Anthony D. Birch

tbirch@luminet.net

www.luminet.net/~tbirch

Draft: 9/23/98

CHAPTER I

I. A Common Sense Approach

We normally understand mental images to be integral parts of our lives. Mental images are common when we recall past experiences. Recollection of unusually pleasant or unpleasant events may bring to mind mental images that have nearly the potency of the original perception. We can also experience vivid images of events that never happened, using our imagination to conjure up images of possible circumstances. Still another aspect of normal mental activity involves the use of mental images to solve problems. For example, a mechanic might mentally visualize how a part fits in a machine before attempting to install the part, or an architect might visualize the facade of a building before making a detailed rendering. From the naive point of view then, mental images unquestionably exist and are functional, integral parts of mental life.

Other evidence, not too far removed from ordinary experience, demonstrates the many benefits and uses of mental images. It has been demonstrated that athletes can improve their performance by using mental imagery. Clinical evidence shows that exercising the imagination through constructive daydreams can have important therapeutic effects for the mentally ill. Imaginary objects or imaginary scenarios also serve extraordinarily useful purposes in abstract thinking and theory formation. Einstein reported that at least one factor in his realization that the velocity of light must be the same in all reference frames came by forming a mental image of what he would see if he observed a beam of light while he pursued it at the speed of light (Shepard 1982, p. 6). Kekule credits his discovery of the structure of benzene to a mental image that occurred to him in a dream in which he imagined atoms joining into chain-like molecules that whirled about until one chain formed a ring (Shepard, 1982, p. 7). Numerous other discoveries in science, such as the structure of DNA, have been associated with moments of insight occasioned by a vivid mental image. These examples indicate that mental images are functionally active constituents of processes that produce real physical and mental effects. It seems obvious that we would be at a loss in explaining many of our intellectual or cognitive accomplishments if we could not appeal to the functional reality of mental images.

To refine the observations of common sense a bit, we can separate mental images into specific categories, already indicated in our examples: memory images, images of the imagination, and images used in problem solving or thinking. These separate categories are to be considered advisedly, however. Although we recognize a distinction between an event remembered, an event imagined, and something thought of, we also accept the fluid interchange of the "materials" -- for want of a better word -- involved in these processes. Images from our memory can become part of our creative imagination, and the images of creative imagination, along with factual knowledge and reasoning skills, can be involved in thinking.

How should we characterize the materials, or mental images, involved in these processes? Again, the common sense point of view, using ordinary vocabulary, generates an answer. We tend to refer to mental images as mental objects, or things having object-like properties. There are several reasons for this. First, mental images, like ordinary objects, often seem relatively stable. We say, for example, that we can recall the same image over and over again. Second, we speak of manipulating mental images in the manner of objects. We can, for example, mentally rotate, move, or restructure mental images. Third, we encounter or respond to mental images in ways that are parallel to, or analogous with, physical forms of interaction. We say we can "see" mental images with the inner eye of the mind, or "feel" or "smell" them with an inner sensibility that is similar to actual sensory experience. Fourth, all these object-like characteristics involved in our use of and encounter with mental images can be easily explained in ordinary language by referring to inner "objects" of the mind. The blue sky with puffy white clouds above the verdant valley that I see with the inward eye of memory must surely be something. Similarly, the imagination must have something to imagine with, or by means of, if it is to function at all. Likewise, it seems that thought processes must have constituents or parts, some of which might rightfully be called objects or images used in thinking. From these considerations, it seems a small step to conclude that there must be object-like mental entities whose presence in the mind accounts for our experience of them.

The common sense view avoids any skepticism that might be generated by referring to mental as well as physical objects. Although mental images seem to be object-like in many senses, common sense still insists that in most cases we are easily able to differentiate ordinary perception of objects "out there," in the real world, from objects inwardly "seen." The common sense view of visual perception is that it unambiguously informs us about what is out there in the real world, what is immediately present to us. Mental images, on the other hand, are understood to be things that are not currently present in the real, exterior world. Ordinary talk about mental images, I suggest, shows that we accommodate this understanding by using a secondary meaning of the verb "see" when we speak of "seeing" mental images. Often, one simply places quotes around the word to indicate our understanding that mental images are something inwardly seen, or seen with the mind's eye, rather than outwardly seen.

In addition to obvious ways mental images are used by the mind, we also have experiences of mental images for which there are less obvious explanations. Dreams are also usually classified as mental images, and their purpose or function in mental life is perplexing. The appearance in a dream image of the solution to an intellectual problem, as in Kekule's case, is decidedly rare. For the most part, if dreams are informing us or being used by the mind to perform specific tasks, these purposes are not self-evident. Unlike memory, imagination, and thought imagery, dream imagery seems to be not so much directed by us as directed at us, following its own course, and sometimes even causing us to mistake dreams for reality, at least for a brief time.

Yet, the existence of powerful dreams is not likely to sway the common man's belief that mental images are easily identified, are not likely to be permanently confused with objects of perception, and are relatively easy to sort into the meaningful and the meaningless. After all, we know that realistic dreams are similar to other forms of mental images that can be directly induced by external causes. Vivid waking hallucinations, for example, may be induced by drugs. Strong visual after-images can be induced by a flash of light. Even when we experience such strong images, we often have some awareness that these are caused by temporary changes in the nervous system, not temporary changes in external reality. Many dreams may be like these forms of induced imagery, originating from a combination of subliminal sensations and automatic interpretive mechanisms that supply a context to sensations. The subliminal sensation of a hand trapped in a blanket, for example, becomes a dream of entrapment. Speculation about their causes, however, is not the only means we have of differentiating dreams or hallucinations from reality. In many cases, the content of such images gives away their type, if not their purpose. Dreams and hallucinations frequently show us things that could not happen in reality, and this prevents us

from mistaking such images for representations of actual circumstances. Like other forms of mental imagery, I suggest that the common man's view recognizes dreams, hallucinations, and after-images as part of the same general class of mental "objects" or visual experiences that are generally known to represent things not actually present in the external world.

There are, then, from the common sense point of view, a number of points to be made about mental images. First, there are several kinds, including memory images, imagination images, dreams, hallucinations, after-images, and images used in problem solving or thinking. Second, there is an intuitive and easily-grasped sense of what mental images are and how mental imaging is distinct from perception. Mental images are often thought of as being objects of a sort, but ones that we see inwardly rather than through ordinary perception. We understand that seeing mental images is only analogous to perception, for "seeing" mental images means the inner seeing of something not actually present to the senses. Third, the experience of having mental images is so common that one is easily led to the supposition that mental images are necessary and universal; images appear to be essential elements in mental activities, without which we could not think.

II. Philosophic Questions about Mental Imagery

Philosophical questions inevitably arise when we attempt to improve on the intuitions of common sense. Let us consider some general philosophic considerations that apply to mental images.

Philosophers often object that the vocabulary of ordinary language, when used to describe mental images, is grossly inadequate. One of the most common objections is that speaking of inwardly "seeing" something is completely uninformative. Philosophers want to know precisely what the quotes around "seeing" mean. If it is not the same as ordinary perception, we need to give some specific differentiating features of this form of seeing, otherwise, we will not have said how, or even if, it can be reliably distinguished from perception.

The same objection applies to the quotes around the word "object." Again, philosophers want to know what sense to give the notion of inner mental, as opposed to physical, objects. First of all, philosophers are concerned with the issue of the reduction of mental states to physical states of the brain. Most philosophers, and many ordinary people, believe that contemporary science should be sophisticated enough to provide a means of relating all mental states to physical changes in the brain. If this can be done, then so-called "mental" objects must be physically instantiated in the brain in some fashion. If mental objects do in fact have some form of spatial or physical existence, then we need to explain this using more precise language, for the common man's language connotes that mental images are objects like pictures, and neither a common sense nor a philosophic view is likely to endorse the supposition that mental images are physical in the same way that photographs or paintings are physical. If they were, they would need a physical location, and no one, presumably, thinks that mental images are literally photographs or pictures located in the heads of those who see mental images. Ultimately, in this scenario, the common sense view degenerates into the absurdity that there are pictures in the head.

Philosophers argue that a second series of objections arise if we take the opposite view, and assume that mental objects are not physically instantiated. This leads to metaphysical problems that again challenge a common sense inventory of objects. If mental images are genuinely non-physical, our ontology must include objects that do not exist in space, but with which we seem to be directly acquainted. This, of course, was one of the considerations that led Descartes to formulate a dualistic metaphysics in which objects could be said to exist either as material or non-material substances. This, in turn, leads to the well-known philosophical problem of how to explain the interaction between the physical and the mental realms.

More philosophic questions arise when we reflect on the role of mental images in thought, even without considering the impact of possible reductionist scientific explanations. We pointed out that Einstein, for example, used a mental image to help him find an insight into the velocity of light in any reference frame. It is clear, however, that merely conjuring up this image does not fully explain how he arrived at his famous equation. If it did, the solution would be as obvious to us as it was to him. The extensive mathematical reasoning Einstein used is not immediately available to us from considering the image alone. We wonder, then, what the mental image really has to do with the thinking process. Perhaps it was not as useful as it appeared to be. Aside from reductionist considerations, the entire common-sense supposition that images in fact play a role in thought and in fact have meaning, could be mistaken.

Many philosophers have made the same sort of point in another way. Consider any sort of image, for example, a picture of a man sitting on a chair in front of a house. How do we know, philosophers ask, what the picture intended to be a picture of? Is it a picture of the man, the chair, or the house? From the picture alone, we have no way of knowing. We need to have a context in which the meaning or intent of the picture is known. In any case, philosophers argue, it is obvious that an image by itself cannot have any specific meaning. Since thinking relies on the generation of specific meanings and interconnecting them, some philosophers conclude that images themselves cannot really be involved in the thinking process.

There are no obvious solutions to the problems produced by initial philosophic reflection. It is evident that the common sense approach, if it is to be retained, requires serious reexamination. In order to set a course for further examination of the issues, let us limit the scope of the project before us. First, we can limit the investigation to visual, rather than aural, tactile, or other types of mental images, since visual images are most often the subject of philosophic debate. Second, we can think of questions about mental imagery as falling into two broad categories:

1. What is the nature of visual mental images?

2. How are visual mental images involved in thought?

By questions concerning the nature of mental images, I shall understand such issues as determining the origin, kinds, manner of existence, and general properties of mental images. By issues related to the role of mental images in thought, I shall understand the broad issue of how mental images fit in explanations of the thinking process. This includes such issues as the necessity of images in thought, the causal effectiveness of mental images, the relationship of images to linguistic forms of expression, and the place of mental images in scientific theories of the mind and brain.

III. Overview of the Thesis Defended

I shall defend, broadly, the common sense approach to mental images. The common sense approach, despite its flaws, can be supplemented with a phenomenological account of intentional states and the formal definitions and distinctions of traditional psychology. This provides the basis for a viable philosophic understanding of the nature and purpose of mental images. An outline of this defense follows, in the form of summaries of subsequent chapters.

In chapter 2, I investigate the history of philosophy and psychology of mental images. This chapter is intended to present, rather than to resolve, some of the received doctrines and problems about mental images. I describe the fundamentals of two camps, the imagists, who understood mental images as central to mental life, and the descriptivists, who claimed images could be eliminated from an ontology of the mental.

In chapter 3, I present a positive account of mental images, based on a compromise between the insights of the imagists and the descriptivists. The positive account of mental images is based largely on what I call the traditional philosophy and psychology of mental images. I argue that the nature of mental images may be understood using introspective evidence, supplemented by descriptive phenomenology and a metaphysics that includes mental objects. This approach, consistent with both common sense and the formal definition of traditional psychology, results in an inventory of mental imagery types, defined as conscious states in which we are aware of a mental presentation of the absent (a non-present "object"). The range of conscious mental imagery types encompasses a variety of phenomenological states, from quasi-sensory states ("given" or "seen" in a manner approximating actual vision), to intentional states in which the image is not so much "seen" as it is posited as a symbol of an intentional content. The notion that mental images are objects, therefore, must be understood analogically, since they lack the properties of ordinary objects and they are not literally experienced as pure visual presentations, but represent the agent's intentional awareness of both visual and conceptual content, provisionally bound together in a single form we call an image. Because they are meaningful, mental images can be among the symbolic particulars which serve as the vehicles of thought.

I conclude chapter 3 with a study of the metaphysics of mental images. The specific claims of this section are limited, but I do attempt to show (1) that we can have an explanation of mental states in terms of their physical causes even though we are lacking a reductive explanation; (2) there is a general-level resolution to the homunculus problem; and (3) there can be a significant task in psychology and in descriptive metaphysics in the absence of a fully-reduced science of the mind. I suggest that in any event, we still need a descriptive metaphysics of mental images that addresses their mode of being as intentional objects that are "in" a subject. Although I leave some metaphysical questions unanswered, I attempt to outline what considerations need to be addressed in a metaphysical account.

In chapter 4, I begin an examination of how two contemporary, opposing theories in psychology define, describe, and experiment upon mental images. This examination spans two chapters. A critique of these views helps differentiate my view from others.

For expositional purposes, it is fortunate that there is presently a debate in cognitive psychology about mental images. This debate has thrown the experimental and theoretical apparatus of cognitive psychology into a stark and revealing light. In outline, the contemporary debate is a relatively straightforward set of opposing claims. The descriptivist side has been supported by Pylyshyn, who has argued that mental images are to be explained by the operation of description-like data structures in the brain; the image-like qualities they seem to have are the epiphenomenal result of more primitive operations that have no features of true images. Kosslyn has supported the pictorialist or imagist side. His is a particularly interesting comparison to the view I defend, because he attempts to rescue some of the intuitions of common sense, specifically, the notion that we inwardly "see" mental images. Kosslyn also asserts that mental images are causally-effective elements in cognition; i.e., they are not epiphenomenal. At the same time, however, he attempts to defend a computational view of the mind. I show that this attempt fails, and that his pictorialism is just epiphenomenalism.

Chapter 5 concentrates on the experimental evidence adduced in support of the descriptivist and imagist views of mental imagery. I argue that with a few exceptions, these experiments do not support any one interpretation since (1) the models for these interpretations are underdetermined by the data; (2) the experimental design of several important experiments is flawed; and (3) there is significant experimental evidence that conflicts with the proposed interpretations. I suggest that the failure of these experiments to yield a cohesive interpretation is due to two factors. First, the ground rules for experimental data gathering adopted by the disputants themselves were extremely restrictive. Second, their evidence reflects part of the nature of mental images themselves, which, under the appropriate experimental conditions, yield an amalgam of descriptional and pictorial qualities, or, more precisely, from the side of the subject, show that subjects employ a variety of strategies to solve problems simultaneously using descriptive designators and imagery symbols in consciousness.

I conclude chapter 5 with a summary view of computationalism and epiphenomenalism based on the example theories I have discussed and the general background assumptions that drive computationalist accounts. I suggest that computationalism leads either to a form of epiphenomenalism that is exceptionally difficult to accommodate to common sense or to a self-contradictory position I call "computer-user dualism."