Goldman1987 Foundations of Social Epistemics

ALVIN I.
GOLDMAN
FOUNDATIONS OF SOCIAL EPISTEMICS*
ABSTRACT. A conception of social epistemology is articulated with links to studies of

science and opinion in such disciplines as history, sociology, and political science. The
conception is evaluative, though, rather than purely descriptive. Three types of evalua-
tive approaches are examined but rejected: relativism, consensualism, and expertism. A
fourth, truth-linked, approach to intellectual evaluation is then advocated: social
procedures should be appraised by their propensity to foster true belief. Standards of
evaluation in social epistemics would be much the same as those in individual epis-
temics, only the objects of evaluation would be interpersonal patterns of judgment and
communication, and institutional practices that bear on opinion formation.
1. INTRODUCTION
Epistemology has historically been preoccupied with individual

knowers and their minds. This preoccupation has a plausible rationale.
Knowers are individuals, and knowledge is generated by mental
processes and lodged in the mind-brain. Thus, despite isolated
attempts to dispense with the knowing subject (e.g., Popper 1972), it is
entirely fitting for epistemology to be concerned with individual
knowers and their minds.
But concentration on the individual to the exclusion of the social
is inappropriate. T h e bulk of an adult's world-view is deeply indebted
to her social world. It can largely be traced to social interactions, to
influences exerted by other knowers, primarily through the vehicle of
language. It is imperative, then, for epistemology to have a social
dimension. How this social dimension should be conceived and struc-
tured, however, is far from apparent. No well-entrenched epistemolo-
gical tradition provides such a structure. This paper lays the foun-
dation for one such structure, for one conception of social epis-
temology.
In a recent book, Epistemology and Cognition (Goldman, 1986), I
develop a conception of epistemology as a multidisciplinary subject,
orchestrated and directed by philosophy but requiring contributions
from other fields as well. I call this multidisciplinary conception
epistemics. Epistemology and Cognition focuses on one branch of
Synthese 73 (1987) 109-144
© 1987 by D. Reidel Publishing Company
110 ALVIN I. GOLDMAN
epistemics: primary individual epistemics. That branch centers on

basic processes of the mind-brain, and is therefore allied with the
cognitive sciences. In this paper I elaborate the conception of the
social branch of epistemics and say a few words about its disciplinary
alliances.
2. THE EVALUATIVE TASK OF SOCIAL EPISTEMICS
To begin thinking about social epistemology, consider the profile of

belief-states of a community, group, or culture. Under 'belief-states' I
mean to include varying degrees of confidence as well as full-fledged
'acceptance'. For simplicity, though, I concentrate on states of binary,
all-or-none belief. One possible conception of social epistemology is
purely descriptive and explanatory. It might describe belief-profiles of
different groups and seek explanations of them in terms of social
interchange. A diachronic perspective is especially inviting: a group's
belief-profile at time t + 1 might be explained by its profile at time t
together with communication transactions during the interval. Many
disciplines and subdisciplines are currently engaged in studying belief-
profiles from approximately this perspective. History, sociology, anth-
ropology, political science, social psychology, and so on explore
science, scholarship, political opinion, and popular culture with a view
to understanding ideational change.
My conception of social epistemology has some affinities with these
disciplines, but the epistemological aim is not coterminous with theirs.
These disciplines have strictly descriptive and explanatory goals, while
the central aim of epistemology is normative, evaluative, or critical. It
does not merely aim to say how and why belief-profiles change or
evolve, but to appraise such changes along some epistemologically
relevant dimensions. Mechanisms of intellectual influence interest the
social epistemologist, but not for their own sake. The epistemologist
wishes to investigate epistemologically relevant properties of the
mechanisms. What those epistemic properties are remains to be
specified.
Despite this difference of emphasis, I do favor a wide scope for
social epistemology. On my view all social influences on belief are of
potential interest to social epistemology, whatever the contents of the
beliefs. Such breadth of scope might be viewed with suspicion in some
quarters. Isn't this approach all too congenial to the so-called 'strong
FOUNDATIONS OF SOCIAL EPISTEMICS 111
programme' in the sociology of science (Bloor, 1976; Barnes, 1977)?

According to the strong programme, social factors are explanatorily
relevant to all scientific activity. Critics of this program reply that a
sociological explanation of scientific belief is in order only when the
target scientist contravenes the norms of scientific rationality. This is
made explicit by Laudan (1977, p. 202) in what he calls the ara-
tionality assumption: "the sociology of knowledge may step in to
explain beliefs if and only if those beliefs cannot be explained in terms
of their rational merits". The arationality assumption establishes a
division of labor between the historian of ideas and the sociologist of
knowledge: the latter may step in only when there is some deviation
from the norm of rationality. In Newton-Smith's (1981) formulation,
sociology is only for deviants.
I take exception to this arationality assumption. There is no reason
why one and the same belief, scientific or otherwise, cannot be
explained both in rational and in social terms. This leaves very wide
scope, in principle, for social explanations of belief. I agree with
Laudan that it is very unlikely that there are laws governing belief that
feature exclusively sociological variables in their antecedents. Social
or sociological factors could not provide a 'complete' explanation of
belief. But there is no reason why social factors cannot influence
belief-formation in conjunction with other, indisputably rational, fac-
tors. Perhaps their influence may be exerted via the mediation of
psychological events that figure in a rational explanation (for example,
in what Newton-Smith (1981), calls a 'minirat' explanation). Some-
thing like this is granted by Laudan too in a more recent paper
(Laudan 1984). He acknowledges the possibility of a 'sociology of the
rational', that would explain, for example, why different conceptions
of rationality evolve in different cultures. I would go further, however,
and point out that rationality might partly consist in certain forms of
social interchange. Some styles of debate and mutual criticism, some
modes of doxastic response to the arguments of others, may be partly
constititutive of rationality. So there is no tension between social and
rational belief-causation.
Let me cite some commonplace types of social explanations that are
compatible with rational explanations of belief. First, suppose a belief
is formed rationally as a function of several competing hypotheses
and the available evidence. The hypotheses surveyed, however, were
generated by a number of different scientists, each working in his own
112 ALVIN I. G O L D M A N
research tradition, partly spurred and influenced by the competition of

rival research traditions. In this case, disciplinary fragmentation - a
certain social fact - is part of the origin of the competing hypotheses
that the working scientist considers. Second, a partial cause of the
scientist's entry into the discipline may be the financial or reputational
incentives of organized science. This reward structure, of course, does
not by itself explain the scientific content of an explanandum belief.
But it is one of many operative causes that ultimately lead to the
belief. Third, the rational formation of a belief involves the application
of a suitable methodology, which the target scientist employs. Whence
this methodology? The scientist may have learned it from his mentors,
who in turn acquired it from other members of the scientific com-
munity, who (perhaps rationally) enforce that methodology as a con-
dition for respect and acceptance within that intellectual matrix (cf.
Whitley 1984). In all these cases, the many causes of the belief include
some social factors; but these in no way undermine a rational
explanation of the belief.
My concurrence with the Edinburgh school on the wide scope for
social factors is potentially misleading, for its adherents often seem to
have a more specific agenda with which I disagree. They seem to think
that scientific beliefs can generally be explained by certain kinds of
social factors, ones that would normally be counted as external to the
proper conduct of science, such as ideological or political interests, or
influences of the general intellectual milieu. Here I part company with
them. In the case studies to which strong programmers often appeal, a
central issue is whether the predominant influences on scientific belief
have been 'external' or 'internal' to the discipline. (See Forman
(1971), and Shapin (1975), for examples.) I generally side here with
the 'internalist' historical accounts rather than the 'externalist' ones.
(See Hendry (1980), and Cantor (1975) for rebuttals of Forman and
Shapin, respectively.) This does not mean that I oppose social
explanations in these cases. Internal explanations can be social
explanations; intradisciplinary influences can be interpersonal
influences. It is a mistake to assimilate the external/internal distinction
to the social/asocial distinction. (It should be acknowledged that this
point is sometimes expressly stated by strong programmers. See Bloor
(1973), pp. 190-91, Bloor 1976, p. 4, and Shapin 1982, p. 197.
However, they seem to have a marked preference for external
explanations.)
Notice that nothing in my epistemological position forces me to

take one or another side of the internal/external dispute. My view of
the historical facts is, therefore, incidental. In fact, it would not
conflict with my epistemological stance even if there were no interes-
ting social explanations of many beliefs. All I wish to defend is the
potential scope of social factors. Details of particular historical events
have no implications for this position.
Other matters of disagreement, however, are more critical to the
theoretical issues of this paper. For example, I think it is important to
deny the strong programme's contention that explanations of a belief
cannot refer to its truth, or the state of affairs that makes it true (on
this point, see Newton-Smith (1981), pp. 252-53). Even more
significantly, I strongly disagree with the Edinburgh school's endor-
sement of epistemic relativism. This topic will be treated at length in
later sections.
I have defended the role of social factors in belief explanation. But
is this critical for the epistemologist, who is concerned primarily with
normative assessment? Shouldn't the epistemologist restrict attention to
questions of scientific methodology, to canons of logic and evidence?
While the epistemologist may acknowledge the social roots of belief,
even scientific belief, why is this acknowledgement epistemologically
significant? How can it affect matters of epistemic evaluation or
appraisal?
Although this question is addressed at length later, some preli-
minary answers may be given here. Our previous examples provide
hints of evaluative subject-matter. Consider the existence of rival
research traditions, of ongoing competition, in a discipline. The epis-
temologist may be interested in evaluating the merits of such com-
petition. The aim here is not to evaluate any specific belief, as rational
or irrational, for example. (Rationality is not the only possible dimen-
sion of epistemic appraisal.) The epistemologist may be interested in
the epistemic consequences of intellectual rivalry. Is such rivalry
epistemically beneficial, and if so, how and why? A similar point holds
for the social and economic reward structures that characterize
modern science and other fact-finding institutions. The existence and
nature of such structures are stressed and explored by sociologists
(e.g., Merton, 1970). What interests the epistemologist are the epis-
temic consequences of such structures.
It should also be stressed that social epistemology is not restricted
to the epistemology of science or scientists. It includes the epis-

temology of the common person. Far from confining attention to
belief-profiles of specialists, it can be concerned with belief-profiles of
entire cultures. Lay people often form beliefs about scientific subjects,
and on policy issues closely linked to scientific matters. But lay people
have neither the time, the interest, nor the training to utilize technical
methodologies that specialists employ in forming beliefs on these
matters. Rather, their belief-formation relies heavily on judgments of
authority. The epistemological question here is: what are the proper
procedures for appeal to authorities? How should community
decisions reflect expert opinion? Here too we have normative ques-
tions about social or inter-personal structures that fall in the bailiwick
of epistemology.
3. RELATIVISM, CONSENSUALISM, AND EXPERTISM
In Section 2 I argued for an evaluative approach to social epis-

temology. But what, exactly, are the objects to be evaluated? And
what is the basis, criterion, or standard by which evaluations should be
made? In this section I consider a variety of possible targets and bases
for evaluation and briefly indicate my preference. My own choice is
explored more fully in Sections 4 and 5.
There are at least three possible targets of evaluation for social
epistemology: (A) beliefs of individuals, (B) social belief-profiles, and
(C) social practices, procedures, and institutions. The candidacy of (B)
and of (C) needs no explanation, but what about (A)? How could an
individual's belief be an object of evaluation for social epistemology?
Quite simply, a belief might be evaluated by reference to some
properties of the community to which the individual belongs. Has the
belief been generated by methods of which the community approves?
Does it agree in content with other people's beliefs? Samples of these
approaches will be examined, as well as approaches invoking (B) and
(C) as evaluative targets.
Let me turn next to alternative bases of evaluation for social
epistemology. One possible basis, as just indicated, is the set of
methods endorsed by the community or group. A second possible
basis is consensus, either the existence or promotion of consensus, or
agreement with consensual opinions. This second, consensualist, basis
bears a similarity to the first basis, except that consensualism concerns
F O U N D A T I O N S OF SOCIAL E P I S T E M I C S 115
agreement in belief content whereas the first basis concerns agreement

in method. The third type of basis is expertism. It would evaluate
objects either in terms of promotion of expertise or in terms of
agreement with expert opinion. The fourth basis I call veritism. It
would make evaluations by reference to the production of true belief.
Combining the three possible objects of evaluation with the four
possible bases for evaluation, a 3 × 4 matrix of possible positions can
be generated, as shown in Figure 1. However, I construct no in-
telligible portraits of social epistemology that correspond to the
OBJECTS OF EVALUATION
Beliefs Social Social Practices or

of Individuals Belief Profiles Institutions
Methods
Accepted by Relativism
One's Group
Z
0 Group
Consensualisml Consensualismz Consensualisms
Consensus
>
0
m Expert
Expertisml Expertism2 Expertism3
< Opinion
True Belief
ProductiOn Veritism
Figure 1.
middle and right squares of row 1 nor to the left and middle squares of
row 4. This leaves us with eight possible positions.
In addition to objects and bases (or standards) of evaluation, an
important question for any evaluative subject is its terms of evalua-
tion. In epistemology, several terms are widely in use, including
'rational' and 'warranted' (or 'justified'). Of the eight possible positions
that have been generated, some are more naturally articulated with
one evaluative term and others with another. I shall comment on the
most plausible evaluative term for each of the eight positions.
Let me now explore these eight positions in some detail. The upper
left square represents a position I call relativism. (Another possible
label is parochialism.) Relativism consists of three theses: (A) There
are no universal, context-free, super-cultural, or transhistorical stan-
dards by which to judge different belief-forming methods. (This thesis
might be called nihilism, or anti-universalism.) (B) Whatever methods
a group accepts are right for them. (C) An individual's belief is socially
warranted (or rational) if and only if this belief is formed (or sustained)
by methods that are accepted by the individual's group.
I am not sure this position is fully endorsed by any identifiable
theorist. Some theorists, such as Barnes and Bloor (1982) and Rorty
(1979), clearly endorse thesis (A), the nihilist or anti-universalist
strand of this position. They also appear to endorse (B). But it is
debatable whether they endorse (C). Rorty, however, favors an ac-
count of rationality and epistemic authority in terms of "what society
lets us say"; he identifies "the community as source of epistemic
authority" (Rorty 1979, pp. 174, 188). So perhaps he accepts the
entire version of relativism I am describing. There are also affinities
with Kuhn (1970), who sees the scientific community's current
paradigm as the only available instrument by which to appraise a
member scientist's beliefs.
One difficulty with relativism is that a believer may belong to
several communities, which have conflicting sets of approved
methods. A scientist, for example, may belong to (a) a general culture,
(b) a certain scientific discipline, and (c) a faction within that discipline
which champions a distinctive methodology. A given belief of his may
conform with his own faction's tenets, but violate methodological
strictures accepted by most members of the discipline. How is the
belief to be evaluated?
The relativist may retreat to the view that the belief is warranted
relative to one community and unwarranted relative to another. But

this adds a further dimension of relativization not in the initial for-
mulation. Suppose we consider this kind of relativism, however, as a
candidate framework for social epistemology. How interesting and
significant a framework would it be?
It strikes me as quite uninteresting and impoverished. What would
the tasks of social epistemology consist in, on this approach? They
would consist in (i) identifying a given community's methods, and (ii)
determining whether any specified individual has complied with or
violated those methods. This combination of tasks is suited, perhaps,
to the historian, the sociologist, or the anthropologist. But it has little
resemblance to any traditional philosophical mission of epistemology.
The relativist may argue that, nonetheless, this is the only defensible
evaluative mission. What grounds are there for such a claim? One
proffered ground might be descriptive epistemic relativism, the doc-
trine that different groups and communities, especially at different
historical periods, have in fact endorsed different intellectual
methodologies. This descriptive thesis is, of course, uncontroversial.
But the present thesis of evaluative relativism hardly follows from it.
Mere historical diversity does not entail that the only proper standard
of evaluation for a believer's performance is the methods of his or her
group.
A second possible rationale for theses (B) and (C) of evaluative
relativism is its thesis (A), nihilism or anti-universalism. According to
nihilism there are no other, super-cultural, standards for evaluating a
believer's performance (or any other potential object of social-epis-
temological evaluation). So the only remaining sensible benchmark
are the methods of the believer's own group. (Surely it makes no sense
to evaluate a believer's performance by the methods of some other
group, especially a culturally or temporally inaccessible group). The
plausibility of evaluative relativism thus hinges, in large measure, on
the defensibility of epistemic nihilism. But I shall argue that nihilism is
mistaken. There are universal, context-free, standards of evaluation.
Let me turn now to the consensualist approach to social epis-
temology, represented by row 2 of our matrix. Clear-cut endorsements
of consensualism are hard to find, though many discussions hint at it.
In the logical positivist literature, and elsewhere, science is often
touted as exemplary because of its intersubjectivity, where this is often
understood as the ability to settle disputes among subjects. Settling a
dispute, in turn, is the generation of agreement or consensus among

disputants. One unambiguous statement of this position is by John
Ziman: "The goal of science.., is to achieve the maximum degree of
consensuality" (Ziman 1978, p. 6).
Three different versions of consensualism correspond to the three
entries in row 2, which differ in their evaluative objects. The column 1
version focuses on individual beliefs. It says that an individual belief is
warranted if and only if it accords with the consensus in the believer's
community. The column 2 version focuses on social belief-profiles as
objects of evaluation. A social belief-profile is deemed rational if and
only if it has a high degree of consensus. The column 3 variant takes
social practices and institutions as evaluative targets. Consensualism
in this form maintains that practices and institutions are rational to the
extent that they produce consensus.
I have mentioned one possible attraction of consensualism" its
connection with dispute settling. One other possible attraction is its
resemblance to coherentism, which might make it appealing to those
who find coherentism congenial. To develop this theme, a theorist
might begin by distinguishing two species of warrant (both applicable
to individual beliefs): individual warrant and social warrant. Each kind
of warrant could then receive a coherentist treatment. Intrapersonal
coherence could be advanced as a theory of individual warrant while
interpersonal coherence - agreement with one's peers - would provide
a theory of social warrant. (However, these types of coherence are not
really the same. Interpersonal coherence is shared belief in the same
propositions, whereas intrapersonal coherence is belief in a set of
different but appropriately related propositions, e.g., consistent pro-
positions, mutually supporting propositions, or the like.) On an alter-
native line of development, coherence could be advanced as a theory
of rationality for individual beliefs, while consensus is endorsed as a
theory of rationality for social belief-profiles.
Whatever the initial force of these attractions, consensualism in any
form is plagued with problems. Let us start with consensualisml which
takes individual beliefs as the target of evaluation. Let us assume that
the term of evaluation, for which consensualism is the candidate
standard, is 'socially warranted'. (The idea here is that a belief might
be socially warranted but not individually warranted, or conversely.
Whether these two types of warrant are supposed to be commensur-
able, and if so how, is moot.) The first problem here is the previously
encountered one of how to identify the relevant community. A cog-

nizer typically belongs to many communities. Belief P may match the
consensus in one community but conflict with the preponderance of
opinion in another. Which community fixes the belief's social warrant?
Second, assuming the first problem can be resolved, suppose that
the cognizer does not believe that P matches the community's view-
point, although in fact it does. Is his belief in P still socially war-
ranted? Or, thirdly, suppose he thinks his belief matches the consensus
but it does not. Fourth, suppose that the cognizer, S, (correctly)
believes that members of his community generally accept P, but he
has excellent reasons for thinking they have been duped. Is S still
socially warranted in believing P?
As the second and third problems indicate, it is questionable
whether actual facts of social consensus determine something called
'social warrant'. It seems, instead, as if a person's warrant is a function
of his beliefs. Actual social consensus or dissensus cannot confer
warrant status if the cognizer is unaware of that consensus. Moreover,
even if the cognizer is aware of consensus, as the fourth problem
indicates, he may have reasons that override it.
These problems raise doubts whether there is even a clear pre-
analytic notion of social warrant. Notice that there is relatively little
use of such a phrase in common currency. It may be misguided to
suppose that there is such a prior notion for social epistemology to
explicate. I shall not invoke such a notion in erecting my own
conception of social epistemics.
A final problem for consensualisml is that an individual cognizer
might be aware of a general consensus on proposition P but also
believe there is a small but authoritative minority who all accept
not-P. It is doubtful that such a cognizer is socially warranted in
believing P.
This last problem recurs for consensualism2, which takes social
belief-profiles as the objects of evaluation. This version is most
naturally formulated as a purported explication of the term 'rational-
ity' rather than 'warrant'. But is a high degree of consensus really
sufficient for the rationality of a belief-profile? Suppose again that a
tiny minority with great expertise all deny P's truth. The great
majority, while aware of this dissent, persist in believing P. Is the
resulting belief-profile nonetheless rational? According to consen-
sualism, it is; but that is counterintuitive.
Another difficulty for consensualism2 is that consensus per se is not

a reliable sign of rationality. It depends on how consensus is reached.
All sorts of methods can yield consensus: brain-washing, the threat of
the rack or burning at the stake, totalitarian control of the sources of
information. Consensus reached by these means does not guarantee
rationality.
This objection weighs equally against consensualism3, which applies
the approach to social procedures or institutions. The power of
procedures or institutions to produce consensus is hardly a sure
indication of rationality. Brain-washing, the tools of the Inquisition,
and official censorship may have the power to yield consensus, but
they are not paradigms of rational procedure. Taking all these objec-
tions together, they are weighty grounds against consensualism as an
approach to evaluative social epistemology.
Since some of the most serious objections to consensualism rest on
its neglect of expert authority, it is natural to turn next to expertism.
Like consensualism, expertism can appear in three guises, depending
on its target of evaluation. Expertisml calls an individual's belief
socially warranted if and only if it accords with expert opinion in that
individual's community. Expertism2 calls a group's belief-profile soci-
ally rational just in case the whole profile reflects the opinions of its
constituent experts. Expertism3 rates a social procedure or institution
rational to the degree that it promotes the general acceptance of
expert opinion.
Two features of expertism should be mentioned as possible sources
of appeal. First, in relying on the opinions of others to form one's own
beliefs, it is obvious that some people should be relied upon more than
others. They may have observed the event in question, or have
relevant technical knowledge. A rational cognizer should take these
factors into account. Second, there is an appealing analogy between
expertism and traditional foundationalism. Foundationalism is a theory
of individual warrant or rationality that stresses the authoritativeness
of certain beliefs, so-called 'basic' beliefs. Expertism could be viewed
as an analogue of foundationalism at the social level, in which the
opinions of experts play the role of socially basic beliefs. From a social
point of view, the beliefs of experts are 'immediately' warranted.
Novices' beliefs can inherit social warrant by relying on experts'
opinions. (The significance of intellectual authority has recently been
stressed by Hardwig 1985, and Welbourne 1981.)
While expertism is more promising than consensualism, it has

several problems and deficiencies. First, there are questions that
parallel those facing consensualism. Is a cognizer's social warrant a
function of experts' actual opinions, or what the cognizer thinks about
their opinions? If the experts actually believe P but the cognizer is
unaware that they do, is he still socially warranted in believing P? If
the cognizer thinks they believe P but actually they do not, is he
socially warranted in believing P? The only natural answer, I think, is
to make the cognizer's social warrant a function of what he believes.
But then one wonders whether this is a genuinely distinctive kind of
warrant, different from ordinary 'individual' warrant. This point can
be pressed by noticing that an expert can be viewed as simply a special
case of a natural sign, or indicator, of the truth-value of P. If a
cognizer believes the putative expert's opinion to be a reliable in-
dicator of P's truth-value, he should take that into account in framing
his own belief. But this is just a special case of reasoning from
indicators. It is not clear that any special kind of warrant - social
warrant - emerges from such cases. This bolsters my previously
expressed doubts about whether there is any separate notion of social
warrant in need of analysis.
The foregoing comments apply to expertisml. What about exper-
tism2? The plausibility of expertism2 partly depends on the inter-
pretation of 'expert'. Does 'expert' mean a person whose opinion (on a
given topic) really is the best available guide to the truth? Or does it
only mean a person who has a reputation for being a reliable truth-
guide? If we take the reputational interpretation, it is not clear that the
rationality of a belief-profile is guaranteed by its reflecting expert
opinion. What if the so-called experts do not deserve their repu-
tations? Then it is not obvious that the community is being fully
rational in agreeing with them. The theory has more plausibility if an
expert is understood in the first sense, as someone who is genuinely a
reliable guide to truth. (Here again, though, one might wonder
whether de facto expertise is relevant. What if the rest of the com-
munity is ignorant of who the real experts are? The community as a
whole just happens to share their opinions by coincidence. Is the
group's belief-profile still guaranteed to be rational? I think not.)
Let us call the first kind of expertism, which appeals to genuine
experts, objective expertism, and the second kind, which appeals to
(deserved or undeserved) reputations for expertise, subjective, or
reputational, expertism. As indicated, I believe that objective exper-

tism has some plausibility. Not so much plausibility, perhaps, as a
theory of 'social warrant', but nonetheless as a (partial) standard for
social epistemology. A little reflection should show, however, that ob-
jective expertism derives its plausibility from a more fundamental val-
ue. It is epistemically desirable for a group to reflect expert opinion
because this increases the likelihood of the group's having true beliefs.
If attainment of true beliefs is taken as the fundamental epistemic
desideratum, then expertism has plausibility as a derivative theory.
I use the term veritism for the approach that rates true belief as the
ultimate epistemic aim. This is the fourth approach I consider. Objec-
tive expertism can be seen as a corollary of veritism, since conforming
to the opinions of genuine experts is a good means to true belief. Even
subjective, or reputational, expertism can be viewed as derivative
from veritism. After all, people need some principle to guide them in
trying to realize veritistic ends. In the absence of anything better, they
might be told to conform their opinions to those of the people they
deem experts. Finally, even consensualism can best be defended on
veritistic grounds. To the extent that societal consensus is a reliable
guide to truth, it makes sense to abide by that consensus. Of course,
societal consensus is not always a good guide to truth, and that is
precisely one of consensualism's weaknesses. But it might make sense
as a derivative rule of thumb, in the absence of anything better. It is
indefensible, however, as a fundamental criterion of epistemic worth
(or of social warrant or rationality).
While objective expertism has merit for the indicated reason, it is at
best a partial approach to social epistemology. First, there are possible
situations in which there are no experts. Here social epistemology
would have nothing to say if it were confined to expertism. It might be
replied that there must always be experts: if not experts in some
absolute sense - people whose reliability on the topic exceeds a
specified threshold - at least people who are comparatively expert -
i.e., more expert than their peers. However, even comparative experts
may not exist. The community may have nobody with any opinion at
all on the topic. We would still want principles of social epistemology
to apply in such cases. Second, whatever expertise already exists in a
community, social epistemology need not be restricted to the adjust-
ment of lay belief to expert opinion. A plausible province for social
epistemology is the creation and improvement of institutions for
increasing the level of expertise. Third, expertism only gives advice to

non-experts. What should the experts themselves do? Are there no
social procedures for becoming expert, or enhancing one's expertise?
Surely such procedures are not precluded; and if they exist, they
belong to the realm of social epistemology. For many reasons, then,
objective expertism does not exhaust social epistemology. (There are
also problems inherent in a more precise formulation of expertism.
What should be done, for example, when experts disagree? Although I
do not try to address this systematically, the following paragraphs are
relevant.)
Before leaving the topic of expertism, it is worth illustrating the idea
with two examples, one of objective and one of subjective expertism.
The example of objective expertism is drawn from a formal treatment
of optimizing group judgmental accuracy, by Lloyd Shapley and
Bernard Grofman (Shapley and Grofman, 1984). Shapley and Grof-
man are interested in the general problem of group decision making,
in particular, how best to assign voting weights to group members to
optimize a group decision. A special case is where a group chooses
whether to believe a given proposition or its negation. (This concerns
the notion of aggregate, or collective, belief, which I have not
expressly treated. I shall accept this notion without further comment.)
Shapley and Grofman show that if individual choices - opinions, in our
special case - are mutually independent, and if the a priori likelihood
that either of the two choices (beliefs) is correct is one half, then a
decision rule which maximizes the probability of the group's getting
the truth is a weighted voting rule. In particular, an optimal voting
rule must assign weights wi so that the following constraint is met:
where Pi represents the probability that member i has a true opinion.

For example, if there is a five member group whose respective
probabilities of being correct are 0.9, 0.9, 0.6, 0.6, 0.6, then one
optimum weighting scheme assigns these members the weights 0.392,
0.392, 0.072, 0.072, 0.072.
Now if one individual has a higher probability of being correct
(having a true opinion) than another, the former has greater objective
expertise on the subject. So the foregoing formula for selecting a
decision rule (a certain type of social procedure) is designed to fix a
group belief by weighting greater expertise more heavily. Shapley and

Grofman prove that such a decision rule is conductive to group
accuracy, i.e., maximizes the probability of getting a true group belief.
In other words, assigning appropriate weights to expertise conduces
to veritistic ends. Although Shapley and Grofman do not give much
explicit attention to truth and veritism - their scheme is more abstract
and general - their model certainly exemplifies the choice of a social
procedure under the aegis of veritism.
The second example I shall give, an example of subjective exper-
tism, is the approach of Keith Lehrer and Carl Wagner (Lehrer and
Wagner, 1981). (Another example of subjective expertism is found in
Stich and Nisbett, 1980.) The Lehrer-Wagner theory looks like a
consensualist approach, since it lays great emphasis on the rational
transformation of initially disparate opinions (subjective probabilities)
into consensus. But the source of rationality is each person's
(repeated) revision of his subjective probability in accord with weights
he assigns to others and their subjective probabilities. Now the weights
a person assigns represent his assessments of the competence, reli-
ability, or expertise, of the members in question. Lehrer and Wagner
deliberately place few constraints on how these assessments are made.
There is no guarantee, then, that the weights (or degrees of 'respect')
have any correspondence to genuine competence or expertise. This is
why I call their approach one of subjective, or reputational, expertism.
This is not an appropriate occasion for detailed appraisal of this
approach. I merely repeat my earlier suggestion that the plausibility of
the approach is best seen as a corollary of veritism. Its limitations are
also best appreciated from a veristic perspective. I conclude that
veritism represents the most adequate general approach to social
epistemology, and is the one I favor for social epistemics. However, I
do not favor it as an account of social 'rationality' or 'warrant'. A
more positive account of its contours is presented in the next two
sections.
4. V E R I T I S M
There is much more to be said in defense of veritism as the principal

approach to social epistemology. The chief rationale, quite simply, is
that the goal of truth is the common denominator of intellectual
pursuits, whatever methods or practices are championed as the best
means to this end. True belief is the shared aim of the Inquisitor and
the scientist, of the creationist and the evolutionist, and of all the
competing research programmes that populate the agonistic arena of
science. The Inquisitor thinks he knows the truth already, that his
victims are well served by being brought to this truth, despite the
severity of the measures. The scientist thinks there is no better road to
true belief than that of free, uncoerced inquiry. However vast the gulf
between these philosophies, they share the goal of truth.
It would be foolish to suppose, of course, that every defender of a
favored orthodoxy preaches it with sincere belief in its truth or
truth-fostering powers. M~ny an ideology, discourse structure, and
conceptual scheme has been embraced in the interest of power rather
than an interest in truth, as sundry theorists such as Marxists, Nietz-
sche, and Foucault have stressed. Still, proponents of such positions
must fly their public claims under the banner of truth, or eventual
truth attainment. They know this is the presumptive aim of intellectual
claims. To admit that one's favored methods or policies are mere self-
or class-serving fictions, that they have no genuine propensity to
conduct people on a path toward truth, is an admission of intellectual
bankruptcy. The goal of truth must be honored even in the breach,
just as the goal of winning must be simulated even by an athlete who
seeks to 'throw' the contest.
Veritism does not deny that people engaged in intellectual pursuits
often have extra-verific motives for what they do and say. Fame,
esteem, and fortune are not even necessarily illegitimate motives; they
are epistemically illegitimate only where they conflict with the pursuit
of truth. But to the degree that social or professional frameworks
encourage the sacrifice of truth to the satisfaction of these extraneous
ends, to that degree such frameworks are deficient from an intellectual
standpoint. Veritism does not claim that every existing practice that
nominally aspires to intellectual ends in fact optimizes their attain-
ment. It only claims that this is a proper benchmark for grading these
practices, precisely because it is their presumptive, or default, goal.
Workers in the social studies of knowledge commonly bracket the
issue of truth. (One exception here is Donald Campbell, whose ap-
proach has points of contact with veritism. See Campbell 1986.) In
analyzing particular cultures and scientific movements, historians,
sociologists, and anthropologists typically shy away from questions of
truth. This is understandable. Light can be shed on the relation
between certain social variables and theory acceptance without ad-

dressing a theory's truth value. However, while truth can be set aside
in the social explanation of belief, truth is not irrelevant in the
epistemic appraisal of ideational practices. Furthermore, while scholars
of the intellect commonly ignore the truth values of the theoretical
commitments of their protagonists, these same scholars themselves
aspire to truth in their own scientific or historical assertions. They aim
at truth in their purely descriptive account of the groups and con-
troversies in quesiton, and they aim at correct explanations of doc-
trinal changes. For example, when Martin Rudwick (1985) traces the
Devonian controversy (in part) through a "gradient of ascribed com-
petences", he is claiming that competence ascriptions in fact played a
causal role in changing bodies of opinion. This is a claim he presum-
ably believes to be true, and he would resist attempts to show the
non-existence or causal irrelevance of the pertinent competence
ascriptions. Thus, while students of science and culture may properly
abstract from the truth of the ideational contents they study, they do
not and should not extrude the question of truth from their own
propositions about the growth, prevalence, and extinction of people's
ideational contents.
In the philosophy of science, some writers question the role of truth
in scientific theorizing. But even these 'empiricists' or 'anti-realists'
acknowledge that a crucial test of competing theories is their ability to
predict and retrodict observational data correctly, i.e., truly. So the
value of truth is again acknowledged, though with some diminution of
the domain.
Other philosophers downgrade the role of truth in science and
culture for slightly different reasons. Richard Rorty argues for replac-
ing objective truth as a goal of inquiry with merely "keeping the
conversation going" (Rorty 1979, p. 377). No doubt it would be
wrong to deny that conversation is sometimes conducted for its own
sake; and conversationalists often seek the satisfaction of affiliative
needs, or emotional release, more than the imparting of truths. The
sheer pleasure of language is also a primary motivator of everyday
talk, as well as literary expression. So not all human communication or
cultural activity aims at truth and truth alone. However, epistemology
is not concerned with all facets of interpersonal or cultural activity. It
focuses wholly on the intellectual sphere, and on intellectual ends;
here the emphasis on truth is entirely in order.
This is not to deny that many an institution or practice can be

evaluated along multiple dimensions, both intellectual and nonin-
tellectual. Judicial practices, such as rules of admissible evidence, may
be judged in part by their effect on accuracy of verdicts. But they can
also be judged by their impact on civil liberties. Practices which are
good or bad from an intellectual standpoint may have different quali-
ties from a political, legal, or moral standpoint. Epistemology concen-
trates on the intellectual perspective without necessarily taking a stand
on how the interest of the intellect should finally be balanced against
possibly competing interests. Some practices, such as guarantees of
freedom of speech, may be justified by reference to both intellectual
and moral considerations; but these considerations should be kept
distinct.
Some philosophers would acknowledge the centrality of truth in a
theory of inquiry but dispute the role of true belief. Popper (1972) has
endorsed the aim of truth, but minimizes the place of subjective states
in epistemology. He propounds a conception of epistemology that is
concerned with a purely objective realm of entitites - problems,
theories, and arguments as such, i.e., objective entities quite apart
from any mental or physical realization. The theory of science and
culture, however, cannot confine itself to 'third-world' objects. The
problems, theories, and arguments that exist 'objectively' are not all
that matter to the state of science at a given juncture. It also matters
who appreciates the problems, accepts the theories, and is persuaded
by the arguments. Considered objectively, both good and bad theories
'exist', as do powerful and completely specious arguments. The critical
question is what impact or reception these theories and arguments
have on scientists and other cognizers. Social epistemology is there-
fore concerned with the promotion of true belief, not the mere
existence of true contents, which may be rejected or ignored by
scientists and laymen alike.
Thus far I have spoken rather vaguely about veritism and the aim of
truth-acquisition. It is time to be more precise. One question concerns
the objects of evaluation under the veritistic approach. As I have
indicated, the chosen targets of evaluation are social practices and
institutions. But these are to be evaluated by their consequences for
social belief-profiles. Social practices are epistemically desirable to the
extent that they promote epistemically preferred belief-profiles. A
second question concerns the precise term of epistemic appraisal
under the veritistic approach. Here I am not entirely decided. I reject

the terms 'warranted', 'justified', or 'rational' as the vocabulary of
choice. They are either inappropriate, excessively vague, or exces-
sively narrow. But I have no single term or expression that captures
the relevant dimension or dimensions of appraisal. Intellectual strength
is the best phrase I can think of; but it is not perfect. 'Intellectual
virtue', or 'epistemic virtue', are other possibilities, though the latter is
lacking in antecedent associations.
Unwillingness to be committed to a unique term of appraisal does
not undermine the veritistic conception of social epistemics, as long as
veritistic standards are formulated by which social practices can be
assessed. There are, indeed, a number of distinct truth-linked stan-
dards, any or all of which can be used to appraise social institutions
and practices. Five different standards can usefully be distinguished:
(1) reliability, (2) power, (3) fecundity, (4) speed, and (5) efficiency.
The reliability of a practice is measured by the ratio of truths to total
number of beliefs a practice would foster. In this reckoning, it is belief
tokens that matter. Ten people believing one proposition determine
ten beliefs. (Perhaps the duration of beliefs should also be incor-
porated into the calculus; but I set that aside.)
By power I mean the problem-solving, or question-answering,
power of a practice (see Goldman 1986, chap. 6). This is measured by
the ability of a practice to help cognizers find and believe true answers
to the questions that interest them. Reliability and power are not
identical. A reliable practice helps prevent error but it does not
necessarily combat ignorance. A practice that generates very few
answers to questions of interest may still have a high truth ratio. But a
high truth ratio is not sufficient for power. A powerful practice is one
that is comparatively effective (at least in the long run) in generating
true answers, and beliefs in these answers, not merely maintaining a
small proportion of false beliefs.
As defined, the term 'power' retains some ambiguity. On one
reading, a practice might qualify as powerful if, for a high proportion
of questions asked, it secures true answers for at least one questioner.
On a second reading, a practice would qualify as powerful only if it
generates knowledge (or true belief) in these answers on the part of
many questioners. To capture the second interpretation, I introduce
the term fecundity.
Fecundity is an important standard for social epistemics. A cul-
ture's scientific practices might be relatively powerful in the sense that

they precipitate the discovery of many scientific truths by a small body
of scientific elite. But public education in the culture may be weak, so
the bulk of the populace is mired in superstitution and ignorance. Such
a culture's practices or institutions are wanting in fecundity, which is
surely one measure of a culture's intellectual strength.
The last two truth-linked standards are question-answering speed
and question-answering efficiency. The former is the relative speed of
getting correct answers. The second standard, efficiency, incorporates
the element of cost. When people want answers to complex or pre-
viously unanswered questions, resources are typically required to get
those answers. More efficient practices are ones that promote answer-
acquisition at lower cost. They might conserve costs by collaboration
and division of labor, for example. The economics of information is
certainly an important topic in the theory of information, and should
not be omitted from the agenda of social epistemics.
How should this array of truth-linked standards be incorporated into
veritism? Two routes might be taken. One would simply acknowledge
this multiplicity of standards and use 'intellectual strength' as an
umbrella term for the set, without trying to weld them into a single
criterion or measure. No complete ordering of possible social belief-
profiles would be contemplated. Alternatively, one might try to effect
a complete ordering, which would require assigning weights to each
separate standard (as well as making them more precise). Some
arbitrariness would be inevitable, but perhaps that isn't a tragedy. In
this paper, I do not try to choose between these strategies. I continue
to speak loosely of veritism without opting for any resolution of this
problem.
Despite the resulting vagueness, veritism retains plenty of content
by contrast with alternative approaches. One contrasting approach not
thus far mentioned is adaptiveness. A social practice might be rated
positively because it promotes survival or fitness, either the survival
and fitness of individual group members or the survival of the group
qua social entity. Functional explanations in sociology and anth-
ropology frequently take this approach, and it might be invoked as an
evaluative stratagem as well. A religious practice, for example, might
be graded highly because of its favorable effect on social solidarity.
However, this sort of grading is not expressive of the intellectual
dimension, the dimension traditionally associated with epistemology's
mission. There is no tight connection, at least no necessary con-

nection, between intellectual strength and adaptiveness. The science
of nuclear fission is intellectually powerful, but it may be inimical to
adaptiveness. For good or ill, intellectual strength is a virtue (or set of
virtues) distinct from adaptiveness.
I should emphasize that there is nothing fundamentally new about
social epistemics, conceived veritistically, except the name. The
philosophical task is that of giving a rational reconstruction of what
people have long said and done in the socio-intellectual arena.
Organized science establishes institutional practices to avoid fraud and
fabrication. The adversary process is instituted in the Anglo-American
legal system to promote more accurate fact-finding. The free market-
place of ideas is defended in the expectation that free competition is
the best means of uncovering truth. All these familiar practices and
institutions are traditionally and plausibly rationalized on veritistic
grounds. So they are practices that arise from the execution of social
epistemics, though not under that label. My attempt to formulate this
conception is just an attempt to reconstruct and systematize an enter-
prise that already has a long history.
5. T H E SCOPE OF SOCIAL EPISTEMICS
Although I have said a few words about objects of evaluation for

social epistemicS, a fuller treatment is needed to fix the relationship
between social and individual epistemics. A useful way to proceed is
to start with a taxonomy of belief determinants. Not all belief deter-
minants are proper objects of epistemic evaluation, but it is instructive
to note which of them are, and of these which are objects for social
epistemics.
Here is a taxonomy of factors that can influence beliefs. This
taxonomy has no privileged status, but it can help carve out the
territories of the parts of epistemics. The starred items represent types
of factors that are evaluative objects for social epistemics.
(1) Environmental objects and events
(2) Basic cognitive processes
(3) Non-social belief-forming methods
*(4) Social belief-forming methods
*(5) Patterns of communicational behavior
*(6) Institutional rules and structures

(7) Nomological regularities
Let me clarify each of these taxa, taking them in order.
Environmental objects and events serve as perceptual stimuli, and
give rise to beliefs via perceptual processes. They are, therefore,
prime causal factors in the determination of belief. However, no
epistemology grades environmental stimuli as epistemically good or
bad, permissible or impermissible. So they are not objects of evalua-
tion for any part of epistemics.
The second category, basic cognitive processes, includes processes
of perception, memory, attention, concept formation, problem solving,
learning and reasoning. By 'basic' processes, I mean (roughly) natural
or native processes, not algorithms, techniques, or heuristics that are
acquired through experience or explicit tutelage. In Goldman (1986) I
have argued that basic processes are suitable objects for epistemic
evaluation. In particular, they comprise the domain for primary in-
dividual epistemics.
The third category, non-social belief4orming methods, is meant to
delineate the province of secondary individual epistemics. A method,
as I use the term, is a learnable algorithm, technique, or procedure for
arriving at beliefs. It can either be a highly generic, topic-neutral,
procedure or a very task-specific one. Methods range from proof-
techniques in logic to procedures for performing and interpreting a
spectrographic analysis. A special class of methods are ones that
appeal, in whole or part, to the assertions, utterances, and beliefs of
other people. These are what I call social methods, which are excluded
from the third category. I want to reserve the class of social methods
for the sphere of social - or 'tertiary' - epistemics. Only non-social
methods constitute the objects of appraisal for secondary epistemics.
The fourth category consists of social belief-forming methods:
modes of doxastic response to the opinions and messages of others.
Within this category are patterns of doxastic response to various
argumentative and rhetorical styles, methods of assessing people's
trustworthiness, and so on. What distinguishes social methods is that
their inputs are opinions or communicational acts of other people, or
the cognizer's beliefs about these. The outputs of social belief-forming
methods need not have any social content. (On the other hand,
non-social belief-forming methods can have social contents in their
outputs.) I acknowledge that the resulting demarcation between social

and non-social methods may not be very sharp; some methods may
mix social and non-social inputs. But we should not expect the
boundary between social and individual epistemology to be perfectly
well-defined.
(Since the boundary between social and non-social methods is not
sharp, perhaps all belief-forming methods could be treated under
secondary epistemics. It would then turn out that part of social
epistemics would be subsumed under secondary epistemics. This is a
conceivable approach, though not my'preferred one.)
Social belief-forming methods do not exhaust the objects of social,
or tertiary, epistemics. (The term 'tertiary' is not intended invidiously,
of course, it is just a natural complement to the other numeric labels.)
Social epistemics also seeks to evaluate patterns and strategies of
communicational behavior. What verbal techniques does the speaker
use to try to persuade his audience? What patterns of questioning,
challenging, rebutting, and criticizing does he employ against an
interlocutor? What communicational media and formats are available
for his use, and which does he select? These various patterns and
strategies comprise the fifth category of belief determinants.
The sixth category features institutional rules and structures that
affect the flow of communication and fix the processes of collective
belief-formation. These can play a critical role in determining which
messages are sent by which sources, over what communication chan-
nels, and to what class of potential receivers. Except for informal,
face-to-face situations, where speech opportunities are largely un-
constrained, potential message senders need to pay or vie for the use
of communication channels. Advertisers must pay for commercial
space; speakers in parliamentary bodies must be recognized by the
chair; public figures must produce messages that are deemed news-
worthy; authors and researchers must have their work accepted by
publishers and professional journals. All the institutional rules and
structures that govern the choice of message conveyance play a
critical role in information flow, and ultimately in belief production.
In certain specialized settings elaborate formal rules control in-
formation flow. Take a judicial system, for example. It aims, among
other things, to produce verdicts of guilt or innocence. The American
judicial system typically involves several parties: a judge, a pair of
opposing legal teams, and a jury. Communications are made under a
set of formal procedures, presided over by the judge. Juries then

debate the case and produce a verdict. Among the rules incorporated
in this institutional set-up are rules governing: (i) the size of the jury;
(ii) the process of juror selection and exclusion; (iii) jury decisions
(e.g., unanimity versus majority rule); (iv) admissible evidence; (v) the
privileged attorney-client relation and advocacy responsibilities; and
(vi) the burden of proof and the instructions given to jurors (e.g., to
convict only when guilt is established "beyond reasonable doubt"). All
such institutional rules and structures are candidates for evaluation
within social epistemics.
The seventh category is not really a class of causal factors, but laws
governing the causal network by which the preceding factors are
linked with beliefs. In addition to psychological laws, there might be
laws of economics that subsume the buying and selling of information,
or principles of cultural transmission analogous to principles of biolo-
gical inheritance (see Cavalli-Sforza and Feldman 1981, and Boyd
and Richerson 1985. None of these laws or regularities would be
evaluated by social epistemics. But they would be of interest to the
social epistemologist in trying to determine the effects of this or that
set of belief-influencing factors.
As explained in Section 4, the standards for evaluating the relevant
social practices are verifistic ones. Social epistemics would assess alter-
native practices in categories (4), (5), and (6). It would enlist the help
of sundry social and behavioral sciences in ascertaining the impact of
alternative practices on social belief-profiles. It would then rate these
practices on veritistic grounds.
In traditional epistemology and philosophy of science, emphasis has
been placed on specification of optimal or ideal methods and prac-
tices. But my proposal for social epistemics (as well as for other
portions of epistemics) would advocate an assessment of all possible
methods and practices. It is important to identify sub-optimal practices
because the optimal ones may not be feasible for every group; or they
may be costly in terms of some competing, non-veritistic values. To
assess whether verific interests should be sacrificed to another value,
one would like to know how serious the veritistic loss is likely to be.
(Indeed, one of my so-called veritistic standards - efficiency - already
requires a calculation of verific gain per some sort of unit of cost.)
How is my delineation of social epistemics related to traditional
definitions of the scope of epistemology, philosophy of science, and
other studies of science? Epistemology and philosophy of science have

tended to focus on categories (2) and (3), though without a systematic
distinction between native and learned procedures. Philosophy of
science has standardly focused on what I would call methods, especi-
ally methods of collecting and assessing evidence. In my categoriza-
tion, then, traditional philosophy of science is (largely) secondary
epistemics. Other disciplines that study science - especially the history
and sociology of science - come closer to social epistemics insofar as
they explore the kinds of practices and institutions in categories (4),
(5) and (6). However, these disciplines characteristically ignore
evaluative questions, and especially abstain from veritistic forms of
evaluation.
6. AGAINST NIHILISM
Veritism stands in clear contrast to evaluative relativism, in particular,

to the doctrine of nihilism or anti-universalism that is embedded
within relativism. It is to be expected, then, that relativists would find
veritism unacceptable. Other theorists can also be expected to ques-
tion veritism on various counts. Let us therefore anticipate some
points of concern and controversy.
Barnes and Bloor (1982) enunciate their relativistic position as
follows: "For the relativist there is no sense attached to the idea that
some standards or beliefs are really rational as distinct from merely
locally accepted as such. Because he thinks that there are no context-
free or super-cultural norms of rationality he does not see rationally
and irrationally held beliefs as making up two distinct and qualitatively
different classes of things" (Barnes and Bloor 1982, pp. 27-28).
Actually, the Barnes and Bloor formulation does not precisely clash
with the theses of social epistemics, because they dispute the pos-
sibility of evaluating beliefs in terms of rationality, while social epis-
temics seeks neither to evaluate beliefs per se nor to evaluate them in
terms of rationality. However, the denial of context-free or super-
cultural norms can be carried over to social practices and institutions.
Relativism may deny that there are any context-free or super-cultural
criteria or standards for judging social practices and institutions.
One point that seems critical to the present form of relativism is the
claim of an irreducible plurality, or multiplicity, of evaluative stan-
dards. Of course, veritism also posits multiple standards: reliability,
power, fecundity, speed, and efficiency. However, each of these is

supposed to be super-cultural, or culture-neutral. So veritism clearly
conflicts with the anti-universalistic component of relativism.
There is another important respect, though, in which veritism can
be pluralistic. Veritism can admit that many alternative social prac-
tices or institutions may be equally reliable, equally powerful, equally
fecund, and so on. Veritism is not committed to a uniquely optimal set
of social practices. (See Goldman 1986, Section 4.3, for an analogous
point about pluralism in individual epistemology.) Since traditions and
conditions may vary from culture to culture, it may also turn out that
what is veritistically optimal for one culture is not optimal for another.
Veritism is compatible with pluralism in another way too: it does not
require every individual to use the same methods. Let a group method
M* be a set of methods (or method bundles) {M1 . . . . . M i , . . . , Mn},
where M~ is the method (or method bundle) used by member i. A
group method is homogeneous just in case all members of the group
use the same method (or method bundle). A group method is hetero-
geneous if two or more members use different methods (method
bundles). Now there is nothing in veritism that precludes hetero-
geneous group methods. Indeed, it could turn out that heterogeneous
group methods are preferable on veritistic grounds. Alternatively, we
can think of a group method as a policy of encouraging or promoting
individuals to use various methods. Then Feyerabend's anarchism
(Feyerabend 1975), which ostensibly opposes all methods, can be
reconstrued as an endorsement of a single group policy, viz.,
"Encourage the proliferation of diverse methods among different
members." On either construal, veritism can in principle agree with
Feyerabend's anarchism, or with Sarkar's approach of method multi-
plicity (Sarkar, 1983).
However, the nihilist might not be assuaged by veritism's com-
patibility with pluralism. The very idea of any super-cultural standard
might be regarded as indefensible. What, then, is my defense of
veritistic standards? Very simply, truth-linked standards are the stan-
dards implicit in the process of inquiry, and in the very notion of
'intellectual' aims and assessments. Individual intelligence is (at least
partly) a matter of facility at problem solving, of finding and believing
true answers to the questions one wants to answer. (See Goldman
1986, Chapter 6.) Similarly, social intelligence is a matter of social
practices and institutions that foster social problem solving. Getting
true answers is, roughly speaking, what defines the aims of the
intellect. So all cultures, communities, or groups that possess the
notion of intellectual appraisal eo ipso accept truth-linked standards,
however much they differ about the methods for applying those
standards. Needless to say, few cultures, communities, or groups have
exclusively intellectual aims. Science and academe are the distinctive
subcultures in which intellectual aims are paramount. In most other
communities and groups, intellectual aims are subservient to other
aims. But insofar as intellectual aims are distinctly identified, truth-
linked standards come into play. This ~s not a point on which cultures
differ from one another.
There are, however, three quite different sources of discontent with
veritism I have not yet identified. These are associated with charges of
(A) circularity, (B) emptiness, and (C) uselessness.
The circularity critique runs as follows. Since 'truth' must be defined
in terms of intellectual methods, it cannot, on pain of circularity,
appear in a criterion for such methods. This sort of objection has been
lodged by Putnam (1983) in discussing the reliability approach to
rationality (or justification). But this claim presupposes that truth is
definable epistemically, or nonrealistically. If realism about truth is
correct, as I argue it is (cf. Goldman 1986, Chapter 7), the point
collapses. On a proper understanding of 'truth', it is not defined by
reference to methods of truth-determination, or verification. Hence,
there is no circularity in having truth-linked criteria of method ap-
praisal.
The objection from emptiness runs a bit differently. All veritistic
standards, it is claimed, assume there are truths or facts, independent
of what groups of people believe. But there are no such facts.
So-called facts are mere constructions by groups of inquirers. On one
variant of this theme, the alleged constructive process can actually be
observed in the scientific laboratory. Science is not a matter of
discovering pre-existing facts; it is just a process of negotiation. Facts
are things to be negotiated, or fabricated, not discovered. So goes the
argument of scientific 'constructivists', such as Latour and Woolgar
(1979) and Knorr-Cetina (1981).
A full discussion of this claim cannot be undertaken here. In the
present context I rest content with two points. First, the sociologist
who studies life in a scientific laboratory does not observe the nego-
tiation of scientific facts. What is observed is only the negotiation of
scientific assertions or beliefs, i.e., what the scientists agree to say or

believe about the facts. That there are 'negotiation' processes in social
belief-fixation hardly demonstrates that there are no facts of the
matter independent of this negotiation.
Second, the prevalence of negotiation is readily explained in a
fashion congenial to veritism. Scientists negotiate with each other
because they accord one another a certain degree of respect as
experts, as people who stand a chance of being right comparable to
their own. They have every reason to heed one another's opinion, and
try to arrive at a cognitive judgment that meets, as far as possible,
with general approval. The negotiation process is best understood,
then, on the assumption that there are facts in the world, and that the
several members stand a better chance of 'getting' the facts if they
heed the opinions of others.
The third objection against veritism is that of uselessness. The
whole point of a criterion in the realm of social epistemology, it is
argued, is to use or apply the criterion. This means determining the
(degree of) intellectual goodness or badness of different social prac-
tices. If there is no mutually agreed upon way of applying a criterion,
it is useless. Yet this is precisely what obtains for veritism if, and to the
extent that, descriptive relativism is true. Where communities differ in
their intellectual practices, they will differ about how to apply a
veritistic criterion. Any attempt to reach a consensus, or settle a
dispute, about the veritistic properties of a specified social practice
will be in vain. Hence any veritistic criterion is useless. This objection
merits treatment in a separate section.
7. THE PROBLEM OF APPLICABILITY
Let me first clarify what I mean by a 'criterion', or 'standard', of

appraisal. I mean simply a condition that specifies a characteristic or
magnitude that makes the objects in question (comparatively) good or
bad. The term 'criterion' is sometimes used in a different sense, in the
sense of a verification procedure, or way of telling, whether an object
possesses a given characteristic or realizes a given magnitude to a
specified degree. But the term is not so intended here. Specification of
a good-making characteristic is a conceptually prior task, to be
distinguished from the task of finding verification or determination
procedures. (For further clarification of these points, in the context of
a theory of justification, see Goldman 1986, Section 4.2.)

Given this understanding of a standard, or criterion, it is no criti-
cism of veritistic standards that they are not themselves determination,
or application, procedures. However, granting that a criterion is not
intended to be a verification procedure, wouldn't it be a salient defect
of veritistic criteria if they were not applicable, if there were no
associated procedures for determining whether any selected social
practices satisfy them (to this or that degree)? Doesn't the nonap-
plicability critique still carry weight?
In considering this critique, we need to look more closely at what is
meant by 'applicability', or 'determination procedure'. One thing that
might be meant by a 'determination procedure' is a decision procedure.
This, in turn, has two possible interpretations. In one common and
very strong interpretation, a decision procedure is a method which
guarantees both (A) that the user will arrive (after finitely many steps)
at an answer (belief) about whether any targeted object satisfies the
standard, and (B) the resulting answer (belief) will be correct.
Is it a reasonable constraint on a standard that it have an associated
decision procedure? No, this is an excessively stringent requirement,
certainly on the strong reading of 'decision procedure'. Even first-
order logic does not have a decision procedure in this sense. Fur-
thermore, it is doubtful whether any proffered standard for social
epistemology would have a decision procedure. Consider even the
standard of consensualism, where applicability seems quite promising.
Is there a procedure associated with consensualism that guarantees
that it will yield correct answers to the question of whether a given
group agrees (unanimously or to any specific degree) on a given topic?
In light of the possibility that some people may suppress their true
opinion - or just in light of the general problem of ascertaining
people's beliefs - there is no such guarantee. So veritistic standards do
not seem inferior to other possible standards in respect of (strong)
applicability.
Suppose we weaken the sense of 'decision procedure' by deleting
clause (B), the requirement of correctness. What we are then left with,
however, is much too easy to satisfy. If there is no requirement of
correctness of application, an acceptable determination procedure
might simply be, "Answer 'yes' to all even-numbered queries of
application and 'no' to the remainder."
Presumably, then, some sort of correctness desideratum should be
part of a determination procedure. But how much correctness, and

how universal must the scope of application be? These questions
indicate that the very notion of a determination procedure is rather
vague. Until this vagueness is reasonably resolved, there is no plausi-
ble demonstration that veritistic criteria cannot have associated
determination procedures.
In formulating the envisaged nonapplicability objection, there was
another strand in the argument that I have not yet confronted. There
was the suggestion that an acceptable standard ought to be mutually
applicable, i.e., applicable by different communities to their mutual
satisfaction. But can veritistic standards be used to settle disputes, or
reach a consensus, about specified social practices? The picture
presupposed here is something like this. An assemblage of represen-
tatives from different intellectual traditions confront the epistemolo-
gist with the entreaty: "Give us a criterion for choosing (or ranking)
social practices that will lead us to (complete) agreement in this
choice." The contention is that veritistic standards do not comply with
this entreaty.
But why should the epistemologist accept this version of the chal-
lenge? Why suppose there is any criterion satisfying the condition that
even people from highly disparate intellectual traditions will come to
(complete) agreement when applying it? Different intellectual tradi-
tions may have all sorts of idiosyncracies and foibles. They may do
(what they call) 'science' in different fashions; they may even use
different logics (e.g., classical versus intuitionistic). So why suppose
that any single criterion could be used to settle disputes? Furthermore,
why is it necessarily a deficiency in a criterion that it could not settle
all disputes? As long as it could be used correctly, by people who
antecedently employ suitable processes and methods, what more can
reasonably be required? Obviously, the use of any criterion requires
employment of previously available processes and methods (at least
non-social methods). If the processes and methods certain people
bring to the criterion are bad enough, they won't apply the method
correctly, and the results of their applications will conflict with the
applications of others. But why is that a fault of the criterion?
The objector may still not be assuaged. All right, he may reply,
perhaps total agreement on application is too much to expect for any
criterion. But what about the possibility of producing a moderate
amount of agreement? Isn't this a good,making trait of a criterion?
Two comments are in order here. First, it has not been shown that
veritism fails to satisfy this more relaxed requirement. It will all
depend on the range of intellectual traditions envisaged as participants
and on how much agreement is a 'moderate' amount. Depending on
how these parameters are fixed, the requirement may or may not
continue to seem reasonable. If other candidate criteria continue to
run afoul of the requirement, which they may, its reasonability would
be moot.
I have a more serious objection to this maneuver, however. It seems
to me fundamentally misguided to impose usability or applicability of a
criterion as a test of (comparative) adequacy. Why should we consider
replacing veritism - which has many arguments in its favor - with
some other criterion, simply because the latter is more usable?
The point can be illustrated by the familiar story about a man who is
poking around on the ground under a lamppost. A friend approaches
and asks him what he is doing. He answers that he is looking for his
watch, which he lost in the bushes across the street. When the friend
asks why is looking here, when he lost the watch over there, the man
replies, "because the light is better here". The moral of the story is
obvious. The ease of applying a criterion, or its propensity to yield
greater agreement, is no reason to suppose it is a better, or correct,
criterion. Ease of application has no clear relevance to correctness,
just as better light under the lamppost is not a relevant reason for
looking there.
Although I refrain from building any particular determination-
procedure into the content of veritism, it should not be inferred that I
have no preferred group of determination-procedures of my own,
which I expect to be shared by many of my readers. On the contrary, I
presume that the best available set of procedures for determining the
veritistic properties of social practices are various procedures of the
sundry empirical and mathematical sciences. It is precisely because
such sciences are needed to apply the veritistic standards that I regard
social epistemics as an alliance between philosophy and various
scientific disciplines.
This acknowledgement, however, may be seized upon to register a
new criticism against veritism, a regress charge. Suppose a group
wishes to decide which social practices to adopt, and consults veritistic
critieria. Won't it have to use some social practices in order to apply
the criteria, such as the social practices embedded in current social
science? But which practices is it allowed to use? Here, again, it looks

like it should consult the criterion. But 'consulting' the criterion means
applying it; and this requires the use of social practices. So a vicious
regress seems to loom.
Two points are germane to this problem. First, the problem as stated
holds as much for nonveritistic as veritistic criteria. It does not cut
invidiously against the latter. Second, the regress can really be aver-
ted. For one thing, judgments about the best social methods or
practices do not necessarily depend on the use of social methods or
practices. Individuals might arrive at these judgments privately,
through the use of basic processes and non-social methods only. Of
course, if they seek to implement these practices within a group, there
will have to be some form of communication, hence some social
practice. But judgments about the epistemic properties of social
practices do not require social practices. Finally, it is a mistake to
suppose that social practices must be chosen at every stage in order to
apply a criterion. Social practices may simply evolve, without deli-
berate consultation with a criterion. So there is no cogent argument
for a vicious regress. (See similar remarks of mine in Goldman 1980,
Section IX.)
A final criticism of veritism is one from necessary self-endorsement.
Veritistic criteria, the argument runs, ensure that whatever methods or
practices are already in place will be approved. If we use practice P to
assess the truth-values of the beliefs that result from P, won't we
necessarily conclude that P performs beautifully? But what value does
such a verdict have? None. Since there is no possibility of a practice
impugning itself under veritism, veritism is worthless as a standard of
appraisal. To use Rorty's epithet, it is purely "Whiggish". (Cf. Rorty
1979, Chapter 7; also Firth 1981.)
The first point to make, in reply, is that self-endorsement is not
obviously a flaw. Self-endorsement is a special case of what Robert
Nozick (1981) calls 'self-subsumption'. He argues persuasively, in
various contexts, that self-subsumption is not a defect. Admittedly, if a
practice or method were necessarily self-endorsing, we might question
the significance of any application of it that yields a self-endorsement.
However, veritistic criteria are by no means necessarily self-endorsing~
How might a social practice or method, applied to a veritistic
criterion, yield a self-impugning judgment? Consider a decision rule
la Shapley and Grofman (1984), which incorporates certain voting
142 A L V I N I. G O L D M A N
weights for making collective judgments. Suppose this decision rule

violates the log formula Shapley and Grofman present. It is still
possible that a group using this decision rule should collectively
decide that only decision rules satisfying the Shapley-Grofman for-
mula optimize accuracy, and should be adopted instead. In this situa-
tion, the decision rule would generate a self-refuting, or self-impugn-
ing, judgment.
There are, then, no sound objections to the veriiistic approach to
social epistemology. Given the intuitively powerful reasons in its
favor, it provides an attractive foundation for social epistemics.
NOTE
* Research leading to this paper was undertaken, in part, under a grant from the
National Science Foundation (SES-8204737) and during release time provided by the
Social and Behavioral Sciences Research Institute of the University of Arizona. For
helpful comments on earlier drafts, I wish to thank Fred Schmitt, Holly Smith, and
participants in the NEH Institute on the Theory of Knowledge (summer, 1986).
REFERENCES
Barnes, Barry: 1977, Interests and the Growth of Knowledge, Routledge and Kegan
Paul, London.
Barnes, Barry and David Bloor: 1982, 'Relativism, Rationalism and the Sociology of
Knowledge', in M. Hollis and S. Lukes (eds.), Rationality and Relativism, MIT Press,
Cambridge, Mass.
Bloor, David: 1973, 'Wittgenstein and Mannheim on the Sociology of Mathematics',
Studies in History and Philosophy of Science 4, 173-91.
Bloor, David: 1976, Knowledge and Social Imagery, Routledge and Kegan Paul,
London.
Boyd, Robert and Peter Richerson: 1985, Culture and the Evolutionary Process, Uni-
versity of Chicago Press, Chicago.
Campbell Donald: 1986, 'Science's Social System of Validity-Enhancing Collective
Belief Change and the Problems of the Social Sciences', in Donald Fiske and Richard
Shweder (eds.), Metatheory in Social Science, University of Chicago Press, Chicago.
Cantor, G. N.: 1975, 'A Critique of Shapin's Social Interpretation of the Edinburgh
Phrenology Debate', Annals of Science 33, 245-56.
Cavalli-Sforza, L. L. and M. W. Feldman: 1981, Cultural Transmission and Evolution,
Princeton University Press, Princeton.
Feyerabend, Paul: 1975, Against Method, New Left Books, London.
Firth, Roderick: 1981, 'Epistemic Merit, Intrinsic and Instrumental', in Proceedings and
Addresses of the American Philosophical Association 55, 5-23.
Forman, Paul: 1971, 'Weimar Culture, Causality and Quantum Theory, 1918-1927:
F O U N D A T I O N S OF S O C I A L E P I S T E M I C S 143
Adaptation by German Physicists and Mathematicians to a Hostile Intellectual

Environment', in R. McCormmach (ed.), Historical Studies in the Physical Sciences,
No. 3, University of Pennsylvania Press, Philadelphia.
Goldman, Alvin I.: 1980, 'The Internalist Conception of Justification', in P. French, T.
Uehling, and H. Wettstein (eds.), Midwest Studies in Philosophy V, University of
Minnesota Press, Minneapolis.
Goldman, Alvin I.: 1986, Epistemology and Cognition, Harvard University Press,
Cambridge, Mass.
Hardwig, John: 1985, 'Epistemic Dependence', The Journal of Philosophy 82, 335-49.
Hendry, J.: 1980, 'Weimar Culture and Quantum Causality', History of Science 18,
155-80.
Knorr-Cetina, Karin: 1981, The Manufacture of Knowledge, Pergamon Press, Oxford.
Kuhn, Thomas: 1970, The Structure of Scientific Revolutions, 2nd ed., University of
Chicago Press, Chicago.
Latour, Bruno and Steven Woolgar: 1979, Laboratory Life: The Social Construction of
Scientific Facts, Sage Publications, Beverly Hills.
Laudan, Larry: 1977, Progress and Its Problems, University of California Press, Ber-
keley.
Laudan, Larry: 1984, 'The Pseudo-Science of Science?', in James Brown (ed.), Scientific
Rationality: The Sociological Turn, D. Reidel, Dordrecht.
Lehrer, Keith and Carl Wagner: 1981, Rational Consensus in Science and Society, D.
Reidel, Dordrecht.
Merton, Robert: 1970, The Sociology of Science, University of Chicago Press, Chicago.
Newton-Smith, W. H.: 1981, The Rationality of Science, Routledge and Kegan Paul,
London.
Nozick, Robert: 1981, Philosophical Explanations, Harvard University Press, Cam-
bridge, Mass.
Popper, Karl: 1972, 'Epistemology Without a Knowing Subject', in Objective Know-
ledge, Oxford University Press, Oxford.
Putnam, Hilary: 1983, 'Why Reason Can't Be Naturalized', in Realism and Reason,
Cambridge University Press, Cambridge.
Rorty, Richard: 1979, Philosophy and the Mirror of Nature, Princeton University Press,
Princeton.
Rudwick, Martin: 1985, The Great Devonian Controversy, University of Chicago Press,
Chicago.
Sarkar, Husain: 1983, A Theory of Method, University of California Press, Berkeley.
Shapin, Steven: 1975, 'Phrenological Knowledge and the Social Structure of Early
Nineteenth-Century Edinburgh', Annals of Science 32, 219-43.
Shapin, Steven: 1982, 'History of Science and Its Sociological Reconstructions', History
of Science 20, 157-211.
Shapley, Lloyd and Bernard Grofman: 1984, 'Optimizing Group Judgmental Accuracy
in the Presence of Interdependencies', Public Choice 43, 329-43.
Stich, Stephen and Richard Nisbett: 1980, 'Justification and the Psychology of Human
Reasoning', Philosophy of Science 47, 188-202.
Welbourne, Michael: 1981, 'The Community of Knowledge', Philosophical Quarterly 31,
302-14.
144 A L V I N I. G O L D M A N
Whitley, Richard: 1984, The Intellectual and Social Organization of the Sciences, Oxford
University Press, Oxford.
Ziman, John: 1978, Reliable Knowledge, Cambridge University Press, Cambridge.
Department of Philosophy
University of Arizona
Tucson, Arizona 85721
U.S.A.

Goldman1987 Foundations of Social Epistemics

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Goldman1987 Foundations of Social Epistemics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Goldman1987 Foundations of Social Epistemics

Uploaded by

Copyright:

Available Formats

ALVIN I.

FOUNDATIONS OF SOCIAL EPISTEMICS*

ABSTRACT. A conception of social epistemology is articulated with links to studies of

Epistemology has historically been preoccupied with individual

epistemics: primary individual epistemics. That branch centers on

2. THE EVALUATIVE TASK OF SOCIAL EPISTEMICS

To begin thinking about social epistemology, consider the profile of

programme' in the sociology of science (Bloor, 1976; Barnes, 1977)?

research tradition, partly spurred and influenced by the competition of

Notice that nothing in my epistemological position forces me to

to the epistemology of science or scientists. It includes the epis-

3. RELATIVISM, CONSENSUALISM, AND EXPERTISM

In Section 2 I argued for an evaluative approach to social epis-

agreement in belief content whereas the first basis concerns agreement

Beliefs Social Social Practices or

relative to one community and unwarranted relative to another. But

dispute, in turn, is the generation of agreement or consensus among

encountered one of how to identify the relevant community. A cog-

Another difficulty for consensualism2 is that consensus per se is not

While expertism is more promising than consensualism, it has

reputational, expertism. As indicated, I believe that objective exper-

increasing the level of expertise. Third, expertism only gives advice to

where Pi represents the probability that member i has a true opinion.

group belief by weighting greater expertise more heavily. Shapley and

There is much more to be said in defense of veritism as the principal

between certain social variables and theory acceptance without ad-

This is not to deny that many an institution or practice can be

under the veritistic approach. Here I am not entirely decided. I reject

ture's scientific practices might be relatively powerful in the sense that

mission. There is no tight connection, at least no necessary con-

5. T H E SCOPE OF SOCIAL EPISTEMICS

Although I have said a few words about objects of evaluation for

*(6) Institutional rules and structures

outputs.) I acknowledge that the resulting demarcation between social

set of formal procedures, presided over by the judge. Juries then

other studies of science? Epistemology and philosophy of science have

Veritism stands in clear contrast to evaluative relativism, in particular,

power, fecundity, speed, and efficiency. However, each of these is

scientific assertions or beliefs, i.e., what the scientists agree to say or

7. THE PROBLEM OF APPLICABILITY

Let me first clarify what I mean by a 'criterion', or 'standard', of

a theory of justification, see Goldman 1986, Section 4.2.)

part of a determination procedure. But how much correctness, and

science? But which practices is it allowed to use? Here, again, it looks

weights for making collective judgments. Suppose this decision rule

Adaptation by German Physicists and Mathematicians to a Hostile Intellectual

You might also like