Consciousness and Perspectival De Se content

Abstract

Most people think indexical thought has special content (de se content). However, it has been acknowledged that classical examples, such as those offered by Perry and Lewis, are insufficient to establish this conclusion. Ongoing discussions typically focus on first-person beliefs and their relationship to the explanation of successful behavior and linguistic practices. Instead, I want to direct attention to the phenomenal content of our conscious experiences and the largely neglected contribution that its comprehension can make to the way in which first-person representation is understood. This paper offers a novel argument for both the existence and a specific interpretation of the essential indexicality thesis based on widely accepted assumptions regarding the relation between phenomenology and representation. Specifically, I will argue that if (i) two different individuals can have token experiences with the same phenomenology and ii) two token experiences of ordinary subjects with the same phenomenology have the same phenomenal content, then phenomenal content is de se content of a particular kind: perspectival de se content.

1 Perspectival De Se content

Let us call ‘indexical thoughts’ those mental representations the subject sincerely expresses, or is disposed to sincerely express, using an indexical expression. Examples include the belief that I am a philosopher, the perception that there is a computer here, and the desire that the war ends now. The canonical way to individuate a thought is to determine the kind of attitude it is and the content that it has (e.g. Fodor 1987, p. 11). The belief and the desire that the war ends now involve the same content but a different attitude. Differences among thoughts that involve the same attitude, among beliefs, for example, are explained by a difference in content.

If one accepts such an approach, examples like Perry (1979) messy shopper or Lewis (1979) two gods have led to the consideration of indexical thoughts as special due to traditional theories of content failing to adequately account for them. However, recent acknowledgments suggest that despite the persuasiveness of these examples, they alone are insufficient to establish this conclusion ((Cappelen & Dever, 2013; Devitt, 2013; García-Carpintero, 2017; Ninan, 2016; Magidor, 2015; Shaw, 2020; Torre, 2018). See (Morgan & Salje, 2020 for a recent review.) Additional arguments are required to further support this claim.

Traditional theories disagree about what contents are supposed to be—e.g., complexes of objects and properties, Fregean senses, classes of possible worlds, or sentences in the language of thought. Nonetheless, they agree they are the kinds of entities that have truth values and that satisfy two additional conditions (Kölbel, 2013; Ninan, 2016):

Global::

Contents have a global truth value, as opposed to having truth values only relative to subjects or times: given a complete description of the world, we can assign them a truth value. In other words, they are assigned the same truth value for any location or point of evaluation (e.g., space, time, or subject) within a world.

Portable::

They are portable. Different mental representations can share the same content, which can be entertained by different subjects at different places and times.

Let us call ‘de se contents’ those contents traditional theories cannot accommodate because they are either not-global or not-portable.¹ While this paper concerns the content of conscious experiences, we’ll introduce the debate on the nature of de se representation through belief, avoiding controversies that will be discussed later.

Consider the beliefs two different subjects, Xóchitl and Zahira, would express by uttering:

• (1) Xóchitl is wearing a hat on 01/10/2023 at 5 p.m. GMT.

This is a clear example of Zahira and Xóchitl having the same thoughts and, therefore, sharing the same content, which satisfies portable. Moreover, we can assign these beliefs a truth value once we fix the world. It is true iff Xóchitl is wearing a hat in the actual world on 01/10/2023 at 5 p.m. GMT. Hence, global is also satisfied. Contrast these beliefs with one Xóchitl would express by uttering (2):

• (2) I am wearing a hat on 01/10/2023 at 5 p.m. GMT.

Xóchitl can rationally accept (2) without accepting (1). If this difference in cognitive significance has to be explained by a difference in belief, and thoughts are canonically individuated, then they must have a different content—the attitude and the subject remain the same in both cases.

Skeptics of de se content argue that this phenomenon is analogous to other cases where beliefs with the same reference vary in cognitive significance (Cappelen & Dever, 2013; Magidor, 2015). If, for example, we can explain why the Joker attacks Bruce Wayne only when he believes that Bruce Wayne is Batman without giving up traditional understandings of content, we do not have to abandon portable or global to characterize indexical thought either.

Those who accept de se contents are divided into two groups based on their views regarding the necessary amendments for an adequate theory of mental content. Globalists might agree that the difference in cognitive significance in Xóchitl’s beliefs is similar to other cases involving co-referential terms. However, they deny that the content of indexical thought is portable. For example, Frege’s suggestion that first-person thought depends on “a particular and primitive way, in which he is presented to no-one else”((Frege, 1956), p. 298). This means that Zahira cannot have the same belief as Xóchitl because no one else can represent her under such a“mode of presentation.”Those who appeal to reflexive contents and deny the possibility of different thoughts having the same reflexive content, such as the singular proposition that has the token thought as a constituent (‘the thinker of this very thought is wearing a hat’), further exemplify the globalist position (See (García-Carpintero, 2017) and (Perry, 2001a, b) for an illustration of different versions thereof).

Contrastingly, portabilists dispute the global nature of content. Lewis (1979) exemplifies this by proposing that beliefs are self-attributions of properties, such as wearing a hat. In this view, if Xóchitl is wearing a hat but Zahira is not, the self-attribution is correct for Xóchitl but not for Zahira. Lewis’s understanding of beliefs can be straightforwardly translated into a canonical relation between a subject and a truth-evaluable content by letting contents be classes of centered—worlds ordered pairs of worlds and centers (e.g., subjects and times).² Therefore, Xóchitl and Zahira may entertain the very same content, and portable is thus respected. However, a truth value can be assigned only relative to a subject and a time, which is a straightforward rejection of global. A reflexive theory that holds Xóchitl and Zahira believe the same content can also illustrate this view.³

A fundamental aspect of the portabilist approach is that the content is not a singular proposition. Instead, it excludes the particular individual one would take the indexical expression to refer to—e.g., a subject, a place, or a time—from the content itself. This exclusion allows for the possibility of the content being portable and having the same content when different particulars are involved. In this sense, one can think of this content as encompassing a perspective that different particulars can occupy. I will call de se content that is portable and not-global ‘perspectival de se content.’

Moreover, whereas global contents are true or false depending exclusively on how the world turns out to be—hence, the same content can be true in some worlds and false in others—perspectival content is more fine-grained. It depends on who the subject (or the token mental state, place or time) turns out to be. As a result, the same perspectival content might be true for some particulars while being false for others, even within the same world.⁴

This paper does not focus on the content of beliefs but rather on the content of conscious experiences. The content of conscious experience is often expressed by deploying an indexical expression. The question that we must answer is how this content should be understood. In their influential work on de se skepticism, Cappelen and Dever (2013) note that not many arguments attempt to show that the content of conscious experience is de se.⁵ This work contributes to filling this gap.

I will argue that experiences whose content is expressed using indexical expressions have perspectival de se content. If experiences with the same phenomenology have the same content, then two different individuals can have experiences with the same content (portable). However, since the truth value of the content changes from one subject to another within a world, this content cannot be global. The rest of the paper discusses the details and scope of the argument, including the idea that two experiences of different subjects can share the same phenomenology in §2 and the relation between phenomenology and phenomenal content in §3. The argument itself is presented in §4, and possible objections are discussed in §5.

2 Same phenomenology

When one looks at an apple, one has a specific experience; there is something it is like for one to look at the apple. The phenomenal character or phenomenology of an experience is what it is like for the subject to have the experience. Our basic, pre-theoretical understanding acknowledges that two numerically different experiences can share identical phenomenology: the experiences of looking at an apple before and after blinking one’s eyes. This is also true in the case of token experiences of different subjects. As children, we have wondered if other people see things the same way we do and, although not using technical terms, whether someone else’s experience when looking at the grass might have the phenomenology of our experience when looking at a ripe tomato.

Same Phenomenology:

There can be two subjects A and B, having - maybe in different circumstances—two token experiences, $e_{A}$ and $e_{B}$, such that $e_{A}$ and $e_{B}$ have the same phenomenology.⁶

To understand Same Phenomenology, we must clarify what it means for two experiences to have the same phenomenology. This is typically defined as the experiences being phenomenally indistinguishable under ideal conditions—abstracting away from discriminatory and memory limitations. This characterization invites questions regarding interpersonal comparison (Frege, 1956) and the impossibility of verifying claims about sameness in phenomenology between individuals (Schlick, 1959). However, the coherence of interpersonal indistinguishability has been carefully defended by Shoemaker (1982, 1996). Even Stalnaker (2000), who defends the ‘Frege-Schlick’ doubts, concedes the intuitive force of the idea that two people who are physical duplicates undergo experiences with the same phenomenology. Stalnaker’s concession mixes, notwithstanding, the pull of two independent intuitions: one regarding the possibility of two different individuals having two token experiences with the same phenomenology and another regarding phenomenal internalism. It is thus necessary to distinguish the two to avoid unnecessary restrictions in the scope of this paper.

Phenomenal internalism is roughly the view that phenomenology supervenes on the individual’s internal states. It contrasts with phenomenal externalism, the idea that external features partly constitute phenomenology. The interesting intuition for current purposes is neutral on the details of the debate between phenomenal internalism and externalism. Hence, the intuition regarding shared phenomenology can be reframed by thinking of the experiences of two individuals who are not only internal duplicates but also in identical situations concerning the required external properties.

Stalnaker also questioned whether the intuition that different individuals can have experiences with the same phenomenology comes from our pre-theoretical understanding of phenomenology or from the assumption that physicalism is true—and hence that everything, including phenomenology, globally supervenes on the physical. I think that there are good reasons to think our pre-theoretical notion of phenomenology supports this intuition, as it is shared by those who reject physicalism. For example, in one of the most famous thought experiments against physicalism, Jackson (1982) argues that all physical knowledge is insufficient to gain knowledge regarding phenomenology. He uses the example of Mary, who has complete physical knowledge of color vision but obtained within a black-and-white room. According to Jackson, Mary only learns the phenomenology of color experiences when she is released and sees the color for the first time. Jackson stresses that “the knowledge Mary lacked [...] is knowledge about the experiences of others” (p. 292). Chalmers, a prominent critic of physicalism, argues that it is not plausible that, in the actual world, two individuals who share their behavioral dispositions have experiences that differ in phenomenology. In the actual world, sufficiently fine-grained functional duplicates are phenomenal duplicates (Chalmers, 1995. Cf. Sebastián & Martínez, 2024). As functional duplicates are clearly (nomologically) possible, Chalmers’ argument supports Same Phenomenology.

This discussion spells out the support and scope for Same Phenomenology. The intuition that supports it concerns whether our worldmates can have, at least in some circumstances, experiences like ours. Hence, we can take the possibility involved in Same Phenomenology to be nomological. Same Phenomenology is widely accepted in current discussions about the nature of conscious experience. While different theories may have different conditions for shared phenomenology, very few reject the possibility altogether, as it would come at a significant cost.⁷

3 Same phenomenal content

In having an experience, we become aware of things; in this way, conscious experiences convey certain information to us, reducing uncertainty regarding our environment and ourselves. For example, suppose we don’t know whether a red apple is on the table. In that case, such uncertainty can typically be dissolved by the information provided by the experience one has when looking at the table. The information thus conveyed is called ‘the content of the experience.’ Such content imposes conditions on reality, reflecting what it takes for the experience to be accurate or correct. In the previous example, the experience would be inaccurate if the object on the table happened to be a yellow pineapple, and the information gathered would be false or incorrect. The view that experiences have content in this sense is widely accepted, especially in the case of perception (Byrne, 2009; Nanay, 2014; Schellenberg, 2018; Siegel, 2010).

There is an intimate relation between the information the experience conveys, the content of the experience, and its phenomenology. In particular, there is some common information different token experiences seem to convey to their subjects when they have the same phenomenology. This has opened a route for theorizing and investigating the phenomenology of experience by studying what experiences convey to us. As Chalmers notes, “given a specific phenomenology, it seems that if a mental state has this phenomenology, it must also have a certain specific representational content” ((Chalmers, 2004), p. 157). Call the representational content shared by experiences with the same phenomenology ‘phenomenal content’. Thus:

Phen$\rightarrow$ PCont:

Necessarily, if two token experiences have the same phenomenology, they have the same phenomenal content.

This entailment is neutral on possible relations of identity, determination, or grounding that might hold between phenomenology and representational content. As a result, it is not surprising that Phen$\rightarrow$ PCont has been supported by philosophers with a wide range of theoretical perspectives, ranging from those who seek to explain phenomenal content by appealing to phenomenology to those who seek to reduce conscious states to representational ones.⁸

However, Phen$\rightarrow$ PCont is not without its challenges. Some, like (Speaks, 2009) believe that de re thought should be grounded in the content of perception. If I’m looking at an apple, and my de re belief about it is to be based solely on the content of the experience, then the particular apple has to be part of it. This poses a challenge to Phen$\rightarrow$ PCont: if the apple is replaced by a qualitatively identical one, I’ll have an experience that is phenomenally indistinguishable from the previous one, yet the content changes. However, it is highly debatable whether de re thought must be grounded exclusively on the content of experience or also in some other relations like causal or teleofunctional (Lewis, 1979; Sect. XII-XIII). Moreover, Phen$\rightarrow$ PCont states that token experiences with the same phenomenology have a shared representational content (phenomenal content). This is consistent with there being some further representational content that different token experiences with the same phenomenology might not share, which, in turn, grounds de re thought.⁹

An independent reason to reject Phen$\rightarrow$ PCont can be derived from particular views on the nature of representational content. For example, if one thinks that physical duplicates share phenomenology but not the corresponding representational content, which depends on features external to the individual (Block, 1990). The internalist intuition regarding phenomenology is widely shared, and some have argued that it finds support in empirical research in psychophysics and neuroscience (Nanay, 2014; Pautz, 2021). On the other hand, there might be different ways to motivate externalism regarding representational content. For example, following teleological theories of mental content (Millikan, 1989; Neander, 1991; Papineau, 1984), one might have reductivist aspirations and think that representational properties depend on having a specific evolutionary history, or one might believe that high-level properties—such as being water or an apple—enter phenomenal content (Di Bona, 2017; Siegel, 2010) and that representation requires causal interaction. In either case, Phen$\rightarrow$ PCont is defeated if one endorses phenomenal internalism because phenomenal duplicates might differ in their evolutionary histories or causal interactions. A radical position would deny any dependence or grounding relation between any kind of representational content and phenomenology, along the lines recently defended by Papineau (2021). He thinks that conscious states represent as a matter of contingent psychological facts and correlations that hold between some states with the same phenomenology and environmental features. If any of these reasons are compelling, they support the possibility of two experiences with the same phenomenology that do not share any common content, hence the rejection of Phen$\rightarrow$ PCont.

These concerns about Phen$\rightarrow$ PCont can be mitigated by considering a less stringent version of this thesis. Relaxing the entailment expands the scope of the argument in the next section. Those who reject Phen$\rightarrow$ PCont may agree, as Papineau does, that experiences with the same phenomenology commonly convey some information. They endorse the entailment between phenomenology and phenomenal content, so long as it is restricted to the experiences of organisms that share some relevant psychological facts, relations, and phylogenetic history. Humans in our world are ideal candidates for satisfying these restricted conditions, as we share evolutionary history, cognitive architecture, and contingent relations to natural kinds. Hence:

Phen$\rightarrow$ PCont*:

If two token experiences of actual humans with normal cognitive capacities have the same phenomenology, then they have the same phenomenal content.

Phen$\rightarrow$ PCont* is a very minimal demand on the relation between consciousness and representation that allows us to investigate the nature of states with phenomenology in the first place. It is acceptable for a vast range of theories that otherwise disagree about substantial issues. It remains neutral in the debates surrounding phenomenal internalism and semantic externalism. It is compatible with those who think intentionality depends on phenomenology, as defenders of phenomenal intentionality do. And it is also compatible with all sorts of reductive representational theories of consciousness. Moreover, it can be accepted by those who think that phenomenology is not exhausted by phenomenal content and those who think that experiences can have contents beyond phenomenal content.

4 The argument for perspectival (portable and not-global) phenomenal content

To develop our argument, we must consider experiences whose phenomenal content is expressed by deploying an indexical expression—and then evaluate how such content is to be understood. Many philosophers believe this is true for most, if not all, experiences, given that awareness in experience involves some sort of self-awareness. This idea has roots dating back to Aristotle (1968) and has been further elaborated in recent phenomenology¹⁰ and analytic philosophy of mind.¹¹ For instance, in discussions about phenomenal consciousness, a distinction is drawn between the qualitative character, which accounts for the differences in phenomenology between various types of experiences, and the subjective character, which imbues our experiences with a sense of ‘mineness’ or ‘for-me-ness’ (Levine, 2001; Kriegel, 2009b).¹² If we accept the idea of a subjective character of consciousness based on phenomenology (Kriegel, 2009a, b; cf. Nida-Ruemelin, 2017), then reconciling this view with the representationalist approach we are considering, necessitates identifying an aspect of phenomenal content that corresponds to this subjective character.

The debate around precisely understanding subjective character leads directly to the challenge of determining the most appropriate way to express phenomenal content. Some theorists suggest that necessary self-awareness pertains to the experience itself, proposing content expressions like ‘This experience/state is such-and-such.’ Conversely, others advocate for a first-person perspective, leading to expressions like ‘Such-and-such state is mine/for me’ or ‘I am in such-and-such a state or relation.’¹³ For the purpose of our discussion, we can navigate around the intricacies of understanding subjective character, as both perspectives utilize indexical expressions to convey phenomenal content. Our pivotal concern is interpreting these expressions and assessing their connection to potential de se content.

However, some theories within the broadly representationalist approach are hesitant to recognize the subjective aspect as part of phenomenology. Consequently, our argument could gain additional depth and broader appeal by reaching a conclusion that does not exclusively hinge on the subjective character of experience.

The content of perceptual experience, especially visual ones, is often specified using indexical expressions like ‘in front of me,’ ‘behind me,’ ‘now,’ ‘since some time ago,’ ‘here’ or ‘over there.’ If they convey the presence of mind-independent entities, they do so, at least typically, in relation to the subject’s own location. Suppose we let those entities be ordinary objects and their properties for ease of exposition. In that case, my current visual experience looking at an apple by a computer does not merely convey the presence of an apple and a computer but the presence of an apple to the left of a computer from here/my location. Consequently, we can focus on these examples. The analysis will directly apply to understanding phenomenal contents encompassing subjective character, as discussed in Section 5.2. These contents are characteristically expressed using indexical expressions, a point recognized across the spectrum of debate, even by those who challenge the notion of such contents as de se (Cappelen and Dever, 2013, Ch. 8). This consensus prompts the crucial question: can these contents, articulated through indexical expressions, be classified as de se content?¹⁴ I contend that embracing Same Phenomenology along with Phen$\rightarrow$ PCont* imposes a specific reading of the phenomenal content of these experiences that must be portable and hence that their phenomenal content cannot be global: it is perspectival de se content.

As we have observed, our theories should accommodate the (nomological) possibility that different individuals might have experiences with the same phenomenology. Same Phenomenology does not imply that the phenomenal character of any given experience from one subject is necessarily shareable with someone else’s experience. It makes room for the possibility of “unrepetible” experiences. Nevertheless, within this framework, it’s reasonable to assume that not all experiences are unique in this manner. Take, for instance, the visual experience Xóchitl has when looking at an apple by a computer ($e_{X\acute{o}chitl}$), as in the previous example. Even if one allows for the existence of “unrepeatable” experiences, anyone endorsing Same Phenomenology would likely agree that Xóchitl’s experience doesn’t fall into this category. Thus, it is nomologically possible for a different individual to have an experience with the same phenomenology as Xóchitl’s.

Suppose Zahira is an individual whose specific experience, $e_{Zahira}$, shares the same phenomenology as Xóchitl’s, $e_{X\acute{o}chitl}$. To make a claim about the phenomenal content of $e_{Zahira}$ in relation to $e_{X\acute{o}chitl}$, we need a principle linking phenomenology with content such as (Phen$\rightarrow$ PCont). However, exploring the restricted version of the entailment (Phen$\rightarrow$ PCont*) is more intriguing.

In the first section, we noted that various theories might place different conditions on Zahira—and potentially her environment—to have an experience with the same phenomenology as $e_{X\acute{o}chitl}$. Neither these constraints nor the laws of nature seem to prevent her from sharing with us the kind of things that might be relevant to the restriction of the entailment—such as sharing our evolutionary history, our cognitive architecture, and our contingent relations to natural kinds. Consequently, we can safely assume that Zahira is a possible human of the actual world. This assumption leads us to conclude that both Zahira and Xóchitl, along with their respective experiences, $e_{X\acute{o}chitl}$ and $e_{Zahira}$, meet the criteria for Same Phenomenology and Phen$\rightarrow$ PCont*.

Content is portable just in case different individuals at different locations and times, can entertain it. In our example, involving the first person pronoun as an indexical expression, we must consider whether different individuals can entertain the content. The argument for phenomenal content being portable is straightforward in this context. Ex hypothesi, $e_{Zahira}$ and $e_{X\acute{o}chitl}$ have the same phenomenology. Consequently, under Phen$\rightarrow$ PCont*, they must have the same phenomenal content. Call such a content ‘$PCONT_{e}$.’ A sentence like ‘there is an apple to the left of a computer from Xóchitl’s location’ does not aptly describe this content, for it is not plausible that the experience Zahira has when looking at an apple close to a computer conveys something about Xóchitl. Moreover, this would prevent Zahira’s experiences from being veridical when Zahira is indeed looking at an apple to the left of a computer if no such arrangement exists near Xóchitl. For the same reasons, it is also implausible that $PCONT_{e}$ is adequately expressed by ‘there is an apple to the left of a computer from such-and-such location,’ where such-and-such location is the location Xóchitl happens to ocuppy, since Zahira can have an experience with the same phenomenology in a completely different location—for example, if they are in qualitatively identical rooms.¹⁵ Thus, in order to express $PCONT_{e}$ in such a way that it is a plausible content for both $e_{Zahira}$ and $e_{X\acute{o}chitl}$, they would deploy an indexical expression—e.g., ‘there is an apple to the left of a computer from my location’‘—that must be read in such a way that allows both Zahira and Xóchitl to have the same content. This imposes a reading of the content that respects portable . In general, phenomenal content has to satisfy portable if different token experiences entertained by different subjects (or at different times or locations) can have the same phenomenology and there is a connection between the phenomenology of these experiences and its phenomenal content—as provided by Phen$\rightarrow$ PCont or a restricted version thereof. This rules out a globalist interpretation (global but not-portable) of phenomenal content. Whether this content aligns with the skeptic’s perspective, being non-de se (both global and portable), or whether it constitutes perspectival de se content depends crucially on evaluating its global nature.

Content is considered global when it is assigned a consistent truth value at every point of evaluation within a particular world. In other words, once the state of the world is determined, we can establish a definitive truth value for the content. In the case under consideration, where the content is expressed deploying the first-person pronoun, it is natural to take the relevant points of evaluation to be the inhabitants of such a world. Hence, we should examine whether two individuals inhabiting the same world can have representations with the same content but different truth values.¹⁶

Zahira and Xóchitl are wordlmates and have experiences with the same phenomenology and, thereby, the same phenomenal content. Within the framework under discussion, the answer to whether such phenomenal content can have different truth values is a straightforward yes for anyone who accepts that veridical experiences can have the same phenomenology as hallucinations or illusions.¹⁷

Consider a scenario where Xóchitl’s experience is veridical, in contrast to Zahira’s, which is illusory. Zahira is looking at a pineapple to the left of a computer, yet due to a malfunction in her visual system, she ends up having an experience with the same phenomenal character as $e_{X\acute{o}chitl}$. Under Phen$\rightarrow$ PCont*, Zahira’s and Xóchitl’s experiences must have the same phenomenal content $PCONT_{e}$. However, this content holds true for Xóchitl but not for Zahira: there is indeed an apple to the left of a computer in Xóchitl’s case, but not in Zahira’s. Since Zahira and Xóchitl inhabit the same world, fixing the world is insufficient to assign $PCONT_{e}$ a truth value. $PCONT_{e}$ cannot be considered global as it yields different truth values for different individuals in the same world. This leads to the conclusion that the phenomenal content in such instances is portable and not-global, i.e., perspectival de se content.

5 Objections

This section will explore two distinct categories of objections relevant to our discussion. Initially, we will address objections specifically related to perceptual experiences, deliberately excluding considerations of subjective character for clarity. Following this, our attention will shift to objections primarily concerning subjective character, offering a comprehensive analysis of the cases considered.

5.1 Perception and perspectival content

(Cappelen & Dever, 2013; ch. 8, fn.6) propose, in reply to Schroder and Caplan (2007), an explanation of sameness in phenomenology based on a lambda-extracted “content”, which is not a carrier of veridicality—hence the scare quotes. In the example’s case, this would be the property picked up by the predicate P: $\lambda$x.there is an apple to the left of a computer from x’s location. They propose that truth value can only be assigned to particular experiences but not to the type they belong to—individuated by the common phenomenology. $E_{Zahira}$, predicates P of Zahira, and it is veridical just in case Zahira satisfies P; i.e., if she instantiates the property of being such that there is an apple to the left of a computer from one’s location.

The initial response to the skeptic should clarify that within our discussion framework, the attribution of (truth evaluable) content to phenomenology is granted, regardless of the ongoing debate surrounding de se phenomena. This framework posits that certain information, or phenomenal content, is consistently conveyed across experiences that share the same phenomenology, irrespective of the specific details of each experience. They all restrict uncertainties regarding possible alternatives, and they can be assigned a truth value depending on whether this information corresponds to reality. To illustrate, any experience with phenomenology identical to $e_{X\acute{o}chitl}$ would be considered inaccurate if, in reality, there is a computer but no apple present. This supports the claim that phenomenal content, the content associated with the kind, is indeed a carrier of veridicality.

Talk in terms of properties does not preclude this fact, as such property-content imposes conditions on the entity of which it can be adequately predicated and thereby is informative in the required sense. For example, Tye (2021), who has recently defended the view that experiences with the same phenomenology represent the same property, acknowledges that such content can be true—in the case of veridical representation—or false—in the case of a perceptual illusion. According to Tye, the kind of experience one typically has when looking at a green object represents the property of being green and is adequate if the perceived object is green and inadequate otherwise.

Phenomenal content cannot be fully characterized merely by reference to a property. This is because the adequacy of an experience hinges not just on the presence of the property but also on its instantiation by the experiencing subject. In other words, the fact that someone satisfies P cannot make the kind of experience with the same phenomenology as $e_{X\acute{o}chitl}$ adequate unless the individual that satisfies P is also the subject of the experience. This restriction holds irrespectively of the nature of the particular experience in question and who the actual subject of that experience happens to be.

To address this restriction, the skeptic could consider modifying the predicate attribution. This could involve explicitly specifying that the individual to whom the property is attributed must also be the subject of the experience. For instance, the predicate could be formulated as $\lambda$x.there is an apple to the left of a computer from x’s location and x is the subject of experience $e_{X\acute{o}chitl}$. However, this amended predicate cannot capture the phenomenal content of Zahira’s experience, which has the same phenomenology as $e_{X\acute{o}chitl}$. If Phen$\rightarrow$ PCont* holds, then $e_{X\acute{o}chitl}$ and $e_{Zahira}$ have the same content. However, Zahira is not the subject of $e_{X\acute{o}chitl}$.

Another option might be to generalize the predicate, replacing the specific token with the type, as in $\lambda$x.there is an apple to the left of a computer from x’s location and x is the subject of an experience of kind E—where E is the kind of all experiences with the same phenomenology as $e_{X\acute{o}chitl}$. However, this approach is also flawed. If that were the content of Zahira’s experience, her experience would be considered adequate even if she is hallucinating without an actual computer present, as long as another person is having a veridical experience with identical phenomenology.

Addressing the identified issue with Cappelen and Dever’s proposal results in an amended solution that aligns with Lewis’, showing no substantial difference. According to Cappelen and Dever, experiences with the same phenomenal character have a common lambda-extracted content, a predicate that denotes a property. However, any experience with the same phenomenal character as Xóchitl’s will predicate such a property of the subject of the corresponding particular experience and cannot be veridical unless its subject instantiates the property. This imposes a distinctive relation between the subject of the experience and such a property. The properties suggested by the skeptic to be phenomenal “content” are precisely the kind of properties that some portabilists such as Lewis postulate to be the content of de se representations, which involve, additionally, a distinctive relationship between a subject and such properties. He calls such a unique relationship in the case of belief ‘self-attribution.’¹⁸ Properties do not have truth conditions, but the self-attribution of these properties is indeed correct or incorrect depending on whether the subject instantiates the property. As Lewis himself argues, we can reinterpret discussions about properties as content of attitudes within the canonical framework. This framework involves an attitude towards a content that is evaluable for truth. However, this reinterpretation requires a significant shift: substituting the traditional idea of content being true at a world with that of being true at specific locations within that world, such as individuals, times, or places. In other words, we must transition from traditional propositions, which are generally understood as defining classes of worlds, to centered propositions, which define classes of centered worlds.¹⁹

However, Cappelen & Dever (2013, ch. 5) contend that talk in terms of centered worlds does not inherently require introducing special content. In essence, they argue that such an approach does not automatically imply that the content under consideration is de se. Focusing on perception (ch.8), they discuss the case of Maeve, who possesses a unique perceptual system attuned to the environment surrounding another individual, Brigid. They use this example to argue that Maeve’s experience represents how things are around Brigid, denying the need for a first-person feature in perceptual content.

To begin with, it’s important to note that there’s no provided justification to assume that Maeve’s experience represents Brigid’s surroundings, rather than misrepresents how things stand around Maeve. While it is true that in the given example, Maeve’s experience is triggered by the environment around Brigid, we must heed Millikan (2000) caution against confusing the cause of a representation with its content.²⁰

Our understanding of the challenge becomes, nonetheless, clearer if we accept, for the sake of argument, that their example indeed supports the idea that the expression ‘there is an apple to the left of a computer from my location’ does not accurately capture the phenomenal content of Maeve’s experience. However, they interpret the case as evidence that perceptual content does not inherently introduce essential indexicality (p.149) because it can be interpreted non-indexically as ‘there is an apple to the left of a computer from Brigid location’: “Maeve can have perceptual contents that have no absolute truth conditions, but whose role in Maeve’s perceptual system is to be true relative to Brigid” (p.149).

At this point, careful consideration is crucial in understanding this claim. Cappelen and Dever advocate for object-dependent content, where the object itself is part of the content. However, such content cannot be an adequate candidate for phenomenal content because it is perfectly global: once we fix the way the world is, we thereby fix Brigid’s location, and we can assign a truth value to the content.²¹

If Xóchitl’s and Maeve’s experiences have the same phenomenology and satisfy Phen$\rightarrow$ PCont*, their experiences must have the same phenomenal content, which must be portable. However, the content attributed by Cappelen and Dever to Maeve’s experience is not a plausible candidate for the phenomenal content of Xóchitl’s experience, as Xóchitl’s experience does not predicate anything of Brigid. If we want to use a sentence that they both suitably use to express the phenomenal content, we must reintroduce an indexical expression, such as ’there is an apple to the left of a computer from the location where my visual information/this information is registered.’ This content must be portable so that the phenomenal content is the same for both experiences. To achieve this, the phenomenal content must not be “referentially anchored,” to borrow Cappelen and Dever’s phrase (p. 96), and it should not depend on a particular subject (Xóchitl, Maeve, or Brigid) or a specific token piece of information.²² Consequently, for the reasons we have already discussed, the content is not global: Maeve’s experience could be veridical, while Xóchitl’s is illusory. Phenomenal content is perspectival de se content.

5.2 Subjective character and perspectival content

Let’s consider Xóchitl’s experiences, $e_{X\acute{o}chitl}$, and grant it has subjective character. Different theorists may offer different characterizations of subjective character, but for illustrative purposes, let us assume that the expression ‘I am in such-and-such a state’ captures its phenomenal content. The fact that one oneself is represented as being in a certain state reflects the subjective character. Experiences with different phenomenology—different qualitative characters—would convey that the subject is in different states. Imagine Zahira’s experience, $e_{Zahira}$, shares the same phenomenology as Xóchitl’s. Based on the previous section’s argument, under Phen$\rightarrow$ PCont*, $e_{Zahira}$ and $e_{X\acute{o}chitl}$ must have the same phenomenal content; therefore the content is portable. Yet, if it is possible that Zahira is in the state that her experience conveys, but Xóchitl is not, then the content is not-global.

A potential objection to the argument might concede that Zahira and Xóchitl’s experiences convey that they are in the same state, thereby sharing the same qualitative character. However, one might suggest that, due to Zahira and Xóchitl being distinct subjects, their experiences possess different subjective characters despite other similarities.²³ The problem with this objection is that subjective character is granted to be part of phenomenology. Therefore, if the subjective character changes from one subject to another, it becomes not only nomologically but also metaphysically impossible for two different subjects to have experiences with the same phenomenology. This is a difficult position to accept. Moreover, this objection requires independent reasons to reject the intuitive motivations discussed in §2 in favor of Same Phenomenology. The mere emphasis on subjective character does not suffice to reject Same Phenomenology. This is because the idea that experiences possess a subjective character manifest in their phenomenology is entirely consistent with the possibility of two different subjects having experiences with the same subjective character and, consequently, the same phenomenology, as we have seen.

The need for perspectival content becomes even more compelling if one takes conscious experiences to concern themselves. In such a case, an indexical expression like ‘this experience’ is employed in statements like ‘This experience is such-and-such’ to express phenomenal content (Kriegel 2009b; p. 203). Similar to the previous discussion, the fact that the experience itself is represented as being a certain way is intended to reflect the subjective character and differences in phenomenology by the way the experience is represented as being. However, if phenomenal content were object-dependent, as suggested above, then the particular token experience would be part of the content, and two token experiences would always have different phenomenal content. If there is an entailment between phenomenology and phenomenal content, then it would be metaphysically impossible for two token experiences, even from the same subject, to possess the same phenomenology.

Alternatively, one might contest the entailment between phenomenology and phenomenal content (Phen$\rightarrow$ PCont*). This stance implies that experiences, like those of Zahira and Xóchitl or even two token experiences of the same individual, could share subjective character and phenomenology but diverge in phenomenal content. However, relying solely on subjective character does not adequately justify this view: if phenomenal content is perspectival de se we can perfectly accommodate Zahira’s and Xóchitl’s experiences having the same phenomenal content. Moreover, rejecting the entailment is not viable for those pursuing the representationalist project outlined in §3, which includes prominent proponents of subjective character such as Kriegel.

6 Conclusion

The debate on essential indexicality has primarily focused on propositional attitudes, yet it has largely overlooked the potential contribution that studying the content of experience might make to the debate. In this paper, I have argued that if two subjects can have experiences with the same phenomenology and these experiences thereby share content, traditional theories are inadequate in characterizing such content. The argument in this paper supports the claim that phenomenal content is de se in that it is portable and not-global (perspectival de se content).²⁴

The goal of this paper has been to illustrate that the representationalist research program, the predominant approach in the philosophy of perception, necessitates the inclusion of specific types of content. The minimal assumptions about the relationship between phenomenology and content, made independently of the de se phenomenon debate, are not only very plausible and modest but are also broadly accepted in discussions about the nature of conscious experiences. This underscores the significance of our conclusion.

As a final remark, it is pertinent to note that the conclusion that there is perspectival de se content is restricted to phenomenal content, and it cannot be extrapolated to the content of other propositional attitudes without further controversial assumptions regarding the phenomenology of those states. While this paper’s argument allows for the possibility that belief content may not adhere to the de se criteria (portable and global) or it may correspond with a globalist approach (global and not-portable), recognizing the existence of perspectival de se content could potentially reduce the reluctance to accept contents that do not satisfy global in various philosophical analyses. Moreover, the conclusions drawn in this paper may also tip the balance in favor of an explanation of related phenomena, such as immunity to error through misidentification in terms of experience (Recanati, 2007, 2012; Sebastián, 2024), contrasting with alternative accounts (e.g., García-Carpintero 2018; Morgan 2012; Palmira 2020; Peacocke 2014; Wright 2012).²⁵