All Things Are Full of Gods, page 45
HEPHAISTOS: I have no doubt that’s true, but explicable isn’t the same as reasonable. As I’ve said, however, I’m open to revisions in the laws of nature if they should prove needful. But, until we can isolate the mechanisms present at the beginning of life—or what we can reasonably deem to be its beginning, when metabolism and replication became distinct from transient disequilibrium and structural amplification—we’re simply arguing over imponderables. You continue to pose interesting questions, this I concede, but you’re too eager to propose extravagant answers. As Richard Feynman said, “What I cannot create, I do not understand.” And no one, not even any of us gods, knows how to build an organism.
PSYCHE: In this instance, that ignorance is itself enlightening. And I would still insist that the evidence is on my side here.
HEPHAISTOS: Insist away. As you’ve so eloquently argued, our method determines the questions we ask of nature and also the answers we’re capable of understanding; and our method is all too often merely an expression of our prejudices.
HERMES: [Clearing his throat a little more dramatically than necessary:] If I may interject, I think we may be overlooking something rather obvious. I suspect that the issue of the genetic code’s complexity, interesting as it is, is of only secondary importance. The more important issue, it seems to me, is that of . . .
HEPHAISTOS: Yes . . . ?
HERMES: That of language as such.
HEPHAISTOS: [With a terse laugh:] Why, of course. What else could it be?
VIII
Language, Code, and Life
HERMES: Oh, Hephaistos, my brother, surely you must grasp that language is the very epitome of top-down causation. That much, it seems to me, we’ve established. Its entire substance—its syntax and semantics, the symbolic thinking in which it subsists, the intentionality informing and driving it—lies beyond the physical altogether, in that hermeneutical space whose only location is the activity of a living mind. I’ve been clear that I see the notion of the indissoluble structure of language evolving from non-intentional physical processes to be preposterous. No less preposterous, it seems to me, is the notion that the capacity for language—even, say, simply the bare algorithm of Merge—appeared all at once as a mutation in the evolutionary history of a single species. Neo-Darwinist orthodoxy has at least the virtue of humility; it confines its claims to a very exiguous scale indeed: small, fortuitous mutations at the level of the genome, only some of which survive. But the sudden bottom-up appearance in a single isolated gene-pool of a capacity such as Merge, presuming as it does the entire semeiotic economy of language, would constitute a mutation at the system-level, which means at a level of the unimaginably intricate organization of countless systematic functions; and, even then, that organizing activity would all be in service of a faculty whose entirely real ontology—apart from the material substrate of sounds, gestures, and traces—is extra-physical. This is absurd in physicalist terms. And, as it is for language, so I suspect it is for life, for the simple reason that life is language.
As diverting as our debate today has been, we’ve touched upon the semeiotic level of intentionality in organic systems only three or four times, I believe, and then merely in passing. We should really pause to consider the matter a bit more deliberately. Replicating life, as we’ve said, isn’t just the prolongation of dynamic disequilibrious processes in certain persistent negentropic alignments, and it certainly isn’t simply an expansion of complex physical patterns. Life actually communicates itself, from one organism to another, and that communication is a good part of what marks it out as a living system. But how does it do so? Given the way some theorists speak, one might think we were talking about the mere transference by contact of some periodic physical pattern—or quasi-periodic but largely regular pattern—producing something like the pretty images conjured up by cellular automata or the fractal ramifications of crystals. But what we’re discussing is semantic content, which exists and has effects not at the level of the physical embodiments of syntactic structures alone—those are merely the medium of transmission—but instead by being intended and being understood. We’re talking about “information” in its most commonsense acceptation. The genome isn’t a collection of small physical switches that work like springs upon the cell; its whole causal power lies in its legibility as intentional content to an interpretive and cognitive agency, capable of judiciously rendering that content into a vast variety of distinct uses. The content is transcribed into code and then read out from code again, and this is more than a convenient metaphor. There’s a real semeiotics of life, translating organic systems into “digital” information and then translating that information back into proteins and cells and tissues and so forth—albeit with a certain interpretive latitude that’s also proper to the non-physical potentials of language. That semeiotics—that code—carries instructions and templates in symbolic form, and is communicated and interpreted across the disjunction between phenotypes and even sometimes across the disjunction between taxonomic categories; but where that code comes from is impossible to say in terms of the physical sciences as we know them. Perhaps it emanates from the mystic realm of “physical information,” but of course, as we’ve said, information of the kind we’re talking about isn’t some quantifiable element or flow of energy within the constitution of material nature; only its transmission is; in itself, it’s intentional content and nothing more, with formal constraints and a purposive structure, conveyed by what look like symbolic conventions rather than physical entailments; otherwise, it isn’t information. I mean, in addition to the code there is—there must be—an intentional agency that encodes, as well as another that decodes and interprets, and then there’s the use to which that encoded material is put by the organism that employs it . . . employs it judiciously, as I say, and often creatively. So, yes, the astounding complexity of the physical systems in any organism is dazzling to contemplate, but for me all of that’s a subordinate consideration compared to the sheer . . . well, I want to say, the sheer miracle of that code. And, again, I’m not even referring to the specific complexity of the code—which is also marvelous, of course, with all its layers of coding within coding, all its protocols for correcting errors, all its provident redundancies, and so forth—but simply to its semeiotic nature as such, and its dependence upon the hierarchical structure of all language.26
HEPHAISTOS: Well . . . language may be a bit of a stretch. We’re talking about molecular sequences, not a grammar of verbs and nouns and . . .
HERMES: Of course we’re talking about molecules, but we’re also talking about meanings expressed in grammars, at least at the hermeneutical level of the operation. Who cares what the physical platform of transcription is? The very reality that the information in DNA can be transferred to another medium, and even stored on a non-organic platform, marks it out as simultaneously semantic in nature and physically inert in efficacious power. The power of that intentional content depends upon the interpretive agency that translates it. When we talk about text printed in books, there too all we have at the physical level are intrinsically meaningless material phenomena: paper and ink and physical markings without intrinsic content. Texts become “meaning”—verbs and nouns and so forth—only in the intentionality of the author and the reader. So, too, only when the organism writes or reads its genetic code do those templates become a grammar, a symbolic system discriminated between nouns and verbs and the like—between objects of operation, like proteins or tissues, and operational instructions. In any linguistic economy, more to the point, the entire causal power resides in that superstructure of meaning, while the entire superstructure of meaning resides in that hermeneutical space to which we keep returning, and that hermeneutical space itself resides only in intending mind. Such is the top-down indissolubility of language’s hierarchy of functions that (to borrow a bit of scholastic terminology) the more “eminent” aspects of the economy are the ontological “grounds” of the less eminent. Intention is the ontological ground of syntax and semantics, syntax and semantics are the ontological ground of linguistic inscription or expression, inscription and expression are the ontological ground of the physical traces that record the code, and only the very last of these tiers of causal relation inhabits the actual material world.
Let’s remember, after all, that the science of information concerns only measures of so-called syntactic information—though, really, it’s often not really a “syntax” that’s at issue, but just physical quantities. Shannon’s equations, for instance, are entirely concerned with the transmission of “bits” and the degree to which they reduce uncertainty. Here, however, a crucial distinction has been elided. I think I’ve already complained about this, as it happens: information-theory in the sciences speaks of the distinction between the syntactic and the semantic as though it were a distinction between the physical or quantitative vehicle of information on the one hand and the contents of information on the other. But this is a terrific confusion. Syntax, properly speaking, is an order of symbolic relations, not merely the sequence of physical quantities needed to convey some measure of information. The latter might better be called something like information’s material “parataxis” or “seriality” or simply “flow.” And the failure to recognize this difference can make one also fail to appreciate how radical the top-down causal hierarchy of linguistic information truly is, and can even lead to such nonsensical notions as computational or functionalist models of mind, as we saw yesterday. It can even encourage something as odd as the currently fashionable delusion that Artificial Intelligence based on “Large Language Models” would be capable of actual semantic learning and writing and thinking; but, of course, such models are what’s called “autoregressive” processes, which achieve their results precisely by totally inverting the actual structure of language in a way that makes semantic learning or synthesis of any kind impossible. It’s a predictive system based on a massive compilation of “tokens” of linguistic information—not even its syntax, in the proper sense, but only its physical expressions at the “paratactic” level—which generates a mere simulacrum of syntax and semantics. It’s an illusion produced by, on the one hand, the relentless statistical reduction of the aperiodic structure of semantics to as large a set of periodic physical repetitions as possible and, on the other, the relentless statistical inflation of the compressible, periodic level of information into an ever vaster imitation of the real openness of the semantic level. It’s ingenious, but it achieves the appearance of meaning only by the resolute negation of the reality of meaning, sorting its bright little tokens of information into ever more regular and flexible patterns of juxtaposition through the sheer brute quantitative force of its data. Admittedly, if one finds functionalism coherent, I suppose one can believe anything . . .
HEPHAISTOS: You’re wandering fairly far afield, brother.
HERMES: Yes, I suppose so. But it’s relevant even so. My point is that where “code” is concerned, the paratactic order is determined by the higher level of syntactic order; but that syntax is in turn determined wholly by the higher level of semantic meaning; and that meaning depends wholly upon the irreducibly higher level of symbolic thought. Hence, Shannon’s formulations aren’t of any help at all where semantic information is concerned. Meaning can’t be measured in discrete quanta; its structure is hypotactic and its nature intentional, and it’s something in a sense outside of—but also more than merely supervenient upon—the whole quantitative realm of bits and their transmission. In fact, there’s an almost perfect inversion between determinacy and indeterminacy where these two varieties of information are concerned, as there is between physical patterns that merely repeat and the kind of meaning that creates organic reproduction. At the so-called syntactic level, the more determinate the content, the more susceptible it is of compression—of condensation, that is, into an algorithm. A thousand repetitions of a sequence of twenty numbers can be coded by providing just one iteration of that sequence and attaching it to a simple functional instruction for a thousand successive reiterations. A merely periodic sequence isn’t random; it’s precise and determinate, and so an algorithm can be generated that wholly contains all the “syntactic” information present in its uncondensed form. But then, once again, we’re dealing with something entirely simple; we’re back to a process like the morphogenesis of a crystal—a mere structural convention that multiplies itself in a purely geometrical amplification of its initial structural “theme,” as it were, measurable in fully quantitative terms. Conversely, a wholly random sequence of, say, twenty-thousand numbers or letters can’t be usefully compressed or captured algorithmically.
Semantic information, you see, obeys a contrary calculus to that of physical bits. As it increases in determinacy, so its syntactical form increases in indeterminacy; the more exact and intentionally informed semantic information is, the more aperiodic and syntactically random its physical transmission becomes, and the more it eludes compression. I mean, the text of Anna Karenina is, from the purely quantitative vantage of its alphabetic sequences, utterly random; no algorithm could possibly be generated—at least, none that’s conceivable—that could reproduce it. And yet, at the semantic level, the richness and determinacy of the content of the book increases with each aperiodic arrangement of letters and words into a coherent meaning. There’s no way in which that intentional level of semantic meaning can be reduced to that physical level of inherently meaningless “paratactic” information-flow, much less generated by it; structurally the two levels continuously diverge from one another precisely to the degree that each becomes more coherent in its own terms, and so any algorithmic reconciliation of the two in terms of some more comprehensive notion of information is infinitely unattainable. And this presents something of a difficulty for a wholly physicalist account of life, because what’s encoded in a genome and in the organic system to which it belongs, and so what allows for replication and variation, is a replete semeiotic economy, one that’s quantitatively random and resistant to algorithmic compression at the syntactic level and yet utterly and exquisitely precise, meaningful, and determinate at the semantic level. There’s no other way it could function. Even if the code contains algorithms, they subsist within a structure of information that may generate certain transitory local syntactic repetitions—I mean, yes, a rose is a rose is a rose (repeat infinitely)—but that structure can’t in turn be generated by such repetitions. This is why Schrödinger, well before the discovery of the double helix, already predicted that what would be found at the basis of replicating organisms would have the character of an “aperiodic crystal,” as only an aperiodic syntactic medium—one we now know to be composed of such things as strings of deoxyribose and phosphate molecules—is capable of conveying the transcription of life’s code.27 This is why, too, the attempt to ground the origins of life in accidental structural refinements of purely physical systems for processing matter and energy, even when they’re systems of organic metabolism, can never really provide a causal narrative of the transition from the prebiotic to the vital, or get us down to the essential level where the system operates. Organic replication isn’t a closed mechanical process in any sense; it has nothing like the compact and iterative periodic structure of a pullulating crystal; it’s not an engine, but a process of understanding. In the case of life and mind, the distance between structural and strong emergence remains absolute.
That’s why Paul Davies, as we saw, would like to see the laws of information somehow directly written into physics; but that seems a strange way of thinking of it, at least in regard to what we mean by “physics” at present. It’s simply futile to imagine that the laws of life could be enucleated in the form of laws of physics . . . or chemistry, for that matter. The semantic information communicated in life’s coding—or, rather, the formal causality determining that coding—vastly exceeds the intrinsic limits of the physical structure of information by which it’s conveyed. The physical syntax simply can’t generate the semantical meanings; rather, the top-down causality of the semantic level determines the physical ordering that embodies it. I’m sorry to keep recycling the same analogy, but to confuse the code for the chemical constitution of the genome is like mistaking the contents of the text recorded in a book for the book’s paper and ink. And, even if you want to argue the contrary—to advance the fantastic proposition that there’s some incomprehensibly mathematically imponderable but real algorithmic process that allows for the semantic structure to arise from its purely physical syntax—you’ll still be confronted by a semantic barrier to compression: to wit, that the genome, we now realize, is not just the repository of an invariant code, but is part of an interpretive and creative negotiation between the macromolecule and its cellular and somatic context. It’s, as James Shapiro says, a “read-write” medium. Life is a semantic structure that colonizes and informs its physical syntax—it’s a formal cause with a final intentional horizon that shapes its material cause toward an end—and it’s this structure that the cell not only takes from but also imposes upon its genome. The generation of new semantic meaning in that process is potentially interminable, and isn’t dependent on any conceivable algorithm that could arise from the mathematically quantifiable physical information the process employs. Life, then, and life’s evolution are not algorithmic processes; they’re processes of open intention and determinate meaning; they’re language. So let’s not play games here: meaning, intention, semantic content, finality—all of this has its real existence solely in the realm of mind. If life is code, then life subsists only in mind; but life is code; therefore . . . well, you can complete the syllogism for yourself.
