Copyright © 1997 by Terrence W Deacon

All rights reserved

Printed in the United States of America

First published as a Norton paperback 1998


For information about permission to reproduce selections from this book, write to Permissions, W W Norton & Company, Inc., 500 Fifth Avenue, New York, NY 10110.

The text of this book is composed in New Caledonia with the display set in Trade Gothic Condensed Bold. Composition and manufacturing by the Haddon Craftsmen, Inc.



Deacon, Terrence William.

The symbolic species: the co-evolution of language and the brain / by Terrence W Deacon.

Includes bibliographical references and index.

ISBN 0-393-03838-6

1.  Neurolinguistics. 2. Brain—Evolution. I. Title.

QP399.D43 1997

153.6—DC2O 96-31115


ISBN 0-393-31754-4 pbk.

W W Norton & Company, Inc.

500 Fifth Avenue, New York, N.Y. 10110

W W Norton & Company Ltd.

10 Coptic Street, London WCJA IPU

Extracted From: The SYMBOLIC SPECIES, The Co-evolution of Language and the Brain
Terrance W. Deacon

Page 71 and 72 of Chapter 3 [Symbols Aren’t Simple]

....To get a sense of this logic of signs, let’s begin by considering a few ex­amples. When we say something is “iconic” of something else we usually mean that there is a resemblance that we notice. Landscapes, portraits, and pictures of all kinds are iconic of what they depict. When we say something is an index we mean that it is somehow causally linked to something else, or associated with it in time or space. A thermometer indicates the tem­perature of water, a weathervane indicates the direction of the wind, and a disagreeable odor might indicate the presence of a skunk. Most forms of animal communication have this quality, from pheromonal odors (that in­dicate an animal’s physiological state or proximity) to alarm calls (that indi­cate the presence of a dangerous predator). Finally, when we say something is a “symbol,” we mean there is some social convention, tacit agreement, or explicit code which establishes the relationship that links one thing to an­other. A wedding ring symbolizes a marital agreement; the typographical letter “e” symbolizes a particular sound used in words (or sometimes, as in English, what should be done to other sounds); and taken together, the words of this sentence symbolize a particular idea or set of ideas.
     No particular objects are intrinsically icons, indices, or symbols. They are interpreted to be so, depending on what is produced in response. In simple terms, the differences between iconic, indexical, and symbolic rela­tionships derive from regarding things either with respect to their form, their correlations with other things, or their involvement in systems of con­ventional relationships.
     When we apply these terms to particular things, for instance, calling a particular sculpture an icon, a speedometer an indicator, or a coat of arms a symbol, we are engaging in a sort of tacit shorthand. What we usually mean is that they were designed to be interpreted that way, or are highly likely to be interpreted that way. So, for example, a striking resemblance does not make one thing an icon of another. Only when considering the features of one brings the other to mind because of this resemblance is the relation­ship iconic. Similarity does not cause iconicity, nor is iconicity the physical relationship of similarity. It is a kind of inferential process that is based on recognizing a similarity. As critics of the concept of iconicity have often pointed out, almost anything could be considered an icon of anything else, depending on the vagueness of the similarity considered.
    The same point can be made for each of the other two modes of refer­ential relationship: neither physical connection nor involvement in some conventional activity dictates that something is indexical or symbolic, re­spectively. Only when these are the basis by which one thing invokes an­other are we justified in calling their relationship indexical or symbolic. Though this might seem an obvious point, confusion about it has been a source of significant misunderstandings. For example, there was at one time considerable debate over whether hand signs in American Sign Lan­guage (ASL) are iconic or symbolic. Many signs seemed to resemble pan­tomime or appeared graphically to “depict” or point to what was represented, and so some researchers suggested that their meaning was “merely iconic” and by implication, not wordlike. It is now abundantly clear, however, that despite such resemblances, ASL is a language and its elements are both symbolic and wordlike in every regard. Being capable of iconic or indexical interpretation in no way diminishes these signs’ capacity of being interpreted symbolically as well. These modes of reference aren’t mutually exclusive alternatives; though at any one time only one of these modes may be prominent, the same signs can be icons, indices, and symbols depend­ing on the interpretive process. But the relationships between icons, indices, and symbols are not merely a matter of alternative interpretations. They are to some extent internally related to one another.
    This is evident when we consider examples where different interpreters are able to interpret the same signs to a greater or lesser extent. Consider, for example, an archeologist who discovers some elaborate markings on clay tablets. It is natural to assume that these inscriptions were used symboli­cally by the people who made them, perhaps as a kind of primitive writing. But the archeologist, who as yet has no Rosetta Stone with which to decode them, cannot interpret them symbolically. The archeologist simply infers that to someone in the past these may have been symbolically interpretable, because they resemble symbols seen in other contexts. Being unable to in­terpret them symbolically, he interprets them iconically. Some of the ear­liest inscription systems from the ancient Middle Eastern civilizations of the Fertile Cresent were in fact recovered in contexts that provided additional clues to their representations. Small clay objects were marked with re­peated imprints, then sealed in vessels that accompanied trade goods sent from one place to another. Their physical association with these other arti­facts has provided archeologists with indexical evidence to augment their interpretations. Different marks apparently indicated a corresponding num­ber of items shipped, probably used by the recipient of the shipment to be sure that all items were delivered. No longer merely iconic of other generic writinglike marks, they now can be given indexical and tentative symbolic interpretations, because something more than resemblance is provided.

Jumping to Chapter fourteen: "Such Stuff as Dreams Are Made On"

Page 448
.... Characters point to other characters strings point to other strings. In contrast, an animal mind, even if it is of minimal computational capacity, constructs and processes internally generated indices with respect to an external world to which it is partially adapted. Its capacity to generate indexical responses may be limited but the scope of reference is open-ended. Small brains may be capable of only a limited scope of iconic and indexical reference, but it is a mode of repre­sentation none the less. Algorithms that are not adaptive in a deep sense have none of this quality. Not all indexical processes are conscious, how­ever. When reference is circular it too is mere mechanism, because there is nothing else for the indices to refer to but themselves. Where there is nothing other to represent, there is nothing to be conscious of. A program—whether in a computer or in a brain—could be an object of consciousness but not its source. These arguments do not really contribute any new explanation of consciousness; they merely illustrate certain minimal conditions for symbolic interpretation. But it is significant that this indexical/symbolic distinction seems central to the problem of consciousness.
     No matter what else various theorists might claim about the nature of consciousness, most begin by recognizing that to be conscious of something is to experience a representation of it. The subjective experience of consciousness is always a consciousness of something. This is not to suggest that consciousness is something separate from the representation process itself (a view that Daniel Dennett caricatures as the “Cartesian Theater’ perspective).4 It is simply the realization that experiences arise as the brain pro­gressively transforms neural signals which have been modulated by external physical events into correlated patterns of neural activity in other parts of the brain, which themselves modulate other patterns of neural activity and so on, each re-presenting some formal aspect of the initial interaction in an additional neural context. In very general terms, both the neurophysiological and subjective information processes can be described as generating representations and interpreting them in terms of others.
     For example, the pattern of electromagnetic waves reflecting off an ob­ject and entering the retina and the pattern of neural signals ramifying through circuits of the visual areas of the brain are both part of the causal chain on which the experiences of color are based. The color does not in­here in the object alone, nor is it merely a mental phantom. Something in­trinsic to the object is re-presented in the pattern of light waves and again re-presented in the pattern of neural signals. But it is also re-presented in the subjective experience of color. There is no jump from material stuff to mental stuff in this process. The material and cognitive perspectives both
Page 449
recount relationships in form between successive and simultaneous points in a process, and these formal relationships, I submit, can be either iconic, indexical, or symbolic at various stages and levels of the process.
     If consciousness is inevitably representational, then it follows that a change in the nature of the way information gets represented inevitably con­stitutes a change in consciousness. Consciousness of iconic representations should differ from consciousness of indexical representations, and this in turn should differ from consciousness of symbolic representations. More­over, since these modes of representation are not alternatives at the same level but hierarchically and componentially related to one another, this must also be true of these modes of consciousness as well. They form a nested hierarchy, where certain conditions in the lower levels of con­sciousness are prerequisite to the emergence of consciousness at each higher level.
     All nervous systems support iconic and indexical representational processes, irrespective of their size and complexity. They are the basic in­gredients for adaptation. To some extent, I suspect that every living nervous system exhibits consciousness with respect to the iconic and indexical rep­resentations it can support. It’s just that for some, this is a very limited realm. Their interpretive capacity will determine their capacity for consciousness. The differences between species in this regard are not qualitative, but quantitative. In species with more complex brains, representational states will be more numerous, more diverse, have a greater range of arousal am­plitude, and will integrate across signals that cover a greater scope in both space and time. Statistics of large numbers and immense differences in mag­nitude have a way of making quantitative differences appear to be qualita­tive differences. It is therefore easy to imagine that the human difference is a difference of this kind, a significant quantum increase in the capacity. And to some extent it is. It’s just not the only or even the major difference.
     To appreciate why human beings are able to experience conscious states unprecedented in evolution, we do not need to have solved the mystery of consciousness itself. We do not need to understand the mechanism under­lying conscious states in order to recognize that since they are based on rep­resentations, any difference in representational ability between species will translate into a difference in the ability to be conscious of different sorts of things. The formal characteristics of the interpretation process, whether iconic, indexical, or symbolic, will define the elements of a creature’s con­scious universe. So the development of an unprecedented form of repre­sentation—symbolic representation—while not the origin of consciousness, has produced an unprecedented medium for consciousness. This doesn’t
Page 450
deny generic consciousness to other species; it only denies a particular as­pect of consciousness that is based on symbolic abilities. Our brains share a common design logic with other vertebrate brains, and so we also share all those aspects of consciousness that are mediated by the iconic and in­dexical representation that these other species experience. Since iconic and indexical referential relationships are implicit and essential components of symbolic reference, the modes of consciousness that other species ex­perience are an essential ground for consciousness of the symbolic world. We live most of our concrete lives in the subjective realm that is also shared with other species, but our experience of this world is embedded in the vastly more extensive symbolic world.
     The evolution of symbolic communication has not just changed the range of possible objects of consciousness, it has also changed the nature of con­sciousness itself. Common sense psychology suggests that a lot of thinking gets done in the form of talking to oneself, editing and re-editing imaginary future or reconsidered past conversations, even when this also involves writing or typing out these thoughts to see how well the shorthand of imag­ined monologues translates into a coherent argument. Of course, these sorts of internal conversations must be unique to human brains while the majority of the other modes of thought are not. And given that our brains have only recently been “made over” to aid language processing, it is likely that the proportion of neural space and time dedicated to these various men­tal activities strongly favors the nonlinguistic. This does not necessarily imply that other species do not “replay” troubling past experiences over and over again, or that they are incapable of actively imagining possible expe­riences in some immanent future. They simply do not do so with the aid of symbolic reference or linguistic mnemonics. It also does not imply that imagistic thinking in humans lacks symbolic character and symbolic logic, though these forms of cognition are capable of following chains of associa­tion that are also uninfluenced by language.
     The Russian cognitive psychologist L. S. Vygotsky suggested in the 1930s that a significant number of normal human psychological processes could be understood as internalized versions of processes that are inherently social in nature.5 He gave language a central role to play in this because its fundamentally social nature provides a mental tool for gaining a kind of sub­jective distance from the contents of thought, that is, from our own sub­jective experiences. By importing, as it were, an implicit speaker-listener relationship into cognition, we create a tool for self-reflection by a sort of virtual social distancing from our own thought process. Thus we can talk to ourselves as though talking to others. Vygotsky conceived of mental devel­-
Page 451
opment as a process of condensing and streamlining this internalized social process.
Language functions as a sort of shared code for translating certain essential attributes of memories and images between individuals who otherwise have entirely idiosyncratic experiences. This is possible because symbolic reference strips away any necessary link to the personal experi­ences and musings that ultimately support it. The dissociation allows indi­viduals to supply their own indexical and iconic mnemonics in order to reground these tokens in new iconic and indexical representations during the process of interpretation. My imagistic and emotional experience in re­sponse to the episodes described in a novel is distinct from that of anyone else, though all readers will share a common symbolic understanding of them. The “subjective distance” from what is represented confers a repre­sentational freedom to thought processes that is not afforded by the direct recall or imagining of experiences.
     This is crucial for the development of self-consciousness, and for the sort of detachment from immediate arousal and compulsion that allows for self-control. Self-representation, in the context of representations of alternative pasts and futures, could not be attained without a means for symbolic rep­resentation. It is this representation of self that is held accountable in so­cial agreements, that becomes engaged in the experience of empathy, and that is the source for rational, reflective intentions. According to Vygotsky, this sense of self emerges slowly as children mature. It becomes progres­sively more facile in perspective shifting and at the same time consolidates greater control over the other aspects of self that derive from nonsocial sources, such as the experience of pain and effort, the arousal of basic drives, or the physical boundaries of control over events. And, as studies of this process in various mentally and socially handicapped children suggest, the extent to which it is developed depends both on the extent of exposure to relevant social-symbolic experiences and on the symbol-processing capacity of the individual.
     Unlike the interpretation of icons and indices (a process which is uniquely personal and insular within each brain), symbolic representations are in part externally interpreted—they are shared. For example, though each of us supplies the interpretation of the words and phrases we hear and use, on a moment-by-moment basis, the implicit injunctions and constraints that de­termine each individual interpretation are borrowed from the society of users, and the symbolic reference that results is only reliable insofar as each interpretation corresponds with those performed by others. Imagine that Washington Irving’s character, Rip Van Winkle, had remained in his magi-­
Page 452
cal slumber for many centuries. Upon awakening, not only would he be out of touch culturally, but he would find himself constantly misinterpreting the meaning of many still familiar-sounding words and phrases being spoken by those around him. As languages evolve and meanings and patterns of use drift away from older patterns, reference is maintained by continuity but not fidelity to the past. Symbolic reference is at once a function of the whole web of referential relationships and of the whole network of users extended in space and time. It is as though the symbolic power of words is only on loan to its users. If symbols ultimately derive their representational power, not from the individual, but from a particular society at a particular time, then a person’s symbolic experience of consciousness is to some extent society-dependent—it is borrowed. Its origin is not within the head. It is not implicit in the sum of our concrete experiences.
     Consciousness of self in this way implicitly includes consciousness of other selves, and other consciousnesses can only be represented through the virtual reference created by symbols. The self that is the source of one’s experience of intentionality, the self that is judged by itself as well as by oth­ers for its moral choices, the self that worries about its impending depar­ture from the world, this self is a symbolic self. It is a final irony that it is the virtual, not actual, reference that symbols provide, which gives rise to this experience of self. This most undeniably real experience is a virtual re­ality.

     In a curious way, this recapitulates an unshakable intuition that has been ubiquitously expressed throughout the ages. This is the belief in a disem­bodied spirit or immortal “pilgrim soul” that defines that part of a person that is not “of the body” and is not reducible to the stuff of the material world. My ability to appreciate symbolic reference is not “reducible” to the indexical or iconic reference I use to ground my interpretation, though it is dependent on these lower-level modes of reference. Symbolic reference is also independent of any particular interpretive process, and retains its ref­erential invariance despite interpretation by very different iconic and in­dexical processes in different minds. Its virtual nature notwithstanding, it is the symbolic realm of consciousness that we most identify with and from which our sense of agency and self-control originate. This self is indeed not bounded within a mind or body, and derives its existence from outside— from other minds and other times. It is implicitly part of a larger whole, and to the extent that it too contributes to the formation of other virtual selves and worlds, it is virtually present independent of the existence of the par­ticular brain and body that support it. This may seem a shallow sort of dis­embodiment that pales in compared to mystical images of “out of the body
Page 453
experience”—it is more similar to the legacy of self composers leave in their music or great teachers bequeath to their students—but this symbolic as­pect of self is nonetheless the source of our internal experience of free will and agency.
     Abstract symbolic objects, like the Pythagorean theorem, guide the de­sign and construction of innumerable human artifacts every day. Imagin­ing counterfactual conditions, like what I might have done if I were the one who had stumbled on a reported accident scene, can cause me to enroll in medical first-aid training and perhaps some day aid an accident victim. Even imagined worlds—Olympus, Valhalla, Heaven, Hell, the “Other Side”—influence people’s behavior in this world. Indeed, assumptions about the “will” of an ineffable God have been among the most powerful tools for shaping historical changes. These abstract representations have physical ef­ficacy. They can and do change the world. They are as real and concrete as the force of gravity or the impact of a projectile.
     On the other hand, the self that persists to influence others and continue shaping the world independent of the brain and body that originally ani­mated it is detached from the specific iconic and indexical experiences that once grounded it in a personal subjective experience. This is precisely what makes it available for regrounding in the subjective experience of others, of becoming part of the self that controls, and feels, and connects with yet other selves from the locus of a different body and brain. In this regard, this part of personal identity is intersubjective in the most thoroughgoing sense of the term, and is capable of true transmigration, though not necessarily as a unified whole.
     By this twist of logic, or rather its untangling, we also again return to Descartes’ elaboration of the religious insight that only humans have a soul, and that this core of the self derives from a realm that is of the nature of language, pure mathematics, or geometry. Descartes’ insight, currently maligned as archaic and antiscientific, seems to have more than a passing similarity to the notion I have developed here. But his rationalist assump­tions—like those represented in theories of innate knowledge of language, computational theories of mind, or claims that human brains have a special essence that imbues them with intentional ability—reflected an implicit an­alytic or top-down perspective on the nature of symbolic reference. By fail­ing to appreciate the constitutive role of lower forms of reference, iconic and indexical reference, this perspective kicks the ladder away after climb­ing up to the symbolic realm and then imagines that there never was a lad­der in the first place. This leaves symbolic reference ungrounded and forces us to introduce additional top-down causal hypotheses, such as the existence
Page 454
of an ephemeral soul or the assumption that there can be forms of compu­tation or mental language (mentalese) that are intrinsically meaningful, in order to fill in for this missing causal role in the explanation.
     But unlike the eliminative materialist alternative, the perspective I have outlined does not suggest that this top-down experience of self is all epiphe­nomenal nor that some of the claims about the nature of the mind which derive from it are based on mystical notions. The symbolic representation of self is solidly grounded in simpler representations of self, derived from simpler forms of representation, and yet the arrow of cognitive processes points neither from lower to higher nor from higher to lower forms of ref­erence. As symbolic reference and symbol minds co-evolved from the non-symbolic, each level of process drawing adaptive novelty from the other, so do the levels of self-representation that constitute our experience bring themselves into being in a moment-by-moment coevolutionary process. As the symbolic process can be the co-author of our unanticipated brains, so can the symbolic self be the co-author of the component neural processes that support it. We live in a world that is both entirely physical and virtual at the same time. Remarkably, this virtual facet of the world came into ex­istence relatively recently, as evolutionary time is measured, and it has pro­vided human selves with an unprecedented sort of autonomy or freedom to wander from the constraints of concrete reference, and a unique power for self-determination that derives from this increasingly indirect linkage between symbolic mental representation and its grounds of reference. With it has come a more indirect linkage between mind and body, as well. So this provides a somewhat different perspective on that curious human intuition that our minds are somehow independent of our bodies; an intuition which is often translated into beliefs about disembodied spirit and souls that per­sist beyond death. The experience we have of ourselves as symbols is in at least a minimal sense an experience of just this sort of virtual indepen­dence—it’s just not an independence from corporeal embodiment alto­gether. Though this might seem like a weak consolation in comparison to the freely transmigrating homunculus of mythical tradition, we should not underestimate the miraculous power of symbols to break down even vast barriers of space, time, and idiosyncratic experience that would otherwise separate us impenetrably.
     As we have seen, the symbolic threshold is not intrinsic to the human—nonhuman difference. It is probably crossable to some extent in many different ways by many species. This means that we are not the only species that could possess such a “pilgrim soul,” to use William Butler Yeats’s elegantly descriptive phrase. It was a Darwinian accident, or miracle, of nature that this ability arose once and persisted for so long; but it has provided each of us with the opportunity to participate in bringing new “souls” into the world, not by procreation, but by allowing our own symbolic selves to be shared by other human beings, and perhaps by other animals, or perhaps eventually even by artifacts of our own creation.