1. Externalism and Transparency

Externalism (in the sense that I am using it) is the view that sensible qualities are instantiated in external, mind-independent objects. (This view is distinct from other externalisms such as externalism about mental content or justification.) Its defenders include Bill Brewer (Reference Brewer2004, Reference Brewer2011), Tyler Burge (Reference Burge2010), Alex Byrne (Reference Byrne2001), John Campbell (Reference Campbell2002), Brian Cutter (Reference Cutter2018), Fred Dretske (Reference Dretske1995), Gilbert Harman (Reference Harman1990), Mark Johnston (Reference Johnston2004), Harold Langsam (Reference Langsam2011), M. G. F. Martin (Reference Martin and Sainsbury1997, Reference Martin2002), John McDowell (Reference McDowell1994), and Michael Tye (Reference Tye and Crane1992, Reference Tye2009). The primary motivation for externalism is that perceptual experience seems to be transparent—that sensible qualities seem, in experience, to be externally located rather than internal to our minds.

The claim that experience is transparent has been put in various ways. Here is how Gilbert Harman (Reference Harman1990: 39) puts it: ‘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 experiences . . . Look at a tree and try to turn your attention to intrinsic features of your visual experience. I predict that you will find that the only features there to turn your attention to will be features of the presented tree’.

Michael Tye (Reference Tye and Crane1992: 160) similarly says, ‘Standing on the beach in Santa Barbara a couple of summers ago on a bright sunny day, I found myself transfixed by the intense blue of the Pacific Ocean . . . I experienced blue as a property of the ocean not as a property of my experience. My experience itself certainly wasn't blue’. (See also, for example, Loar Reference Loar, Smith and Jokic2003; Dretske 2003; Cutter Reference Cutter2018; Speaks Reference Speaks2014; Martin Reference Martin2002; Byrne Reference Byrne2001).

Many philosophers claim that perceptual experience is transparent. They are not always clear about what all, and what exactly, they take sensible qualities to seem like in perceptual experience. But the core of their claim is that, in perceptual experience, sensible qualities (such as colors, smells, textures, pitches, flavors) seem to be instantiated in mind-independent objects. Various philosophers, including non-transparency theorists, make additional claims about the way sensible qualities seem in experience, such as that they seem to be immediate (Campbell Reference Campbell2002; Martin Reference Martin2002), intrinsic (Tye Reference Tye2000: 153; Johnston Reference Johnston1992: 223; Yablo Reference Yablo1995: 489; McGinn Reference McGinn1996: 542; Chalmers Reference Chalmers, Gendler and Hawthorne2006: 66; Cutter Reference Cutter2018: 41), or categorical (Yablo Reference Yablo1995: 490; Boghossian and Velleman Reference Boghossian and Velleman1998: 86; McGinn Reference McGinn1996: 545; Cutter Reference Cutter2018). But in what follows, I focus on the above core claim.

Not everyone who claims that perceptual experience is transparent accepts (or even talks about) externalism. But everyone who accepts externalism claims that perceptual experience is transparent in the sense described above. Indeed, as Cutter (Reference Cutter2018: 39) says, ‘the primary motivation’ for externalism draws from the claim that sensible qualities seem to be instantiated in mind-intendent objects.

But now, as externalists typically acknowledge, that things seem thus-and-so does not entail that they are thus-and-so. Externalists nonetheless consider perceptual appearances to be a motivation for externalism because they also accept something like phenomenal conservativism—the view that if it seems to one as if P, then one has at least prima facie justification for believing that P (Huemer Reference Huemer2001)—at least as applied to sensible qualities (see Cutter Reference Cutter2018). So then this is how externalists typically argue: In perceptual experience, sensible qualities seem to be features of external, mind-independent objects (transparency); we are prima facie justified in believing that things are as they seem (phenomenal conservativism); therefore, we are prima facie justified in believing that sensible qualities are features of external, mind-independent objects. Externalists then argue that we do not have compelling evidence against this claim. So they conclude that externalism is justified and, indeed, true.

That is their argument for externalism. This is not yet an argument for any full theory of experience like naïve realism or externalist representationalism, as such an argument requires additional resources showing that the above considerations fit best with such a theory (more on this below). But the above argument, if sound, does support—and, indeed, is philosophers’ primary motivation for accepting—externalism.

I have no bones to pick with either the transparency claim or phenomenal conservativism. I am happy to grant that in experience sensible qualities seem to be features of mind-independent objects, and I laud externalists’ insistence on giving phenomenology its epistemic due. So I embrace these assumptions henceforth.

But these assumptions do not lead where externalists think that they lead. Externalists think that the above considerations fit best with either naïve realism or externalist representationalism. That is, they think that either naïve realism or externalist representationalism do a better job than competing theories of vindicating the relevant appearances, since these theories imply that sensible qualities really are instantiated in mind-independent objects as they appear to be. (Tye, who is an externalist representationalist, does at least acknowledge that transparency considerations are also consistent with the sense-datum theory [2000: 47].) But, as it turns out, one need look no further than externalists’ own premises to find an objection to these two theories about perceptual experience that they favor. And one need look no further to find support for a theory that they abhor—namely, the sense-datum theory.

Or so I argue below. In the next section, I describe three views about the nature of perceptual experience—two that externalists like, one that they most definitely do not like. Then, in the section after that, I argue that the primary motivation for externalism in fact supports the view they do not like: the sense-datum theory.

2. Three Views about Perceptual Experience

Most (if not all) externalists are either naïve realists or externalist representationalists. For example, Brewer (Reference Brewer2004, Reference Brewer2011), Martin (Reference Martin and Sainsbury1997, Reference Martin2002), McDowell (Reference McDowell1994), and Langsam (Reference Langsam2011) are naïve realists; Byrne (Reference Byrne2001), Cutter (Reference Cutter2018), Dretske (Reference Dretske1995), Harman (Reference Harman1990), and Tye (Reference Tye and Crane1992) are externalist representationalists.

Naïve realism is the view that perceptual experiences are constituted by one's awareness of external objects and properties. On this view, when I see a red-tailed hawk while eating pistachio ice cream and petting my dog, that hawk, ice cream, and dog are constituents of my experience. And the properties that I experience them as having really are properties of those external objects. The hawk's tail, which I experience as orange-ish, really is orange-ish. Likewise, the taste of my ice cream and feel of my dog really are features of the ice cream and the dog—the ice cream really does have a sweet, nutty flavor, and the dog's hair really is soft and fluffy.

Externalist representationalists agree with naïve realists that external objects really do have the properties that we perceive them to have (at least in normal cases of perception). And they also agree with naïve realists that the qualitative character of our perceptual experiences is grounded in, or derives from, the ways those external properties and objects really are. But externalist representationalists disagree with naïve realists about what constitutes our perceptual experiences. Again, naïve realists say that our perceptual experiences are constituted by our awareness of external objects and properties. Externalist representationalists deny this, instead saying that our perceptual experiences are constituted by our representing those properties and objects. So when I experience the hawk's tail, the ice cream's flavor, and the dog's fur, what is going on is that I am representing these objects as having certain colors, tastes, textures, and so on.

As I have said above, most (if not all) philosophers who are moved by the considerations laid out in the previous section are either naïve realists or externalist representationalists. Not sense-datum theorists. The sense-datum theory, which is most closely associated with Bertrand Russell (Reference Russell1912, Reference Russell1913, Reference Russell1914), is the view that perceptual experiences are constituted by one's awareness of sense data—that is, objects that our sensory systems generate due to their interaction with the world, and that we each perceive in our own private space. So when I have my hawk-ice-cream-and-dog experience, what is going on is that things in the physical world are causing my sensory systems to generate these sense data that I then perceive. And these sense data really do have the sensible qualities they seem to have. My ‘hawk’ sense datum, for example, really is (partly) orange-ish.

Besides Russell, H. H. Price (Reference Price1932), C. D. Broad (Reference Broad1923), G. E. Moore (Reference Moore1953), Frank Jackson (Reference Jackson1977), Howard Robinson (Reference Robinson1994), John Foster (Reference Foster2000), and Brian O'Shaughnessy (Reference O'Shaughnessy and Smith2003) defend the sense-datum theory. Although it is no longer very popular, the sense-datum theory used to be the dominant view in philosophy. Thomas Reid ([Reference Reid1785] Reference Reid2011: 76) said, ‘All philosophers, from Plato to Mr. Hume, agree in this, that we do not perceive external objects immediately, and that the immediate object of perception must be some image present to the mind’.

There are many possible versions of the sense-datum theory. But on a traditional version of it—the version that I focus on here—sense data, despite being generated by the mind, are not themselves in or part of the mind (contrary to what some philosophers assume about sense data). Early sense-datum theorists in particular insist that sense data are wholly distinct from our minds. Russell (Reference Russell1912: 67), for example, emphasizes that ‘acquaintance with objects’ such as sense data ‘essentially consists in a relation between the mind and something other than the mind’. And Russell (Reference Russell1913: 78) says, ‘[I] hold that the sense-datum is certainly something other than the subject, something to which the subject's relation is just as “external” as to the physical object’ (see also Moore Reference Moore1953: 30–31). So, according to this traditional version of the sense-datum theory, sense data—these objects that we are aware of in experience—are distinct from our minds. In this way, they are mind-independent. They are like images cast upon a (private) screen, or items in a virtual-reality world.

Some philosophers might be surprised to learn that, on Russell's traditional version of the sense-datum theory, sense data are mind-independent. For they might have thought—and maybe were taught this—that sense data are (and must be) thoroughly mental items and, thus, must be mind-dependent. Indeed, some philosophers assume as much in their arguments (see, for example, Martin Reference Martin2002; Valberg Reference Valberg and Crane1992; Huemer Reference Huemer2011). But this assumption is mistaken. (And, as I show below, it is an important mistake.) For, again, on a prominent, traditional version of the sense-datum theory—the one I focus on in what follows—sense data are mind-independent.

Naïve realism, externalist representationalism, and the sense-datum theory each has its own theoretical strengths and weaknesses. But I do not weigh them here (though below I do say more to fend off dismissiveness toward the sense-datum theory). For my purpose in this article is just to point out an underappreciated mistake, and a predicament, for philosophers who rely on certain motivations (that is, transparency plus phenomenal conservativism) to support a certain view about sensible qualities (that is, externalism), thinking that this view and those motivations fit best with a view about the nature of perceptual experience that they like (that is, either naïve realism or externalist representationalism). The mistake is in thinking that those motivations fit best with either of the views that they like. They do not. And the predicament arises because those motivations in fact fit best with a view that they do not like (that is, the sense-datum theory). Or so I argue in the next section.

3. Transparency Revisited

There are various kinds of sensible qualities. For example, there are colors, shapes, textures, pitches, flavors, smells, and temperatures. These various kinds of sensible qualities show up in various kinds of cases. There are ordinary cases of veridical perceptual experiences, like the ones I have been describing. Then there are also less ordinary cases—hallucinations or illusions, for example. These cases are different in important ways from ordinary, veridical cases. But, like those cases, they feature sensible qualities (or at least they seem to). In fact, in some such cases, things seem just like they do in cases of ordinary perception. Thus, insofar as we are taking the appearances seriously, we must consider these kinds of cases as well.

Externalists take phenomenological appearances seriously. And they claim to get the cases right. That is, they claim that externalism is most faithful to the way things seem in experience. And since they also accept that we are prima facie justified in believing that things are as they seem, they cite phenomenological faithfulness as a good reason to accept externalism. Indeed, this is their primary motivation for their view. Externalists then pair their view with either naïve realism or externalist representationalism, because they believe that one of these views is best suited to vindicate the appearances that they care about.

However, in this section, I argue that the sense-datum theory is in fact better suited to vindicate the way sensible qualities seem in experience than either naïve realism or externalist representationalism. I argue that it does at least as well at capturing the way sensible qualities seem in veridical perception, and that it does much better at doing so in cases of hallucination. So I argue that it is the sense-datum theory, not externalists’ actual views, that gets the cases right. Thus, insofar as externalists are right to privilege the way thing seem in experience, we should favor the sense-datum theory over naïve realism and externalist representationalism. And given that these considerations are externalists’ primary motivation for their view, externalists in particular will have reason to change their ways and accept the sense-datum theory.

Start with cases of ordinary, veridical perception. Take a very simple case: I see a red ball in front of me. The transparency claim implies that the sensible qualities that I see—the redness and roundness, for example—seem to be instantiated in a mind-independent object. And, if externalists are right, this gives us reason to believe that the redness and roundness are, in fact, instantiated in a mind-independent object.

Both naïve realism and externalist representationalism deliver this verdict. That is, they imply that the redness and roundness I see really are instantiated in a mind-independent object. But so does the sense-datum theory. It implies that the redness and roundness I see are instantiated in a mind-independent object—namely, a sense datum. For, recall, according to the sense-datum theory, sense data are wholly distinct from our minds—they are no more occupants of the mind than are rocks, tables, and chairs. And, according to the sense-datum theory, sensible qualities are instantiated in these sense data. Thus, all three views under discussion deliver a verdict in line with externalists’ transparency-inspired judgment (which reflects their primary motivation) that sensible qualities are instantiated in mind-independent objects. Therefore, as far as externalists’ primary motivation goes, all three views handle cases of veridical perception equally well.

Maybe you think this is a cheat. That is, maybe you think that the sense-datum theory cannot really be in line with externalists’ primary motivations, because you think that sense data are not really external objects in the way externalists mean it. I will not try to guess what externalists mean (they are not always clear). But, regardless, what matters is that, on the sense-datum theory, sense data really are external to our minds in exactly the way that is relevant to externalists’ transparency motivation—they really are wholly distinct from, not a part of, and external to our minds. Therefore, the sense-datum theory satisfies this constraint.

As I suggested above, this may come as a surprise. For the transparency claim is often assumed to support views like naïve realism and externalist representationalism, not the sense-datum theory. In fact, some use it expressly to criticize the sense-datum theory (for example, Martin Reference Martin2002; Valberg Reference Valberg and Crane1992; see also Crane Reference Crane1992; Huemer Reference Huemer2011). However, these philosophers assume that sense data must be in the mind and, indeed, mind-dependent. So they assume that the sense-datum theory obviously runs counter to transparency. But, according to a traditional version of the sense-datum theory, sense data are wholly distinct from the mind. They are mind-independent. Recall that Russell (Reference Russell1912: 67), for example, says that acquaintance with objects such as sense data ‘essentially consists in a relation between the mind and something other than the mind’. And Russell (Reference Russell1913: 78) says, ‘[I] hold that the sense-datum is certainly something other than the subject, something to which the subject's relation is just as “external” as to the physical object’. He also explicitly denies that sense data are in the mind (Russell Reference Russell1912: 67; Russell Reference Russell1913: 79). And he allows that it may be possible for sense data to exist unperceived. Other traditional sense-datum theorists agree. G. E. Moore (Reference Moore1953: 31), for example, distinguishes between ‘the sense-data which I see’ on the one hand, and the mind and its activities (“my seeing of them”) on the other. And he says that it is ‘at least conceivable’ that sense data continue to exist after we stop perceiving them (see also Moore 1903: 35–56; Price Reference Price1932; Jackson Reference Jackson1977). He also says that it is conceivable that sense data are located on the surfaces of material objects. (Frank Jackson [Reference Jackson1977: 102–3] goes further, saying that sense data are, in fact, in physical space at various distances from perceivers.) Thus, on a traditional version of the sense-datum theory, sense data satisfy the transparency constraint—they are external to and independent of the mind.

You might think that, although sense data may not be in or part of the mind, they must be mind-dependent in another sense—namely, in that they could not exist independently of the mind; if one's mind ceased to exist, then, necessarily, so too would any sense data it generated. This is not true of ordinary external objects like rocks and tables—they can exist sans minds. So you might think that sense data are mind-dependent in a way that ordinary external objects are not. So you might think that sense data are not really external objects.

To this, a sense-datum theorist might reply that sense data are not, in fact, mind-dependent in the above sense—that they could possibly exist independently of the mind. As we saw, Russell, Moore, and other sense-datum theorists seemed to believe this. And, at any rate, this view is at least open to sense-datum theorists.

However, I prefer a different reply. It is that whether or not an object is mind-dependent in the above sense is not part of how it appears in experience, so it is not relevant to the present question of which view fits best with the appearances. My experience of a red ball in front of me, for example, does not testify as to whether the ball could exist without a mind. It is unclear what it would even be like for it to. What would it be like for a red ball to appear, phenomenologically, as though it could or could not possibly exist without a mind? These are modal facts about what could be the case if one were not experiencing some object. It is hard to see how these facts could be part of the way an object in fact appears in experience. And, regardless, that is just not how objects do appear in experienceFootnote ¹—which, again, is important because it is the appearances—how things seem in experience—that externalists draw on to formulate the transparency claim. So externalists cannot cry foul in virtue of sense data necessarily ceasing to exist without a mind (if that is indeed the case). For that has nothing to do with how things seem, phenomenologically.

Another way you might resist the idea that sense data are genuinely external is by saying that, unlike the ordinary external objects that externalists mean to be talking about, sense data are caused or generated by the mind.

But, again, the way that sensible objects are caused or generated is not an item of experience—it is not part of how things seem, phenomenologically. It is not as if sensible properties and objects have some glow, or other sensible marker, that speaks to their origins. And, again, it is the appearances—how things seem in experience—that externalists draw on to formulate the transparency claim. So, as before, externalists cannot cry foul in virtue of sense data being generated by the mind, for that has nothing to do with how they seem, phenomenologically. In experience, sensible qualities seem to be instantiated in mind-independent objects. According to the sense-datum theory, they are. Thus, as far as externality goes (which is what the transparency claim is limited to), the sense-datum theory fits with the way sensible qualities seem in experience.

Maybe you think that there is more to the way sensible qualities seem in experience—that it is not just that they seem to be instantiated in mind-independent objects. For example, maybe you think that sensible qualities seem to be instantiated in physical objects, which are located in physical space, have mass, and so on. Sense data may not be physical—they do not have mass, physical locations, or the like. So perhaps the transparency claim, and the subsequent argument for externalism, can be beefed up to rule out the sense-datum theory.

Before I address the particulars of this objection, I note that these are treacherous waters for externalists who are loyal to naïve realism or externalist representationalism. That is, it is dangerous for them to start accepting further appearance-based inferences about sensible qualities beyond the inference to externality. For, as they are well aware, some of the ways sensible objects and qualities seem in experience are inconsistent with the way things are in the physical world. There are of course illusions and hallucinations (which I discuss below). But there are also facts like this: When I look at a penny from a certain angle, it appears to be elliptical. In the physical world, the penny is circular, not elliptical; yet it really does seem like ellipticality is instantiated. Sense-datum theorists will say that ellipticality is instantiated—in a sense datum (Russell Reference Russell1912: 46; Price Reference Price1932: 55; Broad Reference Broad1923: 237). So they can handle these appearances. But naïve realists and externalist representationalists have to deny these appearances, saying that ellipticality is not, in fact, instantiated. This example is controversial. But there are plenty of others to illustrate this point. For example, when you cross your eyes, you ‘see double’—it looks as though there are two of the object in front of you. But there are not two physical objects out there. There is also perceptual variation—for example, when an object looks one color to me and another color to you. Assuming that physical objects are not (always) multiple colors all over, standard externalists cannot possibly save all of the appearances. This is well-worn terrain. My point is just that externalists hoping to stay loyal to naïve realism or externalist representationalism must be careful when making further appearance-based inferences about sensible qualities beyond that they are instantiated in mind-independent objects.

But back to the objection. The objection derives from the claim that sensible qualities seem, in experience, to be features of physical objects. Is this claim true? It is true that we have sensations of weight, from which we may infer that things in our environments have mass. But sense-datum theorists can account for these sensations as qualities of sense data. (Sense-datum theorists focus more on visual perception—everyone does—but they posit sense data of other modalities as well.) So sense-datum theorists can capture these appearances. It is also true that sensible objects seem to be spatially located. But, again, sense-datum theorists agree that they are spatially located. They usually say it is in a private, nonphysical space (see, for example, Russell Reference Russell1912: 46), though some say that sense data are located in physical space (for example, Russell Reference Russell1914; Jackson Reference Jackson1977: 102–3). Thus, just in terms of the way things seem in experience, it is not at all obvious that the sense-datum theory is leaving anything out corresponding to the apparent physicality of sensible objects.

So the sense-datum theory can handle appearances of physicality. But now also notice that, in some other ways, sensible objects do not seem physical, phenomenologically speaking. Physics tells us that ordinary material objects are mostly empty space. But that is not how sensible objects look or feel. According to various leading theories of space, space is high-dimensional (not three-dimensional), or constituted by flashes, or made up of objects spread out across the entire universe, or are tiny vibrating strings (see Emery Reference Emery2017). But this is not how sensible objects seem, in experience. Furthermore, as dualists emphasize, sensible qualities like colors seem to be importantly different from the structural and dynamical qualities associated with physical qualities (see Gertler Reference Gertler and Kriegel2020). Thus, contrary to the present objection, sensible qualities and objects do not seem, phenomenologically, to be physical in every respect. And the sense-datum theory can explain this. For, on the sense-datum theory, sense data lack the physical features they seem to lack—they are not mostly empty space, high-dimensional, string-like, or the like. So, on this point, the sense-datum theory may do even better, appearances-wise, than naïve realism and externalist representationalism.

What about other appearances? For example, some philosophers say that causality—such as one billiard ball hitting another—enters into how things seem in experience (Siegel Reference Siegel2010). But presumably sense data do not bear those kind of causal relations to each other. So perhaps this is one appearance where sense-datum theorists are at a disadvantage.

Maybe. But there are various things sense-datum theorists could say here. For example, they could deny (as others have) that causality is ever directly experienced. They could instead say that judgments about causation are higher-level interpretations or based on quick inferences, for example. Or they could admit that we experience causality, but deny that causality is a sensible quality like color or shape. Thus, they could stand their ground in the present debate, which is about sensible qualities. So even if appearances of causality do make some trouble for sense-datum theorists, this point is hardly decisive in this context.

Maybe there is some other aspect of the way sensible qualities seem in ordinary, veridical perception that gives an edge, appearance-wise, to naïve realism or externalist representationalism over the sense-datum theory. But it is not obvious what that would be. As I have said, other philosophers (including non-externalists) make claims about the way sensible qualities seem in experience that go beyond the claim of externality, such as that sensible qualities seem to be immediate, intrinsic, or categorical. But the sense-datum theory captures these appearances, since it implies that sensible qualities are immediate, intrinsic, categorical features of sense-data. Some externalists say things like perceptible objects seem to be “ordinary” objects (for example, Cutter Reference Cutter2018: 39), and they also assert that sensible qualities seem to be features of objects like trees, oceans, and sunsets—which sense-data are not. But, to avoid simply begging the question, we need to know what aspects of phenomenology—just the appearances, remember—justify these claims. Again, it is not at all obvious. As far as the appearances go, the sense-datum theory delivers exactly the same verdicts or predictions as its competitors about how things should seem in veridical perceptual experience.

If there is some other, as yet unnoticed aspect of phenomenology that cuts against this conclusion, externalists should bring it to light. Their main claim is just that sensible qualities seem to be instantiated in mind-independent objects. This claim sits well with the sense-datum theory. As do other claims about veridical perception in the offing. Indeed, as things stand, it looks like the sense-datum theory captures the way sensible qualities seem in veridical perception at least as well as, if not better than, its competitors.

So much for the ordinary cases. Now for the extraordinary ones—hallucinations, in particular. Take a very simple case: I hallucinate a red ball. Some hallucinations are so vivid and lifelike that they seem exactly like veridical experiences (for discussion, see Pautz Reference Pautz2021: ch. 1). Suppose that the present case is one of them. So my hallucination of a red ball seems exactly like my veridical experience of a red ball. Thus, given that in the veridical case it seems to me like sensible qualities (such as redness, roundness) are instantiated in a mind-independent object, so too in the hallucinatory case it seems to me like sensible qualities are instantiated in a mind-independent object.

The sense-datum theory accommodates these appearances. For, according to the sense-datum theory, the redness and roundness that I am hallucinating really are instantiated in a mind-independent object—namely, a sense datum. Naïve realism and externalist representationalism, however, do not accommodate these appearances. For, on those views, sensible qualities are instantiated in physical objects like trees, oceans, and balls; but when I hallucinate a red ball, there is no such object present for those qualities to be instantiated in. So defenders of these views have to say something else about why it seems in experience that some object is red and round. Externalist representationalists typically say that, in hallucinations, we represent qualities that are not there—that the contents of such experiences are uninstantiated universals. So they say that although sensible qualities may seem to be instantiated in mind-independent objects in hallucinations (just as they do in veridical perception), in fact, they are not. Naïve realists tend to say, and then emphasize, that hallucinations and veridical perceptions are fundamentally different kinds of experiences. This leaves open what kind of experience hallucinations are. One option is to say that hallucinations involve sense data (Byrne and Logue [Reference Byrne, Logue, Haddock and Macpherson2008: 63–65] call this “Austinian disjunctivism” after J. L. Austin [Reference Austin1962]). But few naïve realists are likely to choose this option (as Byrne and Logue [Reference Byrne, Logue, Haddock and Macpherson2008: 64] point out, not even Austin did). Yet, if they shun sense data, there are no mind-independent objects for these sensible qualities to be instantiated in (at least not obviously). So then it looks like they have to deny hallucinations’ appearances that sensible qualities are instantiated in mind-independent objects.

So the sense-datum theory better captures hallucinations’ appearances. Again, on naïve realism and externalist representationalism, when I hallucinate a red ball, sensible qualities such as redness and roundness are not instantiated in a mind-independent object. On externalist representationalism, these qualities are not instantiated at all. And, unless naïve realists resort to the sense-datum theory on this point, they have to say something similar. But this is out of step with the appearances. For, again, when I hallucinate a red ball, it seems in experience that redness and roundness are instantiated in a mind-independent object. Thus, among our three views, the sense-datum theory is the most faithful to the way things seem in hallucinations.

You might think that it does not matter if a view is not faithful to the way things seem in hallucinations, because hallucinations are ‘bad’ cases—deviant or abnormal—and we should not base our views about perceptual experience on bad cases.

But ‘bad’ in what sense? How are hallucinations bad cases? It is true that hallucinations are not good guides to the way things are in one's physical environment. So in that sense they are bad cases. But, here, in the context of talking about how sensible qualities seem in experience, our concern is not the physical environment per se. Rather, our concern is experience. In particular, it is how sensible qualities seem in experience. And hallucinations are perfectly good cases of that. The ways colors, shapes, tastes, smells, and the like appear in hallucinatory experiences are, at least in some cases, no different from how they appear in ordinary, veridical experience. So the fact that hallucinations do not match (in some sense) with one's physical environment is irrelevant. What we are concerned with here is not what sensible qualities correspond to or match up with; we are concerned with sensible qualities themselves and what they seem like in experience.

Another way to put the above objection is to say that phenomenal conservativism (a key premise, for externalists) only confers prima facie justification for believing that things are as they seem, and then to say that, in cases of hallucination, this prima facie justification is overridden by the fact that something is amiss with the experience.

But the response is the same: Justification for what? Justification for beliefs about the physical world? True, hallucinations do not give us that. But what about justification for beliefs about experience? About sensible qualities? Hallucinations, as abnormal as they may be, are perfectly good sources for justified beliefs about those.

Consider an analogy. Suppose you smell a red rose for the first time. You experience a certain sensible quality—that is, a certain smell—and thereby come to know what it is like. But now suppose that you are in the Matrix (as depicted in the Matrix film trilogy). So, actually, there is no rose there. In some sense, this is a ‘bad’ case of perceptual experience, since your experience is not a good guide to what is around you, physically. But, in another sense, this is a perfectly good case of perceptual experience. For you still have the experience. Those sensible qualities are still there. And you still know what they are like (see Sturgeon Reference Sturgeon1998; Johnston Reference Johnston2004). That smell, along with the shape and color of the rose that you see, are just as much a part of what you experience as in ordinary, veridical perception. (This is part of what makes Matrix-like scenarios so compelling.) And this is precisely what externalism is a view about. It is a view about sensible qualities—about what they are like and where they are located.

Perhaps the worry is not just that things go amiss in hallucination. Perhaps it is also that hallucinations are rare—infrequent oddities that are not to be trusted in any respect, since they do not occur very often.

But the frequency of an experience does not by itself have any bearing on whether we should take its appearances seriously. Every maximally specific experience is rare—no experience is exactly duplicated very often, if ever. So if the frequency of an experience bears on the trustworthiness of its appearances, then we can never trust the way things appear in experience—which runs directly counter to the externalist assumptions that we have adopted. Thus, given those assumptions, it does not matter that hallucinations are rare experiences. What matters is that they are experiences. And they are indeed that—just like any other experiences.

You might still think that hallucinations are odd or weird, which may affect how much credence you are inclined to lend them. But weird in what sense? They need not be weird in the sense that they seem, phenomenologically, unlike other experiences. For some hallucinations seem exactly like other experiences, phenomenologically. Maybe the content of some hallucinations (such as pink elephants) is weird. But not all of them. Some hallucinations are banal (such as a red ball, a pile of leaves). We could always limit our attention to those hallucinations. The sense-datum theory would still have the edge in terms of capturing their appearances.

Perhaps there is another objection here having specifically to do with mind-independence. Perhaps it is that when it comes to hallucinations we have a defeater for the belief that sensible qualities are instantiated in mind-independent objects. Whereas in veridical perception we have a reason to believe that sensible qualities are instantiated in mind-independent objects and no reason to believe otherwise, in hallucination, even if we have some reason to believe that sensible qualities are instantiated in mind-independent objects, we also have reason to believe that they are not—since hallucinated objects are not present in the physical environment.

I completely agree that when I hallucinate a red ball (for example), and I know that I am hallucinating, I have a defeater for the belief that redness is instantiated in a physical object in my environment, since no such object is actually there. However, this does not mean that I thereby also have a defeater for the belief that redness is instantiated somewhere. That there is not a red ball in my physical environment does not entail that that redness is uninstantiated. The sense-datum theory shows us that. And whereas my conviction that there is a red ball in my physical environment would diminish entirely if I learned that I was hallucinating, my conviction that something is red here—even if I am not sure what it is—would remain. For that sensible quality seems present in my hallucination just as it does in an indistinguishable veridical perception.

An externalist could respond by denying that sensible qualities are present in hallucinatory experiences—that although it seems like they are, in fact, they are not. There are two different ways to understand this objection. The first is as saying that although it seems phenomenologically like sensible qualities are present in hallucinations, in fact, they are not. But this strategy is of doubtful coherence, since, on most accounts, sensible qualities just are ways things seem phenomenologically. Furthermore, this strategy denies—completely disregards—the appearances, which is in tension with externalists’ primary motivation for their view.

Another way to understand the above objection is as saying that, although someone who is hallucinating may be inclined to believe that sensible qualities are present in their experience, they are not, and it does not even seem phenomenologically as if they are (see Martin Reference Martin, Gendler and Hawthorne2006). This strategy has a number of problems (see, for example, Siegel Reference Siegel2004, Reference Siegel, Haddock and Macpherson2008; Johnston Reference Johnston2004; Sturgeon Reference Sturgeon1998; Pautz Reference Pautz and Nanay2010, Reference Pautz2011). But the point I want to stress is just that this strategy requires rejecting countless contravening self-reports and honest introspective efforts. And such a strategy, aside from being problematic in other ways, undermines the credibility of the very kind of experiential appraisals externalists require to motivate their view in the first place.

So, either way, it is unwise (or unfair) for externalists to ignore hallucinations. And, again, the sense-datum theory is more faithful to the way things seem in hallucinations than either naïve realism or externalist representationalism. For, among these three views, only the sense-datum theory implies that, true to the way things seem in experience, the sensible qualities of hallucinations are instantiated in mind-independent objects.

And it is worth emphasizing how big the disparity is here—that is, how far off naïve realists and externalist representationalists (but not sense-datum theorists) are when it comes to the way things seem in hallucinatory experiences. It is one thing for a view about perception to imply that some sensible qualities are a bit different from how they seem in experience—maybe a different hue, or shape, or size, or even location. But what naïve realists and externalist representationalists have to say is that, in hallucinations, sensible qualities are not instantiated at all. This is a radical departure from how things seem in hallucinations.

To see this more clearly, distinguish between two components of the transparency claim: The instantiation component and the location component. The transparency claim implies that sensible qualities seem to be instantiated, and it also implies something about where they seem to be instantiated—namely, in mind-independent objects. Externalists object to various views about perception by saying they run afoul of the appearances with respect to location. But, in cases of hallucination, externalists’ favored views run afoul of the appearances with respect to both location and instantiation. For these views imply that hallucinated sensible qualities are not just located somewhere other than where they seem to be; they imply that these qualities are not instantiated anywhere. Thus, even if externalists are right that various views about perception are at odds with the appearances, their own views about hallucinations are even more at odds with the appearances.

Consider the point from a different angle. Some representationalists (not just externalists) say that perceptual experience is a lot like belief, and that having a perceptual experience of something that isn't really there, as in hallucination, is a lot like having beliefs about something that does not really exist (for example, Santa Claus). But whatever the merits of this suggestion are, surely it is at odds with the way things seem in experience. For, unlike just believing things about, say, Santa Claus, when I hallucinate Santa Claus, something seems to be there—I am presented, indeed confronted, with sensible qualities; it seems like something I can touch, feel, and point to (see Johnston Reference Johnston2004). So even if perceptual experience is a lot like belief, surely that is not how things appear. The claim that hallucinated sensible qualities are not instantiated anywhere—which both naïve realists and externalist representationalists accept—is indeed a striking departure from the appearances.

Some philosophers contend that it is beyond striking—that it is simply unbelievable. Price (Reference Price1932: 3), for example, says, ‘When I see [or hallucinate] a tomato . . . I cannot doubt that there exists a red patch of a round and somewhat bulgy shape’. Price says that he cannot doubt—that he is certain—that the no-instantiation claim is false. Now, whether or not we go as far as Price, it is fair to say that the no-instantiation claim is a departure from how things seem in hallucinatory experiences, and a radical departure at that.

At this point, naïve realists and externalist representationalists might return to the earlier point about the rarity and oddity of hallucinations, but in a more dialectically oriented way. Specifically, they might grant that the sense-datum theory handles hallucinatory appearances better than their preferred theories, and even grant that this is a point in favor of the sense-datum theory, but then they might suggest that it is not much of a point in favor of the sense-datum theory because of how rare and odd hallucinations are. Naïve realism and externalist representationalism save most of the appearances. Maybe that is good enough.

I am not so sure. In a subfield—philosophy of perception—where single counterexamples are often considered damning, and in a debate—over externalism—where phenomenological appearances are given special weight, and in a context—when considering hallucinations—where there is such a big disparity in how well the different theories handle the cases, I do not think that externalists can so easily brush hallucinations aside. However, with that said, remember that my goal is not to argue that at the end of the day everyone should be sense-datum theorists. Rather, my goal is to argue that the appearance-based considerations that externalists draw on fit better with the sense-datum theory than their favored theories and, thus, those whose primary theoretical motivations are those appearance-based considerations ought to be sense-datum theorists.

Again, externalists’ primary motivation for their view is about the appearances. They claim that sensible qualities seem in experience to be instantiated in mind-independent objects and that this is a good reason to accept externalism. It is just that the sense-datum theory fits better with these motivations than externalists’ actually favored views. It does at least as well at capturing the way sensible qualities seem in ordinary perception, and it does much better in cases of hallucination. Thus, insofar as externalists are right to privilege the way thing seem in experience, we should favor the sense-datum theory over their actual views—naïve realism and externalist representationalism. And given that these considerations are externalists’ primary motivation for their view, they in particular have reason to accept the sense-datum theory.

Notice that I am not just pointing out (as others have) that naïve realism and externalist representationalism have a hard time explaining the qualitative character of hallucinations. After all, a lot of views about perception have this problem. Rather, I am appealing to externalists’ very own motivation for their view—based on the transparency claim—and arguing that, given that this is their primary motivation, externalists in particular should be sense-datum theorists, not naïve realists or externalist representationalists.

Notice also that I am not just pushing the standard sense-datum-theorist line that for any way things appear in experience, there must be something that is that way. I acknowledge the fallacy in this general inference. What I am rather doing is, again, appealing to externalists’ more specific, narrow claim that in experience sensible qualities seem to be instantiated in mind-independent objects, and arguing that, given that their primary motivation is based on this claim, externalists in particular should be sense-datum theorists.

But now, with that, consider one last objection (one that may also have been your first objection, nagging at you from the title on). It starts with a question: Is not the sense-datum theory crazy? Is it not totally implausible? Are we not supposed to mock the sense-datum theory, and maybe even feel good about doing so, because at least we do not entertain notions like that anymore? So then, put more directly, the objection is that externalists need not take the sense-datum theory seriously because it is a non-starter view.

The sense-datum theory actually has more going for it than is commonly assumed (see Pautz Reference Pautz2021, for a nice overview of its pros and cons). Plus, every view about the nature of perceptual experience has serious problems and implausible implications. I have already mentioned one implication of naïve realism and externalist representationalism that I, for one, find incredible—namely, that sensible qualities are not instantiated (at all!) in hallucinatory experiences. This problem applies to other views about perception as well. Again, every view has problems. So the sense-datum theory is not alone in that regard. And, as I have said, it also has some things going for it, such as that it does a good job of capturing the phenomenological appearances in both veridical and hallucinatory experiences. So the sense-datum theory cannot be so easily written off.

However, at the end of the day, my aim is not to argue that the sense-datum theory is true, or that everyone should accept it. Rather, my aim is to argue that externalists in particular should accept the sense-datum theory given the primary motivation for their view. My argument presupposes that externalists are right that sensible qualities seem in experience to be instantiated in mind-independent objects and that we should trust these phenomenological appearances to the extent that we select views about perceptual experience on their basis. But these presuppositions could be wrong. Externalists’ faith in phenomenology could be misguided. Transparency-type considerations could be playing an outsized role in the philosophy of perception. And so if accepting the sense-datum theory is too high a cost for externalists—perhaps because they think the view is crazy or unbelievable—then they should focus on other motivations, or find new ones.