Home Page         Papers by Peirce         Peirce-Related Papers


On Peirce's Conception of the Iconic Sign
(Ver. 2.0 of October 9, 1997)

Joseph Ransdell

Dept of Philosophy, Texas Tech University
Lubbock, Texas 79409, U.S.A.

Ransdell's home page

This is a revised version of a paper originally published in Iconicity: Essays on the Nature of Culture, Festschrift for Thomas A. Sebeok, ed. Paul Bouissac, Michael Herzfeld, and Roland Posner (Stauffenburg Verlag (1986).

The changes from the original version are relatively minor, but enough to make it necessary to treat the present version as a distinct text for purposes of exact reference. Since there is no normal pagination on a web page, I assign in lieu of that paragraph numbers, included in brackets and placed flush right, just above the paragraph, for purposes of scholarly reference.

      Peirce was a keen observer of phenomena of semiotical interest and also exceptionally knowledgeable in the prior semiotical traditions, which stem back to antiquity, but his conception of the iconic sign--like his other semiotical conceptions--was developed neither by generalizing from cases and kinds of iconicity he had observed nor by appropriating a pre-existing theoretical conception of this sort of significance or meaningfulness. It is no doubt true that his personal observations and encyclopedic learning strongly affected his way of conceiving these things, but he was convinced from the beginning, and remained convinced, that semiotic had to be developed "architectonically", meaning that it should be developed rigorously from a few relatively simple though highly abstract principles which could be used recursively--that is, reapplied repeatedly to their own products--to yield a highly systematic (and potentially infinite) network of conceptions of which none would be fully comprehensible apart from its systematic inter-relationships with other conceptions.
      This does not mean he supposed the cognitive value of such a theory to be a matter only of internal coherence and other intra-systemic properties, such that purely a priori considerations could establish its cognitive worth, but rather that he believed semiotic should truly be a theory, in the sense of a comprehensive and coherent intellectual "vision" of the phenomena under investigation which would be put to the test as a whole when attempts were made to apply it (and therefore have to undergo systemic modifications and corrections as a whole if and when its inadequacies were revealed in being put to the test of use), as distinct from being a collection of eclectically derived generalizations to be put to the test and modified one by one. This was a conviction he followed consistently across the half-century or so he devoted to the development of semiotic in one respect and another.
      Yet a widespread impression to the contrary still persists, shown by the way in which isolated semiotical conceptions and distinctions, sometimes with little more than a superficial verbal similarity to his, are attributed to him and then adopted, modified, or critiqued with no apparent awareness of the systemic considerations that are an essential aspect of the substance of his ideas. In view of this, it seems best if, in explaining some things about Peirce's conception of the iconic sign, I focus initially upon its logical place and function within his theory generally, though I will also try to convey some idea about what its uses may turn out to be for those who find his overall approach to the relevant phenomena congenial and promising.
      To begin with, then, what does Peirce have in mind when he thinks of things as being "signs"? His many formal definitions of "sign" and cognate terms are not the proper starting point for understanding this; for they are theoretical formulations--more or less rigorous, abstract, and exact, depending upon whom he supposed himself to be addressing in any given writing--and should properly be used rather as touchstones and tests of one's understanding of his theory as it develops. Ironically, though, because of the extensive theoretical use since his time of the word "sign" (and its equivalents in other languages), we are actually more poorly positioned now to understand the pre-theoretical idea of which his technical definitions are careful refinements than we would otherwise be; for the various--and often highly idiosyncratic--technical usages of this word in the past 60 or 70 years have often departed so radically both from its colloquial senses and from the theoretical usages developed up to Peirce's time, which were usually fairly close to colloquial usage, that it has become difficult to perceive what is actually rather obvious, on a common-sense level, about what it means for something to be a sign, namely, it is anything--whether one meets with it in a dream, or in imagination, or in memory, or in sense-perception, or however--which is such that, if one responds to it appropriately, it will thereby reveal, disclose, make manifest, make apparent, make experientially present or available something about something.
      Perhaps our response to what the sign reveals will result immediately in some sort of behavior on our part, or perhaps not, depending upon what it reveals, but the basic idea here is revelation, not stimulation. The reader can verify this by consulting any comprehensive dictionary and looking up "sign" or its equivalents, where I believe it will be found that what is constant in nearly every usage listed, perhaps even without exception, is the idea of something with a revelatory or manifestational power. For example, even the Greek word "sema", which means tomb or grave in one of its usages, does not merely refer to a place of interment, but refers to it as a place marked as one where someone has been interred, thus as revelatory of what may be found there.
      When the sign-value of something is not immediately clear to us, so that some more or less conscious intellectual activity is required to reveal what that value is in the particular case, or when we have taken the apparent significance of something at face value but then learned subsequently that we erred in doing so, we naturally think of our cognitive response to the thing as being interpretational in character, so that even in those cases where we are unaware of ourselves as actually engaging in any interpretational activity we must, nevertheless, grant that to so characterize our response to a sign as such is appropriate. Now, if we take, on the one hand. the objective content of our understanding, i. e. of our interpretation, as such (as distinct from our understanding or interpretation considered as a "subjective" or "mental" occurrence or state), and, on the other hand, that about which the sign has something to "tell" us or reveal to us, we have the basic pre-theoretical ideas which Peirce refined into the theoretical conceptions of the interpretant and object of the sign, respectively. The sign itself then being that which mediates--functions as a means and a medium--between that which it is capable of revealing something about and what it is capable of revealing about it.
      Suppose, now, that there was a sign so comprehensively adequate in its revelatory power that what it could reveal to us about its object was all there is to reveal about it: in such a case what the object is and what the sign could reveal to us about it (the total proper interpretant of the sign) would be substantially identical, assuming that what something is, and what it can in principle be known to be, are identical. In this extreme case, then, the object and the interpretant would differ only in the sense in which there is a difference between something considered in itself and something considered under the aspect of being known; hence in actual cases, where the interpretant is only a partial revelation of the object, there is a real but correspondingly partial identity of interpretant and object--though, again, as qualified by the difference between the thing in itself and the thing as it is known to be. I think this, too, can fairly be said to be implicit in our ordinary or colloquial idea of what it is for something to be a sign, since we surely do not ordinarily think that a sign has done its job unless it has put us into experiential contact with the object itself.
      But thus far we have only the vague pre-theoretical idea of a sign, and it involves--besides the idea of the sign itself, the object which it is capable of revealing, and that which it is capable of revealing about it (i.e. its interpretant)--the further ideas of someone interpreting the sign, and of an act of interpretation. However, Peirce believed that, for theoretical purposes, it would be both possible and desirable to omit from the conception of the sign relation in its generic form any reference either to an interpreting agent or to an act of interpretation, without thereby denying that it might be pertinent in some special discipline or area of application to include these further conceptions because the subject-matter of investigation therein is specially conceived in such a way as to require it.
      In so conceiving the sign relation in its generic form, Peirce enormously broadened the scope of possible application of a theory of signs, for there are many phenomena which may be amenable to a semiotic that does not assume any interpreting agent but only makes available a way of describing analytically the various forms and possible modifications of such processes: for example, there are the various intra-organismic processes, which biologists tend to think of in terms of the transmission of information; there are socio-economic-political processes to which the idea of individual interpreting agents and their individual acts of interpretation may be impertinent; and one may even speculate on the possibility of a semiotical analysis of some puzzling physical phenomena, e.g. those arising in connection with quantum mechanics, in which the relationship of the observation to what is observed seems peculiarly difficult to get clear on because we have such difficulty in separating conceptually the observation in the sense of what is observed from the observation in the sense of the act of observing, and the latter from the observing agent. (Semiotic is clearly applicable to matters of scientific method, and one special characteristic of quantum mechanics is the peculiar way in which method and subject-matter seem to be logically inseparable at certain points.)
      But although Peirce himself was as aware as anyone of his time could have been of the possibilities of application of such a semiotic, his most basic motivation in abstractively refining the conception of the sign relation into three, and only three, essential terms was, I believe, that of allowing us to reconceive the mind and mentality in terms of the logical form of the process and content of thought as such, as distinct from regarding the mind as something which is doing the thinking, the aim being to eliminate thereby the need for invoking a Cartesian mental substance as agent, with all that this implies.
      There are, to be sure, a number of formulations of the basic sign relation and of more specific semiotical conceptions in Peirce's writings which do involve reference to an interpreter and/or to an act of interpretation, but--without arguing the point here--I will only say that I do not think one can understand Peircean semiotic well enough to make effective use of it if one does not realize that these formulations are either subject-specific, or else, more typically, are what he referred to trenchantly as "sops to Cerberus", intended to provide a temporary bridge between his view and the intellectual prejudices of his presumed readers when he was addressing a readership he thought would be unlikely to understand his conceptions initially if presented in their theoretical purity. (It should be borne in mind in reading Peirce that he--like Frege and Husserl, for example--was battling the rising tide of psychologism which has, since his time, so thoroughly engulfed us that it is difficult for us to realize the extent to which we have come to take its assumptions for granted. By "psychologism" I mean the treatment of a topic in such a way as would be appropriate to the special sciences of psychology, if such sciences were in fact sufficiently developed to treat such matters. The attempt to psychologize logic –- indeed, philosophy in general--was already well under way during the latter part of the 19th Century. Peirce's rhetorical concessions to this increasingly dominant but alien intellectual situation proved to be unwise: not much attention was paid to what he said, anyway, but such concessions have misled many readers ever since.
      In any case, the logical effect of Peirce's refinement of the conception of the sign relation into that of sign, object, and interpretant was that the agency in the sign-interpretant process was thereby shifted to the sign itself, which must now be thought of as generating or producing its own interpretants, each act of interpretant production being called a "semiosis". The deep prejudices of our own time make it difficult for us to understand immediately the full import of this, though. For we are accustomed to thinking that we, as interpreters and users of signs, have it within our power to give meanings to things by pure acts of will or intention: thus we talk about "conventions" of meaning or "conventional" meaning which we arbitrarily establish as if in imitation of a divine fiat; we think we are capable of simply "stipulating" what a word means at will; we talk about "inventing languages" which will somehow be free from the limitations inherent in the words we use in the process of inventing the new ones; and so on.
      Against this, Peirce's semiotical viewpoint implies that all we can really do, as interpreters, is to observe or note the meanings which things already have--the interpretants they can and do generate of their own accord, which constitute the objective content of our acts of interpretation--and that, as users, we can affect or change the meanings of signs only in much the same way we can change or affect the nonsemiotical properties of things, namely, by putting them in connection with one another in ways that may de facto result in changes in their meaning properties as they interact with one another. Thus the matrix of change in meaning--and of such participation in the creation of meaning as we are capable of--is to be found in the compositional process, not in acts of definition as such. It is doubtless true that some intelligible sense can be given to the idea of "establishing a convention" or "stipulating a meaning" or "inventing a new language", consistent with the Peircean view of the autonomy of the triadic semiosis process; but the sense which can be thus given to these ideas will not enable them to do the fundamental logical job which conventionalists commonly have in mind; for it can amount to nothing more in any case than a more or less successful experiment in juxtaposing pre-existingly meaningful entities in such a way as to produce such novel results as the nature of these entities themselves determines.
      In some ways, this conception of the autonomy of the sign is actually much closer to ancient and archaic ("primitive") ways of understanding meaningfulness or significance than it is to the modern view. Words, and other signs, were and are commonly regarded by non-modems as among the most potent of all entities: powers-that-be, which are not to be dealt with in the casual, and often highly irresponsible, manner to which we are accustomed. There is a child's verse--familiar to those who have grown up in the U.S.A., at least, though I have no idea where it originates--which expresses the remarkable naiveté of the modern mind concerning the power of words:

Sticks and stones can break my bones
But words can never hurt me.

What an idea! Childish, of course, and patently absurd when we reflect upon it; but in fact we commonly do both tolerate and practice abuse by words and abuse of words to a degree that a "primitive" would find astonishingly stupid, in view of their known powers. We smile indulgently at the "primitive" fear of word-magic, as if unaware that we ourselves "live, move, and have our being" within a social milieu in which word-magic and sign-magic generally--the arts of the politician, the "public-relations expert", the advertiser, the salesman, and so forth--so pervasively and so irresponsibly rule us that we find it difficult to believe in any social reality at all. The word "modernism" doubtless means many different things, but I suspect that what is at the bottom of the distinctively modern mentality is the belief that we can be the masters of everything, the mastery of words and signs being a relatively easy part of that (they seem so malleable), and the consequent disillusionment, cynicism, and despair which follow in the train of this megalomania. "He whom the gods would destroy they first make blind"--blind, that is, to the gods themselves, the powers-that-be that are superior to us, whether we are aware of that fact or not.

      To be sure, modernism has now largely run its course in the West, withering away even in the U.S.A., though what presently passes for "post-modernism" is usually just modernism deracinated, and its basic assumptions still occlude our vision of alternatives. I think myself that, whatever the merits and demerits of Peirce's sign conceptions generally may turn out to be, when and if they are ever put to the test of application and truly informed criticism, his attempt at a radical reversal of our understanding of the source and agency of meaning is the most profound attempt to date--the most far-reaching in its implications--at moving us into a genuine post-modernism which will, among other things, regain and stabilize the "primitive" insight that, like it or not, we are far more the pawns than the agents of meaning.
      The reader may think I have forgotten my topic: Peirce's conception of the iconic sign in particular. But the reason I have dwelled particularly on the point that all signs--including linguistic signs--are autonomous, in Peirce's view, is that the contrary view, that at least some signs--linguistic ones in particular--owe their meaningfulness or significance to a sort of magical human agency, has tended to expand both in semiotical theory and in interpretational practice to the view that all signs--including iconic signs--are "conventional" in character, and I think it important to emphasize that Peircean semiotic is so conceived from its logical beginnings that there is no way in which such a view can get a foothold in his theory. In fact, there is no basic conception in his theory corresponding to the idea of the linguistic sign, though it is often thought that his conception of the symbol is more or less the same as the conception of the linguistic sign. But this is not so: there are entities which we ordinarily regard as words which are not symbols, and there are symbols which can by no reasonable stretch of usage be rightly called "words".
      The conception of the symbol, along with the coordinate conceptions of the icon and the index, are not technical refinements of pre-existing ideas of more or less familiar classes of signs at all, but are rather conceptions which Peirce derived from a pre-semiotical basis in his theory of the "categories", and they can function in application only to isolate dimensions of the significance in things but not to classify or sort out things into distinct groups in the way that, say, the conceptions of an orange and an apple and a pear can be used to sort out a batch of mixed fruit into separate piles. Thus when we identify some sign as being iconic, for example, this only means that the iconicity of that sign happens to be of peculiar importance to us for some reason or other implicit in the situation and purpose of that analysis, with no implication to the effect that it is therefore non-symbolic or non-indexical.
      To get clear on this and other matters, let us go to the logical beginnings of Peirce's semiotic in his theory of the categories. The word "category" is perhaps best understood for present purposes as meaning a type of predicate or type of property. Peirce continued to believe throughout his long philosophical career that in his first substantive philosophical publication--the essay "On a New List of Categories" of 1867--he had succeeded in demonstrating that all possible properties of a thing (anything whatsoever, regardless of its reality status) are divisible into three types: monadic, dyadic, and triadic, his proof of this being to the effect that in thinking about it we necessarily suppose predicates applicable to it corresponding to properties of each type.
      (1) A monadic property is a wholly intrinsic property of a thing, which means that, in regarding a thing as having such a property, one is making no implicit reference to any second thing. Insofar as we regard something in this way, we are regarding it neither as existent or as non-existent, as real or as unreal, since to regard a thing only in respect of its monadic properties is to regard it as if it has no relationship whatsoever to anything else (including oneself), whereas the ideas of existence and reality pertain to things in their relationships to one another. "Firstness" is his name for the category of properties or predicates of this sort.
      (2) A dyadic property is a two-term relational property. To regard a thing dyadically--under the aspect of the category he calls "secondness"--is to regard it just insofar as it is in a relationship of affecting or being affected by something else: being in interaction with something else. In general, any property which at once distinguishes a thing from a second, thus involving opposition or otherness, while relating them as a pair nonetheless, would qualify as a secondness property.
      (3) The third category of properties--those which he calls "thirdnesses"--are the triadic or three-term relationships. Now, the category of thirdness is the specifically semiotical category, which is to say that the representation relation--the sign-object-interpretant relation--is the generic form of thirdness. In other words semiotic is conceived by him in its broadest scope as being the general theory of the category of thirdness, and his general conception of a thirdness relation is to be found in this definitions of the representation relation, provided the definition is of the requisite degree of abstractness (i.e. is not a "sop to Cerberus"). For example:

[A sign is] anything which determines something else (its interpretant) to refer to an object to which itself refers (its object) in the same way, the interpretant becoming in turn a sign, and so on ad infinitum. (Collected Papers, 2.303)

Before commenting on this further, though, several points about the categories in general should be made.

      First, the overt linguistic form of a predicate is an unreliable guide to the type of property it designates. For example, the linguistic predicate "x is red" stands for a monadic property if what is meant is simply a certain apparent or manifest color quality which the sighted can recognize but the blind cannot. (This is not, of course, a formal criterion: I am only intending to speak suggestively of what is meant by these different categorial conceptions.) But if "x is red" is scientifically defined in terms of the wavelength of the light emitted by it when irradiated from a certain sort of light source under certain conditions, then being red is a dyadic property, for we would then be implicitly regarding the red thing in a possible interactional relationship with the apparatus used in ascertaining the wave length. (I am reasonably certain that Peirce would also regard the property as dyadic if it were defined in terms of the relation to some standard color patch, because in such a comparison, the thing compared and the patch to which it was being compared could be thought of as being in brute causal interaction with one another--or at least this is the way people who work with colors and color relationships commonly think of color comparisons, namely, as involving visually perceptible interactions of the things in respect to their color. However, this sort of case is actually rather more complex than that, and there are other ways of construing the matter as well.) But if "x is red" means that the subject of predication is of a certain ideological persuasion, the property would be a triadic one, since the predicate in this case involves reference to a certain idea or ideal, and the most reliable heuristic principle one can follow, in practice, in determining whether or not a predicate is of the triadic type--that is, whether the property is triadic in logical form--is to ask oneself whether or not there is an implicit reference in it to something which is of the nature of an idea or ideal, since such entities can have only a representational mode of being.
      (To avoid a possible confusion here, I should remark that all predicates are representations: the question is whether or not the predicate represents something which is itself of a representational character. Again, this is not a formal criterion, but only a suggestion about how to ascertain what type of predicate one is dealing with.)
      Second, Peirce argued that there is no fourth or more complex category of properties, any apparent exception supposedly always being reducible in principle to some combination of the other three.
      Third, Peirce believed he had demonstrated that thirdness logically presupposes secondness, and secondness logically presupposes firstness, though not conversely. This does not mean there can be an object of experience which is solely monadic, or even one which is dyadic but not triadic, for he held that all three categories are applicable to every object which we can possibly experience. But the supposed fact that every possible object of experience has properties of all three types follows from what it means to be an object of experience, not from the presuppositional principles. (Peirce's category theory is a logical descendant of Kant's, in this respect, just as it is a logical descendant of Aristotle's in respect of its being a classification of the types of predicates.) The presuppositional ordering of the categories provides a most helpful structural clue, by the way, in understanding many of the systemic inter-relationships in Peirce's theory.
      Fourth, the fact that, according to Peirce, any object of experience necessarily has properties of all three categories does not imply that semiotic is the single all-embracing science; for our interest in a thing, scientific or otherwise, might be limited to its dyadic and/or monadic properties. It does mean, though, that there is nothing that cannot, in principle, be regarded from the semiotical perspective.
      Fifth, although it follows from the third point above that there is no such thing as something which is only a sign, it is worth emphasizing that everything which is a sign necessarily has non-semiotical properties as well. For some of the commentary on Peirce seems to involve an implicit assumption that, if something is a sign, then it somehow has a peculiarly insubstantial ontological status. This presumably originates in a misconstruing of Peirce's repeated insistence that all thoughts are signs (or that all thought is in signs), which some commentators seem to regard as meaning that signs are only thoughts, in some ghostly sense of "thought". As I remarked earlier, it is not easy for us to perceive immediately the real import of a radical inversion of our basic intellectual assumptions. Thus whereas Peirce was trying to develop a way of regarding things in such a way that we can dispose, once and for all, of the idea of "the ghost in the machine," to use Gilbert Ryle's felicitous phrase, he is understood as if he were saying that the machine is only a ghost!
      Returning now to the formal definition quoted above, it is important to understand that what is being defined is the representation relation, for although it is worded in such a way as to encourage one to think it is the word "sign" which is being defined, the words "sign", "interpretant", and "object" are merely labels for the three ordered terms of the triadic relation, which is, in this definition, being defined rather by use of the words "determines", "refers to", and "in the same way". (The phrase "ad infinitum" is redundant since the import of the definition is that the interpretant is itself another sign.) That is why there is no circularity involved in the implied fact that what a sign generates is another sign, for what this means is simply that anything that can play the role of the interpretant in one actualization of the relation can--and indeed must--play the role of the sign in a further instantiation of it. It is also implicit in the definition that the sign is itself an interpretant of a logically prior sign of the same object. Moreover, it is also implicit--if not readily apparent--that the object is itself a sign (which, to repeat a point made above, does not mean that it is nothing but a sign).
      I will forego explaining here why these implications are involved in order to focus attention rather on the import of the definition, which is that when we regard things from the semiotical point of view, we regard them as links in a chain of signs, which are always interpretants of logically prior signs, and whose interpretants are themselves signs, and so on, ad infinitum in both directions. But the chain can nevertheless be thought of as originating in an entity--the semiotical object--which, because it is itself of the nature of a sign (in addition to whatever else it may be), is capable of generating ancestrally the entire chain as a sequential manifestation of itself, and indeed is capable of generating any number of other chains which can be regarded as further sequential self-revelations.
      Does such a chain ever end? Can it have an absolute beginning? Such questions do not seem to me to be so important as they did when I first began trying to understand Peirce's theory, one reason being that the answers--predictably enough, I should think--take the form of "Yes, if you mean such-and-such; No, if you mean such-and-such," and are really only matters of drawing certain distinctions which, once drawn, cease to be of much interest, though it is necessary to be able to draw them. But let it suffice for present purposes to say that there is nothing paradoxical or even anything that strains common-sense in these matters.
      The second reason for these questions no longer seeming to me to be of paramount importance is that, when it comes to the application of the theory, the mind-boggling infinities involved in the abstract conception translate into practice simply as meaning that there is no such thing as a final or definitive or exhaustive or absolutely basic semiotical analysis of anything, but only analyses which are in fact sufficient to satisfy whatever more or less definite intellectual need or needs motivate one to undertake such an analysis to begin with. There can be final answers only to final questions, but there are no final questions in matters semiotical, anymore than there are final questions in physics or in any other field. For every question presupposes something unquestioned, to fix the subject-matter being put into question, and there is nothing unquestioned which could not, in principle, be put into question.
      Let us turn now to the divisions of types of signs in order to locate the iconic sign within this larger framework. (I will restrict myself to what seems minimally necessary.) The basic method of sign division is always the same: take something assumed to be a sign, consider it either as it is in itself, or in relation to its object, or in relation to its interpretant, and apply or reapply the categorial conceptions to it in one way or another. Given something which is a sign, we know--in virtue of the categorial "deduction" in the 1867 paper referred to earlier--that in addition to standing in a triadic relationship, it will also have dyadic and monadic properties, and properties of any of these types may have something to do with its particular sign-value, i. e. with some particular identity it has as a sign. (It can have many such identities, as well as many nonsemiotical identities.) Thus if we are interested in some particular semiotical identity the sign has, in virtue of a monadic property of it, we are, then, interested in it as a qualisign; if it is in virtue of a dyadic property of it, as a sinsign; and if in virtue of a triadic property, as a legisign. For example, if we suppose the following to be an assertion actually being made:

Because of the long religious fast, John was unable to stand fast or even to run fast when the enemy attacked.

then there is one qualisign "fast", there are three sinsigns of that configuration, and there are three distinct legisigns corresponding to those three sinsigns. There is one qualisign "the", there are two sinsigns of that configuration, but there is only one legisign embodied in those two sinsigns. If the first occurrence of "fast" were replaced by "abstinence from food" then the same three legisigns would be embodied but the one which was previously embodied in the first occurrence of "fast" would now be embodied in the synonymous replacement phrase. (The reader presumably gets the idea, but in order to be really precise on this, without being unduly prolix, I would have to make use of three distinct notational devices for referring to a sign as such, instead of making do with the usual single device of quote marks.) I define this trichotomy--which one should get clear on before trying to get clear on the icon/index/symbol distinction--in such a way as to make clear that one and the same actually occurring entity, functioning semiotically, can be regarded as being at once a qualisign, a sinsign, and a legisign, rather than defining it in the more usual way in terms of the distinction between a quality, an actually occurring or existing individual thing, and a law (rule, habit, custom, etc.), because a definition of the latter sort can be misleading if it is not understood that Peirce regards neither qualities (properties of the category of firstness) nor laws (relational properties of the category of thirdness) in logical detachment, as it were, from occurrent instances of signs.

      Thus as regards qualities, these are monadic properties of objects, and his way of regarding a monadic property is to think of the monadic predicate which properly expresses it in much the same way contemporary logicians think of what are usually called "propositional functions". Take, for example, a proposition like "John is red" (meaning his skin is red, as from a sunburn). If we eliminate the reference to John--and to anybody or anything else--what is left is something of the verbal form "x is red", which is a propositional function (Peirce's term is "rheme"), which may be thought of as a proposition whose subject term is devoid of any positive content. (Borrowing the terminology of mathematicians, contemporary logicians would speak of the sign "x" as being a "variable" and would regard "John"--or any other referring expression that could replace the "x"--as being a "constant.")
      Now, the way Peirce regards the contentless subject term (the "variable") semiotically is as being a sign so extremely indeterminate in its significance that it is impossible to identify the object in any way other than as being qualified by the property expressed in the predicate, so that what "x is red" stands for is simply an object whose only specified property is that of being red. In other words, what Peirce means by a quality is an object regarded as qualified only by a monadic property, as distinct from thinking of a quality as an entity somehow detached from any object whatever. Thus if we regard any object only monadically, we are ipso facto regarding it simply as a quality, which does not preclude the possibility that the same entity can also have existential characteristics as well (as indeed it must, though not necessarily existence in "the real world").
      Similarly with regards to the legisign. To speak of something as being a "law" or "rule" (or with the use of any of several other words more or less synonymous with "thirdness" for him) is to speak of a triadic relation, which can be construed dynamically as a general form of relating, and he conceives triadic relations as being located in their relates as active or passive powers of them. Thus he does not regard laws, rules, etc. as entities existing independently of any objects, somehow awaiting application to them when appropriate, but rather as objects regarded particularly in respect to some triadic ordering power (active or passive) which they have. The same object, then, which, from one categorial perspective, is a quality, and, from another, an existing individual, can also be thought of, from a third categorial perspective, as an embodied power. Reflection on the earlier example will indicate, moreover, that there is not in general any necessary implication running from the qualisign identity of a thing to its sinsign identity, nor from either of these to its legisign identity. On the other hand, given that a certain thing embodies a given legisign, that thing does have a certain sinsign identity in virtue of instantiating or actualizing the legisign, and is said to be a "replica" of it in that respect--though that same sinsign may also have identities as a sinsign which have nothing to do with its status as a replica.
      Now let us consider the icon/index/symbol distinction, which is derived by considering the sign in relation to its object. (More specifically, its "dynamical" object as distinct from its "immediate" object, the former being its object as it is in itself, regardless of any particular representation of it, the latter being the object as it appears within a semiosis sequence--i. e. as it is represented to be--for the purpose of referential identification. For present purposes we need not work with this distinction further, though, nor with a correlative distinction between three kinds of interpretants.) If the sign's representative clue is based on, or grounded in, a similarity (resemblance, Iikeness) to its object, then it is, of course, iconic. If it is based on a dyadic or existential relationship with its object, then it is indexical. And if it is based on nothing but the fact that it has the power to generate an interpretant sign of itself in which it will be interpreted as being a sign of that object--that is, if it is based on nothing but the fact that it has the power to generate an interpretant sign of itself in which it will be represented as a sign of that object--then it is a symbol. In the case of the index, the interpretant sign of it, as such, will represent it not simply as a sign of the object but as being in existential connection with it. In the case of the icon, the interpretant of it will, as such, represent it both as being a sign of the object and as being like it.
      This is to say, that in identifying something as an index or an icon, respectively, one is committed to certain things one is not committed to in the case of the symbol: in the one case, one is committed to the claim that there really is a certain dyadic or existential interactive relationship between sign and object; in the other case, one is committed to the claim that sign and object really do have a certain property in common (since Peirce construes resemblance, likeness, or resemblance as identity in respect to some property). Now to say that the sign and the object have a property in common or are identical in respect to that property does not mean that they are absolutely indistinguishable. Two objects with perceptibly distinguishable shades of red, for example, could nevertheless function as iconic sign and object relative to the property of being red, provided the difference was not great enough to be of any practical import in the semiosis process in question.
      An icon proper is always a qualisign--that is, it is always a monadic property of an actually occurring sign--though the sign embodying it can be called "iconic" (or a "hypoicon") in virtue of doing so. Thus, if we are regarding the actually occurring sign as something actually occurring, we are regarding it as a sinsign, and as an iconic sinsign in particular, if its iconizing function is what we are particularly concerned with in connection with it. Or if we are regarding the sign particularly as regards its power of generating an interpretant as such, then we are regarding it as a legisign, and as an iconic legisign, in particular, if we are especially concerned with its controlling role in iconization. Now, a sign embodying an icon proper is not necessarily an iconic legisign (that is, an icon proper is not necessarily "governed" by an iconic legisign), but, supposing that it is--in which case the sign as sinsign is a replica of itself as legisign--then a sign which interprets it as such (i. e. a proper interpretant of it) will represent it as being iconic of something which the legisign has determined the interpretant to represent iconically as well and in the same respect (and the interpretant's interpretant is similarly determined, and so on, ad infinitum). In general, to say that a certain actually occurring sign is a replica of the iconic legisign of something, is to say that the sign necessarily has an interpretant which represents it as doing so (and which itself represents that same thing iconically in the same respect), whereas, if it is not identifiable as being a replica of that iconic legisign, supposing it is fit to iconize that thing in that way, it may or may not actually have such an interpretant, and if it does, then it will be in virtue of the functioning of some other semiotical factor.
      One use of the conception of the iconic legisign would be in the description of semiotic processes in which a certain form or quality persistently maintains itself in the serial process, such that each successive iconic interpretant either leaves off, or acquires, something extraneous to the icon proper embodied in each member of the series (in virtue of which the series is iconically unitary), or else simply duplicates the prior sign in its iconic aspect. For example, in trying to understand a certain subject-matter--Plato's image of the cave in the Republic, let us say--one may have the initial conviction that it contains some formal structure of great usefulness, if one can effectively isolate what is essential in it, and one may then go through a process of "stripping away" from this image or iconic sign aspects of it which are not a part of the icon proper which one wants to isolate conceptually: thus, its being a cave is stripped away (omitted in the next interpretant), leaving the more abstract idea of its being a place where the sun doesn't shine; the idea of the sun is stripped away, leaving the more abstract idea of a source of light; the idea of visual light is stripped away, leaving the more abstract idea of a medium in which things can reveal themselves as what they are, and so on, until one is satisfied that the "essential meaning" of the image has been sufficiently isolated.
      At no point will the "pure" icon, the icon proper, ever be completely denuded or stripped to its essentials, in the sense that the iconic sign embodying it will contain no qualitative features extraneous to the icon proper. But it might become sufficiently "denuded" to be amenable to being "re-clothed", as it were, by being concretized in an image of, say, a passage from confusion to clarity, as Plato intended it, or of anything else which it would turn out to be fit to iconize effectively. (Note, though, that when I say "one may go through a process of stripping away from this image, etc.", I am speaking psychologistically: to speak more rigorously, I would instead have to speak only of the chain of semiosis itself as involving a continuing identity of the icon proper throughout each successive movement of the process, the extraneous factors being successively omitted, or added, without suggesting the idea of a mind external to the process which is doing the stripping away from without, as it were.) A rather different sort of semiosis process, for the description of which the conception of the iconic legisign might be of some use, would be any biological process in which the serial order involves the continual reduplication of some form or quality--icon proper--in varying guises throughout the course of the process, as, for example, in genetic transmission and in the growth of the individual insofar as he or she is controlled by genetic factors.
      In general, what we seem to have here, in the conception of the iconic legisign, is something rather like the ancient idea of the "formal cause", which originates with Plato--or perhaps earlier, with the Pythagoreans--and is expressed most emphatically in the Phaedo, where Socrates says that he believes that the best way of understanding why something is, say, kalos (beautiful, fine, noble) is to understand that it partakes of the kalos itself (that is, the eidos or form of the kalos), and so similarly for other attributes of a thing. The point to this is missed if one supposes that Plato was somehow mesmerized by a barren tautology in saying that, if anything is F, then it has F in it ("partakes" of F), whereas the point is to assert the importance of recognizing the role of the persistence of immanent form in the process of change. But however this may be, the conception of the iconic legisign is implicitly being appealed to whenever one regards a semiosis process as having an inherent tendency to maintain a certain formal or qualitative identity throughout the series of transitions in semiosis from sign to interpretant sign. Iconic legisigns will usually be functioning together with indexical and symbolic legisigns as control factors in semiosis processes, though, with the indexical legisigns functioning to maintain sameness of reference or origin throughout, and the symbolic legisigns functioning to provide the contours of the telic or purposive or directional aspect of the process. (The sorts of processes I suggested above as possible examples of the functioning of the iconic legisign also involve some indexical and symbolic legisign factors as well, though we will not go into the topic here of how these other species of legisigns function and how they can do so co-operatively.)
      It would be natural to think of a legisign as being a rule of interpretation to which an interpreter can appeal in identifying the proper interpretant of the legisign's replicas, and Peirce himself sometimes speaks of legisigns as being rules. But although it is sometimes helpful to think of legisigns in this way, it can also be misleading if one is not clear on the fact that a legisign does not preside over a certain subject-matter from without, as it were, but is rather an immanent principle of control within the subject-matter. (This is why I have insisted here that legisigns should be understood primarily as being active powers in actually occurring signs: insofar as a given sign is a legisign, it is a ruler, not something subject to rule.) There is nothing about any given sign considered in itself--no intrinsic marks or characteristics--that could be sufficient to warrant identifying it as replicating a given legisign, and therefore one cannot ascertain what the proper interpretant of a sign is simply by appealing to a rule under which that sign can be shown to fall. The justification for identifying a sign as a legisign (or, what comes to the same thing, as a sinsign replicating a legisign) lies rather in global characteristics of the semiosis process of which it is a part, taken in conjunction with the general heuristic principle--implicit in any sort of research, semiotical or otherwise--that one is to so construe the phenomena as to make it maximally intelligible.
      Thus in the case of the iconic legisign in particular, if the elements of a given process or serial order of things linked by sign-interpretant relations appear to display a tendency to replicate persistently a common form (perhaps in various extraneous guises), then one will have reason to suppose that it may be appropriate--that is, intellectually profitable--to identify the signs as, among other things, iconic legisigns, transmitting successively not only the common form (the icon proper) but also the property of being iconic legisigns, since it is the transmission of this property from sign to interpretant sign which constitutes the persistency in the recurrence of the common form. However, insofar as the chain of semiosis is controlled only by an iconic legisign, it will probably have a certain random or "wandering" character, inasmuch as there is nothing operative to correlate the persistent identity of the property iconized with a persisting sameness of indexical reference. For example, a rumor about someone may "travel" in such a way that who that someone is keeps changing in successive versions of the rumor (whereas if the person stays the same but the story keeps changing, it would be a case of control by an indexical legisign, without any correlative control by an iconic legisign). Or again, a certain mythic form may keep recurring with changes in the protagonists. Another possible example that comes to mind is the tradition of interpreting the icons in the Book of Revelation: the referents of these images changes from generation to generation, as the supposed predictive power of that curious work keeps failing to demonstrate itself.
      Now, the material mechanism which accounts for this property of a process is quite another matter: a semiotical description of a process does not in itself explain the process in the same sense in which, say, an account of the physical interactions and forces involved would explain it. There is surely a sense in which a semiotical description is explanatory, but it does not compete with or replace explanations in non-semiotical terms, supposing such explanations are available and desired (as they may not always be). Thus, for example, a semiotical description of the mode of functioning of a computing machine would say nothing one way or another about the physical processes that actualize its operations. If, say, a computing machine were constructed out of levers and gears and the like that exactly duplicated the functioning of the nervous system of a gnat, the semiotical description of both gnat and machine in this respect would be identical, though the explanation of how the machine worked, in terms of levers and gears and the like, would be quite different from an explanation of how the gnat's nervous system worked, in terms of the electrical and chemical processes involved. The general point here is that the reason for using the concept of the legisign in a semiotical analysis is not to invoke something which accounts for this or that feature of the process under analysis (such as, say, the persistence of a form), but rather to be able to describe the process in such a way that the feature is effectively represented in semiotical terms, for whatever further purpose that may serve.
      But not all iconic signs are iconic legisigns, and I wish now to turn to another important aspect of iconicity, namely, the relation of the icon proper to that which it iconizes. It is important to bear in mind at this point that, although the icon proper is only a qualisign, and therefore never occurs apart from a sinsign embodying it, it is nevertheless possible to consider it in abstraction from its sinsign embodiment, as an entity thereby considered as if unrelated to anything else--except, of course, its semiotical object. But then the sign-object relationship is a very special one in the case of the icon proper. For just insofar as its object is considered only as being the object of the icon proper, it, too, is being considered apart from any relation to anything else; and since. moreover, any properties it has which are not represented by the icon proper are also being conceptually ignored, it follows that there is, insofar, no way to distinguish the object from the sign. In other words, the relationship of the icon proper and its object, just insofar as it is its object, must be one of identity; for the iconic relationship is grounded in a property identity of sign and object, and we are, ex hypothesi, prescinding from every logically extraneous property in considering them simply as icon proper and object. This peculiar sign-object identity has far-reaching implications.
      An accurate individual roadmap of a highway system can function as an iconic sinsign of that system; and although the qualisign which is the configuration of the lines on the map will not be exactly like the configuration exhibited by the highway system itself (since the map will not show every slight curve in the highway, for example), the difference is of no importance as regards the respect in which the user of the map is interested: it is enough if, say, the map-qualisign is sufficiently like the highway configuration to enable the driver to ascertain where a turn onto another highway is to be made. Map and territory are--to the extent that the map really does map that territory--identical in the relevant respect, though not, of course, in all respects. As I remarked earlier, the qualisign which is the icon proper need not be absolutely indistinguishable from the property of the object which it iconizes, in order to be semiotically indistinguishable from it. Hence, there is a genuine--if only partial--identity here, from the semiotical point of view.
      There is nothing counterintuitive in recognizing a real identity here: on the contrary, it is surely just what we ordinarily take for granted when we point to a map and say something like "You go from here to here, and then you turn this way, continuing in this direction until you come to here and . . ." It is surely far more commonsensical--and theoretically felicitous--to hold that the demonstrative pronouns being used are directly designating the territory itself, when they are used to designate the points on the map representing those places on the territory, assuming the map truly is iconic of the territory in the relevant respects.
      But the identity of iconic sinsign and object need not be only partial; for one could, for example, use the territory itself as map of itself by flying over it and observing the highway configurations in just the same way and for just the same purpose as one would look at an ordinary roadmap of it before starting out. Or again, the pen presently before me on my desk is like innumerably many things in one respect and another (e.g. Iike a rocket in shape, like a certain automobile in color, like several other pens which I own in a great many respects, and so on), and one of the things it is like is itself. Hence it can be regarded as iconically self-revelatory: immediately presentative of itself, though nonetheless still representative in character. But this is true of anything one perceives, which means that a semiotical epistemology would seem to be free from the main problem traditionally associated with the doctrine of representative perception, namely, that the representation ("idea") of the object seems necessarily to occlude the object itself, making of it an unknowable Ding an sich. For the object can be regarded as immediately present, in the experiential sense of "immediate", while being regarded at the same time as mediated logically through a sign. (We will be returning to this topic shortly.)
      As regards any difficulties in the self-referential aspect of self-representation, they can perhaps always be resolved by treating a given entity at one moment as an iconic sign of itself at the temporally adjacent moment--which can be as brief as you like (e. g. more brief than the time involved in perceiving something)--by making a division of it, for theoretical purposes, into two distinct entities at the two moments; for to make such a theoretical division does not entail making any division of it for any practical purposes. Thus the pen I immediately perceive before me can be regarded as two distinct sinsigns at any two distinct moments, for the purposes of epistemology, but this does not preclude my counting both pens as the same pen for any other purpose. (The concept of self-representation is far too important to be dealt with in such a summary fashion, though, and these remarks are intended as suggestive only. The most profitable resource for investigating the concept of self-representation semiotically would perhaps be found in the legal tradition, particularly in the history of the development of procedural rules governing the representational role of lawyers.)
      Perhaps the most important thing to understand about iconic signs generally is that their distinctive function is to make their objects immediately available as they are in themselves, in this and/or that respect (perhaps, even, in every respect), which is quite different than that served by indexical or by symbolic signs. (This is their function from the point of view of text analysis; from the point of view of process analysis, iconicity is duplication or replication, partial or total.) Unfortunately, the word "symbol" has been and still is used extensively both to refer to signs regarded as "conventional" and to signs functioning primarily as iconic (as is usual when literary, mythic, and religious "symbolism" is spoken of); nor is it uncommon to find uses wherein it is impossible to say what is intended, as well as uses that conflate these considerations incoherently.
      I am not thinking here of semiotic theorists who contend that all signs are conventional, but rather of writers in various fields who have no theoretical ax to grind in this connection but who treat the phenomena with which they are concerned as exhibitive in semiotical character while at the same time talking of "rules", "codes", "decipherment", and the like, which are expressions appropriate rather to so-called "conventional" signs. Roughly, whenever recourse to a code or rule of interpretation is required in order to understand what a sign means, one is insofar dealing with something symbolic (in Peirce's sense of "symbolic"), whereas insofar as something is iconic, it represents (exhibits, displays, shows) in itself some relevant feature or features of its object, so that it is a matter of perceiving the relevant properties of the object--indeed, perceiving the object itself, partially or wholly--in the sign rather than of moving from the sign according to some rule which "arbitrarily" correlates sign and object. It is certainly true that in order to perceive the object in the sign one often must have learned how to perceive a sign of that sort, and this may involve following culturally funded rule-like perceptual techniques. But what is pertinent is not what is involved in learning how to interpret signs, but rather what the role of the sign is in the revelation of the object. Is it capable of standing in place of its object--of being "as good as" its object--for some purpose, because it is sufficiently like it in the relevant respect? If so, then it is, to this extent, capable of being iconic of it.
      What tends most to confuse the matter is perhaps the fact that the iconic aspect of a sign is always only one of the semiotically relevant factors involved in any semiosis process actually under analysis (or in a text being explored for its semiotical characteristics). By itself, the iconic sign is mute, in a certain sense: in and through itself it reveals, displays, exhibits, shows forth, manifests, presents, presences, makes something immediately available (if only partially), and so forth, but simply as iconic it neither indicates which object it is, nor says anything about it. It can be learned from but it doesn't teach. This is why, for example, the Platonic Socrates could deny being a teacher: he made of himself an iconic sign of a certain form of life (or, if you prefer, Plato made that of him), and although that talkative man could hardly be said to be mute without qualification, he was indeed mute when it came to making explicit claims about his own significance--until the time of his trial, when, without ceasing to be an iconic sign (of philosophy), he also said explicitly what it was that he iconized. Moreover, Plato's dialogues are, in the same qualified sense, mute, as are all dramatic exhibitions as such.
      On the other hand, it is also true that the circumstances and manner in which something is exhibited can fairly be taken as equivalent to a saying of something, which is why commentators on Plato can reasonably (though often rather naively) talk about what Plato said, claimed, urged, argued for, etc., in his dialogues, even though he never says a word in them in propria persona. Similarly, Socrates can also be regarded as implicitly having made claims about his own significance as an iconic sign long before the trial, given the way in which he persisted in forcing himself upon the attention of others. Moreover, in an attenuated sense, at least, every iconic sinsign--even an inanimate sign--can be regarded as asserting itself, simply in virtue of such things as "standing its own ground" in an existential field, in generating an interpretant, in impressing itself upon the attention, and so forth. Since saying, asserting, claiming, etc., essentially involve indexical and symbolic aspects, this means that iconicity always occurs in connection with indexicality and symbolism, at least in weak form, and Peirce did in fact claim that all occurring signs involve all three functions to some degree.
      It is sometimes argued that the iconic is a special case of the symbolic, on the ground that a likeness between two things is really a matter of "convention", the further premise being that the symbolic can be equated with the conventional. But this will not do. For even if it were granted that likeness is a matter of convention--which I would not myself grant, except in a highly qualified way--it still would not follow that iconicity is conventional. The mistake here is in identifying iconicity and likeness; whereas their relationship is not identity but presupposition: iconicity presupposes likeness of sign and object, but likeness is not itself a semiotical relation. Thus even if likeness were a matter of convention, it would still be a matter of fact and not a matter of convention that a given entity functioning as sign conformed to that convention and is, therefore, fit to be iconic. The conception of iconicity does not itself presuppose anything about what constitutes likeness. Moreover, the symbolic and the conventional cannot be identified, in any case, at least as regards the way in which Peirce construes the nature of symbolism.
      Most discussion of iconicity to date has concentrated chiefly upon such things as maps, portraits, diagrams, physical models, photographs, and the like, rather than upon the iconicity involved in the perception of all objects. Indeed, one gets the impression from some writers that these sorts of objects are supposed to be somehow especially paradigmatic of iconicity. In fact, however, these special types of iconic signs are special only because their very names--"map", "portrait", "photograph", and so on--analytically imply their iconic function. They are not in any other sense paradigmatic, and they have no peculiarly central role to play in semiotical analysis. What these sorts of iconic signs do have in common, though, is that the use of them as iconic signs supposes that they have themselves been immediately perceived as sensory objects in their own right prior to their use as representative of something else. This means that they are both self-representing and other-representing iconic signs, which is to say that the icon which they embody is doubly referred, to themselves and to some other object as well. That is, in such cases the icon is to be thought of as being truly representative both of a certain sensory object immediately perceived (for example, the map) and of some object not in immediate perception (for example, the territory).
      The term "immediacy" is prone to ambiguity in this connection, however. Peirce says:

Icons are so completely substituted for their objects as hardly to be distinguished from them. Such are the diagrams of geometry. A diagram, indeed, so far as it has a general signification, is not a pure icon; but in the middle part of our reasonings we forget that abstractness in great measure, and the diagram is for us the very thing. So in contemplating a painting, there is a moment when we lose the consciousness that it is not the thing, the distinction of the real and the copy disappears, and it is for the moment a pure dream--not any particular existence, and yet not general. At that moment we are contemplating an icon. (Collected Papers, 3.362)

Thus there is a real sense in which even the other object--for example, the territory represented by the map--can be said to be immediately perceived, namely, in the sense that if the representation is a true one, the relevant properties of the represented object are just as immediately present in consciousness as are the properties of the representing object since they are, ex hypothesi, the same properties. Yet, on the other hand, there is also a sense in which the represented object is only mediately perceived, inasmuch as it is not sensorily perceived but is perceived rather through the mediation of something which is sensorily perceived.

      But then there are cases--quite common--where it is not easy and perhaps even impossible to say whether a given object is iconically other-represented or self-represented. Consider, for example, the case of the television coverage of some current event--a football game, for example--transmitted and reproduced on the viewer's screen as it occurs. Suppose television technology were improved to the extent that the image on the screen had a three-dimensional appearance, natural color, sharp definition, and were large enough to encompass the total field of vision. In spite of the fact that the perception of the event via the screen might be experientially indistinguishable from the visual perception one would have sitting in the grandstand, the screen would nevertheless be an other-representing iconic sign of the football match itself. But suppose technology were also to make it possible to dispense with the screen in favor of some apparatus which, let us say, connects with and stimulates the optic nerves directly in such a way that qualitatively the same visual experience occurs as would occur if one concentrated one's attention on the television screen--or, as far as that goes, if one simply watched the game from the grandstand. What would the iconic sign be in such a case? Would we say that the football match was immediately perceived, since there was no mediating screen image? The perception would certainly be physically mediated through the television apparatus, eyeballs, optic nerves, etc.. But then physical mediation (light waves, eyeballs, optic nerves, etc.) is present even in looking at the football match from the grandstand. Surely mediation in the logical sense is quite different from mediation in the physical sense. But if we were to say that the perception is logically immediate in the case of the eyeball-attachment device, why should we not also say that in the case of seeing the event on the television screen? The experiences are by hypothesis qualitatively identical in the two cases, the difference Iying simply in the physical means by which they are brought about. Other examples, not of a fantastic sort, which might be worth pondering would be, for example, seeing something in a reflecting telescope, seeing something in a mirror, seeing something through eye-glasses; hearing a sound through a loudspeaker, through a megaphone, through a hearing-aid, and so on.
      What all this suggests is not that there is no real distinction between an other-representing and a self-representing iconic sign, but rather that the distinction is perhaps not so obvious as it may appear to be prima facie, and that there will be a range of borderline cases not clearly classifiable as being of one or the other type. It also suggests that the term "immediate perception" may be misleading. There is no immediate perception in the logical sense of the term "immediate". All perception is mediated in the sense of being representative. Hence what Peirce calls--and would usually be called, or regarded as--"immediate" perception would be better called "direct" perception, which differs from "indirect" perception as self-representation differs from other-representation. An adequate philosophical development of this would involve working with the coordinate conceptions of the indexical and symbolic signs, though, and with other distinctions of sign-type as well--assuming one wanted to develop it in a thorough-going semiotical way--and we cannot venture into such further considerations here. Let it suffice for present purposes simply to note that while the iconic function can be "prescinded" from--that is, Iegitimately considered in abstraction from--the indexical and symbolical functions, at least sufficiently for us to be able to see the logical job which the iconic sign is doing in this connection, no actual case of iconic representation can be adequately or fully explicated without articulating the other types of representation which are functioning co-operatively with it.
      Now let us consider the case of memory awareness, that is, the recall of some event in one's past experience: not the propositional knowledge that such-and-such occurred, but rather the experiential recall in memory of some part of one's past experience. This would be a species of perception, in the broad sense of "perception" pertinent here. Supposing the attempt to recall is successful, we may say of memory just what was said of sense perception: the intended object (that is, the remembered event itself) and the iconic sign are at least formally identical. Are they materially identical or diverse? Since a memory "image" is not a material entity, I see no basis for positing material diversity here and, therefore, no reason for not saying that the object of which we are aware in this sort of memory is the very past object or event itself. In this sense, memory perception of this sort is direct and so we have direct as well as evidential access to the past.
      One consideration (rather weak, I should think) which might be thought to militate against this would be the fact that one is only aware of a very limited number of aspects or characters of the object in memory. But then sense-perception is highly schematic or abstract in that sense as well: in fact, a memory awareness of an event can be more accurate and comprehensive than the original sensory perception of it (as is sometimes the case when the sensory experience occurred under conditions of stress or great excitement, for example).
      Another objection might be that the time of the object or event remembered is always different from the time of the remembering of it. But then this is true of sensory perception as well. For example, the date of the sensory experience of a nonsolar stellar phenomenon is always at least four years later than the date of the phenomenon itself, and, indeed, one can sensorily perceive stellar phenomena which occurred long before one was born. In any case, all sense perception involves a time lag due to the mediating function of the nervous system, which, however brief it may be, still determines different dates for the experience and that which is experienced. Since, strictly speaking, we never perceive anything through our senses but the past, the mere fact that the intended object is in the past cannot itself be an objection to saying that veridical recall is direct perception of the recalled object.
      Finally, as in the case of sensory perception, we may, if we like, speak of the memory content as being an iconic sign of the intended object (the event remembered), and thus we may speak of the successfully remembered object as a self-represented iconic sign. In general, sense perception and memory perception (of the sort in question) are essentially the same as regards the iconic representative function. There are other important differences between them, of course, but the important point for our limited purposes here is that the combination of the doctrines of representative and immediate (direct) perception can be seen to apply to both of them in the same way. I think it can also be seen that the perception of the future--that is, the foreseeing, as distinct from the foresaying, or prediction--may be amenable to the same sort of analysis, though, of course, there are important differences to be explicated as well, and I don't mean to suggest that there will be no difficulties in working out such an explication in a rigorous way.
      From the epistemological point of view, then, the concept of iconic representation can be understood as functioning to synthesize the traditional conceptions of representative and immediate (direct) perception into a single new conception which cannot strictly be identified with either of its ancestors in their traditional form, but which manages to incorporate the basic rationale of both. Historically, the basic epistemological function of the conception of representative perception has been that of accounting, in a general way, for the possibility of error, but it has characteristically done so at the expense of leaving the object as an "I know not what". The doctrine of immediate perception, on the other hand, has the virtue of avoiding the latter problem, but it is faced with the difficulty of explaining how error is possible, if our perception of the object is not mediated by anything which can be misrepresentative. Peirce's conception of the iconic sign appears to retain the virtues of both doctrines while eluding certain problems connected with each of them, though of course there are many further questions which must be raised in assessing the epistemological adequacy of this unusual synthesis.
      I would like finally to draw attention to a closely related problem that has arisen in the interpretation of Peirce's thought to which the foregoing considerations are directly pertinent, namely, the problem of the relation between the ideal "ultimate opinion" in terms of which Peirce sometimes defines or characterizes truth, and the object of that opinion, the reality it represents and reveals. This object is called "the real (or dynamical) object", and the question is whether or not it is something other than that opinion, and if so, in precisely what sense. This question has been implicitly answered, I believe, since Peirce holds that all cognition is perceptual, in the sense that it always involves (logically, not psychologically) an iconic presentation of the cognized object. For an opinion, semiotically construed, is a symbolic entity, and Peirce held that

every symbol must have, organically attached to it, its Indices of Reaction and Its Icons of Qualities (Collected Papers, 5.119)

And that:

A symbol is a sign naturally fit to declare that the set of objects which is denoted by whatever set of indices may be in certain ways attached to it is represented by an icon associated with it. (Collected Papers, 2.295).

The pragmatic maxim is, I take it, a sort of corollary to this. But, in any case, what is important here is that, in view of the epistemic function of the icon, this means that the real object is in the ultimate opinion--and in any actual opinion just insofar as it is identical with the ultimate opinion--in the sense that it is representatively present in the iconic dimension of the opinion. In the case of the comprehensive ultimate opinion--"the phantom ultimate issue of the aggregate of all possible inquiries" (Collected Papers, 8.104)--the real object is wholly self-represented iconically since nothing falls outside the scope of that opinion and there can therefore be no indexical reference to a materially distinct entity external to it. (Indexical reference always occurs within an existential order.) In the case of any particular and actual opinion the object may be either self-represented or other-represented, but it follows from the above that, for any object, there is always a possible true opinion in which that object is iconically self-represented or "directly" present. (This may not be a possibility for you or me, however, given our human sensory limitation.) In short, all objects without exception, are directly perceivable in principle, and of course many of them are in fact directly perceived, in the sense given to that here through the notion of iconic self-representation.

      In view of this, we clearly cannot say that the real object is other than the final opinion; for an opinion, semiotically construed, is a complex representation relation involving symbolic, indexical, and iconic functions, and the object is in that relation in its iconic part. On the other hand, as a logical part of the opinion, it is not identical with the opinion itself. But I have discussed this matter elsewhere in connection with certain matters of textual exegesis (in my paper "On the Epistemic Function of Iconicity in Perception", in Peirce Studies No. 1), and there is no reason for me to do so again here.

END OF:  Ransdell, "On Peirce's Conception of the Iconic Sign"


Do you think the author has it wrong? If so and you want to contribute a critical comment or commentary, brief or extended, concerning the above paper or its subject-matter, or concerning previous commentary on it, it will be incorporated into this webpage as perspicuously as possible and itself become subject thereby to further critical response, thus contributing to Arisbe as a matrix for dialogue. Your contribution could also be of the nature of a corroboration of the author, of course, or be related to it or to some other response to it in some other relevant way.

The URL of this page is:

Ransdell's home page is:

This paper was last modified November 27, 1997 — J.R.
This page modified for formatting and site information by B.U. August 13, 2012 — B.U.

Queries, comments, and suggestions regarding the website to: