Erkenn (2009) 71:19–34 DOI 10.1007/s10670-009-9175-9 ORIGINAL ARTICLE

Introspective Self-Knowledge of Experience and Evidence Frank Hofmann

Received: 10 March 2009 / Accepted: 10 March 2009 / Published online: 30 April 2009
Springer Science+Business Media B.V. 2009

Abstract The paper attempts to give an account of the introspective self-knowledge of our own experiences which is in line with representationalism about phenomenal consciousness and the transparency of experience. A two-step model is presented. First, a demonstrative thought of the form ‘I am experiencing this’ is formed which refers to what one experiences, by means of attention. Plausibly, this thought is knowledge, since safe. Second, a non-demonstrative thought of the form ‘I am experiencing a pain’ occurs. This second self-ascription is justified inferentially, on the basis of the first, demonstrative thought. Thus, an account of introspective experiential self-knowledge can be developed which is richer and more adequate to the phenomena than pure reliabilism and Dretske’s displaced perception model. There is really such a thing as introspection, but no inner sense.

1 Introduction How do we know that we have a certain experience?—I submit that the right answer to the question, ‘‘How do you know that you have a pain in your foot?’’, is not, ‘‘I feel it’’, but rather, ‘‘I notice it.’’ Having a pain is one thing; noticing it is another. By noticing our experiences we come to know that we have them. Most of our experiences come and go without ever being noticed. But once noticed, an experience has entered the grip of cognition. Some have wondered how to solve the trilemma about introspective selfknowledge posed, most famously, by Paul Boghossian. (Let us take the expression ‘self-knowledge’ in a wide sense, encompassing both knowledge in the strict sense and justified belief. Later we will have to distinguish between knowledge and

F. Hofmann (&) Universita¨t Tu¨bingen, Philosophisches Seminar, Bursagasse 1, 72070 Tu¨bingen, Germany e-mail: [email protected]

123

20

F. Hofmann

justification. But let us ignore this distinction for the moment.)1 The trilemma applies to both introspective self-knowledge of one’s own propositional attitudes and introspective self-knowledge of one’s own experiences. In the following, I will be concerned only with the latter kind of self-knowledge, introspective selfknowledge of one’s own experiences (though I believe the account to be presented can be extended to cover also the former case). For both cases, the trilemma goes like this: Introspective self-knowledge is either inferential, observational (or ‘quasiperceptual’), or based on no ground (or evidence) at all. That introspective selfknowledge is not inferential is taken as obvious by many. That it is not observational (or ‘quasi-perceptual’) has been argued extensively, most notably by Sydney Shoemaker (1994). And that it is not based on any ground (or evidence) at all is quite implausible. For if it is based on nothing at all, then it does not seem to be a cognitive achievement. But, prima facie, self-knowledge is a cognitive achievement.2 So what are we to say?—I submit that the solution to the trilemma lies in distinguishing two steps in the formation of introspective beliefs about one’s own experiences. The first step is a demonstrative thought about one’s experience which is a piece of knowledge. This demonstrative thought can be identified with the state of noticing one’s experience. The introspective thought that I am experiencing this right now (where ‘this’ refers to the relevant intentional object of my experience) is safe and, hence, knowledge. The second step, then, is based on the first one, and it yields an evidentially justified non-demonstrative introspective belief. This is the ordinary, non-demonstrative thought that I am having a pain in my foot. So, on the one hand, introspective beliefs are inferentially justified. The ordinary, non-demonstrative thought that I am having a pain in my foot is inferentially justified, justified on the basis of a demonstrative thought serving as its evidence. This demonstrative introspective thought, on the other hand, is not so easily classified. But most plausibly, it can be classified as neither inferentially based nor observational. It is not based on any ground at all. It is not based on the experience, since the experience itself is constitutive of the demonstrative introspective thought. So the relation is, in one sense, tighter or stronger than the grounding relation between evidence and what is justified by evidence. It is not a contingent or causal relation. (This much is a truth in so-called ‘constitutive accounts of self-knowledge’.)3 It seems appropriate to speak of ‘direct knowledge’. In this paper, I would like to present an account of introspective self-knowledge of our experiences which takes introspection seriously. There really is such a thing

1

See, for example, Boghossian (1989). There, Boghossian poses the problem for knowledge in the wide sense just introduced, encompassing both knowledge in the strict sense and justified (true) belief.

2

Interestingly, when discussing the third option, Boghossian mentions the relevance of attention as a general feature of introspective thoughts which is not exhibited by the usual cases of ‘insubstantial’ knowledge without any ground or evidence (cf. ibid., p. 76). Attention will play a crucial role in the account to be presented.

3

For example, Brie Gertler’s account of introspection and introspective self-knowledge, to which I will refer below, belongs to this class of constitutive accounts of self-knowledge. Cf. Gertler (2001). Similarly, Burge holds a constitutive account of self-knowledge (of one’s thoughts, not of one’s experiences). Cf. Burge (1988).

123

Introspective Self-Knowledge of Experience and Evidence

21

as introspecting on one’s experience—even though this act of introspection is neither observational nor ‘quasi-perceptual’. We do not perceive our experiences in any interesting sense. We notice them, by paying attention. Introspection is an attentional act whereby we notice our experiences or, more exactly, some part or aspect of our overall experience. It results in a demonstrative thought about this part or aspect of one’s experience.4 The account is in line with first-order representationalism (intentionalism) about phenomenal consciousness such as, paradigmatically, Fred Dretske’s or Michael Tye’s representationalism.5 It is also consistent with phenomenal transparency. And it preserves the idea of evidentialism: (epistemic) justification requires evidence; all justification is evidential. Three models of introspection have been proposed by leading representationalists: an inner-sense model (Lycan), the displaced perception model (Dretske) and pure reliabilism (Tye). But all of these models have to face serious problems, I think. The inner-sense model is subject to all the criticisms that have been voiced against the idea of becoming aware of one’s experience in a quasi-perceptual way. There just does not seem to be any such inner perceiving. Leaving inner-sense theories to one side, there remain the displaced perception model and pure reliabilism. But the displaced perception model is quite problematic too, as has been shown by Aydede (2003) and Kemmerling (1999), most importantly. It makes introspection inferential, but in a way which is indefensible. In order to know what experience one is having, one would have to have a connecting belief which allows one to move from a thought about the world to a thought about one’s own ‘corresponding’ experience. But there just does not seem to be any such connecting belief which can do this job in an appropriate way. In the end, a dilemma arises: Either there remains an epistemic gap, or the connecting belief already presupposes introspective self-knowledge. So we would have to take recourse to pure reliabilism, as Tye proposes.6 According to this view, there is no inner perception and there is no need for any connecting belief. Experiences just bring about introspective thoughts and beliefs, and they do so in a highly reliable way. That’s what makes the resulting states instances of self-knowledge. Reliable processes are the grounds of our selfknowledge, and they need not involve any quasi-perceptual states or supporting beliefs as long as they are reliable. What are we to say about this view?—In the end, it seems to me, this view is unsatisfactory. It may be true, and I do not have any strong and conclusive argument against it. But my impression is that pure reliabilism leaves out some important intuitions about knowledge and justification which it would be quite desirable to respect. Most importantly, the idea of evidentialism is not preserved. If pure reliabilism is correct, justification is not 4

In this paper, I will use the term ‘thought’ as referring to a conscious propositional state including the belief attitude, i.e., to conscious belief. The phenomenon of merely entertaining a content, and the distinction between merely entertaining and believing, becomes relevant only when it comes to introspective self-knowledge of one’s own propositional attitudes. Concerning this case, see the remark in the last footnote of Sect. 2.

5

Cf. Dretske (1995), Tye (1995).

6

See Tye (2000), Chap. 3.2.

123

22

F. Hofmann

evidential. And if it is not evidential, no adequate treatment of defeaters will be forthcoming. This is a major drawback of pure reliabilism, in my view. Pure reliabilism is a last resort that we may accept if no better theory can be constructed. But the hope is that we can develop a view which can preserve more of our intuitions and, thus, can be more satisfactory. In any case, this is what I want to do in the following. (I will indicate at the relevant passages where the proposed view has advantages over pure reliabilism.) Important features of the account which I am to propose are: (1) There is no inner sense. (2) The following two desiderata are satisfied: introspective self-knowledge is peculiar and privileged. (3) The only kind of awareness in introspection is thatawareness. The core of the proposal is that introspective self-knowledge has a two-steps structure. The first step is the act of introspection. By paying attention to what one experiences, one forms a thought about one’s experience. This thought is a demonstrative thought which can be expressed by saying ‘I am experiencing this right now’, or ‘This is what I am experiencing right now’. Here, the ‘this’ refers, by attentional demonstration, to the relevant intentional object of one’s experience. The second step is a transition from this demonstrative thought to a non-demonstrative thought which is a self-ascription of the relevant kind of experience. Its canonical expression is, for example, ‘I am having a pain in my left foot’. This is the ordinary, non-demonstrative self-knowledge which is usually, if not exclusively, at the heart of the discussion. The first, demonstrative self-ascription is knowledge. The second self-ascription is evidentially justified, justified by having the first self-ascription as its ground, or reason, or evidence. My plan for the following is to discuss the first and second step in more detail (Sects. 2, 3). Then, I will conclude with an outlook on the larger landscape (Sect. 4).

2 The First Step: Introspection In a very interesting paper, called ‘‘Introspecting phenomenal states’’, Brie Gertler presents an account of introspective self-knowledge of one’s experiences with the following features: (1) Self-knowledge of one’s own experiences is arrived at by introspection—and so is properly called ‘introspective self-knowledge’. (2) There is no inner sense in introspection. (3) Introspection essentially involves attention to aspects of one’s experience and demonstrative reference by attention. Even though I disagree with many aspects of Gertler’s account, I would like to develop a similar account which preserves these three features. There are many points of disagreement. What I find most plausible about Gertler’s view is her idea of demonstrative reference in thought to the content of one’s experience, brought about by demonstrative attention. I will rely on this idea in the following, without committing myself to much of the rest of Gertler’s view. In addition, I do not wish to use introspection in order to argue against first-order representationlism, as Gertler does. Quite the contrary, I wish to defend first-order representationalism, by

123

Introspective Self-Knowledge of Experience and Evidence

23

developing a plausible account of introspection which is coherent with it and supplements it.7 The first step of introspection is attention. Suppose Selma has a pain in her left foot. A part or aspect of her overall experience is this pain. At first, she does not notice her pain, she simply has it. (Her attention may be somewhere else, with some other part or aspect of her experience.) At some point, however, she notices her pain. She pays attention to what goes on in her foot, and she realizes, or notices, the pain quality that she has already experienced for some time. But she does not yet classify it conceptually as a pain. She simply notices what she experiences—the pain quality in her foot—without yet thinking that she is in pain. (Here, the expression ‘pain quality’ is meant to refer to the intentional object of the pain experience.) Still, she is having a thought now, a demonstrative thought about her experience. The act of introspecting, by attention, on one’s experience results in a demonstrative thought, a judgment to the effect that one is experiencing this, where the ‘this’ gets its referent by attentional demonstration. Finally, Selma conceptualizes what she notices herself as experiencing as a pain. She comes to think that she is having a pain in her left foot. We have to distinguish three states. First, the experience. Second, noticing the experience. This noticing takes the form of a demonstrative thought about the experience which refers to the experience’s content by means of attention. Third, the fullblooded, non-demonstrative thought which classifies the experience as a pain, for example. The latter one is the self-knowledge as it is usually discussed in the literature. What is rarely discussed and often overlooked is the second kind of state, the noticing of one’s experience.8 The following phenomenological argument in favor of this tripartite distinction can be given. Phenomonologically, it seems obvious to me that there is this further, distinct state of noticing. Intuitively, one can realize or notice what one experiences without yet being able to conceptually classify it as a so-and-so. Especially with new or complex experiences and with experiences of very low intensity this is a (not too rare) possibility. One does not yet know—and perhaps never will know—what kind of experience it is one is having. One is still in the process of ‘searching’ for the right conceptualization. Nevertheless, one does have some grip on the experience. One grasps what one experiences, and this is a (demonstrative, or indexical) conceptual, cognitive grasp.9 7

It would take another paper to lay out and discuss the other points of disagreement. Let me just mention two more. (1) The account to be proposed does not say that self-knowledge can be certain. (2) I do not think that introspection without introspective belief is possible.

8

It would be an interesting question to investigate what the account has in common with Moritz Schlick’s observation sentences or ‘Konstatierungen’. One of Schlick’s examples is the observation sentence ‘Here now pain…’ (Schlick 1934, p. 96). It is difficult to understand what exactly Schlick has in mind, and different interpretations may be possible. One interpretation is that the sentence ‘Here now pain’ expresses an introspective thought. Then, Konstatierungen would be expressions of introspective thoughts or beliefs. Another example Schlick mentions, the Konstatierung ‘Here now so and so’, is even closer to the first, demonstrative thought. One might take it as an attempt at formulating a sentence which expresses what I have described as the demonstrative thought ‘I am experiencing this right now’.

9

To speak of ‘searching for the right conceptualization’ is to be taken with care. It does not imply that some sort of comparison between the content of one’s experience and the content of certain ‘conceptual descriptions’ takes place, as Laurence BonJour suggests. Cf. BonJour (1999).

123

24

F. Hofmann

This is the phenomenological datum that I take as given. How do we describe it theoretically?—The following description and explanation seems to me very attractive. I propose that the best way to conceive of this grasp is to conceive of it as a demonstrative thought having the form ‘I am experiencing this right now’ or ‘I am having an experience of this right now’. The demonstrative concept denoted by ‘this’ refers to what one experiences, the relevant intentional object. It is picked out by attention. (This attention is just the ordinary kind of attention, used or applied in introspection. There is no special, introspective attention of its own, over and above the ordinary kind of attention. I will come back to this point later, at the end of this section.) The experience itself is embedded, or included, in the act of introspection. Selma’s introspective thought that she is experiencing this has the pain as a constitutive element. The experience fixes the referent of the introspecting thought, it provides its object. (Ontologically, we can classify the object as a quality or property or feature. What we experience are certain qualities or properties or features. For the present purposes, this seems to be an appropriate way of speaking about experiences and their contents or objects.) Let us now consider the epistemological status of introspection. How about ‘firstperson authority’?—For the following discussion I would like to accept the two desiderata that have been put forward quite succinctly by Alex Byrne, peculiarity and privilegedness.10 Introspective self-knowledge is both peculiar and privileged. Its peculiarity consists in a principled first-person/third-person asymmetry. Introspective self-knowledge is always from the first-person point of view. Only the subject herself can arrive at introspective self-knowledge about her own experiences. Introspective self-knowledge is also privileged in that it is safe, or reliable, in a special way. Any account of introspective self-knowledge should preserve these two features, I assume. In order to assess the epistemic status of thoughts and beliefs I propose to take safety as our working criterion. I believe it is more than a working criterion, but for the present purposes it will suffice to take it as such. If a thought or belief passes the test of safety, this will at least be a good reason to think it is knowledge. As an initial characterization of safety the following will be appropriate. The belief that p is safe if and only if it could not easily have been wrong. A synonymous characterization (as I assume) would be: The belief that p is safe if and only if it is true in all close possibilities (possible worlds). What could easily have been the case is what happens in a close possibility. The closeness is to be understood as ‘what could easily have happened’.11 Safety has many advantages over sensitivity, the most important alternative criterion. This has been shown in the meantime by Ernest Sosa and Timothy Williamson, most notably.12 Safety allows for closure and it preserves the possibility of knowing that one is not the victim of an evil demon or an envatted 10

Cf. Byrne (2005).

11

In the end, the safety condition has to be reformulated as a condition on the method by which a belief has been arrived at. This is necessary in order to account for knowledge of necessary truths. (I intend to work out such an account in the near future.) But for the present purposes it will suffice to take safety in the more simple version, as a condition directly on the belief in question. 12

Cf. Sosa (1999, 2002) and Williamson (2000). By now, Sosa has moved away from safety. Cf. Sosa (2007).

123

Introspective Self-Knowledge of Experience and Evidence

25

brain. Safety does not lead into skeptical consequences, at least not as directly as sensitivity. Safety is also intuitively plausible, it seems to me. If, for example, Selma’s belief that she is having a pain in her left foot is true and could not easily have been wrong, then it qualifies as knowledge. Plausibly, a belief like this is often, or in ordinary circumstances, such that it could not easily have been wrong. So we get the intuitively plausible result that Selma’s belief is a piece of self-knowledge, in ordinary circumstances. Let us now consider error possibilities in order to see whether an introspective self-belief is safe. What could go wrong?—There are three cases to be considered: (1) There is no experience at all. (2) There is an experience, but no attention directed at it. (3) There is both an experience and attention directed at it, but no demonstrative reference to what is experienced. A closer look, however, reveals that these are not really error possibilities. Rather, they are possibilities in which the demonstrative thought (that I am experiencing this right now) does not occur at all. If there is no experience, then there is no introspection and no demonstrative thought about what one experiences. One cannot have this introspective thought without having an experience, since the experience is a constitutive part of the thought. The same is true of the second possibility. Without suitable attention, no introspective thought occurs. The relevant kind of (mental) demonstration requires attention directed at what one experiences. Finally, the third possibility is also not an error possibility, since without demonstrative reference the relevant introspective thought is not present. Therefore, in all of these three kinds of cases no error arises. These are not error possibilities, but possibilities (if they are possibilities at all, which we do not have to decide for present purposes) in which the demonstrative introspective thought—for whatever reason—does not occur at all. What, then, are the real dangerous possibilities? It is hard to come up with any possibilities which are remotely plausible. The first one that comes to my mind is the possibility of mistaking a thought for an experience. So imagine that William has a certain thought, say, the thought that he has a pain in his right foot, even though he does not have a pain in his right foot. (The thought may be a conscious belief or judgment, but it could also be a conscious desire or fear.) Now he pays attention, but he mistakes his thought for an experience. His attention does not go at what he experiences but rather to the content of his thought. He demonstratively refers to what he thinks instead of what he experiences, by thinking ‘I am experiencing this right now’. We may count both what one experiences (the contents of one’s experience) and what one thinks (the contents of one’s conscious propositional attitudes) as one’s conscious contents. One can pay attention to various parts of one’s conscious contents. And in William’s case, demonstrative reference (in thought) to a conscious content takes place which is not the content of an experience. So William misclassifies it as a content of experience. (Here, I am assuming that the notion of an experience is rather narrow and does not coincide with the notion of a conscious state. Some have argued that conscious thoughts are experiences, phenomenal states, conscious states with a phenomenal character of their own.13 We do not have to decide here whether 13

Cf. Horgan and Tienson (2002), Kriegel (2002) and Pitt (2004).

123

26

F. Hofmann

this view is correct, as long as there is some way of singling out the narrower class of experiences. For example, experience in this narrower sense may be nonconceptual, whereas the conscious contents of thoughts are conceptual.) What are we to make of this possibility?—It seems to me that we have to admit that it is a real possibility. One can indeed mistake one’s thought for an experience. Ordinarily, this will be very unlikely. But in special circumstances, there may be pressures of certain sorts which can lead up to such a confusion. If one expects to be in pain, because certain strong evidences suggest it, one can come to think that one is in pain without being in pain. And then it seems also possible that one misclassifies the content of this thought as a content of one’s experience. I think we should admit that this is a possibility. But when it comes to the question of whether this possibility undermines knowledge, it seems fair to say that it does not. The possibility is not a close possibility. At least, in many cases it will not be true to say that this possibility could easily have occurred. So ordinarily, William will still know and the first error possibility does not undermine his knowledge. A second possibility is the possibility of multiple subjects. This may be even more unlikely and bizzare. But perhaps it is a real possibility, too. There may be a situation in which there is not one single subject which could be referred to by the first-person concept ‘I’ but, say, two subjects. Perhaps, split brain cases may be of this kind. Then it is perhaps possible that the subject referred to is not the one which is having the relevant experience to which the act of attention goes. I am quite unsure whether this is a real possibility. But, again, we can be generous here, since it is certainly a very remote possibility, if it is a possibility at all. Ordinarily, it will not be an easy possibility which could undermine knowledge. Again, safety yields the right verdict. (Note that if the first-person concept does not succeed at all in referring to some subject, we do not have any error possibility, but only a possibility in which the first step of introspection does not result in the relevant demonstrative thought.) No further possibilities of error come to my mind. I therefore conclude that the demonstrative self-ascription ‘I am experiencing this right now’ is safe and, thus, a piece of knowledge, at least in ordinary circumstances. Even though it is probably not infallible, since there are real possibilities of error, it is safe. In normal or typical circumstances, error possibilities are not closeby and do not undermine knowledge. Thus, the first step of introspection yields knowledge. There may be an impression that the first step does not really yield any interesting piece of knowledge. But a good part of the impression that the demonstrative thought is uninformative or insignificant may be due to a tendency to overlook that the demonstrative thought encompasses the relevant experience itself. As I said, the experience is itself part of, or embedded in, the demonstrative thought. And the experience does have a certain, determinate content (as determinate as the contents of experiences are). It is quite important to keep this in mind. For, the demonstrative thought is supposed to be the evidence for the non-demonstrative self-ascription. So the experience and its content, being part of the demonstrative thought, are part and parcel of the evidence. As we will see in more detail in a moment, the content of the experience helps to justify the final, non-demonstrative self-ascription. In the act of introspection, the experience is embedded within a demonstrative conceptualization.

123

Introspective Self-Knowledge of Experience and Evidence

27

The experience and its content is enfolded by a conceptual state, a thought. The entire state, then, serves as the evidence for the second step, the non-demonstrative self-ascription. We have seen that the first step of introspection yields a demonstrative thought which is safe and, therefore, knowledge. In one good sense, then, the first (demonstrative) kind of self-knowledge is privileged. Indeed, the safety seems to be especially robust. If one knows that this is what one is experiencing right now, by introspecting on one’s experience, error is really almost impossible. This may account for the intuition that self-knowledge is ‘privileged’ in the sense of being a special kind of knowledge which is reliable or safe in a special way. The demonstrative self-knowledge is also peculiar. There is a principled firstperson/third-person asymmetry. Only the subject herself can demonstratively refer to her experience in such a way as to form the demonstrative introspective thought. (One would have to go as far as to consider possibilities such as the multiple subjects scenarios in order to reach the gray area of the first-person/third-person boundary.) The connection is not just contingent, as pure reliabilists want to have it. Knowledge is not merely due to a highly reliable process, connecting the experience and the thought. The experience is constitutive of the demonstrative self-knowledge. This much is the truth in recent so-called ‘constitutive accounts of self-knowledge’. (But it is only true of the first step, the demonstrative self-knowledge, not of the second step). So this kind of self-knowledge is indeed peculiar. The knowledge and what is known are in a more intimate relation than in cases of perceptual knowledge about the world, for example. And the specialness or peculiarity hinges on the firstperson/third-person distinction. It is simply due to the nature of demonstrative reference to one’s experience in thought that the first-person perspective is set apart from the third-person perspective. Only the subject herself can introspectively attend to what she experiences; only she can notice what she experiences in the relevant sense. Therefore, demonstrative introspective self-knowledge passes the test of peculiarity. A few more words about attention may be in order. To speak of ‘introspective attention’ or ‘attention in introspection’ is not meant as implying that there is a special kind of attention which is distinct from the usual kind of attention in perception and sensation. One can perceive or sense various qualities or features with various degrees of attention, roughly speaking, with or without attention. When having a visual experience of two red tomatoes lying on the table in front of one, one can attend to the color of the left tomato. Equally, one can attend to what one feels in one’s right foot. This is ordinary perceptual-sensational attention. When it comes to introspection, one question is: Is there such a phenomenon as paying attention to one’s experience? The proposed view gives a positive answer to this question. We can indeed attend to our experiences in introspection. But the attention used in introspection is not another and different kind of attention. So the second question is: Is introspective attention a kind of attention of its own, distinct from the ordinary perceptual-sensational attention? The proposed view gives a negative answer to this second question. Introspective attention simply goes to the object of the experience, the feature or quality which is experienced—just as the ordinary kind of attention. It is not a distinct kind of attention of its own. Therefore, it does

123

28

F. Hofmann

not detect any further, intrinsic feature of the experience.14 This is just as the transparency intuition says. What we come to know about our experiences by introspection is just what it is that we experience and no further, intrinsic property of the experience, as Moore, Harman, Tye, Dretske, and others have observed.15 So two important theses are part of the account of introspection that I am offering. First, there is only one kind of attention (leaving thought attention to one side), the ordinary perceptual-sensational kind of attention, and it is used in introspection. And second, this is compatible with, and even helps to explain, the transparency intuition. Because ordinary perceptual-sensational attention goes to the objects of experience, and it is used (or re-used, if you like) in introspection, introspection does not detect any further, intrinsic features of our experiences; it only informs us about what we experience. There is no need for any further, special introspective attention.16 Suppose, therefore, that demonstrative introspective self-knowledge works in the way suggested, and that it is privileged and peculiar, as I have attempted to show. How about the second step? How about the non-demonstrative, standing selfascription?17

14 Of course, there is another kind of attention, ‘thought attention’, as one could call it. When introspecting, one is paying attention to the question of what one is experiencing, whereas, when forming perceptual beliefs about one’s environment, one is paying attention to the question of what there is out there. This is a different focus of one’s thinking. And of course, there is thought attention in this sense. But this is uncontroversial. No-one needs to deny that there is this further kind of attention, and no-one should. The only relevant question is whether there is another kind of attention, over and above the ordinary attention involved in sense perception and thought attention—a special kind of ‘introspective attention’ which goes to the phenomenal qualities of one’s states ‘directly’. This is what I am denying. 15

Cf. Tye (2000), Chap. 3, for a detailed discussion of transparency. (There, reference to the other authors can be found). 16 At this point, the view departs from self-representationalism as proposed by Terry Horgan and Uriah Kriegel. It is part of self-representationalism, as I understand it, that there is a special, further kind of attention, one which is used exclusively in introspection and which yields a transition from a selfrepresentational ‘proto-belief’ (which is constitutive of any experience and which is the inner awareness of, or in, experience) to a fullblooded introspective belief which is self-knowledge. See, for example, Horgan and Kriegel (2008, unpublished manuscript), Kriegel (2008). As Kriegel realizes, this seems to threaten the transparency intuition, at least prima facie. But in the end, Kriegel thinks, transparency does not raise any serious difficulty for self-representationlism. Here, I cannot go into a discussion of selfrepresentationalism, simply for reasons of space. Let me just mention that I am not convinced that selfrepresentationalism is entirely consistent with transparency. The view I am advocating here is in line with first-order representationalism, and it can both preserve the transparency intuition without any reservation and deny the existence of any distinct and special introspective attention. These seem to me to be two important advantages. (For a range of discussions of self-representationalism, see Kriegel and Williford (2006). An interesting criticism of self-representationalism related to the issue of attention can be found in Gennaro (2008)). 17 How about introspective self-knowledge of one’s own propositional attitudes?—It seems to me that the first step of the model can be extended in order to cover also the case of propositional attitudes. In this case, one comes to think that this is what one is thinking, while being in some conscious propositional state, e.g., the conscious belief that water is wet. One has the thought (that water is wet) and thinks about it, in a demonstrative way, at the same time. The model, thus, provides a way of understanding what Burge has said about what he calls ‘cogito-like thoughts’. The thought expressed by saying ‘I am thinking that water is wet’, understood in the proposed way, is a demonstrative thought about a thought that one is having at the same time. Plausibly, this yields (self-)knowledge. I intend to work out this extension in a

123

Introspective Self-Knowledge of Experience and Evidence

29

3 The Second Step: Non-Demonstrative Self-Ascription Now it is time to distinguish more strictly between knowledge and (epistemic) justification. Let me start with some general considerations about knowledge and justification. I will assume that knowledge does not require justification. To some extent, this is a terminological matter, but not entirely. One may use the term ‘justification’ in a very wide sense such as to cover all kinds of positive normative (epistemic) status (except for the status of being true), including what has been called ‘warrant’ (that which turns true belief into knowledge), ‘entitlement’, ‘well-groundedness’, ‘appropriateness’, and the like. I will not use ‘justification’ in this very wide sense. Rather, I think there is a kind of positive epistemic status, properly called ‘justification’, which is more special (but still not very special) and which is strictly linked to having evidence and basing one’s judgment or belief on the evidence one has. Justification in this sense is a paradigmatic epistemic achievement of subjects. One accomplishes it if one bases one’s belief on adequate evidence, if one believes what one believes for a good reason. Justification is necessarily linked to good reasons. This strikes me as an obvious intuition about justification, as I understand this notion. But, as I said, this is in part a terminological matter. It is not entirely terminological, though. One can use the term ‘justification’ for a variety of different features, in a more or less encompassing sense. To use it in the sense I just indicated is to make (at least) the claim that there is feature belonging to this variety which is often, even if not always, present in intuitive examples of what is called ‘justified belief’, which is sufficiently interesting or significant, which is truth-related (as has been claimed by many epistemologists), and of which it is an interesting theoretical question whether it is necessary for knowledge or not. I will assume that justication is not necessary for knowledge. Knowledge is accounted for in terms of safety, and safety does not require justification.18 Furthermore, I will assume that justification requires evidence. A belief can only be justified if it is based on adequate evidence or an adequate reason. (I will use the terms ‘evidence’ and ‘reason’ interchangeably.) This sets the view apart from pure reliabilism. A pure reliabilist would hold that (sufficiently) reliable formation of a belief is sufficient for justification. The view I am proposing is that reliable production need not amount to justification, since it all depends on whether the reliable production is, or involves, the basing of one’s belief on adequate evidence. On an extremely liberal understanding of ‘evidence’, even a pure reliabilist might hold that reliable belief production does involve evidential basing. But this is too liberal an understanding of ‘evidence’. In a narrower, and more interesting interpretation of ‘evidence’, a subject’s evidence consists in some of her intentional states (of a certain kind). To base one’s belief on evidence in this sense requires an Footnote 17 continued further paper in the near future. An interesting and somehow similar analysis can be found in Spicer (2004). 18 If one takes sensitivity as one’s criterion for knowledge (along the lines of Nozick or Dretske or, in a contextualist vein, David Lewis), knowledge will also not require justification.

123

30

F. Hofmann

intentional state on which one bases this belief, for example, some further belief, some perceptual experience, or some intuition. And this is certainly a significant phenomenon. Quite often we base our beliefs on beliefs, perceptual experiences, intuitions, and the like, and intuitively this seems to be an epistemic achievement (at least, in cases where the evidence meets some further conditions). In contrast, merely being produced by some reliable mechanism or process need not involve any intentional state which serves as the evidence on which the subject bases the resulting belief. Reliable belief production can be ‘direct’ and need not involve any intentional state. Of course, it may be true that in many cases reliable belief production and evidential basing go hand in hand. But they are not the same thing. So there is a definite point of disagreement between the proposed view and the pure reliabilist view. The argument for holding that having evidence requires certain intentional states is simply this. Justification requires that there is something within the perspective of the subject which speaks in favor of the proposition to be believed. Otherwise, the subject would just ‘blindly’ come to believe this proposition. If nothing from her point of view speaks in favor of the proposition, she cannot base her belief on anything that speaks in favor of it from her point of view. So she will not be justified in believing it. Her belief may be objectively likely to be true. But without a suitable intentional state on which the subject bases her belief, no justification can arise. The case of Norman the clairvoyant, as described initially by Laurence BonJour, is an illustration of this point.19 Norman reliably forms a belief, by some mechanism of clairvoyance that he has been equipped with from a certain point in time on. But he has no evidence for what he believes. The mechanism works without any intentional states that can count as evidence. (Norman has no ‘clairvoayance intuitions’. If he had, things would be different.) Intuitively, Norman is not justified in his belief. The reason for this is simply that he has no evidence for it. Pure reliabilism cannot deal adequately with this case, but the view on offer can. The view respects the principle of ‘evidentialism’ (for justification), as one might call it: justification requires evidence.20 This is a departure from pure reliabilism, and an improvement, I think. Reliability remains a necessary condition of justification, since the evidence must be adequate and the inference from the evidence must be conditionally reliable. But mere reliable production is not enough for justification. This is intuitively more satisfactory than the ‘bleak’ picture of pure reliabilism.21

19

Cf. BonJour (1985), Chap. 3, p. 41.

20

Cf. Conee and Feldman (2004). Let me emphasize that, in contrast to Conee and Feldman, I would like to sketch an evidentialist account wich is externalist. For a convincing critique of Feldman’s internalism, see Greco (2005). 21 Please note that this is not so say that epistemic externalism is false. The view on offer is still an externalist view, since it takes reliability to be a necessary condition, and reliability need not supervene on intrinsic matters and need not be directly internally accessible. This is reflected in the adequacy condition on evidence necessary for justification: justification requires adequate evidence, and adequacy requires at least reliability. Of course, knowledge is adequate evidence, but intentional states which are not knowledge can also be adequate evidence—in contrast to Williamson’s view (cf. Williamson 2000).

123

Introspective Self-Knowledge of Experience and Evidence

31

Having said this much about justification and evidence in general, let us take a look at the second step of introspection. The first step of introspection is the act of introspection yielding a demonstrative thought which is knowledge. The second step is the transition, or inference, from this thought to a second, nondemonstrative thought.22 In our example of Selma, the second thought is Selma’s thought that she is having a pain in her left foot. (This is an example of what is usually called ‘introspective self-knowledge’. The demonstrative self-knowledge that was the topic of the last section is usually ignored.) In the following, I will try to show that this second, non-demonstrative thought is justified, since it is based on the first, demonstrative self-ascription which serves as its (adequate) evidence. (It is adequate evidence, since it is knowledge.) The overall picture therefore is this: With introspection, we start with demonstrative self-knowledge, and based on this knowledge, we arrive at further, non-demonstrative selfascriptions of mental states which are justified. (I will leave open the question whether these self-ascriptions are knowledge, but I certainly do not deny that they are.) These non-demonstrative thoughts classify the experience as an F-experience, or an experience of an F, for example, as a pain-experience, or an experience of red, and so on. The demonstrative concept denoted by ‘this’ is replaced by a non-demonstrative, standing concept, such as the concept of a pain. These non-demonstrative concepts are the so-called ‘phenomenal concepts’. (Whether the indexical concept denoted by ‘this’ is also a phenomenal concept can be left open.) Their primary use is exactly in such cases of introspecting on what one experiences simultaneously.23 In cases where the subject bases her standing, non-demonstrative belief on her act of introspection, she believes that she is having a pain in her left foot because she notices it—because she thinks that this is what she is experiencing right now, where the referent of the ‘this’ is fixed by attentional demonstration. I take it that this ‘because’ indicates the basing relation, and that it requires a causal relation (of the right, ‘normal’ kind). But a causal relation (even if of the right, ‘normal’ kind) is probably not sufficient for the basing relation. In any case, when the subject makes the standing self-ascription because of her introspection, and the introspection is a state of knowledge, then it seems intuitively clear that the resulting self-ascription is

22 I will use the notion ‘inference’ in a liberal, wide sense. Inferences are just transitions between intentional states that are somehow content-driven. There need not be any significant ‘inference rule’ (like modus ponens) governing these transitions. 23 There is a complication here. It may be necessary, in the end, to relativize justification to positions in a proposition. One may be justified in believing the proposition that a is F, as opposed to the proposition that a is G, without being justified in believing it as opposed to the proposition that b is F. The evidence is evidence with respect to a propositional position. With introspection, the position for which the experience (as part of the first, demonstrative introspective thought) provides evidence is not the subject position, where the person referred to by the first-person concept is concerned, but the predicate position. The account on offer does not propose that there is experiential evidence for the application of the firstperson concept. There is no need for such an experiential evidence, since one starts with the first-person thought that oneself is experiencing this, and this thought is simply knowledge (which does not require evidence or justification) and can, therefore, justify the second, non-demonstrative belief with respect to the subject position.

123

32

F. Hofmann

justified. Indeed, the resulting self-ascription is a paradigm of a justified introspective belief—a belief arrived at and justified by introspection.24 Not just every belief about one’s present experience can count as an introspective belief. One could arrive at such beliefs in a number of different ways. (Here I mean the non-demonstrative beliefs.) But only one of them is the one which makes the belief an introspective belief. Justification by introspection (the act of paying attention to what one experiences and thus forming the demonstrative thought that this is what one is experiencing right now, i.e., noticing one’s experience) thus provides the crucial criterion for distinguishing introspective beliefs from other, non-introspective beliefs having the same contents as introspective beliefs. The epistemic status of introspective beliefs depends on their history. They are the ones arrived at by introspection and, thus, they are based on (demonstrative) selfknowledge. This already amounts to an explication of peculiarity for the second step. Since introspection is always from the first-person perspective, and standing introspective beliefs are based on introspection, they inherit the essential connection to the firstperson perspective. Only the subject herself can arrive at the standing introspective belief that she is having a pain in her foot. Therefore, the account can preserve the desideratum of peculiarity. Privilegedness is also preserved. By basing her belief on introspection, i.e., a state of knowledge, the introspective self-ascription is justified. And it seems fair to say that it is justified to a high degree. Of course, there may be cases with lower and higher degrees of justification. But in general, introspective beliefs exhibit a rather high degree of justification. One would have to have quite strong counter-reasons (defeaters) in order to overthrow this justification. This can be explained by the high degree of reliability which is (at least typically or normally) guaranteed by the introspective process. First, the introspective demonstrative thought is very safe. Therefore, it is (very) adequate evidence. Second, the transition between this thought and the standing self-ascription of the experience is of a high conditional reliability. Given that the introspective demonstrative thought (in which the experience itself is embedded) is true, it is highly probable that the resulting standing self-ascription is true. The transition, or inference, is epistemically very good since it is highly conditionally reliable. The contents of the two states are related in the right kind of way. The first is really very good evidence for the second. Therefore, the introspective belief turns out to be based on (very) adequate evidence and is thus highly justified, and it seems appropriate to speak of a ‘privileged’ status. (Still, error is possible here.) In this way, the view on offer can preserve the two desiderata of peculiarity and privilegedness. The two stages of the introspective process exhibit features and are 24

A further point should be noted. We are not to assume that experience is ontologically transparent to the subject. The subject comes to have the justified belief that she is having a pain in her foot. But introspectively she is not able to discern the nature of the intentional object of her pain experience. She can ascribe the pain experience, by applying her pain concept, and the self-ascription is normally justified (on the basis of the demonstrative thought). But she cannot, by mere introspection, tell or recognize what the nature of the intentional object of her experience is (whether it is a particular kind of tissue damage, for example).—Thanks to an anonymous referee for bringing up this issue.

123

Introspective Self-Knowledge of Experience and Evidence

33

related in such a way that both the first and the second introspective self-ascription turn out to be essentially first-personal and of a high epistemic quality.25 The view respects the principle of evidentialism (for justification): justification requires evidence. So it differs from pure reliabilism, but it preserves the idea of reliability. This completes the presentation of the view.

4 Conclusion Let us take a look at the larger landscape. All of what has been said is compatible with (first-order) representationalism (intentionalism) and the transparency of phenomenal consciousness. A representationalist is not committed to pure reliabilism (Tye), nor to the model of displaced perception (Dretske). There is an alternative epistemological position about introspection which is available to the representationalist. And this alternative position respects the intuition of transparency. We need not be aware of any intrinsic features of experiences in order to arrive at introspective beliefs about our experiences, and we are not, as Moore, Harman, Tye, Dretske, and others have emphasized. Introspection is directed at what one experiences, at the (intentional) objects of experience. There is no need for any inner sense. The alternative epistemological position has certain advantages, as I wanted to show. One of them is that evidentialism about justification is vindicated. (We could speak of ‘evidential reliabilism’.) This is an important improvement, I think, since it allows us to bring the topic of evidence, reasons, and justification properly into focus. Justification is related to reasons and counterreasons (rebutting and undermining defeaters). The pure reliabilist cannot explain our intuition that justification can be overridden or undermined by suitable defeaters. (I take it that reasons and defeaters are intentional states. Defeaters are, in Bergmann’s (2006) sense, ‘mental state defeaters’ and not ‘propositional defeaters’.) If a process is reliable, the resulting belief is justified, and no counter-evidence can change that.26 But if, in contrast, justification requires intentional states serving as evidence, then there is (at least) a chance of explaining how various different reasons and counterreasons can work together and strengthen or eliminate justification. The overall intentional state of the subject is relevant to the issue of justification. This needs to be worked out in detail, to be sure. But at least there is an idea of how to deal with the phenomenon of defeaters. And, hopefully, this adds up to the attractions of (firstorder) representationalism. Acknowledgments I would like to thank Wolfgang Barz, Manfred Frank, Thomas Grundmann, Joachim Horvath, Christian Loew, Susanne Mantel and Peter Schulte for valuable discussions.

25

Of course, justification does not exclude knowledge. The second step may lead to standing, nondemonstrative self-knowledge in many cases. I have only left out to investigate whether and when this is so, but I do not want to deny that this may occur often. 26 To add a ‘no-defeater condition’, as Goldman (1979) did, is ad hoc and calls out for explanation. It is hard to see how the pure reliabilist could come up with a satisfactory explanation.

123

34

F. Hofmann

References Aydede, M. (2003). Is introspection inferential? In B. Gertler (Ed.), Privileged access (pp. 55–64). Aldershot: Ashgate. Bergmann, M. (2006). Justification without awareness. Oxford: Clarendon Press. Boghossian, P. (1989). Content and self-knowledge. Philosophical Topics, 17, 5–26. (Reprinted in Privileged Access, pp. 65–81, by B. Gertler (Ed.). Aldershot: Ashgate, page numbers cited according to this reprint). BonJour, L. (1985). The structure of empirical knowledge. Cambridge: Harvard University Press. BonJour, L. (1999). The dialectic of foundationalism and coherentism. In J. Greco & E. Sosa (Eds.), The Blackwell guide to epistemology (pp. 117–142). Oxford: Blackwell. Burge, T. (1988). Individualism and self-knowledge. Journal of Philosophy, 85, 649–663. Byrne, A. (2005). Introspection. Philosophical Topics, 33, 79–104. Conee, E., & Feldman, R. (2004). Evidentialism: Essays in epistemology. Oxford: Oxford University Press. Dretske, F. (1995). Naturalizing the mind. Cambridge: MIT Press. Gennaro, R. (2008). Representationalism, peripheral awareness, and the transparency of experience. Philosophical Studies, 139, 39–56. Gertler, B. (2001). Introspecting phenomenal states. Philosophy and Phenomenological Research, 63, 305–328. Goldman, A. (1979). What is justified belief? In G. S. Pappas (Ed.), Justification and knowledge. Dordrecht: Reidel. Greco, J. (2005). Justification is not internal. In M. Steup & E. Sosa (Eds.), Contemporary debates in epistemology (pp. 257–269). Oxford: Blackwell. Horgan, T., & Tienson, J. (2002). The intentionality of phenomenology and the phenomenology of intentionality. In D. Chalmers (Ed.), Philosophy of mind. Oxford: Oxford University Press. Kemmerling, A. (1999). How self-knowledge can’t be naturalized (some remarks on a proposal by Dretske). Philosophical Studies, 95, 311–328. Kriegel, U. (2002). Phenomenal content. Erkenntnis, 57, 175–198. Kriegel, U. (2008). Self-representationalism and phenomenology. Philosophical Studies. (forthcoming). Kriegel, U., & Williford, K. (Eds.) (2006). Self-representational approaches to consciousness. Cambridge: MIT Press. Pitt, D. (2004). The phenomenology of cognition or what is it like to think that p? Philosophy and Phenomenological Research, 69, 1–36. ¨ ber das Fundament der Erkenntnis. Erkenntnis, 4, 79–99. Schlick, M. (1934). U Shoemaker, S. (1994). Self-knowledge and inner sense. Philosophy and Phenomenological Research, 54, 249–314. Sosa, E. (1999). How must knowledge be modally related to what is known? Philosophical Perspectives, 26, 373–382. Sosa, E. (2002). Tracking, competence, and knowledge. In P. Moser (Ed.), The Oxford handbook of epistemology. Oxford: Oxford University Press. Sosa, E. (2007). A virtue epistemology. Apt belief and reflective knowledge (Vol. 1). Oxford: Clarendon. Spicer, F. (2004). On the identity of concepts, and the compatibility of externalism and privileged access. American Philosophical Quarterly, 41, 155–168. Tye, M. (1995). Ten problems of consciousness. Cambridge: MIT Press. Tye, M. (2000). Consciousness, color, and content. Cambridge: MIT Press. Williamson, T. (2000). Knowledge and its limits. Oxford: Oxford University Press.

123