Open Journal of Philosophy
2012. Vol.2, No.1, 1-9
Published Online February 2012 in SciRes (http://www.SciRP.org/journal/ojpp) http://dx.doi.org/10.4236/ojpp.2012.21001
Mental Content Externalism and Social Understanding
Halvor Nordby1,2
1Faculty of Education and Social Work, Lillehammer University College, Lillehammer, Norway
2Department of Health and Society, The University of Oslo, Oslo, Norway
Received November 17th, 2011; revised December 20th, 2011; accepted December 28th, 2011
Tyler Burge has in many writings distinguished between 1) mental content externalism based on incorrect
understanding and 2) mental content externalism based on partial but not incorrect understanding. Both 1)
and 2) have far-reaching implications for analyses of communication and concept possession in various
expert-layperson relations, but Burge and his critics have mainly focused on 1). This article first argues
that 2) escapes the most influential objection to 1). I then raise an objection against Burge’s argument for
2). The objection focuses on Burge’s claim that a person with a partial understanding of a term in our
community expresses our standard concept because he is willing to defer to our standard understanding,
while his “Putnamian” twin in a counterfactual community does not. The problem with Burge’s argument
for this claim is that he does not consider the possibility that the person in our community and the twin
would defer to the same understanding if they were presented with the same alternatives. Drawing from
widespread dispositional assumptions about meaning, I argue that Burge must accept that they express the
same concept if they would defer to the same understanding. The article closes with an examination on
various ways the externalist may attempt to avoid this problem and concludes that none of them succeeds.
Keywords: Content Externalism; Concept Possession; Partial Understanding; Linguistic Meaning;
Conceptual Role Semantics
Introduction
Tyler Burge’s influential arguments for social externalism
about mental content have their basis in thought experiments
involving three steps. In the first step Burge claims that a per-
son in our linguistic community has beliefs involving our stan-
dard concept despite having an incomplete understanding of the
expression he uses to express that concept. In the second step
Burge imagines that the person in our community has a “Put-
namiam”, non-intentionally specified identical twin in a coun-
terfactual community where the same expression has a standard
meaning that does not correspond to the correct understanding
in our community.1 Except for this difference we are to con-
ceive of the counterfactual community as identical to ours. The
third step is the externalistic conclusion: because of the differ-
ence between the two linguistic communities, the person in our
community and his twin do not associate the same concept with
the expression (Burge, 1979: pp. 77-79).
In their discussions on social externalism, Burge and his
critics have paid most attention to the first step of the thought
experiment, particularly in connection with Burge’s well-
known “arthritis” case involving a patient who mistakenly
thinks that “arthritis” applies not only to inflammations of the
joints, but also to inflammations elsewhere that have spread
from the joints (Pessin & Goldbe rg, 1996; Schantz, 2004). M an y
of Burge’s critics have argued that the patient’s concept is not
arthritis but rather an alternative concept with an extension that
directly matches the patient’s incorrect understanding of “ar-
thritis”, i.e. a concept that applies to inflammations that have
spread from joints as well as inflammations of the joints (Wood-
field, 1982; Loar, 1987; Crane, 1991; Bach, 1994; Nordby,
2004).
This objection centers particularly on incorrect understanding.
Burge, however, also focuses on partial but only incorrect un-
derstanding, which he explains as follows:
One need not rely on an underlying misconception in the
thought experiment. One may pick a case in which the
subject only partially understands an expression. He may
apply it firmly and correctly in a range of cases, but be
unclear or agnostic about certain of its applications or im-
plications which, in fact, are fully established in common
practice (Burge, 1979: p. 82).
In such cases of partial understanding there is no “underlying
misconception”, and so the idea about an alternative concept
that directly matches an incorrect understanding cannot prevail.
Burge’s analyses of partial understanding are not as developed
as his analyses of incorrect understanding, but they deserve
attention both because they are invulnerable to the standard
objection to social externalism based on incorrect understand-
ing, and because it is reasonable to assume that partial under-
standing is a more widespread phenomenon than incorrect un-
derstanding. In fact, in many areas of discourse it seems to be
the rule and not the exception that a partial understanding is
involved.2 Consider, for instance, doctor-patient interaction and
the fact that patients normally have an incomplete understand-
1Burge’s idea of a non-intentionally specified identical twin is adopted from
Putnam’s “The Meaning of Meaning” (1975). Unlike Putnam Burge makes
it clear that his thought ex p eriment f o cuses on me n tal content. Putnam has in
more recent writings made it clear that he thinks content externalism in-
cludes mental content (Putnam 1996).
2The latter cl aim is reasonab le for the followi ng reason: normally , when we
have beliefs about a term such that we do not have a complete understanding
then th ese beliefs are f ormed on th e basis of co rrect informat ion from com-
petent speakers or other reliable sources like a dictionary. None of the ar-
guments in this article will depend on the plausibility of this empirical claim,
but I menti on it since it i s releva nt for the w ider sig nifica nce of the disc ussi on.

H. NORDBY
ing of medical terminology. Does this mean that they possess
alternative concepts, or do they have the same concepts as the
doctors they encounter? Externalism based on partial under-
standing has considerable implications for how we should con-
ceive of communication and concept possession in this and
other forms of interaction between professional experts and
laypersons.3
This article’s central argument is that Burge’s argument for
social externalism based on partial understanding fails to be
convincing even though it avoids the most influential objection
to social externalism based on incorrect understanding. The
problem with Burge’s argument is that there seems to be no
good reason why deference-willingness to normative meaning
should be restricted to the relation between a speaker and his
actual linguistic community. For Burge, the reason why the
person with the partial understanding in our community pos-
sesses our normative concept is that he is willing to defer to the
correct understanding in our community. But why not assume
that there is an alternative concept that better matches his own,
and that he would have, therefore, chosen this alternative un-
derstanding if he were confronted with it?
Granted, since a person with a partial understanding of a
term is agnostic about its full applications conditions, no alter-
native concept can directly match his idiosyncratic understand-
ing as in cases of incorrect understanding like the “arthritis”
case. But in the light of widespread dispositional assumptions
about meaning, Burge seems committed to accepting that if the
person in our community would have chosen to defer to an
alternative understanding even though it does not directly
match his own understanding, then he has an alternative con-
cept. I will argue that the fundamental problem for the exter-
nalist is that as long as the person in our community and his
twin are defined as identical “from the inside”, it is not unrea-
sonable to assume that they would defer to the same meaning if
they were presented with the same alternatives. But then, ac-
cording to dispositional approaches, they also posses the same
concept and thus social externalism is false. I will conclude that
there seems to be no satisfactory solution to this problem for
the social externalist, and that Burge’s argument for social ex-
ternalism based on partial understanding is, therefore, uncon-
vincing.
The “Arthritis” Case and Cases of Partial
Understanding
Burge’s well-known “arthritis” case is the most commonly
discussed illustration of his notion of social externalism. Burge
first asks us to imagine a patient who mistakenly thinks that the
word “arthritis” applies to an inflammation he has developed in
his thigh as well as inflammations in joints. According to Burge,
since the patient is willing to defer to the medical explanation
of “arthritis” he has o ur standard concept of arthritis even bef ore
his understanding is corrected: “The patient believes that stiff-
ening of the joints is a symptom of arthritis, that certain sorts of
aches are characteristic of arthritis ... and so forth. In addition to
these unsurprising attitudes he thinks falsely that he had devel-
oped arthritis in his thigh” (Burge, 1979: p. 77). In the second
step of the thought experiment Burge imagines a twin of the
patient who lives in a counterfactual community where “arthri-
tis” has a standard meaning that corresponds to the patient’s
misconception. The third step is the externalistic conclusion:
“We suppose that in the counterfactual case we cannot correctly
ascribe [to the twin] any content clause containing an oblique
occurrence of the term ‘arthritis’” (Burge, 1979: p. 78).
Why should we accept the first step of this argument? Why
not suppose that the patient expresses an alternative concept
that matches his incorrect understanding of “arthritis”? This
supposition is what many critics of Burge have put forward.
Tim Crane has, for instance, argued that the patient's beliefs
involve the concept tharthritis “which applies to both arthritis
and whatever is the disease he has in his thigh” (Crane, 1991: p.
18). The reason, Crane claims, is that until the patient.
...is able to correct his belief about the meaning of the
word “arthritis”, there is no reason as yet to suppose that
he can discriminate between arthritis and tharthritis. So
his concept—which at the very least must reflect an abil-
ity to discriminate—will apply to arthritis and tharthritis
alike (Crane, 1991: p. 19).
Kent Bach has also argued in a similar way. He asks which
concept the patient possesses, arthritis or some concept that
“just captures the misconception” (Bach, 1994: p. 268). Bach
then claims that the “very inferences [that the patient makes]
constitute the evidence for attributing to the patient some other
notion than that of arthritis” (Bach, 1994: p. 268).
These and many other objections to Burge’s “arthritis” case
have had the following form. 1) The patient’s concept is the
concept that directly matches his incorrect understanding. 2)
There is an alternative concept that directly matches the pa-
tient’s understanding of “arthritis”. 3) The patient’s concept is,
therefore, this alternative concept. Burge attempts to respond to
this objection (Burge, 1979: pp. 110-112), but it has been a
widespread view that this response is unconvincing (Woodfield,
1982; Loar, 1987; Jacob, 1987; Crane, 1991; Bach, 1994; Pes-
sin & Goldberg 1996; Schantz, 2004).
Burge’s critics have often seemed to think that if one has re-
futed Burge’s “arthritis” argument, then one has shown that
social externalism is implausible. The plausibility of social
externalism has, in fact, been equated with the plausibility of
social externalism based on incorrect understanding, like the
“arthritis” case (Pessin & Goldberg, 1996, Schantz, 2004).
However, this inference is unjustified. The reason is that Burge
makes it clear in several places that he thinks that a partial, but
not incorrect understanding, is a sufficient foundation for social
externalism. As an example, he mentions the following case:
[A] protagonist is unsure whether his father has mortgages
on the car and house, or just on the house. He is a little
uncertain about exactly how the loan and the collateral
must be arranged for there to be mortgage, and he is not
clear about whether one may have mortgages on anything
other than houses. He is sure, however, that Uncle Harry
paid off his mortgage (Burge, 1979: pp. 82-83).
It is important to note that as Burge understands this and
other cases of partial understanding, an objection similar to the
objection against the first step of the “arthritis” case cannot
succeed. For when a person has a partial understanding there is
no basis for holding that the person has an alternative concept
that directly corresponds to his misunderstanding. According to
Burge’s distinction between incorrect and partial understanding,
3This expert-layperson distinction is first and foremost associated with
Putnam’s “The meaning of ‘meaning’” (1975), and Burge makes it clear that
his arguments are in many ways based on assumptions Putnam makes
(1979). A rather different way of developing Putnam’s distinction can be
found in Evans (1982).
2

H. NORDBY
if a person has an incorrect understanding then he has an “un-
derlying misconception”, and he does not, then, have a partial
understanding.
It is not entirely clear whether this way of conceiving of the
distinction corresponds to the understanding speakers of Eng-
lish normally have. In an intuitive sense, a competent speaker
might think that someone who has an “incorrect” understanding
can sometimes be on “the right track”, and in this sense also
have a “partial” understanding. As an example, one might refer
to cases like the classical “Blind men and the Elephant” story,
involving a man who caught hold of an elephant’s tail and er-
roneously believed that an “elephant” referred to something in
part shaped as a rope. Some might suggest that this man was on
the “right track”, because part of an elephant (e.g. its tail) is
indeed in the shape of a rope.4
It would fall outside the present argumentative purposes to
discuss in detail how the distinction between partial and incur-
rect understanding should be defined, and how it corresponds to
common sense. The important point here is that Burge’s con-
ception is reasonably clear, and that this is the conception I will
use here. Thus, Burge regards a partial understanding as an
incomplete understanding that does not involve a mistake about
the meaning of the term in question. In the following I will
adopt this way of distinguishing between partial and incorrect
understanding.5
Obviously, this terminological distinction does not imply that
persons with an incorrect understanding do not think that they
have a partial understanding. In Burge’s “arthritis” case it is
natural to assume that the patient thinks there is much he does
not know about the meaning of the term “arthritis”. It is just
that he has, for some special reason, started to believe that ar-
thritis is a disease that can occur outside joints. Yet, before the
patient defers to his doctor’s understanding as Burge assumes
that he does, the patient does not think that his belief is incor-
rect. The patient is nevertheless happy to accept that he does
not have expert knowledge about arthritis, that he is agnostic
about the full application conditions of the term. Similarly,
persons with a partial understanding will normally not consider
themselves to have an incorrect understanding. If we think that
a belief about the meaning of a term is false (and if we are suf-
ficiently rational), we do not form that belief.
Furthermore, persons with a partial understanding may have
partial understandings that differ. Cases of partial understand-
ing can be different both in the sense that they can represent
different approaches to a complete understanding and in the
sense that they can be more or less partial. But all cases of par-
tial understanding are similar in that they are “on the right
track”. As long as this is so, then no objection that appeals to an
alternative concept that directly matches an “underlying mis-
conception” can succeed.
This observation is not dependent on whether or not the ob-
jection succeeds in cases of incorrect understanding. The point
is that even if the objection succeeds in cases of incorrect un-
derstanding, it does not undermine social externalism grounded
in partial understanding. This, at any rate, is what Burge would
seem to think. His systematical defense of social externalism is
restricted to defending the idea that the person in the “arthritis”
case does not possess an alternative concept that matches his
incorrect understanding (Burge, 1979: pp. 89 -103). Burge w r i t e s:
“I shall have little further to say in defense of the second and
third steps of the thought experiment. Both rest on their intui-
tive plausibility, not on some particular theory” (Burge, 1979: p.
88).
Burge also says “I find that most people unspoiled by phi-
losophical training regard the three steps of the thought ex-
periment as painfully obvious” (Burge, 1979: p. 87). Burge as-
sumes that social externalism is plausible unless there are good
reasons for revising our ordinary use of psychological explana-
tions that ascribes concepts that are incorrectly or partially un-
derstood. In the end he summarizes his discussion like this: “Of
course, usage is not sacred if good reasons for revising it can be
given. But none have been” (Burge, 1979: p. 102). The only
objection Burge has considered is the objection that appeals to
an alternative concept that matches the incorrect understanding
in the “arthritis” case; we must, then, assume that Burge thinks
that this objection is the only real threat to social externalism.
Burge does not think that social externalism based on partial
understanding needs an independent, thorough defense.
But is this really so? It is important to bear in mind that so-
cial externalism does not merely rely on how well a person’s
understanding approximates a complete understanding; it also
relies heavily on deference-willingness to normative meaning
in this relation An important assumption in Burge’s “arthritis”
case is that the patient accepts the doctor as an authority on the
meaning of “arthritis”: “[When told about his mistake] the pa-
tient is surprised, but relinquishes his view and goes on to ask
what might be wrong with his thigh” (Burge, 1979: p. 77). The
same point applies in cases of partial understanding. For the
externalist, the reason why a person with a partial understand-
ing has our concept in the first step of the thought experiment,
is that he has an understanding that approximates a complete
understanding sufficiently, and that he is willing to defer to our
normative meaning upon confrontation. And just as in the “ar-
thritis” case, the crucial idea is that the person has our concept
already, even before he learns more about its meaning.
Similarly in the third step of the thought experiment, the
reason why the twin does not possess our concept is that he is
member of a community where the standard meaning of the
term is different from the normative meaning in our community.
The externalist claims that since our understanding is not
available for the twin to defer to, and since he merely has a
partial understanding compared to ours, it cannot be correct to
hold that he has our standard concept.
If we are going to find a convincing objection to social ex-
ternalism based on partial understanding, we must address
some of these assumptions. But appealing to the idea of a con-
cept that directly matches an incorrect application cannot un-
dermine any of them. If this is the only objection to social ex-
ternalism we can find, then social externalism based on partial
understanding remains an eligible alternative with striking im-
plications for how we should think of concept possession, not
only limited to philosophy of mind, but in all areas where ques-
tions about language mastery and the relation between mind
4I would like to express my thanks to a referee for this journal, for mention-
ing this as an example that illustrates, in a very illuminating way, that it is
not obvious how the distinction between partial and incorrect understanding
should be fine tuned.
5Obviously, much more can be said about these concepts, and it is not,
p
erhaps, clear that Burge’s distinction corresponds to all of our intuitions
about the meaning of the expressions “incorrect understanding” and “
p
artial
understanding”. However, discussing this would fall outside the argumenta-
tive purposes here. The important point is that Burge’s distinction is rea-
sonably clear, and that it can be used todistinguish between the two forms
of social externalism.

H. NORDBY
and world are relevant.
In the following I will develop a stronger objection to Bur ge’ s
argument for social externalism based on partial understanding.
The key idea of this objection is that Burge’s concept of defer-
ence-willingness as a relation between a person and his actual
community is too narrow. Consider traditional dispositional
accounts of understanding (Kripke, 1982; Cummins, 1989;
Guttenplan, 1994; Loewer, 1996). According to dispositional
approaches, how a person understands an expression involves
more than how he actually applies it. Considerations about how
the person is disposed to use it in counterfactual circumstances
are also crucial. Let us suppose that a person with a partial un-
derstanding in the first step of Burge’s thought experiment
defers to our understanding. But suppose also that the reason he
does this is that he has not been confronted with other concep-
tions of what the word in question means. How can we be cer-
tain that he would not choose to defer to some alternative un-
derstanding that he thinks better matches his own? This, I will
argue, is a question Burge pays insufficient attention to.
Deference-Willingness and Implicit Conceptions
A distinction Christopher Peacocke has made between pos-
session conditions and attribution conditions of concepts can
serve as a starting point for a deeper examination of Burge’s
assumptions about deference-willingness.
In Peacocke’s influential A Study of Concepts (1992), pos-
session conditions of concept are not simply conditions for
having beliefs involving concepts. Rather, they are conditions
not only for having a concept, but also for having a full mastery
of it (1992: pp. 27-33). The reason Peacocke makes this qualify -
cation of full mastery is that he wants to leave room for what he
calls “attribution conditions”, i.e. conditions for truly ascribing
a concept to someone who does not have a full command of it:
The attribution conditions for red, the conditions under
which something of the form “x believes that _red_” is
true, are much weaker than the possession condition. The
following are jointly sufficient for such an attribution to
be true.
1) The subject is willing sincerely to assert some sentence
of the form “_red_” containing the word “red” (or some
translation of it).
2) He has some minimal knowledge of the kind of refer-
ence it has (e.g. that it is a colo r wor d).
3) He defers in his use of the word to members of his lin-
guistic community (Peacocke, 1992: p. 29).
For Peacocke, this idea of attribution conditions is precisely
based on Burge’s arguments for social externalism. Peacocke
claims that the idea that a minimal understanding can be suffi-
cient for true concept ascriptions “has been extensively de-
fended by Burge. I take his arguments as compelling” (Pea-
cocke, 1992: p. 29).6 Peacocke also writes about his above con-
dition “c” that deference is
… what distinguishes the case we are interested in, partial
understanding (and partial misunderstanding) of a word in
a communal language, from the quite different case of an
individual’s taking over a word from his community and
using it in his own individual, different sense (Peacocke,
1992: p. 29).
Burge makes it clear in several places that he thinks defer-
ence is important for the same kind of reason. Towards the end
of his discussion he sums up the general idea behind step one of
social externalism like this:
Crudely put, whenever the subject has attained a certain
competence in large relevant parts of his language and has
(implicitly) assumed a certain general commitment or re-
sponsibility to the communal conventions governing the
language’s symbols, the expressions the subject uses take
on a certain inertia in determining attributions of mental
content to him. In particular, the expressions the subject
uses sometimes provide the content of his mental states
and events even though he only partially understands, or
even misunderstands, some of them. Global coherence
and responsibility seem sometimes to over-ride localized
incompetence (Burge, 1979: p. 114).
Burge does not say explicitly that he thinks deference is nec-
essary for the reason Peacocke gives, but we may reasonably
suppose that considerations along the same lines must underlie
Burge’s emphasis on deference. Thus, in the third step of the
thought experiment, Burge’s idea seems to be that the standard
understanding in the counterfactual community is not our cor-
rect understanding, and so the twin or any other member of that
community is not able to defer to our understanding. According
to the social externalist, the twin and the other speakers are in
the same kind of situation as the person Peacocke described:
they are using the word in question in their “own different
sense”, compared to the normative meaning in our linguistic
community.
It is important to bear in mind that for Burge, this condition
of deference only applies to persons who do not have a com-
plete understanding. Burge’s arguments are consistent with
accepting that a person in the counterfactual community has our
concept if he has an understanding that corresponds to a com-
plete understanding in our community, but this issue is not
something Burge focuses on in his discussion of social exter-
nalism. More generally, Burge’s aim is not to present an overall
theory of what it is to possess a concept; he is not even inter-
ested in analyzing what it is to have a complete understanding.
In Peacocke’s A Study of Concepts, on the other hand, the at-
tribution conditions are conceived of as an amendment to the
general theory of what it is to have a full mastery of a concept.
In more recent writings Peacocke has developed a more
comprehensive theory of how it is possible to have beliefs in-
volving a concept despite not having a full mastery. According
to this theory, a person can have a concept even though he
merely has an “implicit conception” of its meaning. Further-
more, Peacocke holds that an “implicit conception” can be suf-
ficient for a person to have a concept even though the correct
explanation of the concept cannot be found in the person’s ac-
tual linguistic community. As an example Peacocke mentions
the concept of the limit of a series:
6As said above, Burge’s defense of social externalism is generally focused
on the “arthritis” case and incorrect understanding. Burge, in fact, presents
no argument that exclusively focuses on partial understanding. Many critics
of Burge would therefore disagree with Peacocke when he holds that
Burge’s arguments are compelling.
One of the most spectacular illustrations of this is given by
the famous case of Leibniz’s and Newton’s grappling with
the notion of the limit of a series, a notion crucial in the ex-
planation of the differential calculus. It would be a huge
4

H. NORDBY
injustice to Leibniz and Newton to deny that they had the
concept of the limit of a series… I would say that each of
these great thinkers had an implicit conception, which ex-
plained their application of the phrase “limit of…” in mak-
ing judgments about the limits of particular series of ratios.
What they could not do, despite repeated pressing by critics
and well-wishers, was to make explicit the content of their
implicit conceptions (Peacocke, 1998: p. 49).
Some philosophers have argued that Peacocke has offered a
plausible analysis of cases of this kind (Higginbotham, 1998;
Toribo, 1998), while others have been more critical (Schiffer,
1998; Rey, 1998). While a detailed discussion of Peacocke’s
theory of implicit conceptions would fall outside the present
argumentative purposes, we may draw attention to the impor-
tant point that it is possible to consider the idea of deference-
willingness to normative meaning as having a wider application
beyond those cases where a person defers to speakers in his
linguistic community. According to this broader interpretation,
someone who would have deferred to the correct understanding,
even though it cannot be found in his actual linguistic commu-
nity, should also be thought of as someone who has our norma-
tive concept: “To whom were Leibniz and Newton supposed to
defer? There was no one else who understood the notion better”
(Peacocke, 1998: p. 50). Peacocke’s crucial assumption is that
these theorists would have accepted the modern explanation
of the concept of the limit of a series if they had been con-
fronted with it and had adequate time to think about its mean-
ing.
As a further if somewhat different example, consider a lin-
guist who studies a historical language that has ceased to be a
living language, i.e. one not presently used in any actual com-
munity. The linguist’s knowledge of the language, we may
imagine, draws on archeological material. For many expres-
sions of the language he has formed a competent understanding,
but for a certain expression “c” he be liev es tha t he mer ely has a
partial understanding. He is confident that the expression “c”
applies to a large carnivorous mammal, but he is not sure which
one. Furthermore, no one in his linguistic community knows
more about the meaning of “c” than he does, and so there are no
linguistic experts whom he can turn to. One day, however, he
comes across with some other written inscriptions that settle his
uncertain ty: it become s clear to him that “c” refe rs t o male lions.
Accordingly, the linguist defers to this understanding of “c”.
The problem for the social ex ternalist now begins to emerge.
Remember that for Burge, the crucial idea is that persons with
an incorrect or partial understanding can have our concepts
even before they defer to our understanding. They have our
concepts as long as that they are deference willing—as long as
they would have deferred to our understanding. It is therefore
not clear that the social externalist would be justified in reject-
ing the idea that the linguistic in the above example associates
the concept male lion with the expression “c” even before he
found the further inscriptions. For in this case deference-will-
ingness is also crucial, the only difference is that the norm ex-
ists as written inscriptions and not as a public meaning. But
what if we were to go one step further and assume that there is
not even a hidden written inscription, but merely a hypothetical
explanation of meaning to which the linguist would have de-
ferred?
The point is that we can grant the social externalist that if our
understanding is the only understanding a person with a partial
understanding in our community would be willing to defer to,
then he also has our concept. But this is to beg the real question.
The crucial question is whether we may reasonably assume that
this is the choice he would make were he confronted with other
hypothetical meaning explanations as well.
Deference-Willingness in the Thought
Experiment
Does Burge respond to this challenge? It would be unfair to
Burge to claim that he does not attempt to argue for the first
step of the externalistic thought experiment. But as noted above,
Burge’s discussion of social externalism is first and foremost
devoted to a defense of the idea that a person in our community
can possess our standard concept despite having an incorrect
understanding. Furthermore, Burge’s discussion of deference-
willingness focuses on the internal relation between a person
and the normative meaning in his actual linguistic community.
Burge does not consider as a possible objection to the first step
of social externalism the idea that a person with a partial under-
standing would have deferred to an alternative understanding.
It is not unreasonable to insist that the externalist should ad-
dress this objection. After all, most philosophical accounts of
language mastery and understanding have been dispositional. It
has been a traditional view that questions about concept posses-
sion should be answered not only on the basis of reflections on
how a person actually applies a concept, but also on assump-
tions about how he would have used it (Kripke, 1982; Cummins,
1989; Loewer, 19 97; Fodor, 1998; McManus , 2000; Kusch, 2005).
In fact, a striking feature of Burge’s discussion of social exter-
nalism is that the understanding of the counterfactual commu-
nity, as he describes it, could be a candidate for the alternative
understanding to which the person in our community would
have deferred. Consider again the “mortgage” case. As shown
above, the person with the partial understanding in our commu-
nity is conceived to be a person who “is not clear about whether
one may have mortgages on anything other than houses”
(Burge, 1979: p. 83). The second and third steps of the thought
experiment are described like this: “Imagine our man constant
in the ways previously indicated and that “mortgage” com-
monly applied only to mortgages on houses. But imagine
banking practices themselves to be the same. Then the subject’s
uncertainty would plausibly not involve the notion of mort-
gage” (Burge, 1979: p. 83).
But what if the person in our community were presented with
this alternative meaning of “mortgage”? As shown, Burge
claims that our ordinary practices of ascribing beliefs support
social externalism. Elsewhere he says that he has “...presented
the experiment as appealing to ordinary intuition” (Burge, 1979:
p. 88). But it does not seem evident that the person in our com-
munity would have chosen our understanding, and if he were to
choose to defer to the alternative understanding, then it seems
false that “ordinary intuition” supports the view that he has our
concept.
What if the externalist responded to this challenge by refor-
mulating the understanding of the person in our community so
that he would, on the basis of this understanding, recognize our
public understanding as the normative meaning? A person with
a well-developed partial understanding of this kind, Peacocke
would say, has an implicit conception of the correct explanation
of his concept. Burge has, in fact, also focused on this kind of
incomplete understanding, although not in his discussion of

H. NORDBY
social externalism and not explicitly as implicit conceptions:
Suppose I explicate my word “chair” in a way that re-
quires that chairs have legs, and then come to realize that
beach chairs, or deck chairs bolted to a wall, or ski-lift
chairs, are counter-examples. Or suppose that I learn more
about how to discriminate water from other (possible or
actual) colorless, tasteless, potable liquids. In such cases, I
learn something about chairs or water that I did not know
before. In these cases it is simply not true that the refer-
ence of my words “chair” and “water” must change. Al-
though it is true that my conception—my explication
—changes, it remains possible for me to observe (with
univocal use of “chair”): “I used to think chairs had to
have legs, but now know that chairs need not have legs”.
It remains possible for me to have thoughts about water as
water, knowing that there might be some other liquids that
I could not, by means other than use of my concept water,
discriminate from it. Thus there is a sense in which the
concept, and the translational meaning of the word in the
idiolect, remain the same despite the changed discrimi-
nating ability, or change in explication (Burge, 1989: pp.
182-183).
Burge’s point is that in cases of this kind a person will, on
the basis of the underlying understanding he already has, adjust
his explicit statements when confronted with uses that do not
match these statements. In this sense he already has an implicit
conception that causes him to adjust his meaning explanations.
It is this conception, and not his explicit statements about the
term, that c onst itutes the normative st andard for him.
The idea in the first step of the externalistic thought experi-
ment would then be this: a person has an implicit conception of
the correct meaning of a term in our community. Since the per-
son would recognize our understanding as the understanding
that best matches his idiosyncratic understanding, he would
defer to our understanding even if he were presented with al-
ternative explanations. This means that it is our concept that he
possesses.
However, even if we grant the externalist this line of reason-
ing in the first step of the thought experiment, the same prob-
lem arises in the third step. For if we think of the twin in the
counterfactual community as having this kind of partial under-
standing, then we may plausibly assume that he would have
deferred to our understanding as well, for now he has a well-
developed understanding compared to our normative meaning.
So why not suppose that he would have thought that we are
experts about the application of his term; that we are “produc-
ers” of a word he is a “consumer” of. The problem for the ex-
ternalist thus reemerges. Either he needs to explain why the
significance of deference should be restricted to the relation
between a speaker and his actual linguistic community, or he
needs to show that the twin would have chosen his own com-
munity.
The Fundamental Problem
The fact that it is not obvious that the person in our commu-
nity would have deferred to our community is merely one as-
pect of the most fundamental problem for the social externalist.
For just as we can ask whether the person in our community
would defer to the counterfactual community as Burge de-
scribes it, we can also imagine other communities that are more
or less similar. The basic problem is thus: the social externalist
needs to show that the person in our community and his twin
would not have deferred to the same understanding if they were
presented with the same choices. But as long as they are de-
fined as identical “from the inside” in the Putnamian sense—
that is, as long as they are supposed to have the same idiosyn-
cratic perspective on the world—it is not unreasonable to as-
sume that Putnamian twins would have deferred to the same
understanding. As shown, the crucial idea for Burge is the idea
of a person who has “assumed a general commitment or re-
sponsibility to the communal conventions governing the lan-
guage’s symbols” (Burge; 1979: p. 114). In accordance with
this assumption, if the person in our community and the twin
would have deferred to the same understanding, then it seems
that this understanding is the best candidate for capturing the
nature of their concepts.
It is important to bear in mind that this objection does not
depend on the idea of an entire alternative twin earth commu-
nity to which the speakers would defer. Consider again Burge’s
“mortgage” case. We may imagine a group of speakers in our
community who have chosen to understand “mortgage” in a
way that is fairly similar to the understanding of the person in
the first step of the thought experiment but is nevertheless dif-
ferent from our standard understanding. How can we be sure
that the person would not have deferred to these speakers? We
could then introduce the same group of speakers in the coun-
terfactual community in the third step of the thought experi-
ment and ask the same question about the twin.
This point is significant, since some might object that it is
not plausible to assume that the person with the partial under-
standing in our community and his twin would have deferred to
alternative twin earth communities. But this is to misunderstand
the problem. The crucial question is not whether the person in
our community and his twin would have deferred to the same
twin earth community. Rather, the crucial question is whether
there is some alternative understanding—described in one way
or another—that both would have chosen. From this wider per-
spective, Burge’s argument is unconvincing because it does not
establish that it is reasonable to assume that they would have
chosen different norms of meaning if they had been presented
with the same alternatives.
There are two ways the externalist could attempt to respond
to this objection. The first is to claim that how the person in our
community and his twin would have deferred if confronted with
all the relevant alternatives is not the crucial point; instead,
what is decisive is the fact that they defer to their respective
communities as described by Burge. The problem with this
response is that it is inconsistent with analyses of meaning and
understanding that assume that questions about concept posses-
sion must be answered on the basis of dispositional considera-
tions. To be sure, dispositional accounts of meaning and under-
standing have been developed in different ways (Kripke, 1982;
Guttenplan, 1994; McManus, 2000; Kusch, 2005), but for the
purposes here these further differences are not crucial. My aim
is merely to show that Burge’s argument is unconvincing, and
for this purpose it is sufficient that considerations about how a
person is disposed to apply a term in counterfactual contexts
seem relevant for questions of understanding. There is an im-
mediate, intuitive appeal to the idea that a person’s overall dis-
positions to apply a term are relevant for determining which
concept he expresses. If we can form a clear view of how a
person understands a term not only in actual cases, but also in
6

H. NORDBY
relevant counterfactual ones, then it seems reasonable to as-
sume that his concept has an extension that matches both the
actual and cou n t e r f a c t ua l applications.7
This idea has been central in very many philosophical analy-
ses of concept possession, and not only classical accounts of
straightforward one-to-one applications of natural kind con-
cepts. To see this, we may consider again Peacocke’s above
analysis of implicit conceptions of a concept like the limit of a
series. For Peacocke, the crucial point is that if a person who
has an implicit conception of this concept were confronted with
its correct explanation and, given time to consider it, then he
would defer to this explanation and thereby show that our nor-
mative concept is the one that best matches his understanding.
Peacocke (1998) thinks that this idea is plausible in itself, but
we need not accept Peacocke’s specific analysis of implicit
conceptions in order to accept the more general idea that coun-
terfactual considerations about how a person is disposed to use
a concept should have an intuitive role in theories of concept
possession. The problem is that Burge does not confront this
idea in his arguments.
The other way the externalist can attempt to respond to the
dispositional challenge is to argue that even though a disposi-
tional approach to concept possession is correct, the person in
our community and his twin would not have accepted and de-
ferred to the same explanation of the term in question. To this
response I emphasize that I have not claimed that they would
have deferred to the same meaning. For the purpose of showing
that Burge’s argument is unconvincing, I merely needed to
point to the fact that it is not reasonably clear that they would
have deferred to different meaning explanations. It is simply
difficult to imagine the whole scenario, and especially difficult
to determine what the twin would have done. We do not seem
to have any robust intuitions about this.
This is a general point that applies no matter how the thought
experiment is formulated. Consider, for instance, the suggestion
that the person in our community would defer to our under-
standing (and the twin to the counterfactual community) simply
because it is his actual community. One might seek to ground
this suggestion in general ideas about how a person can feel
attached psychologically to things in the actual world, so that
he would choose to have these things because they are the
things that they actually are. An analogy can make this idea
clearer: a person might be attached to his actual old car in a
way such that he would not prefer to have a new car even
though the old car is noisy, cramped and unreliable.
So why not suppose that the person in our community would
prefer to defer to our understanding for the same kind of reason?
According to this objection, the person in the “mortgage” case
(2 above) might not defer to an explanation of “mortgage” that
represents our understanding if he does not know that it is our
understanding. But the crucial point, some might suggest, is
that he would choose our understanding if he knew that it was
our understanding.
However, even if we grant to the externalist that the person
has this kind of knowledge the problem remains the same: in
order for Burge’s argument to be “intuitively compelling” as he
claims that it is, then it also has to be intuitively compelling that
persons with a partial understanding feel attached to the norma-
tive meaning in their linguistic communities in the same way
that we can feel attached to other things. But this is far from ob-
vious. In the “mortgage” case, it is not clear that the person in
our community case would experience a special attachment to
our correct definition of “mortgage”. We do not seem to be
emotionally or cognitively attached to the standard meaning of
language expressions in the way we are sometimes attached to
actual persons or things.8
It is important to bear in mind that this objection does not
require that we pick out the understanding that best matches
that of the person with the partial understanding. It is sufficient
that we can think of one alternative meaning explanation that
the person might choose to defer to. In the “mortgage” case we
may in fact use Burge’s description of the term’s meaning in
the counterfactual community as an alternative explanation. As
shown above, Burge imagines “our man constant in the ways
previously indicated and that ‘mortgage’ commonly applied
only to mortgages on houses” (Burge, 1979: p. 83).
The difference between this alternative explanation and our
standard understanding is that our standard concept applies not
only to loans to finance the purchase of houses, but also to
loans to finance the purchase all forms of real estate (including
property like land). As Burge characterizes the first step of the
thought experiment, this idea about real estate is very remote
from the person’s understanding. So suppose that the person
was asked to choose between the alternative explanation and
our standard understanding. It does not seem unreasonable to
assume that he would reason as follows: “This alternative way
of understanding ‘mortgage’ seems to capture my understand-
ing better than the comprehensive idea about real estate. I had
absolutely no idea that it is possible to have a ‘mortgage’ on
land. So it seems that my concept of mortgage, as I have under-
stood it, corresponds to this more narrow way of understanding
the term.”
Of course, the person might then choose to change his under-
standing so that it matches our standard understanding, but that
would not help the social externalist. The reason is that the
externalist is committed to making a distinction between a per-
son who defers to an explanation of a word on the basis of an
understanding he has, and a person who defers simply because
he thinks that an explanation is correct (regardless whether or
not he has a partial understanding). Consider as an example of
the latter case a person who is beginning to learn English and
who thinks for some time that “blue” means red. When he dis-
covers his mistake he will defer to the public meaning of “blue”
and change his beliefs about the term, but that does not mean
that he earlier expressed the concept blue by “blue”.
7There are w ell known chal lenges relat ed to the need of specifyi ng cases o
f
application that determine the nature of a person’s concepts (Loewer,1997;
McManus, 2000; Kusch, 2005). For instance, the concept I express by “dog”
can be dog even if I mistakenly apply “dog” to a cat that from a long dista nc e
looks to me like a dog. But the problem of specifying content-determining
applications in a non-circular way does not undermine the general idea tha t a
p
erson’s concepts are individuated both on the basis of relevant actual and
counterfactual behavior. It is this general ide a that Burge does not confront.
8What if Burge simply assumed that the persons in the thought experiment
would choose their respective actual communities? There are two obvious
problems with this strategy. First, if Burge’s premises are either question-
able or insufficiently justified, then social externalism based on partial un-
derstanding is not a very interesting philosophical theory. Secondly, if social
externalism based on partial understanding is going to describe a widespread
p
henomenon (that persons’ concepts are individuated “externalistically”) as
Burge clai ms th at the p o siti on do es, t hen t he as sumpt io ns Bu rg e makes mus t
correspond to how persons with a partial understanding would tend to be-
have in real life.
This is obviously part of the reason why Burge requires that
the person in the thought experiment has a minimal under-
standing. Burge’s idea is that a person with a minimal under-
standing has our concept if he is deference-willing and if his

H. NORDBY
understanding matches our standard reasonably well. The
problem for Burge is that even if we grant that these two condi-
tions are met in the “mortgage” case, it is not unreasonable to
assume that the person would think that the alternative explana-
tion better matches his understanding and that he therefore
would think that his present concept corresponds to this expla-
nation.
We can strengthen this objection further by imagining that
the person was confronted by a group of speakers who under-
stood “mortgage” in the alternative way. The suggestion would
then be that this would make him even more confident that he
has an alternative concept, that the fact that a group of speakers
has a similar understanding would make it even more probable
that he would think that his use of term belongs within what
Wittgenstein would call a “language-game” that does not corre-
spond to the standard definition in a dictionary. Again, the pre-
sent argument does not require that this is obvious. For the
purpose of showing that Burge’s argument is unconvincing, it
is sufficient to point to the fact that we do not have robust in-
tuit-tions about this. Further arguments are needed, but Burge
has not provided any.
This fundamental uncertainty about the preferences of the
persons in the thought experiment relates in particular to the
fact that it is possible for “twin earth” thought experiments to
have an important role in philosophical arguments as long as
we restrict our assumptions to how the counterfactual commu-
nity actually is. It is quite a different matter when we also have
to determine, on the basis of these assumptions, how the twin in
the counterfactual community would have acted if confronted
with yet another counterfactual understanding. In order to de-
termine this, we need to determine how the twin would have
thought about himself, his understanding and his community in
this even more hypothetical situation. The thought experiment
becomes so complex that it seems overwhelmingly difficult to
make robust interpretations without ramifying further substan-
tial and theoretical assumptions.
This issue is not at all confronted in Burge’s arguments for
social externalism based on partial understanding. What Burge
appeals to in his defense of the third step of the thought ex-
periment is the idea that an understanding that is not coextend-
sional with our understanding cannot correspond to our concept:
“The word ‘arthritis’ in the counterfactual community does not
mean arthritis. It does not apply only to inflammations of
joints” (Burge, 1979: p. 79). As shown above, critics of Burge
have argued that the person with the incorrect understanding of
“arthritis” in our community possesses the standard concept of
the counterfactual community because it directly matches his
understanding. This objection cannot succeed in cases of partial
understanding, but I have argued that Burge faces a further
problem related to deference-willingness: in order to make it
reasonably clear that the twin would have chosen to defer to the
standard in his community, then this standard must be defined
to approximate his understanding. But then it is not unreason-
able to assume that the person in our community would have
deferred to this understanding as well.
More generally, it is not implausible that the person in our
community and his twin would have deferred to the same un-
derstanding if they were presented with the same choices. And
if they would, then a reasonable “wide” interpretation of the
externalistic assumption about deference-willingness implies
that they have the same concept. But then social externalism
based on partial understanding is false. The social externalist
could retreat to a “narrow” conception of deference-willingness,
but at the cost of adhering to a problematic and counterintuitive
conception.
Conclusion
Social externalism based on partial understanding escapes
the standard objection to the argument for social externalism
based on incorrect understanding. But social externalism based
on partial understanding faces another problem related to the
assumption about deference-willingness to normative meaning.
Burge claims that the person with a partial understanding of a
term in our community has our concept since he is willing to
defer to our correct understanding, and that his counterfactual
twin has another concept since he lives in a community where
the term has another standard meaning.
I have argued that in order for this argument to be convincing,
it has to be reasonably clear that deference-willingness should
not be understood as a wider dispositional attitude, or that the
person in our community and his twin would not have deferred
to the same understanding if they were presented with the same
choices. The main problem for Burge is that this is not reasona-
bly clear. Burge rests the main part of his defense of social
externalism on what he thinks of as the immediate appeal of the
three steps of the externalistic thought experiment. In this way
Burge accepts the widespread view that philosophical thought
experiments can support substantial philosophical conclusions
if we have strong intuitions about how they should be inter-
preted. The problem is that we do not have such intuitions
about some of the crucial assumptions social externalism based
on partial understanding requires. The externalist needs to pre-
sent further substantial arguments for these assumptions, but
there are good reasons for being skeptical about the possibility
of finding such arguments.
REFERENCES
Bach, K. (1994). Thoug h t an d language. Oxford: Cla rendon Press.
Burge, T. (1979). Individualism and the mental. In P. French, T. Ueh-
ling, & W. Wettstein (Eds.), Midwest studies in philosophy, 4.
Burge, T. (1986). Intellectual norms and foundations of mind. The
Journal of Philosophy, 86, 3-45.
Burge, T. (1989). Wherein is language social? In A. George (Ed.),
Reflections on chomsky. Oxford: Basil Blackwell.
Crane, T. (1991). All the difference in the world. The Philosophical
Quarterly, 41, 1-25. doi:10.2307/2219783
Cummins, R. (1989). Meaning and mental representation. Cambridge,
MA: MIT Press.
Evans, G. (1982). The varieties o f reference. Oxford: Clarendon Press.
Fodor, J. (1998). Concepts. Oxford: Oxford University Press.
doi:10.1093/0198236360.001.0001
Guttenplan, S. (Ed.) (1994). A companion to the philosophy of mind.
Oxford: Basil Blackwell.
Higginbotham, J. (1998). Conceptual competence. In E. Villanueva
(Ed.), Concepts: Philosophical Studies, 9.
Jacob, P. (1987). Thoughts and belief ascriptions. Mind & Language, 4,
301-325. doi:10.1111/j.1468-0017.1987.tb00124.x
Kripke, S. (1982). Wittgenstein on rules and private language. Oxford:
Blackwell.
Kusch, M. (2005). Fodor v. Kripke: Semantic dispositionalism, ideali-
zation and ceteris paribus claus e s. Analysis, 65, 156-163.
doi:10.1093/analys/65.2.156
Loar, B. (1985). Social content and psychological content. In R. Grimm,
& D. Merill (Eds.), Contents of thought. Tucson, AZ: Arizona Uni-
8

H. NORDBY
versity Press.
Loewer, B. (1997). A guide to naturalizing semantics. In B. Hale, & C.
Wright (Eds.), A companion to the philosophy of language. Oxford:
Basil Blackwell.
McManus, D. (2000). Boghossian, Miller and Lewis on dispositional
theories of meaning. Mind & Language, 15, 393- 399.
doi:10.1111/1468-0017.00141
Nordby, H. (2004). Concept possession and incorrect understanding.
Philosophical Explorati ons , 7, 55-70.
doi:10.1080/1386979032000186854
Peacocke, C. (1992). A study of concepts . Cambridge, MA: MIT Press.
Peacocke, C. (1998). Implicit conceptions, understanding and rational-
ity. In E. Villanueva (Ed.), Concepts: Philosophical, 9.
Pessin, A., & Goldberg, S. (Eds.) (1996). The twin earth chronicles.
New York/London: M. E. Sharpe.
Putnam, H. (1975). Mind, language and reality (Vol. 2). Cambridge:
Cambridge Universit y Press. doi:10.1017/CBO9780511625251
Putnam, H. (1996). Introduction. In A. Pessin, & S. Goldberg, (Eds.),
The twin earth chronicles. New York/Lon d on : M. E. Sharpe.
Rey, G. (1998). What implicit conceptions are unlikely to do. In E.
Villanueva (Ed.), Concepts: Philosophical Studies, 9.
Schantz, R. (Ed.) (2004). The externalist challenge. Berlin/ New York:
Walter de Gruyter.
Schiffer, S. (1998). Doubts about implicit conceptions. In E. Villanueva
(Ed.), Concepts: Philosophical Studies, 9.
Toribo, J. (1998). The implicit conception of implicit conceptions. In E.
Villanueva (Ed.), Concepts: Philosophical Studies, 9.
Wittgenstein, L. (1953). Philosophical investigations. Oxford: Basil
Blackwell.
Woodfield, A. (1982). Thought and the social community. Inquiry, 25,
435-450. doi:10.1080/00201748208601979