Necessarily Veridical Hallucinations: A New Problem for the Uninstantiated Property View

Abstract

Philosophers of perception have a notoriously difficult time trying to account for hallucinatory experiences. One surprisingly quite popular move, and one that cross-cuts the representationalism/relationalism divide, is to say that hallucinations involve an awareness of uninstantiated properties. In this paper, I provide a new argument against this view. Not only are its proponents forced to classify many hallucinations as veridical, such experiences turn out to be necessarily veridical. In addition, I show that representationalists who endorse the uninstantiated property view must reject the common fundamental kind claim and adopt disjunctivism, and naïve realists/relationalists must radically modify their disjunctivism: The distinction between ‘veridical’ and ‘hallucinatory’ will no longer track a metaphysical distinction between the relevant experiences.

I. INTRODUCTION

Hallucinations, like veridical perceptual experiences, seem to involve an awareness of objects and properties. Although there are views according to which hallucinations really do involve an awareness of objects—sense-datum theories (Price 1932; Jackson 1977; Robinson 1994; Sollberger 2015), theories that permit non-existent objects (Meinong 1904; Parsons 1980), and theories that hold that hallucinations involve a relation to ordinary objects (Raleigh 2014; Ali 2018; Byrne and Manzotti 2022)—most theorists deny that any sort of object awareness is involved. However, the idea that hallucinations involve an awareness of properties, uninstantiated properties to be precise, is much more popular.

Interestingly, this view finds support within both the representationalist and the naïve realist (or ‘relationalist’) camps. For example, it is endorsed by George Bealer (1982), John Bengson et al. (2011), Matthew Conduct (2012), Fred Dretske (2000), Peter Forrest (2005), Mark Johnston (2004), Colin McGinn (1999), Adam Pautz (2007), David Sosa (2007), and Michael Tye (2014). Tye says:

[Y]ou cannot attend to what is not there. But on my view there is an un-instantiated quality there in the bad cases…an un-instantiated quality is present in hallucination. (Tye 2014: 51)

The view might seem strange initially, but there are many considerations in its favour. First, the uninstantiated property view is extremely well placed to provide a simple explanation of the phenomenal characters of hallucinations. On this view, the reason that our hallucination is of a green apple rather than a red one (say) is that we’re aware of uninstantiated green not uninstantiated red. This also explains how we can know what it's like to experience green in virtue of having a hallucination of green even if we have never veridically experienced this colour. Mark Johnston, Michael Tye, and Adam Pautz have pointed out that if Frank Jackson's Mary hallucinated red, she could come to know what it's like to experience red without ever leaving her black and white room. Just as hallucinations can provide us with knowledge about what it's like to experience certain properties, they also seem able to provide us with new facts about the mind-independent world. For example, we can learn that red is more similar to orange than it is to green (say) purely through having hallucinations of red, orange, and green (Johnston 2004; Pautz 2007; Tye 2014).

Although it has many explanatory virtues and is a well-supported view of hallucination, the uninstantiated property account faces some obvious challenges. The main objection concerns the metaphysical commitments the uninstantiated property view is forced to accept. Uninstantiated properties are abstract universals and are outside the spatiotemporal realm. [See Gow (2018) for an explanation of why proponents of this view are committed to a Platonic ‘otherwordly’ conception of universals rather than the more moderate Aristotelian view.¹] Both Uriah Kriegel and David Papineau have objected to the idea of making our occurrent conscious experiences metaphysically dependent on abstract entities (Kriegel 2011; Papineau 2014, 2021). And it is worth adding that physicalists would do well to stay away from a commitment to non-physical entities playing an essential role in our experiences (Schellenberg 2011a). Although there are some ways of defining physicalism that make room for abstracta, it is probably fair to say that a commitment to non-physical entities playing an essential (and, of course, non-causal) role in our mental lives sits uncomfortably alongside a physicalist world view.

Although I think that the metaphysical commitments of the uninstantiated property view are a concern, it is something about which its proponents have been happy to bite the bullet. The problem I will focus on is a new and I think much stronger objection, and bullet-biting simply isn’t an option. My aim is to show that the uninstantiated property view is not fit for purpose—not only is it forced to classify many hallucinations as veridical, such experiences end up being necessarily veridical.

II. THE PROBLEM: VERIDICAL HALLUCINATIONS

It seems to be fundamental to our concept of perceptual experience that some perceptual experiences are veridical and some are not. Representationalists and naïve realists/relationalists unpack the notion of veridicality in different ways. For a representationalist, an experience is veridical if the content of the experience is accurate or true. Hallucinations and illusions are misperceptions—the error involved arises at the level of experience, through the content of that experience being inaccurate or false. Naïve realists usually deny that perceptual experiences are representational, which means that, strictly speaking, error does not arise at the level of experience at all.² Only the beliefs we make as a result of our experiences can be assessed for error. Non-veridical experiences are those that lead to false beliefs, and so the error involved arises at the level of belief. Notwithstanding the differences between representationalism and naïve realism, proponents of both views agree that we do, at least sometimes, enjoy veridical perceptual experiences, and can agree that, in some sense, these experiences get things right. Similarly, proponents of both accounts will want to claim that some of our experiences are inaccurate, non-veridical, or misleading.³

In this section, I’ll describe a kind of hallucinatory experience that turns out to be veridical if we endorse the uninstantiated property account. However, before I describe these cases, it will be useful to begin with an example of a hallucinatory experience that proponents of the uninstantiated property account are able to classify as non-veridical. Identifying the feature of this experience that allows the uninstantiated property theorist to classify it as non-veridical will help to explain why they are unable to classify other kinds of hallucinations as non-veridical—specifically, when this feature is missing.

II.1 Unproblematic hallucinations

According to the uninstantiated property theorist, having a hallucination of a green apple involves standing in an awareness relation to shape and colour properties. We seem to be related to the property green (for example), and for the uninstantiated property theorist, we really are related to the property green. However, the representational content of our hallucinatory experience of the green apple includes representing the property green as being instantiated by an object in our local environment. Our hallucination seems to present us with an actual green apple and therefore seems to involve object awareness as well as property awareness. The green is represented as being instantiated and located in publicly accessible space. Consequently, the content of our experience will be inaccurate or non-veridical since the properties are not instantiated relevantly by anything in our local environment—they aren’t present in the local, publicly accessible space. The representationalist can therefore account for what has gone wrong in this particular hallucinatory case: The representational content of our perceptual experience ‘says’ that there is a green apple in front of us when there is no such thing. Likewise, the naïve realist can account for the error involved in this example, since the belief we have as a result of our experience includes a commitment to the idea that such properties are present through being instantiated by an object in our local environment, and this belief is false.

II.2 The problem cases

There are many different problem cases, but they can all be categorised together, and helpfully defined as property hallucinations. Property hallucinations are experiences during which we seem to be aware of properties that don’t appear to be instantiated by objects and don’t genuinely seem to be in our local environment at all. After-images provide good examples, and so property hallucinations should be familiar to us all. If we stare for a while at a bright red image and then look at a white wall, we will experience the colour property green. However, it does not seem to us that the wall instantiates green. It doesn’t look as if the property green is instantiated by anything; on the contrary, the green seems to be uninstantiated and free-floating.

Before I explain precisely why property hallucinations present a problem for the uninstantiated property theorist, let me explain why I think we should characterise after-image experience in terms of properties that seem to be uninstantiated. I would hope that this seems intuitive upon reflecting on one's own experience; however, it requires a defence since it is at variance with an alternative view—that after-images seem to be instantiated. There are three quite different versions of this view. I’ll first consider the claim that after-images seem to be instantiated by objects in the local environment, which is defended by Tye (2000) and Ian Phillips (2013). I’ll then discuss Umrao Sethi's (2021) view according to which after-images are property instances, which depend for their existence on their being perceived. They’re not instantiated in the sense of inhering in an object, but Sethi proposes an additional way of understanding instantiation; these property instances count as being instantiated due to their being genuinely there in front of one, at a particular spatiotemporal location.⁴ Although this is a theory about the metaphysics of these experiences, Sethi also claims that after-images seem to be property instances that are genuinely spatiotemporally located. The third view is that after-images involve properties that seem to be instantiated by entities; not (as on Tye's and Phillip's view) by mind-independent entities, but by mind-dependent sense-data.⁵ Let's begin with Tye's view:

[C]onsider first the case in which one sees a red afterimage in front of a much larger background yellow surface without realizing that it is an afterimage. Here one undergoes an illusory experience as of something red and filmy hovering in space in front of something yellow—an experience similar perhaps to that of viewing (in dim lighting) a blood stain on a transparent sheet of glass suspended between oneself and a yellow back- ground surface. Now suppose that one realizes that one is having an afterimage. One is no longer inclined to believe that there is something red suspended in space before one. Nonetheless, at the nonconceptual level, one still undergoes an experience as of something red in front of a yellow background. At this level, one's experience is still phenomenally similar to the veridical experience of the blood stain. (Tye 2000: 85)

Tye claims that as far as the phenomenology is concerned (this is what ‘the nonconceptual level’ refers to) after-images seem to be instantiated. Now, I accept that, for a moment, one might experience the colour property as being instantiated by an object (although I think that the colour would then seem to be instantiated by the wall or background surface rather than by a transparent object between oneself and the wall). However, as soon as one moves one's gaze and the colour property moves with it, one will no longer experience the colour property as being instantiated by anything. It just isn’t possible for ordinary after-images to seem to be instantiated by objects for any length of time. This is simply because they don’t behave like instantiated properties. They move with the direction of our gaze, their vividness diminishes over a relatively short period of time, they are not occludable, and they do not exhibit size constancy (Siegel 2006; Masrour 2013).⁶ What is more, our own behaviour suggests that after-images don’t seem to be instantiated or genuinely located in the local, publicly accessible environment. We don’t expect other people to experience our after-images, and we should if after-images seemed to us to be instantiated in public space. Ian Phillips has argued for a view that is similar to Tye's; he claims that after-images are ‘apparent presentations of public objects’, namely illusory light phenomena (Phillips 2013: 2).

It is important to understand the motivation behind Tye's and Phillips’ claims that after-images seem to be instantiated. Some philosophers have appealed to after-images with a view to denying that perceptual experience is ‘transparent’. They argue that perceptual phenomenology also includes features that aren’t perceptually experienced as being external to us, and say after-image experiences are prime examples of perceptual experiences presenting us with features (or ‘qualia’) that seem to be internally realised (Boghossian and Velleman 1997; Block 2003; Kind 2008; Farkas 2013).⁷ Tye and Phillips are both advocates of transparency and want to deny that after-image experience provides an example of our being presented with properties that seem to be internal.⁸ This is why Tye says that, as far as the phenomenology is concerned, the after-image appears to be instantiated by a transparent plane of glass even when we know it's an after-image. In other words, he argues that the after-image appears to be instantiated by an object in the environment in order to reject the anti-transparentist's claim that after-images seem to be internally realised phenomena. Phillips argues that after-images seem to be light phenomena for precisely the same reason.

I agree with Tye and Phillips that we don’t experience after-images to be instantiated by our minds. Indeed, I’m not sure what it would seem like to experience colour properties as being internally realised.⁹ However, Tye and Phillips present us with a false dichotomy. They assume that our only two options are as follows:

• 1. After-images appear to be instantiated by our minds (as qualia).

• 2. After-images appear to be instantiated by something in our local environment.

Since they want to reject the first option, they defend the second. However, if we reflect on these alternatives, neither seems to capture the phenomenal character of after-image experience. Fortunately, my own characterisation of the phenomenology of after-image experience provides a much-needed third option:

• 3. After-images seem to be uninstantiated colour properties.

After-images don’t seem to be properties of our minds, yet neither do they seem to be properties of objects in our environment. In other words, we can agree with Tye and Phillips that after-images don’t seem to be properties of our minds, or ‘qualia’, but we can deny this without having to make the phenomenologically rather implausible claim that after-images seem to be light phenomena or instantiated by transparent glass. During an after-image experience, we seem to be aware of a colour property that doesn’t seem to be instantiated by anything or to be genuinely located in the publicly accessible, local environment. In other words, Tye and Phillips can achieve their purpose and successfully reject the qualia theorist's view by agreeing with me and endorsing this third option, which better reflects the phenomenology of after-image experience.

We can now consider Sethi's view that although after-images don’t seem to be instantiated in the inhering in objects sense, they do seem to be instantiated in the seeming to be property instances that are genuinely spatiotemporally located sense.

Phenomenologically speaking, yellowness is there for the perceiver in the very same way as it is there in an ordinary case of color perception. We cannot capture the phenomenology of the experience merely by appealing to an uninstantiated universal, yellowness; for this does not capture how the yellowness that the subject experiences is right there in front of her or him, at a particular spatiotemporal location… (Sethi 2021: 428)

Sethi's main aim is to develop an account of the metaphysics of after-images (and sensory properties more generally), and the phenomenological claim above is used to motivate this metaphysical view. As it happens, I think Sethi provides a more plausible account of the metaphysics of after-images than the uninstantiated property view, yet I don’t think the above description successfully captures their phenomenology. I’ve argued that after-images don’t seem to be genuinely spatiotemporally located. This is partly due to their behaviour (as mentioned above), but the strongest evidence comes from the fact that it doesn’t seem to us that our companions should experience our after-images. If Sethi was right and the after-image really did seem ‘there for the perceiver in the very same way as it is there in an ordinary case of color perception’ (my italics), then it would seem to us that our companions should experience the after-image as well. In any case, it is worth noting that just as Tye and Phillips can achieve their purpose (denying the qualia theory and upholding transparency) while agreeing with me that after-images seem to be uninstantiated properties that are not genuinely located in publicly accessible space, it is perfectly possible for Sethi to agree with me that after-images don’t seem to be property instances that are genuinely spatiotemporally located while maintaining that this is what they are in fact. In other words, Sethi's theory about the metaphysics of after-images would not be threatened by conceding my point about the phenomenology of after-image experience.

The third alternative, that after-images seem to involve properties instantiated by mind-dependent entities (such as sense-data), gets closer to capturing the phenomenology of after-image experience than the two previous views: If after-images seem to be mind-dependent sense-data, then their seeming to be mind-dependent would explain why they exhibit all the features I have described (moving with the direction of our gaze, not being occludable, not exhibiting size constancy, and not seeming to be genuinely spatiotemporally located in publicly accessible space).¹⁰ Nevertheless, this proposal is ultimately unable to provide a plausible account of the phenomenology of after-image experience. I agreed with Tye and Phillips that after-images don’t seem to be qualia—it is hard to imagine how colour or shape properties could seem to be internally realised properties of us or our experiences. And after-images fail to seem to be mind-dependent sense-data for the very same reason: Everything we are perceptually aware of seems to be mind-independent. The phenomena that genuinely seem to be mind-dependent—moods, emotions, and pains, for example—don’t perceptually seem to be mind-dependent. I don’t think that anything can perceptually seem to be mind-dependent. I suspect that this fact about perceptual phenomenology explains why naïve realism is considered the common sense view of perception—perceptual experience just seems to consist in a direct and unmediated relation to objects in our environment. The phenomenal relationality of perceptual experience is undeniable, and yet it is hard to imagine what it would be like to stand in a two-place relation to something internal, whether this is an internally realised property or an internal, mind-dependent object.¹¹ After all, even proponents of the sense-datum theory don’t think that perception seems to involve the awareness of mind-dependent sense-data. This is a view of the metaphysics of perceptual experience that its proponents say we should adopt for metaphysical, not phenomenological reasons. I’ve argued that all three views that claim that after-image experience seems to involve instantiated properties are phenomenologically inadequate. Describing after-image experience as seeming to involve uninstantiated properties provides the only way to successfully capture their phenomenology.

We are now in a position to examine why property hallucinations cause such a problem for the uninstantiated property theorist. Imagine that instead of hallucinating a green apple, you are simply having a hallucination of green—a ‘disembodied’ green, if you like. In this situation, it will experientially seem to you that you are aware of the colour property green, although the green will not seem to be instantiated in either of the senses Sethi picks out. The experience does not represent (nor does it lead to the belief) that the green is instantiated by an entity (mind-independent or mind-dependent). Nor does the green appear to be genuinely spatiotemporally located.

Since a commitment to the instantiation of the relevant property during a property hallucination isn’t part of the representational content of the experience, nor of the belief that arises as a result of the experience, then the uninstantiated property theorist is forced to say that there is nothing wrong with this experience at all—it is perfectly veridical. After all, according to the uninstantiated property theorist, it doesn’t just seem to you that you are standing in an awareness relation to uninstantiated green—you really are! Your experience is perfectly veridical. No aspect of this experience is inaccurate or non-veridical. Everything is exactly as it seems.

Phosphene experiences are another example of what I’m calling property hallucinations. When we lightly push on our closed eyes, the swirly colour experiences we have as a result do not seem to be instantiated by anything in our environment. Nor do I think it is correct to say that phosphenes seem to be properties of our experience or objects internal to us. As I’ve said, while a case can be made for the idea that moods seem to be properties of us or our experience, I don’t think anything can perceptually seem this way.¹² Like after-images, phosphenes seem to be free-floating and entirely uninstantiated, that is, uninstantiated in both Sethi's senses. They don’t seem to be instantiated by objects (mind-independent or mind-dependent), and the fact that typical phosphene experience occurs when one's eyes are closed means that they don’t seem to be genuinely spatiotemporally located in publicly accessible space. Again, these experiences present a real problem for the uninstantiated property view. Our phosphene experiences do not represent the colours and shapes to be instantiated by objects or available in public space, nor do our consequent perceptual beliefs. Therefore, it seems that there is nothing wrong with these experiences—we seem to be aware of uninstantiated redness, yellowness, and swirliness (say), and we really are aware of these properties. Our experience is perfectly veridical.

Let's consider another example: the waterfall illusion. After staring at a waterfall (or another source of continuous movement) for a period of time, we find that when we shift our gaze to something stationary, we will experience movement. However, although the experience somehow involves ‘movement phenomenology’ the stationary object is not itself represented as moving. Nor does the experience give rise to the belief that the stationary object is moving. The movement doesn’t seem to be instantiated or genuinely located in the publicly accessible environment at all. Here is Dretske's description of the phenomenon:

Remarkably, though, this movement does not “attach” itself to objects. None of the objects one sees appear to be moving. Yet, one experiences movement…. Frisby describes this as “contradictory,” but it is nothing more than p-awareness of one property (movement) without this movement's being instantiated (as it normally is) in or by some object. (Dretske 2000: 163–4)

We want to classify this experience as non-veridical; it is called the waterfall illusion after all. However, the uninstantiated property theorist who holds that one is genuinely related to the (uninstantiated) property of movement when one is undergoing this illusion will have to say that one's experience of movement is perfectly veridical, and that one's belief that one is aware of uninstantiated movement is entirely true. The waterfall illusion seems to involve the awareness of ‘disembodied’ movement, and, according to the uninstantiated property theorist, it really does! Again, for the proponents of this account, things are just as they seem.

In the above examples, it is not part of the representational content of the experience, or, in naïve realist terms, the content of the perceptual belief caused by the experience, that the properties we seem to be presented with are instantiated (by objects in our local environment or by mind-dependent entities) or genuinely spatiotemporally located. For the uninstantiated property theorist, everything is as it seems. Consequently, they are unable to justify our classifying these experiences as non-veridical.

III. OBJECTIONS

There are two main ways my opponent can respond to the argument above. The first is to resist my conclusion by finding a way for the content of the experience (or perceptual belief) to come out as false, even if we are supposing that the uninstantiated property view is correct and so the experience really does involve an awareness of uninstantiated properties. If it turns out that the contents of property hallucinations are ultimately false, then the experience can still qualify as non-veridical. The second option is simply to accept my conclusion and embrace the fact that property hallucinations are entirely veridical (and so shouldn’t be thought of as property hallucinations at all). I’ll consider each response in turn.

III.1 Option one: property hallucinations are non-veridical

I’ve argued that property hallucinations are best characterised as experiences that seem to involve the awareness of uninstantiated properties; this provides an accurate description of their phenomenal character. (I’ll assume this in what follows since I’ve discussed the issue in Section II.2.) As a result, these experiences must be classified as veridical on the uninstantiated property view—they seem to involve the awareness of uninstantiated properties, and they really do. The first way my opponent can respond to my challenge is to argue that the content of the experience (or resulting perceptual belief) is false nonetheless, and so the experience can be classified as non-veridical after all. There are two ways my opponent could try to achieve this. They could argue that the representational content of the experience (or the content of the perceptual belief) isn’t reflected in the phenomenology. In other words, although the experience seems to involve an awareness of uninstantiated properties, in fact, the content is something quite different. Alternatively, they could argue that although it's true that property hallucinations involve seeming to be aware of uninstantiated properties, this doesn’t fully capture their phenomenology. Once we consider the phenomenology of property hallucinations in more detail, we find that the fine-grained contents of these experiences (or the perceptual beliefs arising in response to these experiences) include false elements, and so the experience in fact turns out to be non-veridical.

Let's first consider the possibility that phenomenology doesn’t determine content. For the representationalist, this would mean denying that we can read off the content of our experiences from their phenomenal character. For the naïve realist, it means denying that the content of the perceptual belief we form as a result of our experience matches the phenomenal character of that experience. Although the phenomenal character of our green after-image experience involves seeming to be aware of uninstantiated green, perhaps the content of the experience or perceptual belief is that the green is instantiated or otherwise present in the local environment. If so, then the content is false and the experience is non-veridical after all.

This route isn’t viable for the naïve realist, who would have to claim that an experience we would characterise as seeming to be aware of uninstantiated green causes one to believe that green is instantiated or genuinely present in the local environment. Remember, a naïve realist who subscribes to the uninstantiated property view already believes that they are aware of uninstantiated green—this is just what their theory states. It would therefore be completely mysterious why they should arrive at a belief that green is instantiated since it goes against both their own theory and the phenomenal character of their experience.

This approach is problematic for the representationalist too, since it entails that we are strangers to the contents of our own experiences. It would seem to us that we are aware of uninstantiated green—this is what we would say if asked to describe our experience—and yet, without our knowing it, our experience is really ‘saying’ that green is instantiated or present in the local environment. Even putting this counter-intuitive consequence aside, the position we end up with isn’t much of an improvement over the view that the content is correct and property hallucinations are veridical. Even if we allow the representational content to come apart from the phenomenology so that the content of the experience can be false, we end up with an experience of which it is true to say both that it is non-veridical (since the content is false) and that everything is as it seems (since its phenomenal character involves seeming to be aware of uninstantiated green). This doesn’t seem any more plausible than saying that property hallucinations are veridical.

Let's turn to the second way my opponent can try to establish the non-veridicality of property hallucinations. This involves claiming that there is more to the phenomenal character of property hallucinations than simply seeming to be aware of uninstantiated properties. For example, if one is experiencing a complex after-image experience involving multiple colours, then the experience will involve seeming to see these colours as standing in certain spatial relations to each other. Note, the claim isn’t that we will experience these colours as standing in certain spatial relations with respect to the objects and properties that are instantiated in our local environment. As I argued in Section II.2, we might for a moment experience an after-image as occupying a genuine location, but this will be short-lived; the unusual behaviour of after-images ensures that we don’t experience them as genuinely located for long. So, the idea would be that the uninstantiated properties seem to be spatially arrayed with respect to each other. My opponent could argue that these properties are not genuinely spatially arrayed with respect to each other, and so the content of the experience or perceptual belief is false, and the experience is non-veridical.

The worry with this manoeuvre is that it casts doubt on the prudence of appealing to uninstantiated properties to explain the phenomenal character of hallucinations in the first place. If my opponent claims that uninstantiated red isn’t really above uninstantiated orange (say), then being aware of uninstantiated properties doesn’t explain the phenomenal character of the hallucination after all; we may have an explanation of why we’re experiencing red and orange, but we have no explanation for why we are experiencing red above orange. What is more, it isn’t clear that my opponent can consistently claim that the uninstantiated properties we’re allegedly aware of during hallucinations are not spatially arrayed. The reasons that initially motivated the uninstantiated property view support the idea that these properties, although not spatially arrayed with respect to the local environment, are spatially arrayed with respect to each other. After all, the claim was that we must be aware of uninstantiated red, green, and orange to explain how we could come to know that red is more similar to orange than it is to green. However, we can also come to know things about spatial relations like ‘above’ and ‘under’ through hallucinating red above orange, so following the uninstantiated property theorist’s own reasoning, we must be aware of these relational properties even though they are not instantiated in the local environment.¹³ The upshot is that if the uninstantiated property theorist denies that we are aware of uninstantiated relational properties along with uninstantiated colour properties, they completely undermine the motivation for their view. If we don’t need to appeal to uninstantiated relational properties to explain the phenomenal character of experiences involving the apparent awareness of such properties, then why should we appeal to uninstantiated colour and shape properties to explain the phenomenal character of experiences involving the apparent awareness of colours and shapes?

Although I have only considered one way my opponent could try to claim that the fine-grained phenomenal character of the experience is such that the content of the experience or perceptual belief turns out to be false, it is sufficient for showing that the general approach isn’t feasible. If proponents of the uninstantiated property view are going to justify their appeal to uninstantiated properties as the best way of explaining the phenomenal character of the experience and how we can come to know certain truths through having these experiences, then, assuming the content matches the phenomenal character, they can’t also claim that the content is false. If the content is false, then the awareness of uninstantiated properties doesn’t explain the phenomenal character after all. Let's move on to the second option.

III.2 Option two: property hallucinations are veridical

The second way of responding to my argument involves embracing the idea that the experiences I have called property hallucinations are veridical. My opponent could begin by pointing out that we already classify certain experiences as veridical hallucinations—the phenomenon is well known—yet while such experiences are interesting, there is nothing particularly confusing, problematic, or implausible about them. The veridical hallucinations already discussed in the literature occur when a subject's hallucinatory experience just happens to match the experience they would be having in that particular situation if they were perceiving normally. Grice describes a case where a subject is hallucinating a clock on a shelf when there really is a clock on a shelf in front of the subject's eyes (Grice 1961: 142). The subject is hallucinating, and yet since their experience matches what they would be seeing if they were having a normal perceptual experience, the hallucination is veridical. Grice uses these experiences to support his causal theory of perception—we can only say that x sees y if x's y experience is caused in the right way by y. [See Lewis (1980) and Wilkie (1996) for discussion.]

However, there is an important difference between these kinds of veridical hallucinations and the illusory or hallucinatory experiences I claim my opponent is forced to classify as veridical. To begin, note that there are two conditions that ordinary veridical experiences meet, which justifies our classifying them as veridical: The experience matches what is there to be experienced—let's call this the matching condition—and the objects of the experience play an essential explanatory role, explaining the occurrence of the experience and (at least to some extent) its phenomenal character—let's call this the explanatory condition. Experiences like the ones Grice describes only meet the matching condition. We call the experience veridical only because it happens to match what is there to be experienced. It is pure coincidence that the subject hallucinating a clock on a shelf would be having the experience of a clock on a shelf if they were perceiving normally. The experience doesn’t meet the explanatory condition: The object in front of the subject (that they would be seeing if they were perceiving normally) doesn’t play an essential role in the explanation for the occurrence of the hallucinatory experience, nor its phenomenal character.¹⁴ We can therefore understand the reason for classifying these kinds of hallucinatory experiences as veridical (they meet the matching condition), without this classification generating anything at all implausible, puzzling, or confusing.

This is not the case for the experiences I’m calling property hallucinations. If my opponent is right that hallucinations are the awareness of uninstantiated properties, then, like ordinary veridical experiences, the experience meets both conditions for being veridical. That is, it is veridical in the sense that it matches what is there to be experienced—as Tye says, ‘an un-instantiated quality is present in hallucination’ (2014: 51); however, it is also veridical in the sense that the objects of the experience (in this case, uninstantiated properties) play an essential role in explaining the experience and its phenomenal character. Since property hallucinations meet both criteria for veridicality, they are unlike the Gricean cases. Therefore, my opponent can’t downplay the significance of the fact that they are forced to classify a kind of hallucinatory experience as perfectly veridical by appealing to the Gricean kind of veridical hallucination.

Of course, my opponents could accept this contrast between Gricean veridical hallucinations and property hallucinations, and continue to embrace the idea that property hallucinations are perfectly veridical. In other words, my opponent could simply say that I’m wrong to call these experiences ‘hallucinations’; they’re just as veridical as our veridical experiences of ordinary objects like laptops and coffee cups. Now, thinking of after-image experiences as being pretty much on a par with ordinary veridical experiences sounds quite strange. We all, even proponents of the uninstantiated property view, think of and refer to the experiences I’ve called property hallucinations as non-veridical experiences, which suggests we have the intuition that it would be implausible if these experiences turned out to be veridical. In addition, we tend to think of veridicality as something that pertains to objective matters of fact which can (at least in principle) be corroborated by other people. If uninstantiated property experiences are veridical, then they’re veridical experiences that can’t be corroborated by other people. This adds to the feeling of implausibility.

However, I don’t want to rest my objection to this response on its being counter-intuitive. And indeed, there is a more serious issue with this idea. I take it that the uninstantiated property theorist will want to distinguish our ordinary veridical experiences from our veridical experiences of uninstantiated properties—after all, they do use the terms ‘hallucinatory’ or ‘illusory’ with reference to the latter experiences. Of course, there is an easy way of distinguishing between the two different kinds of experiences, but it isn’t going to be a comfortable solution for the representationalist. The easy solution is to distinguish ordinary veridical experiences from veridical experiences involving uninstantiated properties in terms of the kinds of entities we need to appeal to in order to explain the phenomenal character of our experience. We only need to appeal to concrete physical objects and their properties when it comes to ordinary veridical experiences, whereas accounting for the phenomenal character of veridical experiences involving uninstantiated properties will require uninstantiated properties, which are abstract entities. This allows a plausible way of distinguishing the two kinds of experiences, but it seems we are left with just that: two fundamentally different kinds of experiences; one kind involving concrete entities and one kind involving abstract entities. Representationalists generally endorse the common fundamental kind claim (and say that all perceptual experiences are, metaphysically speaking, of the same fundamental kind), but here it seems they must move to disjunctivism.

It is worth unpacking this objection a little further since it relies on the idea that representationalists are committed to instantiated or uninstantiated properties being essential to our experiences. Representationalists talk about phenomenal character being explained in terms of the properties represented by that experience, and it might be thought that they thereby escape the stronger claim that these properties (instantiated or uninstantiated) are essential to the experience. (And indeed, internalist representationalists who think that our perceptual experiences are wholly internal would be unaffected by the objection under discussion. However, since internalist representationalists tend not to endorse the uninstantiated property view, we can put this aside.) It seems clear that the representationalists who endorse the uninstantiated property view are indeed committed to this stronger claim. To begin with, many theorists sign up to it explicitly:

[W]e normally “see right through” perceptual states to external objects. (Lycan 1996: 117)

When we introspect, we are aware of the external things and their properties but not of any internal experiences, nor any properties of those experiences, nor any related properties of ourselves…. the phenomenal character of my experience, what it (the experience) is like, how it feels, is a matter of the properties my experience represents. So, phenomenal character is out there in the world. (Tye 2015: 484)

It is clear that the externally located properties are playing an essential and partly constitutive role in explaining the phenomenal character of the relevant experiences. What is more, it should be noted that there would be no reason to appeal to uninstantiated properties to explain the phenomenal character of hallucinations if an essential relation to instantiated properties wasn’t required in veridical experience. After all, if an essential relation to an instantiated property isn’t required to explain the phenomenal character of ordinary veridical perception, then there would be no need to postulate uninstantiated properties to explain the phenomenal character of our experiences when there is no relevant instantiated property to do the job. [I give a more detailed argument for why representationalists who appeal to uninstantiated properties to explain the phenomenal character of hallucinations are committed to disjunctivism in Gow (2018).]

We’ve seen that it will be problematic for the representationalist to simply concede that (what I am calling) property hallucinations are perfectly veridical experiences. Let's consider whether this option is feasible for the naïve realist. Naïve realists are typically disjunctivists already, so the fact that a distinction between veridical experiences involving instantiated properties and veridical experiences involving uninstantiated properties will require disjunctivism might not seem to present a serious challenge. However, my argument will require the naïve realist to radically re-think how they classify experiences. To begin with, naïve realists typically make a distinction between veridical perceptual experiences, which essentially involve relations to mind-independent entities, and hallucinations, which they claim are a fundamentally different kind of mental state. Veridical experiences are genuinely perceptual experiences, and it is because they essentially involve relations to mind-independent entities that they qualify as such. According to the standard naïve realist line, hallucinations aren’t genuinely perceptual experiences since they do not constitutively involve relations to mind-independent entities. All this will have to change for the naïve realist who endorses the uninstantiated property view of hallucination. It will no longer be possible to distinguish between ordinary veridical experiences and hallucinations based on whether the experience involves an essential relation to mind-independent entities; all experiences involve essential relations to mind-independent entities for the uninstantiated property theorist. Naïve realists want to make a distinction between perceptual experiences and experiences that (whatever they are) aren’t perceptual experiences. However, if something counts as a perceptual experience in virtue of involving an essential relation to mind-independent entities, then, on the uninstantiated property view, hallucinations qualify as perceptual experiences too.

I’ve argued that naïve realists who endorse the uninstantiated property view of hallucinations will have to classify both ordinary veridical experiences and hallucinations as perceptual experiences, since they all involve essential relations to mind-independent entities, and this is definitional of perceptual experience according to naïve realists. Importantly, this means that the distinction between ‘veridical’ and ‘hallucinatory’ will no longer track a metaphysical distinction between the relevant experiences—some veridical experiences (those I’m calling property hallucinations) will involve essential relations to abstracta, and so will have the same metaphysical nature as hallucinations (i.e., a hallucination of a green apple). This is very different from standard naïve realism. On the standard view, we have veridical perceptual experiences on the one hand (they involve essential relations to mind-independent entities and lead to beliefs that are true) and hallucinatory experiences on the other (they don’t involve essential relations to mind-independent entities and lead to beliefs that are false). Here, the metaphysical distinction that is based on the structure of the experience coincides with the distinction that is made according to whether the experience gives rise to a true or a false belief. By endorsing the uninstantiated property view, ordinary veridical experiences and the experiences I’m calling property hallucinations will lead to true beliefs and so will qualify as veridical perceptual experiences. This constitutes a radical modification of the naïve realist's view, since categorising experiences according to their metaphysical structure will no longer provide a means of differentiating between veridical and hallucinatory experiences; hallucinations have exactly the same metaphysical structure as some veridical experiences.

We’ve seen that there are some significant drawbacks for both the representationalist and the naïve realist if they take the route of embracing the veridicality of property hallucinations. I’ll finish by outlining one further and particularly worrying concern with this approach. Now, it is always possible to wonder whether our perceptual experience is entirely accurate, or whether an aspect of the experience is illusory. Perhaps the object is really further away than it seems, or perhaps it is really oval rather than spherical (and so on). I have argued that the uninstantiated property theorist is forced to classify many non-veridical experiences as veridical; it seems to the subject of an after-image experience (for example) that she is aware of uninstantiated colours and shapes, and she really is. However, there is something very unusual about these experiences—there doesn’t seem to be any room at all for error. Unlike other perceptual experiences, it doesn’t make sense to wonder whether one is experiencing an after-image accurately or not. After all, the reason proponents of the uninstantiated property account posit such properties is to explain the phenomenal character of the subject's experience; that is, to explain why a subject is experiencing a particular shade of green when they are not relevantly related to anything green in their actual environment. If the uninstantiated properties are going to be able to play this role, then it seems that the subject cannot be wrong about which property they are aware of. If they could be wrong, then the presence of the uninstantiated property wouldn’t adequately explain the phenomenal character of their experience, and so there would be no reason to insist on its existence. It is therefore essential to the uninstantiated property account that the subject is necessarily aware of whichever universal is being called upon to explain the phenomenal character of the experience. The upshot of this is that according to the uninstantiated property account, not only is everything as it seems during a property hallucination, everything is necessarily as it seems. In other words, these experiences are more than just veridical, they are infallible.

Given the familiar problems associated with accounting for an awareness relation between subjects and abstracta—the relation cannot be causal since abstracta exist outside the causal realm—one might have expected one's experience of uninstantiated properties to be less epistemically secure than one's experience of one's local environment. Bizarrely, it turns out that the opposite is true: Given the role uninstantiated properties are being asked to play in determining phenomenal character, perceptual experiences involving these properties are in fact guaranteed to be veridical.

IV. CONCLUSION

The experiences I have discussed in this paper have something important in common. They are experiences during which we seem to be aware of properties that don’t seem to be instantiated and that don’t seem to be genuinely spatiotemporally located at all. Nor do such experiences give rise to the belief that these properties are instantiated or genuinely spatiotemporally located. We want to say that these property hallucinations are non-veridical experiences; this is how we classify them, and we tend to think that experiences that are veridical are those that could, in principle, be corroborated by other people. Yet the uninstantiated property theorist is unable to do justice to our classification of them as such. Proponents of this view hold that the subject is genuinely aware of the very properties they take themselves to be aware of. The content of their experience is therefore entirely veridical, and the content of their subsequent perceptual belief is true. The uninstantiated property view is a theory of hallucination, and yet it is committed to claiming that many experiences we classify as non-veridical are in fact veridical. What is more, unlike ordinary perceptual experiences, those experiences that involve the subject being aware of an uninstantiated property are guaranteed to be veridical. The fact that these experiences are more epistemically secure than ordinary perceptual experiences involving the subject's local environment is problematic. But more importantly, a theory of hallucination that entails that some hallucinations are necessarily veridical verges on the absurd.

Footnotes

REFERENCES