Pitt (2017) has recently argued that reductive representationalism (RR) is committed to an absurd thesis about the nature of hallucination. The absurdity is akin to the ‘paramechanical hypothesis’ Ryle (1949) attributed to Descartes, according to which mental processes are just like physical processes, absent the matter.Footnote 1 By contrast, Pitt sees the RR’ist as being committed to a paraphenomenal hypothesis (‘PH’), on which hallucinatory experiences are just like veridical experiences, except that, in a hallucinatory experience, there is nothing that (phenomenally) appears, or seems, any way to us. Hallucinating green grass is the same thing as seeing green grass—just without the grass.

We have two aims, one smaller and one larger. Our smaller aim is polemical: we will contend that Pitt is mistaken. One of the more well-known forms of RR—the property-complex theory (Dretske 1995; Bengson et al. 2011; Tye 2014a, b, 2015)—is not committed to the PH. Our larger aim is clarificatory. Pitt’s argument, though flawed, is symptomatic of fairly common misconceptions of the aims and structure of the property-complex theory. A remedy is thus overdue. We begin with a brief description of RR and the property-complex theory, before turning to Pitt’s argument.

1 The property-complex theory

Pitt describes RR as follows:

These theorists hold that the phenomenology of perception...can be reduced to a kind of non-phenomenal intentionality, which in turn can be explained in naturalistic causal–informational–teleological terms. The qualitative features associated with an experience are properties, not of the experience, but of the worldly (or bodily) things it represents. The blue that characterizes what it’s like to see a clear sky at noon, for example, is a property, not of one’s experience of the sky, but of the sky. Its relevance to the characterization of the experience of a clear sky at noon is due to the fact that one’s experience represents it, not that one’s experience instantiates it. These views are externalist about qualia. (2017, pp. 735–736)

RR comes in many forms, but as an illustrative case we will focus on the property-complex theory, or PC theory for short. This view consists of two core claims. First: phenomenal character is identical to a property-complex, which may or may not be instantiated in the subjects environment (e.g. Dretske 1995, p. 73; Tye 2014a, p. 54, fn. 13; 56). Second: to have an experience—whether veridical (henceforth, VE) or hallucinatory (henceforth, HE)—with a property-complex as its phenomenal character, the subject must have a mental state of the right sort that represents this property-complex, which thereby makes the subject aware of the property-complex.Footnote 2 So experiences ‘have’ phenomenal characters in the same way a predicate has a semantic value (e.g. a property); not by instantiating it, but by representing it (Tye 2014b, p. 85).Footnote 3\(^{,}\)Footnote 4

Suppose a subject S has a HE as of a blue ball, and later a VE as of a blue ball.Footnote 5 HE and VE, we’ll assume, are phenomenally identical. The PC theory says that the HE and VE will represent the same property-complex, e.g.:

$$\begin{aligned} {B}: {\lambda x(x\, \hbox {is blue} \wedge x\, \hbox {is round})} \end{aligned}$$

The represented property-complexes are structured out of simpler properties (Tye 2014c, p. 306). In the present case, B—the property of being an x such that x is blue and x is round—has being blue and being round as parts. All that distinguishes S’s VE and HE is whether the property-complex is instantiated. In the HE, the property-complex is wholly uninstantiated.

2 Where to put the property?

The route from the PC theory to the PH is predicated on three background theses. The first two are general, applying to all experiences on the PC theory. The third point concerns hallucination.

First, Externalism: the PC theory is an externalist variant of representationalism. The PC theorist drives phenomenal character—what your experience is like—out into the world (Tye 2015). You can become aware that you are having an experience as of a blue ball, but you are not aware of your experience. You are aware of B, and if the experience is veridical, the ball. Nothing else. The qualitative features associated with your experience, then, are properties of (if anything) the worldly (or bodily) things it represents, not of the experience itself. To take a different example, the greenness that characterizes what its like to look at healthy grass is a property, not of the experience of the grass, but of the grass itself.

Second, Common Kind: HE’s and VE’s are fundamentally the same kind of mental event. Compare the disjunctivist: VE’s are fundamentally relations of acquaintance with mind-independent objects and properties, while HE’s are something else altogether (Martin 2004).Footnote 6 The PC theory is, at least by design, resolutely anti-disjunctivist.

Third, No Instantiation: when S has a HE as of a blue ball, the only thing the PC theorist has at her disposal to explain the phenomenal facts—e.g. the fact that it (phenomenally) seems to S as if there is something that is \(\varphi \)—is that S is harboring a perceptual representation of the right sort that represents B, where B is an uninstantiated property-complex. What makes S’s HE phenomenally indistinguishable from her VE of a blue ball is that they both represent the very same property-complex B. Indeed, they are indistinguishable precisely because they are phenomenally identical.

With these points in place, we can reconstruct the problem as follows. Let’s start with two relatively innocuous claims:

  1. P1

    In a HE, it (phenomenally) seems to S as if there is some x that is \(\varphi \).Footnote 7

  2. P2

    In a HE, there is no x that seems \(\varphi \) to S.Footnote 8

On P1: when S has an HE as of a blue ball, it seems to S as if there is something before her that is blue and round. On P2: when S hallucinates, S is aware of B. But as Pitt (Ibid: 737) notes, being uninstantiated, B doesn’t itself look or seem or appear any way to S.Footnote 9 And S’s hallucinatory experience doesn’t itself seem \(\varphi \) either, since S isn’t aware of her experience. She isn’t aware of anything but B. This is what No Instantiation comes to.

So putting P1 and P2 together, what we know is that, at least in a hallucination, it seeming to S as if there is something that is \(\varphi \) won’t hold in virtue of there being an x that seems \(\varphi \) to S. As Pitt points out, in an HE on the PC theory, there isn’t anything out there that is the bearer of the properties were aware of in experience (2017, p. 736). Yet according to Pitt, this actually unearths a deep worry in light of the PC theorist’s commitment to Externalism and Common Kind. Consider the veridical counterpart to P1:

  1. P1*

    In a VE, it (phenomenally) seems to S as if there is something that is \(\varphi \).

What seems blue and round to S when she has her VE is the ball, and by Externalism the phenomenal character of S’s VE just is a feature of the ball. It’s tempting then to think that in P1*, it seeming to S as if there is something that is \(\varphi \) will hold in virtue of there being an x that seems \(\varphi \) to S. And given Common Kind, we should say the same thing about HE’s. Generalizing, we thus get:

  1. P3

    If the PC theory is true, then in any experience E—including hallucinations—it (phenomenally) seems to S as if there is something that is \(\varphi \) in virtue of there being an x such that x seems \(\varphi \) to S (where x is not E).

And then if you conjoin P3 with P1, it follows that:

\(\therefore \):

If the PC theory is true, then in a HE, there is an x such that x (phenomenally) seems \(\varphi \) to S.

Of course, we can now see where this is heading. For when coupled with P2, it also follows that:

\(\therefore \):

PC theory is false.

Pitt does not frame his discussion in precisely these terms. Nonetheless, the argument bears on the point at hand insofar as it makes clear why Pitt thinks the PC theorist is committed to the PH.

To see why, consider P3. This premise should strike the reader as being somewhat curious, reminiscent as it is of Robinson’s (1994, p. 32) phenomenal principle, which is often itself understood as anathema to representationalism given its association with sense-datum theories (e.g. Fish 2010, p. 79). So there is no doubt that the PC theorist will want to reject P3. However, we take it that Pitt’s point is that without P3, the PH follows:

...[In HE’s] theres no place to put the property youd mention in describing what your experience [e.g. of a blue sky] is like. It can’t be the same thing, only without the sky, since the sky was where the qualitative feature you experienced was supposed to be located. (2017, p. 737)

As emphasized, the PC theory seeks to respect Common Kind. Yet if Pitt is right, then without P3, in HE’s there is now nothing external that instantiates the property-complex B, and so nothing that seems or looks some way to S.Footnote 10 As a result, HE’s are about as much like VE’s as non-material mental processes are like physical processes on the paramechanical hypothesis. In other words, they are not alike at all.

And the PH isn’t a problem merely because of Common Kind. Suppose the PC theorist said that the property-complexes were instantiated by the experiences themselves—experiences which, naturally enough, would be present in both HE’s and VE’s. Common Kind would be upheld, but at the expense of Externalism. Thus, the PH seems to cut to the dual ambitions that motivate the PC theory, i.e. reducing phenomenal properties to the properties of external objects while also acknowledging that VE’s and HE’s are fundamentally the same kind of mental event. As Pitt puts it, if the things that have the properties that appear to us are removed as is the case in a HE then the basis for a reductive account of the phenomenality of experience goes with them (2017, pp. 737–738).

3 Non-referential looks

Whether or not the PH is itself absurd, we submit that it is not absurd to think that the PC theorist might be committed to the PH. After all, if the aim of the PC theory is to drive phenomenal character out of the head and into the world, it’s not clear how this could apply in hallucinatory cases. When we hallucinate, there is a head (or just a brain), but no world—at least not in the sense that there is some object outside us that appears to us some way. Thus it is no surprise that Pitt’s worry has some precedent (Thompson 2008), with some going so far as to argue that what PH reveals is that, Common Kind notwithstanding, PC theory is really just a form of disjunctivism (Gow 2018).Footnote 11 Of course, disjunctivism is not without its own problems, but it is certainly not absurd.

A move towards disjunctivism, however, is premature. The PC theorist can reject P3 without committing to PH.Footnote 12 Articulating exactly how is critical to dispelling misunderstandings of the PC theory.

It will help to separate the consequent of P3 into two claims. Recall, P3 said this: if the PC theory is true, then in any experience E, it (phenomenally) seems to S as if there is something that is \(\varphi \) in virtue of there being an x such that x seems \(\varphi \) to S (where x is not E). So, focusing on the consequent, we first have:

  1. (P3a)

    It (phenomenally) seems to S as if there is something that is \(\varphi \).

And then:

  1. (P3b)

    There is some x such that x (phenomenally) seems \(\varphi \) to S.

The consequent of P3 holds that P3a is grounded in P3b. From this it follows that P3a cannot obtain absent P3b’s obtaining. But, according to the PC theory, this is false. Suppose S has two phenomenally matching experiences as of a blue ball, one of which is a VE while the other is a HE. In both cases, P3a will hold. In S’s VE both P3a and P3b will hold. Yet although it is true that P3a obtains in the case of VE, it will hold not because P3b holds, but simply because S is aware of the properties of the blue ball, i.e. B. Here B is instantiated, but its being instantiated is irrelevant to P3a obtaining, or to how things seem to S more generally.

Similarly, in a matching HE, it seems to S as if there is a blue ball because S is aware of the properties (i.e. B) a blue ball would instantiate, not because there exists something that seems blue. The fact that no particular instantiates B, and the fact that P3b does not hold, does not undermine P3a. So, Pitt’s point that the basis for a reductive account of the phenomenality of experience (2017, pp. 737–738) is eliminated whenever we eliminate the object (as in a HE), betrays a misunderstanding. For on the PC theory, objects never mattered to the phenomenal character of experience to begin with, irrespective of whether the experience is veridical or hallucinatory.

Put another way, the PC theorist would reject the PH because the veridicality of a VE—and thus that (e.g.) B is instantiated and that there is something that looks blue and round—is irrelevant to the VE’s phenomenal character. And that holds quite generally; phenomenal characters just are property-complexes. Note that while this implies that external objects are otiose insofar as phenomenology goes, it does not imply that the PC theorist has abandoned Externalism. For all we have said here, all of the represented property-complexes can be mind-independent. And that is good enough for Externalism. When it is said that phenomenal character is driven out into the world, what is meant is that (e.g.) greenness—the property that characterizes what it is like to have (e.g.) experiences of healthy grass—if instantiated at all, it is instantiated by the grass.

So the way to understand Common Kind on the PC theory is like this: the experience involved in veridical perception and hallucination are both fundamentally representations of property-complexes. The fundamental kind of mental event that occurs when S sees a blue, round ball as blue and round—subjectively characterized by it seeming to S as if there is something that is \(\varphi \)—is precisely what is present when S has a phenomenally identical hallucination as of a blue, round ball: a perceptual representation of B.

Thus, pace the PH, HE’s are not just like VE’s absent there being something that (phenomenally) appears, or seems, any way to us. No: what’s true is that HE’s are just like VE’s because both cases are fundamentally a matter of it seeming to S as if there is something that is \(\varphi \). In this way, P3a is analogous to:

(P3a\('\)):

It’s raining.

In (P3a\('\)), ‘it’ is non-referential. It does not refer to some x such that x is raining; it does not commit us to saying that \(\exists \)x(Rx). Thus P3a is not like:

(P3b\('\)):

It’s blue.

In (P3b\('\)), there is some x such that x is blue. In this way, P3b tracks P3b. But, again, P3b doesn’t pick out anything fundamental about having a visual experience as of a blue ball; perceptual experiences, irrespective of their accuracy, are not fundamentally cases of awareness external material objects, or fundamentally a matter of objects themselves seeming some way. When nothing looks blue, we still have experiences as of something blue. And when something does look blue, the thing itself is irrelevant to the phenomenology of your experience as of something blue.

A final point. A corollary of Common Kind is that, for any VE, there is a subjectively indistinguishable HE. Pitt protests that the PC theorists commitment to No Instantiation violates this corollary, remarking that “instantiated pink and uninstantiated pink are experientially distinguishable (2017, p. 738). Now Pitt does not tell us why this is so, but perhaps the idea is something like this. In undergoing most if not all experiences—whether veridical or not—it will seem, phenomenologically speaking, as if we are aware of particulars, including both objects and property-instances. Following Schellenberg (2010, p. 23), call this phenomenal particularity. This strikes us as a datum.

Suppose then that S undergoes a VE with phenomenal particularity; it is part of the phenomenal character of this VE that, when S undergoes it, it will seem to S as if she is aware of a particular. Given the aforementioned corollary of Common Kind, it follows that there is an HE with phenomenal particularity. But now we have a potential source of Pitt’s worry; for how could a HE have phenomenal particularity when (by No Instantiation) it does not represent a particular at all? The PC theorist, recall, explains the possibility of subjectively indistinguishable HE’s and VE’s by their representing one and the very same property-complex; identity of represented property-complex entails identity of phenomenal character (property-complexes just are phenomenal characters), and sameness of phenomenal character entails subjective indistinguishability. Yet as we have seen, this also means that instantiation has no phenomenal import. The objection from phenomenal particularity can be seen as casting doubt on this claim.Footnote 13

This is a complicated matter, but the PC theorist can supply an answer by appealing to Johnston’s (2004) mimicry strategy. The idea here is to posit that all represented property-complexes include spatiotemporal universals. So now, instead of B, S’s experience represents:

$$\begin{aligned} {B*}: {\lambda \hbox {x(x is blue} \wedge \hbox {x is round} \wedge \hbox {x is at location L} \wedge \hbox {x is at time}\, {t})} \end{aligned}$$

According to Johnston, the spatiotemporal relational element within the represented property-complex “mimics spatial and temporal extent and thereby mimics particularity” (Ibid, p. 142). So even though the property-complex represented by S’s HE (i.e. B*) is constituted of locally uninstantiated universals, when S undergoes this HE, in virtue of B* having detailed spatiotemporal properties as its constituents, it will still seem to S as if she is aware of a particular. If that’s right, then phenomenal particularity itself would give us no reason to think instantiation has phenomenal import.Footnote 14

4 Conclusion

In sum, we conclude that Pitt’s argument, as attractive as it may initially seem, betrays a misconception of the PC theory, and also the resources available to the theory. Consequently, the argument fails. Externalist RR—at least the variant under discussion here—if absurd, is not absurd on account of the PH. Of course, there remains the question of whether the PC theory can really account for the phenomenal facts by appealing to property-complexes alone, and whether we really can be aware of uninstantiated property-complexes (cf. Pautz 2007). However, these questions are further questions, orthogonal to the present question of what the PC theory actually says, and whether what it says entails the PH, or must collapse into a form of disjunctivism. On these matters at least, Pitt—and others—are wrong.