This paper aims to dissolve the problem of the compatibility between authoritative self-knowledge of propositional attitudes and content externalism. The basic strategy to do this is to appeal to pair concepts of semantic inferentialism, i.e., the propositional content and the representational content. Inferentialists can explain how the former content can be authoritatively known from the firstperson perspective, and why the latter content is partly determined by external factors. In addition to this, it is also shown how this inferentialistic conception of content succeeds in dodging the attack on authoritative self-knowledge derived from the physical externalism about the mental content. What makes it possible is that the inferentialistic conception of content rejects the traditional but problematic presumption that the intension is the determinant of the extension. In this way, this paper reconciles the authority of self-knowledge of propositional attitudes with the thesis of (physical) externalism.
It is, on the one hand, often stressed that our language has a public character. Content externalism is one of the clear manifestations of the character. On the other hand, it is also difficult to deny that our self-knowledge of mental states (including propositional attitudes) has a certain kind of authority. Moreover, since the self-knowledge is normally achieved relying on no inference, the authority is based on no empirical evidence. Although both of these claims (the public character of our language and the non-inferential authority of our self-knowledge) seems quite plausible respectively, there is, at least apparently, a tension in simultaneously accepting both of them. This is what I call ‘the problem of compatibility’, which is the subject of this paper.
But what exactly is the problem? Although the controversy is complicated, the gist
of the problem is easy to grasp.1 Content externalism claims that the contents of
linguistic expressions are determined partly by certain (environmental or social) factors
external to an individual speaker’s inner state. Once we endorse this thesis concerning
the contents of linguistic expressions, then it becomes almost unavoidable to endorse
the same thesis concerning the contents of propositional attitudes, since those two
sorts of contents are arguably dependent on each other. But the external factors that
1 The following formulation of the compatibility problem is first proposed by
That seems to be a contradiction! But is that really contradictory? The crucial point on which we have to reflect is what the word ‘contents’ means here. To begin with, (1) it is not at all obvious that the contents of which we can have authoritative selfknowledge and the contents that are beyond our grasps are the same. Then, if they are different, (2) in what do the former and the latter contents consist respectively? And finally (3) how are the former contents authoritatively known to us through no evidence and no inference, and why are the latter contents beyond our own grasps?
In my opinion, the answers to all these questions can be given adequately from the semantic inferentialist’s standpoint. Therefore, briefly turning aside from the problem of compatibility, I introduce the main points of semantic inferentialism.2
Semantic inferentialism is concerned with the question of what it is for statements uttered or propositional attitudes held by us to be meaningful or to have contents. And it answers the question basically in the spirit of the use theory of meaning. Traditionally, philosophers have divided meaning into two distinctive categories—the intension / the extension, or in Brandom’s terminology the concomitant aspect / the representational aspect of the propositional content.3 Roughly speaking, many of the philosophers try to explain the former based on the latter. Semantic inferentialists, however, base on our normative practice of inference and invert the order of explanation.4
To begin with, according to inferentialists’ explanation, a statement one utters has
some propositional content when it is inferentially articulated, i.e., when it counts as a
move in the game of giving and asking for reasons. Then, the propositional content of
a claim has two different aspects, which are the inferentialist’s alternatives of the
traditional distinction of intension and extension. First, the concomitant aspect of the
propositional content that the uttered statement has consists in the inferential
commitments5 that the original statement, along with the background beliefs of the utterer,
2 This view is rigorously and exhaustively developed in
Second, in order to understand the representational aspect of the propositional content of a statement (in short, ‘representational content’ of a statement), we should attend not merely to the inferentially articulated dimension, but also to the socially articulated dimension of our game of giving and asking for reasons. To begin with, the representational content of a claim, contrasted with the propositional content of the claim (i.e., what is said by the claim), is the object that the claim talks about. And what the object is like is determined by the identity condition of the object. Then the representational content of the claim consists in the inferential roles of various true identity statements that describe the identity condition of the object. But what inferential roles do these identity statements play? Their roles are enabling us to make new substitutional commitments7 through substitutional inferences, i.e., the inferences that draw a consequence by simultaneously replacing a certain term occurring in a premise with another term based on an identity statement. Therefore, grasping the representational content of a statement consists in the ability to derive various substitutional commitments from the original statement together with those true identity statements through making substitutional inferences. But for what purpose does such an ability serve? We employ this kind of ability in our social and communicative practice. Grasping the representational content of a claim is indispensable for assessing the claim made by another person from de re standpoint and for extracting useful information from the claim. In this sense, the representational content of a statement is not only inferentially but also socially articulated in our inferential practice.8
Finally, all the above explanations can be, mutatis mutandis, applied to the explanations of the contents of beliefs. Additionally, for the sake of argument, I presuppose that the contents of the other propositional attitudes than beliefs (e.g., desires and intentions) can be also explained in similar ways. 3 The Basic Line 6 See Brandom (1994, pp. 168-70, pp. 186-90). Strictly speaking, a third factor, i.e., the incompatibility relation between commitments and entitlements, is also needed in full explanation of inferential articulation of the propositional content. But, for reasons of space, we have to leave it out. 7 The distinction between the inferential commitment and the substitutional commitment consists in the difference between de dicto viewpoint and de re viewpoint. An inferential commitment of a claim is an inferential consequence of the claim together with the speaker’s auxiliary background beliefs. So, the speaker herself is prepared to acknowledge the commitment from her own standpoint. Oppositely, a substitutional commitment of a claim is the commitment undertaken by a interlocutor from de re viewpoint. The substitutional commitment is undertaken through the substitutional inference based on an identity statement to which the interlocutor is committed, regardless that the speaker may acknowledge the commitment or not. 8 See Brandom (1994, pp. 136-40, pp. 370-6, pp. 495- 520).
Now we can come back to the series of questions concerning the compatibility problem. The basic line of my position is very simple. Let me answer briefly the questions (1) and (2) at a stroke. In my opinion, the two sorts of contents, i.e., the contents that one can authoritatively know without relying on any evidence and the contents that are in a sense beyond one’s own grasp, are different. And my suggestion is that, from the semantic inferentialist’s viewpoint, we should regard the former as ‘propositional contents’ and the latter as ‘representational contents.’9
But how can we authoritatively know the propositional contents of our own
attitudes based on no evidence? As we saw in section two, grasping the propositional
content of one’s own belief is grasping the inferential commitments and the
entitlements connected to the belief. And the inferential commitments of a belief, on the one
hand, are the statements that the belief, together with other auxiliary background
beliefs of the believer, makes her prepared to acknowledge.10 On the other, the
entitlements of the belief are the statements that the believer is prepared to acknowledge
as justification of the belief. For our purpose, what is important in these
characterizations of commitments and entitlements is that both of them contain the normative
phrase of ‘(the believer is) prepared to acknowledge….’ In virtue of this feature, we
9 But I have to note that this direction has been already suggested in Brandom (1994, p. 507).
So what I try to do below is to flesh out his idea and thereby to apply it to an attempt of
dissolving the compatibility problem.
10 Strictly speaking, there can be an inferential commitment that a subject is not prepared to
acknowledge. Brandom pointed out that there are two ways of undertaking an inferential
commitment.
To begin with, let us turn to the explanation of our grasping the inferential commitments of our own beliefs. Suppose that B(P) is one of my beliefs, and Cs are the inferential commitments associated with B(P). As we saw in the above paragraph, a statement S is C if and only if B(P), together with other auxiliary background beliefs of mine (call them ‘B(Qs)’), makes me prepared to acknowledge S. However, in fact, being prepared to acknowledge S is equivalent to believing that S. So, the above biconditional can be put in the following way; a statement S is C if and only if B(P), together with B(Qs), makes me believe that S. In other words, A statement S is C iff given that B(P) and B(Qs), I believe that S. This means that in order to judge whether a given statement S is C or not, I only have to judge whether I believe that S or not given that B(P) and B(Qs).
Here, it is helpful to remember the ‘Transparency Thesis’ about self-ascription of beliefs, which is first proposed and named by Roy Edgley, and is afterward developed by Gareth Evans and Richard Moran.11 According to the thesis, in order to judge whether I believe that S or not, all I have to do is judging whether it is the case that S or not. (The relevant question is directed not to myself, but to the world.) Why? For asserting the following sentence (what is called ‘Moore’s sentence’) sounds awkwardly irrational; ‘It is the case that S, but I do not believe that S.’ Note that the Transparency Thesis holds only in the case of self-ascription. (It is not at all irrational to assert that it is case that S, but she does not believe that S.) Now, let us apply this thesis to the question at issue. In order to judge whether I believe that S or not given that B(P) and B(Qs), all I have to do is judge whether it is the case that S or not, given that P and Qs. As long as I am equipped with minimal rationality and with the normal ability to make inference concerning the world, I can judge whether a given statement S is C or not, based on this kind of transparency procedure. Therefore, we can authoritatively know what inferential commitments our own beliefs have.
This type of explanation, mutatis mutandis, can be applied to the grasp of the entitlements of one’s own beliefs. In addition to the previous notation, suppose that Es are the entitlements associated with B(P). Then, a statement S is E if and only if I am prepared to acknowledge that E is a justification of B(P). But again, for the synonymy of ‘being prepared to acknowledge’ and ‘believe’, we can translate the above biconditional into the following form; a statement S is E iff I believe that S is a justification of B(P). Through the transparency procedure, the right side of this biconditional can be judged by judging the question of whether it is the case that S justifies P or not. Consequently, we, who have minimal rationality and the ability to assess evidence, can authoritatively know what entitlements our own beliefs have. Now, we obtain the explanation of the authoritative knowledge about the propositional contents (the inferential commitments and the entitlements) that our own beliefs have.
Next, why are the representational contents of one’s attitudes beyond one’s own grasp? According to the inferentialist’s view, grasping the representational content of a claim —grasping the object that the claim talks about—demands of us to grasp the substitutional commitments that the original claim plus the relevant true identity statements make us undertake. Here, the adjective ‘true’ is essential. For instance, my belief that the Morning Star is very bright is, needless to say, directed to the Morning Star. Of course, I know that. But this belief can be also said to be directed to Venus since it is true that the Morning Star = Venus, whether or not I know it. In other words, my belief that the Morning Star is very bright conveys some information about Venus too. I, however, may not be able to utilize the information since the truth of the relevant identity statement is objective, and I have to make certain empirical inquiry to know whether it is true or not. In a word, a posteriority of true identity statements makes a lot of room for my not completely grasping the representational contents of my own beliefs.
These are the basic line of my argument. And they seem to suggest a way to reconcile authoritative self-knowledge with content externalism. For our picture can admit both that we can, in the propositional sense, authoritatively know the contents of our own propositional attitudes, and that the contents of our propositional attitudes are, in the representational sense, determined by certain external factors. In the following section, we will turn to investigate at length exactly what implications this line of thought has to content externalism. 4 Implications to the Physical Externalism12
Now, let us consider Putnam’s physical externalism (see
In order to be more rigorous, the following two points are worth noting. The first is
that when Putnam uses the word ‘mental state’ above, he uses it in the ‘narrow’ sense,
i.e., that the mental states that he supposes are individualistic in the sense that they
supervene just on subject’s inner states, not on the physical / social environment of
12 Although the social externalism proposed by
From this argument, Putnam concludes that we have to reject either of the two familiar claims, i.e., (1) the claim that the speaker’s mental states determine the intension of a word, or (2) the claim that the intension of a word determines its extension. As is widely known, Putnam selected the first option of rejecting the claim (1). This means that the understanding of word’s meaning— connecting an intension with a word—is not a matter of the mental since the connection obtains partly by the causal link between a speaker and her environment. Thus, the physical externalism follows.
Although Putnam himself applies his argument only to the meanings of words, his
argument, as
Roughly speaking, our reply to this problem is rejecting claim (2) and sustaining claim (1), which is just the opposite of Putnam’s position. In his argument, the rejection of (1), which is equivalent to endorsing content externalism, depends on the 13 For the argument, here I presuppose that the phenomenal states are generally narrow. But this presupposition is not indispensable for my discussion. rightness of (2). However, is claim (2) really essential for our concept of content (or intension)? Isn’t there any other conception of content (or intension) that does not imply claim (2)? Here, the standpoint of semantic inferentialism is crucial. As I pointed out in section two, inferentialism reverses the traditional order of explanation between the extension and the intension. Traditionally, philosophers tend to explain what the intension is based on its relation to the extension, presupposing an independent explanation of what the extensions is. And the view that the philosophers widely shared was that the intension is something giving the sufficient condition of the extension (although it is open exact what the intension consists in). However, the inferentialist radically departs from this way of thinking. Inferentialist’s alternatives of the intension and the extension, i.e., the propositional content and the representational content, are both explained in terms of our inferential practice. What is crucial for our discussion is that inferentialist’s explanation of the two concepts does not imply the traditional claim (2). I want to elaborate on this point below.
According to inferentialism, on the one hand, grasping the propositional content of a statement (or a belief) is grasping the inferential role that the statement (or the belief) plays, namely the inferential commitments that the subject who asserts the statement (or holds the belief) undertakes and the entitlements that she regards as support for the statement (or the belief). On the other hand, grasping the representational content of a statement (or a belief) consists in grasping the substitutional commitments derived from the statement along with the relevant true identity statements. It follows from this explanation that grasping the propositional content of a statement (or a belief) does not necessarily contains fully grasping its representational content. For, as I pointed out in section three, the relevant true identity statements may be a posteriori.
For example, consider the statement,
And imagine a rational person, Nancy, who asserts this statement with a good grasp of its propositional content (its entitlements and its inferential commitments): e.g., she grasps what perceptual statements she can use to support the statement, and she grasps that by asserting the statement she is committed to the statement that the Morning Star is a certain large substance flouting in the space etc.. But, in this case, it is possible that she refrains from undertaking the commitment to the following statement,
Since the following identity statement is undoubtedly a posteriori, she may be ignorant of it.
The Morning Star = the Evening Star. (i) Nevertheless, we, who has already known that (i) is empirically true, know that her statement (a) is directed not only to the Morning Star, but also to the Evening Star. So we can derive the commitment to (b) from the commitment to (a) together with the commitment to (i). In other words, the inferential role of (i), in this context, is enabling us to make substitutional inference from (a) to (b). Nancy, however, cannot make such a substitutional inference. Consequently, she does not grasp the substitutional commitment derived from the proper substitutional inference, and therefore, she does not fully grasp the representational content of her own statement (a).
To summarize, the inferentialist defines the propositional content of a statement in terms of the entitlements and the inferential commitments that a rational utterer connects to the statement. So, in uttering the statement, the rational utterer, as I argued in section three, always grasps the propositional content of her statement. On the contrary, grasping the representational content of the statement consists in grasping the proper substitutional commitments (like (b)) that can be obtained through proper substitutional inferences based on true identity statements (like (i)). But since the truth value of the identity statements is objective, and knowing the value is an empirical matter, it is possible that the rational utterer does not know which identity statement is true. Therefore, the representational content of the statement can be unknown to the utterer. In a word, grasping the representational content of a statement demands the utterer of empirical inquiry although grasping its propositional content does not.
If grasping the propositional content of a statement does not necessarily contains grasping its representational content, then it follows that its propositional content is not something providing the necessary and sufficient condition of its representational content. This is just the same as rejecting the above claim (2). Consequently, the inferentialist can dissolve the compatibility problem, in full agreement with Putnam on his externalist insight that Emily’s belief that water is transparent is directed to XYZ, not to H2O. Since the inferentialist rejects (2), accepting the authoritative selfknowledge about propositional contents of one’s own belief is no bar to the belief’s representational content being partly determined by her physical environment, of which she may be completely unaware. Indeed, the inferentialist, as we saw in the above paragraph, positively asserts that the representational content of a propositional attitude varies depending on her physical environment.14 For the inferentialist too, the representational content is an empirical matter.
In my diagnosis, the problem of compatibility between self-knowledge and externalism arises, due to a bias traditionally shared in several conceptions of intension or mental content, i.e., the bias that the intension (or the mental content) somehow determines the extension. Inferentialism, however, rejects the bias, and thereby can reconcile the distinctive self-knowledge with the externalist’s insight. This shows an important advantage of inferentialistic conception of meaning and mental content. For each of the distinctiveness of our self-knowledge about propositional attitudes and the public character of our language is not easy to throw away. The large quantity of papers concerning this topic shows the seriousness of this dilemma. But inferentialism dissolves the dilemma by offering two new conception of meaning (or mental con14 Brandom is much self-conscious of this externalistic character of his view. He focuses on this topic in Brandom (1994, pp. 631-3, pp. 645-649). tent), i.e., the propositional content and the representational content. In other words, by offering them, inferentialism neatly accommodates both the subjective, authoritatively known aspect and the public, empirically known aspect of meaning (or mental content). In this sense, inferentialist’s conception of meaning (or mental content) brings us a significant insight.