Introduction: The Problem of Perception and the Three Accounts of Perception

Perception is considered as the paradigm way of connecting with the world of objects, events, and properties. The objects, their properties, or the events are taken as mind independent or real. And perception seems to be the way of presenting or acquainting the perceiver with these external objects. At least commonsensically or pre-theoretically, this appears to be how we understand perception. Perception also connects the subject with the world around in a manner in which cognitive mental states like belief cannot. The perceptual presence is characterised by an immediacy and richness of content which beliefs do not have. For instance, one may think about sunrise, but thinking doesn’t present the event of sunrise in the manner in which seeing the sunrise does. The character of a perceptual experience and the character of a belief state are strikingly different in this sense. Furthermore, perception also acts as the ground of many of our beliefs. Perception is characterised by a sense of “being in presence” or “being acquainted to” which makes perceptual connection with the world more intimate and direct and connection through belief. Thus, perception has a crucial role to play regarding our knowledge of the world.

Despite this fundamentality of perception in our relation and knowledge of the world, philosophers have been wary about the nature of perceptual experience and its relation to the real world. Philosophers have pointed to subjective experiences like illusion, hallucination, to caution us against the common-sense trust we have about immediacy of perceptual experience just alluded to in the last paragraph. As subjects we are susceptible to or undergo several forms of subjective experience which appear like perceiving things in the world, but in fact this is not the case. Illusions are a form of experience where there is some mismatch in properties experienced and properties actually present. Like, a straight stick dipped in water appears bent. This means our experience depicts the world to be having a bent stick at a certain place and time even though actually there is a straight stick. Hallucinations are a form of experience where the mismatch between appearance and reality is even more striking. In hallucinations, a subject can have an experience with no corresponding grounding on the real world. For instance, a person under hallucinogenic drugs might experience a pink elephant standing in front of her, even though there is nothing like an elephant in front of her. For the purpose of this paper, we could in principle contrast perception with either illusion or hallucination. However, we will primarily contrast perception with hallucination as hallucinations provide a striking contrast to perception. Also, in this paper we will use the term perception to strictly denote veridical experiences or “good cases”, but we will use perceptual experiences to denote a class of experiences which include perception, illusion, and hallucination. In other words, perceptual experiences could be veridical (i.e. perception) or non-veridical (i.e. illusion or hallucination), but perception only refers to veridical experiences.

The argument from illusion (Ayer 1953, p. 3) and the argument from hallucination (Robinson 1994, p. 87) point out that in non-veridical situations, like illusions and hallucinations, the experiencer is not acquainted with a real object. Furthermore, qualitatively or experientially such non-veridical experiences seem (or can be) like veridical perception. That is, one’s hallucination of an elephant may be qualitatively just like one’s perception of an elephant. Thus, it makes sense to claim that both in case of hallucination and in case of perception one is acquainted with objects of the same kind which explains why they are qualitatively identical in nature. However, as the object in question cannot be real in a non-veridical case like a hallucination, the argument goes, even in case of perception, the object is not real. This has been the way to argue that all perceptual experiences—perception, illusion, or hallucination—are relations to mind-dependent objects or what they are often called sense-data.

If the sense-data theorists are correct, then our common-sense notion of perception is radically mistaken. We do not have direct access to the world of real objects. Rather we, that is each individual perceiver, are directly perceiving one’s own subjective sense-data. Subsequently, difficult questions arise about our relationship with the world. If the objects of perception are not mind independent, then how can we gain knowledge about the real world? Are the objects of perception or the sense-data then copies of real objects? But then as we cannot perceive real objects per se how do we know what we perceive are copies of what we do not perceive? Sense-data theories, in fact, are associated with a twofold problem. On the one hand, it raises sceptical worries about our ability to know the world. And, on the other hand, it puts to doubt the very existence of a mind-independent world. If everything we perceive is only mind-dependent objects, then what good reason do we have to posit a mind-independent reality at all? Maybe, the reality is exhausted by the mind itself.

Given these problems, attempts have been made to establish realism about the object of perception and quell scepticism about our knowledge of the real world. One view, perhaps the mainstream view in philosophy of perception, rejects the sense-data view by introducing a content–object distinction (Searle 2015). The representational view, sometimes referred to as the content view, claims that perceptual experiences are underpinned by perceptual content. For instance, when looking at a white cat, the perceiver’s perceptual system creates a content that “there is a white cat” in front of the perceiver. However, the perceiver’s perceptual system can produce the same content even in the absence of a real cat. As it might do while undergoing a hallucinatory experience. In both cases, the content of the experience is identical. Thus, in both cases, it seems to the perceiver that “there is a cat in front” even though such is true in one case and not so in the other. According to the content view, what explains qualitative commonality of perception and hallucination is then the common kind of content they share. The object, the real mind-independent thing, within the representational theory serves as the condition of satisfaction (Searle 2015, pp. 41, 57). It is that which makes the perceptual experience concerned a case of perception, illusion, or hallucination. Whereas in perception there is an object in the real world as purported by the content, in hallucination there is no object as purported by the content. Unlike the sense-data theories which explain the phenomenal character of experience (qualitative character) in terms of the object or sense-data, the representational view explains the phenomenal character in terms of the content.

The other prominent realist alternative is what is often termed as naïve realism (Martin 2004; Soteriou 2016). The naïve realists claim that they are motivated to defend the common-sense notion of perception. They point out that when we introspect upon experience, it does not seem to reveal any mental objects or mental content. Rather, it continues to reveal the external objects. That is when a perceiver turns his or her natural unreflective directedness from the world towards the experience itself he or she still finds the same external objects. Thus, introspection reveals, claims the naïve realist, that external objects are constitutive of the experience. For the naïve realist then, the trees, clouds, birds that we may see when looking outdoors are not representations of something real, as the representational view would maintain, neither are they mental objects, as the sense-data theorists maintain; rather, these are constitutive of the very experience the experiencer is having (Martin 2002, p. 395).

For the naïve realist perception constitutively involves the object as part of the experience. Furthermore, the external object itself determines the phenomenal character of the experience. According to this view, the phenomenal character of the experience is nothing but the presence of the external world to the subject. However, as hallucinatory experiences lack real objects as constituent of the experience this entails that, according to the naïve realism, hallucinations lack phenomenal character (Martin 2004, p. 39). Thus, hallucinatory experiences are a fundamentally different kind of experience to veridical perception. The difference is not just epistemological, i.e. one is veridical and the other is not, but also metaphysical, i.e. one constitutively includes real objects but the other does not. One may be surprised by the naïve realist claim that non-veridical experiences lack phenomenal character and wonder how they explain the sheer fact of hallucination or illusion. To this, the naïve realist response is that non-veridical experiences are subjectively indistinguishable from a veridical perception. And this negative epistemic property of subjective indistinguishability explains why a hallucinating subject might think or behave as if he or she is perceiving. But the subjective indistinguishability does not entail a common content or object. For instance, the mere fact that the subject cannot distinguish between hallucinating a cat and perceiving a cat doesn’t entail that there is something positive in common in between the two experiences.

Following this debate, in the rest of the paper, our attempt will be to critically evaluate the claims set forward by naïve realism. Firstly, we oppose the claim that naïve realism is a commonsensically attractive position. A philosophical position being commonsensically appealing doesn’t make it correct, but one may hold that it provides it certain merit deserving philosophical attention. Here we will try to show that naïve realism may have certain commonsensical appeal when it comes to how we understand veridical perception, but it is not commonsensically appealing when we consider perceptual experiences—the veridical and non-veridical cases—as a whole. In fact, we will try to show the introspective basis of naïve realism is not genuinely appealing to common sense even in case of veridical perception. Secondly, we will focus on how naive realist account of perceptual experience is too narrow and does not adequately explain the nature of perceptual experience. Having critically examined some of the problems of naïve realism we will try to focus on another aspect related to naïve realism which might appear to be more promising. Alongside its claim of (constitutive) realism naïve realism is also characterised by its notion of primacy of perception over non-veridical perceptual experiences—illusion and hallucination. In this regard, we will highlight the plausible epistemological merit of naïve realism, but also critically examine the notion of primacy to find out whether or to what extent the notion of primacy of perception is tenable. We will use a celebrated thought experiment ingeniously for this purpose. Also we will bring insights from recent works of cognitive science to come to an informed conclusion. Having considered naïve realism from the perspective of introspection as well as the notion of primacy of perception we will end the paper with the conclusion that naïve realism is a problematic thesis regarding perception on several counts and thus may not be the preferred model of understanding perception and perceptual experience.

Introspection and the Problem with Naïve Realism

One of the central ideas of naïve realism is its commitment to common sense. This comes out all the more clearly in one of the most important naïve realists, M. G. F. Martin’s account of perception. Martin’s motivation for holding naïve realism is its appeal to common-sense or naïve notion regarding perception. The naïve view about experience comes about by taking a first-person point of view about experience as the starting point for theorising. Martin thinks that naive realism is “the best articulation of how our experience strikes us as being to introspective reflection on them” (Martin 2004, p. 42). For him, the reason why naïve realism is appealing to common sense is provided by introspective reflection on experience. Introspection serves as the basis for the claim that in perception, the person directly experiences the real object. That is by reflecting upon the experience, the subject only encounters the objects of the world rather than any subjective mental objects or mental content. Thus, what we find here is that naïve realism makes its core claim—that external objects are constitutive of the experience in case of perception—on the basis of introspection.

But here we may point out that introspection is a double-edged sword. Though it appears to present an attractive account of perception, it becomes problematic when non-veridical experiences are taken into consideration. If we apply introspection not just to understand the nature of perception, but also to understand perceptual experience in general, we find that the common-sense understanding that we receive from introspection is saying something in opposition to Martin’s claim. To illustrate the point, suppose we were to introspect upon two experiences which are subjectively completely indistinguishable even though one is a case of perception and the other a case of hallucination. Further, suppose we do not know that one of the experiences is hallucinatory. Now in such a scenario, purely on the basis of introspection if we are asked to decide on the nature of the two experiences—whether they are of the same kind or of different kinds—we should say both these experiences are of the same kind. This is intuitively quite clear after all; introspectively, we lack any mark to tell the difference between the two cases. What this shows is that introspective reflection upon a perception and a subjectively indistinguishable hallucination results in a common-sense claim that perception and hallucination (or perceptual experience in general) are experiences of the same kind. However, this common-sense claim goes against naïve realist view that perception and hallucination are fundamentally different kinds of experience. In other words, introspection upon experiences can result in conflicting notions about the nature of experience. On the one hand, it may be supposed to support the realist claim of naïve realism, but on the other hand, it may be shown to contradict the disjunctive claim of naïve realist claim—that veridical and non-veridical experiences are constitutively different in kind.

What we are trying to emphasise here is that one may (as the naïve realist does) appeal to introspection to justify the claim that objects of perception are external. But that very introspection would also be the basis of the justification that perception and hallucination are experiences of the same kind, a conclusion which is fatal to the naïve realist argument. Further, it is also questionable whether introspection upon veridical perception reveals that the objects are constitutive of the experience. Introspection from naïve perspective should reveal that the concrete object is there in the external world; it is not certain that would reveal that the experience contains the object in any sense. If it is true that by experience we mean that it is something which is ontologically subjective and private, in contrast to the object by which we refer to something ontologically public and objective, then one may hold that introspection does not reveal the ontologically external object as a constituent of ontologically subjective experience.

Another problem with the naïve realist position is in the manner it uses introspection for its explanatory purposes. Firstly, introspection is the basis of the naïve realist claim that in perception the objects are constitutive of the experience. This implies, according to the naïve realist position, introspection has access to ontologically objective, mind-independent reality. Introspection thus, if we were to accept this view, relates the subject with objects in the external world. However, this above naïve realist view doesn’t seem to go well with its position regarding non-veridical experiences like a hallucination. Introspection on a hallucinatory experience cannot reveal the absence of the real object in the experience as for a hallucinatory subject the experience mistakenly seems to be one of being presented with real objects. The question then arises: How come introspection has the ability to detect what is the case in perception but lacks that same ability during hallucination? Or to put it in another fashion, how introspective function of the mind can correctly determine the veridicality status of the experience in perception but cannot do so in hallucination? Why does it fail to determine the absence of object in hallucination?

This problem, however, would not arise if one takes a representational view to perception. This is because representational view explains indistinguishability to something robust in common (Searle 2015, p. 27). The representational view can coherently maintain that the subject cannot distinguish the two experiences as both the experiences are underlined by the same kind of content.Footnote 1 As a result, the question of why introspection cannot detect the absence of object in hallucination is answered adequately. In contrast, the naïve realist position entails that introspection has the ability to determine the presence or absence of object in experience during perception, but lacks this ability to determine the presence or absence of object in hallucination. Thus, naïve realism may be accused of providing equivocal function to introspection. This is fallout of naïve realism’s holding the position that indistinguishability does not entail commonality. Any merit of the epistemic frugality of holding indistinguishability doesn’t entail commonality is thwarted by the problem arising thereof, i.e. the resultant obligation to provide equivocal function to introspection case of perception and hallucination.

Perhaps the problem in naïve realist notion regarding the use of introspection to argue for naïve realism stems from a mistaken notion of introspection. Fred Dretske (1995, pp. 43–44) provides us with an understanding of introspection which shows what is essentially wrong with naïve realism’s treatment of introspection. Dretske, who is one of the major proponents of representational thesis regarding perceptual experience, maintains that perceptual experience is a representation. Further, he also claims introspection is also a representation. However, he highlights the crucial difference between the two forms of representation. Perceptual experiences (both veridical and non-veridical experiences) according to Dretske are systemic representations (Dretske 1995, p. 15). By systemic representation, he means that representational functions of the perceptual system are fixed by the biological functions of the sensory system. The representational function of the sensory system is to represent objective properties such as shape, colour, taste. On the other hand, introspection, Dretske maintains, is an acquired (or conceptual) representation. It is a representation of the mental state concerned. If one is introspecting on perceptual experience, then introspection is representation of the perceptual experience one is undergoing. However, the crucial point to note is that introspection is not a sensory representation whose function is to represent the objective properties of the external world. Introspection does not represent through sensory modalities such as sight, touch, smell. It represents that the subject is experiencing so-and-so. Introspecting upon experience is conceptually (doxastically) representing that one is experiencing so-and-so. That is, it is a conceptual representation of a sensory representation (experience). This shows that introspection is not an experience. Introspection has no direct representational relation with the external objects. Rather, introspection represents that one is having certain experience, or thought, etc. And this also why introspection cannot distinguish between veridical and non-veridical experiences, i.e. the presence or absence of real object. Perceptual experience when veridical, i.e. perception, represents that so-and-so is the case, say, there is a cat sitting on a mat. Likewise, perceptual experience when non-veridical, i.e. hallucination, can represent similarly, that is there is a cat sitting on a mat. Now introspection on both the above-mentioned veridical and non-veridical experiences would yield the same representation. In neither case does introspection per se relates the subject to the external objects. It only represents (in doxastic, non-sensory terms) that, “I am perceptually experiencing that there is a cat sitting on a mat”. Introspection, to emphasise, is not experience which can represent or relate to external objects. It rather can represent what some other mental states, like perceptual experience, thoughts, moods, etc., the subject is undergoing. Once we understand introspection in this light we can avoid the equivocal use of introspection which naïve realism commits. The question raised against naïve realism, in the last paragraph, why introspection can detect the presence or absence of object in perception but not in hallucination, thus cannot be raised if we accept representational view about perception. A representationalist maintains that both the veridical and non-veridical experiences are underpinned by the same kind of content. As a result, introspection which only represents the mental content available to it represents that one is experiencing so-and-so.

We have mentioned earlier the naïve realist also maintains the unusual claim that hallucinatory experiences lack phenomenal character. Now one may wonder on what basis the naïve realist makes this claim. It cannot be after all on the basis of introspection. After all, introspection on non-veridical experiences does not provide any discernible difference from veridical experience. It could only mean that the naïve realist has a preconceived notion about phenomenal character—that it is the presence of the real object itself (Martin 2004, p. 83). This is an unusual claim to say the least. Later in this paper, we will briefly point out why it does not go well with scientific understanding on perceptual experience. But here we can point out a strange philosophical paradox that results from accepting this view of naïve realism. If we accept that hallucinations, lacking relation with the world, are not phenomenal experiences, then the subject during hallucination is a philosophical zombie—a subject having beliefs and judgements nevertheless lacking consciousness. In fact, Martin admits that his position amounts to accepting hallucinating subjects are philosophical zombies (Martin 2004, p. 83). But consciousness as we know is the “what is it like to” character of experience (Nagel 1974). And it seems too strange to believe that hallucinating subjects lack the subjective feel associated with consciousness. At least typically speaking a subject describing hallucination doesn’t report her experiences to lack qualitative feel or “what is it like to” aspect of experience. Thus, we see here the naïve realist position that subjective indistinguishability does not entail commonality may be logically tenable, but when we try to see the full implications of this position, we encounter philosophically problematic or even absurd positions. In contrast, the intuition of representational theory (and sense-data) which believes subjective indistinguishability is underpinned by commonality appears to have a better tenability. Representational view takes something positive in common between the case of hallucination and perception to explain why the qualitative character of the experiences are the same and hence doesn’t entail the flabbergasting conclusion that hallucinating states of mind lack consciousness.

Moreover, introspection, it might be argued, rather serves the basis for the claim that hallucinations present one with a phenomenal experience. If we agree, as do both representationalists and naïve realists, that the objective difference between perception and hallucination lies in the external world, then it makes sense to claim that what explains the commonality or what explains the lack of distinguishability is the common phenomenal character. Suppose one hallucinates that “a cat is sitting on a mat”. The subject may mistakenly take this experience to be veridical, or the subject may be knowing that this is only a hallucination. But either way, the subject experiences something with a phenomenal character—say that of a white cat sitting on a red mat. This phenomenal commonality, after all, is why the subject is introspectively unable to distinguish this case from a veridical case. When the hallucinating subject comes to know that she is or was hallucinating—a white cat sitting on a red mat, for instance—typically she is surprised. She is surprised because the experiences phenomenally seemed the same to her. It seemed there were a white cat of a certain size and a red mat of a certain texture present before her. The phenomenal character of her experience representing “whiteness” or “redness” or the shapes and the textures match with a veridical experience. And thus she is unable to introspectively distinguish her hallucination from an actual perception with the same content.

Empirical studies on patients with Bonnet syndrome—patients with very little or no vision—show that such patients are capable of having a wide range of hallucinatory experiences (Searle 2015, p. 165). Such patients do not report to be having phenomenally absent experiences. If anything they report having experiences often richer than their actual perceptions (when they were endowed with vision). It seems mistaken to consider that such experiences are non-phenomenal. Surely, such experiences are non-veridical. After all such patients report experiencing things which are not there and such patients as we have mentioned lack vision. However, the phenomenal quality of the experience is not compromised by the non-veridicality of the experience. Whether it is the case of a subject with otherwise normal visual ability or a person suffering from Bonnet syndrome a person when hallucinating seems to be having phenomenally rich experiences in the absence of real objects. Real objects, we may maintain here, do not provide phenomenal character to experiences. Real objects, however, do provide an important role in determining veridicality condition of the experience. They serve as a necessary condition to veridicality of the experience.

Here we may provide a short summary of our position so far. We have raised questions regarding the use of introspection to establish the naïve realists’ thesis. Firstly, introspection is knowledge of ones experience, but it is not being aware of the presence of external world per se. However, the naïve realist accepts that introspection can be the basis of the claim that external objects are constitutive of experience. Furthermore, we have tried to show how this naïve position sits uneasily with naïve realists’ view regarding non-veridical experiences like illusion or hallucination. The naïve realist is forced to maintain that introspective ability differs between a case of veridical perception and hallucination (or illusion). In the former case, it successfully and even paradigmatically picks up the real objects, whereas, in the latter case, it fails to detect the absence of the objects. Moving ahead, the naïve realist also maintains that hallucinations lack any phenomenal character. Presumably, the naïve realist has to accept this strong unintuitive position to hold on to metaphysical disjunctivism—that is, veridical perceptual experience shares nothing robust in common with non-veridical perception. However, this position appears both introspectively and scientifically unattractive. Rather, it seems reasonable to believe that hallucination has phenomenal character given what we know and understand about hallucination.

Analysis of the Supposed Primacy of Perception

We have seen that the sense-data theory and the representational theory about perceptual experience share the view that veridical and non-veridical cases are underpinned by something robust in common between them.Footnote 2 Common object (for sense-data views) or common content (representational views) explain the primary characteristic of experience. The nature of the experience per se, notwithstanding its relation with the external world, is the same in case of both perception and hallucination. It means that both perception and hallucination are provided explanatory treatment of the same nature. One of the guiding motivations behind naive realism is to avoid giving the same kind of explanation for both kinds of phenomena—perception and hallucination. The naive realist believes we need not take the two phenomena as equal or provide them equal explanatory treatment. Rather, there is a case for thinking perception is the norm, the primary, the paradigm, whereas hallucination is the exception, secondary, or subordinate in some sense. The idea may be that, if it is true that the fundamental function of experience is to connect the subject with the world outside, then a form of experience which fails to do that (hallucination) but only misleadingly appears to be doing that cannot be considered in the same breath as a form of experience which is doing its function properly (perception). From a naïve realist perspective, this is a crucial intuition which the representationalist and the sense-data theorist fail to take account of. As a result, these theories are compelled to give perception similar explanatory treatment as it does to hallucination. This has the unfortunate consequence of depriving perception of what seems fundamental to it—the immediate presence of external objects as constitutive of the experience. By allowing hallucination the same explanatory space as perception, the representationalists (and the sense-data theorist) have to settle for compromises in explaining perception. The sense-data theorists have to accept the uneasy view of “mental objects” which might further snowball into some form of Berkeleyan idealism, whereas the representationalists, as the naive realist may allege, have to artificially bring in a new category that of experiential content, over and above the object of perception. These experiential manoeuvres naïve realists allege bring in the so-called veil of perception (Martin 2002, p. 396) between the perceiver and the world, the “veil” being mental objects for sense-data theories and “content” for representational theories which restrict the immediacy between perceiver and the object in the real world.

The naive realist seeks a new way out which avoids creating such artificial conceptual constructs like “mental objects” or “mental content” to explain perception—the paradigmatic relation of the subject to the world. Naïve realists refrain from giving positive characterisation for hallucination (or illusion) as well as for perceptual experience in general. Naive realism only provides a positive characterisation of perception as a form of experience which is fundamentally constituted by the external objects. In contrast, for perceptual experience in general, it simply provides a negative characterisation that perceptual experience is subjectively indistinguishable to perception. Thus, there is no robust object or content that perception shares with other forms of perceptual experience; only aspect of commonality between the perceptual experiences is the negative epistemic property of indistinguishability. This manner of characterisation ensures explanatory priority in favour of perception. Perception has the metaphysical property of constituting real objects as constitutive of the experience itself. But indistinguishable hallucination does not share such metaphysical status. Lacking objects as constituents, a hallucinatory experience is characterised only in terms of perception—that it is indistinguishable from perception. Commentators have described this position as a view in which non-veridical experiences are seen as being “parasitic” to perception (Soteriou 2016, p. 170). Our attempt will now be to critically examine this important naive realist notion of primacy of perception and see how it holds up to critical evaluation.

We have already discussed some of the problems associated with the naïve realist view. There is also an additional criticism which can be levelled against the “parasitic” conception just described. Susanna Siegel has criticised this frugal manner of describing perceptual experiences. According to Siegel indistinguishability with veridical perception is not sufficient to characterise perceptual experience in general (Siegel 2004, p. 93). Siegel considers certain cases of non-veridical experiences where the notion of indistinguishability with veridical perception does not work. Siegel gives the example of Müller-Lyer lines to explain the above point. In Müller-Lyer optical illusion we have two lines of equal length having “shafts” of arrows in tail-ends of each line. The “shafts” in one of the lines are pointed in an inward direction (like a head of an arrow), whereas the “shafts” in the other line is pointed outwards (like the tail of an arrow). Now despite the fact that the two lines are equal in length, on seeing the lines, subjects undergo an optical illusion where one line (the one with outward pointing shafts) looks longer than the other one (the inward pointing one). Now, Siegel says we cannot characterise this illusory perceptual experience merely on the basis of indistinguishability with a veridical perception. This is so because it seems there is no possible veridical experience to which such non-veridical experience is indistinguishable to. After all, how do we draw two lines, such that the veridical perception of these two lines becomes indistinguishable with the illusory perceptual experience of Müller-Lyer lines?Footnote 3 If Siegel is right, then the whole naïve realist enterprise of defining non-veridical experiences only in terms of indistinguishability, or the “parasitic” notion of perceptual experiences may be mistaken and thus provide us one more reason to not accept naïve realism.

However, even if the “parasitic” notion is too narrow and fails to provide sufficient characterisation of perceptual experiences in general—perception, illusion, and hallucination—the notion of explanatory priority of perception has a certain ring to it if we see the matter from another point of view. Generally speaking, our descriptions of hallucinations (and illusions) are cases where we mistake such experiential content for actual perception, where there are actual corresponding objects and/or properties in the real world. In so far as this idea is acceptable we may tolerate a certain priority of the real world and thus veridical perception over other non-veridical perceptual experiences. The idea is without a real world there could be no veridical perception and without perception, there could be no hallucination (or illusion). Perception thus has a primacy as it can take place without hallucination (or illusion), but hallucination may not occur without the ability to perceive. Figuratively, one may say perception provides the subject with a “primal touch” of reality without which purely mental creations as hallucination or imagination cannot exist. One may here consider the case of patients with Bonnet syndrome, mentioned earlier, and wonder that such patients do hallucinate even in the absence of the ability to perceive. But it is possible to reply saying that such patients had undergone perception in past. Even if they are no more capable of perceiving, they still retain the capability to undergo perceptual experiences in virtue of world relating veridical perceptual experiences of their past. Thus, the conjecture that perception of the real world has a certain priority in explaining the perceptual experience, in general, is not troubled by the facts of Bonnet syndrome.

Here we may consider a thought experiment of a congenitally blind person who has never perceived and thus lacks the so-called primal touch of perception. The relevant question is: Can such a person hallucinate? Can a person who has never perceived undergo experiences with the phenomenal character? Or consider Frank Jackson’s famous knowledge argument scenario (Jackson 1982, p. 130; Jackson 1986, p. 291) where Mary is a brilliant scientist who knows everything about colours but has never seen colour in life as she has lived all her life in a non-coloured environment. Jackson’s famous question is: When Mary encounters a red object for the first time after coming out of that restrictive environment, will she not know something new? However, we can ask here a different philosophical question, couldn’t Mary hallucinate a red tomato (for instance) when she was in her black-and-white environment and had never encountered anything coloured? If she couldn’t, then it would show that indeed perception is primary and without a primal touch of reality there could be no hallucinatory perceptual experience of a red tomato, for instance. If a congenitally blind person necessarily lacks visual qualia in her experiences or if Mary cannot hallucinate a red object having never seen anything coloured, then it would appear that perception has a primacy with regard to the phenomenal character of experience.Footnote 4

To begin with, there seems to be no logical impossibility regarding the plausibility of a congenitally blind person hallucinating visual qualia or in Mary hallucinating a red tomato despite never having seen a red tomato. It is definitely possible to conceive such a scenario, and it may even be possible for neuroscientists (perhaps in future) to produce visual hallucinations even in a congenitally blind person. Thus, the supposed primacy of perception with regard to phenomenology is not metaphysically necessary. Perhaps the supposed primacy may be better understood in terms of perception being the norm and non-veridical experiences like illusion or hallucination being the deviations or exceptions of the norm. Normally, corresponding to our perceptual experiences there are actual objects in the reality. And thus in having a perceptual experience we are justified in taking the experience at face value (Austin 1962, p. 115). This common-sense notion of the primacy of perception may be acceptable, but as we have seen, it cannot be used to make strong metaphysical claims as has been done by the naïve realists. A further possibility for the primacy claim may open up if we consider the evolutionary and biological aspects of the perceptual experience. If it is true that perception evolved out of an evolutionary need to track reality, then it makes sense to consider hallucinatory experiences which fail to track reality as “misfires” of the perceptual ability. Within this framework, the biological function of our perceptual apparatus is to track reality which is provided by veridical perception. Hallucinations being “misfires” are thus explanatorily secondary to perception. There is much to appreciate about this framework of understanding perception, but in this paper, we will not be able to go into further details.

Before concluding the paper, we need to discuss a recent development in the understanding of perception from the viewpoint of cognitive science. Using this new way of conceiving perception, we would sketch out a further problem for naïve realism. Perceptual experience is often seen as a causal relation to real objects. In this understanding, light rays reflecting from distal objects like a chair or a table enter the eye and commence the perceptual process. The rod and cone cells create electric signals and encode the retinal image and send it to the brain through the optic nerve. The brain detects patterns from the sent electrical signal and constructs a perceptual scene—a rich phenomenal scene with colours, shapes, and so on. This model of perception has been called “passive accumulation model” where the brain reconstructs a passively obtained retinal signals into an experience (Clark 2013, p. 470). Within this framework, non-veridical experiences like hallucinations could be seen as unusual creations of the brain. Some well-known philosophers and cognitive scientists (Clark 2013; Seth 2015) are trying to upturn this picture of perceptual experience in which hallucination is seen as anomalous perceptual experience. Instead, they have argued that perception should be better seen as “controlled hallucination” (Clark 2013, p. 494). These cognitive scientists maintain that perception is an instance of predictive processing where the brain predicts a perceptual scene using top-down information like prior beliefs and memories. In a normal scenario of a veridical perception, such predictions are corroborated by incoming sensory signals such that any deviations or mistaken prediction is adjusted in response to contradictory sensory signals. Perception in this view is a continuous hallucination which the person undergoes. However, such “hallucinations” are controlled by sensory signals entering the sense organs, like eyes. Thus, sensory signals act as a check and a basis for continuous update and subsequent better prediction of the scene of perception. Thus, hallucination, according to this predictive processing model is not an anomaly among perceptual experiences. Rather, such top-down influence is marked across both veridical and non-veridical experiences. What distinguishes veridical perception from hallucination, in this understanding, is how the experience is shaped by the incoming sensory signals in case of perception whereas how the brain has a free reign, so to speak, in case of hallucination in creating phenomenal experiences. Thus, a case of pure hallucination from a case of perception is different only in that in one case (perception) the hallucination is controlled and adjusted according to incoming sensory information, whereas in the other case (hallucination) no such bottom-up corroboration occurs.

If this predictive model of perception as “controlled hallucination” is correct, then it provides a fresh impetus to our criticism of naïve realism. According to this increasingly common view in cognitive sciences (Clark 2013; Rao and Ballard 1999; Kveraga et al. 2007; Seth 2015), both veridical (perception) and non-veridical (hallucination) cases use active generative top-down models to create the perceptual experience.Footnote 5 The philosophical significance of this model derives from the fact that both veridical and non-veridical experiences use similar top-down mechanism to produce perceptual phenomenology. As similar top-down mechanism is responsible for phenomenology, it makes sense to argue that veridical and non-veridical experiences do not differ metaphysically in so far as perceptual phenomenology is concerned. If this viewpoint is correct, then once again naïve realism is wrong to consider that perceptual experiences are a disjunction between two metaphysically different forms of experiences—veridical experiences (perception) and non-veridical experiences (illusion and hallucination).

Furthermore, naïve realism is wrong in thinking that hallucinations are to be explained in terms of perception, i.e. hallucinations are parasitic on perception. From the naïve realist perspective, hallucinations lack any positive characteristic. What allows hallucinations to qualify as perceptual experiences is the epistemologically negative characteristic that hallucinations are indistinguishable from perception. However, if we accept the generative model of perceptual experiences, discussed above, we find something quite different. Hallucinations can actually be given a positive characterisation. They involve generating top-down models which result in perceptual phenomenology. Indeed, this characteristic is also shared by perception. Perception too involves generating top-down models to produce perceptual phenomenology. What distinguishes perception from hallucination is that only in case of the former sensory signals act as corrective data on the basis of which top-down generative models created by the brain are tweaked. Thus, perception is a form of controlled or, perhaps, corrected hallucination. This model, far from understanding hallucinations as parasitic, on perception in fact invites us to consider perception in terms of hallucination.

To avoid any confusion it may be noted here that the claim that perception is to be understood in terms of hallucination is only restricted to perceptual phenomenology. There is indeed a categorical difference between perception and hallucination in terms of veridicality. The former is veridical, while the latter is not, which in turn means the former is related to real world, while the latter is not. However, the emphasis of the claim that “perception is controlled hallucination” lies in the nature of perceptual phenomenology. It means that the phenomenal character of perception is produced due to top-down generative models of the brain just as it is so produced in hallucination. That is perceptual phenomenology even in veridical cases is not constituted by the real objects (as naïve realism claims), but is generated by the brain. And it is in this sense we have to consider the naïve realist claim of difference between perception and hallucination with regards to phenomenology is problematic.

Conclusion

We have critically discussed the problems of a naïve realist conception of perception and perceptual experience. One major issue is that the naïve realist claims that introspection is the basis for naïve realism. The naïve realist thinks that introspection shows that external objects are immediately present and constitutive of the experience when one is undergoing perception. However, we have argued here that introspection is not the kind of faculty on the basis of which realist claim should be made. Introspection is perhaps best seen as representation of experience (Dretske 1995). Thus, it is not clear why introspection should be considered as the ground for the naïve claim that external objects are constitutive of the experience in perception. Furthermore, even if we were to grant this naïve realist claim, there are further problems in store. The naïve realist takes the unusual position of fundamentally separating perception from hallucination (and illusion) in kind. However, a critic of naïve realism may well point out that introspection itself may be the basis of the claim that there is something in common between perception and hallucination. We have also been critical of how the naïve realist uses introspection in its explanatory project. While, on the one hand, introspection is supposed to reveal the immediacy of the object of experience in perception, but on the other hand, introspection is unable to serve in any positive explanatory role in case of hallucination. We also find that the naïve realists claim that hallucinations lack phenomenal character highly improbable given what we know about hallucinations.

We have seen that current understanding in cognitive sciences also seems to deny the naïve realist claim that there is nothing substantial in common between perception and hallucination (or illusion). Scientists increasingly see perception not as a passive but as an active process, where the brain actively predicts reality based on the available sensory and non-sensory information. The active role of the brain in creating a perceptual scene shows that perception is not unlike hallucination in kind; rather, it is a special form of hallucination. It is a form of hallucination which is controlled by the incoming sensory signals. Thus, the naïve realist manner of metaphysically separating perception and hallucination may not behold if our current understanding of the perceptual process is correct.

Given the problem with naïve realism, it seems representative realism is the right way to understand the nature of experience as well as a perceiver’s relation with real objects. Perceptual experience purports to present the real world around the subject. The world as represented in perceptual experience is teaming with qualities such as shapes, sizes, colours, odours, tastes. These qualities are purported to be belonging to the real world. However, this manner of representing the world is dependent upon the sensory modes available to the subject. Given the perceptual endowments, an organism is supposed to represent the world in a certain manner if it is to veridically represent the world. However, just as speech can “misfire” (Austin 1962, p. 16) similarly experience can also be hallucinatory (or illusory). Experience may purport the environment to be a certain way—having certain colours, shapes, etc.—even though that may be completely hallucinatory. Thus, experiences can go wrong or can be non-veridical. However, the non-veridicality or veridicality of experience doesn’t make it qualitatively different from experience per se. In both cases, the experience represents the world or the environment to be of a certain way. If there is any merit to the notion of the primacy of perception, I think, it needs to be appreciated in this context. Perceptual experience purports to represent the external world to the subject. Thus, whether experience succeeds and represents veridically, or fails and represents non-veridically, it functions to represent an external environment, in a certain manner, to the subject. To the extent, it holds that experience is in the business of representing the external world to the subject; it may make sense to think the perception is primary among perceptual experiences. If all perceptual experiences purport to the represent the world, then in that sense all perceptual experience “attempts” to the veridical or attempts to be perceived. This feature of experience can be compared with belief. Beliefs represent reality to be of a certain way. Even though beliefs can be both true and false, beliefs are propositional attitude where something is regarded as true. Similarly, experiences, whether veridical or non-veridical, purport to be veridical perception. However, even if this weaker notion of the primacy of perception is true, it doesn’t show by any means that as experience per se, perception and hallucination are different in kind. It is needless to say that this weaker notion of primacy does not prove the naïve realist notion that hallucination lacks phenomenal character. Thus, we may conclude that naïve notion of realism about perception is too strong and difficult to accept in the light of this discussion.