Structural Phenomenology:

An Empirically-Based Model of Consciousness

 

 

 

 

 

 

 

 

 

 

 

 

 

by

Steven Ravett Brown

 

 

 

 

 

 

 

 

 

 

 

A Dissertation

Presented to the Department of Philosophy

and the Graduate School of the University of Oregon

in partial fulfillment of the requirements

for the degree of

Doctor of Philosophy

 

August 2003


 

 

 

“Structural Phenomenology: An Empirically-Based Model of Consciousness,” a dissertation prepared by Steven Ravett Brown in partial fulfillment of the requirements for the Doctor of Philosophy degree in the Department of Philosophy. This dissertation has been approved and accepted by:

 

 

 

 

____________________________________________________________

Dr. Mark Johnson, Chair of the Examining Committee

 

 

 

________________________________________

Date

 

 

 

Committee in Charge:  Dr. Mark Johnson, Chair

                             Dr. John Lysaker

                             Dr. Don Levi

                             Dr. Don Tucker

 

 

Accepted by:

 

 

 

 

____________________________________________________________

Dean of the Graduate School


 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

© 2003 Steven Ravett Brown


An Abstract of the Dissertation of

Steven Ravett Brown              for the degree of    Doctor of Philosophy

in the Department of Philosophy               to be taken                    August 2003

Title: Structural Phenomenology: An Empirically-Based Model of Consciousness

 

 

Approved:  _______________________________________________________    

                                                  Dr. Mark Johnson

 

 

 

In this dissertation I develop a structural model of phenomenal consciousness that integrates contemporary experimental and theoretical work in philosophy and cognitive science. I argue that phenomenology must be “naturalized” and that it should be acknowledged as a major component of empirical research. I use this model to describe important phenomenal structures, and I then employ it to provide a detailed explication of tip-of-tongue phenomena.

The primary aim of “structural phenomenology” is the creation of a general framework within which descriptions of experiences may be organized. The work of Husserl, Gurwitsch, the Gestalt psychologists, and many contemporary philosophers and cognitive scientists reveals several basic parameters underlying subjectivity.

Chapter I argues that Husserlian methodology possesses problems both of praxis and of internal logic, and that its phenomenological descriptions cannot have the certainty he claimed. Consequently, an adequate phenomenology must incorporate empirical studies. This conclusion enables explicit transitions between empirical investigations and phenomenological insights.

Chapter II introduces the theoretical framework underlying my model. I identify four parameters applicable to all experiences: 1) the degree of volitional emphasis with which something is experienced, i.e., the intensity of our focus on it, 2) the degree of non-volitional emphasis, i.e., the degree to which it is salient, 3) a variant of intentionality I term “directionality”, and 4) the property of recursion. Experiences are embedded within a complex set of relationships that unify and direct a layered phenomenal structure. I support these claims with evidence discovered over the past two centuries of research.

Chapter III applies my model to the tip-of-tongue (TOT) state, in which difficulty remembering is accompanied by a sense of active searching. I show that a phenomenological description of the TOT experience is dependent on cognitive data, and that a phenomenological analysis is necessary to properly interpret these data.

By showing how structural phenomenology offers a perspective from which to elucidate the results of experimental studies, I hope to clarify and establish the explicit role of introspection in empiricism, and of empiricism in phenomenology.


CURRICULUM VITA

 

STEVEN RAVETT BROWN

714 Ingleside Drive

Columbia, MO 65201-5927

srbrown@ravett.com

(573) 449-9204

 

 

EDUCATION

 

Ph.D., Philosophy, 2003. University of Oregon, Eugene, OR. Dissertation: “Structural Phenomenology: An Empirically-Based Model of Consciousness.” Advisor: Mark Johnson

 

MA, Human Sciences, 1995. Saybrook Institute, San Francisco, CA, Department of Human Sciences

         

BA, Music Composition, 1986. University of Washington, Seattle, WA, Department of Music

 

ABD, Experimental Psychology, 1976. University of Tennessee, Knoxville, TN,   Department of Psychology

         

BS, Physics, 1968. Duke University, Durham, NC, College of Arts and Sciences.

                

AOS

 

Phenomenology, Philosophy of Mind, Consciousness Studies, Cognitive Science

 

AOC

 

Philosophical Psychology, Philosophy of Language, Philosophy of Science, Aesthetics

 

PUBLICATIONS

 

Brown, S. (2003) Some implications of the Gestalt conception for phenomenological methodologies. Journal of Consciousness Studies: in review.

 

Brown, S. (2002) On the mechanism of the generation of aesthetic ideas in Kant's Critique of Judgment. British Journal for the History of Philosophy. Accepted for publication in 2004.

 

Brown, S. (2002) On conference styles: personal reflections provoked by ASSC-6. Invited Commentary on the Association for the Scientific Study of Consciousness Sixth Conference. Journal of Consciousness Studies, 9, 7, pp. 50-53.

 

Brown, S. (2002) Emotive schemas: an integrative approach to expressivity in music. Metaphor and Symbol: in review.

 

Brown, S. (2002) Peirce, Searle, and the Chinese Room argument. Journal of Cybernetics and Human Knowing, 9, 1, pp. 23-38.

 

Brown, S. (2000) Tip-of-the-tongue phenomena: an introductory phenomenological analysis. Consciousness and Cognition, 9, 4, pp. 516-537.

 

Brown, S. (1999) Beyond the fringe: James, Gurwitsch, and the conscious horizon. Journal of Mind and Behavior, 20, 2, pp. 211-227.

 

Brown, S. (1996) The role of metaphor in natural languages: a theoretical inquiry. Abstract. Consciousness Research Abstracts. Thorverton: Imprint Academic.

 

 

 

Academic Appointments

 

Teaching Assistant. University of Oregon, Eugene, OR. 1998-9. Department of Philosophy. Courses taught included ethics, Greek philosophy, and 19th century philosophy.

 

Lecturer. Rennselaer Polytechnic Institute, Troy, NY. 1979. Department of Psychology.

Complete responsibility for teaching and administering introductory psychology courses.

             

Teaching Assistant. University of Tennessee, Knoxville, TN. 1974-5. Department of Psychology.

Taught classes in experimental methods and design, and elementary statistics.

 

Non-Academic Positions

 

On-line philosophy consulting. Web Service: Ask A Philosopher. 2001-present. Sponsored by the International Society for Philosophers.

 

Mathematics tutoring: high school students. Columbia, MO. 2000-2001.

 

Database instructor and trainer; freelance designer/programmer. San Francisco, CA. 1990-1996.

 

Microcomputer consultant; freelance consultant and instructor on microcomputer technology. San Francisco, CA and Seattle, WA. 1987-1990.

 

Presentations

 

Brown, S. (2002) Structural Phenomenology, Its Relationship to Language. Presented at the Association for the Scientific Study of Consciousness Conference: Consciousness: Consciousness and Language. Barcelona, Spain.

 

Brown, S. (2002) Does Phenomenology Rest on Paradox? Presented at the Fifth Annual Conference: Toward a Scientific Basis for Consciousness. Tucson, Arizona.

 

Brown, S. (2001) Structural Phenomenology: A Top-down Analytic Methodology. Presented at the Association for the Scientific Study of Consciousness Conference: The Contents of Consciousness: Perception, Attention, and Phenomenology. Durham, North Carolina.

 

Brown, S. (2000) Phenomenological Description: An Extension of Gurwitsch's Multi-Dimensional Approach. Presented at the Center for Semiotisk Forskning, Aarhus, Denmark.

 

Brown, S. (2000) Conscious Inhibition: A Synthesis of the Phenomenological and Cognitive Approaches to Mind. Presented at the Pacific Division of the American Philosophical Association. Phoenix, Arizona.

 

Brown, S. (2000) Tip-of-the-Tongue and Conscious Inhibition. Presented at the Fourth Annual Conference: Toward a Scientific Basis for Consciousness. Tucson, Arizona.

 

Brown, S. (1998) The Transduction Hypothesis. Presented at the Association for the Scientific Study of Consciousness Conference: Neural Correlates Of Consciousness:

Empirical And Conceptual Issues: Bremen, Germany.

 

Brown, S. (1997) Tip-of-the-Tongue Phenomena: An Introductory Phenomenological Analysis. Presented at the Fifth International Cognitive Linguistics Conference. Amsterdam, The Netherlands.

 

Brown, S. (1996) The Role of Metaphor in Natural Languages: A Theoretical Inquiry. Presented at the Second Annual Conference: Toward a Scientific Basis for Consciousness. Tucson, Arizona.

 

 

PROFESSIONAL SOCIETIES

 

American Association for the Advancement of Science

Association for the Scientific Study of Consciousness

Society for Philosophy and Psychology

American Philosophical Association

International Neural Network Society

The Cognitive Science Society

 


 

ACKNOWLEDGEMENTS

 

 

I would most especially like to acknowledge the support of my wife, Shanna Helen Swan, emotionally, intellectually, and financially. This might have happened without her; but neither so quickly nor easily. I would not have obtained this degree from the University of Oregon, in particular, if it were not for her motivating me to reach out to an unfamiliar world. This dissertation, then, is not only my child, but hers also.

 

I also wish to acknowledge the unusual flexibility and tolerance of Mark Johnson in admitting a fairly abnormal candidate to a doctoral program in Philosophy, and in recognizing my sincerity and motivation in the face of both external criticisms and my own limited vision. His continuing support was invaluable and his ideas were always stimulating and rigorous. In addition, Dr. Johnson’s embracing of a variety of areas, approaches, and collaborations in the field of philosophy is both remarkable and courageous.

 

 


 

TABLE OF CONTENTS

 

 

Introduction__________________________________________________ 14

Chapter one____________________________________________________ 21

CRITIQUE OF CLASSICAL PHENOMENOLOGY__________________ 21

I. Historical Background and General Issues_______________________ 21

II. Apodicticity and Axiomatization – the Ideal Case?______________ 25

III. Is Science Axiomatic? What Kind(s) of Inquiry Should It Be?_____ 28

IV. Phenomenological Methodologies and Why They Are Covertly Empirical_________________________________________________________ 38

V. But What of Phenomenological Data? Why That Data Is Not Apodictic_________________________________________________________ 44

VI. The Big Problem: Phenomenological Methodology Is Inadequate to Investigate Its Own Assumptions____________________________________ 49

1. Introduction and Outline of the Argument____________________ 49

2. Background: The Constancy Hypothesis and the Origin of Gestalt Psychology_____________________________________________________ 52

3. Gurwitsch’s Gestalt Psychology_____________________________ 55

4. Modern Gestalt Psychology Has Refined But Not Radically Altered Its Principles____________________________________________________ 58

5. First Statement of the Problem: Husserl’s Essentialism_________ 62

6. Examples of Some Problems with Essentialism________________ 66

7. The Heart of the Problem: Husserlian Methodologies are Atomistic and Atomism Is Not Compatible With Gestalts____________ 70

VII. Now What? The Case for Empiricism and the Gestalt___________ 74

Chapter two___________________________________________________ 80

The Structural Analysis of Consciousness_____________ 80

I. General Introduction_________________________________________ 80

II. The Structural Analysis of Consciousness: Introduction to the Parameters_______________________________________________________ 85

III. Some Introductory Examples and Analyses____________________ 91

IV. The Empirical Basis of Intensity_____________________________ 101

V. The Empirical Basis of Recursion_____________________________ 112

VI. The Empirical Basis of Directionality_________________________ 121

VII. A Short Summary_________________________________________ 130

Chapter three________________________________________________ 132

Tip-of-the-Tongue Phenomena and Structural Phenomenology_______________________________________________ 132

I. Introduction________________________________________________ 132

II. Explication of the TOT State: Definitions and Clarifications______ 135

III. Explication of the TOT State: The Verbal Conception Extended__ 138

IV. The TOT State in Detail_____________________________________ 143

1. Etiology_________________________________________________ 143

2. Components_____________________________________________ 148

3. Its Resolution____________________________________________ 152

V. The Phenomenon of Protension and Its Universality____________ 153

VI. Protension and the TOT: The Necessity for a Goal-Directed Process_________________________________________________________________ 159

1. Introduction: There Are Goals______________________________ 159

2. General Specifics of Goal Convergence: Goals Aid Retrieval___ 161

3. General Specifics of Goal Convergence: Goals Aid Error Evaluation and Correction_______________________________________ 163

4. Some Specific Generalities from These Processes: Directionalities Describe the Details_____________________________________________ 165

5. Some Specific Generalities from These Processes: On Goals and Gestalts_______________________________________________________ 171

6. Consequences: General Structural Principles_________________ 174

VII. The TOT State: Implications from Structural Phenomenology__ 174

VIII. The TOT State: Closing Remarks___________________________ 178

Chapter Four_________________________________________________ 182

Conclusions_______________________________________________ 182

I. What I Have Attempted to Do and Why, Generally______________ 182

II. What I Have Attempted to Do and Why, More Specifically______ 185

III. What I Will Attempt to Do: Future Directions_________________ 188

 

 


 

LIST OF FIGURES

 

 

Figure 1: Illusory Figures............................................................................................. 60

Figure 2: Dalmatian....................................................................................................... 95

 

 


 

 

 

 

Introduction

 

If we are to approach the study of mind, how are we to do so? Perhaps the earliest efforts were due to various religions attempting to foster the attitudes of worship, reverence, receptiveness to religious feelings and/or revelation, or their claims concerning the best attitudes toward various aspects of life and how to acquire them. In order to accomplish these aims, religious practitioners needed insight into their own minds and that of potential converts. Much of clinical psychology and various therapeutic approaches are also oriented toward this healing and controlling aspect of mind. These approaches have been concerned with mental experiences, primarily the results of introspection, and have attempted to analyze those experiences into types, subtypes, and so forth, and note the interactions and relationships between those classes of experiences.

Let us consider, most generally, three possible approaches to the study of mind. The traditional analytic philosophical approach, usually termed “philosophy of mind”, attempts to present and resolve general questions about mind in the abstract and its relationship to the world. Modern analytic philosophy, insofar as it is concerned with mind, has taken the approach of attempting to abstract and formalize various mental characteristics and properties. This approach has perhaps culminated in the construction of the digital computer, which was initially envisioned as a realization of the principles of logical thought.

Second, we may approach mind in terms of its embodiment, in very focused and specific terms. The study of what might be termed the “mechanics” of the mind – that is, the construction and validation of abstract systems attempting to model mental processes - has progressed enormously since the time of Tarski, Carnap, and other pioneers of formal logic who conceived this approach (e.g., Tarski, 1956; Carnap, 1961). Contemporary empirical approaches tie the analysis of mind to experimental procedures, to behavioral analysis, and to the elaboration of models designed to analyze and predict experiences. That is, this approach investigates the interaction and generation, by the brain, of the mind. This area, starting with Watson and Skinner, has moved, after Chomsky’s devastating critique of Skinner (Chomsky, 1967), to a realm which treats specifically of the mental, but couched in terms and explored in investigations designed to evoke and involve the physical. For example, there is a huge literature on the phenomenon of “attention” which involves extremely ingenious and precise experiments involving the timing of reactions, perception, orientation, and so forth. There is an equally large literature investigating “memory”, involving equally ingenious experiments with various arrangements of words and objects, contextual effects on memorization, and many more areas. Both of these above areas are also tied to experiments in physiology, psychopharmacology, and neurology: to the brain. Contemporary cognitive science is still heavily influenced by the descendents of the analytic philosophers, viz., the artificial intelligence theorists. This conception of mind, based on formal operations and atomistic elements, is alive and well today, and although there are attempts to realize less atomistic approaches they are still in the minority[1]. Gestalt psychology was the precursor of some of this material, and it does survive today, albeit greatly changed and elaborated, as we shall see.

However, a third area, which I mentioned briefly above, has been largely neglected as an explicitly empirical arena. That area is the analysis of mind as experience, that is, the application primarily of various introspective techniques to elaborate and interrelate our thoughts, feelings, sensations, in short, whatever it is that we experience consciously. I will not at this point touch on the question of whether there are “subconscious” or “unconscious” experiences or whether the term “mental” may be applied to processes of which we are at some point not conscious. The use of introspection, by and large, for a variety of reasons, has been an implicit and indeed an unmentioned and even hidden aspect of many fields. That is, in the early part of the 20th century introspective approaches were explicitly employed in psychology, although subsequently, for reasons I will touch on later, they fell into disrepute. Since then, introspection, despite its wide use in a variety of fields, has become “out of fashion”, largely unmentioned and unmentionable. The study of the contents and structure of experience, while it has significantly progressed since Husserl, Gurwitsch, and others, has not been recognized as such, nor, until very recently, been seen by other fields as a valid domain in its own right.

Would it not be easier to study behavior, or to abstract from the mental to formal systems, as analytic philosophy as done, to confine ourselves to the “mechanics”? Pure behaviorism, as the sole route to understanding the mind, has been thoroughly discredited. Chomsky, Fodor, and others have argued persuasively that behaviorism is simply insufficient (e.g., Chomsky, 1967; Fodor, 1975); that mental constructs are necessary in both theoretical and empirical studies of the mind. What about confining studies of the mental to abstractions employing formal constructs, as Chomsky, Fodor, and many others would have us do? Formal logic, generative grammar and its modern descendants, artificial intelligence, and thus any abstract approach to the mind does not need, one might claim, to descend to the level of “raw feels”, individual qualities, i.e., pure introspective data.

Why, in other words, explicitly employ introspection and its results? There are at least two general reasons for this, I believe. First, the constructions used in more abstract systems must be validated. Surely this can only be accomplished through introspective studies. Thus, we find that in linguistics it is common to refer to one’s “intuition” about the grammaticality of a construction, about the “proper” term to employ in a sentence, and so forth. In cognitive science, one must abstract from something, namely, if not behavior, the introspections of one’s subjects. That is, when one obtains data from subjects in cognitive experiments, one typically either obtains something like reaction times – behavioral data – or data from, for example, verbal reports of the subject’s memories, their judgments of various psychophysical qualities, or their judgments of category membership – all of the latter involving introspection. Those studies, and similar ones, do then employ introspection, but under the rubric of “raw data”, “individual subjects’ results”, or some such. However, this is merely introspection by another name (on this, see also Jack and Shallice, 2001). Second, it is ultimately the mental that we wish to describe, at least inasmuch as we are interested in the meanings of words, in our emotions, sensations, and in the processes by which we arrive at those meanings, feelings, and other aspects of mental life. Why pursue psychology at all, first, if we assume no mental life, and second, if we do not then attempt to describe that mental life? We are, in that case, merely reverse-engineering biological computers. But if that former pursuit is at least part of psychology’s[2] ultimate aims, then we must, ultimately, anchor our data, our constructs, and our theories, as well as our predictions, to that very mental life we are attempting to explicate.

More specifically, if we look at the various fields involved with what I have termed “mechanics”, we find that visual perception, for example, has made great strides in extending gestalt theory, although vision studies are not usually considered studies of the mental. Yet inasmuch as that field concerns itself with liminal data, with data on color discrimination, with interactive data such as Stroop tests, and many other data based on visual experiences, it must begin with the results of subjects’ introspections. Linguistics, which, insofar as grammar and semantics is concerned is directed at our experiences of word meanings, of the “feeling” of the correctness of a grammatical construction, and so forth, has developed rigorous techniques of introspection and evaluation of introspective data; it is, I will claim, the child of phenomenology, if one largely[3] unaware of this particular parent. And the sub-discipline of cognitive linguistics has in fact explicitly attempted to analyze, to some extent, the content of consciousness (e.g., Fauconnier and Turner, 2002; Turner, 1996). Yet even here its practitioners do not normally see themselves as the inheritors of the tradition of Meinong, Husserl, James, Gurwitsch, and others, although they are continuing the investigation of exactly what those philosophers began. Thus, even though I will argue in detail below that Husserl’s aims overreached the capabilities of the subject, I vigorously maintain that he nonetheless contributed enormously to the very important area of what might be likened to the grammar and semantics of the analysis of our conscious experience. Further, to continue the expansion of this area of study, phenomenology, on the one hand, must recognize that, as in linguistics, techniques of introspection and evaluation can be both realizable and valuable if they are thought of not as esoteric and unique philosophical methodologies providing ultimate ontological knowledge, but as subsets of a large array of techniques that have been developed in a variety of fields which enable us to study the mental, experienced, aspect of mind. On the other hand, fields such as cognitive science need to recognize that one goal of their endeavors is the explication of the structure of the mental, and that phenomenology, in content and technique, can contribute to this explication (“The starting point for work on consciousness is introspection and we would be foolish to ignore it”[Block, 2001, p. 203]).

Phenomenology, then, aims, at least in part, at identifying, classifying, and analyzing what we might term the components of our mental experiences – our thoughts, sensations, feelings, and so forth - sometimes with the goal of predicting them, sometimes merely with the goal of classifying them in order to interrelate them and to relate them to other aspects of life. In addition, some branches of phenomenology claim access to privileged knowledge about the world. However, I would like to suggest, in this essay, that the phenomenological and the empirical may be quite usefully joined, and I argue, below, for the utility of some aspects of phenomenology. The application of the scientific method, involving the replication of observations and experiments and the testing of theories, in short, with systematicity and consensus, has finally begun to enable some limited degree of predictability in the phenomenological arena. Conversely, introspective inquiries are now, as I will illustrate, virtually ubiquitous in the empirically oriented studies of the mind. But my essay will not be confined merely to pointing out what is, really, a fact which is rather obvious to anyone studying the areas of cognitive science, linguistics, computer science, and so forth, i.e., the ubiquity of introspection in those fields.

The phenomenological movement has seen itself as possessing a unique methodology which enables its members to answer to some of the most pressing questions in philosophy by either discarding or reorienting some of the most accepted claims and assumptions of the Western philosophical tradition. Starting with Husserl, phenomenology has addressed problems which it sees as resulting from these claims by recasting and in some cases denying them. Husserl spent much of his time attempting to show that much of traditional philosophy and its modern children, the scientific method and the field of psychology, have taken pathways which inevitably result in incomplete and even problematic pictures of reality. However, there is a great deal of skepticism today about Husserl’s claims and the claims of phenomenology in general. The promise of phenomenology, to thoroughly revise philosophy and to base it on certain and clear truths, has not, by all accounts, come to pass, nor, if its numerous critics are correct, can it ultimately do so. It is my belief that for the most part, this skepticism is justified. From the beginning of the last century, roughly, the progress in sciences now termed psychology, cognition, and artificial intelligence, have, despite notable failures and shortcomings, far surpassed what Husserl thought science capable of, and the empirical disciplines now investigating various aspects of the mind encompass not merely an enormous variety of subjects, but subjects which Husserl could not conceive science as capable of studying, for example, individuals’ sense of the meanings of their lives. Further, experimental methodology in the sciences has been expanded and refined so that empirical studies can not only include such topics, but study them with unprecedented accuracy.

In light of such problems, why take phenomenology seriously? As a route to a new philosophy it does indeed seem a dead end; as a investigation of mental contents, much of its purview has been surpassed by cognitive studies. If phenomenology is to a great extent not philosophy, or is a philosophical movement which has failed in great measure, then what function can it serve? Most generally, I will answer this question by arguing that phenomenology must be “naturalized”, i.e., it must become a major aspect of the explicit integration of introspection into empirical research.

In order to explicitly and clearly unite introspective studies with empirical studies, several steps are necessary. First, the claims of modern phenomenology for the uniqueness of a rather specialized set of techniques in terms of their structure and findings must be evaluated and put into the perspective of an empirical science which has altered and matured since those techniques were formulated. I will argue that the practices of traditional phenomenology are neither clear, easily communicable, nor, finally, substantially different from those of the empirical sciences.

Second, modern phenomenology makes very strong claims about the privilege of its techniques[4] over those of the empirical sciences. I evaluate those claims, and in the process show them to be either doubtful or simply erroneous.

Third, I will argue that in the light of developments in Gestalt theory, the assumptions underlying the above claims lead, in fact, to a vicious circle when phenomenology itself is called on to support or defend them.

These first three steps will be taken in Part I of this essay. In other words, Part I will essentially consist of two sections. First, I will attempt to indicate the points in Husserl’s, and thus traditional phenomenology’s, approach which fail in accomplishing his aims, and argue that it is just those areas which imply that phenomenology must be turned from its previous course to one bringing it under the umbrella of contemporary empirical studies. This section deals with a general explication of Husserl’s aims, with modern philosophy of science, with intersubjectivity, and finally with apodicticity. The next section will show how consideration of some of the problems in Husserl’s treatment led Gurwitsch[5] to attempt to fuse the empirical study of gestalts with Husserl’s phenomenology. However, we will come to see that Gurwitsch’s approach, in its attempts to conserve many of Husserl’s insights which I will have previously criticized, falls short of fully bridging the areas of traditional phenomenology and empirical studies of mind. In addition, we will find that, contrary to his desire to strengthen Husserl’s position, that latter position is substantially weakened by Gurwitsch. Thus, as a consequence of Gurwitsch’s consideration of Gestalt principles, Husserl’s classical phenomenology will be discovered to involve a vicious circle inasmuch as it makes claims to apodicticity. On the other hand, I consider Gurwitsch’s contribution, the then-radical idea that phenomenology must depend to some limited extent on empirical studies and its detailed development, which served both to alienate him from the philosophical establishment and to initiate the synthesis of phenomenology and cognitive psychology, one of the most important in the field.

Part II will start with the fourth step, that of justifying the general introspective approach, and some of the results of phenomenology in the light of the empiricism of the latter half of the 20th century. I will support, with modern empirical data, some of the claims of Aron Gurwitsch concerning the applicability of Gestalt theory to phenomenology, although I will have shown in Part I that his arguments concerning phenomenology’s privileged position are in error. Part III, the fifth step, will consist of a more detailed explication of the empirical phenomenology[6] resulting from the first two parts, and its relevance to some specific areas in modern cognitive science, to illustrate the usefulness of the reciprocal application of phenomenology to cognitive science, and cognitive science to phenomenology. As a concrete illustration, I will employ insights from both phenomenology and cognitive science to describe and in part explain an experience characterized by William James as the tip-of-tongue phenomenon.

Part IV will consist of my own theoretical work in structural phenomenology, which starts with an approach similar to [7] that of Gurwitsch. I will show that a top-down analysis based on data derived from introspection and from gestalt considerations can lead to an extremely detailed, yet quite general, analysis of conscious experiences which is applicable to a wide variety of areas, including the cognitive sciences.

It is not possible for me to do justice to Husserl’s output, which is both enormous and varied, not merely in scope, but in the alterations of his position over the course of his career. However, I feel that it is necessary, at the beginning of an essay intended to support a change of direction in phenomenology, to deal with what I see as major problems in the position of phenomenology’s founder. Thus, while I will touch on an array of issues, proceeding from general to particular, my topics will actually be quite limited relative to the scope of Husserl’s work. However, the issues I will deal with are some of the broadest and most important, and it is necessary that he be critiqued from a modern standpoint so that his results may be adapted to the exponential increase of knowledge and methodology in the last several decades. These issues include that of Husserl’s view of science versus the viewpoints of some modern philosophers; questions about the implications of intersubjectivity and how that relates to both the knowledge and the communication of phenomenological studies; Husserl’s notion of apodicticity and how that must be modified in the light of both logic and modern knowledge; modern implications of the constitutional dynamics of the mind; and finally the notion of essences, and how the traditional essentialism still largely embraced by phenomenology (and other fields) must be seriously reconsidered. I will employ, instead of extended arguments, short indications of how those arguments should proceed, and in addition references to those commentators on Husserl who have provided much fuller expositions of those arguments. This series of short summaries will serve, I hope, as an introduction to one of the main themes of this essay, viz., that work subsequent to Husserl has demonstrated that what might be broadly termed “gestalts” must dominate analyses of phenomenal consciousness[8]. Further, the nature of these entities has profound consequences for the work of Husserl and his successors, and, along with other factors, leads to the necessity for integrating phenomenology into the empirical studies of the mind which have broadened in scope so dramatically since Husserl’s time.


Chapter one

 

Critique of Classical Phenomenology

 

I. Historical Background and General Issues

 

Husserl’s conception of the ideal science[9] was the classical picture, held until Kuhn’s (Kuhn, 1964) well-known critique of the classical view of science, and Gödel’s notorious undecidability theorem (Gödel, 1992). Until those and similar critiques opened a floodgate of criticism towards the traditional conception of science, the structure of an ideal theory was understood to be a hypothetico-deductive system, i.e., a system employing a set of well-defined postulates and operations which are elaborated to deduce and explain its purview. Thus, Husserl states, “No reasonable person will doubt the objective truth or the objectively grounded probability of the wonderful theories of mathematics and the natural sciences.” (Husserl, 1965, p. 74; see also Seidler, 1977, p. 308). Popper writes, “… the form of a rigorous system is aimed at. It is the form of a so-called ‘axiomatized system’…” (Popper, 1968, p. 71). One of Husserl’s primary aims, then, was to put philosophy on this footing. He criticized philosophy as having “a lack of clarity of perfection in the systematic ordering of proofs and theories” (Husserl, 1965, p. 74). Underlying his attempt to put philosophy on a footing comparable to mathematics was his conviction that one needed to establish clear, indeed, apodictic (“necessary, a priori, and absolutely certain, indubitable evidence” [Levin, 1970, p. xviii])[10], elements for philosophy, analogous to the elements of, say, Euclidean geometry, and that without such basics, philosophy would be at best unclear and at worst no more than opinion. Thus, he states, “One knows and approves of the mathematical style of thinking…. It is toward this style that we orient our concept of the a priori.” (Bernet, et al., 1999, p. 79). If Husserl’s aim is the correct one, and the ideal system is axiomatized, then one must first determine what kinds of entities make up the basic elements of a philosophy, and one must also determine how to find or describe those entities.

Strongly influenced by Kant, desiring to resolve the Cartesian doubt (see below), Husserl took as his starting point what he considered to be the universality of some of our intuitions. That is, in order to address the Cartesian skepticism about the dubiousness of evidence for objectivity and the foundations upon which to base such evidence, Husserl took the (supposedly) self-evidently given truths of mathematics and logic as not merely the exemplars, but as the starting points for apodicticity in philosophy. As a mathematician, not only did he feel that mathematics provided the best exemplar of a system, and evaluated all other attempts at systematicity by that standard. He thought that the necessity for induction was a near-fatal flaw in empiricism, inasmuch as an empirical area proposed to be a science. Thus comments such as, “It is the proper achievement of the phenomenological reduction… to keep methodically to the pure givenness of consciousness” (p. 61) follow from that rationale. In a similar vein, Smith comments:

 

Our knowledge of such a priori propositions is gained by means of what Scheler calls an 'intuition of essences' of the sort that is involved, for example, when we grasp the colour red and grasp that it is different from green or blue, or when we grasp the essential interconnection between red and visual extension. We do not have to observe and check and carry out inductions in order to grasp that red is different from green, or that jealousy is different from greed (Smith, 1996).

 

One must bear this viewpoint in mind when reading Husserl’s various critiques of philosophy and psychology (e.g., Husserl, 1965; Husserl, 1970; Husserl, 1998b, pp. 33-50).

More specifically, given the conviction that one can begin with apodictic intuitions similar to mathematical and logical conceptions, philosophy’s content should not be based on facts, i.e., on empirical – inductive – knowledge, or anything else which might be uncertain. Its content, as well as its structure, must be as unquestionable as that of mathematics. “The mathematician abstains in principle from every judgment concerning real actuality and, instead of actuality, concerns himself with ideal possibilities and their related laws” (Bernet, et al., 1999, p. 79). And so, ideally, then, must the philosopher. Thus, in their dependence on “real actuality”, lay, according to Husserl, the weakness of the empirical sciences. As we shall see, Husserl attempted to explicate, employing the concept of the “lifeworld”, just how this uncertainty enters not merely science, but virtually all varieties of experience. He can thus claim, “…historical reasons can produce only historical consequences. The desire either to prove or to refute ideas on the basis of facts is nonsense…” (Husserl, 1965, pp. 126-127). Further,

 

if it is to be called ‘knowledge’ in the narrowest, strictest sense, it requires to… have the luminous certainty that what we have acknowledged is, that what we have rejected is not…. We also speak, e.g. of an act of knowing where the judgment we pass is associated with a clear memory that we previously passed a judgment of precisely the same content accompanied with inner evidence (Husserl, 2001a, p. 17).

 

This is the origin of Husserl’s approach to abolishing the Cartesian doubt. The elaboration of this answer to Descartes was indeed ingenious. Husserl abolished any consideration of the “actuality” of objects, in order that philosophy be able to treat not merely what Descartes believed apodictic, viz., one’s mental contents, but also their referents, i.e., actual objects, as apodictically certain. A mental act termed “bracketing” or “epoché” (epoch: e.g., Husserl, 1998a, pp. 219-220; Husserl, 1998b, p. 34) was aimed at taking the intuition of actuality out of play, in a sense (“The aim hereby is to bring into view the pure, immanent, constitutive subjectivity which would be ‘left over’… even if the world did not exist…” [Bernet, et al., 1999, p. 67]). This act, if successful (and I will have much more to say on that point below), is then envisioned as effectively altering the status of objects to that of exemplars of ideal entities, similar to the way that specific examples of figures or formulas exemplify more abstract mathematical ideas. One can then proceed to intuit or intuitively grasp the meanings[11] of those objects (or, more precisely, the essences of abstractions of particular objects – analogously to grasping the essence of triangles from specific drawings of triangles[12] – and see next paragraph) just as one grasps the meanings of mathematical terms. Objectivity then is freed from the Cartesian doubt, and one can (I assume) then resupply the component of actuality to the exemplars, in the certainty that one has grasped their “essence”. Husserl realized that one cannot utterly ignore or forget that some objects are “real” and some are “mental”, but he decided that one could suspend or ignore that property and focus instead on all other characteristics of the phenomena. Thus the Cartesian doubt might be, in effect, bypassed, and certainty obtained in the “givenness” of objects, i.e., in what we are presented with, whatever its origin.

Specifically, once particular objects have been bracketed through the epoché, the next step is to show precisely how, from these particulars, abstractions comparable in certainty and clarity to those of mathematics may be derived. To this end, Husserl outlined the idea of the method of “eidetic variation” (Levin, 1970, p. 84), “essential seeing” (Husserl, 1977a, pp. 339-347), or “free variation” (p. 340). In this method, specific phenomenal objects[13] are altered “freely” in a variety of ways (e.g., see also Ihde, 1977, pp. 86-87, pp. 100-103). The results of that process of variation, those diverse objects, enable one to arrive at[14] the commonality – or alternatively, to generate the abstraction - underlying those variations[15]: the “eidos”, or “essence” of a phenomenon. As an example, one might want to find the essence of lamps. Now, in order to avoid the problem of circularity, one could not, strictly speaking, set out to find that specific essence [16]. Instead, one must let variations start with a particular lamp, and vary, say, the appearance of that lamp, imagining it as taller, differently colored, and so forth. As the variations increase, one will spontaneously become conscious of some central core, or an abstraction, i.e., an “essence”, which may indeed be that of lamps… or perhaps of lighting in general, or of desktop furniture. This spontaneity, not merely in the creation of variations, but in the grasping of some essence thus revealed, as I have said, is recognized as necessary by Husserl in order to avoid the problem which results from starting with a particular idea of an essence. If that latter were the case, the method of arriving at essences – eidetic variation - would be invalidated, since the essence guiding the variations would itself have no generating methodology, or a regression would start. I will touch on some of the problems with this methodology as I proceed. I must emphasize here that despite the essence corresponding to a type, it is clear that essences must be present – at the least, after their realization - in all the individual phenomena. Otherwise one could not[17] recognize those phenomena as such, i.e., as having that essence[18].

Suffice it to say that the essence also seemed to solve the problem of relating the stability of nature to the flux of phenomena in consciousness (i.e., “how there can be a science of essentially fluid objects” [Seidler, 1977, p. 316]), since it (according to Husserl) remains constant and is present within all exemplars. In addition, since these essences are derived, given the initial bracketing, without the assumption of objective existence, they are independent of that assumption, and thus the resulting system is non-empirical, analogous to mathematics. These essences, in this view, are the equivalents of something like lines or planes in geometry, i.e., axioms, abstractions from experience which can then be employed as elements first, in intuitively true and certain (i.e., apodictic) philosophical descriptions, and second, in[19] the deductive generation of axiomatized philosophical systems. Husserl, however, stopped short of that last step, contenting himself with the general explication of methodological issues and the beginnings of what can be termed “Husserlian” descriptive phenomenology. He explicitly states that “deductive theorizings are excluded from phenomenology… non-intuitive modes of procedure of any kind, only have the methodic function of leading us to the matters in question upon which a subsequent direct seeing of essences must make given.” [my Italics] (Husserl, 1998b, p. 169). That is, “philosophy, as the foundational science, has to be morphological and not nomological, descriptive and not deductive” (Mohanty, 1978, pp. 300-301). It is the determination of essences upon which phenomenology is focused; subsequent philosophizing may then, one assumes, make use of those essences deductively.

There is an enormous multiplicity of assumptions behind this methodology, virtually all of which, I will argue, can be cast into doubt, if not simply refuted, given empirical and philosophical advances since Husserl. I will proceed through these assumptions and provide brief critiques of many of them. I will spend more effort explicating the problems stemming from the gestalt nature of perception and cognition, since this latter data serves both to cast severe doubt on the apodictic nature of the results of his methodologies[20], and to introduce an empirical approach to phenomenology similar to Gurwitsch’s.

 

II. Apodicticity and Axiomatization – the Ideal Case?

 

I shall start by very briefly considering Husserl’s position vis-ą-vis mathematics and axiomatized systems [21]. If mathematics is not, even ideally, organized as an axiomatized system, as Husserl held it to be, then to embrace that kind of system as an ideal for knowledge would simply seem to be incorrect. Husserl’s ideal system is based on what he termed “apodictic” insights, that is, the intuitive grasping of certain essences, i.e., characterized by their “absolute unimaginableness (inconceivability) of being otherwise” (Seidler, 1977, p. 311). Thus, “to constitute an object is to see it, to present it in a luminous intuition, absolute, adequate, and apodictic…” (Levin, 1970, p. 30). “Apodictic insight has the characteristic, therefore, of being what Husserl calls ‘necessity’…” (p. 41). Further, apodictic knowledge is necessarily adequate and complete: ”An apodictic insight (judgment) is the outcome of a special process of modalization [eidetic variation] performed upon an adequate evidence… only complete (adequate) evidence could demonstrably sustain the insight that it is final, indubitable, apodictic” (p. 84).

However, to start with, I will illustrate below that mathematics is not obviously apodictic. Given that, one must question the possibility and indeed the desirability of phenomenology’s being such. If mathematics itself, Husserl’s model for certainty via the intuitive clarity and universality of its concepts, is not necessarily apodictic, how can phenomenology’s insights be? Later, we will see that it is indeed possible to argue that neither Husserl’s, nor any type of introspection, may be considered apodictic.

Let us begin by asking whether mathematics is axiomatized. What does one mean by “axiomatized”? An axiomatic system starts with a finite number of well-defined symbols, terms, and operations, which are applied to a well-defined set of consistent postulates (axioms). From these, through the processes of deduction and induction, a consistent system is generated. Now, without going into questions involving the nature of well-defined terms, the origins of the meanings of the initial terms, the nature of the deductive and inductive processes, and so forth, all of which are quite complex problems which have long histories of debate, one may simply ask, “What is ‘mathematics?’” If mathematics is defined as one particular set of such axioms, worked out into theorems, we might well consider that mathematics is axiomatized, despite considerations such as Gödel’s proof. But “mathematics” refers, at minimum, to the set of all such systems. That set is not axiomatized, since there are subsets within it which have mutually contradictory[22] axioms. Then one might ask whether phenomenology should conform to the structure of one such set, so that it may proceed without contradictions[23]. Husserl provides no answer to this, but it is easy to maintain that this cannot be the case; even considering just one person’s phenomenal and conceptual content, there are contradictory ideas[24], and this is well-known to be the case between persons. For example, suppose that a Marxist and a Catholic were asked to arrive at the essence of the concept of religion, of a god, of society, or of morality. Could one seriously expect them to agree on identical, or even similar, essences, i.e., results of the method of eidetic variation, for these concepts? What of a single person who is now a Marxist and was previously (say, 10 minutes ago, just before changing their mind) a Catholic? Thus, there cannot be a single non-contradictory set of essences in phenomenology. But then phenomenology cannot be axiomatized in the sense of being a single axiomatic system, and deductive/inductive operations will not, by themselves, serve to work out the implications or consequences of phenomenological essences. That conclusion alone, simple as it is, casts doubt on Husserl’s program: what other operations are there which produce the certainty of results which Husserl desired, aside from taking all statements in such a system, not as derived or inferred theorems, but as individually determined through apodictic insight? In that case, one would have to first derive the insight through such inferences, then apply one’s “seeing” to it in order to determine or to confirm its apodicticity. Yet that process of derivation, if based on the mathematical model, cannot result in contradictions. We see, then, why Husserl may have limited his phenomenology to description; it is now clear that phenomenology cannot be an axiomatic system in the mathematical sense[25].

Next, are mathematical ideas apodictic? Again, let us consider the relationship between systems, in this case, two contradictory geometrical systems: Euclidean, i.e., planar, geometry versus a non-Euclidean, e.g., spherical geometry. These are well-known examples, today, of internally consistent, mutually contradictory systems, both of which describe the world, i.e., actuality, accurately within certain parameters. Euclidean geometry, as far as is known at present, describes the universe both on small (i.e., human) scale and as a whole (the universe is now thought to be flat), while spherical geometry describes space on an intermediate scale near a gravitational field. The first follows from one formulation of a postulate[26], roughly, that perpendicular lines intersect at right angles. The second follows from another, roughly, that perpendicular lines intersect at greater than right angles. Are both of these clear, certain, intuitively grasped, inconceivable as being otherwise, i.e., apodictic? If that is true now, it certainly was not prior to the work of Riemann, Lobachevsky, and other modern mathematicians, when the first version of the “perpendicular postulate” was the only one accepted unquestioningly. In fact, it was only the relevance of the second version to general relativity that caused it to be taken seriously. What can “apodictic” mean, in this context, when two[27] mutually contradictory systems are internally consistent, and both correspond to actuality? It is interesting to consider that if these two geometrical systems are now both apodictic, but were not[28] previously, then the nature of that kind of “certainty” is an empirical one[29]. Further, if mathematics can so easily be shown not to be apodictic[30], what of phenomenology, in which mutually contradictory (as I indicated above) essences are derived, as we shall see, from difficult, vaguely described, and theoretically questionable procedures? If mathematics itself is not based on apodictic truths, then the broad aim of Husserl’s project, i.e., providing a basis for philosophy in a certainty comparable to mathematics[31], would seem to be somewhat more doubtful. That is, if mathematics is not apodictic, why should phenomenology be? I will treat this latter question in more detail below.

 

III. Is Science Axiomatic? What Kind(s) of Inquiry Should It Be?

 

Let us move now to a brief consideration of the actual nature of science, as praxis, and compare that with Husserl’s understanding of science. That latter involved the notion of perspective or viewpoint, in the sense that all individuals experience the world from a particular perspective: the “life-world” (Lebenswelt). I will quote at length here because of the importance of this conception.

 

As scientific themes, nature and mind do not exist beforehand; rather, they are formed only within a theoretical interest and in the theoretical work directed by it, upon the underlying stratum of a natural, pre-scientific experience… it is necessary to begin with this concretely intuitive unity of the pre-scientific experiential world…. If one had always returned to the complete original concretion of the world, as it is always experienced in naēve originality, and if… one had never forgotten this concretely intuitive world as their field of origin, the absurdities of naturalistic psychology and socio-cultural science would not have been possible… (Husserl, 1977b, pp. 40-41).

 

Or, as Bernet puts it,

 

That is to say, the world which is itself experienced primordinally and which is able thus to be experienced by the individual subject in abstraction from the traditional, intersubjective system of communication. …typical, vague, primary universality, which is sufficient in everyday life. …the world situated prior to all sciences and their theoretical intentions, as a world of pretheoretical intuition (Bernet, et al., 1999, p. 221).

 

Again, we see that this concept is derived, in part, from an understanding of science and the scientific process, viz., one “descriptive of the positivism at the beginning of the century”, Seidler, 1977, p. 312, which has radically altered since his time[32]. The problem, as Husserl saw it, was that scientists, as a result of their emotional and intellectual inclinations, their training, and their participation in the enterprise of scientific investigation, have a particular set of perspectives on the world, inasmuch as they function as scientists. These various perspectives may contrast with their and others’ “primordinal”, i.e., immediate, everyday experiences, yet the abstractions that science deals with are based on, indeed, derived from, those experiences. But since they are one particular (“objective”) type of abstractions, science is unable to employ phenomenological methodology to ascertain the true universals of experience. Thus, “Any scientific description of the world is essentially incomplete in that it inevitably omits major dimensions of our life-world experiences” (Gutting, 1978, p. 44). Husserl “…became sensitive to the fact that these sciences had nothing to say with regard to… questions concerning the sense and meaning of life”, and claimed that science is “a theoretical-logical substruction, the substruction of something fundamentally unable to be perceived, something fundamentally unable to be experienced in its own being-itself” (Bernet, et al., 1999, p. 225).

One might claim, reading his Cartesian Meditations, that Husserl, through the phenomenological process of defining science through intuition, i.e., “‘immersing ourselves’ in the scientific striving… in order to see clearly and distinctly what is really being aimed at” (Husserl, 1995, p. 9), and in speaking of “grounding” and “evidence” (pp. 10-11), has indeed come upon a rather modern notion of science. We are, however, finally confronted with statements supporting the requirement of an “absolute foundation, and absolutely justified” as what “furnishes guidance in all sciences” (p. 11), a conclusion which merely reiterates the points above. In addition, as we shall see in more detail below, Husserl’s conception of a particular science, psychology, was also derived, reasonably enough, from the practice of psychology in his time. That practice involved the determination of “psychophysical facts and norms… without a systematic science of consciousness” (p. 93). Needless to say, psychology has fundamentally altered in the intervening seven or eight decades.

There are two points here that need to be distinguished and analyzed separately. One, having to do with the origins of scientific concepts, maintains that those concepts are derived from and traceable back to concepts in the lifeworld. The second point has to do with the subject-matter of science, and we will look at this below. To start with, even theoretical concepts, although they might not in themselves be directly traceable to life-world experience, must be potentially, at least, shown to be derivative.

 

Husserl’s claim requires only that whatever theoretical entities we do retroductively infer or posit must have as their evidential basis the self-evident showing-itself of something in originary intuition in life-world experience…. Thus, Husserl’s characterization of scientific method can accommodate a realist view regarding the observational and existential status of retroductively inferred theoretical entities. (Belousek, 1998, pp. 83-85).

 

“Retroduction” here refers to the process of inferential reasoning first termed such by Peirce (Peirce, 1998a, pp. 45-56), and later elaborated by others (e.g., Hanson, 1965; Forstater, 1997). It may be said to characterize a type of inference by analogy as follows [33]. We can note the various lengths of the shadows, in sunlight, of an upright stick. Through a combination of induction and deduction, we infer that the shadows will change in length and angle by such-and-such over such-and-such time, but we retroduct that therefore light travels in straight lines. This latter conclusion is neither an induction nor a deduction, but is the result of a process of reasoning by analogy, a conceptual leap. Indeed, Belousek makes the point that there is a creatively inferential component to retroduction, i.e., that while what it posits is “indicated” by the phenomenon, that this “cannot be guaranteed by the self-evidence of the phenomena” (Belousek, 1998, p. 85). That is, in cases where retroduction creatively “extends meaning-reference”, a radically new meaning is created, one which does not refer in any way to previous entities, but which refers to newly posited entities. There is then a kind of feedback from newly created theoretical entities, in this view, to the perceptions on which they are based, which radically changes the meanings of those perceptions. When that occurs, those meanings cannot be traced back to “originary” life-world concepts. Thus, Belousek presents the example of a track in the cloud-chamber (p. 84), a curved line which does not merely indicate a particle, but becomes, conceptually and perceptually, that particle’s track. This example illustrates how the meaning of that track refers, not to previous entities, but to the new entity, the particle whose track it is. Creative meaning alteration, then, bootstraps itself away, so to speak, from originary life-world meanings, supported not by those but by other theoretical constructs. According to this argument, it is the case that there are concepts both in science and mathematics which, although derived from previous concepts in those areas, are not directly traceable from them by logic or conventional inference. That is, Belousek is claiming that science is indeed traceable to the life-world, but not necessarily through any clear inferential process, and that Husserl is mistaken in his conclusion that this is not the case.

But this kind of argument can work either way. Let us suppose that Belousek’s argument is incorrect, and that in the case of creative retroduction one cannot find means, or paths, through inferences, to tie science to the intuitions of the life-world[34]. What we might ask then, is how “abstract”, how difficult, scientific bracketing actually is. That is, if bracketing can be shown to be easily attainable, and regularly practiced in normal human life, then one could claim that science is not limited by its bracketing of certain life-world parameters, but is indeed made more flexible by the possibility of multiple types of bracketings, and this flexibility would merely echo that found in the normal life-world. Science might proceed in some cases by bracketing one type or group of life-word conceptions, and in other cases by bracketing not that former set, but another type, and in yet other cases by not bracketing at all. If the act of bracketing could be shown to be this simple and flexible, then science could potentially become more flexible than phenomenology, which, according to Husserl (above) should explicitly retain all life-world connections. And we may indeed inquire whether, despite the fact that the act of bracketing was originally designed to free phenomenological insights from Cartesian doubt, that freedom is so radical, except to philosophers? Even if, through bracketing, one accomplishes that task, I see no reason to claim that suspending belief in some object’s existence, or even in the existence of the world, cannot be traced, through unbroken, straightforward chains of inferences, to the normal life-world, and found to be a simple extension of normal human thinking. To assert, on the one hand, that the truly bizarre concepts and insights of, say, quantum theory, are traceable by such means to more foundational scientific concepts, and yet that an easily-expressed suspension of belief is not, is to make a statement which seems mere rhetoric.

I can be quite specific in my claim. I will infer a possible path from common experience to bracketing, as follows. First, one starts with mistaken perception: one sees a shape in shadow or a distant person, and mistakenly “recognizes” an animal or a friend. One may then realize (as a result, say, of moving toward them) that this is a mistake, and see the shadow or the stranger as they are in actuality. These are common experiences; we have all had them. From these, one realizes that what one sees a) can be a mistake, and b) may not be real. One then generalizes from these and similar events, through classical induction, first, to more common perceptions, then to reality as a whole, e.g., that our world is (or may be), i.e., has some “evidential basis” for being, “unreal”, “illusion”, and so forth. One may then conclude that one should at least suspend belief in its reality, analogously[35] to suspending belief in some uncertain specific, e.g., the animal in the shadow, above; the latter is also a common mental act. Thus, in a few clear steps, one can infer an act of bracketing very close to, if not identical with, Husserlian bracketing. If the act as described above is not precisely Husserlian, or not sufficiently so for some critics, it should be modifiable to be so, since it is at least within the same class of acts. Other possible inferential pathways might start from drug experiences, insanity, or religious experiences, all common throughout history and all cross-cultural. The onus is now on the Husserlian to either show that these possibilities are untenable as starting-points to derive the epoché or to show that science does not, indeed cannot, employ this and similar simple acts of bracketing. That is, if bracketing is indeed so simply derived from situations in the normal human life-world, then we must require Husserlian phenomenologists to show one of two things. They must clarify their claims that somehow, in the case of science, bracketing does somehow remove (“suspend”) one from their connections to that life-world. Alternatively, in the face of the above demonstration, they must demonstrate how it is that the act of bracketing is somehow easy in the case of science, and does nonetheless create the suspension they claim, yet is not easy - but does not create this suspension - in the case of phenomenology.

In either case, then, it seems that science is not separated from the life-world by any kind of fundamental gap.

 

We thus come to the second point. The subject-matter of science, we may now understand, need not be limited to “theoretical-logical substructions”, nor need it neglect “questions concerning the sense and meaning of life”. In order to approach the question of the subject-matter of science gradually, I will first take the position, from Kitcher, that science reflects a search for truth, where truth is understood in the context of a naturalized correspondence theory. Kitcher, in setting forth the basis of such a theory, states,

 

We explain and predict the differential successes of our fellows in coping with the world by supposing that there are relations between the elements of their representations and independent objects… it allows for the contents of our perceptual beliefs to be partly determined by our prior cognitive state, and it enables us to understand our seeming ability to achieve greater cognitive harmony in terms of increased match between our representations and an independent reality. (Kitcher, 1993, pp. 130-132).

 

More pertinent to the point of evaluating the axiomatic structure of science and scientific reasoning, and more generally, that of rationality itself, we might consider Kitcher’s statements:

 

I conceive of rationality as a means-end notion. Concepts of rationality are generated by thinking of entities (people…science…)… as meeting some criterion of good design (maximization of expectation…)… relative to a set of goals. (p. 179).

 

Frequently, rationality is taken to be constituted by a set of rules whose status is independent of their tendency to promote any ends. Explanations by appeal to [this latter conception of] rationality require showing only that a particular set of statements conforms to the rules.… This flawed apsychologistic conception of epistemology invites… critiques…. [among which is that] the divorcing of rationality from any tendency to promote epistemic ends fosters relativism. (footnote 3, p. 179).

 

Given this viewpoint, it would indeed seem that science is not oriented toward the creation of ideal mathematical abstractions, as Husserl believed. On the contrary, as Gutting states,

 

The emphasis of modern science is quite the reverse [of Husserl’s claims]. Idealizations are of interest to the scientist only insofar as they provide a convenient way of approaching the complexities of empirical reality…. It is the ideal models that are regarded as imperfect approximations to the concrete phenomena. (Gutting, 1978, p. 44).

 

Whether or not it is true, as Gutting maintains (p. 45), that science operates with the principle that the world is ultimately structured mathematically[36], it remains the case that for science, the idealizations actually employed are approximations to the reality, mathematical or not, that they attempt to describe. For Husserl, in contrast, those idealizations are the ultimate aim of science (p. 46). Gutting also (in addition to Belousek) criticizes Husserl’s claims that the life-world is the “only possible basis for immediate experience of the world” (p. 53), and thus the only originating point for the conceptual structure underlying science. “The realist must hold that there is no single privileged conceptual framework in terms of which immediate experience must be given” (p. 53). The gist of his argument is that, contrary to Husserl’s understanding of science, the scientist does not attempt to reproduce but to explain the world. Thus, subjective experiences like colors, smells, and so forth are not neglected by science, but rather approached as areas that need analysis and explanation. And these explanations may or may not employ subjective accounts[37]. This does not indicate, however, that science neglects the life-world; quite to the contrary, science is occupied (to a great part) in explaining it. But those explanations must, in some cases, go beyond the naēve contents of that life-world in order to accomplish that aim. Further, the lifeworld of the scientist, particularly of the psychologist, is today neither singular, as Husserl’s conception would lead us to believe (and see Kitcher, p. 89, below), nor is it restricted to abstractions at some remove from experience, devoid of considerations of meaning.

These arguments further support my contention that axiomatic structures are not necessarily employed in scientific praxis. In fact, one might argue that no formal structures whatsoever are necessary for practicing science. Kitcher conceives of science, in general, in terms of “goodness of design for achieving goals” (Kitcher, 1993, p. 179), whether those goals are epistemic or practical. “Rational” in this context refers to the above functional concept, not to a formal (axiomatic) concept, and relates the achievement of one’s goals to the means which are more efficient, faster, and/or more fully realized than by other means. It may be that in many cases axiomatization is indeed the most efficient means to demonstrate relationships or even to reach conclusions, but that is not necessarily the case. It is possible that induction, for example, particularly “eliminative induction” (e.g., pp. 233-255), or what Peirce terms “abduction” – a type of hypothesis construction - (Peirce, 1998b, p. 95, and note also my comments on retroduction above), is most efficient in some contexts [38]. In addition to the classic processes of induction and deduction, the acceptance of authority is another means of generating support for hypotheses, and this also functions as a rational, efficient, and largely accurate means of reaching conclusions in science (e.g., Kitcher, 1993, pp. 221-222). Thus, in both reaching and supporting goals in science, a wide variety of cognitive and social processes are employed.

One might object that science is not these various practices, but is the set of conclusions, organized and structured as an axiomatic system, that results from them. However, first, this is clearly at best incomplete; science is more than its conclusions, even given that we concede that science has reached conclusions[39]. Science must include, at the very least, support for those conclusions, or forfeit any empirical claims, and that support must incorporate methodology. Second, assuming for the sake of argument that the above objection has merit, where, actually, do we find such scientific content, formally organized, outside of restricted and usually artificial demonstrations in textbooks? It is easy to take examples from physics, supposedly the paradigm for axiomatic science, and note that Einstein’s thought experiments, for example, initiating the theory of relativity, had to do with illustrations employing lamps, trains and elevators, hardly well-defined axioms. Alternatively, we may note that despite mathematical notation, the manipulation of concepts in physics does not always proceed, as noted above, through deductions and inductions from axioms. One might argue that ultimately, a physical theory, as the ideal case, is formulated as such a system. But what does that actually amount to?

If we calculate the orbits of the electron in the hydrogen atom, for example, employing Schroedinger’s equation, which can be performed fairly exactly in this two-body system, we obtain, through reasonably straightforward mathematical operations, an equation solving the system. But what does this equation mean? The debate on the meaning of the wave-function in quantum mechanics has not ended to this day. As far as the deductive[40] elaboration of a set of axioms, we have established the energy levels and general configuration of a system which, in physical terms, has become more puzzling through that elaboration than before it[41]. Yet in the functional terms set out by Kitcher, we have accomplished something quite meaningful through this description of energy levels, in that we can now expect the absorption of particular frequencies of light by that system. But that is a prediction which must subsequently be verified through experiment. The axiomatic result, analogous to proving a theorem in geometry, Husserl’s ideal, has generated an elaborate formalization which, rather than reaching a definite physical conclusion - the proof of a theorem in physics - raises more questions than it started with, both in terms of understanding and of elaborating that particular formalization. In terms of the progress of science, however, it is ideally suited for the advancement of physics, qua empirical science.

We might take another example, more closely resembling the axiomatic ideal, and examine the mathematics describing asteroids or billiard balls. Those mathematics do indeed describe these systems, if taken in isolation and for special cases[42], and we may well concede that some aspects of those cases have in a sense approached Husserl’s ideal. But even so, whether physicists will allow that we have described the “real world”, actuality, all will admit that the value of such descriptions lies at least as much in the utility of that description’s prediction, say, of whether and when such-and-such an asteroid will be visible from or will hit the earth, than in their constructing a picture which corresponds to actuality, per se. To put it another way, in order to keep physics, and science, advancing toward the goal of revealing truth, descriptions which generate questions are considerably more useful than those which, while only creating descriptions, do not lead to insights which generate further inquiry. That kind of goal, however, returns us to Gutting’s or even more clearly to Kitcher’s functional explanation of science, not to Husserl’s.

I will not provide more examples along these lines; Kitcher has done so much more competently than I. While it may amount to beating a dead horse, one more set of quotes serves to summarize much of the above:

 

Legend [i.e., the traditional conception of science] conceives the growth of knowledge in terms of theories…. I have tried to build up a different type of framework… born of conviction that even the usual refined substitutes for Legend’s blanket notion of theory are at a great remove from scientific work…. Philosophers have usually treated the community as if it were a single knower whose initial state is something like consensus practice. (Kitcher, 1993, p. 89).

 

To conceive of Wissenschaft as possessing, or indeed even capable of such monolithic structure, given that the above critiques are at all reasonable, is simply incorrect. Kitcher’s position, as I have attempted to indicate, in its emphasis on flexibility, in its valuing of function over form and a plurality of approaches, illustrates the conflux of methods science employs for converging on truth. There is, therefore, no bar at all in this conception to the employment of the results of mental acts as the content of a scientific investigation, nor is there restriction of that investigation, or indeed its results, to any formal structure.

Up to this point, then, I have argued that first, Husserl’s conception that science is separate from the “life-world” is incorrect. Either science is in fact dependent on and traceable to life-word viewpoints and conceptions, or, equivalently, that from within the life-world, multiple types of bracketings are possible, some of which are equivalent to those Husserl considered “scientific”. Second, science is not and should not be an axiomatic system, and Husserl’s conceptions of both the ideal philosophy and the ideal science are based on erroneous ideas of both science and of axiomatic[43] systems. Third, science is much more flexible in its methodology and content than Husserl understood it to be. While it may include, as minor aspects, some axiomatizations, by-and-large its purview is much greater, and includes a wide variety of types of thinking, formalizations, and content treated through systematic empirical means (see also below).

 

Husserl’s next point, that science is directly based on a pretheoretical worldview, is questionable on several grounds. First, the concept of a “pretheoretical” worldview or intuition, in the light of decades of study in cognition, ranging from Piaget (e.g., Piaget, 1971, and see below) through contemporary studies (e.g., Johnson-Laird, 1994; and Gutting, above), may not even be a coherent one. These studies point to strong evidence that from childhood on, various types of inference, modeling, and other activities commonly considered building-blocks of theories are ubiquitous in thinking and perception. Thus, the term “pretheoretic”, in the light of those studies, seems a conceptual mistake, since there literally seems to be no time in human development, after the beginning of the function of the central nervous system, in which we do not produce and test theories to some extent.

To employ the term to distinguish between “scientific” and “non-scientific” theories is also erroneous. Even an oversimplified example, like repairing an auto, illustrates this. Given a strange sound, one opens the hood, looks around, listens, applies one’s experience of similar sounds, and asks other mechanics what they think. An older, experienced mechanic may have had a lot of bad carburetors lately, and one forms a theory on the basis of that person’s authority, to the effect that the problem is the carburetor. However, after looking, nothing is found wrong there. One then investigates further, and concludes that it may be the valves. After further investigation it is in fact found that they are bad. This is a classical, if simple, example of empirical scientific praxis, which has resulted in a correspondence between one’s conception of what was wrong and what actually was wrong: we have determined the truth. Even in this example, where there was a definite answer drawn from a small set of possibilities, one has had to perform a huge variety, not merely a huge amount, of physical, cognitive and social operations, necessitating a vast knowledge of human relations, tools, autos, and so forth. A mistake, due at least in part to social norms (acceptance of authority), was corrected. Should making a mistake, formulating the wrong theory, mean that one could not make progress, do research, arrive at the truth? Clearly not. Following procedures very painfully worked out for correcting mistakes, very similar to those in any scientific investigation, the truth about the auto’s malfunction was determined. Where then is the boundary between science and everyday theory-formation and problem solving?

And while one may readily concede that scientific investigation, certainly as practiced by individuals, has biases, Kitcher, for example, presents multiple examples and mechanisms illustrating the means by which the explicit and implicit knowledge that scientists bring to investigations, and whatever assumptions any contemporary science is rooted in, are ultimately able to be questioned and investigated through science itself (Kitcher, 1993, pp. 219-302). Thus, it is actually science, rather than pretheoretical intuition, which actively compensates for bias and theoretical predilections (and see below for a critique of eidetic variation). One could claim that this very systematic compensation might serve as a rough definition of the validation practices within science. In summary, there is an enormous literature concerning various aspects of bias (e.g., Lynn, 1986; Evans, 1989; Cohen and Freeman, 1998; Schroyens, Schaeken, et al., 1999), which indicates that science is quite capable of compensating for a variety of worldviews.

Further, science is not limited in the content of its investigations to physical actuality but may also investigate ideas, concepts, and even Husserlian essences (e.g., Giorgi, 1985, and below). Nor, similarly, is it limited in its methods to physical manipulations of objects. On the contrary, as I have indicated above, it may even employ, among other methods, if it seems useful to do so[44], the act of bracketing (but see the critique of that methodology below). At this point in time, it is notably easy to present specific examples of science’s examination and employment of feelings, mental acts, and even “questions concerning the sense and meaning of life”. Thus, Lopez and Guarnaccia, for example, deal explicitly with suicide, culture, and feelings of depression and anxiety. “Selected research on anxiety, schizophrenia, and childhood disorders is examined, with particular attention given to the study of ataque de nervios, social factors affecting the course of schizophrenia, and cross-national differences in internalizing and externalizing problems in children.” (Lopez and Guarnaccia, 2000, p. 571). Similarly, Ajzen describes his article thusly: “This survey of attitude theory and research published between 1996 and 1999 covers the conceptualization of attitude, attitude formation and activation, attitude structure and function, and the attitude-behavior relation.” (Ajzen, 2001, p. 27). These are recent articles concerning suicide, selected more or less randomly from an index of such works. They are empirical, controlled, peer-reviewed, and presumably, methodologically, at least, replicable studies of affect and motivation. They certainly relate to persons’ coping with the “meaning of life”. Need more be said about science’s actual and potential scope, as presently conceived and practiced?

If examples of science’s analysis of pure phenomena are desired, one may cite introspective studies ranging from Ebbinghaus’ studies of memory (Ebbinghaus, 1964), through the Gestalt psychologists and their modern descendents (e.g., Köhler, 1962; Lesher, 1995; Palmer, 1999), to a wide variety of cognitive work. To put it another way, the term “phenomena” refers to anything upon which we can introspect. When Ihde, for example, writes of “visualizing” a cube in order to perform eidetic variation or bracketing on that phenomenon (e.g., Ihde, 1977, pp. 100-101), that visual image and its meaning[45] is one that may also be studied introspectively in cognitive psychology, through fMRI in the laboratory, in social studies, and in many other scientific disciplines. Science has extended itself into the phenomenological arena; rather than divorcing itself from the life-world, it has brought its purview into that world.

Thus Husserl’s picture of scientific investigation, a reflection of the understanding[46] of his times, was fundamentally flawed. If that is the case, and if, in addition, phenomenology is necessarily empirical, as I will argue below, then phenomenology must be considered an aspect of science, rather than the other way around.

 

IV. Phenomenological Methodologies and Why They Are Covertly Empirical

 

I will now move to a more detailed examination of Husserlian phenomenology. As I mentioned above, there are two techniques that, according to Husserl, give phenomenology its uniqueness. The first is the epoché, the second the method of eidetic (or “free”) variation. Later, I will examine these in detail in the light of Gestalt psychology, to lead toward (and past) Gurwitsch’s phenomenological position. Now, however, I will assume, to start, that Husserl may be correct in his evaluation of the necessity and even the general nature of the results of these methods, and I will evaluate some implications of these procedures for phenomenology. But in order to evaluate the Husserlian methodology, one must ask about the applicability of those methods. That is, one can assume that Husserl, employing his methods, arrived at a variety of insights into his own experiential content, at a minimum. But he was only one person. Are his methods applicable to others? Are they communicable? Are the results of multiple persons employing them comparable? These are merely the starting-points for a wide variety of questions concerning the applicability and verifiability of the phenomenological method.

Procedures of verification (I include both reliability and validity under this term) have had a long and contentious history. There is a huge literature on verification, from relatively simple procedures involving duplicating experiments in the physical sciences to more complex procedures in the social sciences. The latter fields are more complex, since, very roughly, individual human variation, experimenter bias, and ambiguous results lead to problems in replication of conditions and in the interpretation of results. These complications are well known, and I do not believe that it is necessary for me to cite much literature here (but see for example Anastasi, 1969, for some specific procedures and tests; Goldman, 1986, for theory).

Once one admits that phenomenological methodology needs verification, one inexorably starts down this road. One must admit the possibility that phenomenological methodology must accommodate itself to the methodologies of the sciences, at least insofar as verification goes, because phenomenology has opened itself to the necessity of empirical data in at least that respect. Yet to deny the necessity of verification is to maintain that, first, phenomenological training, however that is accomplished, is uniformly successful, and second, that different phenomenologists have little difficulty understanding each other’s results, i.e., not merely that each arrives at the same “essence” or “eidos” for some phenomenon that other phenomenologists have, but more importantly, that each knows that the others’ results are identical to theirs, or, if they differ, how they differ. If these conditions do not hold, and no verification procedures are employed, what point is there for anyone but Husserl to do phenomenology? Thus, Seidler asks, “What can be the grounds for settling disagreements if two such subjects disagree on a single ‘essence’?” Seidler, 1977, p. 318).

Even if one went so far as to claim that phenomenological methodology is analogous to a craft which could only be passed from master to pupil through direct contact, lessons, and example, an empiricist would counter that in order for teachers to establish that their students had indeed learned the method, validation procedures would be required, procedures which were, in fact, performed through a variety of empirical means. Educational evaluation and testing is a well-researched field (and see e.g., Giorgi, 1985; Washington and Biro, 2001; and Overgaard, 2001, on the methodology of evaluating verbal phenomenological reports). Another possible objection to this type of criticism might be to cite the work of, say, Scheler, above, as verification through consensus obtained after publication, i.e., that other phenomenologists read his papers and (presumably) agree with him. Yet, first, that is indeed a procedure within the praxis of empirical verification, and second, once even that extent of obtaining consensus is admitted, then one immediately considers following it with such standard empirical procedures as the replication of published results, and so forth.

One may, in short, question the effectiveness of the epoché as a mental act, and explore the implications of that question. The epoché, as an idea, is an ingenious notion, but to actually maintain that one can entirely take consideration of objective existence “out of play” seems to impute control beyond the capacity of the (sane) human mind. To conceive of so doing and to actually perform the act are two very different things. Yet this objection, and its counter, per se, will end as no more than a shouting match between Husserlians and their critics. That type of dispute is what has cast introspection, as a methodology, into such disrepute, particularly after Titchener (see for example Gardner, 1985, pp. 106-109).

However, we might reformulate the objection by examining it more closely. The epoché, on the face of it, is indeed a methodology, i.e., one exercises the mental act of “bracketing” particular properties. However, if one asks what the methodology of actually applying or implementing this bracketing may be, that is, how, precisely, as a mental act, might one implement bracketing, one finds statements such as the following:

 

’The first and basic methodological component in the theory of cognition’ is the ‘skeptical position-taking [Stellungnahme], the absolute epoché which recognizes no pregivenness and sets its non liquet [‘it is not clear’] as an abstention from judgment over against all natural cognition’ (Bernet, et al., 1999, p. 67).

The attempt to doubt anything intended to as something on hand necessarily effects a certain annulment of positing… we, so to speak, “put it out of action” we “exclude it,” we “parenthesize it”…. The “excluding” is brought about in and with a modification of the counter positing, namely the “supposition” of non-being… (Husserl, 1998a, p. 59).

 

For what may be a clearer exposition of the epoché we may turn to Pietersma:

 

The phenomenologist, then, will not allow his everyday convictions and beliefs, no matter how well founded he may think them to be, to influence him in such a way that the man believes nothing at all [concerning the existence of griffins]…. No beliefs the phenomenologist has with regard to the world should come into play…. Convictions of the kind mentioned are set aside… (Pietersma, 1979, pp. 37-38).

 

But how does one do this? One must “set aside,” “put out of action” convictions, “not allow,” “exclude” them from influencing one; how can this possibly be effected? Further, if we assume, for example, that Husserl had done this, how do we know that others have, in similar manner, as effectively, and for the same types of contents and actions?

In other words, despite my derivation of a possible inferential pathway to an act similar to Husserlian bracketing, above, one may still inquire as to where one might find a description of the training necessary for this act (i.e., the phenomenological “abstention from judgment”), and how successful it is in imparting this particular skill, without, so far as I am aware, being satisfied. It is, for example, fairly easy to get someone to accomplish the mental acts termed “visualizing” something; this request is understood and carried out intuitively by most people with varying degrees of success. In addition, there are exercises that one may undertake to improve one’s skills at visualization, and straightforward tests to verify their effectiveness. In contrast, the term “bracketing” or “epoché” purportedly conceals a mental technique, which, since it is claimed not to be intuitive, escapes all but phenomenologists (as Husserl states quite explicitly, e.g., Bernet, et al., 1999, pp. 60-61). Ihde (1977, pp. 42-54) has a fairly extended treatment of a technique involving the redirection of one’s attention, but he cites no studies verifying that this technique actually is effective, nor does he claim that this trains one to “bracket”, rather, it seems to be a technique for introducing one to the “method of variation”, a different class of mental acts. Moustakas (1994) does supply techniques, but despite his avowed anti-science bias (e.g., pp. 46-47), he actually requires and presents methods which, while termed “phenomenological research”, amount to classic empirical studies (e.g., pp. 121-175) which, despite their use of introspective data, are easily subject to a variety of methodological verification procedures. My claim here is supported by the work of Giorgi, 1985, who is overt in asserting that phenomenological research is empirical and (should be) intimately connected to psychology, and by his pupil, Hurlburt, who has systematized and extended Giorgi’s methods and carried out detailed empirical phenomenological analyses (e.g., Hurlburt, 1997; Hurlburt and Heavey, 2001; Hurlburt and Heavey, 2002a; Hurlburt and Heavey, 2002b). How indeed, without verification, can one maintain that the epoché, however helpful it might be if attained, is in fact attainable?

This same point is also brought up by Roy, et al.: “The failure to make reduction into a concrete method… is arguably the most salient weakness of current appeals to Husserlian phenomenology” (Roy, et al., 1999, p. 74). One might add, given the above, that even if reduction were made into a concrete method, how would we know that method was effective, to what degree it was effective, and in what circumstances? As Lind summarizes it,

 

Protocol collation seems to be caught in an epistemological dilemma…. As subjects, nonphenomenologists generally lack the reflective skills or sensitivity to discern phenomenological structures on their own. Therefore… conclusions, though grounded in multiple experiences, would contain little insight into the phenomenological character of the theme under investigation. If, on the other hand, the researcher were to engage in active reflection… his own interpretation of the data would again be subject to the pitfalls of inaccuracy, theory-ladenness, and idiosyncrasy or ethnosyncrasy. (Lind, 1982, pp. 89-90).

 

The phenomenological bracketing and self-awareness, i.e., the movement beyond the “natural attitude”, the putative starting point of phenomenological explorations, to the full phenomenological reduction, then, exposes us to two classes of doubt. First, based on the act itself, one may question whether one has indeed thoroughly uncovered all of one’s hidden assumptions, i.e., has actually rid oneself of the “natural attitude” and successfully accomplished the epoché. Even given that one could find a precise statement of the mental states or attainments of one who has, so that we could have a clear idea of the desired end-state of bracketing, how does one know that they have actually reached it? Which and how many assumptions must be uncovered in order to free oneself of some remnant of a “natural” or indeed a “scientific” attitude? One can, of course, apply the same question to the above-mentioned method of variation. Second, if such studies were in fact to be implemented, the necessity of verifying their efficacy would seem to strike a blow to the heart of Husserl’s position on the primacy of the phenomenological viewpoint. That is, the above question about knowing the effectiveness of the putative training procedures for bracketing need not question the ideal effectiveness of those procedures; the important point is that it questions our knowledge of that effectiveness.

Aside from methodology, what of the results, viz., phenomenological knowledge or content? Dreyfus, for example, is quite explicit in claiming that Husserlian phenomenology is both empirical and non-formal (Dreyfus, 1996, § 45, 46, 49, 50), although his explicit reference to the latter deals with Merleau-Ponty. However, the degree of consensus that would be necessary for phenomenology to claim status as an empirical science even on the level of, say, cognitive psychology (much less on the level of physics), is approached in the psychological and physical sciences only in the case of long-established, experimentally verified facts. Note that I am not saying that phenomenology is not or should not be empirical; but I am denying that it can claim the status of a science, for the following reasons. I am not aware of consensus anywhere in the literature of phenomenological investigation; on the contrary, that literature is notoriously difficult to interpret, not to mention duplicate. “Writers calling themselves phenomenologists disagree on nearly every point” (Lind, 1982, p. 86). One might deny the effectiveness, even theoretically, of verification insofar as phenomenology goes, but, again, where does that leave the intersubjective interpretation of phenomenological contents, the results of the methodology?

More generally, given the existence of other minds, the status of our insight into both our explanations of and our understanding of other persons, i.e., the nature of our knowledge of the “match between our representations” of them and their “independent reality” (Kitcher, 1993, pp. 130-132), follows from the nature of the above conception of truth. The outline of this aspect of my argument is, in effect, a variant of the argument involving intersubjective doubt. Ascertaining characteristics of other persons, whether from the standpoint of verifying their comprehension of a methodology or from a more general desire to investigate their phenomenal consciousness, entails the acceptance of the possibility of error in such judgments, a “differential success” in knowing others. But it also entails the acknowledgement that we can learn, even with error, about others. How, then, do we carry out our investigations, and how do we verify their results? Can such an enterprise ultimately be productive, i.e., can we “advance” our knowledge of others? When we do find Husserl’s answer to this question, it is in fact an empirical one: the existence and nature of other minds is understood through “similarity” and “associative coinciding par distance” (Bernet, et al., 1999, pp. 159-160) with our own, and verified by observing other’s behavior:

 

The other psychic determinations are proven or confirmed by the fact that they stand together with the originally perceived corporeality in a nexus of continuous, reciprocal motivation… the confirmation of the other as a being with his own immanent experience has in toto the character of a concordance of interpretations which are joined together. (pp. 162-163).

 

The “concordance of interpretations” above is merely the “unity” of a set of “experiences”[47] of another. I can see no difference between the above and a straightforward empiricism; not only does conventional empiricism hypothesize that our notions of other minds derive from perceived (consciously or not) similarities between our bodies and actions and those of others, but recent work in neuroscience on so-called “mirror neurons” (e.g., Rizzolatti, et al., 1996) provides the beginnings of our understanding of its neural instantiation.

Further, Marbach, in an article synthesizing Piaget’s and Husserl’s approaches to phenomenology, states that in order for one to evaluate the “world-constitution” of infants and of animals, phenomenology must “rely on empirical evidence which is as firmly established as possible” (Marbach, 1982, p. 467). Thus, “there is a modified empathy concerning levels of development” (p. 459) with respect to children. But neither Marbach nor Husserl can claim that there is a sharp boundary between children and adults. However, if there is in fact a continuity between adults and children, where do we draw the line as far as the necessity for empirical evidence goes? Although we may be at the same level of “development”[48] as another adult, we may not be at the same level of education, of conceptual ability, and so forth. Do not those differences, then, also imply the necessity of some sort of empirical aids to fathom individuals’ conscious experiences? Indeed, the history of psychology strongly supports this. But then we are back to the necessity of empirical data to bridge the intersubjective gap.

Yet to seriously consider these and other processes as empirical places Husserlian phenomenology in a dilemma, since Husserl vehemently denied the relevance of empirical methodologies to the phenomenological process, as the latter was going to serve as the source of the apodictic basis of an axiomatic philosophy. Thus, exploring either phenomenological methodology or phenomenological knowledge leads to a form of skepticism which might be termed “intersubjective”, or to processes of verification which necessarily partake of the empirical studies Husserl is attempting to transcend.

 

V. But What of Phenomenological Data? Why That Data Is Not Apodictic

 

One might still argue, however, that the phenomenological methods, viz., the epoché and eidetic variation[49], create an intrinsic difference between the subject matter of science and that of phenomenology. With these methods, one might claim, one obtains non-empirical knowledge. One pays attention to the phenomena only, without regard for their actuality; one disregards, in a sense, the end of the intentional arrow. If that is the case, then despite the above critique, one might argue that phenomenology merely needs to encompass one particular type of scientific methodology, viz., that related to education, i.e., how to teach and evaluate the learning of phenomenological methods – analogous, perhaps, to teaching geometry – and leave it at that. Phenomenology’s peculiar enterprise then remains largely untouched, with the proviso that in some cases its results must be further investigated in order to verify them. That is, while the phenomenologist might admit that even though knowledge – if we might term it thus - acquired through the epoché and eidetic variation must be verified by other phenomenologists, that knowledge, since it is apodictic, cannot be obtained in the first place by empirical methods, and thus by science, but must be reached, initially at least, through methods unique to phenomenology.

This claim amounts to maintaining that phenomenological insights, in contrast to empirical knowledge, may be known to be apodictic. Thus, given that the arguments above are correct, the apodicticity of phenomenological knowledge is the last bastion of its claim to uniqueness. Yet we shall see that this claim too may be severely questioned. Earlier, I defined the term “apodictic” and presented a short and superficial argument indicating that there might be problems with its suitability to mathematics. Now, however, I will look in great detail at apodicticity, especially as understood by Husserl, and will end by maintaining that there is no sense in which experienced phenomena are apodictic. There are two main lines of criticisms of apodicticity. The first has to do with skeptical responses to claims of certainty, the second with the nature of apodicticity as a type of judgment.

One class of skeptical responses to apodictic claims made for phenomenal experiences are perhaps best summed up by Dennett, in his article “Quining Qualia” (Dennett, 1994). This class of responses is taken from a position derived from Wittgenstein’s private language argument (e.g., see Wittgenstein, 1988, §566-572). Dennett presents what he terms “intuition pumps”, i.e., brief arguments designed to call into question the position that our experiences, understood as classical qualia, are knowable in any certain terms. If that is the case, then the existence of qualia, in any coherent sense, is doubtful. As Dennett describes them, qualia are phenomenal experiences that have all of the following properties: they are 1) “ineffable”, 2) “intrinsic”, 3) “private”, and 4) “directly or immediately apprehensible in consciousness” (Dennett, 1994, p. 47). “Ineffable” is taken to mean that “one cannot say to another”, i.e., cannot communicate in any manner, in “what way” one is currently experiencing. “Intrinsic” is not explained except to say that it implies that qualia are “atomic and unanalyzable”; I assume it has something to do with the next property, that of being “private”, which Dennett says implies that “all interpersonal comparisons of these ways of appearing are (apparently) systematically impossible” (p. 47). The last property is, at this point, reasonably clear, I hope[50].

The gist of Dennett’s argument has to do with memory. If there is no objective (i.e., public) record of the past, then there is no way to verify whether a quale we are now attributing to some object (e.g., the taste of a brand of coffee) is the same as it was an hour ago, except through our (fallible) memories. If we now believe that we do not enjoy some brand of coffee which we did enjoy an hour ago, we do not know whether the taste has changed[51] or our evaluation of the taste has changed[52]. This is certainly not a refutable skeptical stance, if one takes it seriously, and it may indeed follow from this that the notion of qualia is incoherent, and thus that their existence is doubtful. However, taken seriously, this position also leads to questioning the public, i.e., consensual, basis of records. That is, a public record should be no more reliable than our memory of qualia, since the former depends on the testimony of others, which is as uncertain as our own, since it depends on their memory, or alternatively, depends on assuming that written records, for example, have not altered, been spontaneously created, are also remembered accurately, and so forth. The “objective” world, in other words, is just as vulnerable to this type of skepticism as the subjective, as Descartes realized long ago. If Dennett’s argument is taken to the limit, then, he is hoist by his own petard.

In addition, in denying the reality of qualia, even in the easily-doubted strong sense above (which I do not espouse either), Dennett lays himself open to another objection. Siewert (1998) points out that it is not permissible to move directly from the position that one cannot know some property to the claim that therefore it (i.e., a quale) does not exist. As Siewert puts it[53], Dennett’s conclusion is that “where these claims are concerned, the rule ‘possible, only if warrantable’ holds” (p. 167). And this conclusion is not logically justified.

That aside, if we do not take Dennett’s claims about the implications of his property attributions of qualia quite so literally, we arrive at the position for which I have argued above, viz., that the determination of phenomenal properties is an empirical matter, and one which science is already bent towards clarifying. Whether one feels that one must, given this latter position, follow Dennett’s lead in claiming that phenomenal experience is then a type of reporting of internal states[54] seems irrelevant, initially, at least, to the general issue of their empirical determination. In addition, because of the ambiguity and problems with the meaning of terms like “reporting” (“deliverance”) and “states” (i.e., “property”) here, I will only touch on the issue of the actual model of mind that may be implied by this position, and not until later in this essay. That is, Dennett is of course arguing the above with the goal of subsuming the subjective under a model driven by an understanding of the mind as a type of computing device. I do not believe, however, that it is necessary to make this assumption[55] in order to claim that a) qualia are not necessarily immediately and transparently accessible to consciousness, and b) that nonetheless it is possible, employing a variety of methods, some, at least, derived from the various sciences, to ascertain reasonably clearly what one has experienced an hour ago, and what one is experiencing now[56].

Levin is a critic of Husserlian apodicticity who would, I believe, support my claim (a), above, if that claim is taken in a very particular sense. That is, Levin approaches this issue from a phenomenological viewpoint (Levin, 1970), and the sense of “qualia” applicable to his critique would correspond to a Husserlian essence. The early Husserl takes “transcendent” to refer to objects in the world, i.e., objects with both spatial and temporal properties, and “immanent” to refer to phenomenal (subjective) objects, those without spatial properties (Levin, 1970, p. 15). Later (e.g., pp. 18-20), Husserl extends “transcendent” to refer to any essences whatever, and “immanent” to refer to non-essential subjective phenomena, the momentary flux of consciousness, the “living, streaming present” (p. 60). Now, as far as objective entities go, we are uncertain about them because they always entail an infinite amount and type of perspectives and implications that we cannot take into account from any particular viewpoint or understanding, and after any finite amount of time.

 

The ‘real’ object is necessarily given perspectivally… [it] brings together and orders a harmonious, continuing series of experienced ‘adumbrations’… the ‘real’ object is an ideal unity because, as object, it bears the sense of being more and other than the acts of consciousness that relate to it; and because it is a synthesis… with certain other predelineated and horizonally anticipated, but as yet unfulfilled intentional acts… and since these motivations… yet extend into the infinity of the objects’ history, it is necessary to affirm the partial indeterminacy which defines such ‘real’ objects. Transcendent perception (perception of things in the world) is always imperfect, incomplete, inadequate (pp. 52-53; also, e.g., p. 92).

 

There is, then, an inherent indeterminacy and doubt concerning those kinds of transcendent objects [57]. Now, in Husserl’s later writings, the categories of transcendence and immanence get extended to apply also to “subjective” entities (e.g., p. 154). That is, a transcendent subjective entity is an essence arrived at through the process of eidetic variation[58] where this could be applied to either “objective” or “subjective” entities, while an immanent subjective entity is one of the singular phenomena involved in the constitution of this essence. Levin maintains that these characteristics also pertain to immanent objects, for two reasons. One has to do with the infinite potential for variations inherent in the constitution of any objects whatsoever constituted or synthesized through the method of free variation (or for that matter any inferential method), the other with the nature of the essence or eidos arrived at through those processes.

 

And quite in keeping with their transcendent nature, essences, though ideal (in the special sense of being contrasted with spatio-temporal ‘reals’…) declare themselves as enduring objective concerns through ‘horizons’ of intentional meaning (p. 155).

 

It is just those same kinds of horizons which, as in the case of “spatio-temporal” objects, relegate the status of essences of any sort to indeterminate (see also pp. 172-177). Levin argues that because the process of eidetic variation within subjectivity, i.e., among immanent essences, is subject to the same infinite depth as that involving objects, the same uncertainty must apply (“seemingly overlooking the fundamental sense of transcendence we have noted, [Husserl] divides essences into immanent and transcendent, and maintains the possibility of… apodictic knowledge of the former kind” [p. 155]). That is, Levin is questioning Husserl’s claim that while transcendent essences cannot be fully known, immanent essences can be. The universal possibility of eidetic variation, according to him, should introduce the same indeterminacy in immanent essences as spatio-temporality, according to Husserl, does in transcendent essences.

Thus, we cannot know when the process of eidetic variation should end, and whether the next variation will alter radically the nature of the essence being constituted. For if we did know that, then we would be somehow know that further variation “would result merely in an amplification, a confirmatory filling out of the essential structure already discerned” (p. 181). And what could possibly justify that knowledge? “It would not seem possible to adjudicate in advance, in a wholly a priori and noncontingent manner” (p. 181) when some modification could alter the essence being constituted and create a novel one. As Levin argues, if apodicticity could be ascertained from inadequate knowledge, we would have to assume that further knowledge is “somehow non-essential and… known to be thus” (p. 181). Thus, as far as essences go, whether we term them either “immanent” or “transcendent”, we must take an attitude toward them similar, in this respect, to the attitude taken toward objective transcendences, i.e., that they are always dubious to some extent (“the grounds on which Husserl denies apodicticity to perception of material objects… should obtain… for the immanent perceptions” [p. 135]).

Further, in order to know that an essence is apodictic, we must explicitly (doxically) make that judgment: “apodicticity is the result of a reflective operation (critique) performed upon an evidence already compelling” (p. 84). That act of judgment implies that another process must be applied to the (presumed) apodictic essence to ascertain that it is in fact apodictic . Thus, to actually make the judgment that an essence is apodictic, according to Levin, we are involving ourselves in an act which implies the same kind of uncertainty as the initial acts of free variation and constitution of the essence. “The judgment that an evidence is apodictic logically presupposes an eidetic variation performed upon some originally given evidence… ‘apodicticity’ could only designate an evidence constituted in a very special way through reflection” (p. 97). But if the initial constitution of any essence, transcendent or immanent, is itself intrinsically incomplete and thus dubitable, the further constitution of the judgment of the apodicticity of that essence then only adds to the dubiousness of that judgment.

Levin then brings up the issue of the intersubjective validation of apodicticity (e.g., see pp. 212-213). It seems to have been taken for granted by Husserl that the judgment of the apodicticity of an essence could easily be duplicated or repeated by any skilled phenomenologist (but see my comments above). Levin also seems to assume, despite contradictory evidence, that apodicticity (and that judgment) is, or should be, intersubjectively uniform. That is, Levin’s examples, exploring the possibility of deviant apodictic judgments, take the non-universality of apodicticity as an abnormality, even within a community of similar individuals. I have touched on this point above, and it seems that abnormality should be inferred from an observed uniformity of judgment, rather than the opposite. One might put it that the spatio-temporal and “horizonal” uncertainty of the constitution of the individual essence is spread intersubjectively, when we ask a community of phenomenologists to agree. We must then inquire, as I have mentioned, as to what, beyond the individual intuiting of the nature of the essences, must occur to confirm or disconfirm these judgments.

Finally, according to Levin, the only possible candidates for any kind of apodicticity, given the fact that judgments of any sort are reflective conclusions, whether they concern subjective or objective transcendences, are the immanent phenomena, the “living, streaming momentary present” (p. 96). These, according to Levin, while they may be considered “evidentially adequate, absolute, and indubitable” (p. 96), may not be considered apodictic in Husserl’s doxic sense. That is, they are always, as a stream of varying phenomena, the ground underlying the subjects of judgments and can never be, because of their temporal flux, either the subjects for, nor the objects of judgments. Only objects constituted from this flux can fulfill either of those roles.

However, I do not believe that Levin is correct in taking this initial, most basic, stream or flux of phenomena as any sort of “indubitable” ground whatsoever, because here, if anywhere, it would seem that some version, at least, of Dennett’s above criticisms would be applicable. Given that they are not able to be doxically judged as apodictic, Dennett’s point that our moment-to-moment memories of them are unreliable seems valid.

 

VI. The Big Problem: Phenomenological Methodology Is Inadequate to Investigate Its Own Assumptions

 

1. Introduction and Outline of the Argument

 

If these above arguments are correct, there is nothing which can be known to be apodictic, and one of the major thrusts of Husserl’s great project has failed. Why then continue with phenomenology, or why not merely conceive of it as a branch of psychology? As to the former, I believe that phenomenology has a great deal to contribute, both methodologically and in content, towards the analysis of subjectivity, and I will present many arguments and illustrations to that effect in this essay. As to the latter, since I do argue for the naturalization of phenomenology, I am indeed arguing that it must be considered a branch of psychology, but one that should extend that latter area into the explicit embracing of introspective studies.

But where do we proceed from here? The next step is to show specifically where empirical studies conjoin with phenomenology. That is, it is clearly insufficient merely to claim generally that phenomenology should be an aspect of the empirical study of the mind. What I would like to show, then, is where, precisely, it is necessary to introduce empirical data and methodologies into phenomenology, in order to resolve questions that arise from phenomenology itself. I will argue that Husserlian phenomenology’s own methodology gives rise to problems that would in fact render it ineffective insofar as Husserl’s original intent was concerned, but that those problems, while not resolvable as Husserl would have wished, can be answered in such a way that phenomenology’s empirically-oriented aspects are largely unaffected. More specifically, I will argue that certain implications of phenomenological methodology (viz., the epoché and the method of variation) lead to problems insofar as the unity of phenomenal experiences are concerned, and that the resolution of those problems lies in the concept of the gestalt, which is however an empirically-derived and testable model. If phenomenological unity is conceived in terms of gestalt unity (as Gurwitsch also claimed), then not only are phenomenology’s ontological claims irrelevant, but phenomenology must concede a necessary connection to empirical psychology through the agency of the gestalt.

 

The investigation of the assumptions held by Husserl, and thus what I will term “classical” phenomenology[59], i.e., the philosophical school still holding to the correctness of Husserl’s methodological claims, raises both methodological and logical problems. One of these problems involves empiricism, in the following sense. When Husserl’s assumptions are investigated empirically, we will find that some of them are simply incorrect. Yet that investigation, as one depending on the results of conventional empirical studies, could not have been conducted phenomenologically. Gurwitsch, who explored investigations of this type (as well as did, for example, Merleau-Ponty and Piaget), draws heavily on the experimental findings of the Gestalt psychologists, and I will cite evidence concerning the incorrectness of Husserl’s assumptions from their modern descendents, and others. According to the principles of phenomenology, however, this type of investigation is at least questionable, if not invalid, particularly when applied to phenomenology, which is supposedly a more fundamental type of inquiry. Yet those empirical investigations show quite clearly that some very basic and important assumptions made in phenomenology are, as I have said, incorrect. This must lead us to conclude that empirical investigations are indeed relevant to phenomenology. But that leads to considering phenomenology and empiricism as at least being on an equal footing.

The second issue is one discovered as a logical consequence of the above investigations. Once we have determined that the assumption of the constancy of phenomenal components, the “constancy hypothesis”, which I will explicate in detail below, is without basis, then phenomenology is faced with a severe dilemma, in that its methodologies, touted as the fundamental differences between, and bases for the advantages of, phenomenology over empiricism, seem to imply a fundamental inadequacy in phenomenology.

My argument below will consist of the following steps:

1) In order to utilize the phenomenological methods of the epoché and/or the method of free variation, one must assume that there are “core” or “essential” elements to virtually any experienced phenomenon (or, as I will explain below, to sets of essences) which are unaltered by varied and profound alterations in their surrounding contents.

2) But if this atomistic (Husserlian) picture of phenomena is true, then Gurwitsch and others must be wrong in their assertions (below) regarding the universality and the essential interconnectedness of gestalts.

3) But modern experimental evidence and theory both back up those latter conclusions.

4) If that is the case, then there must be at least some incidences where the alterations[60] implicit in the above phenomenological methodologies do in fact alter virtually all components, including the putative essences, of certain phenomena.

5) But then an investigation a) of the existence of gestalt properties of phenomena in general (i.e., whether phenomenal components are or are not atomistic), and/or b) of any essential components of any phenomenon, cannot be carried out through either the epoché or the eidetic reduction, since both of those assume the atomism refuted by empirical data.

6) Therefore classical phenomenological methods cannot ascertain that there are “essences” without circularity.

7) Further, if the above argument is valid, Husserl’s (and Gurwitsch’s) claims that phenomenology, through the discovery of the essences of phenomena, can put philosophy on an apodictic and “scientific” basis, were incorrect.

8) And more generally, if the methodology based on Husserl’s atomism is not merely practically difficult, but theoretically incorrect, then the metaphysics based on that methodology is cast into serious doubt. In addition, we may need to rethink essentialism in other contexts, i.e., in the applications of mathematics, logic, and linguistics; and the fact that empirical studies enabled this paradox to be uncovered is strong indication that phenomenology and empiricism should be considered reciprocal, freeing phenomenology from the above methodological problem.

Since it is consideration of the nature of phenomenal wholes, both through Gestalt theory and phenomenology, which will bring us to this conclusion, it is the nature of the gestalt which will lead us toward a structural, empirical phenomenology.

 

2. Background: The Constancy Hypothesis and the Origin of Gestalt Psychology

 

The “constancy hypothesis” is fundamental to Gurwitsch’s conception of these issues, and indeed to his conception of Husserlian phenomenology. Gurwitsch held, and I agree, that this hypothesis is untenable in the light of extensive experimental evidence[61]. Gurwitsch also had theoretical reasons for rejecting the constancy hypothesis, which I will detail below. The constancy hypothesis, according to Gurwitsch, states that “if the same neural element… is repeatedly stimulated in the same manner, the same sensation will arise each time” (Gurwitsch, 1966, p. 5). More explicitly,

 

Sense-data are not modified nor are they qualified by the sensory facts of a higher order [see above and below] which they found and support…. In the theory of production, the constancy-hypothesis seems somehow concealed…. The constancy-hypothesis also is implied in Piaget’s theory of the schemata as arising from the assimilating and accommodating activity (Gurwitsch, 1964, pp. 90-91).

 

Further,

 

Whenever the immediate data seem to conflict with this hypothesis, reference must be made to the effects which are produced by the same stimulus if it comes in to play isolatedly… called normal…. Anomalies are explained by reference to the intervention of facts usually conceived of as belonging to a higher level – such as judgment. These anomalies originate, not in the elementary sensory data themselves, but rather in the interpretation which these data are given (Gurwitsch, 1966, p. 5).

 

“Sensory data” as Gurwitsch uses the term are what might also be called the totality of apprehensions or sensory impressions, that is, the experiences of objects, of shapes, colors, of music, of sounds, and so forth (e.g., Gurwitsch, 1964, pp. 87-90). As Gurwitsch notes, 19th century psychologists such as Helmholtz, Weber, Fechner, Müller, and others held the above viewpoint (Gurwitsch, 1966, p. 5). The problem then arises as to how to explain the unity, the Gestalt-quality, of various phenomena, e.g., visual phenomena, including grouping and figure-ground effects, aural phenomena, such as melodies, and so forth. He notes that von Ehrenfels (von Ehrenfels, 1890) first employed the term “Gestalt-quality” (Gurwitsch, 1966, p. 6) to refer to these phenomena. A musical melody, in this light, becomes a somewhat paradoxical phenomenon, in that “the melody appears as a sensory or quasi-sensory impression which does not arise from any stimulus” (p. 7). That is, since the constancy hypothesis, above, implies that any sensory impression has a stimulus giving rise to it, the lack of a specific stimulus for the melody per se is puzzling for this conception of sensation. The Gestalt-quality solves this problem, but not, at this stage, in a very satisfactory manner. For one thing, it is conceived of as influenced by (“conditioned by” [p. 7]) the specific sensory impressions from which it arises, but as basically independent of them, as, in effect, a kind of higher-level abstraction which, while a sensation, is somehow “quasi-sensory” (p. 7). Thus, according to Gurwitsch, its origin as an experienced phenomenon is not clearly conceived.

Husserl, however, takes this conception further than the psychologists above, in his notion of the “quasiqualitative Momente” or “figurale Momente” (Gurwitsch, 1966, p. 9; Husserl, 2001b, §51), which are “immediately perceivable wholes” (Gurwitsch, 1966, p. 9). Although Husserl may have formulated this principle independently, his understanding is clearly influenced by the intellectual climate of the time, so much so that his conception is very similar to that of von Ehrenfels, and virtually identical to that of Stumpf (e.g., Stumpf, 1890). Both Husserl and Stumpf held that the experience or character of unity “is inherent to what is perceived, is one of the sensory features of it and is part of its constitution” (Gurwitsch, 1966, p. 9). “Prior to every activity of categorical thinking, the elements are given as forming a group” (p. 9), i.e., as part of the sensation, the perceivable whole is as explicable as any other sensation; the perception of a “swarm” of bees is just as much a part of the experience as the individual bees, for example. Stumpf, according to Gurwitsch, introduced the idea of Verschmelzung in order to explain (among other things) the phenomenon of melodies, and Husserl’s corresponding conception modifies this idea to the extent of giving “to the concept an even wider meaning so as to restrict it no longer to simultaneous data” (Gurwitsch, 1964, p. 78). The idea of Verschmelzung, then, is that of the experienced unity caused by a set of simultaneously occurring related phenomena, such that their relationship is precisely that unity (e.g., a musical chord, a swarm of bees). Stumpf is careful to distinguish his concept from that of a simple fusion of components, in which the components lose their individual qualities, and conversely, when analyzed, separate back into “distinct simultaneous sensations” (p. 79). The relationship is a new and different entity, although dependent on its components.

Gestalt-qualities, in this early conception, although they are caused by “elementary sensory facts”, correspond to “no objective stimuli whatever, and consequently, no excitations in the receptive sense-organs, but… do not lose the character of sensory immediacy” (Gurwitsch, 1966, p. 10). In addition, according to Gurwitsch (p. 10), Husserl did not conceive of the figurale Momente as the result of mental operations on the basic sensory stimuli. The figurale Momente were phenomena arising in particular circumstances, and no further explanation was given by Husserl of their generation, except that they “are a consequence… of a fusion (Verschmelzung) among the elements and the relations of these elements” (p. 10). In summary, Verschmelzung bestows

 

…experienced or sensed unity upon such sense-data as enter into this relation… but, as Stumpf points out, Verschmelzung does not modify or qualify the sense-data… the sense-data… are not only unaltered by analytical discrimination, but also are experienced exactly as they would have been if they were not given in the relation of Verschmelzung (pp. 79-80).

 

Husserl’s conception is in this regard identical to Stumpf’s. That is,

 

… that the elements happen to fuse with one another and form a group whose unity is immediate and perceptual does not mean that they undergo any modification whatever; in their fusion, they do not differ from what they would be if they were taken in isolation (Gurwitsch, 1966, p. 10, my Italics).

Sensory qualities of a higher order, qualities founded upon ordinary sense-data, are incidental and adventitious to the founding elements in that these elements are not affected by the quality they found, nor by the unity which the founded quality bestows upon them (Gurwitsch, 1964, 84).

 

The constancy-hypothesis is, then, an intrinsic aspect of the above conceptions. Since Gurwitsch rejects the constancy hypothesis, and this conception of the fusion of qualities relies on that hypothesis, Gurwitsch must (quite reasonably, I think) reject it. For Gurwitsch, there are two problems in the above. First, the group quality, or Gestalt-quality, when it arises, remains in some sense apart from the fusion which generates it, and further, the fusion itself does not alter the character of the fusing elements; the sensory totality is atomistic. Second, the Gestalt-quality is an abstraction from sensations, giving rise to a phenomenal type duality, i.e., the result of fusion is a type of experience different in kind from other sensations. For my purposes, the second problem is relatively minor, but the first, as I have pointed out, may have devastating consequences for phenomenological methodologies.

One fundamental problem, as Gurwitsch saw it, was to account for the organization of sensations without postulating some organizer which is independent of those sensations. According to Gurwitsch, any faculty or processes which stand outside, i.e., which are of different type (“higher-order” [Gurwitsch, 1964, p. 90]) than the sensory faculties, and which organize or structure them entail one or more of several possible problems. Whether that organizing faculty originates in the intellectual properties of schemata, as Piaget (according to Gurwitsch) would have it, in the abstractions of the figurale Momente of Husserl, or indeed in the processes which isolate aspects of order from a largely chaotic sensory stream, as James hypothesized, Gurwitsch argues that such a faculty, separate from sensation, leads either a) to a radical differentiation of organizing schema or principles from the sensations they operate on, and a subsequent atomistic phenomenalism, resulting in both embracing the constancy hypothesis and in a false dichotomous typology of sensation; b) to a regress of organizing hierarchies, where the origin and formulation of the particular organizing or isolating processes are themselves unaccounted for except through other, similar, processes; or c), as Arvidson notes, the problem of the transience, i.e., the lack of stability, of such organizational processes (Arvidson, 1992, p. 57), since, once applied, only sensation might maintain them, except that they are not sensation. It was a) and c) which were most problematic for Gurwitsch.

I might also note that despite the fact that Gurwitsch takes, in The Field of Consciousness, most of his examples of Husserl’s conception from the latter’s Logical Investigations, one can find support for the above in some of Husserl’s other works, for example, The Idea of Phenomenology. Thus, in that latter work, Husserl claims that one can visualize colors, and that they can “be reduced through the exclusion of all transcendent significance” (Husserl, 1970, p. 54), but that nonetheless “perception posits existence, but it also has an essence which as content posited as existing can also be the same in representation” (p. 55, my Italics).[62] The mutual independence of sensations is affirmed.

 

3. Gurwitsch’s Gestalt Psychology

 

The next development of Gestalt psychology, however, implies answers to both of these problems. Before I go into that, however, I would like to mention briefly that Gurwitsch is specifically employing the phrase “psychological point of view” (Gurwitsch, 1966, p. 11) to describe the school of Graz, and the work of von Meinong, Benussi and others, which initially developed and formulated the refinement of Gestalt theory which he terms “production”, viz., “a mental operation, an intellectual activity of a certain kind which resembles the act of grouping parts into a whole” (p. 13). Descriptions of this process, published in 1907 by Benussi, 1907 and von Meinong, 1899, sound surprisingly modern; and von Meinong could serve as a model for the present movement to “naturalize” phenomenology by uniting philosophy and psychology, in that he explicitly intended to employ his psychological theory as an application of philosophical concepts (Gurwitsch, 1964, p. 60).

What we find in this next phase of Gestalt psychology is the acknowledgement that whether a sensation’s components are “abstract” or not, they are all experienced equally as aspects of that sensation. Thus the experience of a melody (the gestalt) corresponding to or generated by a sequence of notes (the components) is as much an aspect of that sensation as the experience of the notes themselves, i.e., as immediate, as “both homogeneous and altogether a matter of sensibility” (p. 89; and see the examples of illusory contours, below). Given this insight, the Gestalt-quality cannot be conceived as phenomenally separate from any other aspect of the experience, nor can it be conceived of as a type or kind which is different from that of sensation. But it still must be acknowledged as different in some respect, and this respect is now conceived of functionally. That is, the Gestalt-quality is a “functional concept” (p. 89), an “internal condition” (p. 95) as much responsible for sensations as are external conditions. Upon Köhler’s abandonment of the constancy-hypothesis (e.g., Köhler, 1913; Köhler, 1962), “all features displayed by perception must be treated on the same footing” (Gurwitsch, 1964, p. 91), and they are dependent on “a plurality of variables” (p. 95), which include internal ones.

We have here the very modern conception of an interaction between sensory input and higher-level processes, where both modify the other, an interdependence resulting in a unified sensory experience. Implied by this interdependence are several consequences. First, there are generative higher-order mental processes which are dependent on neural processes, a concept taken for granted now, but fairly extreme then (i.e., circa 1913). To take a relatively simple example, there are common visual confusions involving apparent size, such as when an unfamiliar near object, next to a familiar, large, far object is seen as being as large as that far object[63]. The far object exerts a “higher-order” influence, since it is not simple visual sensation, but our memory of that object’s size which influences the perceived size of the other object. Second, there are components of sensations, e.g., our sense of distance from the above objects, which are dependent on these generative mental processes, implying that sensations, i.e., Gurwitsch’s “sense-data”, are more than “qualified by” higher-order aspects. Instead, they partake of them. But further, according to Gurwitsch, a percept, depending on both “external and internal conditions” (p. 95), varies, as those conditions vary, as a unified whole, a “homogeneous entity” (p. 95). In the above example, the unfamiliar object is seen, too large, as a whole, rather than some of its features being seen as large and far and some as small and near. This conception, then, takes the radical step of denying an atomistic mentalism. When this step is taken, sensations per se cannot be asserted to be atomistic in the sense that Husserl would want, viz., that their components are unmodified by manipulations of any aspects, either higher-order (“quasiqualitative Momente”) or sensory aspects of an experience, since those latter aspects are all part of the phenomenal mix, so to speak; they are all “on the same footing” (p. 91; see also, e.g., Gurwitsch, 1966, pp. 233-234).

Gurwitsch now must back up this assertion with data, and he does so with a very simple example, that of two dots seen side-by-side, against a uniform background.[64]

He points out that these two not only may be seen as two separate dots or as two forming a pair, but may be seen as the end-points of a short line segment, or as the end-points of two indefinitely long lines extending in opposite directions from those two dots. In this situation, then, there are not merely just the two dots, but (at least) four possible systems of organization of the visual field stemming from those dots, one (system) of which is usually dominant. That is, although we know that we can, or have just, seen the two dots as the ends of a short line segment and simultaneously (with that knowledge) see them merely as a pair against a uniform background, we do not simultaneously see a pair of dots against a background and see those dots as the ends of the line segment. Just as important is that all of these systems, whichever is perceived, present themselves as figures against grounds, where the figure is seen as structured in some manner, and the ground as significantly less structured. The contour of such visual figures “belongs entirely to the figure and has no significance for the ground” (Gurwitsch, 1964, p. 111). As Gurwitsch points out, this general characteristic, the figure-ground structure, is universal not only in visual but in all “perceptual phenomena” (p. 112); nor is it confined to perceptual structures (p. 113).

Given all the above, Gurwitsch concludes that “such a configuration cannot be considered as built up out of the ’parts’ of which it consists, if these parts are regarded as independent and self-contained elements” (p. 114). In addition, we see from the above example that the structures imposed on “simple” components are actually aspects of the perception of those components: the dots may be paired, they may be the ends of line-segments, and the dots, and perhaps also the line segments, are set “against” an undifferentiated, continuous background that is seen to extend behind the segments and the dots. Gurwitsch’s next point is that the components of a Gestalt are participants in this structure, to different extents depending on their functions in that structure. Some are more important to the creation and maintenance of that structure than others (p. 115).

 

4. Modern Gestalt Psychology Has Refined But Not Radically Altered Its Principles

 

Gurwitsch worked out these ideas from the early part (circa 1929) until the middle (circa 1960) of the last century. One might well ask how well they have stood the test of time. Is Gestalt psychology still taken seriously as an experimental/theoretical paradigm? If so, have its ideas changed significantly? It is well known that many of the neurophysiological ideas of Gestalt psychologists were primitive and largely incorrect (e.g., Köhler, 1971, pp. 237-251; Koffka, 1963, p. 62). But what of the general principles as I describe them above, relating to the unity of experiences, the existence of experienced phenomena resulting from high-order, i.e., functional processes, and the indistinguishability of those latter experiences from other sensations?

In fact, virtually the same principles, now hypothesized as instantiated in distributed neural networks, are routinely invoked, and the term “gestalt” is still employed to refer to such unified perceptual, and even cognitive, experiences. That is, it is now accepted as fact that the visual modality of the central nervous system, to take one example, employs both local and top-down neural processes which result in the generation of complex and unified experienced patterns. These patterns can occur, as visual experiences, even in the total absence of visual stimuli corresponding to them, as Gurwitsch anticipated. It has been established, for example, that people can clearly see, i.e., have the visual experience of, figures and outlines of figures which “fill-in” between isolated points, and that this experience is not abnormal, but can be induced in human subjects at will. The analysis of certain “subjective” or “illusory” contour figures[65] which are actually generated from the filling in of absent contours is still actively being researched (e.g., Mendola, et al., 1999; Lesher, 1995; Shipley and Kellman, 1992; Idesawa and Zhang, 1999). This class of phenomena so clearly provides support for Gurwitsch’s claims that I would like to elaborate on it somewhat.

Lesher provides multiple drawings in his article which clearly evoke these phenomena in a reader. Briefly, he states that “an illusory contour is defined as the percept of a clear boundary in regions where there is no corresponding luminance gradient” (p. 280). One of the simplest drawings is that of four circles arranged at the corners of a nonexistent (i.e., not drawn) square. If the circles are unbroken, we merely see, in effect, four rather large dots. If, however, the circles have right-angled wedges cut out of them at the locations on which the corners of a square would rest if such a square were actually present, and the circles are close enough, virtually all observers literally see a square, while simultaneously being aware that there is actually no square drawn on the page. There are many other illusory contours which can be induced and which Lesher and Idesawa present: lines, circles, triangles, and more complex figures which are clearly seen, and, paradoxically, are simultaneously clearly not seen. Both Lesher and Grossberg (Grossberg and Mingolla, 1987), and others, have created theories modeling the neural bases of these effects. We are, therefore, not dealing with pure phenomenology nor with merely observational empiricism here, but with full-blown, theoretically and empirically-based, extensively researched confirmation of Gurwitsch’s contentions that

a) the constancy-effect does not hold, that

b) our perceptions are not simply of “real-world” stimuli, that

c) higher-order effects are experienced as immediate, clear, and reproducible aspects of sensations, and that

d) those sensations are thoroughly holistic[66],

in that the illusory contour is utterly dependent on the components, while the components themselves (e.g., the circles without wedges, above) are experienced as generating aspects of those contours, i.e., as “defined and determined” (Gurwitsch, 1964, p. 130) by the whole (see the figure on the next page).


Figure 1: Illusory Figures

The effect of gestalt grouping on such basic visual phenomena as persistence and extinction of figures and contours is also established (e.g., Ward, et al., 1994); Humphreys’ summary of much of the literature on visual binding and grouping (Humphreys, 2001) indicates that there are multiple types of processes involved, ranging from low-level grouping resulting from local processes on the retina, to high-level processes involving the interaction of stimuli and conscious attention. Peterson and Kim (2001) still employ the “Rubin vase-faces display” (p. 330), as does Gurwitsch (1964, p. 118), to illustrate figure-ground effects, and state that “grounds are not shaped by any contours they share with figures; they appear to simply continue behind the figures near those contours” (p. 329). This quote, quite original (as far as I know) to the authors, might have been lifted from the early Gestalt literature or from Gurwitsch, above. From the dates on the above papers, merely a selection of the huge literature available, it is easily seen that Gestalt principles are quite alive, actively being researched, and have been established with firm neural (e.g., Lesher, 1995; Martinez and Alonso, 2001; Grossberg, 1997) as well as experiential bases. Since most of the ideas that I will utilize in my own model of the structure of consciousness are also based on these principles, it definitely behooves me to establish their legitimacy.

 

Although I have concentrated so far on visual perception, it is not merely in that modality that Gestalt theory may be employed. There is active research on the relationship between music and Gestalt theory, for example. Shepard (1999) claims that gestalt grouping principles are involved in music perception; Dowling (1994) has hypothesized that melodic contours are musical gestalts in conjunction with scales. That is, he found that not only were melodies heard as unitary experiences, but altering the key in which they were heard changes our ability to remember them (e.g., p. 186). Contour, interval sizes, scale and rhythm form a gestalt which characterizes a melody despite some degree of alteration of those parameters (e.g., pp. 180-182). Terhardt (1987) goes so far as to suggest that music is processed in hierarchical nested gestalts (e.g., pp. 160-161). The Gestalt concept, then, is applicable not only to the visual, but to other perceptual modalities. Thus, in addition, Tsur (2000) has recently investigated the figure-ground relationship in music as it relates to other arts, such as poetry. Similarly, Aksentijevic, et al. (2001) and Kubovy, et al. (Kubovy and Van Valkenburg, 2001) hypothesize cross-groupings between the visual and auditory modalities. In addition, I will argue later that it is desirable to extend this applicability to more abstract cognitive and linguistic realms.

Gurwitsch emphasizes the holistic characteristics of gestalts even to the extent of criticizing William James, whom he greatly admires. He argues that James does not take sufficiently into account the extent of part-whole interactions. While James considers that presently experienced mental states are influenced by those preceding and following, he does not go far enough in his characterization of the nature of that influence. Gurwitsch insists that a “datum has its phenomenal identity only within [a] contexture” (Gurwitsch, 1964, p. 130), and that James does not take that context as “the definition and determination” (p. 130) of a substantive part of a whole, but maintains (according to Gurwitsch) that only “a certain shading of the presently experienced mental state” (p. 131) occurs. Gurwitsch is attempting to emphasize the reciprocal determination of the part and the whole, a determination which may entail influence well beyond the “shading” of either, as we have seen from the illusory contour example above.

 

5. First Statement of the Problem: Husserl’s Essentialism

 

Below, I will attempt to explicate more thoroughly some aspects of the problem I have outlined above. In doing so, I am going to only very briefly address Husserl’s writings on the problem of constitution, because I wish to approach that issue primarily through Gurwitsch’s treatment of it, and because Husserl’s treatment is spread through an enormous quantity of writings on other issues. Husserl starts dealing with this topic as early as the Logical Investigations (Husserl, 2001a; Husserl, 2001b), and addresses some of the dynamic aspects quite extensively in his work on time consciousness (Husserl, 1990), and in other works, but I simply do not have space to treat in depth his writings on this issue. I will dip into them, briefly, but will rely mainly on Gurwitsch, his pupil, and other commentators to summarize Husserl’s viewpoint.

Let us therefore turn to the specific issue of the constitution of phenomenological objects. I am taking Husserl’s term “constitution” (konstitution – e.g., Zahavi, 1992, p. 120; Sokolowski, 1964) to refer both to the dynamic processes of generation and/or formation of unified objects. One might conceive it, temporally, as the unification of a variety of experienced phenomena (“primal apprehensions”: Sokolowski, 1964, p. 540) into a single phenomenon with discernible components. Thus, we may say that an object is constituted actively as it is apperceived (e.g., Zahavi, 1992, p. 113), or that we may investigate the more-or-less static constitution of some given perception. It is easy enough to find examples of this in perception, where visual components form a unified (static) image, or auditory components a (temporally constituted) melody. In addition, the process of abstraction, in which several specific examples are generalized and united under a single abstract concept, is also an example of this set of processes which I will term, among other phrases, “gestalt unification”, and which Husserl termed, among other names, “constitution”. The issue of the nature of the constitutive processes has profound implications for Husserl, as we shall see. It was in fact the inadequate treatment of constitution by Husserl that led Gurwitsch to embrace Gestalt psychology, which in turn led him to create a phenomenal description quite at odds with some aspects of Husserl. Before I go into detail on this issue, I will present an argument which reveals severe problems both for Husserl and ultimately for Gurwitsch.

Inasmuch as it is possible to speak of Husserl having a clear position on the nature of the components of phenomenal experience, given the changes in his thought as he matured, he was always an atomist of some stripe, as we have seen above. That is, for Husserl, the components of phenomena were essentially unaltered as they were concatenated or grouped with, or separated from, other components. In fact, this is a necessary implication of his positions on the epoché and the method of variation. First, the epoché is a technique which, ideally, takes the component of existence present in certain phenomenal objects, viz., the experience of an object as an “objective” entity, and sets it aside (“brackets” it), as we have seen above, in order to discover the “pure” phenomenological perspective on that object. That is, that component[67] of the object is, if not eliminated, at least altered. Implicit in this, then, is the conception that in all other ways, that object will be essentially unchanged by that alteration. Otherwise the operation of bracketing would result in far more than this one change, and Husserl vehemently denies this, as we have seen above.

While he does maintain that there are radical implications, i.e., for philosophy in general, of this change, he denies, as we have seen, that the phenomenal change itself, the change in the object as a result of the operation of bracketing, is radical in the sense that any other aspects or components of that object are altered. But in order for this to be true, the object’s other components must then be, effectively, atoms which are merely conjoined with the idea of its objectivity, i.e., concatenated or grouped as independent, even if interrelated, components, in the totality of the experienced phenomenon. Let me put this more simply. Husserl maintains that bracketing makes irrelevant the objectivity of an experience, but alters nothing else about that experience. That implies that an object’s other components must be unaltered by the change in that former component, and that in turn implies that unless the component of objectivity is somehow radically different from other components in how it is combined with other components, all components of phenomenal objects must be similarly independent of one another. But Husserl does not say that this component is different in that respect[68], and in fact, he maintains that phenomenal objects are unchanged in all other respects after bracketing. That latter position is actually necessary if he wants the effects of bracketing to be limited to that suspension, which he most emphatically does.

Second, the logic is the same when applied to the method of free variation (eidetic variation), the methodology employed to actually discover an object’s essence. Here there is an even more thorough shifting, changing, and alteration of the object’s components (e.g., Ihde, 1977; Husserl, 1995, pp. 70-71). But again, Husserl’s strongly held position is that despite what can be remarkable changes in a phenomenal object as a result of these variations, there is a “core” or “essence” which is unchanged through the possibly radical addition, subtraction, and alteration of those components[69]. More specifically, it is an implicit assumption of Husserl’s, not merely that the kind of relationship realized in the various exemplars of an essence implies a radial relationship between those exemplars and the essence (i.e., that they overlap at a common “essential” core, or, alternatively, that an identical “essential” set of components is present throughout all possible deriving sets of exemplars[70]), but that those exemplars, characterized by altered sets relative to each other of components, retain at least the common essential components intact, unaltered by their phenomenal context. But how could this be, unless that essence, and indeed all the components, were independent of each other in the above sense, i.e., unaltered by various combinations and interrelationships[71]? For if the essence or core components of phenomenal objects were altered by variations in the other, non-essential, components, then there would be in fact no essence, and Husserl would be faced with problems in both his positions on the formation of abstractions and with his claims concerning the foundations of his metaphysics: phenomenology would not be able to discover the essences of objects, and his solution to the Cartesian dilemma would be unfounded. The very basis of Husserl’s metaphysics, and of his claims of phenomenology’s uniqueness and profundity, then, rest on a pervasive atomism.

Gurwitsch makes what seems, in this context, an odd claim. In contrast to his assertions that gestalts are holistic, with interdependent components, he insists that what Husserl terms the “central noematic nucleus – that which is intended, taken exactly as it is intended” is invariant “with respect to variations concerning noematic character” (Gurwitsch, 1964, p. 180). I must admit that I do not understand how this can be in harmony with his[72] general position. “Noematic character” here is taken as variations in one’s perspective: “my idea of Greenland differs from that of an arctic explorer, though the object is the same” (p. 178). “Object” here refers, of course, not to the material, “real” Greenland, but to the ideal, the essence, the result of the process of eidetic variation. Now, for Gurwitsch to remain a Husserlian, which is his aim, he must indeed maintain that this invariance holds universally. But if we are to be consistent with the above conception of gestalts and conceptual structure, we must deny that it is necessarily the case. The issue here is rather black and white. If there is even one instance in which the eidos, the noematic nucleus, the essence of the phenomenon, varies with variation in perspective, then the claim that the phenomenological reduction is an ontological and/or epistemological source of apodicticity is simply not true. This is not a situation in which one may allow a few exceptions, if one is, as are Husserl and Gurwitsch, making strong claims. From my empirically-oriented point of view, it is perfectly permissible that the noematic nuclei of many concepts may either vary or be invariant under alterations of “character”, and it may indeed be the case, for example, that there are invariant nuclei of concepts in most formal languages[73]. But I am not making claims about certainty; to the contrary, I am denying certainty. The onus, then, is on the Husserlian to refute both Levin’s general objections, above, and my empirically-based objections (continued below).

One might object that if the above atomistic conception were not the case, then the intuition or experience of identity would have no basis. But on the contrary, there is no need to assume that different experiences brought under the rubric, say, of redness must be identical; they could be similar, but different. That is, one might claim, for example, that the experience of identity is based on either a formal, algorithmic, procedure or on some non-formal intuition. If it is based on a formal procedure, then one is claiming that the mind proceeds in this instance as a digital computer does: a procedure compares the features of two elements, and if and only if those features are the same, the elements are experienced as the same. There is, however very little evidence that the mind, or brain, operates in this fashion, except in unusual circumstances (i.e., in the conscious enumeration of features in very visually confusing contexts); there is in fact, as I will show, evidence to the contrary[74]. Alternatively, it may be that an intuition based on other, more approximate (e.g., analog), processes may be more likely to provide us with a reasonably accurate experience of identity. If this latter is true, then the intuition of identity and the intuition of similarity would seem merely to differ in degree rather than type; and if that is true, then there is no necessity, again, for essences. Further, experiences felt as similar may be related, for example, as Wittgensteinian families. That is, the idea of an essence may in many cases be replaced by that of a “string” or even a “group”, and there may be no central overlap, thus no “core” at all for a variety of related experiences (see also Levin, 1970, footnote 70, p. 184-185). Further, as I will argue below, there is now strong experimental evidence (e.g., Smith and Sloman, 1994; Sloman, et al., 1998; Sloman and Ahn, 1999; Rips, 2001; Gennari, et al., 2002; Sloman and Malt, 2003) that there is no “core” or central set of components to many concepts, i.e., very little if any functional difference between “necessary”: essential, or “characteristic”: similarity-based, features. As Thibaut puts it, “Smith and Sloman obtained a small dissociation [between necessary and characteristic features] only in the sparse condition and under think-aloud instructions” (Thibaut, et al., 2002, p. 648).

 

6. Examples of Some Problems with Essentialism

 

Let me present some examples to concretely illustrate this problem. Suppose that we want to ascertain the essences of lamps and the essences of tables. Following Husserl, we would employ the method of free variation on examples of lamps, and after varying many characteristics of lamps, attempt to intuit or grasp the overlap, the unchanging aspects common to those variations, in order to grasp the essence of lamps. Similarly, we would at some later time apply the same methodology to tables, and could undoubtedly come up with an enormous number and variety of possible tables, during which we would attempt to grasp, as with lamps, their essential core qualities.

Now the problem that could arise would be if, in our quest for more variations on lamps, we hit on something that, quite unintentionally, we also found in our quest for variations on tables: a kind of lamp-table. The first problem, then, is whether we actually found something with two essences, with only one essence that we mistakenly thought had another (the lamp-table was really only a lamp), or with neither essence. How do we settle these issues? But there is another problem, perhaps even more severe. Suppose that the lamp-table simply had no characteristics in common with the first instances, the instances of either lamps or tables with which we started our searches, but did have characteristics in common with the variations immediately previous to itself. That is, in one context, we understand it as a lamp, in another context, we understand it as a table; and only afterward, comparing those contexts, are we aware that this is, in all its components, identically the same object in both cases. That is, suppose that there was no overlap of characteristics between the lamp-table and either the initial lamp or the initial table? Must one dismiss this possibility as fantasy? If not, we would then be in a very strange position, if we were Husserlian phenomenologists, of admitting that the concept of lamp, let us say, had the structure of a Wittgensteinian family, with no core, but merely successive overlapping characteristics which instantiated the concept; and we would also have to say that these characteristics varied with context: in a “lamplike” context the characteristics which give rise to (i.e., cause us to experience) a table in a “tablelike” context, give rise to a lamp. Before this latter hypothesis can be dismissed as wild fantasy, we must somehow account for Sloman et al’s results, above, where they found precisely that: no central core of concepts within particular conceptual classes. Further, Rips points out that an “interaction view”, for which he cites experimental evidence (e.g., Rips, 2001, p. 846-848), allows for natural kind membership with no essential properties. In this viewpoint, “an object’s membership in a natural kind depends on whether the object instantiates the laws for that kind” (p. 847). 

Let us take another example. Suppose that one person wants to intuit the essence of the number six (i.e., “sixness”), another the essence of “fiveness”. This is, or should be, an extremely clear-cut case of an abstract or ideal object, one which, as a number, has indeed given rise to speculation about the existence of ideal objects. In order to accomplish this intuition, we might, I assume, proceed by asking a (phenomenologist) friend to view and/or visualize examples of six objects: six cards, six flowers, six mountains; and find, eventually, that the common core was sixness. Later, to intuit fiveness, we would proceed similarly, with another (hopefully a phenomenologist, although they are few and far between) friend employing sets of five objects. Now, suppose that we offered to these two people, after they had respectively examined multiple examples of six and five objects, the same set of piles of sand as an example late in this sequence of exemplars of sixness and fiveness. Suppose we arranged the piles so that there were five, but one of those five was spread-out enough that one might confuse it with two piles. I find it quite conceivable that one person might see six piles, and another five, depending on context, i.e., depending on what number of other objects they had been seeing. What, then, first, of the essence of the numbers, and second, the fact that the same characteristics are experienced as giving rise to “sixness” in one context, and “fiveness” in another? On the surface, a Husserlian could dismiss this as a common type of ambiguity which would not affect the phenomenological project. Yet one must ask, as I have previously, just how close we must approach similar sets of exemplars, and different essences, before we feel the need for verification of the process of variation itself. That is, if fiveness, say, has unclear referents, then we are entitled to ask of what it is the essence; and further, if what is presumably the same essence – fiveness – refers to different exemplars for different people, in what sense is it an essence?

To put this more generally, let us take the classical example of the tree in the garden. The phenomenological stance maintains that although one can take a variety of perspectives toward the tree: one can move around it, imagine it growing, denuded of leaves, and so forth, that despite all of these various intentional stances, the tree remains the same tree, and we experience the tree, through all those perspectives, as the same tree. Thus, according to that stance, there needs must be an essence which remains constant through all the perspectival changes, and that essence provides the anchor, so to speak, for the identity of that tree. However, I maintain, that to the contrary, what we are experiencing during those perspectival alterations may have no constant element at all, except for one: our intuition, explicit or implicit, that this is the same tree. That intuition of sameness does remain constant, or relatively so, throughout the possible variations that our experience of that tree goes through. We must then ask, first, is the tree the same, in the sense that some of its components, and most particularly an essential set of those, remains constant through the alterations? But this does not follow, nor does it follow that such a set is necessary for the intuition of sameness to be present. Recall the argument above employing Wittgensteinian families. Such a progression through components might well maintain an intuition of sameness; at any rate, there is no argument I know of demonstrating that this is not possible. Second, is that intuition of sameness correct? But why should it be? It is merely a feeling, and we will see in the last Chapter of this dissertation that such feelings can indeed be in error, and/or due to an unexpectedly wide variety of factors. Thus, altered sufficiently, one perspective on the tree may not have any components identical to the perspective with which one started, although we may experience the intuition that it is the identical tree. That is, it is possible to dissociate the intuition of identity from the actual experience of identity.

A simple example will suffice to illustrate that point. If we hide most of an object behind a screen, so that one sees only its rear, and then push the rear so that it disappears behind the screen, while simultaneously the front of an object appears at the other side of the screen, in the direction of motion of its “rear”, we will almost invariably assume, even if the “front” is widely disparate from the “rear”, that it is the same object emerging from behind the screen. There may be no phenomenal similarity or connection between these two objects, aside from similar velocity and direction of motion, and their masking by the same screen. What then is the intuition of identity here based on, in terms of the object’s components, i.e., an unchanging core or essence? Surely its essence cannot be merely the similar velocity and direction of motion of its “front” and “rear”.

 

Now, the above arguments directly address one type of essentialism, involving common components which are then sequestered, so to speak, to realize an essential set of such components. Yet it is possible to object to this conception of essentialism and maintain that in fact what is occurring is that whether or not there are common, core components to individual perspectives or individual exemplars, nonetheless an essence is generated, so to speak, from the variant but similar components of the various exemplars. Thus, in this conception, the essence is a set of components which arises more-or-less spontaneously from the exemplars yet is in a sense independent of them, in that it is not present in any particular exemplar, but instead requires a set of them to arise.

But the same kind of problem as with the previous conception of essences holds here, in a slightly different form. If we take a step up in abstraction, and consider the superset of sets of essences of the same objects, generated over time and over subjects, i.e., the repeated generation of the “same” essence at different times by the same person, or the generation of the same essence by different people, we find a situation in which the regress stops. Here, we must concede that different exemplars of the same essence must have common components. One cannot here argue that this regress continues, and that there is an essence generated, in a next regressive step, from a set of essences. In fact a set of essences must, to support a classical phenomenological position, manifest just the type of set of common components which the previous conception of essentialism, above, demanded in the sets of exemplars.

This condition is similar to those I employed in analyzing problems with phenomenological methodology. In order for essences to be communicable, useful, reproducible, and/or verifiable, they must be able to be duplicated, even if only by the same person, in different instances. To ascertain this it must be ascertained that the essence of the tree in the garden today is in fact the same, to some reasonable extent, as the one found yesterday. And thus there must be a direct comparison rather than the generation of yet a further, higher-order, essence over multiple instances of “lower-order” essences, and that comparison, given an atomistic epistemology and/or metaphysics, must assume not merely similarity of components, but the actual identity of some set of core components, just as in the previous argument. The only difference is that in this case the core components are of the set of essences. We are then faced with the same problem, at the level of essences, as we had earlier at the level of exemplars, viz., we must assume that those core components are atoms which do not vary in different contexts, where each context here is an essence which might be slightly different from another instance of the same essence. But again, all we really know is that our intuition of identity is constant, not that the components of which we have that intuition are constant.

One might object that in fact essences do have to be identical; that the multiple essences, in the above sense, of a particular tree, or even more strongly, of numbers and other formal objects, e.g., fiveness, whatever the exemplars giving rise to it, must, as essences, be identical with each other. The problem here is not with essences in purely formal systems; they may indeed be identical when they refer to the same formal entities. Rather, it is the applications of such systems, and essences, back to the exemplars from which they sprang, the empirical applications of these essences in formal systems, their reference to real-world situations, that again give rise to the same problem. For here again the identical core or common components are not sufficient. We need merely think of the example, above, of fiveness and sixness. As purely formal essences, applied only in formal arithmetic, one can surely claim, at least, that fiveness is, or should be, always identical. Yet when we attempt to apply it to situations where one person has derived fiveness from one set of exemplars, and another person has derived sixness from the same set, as with the piles of sand above, we find that in fact there is not an identical set of components to the essences, and my objection (and see the general argument below) applies.

Consider also the example I gave earlier, involving non-Euclidean geometries. Empirical considerations effectively split the essence of the parallel postulate (or of parallel lines). Where there was first one essence, and the formal system of Euclidean geometry proceeded unaltered and unquestioned, upon consideration of data relating to space-time geometries, that essence was altered to (at least) three. Similar arguments apply to the number of degrees in the angles of triangles on planes, spheres, and pseudo-spheres (surfaces of negative curvature).

Thus, until an abstract formal system relates to real-world applications or to the exemplars from which it is, and could be, derived, we may indeed consider that Husserl and similar essentialist positions might be correct. Some formal systems might be understood as operating with concepts or essences which are simple, atomistic, and identical in their components. But as soon as those systems either refer to the world from which they originated, or as soon as those systems are applied to that world, they can no longer be assumed to be employ atomistic, essentialist essences. The differing components relating to the empirical alter the components relating to the formal, and essentialism must be radically rethought.

I am not, however, attempting to make this particular issue black and white; I am sure that in many cases there are core concepts behind ideas or appearances. This argument does not refute essentialism, in general. But, if correct, it does limit essentialism to formal systems, and perhaps to some restricted empirical cases. What I hope to have indicated in the above examples is that phenomenology must employ empiricism to resolve issues of ambiguous phenomenological analysis when traditional phenomenological methods fail.[75]

 

7. The Heart of the Problem: Husserlian Methodologies are Atomistic and Atomism Is Not Compatible With Gestalts

 

But if my reasoning is valid, then what of the constitution of objects? Given the above reasoning, Husserl and those of a similar ilk must maintain that phenomenal objects are sets or collections of independent components and their relations. A Husserlian is faced, then, not only with explaining the unity of objects (and I agree with Gurwitsch that this surpasses mere fusion, i.e., Verschmelzung [e.g., Gurwitsch, 1966, p. 252]), but further, with explaining how it is that the components of objects do in fact alter with changes in phenomenal context. Husserl has no real answer to these issues, as we have seen above. It was, as I have mentioned, these problems that were some of the driving forces behind Gurwitsch’s (Gurwitsch, 1964; Gurwitsch, 1966) reformulation of phenomenology, and that motivated his embracing Gestalt psychology. Mirvish, 1995, puts this problem into a historical perspective, noting, as did Gurwitsch (above), that Husserl’s approach contains assumptions carried over from introspectionist psychology. Although Mirvish realizes that these lead to the implicit notion on Husserl’s part that the alteration of a concept, e.g., by bracketing, must leave the remainder of that concept (the “residue”) intact, he does not work out the further implications of that inference for Husserlian phenomenology. Merleau-Ponty, however, does seem to realize these consequences (Merleau-Ponty, 2001). Toadvine, in his characterization of Merleau-Ponty’s critique of Gurwitsch, states:

 

Merleau-Ponty suggests that eidetic analysis falsifies transcendence by transforming it into relations between essences.… The relations between things within the perceptual field, and the relation between theme and field, cannot be accounted for in terms of noematic structures. Perceptual identity is based on a carnal grasping of the whole perceptual field; it is not based on a synopsis within consciousness of previously separate elements…. The essence is not a positive element…. The unity of the thing is of a piece with the unity of the entire field…. Hence the eidetic method is in reality an idealistic variant of the constancy hypothesis. [my Italics] (Toadvine, 2001, pp. 199-200).

 

I want to emphasize the importance of the implications of the above points. Suppose that a classical (Husserlian) phenomenologist wishes to investigate experienced phenomena, very generally, in order to determine whether the components of phenomenal experiences do indeed vary independently, as Husserl claimed, or whether they are mutually dependent, as the Gestalt psychologists and Gurwitsch claim. How is this investigation to be conducted? It cannot be conducted through empirical studies, because then phenomenology would, first, be guilty of “psychologizing”, and second, because the methods of phenomenology (viz., the epoché and free variation) would not then be employed. One would in that case not actually be conducting a phenomenological investigation. But if either or both of the epoché or method of variation were utilized to investigate this general issue of independence, then the phenomenologist would be caught on the other horn of the dilemma: assuming what one is attempting to show. For those methods rely, as we have seen, precisely on the hypothesis that would be investigated: the independence of components. I have argued that although Gurwitsch was not fully aware of the implications of this dilemma at the heart of phenomenology, his arguments against what he terms the “constancy hypothesis” support my claims here. Gurwitsch, however, was very aware of the initial problem relating to the independence of components, and The Field of Consciousness was in part his answer to it.

One might conceivably reply that classical phenomenology can comfortably rest on circularity. This might be true of Heidegger’s variant of phenomenology, but can it be maintained for Husserl? I claim that this cannot be true. For example, Husserl states, “As [mental acts] are essentially related to one another, they display a teleological coherence and corresponding connections of realization, corroboration, verification and their opposites… They logically bring together acts…” (Husserl, 1970, p. 60). Husserl is employing, at least in part, logical criteria as fundamental relationships in the constitution of objects. How then could he have accepted circularity in his methodological criteria?

 

In summary, we have seen that in order to utilize the phenomenological methods of the epoché and/or the method of free variation, one must assume

Š      that there are “core” or “essential” elements to virtually any experienced phenomenon (or particular essence) which are unaltered by, in the latter case at least, varied and profound alterations in their surrounding contents.

Š      But if this atomistic picture of phenomena is true, then Gurwitsch must be wrong in his assertions regarding the profound and essential interconnectedness of gestalts.

Š      But modern experimental evidence and theory both back up Gurwitsch’s assertions.

Š      If that is the case, then there must be at least some incidences where the alterations implicit in the above phenomenological methodologies do in fact alter all components, including the putative essences, of certain phenomena.

Š      So those essences so altered cannot be determined by phenomenological methods, nor can those phenomena be ascertained to have, in fact, essences at all; and we do not in fact know how large or extensive is the set of such (possibly) non-essential phenomena.

Š      Thus, theory and methodology based on Husserl’s atomism must be not merely practically difficult, but theoretically incorrect.

 

Further, Husserl’s hypothesis that phenomenology can discover the essences of phenomena and thus put philosophy on an apodictic and “scientific” basis is shown to be incorrect, since if even one single phenomenon had no essence theoretically determinable by phenomenological methods, those methods would, as the revealers of ultimate truth, be invalid.

But the above conclusions are certainly true of Gurwitsch’s variant of Husserlian phenomenology also. Despite his desire to employ both the above methods, his own conception of the nature of gestalts, backed up with extensive evidence, demonstrates that he must also abandon both those methodologies as sources of apodicticity.

Gurwitsch’s insight was that since gestalts[76] did not follow the principles of the constancy hypothesis, even approximately, the result of the phenomenological reduction, viz., the epoché, is in fact instantiated when we experience, and explored when we investigate, gestalt-qualities. That is, the gestalt is, in his conception, a primitive or prototypical (“incipient phenomenological reduction” [Gurwitsch, 1964, p. 169]) form of the grasping of ideas we find as a result of the epoché. He cites Koehler’s claim that the “perceptual world in which we live and act [is] the basis from which every science… must start” (p. 169) as further justification of a connection to Husserlian thought. However, there are severe problems with that connection in addition to the contradiction at the heart of this reasoning which I have described above.

Given the above, then, we must ask, first, what is to become of Gurwitsch’s approach and his results? Second, what is to become of phenomenology in general? Whatever may become of Husserlian phenomenology, I do not believe that a naturalized phenomenology need be abandoned. As far as Gurwitsch goes, given that his analysis of the nature of gestalts is correct, and indeed we have found it to be supported by modern investigations, one can, then, apply his insights into the structure of experienced phenomena and to consciousness as a whole in a manner quite compatible with his analyses in The Field of Consciousness. We can approach phenomenology from the gestalt standpoint, as long as we do not regard phenomenology as an apodictic philosophy. Thus, the epoché and the eidetic reduction become techniques which are additions to the huge array of introspective techniques and various methods of investigating first-person experiences already employed by psychology, cognitive science, linguistics, and consciousness studies. Why not completely absorb what might be thus termed “structural phenomenology” into psychology, in that case? In a sense, that should happen. In another sense, there have been no efforts at systematic and explicit first-person analyses, in general terms, since the disastrous enterprise of Titchener (e.g., Titchener, 1901) and his cohorts. Given that contemporary insights can be integrated with Gurwitsch’s, the latter’s analyses must be modified somewhat, but we will find that a structural phenomenology based on this synthesis can be extended significantly farther than Gurwitsch’s.

There are other approaches to phenomenology which employ empiricism. The most well-known is probably the early and middle work of Merleau-Ponty (e.g., Merleau-Ponty, 1968; Merleau-Ponty, 1970). Gurwitsch both utilizes and criticizes the latter’s approach (see below). William James (e.g., James, 1950b; James, 1950a; James, 1996) might be considered in this light also, and Gurwitsch, a great admirer of James, cites him extensively. In the course of establishing my direction through Gurwitsch and gestalt theory, I will necessarily mention their, and others’, ideas.

 

VII. Now What? The Case for Empiricism and the Gestalt

 

In order to proceed, then, past Gurwitsch and Husserl, we must take into account the above discussion in the following manner. First, we must concede that the pressing question for Husserlian thought, viz., the nature of the existence of the world and its constitution by experienced phenomena, must go unanswered. The uncertainties in Husserlian practice, in the nature of apodicticity, and in Husserlian methodologies require us, as I have argued, to bypass these questions. Thus, Gurwitsch’s criticism of Merleau-Ponty, that “no transcendental question is raised… as to the constitution of the pre-objective world… he accepts it in its absolute factuality” (Gurwitsch, 1964, p. 171), is one which I cannot support.

The arguments above certainly do not refute Husserlian metaphysics; how could they? But they necessitate our ignoring questions about “the constitution of the pre-objective world” as irrelevant and unresolvable, at least so in the context of classical, and thus also naturalized, phenomenology. As I mention above, Gurwitsch had the insight that the abandonment of the constancy hypothesis was an “incipient phenomenological reduction” (p. 168) in that it detached one, in effect, from the idea that sensory data, as he termed it, necessarily corresponded to real-world, i.e., transcendent, phenomena. To describe a gestalt purely as a sensory experience was to engage in phenomenological description[77], because the gestalt possessed internally-generated aspects and cohesion which necessitates description “exactly as it presents itself through the very perception without any reference to an extra-perceptual reality” (p. 170). He makes this declaration:

 

As a consequence of this relinquishment [of the constancy hypothesis], the description and investigation of what is given to consciousness is emancipated from considerations concerning constellations of stimuli…. No knowledge of objective things and events, no considerations as to what “must” occur, given certain stimuli, and the relations between them, must influence pure description. The latter must not be modified nor obscured by what we learn about the world from the natural sciences. (Gurwitsch, 1966, p. 113) [Brackets added]

 

His logic seems to proceed thusly: 1) “in the analysis of a given perception, we deal with the thing as it appears”; 2) “one is then immediately confronted with the problem of the relationship between the thing as it appears and the thing as it really is”; 3) “Gestalt theory thus lead[s] to the problem of accounting for real things in terms of things as experienced” (Gurwitsch, 1964, p. 170). We can immediately see, however, that 3 does not necessarily follow from 1 and 2. In fact, the course of cognitive science has instead resulted in: 3a) Gestalt theory thus leads to the problem of accounting for experience (viz., gestalts) in terms of real things, i.e., the central nervous system, or, broadly, the human organism, as it interacts with the world. We find, in current literature, quotes such as the following:

 

This article develops the FACADE theory of 3-dimensional (3-D) vision and figure-ground separations… the model describes how geometrical and contrastive properties of a picture can ether cooperate or compete when forming the boundaries and surface representations… (Grossberg, 1997, p. 1); …a gap remains in our understanding of how visual percepts arise from neurobiological properties of identified neurons. A step towards closing this gap is made herein by modeling how perceptual groupings might emerge from interactions of cells with known receptive-field properties... (Grossberg, et al., 1997, p. 106).

 

And so forth; there are many other articles which I could cite dealing with the neurobiology of gestalt formation. To put this in another perspective, Gurwitsch, in describing relinquishment of the constancy hypothesis, seems to take the fact that the human sensory field contains, even creates, sensation as license to claim its independence of “objective things”, especially if those things are considered from a “scientific” point of view. Yet we have seen that just that latter point of view has resulted in explanations, even from the relatively crude basis of current digital technology, of many of the holistic gestalt properties which Gurwitsch cites as triumphs of the abandonment of the constancy hypothesis.

Now aside from the logical problem here – and, after all, that 3 does not necessarily follow does not mean it is incorrect; 3a does not follow logically either – Gurwitsch still has the problem I described above relating to the epoché[78]. But as we have seen, given the abandonment of the constancy hypothesis, while it may follow that the analysis of sensations in terms of gestalts is a phenomenological analysis, it cannot be one on Husserl’s terms. And what that means, then, is that while one may compare the experience or description of a gestalt to a phenomenological reduction, it remains a comparison only, because the holistic nature of the gestalt precludes the atomism necessary for a true epoché. But in a sense this is a good thing. If indeed a gestalt description were precisely an epoché, then there would have been something wrong either with the necessity for abandoning of the constancy hypothesis[79] as an intrinsic aspect of gestalts, or with utilizing the gestalt (since it is a whole) as a phenomenological description. But the former is highly unlikely to be true; we have seen that both older and modern data supports this conclusion.

These points raise the question as to whether the latter is true, i.e., is the gestalt, generally, an accurate descriptive scaffolding, an accurate basis for structuring phenomenological description? The rest of The Field of Consciousness attempts to answer this question in the affirmative, but it does so by comparing gestalt structure with Husserl’s analysis of consciousness. Yet it seems to be that the latter is in some respects, at least, deeply flawed. And in fact if I, now, answer this question in the affirmative, I am, given my conception of the gestalt, also sustaining a criticism of Husserl and of Gurwitsch. On the other hand, that criticism does not, at this point at least, concern any specifics of their analyses of consciousness[80], but of their metaphysical and epistemological claims to apodicticity and to certain claims, some of which I have explicated above, concerning particular types of knowledge and of the nature of science.

Why, then, do I consider the gestalt to be the paradigm most useful in describing experienced phenomena: the contents of consciousness? My reasons are similar to those of Gurwitsch. While I do not agree that experiencing or describing gestalts entitles one to any of the metaphysical or ontological claims of Husserlian phenomenology, and further, that whatever epistemological residues of Husserlian apodicticity that lie in introspective claims have been virtually erased by the considerations above, it is nonetheless my contention that the gestalt does describe experience qua experience, for several reasons. First, there are aspects of experience which are internally created[81], such as the illusory contours, above, and it is precisely the gestalt, as a construct and a conception, which accounts , albeit somewhat differently than Gurwitsch’s early studies indicate, for those contributions offered from us to the world.

Second, our phenomenal experience is holistic[82]. I will characterize a phenomenal whole as follows. Such a phenomenon has, first, the property that when any of its components or aspects are altered, all of its other components are (or may potentially be) altered. Second, a holistic phenomenon (which I will alternately term a “gestalt”) has the property of being a unitary experience. That is, there is a phenomenal component to that gestalt which is experienced as relating to all other components as superset to subsets, or as object to object’s components. The gestalt has a single “quality” which results from, but is not equivalent to, the interrelationships and interactions of its components. These two properties are, I claim, the determining characteristics of phenomenal wholes or gestalts.[83]

It is perfectly true that we see components in our visual experiences, for example, and that those components can be listed separately, combined as separate elements, and so forth. Yet when they are experienced reasonably closely together, spatially, temporally, or functionally, even if they do not coincide, they effectively commingle and influence one another: they create not merely figures, i.e., strong and local interrelationships between the elements which give rise to new elements, but a context, i.e., an environment in which the elements relate both to the whole formed from all of the elements and to co-present elements which have not joined the figure. I will argue below that the holism of gestalts may be in part derived from, and in part equivalent to, more fundamental parameters of consciousness.

What we will find, in Gurwitsch, and as what I might term a “first-order” analysis of the structure of consciousness, is something like the above - what is usually termed a “searchlight” configuration[84]. We find a central “core” or “theme”, usually with perceptual or conceptual content, a “surrounding” field (or “thematic field”) of associated and implied meaning for that core, and a “background” of meanings, relations, and sensations[85]. This configuration is easily justifiable by employing Husserl’s analysis of consciousness; and Gurwitsch employs just that analysis as a major aspect of his own justification, with support from gestalt studies. However, as I have repeatedly said, I cannot employ Husserl to justify either my use of gestalts or the above “searchlight” configuration. However brilliant a psychologist Husserl may have been (and indeed I consider him Freud’s peer), he was only one person, and only performed, at best, single-subject experiments (i.e., on himself). There is, it is true, a great deal of literature (e.g., Ebbinghaus, 1987) justifying this type of experimentation, but nonetheless single-person studies are ultimately insufficient to justify anything but more studies, as did Ebbinghaus’, and the uncertain and even contradictory results of classical phenomenology do not lend themselves to inspire reliance on no more than Husserl’s efforts. Since Gurwitsch draws much of his support directly from Husserl, I must also look elsewhere than Gurwitsch. Where do we turn, then, but to experimental and cognitive psychology?

But first I must make clear what it is I am looking for. The existence and structure of gestalts and their interdependent holistic character are clearly supported in the literature, and I have cited some small amount of that literature already. So I will not further support the existence[86] of gestalts. Their very general structural characteristics of figure and ground, and their holism are properties for which I have cited some evidence, but which need, I believe, further support and much more detailed explication. In addition, justification for the focused, searchlight type of structure described in Husserl and Gurwitsch, and second, evidence of the pervasiveness of gestalts are both necessary. It is all very well to claim that sensory phenomena are gestalts, and I will bring more support for the virtual universality of that claim, but consciousness in the sense I am concerned with, i.e., experienced phenomena, does not consist only of sensations. What of cognition, of language and symbolic thinking in general? In what sense, if any, does language have gestalt structure? If we cannot say that language (as experienced) consists of gestalts, and in what sense it does, then we cannot claim that consciousness generally possesses gestalt structure, since so much of what we experience is linguistic: internalized speech, the written word, spoken or heard utterances.

Let us make two rather simple assumptions. First, the neurological system giving rise to mind tends to unite and to abstract, i.e., to combine components[87] into other components which tend to lose specific characteristics and to preserve those common[88] over the total set. Second, that system is conservative, and tends to integrate new components into pre-existing structures and to preserve those structures. Given those assumptions, what would one expect if some anomalous thought, perception, object, etc., occurred simultaneously with some phenomenon with which it did not fit? The answer to that question provides us with what is, in effect, the most “top-down” or abstract statement of gestalt laws: a) it would be ignored, or b) it would be experienced as another, different structure, and perhaps felt as strange, startling, or discordant, or c) it would cause a radical alteration in the gestalt in order to be integrated with that gestalt[89]. But this is in fact what we do experience. We do not experience an anomaly being accepted with none of the above consequences, except in the case of children, who are still developing gestalt structures and boundaries. These results are exactly what we would expect if our phenomenal experience consisted of unified wholes, and just what we would not expect if it consisted simply of concatenations of components.

So I will proceed, then, with my own justification of the gestalt as the structural unit of phenomenological experience and investigation; and while that justification will draw on Gurwitsch, it will neither be dependent on his analyses nor will it exactly duplicate his conclusions. In order to justify using gestalts, however, I must first present what I consider to be the fundamental structural parameters of consciousness. We will then see that the gestalt conception fits nicely with this analysis, and the results of that analysis justify what I will accept and what I will reject in Gurwitsch and other phenomenologists. Thus, in the next Chapter I will present and justify those parameters. I will begin with very general considerations relating to the phenomenological approach, mainly as presented by Gurwitsch. However, I will soon afterwards, and for the remainder of the Chapter, employ specific examples from modern empirical findings to explicate and justify the parameters of my model.

 


Chapter two

 

The Structural Analysis of Consciousness

 

I. General Introduction

 

This chapter will serve as an in-depth introduction and explication of the parameters or dimensions of my model for phenomenal consciousness. It will be written in the same manner that I am modeling consciousness, i.e., top-down. That is, I will start by making very general claims and rather rough explanations of what I consider the four parameters present in all conscious experiences to be. There is no doubt that at first these explications will be incomplete; this may be frustrating at first, but I do not feel that I can begin as a formal system begins, with complete and analytic definitions, first because I am not intending to create a formal system, and second because understanding the parameters I will be employing requires a fairly thorough grounding in both empirical and theoretical ideas from several areas of study. As I proceed, those ideas will hopefully become both clear and organized.

In this section I will provide a general background to the issue of a structure for consciousness, and an attempt to indicate that it is reasonable at least to contemplate this structure. In the next section, I will introduce the dimensions that I will employ to model and to describe phenomenal consciousness. That introduction will be general and rough, and in addition I will include some general arguments against my position, in an effort to provide something like an “antithesis” through which an understanding of the limits of what I am attempting may be contemplated as I proceed. I will then proceed fairly systematically through historical and contemporary data and arguments for each parameter in turn.

 

I am taking the position that it is not enough to have an “intuition” that consciousness has a focus; that it is not enough to accept the claims of Husserl, Gurwitsch, James, and a few others that this is the case. Indeed, even if one were to take those claims as evidence, how strong would that evidence be? Merely the first-person case-studies of a few people. More generally, I wish to go beyond the statements of a few phenomenologists, and beyond what might very well be termed “folk philosophy” to determine whether there is actual evidence, from controlled, replicated, well-performed empirical studies, that phenomenal consciousness has a particular type of focal structure. Gurwitsch’s idea of the noema (e.g., Gurwitsch, 1964, pp. 176-181), i.e., the focal area of consciousness (“the perceived as such”, p. 176; “the meaning”, p. 176), can serve as a reasonable beginning for our investigation. But it must be understood that I will take “noema” in a very particular sense, one which only serves to indicate that consciousness has a general structure, consisting in part of a “central” area, so conceived because of the analogy between the brightness of a light or the focal area of a lens and the focus of our experiences. I will therefore reserve the term “noema” as a technical term referring to Gurwitsch’s[90] conception of consciousness. Although I have argued that this conception is incorrect in its ontological and much of its epistemological inferences concerning the noema, he has correctly intuited the structural aspect of consciousness to which it refers.

This central focus of consciousness consists of our clearest, our most salient, focused, cohesive and intense[91] experiences. The analogy to the brightest area under a light, or to the clearest area under a lens, is what gives rise to both the “searchlight” and the “central” metaphors, and these have been employed for describing both consciousness and attention. Thus, metaphorical understandings[92] of the term “attention” (e.g., Arvidson, 1996; Hardcastle, 1998; Fernandez-Duque and Johnson, 1999), have led to a variety of investigations of this and related phenomena. Fernandez-Duque and Johnson, 1999 (e.g., p. 1) describe several types of metaphors: the “spotlight”,[93] the “filter”, and the “zoom lens”, a refinement of the spotlight. In addition, they speak of the relation of attention to a “premotor action” metaphor. Treisman sometimes describes attention as a “window” (e.g., Treisman, 1993, p.22). Hardcastle writes of the “bottleneck” metaphor (Hardcastle, 1998) and also of the “motor decision squeeze” problem, where attention is limited in content because conflicting actions cannot be simultaneously enacted[94]. Arvidson describes the “spotlight”, “beam”, and “illumination” (Arvidson, 1996) metaphors, contrasting them with processes he terms “singling-out” and “restructuration”. Since there is no literal searchlight in our brains (or minds), nor a lens, it is easy to point out inadequacies of this metaphor, and both Gurwitsch and Arvidson do so. Thus, Gurwitsch states that it misleads us by implying that there is “a beam of light being cast upon a certain content while a chaotic confusion of other contents fills the regions of shadow and darkness” (Gurwitsch, 1966, p. 202). I must confess that in all my reading I have not found anyone claiming the existence of anything like a beam of light which casts shadows in the mind, even as metaphor, with the possible exception of certain passages in William James. One can find statements such as the following: “Information inside the spotlight is processed more quickly or more efficiently, whereas outside the spotlight, information is processed ‘less, or differently, or not at all’” (Brefczynski and DeYoe, 1999, p. 370); but that is a very different claim. Arvidson makes the point that the spotlight metaphor tends to imply that certain experienced contents are “emphasized” and others “deemphasized” (Arvidson, 1996, p. 79), but we shall see that there is empirical evidence to support this implication[95]. His major objection to this metaphor, however, is that it does not account fully enough for organization in terms of relevance, and I agree. But all metaphors are inadequate in some respects, and while it is useful to point these out, it is not necessarily useful to therefore discard the metaphor. I believe that the focus metaphor is a very valuable one, and I will continue to employ it, with some caveats.

Metaphors[96], while useful, are, without specific supporting data, merely intuitions aiding exploration. What can we conclude from the metaphors above? First, we can see that anything like a precise definition[97] of consciousness, attention, focus, or salience is probably unrealizable, from this source, at least. Second, we find two common threads. The “lens” or “spotlight” description is perhaps most similar to the model I will present[98]. It has to do with the reduction in number and/or scope of the contents of consciousness, and an accompanying increase in clarity, intensity, and number of the remaining contents. But generally, the filtering metaphors express the same phenomenology as the spotlight metaphors. That is, whereas the lens metaphor implies that “unlit” content is not discarded, merely dimmed in some way, and the filtering metaphor might imply that filtered content is eliminated, one’s conscious experiences, conceived either way, consist of a primary “area” of focus, clarity, or detail, with a “surround” of less detailed, clear, or focused content[99]. The motor conflict explanation also results in similar conscious experiences and unconscious processes similar, I would expect, to those of the filtering metaphor. It is the processes through which we arrive at the same result, i.e., the clarity of the experience of fewer contents, that are basically different, and it is those processes that the various metaphors aid in explicating.

Phenomenologically, however, we can extract at least one very important datum from these metaphors: that there is, normally, one focus for consciousness[100]. All these metaphors share that characteristic, and this provides phenomenological evidence[101] for the blending of what I will term, below, “salient” intensity and “focusing” intensity. That blending supports a unified, gestalt-type model, and indicates that although there are different neural processes realizing salience and focusing (e.g., see Maddox, et al., 2002, and below), the intensity resulting from those processes is ultimately, except in very special circumstances, experienced as one phenomenon.

In addition, of course, the usefulness, indeed, the necessity of metaphors lies in their implications for designing experiments to investigate the processes underlying consciousness. Thus, while it is essential to be aware of them and their implications for theory and experiment, one must attempt to more directly tease out data concerning attention and consciousness from individual studies. We will find this data, among other places, in the multitude of studies of the phenomenon of “attention”. The literature in this field is enormous, and I cannot possibly treat it in detail, but it is to that literature that I will initially turn for support of my model.

Before I proceed, however, I feel that I must re-emphasize that the point and goal of this dissertation is first, to describe how to describe, and second, to describe, phenomena, i.e., conscious experiences. It is of course well known that there are visual and other processes that are unconscious[102]. Consequently, and confusingly, in the wide variety of empirical literature involving gestalts, features, grouping, attention, and so forth, there is no fine division made concerning where experience ends and unconscious processes begin. Vision researchers, for example, speak of features as elements that the visual system[103] recognizes and/or processes, and commonly conflate those with features in the phenomenal sense, i.e., as components of visual experiences. A feature, for this literature, might be experienced, or it might be an aspect of an unconscious process conveniently labeled “feature” because it functions as one, supports the production of experienced features, or for some other reason. This bears on the important issue of just what a whole is; and my answer is unambiguously, emphatically, and straightforwardly: in this essay, an experienced whole. I will do my best to clear these ambiguities up as I proceed.

In addition, I will be concerned with the will, i.e., with volition. Many of the studies of vision and other systems deal, experimentally at least, with volition, in that subjects must be instructed as to what to look for, where to look, and so forth. We will find even more ambiguity, perhaps, in dealing with volition in the experimental literature than with consciousness per se. Yet I will argue that volition is quite central to the structure of consciousness. When I use the term “volitional”, I am referring to acts directly controlled by the conscious will. If we “volitionally” alter a figure-ground relationship by changing context, i.e., by imagining ourselves in a different situation in order to see something differently, what is actually being volitionally changed, by my criteria, is that context, and not the resulting figure-ground alteration. We have taken advantage of something we have noticed about our perceptual/cognitive interactions in order to change, indirectly, our perceptions by altering our cognitions. However, if we alter a figure-ground relationship by concentrating our gaze on what was the ground in order, by focusing our attention on it, to turn it into the figure, then we have volitionally changed that perceptual relationship. Separating these kinds of effects requires not only carefully analyzing the experimental structures, but paying close attention to the instructions given to subjects. One reason I emphasize this difference, of course, is that I am primarily concerned with the phenomenology of volition as well as its functions.

 

II. The Structural Analysis of Consciousness: Introduction to the Parameters

 

I will model consciousness, i.e., phenomenal awareness, from the top down. That is, instead of starting with specific contents or even types of contents, a mode of analysis which tends to give rise to evaluations such as James’ “blooming buzzing confusion”, I will start with very general characteristics of phenomenal consciousness derived from the metaphors described above, from empirical data, and from intuitions similar to those of Gurwitsch, and endeavor to show their virtual universality and their usefulness in structuring specific phenomenal contents. The property, or alternatively, the dimension or parameter, with which I will start I term “intensity”. For reasons that will become clear, I will not employ the term “attention for this property. This dimension in turn consists of two sub-divisions: 1) “focus”, roughly corresponding to the degree one consciously, i.e., willfully, through acts of volition[104], attends to phenomena, and 2) “salience”, roughly corresponding to the degree to which phenomena are non-volitionally experienced as intense, cohesive, contrasted, and/or enhanced[105]. All phenomena of which we are aware, I am claiming, are structured to various degrees, most generally, by these parameters, and that structuring extends throughout phenomenal awareness. By “structured” I mean not only that all phenomena have some degree of focused and salient intensity, but that the degree of intensity a phenomenon possesses alters it as an experience. This property, intensity, as a synthesis of focus and salience, aside from its being characteristic[106] of all conscious phenomena, is such that consciousness, as a whole field, is structured by it[107]. This is the sense in which I am modeling it as a universal parameter, property, or dimension of consciousness. To incontrovertibly demonstrate that intensity is absolutely phenomenally universal is, I believe, impossible. I will argue, however, that a very good case can be made for that universality, based on both theoretical and empirical grounds.

As I proceed, I will argue that properties of conscious phenomena may in addition be elaborated in the following ways: phenomena are recursively structured, i.e., individual components or features of experiences are internally organized by the parameter of intensity[108], and they are “directional”, i.e., the internal focus and salience relations of phenomena have properties similar to but not identical with intentionality[109]. We thus have a total of four parameters or dimensions[110] that I claim are universal to all conscious experiences: salience, focus, recursion[111], and directionality. I will elaborate the meanings of and evidence for these parameters extensively as I proceed. Later, we shall see how these general parameters of consciousness, through these latter relations, can give rise to the holistic properties of both sensory and higher-level (“cognitive”) gestalts. I should note here that this claim about intensity, salience and focus is a departure from Gurwitsch’s characterization of consciousness. He speaks of consciousness as having the dimensions of theme and thematic field[112]. I will show, later, that these need not be considered dimensions in any fundamental sense; indeed, that they are consequences of the parameters above.

These are of course strong claims, and to start, I will counter them with several disclaimers. First, as with any, even partly empirically-supported theory, this must be considered a model. It is quite possible that there are other, even more fundamental ways to structure consciousness; for example, one might claim that cohesion, which I consider, roughly, to be one cause of salience, and also a characteristic of salient phenomena, is a fundamental parameter[113]. Both salience and cohesion, as categories of relatedness, may be directly experienced, but I think that the experience of salience is much easier to support empirically. Second, it is also possible that the parameters I am choosing might be characterized such that they could split into several others. For example, I consider “directionality”[114] a property of cohesiveness, and thus of salience. However, it might be possible to employ that and something like “relatedness” as two parameters instead of the single property of salience. I will however argue against this option.

Additionally, it has been proposed (e.g., Norman, 2002) that the property of focus is in fact the experience of two separate processes: that of focusing attention and that of holding it on the phenomena selected. I believe that this may be explained as combinations of focus and salience, but again, it is virtually certain that focusing is realized and mediated by different neural processes than salience. However, in virtually all circumstances, I will present evidence that salience and focus, as I will characterize them, are not independent, and thus not separate dimensions in either the mathematical or logical sense of that term; and that data and theory in the areas of dichotic listening and visual masking demonstrate their phenomenological interdependence.

Throughout, I will be claiming that salience and focus, while separable in extreme circumstances, are normally phenomenally united into the dimension of intensity. However, there is evidence that processes resulting in salience and those resulting in focus are not merely different neurally, but may in some instances function largely independently. Thus, Maddox has developed an experimental setting in which these must function in opposite directions in order to perform accurate discriminations (Maddox, et al., 2002). He has found that in such situations, salience and focus can be made to function separately. There is then neurological evidence (p. 325) and cognitive evidence (e.g., pp. 336-337, for summary) that these two systems can be, if not completely separated, very nearly so, in special circumstances. Thus, a model employing an overarching phenomenological parameter of intensity may be an approximation. It may very well be that the experience of salient intensity is a different phenomenon than that of focused intensity, aside from specific contents of such experiences, such as bright salient colors versus important, but unobtrusive, features upon which we focus. But I do not believe this is true, first, because of the difficulty of separating these aspects of intensity, and second, because in the metaphors above, all describe consciousness in terms of one focal center and not two, as I have mentioned.

One might also question the above as follows: is the property of intensity, the fusion, as I will argue, of focus and salience, truly phenomenal? We might claim, given the variety of metaphorical understandings of what is termed “attention”, that the above parameters are in fact merely metaphorical, i.e., abstractions experiences, or inferences about unconscious attributes. Of course it is inescapable that they are, to some extent. Yet the fact that we are directly aware of our efforts to focus, and of the results of those efforts as results of focusing; and that we are, similarly, directly aware of increases and decreases of salience, e.g., in visual “pop-outs”, yet unaware that we have brought them about, seems to argue for a more literal understanding of these phenomena. We might consider intensity as relating to the act of willing and its effect on experiencing in general, in this manner: whatever it is we will is, by that act, enhanced in consciousness (accompanied by the recession of other contents), and we also note similar enhancements and recessions when we do not will. That experience of enhancement and its opposite is the actual parameter I am describing, divided into classes which are the results, or not, of the will, termed, respectively, focus and salience. Alternatively, we might consider willing, per se, as the parameter, and its most general effects on experiencing, starting, most generally, with the phenomena of enhancement and recession, to be the basis of this model. The problem I have with this latter formulation is that there are undoubtedly unwilled alterations in intensity, either externally or internally caused, and given that, I cannot claim that either focused intensity or willing is universal to all contents of consciousness. What we will find after all the dust has settled over the data is that processes giving rise to salience relations create a fairly detailed picture of the world, very quickly, but an incomplete one nonetheless, and one which is very inflexible. Focusing processes complete this picture by adding structure, detail, and by providing great flexibility in altering our experiences[115]. Further, we will find that volitional processes are not all or nothing. Not merely the resulting enhancement, but the degree of one’s exercise of the will, i.e., the degree to which one consciously directs one’s attention, is, perhaps surprisingly, quite variable: a continuum, in normal phenomenal experience, as we shall see. Similarly, the degree of non-volitional alteration of enhancement of intensity or contrast is also highly variable. Thus, focusing and salience can both be modeled as interacting continua within the dimension of intensity.

More generally, why pick “fundamental” parameters at all? Why not just list all possible parameters, and have various phenomena dimensioned, so to speak, along whatever seems to work in some particular context? Aside from the rather arbitrary approach and the disorganized results implied by this kind of analysis, there is strong evidence that on a very basic neurological level, processes resulting in salience, viz., processes which result in the creation of a variety of groupings, relationships, and abstractions, are virtually ubiquitous in the central nervous system. These processes normally result in the simultaneous creation of phenomenal structures cohesive to varying degrees, a result of characteristics of parallel distributed processing (PDP) neural networks[116]. That is, salience seems to result, for the most part, from extremes in what are termed “convergent” neural processes. These processes take multiple inputs to single - or small sets of closely-related - neurons and convert them to, usually, a lesser number of output streams which integrate those inputs in ways depending on various characteristics of that particular neural net. Convergent processes are ubiquitous in the CNS, as has been known for decades, and, for example, produce neural responses to visual abstractions ranging from line angles to complex objects. Salience, then, in part results from one or more of several possible extremes in these processes involving dimensions such as brightness, color, or shape, implying relationships between objects on these and similarly functioning properties[117]. These structures, as a result of that varied cohesion and their simultaneous juxtaposition and inter-relatedness against differentially-cohesive structures, will vary in salience. There is also evidence that focus, as I will describe it below, is a function of qualitatively different activation patterns of the central nervous system than is salience.

In summary, a combination of temporal coincidence in neural firing and directed activation by the complex termed the “Extended Reticular Activating System” (“ERTAS”, e.g., Baars, 1993a), and/or by certain patterns of activation in the frontal and dorsal parietal lobes seem responsible for conscious focusing; and activation in the ventral parietal lobes and other locations seems responsible for salience (e.g., see Corbetta and Shulman, 2002, pp. 202-206 for neural activity related to focusing; and pp. 208-211 for neural activities related to salience). Indeed, the overlap, functionally and phenomenologically, between salience and focus is interesting in this context. In addition, since I can and will employ these parameters, as aspects of intensity, to explain characteristics which Gurwitsch and others have considered fundamental, I believe this indicates that I have made a good stab at finding such basics.

Why pick the kind of parameters I am employing? We find many different analyses of consciousness, and usually the characteristics termed “dimensions” are structures such as Freud’s “ego”. But these are contents, or groups of contents, of consciousness, present only in some subset(s) of phenomena. And indeed, it must be possible to partition the contents of consciousness into virtually any number and type of categories. However, I am attempting a more general, top-down analysis, one which supercedes any specific content or even category of content, in that this parameter, intensity, is experienced, to varying degrees, as a property of all phenomenal contents. While it may arise from extremely complex underlying processes and result in highly complex experiences, this property, phenomenologically, is itself a relatively simple class of experiences, structuring consciousness, present at all levels. Its aspects, focus and salience, are also recognized as fundamental. Thus, Nothdurft describes visual salience as follows:

 

Salience can be looked at as a property on its own. The salience of a target affects its visibility and facilitates its detection irrespective of most of its specific features. Salience is not necessarily associated with the feature properties of the target itself but is strongly related to the way it is embedded in visual context (Nothdurft, 2000, p. 3198).

 

The “pop-out” or “stand-out” phenomena (e.g., see Nothdurft, below) that I will describe are, I claim, fairly direct manifestations of our apprehension of salience, or as Fechner put it, of “vividness”. Further, the tip-of-tongue phenomenon, which I will analyze extensively, will, I claim, demonstrate that salience is directly apprehended on a cognitive level through various feelings, including those of knowing and familiarity[118]. Focusing, on the other hand, as the other aspect of intensity, needs little justification as a “property on its own”; we are necessarily cognizant of the volitional aspects of experience.

To summarize, data from enormous numbers of empirical studies support the existence and nature of these parameters. Those studies will establish, first, that they are in fact consciously experienced. Second, they will provide evidence that they are intrinsic aspects of sensory (and other[119]) functioning. Third, they will support the claim that they are the results of general properties of our nervous system. Since these properties are both functional and phenomenological, they illustrate one of the points I am attempting to make in this essay, that phenomenological data can deepen analyses from several areas of cognitive science.

 

III. Some Introductory Examples and Analyses

 

Now that I have set the stage with various claims and disclaimers, it is time to consider more precise explications of those terms. In order to describe the parameter of intensity, I will describe its two overlapping aspects, focus and salience, separately. As I do so, it will become clear, I believe, that they are nearly always blended in the experience of intensity[120].

Before I cite modern data supporting this model, I would like to mention some of the early studies of attention, in order to note that even in the beginnings of investigations, the focal model, the parameter of intensity, and the interaction of intensity and volition were employed to describe the experiences of attending and attention. These parameters were found useful as early as the mid-19th century. Geissler, for example, in his review of the history of theories of attention[121], mentions the work of several of the earliest figures in that field. We find statements such as the following:

 

He [Lotze, in 1852] seems to have been the first to introduce… the analogy of attention and inattention to the Blickpunkt [focal point] and Blickfeld [field of vision] of vision. He says: “the mind is not so constituted as to experience all its (simultaneous) contents with equal clearness and attention. It is rather to be compared to the retina of the eye….” (Geissler, 1909 , p. 476).[122]

 

But it is the terms “clearness” and “attention” which are most important in this context. Geissler continues,

 

Fechner [in 1860] had been led… to distinguish sharply the limen of sensible intensity from the limen as influenced by the degree of attention. As early as 1860 he wrote, for instance: “… Hence the intensity of the idea and the strength with which we think or perceive it must be in some way distinguishable from each other”…. In 1877 he was still more explicit… “If we perceive a sensory phenomenon or represent it to ourselves… the intensity of our conscious activity is then determined on the one hand by the degree of attention with which we perceive the sensory or the memory image, and on the other hand by the vividness or intensity which pertains to the phenomenon itself…. In such cases we can quite well distinguish how much is due to the degree of attention and how much intensity belongs to the phenomenon as such.” (Geissler, 1909, pp. 476-477).

 

In this quote, Fechner employs the phrase “vividness or intensity” instead of “clearness”, but the intent seems the same. He states quite clearly that conscious experience has the parameter of intensity, which is caused either volitionally (“degree of attention”) or non-volitionally (“pertains to the phenomenon itself”). I must admit that the similarity between this description and mine took me by surprise when I discovered it; it would be difficult to put it more clearly than Fechner did in 1877 (Fechner, 2001)[123].

As far as volition as a parameter – or justification for the above division - is concerned, we will find empirical distinctions, below, made between willed and non-willed acts. Those distinctions relate to the parameters of salience and focus, and that is why I am introducing discussion of volition at this particular point. That is, we will find that the salient component of intensity is in most cases almost completely nonvolitional, and that, in contrast, the focusing component is in most cases almost completely volitional. An introduction to the phenomenology of willing is thus necessary at this point.

Phenomenologically, however pervasive one considers volition and volitional states[124], it is nonetheless true that there are experienced differences between the exercise of the will as a phenomenon, its results, and other non-volitional or spontaneous phenomena. Thus, Geissler translates Wundt (see Wundt, 1889) as follows:

 

“Those strain sensations which, with the same attention, accompany the external volitional act as well as the direction of the will to the particular sense departments, form a complex of qualitatively related sensations…” (Geissler, 1909, p. 480)[125].

 

Wheeler sums up his view of the phenomenology in this manner:

 

Kinaesthetic sensations are with us always in mental life…. The extreme variations in past descriptions of the will consciousness both in its broader and narrower aspects have been due to various interpretations of a consciousness which is so largely made up of kinaesthetic sensations. From these experiences we get our notions of striving, strain, activity, force, conation and the like…. The chief cause for the great variability of these descriptions lies in a further attempt to find evidence of… some unique mental process… the unique mental process, we believe, is nothing more than kinaesthetic sensation (Wheeler, 1920, p. 359).

 

Note that Wheeler, like Geissler, is not saying that experiences of willing are literal sensations of the muscular tensions present in our body as we will, or that we need to physically strain to will something. The sensation is kinesthetic, yet not an apprehension, just as the visualization of a scene not present may be described as seeing yet not apprehending, or the recall of a melody similarly. It is “mental life” he is describing. Searle states that:

 

I do not sense the antecedent causes of my action in the form of reasons, such as beliefs and desires, as setting causally sufficient conditions for the action; and, which is another way of saying the same thing, I sense alternative courses of action [my Italics] open to me (Searle, 2000, p.2)…. One of the most common experiences in our lives is that of moving our bodies by our conscious efforts. For example, I now intentionally raise my arm, a conscious effort [my Italics] on my part (p. 5).

 

Brown describes the phenomenology of willing in terms of the “somatosensory components”, the “action structure” and the “purpose or goals” of conscious phenomena (Brown, 1989, pp. 110-115).

Mitchell and Hunt attempt a theory relating the effort of will to contemporary theories of attention. This paper may be the best in the empirical literature relating the experience of volition to a wide variety of measures of recall and performance. They relate “cognitive effort” to attentional capacity (Mitchell and Hunt, 1989), speak extensively of “mental energy” and effort (p. 339) and their relationship to memory retrieval and a variety of studies of attention. However, their conclusion, that effort engages more “resources” (p. 346), suffers from a lack of analysis of the underlying metaphor utilized. There is no explanation of resources in any physiologically meaningful terms. One is thus left with an enormous amount of data which refers either to a rather vague theoretical construct, or to a circular hypothesis. Their conclusion, then, that “cognitive effort is an important concept in attention which can be brought to bear on memory performance to describe… limiting conditions” (p. 346) does not seem warranted, at least inasmuch as that paper is concerned. But one must applaud their efforts.

I have found very few phenomenological descriptions of willing which go much beyond the above, although those examples, while representative, are not of course exhaustive. The philosophical literature on free will is in the main replete with analyses and speculations about causality, about what “freedom” means, and so forth. Fascinating and complex as these metaphysical issues are, they are simply not relevant to the aims of my investigation. Whether or not we “actually” “possess”, in some sense of those terms, free will is irrelevant to the phenomenal fact of willing or volitional mental acts being different from other acts; and fact they are, or people would not have spent the time they have in attempting to explicate their metaphysical and ethical implications. What we will find, oddly enough, is that phenomenal accounts of willing pervade the experimental literature, but that they must be teased out, so to speak.

 

How can one characterize focus and salience, as phenomena? Let us take as our first extended example the classic drawing of black splotches, which on inspection resolves itself into a Dalmatian on a background of such splotches.

 

Figure 2: Dalmatian

 

Previous to that resolution, those splotches are experienced as less interrelated toward each other, and unrelated[126] toward a hypothetical Dalmatian, than after we see a subset of them as a Dalmatian. Afterward, that latter subset is highly cohesive, in that its components are strongly interrelated toward each other, and simultaneously much more strongly differentiated than before from the other (non-Dalmatian) splotches. We not only see the Dalmatian, but we see that it is composed of splotches, and in addition, the individual splotches are now radically differently characterized, even as individuals. Thus, one might now be seen as “part of a leg”, and so forth. In addition, whereas before the Dalmatian appeared, some of the splotches comprising it may have appeared more intense than others, on the basis, perhaps, of their individual shape or location, afterwards any difference in intensity within that set of splotches is due more to a difference relative to the Dalmatian as a whole: perhaps something about its head “catches our eye”. The Dalmatian is a figure set on a ground of splotches.

Grounds will usually be characterized, we will find, by lower internal cohesion[127] and more uniformity, and thus less overall intensity, than figures. There is recent evidence that the figure-ground phenomenon may require some degree of focusing (e.g., Hollingworth and Henderson, 2002, p. 132) to be generated[128]. But that degree is usually small, in the sense that subjects must be sensitized or trained to be aware of it in experiments. Thus, the experience of figure-ground is primarily a non-volitional one, i.e., one that relates to salience, but one which nonetheless may require some focusing. Salience and focusing are unified, synthesized, then, into the experience of intensity even in the low-level phenomenon of figure-ground discrimination.

Further, we see a distinct difference between the set of splotches comprising the Dalmatian and those not, in that the latter first, are less cohesive than the former, second, are still individuals[129], and third, are now the figure’s ground. In addition, the phenomenon of “illusory figures” (see below) may now occur, on the basis of the cohesiveness of the Dalmatian. For example, we see that the Dalmatian has a tail; but there is no actual tail, except for a small space between two splotches in the appropriate area. Intensity in this case has to do with what is termed, visually, “feature binding” (e.g., Davis, 2001), and we might additionally employ something like “the closeness of a match to a template” or the degree of phenomenal interrelatedness of a set of components to explain many instances of higher intensity. Thus, if some visual phenomenon has what could be described, after Nothdurft, as “local feature contrast”, or as the “distinctness of a target from nearby non-targets” (Nothdurft, 2002, p. 1287)[130], it will be experienced, usually, as salient, and more intensely.

But we might also consider another cause, primarily of salience, which seems the opposite to the example above. Suppose that we had a field which was clear and detailed, except for one area, in which we expect to see someone’s face. Instead of a face, we see an incoherent blur. Now in looking at that picture, we will have, at some particular time, one of only two or three possible experiences. We might, without being aware of it, supply a face for that person, and experience seeing that face – an example of “filling in” (e.g., Pessoa, et al., 1998). Alternatively, our gaze might be drawn[131] to the anomaly, the blur where a face should be, and that blur, seen as such, would be the most salient aspect of the scene. But its intensity, in that case, would be in spite of the face’s lack of internal cohesion; it would be due to the anomalous nature of that lack in the overall context. Thus, it is not always necessary that a figure be more internally cohesive than a ground[132]; if its lack of cohesion is enough of an anomaly, then that very lack would become a single outstanding feature, similar to a bright color, which might cause it to pop-out. A third alternative might be a kind of blindness: we might not be aware of either the face or the blur[133]. This is something like the “change blindness” experiments (e.g., Mack and Rock, 1998; Beck, et al., 2001; Rensink, 2002), in which non-cohesive features are suppressed[134] in order to retain cohesion. It seems, then, that salience does not merely occur, but it also functions in visual phenomena. A visual feature which stands out, i.e., is more intense, for example, does so, roughly, because it lacks or alters the cohesion of some context. A red X which stands out from a field of green Xs, a shape which differs from others surrounding it, will also generally stand out.

While salience and thus intensity may be due to greater internal cohesion on the part of a subset of components in the field of consciousness, there are other factors that can compensate for that, usually having to do with strongly anomalous patterns. But even here one can find an explanation in terms of the intrinsic propensity of the nervous system to create pattern; when that process is blocked, the system attempts to compensate, first by highlighting, in effect[135], the blockage, second by attempting to create a pattern to integrate it into the overall context. That initial stage of “highlighting” not only produces salience, it causes feelings such as those described by James and others related to the tip-of-tongue experience. One might conceive of salience, then, as an experience involving the degree of internal phenomenal relatedness. These two characterizations , viz., in terms of distinctness and relative inter-relatedness, are attempts that I am making to clarify causes and contexts of this same dimension or parameter of intensity, i.e., salience, and should be taken as such. As Craik puts it, quite generally, “depth, elaboration, and congruity describe aspects of the encoding process, whereas distinctiveness describes the eventual product of these processes”(Craik, 2002, p. 307).

 

Although many examples in the visual literature use the term “salience” to mean something like the strength with which a visual feature “stands out” in relation to other features, this description may easily conflate the property I am terming “salience” with the property I term “focus”, and thus with intensity, as well. Focus, as I have said, describes the degree of phenomenal distinctiveness, clarity, emphasis, i.e., intensity, which is volitional. This is a complex idea, because we have on the one hand the process of focusing or attending, and on the other we have its result: the increased intensity and richness[136] with which we apprehend or understand something on which we focus. Certainly, the aim of focusing cannot, generally, simply be that act itself; it must in part be useful because of the resulting details we grasp. Thus including both increased components and intensity as implications of the parameter of focusing, although adding to the complexity and difficulty of the analysis, is, I believe, essential. It is clear that one may, even phenomenologically, easily conflate the two properties of focus and salience, simply because a salient apprehension tends to draw one’s attention, i.e., one tends to try to focus on it. But it should be remembered that one can direct one’s attention from as well as to such a phenomenon.

Thus, one sees a picture of a cow in a field wearing a hat, and one’s attention is immediately drawn to the hat. A cow wearing a hat is very odd, for many reasons, and the hat, in that context (on a cow in a field), is extremely incongruous. It is of course that incongruity which, at least in part, causes it to “stand out”. Note that the hat stands out not because of some low-level visual phenomenon – it is not, let us say, wildly differently colored or shaped than either the cow or the rocks in the background – but because of cognitive properties, i.e., what we know about hats and cows. Nonetheless, surely we will all admit that it does, spontaneously and vividly, stand out in that context. Now, on the one hand, one might talk about the degree to which that hat spontaneously stands out in the general context as related to its phenomenal salience (i.e., very high relative to the cow), and on the other hand, one might talk about the degree to which we have deliberately fixed or directed our attention on that hat (and its increased richness, etc.) as its phenomenal focus. It is possible that the hat might have low intensity, i.e., both low salience and low focus, as in a context when, after seeing many pictures of cows with hats, our interest in the hat, our attention paid to it, is low, and we are, in addition, looking around the picture for other interesting content. In contrast, if in the picture the field in which the cow is standing were filled with hats, we might, after having difficulty picking out the one on the cow’s head, break out in laughter and focus our entire attention on that particular hat. It would have high intensity due to our focusing on it, rather than to its salience. Thus, in sum, focus relates to both the effort we might expend in focusing, concentrating, or attending on something and to the resulting intensity; whereas salience is a more passive property of phenomena. We can see from the above just how complex is the phenomenon of intensity as the combination of focus and salience; in similar contexts one or another may be the predominant component of the cow’s intensity.

Examples illustrating the differences between, and the interdependence of, focus and salience can be found in the visual literature. Nothdurft, for example, studied the difference between focus and salience as reflected in search times between “targets with high feature contrast” vs. those “non-salient” targets that do not draw one’s attention (Nothdurft, 2002). He finds that “search for non-salient targets that do not attract focal attention is slower” (p. 1287). Here we see a difference between salience and focus. A salient visual feature is one that is involuntarily intense, that attracts attention, and that stands out from nearby visual features. When we are asked to focus on a non-salient visual feature, a salient one might distract us; in other words, it is still salient, still intense. But if we focus sufficiently even on a non-salient feature, we are capable of “overriding” or “blocking out”, so to speak, the non-volitional intensity of the salient one and of only seeing the one on which we are focusing. And so there is a sense in which focus and salience are always interacting. Is the experience of intensity the same when our attention is drawn by a salient feature and when we deliberately focus on a non-salient one? That is, is focal intensity the same experience as salient intensity? Because of these interactions, because focus can either inhibit or enhance salience, and because of the phenomenological evidence of one center of consciousness from all the metaphors of which I am aware[137], I am arguing that it is, i.e., that the dimension of intensity is the same, whether salient or focused.

Such examples usually relate to low-level visual phenomena, viz., pop-ups and the like, and “top-down” interactions. Another example of what might be termed “cognitive salience” is the following: we can understand a car from several perspectives. One of these is as a means of transportation. When we do this, the car is grasped, and apprehended, in terms of functions and appearances relating to it as a transportation device: the steering wheel is literally seen as a component which turns the car, for example. But if we focus on understanding the car as a machine comprised of a variety of materials, i.e., as metals, glass, plastics, and so forth, we now both grasp and apprehend the body of the car primarily as a shape of metal, and the steering wheel as a circle of plastic. It is of course still a steering wheel, and we still understand that it steers the car, but its material composition comes to the forefront of that gestalt. Those general aspects of both the car and the steering wheel have increased in focus as a result of our altered focus on the car. Now, that increase in focus, our conscious refocusing on the composition of the car, has, so to speak, dragged certain connotations from within the depths of the previous gestalt into the forefront, and the whole must restructure itself accordingly: the saliences of its components also change. For example, we now may apprehend the steering wheel primarily as a shape that moves in space in some particular way to direct the car or primarily as a torus comprised of a lightweight substance. That latter change has been brought about by the former, and in that sense focus and salience are not independent.

It is easy enough to think of similar examples where a change in salience would produce a similar change in focus: if the car were hit with a sledgehammer we would become suddenly and dramatically aware of, and interested in, its metallic composition. Thus, focus and salience are simultaneously what might be termed independent and dependent variables: they may vary independently initially, but a change in one ultimately – and rapidly - results in a change in the other. When would that not be the case? It seems that we can bring something like a separation about in very artificial situations such as Maddox et al., constructs in their study (Maddox, et al., 2002), where “perceptual” and “decisional” (e.g., p. 328) aspects of attention are contrasted through mixing relevant and irrelevant cues in visual discrimination. They had subjects make discriminations where spatial cues and some intrinsic stimulus property (line length) were either relevant or irrelevant to each other. They found that decisions made about a property after seeing the stimulus were affected by its relationship to the spatial cues, depending on whether the spatiality of the property was relevant or not, but that even when the property, the line length, was in an uncued location, judgments could be made about it. So focused, volitional decisions were affected by salience processes differently, depending on post-presentation instructions, but were independent to the extent that salience did not totally govern the ability to make volitional discriminations (e.g., p. 336).

 

IV. The Empirical Basis of Intensity

 

As Posner makes clear, the long (at least 150-year) history of attention renders the field difficult to synthesize, in part because “attention” was employed in studies of several disparate categories of phenomena (Posner, 1982). The term has referred to the performance of tasks and their mutual interference (pp. 170-172), to the subjective experience of attending and its limited capacity (pp. 172-175), and to the neural substrate and its functions as they relate to the former categories (pp. 175-178). Phenomena such as the mutual interference of tasks and the phenomenology of attending, especially as they relate to the capacity[138] of that function, are directly related to what I have been terming “focus”. Thus, empirical investigations of focus have employed the term “attention”, a word which is easily ambiguous as to whether it refers to what might be termed “mere” behavior, e.g., orienting responses, or to phenomenal experiences; and even if it references the latter, it is still ambiguous as to whether it refers to voluntary, controlled, conscious processes or to involuntary processes. In addition, we find studies of a “general warning signal” (Moray, 1959, p. 60), of “input attention” (Johnston, et al., 1995, p. 366), or of the “stimulus-driven… attentional capture” (Theeuwes and Godijn, 2002, p. 764) of attention, e.g., the pop-out phenomenon that I have previously mentioned, which concern what I have termed “salience”. Thus, both aspects of the dimension of intensity are directly related to the field of attention as it has been studied for over a century.

The idea that one can focus one’s attention seems obvious; we do this easily and intuitively. However, aside from the phenomenological fact of this process and the general experience of one thus “shifting” one’s attention to another phenomenon, the question of what is occurring both functionally and experientially has not been precisely and systematically addressed by phenomenologists. Interestingly enough, it has been cognitive scientists who have extensively and methodically pursued what is to a great extent a phenomenological investigation of a variety of questions about conscious and unconscious, willed and unwilled, attention.

We may take for granted that given the climate in academia in the 60s, when behaviorism was just beginning to be challenged, consciousness was still something to be approached extremely gingerly, and that “attention” substituted to some extent for “consciousness” through the 60s and 70s. That is, in this literature, terms such as “attention”, “focusing our attention”, “attentional focus”, “attending to”, and so forth, remain throughout ambiguous as to whether they refer to consciousness. Attention, or attending, is in many instances distinguished from focus, or focusing, as I employ the term, even in contemporary experimental literature on attention, and thus may not in fact always refer to processes which are conscious, much less those which are volitional. We still today find passages such as the following:

 

Distributed attention refers to a condition in which the subject’s attention covers the entire visual field, processing all stimuli in parallel. In contrast, when the characteristics of the task demand the suspension of well-known behavioural routines and a high level of information processing, it is necessary to concentrate the available attentional resources on a circumscribed area of the visual field and process all the selected stimuli in a serial mode. This type of attention is called ‘focused’. (Maringelli and Umilta, 1998, p. 226).

 

I assume the above is deliberately opaque as to whether it refers to consciousness. We have, to take but one example, the visual experience of objects in space, and the aural experience of different sounds which are to some extent experienced spatially. When we focus on some sensory object, are we to any extent doing so “preattentively”, i.e., before we are conscious of it? Are we focusing on the object per se, or on its location in space, and are we conscious, or equally conscious, of both those properties? Similar questions might be asked of many properties of uni- or multi-modal sensory objects. Further, when we focus on concepts, e.g., when we think “in words”, is that focusing related to conceptual objects or to the spatial and temporal progression of the thought?

Systematic empirical investigations of attention started in the 19th Century, as we have seen. Very early in the twentieth century (e.g., Titchener, 1901; Geissler, 1909; Dallenbach, 1913), most investigations were conducted under Titchener’s introspective protocols. Then behaviorism swung the pendulum in the opposite direction. After the decline of introspectionism, World War II stimulated strong interest in signal detection and coding (see e.g., Posner, 1982, p. 169). In the ensuing years, the combination of behaviorism and the signal detection paradigms virtually eliminated anything like introspective studies of attention, much less of consciousness. One finds a virtual absence of references in this area until the middle 60s, and the field did not really gain momentum until the 70s. In the 60s and 70s investigations were conducted employing the newly-created signal detection/information processing paradigm, which is largely unchanged - although greatly elaborated - today. Thus, Shepard, Neisser, Estes, Posner, Garner, and Treisman, for example, were some of the first to treat the mind as an information-processing device (Shepard, 1964; Neisser, 1967; Estes, 1972; Posner and Boies, 1972; Garner, 1974; Treisman, 1977).

Posner starts his investigations of attention and consciousness [139] by attempting to find data to support what he terms three “types” of attention: alertness, selectivity, and a limited processing capacity (Posner and Boies, 1972, pp. 391-392). The first is an interesting characteristic, in that since one may speak of a “general” or “heightened” alertness, that might be considered another parameter or dimension of attention than focusing or salience. Yet on consideration, I believe that since one may be alerted to sensations and/or characteristics ranging from very specific to very general, the term “alertness” is in fact a synonym for focusing. One can easily focus on their general environment, or on sounds, or on some specific sound, and so forth, and this kind of alertness is almost always volitional. Thus, Posner sets up experiments in which he “warns” subjects about impending stimuli, and asks, “Can the warning function and the selective function… be separated?” (p. 393). The “selective function” is “either a tuning process which blocks input from unselected sources… or a general alerting function which enhances input” (p. 392). Selectivity, then, mixes what is later termed “salience” with what I am terming “focusing”. And Posner concludes that

 

Attention in the sense of central processing capacity is related to mental operations of which we are conscious, such as rehearsing or choosing a response, but is not related to the contact between the input and long-term memory that leads to [for example] the letter name (p. 407).

 

These early results are reasonably consistent with my general claims. We are conscious of focusing but not of the assignment of names to familiar objects, at least so far there are no obvious problems. I would also like to point out that the internal structuring and cohesion of letters is a very rapid and non-volitional process. Posner states that for the letters he presented to subjects, “the encoding function appeared to be somewhat steeper over the first 150 msec” (p. 406), implying that within that interval of about 1/10 second letters have been recognized[140] by our visual system.

Treisman’s fascinating paper (Treisman and Gelade, 1980) presents us with one of the first theories claiming that “focused attention” (e.g., p. 98)[141] is required for what I am terming cohesion. Yet she is unable to make this claim unambiguously, since it is manifestly untrue, as she herself admits (p. 99). She qualifies it by saying that focused attention is necessary for “correct perception… although unattended features are also conjoined prior to conscious perception” (p. 98). Now, however, we have a position considerably more difficult to clarify. In these early experiments, we find that letters are counted as single-feature items, where the feature is “shape”, and that color is another feature. Yet it is clear that letters are complexes of simpler features. Overall, her series of nine experiments supports some degree of non-volitional processing, but since there is, first, little control over the complexity of features within a particular dimension (e.g., shape), and second, no fine control over what aspects of “focusing” are volitional[142], it is very difficult to say to what degree the “perception” she measures is volitional and to what degree it is non-volitional. Treisman and Gelade’s is one of the first studies, however, which indicates that some aspects of one’s sensory environment are seen “in parallel”, i.e., more-or-less simultaneously, and given internal structure, to some extent, during that process; while others are seen “serially”, i.e., a few at a time, and that further structure is added during that latter process (e.g., see pp. 132-133). Thus letters, despite being comprised of multiple features, are seen simultaneously, while letter/color complexes are processed serially. There are other possible confounders of processing temporality in these studies, for which later studies have to a great extent compensated. For example, the early studies could not discriminate between the time it takes for communication between neural modules or systems responsible for the different “dimensions” of a sensation (which could result in the some of the delays after the initial parallel processing within a dimension). In addition, the time between successive saccades to different areas of the visual field, which could result in delays, is also not taken into account. Neither of these latter delays would be volitional. Thus, these experiments, attempting to demonstrate, through delay measurement, that volitional processes are necessary for attention are insufficiently rigorous.

I am spending time on these studies because first, they illustrate the difficulties involved with delineating aspects of attention, and second, to illustrate that when attention is studied, the properties on which I am basing my model fall out of the experiments fairly easily, even though precise characterizations are difficult. Later, Treisman and Gormican refined their results, dealing explicitly with consciousness and attention (Treisman and Gormican, 1988). They state, “we suggest that voluntary responses in all search tasks depend on the same processing levels that also result in conscious awareness” (my Italics; p. 43). Thus volitional processes are considered conscious[143]. In a later paper (Treisman, 1993), she addresses the module issue above, finding that experiments support within-module parallel, non-volitional processing, while between-module (e.g., combining shape and color) processing seems to necessitate some sort of focusing, usually volitional. Yet again we have the question of Treisman’s use of the term “focus”, since it might be taken to include the attraction of attention by pop-outs[144] or, again, the patterns of saccades, both of which are non-volitional processes. In fact, she finds that different criteria for object recognition and feature processing, having to do with conjunctions of present vs. absent characteristics, result in the necessity for different types of processing, ranging from inhibitory non-volitional attentional processes to volitional selective focusing (pp. 17-22). Thus what seem fairly clear-cut results relating to modularity or dimensionality again become somewhat ambiguous, certainly very complex. Treisman introduces her own classification of attention in this paper. She identifies unconscious (“preattentional”) processing of features and “divided attention” when focusing, with the latter necessary when there are “separate representations for figures defined by darker and by lighter contrast, with focused attention required to combine across representations” (p. 16). This is a specific example of the modularity issues above. In addition she makes explicit “the idea that feature-coding remains parallel and global up to the level that defines surfaces” (p. 16), and finally, makes

 

a distinction between preattention (inaccessible to awareness…), inattention (… results of preattentive processing… retrieved once attention is redirected), and divided attention (that… allows conscious access to global properties) (p. 16).

 

The distinction between “preattention” and “inattention” is one I would assign to increased processing within the salient or non-volitional aspect of consciousness, but otherwise her results seem to support my model. There is backing for the salience/focusing division, for the unification of those processes, and there is support for the nonvolitional/salience and volitional/focusing relationships. However, there is an issue relating to the clarity of her results concerning volition. Thus, her use of a phrase like “divided attention allows” introduces some ambiguity into that context, from my perspective. But especially given her earlier (1988) paper, I believe that focusing for Treisman, as for myself, is volitional. Note that throughout, her position is that salient (non-volitional) processes are unconscious but that their results may be conscious, and that they result in cohesive structures. Moreover, volitional processes and their results are both conscious, and elaborate, combine and abstract those salience structures.

Palmer and Rock noted that Gestalt groupings require the pre-existence of objects (Palmer and Rock, 1994). They hypothesized processes which underlie classical Gestalt groupings, creating the objects to be grouped. That is, since the Gestalt processes require objects to operate on, the actual object creation must occur prior to them. They modeled object creation on the basis of first, the initial, low-level creation of edges, surfaces, textures, and so forth which occurs starting literally at the retina (e.g., p. 34), second, the establishment of regions of “uniform connectedness (UC)” (“a connected region of uniform visual properties… strongly tends to be organized as a single perceptual unit”, p. 30)[145], which then provide the basis for figure-ground distinctions (p. 39), and finally, objects (p. 38). This relates directly to my model, as follows:

 

Recent results found by Mack, Rock, and their collaborators… found that some process of element individuation – that is, designating figures against ground – appears to occur without voluntary attention, but that classical grouping does not. (p. 39).

 

Thus, the volitional/nonvolitional distinction is present and relates to perceptual functions at low levels of perceptual organization; second, it is intimately involved with Gestalt processes; third, our lack of awareness of these fundamental processes, and in addition, the ambiguity in the results[146] indicates the blending of these properties, as I have claimed.

Watson and Kramer performed a series of experiments which supplement Treisman’s location-based focusing (Watson and Kramer, 1999). They attempted to show that a variety of criteria for object generation, based on Palmer and Rock’s ideas above (Palmer and Rock, 1994), are responsible for the unconscious, non-volitional creation of objects by the visual system.

 Those objects are subsequently assigned relative locations, as we shall see from other studies. In this series of experiments, however, Watson and Kramer support both Palmer and Rock’s claims, and also several important claims I have made about consciousness. First, there are complex non-volitional processes responsible not merely for the creation of objects, but for intra-object structure. Objects, even at the initial salient level, have internal structure. Thus,

 

The UC [uniform connectedness] operator segregates the incoming visual information into distinct UC regions, which are contiguous regions with… color, texture, and luminosity. These segregated UC regions are the entry-level representations… also available to the grouping operator, which employs classical Gestalt grouping principles… to form larger grouped-UC representations that are also available for selection. (p. 34).

 

Further experiments support the segregation of different properties into objects even if adjacent objects possessed similar properties. That is, the object contours , their outlines, in effect, determined what were seen as objects over and above their colors or textures. It is only when the outlines determine the objects very clearly that those other properties come into play: “Object parts increase in salience with increases in the magnitude of concave discontinuities of object boundaries” (p. 41). So if we have a collection of red and blue squares, and red and blue round figures, that totality will be seen, first, as two kinds of objects, and second, primarily as objects determined by their shape: square and round, and only secondarily as red and blue. Note also that all of these results show not merely that objects are being formed, but that those objects have internal structures (“object parts”) which internally vary in intensity[147]. Objects vary among themselves in intensity, and in addition their components vary in intensity. The effects of conscious awareness in the control of object creation is clear, but it is not clearly separable from that of non-volitional object creation:

 

In some respects, these results… are similar to the results of previous studies that have reported top-down effects on object-based attentional selection…. Subjects were more accurate in tracking five continuously moving dots among five moving distractor dots if they had been told to interpret the target dots as vertices of… objects…. [similarly, there was] a same-object effect for contour judgments on ambiguous figure-ground displays when subjects were told to imagine the contours to be on a single object. Thus… top-down factors, in the form of expectancies and instructions, are sufficient to encourage object-based attentional selection (p. 39, my Italics).

 

This passage was written to support the idea that objects are formed independently of location, as a result of volitional effects (focusing) on “stimuli” (i.e., on salient objects). Thus, “top-down” effects serve to voluntarily and cognitively generate objects. From my point of view, however, what is strongly corroborated is that there is clear evidence, first, for both salience and focusing effects, and second, for a thorough integration or synthesis of those effects. Focusing may result in radical alteration of the results of salient processes, and Watson and Kramer’s experiments explicate and support both functional and phenomenological interactions between salience and focus[148].

The location vs. object-based hypotheses concerning the focus of attention have become an ongoing controversy. The resolution of this controversy seems to be the incorporation of both hypotheses into one, as we shall see. The spotlight metaphor has facilitated experiments showing, for example, that small deviations from foveation, as little as 3-4 degrees, can result in significantly fewer details seen and remembered (e.g., Eriksen and Hoffman, 1972). This and other results have reinforced the idea that location is what drives intensity. On the other hand, many experiments have shown that characteristics of objects, common across separated objects, may draw more attention than different characteristics within the same object. Object-based theories are supported by that and other data. As far as my model is concerned, it would seem that object-based theories would be preferable, since it is in fact objects that I am concerned with in inter-relationships of intensity between contents. I believe, however, that syntheses such as Baylis’ or Logan’s (Baylis and Driver, 1993; Logan, 1996), where location is not an absolute, but is object-relative, and thus the hierarchy of object relationships includes location, probably correspond most closely to actual processes.

Older studies of attention, follow-ups of those studies, and studies of object creation, as we see above, tend to support my model. What of more recent studies directly concerned with attention? A possibility for investigating phenomena directly related to attention would be to contrast them with phenomena in situations where one’s “attention wanders”, or where one discovers that one has not been paying attention as closely as one believed. The phenomenon of “change blindness” occurs when an alteration to a feature or an anomalous feature, sometimes of a radical nature, introduced into a scene, can go apparently unseen by many observers – under “conditions of inattention” – i.e., if they were not expecting to see anything abnormal, and were, preferably, attending to some other location in the visual field. Such studies involve situations ranging from experiments on the pop-out effect to people in gorilla suits wandering across basketball courts. These anomalous figures are unnoticed by about 25% of observers (e.g., Mack and Rock, 1998, p. 13). In these experiments, “focus” may refer either to volitional effects on, say, the location of one’s gaze, or it may refer to the locations of saccades, which, while non-volitional in origin, are guided and directed by a variety of volitional processes, ranging from expectations to detailed visual control[149]. The phenomenon of change blindness seems to indicate that what I am terming “focus” is necessary for virtually all Gestalt effects to be generated. According to Mack and Rock, without focusing on objects, we have very little intermittent grouping by proximity, similarity, texture, and shape, among other factors (Mack and Rock, 1998, pp. 11-13). However, motion, color, location, and some few other characteristics do not seem to require focusing. This position is similar to Palmer’s ideas about “uniform connectedness” above (Palmer and Rock, 1994), since it implicates focusing primarily after objects have been constructed. If this is true, then the case is even stronger for the interaction of focus and salience in the production and consciousness of objects.

Part of the difficulty in interpreting change blindness experiments results from determining what causes an interruption to what we are attending, and the extent of that interruption. Rensink, citing such experiments, states (Rensink, 2000) that focus is required to create objects, and that only “proto-objects”, which are “volatile” are pre-attentively[150] formed. When attention is diverted to another location, the “stable” and more complex objects “dissolve back into… constituent proto-objects” (p. 20). Irwin, who, in an early paper, viewed such information as transient (“the perception of a stable and continuous visual world across eye movements is not accomplished by accumulating and integrating the visible contents of successive eye fixations in a spatially defined integrative visual buffer…” [Irwin, 1996, p. 96]), did not seem to consider that such a “buffer” might not be spatially defined. More recently, Irwin has modified his position somewhat to take into account some amount of accumulation of information over fixations (Irwin and Zelinsky, 2002).

One problem, as Hollingworth and Henderson point out in their critique of this and other “visual transience hypotheses” (Hollingworth and Henderson, 2002, p. 116), comes from another area of vision research. Data from long-term memory (LTM) studies “indicates that long-term picture-memory can preserve quite detailed information… humans possess a prodigious ability to remember pictures presented at study” (p. 116)[151]. The pictures in the particular study they cite were presented for only five or ten seconds each, and subjects were tested for recognition of subsets of 2500 pictures (Standing, et al., 1970, p. 73), and had an accuracy of over 95%. The literature in this area dates at least back to the 60s, and strongly demonstrates that “LTM for scenes is not limited to the gist of the scene or to the identities of individual objects” (Hollingworth and Henderson, 2002, p. 116), since even for recognition recall, details must be remembered (i.e., stored in memory) to discriminate within such large sets. Further, the several experiments by Hollingworth and Henderson “do not support the object file theory of transsaccadic memory…. Instead, these data support a view of scene perception in which visual representations accumulate in memory from fixated and attended regions of a scene” (Hollingworth and Henderson, 2002, p. 130). Further, when the actual numbers are examined, it seems that there is almost never complete change blindness, and that “Depending on the particular variant of this task, between 25% and 75% of the observers failed to notice the unexpected object” (Most, et al., 2000, p. 2). Thus, as many as 75% of observers did notice unexpected changes in scenes. Given all this, Rensink’s[152] claims (e.g., “changes in an image of a real-world scene become difficult to detect when made during a flicker, blink… or other such interruption…. Little detailed information is being accumulated.” [p. 18]) seem far too strong. However, change blindness does exist, and Hollingworth and Henderson explain it as resulting from several types of errors of perception (p. 131). They further argue that information involving both object properties and relative object location within scenes does indeed accumulate in long-term memory, but gradually, over several saccadic and/or non-saccadic fixations. Their position, then, is somewhat similar to Palmer’s and to Irwin’s, but allows for reasonably rapid and hierarchical long-term storage and retrieval. Hollingworth and Henderson partially summarize their results as follows:

 

The retrieval of LTM [long-term memory] codes for previously attended objects and the comparison of this information with current perceptual representations is strongly influenced by the allocation of visual attention and thus by fixation position. Access to… an object file in VSTM [very short-term memory] is proposed to be dependent on attending to the spatial position at which the file is indexed. (p. 132).

 

That is, noting change (i.e., comparing previously attended objects to “current perceptual representations”) is very strongly influenced by the location of the fixation areas of one’s gaze. And so if those latter areas do not coincide with an alteration in the scene, change will not be noted, since the above comparison will not be made. Thus, given this hypothesis, change blindness is compatible with the accumulation of information over saccades.

There is a great deal of conceptual content in the above passage involving signal detection and digital computer metaphors with which I do not agree (e.g., “codes” and “object files”), but the gist of their meaning seems accurate to me as it relates to what we have seen of the phenomenology of vision. In addition, when they speak of attention, they refer to “the nature of the representations produced when attention and the eyes are oriented to an object” (p. 132). But that orientation may or may not be volitional. To the extent that it is not, the structure that is generated upon the retrieval of “object files” is the structure generated by salience processes, and the degree of focus controls the extent of further structuring. However, Hollingworth and Henderson’s directions to their subjects should have biased them toward volitional focusing: “participants were instructed to monitor each scene for object changes… and to press a button immediately after detecting a change” (p. 120). Changes in the “objects” in these experiments were expected, although not insofar as their details were concerned. Thus, again, the intensity seems a synthesis of the effects of salience and focus.

Logan attempts another synthesis of object- and space-based theories in his “CODE” model (Logan, 1996). Initially, object- and space-based information are blended into what might be termed “global” properties, and gradually discriminated and separated as visual information becomes detailed or local[153]. This kind of hierarchical discrimination seems quite in line with developmental theories, and his use of analog properties consistent with neural processes. One of the most interesting of his conclusions is that the focus of attention, modeled in his study as a spotlight, is inseparable from the objects in that focus: attention cannot exist without its objects (p. 635). His conclusions, of course, are supported by experiment and not merely predictions of his model. These conclusions are consistent with the branch of phenomenology favored by Gurwitsch, and opposing Husserl, and one which I also prefer. That is, there is no consciousness apart from its objects, and the “shape” of the spotlight is also determined by its contents. In my terms, intensity is thus a parameter or characteristic of phenomenal objects rather than a phenomenon or dimension somehow existing independently, within which objects are placed. I strongly favor that hypothesis.

It seems, then, that up to this point the data indicate that salient processes produce structure, but not enough to evoke meaning, since objects and figure-ground remain uncertain. But consider the following study in the picture-memory field. Loftus and Hogden, in a paper explicitly relating phenomenology to cognitive science (Loftus and Hogden, 1988), present, in one of several experiments, subjects with “132… naturalistic color pictures… depicting seascapes, landscapes, cityscapes, and weddings” (p. 153) in a “study phase” in which they were shown for 40 msec each, with 3 seconds between pictures. The subjects were able to reasonably correctly (70%) tell whether or not they had seen the picture previously in a subsequent “test phase” (pp. 154-155). In 40 msec, then, subjects are obtaining, and storing (in the three-second interval between images), enough information to generate fairly accurate recognition of over one hundred complex pictures. Since the pictures were similar, no simple algorithms (“this is a landscape” vs. “this is a triangle…”) were able to differentiate them. Clearly, volition and fixation were involved, but the brief presentation and the large number of similar pictures provide strong support for neither a sharp boundary between salient and focusing processes nor for object transience. Further, I will describe evidence below from two studies (i.e., Buchanan and Westbury, 2001, and Li, et al., 2002,) combining the two paradigms above, viz., the picture-memory and the object discrimination studies, which indicates that these seemingly contradictory results are both correct.

Given that my model turns in part on the distinction between salience and focusing, should I welcome such ambiguities? What are shown by the above studies are not, I claim, problems with the parameters of salience and focusing as aspects of the dimension of intensity. The ambiguity in the field is partially a result of its neglect of the phenomenology of salience and volition; and there are at least two consequences of that neglect. First, insofar as salient processes are independent of focusing processes, there should be some indication of whether that independence is reflected in consciousness. The indications are that our consciousness of such independence should be minimal. What implications does this have for consciousness studies? Second, phenomenological investigation should be employed to aid in delimiting, inasmuch as possible, the separate and combined contributions of these parameters, and of volition. An explicitly phenomenological investigation, since volition seems more characteristic of focusing, might very well provide experimenters with more information about contributions of top-down versus bottom-up processes.

 

V. The Empirical Basis of Recursion

 

One of the purely abstract considerations that has driven theories about the relationship between parts and wholes is the question of how it is that a part is recognized as such when apprehended, and further, how it is recognized as part of its particular whole. In the first section of this essay, I examined the rationale for gestalts on this basis: that a simple addition or concatenation of components was insufficient to generate a whole. Husserl and Gurwitsch were quite aware of this problem, and it resulted in part in Gurwitsch’s embracing Gestalt psychology[154]. I would like to reexamine it from the standpoint of the components of the objects of attention. Consider the components of any visual object. We have seen that non-volitional processes continue the generation of various interrelationships between components at least in part generated by other low-level (i.e., conscious and non-volitional, and unconscious) processes. We will see below that the act of focusing on components of objects continues those processes of elaboration and discrimination: we can see the vertices of a triangle as small areas, or pick out an object from a scene. In all these cases, however, we have concentrated on that particular component. We are aware of it in the context of the object of which it is a part. Thus, the vertex of a triangle is always of the triangle, once we have seen it as such, and not merely two line segments joined at a point, and so forth. How is this accomplished phenomenally? Surely the object itself, e.g., the triangle, must in some manner be present in its components. If that were not the case, then why would we not apprehend the components as separate objects, outside their previous context (the object in which they were embedded)? But this does not happen, even if that context has vanished; in that case, we provide the context from memory. Given this, we must concede that the object is in some sense present in its components[155].

Consider the consciousness of an object, one that we see or visualize. As a simple visual presentation, similar to the laboratory presentations in classical studies of attention, there is very little meaningfulness - in the sense of rich networks of components - to single letters, to words in odd contexts, or fragments of pictures. Meaningfulness, inasmuch as it is experienced, is, I claim, absolutely dependent on such complex systems of evocations. Indeed, I maintain that those systems realize meaningfulness. Given the extensive literature on meaning[156], I feel I must set forth some rationale for that claim. Functionally, I have no objection to most of the wide variety of conceptions of meaning. That is, meaning as “reference”, meaning as “use”, meaning as some kind of public symbolic definition… all these are, I believe, useful and functional concepts in their particular contexts. So it is largely irrelevant to me, finally, whether the term “meaning” is characterized as one of these many formulations; and if someone wishes to constrain me to substitute another term: “connotations”, perhaps, for what I am describing, I am willing to go along.

I must say, however, that I find most of the non-recursive[157] symbolic definitions of meaning[158] puzzling, from a subjective standpoint. That is, for any individual who is, say, reading a book purely for enjoyment, paragraphs, sentences, and even individual words or short phrases must be considered to have meaning, in some sense of that term, within that person’s moment-to-moment stream of consciousness. Otherwise, why read for pleasure? Should we consider that someone in this situation is reading, say, meaningful words (i.e., words meaningful to them - surely we must concede that), but that they are conscious, not of those meanings, but something else? This conception of meaning, were anyone to hold it, would seem very odd to me. What then would be the point of characterizing the words that person was reading as meaningful to them, i.e., as having meaning for them as they read? Given that consideration, symbolic conceptions of meaning, and particularly public ones, seem very odd. Surely as we read, e.g., “fire engine”, we do not have a symbolic string like “red truck used to put out fires, with ladders…” and so forth, going through our minds. If so, we would, first, be caught in a regress, since each subunit of that phrase would need its meaning elaborated in the same fashion, and second, we would never finish reading our book, since the time it took to read “fire engine” would include the time necessary to hear, or visualize, or process in some manner the next symbolic stage, which, given that it is a string of symbols whose meaning – by whatever definition - is in part determined by its order, could not be simultaneous; and so forth through the regression. It must be, then, that what happens when we read “fire engine” is that nearly simultaneously[159] we visualize a long red truck, perhaps hear a siren… and so forth[160]. And this, in general, is what I will consider to be the meaning of “fire engine”[161].

In a laboratory setting, then, the mere appearance of an object, as that of a shape, with texture, curves, color… and so forth, may have some meaning because of the laboratory context itself, and from resemblance to real-world objects. When a person looks at, e.g., a smooth colored sphere on a computer display, it may bear only a remote resemblance to a basketball or a marble. We see few minimally featured spheres in normal contexts, and we must thus “borrow” meaning in order to see that sphere as rubbery, or as solid and rigid[162]. Thus, the evoking of meaningful objects at least in part provides the meaning for a simple figure which resembles them. But the resulting meaning may still be minimal, and must to great extent be regarded as due to that resemblance.

Further, we see, as a marble, let us say, a colored sphere pictured on a display, and so part of what we see, e.g., the glassiness of a marble, is a component of that sphere’s evocations and thus a component of that sphere. That is, the glassiness of the marble is a component of a component of a gestalt[163]. The sphere itself is a part, a component, of the display: we are looking at a screen with a sphere on it, which latter object we have decided is a marble. But we also know that we do not see a marble; we see a computer display. So the marble is not a marble, it is a pictured object we see as a marble, and that object, as marble, must in turn re-evoke the sphere which is an object on a computer display, unless we are truly hallucinating, in order that we are conscious of the marble as an arbitrary assignment of meaning, i.e., a pictured marble[164]. But if this is the case, and the arguments above are reasonable, then this evocation occurs in part because of a kind of recursion which is phenomenological, where the simple object references another, and that second re-evokes the first, and so forth[165]. But for that second to be meaningful, it must have multiple and complex evocations itself, much more than the first object, in fact. And so we find ourselves forced to consider – as present to some degree – a rich, deep, and recursive phenomenological system even in laboratory situations.

But how, exactly, is the sphere a component of the marble? There are several possible explanations, some of which can be fairly easily eliminated. I will consider the phenomenological picture in more detail, because a particular phenomenological explanation, below, will lead us fairly naturally, I believe, into the experimental literature. There are, as far as I know, three ways in which such phenomenological recursion might occur. First, an object (e.g., the sphere above) could be duplicated within its components. We would then have a recursive system in which the original object, like a reflection in a pair of mirrors, endlessly reiterates and regenerates itself. The advantages of this conception are that it does solve the context problem, since the sphere, for example, would quite simply be a component of the marble, the marble of the sphere, and so forth. In addition, through the recession and multiplication of components, a true continuous field, similar to the creation of the number line, is generated. Yet there are problems with this conception. First, although we are in a sense employing a nearly mathematical concept of recursion here, we are not using mathematical symbols, and the object, recursed in its components, cannot be the actual object that provided the original context; it is a new, if identical, object. This is, after all, an attempt at a phenomenological analysis, and one in which the original object is duplicated (“n-“ plicated, actually, where “n” is some large number) in this manner is simply not true to our experiences. We do not normally experience the equivalent of a hall of mirrors, where each object reflects and reduplicates itself repeatedly in other objects, creating an infinite recession of identical objects. Second, this is not consistent with the experimental results I have so painfully elucidated above (and that I will present below). Normally, subjects do not report regressions of this type. Third, hypotheses about neural mechanisms for this would be very difficult to formulate, since neural dynamics can support only so much structure.

The second alternative is temporal. If the object and its components oscillate between each other, so quickly that they appear as one entity, and interact at the same speed, we might be conscious only of one unitary set. This is perhaps physiologically possible, given reverberating neural circuits, and it might solve the contextual problem, but we may be begging the question of the nature of consciousness if this is a conception of phenomenal objects being shown on a kind of internal movie screen, so fast that a kind of flicker fusion unites them. Who then is watching this movie? If this latter is seen as the solution, it is an inelegant one, and one that creates more questions than it answers, hidden in the word “appear”. I am not willing to accept it without a great deal of evidence. However, a reverberating circuit might accomplish somewhat the same end if this type of fusion was not necessary, and instead object/component identification were tied to the speed of their neural realizations. Then circuits firing in near-synchrony might, through some unknown mechanism, as a result of that coincidence, be united into object and components. In this case, a rapid enough alteration between object and component might bind them together phenomenally by creating the object/component relationship. At least this poses no more problems than the next alternative.

The third alternative is to direct the recursion upwards, in effect. That is, we can have the components of the components be precisely the original object, the context, in effect, in which the components are embedded. In the above example, the sphere which is a component of the marble (which is a component of the sphere) is not a duplicate of the original, it is the original gestalt. This is, actually, a more mathematical solution than the first. It would require, as neural dynamics, that different sets of neurons, one realizing the object and another a component of that object, mutually generate and/or reinforce each other, and the result of that generation, a kind of superset of neural dynamics, be an entity in the same sense that its components (the original object and its components) are entities. If we assume that there is some means by which the latter can happen, then we must also assume that the former can occur by the same means, since they would both be realized, in general, by the same kind of neural events. Here the experience of phenomenal continuity is at least in part caused by the fusion of neural events. We do not know how that fusion is brought about, but since it must occur in those smaller neural sets (i.e., the original object and its components), it seems plausible that it occurs in larger sets[166]. Phenomenally, we have a situation in which the simultaneous apprehension (and/or memory) of the object and component binds them together because of preexisting object-component relationships. Once established, mutual evocation of both the objects and their relationships continues[167].

So it seems that there are at least two reasonable means by which recursive and unified holistic objects, gestalts, can be generated, and that these involve the creation and fusion of fairly large neural sets from mutually generating subsets. The most efficient solution might be a combination of the two above, so that both spatial and temporal factors could influence the synthesis of neural events. Phenomenally, it seems necessary that both object and component be present for the component to be understood as belonging to a particular object, whichever of those is primarily focused on. Is there evidence for any of this? In fact, there is evidence for phenomenal recursion[168], as we shall soon see.

I have mentioned the study by Li, et al., 2002), which shows a clear difference in the processing of complex, highly meaningful visual images compared to simple images. This difference consists in the marked relative ease and greater speed of processing meaningful images. That difference may explain the picture-memory results above, which seem to contradict the classical attention studies. I will return to this study below.

A study by Vecera, et al. (2001), - who are aware of the component identity problem - provides empirical support not merely for object-based theories, but for the recursive nature of an object’s components. They find that not only can attention select parts of objects, but it can simultaneously select the object and some of its components. They state:

 

We observed both the part-based and the object-based attentional effects concurrently. These results indicate that our stimulus objects were perceived as objects per se and not as conglomeration of parts or features…. We do not seem to lose the identity of an object when attending to the parts of that object…. How can part-based attention be reconciled with whole-object processing? Part processing and object processing need not be mutually exclusive, as indicated by connectionist models of part and object processing. (Vecera, et al., 2001, pp. 318-319).

 

In another study, employing extensions of Gestalt-theoretical concepts, Hoffman has not merely hypothesized that parts of objects, as well as the objects themselves, may be what he terms “salient” (Hoffman and Singh, 1997), but has formulated rules for that part-salience for both two- and three-dimensional objects (e.g., pp. 62-63), and has supported those rules with a number of experiments. Here we have a direct theoretical and experimental demonstration of the recursive structure of consciousness. His conception of salience is an interesting functional one: “… parts help us index our memory of shapes. Their salience determines, in part, their efficacy as an index” (p. 32). Salience – which I would equate to intensity here[169] - is thus determined by function: the effectiveness of a part to help determine the shape – which aids in determining the meaning - of an object. And we might extend this conception to the field of consciousness as a whole: the intensity of an object may be determined by its effectiveness in making meaningful the context in which it is itself a part.

Schroyens has studied parafoveal information processing during reading (Schroyens, Vitu, et al., 1999), and found that the reading time of a word is shortened when it was previously available in parafoveal vision, and that the difficulty of the parafoveal word influences the reading time of the foveal word, and vice versa (e.g., p. 1038). Thus, it seems that not merely shape recognition, but many more aspects of meaning are extracted to some extent even when a word is not the primary focus. Meaning is thus a component, a part, of what itself is a part (viz., the meaning of the parafoveal word is a component of the meaning of the foveal word) of the whole context. This is a remarkable demonstration of what might be termed second-order recursion of higher-order components, i.e., meanings. This could not be purely an effect of salience; it required multiple saccades and deliberate concentration on particular words, and so is due to a combination of salience and focusing.

Neuroimaging data also supports this type of recursion. O’Craven, et al. find that “attending to one attribute of an object… enhanced the neural representation not only of that object but also the other attribute of the same object… compared with attributes of the other object” (O'Craven, et al., 1999, pp. 584-585). Thus, increasing intensity toward an object increases awareness of more than a single unitary phenomenon, i.e., of mutually enhancing attributes. What this indicates, in other words, is that the components of a gestalt evoke the whole, and the whole in turn evokes the components, similarly to my model.

Fabre-Thorpe’s studies of picture-memory also support these claims. He considers picture-memory a “specific mode of visual processing” (Fabre-Thorpe, et al., 2001, p. 175), having to do with complex images, which seems to run automatically up to some maximum speed. The accuracy of identifying particular images – e.g., of animals – among similar distractors was astounding: roughly 97% accuracy in identifying 200 images among 1200 presented, after the target images were thoroughly learned. Some of the distractors were make-believe animals, in order to further confuse identification. The time taken to identify the correct targets was under half a second (p. 172). This is strong indication, I believe, that extremely complex gestalts can be retrieved at high speeds, i.e., that we can be conscious , both in memory and in perception, of multiple components simultaneously, and that these gestalts are meaningful.

VanRullen and Thorpe, employing event-related potential measurements, have separated the processing of complex scenes into automatically-running and volitional processes (VanRullen and Thorpe, 2001, pp. 458-459). The first, non-volitional, set of processes did not in fact seem accessible to consciousness; subjects were conscious, it seemed, only of the results of the first blended with the second. This paper, then, supports both my major structural hypotheses: that there are volitional and non-volitional processes, and that they are blended, in phenomenal awareness. Their finding that salience and focus are separate parameters of visual processing is a gratifying support for my model, and so is the difficulty of separating them.

Li, et al. (2002) have explicitly compared picture-memory processing and the simple object processing  employed in conventional attentional investigations, and also found that there seem to be intrinsic differences between the two modes of apprehension. When subjects were shown figures upon which they focused their attention, and simultaneously pictures displaced peripherally, they were reliably able to differentiate between animals and vehicles, for example, in the latter pictures. In comparable tests, they were not able to discriminate large Ts from Ls when the latter simple shapes were peripheral. Thus, it seems that complex shapes which are highly meaningful are processed more easily than simple shapes. The involvement of “higher-order” processes involving meaning seems the only explanation here, i.e., “It thus appears that a sophisticated high level of representation (e.g., semantic) can be accessed outside the focus of attention” (p. 9599). They offer no explanation for this effect, and indeed it goes counter, as they are aware, to the classical literature in attention (p. 9601)[170]. But it supports the picture-memory literature, and in addition supports a conception of consciousness in which parts or components of complex and meaningful wholes are themselves seen as meaningful, and one in which that meaningfulness promotes the recognition of the whole.

In addition, Nairne’s recent survey of the short-term memory literature, and his defense of a cue-driven interference hypothesis (e.g., Nairne, 2002, p. 72, p. 75), is further support for recursion. Consider a situation in which a word is repeated. The usual result is a loss of meaning for that word. This is analogous to Nairne’s mention of the effects of “release from proactive interference”(p. 70). When lists of similar words, i.e., words from the same “conceptual class”, are presented to subjects, they find it increasingly difficult to remember them until the words become dissimilar, i.e., are taken from a different class. At that point, remembering again becomes easy. This seems to be the result of the interference of the similar evocations of the words in the same class. But the implications of this are quite interesting. First, it is very difficult to see how words’ meanings can be simple propositions[171], since in that case evocations of words from a similar class should merely reinforce one another, and indeed the repetition of the same word should strongly reinforce itself. Yet the opposite happens. Further, if the evocations of words merely re-evoked the word with which they were associated, then again, it is difficult to see why interference would take place, since the original word would be, again, reinforced. However, if the evocations themselves evoked other components, the proactive interference is easy to explain; multiple layers of evocations are interfering with each other. But if this is true, it is strong support for the existence of recursive levels of structure in gestalts. We will return to word-association studies below.

I believe that the above studies, taken together, provide strong indication of a complex, recursive, organized, and meaningful structure which is intrinsic to the field of consciousness, and that it actually aids our discriminations and manipulations. Meaningful structure in this sense enables simplification, perhaps because, first, it contributes to the formation of large-scale[172] gestalts, and thus aids the central nervous system in carrying out the ubiquitous processes of abstraction and discrimination. Second, in accordance with my claims above, meaningfulness is related to the number and type of components, and those components, in order to possess the recursive properties I am arguing for, themselves re-evoke the object of which they are parts.

 

VI. The Empirical Basis of Directionality

 

We now have one more property of phenomenal consciousness to investigate, relating to what might be termed the “meta-relationships” between phenomenal objects. This is the property I have been terming “directionality”. We have seen evidence for other dimensions of conscious experiences: that one can focus to different extents and that there are differing degrees of salience of various objects, both simultaneously and sequentially, resulting in various degrees of intensity. Recursive structure and thus, the recursion of the dimension of intensity, has also been supported. It is becoming clear, then, that the components of gestalts, and gestalts themselves, are interrelated to various degrees, and that focusing on some aspect of a gestalt entails that particular other components, along with those focused on, become more intense. For example, when we focus on the Dalmatian in the illustration above, we see its body parts in preference to seeing the lawn it is standing on, and when we focus on that lawn, we see its light and shadow in preference to seeing the Dalmatian. This might seem a trivial consequence of the existence of phenomenal objects: how else could there be objects unless they and their components were bound in such a fashion? Yet we might turn that reasoning around, and ask whether that same binding, viz., the orientation or directionality of these relationships, might create objects as well as be generated by them. Thus, as we mature and we learn about various objects, they become elaborated at least in the simple sense of acquiring more components as we interact with them. But further, those components, as aspects of single objects, must be interrelated with each other and to components of other objects. These points seem so trivial and inevitable that I hesitate to cite evidence for them; one can easily think of instances from Piaget, for example[173].

But if we do elaborate objects as we interact with them, then the literal creation of phenomenal objects[174] is at least in part the result of such interactions. Thus, as we study a musical instrument, what was, for example, a row of keys becomes differentiated into particular keys, with particular functions and characteristics. But beyond that, the action of pressing a key can become differentiated into a class of actions, different types of key presses, each with different sonic and emotional effects. Each type of keypress, then, has become an object which we could, and indeed do, when playing the instrument, view as an individual. Yet all began as the single action of key pressing. How then are these separate objects formed? The general answer is easy enough: a variety of interactions of different types leads to groupings and classifications on the basis of those types. I cannot now ask here how in detail any specific differentiation comes to be[175], but I must consider how to describe them, and their phenomenal alterations and formation, in the most general fashion. The structure, and the change in structure, of the interrelationships of components such as these is as important, I claim, in creating phenomenal objects as is the addition and alteration of content. That is, what I am terming the “groupings” of objects is an issue which is central to the formation of phenomenal objects, as we shall see.

Let us investigate directionality more closely. There are two ways in which we might understand this phenomenal parameter. Consider a particular visual object. We may see it as primarily related to some other object in the field of vision, e.g., as a fire engine might be related to the fire hydrant it is parked near. We could consider not merely the strength of the relationship – the reciprocal[176] intensities - between those two, but in addition, since that fire hydrant is the primary object at that moment relating to the fire truck, that we have a direction of relationship, an asymmetry of intensity, relative to other phenomenal objects[177]. That is, the fire hydrant is more intense, say, as we regard the truck, than is a bystander, and so forth. But aside from the strength of this relationship relative to similar relationships to other objects, we also have what might be termed an absolute relationship. That is, there is always a relationship between the fire engine and that fire hydrant in that context. In this sense, a direction of relationship – a directionality - between the engine and the hydrant is not relative to other objects, but is part of the context, whatever its momentary and relative intensity. Further, as we reflect on directionality, we must consider the origin, so to speak, of the direction. We are primarily considering the fire truck; relative to that, the hydrant is most intense. It might be different were we to consider the hydrant, in the same context, as our primary focus; the fire truck might not be the component most intensely related to it[178]. Thus our directionality is not merely asymmetrical relative to other components, but it may be asymmetrical relative to the same components, considered from different centers, so to speak.

Now consider the internal relationships of the components of the fire engine: the ladder, the wheels. Here again we have moment-by-moment directional relationships: as we see the side of the engine, we are, as we consider and/or apprehend that side, most aware, let us say, of the wheels. Again, there are both the relative and absolute strengths of these relationships. We may speak, then, of the directionality of relationships between gestalts, and of the microdirectionality of relationships within gestalts. Note that these are similar to, but not identical with, vectors. There is no “state space” or “component space” phenomenally, within which to orient vectors (and see below). Our phenomenal space, when we experience it, is retinotopic[179]. In such a space, we might characterize “direction” in either absolute or relative terms; if we assume that two components of a gestalt are related to a third, then that relationship will hold in many settings, although it may wax and wane, so to speak, or disappear in some contexts.

I feel that some cautions are in order here. In the above, I am speaking of “components” and “relationships”. While I do claim that these are phenomenal, I am not deliberately being atomistic; I just do not see any other way to speak of these matters. Indeed, there is data indicating a symmetrical reciprocity of the directionality, and that this symmetry argues for their Gestalt-quality, i.e., for the unity of the components into a single phenomenal entity. That is, associationists studied forward and backwards associations for decades and found asymmetries, but a modern analysis of these studies has found that such asymmetry is most likely an artifact of the studies and their statistics (Kahana, 2002). This finding supports the position that we experience gestalts, and thus that atomism is incorrect. Further, this symmetry also weighs against a vector conception, since vectors are classically unidirectional, and these must be bi-directional. Yet since these relationships do have strengths, intensities, magnitudes of a sort, and since they are between particular components, it is, I believe, possible to think of them as bi-directional “proto-vectors”[180]. In addition, there is another factor which leads quite naturally to such a directional conception, i.e., the temporality of these relationships.

Thus, although I have not been explicit, I have mostly been speaking about components experienced simultaneously. But again, given the data, we must be extremely cautious. Just what is phenomenological simultaneity? Remembering the studies, above, on saccades and features, we have seen that successive saccades, and indeed successive acts of focusing, serve to aid in the elaboration of the objects in the phenomenal field[181]. But what this implies is that normal phenomenal simultaneity is not necessarily simultaneity as found in a laboratory; that the image we experience as being “in the instant”, i.e., with all its components simultaneously present to our consciousness, is in fact the result of multiple successive returns to and from the same components.

Can we know whether the directionality between two or more components is one of simultaneous or one of successive evocations? Let us take the example of the Dalmatian above. In order to see it as a Dalmatian when we see its tail, for example, we must, as I have said, experience the whole in the part: the Dalmatian as a component of its tail, and this must happen phenomenally simultaneously. Now, it may be the case that this simultaneity is in fact a Husserlian “running off” (e.g., Husserl, 1990, p. 29), i.e., a phenomenon in which past experiences blend seamlessly into present experiences, and certainly this must happen to some extent[182]. But I claim that we do not, and indeed cannot, know precisely to what extent this is true in any given case, except perhaps in controlled laboratory conditions. We have seen, from the above studies, that to varying extents we might have simultaneous, near-simultaneous (i.e., saccadic), pre-volitional (i.e., multiple salience-induced focusings), and volitional (multiple focusings) components, all within one object experienced as possessing multiple components. The object we finally experience blends all this. The upshot of this rough analysis, then, is merely to return to the assertion that we experience multiple components, inter-related, some simultaneously and some as temporally following evocations[183].

But given that, we may understand directionality in these two senses. In the phenomenal present we experience the bi-directional proto-vector quality of the mutual inter-directionalities between and within gestalts. Over the course of time, as those components change in focus, salience, and quality, we must experience corresponding changes in directionalities; and of course the same holds between gestalts. Returning to the paragraph above, in which I am writing of the generation or creation of objects through directionality, we now understand in more depth, perhaps, first, that this is conceivable, and second, how this might be described, in general terms, at least. I want to point out that I am being careful here to say “described”, and to remain within that domain. There are enormous numbers of theoretical treatments relating to the generation and creation of phenomenal objects, i.e., “concepts”, “images”, and so forth. Most of the processes involved, if attention is paid at all to the experiential issue, are acknowledged to be unconscious[184]. Although I will mention, later, theoretical work involving gestalt generation and manipulation[185], at this point I am quite deliberately being very careful to avoid entangling myself in anything beyond phenomenological description and empirical data. Be that as it may, I will consider, below, some implications of directionality for the creation of phenomenal objects.

First, however, I will address the empirical issue. Is there data supporting the above assertions[186]? We might start with studies of word associations. These days such studies may be dismissed as insufficiently functional, i.e., as not taking into account the multiplicity of reasons that words are related; and in addition as making the perhaps unwarranted assumption that meaning, or components of gestalts, is somehow related to verbal evocations. However, despite these objections, word association is still extensively studied, and in fact employed to evaluate the structure of meanings, as we shall see. Furthermore, while simply noting associations does not take account of possible functional accounts of the causes of those associations, that alone does not invalidate employing them, inasmuch as they do relate to what I would consider gestalt structure[187], and to what, for example, Buchanan and Westbury term semantic “neighborhood size effects” (Buchanan and Westbury, 2001, p. 531, and see below).

If we examine some of the early studies, we find what seems obvious and trivial: particular words have sets of associated words which are different from other particular words[188]. But trivial as this may be, if we make any sort of assumption connecting words - evoked and/or associated - to the meaning of the initial word evoking them, then we must admit that there is evidence for the directionality I have been describing. I might, for example, site Deese’s studies as a paradigm case of data demonstrating such biases (e.g., Deese, 1965, p. 49). Here, additionally, we see an asymmetry which might be taken to indicate that directionalities are indeed true vectors. That is, analyzing the table on that page, one finds that the associational strength from most words (a) to other words (b) (forward associations) is not the same as that from (b) to (a) (backward associations). However, Kahana, as I have said, has determined that these individual asymmetries, found in many associational studies, are artifacts :

 

The author examined the correlation between forward and backward recall at the level of individual pairs of items. Both this correlation and the correlation between recall of pairs tested in the same direction were near unity. (Kahana, 2002, p. 823).

 

In fact, it is fortunate for a gestalt-based model that they are symmetrical, because, as Kahana mentions, such individual asymmetry would argue[189] for an atomistic conception of components, while symmetry in these associations argues for a tighter-knit structure, more like the gestalt conception (p. 823, p. 835). On the other hand, that analysis of the reciprocity must have necessarily been over many subjects and many experiments. I might still claim that for one person, in one context, directionalities are asymmetrical, as in my fire truck/hydrant example above. In fact, there are two contemporary studies whose results seem to contradict Kahana. Both Sloman, et al. (1998) and Ahn, et al. (2002) found that the relationships between features of concepts, not necessarily relationships between associated words, are usually asymmetrical in their mutual dependencies (e.g., Sloman, et al., 1998 , p. 205; Ahn, et al., 2002, p. 114). I will return to Sloman’s study, in particular, in the next chapter.

Continuing the tradition of word-association studies, Galizio finds that category clustering[190] of words is a “highly robust” finding (Galizio, et al., 2001, p. 609). That is, he found that the grouping of remembered (free-recalled) words into meaning-organized clusters reflecting category membership occurs not only for “natural language categories” but for artificially-constructed categories (p. 609). Again, if words are taken to reflect underlying meanings, as indeed is the implication of their grouping into categories, then this finding also supports the kind of intensity biases or evocation strengths I argue underlie directionality. That is, given that, “words were recalled in clusters reflecting category membership” (p. 609), we may conclude that such clusters, reflecting groupings of meanings, and, equivalently, groupings of gestalts, are likely to occur not only in laboratory situations involving free recall but in real-world situations. Given that a particular meaning evokes another preferentially, we have with that evocation established an empirical basis for what I have termed “temporal” directionality.

As far as atemporal directionality is concerned, Pexman, et al.’s fascinating study (Pexman, et al., 2002) has implications not only for directionality and recursion, but for the picture memory findings above. They found that words with the greatest number of features - where “features” were indicated by the number of associated words, which in turn was taken to indicate the “richness of a semantic representation” (p. 547) - are more easily recognized than those with fewer features. Words were evaluated for number of meanings as well as number of features, and those with only one meaning were selected. It was hypothesized that feedback from the “semantic activation” of the features[191] would improve word recognition, and this hypothesis was supported by the data. Pexman also points out (p. 544) that this result indicates that features, as represented by words, do exist as aspects of meaning, and that semantics does influence phonology, as least insofar as word recognition is concerned. The implications for picture memory are in accordance with Li, et al’s results comparing simple with complex figures, as I have mentioned[192]. That is, we can now consider that the greater the reciprocal activation, i.e., evocation, of a gestalt and its components, the more that gestalt is tied together, so to speak, into a meaningful whole[193]. Further implications for directionality as a phenomenal parameter follow: the more intra-gestalt microdirectionality relative to inter-gestalt directionality, the more closely knit, meaningful, and easily recalled that gestalt will be.

Buchanan and Westbury, in fact, take for granted that different concepts evoke other concepts (i.e., “features”) to different extents. They go further, and differentiate between “associational” and “object-based” models of semantics on the basis of the types of semantic neighborhoods” (p. 532), or as I would term it, the classes of directionalities, in the evocations (again, the “features”) that they exhibit. These models predict, to some extent, different semantic neighborhoods, where a semantic neighborhood is defined by the words, or features, which preferentially cluster nearest to the word being investigated. Object-based models, according to them, are “formulated with respect to properties of the objects to which a word refers”[194], while associational models are “defined by properties of the language used to refer to those objects” (Buchanan and Westbury, 2001, p. 531). We can thus see that directionality is not even an issue in these models; they both assume the kind of preferentiality of which I am speaking. Buchanan favors an associational model because of the ease of experimentation, and demonstrates fairly clearly[195] that such a model does a good job of predicting word recognition (e.g., pp. 539-541). But it is not the details of evoked types that I am concerned with at this point, merely that the evidence is so  strong for such biases that one can make a reasonable attempt at differentiating evocational hypotheses.

Moving from associational studies, we find that Mozer, in modeling a “regularity principle” for the formation of phenomenal objects, states,

 

Regularity in the relations among different parts of an object is weaker than in the internal structure of a part. This principle can be applied recursively to define part-whole relationships among elements in a scene. (Mozer, 1999, p. 52).

 

His model (a computer simulation tested against human subjects) supports this claim (and see also Mozer, et al., 1992). There are several implications of this position. First, the creation of phenomenal objects relates to the predominance of internal directionality within its components. That is, simply, an object’s components preferentially refer to and evoke each other rather than other objects or other objects’ components, the same conclusion I had drawn above. As can be seen, Mozer has a very similar conception of recursion to mine. The implications for the creation of objects, then, are that changes in mutually inter-referring directionalities of components can, among other things, create or dissolve phenomenal objects. I will have more to say about this below.

In addition, the well-known phenomenon of “chunking” also supports my model. As Gobet puts it, “a common definition of a chunk… is a collection of elements having strong associations with one another, but weak associations with elements within other chunks” (Gobet, et al., 2001, p. 236). A chunk, then, seems to be precisely what I am considering as the result of mutually inter-referring directionalities. As such, and given the picture memory data and Pexman’s work, it should be no surprise to find, first, that chunks may be variable in the number and types of their components, and second, that their very complexity may aid in their retrieval. Thus, we might speculate that as chunks get larger (i.e., possess more components) their increasing size is offset by the increasing feedback-based activation effects, insofar as their retrieval is concerned[196].

We may also return to Sloman, et al’s studies (e.g., Sloman, et al., 1998; Sloman and Ahn, 1999; Sloman and Malt, 2003). That is, in his investigations into whether concepts possess essential features (see Chapter 1), Sloman finds that “natural kinds exhibit clusters of correlated properties” (Sloman and Malt, 2003, p. 11); that “a feature [which I also term a “component”] is conceptually central… to the extent that other features depend on it” (Sloman and Ahn, 1999, p. 533); that “the centrality of a feature represents the degree to which the feature is integral to the mental representation of an object, the degree to which it lends conceptual coherence” (Sloman, et al., 1998, p. 190). All of these findings support the model I am proposing, in terms of directionality, recursion, and intensity. What they are claiming is that if there are two features, both of which have many features associated with them, both in the same concept, then those two features will be central to that concept. But that implies that those two features will most likely strongly evoke each other, because of overlapping associations. And that, in turn, supports the idea of directionality as just that kind of bias, where two (or more) features preferentially a) evoke each other (i.e., they are temporally directional), and b) are preferentially paired with each other (i.e., they are atemporally directional).

 

VII. A Short Summary

 

That ends the section dealing with empirical support for my model, but certainly not from lack of further data. In summary:

First, phenomenal consciousness is a unified whole, a gestalt, consisting of various types of components: phenomenal objects, interrelated in such a way that alterations to any component potentially alter many or all other components. Consciousness can be characterized, divided or parameterized[197] with reference to the intensity with which we experience phenomenal objects through the volitional type of act of increasing or decreasing focus on them, and the non-volitional type of act of increasing or decreasing their salience. Both of these acts, and their results, are among our conscious experiences.

Second, I consider that phenomenal objects, while they are gestalts, have components, themselves gestalts, which are, in effect, differentiable areas within the unity which is a given object.

Third, the internal structure of these gestalts is itself characterized by areas of greater and lesser focus and salience.

Fourth, the components of the field of consciousness, and of any individual gestalt within that field, preferentially refer to or evoke other components, both within that same gestalt and of other gestalts.

Thus, the structure which best fits my model is only superficially one of a lens or focused light. I would prefer the metaphor of a turbulent vortex, an inflow which fills an overflowing container. As with liquid turbulence, we have smaller vortices in the flow which are to greater or lesser extents stable, depending on the stability of their sources and on the surrounding flows. Each phenomenal object then corresponds to such a smaller vortex, with its own set of turbulences. A ripple or disturbance anywhere in the flow alters the whole, to various extents. But even this metaphor fails to capture its internal stabilization, active processes which feed back into the flow to maintain its structure. The rationale for this metaphor will be more clear toward the end of the next Chapter.

Chapter three

 

Tip-of-the-Tongue Phenomena and Structural Phenomenology

 

I. Introduction

 

We have seen that phenomenal consciousness can be described with several general structural parameters. Now I will apply this framework to both analyze and describe the phenomenology of the tip-of-the-tongue (TOT) state.

Why will I study TOTs? There are three groups of reasons why they are extremely valuable to study in the context of structural phenomenology. First, TOTs have been fairly extensively studied, both phenomenologically and empirically. There are both empirical data and phenomenological descriptions available of the same phenomenon, and it thus relatively straightforward to relate those literatures, in contrast to many other subfields of phenomena. Second, TOT states are metacognitive. That is, they are about other states of consciousness. Thus, a “feeling-of-knowing” is a state in which one judges that one knows something, and in which that judgment may not only be realized as a proposition, e.g., “I know that x”, but as a feeling akin to a feeling of recognition or a feeling of déją vu. Both of those latter may be expressed propositionally, yet they are not always, perhaps not even usually, experienced as explicit propositions. Instead we term them “feelings”, and experience them as clear and explicit wholes. I will have a great deal more to say about the phenomenology of such states in this chapter. But it is the case that they are states which are intrinsically involved with consciousness, and which cannot be studied, indeed cannot be conceived, without accompanying phenomenological information. As such, investigating them virtually forces bridging between the phenomenological and the empirical arenas. Third, TOTs, as they have been conceived, are primarily verbal. I will comment on this below, to the effect that this is merely one aspect of possible causes of TOT states, but nonetheless, that is how they originated as a field of study, and that is primarily how they have continued to be studied from the last century to the present. As such, they provide a means of extending phenomenology, which has largely concentrated on studies of nonverbal phenomena, mostly of visual sensations, to the verbal arena. While most of these latter studies have been expressed verbally, their subject matter has not primarily been language, with an exception, perhaps, being the field of semiotics. In any case, the TOT brings cognitive studies of language and language production and comprehension together with phenomenological studies in a very unique way, which encourages the extension of phenomenology into empirical linguistics.

Now, what can the particular brand of phenomenology which I am attempting to develop here contribute to the study of TOTs?

First: one of the interesting properties of TOT states, as I mentioned above, is that they are intrinsically self-referential, self-reporting phenomena. The course that has been taken, both in cognitive studies and in phenomenological reports of the TOT phenomenon is to list and analyze its contents. But I am not too concerned with specific contents, as I have said. I will indeed examine much of what has been written about TOT contents, but the purpose of that examination will be either to set them aside for others to explicate or to enable the analysis of structure. What I will be claiming, as a result of my analysis, is that at least part of the content of TOT states is precisely an awareness of their structure; that, indeed, metacognition in general involves an awareness of that structure, manifested as aspects of the “feelings-of-knowing” (FOKs)[198] and the like. One empirically-testable consequence of this hypothesis is that we should not be able to know that we do not know that we do not know something; but we could know that we know that we do not know something[199].

Second: I have argued that consciousness is structured around a dimension of intensity, for one, and that this dimension is comprised of two components: salience and focusing, blended imperceptibly in normal consciousness into that one phenomenal property. I will relate the dynamics of the TOTs state to that structure in the following general manner.

Salient, non-volitional, processes begin the structuring of a conscious experience. They are followed – roughly speaking, as we have seen - by focusing processes. What if, for some combination of reasons, to be discussed below, either 1) those salient processes did not proceed, proceeded incompletely, or proceeded erratically, or 2) the salient processes proceeded, but the volitional processes could not, or 3) neither of those processes proceeded normally? We would probably experience a gestalt which was abnormal in some respects[200]. In addition, since focusing is volitional, we are consciously attempting to focus on this gestalt, and in consequence elaborate and complete it. If we were not able to accomplish those goals, we would, first, become aware that we were not succeeding, second, become aware, to some extent, of why we were not succeeding, and third, attempt to complete either or both of the volitional and the non-volitional processes.

So the above is a fairly specific set of possibilities: 1) that because TOT states are the awareness of the incomplete volitional processes, the non-volitional processes will have been completed before them; 2) that because the non-volitional processes are complete, the TOT is solely the result of incomplete focusing processes; 3) it is a combination of these factors. Given that we can study TOTs, and given that we can study the results of non-volitional vs. volitional processes, we can, in theory, confirm or disconfirm one or several of these hypotheses. Note that, in part at least, due to the nature of the TOT, phenomenological investigation is required to establish and to verify or disconfirm these hypotheses. Thus I am now in a position to claim that structural phenomenology has enabled a) specific hypotheses, b) predictions from those hypotheses, and c) the possibility of empirical investigations which confirm or disconfirm those hypotheses: Popperian criteria. We will find, given the literature and the phenomenology of TOTs, that either hypothesis 1 or 3 above may be true, 2 is almost certainly not, and that 1 is the most likely hypothesis.

In summary, I have laid out above several possibilities for the order and combination of nonvolitional and volitional processes involved in the TOT state.

In addition, I will claim, as a result of structural considerations, that the TOT state can be generalized beyond the standard verbal recall situation. That is, normally, a TOT state involves forgetting a proper noun, but remembering nonverbal aspects related to that noun: recalling a person’s appearance is tied to forgetting their name. I will claim that this TOT state is actually one of four possible situations, corresponding to the various permutations of the nonverbal/verbal modalities.

I will claim, as a result of the empirical work on TOT states, that metacognitive states, e.g., feelings of familiarity, are clearly knowable, discriminable, and measurable, and no more vague, evanescent, or “transparent” than are most other conscious phenomena.

I will also claim that the current understanding of the TOT phenomenon is incomplete, and that in order to fully understand its characteristics and structure one must account for it in dynamic terms. That is, I will claim, most generally, that the TOT state is not merely a set of metacognitions resulting from and searching for a memory, but that it is the result of the interruption of one goal-directed process, and the replacement of that one with another. As a result, the goal(s) of the retrieval and of the TOT state are directly manifested in that state’s characteristics, both structurally and in its content. There are several specific claims that I will make that result from this general one, which I will explicate below.

 

II. Explication of the TOT State: Definitions and Clarifications

 

Before I can begin to present the empirical work on the TOT state, there are several possible confusions that should be immediately cleared up. The tip-of-the-tongue phenomenon is not the processes or states causing that feeling - a feeling, roughly speaking, of imminent recall - to occur. The TOT is a feeling, or more precisely, as we shall see, it is a dynamic gestalt comprised of several feelings, particular contents, and particular types of structures, which are readily and measurably distinguishable from each other and from other metacognitive contents (and structures, as I claim above) of consciousness, such as other types of feelings-of-knowing (FOKs), feelings of familiarity, feelings of intuitive correctness, logical correctness, feelings of recognition, feelings of error, of similarity, of dissimilarity, and so forth. As we shall see, there are specific processes that cause the TOT state[201], and many hypotheses concerning the nature of TOT etiology. Those processes, e.g., processes of interference, of blocking, of inference, of evaluating familiarity, and others which I shall describe, are not the TOT state or any of the components comprising that state. I do not believe that, at this point, any of the hypotheses about TOT etiology that have been empirically studied[202] are demonstrably incorrect; it is my opinion that many of them, singly or in combination, may operate in particular contexts to produce the TOT state. Since these processes are for the most part unconscious, they will not actually be my concern except inasmuch as they might influence the phenomenology of the TOT state.

Several authors speak of “feelings-of-knowing” (FOKs). This category is unavoidably conflated with TOTs. Nelson’s characterization of FOKs is the statement that they are “judgments about whether a given currently nonrecallable item is known and/or will be remembered on a subsequent retention test” (Nelson and Narens, 1994, p. 16). This definition is based on a temporal model of metacognitive monitoring processes which proceeds from ease-of-learning (EOL) judgments to judgments of learning (JOL) to FOK judgments to judgments about confidence in one’s retrieved answers (CIR), as one attempts to learn and to evaluate one’s own progress in that attempt (p. 21). One problem with this model is that it is clearly not exhaustive as to metacognitive states. Thus, we may make judgments about the similarity and dissimilarity of a variety of phenomena, and about whether such judgments are accurate or not; we may make judgments about what we are feeling or about how strong our emotions are. We may make judgments about the accuracy of some intuitive answer we have just found, or about whether we are perceiving something correctly… and on and on[203]. Indeed, the number of such judgments is not merely huge; in addition, such metacognitive judgments are virtually ubiquitous in our mental life, so much so that aberrations in them may underlie many forms of schizophrenia. This is certainly true, for example, in Capgras and Cotard delusions, in delusions of being controlled externally[204], in self-boundary delusions, and others (e.g., Langdon and Coltheart, 2000, pp. 207-211). It is also the case that various aphasias, anomias, and hyperlexias involve impairment of these processes (Glosser, et al., 1996; Aram, 1997; Weekes and Robinson, 1997; Romani and Martin, 1999; Vigliocco, et al., 1999; McCarthy and Kartsounis, 2000; Gorno-Tempini, et al., 2001; Westmacott and Moscovitch, 2001; Caza, et al., 2002). The neurology of metacognitive judgments seems to preferentially involve the prefrontal cortex (e.g., Gehring and Fencsik, 2001; Kikyo, et al., 2001), and thus damage to various prefrontal structures will impair such processes to varying extents. I will return to some of these later, when I consider TOT etiology in more detail.

There are relatively few studies designed to explicitly investigate the FOK/TOT difference. Wellman defined TOTs in very young children as the ability to judge whether they had seen an item, and FOKs as the ability to judge whether they would be able to recognize an item (Wellman, 1977). Significant age-related differences were found in their prediction of their own future recognition (FOK), but not in their judgment of whether they were currently recognizing items, i.e., of their recall (TOT). This study is consistent with my characterization of FOKs (below), and we will find that it also supports Schwartz’s conclusions about TOT etiology[205]. Hart also investigated the difference in terms of the recognition/recall dichotomy (Hart, 1965; Hart, 1967b; Hart, 1967a). When current definitions of the TOT state are compared with the several FOK characterizations above, it may be difficult to find clear distinctions, since judgments of current versus future recognition may not reflect recognition/recall characterizations in current memory theories. I do not, however, feel that this is worth making an issue about, given that we have understood the reasons why, but in any case the distinction I would like to emphasize is not so much one based on judgments of the contents of one’s memory versus one’s recognition, as one based primarily on the feelings (e.g., of frustration, emotionality, imminence) associated with those judgments. This latter delineation is in fact explicitly made by Widner, et al. (1996, p. 527), who also found support for the same inferential etiology as Schwartz. Yaniv’s study made much the same distinction (Yaniv and Meyer, 1987).

The upshot of this is that FOKs may be better portrayed, I believe, as the class of metacognitive states involving judgments of the accuracy and/or reliability of one’s memory, since both recognition and recall (in the above contexts) are memory-dependent. Note that this does not include all metacognitive states by any means; I am excluding, to take just one example, judgments of the reliability and validity of the knowledge of one’s immediate perceptions, primary ingredients in some schizophrenias. But it captures, partially at least, the idea that a FOK (i.e., a feeling-of-knowing) concerns one’s knowing. In addition, notice that this characterization, and Nelson’s, do not restrict themselves to words, nor indeed to symbols of any sort. Given this analysis, TOTs are a subset of FOKs.

Now that we have begun to differentiate the TOT state from other similar states, I can better characterize it as a unique phenomenon. It was probably William James who brought TOT phenomena, per se, to the attention of psychologists, with little more than the quotes below:

 

Suppose we try to recall a forgotten name… There is a gap therein; but no mere gap. It is a gap that is intensely active. A sort of a wraith of a name is in it, beckoning us in a given direction…. If the wrong names are proposed to us this singular gap acts immediately so as to negate them…. And the gap of one word does not feel like the gap of another, all empty of content as both might seem necessarily to be when described as gaps…. There are innumerable consciousnesses of emptiness, no one of which taken in itself has a name, but all different from each other (James, 1950b, p. 251-2).

 

James was followed by several investigators, most notably Brown and McNeill, who characterized the TOT state as one in which “complete recall of a word is not presently possible but is felt to be imminent” (Brown and McNeill, 1966, p. 326). In addition, they stated,

 

The ‘tip of the tongue’ (TOT) state involves a failure to recall a word of which one has knowledge. The evidence of knowledge is either an eventually successful recall or else an act of recognition that occurs, without additional training, when recall has failed. The class of cases defined by the conjunction of knowledge and a failure of recall is a large one. The TOT state, which James described, seems to be a small subclass in which recall is felt to be in imminent. (p. 325).

 

Later, Brown characterizes it thusly, “on occasion, however, memory falters: we are sure that the information is in memory but are temporarily unable to access it” (Brown, 1991, p. 204).

It is not generally known, I believe, that the above phenomenon, which I will term a “nonverbal-to-verbal TOT”, was ascertained by Schwartz to be virtually ubiquitous over human cultures (Schwartz, 1999, pp. 381-382; Schwartz, 2002c, pp. 22-28), and is described almost universally (in 45 out of 51 languages he surveyed) with metaphors involving the tongue. James, then, originated neither the idea of the TOT nor this particular type of description. Schwartz’s definition, “A TOT is a strong feeling that a target word, although currently unrecallable, is known and will be recalled” (Schwartz, 2002c, p. 5, my Italics), is one of the most precise. This definition implies the three felt components of the TOT state which Schwartz identifies, discriminates, and measures in subjects: “strength, emotion, and imminence” (e.g., p. 20).

 I will generalize Schwartz’s definition as follows: A TOT is a strong feeling that a target memory, of whatever modality, although currently unrecallable, is known and will be recalled. The alteration of that one word: “memory”, for “word”, opens up enormous possibilities for both experiment and theory, as we shall see. I will demonstrate below that it is quite possible to systematically generate examples of different processes than the conventional verbal processes, resulting in various types of lapses of memory, all of which may give rise to a TOT state. Further, it will become clear as I proceed that this more general conception follows from the structural analysis of meaning I began earlier. For now, I will merely ask the reader to recall the analysis of meaning I presented in the last chapter. If a word can be an aspect of a gestalt, then, as a component of a gestalt, there is a certain reciprocity or symmetry of relationship which it bears to other components. If one set, let us say, of the components of a gestalt evokes a word, then another set, including the word, may evoke the first, or yet another set.

Thus, the extended conception of the TOT, explicated below, is an empirically-testable prediction of structural phenomenology[206].

 

III. Explication of the TOT State: The Verbal Conception Extended

 

The conventional TOT is the result of a nonverbal-to-verbal set of processes, which we experience as a progression from nonverbal experiences to a verbal experience[207]. We try to remember a person’s name; we can remember their face, their profession, when we met them, and so forth; in other words, we remember the nonverbal components of the gestalt evoked by or associated with that person, and we attempt to retrieve the verbal component(s) of that memory: their name, in this case. Furthermore, a variety of modalities my evoke TOTs for names. For example, Riefer studies TOT states brought about by hearing music and attempting to name the theme (Riefer, 2002); Lawless (Lawless and Engen, 1977) and Herz, 1998), study odor cues for words. In addition, we can understand verbal-to-verbal TOTs in this conventional manner. That is, knowing an acquaintance’s last name, we may try to remember their first name, and so forth.

Now let us permute this situation. We try to remember, say, under what circumstances we met someone. We clearly remember their name, and a friend, says, incredulously, “But that’s Nancy X__, you met her just last week, surely you remember?” We shake our head sadly. No, we do not remember anything except her name, and that perhaps her hair was blonde… but we experience a TOT state: we are fairly sure we will remember other things about her: her, as a person, in time. Our friend prompts us: “The restaurant we had lunch at…” Then, perhaps, we have the “aha” experience: yes, we remember her face, her clothes, her profession, her personality… and so forth. What is this but the nonverbal analog, the inverse, in a sense, of the conventional, verbal TOT, i.e., a verbal-to-nonverbal TOT? Perhaps we have, during this process, analogously to remembering the first letter of her name, remembered her hair color, but nothing else at that moment. Verbal-to-nonverbal TOTs have in fact been described in several settings. Thus, there are several studies of “slips-of-the-pen”, in which the meaning of a Japanese or Chinese character is given either vocally or in one writing system, and the subjects are required to produce the corresponding character in the logographic script in question (e.g., Nihei, 1988; Yamada and Takashima, 2001). Yamada, for example, found that the primary cues were semantic rather than visual similarities (pp. 184-190). One might however question the nonverbal aspect of these studies; they do involve types of writing.

There are other studies, however, which clearly involve verbal-to-nonverbal lapses. Some are clinical investigations of hyperlexia. Thus, Aram speaks of the dissociation between “decoding and comprehension skills” (Aram, 1997, p. 1), where children read (i.e., recognize and pronounce) but do not understand words. Glosser provides further support for phonological/lexical processing without semantic processing in her study of a hyperlexic child (Glosser, et al., 1996). Westmacott and Moscovitch, 2001) present an explicit example of nonverbal TOTs (p. 589) in a severely amnesiac subject. Since such syndromes are present clinically, and since there are bases for it neurologically and cognitively (e.g., Mattson and Baars, 1992; Glosser, et al., 1996; Gorno-Tempini, et al., 2001), then we should expect instances in normal people also, just as we find instances of the reverse effect[208]. In addition, in several studies of normal subjects, Englekamp (e.g., Engelkamp, et al., 1990; Zimmer, et al., 2000; Engelkamp and Zimmer, 2002) finds that the free recall of words is strongly influenced by the performance of related actions. “Action memory” is item-specific, and seems to be coded independently of “script structures” (Engelkamp and Zimmer, 2002, p. 95). It is thus quite conceivable, and consistent with the above, that an action might be remembered independently of, and previous to, the word describing it.

Let us permute this again. Suppose we do not remember, and have no interest in remembering, a person’s name. But we desperately want to remember what we ate when we met her, because was a wonderful meal for which we would like to find a recipe. And so we try to reconstruct her appearance, the smell of the food, the ambience of the restaurant… and finally, perhaps, we remember the marvelous way they seared the tuna with black pepper. This is an example of a nonverbal-to-nonverbal TOT. The complete memory of the tuna’s taste and appearance could have been proceeded by the memory of Nancy X__’s clothes, for example, or the topic of our conversation. Let us take another example. Suppose we went to the Hirshhorn Museum and Sculpture Garden, at the Smithsonian, and saw a sculpture which happened to be Rodin’s Crouching Woman. But we do not remember, or did not see, the name. We know that we liked the particular sculpture that we saw in that hallway at about 3pm, and that it was of a woman, but we cannot visualize the sculpture. Was she sitting, kneeling…? A friend says, “It is dark bronze colored, her arms are around her body….” We visualize a woman like that, and suddenly we see it: she is kneeling, but in a twisted position. We have employed an image to evoke a visual memory, rather than a name. The cue image was an aspect of, or visually related to, the memory we wanted to evoke, and that similarity helped us remember the desired image, just as the similarity of remembering the first letter of a word helps to evoke the word.

The action spoonerisms studied by Mattson, where segments of actions are reversed analogously to the reversal of syllables in spoonerisms (e.g., Mattson and Baars, 1992, pp. 172-182) are not TOTs, may indicate a similar mechanism for nonverbal action errors and linguistic errors. More to the point, Reason draws a parallel between “action slips” in which a habitual action intrudes on a less familiar one, and the intrusions of incorrect words during TOT states (Reason, 1992, p. 82). Further, Chainay has found that actions are more reliably associated with objects, and “contextual/semantic decisions” to words, i.e., that there is less forgetting of actions associated with an object. Thus, 1) a pot and the act of pouring or 2) “pot” and “pouring” are better associated than 3) a pot and “pouring” or 4) “pot” and the act of pouring (Chainay and Humphreys, 2002). In Lawless and Engen’s (1977) study, memory for odor-picture associations was investigated, but not, unfortunately, any associated TOT states[209]. One might object that there are fundamentally different processes governing word searching versus, say, visual searching, and thus that TOT states must be unique to the verbal modality. However, Chuah’s comparison of word and visual memory search and recall times supports the idea, also posited by Cowan (Cowan, et al., 1998), that a modality-independent “central executive” (Chuah and Maybery, 1999, pp. 376-377) is responsible for such searching. Similarly, Munnich states, “We conclude that spatial language and spatial memory engage the same kinds of spatial properties, suggesting similarity in the foundations of the two systems” (Munnich, et al., 2001, p. 171). This commonality is also consistent with the neuroanatomical evidence above, and indeed with much of the work supporting cognitive linguistics’ hypotheses relating language and space[210]. Thus, evidence supports a similarity between processes governing these types of verbal and nonverbal memory.

More examples of these various types of nonverbal TOT precursors are easy to construct. If we are required, on a math exam, to use the Central Limit Theorem to solve a problem, we certainly know its name, but we may very well draw a blank at being able to employ it, unless prompted. If we attempt to remember the directions to someone’s house, that memory may be verbal, a succession of street names and turn directions. Or it may be visual, as the memorized route on a map. We may suffer from kinesthetic TOTS, as we attempt to remember our body’s movements in a Yoga class, from verbal (verbal-to-nonverbal) or from visual (nonverbal-to-nonverbal) prompts. Those memories may thus be evoked verbally or nonverbally, depending on the situation and on our preferred modalities of memory. Further, we may have the same “strong feelings” that a) we know these things, and b) that we will recall them, whether we are experiencing verbal or nonverbal TOT feelings.

Why, then, has there been so little work with nonverbal TOTs, to the extent that the most recent book in the field (i.e., Schwartz, 2002c), by a very experienced investigator, excludes them, by definition? I will not even attempt to speculate[211]. But I will add that Schwartz mentions that in the seminal study of TOTs, Brown and McNeill, 1966) found that there were

 

224 words that were classified by participants as sound matches to an unretrieved TOT target word. They also found 95 words that were similar in meaning to a TOT target. For example, for the word sampan, the participants provided… the following semantic matches: barge, houseboat, and junk (Schwartz, 2002c, p. 10).

 

That is, of a total of 319 evoked words, roughly one-third (0.29) of those were not conventional verbal TOTs, i.e., were not related phonologically or by letter matching to the target, but were related semantically. In addition, since Brown and McNeill did not ask the subjects whether they were, prior to naming it, visualizing a barge, etc., rather than visualizing (or hearing) the word “barge”, we have no knowledge at all as to whether their initial evocations were nonverbal, and they were forced by the context to express them verbally[212].

In summary, the above provides a suggestion as to what a TOT state is, and what may produce that phenomenon. In addition, I have tried to indicate that the TOT phenomenon is not specific to the verbal modality, but is a general phenomenon related to memory retrievals of many types, and that this conception follows from the structural model I have introduced in this essay. If my arguments are correct, there are profound implications for several theories of verbal production (e.g., Burke, et al., 1991; Levelt, et al., 1999), which hypothesize a linear progression from semantic to phonological processes. These theories must be reformulated from linearity to include parallel semantic/phonological[213] processes to account for the above nonverbal TOTs.

The next step I would like to take is to explicate in more detail some of the approaches taken to account for the TOT state, in order to prepare the ground for relating those theories to structural phenomenology. What we will find, roughly speaking, is that the progression into the TOT state is a progression from salient processes to focusing processes. We start in the flow of speech, of memory, of visualizing, of acting, as above. That flow is stopped or blocked by an unconscious conflict or error, by inhibition, or by lack of activation, depending on which etiology or combination of etiologies applies in that particular situation. Since the error halting the previously ongoing flow is unconscious, i.e., salient and nonvolitional[214], we must now become conscious of the halt in the flow of speech, memory, or imagery, and further, conscious of it as an error. This initial consciousness might be compared to the pop-out phenomenon in stimuli, but it might be better described as a “pop-in”, in which we are conscious of a sudden absence, e.g., of the name, rather than the sudden presence of a sensation or stimulus. But at least one result is the same: we focus on the absence, the “pop-in” in order to enrich and fill it in, just as with the pop-out. Normally this succeeds, perhaps with some minimal conscious effort, and the gestalt[215] is filled in, just as a pop-out is elaborated when focused on. But in an extended TOT this does not happen; continued interference, insufficient activation of the goal, or insufficient inhibition of errors: a continuation of erroneous processes[216], blocks or renders insufficient the inferential/activation processes. To resolve the TOT state, we might, among other possibilities, employ external or internal cueing, activation through continuing focusing effort, or we may use inhibitory processes, triggered by the awareness of erroneous fill-ins[217] to eliminate spurious blockers, intruders, conflicts, or inference results. Or we might simply ask someone for help.

 

IV. The TOT State in Detail

 

1. Etiology

 

Now I will fill in some of the gaps in the above outline. Since this is not an essay primarily concerned with the TOT, I clearly cannot cover the complete history of TOT studies nor can I give a complete rundown on hypotheses of TOT etiology. Schwartz, and Brown (e.g., Brown, 1991; Schwartz, 2002c; Schwartz, 2002a) have admirably covered that ground. What I will attempt here is to highlight some of these theories in order to ease their later relationship to the structural hypotheses I have introduced in the previous sections. The result of that will be strong evidence that TOT states are the result of nonvolitional, salient processes, while their resolutions may be the result of either nonvolitional or volitional processes.

 TOTs result from memory retrieval problems. Those problems can have many possible causes, and the TOT literature is filled with speculation about various mechanisms and explorations of a variety of ways memory retrieval could go wrong. I will briefly review many of these below. I do not believe that any one of these explanations is entirely correct; indeed, I believe that it is likely that all of them are correct, in different contexts. There are however two different classes of theories on the origin of the TOT as a conscious experience, and they are not consistent. But that inconsistency does not preclude multiple specific models within those classes being correct; it does however indicate the probable existence of multiple simultaneous mechanisms of memory errors. I will outline these various alternatives, then fill in some details.

These two classes of retrieval theories, and thus of possible retrieval errors, are explicated by Schwartz (e.g., Schwartz, 2002c) and Brown (1991). I will use Schwartz’s terminology simply because it is the most current. He terms the two categories of memory retrieval theories “direct” and “inferential” (e.g., Schwartz, 1999, p. 384, p. 388). Direct theories hypothesize processes which are more passive or reactive than active, with memories being retrieved by means of activations (e.g., Burke, et al., 1991; Levelt, et al., 1999, pp. 3-4) within large networks of nodes which represent, in these models, abstractions of relevant (i.e., memory-oriented) PDP networks. The neural networks thus represented are of course extremely complex, and connect large regions in the hypothalamus, cortex and prefrontal lobes, to name just a subset of the systems involved. I have mentioned some of the studies in this area here and above; there are many more[218]. Thus, possible mechanisms for retrieval errors are problems with the spread of activation due to decay of traces with age or disuse, or through possible transmission deficits (e.g., Burke, et al., 1991; James and Burke, 2000), or through competition with traces activated in tandem (e.g., Baars, 1992; Levelt, et al., 1999, pp. 8-9). Competition causes “blocking”[219], through mechanisms which are not usually clearly elucidated[220], but could involve, for example, top-down activation of particular patterns (e.g., Ahissar and Hochstein, 2000; Miyashita and Hayashi, 2000; Hodsoll and Humphrey, 2001), or inhibitory effects caused by similar but non-identical activations inhibiting each other (e.g., Metcalfe, 1993; Anderson and Spellman, 1995; Theeuwes and Godijn, 2002; Tipper, et al., 2003). Blocking, then, might in some cases be considered a transitional area between the direct and inferential theories.

Inferential access theories are more unified, and more vague, insofar as their explanations of memory retrieval and error processes are concerned. One of the motivations for these theories is the idea that the phenomenology of one’s access to memories is not necessarily that of the process which accesses them, nor of the memories themselves. That is, Schwartz speaks of the “doctrine of concordance” (Schwartz, 2002c, pp. 15-18), first formulated by Tulving, 1989), in which “phenomenological experience… is the feeling that accompanies the cognitive processes” (Schwartz, 2002c, p. 15), and maintains, quite reasonably, that one can question that correspondence. Thus, the feeling that we are nearly accessing a memory, but cannot quite complete that access, which is an aspect of the TOT experience, does not necessarily reflect the actual procedure of access, complete or not, nor indeed the partial access itself, but may instead reflect a metacognitive process which is monitoring the actual memory access attempt. Thus, memory retrieval may involve more than merely activating traces, and in addition, the phenomenology of memory retrieval may itself involve inferential processes about memory retrieval itself. There may in fact be two layers of inference, one involving the process of retrieval and one involving the products of that retrieval, i.e., “TOTs are not based on sensitivity to inaccessible but activated targets… rememberers infer the target’s existence from a host of clues” Schwartz, 2002c, p. 65. Alternatively, persons might employ partial information about cues (e.g., Metcalfe, et al., 1993, on cue familiarity) to guide processes of recall, which may be some sort of generative processes or processes virtually identical to conscious inferential processes. In fact, those latter also contribute, in some cases, to memory retrieval, and thus to memory failure as well. The neurological mechanisms behind these processes are basically unknown, except that they probably are governed by the prefrontal cortex. Unfortunately, the descriptions of these processes are usually made in terms of our conscious inferential processes (e.g., Schwartz, 2002c, p. 65), which leads to confusion as to their exact nature.

More precisely, the difference between the inferential processes involved in arriving at conclusions about memory access or about accessing memories, and the conscious inferential processes that we employ to decide, say, what kind of bird we just saw at the window, is not explained in this theory. If those former processes are conscious, then they are merely the inferential processes which we employ normally to make informed guesses and draw conclusions about our memories, the world, and so forth, and their production of the TOT state in the particular context of making inferences about memory seems completely mysterious, since they do not seem to cause this state in other, explicitly conscious, inferential contexts[221]. Thus, they cannot be those conscious processes, as such, and must be unconscious and associated with a particular context, that of the monitoring of recall. But if that is the case, in what, precisely, do they consist? And where is the evidence for them? After perusing the literature, I can find mention of only one such indicator, that of “cue familiarity” (Metcalfe, et al., 1993; Schwartz, 1999; Schwartz, 2002c). But cue familiarity, “a strong feeling elicited by recognizing a familiar cue” (e.g., Schwartz, 1999, p. 388), first, is not a process of inference, although it may be the result of inferential processes, and second, bears a suspicious resemblance to the feelings hypothesized as elicited in “direct-access” theories. In fact, speaking of cue familiarity, Koriat states, “from a phenomenological point of view, the experience associated with a positive FOK or TOT is often quite similar to what is implied by the trace-access view” (Koriat, 1994, p. 122), and goes on to speculate that cue-familiarity feelings may result from “a global, automatic, and effortless” (p. 122) apprehension of the results of the metacognitive monitoring processes on memory retrieval[222]. He contrasts Burke’s and others’ direct-access position, which he sees as implying that no false information can be retrieved[223], with this position, which allows accessing “incorrect clues” (p. 125). I see no logical problems with this hypothesis, but if one may have a direct apprehension of the results of metacognitive processes, why not, in addition, a direct access of traces? In that case, both hypothesis could be correct, and perhaps one or the other might predominate in different contexts[224].

  On the other hand, if cue familiarity is the result of inferential processes, i.e., if it is the conclusion of temporally-extended processes of unconscious reasoning, of whatever sort, we are back where we started, attempting to find this inferential basis. None is supplied, at least none that clearly distinguishes these processes from normal, conscious inferential processes; and we have seen the problems those latter imply. In fact, I would venture to suggest an experiment. One might provide two sets of cues to subjects in TOT states. One set would contain, covertly - i.e., hidden in such a way that subjects would not immediately think of them as inferential starting points - contradictory information, such that inferential processes employing them would be stymied, or radically slowed. The other set of cues would (covertly) contain information which, although too complex to “get” immediately and intuitively, would, upon inference, lead properly to a conclusion. The expected outcome, if there were unconscious inferential processes, would be, of course, that the latter set of cues would lead to quicker and/or more frequent TOT resolutions than the former, on the average. An experiment of this sort has never been performed, as far as I am aware, although Rozenblit and Keil, in their wonderful study of overconfidence and knowledge (Rozenblit and Keil, 2002), have done something similar in a different context.

But there is another problem with the direct-access cue-familiarity hypothesis, and that is the “illusory TOT” (Schwartz, 1998, p. 626). Here, employing TOTimals, i.e., drawings of imaginary animals with fanciful names designed (as we have seen above) to induce TOTs, Schwartz found that one may reliably induce TOT states to drawings for which one has never known the names. It is still possible, as Schwartz is aware (Schwartz, 2002c, p. 119), that some kind of partial access may explain these results, but it also seems a reasonable hypothesis that some TOTs are the result of metacognitive monitoring of memory processes; inaccurate monitoring, in this case[225].

The TOT, then, is the result of processes which operate pre-consciously, or as I have put it, nonvolitionally. And indeed if this were not true, then we would find several things. First, TOTs would be consciously and voluntarily inducible, and they are not. In fact, it was only with Smith’s TOTimal idea (e.g., Smith, et al., 1991; Smith, et al., 1994; Schwartz and Smith, 1997) that their induction became reasonably controllable[226]. Second, the processes that would produce them would be transparent, as transparent, at any rate, as are conventional inferential processes, and again that is not the case. Third, they would not be experienced as surprising and frustrating, nor as an interruption of the flow of speech, action, or thought, and all of those are the case also. The retrieval errors and interruptions that cause and evoke the TOT state are, then, the results of salient, nonvolitional processes.

 

2. Components

 

Now, in what exactly does the TOT state consist? What are the components of TOT states[227]? If we take James (above) at face value, this would largely seem a futile analysis, and there are several contemporary commentators on the TOT who might agree, to greater or lesser extents[228]. However, there is very strong evidence that the TOT state is comprised of several components which are a) reliable, i.e., present in virtually all instances to greater or lesser degrees, b) measurable, even quantifiable to some extent, and c) valid, in that they relate both to TOT phenomenology as it has been informally reported, and to the functional aspect of the TOT state as a metacognitive indication of memory error: a prediction of recall. I will primarily draw from Schwartz’s work (e.g., Schwartz, et al., 2000; Schwartz, 2002c; Schwartz, 2002a) to support my claims, because it is virtually unique in this area.

In summary, Schwartz finds that TOT phenomenology includes three general components: “the experience of strength, imminence, and emotionality” (Schwartz, 2002c, p. 37). He also mentions “feelings of relief that may follow TOT resolution” (p. 37). There are also non-phenomenological, but consistent aspects to the TOT experience, for example, the length of time one experiences a TOT, the various types of possible resolutions, and, as we have seen above, whether it is verbal or nonverbal.

The “emotionality”, as Schwartz terms it, of the TOT experience is characterized by negative emotions such as frustration (Schwartz, et al., 2000, p. 19), or by a general emotional “arousal” (Schwartz, 2002a, p. 73), which Schwartz did not ask the subjects to specify further. In the former study, there was a small positive correlation between emotionality and TOT resolution (Schwartz, et al., 2000, p. 25). In the latter study, it was negatively correlated with TOT resolution (Schwartz, 2002a, p. 80)[229]. But in either case, subjects had no problem with rating the TOT on this phenomenological dimension, i.e., emotionality, high or low, was a readily introspectable and consistent type of component of the TOT state[230]. Now, the actual experience that subjects have is not of course a general one of “emotionality”, but one of some specific emotion or set of emotions: perhaps frustration, perhaps pleasurable excitement, annoyance, and so forth. Emotionality, per se, then, is actually the intensity of whatever specific emotion they are experiencing, and is thus a direct apprehension of the dimension I have been advocating.

The “strength” of the TOT is never defined precisely by Schwartz, yet subjects are able to rate TOTs on this dimension, and in fact strength is correlated with the subject of the TOT, i.e., what it was that was forgotten: proper nouns elicited the strongest TOTs (Schwartz, 2002a, p. 79). Since a TOT is the experience of not remembering something, we might claim that this is, then, the intensity of that evaluation of forgetting[231], and, given the data, that it is easiest to realize or most clearly apprehended that one has forgotten, when one has forgotten a proper noun[232]. Strength was not correlated with the likelihood of resolution (p. 78), but it was correlated with false resolutions (Schwartz, et al., 2000, p. 24), i.e., resolving the TOT with a word which is incorrect, but which one believes is correct. This is the weakest of Schwartz’s components of the TOT insofar as he has analyzed it to date, in terms of its content. We might claim that a TOT state is strong if there is an intense feeling of certainty that we have forgotten. We might claim that it is strong if we are clearly focused on that certainty, no matter what its intensity is, i.e., if that feeling has few components and/or few distractors. We might claim that it is strong if we are absolutely sure we have forgotten, in contrast to feeling that we can easily remember. This factor, then, is somewhat ambiguous as to its exact content, the richness of that content, and its relationship to the dimension of intensity. I will however add to this later.

Finally, a TOT’s “imminence” is “a judgment of proximity to resolution” (Schwartz, 2002a, p. 80), and as such, if metacognition is involved in the generation of TOT states, should be, and is, positively correlated with TOT resolution (Schwartz, et al., 2000, p. 25; Schwartz, 2002a, p. 80). Imminence is not universal to all TOTs. There are TOT states which do not feel imminent (Schwartz, et al., 2000, p. 25). This phenomenal component further supports the role of the TOT as a judgment, or as correlated with a judgment, of recall. Now, we might ask just what it is to have a feeling of imminence, and particular examples might be the following. Suppose we watch a faucet, slowly dripping. As each drop forms, we anticipate its complete formation and its fall; if this were suddenly not to happen for no apparent reason we would experience some degree of surprise, perhaps a mild disorientation, some puzzlement. The anticipatory certainty, then, of the next drop’s fall was the feeling of imminence in this case, and it might in part be comprised of a visualization of the drop falling, i.e., of the anticipated sequence of events running to completion, with a feeling that this sequence would occur. But it is that latter feeling which is so hard to pin down, in the sense that it does not present itself as do sensory apprehensions. Yet it does present itself clearly and unmistakably, as a particular experience which is distinctively different from other experiences; and this can be understood best if we consider the implications of its absence.

What would we feel about our surroundings if we did not have this anticipation, the protension[233], of events to come running on expected causal paths? Our lifeworld, in Husserlian terms, would be comparable to that of an accident or disaster victim, but in perpetuity, where the unanticipated happens: one’s life is derailed, so to speak. Yet for the accident victim, life eventually returns to its normal course, at least in the sense that the immediate future can be anticipated. If one did not have that feeling, then one would be in a perpetual state of disaster, in effect, where the falling of a drop of water was as much a causal catastrophe as would be its sudden change into an insect or a small homunculus to the normal apprehension. The example, above, of Capgras syndrome is similar to this, where despite the evident identity of a person over time, when one is unable to generate the judgment that the identical person has returned to one’s presence, one cannot accept that identity.

Considered in this manner, judgments such as the feeling of imminence are demonstrably not vague, and they are clearly distinguishable from other feelings or apprehensions, whether emotional, sensory, or metacognitive. I am emphasizing this point for the following reason. If we consider James’s quote, above, we find the misapprehension which I am attempting here to correct, that feelings of imminence, of absence, and other nonsensory[234] apprehensions, many of which are metacognitive judgments, cannot be as clear, distinct, or distinguishable from each other or from other experiences as are, for example, sensations such as sounds or colors. This is simply wrong, and the fact that Schwartz, for example, is able to successfully ask subjects to make