SANFORD GOLDBERG

COMMENTS ON MIRANDA FRICKER’S EPISTEMIC INJUSTICE

ABSTRACT

Miranda Fricker’s Epistemic Injustice is a wide-ranging and important book on a

much-neglected topic: the injustice involved in cases in which distrust arises out
of prejudice. Fricker has some important things to say about this sort of injustice:
its nature, how it arises, what sustains it, and the unhappy outcomes associated
with it for the victim and the society in which it takes place. In the course of
developing this account, Fricker also develops an account of the epistemology
of testimony. Focusing my attention on that account, my central claims are
two. First, at least some of Fricker’s arguments against existing (inferentialist
and non-inferentialist) views in the epistemology of testimony are less than fully
persuasive, and the (non-inferentialist) view she ends up endorsing is not all that
different from the views she criticizes. Second, her reasons for harboring doubts
regarding the role of a principle of default entitlement within a non-inferentialist
account are not persuasive. Neither of these claims affects the overall argument
Fricker is trying to run. Rather, they suggest that Fricker may have picked more
fights than she needed to in the epistemology of testimony. If so, we have reason
to detach Fricker’s important work on epistemic injustice from some of the
details of the story she tells regarding the epistemology of testimony.

Miranda Fricker’s new book, Epistemic Injustice, is a timely discussion of a topic

that has been neglected in the epistemological literature. Among other things, it
makes the case that there is a sort of distinctly epistemic injustice – an injustice
pertaining to a person in her capacity as a knower – that is done to a speaker when
the credibility of her say-so is downgraded by a hearer out of prejudice. This sort
of downgrading of a speaker’s credibility reflects the hearer’s prejudicial attitudes
towards one or more of the groups to which she takes the speaker to belong.
I think that Fricker’s main thesis on this score is correct and important, and that
the book’s extended argument on this score serves as a corrective to the more
narrowly-focused epistemological literature on the topic of testimony.
In my brief comments I will not be addressing the big picture claims Fricker is
making. For one thing, I find myself in broad agreement with most of what Fricker
has to say at the big-picture level regarding testimonial injustice. For another,
what I hope are my most useful comments are directed at the story Fricker offers
138 E P I S T E M E 2010 DOI: 10.3366/E1742360010000870
 COMMENTS ON MIRANDA FRICKER’S EPISTEMIC INJUSTICE
concerning the epistemology of testimony. I will be arguing for two related claims.
First, at least some of Fricker’s arguments against existing (inferentialist and non-
inferentialist) views in the epistemology of testimony are less than fully persuasive,
and the (non-inferentialist) view she ends up endorsing is not all that different from
the views she criticizes. Second, her reasons for harboring doubts regarding the
role of a principle of default entitlement within a non-inferentialist account are not
persuasive. Neither of these claims affects the overall argument Fricker is trying to
run. Rather, they suggest that Fricker may have picked more fights than she needed
to in the epistemology of testimony. If so, we have reason to detach Fricker’s
important work on epistemic injustice from some of the details of the story she
tells regarding the epistemology of testimony.
In the part of the book aimed at testimony, Fricker’s ambition is to establish
the existence of the phenomenon she calls ‘testimonial injustice’. The paradigmatic
case of this sort of injustice is one she calls an identity-prejudicial credibility deficit, which
she characterizes as “a credibility deficit owing to identity prejudice in the hearer”
(28). All hearers assess testimony for credibility; the identity-prejudicial credibility
deficit arises when the hearer downgrades her assessment of the credibility of a
piece of testimony, where this downgrading reflects the hearer’s own prejudice
against a social group in which she places the speaker. Fricker argues persuasively
that such prejudice affects the speaker in her capacity as a knower: among other
things, having one’s credibility systematically called into question will corrode one’s
own self-confidence (and so one’s disposition to self-trust), and it can (and typically
will) have a negative affect on one’s role in projects involving communication or
social coordination. Fricker advances a solid case for thinking that this sort of
injustice is a significant feature of the lives of those who are (or are perceived to
be) members of social groups that are themselves the target of prejudice.
In the course of making her case for the existence and significance of this sort of
testimonial injustice, Fricker addresses the two main positions in the epistemology
of testimony. She introduces this topic as follows:
Accounts of the epistemology of testimony can be seen as falling into two broad
varieties: inferentialist and non-inferentialist. There is room for a diversity of views . . .
but a key motivation for any will be the author’s inclination vis-à-vis inferentialist and
non-inferentialist pictures of the obligations upon a hearer if she is to gain knowledge
from her interlocutor. (61)
She goes on to summarize the two positions as follows:
Roughly speaking, then, in the epistemology of testimony it can seem as though we
must plump for one of two epistemological stories. One story presents the hearer as
gaining knowledge only if he rehearses an appropriate inference. The other story seems
to present the hearer as gaining knowledge by way of one or another default of uncritical
receptivity such that he is entitled to accept what he is told without exercising any critical
capacity. (62)
Fricker goes on to argue that, while the phenomenology of acceptance of
another’s say-so would appear to support the latter (“non-inferentialist”) view,
E P I S T E M E 2010 139
Sanford Goldberg
considerations of hearer responsibilities in the consumption of testimony appear
to favor the former (“inferentialist”) view. She ends up trying to defend a version
of non-inferentialism that nevertheless imposes a substantial requirement on hearer
responsibility.
I have some misgivings about this part of Fricker’s discussion. Some of my
misgivings pertain to the terms in which she sets up the debate. One misgiving
I have regarding the way Fricker sets up the debate is that she appears to be
running together two distinct issues. More specifically, it is not entirely clear
whether she regards the dispute as concerning the conditions on acquiring knowledge
through testimony, or rather the conditions on being justified in accepting (entitled to
accept) what one is told. Although these are separate questions, they appear to
be run together in her discussion. Thus the non-inferentialist view is described as
“present[ing] the hearer as gaining knowledge by way of one or another default of
uncritical receptivity such that he is entitled to accept what he is told without exercising
any critical capacity.” And later she describes non-inferentialism as a view about
“the obligations on a hearer if she is to gain knowledge from her interlocutor” (61),
and as a view that postulates a kind of “default of credulity” in its account of
“empirical justification” (62). Since the question about the condition on acquiring
knowledge from another’s say-so is distinct from the question about the condition
on being entitled to accept (justified in accepting) another’s say-so, one ought to
be clear about which view one is advancing about which epistemic status. This is
especially important given that some authors in the literature (e.g., Faulkner 2000)
have argued for what they call ‘mixed’ or ‘hybrid’ views in the epistemology of
testimony.
A second concern I have with the way Fricker sets up the debate concerns
how, if at all, Fricker’s discussion connects to the main current discussions in
the epistemology of testimony literature.1 The main current discussion in the
literature concerns the issue of reductionism. Here the question is this: does the
sort of justification enjoyed by testimonial belief reduce to other, more basic
sorts of justification, or is it sui generis? Such an issue is often addressed by way
of asking whether the doxastic justification of testimonial belief requires that the
hearer have positive reasons to think that the piece of testimony she observed is
credible, as so-called reductionists maintain, or whether it suffices that there be no
reasons to think that the testimony is not credible, as so-called anti-reductionists
maintain. To be sure, the issue of the doxastic justification of testimonial belief
ought to be distinguished in its turn from the issue of a subject’s entitlement
(or “epistemic right”) to accept a piece of testimony: thus one might think that
positive reasons are needed for a subject to be entitled to accept a piece of
testimony, yet deny that doxastic justification supervenes on these reasons. (Such,
so far as I can tell, is the view developed in Faulkner (2000), although what I call
‘doxastic justification’ he calls ‘warrant’.) It is not obvious how Fricker’s discussion
of the inferentialism/non-inferentialism debate relates to the reductionism/anti-
reductionism discussion. From her discussion it would seem that she regards
140 E P I S T E M E 2010
 COMMENTS ON MIRANDA FRICKER’S EPISTEMIC INJUSTICE
these two debates as two sides of the same coin – that all and only reductionists
are inferentialists, and that all and only anti-reductionists are non-inferentialists.
But the parallelism is not obvious. Consider the possibility of an inferentialist anti-
reductionism, according to which a testimonial belief is doxastically justified only
if one forms one’s belief on the basis of a defensible inference, but where the
inference in question does not cite positive reasons for regarding the testimony
as credible. (Perhaps the inference is from S said that p to p; or from S said that p and I
have no reasons not to trust her to p.) Such a view holds that a hearer’s recognition that
S said that p counts as a reason for the hearer to believe that p, but that the status
of S’s say-so as such a reason does not depend on the hearer’s having any positive
reasons to regard that say-so as credible. The very possibility of such a view, which
appears closely related to the position defended in McDowell (1994), suggests that
the two debates do not line up quite so cleanly.
These worries about the way Fricker frames the debate are related to other, more
substantial worries I have about her position. She aims to develop her own non-
inferentialist position in reaction to what she sees as the shortfalls of inferentialism
as well as the difficulties facing extant versions of non-inferentialism; I worry that
at least some of her criticisms of earlier views are unwarranted, and that as a result
her own non-inferentialist view suffers from a relative lack of motivation.
I begin first with her generic criticism of inferentialism. (I am not sure how
much stock Fricker puts in this criticism, but given that she mentions it often,
I think it worth raising.) Fricker often suggests that inferentialist positions run afoul
of ‘the phenomenology’ involved in the accepting of testimony (61–4 and 80). By
this she appears to have in mind that we typically accept testimony in a more or less
‘direct’ way, one that does not involve reflecting on our reasons for doing so. Since
inferentialism, on her presentation of that view, “presents the hearer as gaining
knowledge only if he rehearses an appropriate inference” (62; italics added), it would
appear that inferentialism is in trouble.
This criticism does not appear to pose a very serious threat to any plausible
version of inferentialism. After all, inferentialism is a thesis in epistemology, not
psychology: its claim is that what justifies a hearer in accepting a piece of testimony
are her reasons for accepting it. (Or perhaps inferentialism is a thesis about the
conditions on the entitled acceptance of testimony; as I mentioned above, it’s not
clear precisely what inferentialist view Fricker has as her target.) The fact that a
hearer accepts a piece of testimony without having those reasons consciously in
mind as she does so is something that any inferentialist should gladly concede.
The inferentialist merely insists, as a condition on, for example, the hearer’s having
formed a doxastically justified testimonial belief, that those reasons be accessible
to the hearer, in the sense that she (the hearer) could cite them as part of a would-
be justifying inference, if the need to do so were to arise.2 My own opinion is that,
since inferentialism is an epistemic thesis, it should not be saddled with a view about
the psychology of acceptance. Since the phenomenology of testimonial uptake is a
psychological matter, I conclude that the objection from phenomenology would
E P I S T E M E 2010 141
Sanford Goldberg
appear to be in trouble from the start. This is not to say that inferentialism is
ultimately acceptable, only that it cannot be shown to be unacceptable by appeal
to claims about the non-inferential phenomenology of acceptance. (I will return to
this below.)
Next I move to one of Fricker’s criticisms against extant non-inferentialist
positions. The criticism in question is directed against those versions of non-
inferentialism that are formulated in terms of a principle of (defeasible but) default
entitlement to accept the word of another. Fricker begins by acknowledging that
such a principle can be advanced on the basis of empirical considerations – those
having to do with our de facto tendencies towards trust on the side of the hearers,
and towards truthfulness on the sides of the speaker – or on the basis of a priori
considerations – those having to do with, for example, the role of uncritically
accepted testimony in the learning of a first language,3 or else the role of rational
sources in the pursuit of truth.4 Fricker’s generic criticism is directed against both
sorts of default-entitlement view. She writes, “Both empirical and a priori default
accounts seem to assume that if no inference is made by the hearer, then her
reception of her interlocutor’s word must be uncritical” (66). She repeats this
criticism in several other places. She writes, “A general default of accepting the
word of others critically unmediated would be justificationally lax” (65). This sort
of criticism is sprinkled throughout her discussion: she describes the default-
entitlement view alternatively as
present[ing] the hearer as gaining knowledge by way of one or another default of
uncritical receptivity such that he is entitled to accept what he is told without exercising
any critical capacity;
effectively represent[ing] the hearer’s critical faculties as in snooze mode at the very moment
when she takes in knowledge from an interlocutor;
represent[ing] the hearer as having his critical faculties in snooze mode unless and until he is
alerted to some cue for doubt that flicks a switch to reawaken his critical consciousness
(62, 65, 66; italics added).
Thus it is clear that she is unsympathetic to a default-entitlement view, on the
grounds that it is “justificationally lax” for allowing the hearer’s critical faculties
to be “in snooze mode”.
It is worth noting that there are various default-entitlement views, of varying
strengths, and that this “snooze mode” objection would appear at best to be
effective only against the very strongest of these views.5 Thus consider the
following three distinct views, all of which appeal to a default-entitlement to accept
testimony:
(1) For all speakers S, testimonies that p, and hearers H, H is entitled to accept
S’s testimony that p, provided that there are no relevant defeaters (against p
or against the credibility of S’s say-so).
(2) For all speakers S, testimonies that p, and hearers H, H is entitled to accept
S’s testimony that p, provided that (a) there are no relevant defeaters, and
142 E P I S T E M E 2010
 COMMENTS ON MIRANDA FRICKER’S EPISTEMIC INJUSTICE
(b) H would likely have picked up on any relevant defeaters, had there
been any.
(3) For all speakers S, testimonies that p, and hearers H, H is entitled to accept
S’s testimony that p, provided that (a) there are no relevant defeaters, and (b)
H recognizes that there are no relevant defeaters.

Of these (1) is the most permissive view: any case that satisfies the conditions
provided in (2) or (3) will satisfy the conditions provided in (1), but not vice versa.
And on the plausible assumptions that ‘recognizes’ entails ‘knows’ and that there is
a safety condition on ‘knows’, (2) is strictly more permissive than (3). Consider in
this light the “snooze mode” objection. Perhaps this objection scores points against
(1): since (1) merely requires the de facto absence of relevant defeaters in order
for the hearer’s acceptance to be entitled, it appears to sanction entitled acceptance
even in cases in which the hearer is credulous in her acceptance (her critical faculties
are in “snooze mode”). But it is far from obvious that the same point can be made
against proposals (2) or (3). After all, neither (2) nor (3) sanctions an acceptance of
testimony unless the consumer of testimony was properly sensitive to the possible
presence of relevant defeaters. How can the consumer be sensitive in that way
if her critical faculties are in “snooze mode”? After all, relevant defeaters do not
simply “announce themselves” – one has to be relatively vigilant, “on the lookout”
for their presence. It is true that the operation of the faculties that render a hearer
appropriately sensitive might not be accessible to the hearer herself: she might not
be able to tell why it is that a piece of testimony “just doesn’t seem right” to her.
But even in such cases, it would be false to say that the hearer’s critical faculties are
“in snooze mode”.
This is important because many empirically-minded versions of the default-
entitlement view have explicitly discussed the need for subpersonal, automatic
processes in the hearer to be sensitive to the presence of indications of deceit or
incompetence in the speaker. (Indeed, Fricker herself cites one such discussion
in Coady (1992, 47).6 ) As versions of a default-entitlement view, such empirically-
minded accounts are closer to (2) or (3) than to (1). But perhaps Fricker intends her
“snooze mode” objection to bear against those versions of the default-entitlement
view that seek to ground the entitlement on a priori considerations. Here a view
like Tyler Burge’s (1992) a priori defense of a default entitlement principle comes
to mind. But while Fricker does devote a good deal of time to criticizing Burge’s
version of a default-entitlement principle, it is unclear to me whether her “snooze
mode” objection scores any points against Burge’s view.
Fricker correctly notes that Burge does not spend much time discussing “the
remaining obligations on the hearer in terms of his on-stage display of sensitivity
to whether or not the default holds in any particular case.” (Fricker 2007, 68)
Nevertheless, it is worth pointing out that Burge does have much to say that
bears indirectly on this question. His 1992 discussion of preservative memory is
instructive, since it is on an analogy with the epistemic role of preservative memory
E P I S T E M E 2010 143
Sanford Goldberg
(in the acquisition of knowledge through proofs) that he models the epistemic
role of testimony (in the acquisition of knowledge through accepting the word of
another). He highlights the importance of properly-functioning cognitive processes
at play in the background when he writes,
Even in empirical reasoning, memory has a purely preservative function that does not
contribute to the force of the justification, but simply helps assure the proper working of
cognitive capacities over time. (494; italics added)
What is more, Burge explicitly acknowledges that when one’s cognitive capacities
do not work properly, this can affect one’s entitlement in the case at hand.7
This is seen in his acknowledgement of one sort of condition under which one’s
entitlement to rely on preservative memory is defeated, and so will have to be
defended by further reasoning (perhaps involving substantive memory). He writes,
When a purely preservative instance [of memory] is reasonably challenged, because
memory has proved unreliable, one may have to rely on substantive memory. (494; italics
added)
The important point here is that in the face of (the allegation of) relevant failures
of memory, the capacity of preservative memory to preserve one’s entitlement
to a memory-sustained belief is called into question, with the result that one
might have to rely on “substantive memory” in order to (defeat the defeater and)
recover an entitlement for the belief in question. More generally, Burge’s position
acknowledges something to the following effect: when a given cognitive system is
operating in what for it are normal conditions, yet it is unreliable in its production or
sustainment of belief, the default entitlement associated with the subject’s reliance
on that system is defeated.8 On the twin assumptions, first, that normal conditions
for the cognitive system(s) implicated in the reception of testimony are conditions
involving the regular run of speakers, and second, that such systems would be
unreliable under these conditions if their critical faculties were in “snooze mode,”
this would suggest that Burge’s version of the default-entitlement view, like that in
Coady (1992), is closer to (2) or to (3) than it is to (1). In that case, Fricker’s “snooze
mode” objection fails to score points.
I move on now to a third and final point. In addition to my concerns over the
terms in which she sets up the debate, and my concerns regarding her criticisms
of extant versions of inferentialism and non-inferentialism, I would like to raise
the issue regarding the proper category in which to place Fricker’s views in the
epistemology of testimony. In particular, while she advertises her position as one
that is non-inferentialist, it is unclear to me whether Fricker’s position should so
count. Although I raise this as an issue rather than a concern – it has to do in the
first instance with proper taxonomy among positions within the epistemology of
testimony – I think that the issue does raise some questions about the nature of the
position she aims to occupy.
Fricker’s overall position appears to be this. To be justified in accepting (entitled
to accept?) the word of another, one must critically assess the credibility of the
144 E P I S T E M E 2010
 COMMENTS ON MIRANDA FRICKER’S EPISTEMIC INJUSTICE
proffered testimony. However, one can be critical in the relevant sense merely in
virtue of being (counterfactually) sensitive to the presence of relevant defeaters,
in the sense that, were such defeaters to be present, one would register this
and would begin to employ more consciously critical capacities in assessing the
testimony’s credibility. Fricker’s model of all of this, in what she terms her ‘virtue
epistemology account’ of the epistemology of testimony, is to treat our assessments
of a testimony’s credibility as a species of perception and perceptual judgment. On
her view, a hearer perceives that a given piece of testimony is (or is not) credible, and
this perception itself is the manifestation of the hearer’s critical assessment of the
testimony.
There is much to admire in this position. (I return to this below.) But it is worth
pointing out that even if Fricker is correct in all that she says here, she still has
not furnished us with a version of non-inferentialism. To see this, consider the
content of a perceptual judgment regarding the credibility of a piece of testimony
(or a testifier). The content of such a judgment would be something like this:
this piece of testimony that p is (is not) credible. (Alternatively, it might be: this testifier
is (is not) credible, or perhaps this testifier is (is not) credible on this occasion of testifying.
I disregard these complications in what follows.) Note, though, that it is one thing
to say that a hearer perceives that a given piece of testimony that p is credible;
it is another to say that the hearer is justified in accepting the testimony that p.
How does the former relate to the latter? One natural suggestion is that the hearer’s
perceiving that the testifier/testimony that p is credible constitutes a reason that justifies
her in accepting the testifier’s testimony that p (and, by extension, justifies the
hearer in coming to believe that p). But this would be a version of inferentialism,
which Fricker is at pains to reject. So assuming that a hearer’s perceiving that this
testifier/piece of testimony that p is credible is not a reason on the basis of which the
hearer accepts (and is justified in accepting) the testifier’s testimony, how then
does it relate to the hearer’s acceptance of (and her justification in accepting) the
testimony?
It is in answer to something like this question that Fricker cites a passage from
Robert Audi. Speaking of a situation of an extended conversation with a stranger
in which the hearer slowly begins to regard the speaker as trustworthy, and so
slowly begins to develop the disposition to take the speaker at her word, Audi
wonders what might explain this slow shift in the hearer’s attitudes and acceptance
dispositions. He writes,
One possibility is an unconscious inference, say from the general credibility of [the
stranger’s] account to the conclusion that this proposition, as an essential part of it,
is true. But perhaps the cognitive influence of my standing beliefs, such as a newly formed
belief that she is credible, need not proceed through an inference from them. Another
possible explanation is more moderate: even apart from my forming beliefs about her
credibility, her eventually becoming, in my eyes, a quite credible person, can in some
fairly direct way produce in me a general disposition to believe her (Audi 1998, 133, as
cited in Fricker 2007, 70; italics in original).

E P I S T E M E 2010 145
Sanford Goldberg
It is noteworthy that Audi here is addressing what appears to be a psychological
question: what explains the hearer’s changing reactions to the speaker? There is
something quite attractive about Audi’s answer to this question: we need not advert
to any inference, even to unconscious inference, to explain the hearer’s changing
reactions towards the speaker over the course of the conversation. But – and
this is very important – it is another question entirely how best to account for
the epistemological dimension of the exchange. Even if we accept Audi’s “more
moderate” explanation for the psychological question, it remains open to the
epistemologist to hold that it is in virtue of the hearer’s reasons for trusting the
speaker that she (the hearer) is justified (or not, as the case may be) in accepting the
speaker’s say-so. Indeed, this is precisely the sort of position Audi himself appears
to have settled on: a non-inferentialist account of the psychological process of
acceptance (what Fricker sometimes calls the “phenomenology of acceptance”),
but an inferentialist account of the epistemology of acceptance (holding that a
speaker is not justified in accepting a piece of testimony unless she has adequate
undefeated reasons for acceptance – reasons that the hearer could employ in an
inference, whether or not she actually carried out such an inference).9 Thus the
appeal to the Audi quote above leaves a key question unaddressed: what does
Fricker take to be the epistemological connection between a hearer’s perceptual
judgment that a given testifier/piece of testimony is trustworthy, on the one hand,
and the hearer’s acceptance of the testimony, on the other?
Although Fricker is less than fully explicit on this matter, here is one possible
way she might approach this question. A subject’s reception of information
through testimony, like her reception of information through perception, involves
her reliance on a cognitive system that is sensitive to the presence of defeaters.
Precisely for this reason, beliefs formed by the subject through her reliance on
that system are reliably formed. This reliability is not perfect, of course. But its
degree of reliability passes a threshold required for the relevant epistemic status
(justification or entitlement). In the testimony case, the reliability in question is
attained in virtue of the subject’s perceptual capacity for non-inferential uptake
of conditions that predict competence and sincerity in the speaker. (It is this that
constitutes her sensitivity to the reliability of the testimony). And the fact that the
hearer is sensitive in this way to the speaker’s reliability – together (perhaps) with
the fact that the cognitive system that underwrites this sensitivity is a natural part of
human psychology – entitles the subject to rely on that system in belief-formation.
The entitlement is presumptive: it can be defeated, as when in a particular case
(involving “normal conditions”) the subject fails to be sensitive to the presence of
relevant defeaters, or when in a particular case (involving “normal conditions”) the
system in question is not properly functioning (in the sense that had there been
defeaters, she would have missed them etc.).
Notice, though, that the foregoing is a version of the “default entitlement”
formulation of the non-inferentialist view, and Fricker claims to reject this kind
of non-inferentialism. We must then ask: are there other ways Fricker’s ‘virtue
146 E P I S T E M E 2010
 COMMENTS ON MIRANDA FRICKER’S EPISTEMIC INJUSTICE
epistemology account’ might endorse non-inferentialism yet manage to avoid
having to postulate a principle of default entitlement? Here is another possibility.
The mature hearer perceives that a given testifier/piece of testimony that p is
credible. But (this line of thought continues) to perceive that a given piece of
testimony that p is credible is also (at least in some cases?) to be in a position
to perceive that p itself, albeit indirectly, through the testimony.10 Then, insofar as
one’s justification for one’s perceptual beliefs is non-inferential, one’s justification
for one’s testimonial beliefs is correspondingly non-inferential (testimonial belief
being a case of extended perceptual belief). By extension, then, we might say that
one is entitled to accept (is justified in accepting) testimony that p whenever the
testimony enables one to perceive that p.
But even if Fricker were to adopt this sort of model, it is by no means clear that
this would allow her to avoid the appeal to a principle of default entitlement. Let
us use ‘telling’ to designate happy cases of testimony, those in which the testimony
is both reliable and veridical. Then the claim on the table is that one is entitled
to accept (is justified in accepting) testimony that p whenever (i) the testimony
that p amounts to a telling that p, and (ii) one’s understanding of the testimony
and one’s perception of it as reliable entitle one to assume (i). (Condition (ii) is a
condition on hearer responsibility; it is needed in order to avoid the counterintuitive
implication that any hearer, no matter how gullible or lucky in recovering the
attested proposition, is entitled to accept any testimony that is both reliable and
veridical.) Now take a case in which a hearer confronts a piece of testimony she
regards as both true and reliable. With respect to any such case, we can raise the
question whether what the hearer takes to be a case of true and reliable testimony
that p really is such. After all, not every case that a hearer regards as credible
testimony that p turns out to be such. For this very reason the hearer’s assumption
that (i) holds, needed (according to this model) for the hearer to be justified in
accepting the testimony, is itself fallible. What, then, should we say about the
hearer’s entitlement to such a fallible assumption?11 My sense is that the hearer is
default-entitled to rely on it, with defeat of this entitlement contingent on positive
reasons for thinking that (i) fails to hold in a given case. Notice, though, that
if the assumption does have this status, then the present attempt to model the
epistemology of testimony is not one that would enable Fricker to avoid the need
to appeal to a principle of default entitlement.
It appears, then, that two of Fricker’s commitments are in tension with one
another: one is her commitment to non-inferentialism, and the other is her
commitment to avoiding the appeal to a principle of default entitlement. If these
two commitments should turn out to reveal an incompatibility – something I do
not claim to have shown here – then Fricker would be faced with a dilemma. She
would then have to acknowledge one of two things: either that her account is a
version of inferentialism after all; or else that her non-inferentialist account makes
an ineliminable appeal to a principle of default entitlement (her objections to such
principles notwithstanding).
E P I S T E M E 2010 147
Sanford Goldberg
I want to conclude by emphasizing that I do not intend this as a hostile dilemma;
my sense is that either way, she can claim to have advanced the debate. If she
opts for the former horn of the dilemma, she can claim to have advanced the
case for inferentialism by blunting the main objection against inferentialist views.
To the criticism of inferentialism that it over-intellectualizes what is involved in
being justified in accepting the word of another, Fricker could respond that her
version does no such thing, but instead regards our reasons for accepting testimony
to be themselves nothing beyond perceiving that the testimony is credible. On
the other hand, if she opts for the latter horn, endorsing a default entitlement
version of non-inferentialism, she might claim to have advanced the debate by
filling in the details of the background cognitive processes in virtue of which
we have such a default entitlement to accept the word of others. I would guess
that she would opt for the latter: with the exception of its appeal to a default
entitlement principle, this option conforms closely to her characterization of her
own position. At the same time, I would think that she should be more attracted
to the inferentialist picture (at least as it is presented here) than she appears to
be. After all, as presented above, the inferentialist picture merely requires that the
hearer possess reasons for acceptance – it does not require that she consciously
have them in mind as she considers whether to accept the testimony. And it would
appear that Fricker’s position can regard ordinary hearers as satisfying this demand.
After all, Fricker writes that “. . . in exercising a sensitivity to the status of the
default of acceptance, the hearer can only be exercising a sensitivity to . . . the balance
of reasons for and against accepting what she is told.” (69; italics added) Thus it
remains unclear to me whether Fricker’s settled opinion is best and most charitably
construed as a development of non-inferentialism, or as a defense of a novel form
of inferentialism. My only claim here is that she has not given an argument that is
decisive between these two options.12

REFERENCES

Audi, R. 1997. “The Place of Testimony in the Fabric of Knowledge and Justification.”
American Philosophical Quarterly 34(4): 405–22.
Audi, R. 1998. Epistemology: A Contemporary Introduction to the Theory of Knowledge. London:
Routledge.
Burge, T. 1992. “Content Preservation.” Philosophical Review 102(4): 457–88.
Coady, C. A. J. 1992. Testimony: A Philosophical Study. Oxford: Oxford University Press.
Davidson, D. 1984. Inquiries into Truth and Interpretation. Oxford: Oxford University Press.
Faulkner, P. 2000. “The Social Character of Testimonial Knowledge.” Journal of Philosophy
97(11): 581–601.
Fricker, M. 2007. Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford
University Press.
Goldberg, S. 2007. Anti-Individualism: Mind and Language, Knowledge and Justification.
Cambridge: Cambridge University Press.

148 E P I S T E M E 2010
 COMMENTS ON MIRANDA FRICKER’S EPISTEMIC INJUSTICE
Goldberg, S. and D. Henderson. 2006. “Monitoring and Anti-Reductionism in the
Epistemology of Testimony.” Philosophy and Phenomenological Research 72(3): 576–93.
McDowell, J. 1994. “Knowledge by Hearsay.” In B. M. Matilal and A. Chakrabarti (eds.),
Knowing From Words: Western and Indian Philosophical Analysis of Understanding and Testimony,
pp. 195–224. Dordrecht: Kluwer.
Millikan, R. 2000. On Clear and Confused Ideas: An Essay about Substance Concepts. Cambridge:
Cambridge University Press.
Stevenson, L. 1993. “Why Believe What People Say?” Synthese 94: 429–51.

NOTES

1 I thank John Greco for this point.

2 In fact, as I will go on to argue below, something like this position – a non-inferentialist
account of the psychology of testimonial uptake, together with an inferentialist account
of the epistemology of acceptance – is occupied by Audi (1997), and Fricker’s own
account comes very close as well. See my comments in the last paragraph of this
commentary.
3 See, for example, Coady (1992) or Stevenson (1993), both of which take inspiration
from comments sprinkled throughout Davidson (1984).
4 See Burge (1992).
5 I thank an anonymous referee from this journal for indicating the need to make this
point, and for suggesting the various versions of the default-entitlement view.
6 See also Goldberg and Henderson (2006) and Goldberg (2007) for further discussion.
7 And to be fair to Burge, it is important to frame the discussion at the level of entitlement
rather than justification, since he takes great pains to distinguish them. He writes, “The
distinction between justification and entitlement is this: Although both have positive
force in rationally supporting a propositional attitude or cognitive practice, and in
constituting an epistemic right to it, entitlements are epistemic rights or warrants that
need not be understood by or even accessible to the subject.” He goes on to make
clear that his claim is one at the level of entitlement: “We are entitled to rely, other
things equal, on perception, memory, deductive and inductive reasoning, and on – I will
claim – the word of others.” (458)
8 This is not to say that Burge’s epistemology is reliabilist; it isn’t. It is only to say that
Burge allows that, at least in conditions where a properly functioning belief-forming
process could be expected to be reliable, de facto unreliability in the methods used is
something that can defeat an entitlement.
9 Some textual evidence, from Audi (1997) (italics added):
. . . [A]t least normally, a belief based on testimony is thereby justified (i.e., counts
as testimonially justified) provided the believer has overall justification for taking the
attester to be credible regarding the proposition in question. Having this justification
implies a capacity for inference, say about the attester’s reliability, but not making an actual
inference, conscious or unconscious. (412)
I cannot acquire justification for believing something on the basis of testimony unless
I have some degree of justification for believing that the attester is credible, as well
as for certain other propositions, such as that I heard the testimony correctly. This
E P I S T E M E 2010 149
Sanford Goldberg
justification cannot come entirely from testimony . . . . Other grounds of justification,
such as perception or memory, must at least tacitly cooperate. But their cooperation
can be justificational without being inferential: they need not produce in me beliefs
of premises from which I infer that the attester is credible; they simply give me a
justification for framing such premises if I need them. (413)
(In this quote just present, Audi’s use of ‘inferential’ is psychological : in claiming that the
cooperation of perception or memory can be ‘justificational without being inferential,’
he means to say that the hearer does not need to engage, consciously or unconsciously,
in the drawing of any inference, although the fact that they justify implies that they
furnish the subject with the materials from which she could devise premises in an
inference, were such needed by her.) And finally this:
Often, when we hear people attesting to various things, we just believe these
things, non-inferentially and even unreservedly. But this natural psychological process
yields knowledge and justification only when certain epistemic conditions are met: there must
be grounds, from another source, for knowledge and justification, even if there
need be no knowledge or justified beliefs of the propositions warranted by these
grounds. In the case of testimonially based knowledge, there must be knowledge,
even if not necessarily justification, on the part of the attester, whereas in the
case of testimonially based justification there must be justification, even if not
knowledge, on the part of the recipient. The first requirement concerns the attester’s
epistemic situation with respect to the proposition attested to; the second concerns
the recipient’s epistemic situation with respect to the attester, or the proposition,
or both. Together, the requirements indicate how, although, psychologically speaking,
testimony is a source of basic beliefs, it is not, epistemically speaking, a basic source of knowledge or
justification. (414)
10 For a version of the view that the perception of testimony that p is an indirect way to
perceive that p, see Millikan (2000), and also, arguably, McDowell (1994).
11 Those who favor McDowell’s approach to epistemology might think to challenge
the legitimacy of this question, as betraying a desire for a “common factor” between
the veridical case (telling that p) from the illusory case (testimony that p that does
not amount to a telling that p). Although I have my concerns about McDowell’s
“disjunctivist” epistemology, I do not have time to go into this here. Suffice it to say
that the argument I am discussing in the text ought to convince those not convinced by
McDowell-style disjunctivist epistemology.
12 I would like to thank Miranda Fricker, for the gracious way with which she corrected
my misunderstandings of her position (any remaining misunderstanding is entirely my
fault); John Greco, for discussions on these topics; and an anonymous referee for this
journal, for helpful comments on an earlier draft.

Sanford Goldberg is Professor and Chair of the Philosophy Department at

Northwestern University. He works in the areas of epistemology, philosophy of
mind, and philosophy of language. His recent work includes Anti-Individualism
(Cambridge University Press, 2007) and Relying on Others: An Essay in Epistemology
(Oxford University Press, forthcoming).

150 E P I S T E M E 2010
Copyright of Episteme is the property of Edinburgh University Press and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.