Is there introspective evidence for phenomenal intentionality?

Abstract

The so-called transparency of experience (TE) is the intuition that, in introspecting one’s own experience, one is only aware of certain properties (like colors, shapes, etc.) as features of (apparently) mind-independent objects. TE is quite popular among philosophers of mind and has traditionally been used to motivate Representationalism, i.e., the view that phenomenal character is in some strong way dependent on intentionality. However, more recently, others have appealed to TE to go the opposite way and support the phenomenal intentionality view (PIV), according to which intentionality is in some strong way dependent on phenomenal character. If this line of argument succeeds, then not only TE does not speak in favor of Representationalism, but it actually speaks against it, contrary to the philosophical common-sense of the last two decades. Moreover, the representationalist project of naturalizing phenomenal character turns out to be seriously undermined on the same intuitive grounds that were supposed to make it plausible. In this paper, I reconstruct and discuss the line of argument from TE to PIV and argue that our introspective intuitions (TE) do not push us in the direction of PIV. On the contrary, the line of argument from TE to PIV is (at best) simply too weak to force us to conclude that intentionality depends on phenomenal character in the sense required for PIV to be true.

1 Introduction

The so-called transparency of experience (TE) is the intuition that, in introspecting one’s own experience, one is only aware of certain properties (e.g., colors, shapes, etc.) as features of (apparently) mind-independent objects. TE is quite popular among philosophers of mind. Traditionally, it has been used to motivate the representationalist claim that phenomenal character (experience’s what-it’s-like-ness) is in some strong way dependent on intentionality (experience’s having representational content) against the traditional separatist orthodoxy, according to which, instead, no inherent connection obtains between phenomenal character and intentionality.

Representationalism has arguably been the mainstream view of consciousness over the last two decades. Despite this, some theorists have recently opposed this view and have argued that intentionality is in some strong sense dependent on phenomenal character. This is the phenomenal intentionality view (PIV). Interestingly, the PIV supporters usually appeal to TE as well and argue that it offers direct evidence for PIV: appreciating PIV’s truth would simply be a matter of attentively paying attention to what we gather from introspection on our own experiences (e.g., Horgan et al. 2004)—namely, TE. They thus build up a line of argument that leads from TE to PIV.

While introspective considerations seem to be always (and quite naturally) in order when it comes to discussions of phenomenal character, it might seem prima facie curious that two opposite camps appeal to the same intuition as evidence in support of two opposite theses. So, since associating TE and Representationalism is philosophically common-place, a question naturally arises as to whether TE really counts as evidence in support of PIV, that is: whether it really supports the phenomenal intentionality claim.

Answering this question is important because, if the line of argument from TE to PIV really is successful, then a powerful case in support of PIV can be made. For, if TE is actually what we gather from introspection, then it seems that considering our introspective intuitions alone—and I will clarify exactly how “alone” is to be understood later—pushes us to conclude that PIV is true. This is relevant not only in the light of the widespread consensus surrounding TE, but also in consideration of its important consequences. That is, first, the philosophical common-sense of the last 20 years, i.e., that TE makes a strong case in support of Representationalism, turns out to be overturned. Second, more importantly (and somewhat ironically), the representationalist mainstream about phenomenal character (along with its proposal for the naturalization of phenomenal character) is overturned as well, on the very same intuitive grounds that were supposed to make it a plausible and appealing option.

In this paper, I reconstruct and discuss the line of argument from TE to PIV and offer a negative reply to the question above: our introspective intuitions (TE)¹ do not push us in the direction of PIV. On the contrary, the line of argument from TE to PIV is (at best) simply too weak to force us to conclude that intentionality depends on phenomenal character in the sense required for PIV to be true. So, the PIV supporters are wrong.

Schematically, here is how I will proceed. After offering a quick survey of the debate on the relations between intentionality and phenomenal character (Sect. 2), I will argue for the following two substantive claims:

(1):

Representationalists and phenomenal intentionalists appeal to the same introspective datum, TE.

(2):

Arguing from TE to PIV leads us to the following trilemma: either (a) the argument is too weak or (b) it is not from TE or (c) it is to be rejected—where (a) is the best case and (b) and (c) are the worst cases.

In Sect. 3, I will be concerned with (1); in Sects. 4–7 I will deal with (2). The upshot will be: as far as our introspective intuitions (TE) are concerned, at best we are not forced to buy into PIV; at worst, buying into PIV runs us into serious trouble (Sect. 8). So, the line of argument from TE to PIV is not only far less powerful than the PIV supporters would like it to be, but it is also quite dangerous and, thus, not so convenient, after all—despite what it looked like prima facie.

2 Setup

Let me start with offering more details on the terminology and the background. This will serve us to better characterize the views on the table and appreciate the depth of the challenge raised by PIV.

2.1 Intentionality and phenomenal character

A traditional distinction in the philosophy of mind is between phenomenal character and intentionality. Consider an example. At the moment, I am seeing a red car parked across the street. Now, what can I say about this experience? In the first place, it seems quite natural to say that it “tells” me something about how things are in the world. For example, it tells me that something is red and with a certain shape, that it is in front of me now occupying a certain portion of the space, and so on. As long as it is in the business of telling me something, my experience represents that something is a certain way. What an experience represents is its representational content, which is usually understood in terms of conditions of satisfaction. Roughly, the idea is that, in telling us something about the world, our experiences are providing us with the conditions at which they are accurate (or inaccurate). So when I have an experience as of a red car, that experience tells me that the world is such that there is something red, with a certain shape, etc., in front of me now. If there is something in front of me now that is red, has that particular shape, etc., then my experience “tells me the truth” about the world: it is accurate. Otherwise, it is inaccurate.² Having a certain content requires experience to instantiate certain representational properties—e.g., the property of representing that something is red, the property of representing that something is occupying a certain portion of space, etc.³ Usually, experiences (or more generally mental states) that have representational contents are said to be intentional or, equivalently, to have intentionality.

However, there is not just something that my experience represents, but also something that it is like for me to be in it. In other words, my experience as of a red car not only represents things as being a certain way, but it also feels a certain way. When I look at the red car, I have a red-experience with its own peculiar red-feeling, which is qualitatively different from (say) the green-feeling of a green-experience. Likewise, seeing a red car feels to me qualitatively different from (say) seeing a red bike or from having a terrible toothache or getting angry at somebody. This what-it’s-likeness—the qualitative character that seems to inherently accompany every conscious experience—is the phenomenal character of the experience itself. Just as intentionality involves the instantiation of representational properties, phenomenal character requires phenomenal properties to be instantiated by the experience.

2.2 Separatism versus inseparatism(s)

If the prima facie distinction between intentionality and phenomenal character is legitimate, it seems fair to say that, even though experience is one and unitary, it still exhibits two prima facie different dimensions: an intentional and a phenomenal one. But, then, a question immediately arises: How are these two dimensions related—if at all?

Philosophers have divided as to the answer we should give to such a question. According to some, phenomenal properties are “representationally inert.” In other words, they are not (necessarily) involved in the business of representing; on the contrary, they are introspectible, non-representational, intrinsic qualities of the experience itself. Conversely, representational properties do not contribute to phenomenal character. The upshot, then, is that phenomenal character and intentionality are mutually independent and, as such, separable: although they often happen to occur together, there is no internal or necessary connection between them. This is Separatism, and it used to be the orthodoxy in the philosophy of mind (e.g., Nagel 1974; Peacocke1983; Block 1990).

However, during the last two decades, things have changed quite radically: philosophers have largely favored Inseparatism over Separatism.⁴ Inseparatists maintain that intentional content and phenomenal character are in fact non-separable: they do not merely happen to be there together at the same time in our experience, they are (in some way or another) necessarily connected. And so are the phenomenal and the representational properties of the experience. So, there is no such thing as representationally inert phenomenal properties.

This has given rise to further questions, choices and divides. In particular, if phenomenal character and intentional content are non-separable, it seems natural to ask what sort of relation binds them together in such a strong way. The different replies to this question generate different options on the inseparatist menu.

Representationalists (e.g., Harman 1990; Dretske 1995; Tye 1995, 2000) hold that, e.g., the fact that there is something it is like for me to see a red car depends in some way on the fact that my experience represents that something is red, has a certain shape, etc. So, the intentionality of the experience is prior to its phenomenal character. For the purposes of this paper, “priority” is a quite generic term that covers a set of anti-symmetric dependence relations stronger than supervenience, e.g., in-virtue-of, grounding and reduction.⁵

Others support PIV: phenomenal character is prior to representational content (e.g., Siewert 1998; Horgan and Tienson 2002; Loar 2002, 2003; Horgan, Tienson and Graham 2004; Pitt 2004; Kriegel 2007, 2008, 2011, 2013). So, experience’s instantiating certain representational properties ultimately depends on its instantiation of certain phenomenal properties. On this account, the fact that I am having an experience with a reddish phenomenal character when I look at the car makes it the case that my experience represents that redness is instantiated by something in front of me now.

Finally, still others claim that there is no priority at all: phenomenal character and intentionality are mutually dependent—presumably, equivalent, if not one and the same thing (e.g., Chalmers 2004; Pautz 2008, 2013). I call this view Egalitarianism.

In this paper, I will set aside the debate between Separatism and Inseparatism, to better focus on what happens within the inseparatist field. More specifically, I will explore the connections between PIV and TE and the viability of a line of argument leading from the latter to the former.

2.3 PIV and the representationalist way to naturalization

Over the last two decades Representationalism has been quite popular among philosophers. Plausibly, this is because Representationalism seems to put us in a quite convenient position for providing a full naturalistic account of phenomenal character—on the assumption that we already have satisfying naturalistic accounts of intentionality available (Dretske 1981, 1988; Millikan 1984, 1989; Fodor 1990). Roughly, the idea is: If representational content is prior to phenomenal character, then the road to the naturalization of the latter, via a naturalized account of the former, appears to be quite straightforward. It does seem that one “only” has to properly characterize the way in which phenomenal character depends on intentionality. Thereby, one kills two birds with one stone: one buys the naturalization of intentionality and, for basically the same price, one gets phenomenal character naturalized as well.

Given this, it should not be hard to see what sort of challenge PIV raises and the deep consequences for the debate on phenomenal character and its naturalization. For, should PIV turn out to be true, the representationalist project of reducing phenomenal character to naturalized intentionality would be (at minimum) seriously undermined, for at least two reasons. First, if the proponents of PIV are right, then phenomenal character cannot be cashed out in terms of intentionality; rather, it is the other way around. Second, if intentionality strongly depends on phenomenal character, not only do the antecedent difficulties concerning the naturalization of phenomenal character remain untouched, but some further difficulties also arise in regard to the very possibility of achieving something like the naturalization of intentionality itself. For, given PIV, we cannot have intentionality naturalized without first having phenomenal character naturalized.⁶

3 The transparency of experience

In this section, I will introduce TE and show that it is the introspective intuition to which both representationalists and phenomenal intentionalists appeal (Sect. 3.1). Then (Sect. 3.2), I will add some further quick considerations on TE, its scope, and its capacity to rule out Separatism.

3.1 TE, Representationalism, and the phenomenal intentionality view

TE enjoys the favor of many philosophers. Arguably, Moore (1903) put forward the idea first, but TE reached the peak of its popularity in the 1990s, when Harman (1990) and Tye (1995) vigorously reclaimed it. Here is what the transparency datum amounts to. Suppose I focus on the red car-experience I am having and, in particular, on its reddish phenomenal character. Now, no matter how hard I try, it seems introspectively impossible for me to be aware of that reddish character without also being aware of the redness of the car.

In Harman’s and Tye’s own words:

When Eloise sees a tree before her, the colors she experiences are all experienced as features of the tree and its surroundings. None of them are experienced as intrinsic features of her experience. Nor does she experience any features of anything as intrinsic features of her experience. And that is true of you too. There is nothing special about Eloise’s visual experience. When you see a tree, you do not experience any features as intrinsic features of your experience. Look at a tree and try to turn your attention to intrinsic features of your visual experience. I predict you will find that the only features there to turn your attention to will be features of the three. (Harman 1990, p. 39)

Try to focus your attention on some intrinsic feature of the experience that distinguishes it from other experience, something other than what it is an experience of. The task seems impossible: one’s awareness seems always to slip through the experience to blueness and squareness, as instanced together in an external object. In turning one’s mind inward to attend to the experience, one seems to end up concentrating on what is outside again, on external features or properties. (Tye 1995, p. 135)

In short, nothing that looks like an intrinsic, non-representational quality of experience seems to be revealed by introspection, but only features of (apparently) mind-independent objects. So, phenomenal character and representational content are non-separable: if one tries to introspect and focus on the phenomenal character of one’s visual experience alone, one cannot be introspectively aware of such phenomenal character without also being immediately aware of what one’s visual experience represents.

Hence, TE minimally suggests that phenomenal character supervenes on representational content. For our purposes, A supervenes on B if and only if there cannot be differences in A without differences in B, where “cannot” expresses metaphysical necessity. At minimum, then, TE suggests a supervenience claim. Notice that supervenience can be a symmetric relation, so “phenomenal character supervenes on representational content” is to be read in such a way that does not exclude the other direction.

What about phenomenal intentionalists? At least some of them appeal to TE as well. Take, for example, the following passage from Horgan and Tienson (2002):

Mental states of the sort commonly cited as paradigmatically phenomenal have intentional content that is inseparable from their phenomenal character. Let us consider some examples: first, experiences of red as we actually have them. You might see, say, a red pen on a nearby table, and a chair with red arms and back a bit behind the table. There is certainly something that the red that you see is like to you. But the red that you see is seen, first, as a property of objects… [R]edness is not experienced as an introspectible property of one’s own experiential state, but rather as a property of visually presented objects. (Horgan and Tienson 2002, p. 521)

Even though it does not make use of transparency-talk, this passage seems to state something like TE. Indeed, Horgan and Tienson’s point is exactly the same as the one made by Harman and Tye: in introspection, the features usually considered as components of phenomenal character—colors, shapes, etc.—are only available as properties of (apparently) mind-independent objects.⁷ Thus, phenomenal character does not seem separable from the content of the experience.

Kriegel (2007) openly notes that TE is perfectly compatible with, and provides evidence for, PIV:

It may be thought odd to cite the transparency of experience in defense of phenomenal intentionality. The transparency claim is often adduced on behalf of the attempt to naturalize phenomenal character by combining an intentionalist account of it with traditional naturalistic accounts of intentionality in terms of the representation relation. By contrast, the notion of phenomenal intentionality is often touted in the context of rejecting this program. However, regardless of the wider motivations commonly at play in these discussions, the transparency of experience—the fact that phenomenology appears, from the first-person perspective, to be inherently intentional—is surely evidence for the existence of phenomenally constituted intentionality. (Kriegel 2007, p. 321)⁸

Thus (at least some of the) phenomenal intentionalists appeal to TE.

3.2 TE, its scope, and Separatism

The fact that TE minimally suggests a supervenience claim seems already enough to undermine Separatism. For, if the latter is true, it should be possible for us to keep phenomenal character and content separated and introspectively spot the non-representational, intrinsic phenomenal features of experience as separate from the representational ones. But, this is precisely what TE rules out: TE tells us that it is introspectively impossible to keep phenomenal properties and representational properties separated from each other. So, our introspective intuitions seem to offer evidence against the separatist’s account of phenomenal properties as representationally inert. Therefore, it seems plausible to conclude that TE provides us with a prima facie strong case against Separatism.

Of course, Separatists have tried to resist this argument in many ways and pointed at various cases of experiences (both perceptual and non-perceptual) in which TE seems to fail. For example, many have argued that the affective phenomenal character of pain or emotion does not seem to be entirely or satisfactorily re-describable in terms of (apparently) mind-independent object and their properties (De Sousa 2004; Aydede 2006; Block 2006). To some, moods and orgasms do not appear representational at all (Searle 1983; Block 1996, 2003; Deonna and Teroni 2012): when one is depressed, for example, one’s experience does not seem to represent anything, and the qualitative features involved in the experience do not seem to qualify any object. Finally, cases like blurry or double vision have been advanced as examples of non-transparent visual experiences (Boghossian and Velleman 1989; Smith 2008). Supporters of TE have offered their own replies and have tried to accommodate these cases (see, e.g., Crane 1998; Tye 1995, 1997, 2005, 2008; Martínez 2011; Mendelovici 2013; Kline 2015; Lycan 1996).

Be that as it may, I will set these issues aside here. In what follows, for argument’s sake I will take for granted that TE stands, at least in the case of vision. More precisely, I will assume that visual experiences are transparent and, as such, do militate in favor of Inseparatism, at least as far as vision is concerned. The idea is: If TE does not speak in favor of PIV in the case of vision, a fortiori it does in no other case. From now on, then, I will be using the terms “experience” and “visual experience” as interchangeable.

4 Two strong interpretations of TE

As noted, the minimal interpretation of TE in terms of supervenience seems to offer some grounds to reject Separatism. But might TE suggest something less minimal and speak in favor of a positive view, as opposed to only speak against Separatism? After all, the minimal interpretation does not exclude (at least) two further, stronger interpretations that are mutually exclusive:

(SI₁):

Phenomenal character is ultimately grounded in the representational properties of the experience. (This interpretation presupposes the priority of intentionality.)

(SI₂):

Phenomenal properties ultimately ground the kind of content we are experientially and introspectively aware of. (This interpretation presupposes the priority of phenomenal character.)

Ultimately, this comes down to asking whether TE speaks in favor of Representationalism or of PIV, since (SI₁) and (SI₂) are nothing but the representationalist’s and the phenomenal intentionalist’s claims (respectively).⁹

Notice: sociologically speaking, such a question arises only with the rise of PIV. Indeed, before PIV came out, there was no question that (SI₁)—and thus Representationalism—was the right interpretation of TE. Philosophically, this was common-sense. There are at least two reasons for that. First, the fact that one cannot be introspectively aware of phenomenal character without being also aware of representational content has naturally been taken as pushing the idea of a certain priority of the latter over the former. (This might sound particularly natural if one’s concern is finding a way of getting phenomenal character naturalized.) Second, TE suggests that an internal relation obtains between phenomenal character and representational content. Before PIV, no other way of looking at that relation was out there but Representationalism, which simply was the only game in the inseparatist town.

However, after the rise of PIV, things have changed. Recall the passage from Kriegel quoted above: he claims that TE offers direct evidence in favor of PIV. This way of reasoning is exactly the opposite of the one described right above. The thought here seems to be something along the following lines: the fact that introspection of phenomenal character reveals nothing over and above representational content suggests that there is at least a certain kind of content that is ultimately constituted by/grounded in phenomenal properties.

I will offer more details to better spell out and fully unpack the line of reasoning behind the phenomenal intentionalist’s reading of TE momentarily. Let me just conclude this section with the following remarks. The rise of PIV as a theoretical option, and of (SI₂) as a possible strong interpretation of TE, does not necessarily make (SI₁) wrong—and thereby Representationalism false. It just calls into question and blocks the traditional quick and easy move from TE to (SI₁). To prove (SI₁) wrong, one has to show that (SI₂) is the right interpretation of TE. This amounts to showing that TE speaks in favor of PIV. So, the question is: Does TE speak in favor of PIV? As anticipated, I am going to offer a negative answer: TE does not speak in favor of PIV against Representationalism.

5 The argument from introspection alone

The main arguments from TE for PIV have been offered by Horgan and his collaborators (Horgan and Tienson 2002; Horgan et al. 2004). Thus, my discussion will be mainly focused on them. However, the target of my criticism is actually wider. For the idea that our introspective intuitions push us in the direction of PIV is quite widespread among the phenomenal intentionalists, and so is the line of argument from TE to PIV. Let me begin (this section) with what I will call the Argument from Introspection Alone (Horgan et al. 2004). Then (next section) I will move to the Argument from Phenomenal Duplicates (Horgan and Tienson 2002).¹⁰

5.1 The argument reconstructed

The most straightforward way of building up an argument from TE to PIV is simply to say that merely appreciating TE leads us immediately to conclude in favor of the priority of phenomenal character: it is equivalent to, or at least sufficient for, realizing that phenomenal character is prior to intentionality. This is what Horgan, Tienson and Graham (HTG) claim:

To begin with, we believe that a strong case can be made for phenomenal intentionality […]. One important line of argumentation involves focusing the reader’s introspective attention on certain specific actual or potential experiences, as a way of prompting appreciation that such experiences reveal the presence of one or another specific kind of phenomenal intentionality. (2004, p. 301)

Prima facie, this might sound a bit puzzling. We have seen that TE immediately suggests a supervenience claim. But whether it suggests something stronger, like a priority claim, is precisely what is at issue here. It is something that does not seem as manifest as HTG are suggesting, but rather something that has to be shown.

However, it might well be that, when regarded in a non-naïve, theoretically sophisticated way, TE does lead us to conclude that phenomenal character is prior. This is probably what HTG have in mind. For example, consider the somewhat popular line of argument in favor of the priority of phenomenal character suggested by Siewert (1998):

Siewert’s Argument

(P1):

Experience is assessable for accuracy in virtue of its phenomenal character

(P2):

Being assessable for accuracy is having representational content.

Therefore,

(C):

Experience has representational content in virtue of its phenomenal character.¹¹

It is plausible that HTG have in mind something like Siewert’s Argument. In their version, the truth of (P1) simply becomes introspectively manifest on the basis of TE.¹²

Consider again my experience as of a red car. Among other things, that experience has a reddish phenomenal character. According to TE, having such a reddish phenomenal character is nothing but presenting redness as instantiated by some (apparently) mind-independent object. In other words, the condition of satisfaction of my experience—i.e., that there is something red in front of me—is already there when I introspectively consider phenomenal character. (More precisely, it is the only thing I that I can find, in being introspectively aware of phenomenal character.) For HTG, this fact immediately makes introspectively manifest (at least to someone who has the relevant theoretical tools) that my experience is assessable for accuracy in virtue of its phenomenal character. Thus, since being assessable for accuracy simply means having a certain content, my experience has representational content in virtue of its phenomenal character.

So, here is the general structure of HTG’s argument:

Argument from Introspection Alone

(P1):

[It is introspectively manifest that] experience is assessable for accuracy in virtue of its phenomenal character.¹³

(P2):

Being assessable for accuracy is having representational content.

Therefore,

(C):

Experience has representational content in virtue of its phenomenal character.

The crucial premise here is (P1). So let us look more closely at (P1) and its justification.

According to (P1), TE is what makes it immediately manifest that one’s experience is assessable for accuracy in virtue of its phenomenal character. Roughly, the idea is the following. When one introspects, one deals with phenomenal character. Since introspection reveals that in virtue of which one’s experience is assessable for accuracy (its conditions of satisfaction), and what one introspects is phenomenal character, experience is assessable for accuracy in virtue of its phenomenal character. To put this in the form of an argument:

Justification of (P1)

(#1):

Experience’s conditions of satisfaction are introspectively available. [TE]¹⁴

(#2):

What is introspectively available is phenomenal character. [Assumption]

(#3):

Experience is assessable for accuracy in virtue of its conditions of satisfaction. [Assumption]

Therefore,

(P1):

Experience is assessable for accuracy in virtue of its phenomenal character.

If we now add (P2)—the claim that being assessable for accuracy is having a representational content—we obtain HTG’s conclusion. From now on, I will be working with this more fully unpacked version of the Argument from Introspection Alone.

5.2 The argument assessed

If HTG’s argument works and is really from TE alone, then the latter really leads to the priority of phenomenal character. However, as I am going to argue now, TE alone does not suffice to get the claim that experience is assessable for accuracy, and thus has content, in virtue of its phenomenal character in a sense that makes true PIV and excludes Representationalism. This will take two steps. First, I will show that a further, controversial assumption about introspection is required and, thus, that the argument is not from TE alone (Sect. 5.2.1). Secondly, I will argue that, even if we grant the controversial assumption, the argument is still too weak to lead to the desired conclusion (Sect. 5.2.2).

5.2.1 The argument is not from TE

The first thing to observe is that the argument above needs some further independent assumptions—(P2), (#2), and (#3)—in addition to TE to get to the desired conclusion. However, it is unreasonable to expect an argument that does not assume literally anything more than an introspective datum. A reasonable requirement, then, is the following constraint (K) an argument has to meet in order to be from TE alone:

(K):

Premises that are not justified/justifiable on the basis of TE are allowed as long as they are largely uncontroversial.

By “largely uncontroversial” here I mean something like the following: the premise in question should be largely accepted/able by, and so not an object of disagreement among, at least the people involved in the debate that is at stake here.

Why (K)? In short, because the argument has to be from introspection alone, and in particular from TE, and not from something else. So, we want to avoid the case in which one might agree with TE and still reject the conclusion of the argument. For, in that case, it would be no longer clear that TE is what is doing the real job of leading us to the conclusion of the argument.

We can now apply this constraint to (P2), (#2), and (#3).

(P2) follows from the definition of representational content we are working with, so it is innocuous in this context. (#3) seems innocuous as well, since it is true that we can assess our experience for accuracy because it has conditions of satisfaction. Finally, (#2) is a claim about introspection and the target of our introspective awareness. Prima facie, it looks okay: it seems plainly true that, in introspecting our experience, we are in some way dealing with its phenomenal character; nobody would really deny that. However, the point here is a bit subtler. For there is a real substantive question that can be raised as to what kind of awareness the locution “deal with” stands for: Is it direct awareness or indirect awareness? This is (highly) disputable.

Some, representationalists included, maintain that awareness of phenomenal character is indirect. For example, in several places, Michael Tye claims that we become introspectively (indirectly) aware that experience has phenomenal character because in introspection we are (directly) aware of its representational content (e.g., Dretske 1995; Tye 2000: Ch. 4, 2008; Byrne 2010; 2012). On this view, what is introspectively available is ultimately nothing but representational content. In contrast, (#2) suggests exactly the opposite, namely, that we are directly aware of phenomenal character.

Hence, acceptance of (#2) seems strongly dependent on one’s antecedent beliefs about introspection and the direct target of introspective awareness. We do not have to settle this issue here. Given our purposes, it is sufficient to remark that (#2) is a highly controversial claim, whose acceptance is not obvious but seems to rest on certain specific intuitions/beliefs/theory one has about introspection and introspective awareness. Given (K), this is already enough to conclude that (#2) cannot be assumed if we want the argument to be from TE alone.

5.2.2 If the argument is from TE, then it is too weak

The fact that the argument is not from TE alone is a strong clue that TE is too weak to yield the priority of phenomenal character. However, to get something more than a clue and closer to a proof, let us turn this argument into an argument from TE alone and show that it does not lead to a conclusion as strong as the priority of phenomenal character.

The most charitable way of doing this is to grant that (#2) is uncontroversial, indeed true, and to make sure that (#2) is fully compatible with TE. For in principle, there are (at least) two interpretations of “phenomenal character:”

Strong interpretation:

Phenomenal character as the instantiation of intrinsic properties of the experience that are representationally inert. (This is the sense of “phenomenal character” that the separatists favor.)

Weak interpretation:

Phenomenal character as the instantiation of properties that are ultimately responsible for both the phenomenal and the representational features of experience.

I will use “phenomenal character*” to refer to the strong interpretation and “phenomenal character**” to refer to the weak one—I will keep using “phenomenal character” as a neutral expression.

Since we have to stick to TE, we cannot assume that what we are introspectively aware of is phenomenal character*. So, we have to assume that we introspect phenomenal character**. Since this interpretation must be preserved in the whole argument, here is what we get:

Argument from introspection alone fully unpacked (modified)

(P1′):

Experience’s conditions of satisfaction are introspectively available. [TE]

(P2′):

What is introspectively available is phenomenal character**. [Assumption]

(P3′):

Experience is assessable for accuracy in virtue of its conditions of satisfaction. [Assumption]

(P4′):

Being assessable for accuracy is having representational content. [Assumption]

Therefore,

(C′):

Experience has representational content in virtue of phenomenal character**.

Is (C′) equivalent to PIV? Surely, we can go from PIV to (C′). Indeed, the fact that experience represents a certain content in virtue of instantiating certain phenomenal properties implies that experience has such a content in virtue of certain properties that are responsible for the phenomenal features of experience but also for the representational ones. However, the other direction (from (C′) to PIV) is not guaranteed, since (C′) is compatible with a representationalist account as well. Let me now illustrate how.

At first sight, the representationalist seems to face a major difficulty in explaining how representational properties give rise to the phenomenal features of our experience—whereas given what we have said so far, it seems more intuitive to go from the phenomenal properties to the representational ones. However, such a prima facie difficulty disappears as soon as we consider the following distinction between pure and impure representational properties (Chalmers 2004; Crane 2009; Lycan 2015). A pure representational property is the property of representing a content—for example, representing that the car is red. An impure representational property is the property of representing a content in a certain manner—perceptually, emotionally, or (even more specifically) visually, auditorily, etc. An example would be the property of visually representing that the car is red. According to standard versions of Representationalism (Tye 1995, 2000; Lycan 1996), experiences have phenomenal properties in virtue of instantiating certain impure representational properties, which are responsible for both the representational and the phenomenal features of the experience.

So, we have an alternative explanation of (C′) that uses representational properties, as opposed to phenomenal properties, as primitives. Hence, we should conclude that the argument, construed as relying on TE alone, leads us to a conclusion, (C′), that is simply weaker than PIV.

To recap, firstly, we have seen that HTG’s argument is not from TE (Sect. 5.2.1). This instilled in us the suspicion that TE is too weak to speak in favor of PIV. So, we have reconstructed the argument so it flows from TE. The upshot was that, from introspection alone (TE), we could only conclude that experience has representational content in virtue of its phenomenal character**, which does not exclude Representationalism (Sect. 5.2.2).

6 The argument from phenomenal duplicates

One might accept that TE is too weak to support PIV by itself, and still think that the real problem is not TE’s alleged weakness, but the additional set of premises we accompany TE with. So, perhaps if we engineer an argument that combines it with the right non-introspective assumptions, TE really supports PIV. Horgan and Tienson (2002) seem to offer such an argument: it still relies on TE, but makes use of a different argumentative machinery. More precisely, it appeals to phenomenal duplicates. Two creatures are phenomenal duplicates “just in case each creature’s total experience, throughout its existence, is phenomenally exactly similar to the other’s” (Horgan and Tienson 2002, p. 524). Roughly, the idea is this: If entirely phenomenal creatures have intentional experiences, then we can conclude that the instantiation of phenomenal properties determines at least a certain kind of representational content—the one shared by me and every phenomenal duplicate of mine. The appeal to phenomenal duplicates, thus, is an attempt of “filtering out” every non-phenomenal feature of experience to see whether we manage to get intentionality out of that.¹⁵

Intuitively put, the argument goes as follows. Consider, once again, my experience as of a red car: this experience has a red car-ish phenomenal character. If so, then the duplicates will have an experience with a red car-ish phenomenal character as well, since their experiences are phenomenally indistinguishable from mine. Now, as we know from TE, that my experience has a red car-ish phenomenal character means nothing but that certain properties, like redness, a certain shape, etc., appear to be properties of (apparently) mind-independent object. Since my duplicates and I share the same phenomenal character, this is to be true of their experiences too. So, precisely like me, in undergoing an experience with a red car-ish phenomenal character, the duplicates will be having experiences that are as of an external world, as of objects, as of properties of objects, etc. But, then, the duplicates have experiences that not only have the same phenomenal character, but also the same representational content as mine. If so, it seems that the instantiation of certain phenomenal properties alone determines experience’s having a certain representational content—as PIV claims.

In other words, the idea is that the following claim about phenomenal duplicates’ mental lives is sufficient for PIV:

(PD):

There is a kind of representational content shared by any two phenomenal duplicates.

Here is the Argument from Phenomenal Duplicates, then:

Argument from Phenomenal Duplicates

(P1):

There is a kind of representational content shared by any two phenomenal duplicates. [TE plus the conceivability of the phenomenal duplicates]

(P2):

If there is a kind of representational content that is shared by any two phenomenal duplicates, then experience’s instantiating certain phenomenal properties determines a certain kind of representational content. [Assumption]

(C):

Experience’s instantiating certain phenomenal properties determines a certain kind of representational content. [From (P1) and (P2)]

Notice that, unlike the previous one, this new argument does not claim that PIV is simply phenomenally manifest given our introspective intuitions (TE). The idea here is rather that, in order to conclude that PIV is true, we have to play a bit with what we gather from introspection and perform the phenomenal duplicates thought experiment.

So, the Argument from Phenomenal Duplicate does not seem to aim to be from introspection alone. Still, it is surely an argument that strongly appeals to and heavily relies on introspection and TE, in the following sense. First, the material we need to perform the phenomenal duplicates thought experiment comes from introspection—and this material is nothing but TE itself. Second, TE is what secures the thought experiment’s outcome, i.e., that there is a kind of representational content shared by me and my phenomenal duplicates. It is thus crucial for the argument. So, should this new strategy be successful, we would still have serious reasons to claim that TE speaks in favor of PIV.

As just noted, (P1) is justified by TE in combination with further assumptions concerning the conceivability of the phenomenal duplicates that are not problematic in the present context, because a representationalist would agree that phenomenal duplicates are conceivable. Thus, (P1) is fine. The problems are with (P2): I will now argue that (PD) is not sufficient for PIV.

(PD) is equivalent to the following claim:

(PD′):

Necessarily, for any two pair of experiences E and E’ of any two phenomenal duplicates, if E and E’ instantiate the same phenomenal property P, then there is a representational property R that they both instantiate.

(PD′) states that a modal correlation obtains between the instantiation of phenomenal properties and the instantiation of representational properties: if (PD) is true, then (at least) certain representational properties supervene on phenomenal properties. Yet, supervenience (even in its strongest form) is not sufficient for priority: grounding or even reduction are much stronger relations. As McLaughlin and Bennett (2014) correctly note:

Supervenience is not a relation of ontological priority; the supervenience of A on B does not guarantee that B-properties are ontologically prior to A-properties. One way to see this is to note that ontological priority is irreflexive and asymmetrical: nothing can be ontologically prior to itself, or be ontologically prior to something that is also ontologically prior to it […]. But as we have seen, supervenience is reflexive and not asymmetrical. (McLaughlin and Bennett 2014: §3.5)

If so, then one could well accept (PD) without thereby being forced to accept PIV. As it happens, this is precisely what representationalists do: they agree with (PD) but not with PIV. On the contrary, they offer exactly the opposite explanation: in figuring out the phenomenal duplicates scenario, we get a modal correlation between phenomenal properties and (impure) representational properties because the latter are prior to the former. This explanation is perfectly compatible with (PD).

To sum up, the main problem with the Argument from Phenomenal Duplicates is that the phenomenal duplicates thought experiment is not sufficient to establish PIV’s truth. In other words, (P2) is false. Indeed, if (PD) is true, then phenomenal properties necessarily bring along certain representational properties. Yet, this does not mean that they ground those properties: it may just be that phenomenal properties are identical with, or even grounded in, those representational properties.¹⁶

We can appreciate the point in another way. Recall that according to TE, we cannot be aware of phenomenal character without also being aware of representational content. If we really want to stick to this introspective datum, we have to admit that what we are calling here “phenomenal duplicates” are in fact phenomenal intentional duplicates—i.e., creatures with experiences that instantiate properties which are ultimately responsible for both the phenomenal and the representational features of those experiences. Now, such duplicates simply constitute too weak a case for PIV, since we are not guaranteed that they have experiences that represent in virtue of their purely phenomenal nature: it might well be that their experiences represent in virtue of instantiating certain (impure) representational properties that also ground the phenomenal features.¹⁷

To exploit again the phenomenal character*/phenomenal character** opposition: it seems that to prove that PIV is true, one should find a way to show how we get representational content from phenomenal character*. Yet, TE is simply too weak to do that¹⁸: it does allow us to see that there is a strict relation between phenomenal properties and representational properties, but it does not allow us to isolate phenomenal properties—and this is required, if we want to show that they ground representational content.

Here is the upshot, then. The Argument from Phenomenal Duplicates combines TE with some modal assumptions concerning the conceivability of the duplicates and a too strong, and ultimately false, claim: (P2). So, the phenomenal duplicates thought experiment does not help us extract PIV out of TE—it is not really a good solution to our problems. The problem is not just one of engineering or finding the right combination of premises to combine TE with, but concerns TE itself.

7 An objection: the priority is epistemic

Let me consider the following objection:

Alright! Maybe there is no way to determine a metaphysical priority of phenomenal character on the basis of TE alone. However, when we advance introspective considerations, we offer phenomenological descriptions of our experience, in the first place. And it seems that, on the basis of those phenomenological descriptions of our experience, we are in fact able to assign conditions of satisfaction, and thereby intentionality, to it. So perhaps, on the basis of TE alone, we can establish at least an epistemic priority of the phenomenal concepts we make use of in our phenomenological descriptions, since they seem to be enough to capture the intentional features of our experience.

Very quickly, here is my reply. It is true that in the introspective reports under consideration we make use of phenomenal concepts. Yet, we use representational notions too, such as “conditions of satisfaction,” “assessable for accuracy,” etc., and these notions seem to play a crucial role in our description of experience as intentional. So, as far as our introspective descriptions are concerned, since the descriptions of phenomenal character necessarily involve the use of representational concepts, one might well say (1) that it is precisely on the basis of those representational notions that we are able to capture the representational features of our experience and (2) that phenomenal character can be fully captured in intentional terms.

In order to show the priority of phenomenal concepts, the phenomenal intentionalist should offer a re-description of the intentionality of the experience in terms of phenomenal concepts alone, or show how the representational notions can be analyzed in terms of phenomenal concepts. But they do neither of these things—and it is hard to see how they could, relying on TE alone.

8 Taking stock

Let us now take stock. After describing the background and clarifying the terminology (Sect. 2), I introduced TE and showed that it is the introspective intuition to which both representationalists and (at least some) phenomenal intentionalists appeal to (Sect. 3). This secures my first claim:

(1):

Representationalists and phenomenal intentionalists appeal to the same introspective datum, TE.

TE minimally suggests a supervenience claim, which seems already enough to rule out Separatism—in this paper, I did not call this into question.

In Sect. 4, I distinguished two further possible strong interpretations compatible with the minimal one, (SI₁) and (SI₂), that are mutually exclusive:

(SI₁):

Phenomenal character is ultimately grounded in the representational properties of the experience.

(SI₂):

Phenomenal properties ultimately ground the kind of content we are experientially and introspectively aware of.

(SI₁) and (SI₂) are nothing but (respectively) Representationalism and PIV. Given the standard association of Representationalism and TE, the question immediately became whether TE speaks in favor of PIV.

To answer this question, I discussed the two arguments proposed by Horgan and collaborators that flesh out the line of argument from TE to PIV. Here are the results.

As regards the Argument from Introspection Alone, in the best case—i.e., the case in which (#2) is true and the argument is genuinely from TE alone—, we get the conclusion that experience has representational content in virtue of phenomenal character**. This is compatible with Representationalism and thus too weak to speak in favor of PIV.

The worst cases, the ones that we set aside or were not even considered in our discussion for sake of charity, are instead the following three:

• 1. If (#2) is just controversial, then the argument is not from TE.

• 2. If (#2) is false, then the argument is to be rejected.

• 3. The argument assumes the strong interpretation of “phenomenal character”: it is incompatible with TE and thus to be rejected.

On the other hand, the Argument from Phenomenal Duplicates claims that the phenomenal duplicates thought experiment is enough to establish PIV’s truth. But, as we have seen, this is simply false: the duplicate scenario is not sufficient for PIV’s truth. So, the argument is to be rejected.

This leads us to the second claim I am defending here, that is:

(2):

Arguing from TE to PIV leads us to the following trilemma: Either (a) the argument is too weak or (b) it is not from TE or (c) it is to be rejected—where (a) is the best case and (b) and (c) the worst cases.

9 Conclusion

What is the upshot, then? The discussion has highlighted that TE has two faces: it suggests a strong connection between phenomenal character and representational content; yet, precisely because of that, it does not allow us to introspectively consider phenomenal character as separated from representational content. And this seems to be somewhat in tension with the phenomenal intentionalist’s desiderata and needs. For, to establish the priority of phenomenal character, we seem to need a clear way to go from phenomenal properties to representational properties.

The conclusion I suggest we should draw is that the line of argument for PIV that relies on our introspective intuitions (TE) in such a strong way is not really a good strategy for the phenomenal intentionalist: at best, it does not get her where she would like; at worst, it might generate serious trouble.

Now, one might be tempted to see this as a somewhat indirect argument against PIV and in favor of Representationalism. The thought might be: Since TE does not speak in favor of PIV, it is in favor of Representationalism. Let me block this immediately. This is not what my conclusion suggests. First of all, that TE is too weak to speak in favor of PIV against Representationalism does not mean that there is no good argument for PIV at all: it just means that arguing from TE is not a good strategy to support PIV. Secondly, not speaking in favor of X is not one and the same as speaking against X—and here I’ve only been concerned with the former (as it concerns TE speaking in favor of PIV). Given my conclusion, the two strong interpretations, (SI₁) and (SI₂), are still there as legitimate options. In order to prove that PIV is not really viable given TE, one would have to show that the latter speaks in favor of Representationalism. After all, it might well be that TE does not speak in favor of either.

This surely opens further questions concerning TE’s strength and its capacity for supporting any claim about priority and, more generally, about the existence of a priority relation between phenomenal character and intentionality. These further questions cannot be answered here, but are material for another paper.