Philosophical Studies

, Volume 163, Issue 3, pp 819–826

Susanna Siegel’s the Contents of Visual Experience


    • Department of PhilosophyUniversity of California

DOI: 10.1007/s11098-012-0013-6

Cite this article as:
Campbell, J. Philos Stud (2013) 163: 819. doi:10.1007/s11098-012-0013-6

1 The Content View

In The Contents of Visual Experience, Susanna Siegel (2010) gives a defense of what she calls ‘The Content View’. On this view, sensory experiences have truth-conditions, or accuracy conditions, to the effect that various properties are instantiated in one’s surroundings. She says ‘The notion of experience is tied to the idea that experiences have contents, where contents are a kind of condition under which experiences are accurate, similar in many ways to the truth-conditions of beliefs. I call this thesis the Content View’ (p. 4). Later she says the Content View is this: ‘All visual perceptual experiences have contents’ (p. 28). Siegel explains the notion of content she is using here as follows: ‘Contents are true or false, and the contents of an experience are conveyed to the subject by her experience. The sense in which experiences have contents (according to the Content View) thus picks up the strand of ordinary usage that takes contents to be things conveyed by sources of information (as when we speak of the contents of a newspaper story)’ (p. 28). As we will see, this notion of content being ‘conveyed’ to the subject is quite central to her account. What does it mean? Siegel says:

We can distinguish between three ways in which a content can be conveyed to the subject by her experience. First, a content is conveyed by experience if it would be content of explicit beliefs that are natural to form on the basis of visual experience. Second, a content is conveyed to the subject by her experience if it enables the experience to guide bodily actions. For instance, suppose you see the door but don’t form any explicit beliefs about the shape of its doorknob, yet you adjust your grip in advance of touching the doorknob as you reach for it. This could be a case of visual experience guiding action. Finally, a content is conveyed to the subject by her experience if it is manifest to introspection that it is a content of experience (p. 51).

According to Siegel, the Content View is not really escapable in the analysis of visual experience. ‘[O]nce the role of properties in phenomenal character is acknowledged, the Content View is unavoidable.’ (p. 29). She explains that once it is acknowledged that properties do have a role in the phenomenal character of experience, we can say that an experience is accurate only if the properties presented by the experience are instantiated. And once we have reached this point, we can go further and say that there are going to be conditions such that the experience is accurate if and only if those conditions are met. ‘In general, it is plausible that whenever there are some conditions C such that X is accurate only if C, there are some (perhaps stronger) conditions C* such that X is accurate iff C*’ (p. 53). So from the mere acknowledgement that properties figure in the phenomenal character of experience, it follows that experiences have representational contents.
I think that we can distinguish between different ways in which the properties of the things we see figure in our visual experience. Siegel operates with a monolithic notion of ‘perceiving a property’, and I think that once we reflect on the distinctions, we find that there are two different phenomena here. We have to distinguish between:
  1. (1)

    A property figuring in the phenomenal character of visual experience, and

  2. (2)

    A property of a seen object being accessed by the perceiver.

In general, the problem with Siegel’s account is that she does not give due weight to the role of visual attention in our accessing the properties of objects. Once you get it that properties can figure in visual experience at a level that is prior to their being accessed by the perceiver, you realize that it is one thing for a property to be figuring in visual experience, and another thing for that property to be ‘conveyed’ to the perceiver, in anything like the sense explained by Siegel. The accessing of a property by the perceiver may indeed be a matter of representational content; but the figuring of a property in visual experience is more basic than its being accessed by the perceiver; ‘conveyed’ in Siegel’s term. We have to separate a property figuring in vision at the level of phenomenal character, from a property figuring in vision at the level of representational content.

2 Selection Versus Access

I don’t know of a better way to explain the distinction I have in mind here than by looking at Huang and Pashler’s (2007) distinction between the role of properties in the visual selection of objects, and the way in which we visually access the properties of the objects we select. Huang and Pashler operate with a picture of visual attention on which it is underpinned by an underlying architecture of Treismanian feature-maps. The first phase in visual attention is selection of a region of space. At this phase, you may select a region by specifying a particular property. For example, you can often differentiate an object from its background by its color. So you might, for instance, select the red region: the overall set of locations in your visual field at which you find redness. As they put it, the regions selected are three-dimensional, ‘shrink-wrapped’ to fit the objects with which one is concerned. Once you have selected a particular region, you can go on to access its various properties. This is a matter of the visual system making explicit, for further computation or reasoning, the various properties possessed by the region or object (Fig. 1).
Fig. 1

Boolean map theory depends on an underlying architecture of Treismanian feature maps. In the example shown, a feature, redness, is used to select all and only the red areas. A first consultation of the feature maps is made, to find which regions have red at them. This yields a Boolean map, which at this stage is not thought of as labeled. The underlying feature maps are then consulted again to determine what features are found at the region selected, and the labeling then makes those features explicit. Huang and Pashler locate the level of conscious experience as post-access; the present author argues that conscious experience of the property may be required even when the property is not accessed, as the basis of selection of the region or object one consciously encounters (from Huang and Pashler 2007). (Color figure online)

Notice that it is one thing to use a property to select a particular region of the visual field, and another thing to access that property of the region you have selected. So far as the mechanics of the Huang and Pashler model are concerned, you might perfectly well be able to use a particular property to select a region of your visual field, but be unable to access that particular property of the region selected (though you would doubtless be able to access other properties of that region). As we shall see, for a property to figure in the phenomenal character of experience seems to be a matter of its being possible to use it as a basis for selection of a region or object. But for the property to be ‘conveyed’ to the subject seems to be a matter of the subject being able to visually access that property as a property of the region selected.

Siegel’s argument is that once we acknowledge the role of properties in the characterization of visual experience, we have to accept the Content View. For a property to figure in the characterization of visual experience just is a matter of that property being ‘conveyed’ to the subject; and then the visual experience can be said to have a particular content, namely, that the properties ‘conveyed’ to the subject are instantiated. But what I will be arguing is that this conflates two different roles for properties in vision. One does have to do with the conscious experience, and that is the role of properties in selecting a particular region or object. The other does not particularly have to do with conscious experience, and that is the way in which properties of the objects seen are ‘conveyed’ to the subject, or ‘accessed’ by the subject. Once we separate these different roles for properties in vision, it becomes evident that it is one thing for a property to figure in visual experience, and quite another for vision to ‘convey’ to the subject a representational content to the effect that the property is instantiated.

3 Two Ways of Seeing Color

Consider the Ishihara tests for color vision. Here one is presented with a figure, let us say a numeral, perhaps a ‘5’, separated from its background only by its color, suppose it’s red. We have an array of dots and blobs of randomly varying luminances; the only property systematically distinguishing the 5 from its surroundings is the color of the thing. If you pass the test, if you can see the 5, this can only be because you are experiencing the color of the thing. If you can see the 5, that only be because you have color vision and color experience.

Passing this kind of test for color vision, being able to see the 5, is consistent with your being unable to give a verbal report of the color of the object. This is obvious from the case of animal color vision. A tiger padding through the veldt may be able to distinguish its prey from the foliage because of the color of its target; but that does not mean that the tiger has any interest in the color of thing, it may not even be able to attend to the color of the object. All it cares about is the object itself. The tiger may be incapable of attending to the color of the object, even though it uses the color of the thing to select the object from its background. A parallel point applies to humans: the use of color to discern the object against its background is one thing, and the ability to form beliefs about the color of the object is another. There are various ways in which we can illustrate this point. Human children have a lot of difficulty in learning color words. Even after more than 1,000 trials, a child who already has hundreds of words in their vocabulary may still be quite unable to go on to apply the words ‘red’ and ‘green’ (Rice 1980; Franklin 2006). Darwin thought that his children were color-blind, because of the great difficulty they had in learning color words. There is even a word for the phenomenon, ‘Farbendummheit’ (Davidoff 1991, p. 149). Yet ordinary color vision is in place in children from a very young age. As early as 4 months old, categorical color perception seems to be in place in children (Franklin and Davies 2004). Of course, you might point out that the capacity to give verbal reports of the colors one sees is one thing, and the ability to conceptualize the colors one sees is another. Merely establishing that young children can see colors without having the ability to give verbal reports of them does not of itself show that these children do not have a capacity to conceptualize the colors they see. There does not seem to be decisive evidence in support of this idea (cf. Kowalski and Zimiles (2006) for discussion and further references). But even if it is true, the basic point is still be in place, that the ability to use colors to define the objects one sees is one thing, and the ability to conceptualize those colors is another. One could be using colors to define the objects to which one attends without having any ability to attend to the colors themselves, and, consequently, without having any ability to conceptualize those colors.

Of course, a personal-level representation of the color in experience might be functioning to allow the subject to perform a number of other operations than verbal report. But the same basic point still applies: the use of the property to define the object as figure from ground is one thing, and the use of the property in these further computational tasks is another. Ordinarily we are, for example, able to engage in a variety of sorting tasks with regard to color. Given a stack of variously colored chips, you might, for instance, be able to sort them into piles of like-colored chips, or to sort all the green ones into a sequence from light green to dark green. Performing this kind of task is one use to which you might put a personal-level representation, supplied by vision, of the colors of the chips. But the mere use of the color of a chip to differentiate the chip from its background does not of itself imply that one will be capable of these kinds of matching and sorting tasks.

In Siegel’s terms, the point is that color may be figuring in one’s sensory experience, even though there is no sense in which the color of an object is being ‘conveyed’ to the subject by the sensory experience. The subject may incapable of forming beliefs about the color of the object on the basis of that sensory experience. The subject may not be using the color of the object to guide any actions they perform; merely seeing the ‘5’ against its background does not mean that one is capable of using the color itself as a guide to action, whatever that might come to. And, of course, even if one is seeing the color of the thing, in the sense that one can use it color to define the object against its background, one may be quite incapable of introspecting that one is having an experience with that color as a content. So seeing the color is one thing, and having a representation of the color of the object ‘conveyed’ to the subject is quite another.

To sum up. The capacity to use a perceived property to define an object one experiences visually does not of itself mean that one has a personal-level representation of that property. There may be no personal-level operations to which one could be subjecting the representation of that property: one may not be using any such representation to guide one’s actions, or in the formation of belief or the making of verbal reports, and it may not be accessible to introspection. That property may nonetheless be figuring in one’s visual experience. In fact, if you consider the case of perceiving the 5 in an ordinary test of color vision, it seems unimaginable that one could experience the 5 without experiencing its color. The whole point of the color-vision test is that nothing except color differentiates the 5 from its background. Take away the difference in color, and you simply have an array of variously shaped and sized blobs of randomly varying luminances. Suppose you try to imagine this visual experience: you look at a stretch of wall that is manifestly uniform in color, and you see a 5 on the wall, the ‘5’ being differentiated from its surroundings only by its color. I do not think that there is a formal contradiction in this idea. But such an experience seems altogether unimaginable; we can’t visualize it. There might, of course, be a case in which a subject could succeed in some blindsight-style ‘guessing’ that there is a ‘5’ on the wall, even though the subject has no color experience that differentiates the ‘5’ from its surroundings. But what seems flatly unimaginable is that one could have conscious visual experience of the ‘5’ even though one has no visual experience of the color that is the only thing differentiating the figure from its ground. But if this is true for the case of the wall uniform in color, it is of course also true for the case in which color is all that is systematically differentiating the ‘5’ from a background patterning of blobs of randomly varying luminance.

Suppose now that we consider what happens when you do develop the ability to access the colors of the things you see. Does this change the nature of your visual experience? On the face of it, it does not. The richness and density of your color experience is just as it was before you became able to access the colors of things. You have now developed an ability to form specifically visual representations that do ‘convey’ to you, in Siegel’s term, the colors of the things around you. You can use those visual representations of color in belief formation, in guiding action—say, if you are trying to control the colors in a room you are designing—and in forming introspective beliefs about the nature of your current vision. But those visual representations that reflect your access to the colors of things are not themselves constitutively affecting your more basic color experience. (An analogy may be helpful. Consider someone who has ordinary spatial vision, who now has to take on the role of bodyguard to some particular individual. The bodyguard will have to develop a whole range of spatial perceptual skills that they did not have previously; they will now be able to access spatial relations that they perhaps could not access with any facility before. Who has a clear shot at the target? Who is between the target and the only safe exit? And so on. The ability to form and use these higher-level visual spatial representations is indeed a perceptual skill. But it may leave the subject’s spatial experience itself intrinsically unchanged.) In giving this analysis, I am incidentally parting company with Huang and Pashler, who suggest that ‘consciousness’ is a matter of which properties of the thing one is currently accessing. That suggestion is, I think, just a mistake: it would mean that one could consciously see the ‘5’ even though its color was making no difference to one’s phenomenal experience, and as we saw, that does not seem to be the right result.

So far I have been discussing a very simple case in which one uses a property to define an object; it’s ‘simple’ in the sense that only one property is involved and we know what it is. Of course, in general, we use groupings of properties to differentiate objects from their backgrounds. But the general point always applies: using a property, or group of properties, to differentiate an object from its background is one thing, and being able to access that property or group of properties, for the purpose of verbal report, or some other computation, is another. Conscious experience of a property seems to have to do with the possibility of using that property (perhaps in conjunction with others) to select an object as figure from ground. Having that property of the object ‘conveyed’ to one by a visual representation seems to be a quite different phenomenon.

4 The Content View and the Relational View of Experience

Siegel points out that various philosophers (myself included) have argued on behalf of relational or naively realistic views of visual experience. On these views, sensory experience is a matter of being related to the objects and properties in one’s surroundings. On the face of it, a view like this is in sharp contrast to the representational view that Siegel advances. Siegel’s response is not to try to rebut the relational or naïve-realist views of sensory experience. Rather, she argues that in fact, on any plausible construal of what the relational or naively realistic view of experience is, it is quite consistent with the view that visual experience is representational. Really, an admission that we do ordinarily perceive properties is all that is required to establish the correctness of the view that experiences are representational. People have often thought that the kinds of contents that experiences have can be had by those experiences whatever is going on in the non-mental world. On this internalist view, you can, for example, have a visual experience with the content, ‘that dog is piebald’, even if there is no dog there and nothing piebald. You can be having a visual experience as of a particular object having various particular properties even if that object is not there to be seen and in fact there is nothing there with those properties. But Siegel argues that this kind of internalism is not essential to the Content View; the Content View can ‘be divorced from the internalist outlook and its construal of contents’ (p. 6). We could have a view on which visual experiences have ‘contents’ even though we are ‘disjunctivist’ about those contents, in the sense that we don’t think an experience can have the content it does no matter what is going on in the subject’s environment. We could say that if the dog isn’t there, you can’t have an experience with a content relating to ‘that dog’, and if there isn’t anything piebald there, you can’t have an experience of piebaldness. It is possible ‘to formulate content disjunctivism in a way that individuates the contents of strongly veridical experiences by perceived properties, as well as by perceived objects’ (p. 74). Content disjunctivism in this sense is, according to Siegel, still a version of the Content View.

Things look rather different, though, once we acknowledge that there is a difference between (a) a property figuring in the phenomenal character of one’s experience, and thus being capable of serving as a basis for the selection of an object as figure from ground, and (b) a property being accessed on the basis of attention to the object selected. The fact that a property figures in the phenomenal character of one’s experience does not of itself mean that it makes any sense to talk about the experience as being ‘accurate’ or ‘true’. This is somewhat masked by the fact that humans do unselfconsciously access the colors of the objects they see, and it certainly does make sense to talk about true and false, accurate or inaccurate, at this level. But this is not the level of conscious experience itself. Consider again our tiger on the veldt. Suppose that it uses such color vision as it has only as a way of distinguishing objects from their backgrounds. Does it make any sense to suppose that the tiger—the organism as a whole—is or is not being subjected to illusions of color? Color is entering into the phenomenology of its sensory experience, otherwise color could not be making a difference to which objects it can consciously perceive. But if color is not being ‘conveyed’ to the tiger in any of Siegel’s senses—it does not form beliefs about the colors of things, it does not use color to guide its actions, and it does not form introspective beliefs about what color experiences it is having—then it makes little sense to talk about the tiger itself—the whole organism—being subjected to illusions or given true information about the colors of the things it observes.

This way of thinking about the properties encountered in experience is, however, consistent with a broadly relational or naïve realist view of perceptual experience. On this view, the encounter with the properties around one, and the selection of regions or objects on the basis of their possession of those properties, is the basic layer of perceptual experience. Accessing the properties of the objects one encounters is a third stage. It is only at this third level that we have perceptual representation. The trouble with representationalist accounts of perception is that they collapse the basic level of sensory experience into the higher-level, purely cognitive stage at which the properties of objects are explicitly accessed.

Copyright information

© Springer Science+Business Media B.V. 2012