Get Complete Project Material File(s) Now! »
The Argument from Hallucination
In the previous chapter we have seen that, despite its initial appeal, relationalism is often considered a non-starter in virtue of what Valberg calls the ‘problematic reasoning’, featuring one or another version of the argument from illusion or hallucination. Towards the end of the last chapter I observed that, although the argument has been presented in various forms throughout several centuries, it is often difficult to understand precisely what the premises involved are supposed to be and what form the reasoning takes, so much so that Dummett suggested that the so-called ‘argument’ is not an argument at all, but rather a starting point indicating the need to explain how illusions are possible, how it is possible for things to appear other than they are.18
However, although often incomplete and unsatisfactory, reconstructions of the argument abound both among early modern proponents of mental representations (Hume 1748) and the sense datum theorists (Russell 1912, Ayer 1940, and more recently Robinson 1994 and Foster 2000), and among those who have reconstructed the argument in order to respond to it (Snowdon 1992; 2005a; ; French and Walters forthcoming among others).
In this chapter I will unpack the argument, by spelling out its structure and its not always explicit premises. As we have seen in the previous chapter, the argument has two steps: a base case and a spreading step. Those steps need to be carefully examined separately. First, I will ask whether we can establish that, in some cases, a relationalist account cannot be true, and if so, what the cases are for which this can be established. Then I will consider the arguments put forward to spread this conclusion to all cases.
Before starting assessing the argument, a clarification is in order. Traditionally, the argument was not directed against relationalism, rather direct realism. Because of the prevalence of epistemological concerns in those debates, asking whether perception really puts us in contact with the objects that we take to exist in the world around us had priority over the question of whether objects play a role in determining the phenomenal character of perception. This is not to say that the two questions are not connected, and that concerns about the characterization of the conscious character of experience did not feature in these debates. Early sense-datum theorists, at least, were very concerned with how experience strikes us, as my discussion in the first chapter, largely indebted to Moore, shows. However, the argument can be turned against relationalism. If we are not directly aware of physical objects, it cannot be the case that the phenomenal character of perception is constituted by those objects. This explains why the relationalist account of experience failed to emerge until very recently, when many objections to this traditional argument from illusion emerged.19
Moreover, the argument didn’t only aim to show that we do not directly or immediately perceive material or physical objects around us. It also concluded that instead, we are only aware of mental representations of them (Hume) or sense-data (Moore, Russell, Broad, Price). So there is a negative and a positive conclusion to the argument. The negative conclusion claims that the direct object of perception is not what we initially take it to be. The positive conclusion claims that what we are in fact related to are sense-data or mental representations. For our present purposes, we are merely interested in the negative conclusion.
The aim of this chapter is to formulate a compelling argument that targets specifically relationalism. Unpacking the structure of the argument to make it valid and unpacking what underwrites the different assumptions will require a considerable amount of exegesis. However, I will conclude the chapter with a formulation that, it seems to me, seriously threatens to overturn relationalism. Whether the assumptions involved really are to be accepted will be a question for the rest of this thesis.
The Base Case: Conflicting Appearances, Illusions and
The first part of the argument seeks to establish that, for some base case, the phenomenal character of experience cannot be accounted for in terms of the obtaining of a psychological relation to the objects we seem to perceive. Traditional presentations of the argument lump together different cases, to which the term of ‘illusion’ is often applied indiscriminately: (1) cases of conflicting appearances, (2) optical illusions, and (3) hallucinations.
I will consider these three groups separately and examine how, for each case, the argument attempts to demonstrate that relationalism is untenable.
Conflicting appearances and Paradigmatic Looks
Most of the examples featuring in presentations of the argument from illusion by sense-datum theorists are cases of conflicting appearances: cases in which an object appears to have different properties to different subjects or to the same subjects in different conditions. These experiences are very common and involve ordinary optical phenomena such as refraction, perspectival changes, and changes in lighting conditions. For instance, in his presentation of the argument from illusion, Ayer mentions: a coin which looks circular from one point of view and elliptical from another; a straight stick which appears bent when it is seen in water; mirror images; mirages; change in the experienced taste of food due to the condition of the palate; subjective change in perceived temperature; and a coin which seems larger when it is placed on the tongue than when it is held in the palm of the hand : 3).
Hume also presents the argument from illusion using a case of perspectival change:
The table, which we see, seems to diminish, as we remove farther from it: but the real table, which exists independent of us, suffers no alteration: it was, therefore, nothing but its image, which was present to the mind. (Hume 1748: 201)
For simplicity, and to avoid problems that are not central to the argument from illusion, we may replace Hume’s dynamic property with a static property, such as having a certain colour.20 So we can use the example of seeing a white table under a light that makes it appear yellow. I will also focus only on the negative conclusion that we do not perceive the table, disregarding the positive proposal that Hume aims to establish: i.e. that what we perceive instead is a mental image. This gives us the following, very simple, argument (see Snowdon 1992: 70):
1. Whatever I perceive is yellow
2. The physical table is not yellow
3. (from 1 and 2) What I perceive is different from the physical table. Robinson claims that the argument is valid, as the conclusion
follows from the two premises in virtue of Leibniz’s Law (Robinson 1994: 32). However, in a recent paper, French and Walters (forthcoming), drawing on a remark by (Snowdon 1992: 74), argue that the argument, as it is presented, is invalid. The conclusion doesn’t follow from the two premises only in virtue of Leibniz’s law. What follows from Leibniz’s Law is simply that the yellow object is not the physical table, but this is compatible with one seeing the yellow object and the table. In order for the argument to be valid, one needs to add a further premise, which Snowdon calls the ‘Uniqueness Assumption’. The ‘Uniqueness Assumption’ claims that:
(UA): ‘there is, in a particular direction of attention, as it were, a unique, single, [directly] perceivable thing’ (Snowdon 1992: 74).
Snowdon finds the principle quite plausible, but French and Walters argue that the Uniqueness Principle, as it stands, is clearly false, because it entails that if one seems to see a wall behind a table that appears yellow when it is in fact white, one should conclude that one doesn’t see the wall either, which seems hard to defend. The Uniqueness Principle also claims more than what the base case aims, and needs, to establish, for the base case seeks only to establish that one is not aware of the object one putatively sees as yellow (the physical table), not that one is not aware of any physical object. Craig and Walters replace the Uniqueness Principle with a principle in the vicinity, the ‘Exclusion Principle’: (EH): If in an illusion S is aware of an object which is non-identical to the ordinary object O that the subject is putatively perceiving in an illusory way, then S is not aware of the ordinary object O.
Contrary to the Uniqueness Principle, this principle is not obviously false, but-—Craig and Walters argue—it is not obviously true either. They consider possible justifications for the Exclusion Principle, including the idea that the object one directly perceives spatially excludes or occludes the physical object, but they conclude that none of these reasons is conclusive.21 In this context we don’t need to establish whether the ‘Exclusion Principle’ should be accepted. It is enough to notice that the argument, as it is generally presented, is invalid, and that the further premise that needs to be added is far from obvious and a satisfactory justification for it has yet to be provided. Due to the widespread failure to recognise the need of such a further premise little or no attempt has been made to justify it.
Even if one accepts the Exclusion Principle, one might still take issue with the first premise. It is not obvious—a relationalist may say—that the first premise is true. It is not clear that, when I look at the white table showered with yellow light, I see something yellow, and I do not see something white. Austin (1962) famously pushed this line. He argued that describing phenomena of refraction, perspectival changes and changes in illumination as ‘illusion’ is misleading. The ordinary notion of ‘illusion’ suggests the idea of being mislead by experience,22 while most of these cases are ‘too familiar’ (Austin 1962: 3-26) to be misleading. Beyond the critique of the terminological choice, there is a substantive point in Austin’s critique. The argument describes the experience of seeing the white table under a yellow light as a case where one sees a yellow table and doesn’t see a white table. But this is not how things look to us. This description is misleading, as it ignores completely the very well established phenomenon of perceptual constancy. Throughout the day, I see my white desk under all sorts of lighting conditions: under the bright white daylight, or under a yellow light of my desk lamp, in the dusk, and so on. On all occasions, the table looks white to me. Also, when the table in unevenly illuminated, all parts of it still look white. Additionally, it is a well-known fact that coloured light casts shadows of a complementary colour.
The same phenomenon of perceptual constancy applies to many other properties: size, shape, volume of sounds, etc. Consider the example used by Hume. When I look at the table from different distances, the table appears the same size, even though the retinal image of the table varies in size as I move back and forth. It doesn’t look as if the table changes size as I move around it. If it did, the world would appear a buzzing, blooming, incoherent flux of ever changing colours, shapes and sounds, and it wouldn’t seem to us in the first place, as Hume admits, that we see a table, an object that seems to be independent of our perceiving it and which doesn’t constantly change size. So there wouldn’t be a tension between how experience strikes us—as an experience of the table, which doesn’t constantly change size—and what we learn about it after reflection on perspectival variations—that it is an experience of an ever-changing storm of properties, something different from the table we seem to perceive.
Thus, relationalists claim that they can provide an alternative and better account of these experiences than that suggested by the proponents of the argument. Brewer, for instance, claims that perceptual experience is a triadic relation where the third relatum is made up by the circumstances involved in experience, including the subject’s point of view. As we have seen in the previous chapter, this is a very widespread way of conceiving the relation in place in perception among relationalists. For instance, although Campbell (2002) doesn’t explicitly engage with the argument from illusion, he equally commits to a similar triadic relation. This triadic relation accounts for the fact that in some sense the table appears yellow, and still appears white in some other sense. The way the table appears is a function of the colour property possessed by the object, the colour of the light, and my position relative to it.
Brewer’s account of illusions builds on the idea that there are looks that are attributed to objects relatively to the conditions of observation.
When one experiences a table as yellow, when it is in fact white, the table one is perceptually related to presents the look of paradigmatic yellow things, given one’s viewpoint and relevant circumstances. A white ball in red lighting conditions looks red because it looks the way paradigmatic red objects look in ordinary conditions. A straight stick halfway submerged in water looks crooked because it looks the way paradigmatic crooked objects look in ordinary conditions (Brewer 2008; Brewer 2011).
Martin gives a similar account of the immersed stick that looks bent:
In a circumstance in which one did confront a bent stick in good lighting, a paradigmatic circumstance for encountering the shape of being bent, one would be inclined to recognize the object as being bent; it would strike one as similar to bent things, and one would find in it an obvious similarity with bent things and a contrast with other shapes. If the psychological situation a subject is in when he or she truly utters [‘The stick looks bent to me’] is relevantly similar to this paradigm kind of circumstance, then the subject is inclined to find the shape before one as similar to the paradigm of bent things—as more like being bent than anything else. (Martin 2010: 214-215)
For Martin, when we see a straight stick submerged in water, the visually basic properties (what he calls its look) that the stick really possesses are manifest to us. The stick looks bent because, in the current circumstances (when the stick is half-submerged in water), the visual properties that the stick possesses are similar to the characteristic look of 61 bent things. The stick looks like straight sticks look in water, which is in many respects like bent things look. However, they do not look exactly like typically bent things.
Optical Illusions and the Phenomenal Principle
At this point the proponent of the argument from illusion may concede that the cases presented above are not cases of illusion, and that by saying that one doesn’t see the whiteness of the table, but sees the instantiation of yellowness, one misdescribes the phenomenon. However, there are genuine cases of illusion, cases where something really looks to have a property it doesn’t have and doesn’t seem to have the properties it actually has. This is something even Austin acknowledges: there are genuine cases of illusions, such as optical illusions (Austin 1962: 22). These cases might involve the same kind of phenomena involved in the cases mentioned above, such as perspectival change, but are situations artfully created in order to mislead the observer. An example is the class of anamorphic trompe-l’oeil: anamorphic bidimensional pictures that, seen from a certain angle, look like tridimensional objects. Another example is the class of anamorphic sculptures, where a tridimensional object looks to have a certain shape that in fact it doesn’t have (say, that of a cube, or a human face, or a telephone) only when seen from a certain perspective.23 Another example is the class of optical illusions like the Müller-Lyer diagram, in which there are two lines of the same length, one with hashes at the end of the line directed inwards and the other with hashes directed outwards. The line with the hashes directed outwards looks longer than the one with the hashes directed inwards.
In cases of optical illusion, it is not possible to reject the first premise by saying that it relies on a misdescription of the phenomenon. The argument can then be restated with the proper kind of case:
1. When I look at the Müller-Lyer diagram, I am aware of some things that are different in length
2. In the Müller-Lyer diagram the lines are not different in length.
3. What I am aware of are not the lines of the Müller-Lyer diagram. However, it has been argued that the first premise is still suspect. Why should one accept that I am aware of things that are different in length? As we saw in the previous section, it seems possible to account for something’s looking F without assuming that something actually instantiates the property F. Some relationalists apply the account sketched above to cases of optical illusions. Brewer, for instance, applies his account to optical illusions like the Müller-Lyer diagram. Here one cannot say that the lines in the Müller-Lyer diagram seem to be in one sense of the same length and in another sense of different lengths. One line just looks longer than the other, even if we know that the lines have the same lengths. But for Brewer this misleading appearance can be explained, like the cases discussed above, in terms of a triadic relation, where the way the object look is a function of the property of the object, the subject perceiving it, and the subject’s position relative to the object and other factors concerning the conditions of observation:
The [Müller-Lyer] diagram is visually-relevantly similar to a pair of lines, one longer and more distant than its plane, one shorter and less distant—a paradigm of inequality in length. It is therefore perfectly intelligible how someone seeing it might take that very diagram as consisting of unequal lines, regardless of whether she does or not. In this sense: they look unequal in length. (Brewer 2008: 172).24
But why, many have wondered, should we accept that there are some things that have different lengths? All we know is that the two lines look of unequal length. But things, the relationalist can say, can look a certain way without being that way. The relationalist can claim that naïve realist account of perceptual experience […] cannot be directly applied to any case of delusive experience, such as illusions where one does perceive an external object, but misperceives it as other than it really is. (Martin 1997: 95)
The disagreement entails different ways of classifying experiences on one or the other side of the disjunction. Whether one embraces a perception/illusion versus hallucination or perception versus illusion/hallucination version of disjunctivism largely depends on how one spells out the relationalist commitments, for instance whether one allows other factors beyond the object and its properties to contribute to the phenomenal character of experience, or whether one conceives of the relation of acquaintance involved in perception as multi-dimensional (see previous chapter for a discussion of some open questions). If one has a definition of relationalism that rules out its application to illusions, one should accept the base case for illusions and try and block the argument at the level of the spreading step. In this case, illusions will pose the same problem that hallucinations pose to relationalism, and should be treated in a similar way accepting the first premise relies on a fallacious inference from ‘The lines seem to differ in length’ to ‘there are two things which differ in length’. This is a very familiar objection to the argument from illusion. Many philosophers have attributed the acceptance of the first premise to a failure to recognize the intentional nature of perceptual experience, which can present properties without them being instantiated in any object (see in particular Barnes 1944; Anscombe 1965; Harman 1990). This objection led to abandonment of the sense-datum paradigm in favour of an intentional paradigm in the second part of the last Century, together with the dismissal of the argument from illusion.
However, Robinson argues that it would be wrong to treat acceptance of the first premise ‘as little better than a howler’ (Robinson 1994: 54). The first premise relies on a principle that its proponents find very intuitive and phenomenologically well-motivated. Robinson calls this principle the ‘Phenomenal Principle’:
(P): If there sensibly appears to a subject to be something which possesses a particular sensible quality then there is something of which the subject is aware which does possess that sensible quality. (Robinson 1994: 32)
Martin—although he doesn’t subscribe to the Phenomenal Principle— also notices that the Phenomenal Principle shouldn’t be dismissed as the sign of the failure of the sense-datum theorists to realise that mental states are intentional and thus do not require the existence of the object they refer 65 to is explanatorily inadequate. For it doesn’t take into account the fact that these sense-datum theorists recognise the intentionality of beliefs and the like, while denying that, on introspective ground, for the case of perception (Martin 2000: 208). Sense-datum theorists think that the Phenomenal Principle is required as part of an account of the phenomenal character of perceptual experience that best accommodates how experience strikes us introspectively.
So, although some proponents of the argument present the phenomenal principle as obvious, something that ‘cannot indeed be proved, but [which] is absolutely evident and indubitable’ (Price 1932: 62), it is best understood as motivated on the basis of its explanatory power (see (Snowdon 1992b): 73; Martin 2000: 209). The aspect of experience that calls for an explanation is tied to an aspect of perceptual experience that I have considered in the previous chapter as a motivation for relationalism. I noticed that perception strikes us an occurrence that relates us to what we seem to perceive, and that the obtaining of a relation involves the existence of the terms involved. Sense-datum theorists notice that experience strikes us as relational in this way even when it is illusory.
Table of contents :
Chapter 1: The Relational View of Perception
1.1 Trivial and Not So Trivial Claims about Experiences
1.2 How Experience Strikes Us
1.3 Relational Views of Phenomenal Character
1.4 Open Questions and Terminological Distinctions
1.5 A Problematic Reasoning
1.6 The Relationalist’s Commitment to Disjunctivism
1.7 Sense-datum Theories, Representationalism, Qualia Theories
Chapter 2: The Argument from Hallucination
2.1 The Base Case: Conflicting Appearances, Illusions and Hallucinations
2.1.1 Conflicting appearances and Paradigmatic Looks
2.1.2 Optical Illusions and the Phenomenal Principle
2.2. The Spreading Step
2.3 The Indistinguishability Principle
Chapter 3: How to Be Superficial
3.1 The Purported Infallibility of Introspection
3.2 Superficial Properties
3.3 Observational Properties
3.4 The Superficiality Constraint
3.5 Impersonal Indiscriminability
Chapter 4: Hallucinating, for Real
4.1 Instinguishable from What?
4.2 Hallucination: A Fuzzy Notion
4.3 What Counts as Impersonally Indistinguishable from Perception
4.4 The Problem of Hallucination is Not a Phenomenological Problem
Chapter 5: Philosopher’s Hallucinations
5.1 A Metaphysical Possibility
5.2 Experiential Naturalism
5.3 Local Supervenience
5.4 Is there Empirical Support for Local Supervenience?
5.5 The Causal Argument from Hallucination
5.6 Causation and Constitution
5.7 Where Does This Leave Us?
Chapter 6: Introspecting Hallucinations
6.1. An Invalid Argument?
6.2 The Reversed Causal Argument
6.3 Attempts to Avoid the Screening Off Problem
6.4 The Negative Account of IIIHs
6.5 Objections and Clarifications
6.5.1 The Scope of the Negative Account
6.5.2 The Alleged Need for a Positive Account
6.5.3 Very Bad Cases and Unsophisticated Hallucinators
6.6 The Phenomenal Character as a Ghost
6.8 Response to Critics and How to Preserve Superficiality
Résumé de Thèse en Français