Introduction to the Contemporary Imagery Debate

What is the Debate About?

The existing literature in empirical psychology does not make the imagery debate easy to understand. The literature is often contentious and is steeped in the specialized language of contemporary empirical psychology. In addition, the theories often assume prior commitments to materialism and the computational theory of the mind.

Since these factors can create a significant barrier to understanding the debate, I have adopted a slightly different strategy for the initial presentation of the issues. I first attempt to show how the question of the nature and function of mental imagery arises from examples of simple recall. I shall then indicate how the descriptivist and pictorialist positions can be derived from elementary considerations about memory experiences, their relation to information stored in the brain, and the storage of information in propositional or descriptive form.

Explanations of Simple Recall

Explaining the ability to recall information is a test case for examining what it means to have a belief, and in what forms beliefs exist. We will examine a simple case. Consider the question:

Is London the capital of England? (1)

One standard form of explanation of the ability to understand and formulate an answer to simple questions of this nature involves an appeal to the analysis of beliefs in terms of propositions and their properties.

We can distinguish between the form and content of beliefs. The question here, for example, is about a certain content, namely the relation between London and the capital of England. The content is something that exists independently of any particular form or expression. The content can be indicated in various forms, including other languages. "Ist London die Hauptstadt von England?" expresses the same question about the same content. The positive propositional form, "London is the capital of England," addresses the same content. If we ask someone about their beliefs on a certain matter, the question concerns the content, not the specific expression.

To specify the beliefs we have, it seems that we should be able to provide a linguistic expression that indicates the content. Although ordinary, prereflective acts of recall and communication may not involve any consciousness of the theory of propositions, it seems both possible and efficient to provide an analysis of this in terms of propositional form. If our beliefs can be placed in propositional form, simple cases of recall might be illuminated by further analysis of propositions and propositional knowledge.

At this point in the explanation of how we recall and understand, a connection between the propositional view of beliefs and a computational view of the mind can be made. The latter view, needless to say, assumes materialism, so it is assumed that the computations involved in the work of the mind are carried out by physical changes in the brain. Combining this supposition and the idea that our beliefs take propositional form, with the idea that our linguistic expressions derive from the structure and operation of the brain, it seems an evident conclusion that there must be propositional (or perhaps sentential) analogues in the brain that have content. Although there is no agreed-upon theory of how brain structures and neural events can have content, it is, according to this view, obvious that this must be the case, since this fits most conveniently with a completely materialist ontology. The consequence of this idea is that the characteristics of propositional expressions (truth and falsity, translatability, indicating a content) are thought to apply to information structures in the brain.

To consider our example again, it is evident that our conscious access to information does not tell the whole story. In answering such a question we may have absolutely no consciousness of forming a proposition. Under the scenario of computationalism, although we may not consciously express a proposition to ourselves, our belief must in fact be due to a kind of unconscious matching program operating in the brain: the sentence we hear is matched to the corresponding belief structure in the brain. If there is a match, we recognize the sentence as one expressing a proposition that we believe. In most cases, we are conscious only of the sentence as expressing a certain content, and we are immediately conscious of an appropriate form of response.

Recall Involving Images

Now consider the question:

Do German shepards have triangular-shaped ears? (2)

We might answer straight off "yes," because we know this to be true. If this is the case, the situation seems to be like the previous example question (1). But many people answer such a question only after a moment of reflection. They report that this information is something that is checked by means of inspecting a visual mental image. Does the image itself constitute the content of the belief? If so, it seems to be decidedly non-linguistic, or non-propositional in nature.

One explanation of the role of imagery in answering question (2) is to make the example consistent with the sentential analogue view just described. The image that we inspect to retrieve the information must be presumed to derive from sentence-like information in the brain rather than an image. An image could not itself be encoded in a sentence-like format. We must assume that the image itself can not literally embody the information I have retained about the shape of the ears. Properly speaking, our root belief structure must be propositional: we believe that a German Shepard's ears are pointed and roughly triangular in shape. Although we experience an image, this is only because the information we have about the shape of the ears has somehow been disguised as an image.

This explanation is bolstered by an additional argument. To store the image itself would be pointless, since the image does not say anything in and of itself -- it always requires an interpretation. What we need is the interpreted, sentence-like information, that express the content we want to retain. Because of the operation of our visual memory system, it may seem to us that we see an internal image that holds that information. But in fact, the internal picture must have been created by beliefs that we already had that could have been expressed in descriptive form. It could not have been otherwise, for without this prior knowledge, no mental picture could have been generated.

This account may seen to have some plausibility. But it suffers from the following objections: (1) Why is there an internal deception in consciousness? Why would some beliefs that we hold appear to us in the form of an image and others not? What reason can we give for the mind working in this way? (2) It seems the entire account is at odds with the phenomenological facts of the matter. We do have an experience that is rather like looking at something. The image presents itself to us in consciousness for our inspection, and from this we derive the information we need. Based on my inspection of the image, I "see" how the ears look. I then need to frame a linguistic expression to describe adequately how they look. My belief is based on what I see through the mind's eye. Why invent some other account of it?

These objections might incline us to dismiss the propositional account in favor of the introspective account, which appears to be a more honest phenomenology. But the introspective account also creates difficulties. It introduces a new entity: the mind's eye. It requires that we take the mind's eye phenomenon seriously rather than metaphorically or as an epiphenomenal manifestation of other processes. This creates the difficulty of explaining how there can be a separate, internal, perceiving function (or, another mind) within the mind. Also, the account creates a problem for scientific measurement. How can we test for the presence of mental images? Testing linguistic beliefs seems much more straight forward, since verbal utterances are public. Finally, the account designates "the image" as a type of mental entity that apparently bears information, but without an explanation of how images can acquire definitive meanings.

Our examination of a simple case has led us to ponder a puzzle about images: their existence seems self-evident to consciousness, but if they are visual representations of what we know because of propositional analogues stored in the brain, they have no purpose. Why do we have mental images at all? On the other hand, it is irrefutable that we have imagery experiences and it is difficult to believe they are totally without function.

An examination of more complicated cases may provide us with some clues.

Beyond Simple Recall

Once we expand the examples beyond simple recall, many additional issues become associated with imagery. Consider the following example problems.

(1) What letter of the alphabet does the capital letter "n" look like when it is rotated 90 degrees counter-clockwise?

(2) Imagine that your house has a doorway facing the street. You are inside the house, ten feet from the open doorway. The threshold of the doorway is level with the street outside, which runs perpendicular to your line of sight. At this moment, a huge black car tire, ten miles in diameter, begins rolling slowly down the street from left to right. As the tire passes your house, what do you see through the doorway?

These questions involve the use of imagery in problem solving. Are these problems disguised forms of deductions or do they rely completely the existence of a mental image for their resolution? Most people can solve problem (1) easily using imagery; few people can solve problem (2) no matter how vivid their imagery. What accounts for the difference? Perhaps the mind needs both images and a method of propositional deduction to operate. In any event, such examples make it clear that we need to account for more than just memory images; we need a more complex theory addressing their use and manipulation in the mind. Ideally, some description of the structure and processes of the mind would show us how imagery fits in.

The considerations mentioned far allow us to introduce some of the central concerns that drive the imagery debate. These concerns can be understood in terms of a fundamental distinction.

The phenomenal image is what we experience, what we apparently or figuratively see through the mind's eye. Those structures and processes in the mind (understood to be physically instantiated in the brain) that are designated as causes, explanatory factors, or correlates of the phenomenal image are the functional image. As I interpret the imagery debate, the fundamental issue is to explain the nature and function of the phenomenal image, the nature and function of the functional image, and the relation between the phenomenal and functional images.

We can now restate the descriptivist and imagist accounts mentioned above in terms of this distinction. This provides us with a preliminary statement of the two sides in the debate. It should be borne in mind that the actual positions of theorists such as Pylyshyn and Kosslyn are more subtle than what is presented here.

The descriptivist or propositionalist view tries to maintain the view that the information stored in the brain, our beliefs, can always be expressed in linguistic form. Therefore, images have nothing to do with the recollection or manipulation of information. Information is stored in a propositional format (understood in a special sense peculiar to cognitive psychology), and when it is recalled any imagery we experience is just an epiphenomenon, since the actual content of the stored data derives from the descriptive information. In problem-solving and thought processes, the actual work done by the mind depends on descriptive processes, capable of being expressed in propositional form. As for the physical processes involved in recall and thought, it must be assumed that they have the characteristics relevant to their function as propositions: they will be sentential analogues that can enter into relations determined by brain functions. The brain functions that are cognitively operative are not perceptual in nature. Although we may experience images, these are epiphenomena. The phenomenal image therefore has no role in cognition. There are no functional images either since any explanatory factors that we might be tempted to call functional images can ultimately be dispensed with because they reduce to propositions.

The pictorialist or imagist view does not limit knowledge and the manipulation of symbolic information to propositional form. This side asserts that we store information in the form of images and recall them as images. Propositions can be derived from the image, but that does not mean that information is limited to propositional form or that the image serves as a useless intermediary in the functioning of the mind. The pictorialist view holds that the phenomenal image appears to us as it does for a reason: it reflects the structure and operation of the mind, which incorporates a functional image having irreducibly pictorial qualities.

Our purpose in this section has been to present background information necessary for understanding the imagery debate. I have attempted to show that the central issues in the debate, concerning the phenomenal and functional images, can be understood on a fairly intuitive level, without appeal to too much theory. Unfortunately, we can not let matters rest here, for we need to address additional issues that figure in the debate.

Theory and Presuppositions: Additional Layers to the Debate

As we said above, the imagery debate in contemporary psychology has been complicated because of certain assumptions built into the debate. Most theorists in the debate uncritically accept materialism. This is not a problem in itself, since materialism of some sort is generally regarded as preferable to a non-materialist Cartesian approach. But the cognitive science or computational approach also claims to study the mind independently of any material considerations. In much the same way that we can describe the function of various modules in a computer program without reference to the specifics of the physical operation of a computer, cognitive science aims at isolating and describing specific mental computations that underlie our cognitive abilities. In a computer, the various computational states that follow in sequence can each be identified in terms of the specific computational function (e.g., finding a square root, checking the spelling of word) that is being performed. By analogy, physical processes in the brain correspond to various mental functions, that can be discussed separately from the brain processes. In a computer, we can check the physical operations that underlie specific functions. In the case of humans, the opposite is the case: we can easily check (or have indications of) mental functions, but the specific brain processes are generally unknown. The functional approach leaves open the possibility that the mental functions identified can not be successfully connected back to specific brain processes. It is sometimes unclear in the debate what exactly is at stake. Is it enough to simply identify mental functions, or must some specific physical underpinning be indicated or suggested in order to have a viable theory?

Additional confusion has been created by disagreements, implied or stated, about the rules of experimental psychology. Some experimenters, for example, accept the reports of subjects as part of the data or as supporting evidence, while others minimize the importance of subjective reports.

Finally, there is confusion because the entire debate has been structured by the importation of computational terminology. Investigators have attempted to make the explanations of imagery coherent with explanations in terms of computational concepts. The temptation that runs throughout the debate is to appeal to a prior understanding of what such terms as "data," "function," "process," and "execute" mean.

These complicating factors in the debate generate numerous questions:

1. Are images compatible with a computational theory of the mind?

2. Are images primitive constructs within cognitive science?

3. Can images be identified with certain functional states of the brain?

4. Are images handled as computationally holistic entities, or do they have parts?

5. Do images have intrinsic properties that are determined by the functional architecture of the brain?

6. What procedures are necessary for empirical psychology to test the properties of images?

7. Do imagery and vision share functional elements and/or physical sites in the brain?

These are legitimate and important questions and a great deal of energy in the debate has been expended on them. In my work, however, I show that too great a focus on these and related questions has obscured the conceptual, phenomenological, and ontological issues that arise from a broader philosophic perspective concerned with explaining subjective experience.

Return to Introduction to Mental Images Page