Studies, the specter of
the Rational-Social Dichotomy. From Edinburgh and Cardiff to Tucson
and San Diego, there are R*Ds (Rational-Social Dichotomizers) under
plenty of beds, and, even though they protest (Prufrock-style: “that is not
what we meant, that is not what we meant at all”), they must be hauled
out and exposed. Fortunately, philosophers and sociologists may rest easy;
Special Prosecutor Longino is on the case.

It is a pity that Helen Longino’s The Fate of Knowledge is pervaded
by discussions that invite the response of the last paragraph. For at the
center of the book is an insightful presentation of a way in which social
and cognitive aspects of the practice of the sciences might be integrated.
My principal concern in what follows will be to bring this account into



 550

focus, to highlight some of its strengths, and to raise some questions. I’ll
only return to Longino’s obsessive R*D-detection at the end.

2. Some Possible Positions. Let’s start with some common theses about
the aims, achievements and methods of the sciences, and the interrelations
among them. First, a widespread—but not undisputed—claim about truth
as a goal of scientific research:

Veritism: Science aims to accept, and often succeeds in accepting, true
statements.

This is often, but not invariably, combined with:

Correspondentism: The appropriate notion of truth for understanding
Veritism is one that sees truth as correspondence to a mind-indepen-
dent reality.

Further, the joint acceptance of these theses is frequently accompanied by:

Methodism: The acceptance of statements in science is guided by rules
of method.

The package can now be tied together by adding:

Reliabilism: The rules of method are rightly adopted, because follow-
ing them reliably generates and sustains true beliefs.

There are, of course, lots of worries about what the theses mean, why
they are preferable to alternatives, and so forth. For present purposes,
however, it’s useful to note an obvious tension between the first pair and
the second pair: Veritism is concerned with an entity, Science, supposed
to be capable of aims and acceptance; Methodism and Reliabilism are most
naturally understood as applying to people, most obviously to scientists.
On the face of it, some explanation of Veritism is needed, and one natural
elaboration is by way of a three-part commitment to the idea that indi-
vidual scientists have aims and accept statements.

Individualism: (a) Individual scientists aim to accept, and often suc-
ceed in accepting, true statements; (b) Individual scientists aim that
true statements should be accepted by most (if not all) the members
of the community to which they belong; (c) Community acceptance
is to be understood in terms of acceptance by the individuals who
comprise the community.

The principal thrust of (c) is to oppose any idea of an irreducible “social
fact” of community acceptance.

Individualism is not by any means incompatible with the idea that sci-
ence is a social activity. In particular, it can be combined with



    THE FATE OF  551

1. There are many versions of philosophical anti-realism that would question Veritism
or Correspondentism, but that would produce a position isomorphic to the cluster of
theses I’ve conjoined; those versions are equally unsatisfactory from the standpoint of
the sociological critique.

Interactionist Socialism: Individual scientists accept statements as the
result of interactions with others (debate, collaboration, instruction,
and so forth).

Those who adopt both Individualism and Interactionist Socialism are likely
to add an obvious development of Methodism, to wit

Social Methodism: Among the rules of method are rules governing
the interactions among scientists.

And they will see rules of this type as judged by a standard that fuses
Reliabilism with part (b) of Individualism.

There is, however, another obvious possibility. Someone who held In-
teractionist Socialism might also contend that the social interactions in
which scientists are involved (whether they involve only other scientists or
members of the broader society will be left indeterminate) are antithetical
to achieving the aims set forth by Veritism and Correspondentism. This
person would propose

Social Skepticism: The social interactions in which scientists engage
interfere with their acceptance of true statements (when truth is un-
derstood in the correspondence sense).

In more familiar terms, taking the sociology of science seriously spells
trouble for a realist account of scientific knowledge.

One strand in traditional philosophy of science (call it the Realist Pack-
age) combines Veritism, Correspondentism, Methodism, Reliabilism, and
Individualism. A prominent movement in the sociology of science, from
the early 1970s to the present, uses Interactionist Socialism and Social
Skepticism to question the traditional philosophical package(s).1

Yet not only in British politics, but in philosophy of science as well,
recent years have seen attempts to find a “third way.” Moreover, there’s
an obvious way to try: build on to the Realist Package Interactionist So-
cialism and Social Methodism; of course, this will require an explicit re-
joinder to arguments thought to favor Social Skepticism; but this version
of the third way seems to “transcend the Rational-Social Dichotomy.” In
recent years it has been pursued by Alvin Goldman, Susan Haack, and
one of my former selves.

Helen Longino is a self-described follower of the third way. But the
position I’ve just sketched isn’t hers. She does better. The ways in which
she does better have less to do with the acceptance of Interactionist So-



 552

2. I should note that Longino has a thesis she calls “monism,” characterized in various
ways at various places in her book; her formulation on page 67 comes closest to my
usage; I should note, however, that my 1993 book, while in the vicinity of monism, was
a little less crude than the Monism discussed here.

cialism than with her refinement of Veritism and articulation of Social
Methodism. Most fundamental is her recognition of an error that has in-
fected some previous versions of the third way (perhaps most egregiously
in my own 1993). That error is to go beyond Veritism to

Monism: Individual scientists aim to contribute to the attainment of
a single, complete, true account of nature.2

It’s not clear that the idea of a complete true account of nature is coherent,
and any such account is surely not attainable. A better version of the same
general idea is to propose that science aims at the significant truths, and
that significance is a matter of identifying natural kinds, formulating uni-
fying general principles, and so forth. (This is the line I pursued in 1993).
But Longino is right to see that the whole idea is suspect.

Her own position starts with a reaction to Veritism. Longino believes
that scientific knowledge should be understood by focusing on other ve-
hicles of representation than statements. In particular, she is sympathetic
to the idea that visual representations play an important role in the sci-
ences, and to the semantic conception of theories (especially in the version
offered in Giere 1988). Hence, she’d replace Veritism with something
broader, along the lines of

Conformism: Science aims to accept, and often succeeds in accepting,
representations that conform to reality.

The notion of conformity that is at issue here rests on a broadening of the
notion of truth as correspondence. Unlike many philosophers who have
worried about the Realist Package, Longino doesn’t reject Veritism or
conjoin it with some other view of truth (a deflationary account of truth,
or a “don’t ask, don’t tell” approach). Instead she adopts

Extended Correspondentism: Representations conform to reality just
in case they are statements and true in the correspondence sense, or
visual representations and accurate in corresponding to their intended
domain, or models and correspond to the intended domain in appro-
priate respects and to appropriate degrees, or. . . .

Here I’ve left room for the possibility that Longino’s examples of repre-
sentational vehicles don’t exhaust all the types she’d sanction, since she
doesn’t commit herself to the completeness of her list.

Notice that Extended Correspondentism introduces a pragmatic dimen-



    THE FATE OF  553

3. Whether this brings with it the idea that representations that are (in some sense)
correct needn’t be jointly consistent is an issue to which we’ll return below; my own
acceptance of Pluralism emphasizes incompleteness rather than inconsistency.

sion into the account of the desired relation of conformity between rep-
resentations and reality. I have made this explicit with respect to models,
where, following Giere, Longino suggests that models need to correspond
in “appropriate respects and to appropriate degrees”; but the same rela-
tivization to background goals that generate “appropriate” respects and
degrees can easily be allowed in the case of visual representations (we may
think, as Longino does, of maps) and even of statements (a statement
ascribing a value to a parameter counts as true, or approximately true, if
the value ascribed is within an appropriate interval around the actual
value). This allows for the possibility that Monism is far too grandiose as
an account of the aims of science, and what we really want is a collection
of representations adapted to our needs and interests. In fact, Longino
explicitly adopts something like

Pluralism: The desired representations need not fit together into a
single complete true account; rather they may be partial and not nec-
essarily “congruent.”

I’ll want to return to the notion of “congruence” below; for the moment
we can simply understand it as some way of marking out the idea that the
totality of the representations accepted by science, even ideally, need not
fit together into a complete account of nature.3

As I understand it, Longino’s version of the third way proceeds by
adopting Methodism and rewriting Reliabilism to accord with her broader
notion of the representations science aims at (and the broader notion of
correspondence). Given her frequent castigation of other people for pur-
suing social accounts of science in too restricted a way, one might expect
that she would reject Individualism, but so far as I can tell, this is not so.
Longino never attempts to provide a notion of community acceptance that
would be irreducible to acceptance by the individuals who comprise the
community: indeed her response to a problem posed by John Hardwig
concerning the bearer(s) of knowledge in cases where large numbers of
researchers collaborate seems thoroughly in line with Individualism, and
her emphasis throughout is on the importance of interactions among in-
dividual scientists. I therefore interpret her as adopting Interactionist So-
cialism and Social Methodism.

At this point, we come to the last, and to my mind the most interesting,
aspect of Longino’s third way. Given her adoption of Pluralism, Longino
wants to ensure that the collection of representations actually proposed
and accepted in the course of scientific research isn’t restricted through



 554

4. I should note that (as Longino recognizes very clearly), Pluralism has been developed
by a number of people. Patrick Suppes is one important source, and the writings of
John Dupré (1993) and Nancy Cartwright (1999) provide detailed recent elaborations.

the exclusion of some viewpoints from scientific discussion. This leads her
to conclude that Social Methodism has to be articulated in a particular
way, that it must offer an ideal of a scientific community in which different
points of view are expressed and heard. Longino is not so naı̈ve as to
suppose that some condition of simple equality for all voices will do; she
recognizes differences in acumen and expertise—suggesting instead that
the right ideal is one of tempered equality (although, as she admits, she
only gestures towards spelling this out). She is thus committed to:

Democratism: The rules of method have to articulate the ideal of an
appropriate democratic community.

Let’s now take stock. We have the following array of positions. First
the Realist Package, and variants on it that don’t take any social turn
(although they may diverge from Veritism, Correspondentism, Reliabilism,
and so forth); this cluster of views is the stock of traditional philosophy
of science. Second, come the most obvious views in recent sociology of
science, views that conjoin Interactionist Socialism with Social Skepticism
and proceed to draw negative conclusions for traditional philosophy of
science. Third are earlier versions of the third way, such as those that add
Interactionist Socialism and Social Methodism to the Realist Package, at-
tempt to rebut Social Skepticism, but also commit themselves to Monism.
These positions differ from Longino’s preferred alternative not in failing
to transcend some Rational-Social Dichotomy but in not adopting her
broader approach to representations and the Pluralism that flows from it.
For Longino’s version of the third way consists of: Conformism, Extended
Correspondentism, Methodism, Reliabilism and Individualism (so far an
amended Realist Package), Interactionist Socialism and Social Methodism
(theses which, like the denial of Social Skepticism she shares with other
followers of the third way), and, distinctively, Pluralism and Democratism.4

The options I’ve listed are not the only possibilities (indeed, one of my
purposes in this section was to invite philosophers to consider their own
preferred deviations from the positions on which I’ve primarily focused).
But Longino’s seems to me preferable to the alternatives, and, in partic-
ular, to those third way accounts that commit themselves to Monism. Un-
fortunately, I don’t think Longino is entirely clear about why her account
is superior.

3. The Road to Democracy. Let me start with a small point. I’m not sure
how much hangs on replacing Veritism and Correspondentism with Con-



    THE FATE OF  555

formism and Extended Correspondentism. For instead of saying that one
of the aims of science is to adopt models that fit the world in appropriate
respects to appropriate degrees, we might suggest that among the state-
ments science aims to accept is a class describing this kind of fit. That
would still allow for the pragmatism that generates the attractive features
of Longino’s position: science would be seen as aiming at significant truth,
where significance is determined by us and our evolving interests (see
Kitcher 2001, chapters 4–6). Of course, there would still be the challenge
of assimilating the accuracy of visual representations to some notion (or
close analogue) of truth, but it may well be possible to meet the challenge
(for a preliminary attempt, see Kitcher and Varzi 2000).

Longino seems quite clear that not all truths, or visual representations,
or models, are significant (176), and one way to recognize the contrast
between Monism and Pluralism is to suppose that monists view signifi-
cance as an objective, context-independent, matter, while pluralists sup-
pose that what is significant is relative to our cognitive capacities and our
(changing) interests. The analogy with maps, which both Longino and I
like, is useful here. Good maps are held to an objective standard (they
have to conform to the domain mapped) but the character of that standard
(what is omitted, what included, the degree of accuracy required, and so
forth) is set by us and our concerns.

But now we must raise a more serious issue. In advocating Pluralism,
Longino goes beyond the modest view that the bits of nature we try to
represent accurately are a function of us, our capacities and our interests,
to the suggestion that the representations we achieve may not be “con-
gruent” (140), may resist reconciliation (93), or may be “nonreconcilable”
(207). Her discussion of these notions of congruence and reconciliation is
extremely murky. There’s no difficulty in understanding the idea that mod-
els that focus on different aspects of a system may both conform to it or
that maps constructed according to different projections may both be ac-
curate (to specified degrees). No inconsistency threatens in claiming both
that the world is like this map in these respects to these degrees and like
the other one in those respects to those degrees. But inconsistency is
evaded here precisely because the claims are partial: the point is as boring
as the lack of difficulty in identifying Helen Longino as a member of both
Minnesota’s Philosophy Department and of its Center for Advanced
Feminist Studies. Now, of course, good maps may be literally noncon-
gruent, but that’s no problem unless we think that the world has to be like
them in all respects. On a weak interpretation of Pluralism, “noncon-
gruent” is just further emphasis for the partiality of representations.

On the weak interpretation, Pluralism is unexceptionable, although still
strong enough to lead us to Democratism (as we’ll see shortly). But Lon-
gino clearly wants more. Although she seems to be a fan of the Principle



 556

of Non-Contradiction (as I think she should be), she wants to allow for
the acceptability of representations that are “hard to reconcile”—presum-
ably in ways that the adequacy of different maps or the different academic
identifications of Helen Longino are not. Utter mystery descends at this
point. The closest she comes to explaining what is intended occurs in a
response to a counterargument that urges the consistency of all true state-
ments:

The demand for consistency of all true statements is only problematic
if one supposes that statements can be detached from their truth con-
ditions and the contexts in which those are determinable. A contex-
tualist denies that such detaching is possible without constructing a
further or more encompassing context. (94)

There’s a commonplace point here—one supported by an obvious reading
of the following sentences that discuss measurements in the context of
different branches of physics—namely that what appear to be incompat-
ible statements need not actually be so because of the ways in which the
reference of the constituent terms is determined. This should be very stale
news, since the point has been exploited for about three decades in dis-
cussions of the alleged phenomenon of incommensurability (to which Lon-
gino refers in a footnote). Sentences that look incompatible may both be
true because the extensions of tokens of the same type are different. But
there’s no difficulty of reconciliation here. Longino seems to be gesturing
towards something more interesting, but what it is—what else lies behind
her talk of “detaching” statements—is quite obscure.

I think we should only accept the modest version of Pluralism, on which
“noncongruent” is redundant. But this will still provide a route to De-
mocratism. For, if the representations at which we aim are those that an-
swer to changing human interests, it’s important that there should not be
groups whose voices are never heard. In fact, there are two independent
arguments for Democratism, one focusing on the search for significant
truth (or conformity) the other on the search for significant truth (or con-
formity). The latter is Millian in spirit—it arises from the thought that
correct ideas emerge most readily from the clash of opposing points of
view. The former emerges from the rejection of Monism; once the idea of
a “complete true account of nature”—or any other context-independent
specification of the scientifically significant truths—is abandoned, the rep-
resentations at which the sciences aim are those that answer to human
needs (some theoretical, some practical), and it’s important that the needs
of all groups should be included.

Longino’s articulation of Democratism is quite short and very sketchy.
As I’ve already noted, she appreciates the point that vulgar democracy
would be a disaster—there’s a serious problem in letting the uninformed



    THE FATE OF  557

5. Effectively, the female primatologists expanded the set of hypotheses considered,
revealing unexpected complexities in primate social interactions. They did so because
different kinds of events were salient for them.

vote—but it’s not apparent how she proposes to overcome the problem.
In fact, I think the trouble is deeper. Longino has failed to distinguish two
different kinds of problematic situations for which a commitment to De-
mocratism might be the remedy. The Millian Problem arises when the
choice of alternative hypotheses is restricted because of the exclusion of
some group of people from scientific deliberation, so that the hypothesis
that would conform to nature is left out (the idea being that that hypoth-
esis is one that would likely have occurred to the outsiders, but isn’t at all
salient for those who actually engage in the discussion). The Interest Prob-
lem arises when the hypotheses accepted conform to nature in a way that
suits the concerns only of a subgroup of the species (or even of the society
in which the research is done).

I think Democratism is a good way of responding to both of these
problems, but you need to be clear which one you are trying to tackle,
because the ways of articulating the democratic ideal will differ. Consider,
first, a striking example of the Millian problem—prior to the 1970s, when
women began to enter the field in significant numbers, studies of “domi-
nance” in primates were narrowly attuned to the aggressive interactions
among males; female primatologists revolutionized the field by offering
broader perspectives—they showed, for example, how apparently subor-
dinate males were able to achieve significant reproductive success by
“making friends” with females.5 For the Millian problem, the appropriate
way to develop Democratism is to insist on full representation of relevantly
different points of view within the community of researchers (of course,
it’s not easy to identify the notion of relevance here). An obvious example
of the Interest Problem is the current skewing of biomedical research to-
wards diseases that afflict citizens of affluent countries and the relative
neglect of infectious diseases that cause vast amounts of death and suf-
fering. In this instance, the appropriate response is to ensure that the in-
stitutions that allocate resources to various lines of inquiry can no longer
ignore the plight of distant people.

Longino is right to maintain Democratism, but I think she lacks a clear
view of exactly what greater democracy is to achieve. She fails to distin-
guish the Millian problem from the Interest problem. The root of the
trouble here is, I think, her extremely elusive discussion of underdeter-
mination. On the one hand, she seems to think that traditional claims
about underdetermination are “fanciful” (127, quoting Ian Hacking); on
the other, underdetermination is supposed to be rampant, but whether it
is the result of equally good evidence for rival views or divergence in in-



 558

terest is never made clear. The obscurity of her commitments may have
prevented her from saying more about the details of her ideal epistemic
community.

In fact, the latter third of the book is somewhat anti-climactic. After
her brief account of social norms, Longino concedes that she won’t fully
address the important issues she has raised (134). Instead, she concludes
with an attempt to situate her views with respect to positions she regards
as taking the social aspects of knowledge seriously, and with a review of
pluralist approaches in biology (and the history and philosophy of biol-
ogy), whose point is not entirely evident. At the very end, she rightly
remarks that her approach raises new issues for the philosophy of science,
and her penultimate sentence poses the crucial question: “What kinds of
institutional changes are necessary to sustain the credibility, and hence
value, of scientific inquiry while maintaining democratic decision making
regarding the cognitive and practical choices the sciences make possible
and necessary?” (213). That question could easily have been posed some
eighty pages earlier, and I wish that Longino had spent some of the in-
tervening space coming to terms with it.

4. The Obsession with R*Ds. In general, the principal deficiency of The
Fate of Knowledge lies in the space devoted to a superficial engagement
with the views of other authors. This is nowhere more evident than in
Longino’s pursuit of the evil Rational-Social Dichotomy, a pursuit that
often distorts her exposition of her own position. That dichotomy is em-
braced by people who regard social and cognitive explanations as intrin-
sically opposed. Hence it is transcended by anyone who adopts Interac-
tionist Socialism and Social Methodism.

Oddly, however, Longino spends a great deal of time attacking soci-
ologists and philosophers who allegedly commit the dichotomy. I’ll leave
the sociologists to defend themselves (all except Bruno Latour who, since
he has campaigned against the dichotomy in print, is a Good Guy; this
apparently means that we should offer “fruitful” interpretations of his
words—“wildly creative” seems to me the more accurate term). But Lon-
gino argues at some length that Alvin Goldman, Susan Haack, and I are
all guilty, and this leads her to some misreadings of our work. So Longino
overlooks our explicit commitment to Interactionist Socialism and Social
Methodism. How did this happen?

Longino thinks that the philosophers she chastises have a tendency to
“equate social forces with ‘bias’ ” (56) because she overlooks the fact that
we spend a fair bit of time trying to come to terms with Social Skepticism;
and, of course, when one is doing that, one wants to address the question
of whether social forces are inevitably biasing; to address the question is
very different from accepting the thesis; indeed, it’s a large part of the



    THE FATE OF  559

burden of Goldman’s (1999) and my (1993) that social forces aren’t in-
evitably contrary to the rationality and progressiveness of the sciences.

Virtually everything that Longino says about the philosophical work
she criticizes is infected by this odd view, but the nadir is probably her
discussion of the attempts to model scientific communities that Goldman
and I have (independently) made. Longino seems to think that we’re
wedded to the acceptance of neoclassical economics, that we have simple-
minded views of cognitive agents, and a “crass interpretation” of the
sociologists’ notion of credit. To set the record straight: we use a mathe-
matical formalism (one that economists can deploy to different ends), the
utility functions can be specified in a wide variety of ways (and in terms
of preferences that are as rich and wonderful as Longino would like them
to be), and the concept of credit can be introduced in as subtle as fashion
as she’d like; the point of the mathematics is to make unobvious possibil-
ities clear. I suspect that Longino’s preferred approach to the social studies
of science is Geertzian rather than Galilean, but it is a wonderful irony
that someone who is so happy talking about models doesn’t understand
that the models Goldman and I develop aren’t intended as literal descrip-
tions of people: they are supposed to capture interesting facts about the
social organization of science in various respects and to various degrees.

Disentangling Longino’s misinterpretations of my work and that of
others would take another essay far longer than this one. But let me close
on a positive note. Despite her obsession with rooting out R*Ds, Longino
has a valuable and interesting perspective to advance. She is quite correct
in chiding Goldman, Haack, and me for our monist commitments, and
she offers a version of the third way that is worth further discussion and
elaboration.

REFERENCES

Cartwright, Nancy (1999), The Dappled World. Cambridge: Cambridge University Press.
Dupré, John (1993), The Disorder of Things. Cambridge, MA: Harvard University Press.
Giere, Ronald (1988), Explaining Science: A Cognitive Approach. Chicago: University of

Chicago Press.
Goldman, Alvin (1999), Knowledge in a Social World. New York: Oxford University Press.
Kitcher, Philip (1993), The Advancement of Science. New York: Oxford University Press.
———(2001), Science, Truth, and Democracy. New York: Oxford University Press.
Kitcher, Philip, and Varzi, Achille (2000), “Some Pictures are Worth 2 a0 Sentences”, Philos-

ophy 75: 377–381.
Longino, Helen (2002), The Fate of Knowledge. Princeton: Princeton University Press.