1. Preliminary Remarks
According to internalist foundationalists, we can solve the problem of epistemic regress only by admitting foundational beliefs, and the justification of such a foundational belief requires that the believer have a good reason for thinking that it is true. The most important objection to internalist foundationism by far is the dilemma suggested by Wilfrid Sellars. According to internalist foundationalism, a perceptual belief is justified by an experience, so that the believer must apprehend the character of the experience as a reason for thinking that the belief is true. But this view faces the following dilemma. If, on the one hand, the apprehension of the character of an experience takes the form of an apprehension that the experience is of a certain sort, then such an apprehension seems to require justification, and so the regress of justification does not end. If, on the other hand, the apprehension of the character of the experience does not take the form of an apprehension that it is of a certain sort, it is very unclear how such a non-propositional apprehension can provide a reason for thinking that the belief is true.
Laurence BonJour admits that internalist foundationalists are in serious trouble unless they find a solution to the Sellarsian dilemma. However, in his 2001 paper “Toward a Defense of Empirical Foundationalism” and his 2003 paper “A Version of Internalist Foundationalism”, he claims that he has found a way out of the dilemma. The purpose of this paper is to show that his alleged solution is not successful. Admittedly, a few important objections have already been made against him.Footnote 1 But he isn’t faltering in his convictions.Footnote 2 Thus, in this paper, I shall offer new lines of argument to show why BonJour’s account is ultimately untenable.
I shall proceed as follows. In section 2, I shall briefly discuss BonJour’s way out of the Sellarsian dilemma. In section 3, I shall argue that there are two requirements for being a good reason, and that his notion of non-conceptual awareness of sensory content faces a serious dilemma with regard to these requirements. In section 4, I shall discuss his view of conscious occurent belief, because he derives his idea of “constitutive awareness of sensory content”, which plays a pivotal role in his way out of the dilemma, from his conception of conscious occurent belief. Finally, in section 5, I shall discuss his explanatory account, according to which one can infer non-foundational beliefs about the physical world from one’s foundational beliefs about sensory experiences by way of an inference to the best explanation. In particular, I shall argue that his explanatory account faces the problem of representation, which I shall explain in due course.
2. Sellar’s Myth of the Given and BonJour’s Nondoxastic Foundationalism
Internalist foundationalists accept the so-called “doctrine of the given”, according to which a sensory experience can serve as a good reason for accepting a foundational belief, although it is non-conceptual. Wilfrid Sellars argues in his 1963 paper “Empiricism and the Philosophy of Mind” and his 1975 paper “The Structure of Knowledge” that the doctrine of the given is merely a myth. Suppose that an agent, S, believes that P. In order for S to be justified in believing that P, S must have a good reason for thinking that P is true. Let us say that the required reason is R. In order for R to be a good reason for thinking that P is true, R must be the kind of thing that allows S to think that P is true or likely to be true. But, as Immanuel Kant points out, “intuitions without concepts are blind.”Footnote 3 To put it another way, sensory experiences, which are not yet conceptualized, are blind; and such a blind thing cannot serve as a good reason for holding a belief. Therefore, Sellars claims that one’s sensory experience cannot serve as a good reason for holding one’s belief; instead, we should take one’s sensory experience as playing only a causal role. BonJour claims, as noted, that he has a solution to the Sellarsian dilemma on behalf of internalist foundationalism. Let me briefly explain his alleged solution.
First, on BonJour’s view, one’s foundational beliefs are physical-object appearance beliefs. He says:
[W]hat I am conceptually aware of is certain physical-object appearances or apparent physical objects—or, in a slightly more technical terminology, of ways of being “appeared to.” Where the appearance in question is a visual experience, we may say instead that it looks as though there are objects of the sorts indicated, and analogously for other sensory modalities.Footnote 4
Suppose, for example, that S is having a visual experience E of the sort you normally have when there is a blue book in front of you. Suppose also that, on the basis of this experience, S comes to have the following appearance belief:
(B) It looks as if there is a blue book in front of me.
And consider why this foundational belief is justified. What sort of reason does S have for thinking that B is true? BonJour admits that there is no intelligible relation between non-conceptual sensory content and conceptual beliefs in general. Nonetheless, he claims that conscious experience is an exception. On his view, a sensory experience is non-conceptual, but nonetheless it is a conscious state, so that the subject can have a direct awareness of the sensory content by being conscious of it. The question is how the non-conceptual content of E can serve as a good reason for accepting the conceptual content of B. According to BonJour, the relation between the non-conceptual content of a sensory experience and the conceptual content of a belief is not logical. This is because only conceptual items can stand in a logical relation. The relation between the two is not merely causal either, for causal relations are not justificatory relations. What then is the relation between the two? On his view, “it is a descriptive relation, having to do with the accuracy or inaccuracy of fit between a conceptual description and a nonconceptual object that the description purports to describe.”Footnote 5
In the above case, thus, if the conceptual content of B accurately describes the sensory content of E, S’s direct awareness of the sensory content of E provides her with a reason for thinking that B is true. Since S is consciously aware of the sensory content of E, and since she also understands the conceptual content of B, she can recognize the accuracy of fit between the sensory content of E and the conceptual content of B by comparing the two; in such a case, she has a good reason for thinking that B is true. In addition, such an awareness of fit is not a further, cognitively independent judgment, which would then require further independent justification. For an experience like E is a conscious experience, which is constituted by an awareness of its sensory content. To put it another way, S’s awareness of a sensory experience is infallible because it is this awareness of sensory content that makes her experiential state what it is, and hence there is no logical room for this awareness to be mistaken about the content in question. Therefore, S’s awareness of the accuracy of fit between a sensory experience and a belief can provide her with a good reason for thinking that the belief in question is true, without raising a further issue of justification. BonJour says:
Davidson and the others are right in thinking that no intelligible relation of justification can hold between non-conceptual sensory content and conceptual beliefs. I have argued that a special sort of justificatory relation can exist in the specific case where the conceptual belief is a purported description of the conscious, non-conceptual experiential content itself, but the clear implication of that discussion is that this is the only sort of case in which such a justificatory relation between non-conceptual experience and belief is possible.Footnote 6
Along this line of reasoning, BonJour claims that his version of internalist foundationalism can avoid the aforementioned Sellarsian dilemma by grasping one of its horns; for, on his view, the non-conceptual awareness of sensory content can provide the subject with a good reason for holding a foundational belief.
3. The Alleged Constitutive Awareness of Sensory Content and Two Requirements for Being a Good Reason
In this section, I shall argue that there are two requirements for being a good reason, and that BonJour’s conception of direct awareness of sensory content faces a serious dilemma with regard to these requirements.
To begin with, a reason for holding a belief must be the kind of thing that has its correctness conditions. Let us call this requirement “the correctness condition requirement”. For this requirement, we should note the difference between a causal response and a normative response. A mere causal response has no correctness conditions; for example, if someone makes a noise “Ouch” as a mere causal response to some external stimulus, that noise is not the kind of thing that can be either correct or incorrect. By contrast, a normative response has its correctness conditions; for example, if a certain child says “This is red” in response to a visibly blue thing, this normative response is incorrect. Given this distinction between a mere causal response and a normative response, at least from the internalist perspective, the alleged direct awareness of sensory content must be a normative response, which has its correctness conditions. This is because the alleged direct awareness is supposed to provide the subject with a justifying reason for a belief, and a mere causal response cannot be even a bad or incorrect reason.
In addition, from the internalist perspective, for a belief to be justified for a particular person, it is necessary not only that there be a good reason available to the subject that could provide a basis for justification, but also that the subject take the reason to be relevant to the appropriateness of the belief.Footnote 7 And to say that the subject takes a reason to be relevant to the appropriateness of a belief is to say that he or she takes it as a good reason for accepting the belief. The subject can have three cognitive attitudes towards a reason: taking it as correct, taking it as incorrect, or withholding evaluation on it. And something that one takes as incorrect, or something on which one withholds evaluation, cannot be a good reason for holding a belief. Therefore, from the internalist perspective, having a good reason for holding a belief requires that the subject take the reason as correct (or good). To take something as correct is to endorse it. Thus, let us call this condition “the endorsement requirement”.
In the remainder of this section, I shall argue that BonJour’s conception of direct awareness of sensory content faces a serious dilemma in connection with these two requirements. Let me begin by arguing that BonJour can hardly deny the correctness condition requirement.
First, to say that an alleged direct awareness of sensory content has no correctness conditions is tantamount to saying that it is a mere causal response. Consider again the case where S holds a belief that B, namely, that it looks to her as if there is a blue book in front of her. On BonJour’s view, for S to be justified in believing that B is true, she must have an internally available reason for thinking that B is true; and such a reason is her conscious awareness of sensory experience E. Now, given the distinction between a mere causal response and a normative response, the alleged direct awareness of E is either a mere causal response or a normative response. Consequently, if he denies that it has its correctness conditions, this is tantamount to saying that it is a mere causal response. However, from the internalist perspective, a mere causal response, which has no correctness conditions, cannot be an internally available reason which confers justification on a belief.
A related point is worth adding here. In the above case, the conceptual content of B is correct just in case it is a fact about S’s experience that it looks to her as if there is a blue book in front of her. Thus, the fact that it looks to S as if there is a blue book in front of her makes her belief that B true, but the fact by itself does not make her belief justified. In order for the belief to be justified, S’s awareness of E must be a good reason that confers justification on the belief. In other words, making-true is one thing, and making-justified is another. For this reason, from the internalist perspective, the fact that S has a certain conscious experience does not, by itself, confer justification on a belief.
Second, BonJour does apply terms of epistemic appraisal such as “infallible” and “unmistaken” to the alleged direct awareness of sensory content. He says:
Consider a state of, e.g., visual experience, such as the one that I am presently having as I sit at my computer table. … [I]t essentially involves a constitutive, or “built-in,” non-apperceptive awareness of its own distinctive sort of content, namely sensory content. And … such a constitutive awareness of sensory content is in no need of justification and is indeed infallible in the sense that there is no sort of mistake that is even relevant to it.Footnote 8
In the aforementioned case, S’s belief that it looks as if there is a blue book in front of her is correct just in case it is a fact about her experience that it looks as if there is a blue book in front of her. Thus, we may say that S’s awareness of the experience is unmistaken just in case it is indeed a fact about her experience that it looks as if there is a blue book in front of her. This is why S’s awareness of the experience can be a justifying reason for thinking that B is true. An important point for now is that terms of epistemic appraisal such as “infallible” and “unmistaken” cannot be applied to something which has no correctness conditions. To put it another way, the concept of mistake (or infallibility) cannot be applied to something which has no correctness conditions. Thus, if the alleged direct awareness has no correctness conditions, it simply does not make sense to say that it is unmistaken (or infallible).Footnote 9
Third, and more importantly, BonJour admits that one’s assessment as to whether there is the accuracy of fit between the sensory content of an experience and the conceptual content of a belief is normative in a broad sense. He says:
The relation between this non-conceptual content and such a conceptual description thereof … is a descriptive relation, having to do with the accuracy or inaccuracy of fit between a conceptual description and a non-conceptual object that the description purports to describe. And while the assessment or evaluation of the accuracy of a description is not quite the same thing as the logical assessment or evaluation of an inference, it is nonetheless normative and even logical in a broader sense that would have no application to a merely causal relation.Footnote 10
In the aforementioned case, S can take the alleged awareness of E as a good reason for thinking that B is true, if she recognizes the accuracy of fit between the sensory content of E and the conceptual content of B. By contrast, S cannot take the alleged awareness of E as a good reason for accepting her belief in question, if she recognizes the inaccuracy of fit between the two. In this way, S can evaluate the alleged awareness of an experience either as an adequate basis for justification or an inadequate basis for justification. In other words, S can take the alleged direct awareness either as a good reason or a bad reason.
Due to the above reasons, BonJour, as an internalist, can hardly deny the correctness condition requirement. Accordingly, let us turn to the case where he accepts the correctness condition requirement. If he accepts the correctness condition requirement, it directly conflicts with his view that sensory experiences are neither true nor false, and that direct awareness is more primitive than judging. More importantly, a serious dilemma arises in connection with the aforementioned endorsement requirement.
Suppose, on the one hand, that he accepts the endorsement requirement. Given that the alleged awareness of an experience has its correctness conditions, and also given that S takes it as a correct (or good) reason, the awareness of an experience becomes virtually indistinguishable from a belief, so that the former cannot be a justifying reason for the latter. Let me explain this by comparing the following three:
(i) S is aware of the sensory content of an experience.
(ii) S is aware that it looks as if there is a blue book in front of her.
(iii) S believes that it looks as if there is a blue book in front of her.
Suppose that, on the basis of experience E, S comes to hold the belief that B, namely, that it looks as if there is a blue book in front of her. This belief is correct just in case it is a fact about her experience that it looks as if there is a blue book in front of her. In addition, S’s awareness of E is the only justifying reason for her belief. Thus, the awareness of E can serve as a good reason for accepting her belief that B only if what she is directly aware of is about the fact of her experience that it looks as if there is a blue book in front of her. Moreover, due to the correctness condition requirement, S’s awareness of an experience is either correct or incorrect. Therefore, we may say that S’s awareness of the experience is correct just in case it is a fact about her experience that it looks as if there is a blue book in front of her. Here note that S’s awareness involved in the case of (ii) is also correct just in case it is a fact about her experience that it looks as if there is a blue book in front of her. Consequently, S’s awareness involved in (i) and that involved in (ii) have the same correctness conditions, and so (i) and (ii) are merely different ways of expressing virtually the same thing. In addition, due to the endorsement requirement, S’s direct awareness of E cannot be a good reason for holding her belief that B unless she takes the awareness of E as a good reason for the belief. As noted, S’s direct awareness of E is about a fact of her experience that it looks as if there is a blue book in front of her. Thus, what S endorses in this case cannot be something other than the fact that it looks as if there is a blue book in front of her. Therefore, if S’s awareness of E involves her endorsement of this fact, this awareness is virtually indistinguishable from her belief that B. If S’s awareness (i) is virtually indistinguishable from her belief (iii) in this way, the former cannot be a good reason for holding the latter.
Suppose, on the other hand, that BonJour does not accept the endorsement requirement. His denial of the endorsement requirement agrees with his view that direct awareness is more primitive than judging, and so it does not involve any endorsement on the side of the subject. But this option poses a serious problem for him. In order for S to be justified in believing that B, from the internalist perspective, S’s awareness of E must be a good reason for holding the belief. From the internalist perspective, the source of justification must be cognitively accessible to S. In addition, given the correctness condition requirement, the source of justification for a belief has its correctness conditions. If the source of justification is cognitively accessible to S and it has its correctness conditions in this way, from the internalist perspective, it should be possible that S has a cognitive attitude towards it. To put it another way, from the internalist perspective, S must be able to take her awareness of E as a good reason for her belief that B. As mentioned before, one can have three cognitive attitudes towards a reason: taking it as correct, taking it as incorrect, or withholding evaluation on it. And something that one takes as incorrect, or something on which one withholds evaluation, cannot be a good reason for holding a belief. Consequently, to deny the endorsement requirement is tantamount to denying that the subject needs to have a cognitive attitude towards the source of justification for a belief; but a person who denies that the subject needs to have a cognitive attitude towards the source of justification has no good reason to prefer internalism to externalism. But such a position cannot be a plausible internalist view.
For the last-mentioned point, it is worth noting that, as Michael Bergmann points out, internalism about epistemic justification is mainly motivated by the so-called “Subject’s Perspective Objection to Externalism.”Footnote 11 This objection is roughly as follows. Suppose that S has some justification-contributor X in some way for her belief that B. But suppose also that she does not conceive of X as something relevant to the appropriateness of her belief; that is, she is ignorant of the relevant connection between X and the truth of B. Then, although X could be relevant in some way to the appropriateness of her belief, she has no idea why this is so. In such a case, from her subjective perspective, it would be an accident that B is true (if it is true). If it is just an accident from her subjective perspective that her beliefs are true, from her subjective perspective, she has no good reason to think that B is true. According to Bergmann, this objection, or something very much like it, is widely endorsed by internalists such as BonJour, Moser, Fumerton, and Lehrer;Footnote 12 one’s alleged commitment to internalism is subject to serious challenge in case one’s view is vulnerable to this kind of objection. This is because someone who cannot appeal to the main motivating argument against externalism is left without a good reason to prefer internalism to externalism.Footnote 13 Due to this sort of reason, BonJour’s denial of the endorsement requirement would leave him without solid ground from which to complain about externalism.
4. Second-order Belief Awareness
BonJour considers the relation between a first-order occurrent belief and its corresponding first-order belief awareness as being parallel to the one between a second-order belief about an occurrent belief and its corresponding second-order belief awareness; and he derives his idea of the so-called “constitutive awareness of sensory content” from his conception of conscious occurrent belief. Thus, let me briefly discuss his view about the latter.
On his view, if S is aware of the propositional and assertive content of the thought that P, this awareness provides her with a good reason for thinking that she has an occurrent belief that P; in other words, S’s second-order belief awareness provides a good reason for accepting her second-order belief. And, on his view, S’s second-order belief, such as her belief that she has an occurrent belief that P, is almost infallible because there is a clear and internally accessible reason for thinking that it is true. This reason is her second-order belief awareness, that is, her awareness of the propositional and assertive content of the thought that P. As we shall see shortly, this view is implausible, and so BonJour fails to show that the alleged constitutive awareness of content is the ultimate source of justification, which can stop the regress of justification.
To begin, I shall argue that BonJour can hardly deny that S’s second-order belief awareness has its correctness conditions. The reasons why second-order belief awareness is not a mere causal response are parallel to the reasons why first-order belief awareness is not a mere causal response. From the internalist perspective, a mere causal response cannot be a good reason that confers justification on a second-order belief. In addition, S’s assessment as to whether there is the accuracy of fit between the content of her second-order belief awareness and the content of the second-order belief must be normative (at least in a broad sense). For example, S can take her second-order belief awareness as a good reason for thinking that she has an occurrent belief that P, if she recognizes the accuracy of fit between the content of her second-order belief awareness and the content of the second-order belief. By contrast, S cannot take her second-order belief awareness as a good reason for thinking that she has an occurrent belief that P, if she recognizes the inaccuracy of fit between the two. In this way, S can evaluate the alleged second-order belief awareness either as an adequate basis or an inadequate basis for the justification of her second-order belief.
Thus, let us turn to the case where BonJour accepts the correctness condition requirement. As we shall see shortly, a serious dilemma arises in connection with the endorsement requirement. Suppose, on the one hand, that he accepts the endorsement requirement with regard to second-order belief awareness. Given that the alleged second-order belief awareness has its correctness conditions, and also given that S takes it as correct, the alleged second-order belief awareness becomes virtually indistinguishable from a second-order belief for the following reason. Compare the following three:
(i) S is aware of the propositional and assertive content of the thought that P.
(ii) S is aware of the fact that she mentally assents to the thought that P.
(iii) S is aware that she mentally assents to the thought that P.
Suppose that, on the basis of her awareness involved in (i), S comes to have an occurrent belief that P. One thing to note here is that S does not have an occurrent belief that P unless she mentally assents to the thought that P. Another thing to note is that her second-order belief awareness above is the only justifying reason for thinking that she has an occurrent belief that P. Thus, S’s second-order belief awareness above is correct just in case she mentally assents to the thought that P. Thus, (i) and (ii) have the same correctness conditions, and so they are merely different ways of expressing virtually the same thing. Likewise, (ii) and (iii) have the same correctness conditions, and so they are also merely different ways of expressing virtually the same thing.
According to BonJour, S’s second-order belief awareness is distinguished from her second-order belief that she has an occurrent belief, so that the former can be a good reason for holding the latter. But this view is problematic. As noted before, what S is aware of cannot be a good reason for holding a belief unless she takes it as correct; and to say that S mentally assents to the thought that P is tantamount to saying that she has an occurrent belief that P. Therefore, (iii) is virtually indistinguishable from the following:
(iv) S has an occurrent belief that P.
Consequently, given the aforementioned two requirements for being a good reason, S’s awareness of the propositional and assertive content of the thought that P is virtually indistinguishable from her belief that she has an occurrent belief that P.
In addition, on BonJour’s view, S’s belief that she has an occurrent belief that P is almost infallible, because there is no independent fact or situation for S’s awareness to be mistaken about. But this view is also implausible for the following reason. Suppose that S has an occurrent belief that P, and consider how she comes to know this fact. S can consider whether she mentally assents to the thought that P. If she thinks that she mentally assents to the thought that P, she can endorse that she has an occurrent belief that P. The reason is straightforward. If a thought occurs to one, then one can either mentally assent to its content or not. And one’s mental assent is the most important criterion for ascribing a belief to oneself: if one mentally assents to the thought that P then one will think and act on the assumption that P; thus, if one mentally assents to the thought that P, one can self-ascribe the belief that P. One’s belief that P is conscious if one can self-ascribe this belief without recourse to any evidence about one’s behavior.Footnote 14 And one can self-ascribe a belief in this way if one can do so simply on the basis of the fact that one mentally assents to the thought that P. Here note that when one has a conscious belief that P, what one is (introspectively) aware of is just that one mentally assents to the thought that P. Note also that one can easily ascertain whether one mentally assents to the thought that P simply by considering it. If one thinks that one mentally assents to the thought that P after considering whether one mentally assents to it, one can thereby judge that one mentally assents to the thought that P. If so, one can self-ascribe the belief that P. And if one can self-ascribe the belief that P in this way, one has thereby a conscious belief that P. This is why one has a special epistemic access to one’s own current conscious belief. Therefore, S’s belief that she has an occurrent belief that P is almost infallible simply because S can easily ascertain whether she mentally assents to the thought that P, and if S mentally assents to the thought that P then she has thereby an occurrent belief that P. It is not because there is no independent fact or situation for S’s awareness to be mistaken about.
Suppose, on the other hand, that BonJour does not accept the endorsement requirement with regard to second-order belief awareness. On his view, S is justified in believing that she has an occurrent belief that P on the basis of her second-order belief awareness. Thus, from the internalist perspective, the second-order belief awareness must be a good reason for holding the second-order belief. From the internalist perspective, the source of justification must be cognitively accessible to the subject. In addition, given the correctness condition requirement, the source of justification for a belief has its correctness conditions. If so, from the internalist perspective, it should be possible that S has a cognitive attitude towards the source of justification. To put it another way, from the internalist perspective, S must be able to take her second-order belief awareness as a good reason for her second-order belief. Consequently, as argued in the previous section, to deny the endorsement requirement is tantamount to denying that the subject needs to have a cognitive attitude towards the source of justification for a belief; and a person who denies that the subject needs to have a cognitive attitude towards the source of justification has no good reason to prefer internalism to externalism. But again, such a position cannot be a plausible internalist view.Footnote 15
5. BonJour’s Explanatory Account and the Problem of Representation
According to BonJour, empirical justification has a foundation in beliefs about sensory experiences, and beliefs about physical objects and their properties are non-foundational beliefs. In addition, on his view, the conceptual characterization of an experience in terms of phenomenological appearances is epistemologically prior to the conceptual characterization of an experience in terms of physical objects and their properties. He says:
My experience may be such as to incline me to think that a rabbit is present, but the experience cannot by itself reveal that it is actually of the sort that is normally caused by rabbits. A useful way of putting the point is to say that the claims about physical appearances or ways of being appeared to that constitute our primary conceptual formulations of sensory experience must be understood in something like what Chisholm has called the “descriptive, non-comparative” sense of the terms or concepts in question, for only in that sense can the claim to be “appeared to” in a certain way be adequately justified simply by appeal to our constitutive, nonconceptual awareness of sensory content. … [W]e have the ability to grasp or represent the character of our perceptual experience fairly accurately, albeit somewhat obliquely, in terms of such physical object appearances.Footnote 16
[T]here are features of the experiential content itself that are strongly and systematically isomorphic or structurally similar to the correlated situations. … But this seems to indicate in turn that the conceptual characterization of experience in physical-object terms cannot be taken as epistemologically primary after all, that something like a sense-datum or phenomenological characterization is needed if an adequate basis for the inference to the physical world is to be found.Footnote 17
Consequently, BonJourean foundational beliefs about sensory experiences are compatible with various competing non-foundational beliefs about the physical world. The reason is straightforward. Suppose that S has an appearance belief that it looks as if there is a blue book in front of her. This appearance belief is about S’s inner experience rather than how things are in the external world, so that it can be true even when there is no blue book in front of S. In other words, it is possible that S has a visual experience of the sort you normally have when there is a blue book in front of you despite the fact that there is no blue book in front of S, possibly due to adroit manipulation by the evil demon. If foundational beliefs about sensory experiences are compatible with various competing non-foundational beliefs about the physical world, how can one infer non-foundational beliefs from foundational beliefs? BonJour appeals to an inference to the best explanation.
He claims that sceptical hypotheses are less likely to be true than the commonsense hypothesis, according to which one’s perceptual experiences are due to the physical world.Footnote 18 He offers two reasons. First, the commonsense hypothesis has an analog character, because it could explain the spatial features of experiences in analog fashion by the spatial features of the world. By contrast, according to sceptical hypotheses, the features of the world that are reflected in experiences can be completely different from the true features of the world; for example, our experiences manipulated by the evil demon might be unrelated to the true features of the world. Thus, he takes a sceptical hypothesis to have a digital character. Now, it is very unlikely that the complicated features of sensory experiences and the corresponding physical situations are related as a mere correlation. For this reason, he claims that the features of sensory experiences are strongly suggestive of the correlated physical situations, so that it is unreasonable to adopt a sceptical hypothesis until there is some positive reason to think otherwise. Second, the explanatory success of a sceptical hypothesis depends on two things. One is that the sceptical hypothesis can account for the explanation in question by emulating the causal-explanatory structure postulated by the commonsense hypothesis. The other is that the specific translating mechanism postulated by the sceptical hypothesis can do the job of emulation successfully. According to BonJour, the explanatory success of the commonsense hypothesis does not depend on the second condition, so it is less vulnerable to problems than sceptical hypotheses.Footnote 19 On the basis of these considerations, he claims that if one has an appearance belief that it looks as if P then, in the absence of any defeating considerations, one is justified in believing that P. The reason is again that the complicated patterns of sensory experiences are best explained by the commonsense hypothesis, according to which these patterns are systematically caused by the corresponding physical situations.
However, BonJour’s explanatory account above faces the problem of representation, which I shall explain shortly. As Hilary PutnamFootnote 20 and Tyler BurgeFootnote 21 have convincingly argued, the mental contents of at least some of our thoughts are individuated in part by environmental or socio-environmental factors. To illustrate, consider Putnam’s famous Twin Earth argument. Suppose that somewhere in another galaxy there is a planet, called “Twin Earth”, which is an exact duplicate of our Earth except for one thing: what looks and behaves like water on Twin Earth has a different chemical composition, say XYZ. Let us call this different liquid “twater”. Then the concepts water and twater involve different contents, because the former is correctly applied only to H2O, whereas the latter is correctly applied only to XYZ. Therefore, if Oscar on Earth has a thought expressed by his sentence “there is a glass of water in front of me”, it is a thought about H2O, because his use of the word “water” is correctly applied to H2O. By contrast, if Oscar’s exact duplicate on Twin Earth, say Toscar, has a thought expressed by his sentence “there is a glass of water in front of me”, it is a thought about twater, because his use of the word “water” is correctly applied to XYZ. Along this line of reasoning, we may say that at least some of our mental contents are partly determined by relations to the environment, which may in some respect be unknown to us; note that Oscar and Toscar might not be able to distinguish between water and twater.
On BonJour’s view, as noted, the conceptual characterization of an experience in terms of phenomenological appearances is epistemologically prior to the conceptual characterization of an experience in terms of physical objects and their properties. Thus, he takes seriously the idea that physical-object appearance beliefs are compatible with various competing physical-object beliefs. But this view faces the problem of representation of the following sort. Suppose that Oscar on Earth has an appearance belief that it looks as if there is a glass of water in front of him. On BonJour’s view, the concept being appeared to be water is epistemologically prior to the concept being water; roughly put, it is not that liquids that are water are normally appeared to be water, but rather that liquids that are normally appeared to be water are water. But “water” here is a natural kind term that is correctly applied only to H2O. Thus, Oscar’s thought “there appears to be a glass of water in front of me” is a thought about H2O, whereas Toscar’s thought “there appears to be a glass of water in front of me” is a thought about XYZ. But their appearance beliefs are indistinguishable from each other, because their sensory experiences are indistinguishable. Consequently, Oscar cannot infer by way of an inference to the best explanation that there is a glass of water in front of him from his appearance belief; for the concept being appeared to be water understood in the way BonJour suggests is neutral between the concepts being water and being twater. Note that Oscar might not know the distinction between water and twater. Therefore, from his subjective perspective, Oscar could not know that his appearance belief is best explained by the hypothesis that there is a glass of water, instead of the hypothesis that there is a glass of twater. To put the point another way, given the concept being water, we can easily form the concept being appeared to be water; likewise, given the concept being twater, we can easily form the concept being appeared to be twater. But the converse is not the case. If one begins with the concept being appeared to be water in the way BonJour suggests, it is difficult for one to form the concept being water, nor the concept being twater; for the concept being appeared to be water understood in this way is neutral between the concepts being water and being twater. Due to the problem of representation of this sort, BonJourean appearance beliefs cannot be adequate foundations of knowledge about the physical world.
To put the same point another way, one’s appearance beliefs are too weak a foundation to support our knowledge of the world. In the first place, as noted, one’s sensory experiences are compatible with various alternative hypotheses about the world. For example, Oscar’s appearance belief that there appears to be a glass of water is compatible not only with the hypothesis that he is on Earth, but also with the hypothesis that he is on Twin Earth. Therefore, there is no way for Oscar to tell which hypothesis is the best explanation for his sensory experiences. In other words, even though one’s sensory experiences are very well explained by a certain hypothesis, this fact does not show that the hypothesis is correct. In the second place, sensory experiences that a person alone can have are bound to be very limited. But, as far as our present world-view is concerned, there are more facts in the world than the entire human race have found out so far. In addition, the correct world-view must cohere with a lot of people’s various observations about the world. Thus, the correct world-view we seek to acquire cannot be built up merely on the basis of one’s very limited sensory experiences.
One more thing to note is that a competing explanation, which has equal explanatory power, does not have to be a digital explanation. For example, the hypothesis that there is a glass of water in front of Oscar and its competing hypothesis that there is a glass of twater in front of him are equally good at explaining his experiences; for water and twater are phenomenally indistinguishable. Here the competing hypothesis is an analog explanation, because Twin Earth is a nearly exact duplicate of our Earth.
6. Concluding Remarks
In this paper, I have argued for three claims to show that BonJour’s alleged solution to the Sellarsian dilemma is unsuccessful. First, his notion of non-conceptual awareness of a sensory experience faces a serious dilemma with regard to two requirements for being a good reason: the correctness condition requirement and the endorsement requirement. Second, BonJour derives his idea of the so-called “constitutive awareness of content” from his conception of conscious occurrent belief. But this conception also faces an analogous dilemma with regard to the two requirements for being a good reason. Third, his explanatory account faces the problem of representation. On his view, the concept being appeared to be water is epistemologically prior to the concept being water. But one cannot infer by way of an inference to the best explanation that there is a glass of water from one’s appearance belief; for the concept being appeared to be water understood in the way BonJour suggests is neutral between the concepts being water and being twater.
I’d like to thank two anonymous referees for their helpful comments on an early version of this paper.
