The argument from illusion & hallucination

Jeff Speaks

January 10, 2006

1 The idea of an ‘immediate object of perception’
2 The argument from illusion & hallucination
3 Four responses to the argument from illusion
 3.1 Sense datum theories of perception
 3.2 Non-relational theories of perception: adverbialism
 3.3 Disjunctivist versions of direct realism
 3.4 Common kind versions of direct realism

1 The idea of an ‘immediate object of perception’

The first topic we’ll be taking up is the nature of the objects of perception. But before doing this we should get clear on what ‘object of perception’ is supposed to mean.

The core idea here is that perception is essentially relational: to have a perceptual experience is to bear a certain relation (e.g. visual awareness) to an object of the perceptual experience. In this sense, having a perceptual experience is supposed to be more like having a relational property - like standing to the left of someone - than having a monadic property (like having a mass of 80 kg). Why the view that perceptual experience is relational should seem natural.

Once you have the idea of an object of, for example, visual experience, we can distinguish between immediate and mediate objects of visual experience. This distinction has proved difficult to analyze, but intuitively is fairly clear. Suppose that you are watching Paul Martin give a speech on T.V. It seems that you are seeing Paul Martin. But it also seems that you are seeing him by seeing something else: namely, the image on your television screen. Or suppose that you are looking at the Leacock Building from Sherbrooke Street. It seems that you see the building by seeing something else: a part of the building, one of its sides. The distinction between mediate and immediate objects of (visual) perception is the distinction between those things which you see in virtue of seeing something else (mediate) and those which you see, but not by virtue of seeing something else (immediate). When we ask what the objects of perception are, we are asking about the nature of the immediate objects of perception. (For a clear discussion of the distinction between mediate and immediate objects of perception, see Ch. 1 of Jackson, Perception: A Representative Theory.)

Another important distinction: the distinction between objects of perceptual experience and the physical processes which underwrite those experiences.

2 The argument from illusion & hallucination

This question about the objects of perception has a natural answer: the objects of perception are, at least in standard cases, material objects in the perceiver’s environment. (This is sometimes called direct realism.) For most of the 20th century, however, this was not a very popular view about perception. A principle reason for this was that considerations about the objects of illusory and hallucinatory perception seemed to refute direct realism. As Ayer put it in The Foundations of Empirical Knowledge, the argument from illusion is his answer to the question, ‘Why may we not say that we are directly aware of material things?’

There are many different versions of ‘the argument from illusion.’ One of the ways in which these arguments differ is by focusing on either illusion or hallucination. For now we will focus on a version of the argument based on cases of hallucination:

 

1.

All experiences have an object (the relational view of perception).

2.

Hallucinatory experiences do not have objects which are external to the mind.

3.

Hallucinatory experiences have objects which are internal to the mind. (1 & 2)

4.

The objects of experience are the same in the case of hallucinatory and veridical experience.



C.

The objects of all experiences, whether hallucinatory or veridical, are internal to the mind. (3 & 4)

One’s view about the objects of perception largely depends on one’s response to this argument. Today we will distinguish four responses to it; in the next few weeks we’ll spend some more time discussing them.

3 Four responses to the argument from illusion

3.1 Sense datum theories of perception

The most popular response to the argument from illusion for most of the 20th century was to accept the argument, and its conclusion that the immediate objects of perception are always internal to the mind. The view which resulted was called ‘the sense datum theory of perception’, because the mind-internal things which were supposed to be the objects of perceptual experience were called ‘sense data.’

It is not hard to see why so many were persuaded by this argument, and others like it. We’ve already seen why (1) looks plausible; and (2) looks more or less obvious. You might think that (4) looks easier to deny; but pretty convincing arguments can be presented in its favor. These arguments often begin from the fact that hallucinatory experiences can be qualitatively indistinguishable from veridical experiences; given this, it would seem very surprising if the two sorts of experiences had radically different kinds of immediate objects. Wouldn’t one expect such a difference to show up in how the experience seems to the perceiver?

Ayer presents a few versions of this argument in The Foundations of Empirical Knowledge. He suggests that if the objects of veridical and illusory perceptual experiences were different kinds of things, we should expect this to issue in some qualitative difference in the perceptual experience. But it does not:

“When I look at a straight stick, which is refracted in water and so appears crooked, my experience is qualitatively the same as if I were looking at a stick that really was crooked. When, as the result of putting on green spectacles, the white walls of my room appear to me to be green, my experience is qualitatively the same as if I were perceiving walls that really were green.” (6)

Another way to raise substantially the same point is via the fact that veridical and illusory perceptions “may form a continuous series, both with respect to their qualities and with respect to the conditions under which they are obtained.” (8) The example of approaching an object from a distance, or slowly raising the level of lighting in a room.

Although sense datum theory has the argument from illusion in its corner, it also has a number of unattractive features which might make you want to find another way around the argument:

  1. Epistemological worries. It is obvious that, in many cases, perceptual experiences can justify beliefs about everyday objects like chairs, desks, other people, etc. But, on a sense datum theory of perception, the question arises: given that we are only acquainted with sense data in perception, how could perception justify beliefs about external objects? The only obvious answer to this question is: by inference from our experience of sense data. But this gives rise to the question: how is this inference from experience of sense data to beliefs about external objects itself justified? (Not, obviously, on the basis of some observed correlation between sense data and those external objects, since our experience is only of sense data.)
  2. Difficulties in explaining the possibility of thought about external objects. A perhaps more fundamental problem concerns not just our justification for thoughts about material objects, but also the possibility of having such thoughts at all. A form of empiricism which many have thought plausible claims that our ability to have thoughts involving some object o must, at least in paradigm cases, be explained in terms of prior perceptual acquaintance with o. But if sense datum theories are correct, we are never acquainted with material objects. That means that, if this form of empiricism is correct, we can only have thoughts about material objects indirectly, as the cause of this or that sense experience. But this seems implausible.
  3. Internal difficulties with the theory. Further difficulties arise from difficulties in characterizing what sense data are supposed to be. An initial idea is that sense data are supposed to be the objects of perception, and are supposed to have exactly the properties that they seem to have. Why this generates puzzles about these entities; the example of the speckled hen.
  4. Metaphysical worries. You might have further worries if you are attached to materialist views of the mind. Why a view of sense data on which they have the properties which they appear to have makes it hard to identify sense data with physical entities.

3.2 Non-relational theories of perception: adverbialism

One important alternative to the sense datum theorist’s acceptance of the argument from illusion is the idea that the argument goes wrong right at the very beginning, with the idea that experiences are relational. If we do not concede at the outset that perceptual experiences are relations to objects of perception, then we do not have to concede the intelligibility of the sense datum theorist’s questions about the nature of the objects of hallucinatory experience.

Adverbialism is a name for a cluster of non-relational theories of perception. According to the adverbialist, perceiving is not a matter of bearing a certain relation to an immediate object of perception, but rather of having a certain monadic property in a certain way. Take, for example, a visual experience as of a red tomato. Whereas a relational theorist will analyze having such an experience as the bearing of a relation to something - perhaps a red tomato, perhaps a reddish and roundish sense datum - the adverbial theorist will analyze having such an experience as the instantiation of certain monadic property - sensing - in a certain way - reddish-roundishly.

An analogy might make the difference clearer. Consider two properties that you might have:

walking toward Toronto

walking quickly

It seems natural to say that in the first case, to have the property is to bear a certain relation toward Toronto - the relation of walking toward it. But it would seem very odd to say about the second case that to have that property is a matter of bearing a certain relation to quickness - say, the walking-that-way relation. What it is to walk quickly is not to bear a relation to quickness, whatever that might mean, but rather to instantiate a certain monadic property, the property of walking, in a certain way - someone who is walking slowly instantiates the same property, but in a different way.

If this is right, then the question is whether the having of a visual experience is to be analyzed in the same way as walking toward Toronto or walking quickly. The important point as regards the argument from illusion is that only the former sort of analysis entails that there are such things as the objects of experience. For this reason, one of the principle strengths of adverbial theories is often taken to be the fact that it analyzes away our apparent references to ‘mental objects’.

Two objections to adverbialist theories of perception (both from Jackson, Perception, Ch. 3, ‘The existence of mental objects’):

The many properties objection. One can (as the relational theorist might put it) sense several different things, with distinct properties, at the same time. So, for example, one might have a red after-image at the same time as one has a green after-image. The adverbial theorist analyzes ‘A senses something red’ as ‘A senses red-ly.’ The question is: how should the adverbial theorist analyze ‘A senses something red and something green’? At least two possibilities suggest themselves.

  1. The conjunctive version. The adverbial theorist could analyze this as ‘A senses red-ly and green-ly.’ This is the most natural analysis; but it doesn’t work as well for slightly more complicated cases. Consider the following two distinct perceptual experiences:

    I have a red, square after-image and a green, circular one.

    I have a green, square after-image and a red, circular one.

    The problem is that each of these are analyzed by this ‘conjunctive’ version of the adverbial theory as:

    I sense red-ly and square-ly and green-ly and circularly.

    So the analysis is incorrect, since these are clearly different perceptual experiences.

  2. The compound version. It seems that what is needed is some sort of compound way of sensing, so that we can distinguish between sensing redly-squarely and greenly-circularly from sensing redly-circularly and greenly-squarely. So suppose that we introduce such compound ways of sensing. Then we need some interpretation of phrases like

    I sense redly-squarely.

    on which it is not just equivalent to

    I sense redly and squarely.

    The only way to do this seems to be to take ‘redly’ not to be a mode of sensing, but to be, to put it loosely, a mode of shape-sensing. (Equivalently, the adverbial theorist could take ‘squarely’ to be a mode of color-sensing.) For if both ‘redly’ and ‘squarely’ were modes of sensing simpliciter, it would be impossible to distinguish the compound version of the adverbial theory from the conjunctive version. But now consider the two claims

    I have a red after-image.

    I have a red, square after-image.

    which this version of the adverbial theory translates, respectively, as

    I sense redly.

    I sense redly-squarely.

    The problem is that ‘redly’ must be understood as ambiguous between these two analyses: in the first, it stands for a mode of sensing, whereas in the second, it stands for a mode of shape-sensing. But then the analysis fails to explain why the second of the above claims entails the first.

    (Jackson gives a nice example of a case where a word really is ambiguous in this way: ‘impressively’ in ‘He spoke impressively’ and ‘He spoke impressively quickly.’ The adverb modifies the verb in the first, and the adverb in the second. So it is not impossible that ‘redly’ could work in the way that the adverbial theorist says that it does. But the key point is that in the example of ‘impressively’, unlike the example of ‘redly’, the second claim does not entail the first.)

The complementarity objection. This objection is based on a general test for whether a certain property is a monadic property or a relational property. One object cannot both be and not be F (at the same time and in the same respect), nor can an object do something both F-ly and not F-ly at the same time; e.g., something cannot be both red and not red, and cannot walk both quickly and slowly. But one object can be R-related both to something which is F and something which is not F. E.g., I can be to the left of someone who is tall, and someone who is not tall. But now note that I can at the same time have a visual experience of something which is red and something which is not red. This indicates that visual experience is not a matter of sensing F-ly (as the adverbial theorist claims), but rather of standing in a certain relation to an object of the experience.

A third objection might come from the connections between experience and thought. Suppose (for the sake of argument) that we are convinced that propositional attitudes like thought and belief do genuinely have objects, so that an adverbial analysis of these attitudes would be incorrect. We might want an account of perceptual experience to explain how certain agents are in a position to have certain kinds of thoughts or beliefs - i.e., thoughts or beliefs whose objects fall in a certain class. But it is not obvious how we could do this without thinking of perceptual experiences, too, as having objects.

(For a defense of the adverbial theory, see Tye, ‘The adverbial approach to visual experience’ or, for a briefer introduction, see his entry for ‘adverbial theory of mental states’ in the Routledge Encyclopedia of Philosophy.)

3.3 Disjunctivist versions of direct realism

Partly because of these sorts of problems, philosophers have increasingly wanted to maintain a relational view of perceptual experience without lapsing into sense datum theory. One prominent version of this view has been ‘disjunctivism.’ The characteristic move of the disjunctivist in response to the argument from illusion is the rejection of the view that veridical and hallucinatory/illusory experiences form a common kind with objects of the same generic category. (So they would respond to the above version of the argument by denying premise (4).)

Given the above, it should be clear that one of the challenges which faces disjunctivism is the challenge of explaining how veridical and hallucinatory experiences can be indistinguishable while having radically different sorts of objects.

Next week, we will discuss a few versions of disjunctivism.

3.4 Common kind versions of direct realism

Many philosophers have been unpersuaded by disjunctivist responses to the argument from illusion. If those philosophers want to maintain a relational theory of perceptual experience, and wish to be direct realists, they have to find some other way around the argument from illusion. We’ll consider one attempt to do that in a few weeks by, in effect, denying premise (2) of the above argument.