Genuine Triadicity in Computation, Cognition and Consciousness

1 Introduction

At least since 1992, when I prepared an editorial review of Varella and Maturana (1979) and Varella, Thompson, and Rosch (1991) for The Semiotic Review of Books (Lidov 1992), and first encountered notions of autopoiesis and of the organism as a system for approaching homeostasis of information, and perhaps earlier after studying Ray Jackendoff’s demonstration (1987) of the apparent redundancy of consciousness in physical relations, I have wondered whether semiotics might identify particular structures or activities associated exclusively with consciousness. The question seemed to me impossible to pursue, and I doubt it can be securely resolved. This paper explores this question indirectly, but its central concern is to advance a particular conception of the factor in a sign relationship that Charles Peirce called the Interpretant.

I will have little to say about consciousness itself. Consciousness is a space wherein various experiences occur. Beyond signaling here and at the end an interest in that space, my efforts are directed at the analysis of particular experiences – the experience of using signs – that may imply structures that require consciousness as their space of action and that may not be present elsewhere. To such mentally active persons as semiotics scholars, plagued by what our yoga teachers call “monkey brains,” it might seem that all experiences are sign experiences. I think not. It is helpful to remember, even if one thought usually leads to another, that we are quite capable of distinguishing the experiences of qualities and even of certain events from the experience of signs that result from them or give rise to them. Meditation and certain art forms aim with some success to provoke and sustain such non-semiotic experiences. Keeping those experiences off the table for now provides a methodological simplification, and so does my very broad construal of consciousness. For present purposes, consciousness can include: items ^[1] in attention, items available in principle to attention, awareness and the contents of awareness, reportable and non-reportable experiences, human and non-human experiences, forgotten and/or repressed experiences that might be retrieved. What I mean to exclude from consciousness would be such matters as the “experience” of the retina before its information is processed in the brain. The quotation marks are to remind us that in this context, that word may be as vacuous as it would be in talking about the “experience” of a billiard ball when struck by another.

Much of the viewpoint proposed here appears summarily in Elements of semiotics (Lidov 1999/2017^[2]) though that book is primarily oriented to general semiotics while this essay will impinge a bit more on the philosophy of semiotics.

Acknowledging the risk of excessive pre-ambling, it might be useful to some readers that I clarify how I distinguish general and philosophical semiotics and my bias as an author in this regard. I take the task of general semiotics to be the construction of a comparative perspective that helps us understand how representational or potentially representational media differ from each other with respect to their capacities for conveying structure and/or representation, “media” used very broadly here to include graphic arts, written or spoken words, gesture, mathematics, perfume, television advertising, and many others you might think of. Inevitably, general semiotics depends on, informs, and overlaps with philosophical semiotics, where the emphasis falls on questions like: What is representation? or How might we most usefully define “sign”? Although this essay could be said to touch the comparison of media, in this case, say, enzymatic biochemical reactions as one medium as against conscious images as another, it would be a mere sophistry to make much of that, for the question will be whether or not the former are signs at all given a useful sense of “sign.” The distinction I am drawing is rather like those between physics and metaphysics or between psychology as a practice and the philosophy of mind. It is a distinction between domains of enquiry, not persons, for a person may do both. If I, as a writer on semiotics, have any bit of expertise at all, it lies more with the “General/Comparative” than with the “Philosophical.” Should I sometimes seem too hasty in tangling with writings reflecting philosophical experience and historical knowledge that is deeper than mine on the opposite side, I can only hope that the possibility of dialog justifies my risk of foolishness.

2 A preliminary notion of triadicity

Charles Pierce’s (primarily philosophical) semiotics serves as our jumping off point. The lynch pin of my proposal is a tighter restriction on Interpretants than he makes. Peirce is not isolated in pointing at these things he called Interpretants. Frege and others, though construing them in different ways and different vocabularies, point in a like direction. as does the post-structuralist notion of différance. There is a reality behind these construals, whether mind-dependent or not. Though the choice may be partly arbitrary and partly sentimental for me, my inclination to start out with Peirce’s work is also convenient both because of the structure of his proposals and because his work is widely adopted by authors associated with biosemiotics, an intellectual movement some of whose fundamental premises I interrogate here. In our recourse to Peirce, I have the impression that a number of semiotics writers approach his continually evolved ideas as if we were obliged to either accept or reject them whole cloth. Peirce’s notions, developed throughout an extended career, are not systematically unified enough to motivate an all or nothing choice. In any case, such an attitude is patently wrong and would have been disdained by him. Regarding the Interpretant, Peirce was reporting on his observations, inventing a concept, but describing a category that has a reality prior to his description. In that framework we should say: Maybe he got it partly right and partly wrong. I certainly do not want to treat Peirce’s formulations as infallible gospels, but I am very attracted to the preservation of some of them, and one such attraction provides the basis of this paper. This one that I very much desire to hold on to is his idea that semiosis entails an irreducible, triadic relation. My arguments, in a nutshell, are first, that there is a large, important range of operations on information evident in biology, sometimes referred to as semiotic, that do not meet this criterion and need not be considered as semiosis and second, that a semiotics so constructed as to exclude these gains in clarity and purpose.

Here, then, a first source in Peirce:

A Sign, or Representamen, is a First which stands in such a genuine triadic relation to a Second, called its Object, as to be capable of determining a Third, called its Interpretant, to assume the same triadic relation to its Object in which it stands itself to the same Object. The triadic relation is genuine, that is its three members are bound together by it in a way that does not consist in any complexus of dyadic relations. (1896)

Peirce’s insistence that a sign, to be a sign, must engage a genuinely triadic relation, one that cannot be reduced to a complex of dyadic relations, is not, by his own standards, a very clear idea, yet, it feels to me like it almost hits a nail on the head. My work, to get it a bit straighter, might be inspired by an earlier explanation of his:

[…] representation necessarily involves a genuine triad. For it involves a sign, or representamen, of some kind, outward or inward, mediating between an object and an interpreting thought. Now this is neither a matter of fact, since thought is general, nor is it a matter of law, since thought is living. (CP 1.480)

But my task is made more problematic by the ending of the paragraph from which I quoted initially:

A Sign is a Representamen with a mental Interpretant. Possibly there may be Representamens that are not Signs. Thus, if a sunflower, in turning towards the sun, becomes by that very act fully capable, without further condition, of reproducing a sunflower which turns in precisely corresponding ways toward the sun, and of doing so with the same reproductive power, the sunflower would become a Representamen of the sun. But thought is the chief, if not the only, mode of representation. (Peirce 1903: 272–273)

In biosemiotics, the sunflower as a representamen of the sun might receive a warm welcome, and its capacity to respond to sunlight by reproducing itself might well be considered to instantiate sign-action. I want to exclude the sunflower, not a priori and not by vague notion of what is or isn’t “mental” but as the consequence of an earlier principle. The territory I will explore is indicated by Peirce’s passing remarks that “thought is general” and neither a matter of fact or of law. Building on that indication, I offer the following as a preview; it will take further preparation to spell it out. “Nebulous” replaces “general”:

A sign instantiates an irreducible triadic relation in which the third factor, the interpretant, in principle nebulous, specifies the relation (the reference) of the representamen to the object thereby characterizes the object in a further manner conceptually and/or energetically.

Here, I took out the capital letters for “sign,” “object,” and “interpretant” as a signal: Although my derivation from Peirce is obvious, I follow the rules for usage of these terms developed in my Elements, not always identical to his. For present purposes, I don’t think the differences will sow any confusion.

I hope it will become reasonably clear below what I have in mind with “reducible” and “irreducible,” but we have a bit of preparatory business to work through first. Toward that end, I ask the reader to bear with a detour which has the simple purpose of furnishing an alert regarding some difficulties that might be inherent in the notion of a “genuine triadic relation.”

3 Potential difficulties of “triadicity”

Peirce’s formulation draws on the vocabulary of mathematics or mathematical logic. Peirce, himself, is credited with founding contributions to the algebra of relations and clearly was passionate about abstract and axiomatic treatments of his ideas. We can assume he had mathematical notions of relations in mind, but these are hard to be certain about. In mathematics, relation appears as a primitive, that is, undefined notion, and relations are sometimes indicated abstractly simply as a collection of ordered or unordered pairs of terms, or, for tri-part relations, ordered or unordered triples. What Peirce had in mind, a historical question, is not the problem I address here. What “genuine triadic relation” might productively indicate to us is the only question I venture.

Consider the relation evoked by the preposition “between.” I have seen “between” taken as an example of an irreducible three-part relation, but whether it is or not depends on initial premises or the context where the word appears. Let us say for example, that Bill is standing before us between Jim and Jane. Then, we might reduce the “between” to: Bill is to the left of Jane and Bill is to the right of Jim, just two dyadic relations, at least on first view. “Between” here might be merely an abbreviation of the same data. But perhaps the intended reference is not to spatial position but to their ages or their net worths. We can only reduce the preposition, “between,” to a complex of dyadic relations if we know that the preposition’s grammatical objects (or, using an alternate technical jargon, the “arguments” of “between”) Bill, Jim, and Jane are collinear with respect to some sort of line or series, in our case a straight or perhaps a curved line in visual space or in ages or in net worths. If that is given in advance, yes, the reduction stands, but it leaves another possibly irreducible relation in its place. Is “co-linear” an irreducible three-part reduction? I once saw the affirmative assumed but lack enough context to know the right answer. However, if I take an elegant, white dress (sign or representamen) to represent a wedding (object) as a traditional and in some sense sacred ceremony, the interpretant, constituted here by the adverbial “as” phrase, establishes a triad in which the three constitutive pairs do not function independently to yield the same result as the triad. This example is not a definition, and yet, it may convey a “general idea” what irreducible ought to mean in our context. In what follows, I will continue to use the adverbial “as” to point out the interpretant function, and I hope “general idea” will be or become a well-functioning term.

There is a rather simple confusion – readily available to us – which one might trip on when trying to distinguish two-part from three-part relations.

For the notation or description of any two-part relation, we need to identify three elements: two relata and the relation itself. The notation or the description are, minimally, three-part signs (that is, three-part representamina) though in some cases so automated that the interpretant is not present in consciousness. Yet, prior to regarding their sign character, note that the presentation, instantiation, description or notation of a two-part relation must have at least three elements. The three parts of (a, b) are a, b and (,). Three further dyadic relations: (1) “Bill loves Mary.” It takes three words to say it. (2) “Quite a gardener, your grandpa!” Here the coma separates the two relata while the syntactical order, marked as exceptional in English, codes the silent copula. (3) If I place Tom and Jerry before you standing back-to-back with the intention of presenting an instance of “taller than,” some third element is needed to explain why I am doing such a silly thing.

The three parts of the notation make a syntactic form, not a triad of Representamen, Object, and Interpretant. Yet, the resulting entity taken as one totality is a sign. We recapture the semiotic character of the notation when we regard the syntactic structure as a whole as a representamen. In this perspective, the notation as a sign lends itself to semiotic analysis. Two examples:

1) (in a mathematical logic)

the representamen: the syntactic structure, (a, b)

may refer to its objects: a and b

interpreting them as: a “related pair” (or, if you prefer, as an “ordered pair”)

2) (less abstract)

the representamen: the sentence, Bill loves Mary

refers to its objects: Bill and Mary

by interpreting one as a lover and the other as a love object.

My perhaps too-labored point is only to alert the reader not to be misled when thinking about the problem of distinguishing two- and three-part relations by the fact that the syntax required to indicate two-part relations has three parts. The case below might easily be misconstrued in such a potential confusion.

4 Dyadic and triadic relations – a case from the cockroach

To continue with a biological application that may demonstrate the pitfall boded above, and returning to an example I have deployed before (Lidov 1999/2017), I turn to a familiar domestic species, the cockroach.

Have you ever dealt with cockroaches? When you try to swat a cockroach with a newspaper, you may think you see a really uncanny ability to “recognize danger” and evade assault. Zip goes the newspaper, it swings and misses. The roach has already gone somewhere else. Telepathy? No. At the cockroach underbelly tiny hairs extend from “acoustic” nerve cells. The dendrites of these nerve cells reach directly into leg muscles. These hairs are sensitive to air pressure waves. Air pressure waves are excited along the surface the roach is standing on as soon as the newspaper begins to move. The hairs catch the wave. The nerve excites the muscle. A quick early response system of just one cell.

We can “explain” the roach’s reaction without reference to telepathy nor even to cognition. To say “recognize danger” is anthropomorphizing. There is no evidence of cognition, no RE – cognition, and, most important (because it would have been an interpretant), no requirement that the cockroach have any idea what-so-ever of the dangerous fate he or she is evading.

Charles Morris might have insisted that the air pressure wave is a vehicle representing the newspaper – its Object – and the muscle activation an Interpretant; Charles Peirce – in a moment of weakness – might explain that this Interpretant, the muscular action, represents the newspaper to the roach in the same way the airwaves did. But what is gained with these elaborations? We are dealing with a phenomenon that now seems exhaustively explained by blind inputs and outputs. “Recognize danger” implies a three-part relation and sign usage, but do these arise in our description or in the act? Does the overcoat of semiotic terminology illuminate anything further? That question is subtle, perhaps more subtle than I have admitted previously.

Let us nod to the “No” and then double back to cross-examine the “Yes”.

No: The previous biophysical explanation of the roaches’ escape ought to be considered, in the perspective of the “no,” as a good example of what we can have in mind with reduction “to a complex of dyadic relations.” At each step, newspaper→air wave; air wave→nerve stimulus; nerve→muscle, we identify one input generating one output in accord with the physics or biophysics of its context. With the action so captured, semiotic terminology is not called for. We are not dealing with triads, thus not with signs, even though, from our external perspective, the roach performance looks analogous to sign-mediated action and is readily described that way.

Yet, this is no simple or self-evident matter.

Yes: Although our initial reductive explanation of the roach’s defensive maneuver did not entail three-part relations, that reduction might be rejected on ontological or epistemological grounds. At least two, complicated arguments support the Yes. I will just point at the first, but dwell further on and question the second.

1. Charles Peirce built his philosophy on principles that identify semiosis with all kinds of regularity from natural laws to casual habits. The lawfulness, or regularity, of a physical reaction is treated by him as an Interpretant. This development of his principles allows Peirce to contemplate the universe as a kind of universal mind. In that context, the roach reflexes are interpretants by virtue of their lawful regularity; the roach behavior acquires an “intellectual component” (his term, I believe). With this supplement to a purely physical account, the roach escape mechanism is not exposed in any new detail but is integrated into a comprehensive understanding of the character and exfoliation of existence. This view point, which may be essentially theological, is not one I feel I need or need to argue with in this paper. I must observe, however, that it applies to all physical relations (and remember that Peirce was very attracted to the idea that physical laws may evolve) and that this viewpoint eliminates the threshold or threshold zone between semiotic and pre-semiotic realities which will be discussed below.

2. Other ways to get to ‘yes’ appear in biosemiotics, a field attracting diverse philosophical orientations. In his own contributions to the anthology that he edited, Introduction to biosemiotics (2007), Marcello Barbieri dismisses out of hand the type of reduction to binary relations I offered for the roach’s escape as “physicalist” and as destroying meaning (page x). It appears to me that an objection which appeals to an a priori postulate that life is meaningful will ultimately show itself equivalent to or derivable from the Peircean viewpoint of which I just attempted a thumbnail sketch. But biosemiotics provides other arguments that are independent of a prior desideratum of “meaning,” and brings to their defense examples which are far more detailed than mine of the roach and show attention both to their current function and evolution.

5 Initiatives to demonstrate semiosis in elementary biology

I remember fondly, from the 1978 Meeting in Denver of the Semiotic Society of America, the developmental biologist Stanley N. Salthe’s passionate critique of the faddish and simplified reception of the genetic code offered by his colleagues in biology. He criticized their failure to recognize the importance of context in DNA behavior, and “context” was then a major focus for semiotics modeling. For him and others, semiotics was a refuge from group think and facilitated his turn to a fresh articulation of biological principles. We must appreciate this turn in theory that allowed him and others to conceptualize feedback, thresholds, contexts, constraints, conditionality, and information storage as essential components of living systems. Do these imply genuine three-part relations? Salthe’s chapter (2007) in the Barbieri Anthology just cited, might seem to address this question directly, in fact, even in its title: “What is the scope of biosemiotics? Information in living systems.” He emphasizes the bi-directionality of information flows and illustrates the complexity of biological dependence on information.

But information is not reference. I think it will be useful here and in what follows to recall the early idea of cybernetics. Since the first proposals of Norbert Wiener, cybernetics has inspired a range of interpretations, some with the ambition of capturing mental activities and machine intelligence in a common framework and overlapping with or including the province of semiotics. I am certainly not the first to notice this. In “biosemiotic questions,” Kalevi Kull, Claus Emmeche, and Donald Favareau recognize “cybernetics” as a neighboring field, and recognize that, like semiotics, cybernetics has been conceptualized in various ways. They distinguish a “mechanistic cybernetics” from a semiotic cybernetics (45) thus supporting a synthesis. I think a synthesis obscures the most pressing question.

Not to impose purity for its own sake but simply for the purpose of clarifying a very useful distinction, I suggest it is productive for us to consider cybernetics as specifically what Kull et al. call mechanistic, operations controlled by information, such as flipping switches in a computer or activating enzymes, and to oppose this restricted (bio-) cybernetics to (bio-) semiotics. The emerging capabilities of computers were the first incentive for cybernetics. Control by information is not interpretation. The calculator which displays “7” after an input of “3”, “+”, “4” is providing an output but not interpreting (or understanding) the output “as a sum.” In this limited framework, cybernetics is a science of information structures which models inputs and outputs, storage and retrieval, conditionality, and context. It does not include or need “interpretation.” It can be modeled without triadic relations. Similarly, this article proposes to define “semiotics” as a study of signs that participate in irreducible triadic relations. Salthe’s work, mentioned above, provides evidence that cybernetic relations are essential to biology, but not evidence that biological actions generally cross a threshold zone from cybernetics to semiotics.

I propose that we not be hasty in attributing genuinely triadic relations to all life processes. My selection of examples is, I confess, arbitrary; I hope they are nonetheless illustrative.

Evan Thompson’s masterful concordance of contemporary “Western” neuropsychology with yogic philosophies of mind, Waking, dreaming, being (2015), is replete with thinking about consciousness that readily adds incentive to the speculations and questions broached in this article. In the context of an analysis of “self,” this superbly cautious book, cites without fully endorsing it but without any cross-questioning, what is proposed as a demonstration of a sort of “memory” in Escherichia coli bacteria published by Nikita Vladimirov and Victor Sourjik (2009: 1097–1104). Memory, in itself, were the term taken simply to refer to information storage, would not necessarily establish triadicity but it augurs the presence of a triadic formation if memory of x-past represents y-present as a recurrence of x or change in x, and the latter holds here. What Vladimirov and Sourjik point out is that this talented bacterium adjusts its direction and speed of movement (via flagellation) in response to changes of the concentrations of attractants and repellants in its chemical environment but not to the levels of the concentrations, implying, they suggest, a comparison of past and present. This suggestion appears to me to make a mistake of logic. Responding to rates of change and changes of rates does not require comparison with a past state. Rates of change belong to the instant of change, a fundamental point, as I understand it, of Newton’s differential calculus. In this case, chemical methylation provides E. coli with a “machine” that responds to the first derivative rather than an initial linear or quasi linear function of the strength of the concentrate in solution. I picture it as like an enriched thermostat with a sensitive needle for current temperature and a more sluggishly moving frame for acceptable temperature which adjusts, a bit more slowly than the needle, to center on the needle in its new position. Flagellation responds to the current difference between the needle and frame. With our wonderful investigative machinery, we might, I suppose, scoop up a single E. coli cell and determine what its Umwelt was like half a second ago, but that is our game, not the bacterium’s. Semiosis arises not in the action of the cell but in our description of the action.

6 The biosemiotics of Kalevi Kull

Kalevi Kull, semiotician and theoretical biologist of the University of Tartu, inherits and develops the profound tradition of semiotics we associate with that city. In addition to the one mentioned above, an extensive series of articles by Kalevi Kull, alone and with others, examine issues I have touched on from other perspectives and with superb finesse. I am interested to accord attention to these not only because they are in themselves so substantial but also because the latest of them that I have studied arrive at viewpoints that seem to me to be extensively similar to that which will be proposed in this paper although proceeding from vey different premises and definitions. The first publication mentioned below introduces and refines notions of threshold zones between pre-semiotic and semiotic behavior and between types of semiotic behavior. The idea of threshold zones greatly sharpens for us options and differences among theories. The next publication I cite raises the issue of the role of evolutionary history in biosemiotic theory. The last group identifies “confusion” or “free choice” as necessary factors in semiosis. These proposals may be quite in harmony with (and precede by several years) my effort to defend “nebulosity” as a factor that establishes triadic relations; both the differences and similarities are telling.

In “Vegetative, animal, and cultural semiotics: The semiotic threshold zones” (2009: 10) Kull recalls Eco’s introduction of a semiotic threshold:

The concept of lower semiotic threshold was introduced by Umberto Eco in his book A Theory of Semiotics (1976: 6, 19–22). Eco described the lower semiotic threshold as follows: “By natural boundaries I mean principally those beyond which a semiotic approach cannot go; for there is a non-semiotic territory since there are phenomena that cannot be taken as sign-functions. (Eco 1976: 6)

Kull, in collaboration with several colleagues, determined that it was better to speak of “threshold zones” rather than boundaries because they saw semiosis in biology arising – as life seems to have arisen – only in the interplay of a number factors whose coordination cannot be supposed to have achieved at one particular moment (2009: 9–10). In this determination, they were guided by the principle that Kull, though acknowledging Uexküll’s priority, often refers to as the “Sebeok thesis,” that life and semiosis are coextensive.

We should pause to ask whether this principle ought to be called a “thesis” and not a “proposition” or “premise” or “doctrine.” It follows from particular definitions, not from evidence. The definitions I propose in this article are not consistent with it. The premise that identifies all life as involving semiosis has surely contributed to excitement about ethological and biochemical observations, but in the long run, it may be misleading.

That “zones” is an improvement cannot be questioned. So amended, the question, What are and where are these threshold zones? becomes a very clear problem and a fine tool for comparing theories. Kull develops an ingenious response which aligns these zones with Peircean categories. I want to signal an important initial observation, to which we will return later on:

With the introduction of the concept of the lower semiotic threshold, certain problems involving its correspondence to C. S. Peirce’s approach appear. For Peirce, semiosis starts from the situation of lawless chaos; laws then develop as habits. (Kull 2009: 12)

The initial condition of chaos is a condition in which associations would occur at random and then be filtered out as laws develop. Kull does not propose that chaos disappears but implies that it is to some extent superseded. The persistence of randomness is a factor that we must, I think keep track of, but I believe Kull lets it pass without noting its consequences for his later ideas.

Because he adopts Sebeok’s stance at the outset, thus settling the lowest threshold, zones of particular interest to Kull in this paper mark the transition from vegetative to animal semiosis and from animal to cultural semiosis. These terms are motivated by but certainly not identified merely with differences between plants and animals. In distinguishing vegetative and animal semiosis, Kull makes passing reference to a contrast of dyadic and triadic relations. Accepting his terminology, Kull argues rigorously that vegetative semiosis is essentially iconic in its sign relations and that animal semiosis requires indexical relations. ^[3] Vegetative semiosis is characterized by simple correspondences and by the absence within the organism’s umwelt of any unifying continuity, such as space or time that would make these correspondences cohere as a single map. ^[4] Animal semiosis responds to networks of patterns with networks of actions, producing a sense of surroundings differentiated by the boundaries of the self.

Vegetative semiosis is demonstrated by the example of a living cell that conveys information about its environment across its own boundary membrane to determine internal reactions, rather as in the example above of the E. coli. Kull’s demonstration here is also recapitulated in his chapter in the Barbieri (2007) anthology mentioned above: “Biosemiotics and biophysics – The fundamental approaches to the study of life.” Although this demonstration is methodologically akin to Vladimirov and Sourjik’s, it gains additional interest from its description of the role of evolutionary history in establishing and maintaining a relationship which cannot be accounted for otherwise. In the article, the author speaks of a living cell and of a relation between two molecules, Molecule A, outside the cell and molecule B, inside, mediated by a chain of amino acids. He attributes this relation to genetic history rather than chemistry, because a purely chemical description shows us a fortuitous structure that we cannot account for. Evolution across time with consequences now fixed in the DNA explains how the chemical signaling works. In this analysis, the genetically guaranteed historical continuity of life forms serves, at least roughly, the function of an interpretant, as regularity does for Peirce. Again, we must consider levels of description. For an external observer the historical derivation may enhance a view of the cellular action as “communication.” But the enzyme needs no reference to a past state or representation thereof and needs no comparisons or interpretations. Like any object, living or lifeless, e.g. an eroded rock, it carries conditions with past determinations. These determinations are not known to the rock nor, I presume, to the enzyme and have no ongoing function. Like E. coli, the enzyme does fine without knowing. As an element of cellular function, there is no need to appeal to communication, however arbitrary and fortuitous the link of amino acid reactions is. A semiotic description needs the triad, B understands A as X, but in an internal perspective, we simply have A “triggering” B, (a simpler, two-part relation). Communication arises as an element of our description and understanding, not as a cellular process.

In another group of papers, Kull, alone and with others, though not abandoning the Sebeok Thesis, has explored a criterion for semiosis, the criterion of choice, which, prima facie, is logically independent of asserting life (Kull 2009, 2018; Kull et al. 2008). In his construal of choice, choice can occur with the earliest and simplest forms of life. For him, the new criterion reinforces Sebeok’s position but is not derived from it.

In these publications differences of terminologies appear: “confusion” is later replaced by “choice,” but this does not obscure their thrust. I will rely here on the schema as it appears in “Choosing and learning: Semiosis means choice” (Kull 2018). Kull (2018: 1) states first, “The aim of this study is to clarify the relationship between semiosis and learning, and also the mutual relatedness of types of semiosis and types of learning.” Although they are a central interest for the author, I am bypassing two important parts of this paper, his analysis of learning and his exploitation of the idea of choice to compare and unify schools of semiotics often seen as separate from each other or opposed to each other. Both are persuasive; I bypass them simply because they are further from my own theme.

He continues to explain that:

The model we propose consists in a slight redefinition of the general terms of learning, memory and semiosis. We define these terms bringing in the concept of free choice, which has been downplayed by both the computational and the neo-Darwinian approaches to learning. We observe that, in semiotics, the concept of free choice has not received the attention it deserves. (2018: 1)

Besides choice or free choice, foundational concepts are “nowness,” “subjective present,” “possibilities,” and an opposition of algorithmic determination to free choice. The moment of semiosis is the moment of decision between possible objects of reference when an ambiguous Representamen is assigned to one of the Objects possible with the possibilities that are simultaneously present in the subjective “nowness” of the organism. That decision, determining the object, is the Interpretant acting. Though it is not stated explicitly here, it seems clear in the context of the author’s studies that the object can be and often is an action, such as a plant “choosing” to extend a root in one direction rather than another.

Acknowledging the problematic character of “now” in the external world, Kull insists on the subjective character of “nowness,” the time when possibilities are copresent for an organism, and as a caution, he cites from George Frank and Harald Atmanspacher (2009) contending that “semiotics is incompatible with a physicalist concept of time, and with physicalism altogether – ‘there is no present or nowness in fundamental physical theories’ (Franck, Atmanspacher 2009: 212)” (Kull 2018: 455). Despite this caution, and even if my own associations are merely poetic, what this disclaimer brings to mind for me is the proposal of Roger Penrose and Stuart Hameroff that would link consciousness to quantum collapse in the microtubules of nerve cells. (Penrose and Hameroff 2011). Kull’s “nowness” of simultaneous possibilities feels analogous to quantum superposition, but avoiding haste in making anything of that coincidence does seem very sensible. I have read the Penrose and Hameroff papers but have no competence to criticize them.

Free choice is defined negatively in Kull. It is not algorithmically determined; presumably, it is not chaotic, or to use a word that doesn’t appear here, not random. And it appears to correlate with agency in some sense. Kull does not say in positive terms what choice is, but he aligns his position with Raymond Noble and Denis Noble (2018), who directly address precisely that question. For these authors, organisms at all levels of evolutionary development make creative adaptations that originate in stochastically determined behaviors which are filtered by their success or failure. We cannot predict the result but can find rationales for them afterward. “Choice necessarily involves dealing with uncertainty. Low-level stochasticity is the clay from which high-level novelty can develop” (2018 Section V).

The distinction between determinate causes and influential rationales toward which Noble and Noble were guided by the philosophy of Karl Popper also plays a role in my own single attempt to analyze free choice. The two final chapters of my Elements are “codettas” meant to illustrate possible consequences of the theory preceding them, and the first of these considers free will. There I regard chaos as a background factor, but not, as Noble and Noble do, a source of creativity. Simple life forms are not in question in my argument but rather human thought in the context of an elaborate and unbounded semiosphere, and it is in interaction with the latter that I saw free choice arising. To explain how that happens, I was compelled to essentially the same distinction that Noble and Noble make between causes and rational influences.

The central point that Noble and Noble want to prove is that the organism plays an active role in making the determinations that carry forward to its progeny, proposing an “active evolution” in opposition to Darwin’s “passive evolution.” They succeed in showing that the terrain of evolutionary action is not purely genetic, but they do not really persuade us that the broad filtering of random variation to which they attribute biological development is fundamentally different from Darwin’s or that a mode of choosing emerges which is fundamentally different from natural selection among chaotic changes. They make a strong case for adding random behaviors of the organism to random mutation as the source of variety, but that does not establish the emergence of a new species of choice. Kull fundamentally advances the position of Noble and Noble by defining a subjective simultaneity or subjective “now” where choice becomes possible because alternatives are present together at a same time. In respect to the problem of distinguishing random choice from choice which has some identifiable component of non-stochastic but also not wholly determinate cause or some reasonable idea of intention, the precondition of “nowness,” is not yet a solution though it may prove to be a potent pointer.

In searching for of a conception of choice as something other than an interplay of fully determined and fully random elements, I think a revised definition of interpretants does provide a way forward to construe a true alternative to a mixture of “chance and necessity.”^[5] What Kull might hope to capture with the criterion of choice resonates for me with the criterion of nebulosity I propose for full-fledged interpretants. One reason for this resonance is Kull’s need, like mine, to tweak the definition of “interpretant.” A stronger reason is that with “nowness” or simultaneity of possibilities, Kull’s idea evokes a quality that may well seem like the one I will indicate as “nebulosity” although the structure of his concept is much more precise. The major difference in our end results is the location of the lower threshold zone for semiosis. What lies below that threshold for both of us is mechanical action, but I want to include within mechanical actions the very highly developed information structures and controls studied by cybernetics and that occur in all forms of life. Kull has clarified to me (personal communication) that he would expect semiotics to be interested in mechanistic or habituated information structures in organisms when these have been derived (that is, from a phylogenesis) which at some stage included semiosis, marked by some element of choice, a criterion which, he thinks, still invites further study. I do not know whether it will be possible to adopt this principle and still avoid a confusion between evolutionary history (the observer’s perspective) and an analysis of synchronic organic functions (the hypothetical perspective of the organism itself). I need not restate my unease about this ambiguity. Of course, the boundaries of semiotic interests should be wider than the boundaries of semiosis as counter-examples to semiosis cam be as instructive as examples.

Going forward, let us remember that what is at stake are definitions, not facts of the world. I hope to meet my obligation to try to show that mine support useful insights. As a first step in that task, I return to Peirce.

7 Interpretants

Peirce first proposed Interpretants in his youthful paper of 1867, “On a new list of categories.” He argues that to understand X as a sign of Y, we require a third which, in that paper, he first calls “a mediating representation.” He provides three examples. Here is that passage:

Suppose we wish to compare the letters p and b. We may imagine one of them to be turned over on the line of writing as an axis, then laid upon the other, and finally to become transparent so that the other can be seen through it. In this way we shall form a new image which mediates between the images of the two letters, inasmuch as it represents one of them to be (when turned over) the likeness of the other. Again, suppose we think of a murderer as being in relation to a murdered person; in this case we conceive the act of the murder, […]; and thus we resort again to a mediating representation which represents the relate as standing for a correlate with which the mediating representation is itself in relation. Again, suppose we look up the word homme in a French dictionary; we shall find opposite to it the word man, which, so placed, represents homme as representing the same two-legged creature which man itself represents […].

And further,

Such a mediating representation may be termed an Interpretant.

The first example, with b and p, Peirce might later call iconic. The second example, the murder and victim, foreshadows the index. The third, with its legisigns can introduce symbols but to use another term Peirce loved, it is, I fear, degenerate. The first two examples need imagination; not the third, which may be automated, and the third may reduce to a binary relation:

“Man” (in the context English) is equivalent to “Homme” (in the context French).

It is true, as Pierce suggests, that we are likely to conjure up some images, but it is not patent that we need those images to master the translation. In fact, we have machines that translate but so far, no autonomous machines for solving murders.

The alignment of the words “homme” and “man” is a dubious example for that passage because, as is the case of many designated signs^[6] and what Kull, relying on Peirce, calls a “habit,” the relationship has been so automated that we can accept the interpretation as ready-made, frozen in advance. Indeed, Peirce’s bi-lingual dictionary moves us toward reflexes and associations that readily reduce to two-part relations. The only reason I call attention to these degenerate cases is to ask that we not be distracted by them; they pose no obstacle to theory.

Peirce’s first two examples evoke nebulous interpretants. The images are ill defined with little detail and, for each perceiver, idiosyncratic. The images are evanescent and probably not highly illuminated, though the murder might make a vivid flash. I know that Peirce’s concept of the Interpretant seems itself to be nebulous to many folks, but let us try putting the shoe on the other foot by insisting – as a matter of definition – that the nebulosity lies in the interpretants themselves rather than in some concept of them.

“Nebulous,” by etymology, “cloudy,” affords just the right metaphor. The word “cloud” is definite in its form and reference but its objects, clouds, are indefinite. Clouds have no distinct boundaries and shift shape as you watch them. I suggest that semiotics live with the idea that genuine interpretants are in themselves nebulous, that they are, to use Peirce’s words, general ideas.

But are clouds in themselves essentially nebulous? Etymologically the question is ridiculous and to speak of “essence” slips into historical rhetoric. That touch of historical flamboyance can introduce a comparison of chemical-physical and sensory-phenomenal conceptions of clouds. Physics does not need “nebulosity.” For physics, the amorphous and hazy boundaries of a cloud might invite description in terms of probability densities for the possible positions of water vapor droplets. Probability density is a tough concept for some of us, but it is precise and quantifiable. On the other hand, in our experience, nebulosity is a quality and has character but no precision. “[…] neither a matter of fact, since thought is general, nor [….] a matter of law.” Even if these two approaches to description seem to invoke some degree of mutual harmony, they don’t match at all.

Whether or not it appears elsewhere, nebulosity as a quality does appear in consciousness. Qualities in consciousness are potentially elements of thought with functions in organic thinking and are not easily exported to systems of mechanical calculation.

I propose that: A genuine interpretant establishes a conceptually or energetically characterized relationship between a sign and its object which exceeds merely pairing them. An interpretant might represent seven as a sum of three plus four, not merely the output. Turkey represents Thanksgiving as a feast, not necessarily as a day off. In both cases, a nebulous idea is evoked. Of course, “sum,” unlike “feast,” can be accounted for in its mathematical use by precise definition, but the casual and popular use of that concept prepares metaphorical extensions and, in English, enters a vague but very useful confusion with “sum” as a short form of “summary” (as In sum, his reasoning is full of holes).

General ideas, which are at least somewhat nebulous, are real and useful. The notion that our general ideas are nebulous is clearly not a novelty, but to my knowledge, cognitive sciences and semiotics have usually approached this notion by trying to defeat it. Prototype categories, feature sets, fuzzy sets, preference hierarchies, probability densities, conceptual blending, and approximations respond to our conscious experience of general ideas but only by offering more precise models as paraphrases or substitutes. All are worthy inventions, but the original, the nebulous, general idea cannot be reliably substituted by a logical structure. Wittgenstein’s “family resemblance” is not a logical paraphrase; it is a nebulous idea (allowing some imposition of logical relations) that indicates a subtype of nebulosity. Perhaps our general ideas are not fixed either in content or in structure, appearing at one moment like a prototype category and at another moment like a set tentatively defined by features. Or perhaps their structures are volatile and mixed in type. Perhaps the flexibility inherent in nebulous structures allows them to function creatively. We must not abandon the desideratum to “make our ideas clear,” but we can recognize that the possibility of doing so has natural limitations. Thus, I smiled with satisfaction on discovering the copula in the opening sentence of the entry in Wikipedia, our most nebulous authority, on fuzzy sets: It is not simply “is” but the more nebulous “is somewhat like.”^[7] General ideas are not fully immune from logic or logical critique, but they are resistant to logic, and we must be content to be rational when we can not be logical. Logically, we would like to have a less nebulous definition of “nebulous,” but rationally, it is enough to have a sense of an opposition between clear and nebulous.

Requiring by definition that genuine interpretants be nebulous makes the idea that consciousness is required for interpretants sound comfortably similar to a necessary corollary. It is not. Do nebulous items present themselves in consciousness? Yes. Do nebulous items present themselves exterior to consciousness? That question is circular. If I really insist on a qualia, a quality as the critical feature of nebulosity, then the question becomes whether there is such a thing as a qualia that is not in consciousness, a question which may be pointless and which I leave open. Perhaps a better question is whether there is something corresponding to nebulosity that occurs independently of consciousness or that might induce a perception of nebulosity as a quality when consciously experienced. The logical paraphrases listed above didn’t do, but perhaps a loose, flexible, variable mixture of them would come quite close. We could speak of such a “structure” as “proto-nebulous.” I see no reason to exclude proto-nebulous elements from the unconscious or from computers as long as we don’t make any assumption that they induce or are in some sense causes of consciousness. Consciousness is ontologically different from phenomenon analyzed by physics. The closest physics comes (see Evans 2015) is indicating events that may mark its disappearance.

The mental contents we refer to as unconscious are ascribed to unconsciousness by deduction, and here evidence is quite lacking. It seems reasonable to describe images in dreams as often nebulous, but we know them only when those images enter consciousness. If the nebulosity of general ideas is a key to their creativity, then the apparent creativity of sleep, as when we go to sleep with a problem and wake up with a solution, might be evidence that something like nebulous interpretants also act in unconscious thought. An industrial physicist once said to me, “I can’t worry about people stealing my ideas because I don’t know where I got them.” Our primary domain of creativity may well be unconscious, and the work may precede our consciousness of its nebulous results. It makes no sense to presume that our fundamental semiotic theory can encompass unconscious thought other than by far-fetched speculation. It is when speaking of signs, we know first-hand that nebulosity works as a criterion of an interpretant that makes for genuine triadicity. That criterion makes the right starting point and a good place to start. We can adopt it.

May it be that genuine, nebulous interpretants are inherent in some classes of computation or some instances of primitive animal behavior? Certainly, for animal behaviors, the question must stay open. If qualia by definition are exclusively conscious, there is no reason to raise the question with computers. Computers are still mechanical devices (will quantum computers force an amendment?) and subject to exhaustive physical analysis while consciousness is ontologically different. Yet, to authoritatively deny proto-nebulosity to computers, I might need to know how so-called hidden registers function in learning programs, and I do not. That second question can stay open.

In mechanical computation, we find better and better imitations. Years ago, we heard of experiments in some university counseling services that provided automated psychotherapy sessions to students. These accepted a client’s complaint and transformed it into a question. These programs were reported to be helpful by some students. Macintosh’s “Siri” and her A.I. rivals are stunning fakes with to which we feel strongly drawn to attribute consciousness to our delight and amusement, offering the “Siri–Us” problem to cognitive psychology. Of course, we know she is an illusion. Equally striking is the currently new chess playing program, AlphaZero, which a New York Times report said, “clearly displays a breed of intellect that humans have not seen before” and scarier, “Most unnerving was that AlphaZero seemed to express insight. It played like no computer ever has, intuitively and beautifully, with a romantic, attacking style” (Strogatz 2018). Nothing in the descriptions of the mechanics of AlphaZero I have been able to find supports the fantastic language of this description, nor does the appreciative but sober commentary of chess grandmaster Gary Kasporov (Kasporov 2018). AlphaZero is described as a self-teaching system or learning program, and I do not doubt or disparage its powers. Clearly it comes up with principles that have not been known to human chess players, but perhaps it will be more profitable to look to AlphaZero for evidence that can characterize human limitations than to ascribe magical competencies to machines. Something very close to an “interpretant” does seem to appear when computers direct there scrutiny according to conditionals and bidirectional information flows that lead to questions. A computer might so organize information about its optical environment that registering an oval and, within that, smaller highlights, possible eyes, a scanning program could issue instructions to its sensors “Look for a mouth.” It is strenuous to describe such operations and their results without recourse to concepts like representation, hypothesis, and interpretation. So far as I can learn, if you look hard, the computations do reduce to complex inputs and complex outputs connected via binary relations. Perhaps the program is flexible enough to suggest proto-nebulosity. Still, proto-nebulosity does not entail genuine triadicity.

Questions about organisms are more difficult, in part because we cannot look inside. A report from the AAAS Journal, Science, about an experiment with honeybees flamed on the Internet last year (Lukola et al. 2017). Moderately remarkable was the observation that honeybees could master the rather strange task of moving a tiny ball to a designated spot in order to be rewarded with drop of sugar. Much more astounding was the finding that a second bee who had observed the first bee would then enter the ring ready not merely to repeat the performance but to improve on it by modifying the first bee’s method. The intelligence this performance implies requires translation from optical to kinesthetic information as well as the capacity to identify third-person and first-person perspectives with each other.^[8] To concoct an analysis of all that which reduces to binary inputs and outputs would seem enormously difficult to me. And even if a smart team can do it, how would the apparatus fit into a bee brain? Ontologically, bees seem to be more like us than like computers. Perhaps genuinely triadic relations, encompassing nebulous interpretants, simplify the bee calculations. Might these be conscious? We have no good reason to assume the negative.