Editors note: Because of a major snowfall in Savannah, GA during the recent Eastern APA, the Blog of the APA will be hosting papers by people who missed the opportunity to present or who would like a larger audience for their work. If you were on the schedule for the Eastern and would like to take advantage of this opportunity, please fill out the submission form.
By Rami Ali
Accepting hallucinatory perception is accepting the view that all visual hallucinations constitutively involve visual perception. [Fn1. In what follows I constrain myself to visual cases and will omit the prefix ‘visual’. This is because of space, and because the literature largely focuses on visual cases. I also understand ‘perceiving’ and ‘perception’ as success terms, and I assume a naive realism/relationalist view of perception. Nothing in the argument rests on the latter assumption. I assume the view because it helpful to focus on a particular view, and naive realism/relationalism is a particularly demanding views of perception, which also stand to benefit from the following argument.] Because contemporary views reject hallucinatory perception, they deny that perception can be fundamental to all visual experience (henceforth, experience). Instead, experiences are construed as fundamentally representational, or fundamentally disjunctive. [Fn2. For instance, see Pautz 2011.] But rejecting hallucinatory perception is a significant conclusion given the impact of misperception on theories of experience. I argue that one central argument for denying the constitutive role of perception in hallucination, focusing on the existence of ‘hard’ hallucinations, fails to motivate a nonperceptual view of hallucinations.
1 Hard Hallucinations
Hallucinations are typically defined as not involving perception:
In illusion, although a physical object appears other than it actually is, that very object is really perceived; in hallucination, “that” physical object does not exist….If you are misperceiving a part of the carpet as a pink rat, we have a case of illusion, not hallucination. (Smith 2002 p.191)
But why accept this definition? One important reason seems to be hallucinatory cases that occur without perception. Philosophers typically distinguish different types of hallucinations, and it is widely agreed that some variants are particularly resistant to a perceptual treatment, often warranting their own solution. [Fn3. For instance, see Smith 2002, Fish 2009, Genone 2014, and Raleigh 2014.] One example are pure cases. Fish (2009) argues that unlike impure cases, pure “hallucinations … take place in the absence of any background experience of the world” [Fn4. Fish 2009 p.93] and “will therefore not have an acquaintance-based phenomenal character.” [Fn5. Fish 2009 p.93]. Another example are total hallucinations, which involve “the hallucination of an entire scene, such that the subject’s experience bears no relation to her actual environment.” [Fn6. Genone 2014 p.360. These cases are opposed to “the hallucination of one or more non-existent objects in an otherwise normally perceived scene”] A third example are perfect or causally-matching hallucinations. Soteriou (2016) writes “It is theoretically possible, by activating some brain processes involved when a subject genuinely perceives the world, to cause a hallucination subjectively indistinguishable from that perception”. [Fn7. Soteriou 2016]
Pure, total, and perfect cases are formulated differently, and so have different extensions. Fish’s (2009) pure cases are cashed out in terms of phenomenal character, [Fn8. Fish understands pure hallucinations as he does because on his view relationalism involves identifying phenomenal character with the property of acquainting the subject with a given object’s presentational character.] but this might be thought too specific, since some views deny that phenomenal character is constituted by particulars.[Fn9. For instance, see Genone 2014 p.343-344, or Schellenberg 2016.] Genone’s total hallucinations, by contrast, are more general in denying any relation. Perfect cases, by contrast, envision a nonperceptual alternative. If internal processes are sufficient for experiences subjectively indistinguishable from perception, then perception of the surroundings is unnecessary. Perfect cases are therefore both broader and narrower than pure and total cases. They are broader because they include any nonperceptual experiences that are subjectively indistinguishable from perception [Fn10. As in Martin’s 2006 and Fish’s 2009 views of hallucination.], but narrower because they exclude hallucinations that are subjectively indistinguishable from perception. [Fn11. For example, hallucinations that possess a character we associate with imagination, memory, or dreams, or that present objects that perception cannot present e.g. contradictory properties, as experienced in the waterfall illusion.]
To avoid these differences, we can stipulate the category of hard hallucinations. Hard cases involve no relation to the surroundings, possess a perceptual (but not necessarily subjectively indistinguishable) character, and do not involve perception in any sense modality. So, unlike pure cases, they do not stipulate a link to phenomenal character. Unlike total cases, they explicitly involve no inputs from any sense-modality (and so exclude the possibility of perception through crossmodal interaction). [Fn12. Examples of interactions between different senses abound in the literature, and such cases might be thought to involve perception through crossmodal interaction. Some cases involve interference between inputs from two functioning sense modalities e.g. the McGurk effect, the ventriloquist effect, and the sound induced flash illusion, amongst others. Others, like some synesthesia cases, involve inputs from one modality being processed in two modalities (e.g. an object heard might be experienced auditorily and visually). Finally, some cases involve the stimulation of one sense modality generating an experience in another. For instance, vestibular stimulation causes amputees to hallucinate phantom limbs (Lopez, C., Schreyer, H.-M., Preuss, N., & Mast, F. W. 2012).] And unlike perfect cases, they are not constrained to subjectively indistinguishable experiences.
2 The Smooth Transition Argument
Watzl (2010) argues for hallucinatory perception using a ‘smooth transition’ argument. He writes
Contemporary philosophers often consider the bad cases in isolation from the good cases. There is a tomato present in the good case, while there is no tomato present in the bad case….But that is a mistake. There is a smooth transition from the good cases to the bad cases, i.e. a continuum of intermediary cases that lead from the good case to every bad case. [One] … would have to find some point in that transition where our perceptual attention ceases to have an object … and where we thus step from attending to something to merely seeming to attend to something. (p.242)
But Watzl thinks such a point cannot be found. Proceeding through four basic scenario types, Watzl argues that no nonarbitrary line can be drawn between adjacent cases: [Fn13. Watzl 2010 p.243]
- Type 1: cases of perceiving a tomato.
- Type 2: cases of perceiving objects that are qualitatively identical to a tomato, ranging from ordinary solid objects (e.g. a wax tomato) to exotic objects (e.g. light projections).
- Type 3: cases of perceiving diffuse objects. Though parts are located in different places, they appear as a single tomato with the help of lenses, prisms, and other light dispersing objects.
- Type 4: cases of perceiving progressively closer stimuli, first close to the retina, then direct stimulations, and finally penetrating the body and becoming direct stimulations of inner parts.
If this is right, the difference must lie in what we perceive, not in whether or not we do. Hallucination are just relations to unusual worldly objects.
Watzl does not directly address hard hallucinations, but he clearly intends his view to apply to them, since no case is excluded from the smooth transition. Plausibly, such cases involve the internal objects stipulated in Type 4 cases. Watzl accepts such objects, writing
…the smooth transition argument shows there is no clear line between what is external to the subject’s body and what is inside her body…It seems completely arbitrary to say that whether there is something you attend to depends on whether photons are, say, randomly created inside or outside this gelatinous body. (By straightforward generalization the same seems to hold of all other lines one might draw). (p.254)
Even if Watzl is right in rejecting that all perceived objects are external, and that what counts as external to the subject’s body is somewhat arbitrary for some range of objects, without more details on the stipulated internal objects, it is hard to evaluate the proposal’s adequacy. This is because internal objects are problematic, prima facie. If hallucinations involve perceiving internal objects, such objects might interfere in perception by occluding external objects, mediating our access to them, or explanatorily screening off their role by sufficiently determining the experience’s phenomenal character.
This is not to say that all hard hallucinations involve appealing to problematic internal objects. Cases that appeal to envatting machinery, evil demons, or other external entities feeding the subject perceptual information can all be construed as involving either unusual external objects (e.g. an envatting apparatus), proximate objects (e.g. light projections on the envatted subject’s retina), and/or superficially internal objects (e.g. electromagnetic objects on the brain’s surface). [Fn14. This solution is also endorsed by Raleigh 2014.] The first two cases involve no internal objects, while the last appeals to well understood internal objects (i.e. external objects that penetrate the body) that are exclusive to envatting scenarios, and so do not affect perception.
Watzl’s internal objects are problematic only in cases with no external input either insider or outside the subject’s body. One example is the ‘chaos’ case envisioned by Chalmers (2005) and elaborated by Raleigh (2014):
Presumably it is in some sense possible…that the photoreceptors in the eye…might begin to fire even though they have received no readily identifiable …[and] such random firings precisely match the pattern of firings that would occur were the subject to see a lemon. But in such a hypothetical case there would be no equivalent of the machine or demon to be a candidate object of visual awareness … by hypothesis, exactly the same neural activity as in a perceptual visual experience but with absolutely no candidate object of awareness in the environment. (p.23-24)
3 Four Cases
To better understand the adequacy of proposing internal objects for cases like the chaos case, we can consider the case of Ocular, a hypothetical, purely visual being otherwise similar to humans. Plausibly, Ocular’s hallucination is a a hard hallucination in these four conditions, since in each it is plausibly not seeing:
- (1) Eyes Shut: Ocular’s eyes are closed. When they are, no light enters.
- (2) Dark Room: Ocular is in a perfectly dark room.
- (3) Blindness: Ocular’s visual system is damaged, it sees nothing.
- (4) Surgery: Ocular’s visual system is surgically removed.
I will argue that contrary to first impressions, hallucinations in (1) and (2) plausibly involve perception, while (4) precludes hallucinating. (3) is underspecified, under some interpretations it is like (1) and (2), in others like (4).
In (1) and (2) Ocular is still a being with sight. It does not lose its vision because it cannot see under current conditions, no more than we do. More controversially, Ocular is perceiving. It sees its dark surroundings, amongst other things. This is clear if note the difference between the absence of perception, where a subject lacks the capacity to perceive, and the perception of absence, where a perceiving subject is related to a ‘null’ stimulus. [Fn15. For instance, see Sorensen 2008.] Ocular’s case in situations (1) and (2) is of the latter type. Ocular does not lack visual capacities, nor are its capacities inactive. It is merely surrounded by darkened objects. In (1) it sees its eyelids, and in (2) it sees the unlit room. In both cases, the surroundings are completely dark, and so indistinguishable from Ocular’s perspective. [Fn16. In general, all things that are entirely unlit look alike, and are thus plausibly indistinguishable.] When we say Ocular ‘cannot see’ in (1) and (2), we mean it cannot discriminate any features of its surroundings except one: that everything is dark.
One might think this is not right, Ocular lacks access to its surrounding. While perceiving absences is an issue that cannot be fully dealt with here, at least three reasons suggest that this objection is wrong. [Fn17. For a detailed discussion and extended defense of perceiving absences, see Sorensen 2008.] First, the distinction between perceiving absence and the absence of perceiving is intuitive, we readily differentiate possessing sight under unfavorable conditions and altogether lacking sight, so it is worth preserving. Second, Ocular’s experience shares features with perception. Perception informs us about our surroundings, makes us sensitive to changes in them, and provides us with the capacity to visually attend to different parts of the surroundings. In (1) and (2), Ocular’s experience fulfills these conditions. Ocular is informed about its surroundings since it is aware they are dark. It is also sensitive to changes in them, since it notices if e.g. its shut eyes in (1) are opened slightly, or a previously occluded pinhole light is revealed in (2). Finally, Ocular can attend to different parts of the surrounding darkness. It can turn its eyes to look at this or that part of its eyelids, or the room. By doing so it may detect variations in the darkness, or notice a dim light in its periphery come into view.
Finally, (1) and (2) are comparable to ‘positive’ stimuli ganzfelds. Consider Ocular in a room designed and lit in such a way that no part looks any more or less white, or textured, than the other. In that room, Ocular’s experience is similar to that of being in a dark room. It sees the room’s uniform whiteness, it is sensitive to changes in the room, and it can attend to different parts of the room, though everything looks the same. The difference lies only in the properties perceived. The idea that we perceive darkness is also not new. In An Essay Concerning Human Understanding (Bk.2 ch.8 sec.3), Locke writes “the idea of black is no less positive in his mind than that of white, however the cause of that colour in the external object may be only a privation.” More recently, Sorensen (2008) and Phillips (2013) have argued for perceiving different absences like silence.
Perceiving absences helps with the appeal to internal objects because of the role absences can play in explaining why internal objects appear in some cases but not others. We can see this by introspecting our own dark experiences. In a wholly dark room, though our surroundings are uniformly dark, our experience of the darkness is not. Attending to the darkness, we see phosphenes across our perceptual field, they move, make patterns, shapes, appear, disappear, and so on. Such stimuli are perceptually salient features of our experience. We can attend to them, describe them, and even manipulate qualities they possess, like their shape and vivacity (e.g. by moving our eyes, putting pressure on them, or staring at lights). Importantly, these phenomena are not exclusive to dark experiences. They are present in ordinary perception. We attend to them more easily when, for instance, we stare at a blue sky, lack sleep, or sit up quickly.
Phosphenes and related phenomena are internal objects, or at least events internal objects undergo. They contribute to explaining hallucinations in three ways. First, they are fleeting in a way ordinary external objects are not which helps explain why many real hallucinations are erratic. Second, they are features of our perceptual systems with makes them readily available as perceptual objects. Third, they are always present, but not always or even usually perceptually salient. The variation in their salience helps explain why they do not interfere with perception. In darkness, their perceptual impact increases. The reason is that in ganzfelds in general, there is little else to attend to. When looking into the world a large number of stimuli compete for our attention, our world is littered with shaped, lit, textured objects that produce heat, sounds, and smells. Each feature competes for attention. Ganzfelds, by contrast, are a perceptual desert. In them we see a uniformly distributed property, only broken by the occasional flickers of our systems. As such, in ganzfelds, it is easier to see these fleeting phenomena. Indeed, we have empirical reason to associate hallucinations with ganzfelds. [Fn18. For instance, see Wackermann, J., Pütz, P., & Allefeld, C. 2008.] Phosphenes and related phenomena therefore do not occlude, mediate, nor screen off the surroundings. They do not occlude the surroundings because their salience is partly dependent on perceiving the surroundings. They do not mediate our perception of objects since they are not usually salient. And they do not explanatorily screen off surroundings’ role, since seeing the surroundings is a precondition for seeing these phenomena.
A helpful way of thinking about this proposal is to combine two metaphors, one from Campbell (2002), and another from Ffytche (2013). Campbell describes brain processing on the ‘pane of glass model’. On this model, the brain calibrates to make the highly volatile glass (consciousness) transparent. When successful, we perceive the world directly. When unsuccessful, we fail to perceive, or perceive features of the nontransparent glass, and the surroundings poorly. [Fn.19 This sort of calibration is apparent if we think about a case of walking into a regularly lit room from the sunlit outdoors. It usually takes our eyes a few seconds to adjust our perception of the room, and prior to that the room looks more dimly lit than it is.] Campbell’s metaphor helps for cases of perception, but Ffytche’s discussion of the psychiatrist Louis Jolyon West suggests an extension to cases of hallucination:
West provided the analogy of a man looking out of a window from a room containing a fire. In bright sunlight (analogous to sensory input), the man sees only the world outside; however, as night begins to fall, the man begins to see things inside the room reflected on the glass. While the fire burns brightly (analogous to cortical arousal), the man sees the contents of the room as if they were outside the window, but when the fire dies down he sees nothing. (p. 155)
This metaphor nicely illustrates how usually non-salient features of experiences might become perceptually salient under certain conditions (in the metaphor, nightfall).
This brings us to the final two cases. Unlike (1) and (2), (4) is not a case of perceiving null stimuli, but a case where perception is lacking. If Ocular’s system is removed, it cannot process anything visually. But plausibly this also impacts its capacity to hallucinate visually. Suppose Ocular reports hallucinating in (4). Should we accept its report? Prima facie, Ocular is barred from visual experiences if it lacks the requisite mechanisms. As such, it is likely confused and not hallucinating. [Fn20. Here I am bracketing the fact that Ocular’s previous experiences might somehow inform it visually e.g. via memory, since we can conceive a case in which Ocular is (unnaturally) born without a visual system.] This is clear if we compare Ocular’s belief that it is hallucinating to a human believing she suffers echolocation hallucinations. If humans lack echolocation systems, intuitively they are barred from echolocation hallucinations. At most one may be deluded into thinking that one suffers such hallucinations. [Fn21. In my view, Fish’s 2009 account of pure hallucinations reduces them to what are more aptly called delusions. Delusions are cognitive rather than perceptual (like hallucinations), though they certainly affect the way we understand and conceptualize our perceptual world.]
This leaves (3). The situation here is harder to assess because it is underspecified. How we should analyze the case depends on the extent and type of damage to Ocular’s system. At one end, the damage will be so pervasive that Ocular’s system fails at any relevant processing. Such cases would be like (4), involving neither perception nor hallucination. At the other end, the damage will be superficial. What Ocular would see may be impoverished, darkened, or unusual (e.g. features of a burned retina). Such cases are like (1) and (2). Between these two extremes we can expect many variations, and it would be difficult to say, in advance of details, whether each case can be dealt with satisfactorily. Nevertheless, it is plausible that all intermediary cases involve the gradual diminishing of discriminatory capacities. As each capacity fails (for instance the capacity to distinguish shapes or faces), Ocular will be able to discriminate less and less, even in its darkened surroundings (e.g. it might cease to see blue phosphenes with the loss of its capacity to discriminate blue). All such cases would be like (1) and (2), they would involve perceiving absences. If some cases differ from these, without being like case (4), the burden is on opponents to produce particular examples.
Understanding cases (1) through (4) helps understand chaos cases. Such cases are just one more case on Watzl’s smooth transition. They differ from others in arising from spontaneous causal replication. But replicating a neural state, even randomly, is just replicating an experience that makes the surroundings transparent to the perceiver, if we recall the Campbell/West metaphor. If the surroundings are a void, then the randomly generated experience is of the surrounding void. In this sense, chaos cases, like (1) and (2), involve perceiving absence.
A final objection is this: chaos cases, it may be thought, do not merely involve dark surroundings. Rather, they involve stipulating the inadequacy of the surroundings. The problem is that this stipulate is implausible. First, it is not clear that this sort is stipulate is part of what we conceive in the chaos case. Second, the stipulation decides an empirical matter apriori. Can human brains provide a full-blown experience through spontaneous activity, without input from the surroundings? This is decided affirmatively by the stipulation. Little supports this empirically. We can generate experiences with the help of external stimuli applied to patients, but this leaves open that we perceive those stimulations.
Rami el Ali and is Assistant Professor and head of the philosophy program at the Lebanese American University. He works in philosophy of mind and in particular perception, but also has research interests in Phenomenology and the Philosophy of Technology. Currently his focus is on the nature of misperception, and in particular hallucinations.