Transparent introspection of wishes

Abstract

The aim of this paper is to lay the groundwork for extending the idea of transparent introspection to wishes. First, I elucidate the notion of transparent introspection and highlight its advantages over rival accounts of self-knowledge (Sect. 1). Then I pose several problems that seem to obstruct the extension of transparent introspection to wishes (Sect. 2). In order to overcome these problems, I call into question the standard propositional attitude analysis of non-doxastic attitudes (Sect. 3). My considerations lead to a non-orthodox account of attitudes in general and wishes in particular in light of which the problems presented in Sect. 2 disappear (Sect. 4).

The idea of transparent introspection looms large in the current debate about self-knowledge.¹ According to this idea, in determining one’s own current mental states, one does not look inward, but focuses instead on certain “worldly” items, that is, objects, properties or combinations thereof that are, or seem to be, out there in the world. Since transparent introspection seems to be a promising alternative to traditional accounts of self-knowledge, it might be desirable to apply it to the whole range of mental states.² However, a satisfactory application of transparent introspection to attitudes other than beliefs—or non-doxastic attitudes, as I will call them—is still a long time in coming. As far as I can see, this is due to the widely held view that desires, intentions, etc. are attitudes toward propositions. As will be argued, accepting this view makes it difficult to develop a satisfactory account of transparent introspection that covers more than doxastic attitudes. Therefore, I will propose an alternative account of non-doxastic attitudes that better accords with the idea of transparent introspection than the standard propositional attitude analysis, as I will call it.

Since I cannot deal with all non-doxastic attitudes, however, I will confine myself to wishes ³ and try to show that the intentional contents of beliefs and wishes are fundamentally different. My argument is quite simple. The first premise is that the intentional content of an attitude coincides with the semantic content of a sentence which, as uttered by the agent, best expresses the attitude in question. The second premise is that beliefs are best expressed by sentences in the indicative mood, whereas wishes are best expressed by sentences in the optative mood. The third premise is that sentences in different grammatical moods express different kinds of semantic contents (and these differences cannot be captured by the performative analysis of non-indicatives). I conclude that wishes do not have the same kind of intentional content as beliefs: they are not relations to propositions, but relations to postulations—that is, abstract entities that are the semantic contents of sentences in an optative mood. This non-orthodox analysis enables us to apply the idea of transparent introspection to wishes in a satisfying way—and can perhaps serve as a blueprint for applying the idea of transparent introspection to non-doxastic attitudes in general.

Before I motivate and develop my alternative account of non-doxastic attitudes, however, allow me to elucidate the notion of “transparent introspection” in order to forestall possible misunderstandings from the start. In particular, I would like to explain what I understand by transparent introspection and what I take to be its main advantages over competing accounts of self-knowledge. This will be addressed in Sect. 1. The main discussion, then, commences in Sect. 2, where I describe certain problems that are raised in connection with the attempt to extend the idea of transparent introspection to wishes. In order to solve these problems, I will challenge the standard propositional attitude analysis of non-doxastic attitudes and argue for the semantical significance of grammatical mood (Sect. 3). My considerations lead to a non-orthodox analysis of wishes, in light of which the problems presented in Sect. 2 disappear (Sect. 4).

1 The notion of transparent introspection

First of all, let me emphasize that the phrase “transparent introspection” does not refer to the traditional, sometimes called “Cartesian,” claim that if a person is in a certain mental state at t, then this person necessarily is aware, believes, or even knows at t that she currently is in that mental state. This claim has nothing to do with the idea of transparent introspection (henceforth “TI”) as it is understood here, where it may be formulated as the claim that I typically come to know that I am currently in a certain mental state, not by peering inward at the mental state itself, but by focusing instead on certain aspects of the external world, namely, those aspects that are determined by the intentional content of the mental state in question.

TI might be related to, but it is far more general than, the claim that when I try to attend to the phenomenal character of my perceptual experiences it seems to me as though I do so through attending to the features of mind-independent objects.⁴ While the latter claim is restricted to the phenomenal character of one’s perceptual experiences, the former claim, TI, is not so restricted. Something similar holds for the claim, also trading under the label of “transparency,” that, typically, I answer the question as to whether I believe that p by answering the question whether p.⁵ While “transparency” in the latter sense is restricted to beliefs, TI is not so restricted. It could be said, then, that TI is the generic variant of those more restricted claims. Thus, TI corresponds to what Brie Gertler calls “the Transparency model” of self-knowledge.⁶

TI seems attractive for several reasons. For example, Alex Byrne has argued that TI is economical in the sense that it “purports to explain self-knowledge solely in terms of epistemic capacities and abilities that are needed for knowledge of other subject matters.”⁷ This is particularly clear in the case of belief. According to TI, the only thing one has to do in order to realize whether one believes that, say, Tegucigalpa is the capital of Honduras is to wonder whether Tegucigalpa is the capital of Honduras. Wondering whether Tegucigalpa is the capital of Honduras, in turn, does not require any special epistemic capacities and abilities beyond what is needed for knowing that Tegucigalpa is the capital of Honduras. In this respect, TI shows a clear advantage over inner-sense theories, which posit an inner self-scanning or monitoring system that somehow tracks the mental states one is introspectively aware of in addition to our epistemic capacities and abilities that are needed for ordinary empirical knowledge.⁸

Second, Byrne claims that TI explains what he calls peculiar access, i.e., why “one knows about one’s mental life in a way that one cannot know about another’s mental life.”⁹ The reason for this is obvious, since, for example, I cannot come to know about my friend’s opinion about the question whether p just by wondering whether p is the case. In this respect, TI has a clear advantage over Ryle’s (1949) account of self-knowledge, according to which one knows about one’s own mental states in much the same way as one knows about other’s mental states.¹⁰

Third, Byrne suggests that TI has the potential to explain what he calls privileged access, that is, why “our beliefs about our own mental states, arrived at by typical means, are more likely to amount to knowledge than the corresponding beliefs about other’s mental states and the corresponding beliefs about one’s environment.”¹¹ In the case of belief, Byrne’s claim is obviously true, since proceeding from an affirmative answer to the question whether p to a second-order belief to the effect that one believes that p could not possibly lead to wrong results. Byrne nicely elucidates this fact by saying that someone who comes to know about her or his beliefs in the ordinary (i.e., first-personal) way can be said to follow an epistemic rule that reads, “If p, believe that you believe that p.” This rule, as Byrne says, is self-verifying: “If it is followed, the resulting second-order belief is true.”¹² When it comes to mental states other than belief, the case is not so clear. If TI could, however, be extended such that the epistemic rules that one follows in order to know about one’s non-belief-like mental states turn out to be self-verifying as well, then the prospect of providing a satisfying explanation of privileged access in general is not bad. It would seem that, in this respect, TI has advantages over not only Rylean accounts of self-knowledge, but also inner-sense theories, since the former appear to deny the phenomenon of privileged access altogether,¹³ while the latter seem to face considerable difficulties in explaining why the deliverances of inner sense are more likely to amount to knowledge than the deliverances of the outer senses. Something similar holds for expressivism in the spirit of Wittgenstein (1953) and Malcolm (1958)—for this kind of expressivism characteristically claims that assertions to the effect that one is currently in a certain mental state, produced in ordinary circumstances, do not qualify as manifestations of knowledge at all.¹⁴

There is a fourth advantage of TI, which is not mentioned by Byrne. Unlike some traditional views on self-knowledge, TI does not force us to posit any troublesome epistemic relations to likewise troublesome mental items. The Unmediated Observation model of self-knowledge,¹⁵ for example, suggests that at least some of our beliefs about our own mental states are grounded in immediate apprehensions of irreducible subjective factors—be it sense data, qualia, experiential qualities, phenomenal properties, or whatsoever. In contrast, TI implies that we do not need to enter into such relations to such mental items in order to achieve self-knowledge about our own mental states. This makes TI especially appealing for philosophers who are interested in reducing the mental to the physical—for it is tempting to say that the fact that we do not need to immediately apprehend, say, experiential qualities is best explained by the fact that, roughly speaking, there are none. According to certain naturalistically inspired philosophers, what we naively take for the irreducible subjective qualities of our experiences turn out to be nothing more than the properties which external objects are represented as having—properties that are out there in the world and thus equally accessible to everyone else.¹⁶

Though I am sympathetic towards these points,¹⁷ I doubt whether they capture all there is to the appeal of TI. In my opinion, a great deal of TI’s philosophical appeal is due to the fact that TI breathes the spirit of direct realism, so to speak.¹⁸ As traditional philosophical wisdom has it, our mental states form something like an opaque stratum that separates us from the external world. When we perceive, think about, or have feelings toward items in the external world—be it objects, states of affairs, or whatsoever—, we are not directly confronted with those items, but indirectly, by courtesy of some proxy entities—sense data, ideas, concepts, modes of presentation, etc.—which are built into our mental states and lead our awareness, so to speak, in the right channels. Consequently, one’s own mental states are often regarded as enjoying a certain kind of epistemic priority over items in the external world. Suppose that we are aware of an external item o through a mental state Ψ. Proponents of the more traditional view maintain that, while we could not know anything about o if we were not aware of at least those elements of Ψ in virtue of which o is presented to us,¹⁹ we would certainly be able to know something about Ψ even if we were not aware of o. Thus, in the traditional view, the idea of a subject who has effectively “neutralized”—or, as Husserl puts it, “bracketed”—any belief about the external world and thereby precludes any awareness of external objects, but nonetheless succeeds in achieving knowledge about the sense data and modes of presentation which are present to her mind, is perfectly coherent.²⁰

TI turns this conception upside down. To be sure, it does not deny that mental states incorporate proxy entities that direct our awareness toward external objects. However, it denies that mental states enjoy epistemic priority over external objects—for, according to TI, you cannot know anything about the mental state you are in as long as you are not aware of the external object it is about. When you turn your attention away from the external world in order to focus on the proxy entities in virtue of which your mental state has its world-directedness, the prospects of achieving knowledge about your mental state vanish.²¹ Thus, according to TI, there is at least epistemic equality between external objects and mental states. It might be true that we could not know anything about an external object o (of which we are aware through mental state Ψ) unless we were aware of at least those elements of Ψ in virtue of which o is presented to us²²; it seems, however, that we likewise would not be able to know anything about Ψ (or those elements of Ψ in virtue of which o is presented to us) unless we were aware of o. So, according to TI, there is no reason to compare our mental states with an opaque stratum that separates us from the external world. Mental states are not opaque, they are transparent: the external world shines through.

To sum up, TI, far from being universally accepted,²³ is a promising working hypothesis in the field of self-knowledge. Not only does it explain why it seems so difficult to find intrinsic features of perceptual experiences in introspection, it also explains why we answer the question whether we believe that p by answering the question whether p. Furthermore, TI has several advantages over competing accounts of self-knowledge. First, in contrast to inner sense theories, TI is economical in that it does not posit any special epistemic capacities and abilities beyond what is needed for ordinary knowledge. Second, in contrast to Rylean theories, TI takes into account the fact that our epistemic access to our own mental states is peculiar in that the way one knows about one’s own mental states is different from the way one knows about other’s mental states. Third, in contrast to inner sense theories as well as Rylean and Wittgensteinian accounts, TI has the potential to explain why our epistemic access to our own mental states is privileged in that it gives rise to beliefs that are less prone to error than beliefs that arise from other sources. Fourth, in contrast to the Unmediated Observation model, TI is acceptable even for hardened naturalists because it does not commit one to immediate apprehensions of allegedly irreducible subjective items such as sense data or phenomenal qualities. Last but not least, TI is appealing on general ideological grounds, since it breaks with the traditional conception of epistemic priority of the intra-mental over the extra-mental, pleading for epistemic equality instead.

2 The trouble with wishes

In this section, I would like to outline several problems that arise in connection with the attempt to extend the idea of transparent introspection to wishes and that motivate my alternative account of non-doxastic attitudes in general and of wishes in particular. I will confront these problems now, beginning with the least pressing.

Suppose that I believe that it will rain tomorrow. According to TI, I come to know that I believe that it will rain tomorrow, not by peering inward, but by focusing my attention on those aspects of the external world that are determined by the intentional content of my belief. Since the intentional content of my belief is the proposition that it will rain tomorrow and this proposition, in turn, determines a particular state of affairs, namely tomorrow’s rain, I have to focus my attention on tomorrow’s rain.²⁴ Thus, it seems tempting to regard the procedure whereby I come to know that I believe that it will rain tomorrow as though it starts with asking myself, “Will it rain tomorrow?” Since I believe that it will rain tomorrow, I will answer that question in the affirmative, which, in turn, justifies me in proceeding to the “second-order” belief that I believe that it will rain tomorrow. This gives a rough outline of the way that, according to TI, one achieves knowledge about one’s own beliefs: no inward glance, just a focus on a worldly item.

Now, suppose that, in addition to the fact that I believe that it will rain tomorrow, I wish that it would rain tomorrow, since it has not rained for months. How, then, can I determine whether I have that wish? TI implies that I come to know that I wish that it would rain tomorrow, not by peering inward, but by focusing my attention on those aspects of the external world that are determined by the intentional content of my wish. However, according to the standard propositional attitude analysis of non-doxastic attitudes, the intentional content of my wish is exactly the same as the intentional content of my belief, namely the proposition that it will rain tomorrow. Thus, it turns out that in order to know whether I wish that it would rain tomorrow I have to attend to the very same state of affairs to which I must attend in the case of the corresponding belief, that is, tomorrow’s rain. However, this time, it does not seem tempting to regard the procedure whereby I come to know that I wish that it would rain tomorrow as though it starts with asking myself “Will it rain tomorrow?”—for that question would lead me up the garden path. Thus it seems, in the case at hand, that I do not focus my attention on tomorrow’s rain or, should I do so, I would not be able to “introspectively” distinguish between my wish and the corresponding belief.

This conclusion, however, seems to be overhasty. One might say that the mistake here lies in an oversimplifying view of the starting point of TI. It is one thing to focus one’s attention on a certain state of affairs, say tomorrow’s rain; but it is another matter to ask oneself a certain question about it, say “Will it rain tomorrow?” Thus, we are free to suppose that, in the case of my wish that it would rain tomorrow, I do focus my attention on tomorrow’s rain, but ask myself a question that is different from “Will it rain tomorrow?” What is that question?

As far as I know, there are no suggestions yet regarding the special case of wishes in the literature. However, there have been several suggestions regarding the broader category of desires.²⁵ Ignoring differences in detail, these suggestions amount to the idea that one answers the question whether one has a certain desire by answering the question whether the target of one’s desire is desirable—where “desirable,” for obvious reasons, is not meant to refer to the property of being desired by the subject in question, but has to be understood as an umbrella term for positive evaluative properties like being beneficial, being delectable, being diverting, being excellent, being good, being pleasant, being satisfying, being valuable, being worthwhile, etc. Brie Gertler gets to the heart of the matter, when she illustrates this idea by the following example: “When asked, ‘do you want some ice cream?’, I do not look inward to consult my desires, but instead I think about ice cream, to determine its desirability.”²⁶

It might be tempting, then, to say that, in order to determine whether I wish that it would rain tomorrow, I focus my attention on tomorrow’s rain and ask myself whether this state of affairs has a certain positive evaluative property, say, being good. There is a minor complication here, however. While “ice cream” is a mass term that refers to a particular concrete stuff, namely ice cream, “tomorrow’s rain” does not refer to a particular stuff, let alone a particular concrete stuff. Instead, “tomorrow’s rain” refers to an abstract entity, namely the state of affairs that is determined by the proposition that it will rain tomorrow. Now, it should be clear that states of affairs cannot have positive evaluative properties—at least they cannot have them in the sense that concrete stuff, individual things or actions can be good, pleasant, valuable, etc. Of course, we might say things like “It would be a good thing if it were to rain tomorrow.” However, this must not be understood to mean that a certain abstract entity, tomorrow’s rain, exemplifies a certain property, being good. Rather, it says that tomorrow’s rain would be a good thing if it were to obtain. We do not categorically ascribe a positive evaluative property to a certain state of affairs here, but we do so conditionally: we ascribe a positive evaluative property to a certain state of affairs on the condition that it obtains. So, in order to ascertain whether you wish that it would rain tomorrow, you cannot simply ask yourself whether tomorrow’s rain is good—since this would be nonsensical. Instead, it seems that you have to ask yourself whether tomorrow’s rain, if it were to obtain, would be good.

However, even this suggestion is beset with problems. One difficulty is that the proposal at hand seems phenomenologically inadequate. According to this proposal, I determine whether I wish that p by going through at least the following stages: first, I focus my attention on the state of affairs that is determined by the proposition that p; second, I assume (or perhaps imagine) that the respective state of affairs obtains; third, I wonder whether the situation as it is assumed (or imagined) by me is better than the actual situation (where the respective state of affairs does not obtain). However, this not only flies in the face of the widely shared view that, normally, one knows one’s current attitudes directly, that is, without reasoning,²⁷ it also violates the more modest intuition that, normally, one knows one’s current non-doxastic attitudes as directly as one knows one’s current beliefs. While the proposal at hand has three stages, the transparent account of beliefs has just two: first, I focus my attention on the state of affairs that is determined by the proposition that p; second, I wonder whether the respective state of affairs obtains.

However, even this problem is not decisive. It might be said, for example, that phenomenological intuitions about the directness of introspection are misleading or that my tripartite reconstruction of the proposal is uncharitable. Thus, I would like to bring up a third problem that I take to be more decisive, and that is that we can know our wishes through introspection even when they are not in line with our evaluative judgments concerning relevant states of affairs. Consider, first, the following cases that might be brought into play in order to undermine the extension of TI to desires:

I know I really should get out of bed when my alarm rings in the morning, but I just don’t feel like it—I know that I just don’t want to. I often judge that the healthy option on the menu is good for me—that it is valuable in the respect that it is healthy—yet know that I don’t want it. And when I procrastinate, it isn’t just that I know that I want to watch television without judging that it has any value, I also know that I don’t want to work—I don’t want to take the valuable option Ashwell (2013, p. 251).²⁸

Similar cases can be imagined in connection with wishes. Consider an official of the National Rifle Association (NRA) who, though he might recognize that more restrictive gun laws would be of great benefit to his country, does not wish for more restrictive gun laws, since he loves to drive through the town with a loaded gun under his car seat. He might even say, “Of course it would be a good thing if Congress passed more restrictive gun laws. But, well, I don’t wish that the Congress would pass more restrictive gun laws.” Or imagine a dedicated opponent to the death penalty, whose child has just been murdered. He might realize that he wishes that his child’s murderer would be hanged. At the same time, however, he might still feel that hanging the murderer would be a bad thing. He might even say, “Of course, I wish that the murderer would be hanged. But, well, it would be a bad thing, if he were to be hanged.”

It seems that the current proposal—that one answers the question whether one has the wish that p by answering the question whether it would be a good thing if p were to be the case—leaves no room for these (and similar) cases. The proposal at hand predicts, contrary to fact, that the NRA official will believe that he wishes for more restrictive gun laws—for he judges more restrictive gun laws to be good. It likewise predicts, contrary to fact, that the dedicated opponent to the death penalty will not believe that he wishes that the murderer would be hanged—for he judges the death penalty to be bad.

These arguments lead to a fourth problem that stands in the way of extending TI to wishes: it might be said that my examples show the disanalogy between beliefs and wishes to be so fundamental that it renders transparent introspection of wishes altogether impossible. One might argue, “An attitude, say ψing that p, is accessible through TI if and only if there is a question Q about external matters such that a rational subject, who is equipped with the relevant concepts, is disposed to respond affirmatively to Q if and only if she is disposed to respond affirmatively to the question ‘Do you ψ that p?’ Beliefs satisfy this condition for obvious reasons: a rational and conceptually well-equipped subject who is disposed to respond affirmatively to the question whether p will also be disposed to respond affirmatively to the question whether she believes that p—and vice versa. If a subject steps out of line and utters sentences like ‘Of course, it will rain tomorrow. But, well, I don’t believe that it will rain tomorrow’ or ‘Of course, I believe that it will rain tomorrow. But, well, it won’t rain tomorrow,’ we will deem her irrational or conceptually confused. Now, wishes—and probably all non-doxastic attitudes—do not behave like beliefs in this regard. As we have seen, it makes perfectly good sense to utter sentences like ‘Of course it would be a good thing if Congress passed more restrictive gun laws. But, well, I don’t wish that Congress would pass more restrictive gun laws’ or ‘Of course I wish that the murderer would be hanged. But, well, it would be a bad thing, if he were to be hanged.’ We won’t deem the utterer of these sentences irrational or conceptually confused merely due to the fact that he utters these sentences—for we know that, even in rational and conceptually well-equipped persons, wishes and evaluative judgments do sometimes come apart. So it seems that there is just no question about external matters to which a rational subject, equipped with the relevant concepts, is disposed to respond affirmatively if and only if she is disposed to respond affirmatively to the question ‘Do you wish that p?’ Hence, wishes are not accessible through transparent introspection.”²⁹

These objections are problematic, to be sure. However, they are not so severe that they force us to abandon the idea that wishes are accessible through transparent introspection. In my view, the roots of all these problems and difficulties lie in a wrong metaphysical conception of non-doxastic attitudes in general and wishes in particular, namely that all non-doxastic attitudes are attitudes towards propositions. As I have already indicated in the introduction, this standard propositional attitude analysis makes it difficult to see how TI could ever be applied to wishes. Thus, in what follows, I will develop an alternative analysis of wishes, according to which wishes are not relations to propositions, but to what I will call postulations—near relatives of propositions that serve as the semantic contents of sentences in the optative mood. This non-orthodox analysis will resolve all the problems raised in this section: It not only explains why one is able to “introspectively” distinguish between beliefs and wishes that allegedly have the same intentional content, but it is also in accord with the intuition that, normally, one knows one’s current non-doxastic attitudes as directly as one knows one’s current beliefs. Moreover, my analysis will explain why we are able to know our wishes even when they are not in line with our evaluative judgments. Finally, my analysis will show that the fact that there might be no question about external matters which stands in the same relation to “Do you wish that p?” as “Is p the case?” stands to “Do you believe that p?” does not preclude the idea that wishes are accessible through transparent introspection.

These are admittedly sketchy remarks. However, I will state my view on these matters more precisely at the end of my paper, when my alternative analysis of wishes is developed to the full. For the time being, I hope that the presentation of the problems that arise in connection with applying TI to wishes and the promise to solve these problems at a single stroke suffice to motivate the next section: a critique of the standard propositional attitude analysis of non-doxastic attitudes and the development of an alternative account.

3 The standard propositional attitude analysis of non-doxastic attitudes

According to the standard propositional attitude analysis (henceforth “SPAA”), all attitudes, even non-doxastic ones, are relations to propositions. To substantiate their claim, proponents of SPAA typically invoke sentences that specify the attitudes of others from a third-person point of view:

• (1) Paul believes that it will rain tomorrow

• (2) Susan wishes that the world would live in peace

• (3) Tina wonders whether Syria will attack Israel.

The idea is that these sentences mirror the structure of the attitudes they describe: since each of the sentences consists of an intentional verb that connects a name with a subordinate clause, the corresponding attitudes are supposed to be analyzed as a tripartite whole consisting of subject, psychological mode, and intentional content. Moreover, and more importantly, it is assumed that the intentional contents in question are propositions, since the subordinate clauses in (1)–(3) can be transformed into truth-apt sentences: “It will rain tomorrow,” “The world lives in peace,” “Syria will attack Israel.”

A second tenet of SPAA is that psychological mode and intentional content can vary independently from each other. A change of content does not necessarily bring about a change of mode, and vice versa. Again, this seems to be supported by the behavior of (1)–(3), since, by recombining, one can easily create new attitude reports:

• (4) Paul wishes that it would rain tomorrow

• (5) Susan wonders whether the world will live in peace

• (6) Tina believes that Syria will attack Israel

• (7) Paul wonders whether it will rain tomorrow

• (8) Susan believes that the world will live in peace

• (9) Tina wishes that Syria would attack Israel.

Proponents of SPAA typically claim that the attitudes described by (1), (4), and (7) differ only with regard to their psychological mode—their intentional content is the same—to wit, the proposition that it will rain tomorrow.

In my view, the two tenets of SPAA are false: it is not the case that the contents of the attitudes described by (1)–(9) are, without exception, propositions; nor can psychological mode and intentional content vary independently from each other. A first reason why SPAA should seem suspicious is that it models intentional contents on the subordinate clauses of sentences that report attitudes. It would be far more natural to model intentional content on sentences that express attitudes.

Ryle may have been wrong in many respects. Yet he was right in that, when it comes to determining the contents of other people’s attitudes, we regard certain utterances as primary evidence—namely, those utterances which are spontaneous, frank, and unprepared.³⁰ We know, for example, that the former U.S. Ambassador in Germany, Philip Murphy, believes that Germany’s former Foreign Minister, Guido Westerwelle, was inept for office, because Murphy, as we learned from WikiLeaks, said as much in private conversation. The fact that Murphy now declares Westerwelle a “real friend” does not impress us very much, for the words Murphy speaks into the microphones of TV broadcasting companies—in contrast to the words used in private—are carefully chosen for their diplomatic impact. In Ryle’s words, speaking into microphones is a case of studied talk. If we want to know what Murphy really believes about Westerwelle, we have to consult his unstudied talk. ³¹ So, it seems reasonable to suggest that the intentional content of a person’s attitude should be modeled on the semantic content of the sentence that she herself utters (or would utter under appropriate circumstances) in order to express her attitude unambiguously—and not on the semantic content of the subordinate clause appearing in reports on the relevant attitude from the third-person perspective.

Presumably, a proponent of SPAA will react uncomprehendingly. Am I not harping on about a difference that makes no difference at all? Consider the sentence that reports Murphy’s belief that Westerwelle is inept for office:

• (10) Murphy believes that Westerwelle is inept for office.

The subordinate clause of (10), “Westerwelle is inept for office,” has the same semantic content as the sentence by which Murphy would express (and, according to WikiLeaks, in fact expressed) his belief about Westerwelle unambiguously:

• (11) Westerwelle is inept for office.

Thus, it makes no difference at all whether we model the intentional content of Murphy’s belief on (11) or the subordinate clause of (10), since in both cases we obtain the same result: the intentional content of Murphy’s belief is the proposition that Westerwelle is inept for office.

As long as we limit our focus to beliefs, this may be true. However, when it comes to wishes and states of wondering, the situation is different. Which sentence would you use in order to express unambiguously that you wonder whether it will rain tomorrow? Certainly, you would not use “It will rain tomorrow,” but rather “Will it rain tomorrow?” And how about the wish for tomorrow’s rain? Which sentence would you use in order to express that wish unambiguously? Once again, it is not “It will rain tomorrow,” but, presumably, “If only it would rain tomorrow!”³² Thus, the sentences by which we would unambiguously express our wishes and states of wondering are different from the sentences that are obtained by focusing on the subordinate clauses of the relevant attitude reports. While the latter are invariably in the indicative mood, the former are in optative and interrogative moods, respectively.³³

Presumably, the proponent of SPAA will remain unimpressed. “For the sake of the argument,” she might say, “let us grant that intentional content is best modeled on those sentences a person would utter if she were to express her attitude unambiguously. Moreover, let us suppose that sentences that best express wishes are optatives and sentences that best express states of wondering are interrogatives. Even then, it is hard to see why this should undermine SPAA—for the grammatical mood of a sentence is semantically ineffective. Grammatical mood is just an indicator of illocutionary force: the indicative suggests that the speaker makes an assertion, the imperative suggests that the speaker issues an order, the interrogative suggests that the speaker asks a question, and the optative suggests that the speaker expresses a wish. Thus, grammatical mood does not contribute to semantic content. The sentences ‘It will rain tomorrow’, ‘Rain tomorrow!’, ‘Will it rain tomorrow?’, and ‘If only it would rain tomorrow!’ all have the same semantic content. The differences in mood just mirror the different illocutionary forces that might be attached to those sentences under the appropriate circumstances.”³⁴

However, there is overwhelming linguistic evidence that grammatical mood is not only an indicator of illocutionary force. Consider the following example from Michael Pendlebury:

• (12) Rick knows that Sam will play “As time goes by” again.

• (13) Rick knows whether Sam will play “As time goes by” again.

(12) and (13) differ only in the grammatical mood of their subordinate clauses: the subordinate clause of (12) is indicative, the subordinate clause of (13) is interrogative. If grammatical mood were semantically ineffective, then the truth conditions of (12) and (13) should not differ. But they do, since there is a circumstance under which (13) is true, while (12) is false: if Rick knows that Sam will not play “As time goes by” again.³⁵

If grammatical mood makes a semantic difference, SPAA cannot be right. Compare again the belief that it will rain tomorrow and the wish that it would rain tomorrow. As argued before, the content of an attitude should be modeled on the sentence that the subject in question would utter in order to express herself unambiguously. If a subject believes that it will rain tomorrow, she will express herself unambiguously by uttering

• (14) It will rain tomorrow,

whereas if a subject wishes that it would rain tomorrow, she will express herself unambiguously by uttering

• (15) If only it would rain tomorrow!

Since, as we have just seen, grammatical mood is semantically effective, (14) and (15) differ in semantic content. If (14) and (15) differ in semantic content, the attitudes to which they give expression differ in intentional content as well. So, in contrast to what SPAA predicts, the belief that it will rain tomorrow and the wish that it would rain tomorrow do not only differ regarding their psychological mode, but also with regard to their intentional content.

However, it is not yet clear what this latter difference amounts to. One possible answer is given by the so-called performative analysis of non-indicatives (henceforth “PANI”).³⁶ According to PANI, any non-indicative sentence which one utters, performing the illocutionary act of φ-ing, can be treated like an indicative sentence in which the act of φ-ing is made explicit by a corresponding performative verb. On this account, the imperative “Close the window!”, for example, transforms into the indicative “Herewith I order you to close the window.” If we apply this strategy to (15), we obtain:

• (16) Herewith I express the wish that it would rain tomorrow.

If we, for simplicity, abstract from the subtle difference between expressing and having a wish, (16) could be read as “I wish that it would rain tomorrow.” The difference between the semantic contents of (14) and (15) then (and a fortiori the difference between the contents of the attitudes to which (14) and (15) give expression) is as follows: the semantic content of (14) (and a fortiori the content of the belief that it will rain tomorrow) is the proposition that it will rain tomorrow, while the semantic content of (15) (and a fortiori the content of the wish that it would rain tomorrow) is the proposition that the subject in question wishes that it would rain tomorrow.

Obviously, PANI does not go well with TI. TI implies that one comes to know that one wishes that it would rain tomorrow, not by peering inward, but by focusing instead on some worldly item that is determined by the intentional content of one’s wish—presumably something like tomorrow’s rain. If we accept PANI, however, this idea gets lost, since according to PANI, the intentional content of one’s wish does not determine any worldly item at all. Instead, it determines an “inner” item—something like one’s having the wish that it would rain tomorrow.

Fortunately, PANI is not sacrosanct. There are strong arguments against PANI’s approach to understanding non-indicatives.³⁷ The most salient defect of PANI is that it stubbornly denies the intuition that non-indicatives are not evaluable for truth. Suppose, for example, that the day after I uttered the sentence “If only it would rain tomorrow!” it does rain. Then it seems odd to say that what I uttered the day before turned out to be true. It seems more appropriate to say that it turned out to be fulfilled. Thus, it is reasonable to suggest that, rather than being evaluable for truth and falsity, optatives are evaluable for fulfillment or unfulfillment. Something similar holds for the other non-indicatives: imperatives are evaluable for obedience or violation, and interrogatives can be evaluated as answered or unanswered.

If we follow this line of reasoning, we arrive at the idea that the semantic contents of non-indicatives are not propositions—at least not propositions as they are usually conceived.³⁸ Surprisingly, support for this idea comes from Frege.³⁹ According to Frege, both imperative and optative sentences have what he calls a sense, but he denies those senses the title “thought” because they are “not such that the question of truth could arise for [them].”⁴⁰ Unfortunately, Frege does not introduce proper notions for these non-truth-apt senses. In the context of Frege’s inquiry, this is understandable—for he is particularly interested in the semantics of indicatives. In the context of my argument, however, it is desirable to be equipped with proper notions for non-truth-apt senses. In order to avoid confusion with the relevant illocutionary acts, I would like to call the sense of an imperative sentence a directive, the sense of an interrogative sentence an interrogation, and the sense of an optative sentence a postulation.

Directives, interrogations, and postulations, as Frege would say, “stand on the same level as” propositions.⁴¹ So just as we might take propositions for abstract entities that reside in Frege’s “third realm” and to which we can bear the kind of relation which Frege called “grasping,” we likewise might conceive of directives, interrogations, and postulations.⁴² In contrast to propositions, however, the latter do not have the property of being true or false. Rather, they have the properties of being obeyed or violated, answered or unanswered, and fulfilled or unfulfilled, respectively.⁴³

It may help to know a bit more about the nature of these non-truth-apt senses, especially the nature of postulations. So, as a starting point, let us think about the nature of the proposition that is expressed by the sentence “It will rain tomorrow” uttered, say, on the first of April 2014. Subtleties aside, this proposition is a structure consisting of the second of April 2014, the property of being rainy, and the higher-order relation of exemplification—where the higher-order relation of exemplification is to be understood as the relation that holds between an object o and a property P if and only if o is (or has) P. To put it another way, the higher-order relation of exemplification binds the constituents of the proposition—the second of April 2014 and the property of being rainy—together such that the resulting structure can be true or false.⁴⁴ Now, consider the postulation that is expressed by the sentence “If only it would rain tomorrow!”, likewise uttered on the first of April 2014. Taking our cue from Hector-Neri Castañeda, it might be argued that this postulation is a structure that consists of the second of April 2014, the property of being rainy, and the higher-order relation of demand—where the higher-order relation of demand is to be understood as the relation that holds between an object o and a property P if and only if o is not, but ought to be, P. ⁴⁵ To put it another way, the higher-order relation of demand binds the constituents of the postulation—the second of April 2014 and the property of being rainy—together such that the resulting structure can be fulfilled or unfulfilled.

According to my account, then, the semantic content of optatives differs from the semantic content of indicatives: while the latter have propositions as meanings, the former have postulations. Given the close relationship between the intentional content of an attitude and the semantic content of a sentence which best expresses the attitude in question, this means that the intentional contents of beliefs differ fundamentally from the intentional contents of wishes: while the former have propositions as contents, the latter have postulations. Not all attitudes, then, are relations to propositions. Some are relations to directives, some are relations to interrogations, and others are relations to postulations.

Let me draw out some of the consequences of this account—consequences which will be of some importance when we return to the issue of transparent introspection in the next section. When I believe that it will rain tomorrow, I stand in a certain relation to the proposition that it will rain tomorrow. However, this relation is not exhaustively described by saying that I grasp this proposition—for I can certainly grasp a proposition without having the corresponding belief, as for example when I assume for the sake of argument that it will rain tomorrow. So it seems more appropriate to say that, when I believe that it will rain tomorrow, I not only grasp, but also accept the proposition in question. Accepting a certain proposition, in turn, amounts to seeing the world in a special light: when I accept the proposition that it will rain tomorrow the world looks to me a certain way—it looks to me as if the state of affairs that is determined by the aforesaid proposition, tomorrow’s rain, obtains. One might rephrase this idea by saying that, when one believes that it will rain tomorrow, then from one’s first-person perspective it appears as though tomorrow’s rain is a fact.

Now, given that my alternative account is true, something similar can be said about wishes. First, it seems inappropriate to describe the relation in which someone who has a particular wish stands to a certain postulation merely by saying that she grasps the latter—for one can certainly grasp the meaning of the sentence “If only it would rain tomorrow!” without wishing for rain. So it seems more appropriate to say that, when I wish that it would rain tomorrow, I not only grasp, but also endorse the postulation in question. Second, endorsing a certain postulation likewise amounts to seeing the world in a special light: when I endorse the postulation that is expressed by “If only it would rain tomorrow!” the world looks to me a certain way—it looks to me as if a certain state of affairs, namely tomorrow’s rain, should obtain. However, one cannot rephrase this idea by saying that, when one wishes that it would rain tomorrow, then from one’s first-person perspective it appears as though tomorrow’s rain is a fact—for I would certainly not wish for rain if it seemed to me as though it were already settled that it will rain tomorrow. Thus, it is more appropriate to say that, when one wishes that it would rain tomorrow, then from one’s first-person perspective it appears as though tomorrow’s rain is a desideratum. The world, as it were, is represented as featuring a certain lacuna—something that has to be fixed by bringing a certain state of affairs into existence.

One might dismiss the talk of lacunae here as the sort of pompous metaphysical waffle that is often employed in the absence of hard facts. Yet this treatment of lacunae is unfair. Instead, it seems to me that lacunae can play a significant role in the non-orthodox account of meaning outlined above. For example, lacunae correspond to postulations in a way that is analogous to the way states of affairs correspond to propositions: just as the proposition that is expressed by “It will rain tomorrow” is true if and only if a certain state of affairs, namely tomorrow’s rain, obtains, the postulation that is expressed by “If only it would rain tomorrow!” is fulfilled when the relevant lacuna, namely the absence of rain, is fixed by bringing tomorrow’s rain into existence. Thus, if we have gone so far as to recognize semantic and intentional contents which are similar but not identical to propositions, then we should take the idea of lacunae seriously.

To forestall a possible misunderstanding, let me stress that lacunae are not to be identified with unsettled matters of facts—as it might be tempting to think, especially in the face of the wish that it would rain tomorrow. Many wishes are about what happened in the past, and what happened in the past is already settled. Consider Julia, who regrets having married Romeo five years ago, because, say, after marriage, he turned from a sensitive lover into a violent alcoholic. Thus, Julia wishes that she had never married Romeo. In this case, Julia endorses the postulation that is expressed by “If only I hadn’t married Romeo!” It appears to her, then, as though a certain fact should be eliminated from the world, namely the fact that she married Romeo five years ago. Thus, it is not quite precise to suggest—as I did in the two preceding paragraphs—that a lacuna is something that can only be fixed by bringing a certain state of affairs into existence. Sometimes a lacuna can be regarded as something that could be ironed out, so to speak, if it were possible to eliminate a certain fact from the world.⁴⁶ One might distinguish, then, two types of lacunae: lacunae of the first type are like gaps that cry out to be filled, whereas lacunae of the second type are like growths that need to be removed. (One might even speak here of “convex” and “concave” lacunae.)

4 The non-orthodox analysis and transparent introspection

Now, let us return to the question with which we began: Can the idea of transparent introspection be applied to wishes? At least four problems stand in the way of an affirmative answer. The first is explaining how one is able to “introspectively” distinguish between the belief that p and the wish that p, given that the only evidence at one’s disposal is the worldly item that is determined by the intentional content of the respective mental state; the second problem is providing an account that is in accord with the intuition that we usually come to know about our wishes as directly as we come to know about our beliefs; the third problem is explaining how it is possible that one knows that one wishes that p, even in situations in which one is not disposed to judge that it would be a good thing if p were to obtain (or, alternatively: how it is possible that one quite rightly refrains from believing that oneself wishes that p, even though one is disposed to judge that it would be a good thing if p were to obtain); the fourth problem is that, in contrast to the case of belief, there seems to be no question about external matters to which a rational subject, equipped with the relevant concepts, is disposed to respond affirmatively if and only if she is disposed to respond affirmatively to the question “Do you wish that p?”

All these problems disappear in light of my non-orthodox analysis of wishes. First, my analysis explains why one is able to “introspectively” distinguish between, say, the belief that it will rain tomorrow and the wish that it would rain tomorrow. The reason is that the intentional contents of beliefs and wishes differ in kind: the intentional contents of beliefs are propositions, whereas the intentional contents of wishes are postulations. So the difference between beliefs and wishes is already mirrored on the level of intentional content. Moreover, postulations determine worldly items that are different from the worldly items that are determined by propositions: while the latter are states of affairs, the former are lacunae. So the answer to the question as to whether a given attitude is a belief or a wish is decidable “from within” just by looking through the attitude at the corresponding worldly item. If the attitude offers a view of a certain obtaining state of affairs, it is a belief; if it offers, in contrast, a view of a certain existent lacuna, it is a wish.

Second, my analysis is in accord with the intuition that, normally, one knows one’s current non-doxastic attitudes as directly as one knows one’s current beliefs. According to my proposal, one determines whether one wishes that it would rain tomorrow by going through exactly the same number of stages as in the case of belief, namely two: first, one focuses one’s attention on the lacuna that is determined by the postulation that is expressed by the sentence “If only it would rain tomorrow!”; second, one wonders whether this lacuna exists—period.

Third, my analysis explains why we can know our wishes even when they are not in line with our judgments about the desirability of the relevant state of affairs. The reason is simply this: judgments about the desirability of states of affairs have no bearing on how we come to know about our own wishes. As I have just said, one determines whether one has a certain wish by focusing one’s attention on a particular lacuna and wondering whether it exists. Neither focusing one’s attention on a lacuna nor wondering whether it exists requires judgments about the desirability of states of affairs. “But that’s not true,” one might object. “As you explained earlier, a lacuna can be regarded as something that has to be fixed by bringing a state of affairs into existence (or, alternatively, that has to be ironed out by eliminating a fact). Thus, wondering whether a certain lacuna exists amounts to the same as asking oneself whether it would be a good thing if a certain state of affairs were to obtain (or, alternatively, if a certain fact were not a fact at all). So certainly, judgments of desirability are at play here!”

However, this objection is based on a misapprehension of the fundamental difference between propositions and postulations. Recall that, when I wish that it would rain tomorrow, I endorse the postulation that is expressed by the sentence “If only it would rain tomorrow!” Recall further that endorsing this postulation amounts to seeing the world in a special light: the world looks to me as if it were defective in a particular respect, and that bringing a certain state of affairs, namely tomorrow’s rain, into existence would fix it. Now, it is of great importance that the way the world looks to someone qua endorsing the postulation that is expressed by the sentence “If only it would rain tomorrow!” is not identical to the way the world might look to someone qua accepting the proposition that is expressed by the sentence “It would be a good thing if it were to rain tomorrow.”

It might be useful here to consider an analogy to perception. Compare, for example, the visual experience that you get when you look at a red object on your desk and the belief that you form when a friend, who you trust, tells you that there is a red object on your desk. (In order to facilitate imagination, suppose that, in the second case, you are miles away from your desk and on the telephone with your friend, who keeps an eye on your apartment while you are away.) In a way, both the visual experience and the belief are representations of the same worldly condition—the object’s being red—but each represents it in a different manner. While your visual experience represents the condition by means of the color and shape of which you are aware on the occasion of looking at your desk, your belief does the same by means of the proposition that you grasp on the occasion of hearing what your friend tells you about your desk. Thus, the way the condition looks to you through your visual experience is different from the way the condition looks to you through your belief, albeit both seem to aim at the same condition—the object’s being red. Seen through the visual experience, the condition appears as a manifold of color and shape, whereas seen through your belief, the very same condition appears as a certain arrangement of some particular object and the property of being red, namely an arrangement wherein the property of being red is somehow attached to the object in question.

Of course, this is not to say that the endorsement of a postulation is a kind of perception, not to mention that it comes along with an awareness of color or shape. On the contrary, like the acceptance of a proposition—and unlike a visual perception—the endorsement of a postulation is a discursive representation that does not consist in an awareness of phenomenologically salient features. Rather, the crucial point of the analogy is this: albeit both the endorsement of the postulation that is expressed by “If only it would rain tomorrow!” and the acceptance of the proposition that is expressed by “It would be a good thing if it were to rain tomorrow” seem to aim at the same worldly condition—the desirability of tomorrow’s rain—, there is a difference between the ways that condition looks “through” the postulation’s endorsement and “through” the proposition’s acceptance, respectively. It is by no means easy to specify this difference in a non-question-begging way. However, we obtain a rough approximation of the difference when we recall that there is a principled difference between propositions and postulations: while propositions can be regarded as structures that comprise the higher-order relation of exemplification, postulations can be taken for structures that comprise the higher-order relation of demand. Thus, it might be said that, seen through the acceptance of the proposition that is expressed by “It would be a good thing if it were to rain tomorrow,” the desirability of tomorrow’s rain appears as an arrangement of the following three constituents (the first of which is itself a complex arrangement that consists of a certain day—say, the second of April 2014—, the property of being rainy, and the higher-order relation of exemplification): tomorrow’s rain, the property of being desirable, and the higher-order relation of exemplification.⁴⁷ By contrast, seen through the endorsement of the postulation that is expressed by “If only it would rain tomorrow!”, the desirability of tomorrow’s rain appears as an arrangement of the following three constituents (none of which are complex): the second of April 2014, the property of being rainy, and the higher-order relation of demand.

Hence, it should be clear that wondering whether the lacuna exists that is determined by the postulation expressed by “If only it would rain tomorrow!” is not the same as asking oneself whether it would be a good thing if a certain state of affairs were to obtain: while the former amounts to ascertaining whether the world looks to oneself as it looks to someone who endorses the postulation in question—namely as comprising an arrangement wherein the property of being rainy is bound to the second of April 2014 by the higher-order relation of demand—, the latter amounts to ascertaining whether the world looks to oneself as it looks to someone who accepts the corresponding proposition—namely as comprising an arrangement wherein the property of being desirable is bound to tomorrow’s rain by the higher-order relation of exemplification. In short: keeping an eye out for a lacuna is not the same as keeping an eye out for a fact—not even for an evaluative fact to the effect that some state of affairs is desirable.

With the help of these distinctions, we can easily understand how someone who judges that x is desirable might not come to believe that she or he wishes for x, or how someone who believes that she or he wishes for x, does not judge that x is desirable. For this purpose, let us revisit the NRA official and the opponent to the death penalty. According to my diagnosis, the NRA official accepts the proposition that is expressed by “It would be a good thing if Congress passed more restrictive gun laws” but he does not endorse the postulation that is expressed by “If only Congress would pass more restrictive gun laws!” Consequently, the world does not look to him as comprising an arrangement wherein the property of passing more restrictive gun laws is bound to the Congress by the higher-order relation of demand. So when the NRA official commences stage two of transparent introspection, that is, when he wonders whether the lacuna that is determined by the postulation in question really exists, he comes to a negative result. Hence, he does not believe that he wishes for more restrictive gun laws. Something similar holds for the opponent to the death penalty: he endorses the postulation that is expressed by “If only he, the murderer of my child, were hanged!”, but he does not accept the proposition that is expressed by “It would be a good thing if he, the murderer of my child, were hanged.” So when the opponent to the death penalty wonders whether the lacuna that is determined by the postulation in question really exists, he comes to a positive result. Hence, he believes that he wishes that the murderer would be hanged.

One might object that my diagnosis depicts both the NRA official and the opponent to the death penalty as being irrational—for it seems that someone who accepts the proposition to the effect that more restrictive gun laws are desirable cannot reject the postulation expressed by “If only Congress would pass more restrictive gun laws!”, on pain of being irrational. By the same token, it seems that someone who endorses the postulation expressed by “If only he were hanged!” cannot reasonably reject the proposition to the effect that the hanging of the person in question is desirable. But neither the NRA official nor the opponent to the death penalty is irrational. Recall that their statements “Of course it would be a good thing if Congress passed more restrictive gun laws. But, well, I don’t wish that Congress would pass more restrictive gun laws,” and “Of course, I wish that he would be hanged. But, well, it would be a bad thing, if he were to be hanged,” do not seem to be especially puzzling—for we all know that even in rational and conceptually well-equipped persons, wishes and evaluative judgments do sometimes come apart.

However, contrary to what the objection suggests, my diagnosis does not depict the NRA official and the opponent to the death penalty as being irrational. As we have seen, endorsing postulations and accepting propositions are different modes of access to the world: through the latter mode, the world appears to be just the way it is, whereas through the former mode, the world appears to be defective in certain respects. It is hardly surprising then that sometimes the two modes lead to conflicting representations without thereby compromising our rationality. Consider, again, an analogy to perception: In the face of the Müller-Lyer illusion, for example, you might accept the proposition that both lines are of the same length (because perhaps you have measured the length with a ruler), but still have a visual experience to the effect that both lines are of different lengths. This has nothing to do with irrationality; it is just a consequence of the fact that we are endowed with different modes of access to the world that might give rise to conflicting representational states.

Something similar holds for the NRA official and the opponent to the death penalty: one might say that the NRA official has a “cognitive representation” of the desirability of more restrictive gun laws without the corresponding “conative representation”; the opponent to the death penalty, in contrast, does have a “conative representation” of the desirability of the hanging of the murderer, but lacks the corresponding “cognitive representation.” Thus, in my diagnosis, neither the NRA official nor the opponent to the death penalty is irrational. They just receive incompatible messages from different modes of access to the world.⁴⁸

Now, let us turn to the fourth and final problem that seems to stand in the way of extending TI to wishes: there seems to be no question about external matters to which a rational subject, equipped with the relevant concepts, is disposed to respond affirmatively if and only if she is disposed to respond affirmatively to the question “Do you wish that p?” I frankly admit that there really is no question of the relevant sort, but I am not willing to concede that the lack of such a question is disastrous for the attempt to extend TI to wishes.

First, let me explain why I think that there is no question about external matters that stands to “Do you wish that p?” in the same relation as “Is p the case?” stands to “Do you believe that p?” In order to know about your wish that p, you have to focus your attention on a certain lacuna and wonder whether it exists. But no question whatsoever will prompt you to focus your attention on a lacuna. “Would it be a good thing if p were true?”, “Shall p be the case?”, “Is it desirable that p?”, and similar questions that have been suggested in the literature all require indicative sentences as answers, for example: “Yes, it would be a good thing if p were true,” “Yes, p shall be the case,” “Yes, p is desirable,” etc. Indicative sentences, in turn, always express propositions, never postulations. So questions never turn our attention to lacunae, but always to states of affairs.

However, the fact that there are no questions that might turn our attention to lacunae is not especially worrying for proponents of TI. All that counts is that one is able to turn one’s attention to lacunae—be it with or be it without the help of questions. Hence, in contrast to Finkelstein (2003), Goldman (2000), and Nichols and Stich (2003), I do not think that we should make accessibility through TI conditional on whether there are relevant questions about external matters. Instead, I suggest the following criterion:

An attitude—ψing that p—is accessible through TI if and only if there is a sentence S about external matters such that, if S were to be combined with “I do not ψ that p,” the resulting complex sentence would sound as paradoxical as “p, but I don’t believe that p.”

According to this criterion, wishes are accessible through TI—for there certainly is a sentence about external matters that, combined with “I do not wish that p,” leads to a statement that sounds as paradoxical as “p, but I don’t believe that p,” namely: “If only p were the case!” The reason why “If only Congress would pass more restrictive gun laws! But, well, I do not wish that Congress would pass more restrictive gun laws,” sounds paradoxical should be clear from the preceding considerations: the utterer of these sentences expresses her endorsement of a certain postulation and, at the same time, denies that she endorses the postulation in question.

5 Conclusion

By way of conclusion, let us ask how my account of the introspection of wishes fits into the picture of TI that I drew in Sect. 1: Is it economical? Does it account for the fact that one comes to know about one’s own mental states in both a peculiar and privileged way? Is it acceptable from a naturalist’s standpoint? Does it support epistemic equality between the intra- and the extra-mental?

Let us begin with the least problematic issue, namely epistemic equality. As I explained in Sect. 1, “epistemic equality” refers to the claim that the following two ideas are equally incoherent: first, the idea of a subject who is precluded from being aware of any mental states (and elements thereof, including intermediary entities like sense-data or modes of presentation), but nonetheless succeeds in achieving knowledge about the external objects of which she is aware through those mental states; second, the idea of a subject who is precluded from being aware of external objects, but nonetheless succeeds in achieving knowledge about her current mental states. My account respects this intuition of epistemic equality. Here is an illustration: Suppose that you wish that it would rain tomorrow. It seems plausible that you could not know anything about the lacuna of which you are aware through your wish unless you were aware of at least that element of your wish in virtue of which the respective lacuna is presented to you, to wit: the postulation you endorse on the occasion of wishing that it would rain tomorrow. Now, according to my account, you likewise would not be able to know anything about your wish unless you were aware of the respective lacuna.

Does my story account for peculiarity and privilegedness? By drawing on Byrne’s idea of an epistemic rule, my account might be glossed by saying that someone who comes to know about his or her wishes in the ordinary (i.e., first-personal) way follows an epistemic rule that reads, “If the lacuna that is determined by ‘If only p were the case!’ exists, believe that you wish that p.” Like the corresponding rule in the case of belief, this rule is self-verifying in the sense that, if it is followed, the resulting second-order belief is true. The reason is that being under the impression that the lacuna that is determined by “If only p were the case!” exists is not a mental state in its own right that is brought about by the wish that p. It just is the wish that p. Thus, it is impossible both to follow the rule and to get false results. If one is under the impression that the respective lacuna exists, then one ipso facto has the wish that p. Notice that the third-person version of the rule—“If the lacuna that is determined by ‘If only p were the case!’ exists, believe that the person opposite wishes that p”—is not self-verifying. This explains both the peculiarity and privilegedness of the way one comes to know about one’s own wishes.

What about economicy then? Well, it might be tempting to accuse my account of leading to a bloated ontology—after all, didn’t I introduce strange new abstract entities like postulations and lacunae? Yes, of that I plead guilty. But this is not to say that my account is extravagant in the sense that it posits special epistemic capacities and abilities beyond what is needed for ordinary knowledge. Of course, I claim that, in order to introspect one’s wishes, one has to keep an eye out for lacunae. Keeping an eye out for a lacuna, however, is not a special epistemic capacity. Recall that, according to my account, wishes are endorsements of postulations. To endorse a postulation is to adopt a certain perspective towards the world, namely a perspective under which the world appears to one as though it has a particular defect that has to be fixed either by bringing a certain state of affairs into existence or by eliminating a certain fact—in short: the world is represented as featuring a certain lacuna. Keeping an eye out for a lacuna, then, is just the same “activity” in that we engage in the run-up to the formation of a wish. Thus, in order to introspect our own wishes, we draw on the very same capacities and abilities that we use in order to form wishes. This is strictly parallel to the case of belief: in order to introspect one’s beliefs, one has to wonder whether certain states of affairs obtain. Wondering whether a particular state of affairs obtains, however, is not a special epistemic capacity—it is just the same “activity” in which we engage in the run-up to the formation of a belief.⁴⁹ Echoing a phrase once used by Gareth Evans in connection with introspective judgments about one’s own perceptual experiences, one might say that, in introspection of one’s wishes and beliefs, one “re-uses precisely those skills of conceptualization”⁵⁰ that one uses to form those wishes and beliefs. Thus, there is no need to posit an internal self-scanning or monitoring system that tracks one’s mental states in addition to those skills.

It cannot be denied, however, that there is a persistent intuition to the contrary. It seems that there is more to the introspection of wishes than the re-use of those skills that one uses to form those wishes. One might say that, while one is introspectively aware of one’s wish that p, one is not just endorsing the postulation expressed by “If only p were the case!” but also directs one’s consciousness upon this endorsement. How else should one move from the mere wish that p to the higher-order belief that one wishes that p? Thus, it might be tempting to say that, in addition to the first-order mental activity of endorsing a postulation, there has to be a higher-order mental activity that consists in monitoring the first-order activity of endorsing a postulation.

There is something right as well as something wrong about this objection. What is right about it is the fact that the introspection of wishes does not only consist in the re-use of those skills that one uses to form those wishes. Let us again take the example of wishing that it would rain tomorrow. If one’s introspective awareness of this wish only consisted in the re-use of those skills that one uses to form that wish, one would never arrive at the higher-order belief that one wishes that it would rain tomorrow. The reason is that forming the latter belief requires the application of the concept of oneself as oneself and the concept of wish. However, those concepts play no role in forming the mere wish that it would rain tomorrow. Thus, I admit that, when one is introspectively aware of one’s wish that p, one is not just endorsing the postulation expressed by “If only p were the case!” But I deny that the transition from the mere wish that p to the belief that one wishes that p requires a self-scanning or monitoring system that tracks the wish that p. There is a more economical explanation of how one moves from the mere wish that p to the belief that one wishes that p: by adding new concepts to the postulation that is expressed by “If only p were the case!”, namely the concept of oneself as oneself and the concept of wish, one transforms the postulation into a proposition that is expressed by “I wish that p.”⁵¹

Is my account acceptable from a naturalist’s standpoint? I think it is. Of course, by depicting attitudes as relations to abstract entities I have displayed a Platonic stance throughout this paper. However, as I have declared earlier, this was for heuristic purposes.⁵² I am not attached to postulations and lacunae—if they have to make way for entities that are more respected from the naturalist’s point of view, that is just fine with me. A proposal to dispense with postulations, for example, is to say that someone endorses a postulation if and only if he or she stands in a functionally characterizable relation to a sentence of the language of thought.

Before I end, I would like to sketch how my account of TI might be extended to non-doxastic attitudes other than wishes. First, let us look at desires to have—to possess or to own—something (or someone). It seems natural to suggest that these desires are relations to postulations that are expressed by optative sentences of the form “If only x were mine!”, namely, structures that consist of the relevant subject, the desired object or person, the first-order relation of belonging to, and the higher-order relation of demand. Hence, in order to know about one’s own desire to have something or someone, one has to keep an eye out for a lacuna that could be fixed by bringing something or someone into one’s possession. Something similar seems to hold for desires for someone else to do something. One might say that these desires are relations to postulations that are expressed by optatives of the form “If only you were to do x!”, that is, postulations that consist of the addressed person, a certain action, the first-order relation of performing, and the higher-order relation of demand. Consequently, in order to know about one’s own desire for someone else to do something, one has to keep an eye out for a lacuna that has to be fixed by another person’s performing a certain action.

The really hard cases for my analysis are desires to do something. A natural proposal is that they are relations to postulations that are expressed by optatives of the form “If only I would do x!”. However, this is not unproblematic. Consider Fred who has to finish a certain unpleasant task, say, correcting his students’ term papers. Due to his tendency to procrastinate, he does not sit down at his desk and begin to work but turns on the TV. Fred might sincerely express his state of mind by saying “If only I would finish this task soon! But, well, I don’t feel like it.” It seems, then, that Fred endorses the postulation expressed by “If only I would finish this task soon!” but does not have the desire to finish his task soon. Thus, the suggestion that desires to do something are relations to postulations that are expressed by optatives of the form “If only I would do x!” seems implausible.

However, I think that this problem can be solved by an appeal to Davis’s (1986) distinction between volitive and appetitive desires: insofar as Fred endorses the postulation expressed by “If only I would finish this task soon!” he has a volitive desire to finish his task soon; but insofar as he don’t feel like doing it he lacks the appetitive desire for finishing his task soon. Appetitive desires are on a par with cravings, yearnings, longings and urges and, therefore, are not attitudes at all, but lie at the interface between emotions and bodily sensations. Since the proposal at hand is designed to provide an understanding of attitudes, it should not be blamed for not providing an understanding of emotions or bodily sensations.⁵³

But the proposal that desires to do something are relations to postulations is not cut and dried yet. Imagine a non-swimmer walking down the beach, when she suddenly sees a person in danger of drowning. It seems reasonable to suppose that the non-swimmer might wish that she could swim now, but does not have the volitive desire to swim now. Given that both wishes and volitive desires are relations to postulations, this example seems to draw us into contradictions—for doesn’t it force us to say that the non-swimmer both endorses and does not endorse one and the same postulation?

In my view, this problem can be defused by distinguishing between two different postulations that are at play here: one is the postulation expressed by “If only I had the ability to swim now!”, the other one is the postulation expressed by “If only I would swim now!” The non-swimmer might endorse the first postulation, but she does not endorse the second one.

Well, this is not the end of the story—for I haven’t said a word about intentions yet. On the face of it, my analysis so far does not provide the resources to differentiate between volitive desires to do something and intentions. It won’t do, for example, to say that intentions are likewise relations to postulations, since we sometimes do not intend what we volitively desire. For example, I might have the volitive desire to insult my superior right now, but I do not intend to do so, since I know that doing so would have serious consequences for me. We cannot accommodate such cases by disambiguating postulations or by arguing that intentions are not attitudes. For one thing, there just seems to be no basis upon which to conclude that two different postulations are at play here. For another thing, there is no doubt that intentions are attitudes.

Due to this difficulty, I am inclined to admit that the analysis of wishes and desires that I put forward in this paper cannot be transferred one-to-one to intentions. However, I am confident that a more thoroughgoing account of other non-truth-apt senses that I mentioned above, namely directives, might overcome the difficulty. It seems natural to suggest that the intentional content of intentions coincides, not with the semantic content of optatives, but with the semantic content of certain self-directed imperatives.⁵⁴ Consequently, in order to know what one intends, one has to keep an eye out for a worldly item that is neither a state of affairs nor a lacuna, but something third. For reasons of space, however, I cannot pursue this issue any further here.

Let me sum up. I have motivated and developed a non-orthodox analysis of wishes that avoids certain problems that stand in the way of extending the idea of transparent introspection to non-doxastic attitudes. The central tenet of my analysis is that wishes are relations to postulations—the non-truth-apt meanings of optative sentences. To stand in relation to a postulation is to see the world from a specific perspective, namely from a perspective that seems to reveal certain defects—lacunae, as I have called them—that have to be fixed, either by bringing certain states of affairs into existence or by eliminating particular facts. This perspective is fundamentally different from the perspective of someone who accepts a proposition: the way one sees the world qua the endorsement of a postulation cannot be mimicked by any doxastic means, say, for example, by an evaluative judgment to the effect that it would be good if a certain state of affairs were to obtain. So, in order to know whether one wishes that p, one has to keep an eye out, not for a certain fact, not even for an evaluative fact, but for a particular lacuna, that is, the worldly item that is determined by the postulation expressed by “If only p were the case!” This account not only strictly parallels the paradigmatic account of transparent introspection of beliefs, it is also suitable as a blueprint for accounts of transparent introspection of other non-doxastic attitudes.