Biosemiotics, code biology, and operational interpretation

1 Introduction

Biosemiotics is dedicated to studying all forms of communication and signification found in and between living systems. Two primary concerns drive biosemiotics: First, meaning-making-seeking processes namely semiosis, constitute the very essence of living systems, yet the dominant paradigm of biology overlooks them. “Biology is incomplete as a science in the absence of explicit semiotic grounding” (Kull et al. 2011a: 28). Therefore, by focusing on semiosis, biosemiotics complements Darwinian biology, and together, they provide a comprehensive understanding of living systems (Hoffmeyer 2011: 43). Second, originally, Peirce’s aim in proposing semiotics was to “construct a naturalistic but nonreductive account of the human mind” (Short 2007: ix). However, semiotics presupposes interpretation which is teleological. Teleology conflicts with mechanistic explanations in science. As a result, semiotics needs scientific grounding to be a naturalistic account. Biosemiotics may offer tools for grounding sign theories (Kull et al. 2011a: 25).

With decades of development, biosemiotics has renewed its understanding and conceptions in many fields of biology with tools from semiotics and provides new insights and vitality for semiotics in return. Nonetheless, it is not free of challenges, and its development has not been plain sailing. A great pity for the development of biosemiotics was the split between biosemiotics and code biology. Both semiotic biology and code biology share the ontological commitment that meaning distinguishes living from nonliving systems, leading them to propose a new paradigm of life science, biosemiotics. However, the divergences between code biology and biosemiotics caused a split (Barbieri 2007, 2019). Firstly, they differ in how they understand meaning in living systems. Biosemiotics employs Peircean semiotics, which requires interpretation, whereas code biology argues that a code suffices and does not require interpretation. Secondly, they disagree on whether there is interpretation at the unicellular level. Biosemiotics posits that interpretation coexists with life, while code biology disagrees. Furthermore, they have different methodological concerns. Since interpretation is necessary for biosemiotics, it often appeals to a teleological explanation. In contrast, code biology defines codes by arbitrary rules realized through sequences of physical items, aligning with mechanistic explanations. Consequently, biosemiotics may conflict with the mainstream paradigm of modern science, which has excluded teleology from science since Descartes. In contrast, code biology does not share the same worry.

Due to these divergences, at the end of 2012, Marcello Babieri, the founder of code biology, “resigned as editor-in-chief of Biosemiotics and together with eleven colleagues founded the new research field of Code Biology, a field that was explicitly defined in the constitution of our society as ‘the study of all codes of life with the standard methods of science’” (Barbieri 2019). His departure was a significant loss for the field of biosemiotics. Efforts have been made to reconcile code biology with biosemiotics (e.g., Brier and Joslyn 2012; Vega 2018, 2021a, 2021b), but it is hard to say whether these efforts have been successful. In this paper, I will provide another attempt to reconcile the two fields. Contrary to Barbieri’s idea that “a scientific approach to meaning could not survive in that framework [biosemiotics],” I argue that there is no fundamental divergence between biosemiotics and code biology. Through providing an operational understanding of interpretation with Terrence Deacon’s thought experiment of autogenesis, we can explain the origin of the normativity of organic codes which code biology owes us an explanation for.

Accordingly, this paper has two aims: (1) to critically review the debate between biosemiotics and code biology, taking Jesper Hoffmeyer’s account of biosemiotics and Marcello Barbieri’s code biology as exemplars; and (2) based on Deacon’s account of autogenesis and interpretation, to propose the concept of operational interpretation to understand interpretation at the cellular level, which may help to reduce the divergence between the two research programs.

2 Life sciences without living

The mainstream that dominates the life sciences today is neo-Darwinism. It treats living processes within and between organisms as mechanical processes. As a result, biological phenomena are seen as continuous with other physical properties and processes, and thus as not exceptional within the universe. Biological phenomena that were once considered teleological are now reduced to physical processes. Evolutionarily, organisms are viewed as mediums through which genes are replicated from one generation to the next, constrained by natural selection. Developmentally, organisms are treated as nothing more than living machines, responding to instructions from genes. Accordingly, life is considered a lifeless machine according to the paradigm.

For example, the central dogma of molecular biology argues that the flow of genetic information in protein synthesis is a purely mechanistic process.

The Central Dogma. This states that once “information” has passed into protein it cannot get out again. In more detail, the transfer of information from nucleic acid to nucleic acid, or from nucleic acid to protein may be possible, but transfer from protein to protein, or from protein to nucleic acid is impossible. Information here means the precise determination of sequence, either of bases in the nucleic acid or of amino acid residues in the protein. (Crick 1958: 153)

According to the principle, the sequence information is first coded by the sequential structure of DNA. During replication, the sequence information is transcribed and encoded by messenger RNA (mRNA) through template matching (transcription). The mRNA, carrying genetic information, then forms a complex with ribosomes and transfer RNA (tRNA), serving as a template for protein synthesis. This process ensures that proteins fold into the correct structures according to the genetic information represented by the mRNA.

Although there are transfers of genetic information not covered by the central dogma, it still holds a significant place in molecular biology. Nevertheless, my concern here is not whether the dogma can be upheld, but whether it is purely mechanistic. As we can see, there are concepts with obvious teleological connotations, such as code, information, transcription, translation, and correctness. How can these concepts be coherent with a mechanistic explanation?

A common response to this issue suggests that people often use concepts metaphorically, believing they can all be ultimately explained in physical terms. This includes the concepts of code and information, which biologists use in a highly restricted sense of a fairly rich semantic notion (Godfrey-Smith 2000; Griffith 2001; Sarkar 1996). As Godfrey-Smith and Sterelny put it:

[…] there is one kind of informational or semantic property that genes and only genes have: coding for the amino acid sequences of protein molecules. But this relation ‘reaches’ only as far as the amino acid sequence. It does not vindicate the idea that genes code for whole-organism phenotypes, let alone provide a basis for the wholesale use of informational or semantic language in biology. (Godfrey-Smith and Sterelny 2016)

Moreover, some argue that the genetic code exhibits arbitrary characteristics because many other mappings between DNA base triplets and amino acids are biologically possible. Yet, the perceived arbitrariness of the genetic code is superficial, stemming from gaps in our understanding of the complex connections between DNA base triplets and amino acids. As Godfrey-Smith and Sterelny (2016) suggest, the very notion of arbitrariness in this context proves elusive.

A significant challenge to mechanistic understanding in biology is the normative dimension of the genetic code, where processes like transcription and translation are considered either correct or incorrect. Normativity inherently involves teleological concepts. How, then, can biology reconcile normativity within a mechanistic framework? A common response is to invoke natural selection. As Maynard Smith (2000: 179) famously stated, “Where an engineer sees design, a biologist sees natural selection.” Natural selection operates at the population level to shape sequence information encoded by nucleic acids. In the philosophy of science, this is often referred to as the etiological theory of purpose and function (Millikan 1984, 1989; Neander 1991; Wright 1973). According to this theory, specific sequence information exists because it historically led to the production of proteins that positively affected organism fitness. The production of these proteins represents the proper function of the sequence information. Importantly, since natural selection operates as a purely mechanistic process, normativity in biology, as explained by the etiological theory, does not imply teleology. Thus, teleological explanations are dismissed in favor of etiological explanations within this framework.

While widely accepted in biology and cognitive science, etiological theory faces significant challenges. Empirically, it appears overly idealistic. It relies on a simplistic and intuitive understanding of natural selection, which fails to encompass the complexity seen in actual evolutionary processes such as coevolution and frequency-dependent selection (Christie et al. 2022). However, it is not natural selection itself that is flawed, but rather etiological theory’s interpretation of it.

A critical challenge to etiological theory comes from García-Valdecasas and Deacon (2024). Apart from pointing out conceptual inconsistencies within the etiological framework, they argue that natural selection does not perform physical work. “This argument advances the idea that natural selection is a population-level theory that explains statistical changes, rather than a causal theory of force, from which it follows that the theory cannot account for the physical work that enables the persistence and reproduction of living organisations that is key to function” (García-Valdecasas and Deacon 2024: 25). In essence, while natural selection explains patterns in populations, the physical processes that underpin biological functions operate at the organism level. Therefore, García-Valdecasas and Deacon contend that natural selection alone cannot provide a causal explanation for biological functions. They argue that to function effectively, natural selection presupposes the existence of organisms, which are fundamentally responsible for the emergence of normativity. In this view, etiological theory, bolstered by natural selection, inherently implies teleology.

Moreover, beyond the example of the genetic code discussed in this context, biological explanations abound with concepts that inherently suggest teleology, such as adaptation, function, signal, messenger, and even mind-reading (Hoffmeyer 2008, 2010: 43). Despite assertions by mainstream biologists that these concepts are used metaphorically and can be reduced to purely physical terms, this reduction has proven elusive in practice. This discrepancy forms a core departure point for biosemiotics.

3 Biosemiotics

Biosemioticians argue that mechanistic explanations fail because they overlook the teleological and meaningful nature of living systems. “Living creatures are not just senseless units in the survival game; they also experience life” (Hoffmeyer 2008: XIII). In contrast to neo-Darwinism, which seeks to explain away teleology in biology, biosemiotics considers teleology to be genuine. It builds on the Sebeok Thesis, which states, “Semiosis is what distinguishes all that is animate from lifeless” (Sebeok 1988: 1089). Semiosis, or sign process, is a process of meaning-generation involving signs.

The concept of meaning occupies a central place in the notion of semiosis. We can understand it in biosemiotics in two senses. Firstly, it can be seen in a representational sense. A sign represents something to someone. This is the original meaning of semiosis. Secondly, meaning can be understood in terms of value. Semiosis becomes a process through which living systems generate value – every interaction with the environment becomes significant, contributing to their survival and reproduction. Conversely, objects in the environment are not neutral but carry value for living systems. These two senses of meaning are interconnected rather than isolated; they are complementary aspects of the same process. Organisms pursue the value of living through the representation processes. As Emmeche (2011: 94) aptly puts it, “A sign is anything that can stand for something (an object) in some interpreting system (e.g., a cell, an animal, a legal court), where ‘standing for’ means ‘mediating a significant effect’ (called the interpretant) upon that system.” In this view, living systems meaningfully live in a meaningful world through semiosis.

The Sebeok Thesis has two implications that define the aims of biosemiotics. One implication, which is relevant for biology, is that biosemiotics defines living systems through semiosis. Because semiosis involves teleology, the explanations biosemiotics offers for biological phenomena diverge significantly from those of neo-Darwinism. Biosemiotics to a large extent provides an alternative paradigm for biology (Anderson et al. 1984; Emmeche and Kull 2011; Zhou 2018). Therefore, the first aim of biosemiotics is to investigate biological phenomena from “the perspective of sign actions, processes, purposefulness, interpretation, and generality” (Emmeche 2011: 94).

The other implication of the Sebeok Thesis, which is relevant for semiotics, is the assertion that there is no semiosis without the living. Firstly, this means that “Signs and life are coextensive” (Sebeok 1991: 85; Stjernfelt 2002: 337), rejecting the idea of pansemiotics, which suggests that the universe is perfused with signs and everything is semiosis in its nature (Brier 2008: 357). Second, biosemiotics extends the concept of semiosis to encompass “meaningful communication in many other species besides Homo sapiens” (Kull et al. 2011a: 2). Traditionally, semiotics has been primarily a theoretical tool for understanding human language and culture, thereby confining meaningful communication to the realm of culture. This perspective leaves the origins of semiosis unexplained and widens the epistemic gap between nature and culture. By broadening semiosis to include non-human species, biosemiotics suggests that meaning is a natural phenomenon with a natural origin. Then the question becomes “How could natural history become cultural history?” (Hoffmeyer 1996: viii). Biosemiotics may provide a naturalistic explanation of meaning and therefore potentially bridge nature and culture (Anderson et al. 1984; Barbier 2007: xi; Cobley 2017: xii; Sharov et al. 2015). In doing so, this may develop “tools for grounding the sign theories” (Kull et al. 2011a: 25). This is the second aim of biosemiotics.

In order to achieve its aims, biosemiotics primarily employs two theoretical frameworks: Peircean semiotics and Jacob von Uexküll’s theory of Umwelt (1940). Semiosis as a central concept of biosemiotics comes from Peircean semiotics. According to one of Peirce’s most influential definitions:

A sign […] 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. (CP 2.274)

This paragraph presents the formal conditions of semiosis: the presentative, the representative, the interpretative, and the triadic conditions (Liszka 1996: 18–34). Due to space limitations, I will focus on the interpretative condition, which highlights a key divergence between biosemiotics and code biology. This condition requires that for a sign to be a sign, it must be interpreted as such (Liszka 1996: 24). Interpretant is the significant effect upon interpreter produced by the sign. Unlike dyadic models of signs prevalent in linguistics and cultural studies, which posit that a sign comprises a signified and a signifier, Peircean semiotics insists on interpretation by an interpreting system. Clearly, the Peircean triadic model of semiosis implies interactions within and between an interpreting system and its surrounding environment through semiosis. The enactive structure is further elucidated by von Uexküll’s theory of Umwelt.

According to the theory of Umwelt, each organism inhabits its own Umwelt – a meaningful, subjective world centered around the organism where everything holds significance for it. Combined with Peircean semiotics, this perspective posits that all interactions within and between organisms are semiosis. When an organism interacts with an object, the object ceases to be value-free anymore and becomes a meaning-carrier for the organism. The interpretation of the organism plays a central role in this transformation. The organism interprets the object as having significance for it. Consequently, the interaction processes form a functional circle in which meaning is generated. A functional circle is a semiosis realized by the interaction. Functional circles (semioses) constitute the Umwelt of the organism.

According to this framework, processes within and between organisms are seen as semioses rather than merely physical and chemical processes. Biosemioticians contend that biological phenomena can be more comprehensively understood by this framework. As argued in the previous section, traditional approaches to explaining the flow of genetic information in processes like protein synthesis and reproduction encounter difficulties because they often necessitate the use of teleologically involved concepts and struggle to eliminate teleology from explanations. Now let’s explore how we can understand the flow of genetic information within the framework of biosemiotics.

In their groundbreaking work, Hoffmeyer and Emmeche (1991) argue that the flow of genetic information in protein synthesis should be understood as interpretive acts, aka, semiosis. They propose that the concept of information needs to be redefined through Peircean semiotics. If genetic information is understood simply as the determination of sequence, it becomes indistinguishable from other forms of organization in nature. What distinguishes genetic information and other biological information during ontogenetic development is the semiotic nature of genetic information– genetic information function as signs that stand for something within interpreting systems. Although genetic sequences are relatively independent and stable within cells, it should not be understood in this isolated sense. Instead, sequence information should be understood within the whole context of ontogenetic development and reproduction in which it is interpreted. Hoffmeyer and Emmeche formulate semiosis at the organism level this way:

In the epigenetic process, DNA, i.e. the genome, may be seen as just one fragment of an evolutionary stream of signs passed down through the generations. The interpretant selecting such signs among the myriad of internal cytoplasmic differences is the fertilized egg, the zygote. The fertilized egg, the real ‘person’ of biology, can make sense of the elaborate message contained in the DNA, using it to master the epigenetic process, i.e. the construction of the phenotype, the actual organism. (Hoffmeyer and Emmeche 1991: 143)

In this semiosis, “[s]till, it is the zygote which is the subject in the process: It initiates the deciphering of the DNA-message and becomes gradually changed to the embryo in response to the interpretation” (Hoffmeyer and Emmeche 1991: 144). Unlike molecular biology, which interprets ontogenetic development as biochemical processes guided by genetic information, biosemiotics regards it as a form of semiosis. This perspective emphasizes that the process can only occur within an interpreting system, such as the zygote. Because the zygote acts as an interpreting system, the normativity of the genetic code (information) is no longer a mystery.

This perspective clearly departs from a reductionist explanation of genetic information that merely equates it with the sequence of nucleic acids. Instead, biosemiotics interprets genetic information within the framework of semiosis, occurring within the context of the whole organism. Critics may argue that biosemiotics resembles disguised vitalism, as it seems to presuppose rather than explain the mysteries of life. However, Hoffmeyer (2008: 8–12) has argued that although biosemiotics shares some rational concerns with vitalism, it is not a disguised vitalism but proposes a new promising way to investigate life that transcends both mechanistic and vitalist paradigms.

As the explanation emphasizes the centrality of interpretation, Deacon summaries the central dogma of semiotics in contrast to that of molecular biology:

Any property of a physical medium can serve as a sign vehicle of any type (icon, index, or symbol) referring to any object of reference for whatever function or purpose because these properties are generated by and entirely dependent upon the form of the particular interpretive process that it is incorporated into. (Deacon 2021: 539)

However, the centrality of interpretation in biosemiotic explanations led to the split between biosemiotics and code biology.

In the early days of the cooperation between the two schools, Barbieri was clearly aware of the divergences of code semiotics from biosemiotics (Barbieri 2002, 2006, 2007). In his recent response to his critics and endeavors to reconcile code biology with biosemiotics (Vega 2018), Barbieri (2019) argues why he is not satisfied with biosemiotics, once again adhering to code biology. Barbieri’s argument revolves around two key issues – mechanism and interpretation – leading to the divergence from biosemiotics (Barbieri 2019: 23). The mechanism issue is a methodological problem. Barbieri contends that what defines modern science as true science is its reliance on mechanistic explanations. Biosemiotics, by adopting a non-mechanistic approach to life, poses an inherent incompatibility with modern scientific methodology, which emphasizes mechanism. This prevents biosemiotics from being a true science. Barbieri further argues that biosemiotics misconstrues mechanism. For him, mechanism does not entail reductionism, Newtonian determinism, or strict physicalism, concepts which biosemiotics critiques. Instead, mechanism is best understood as scientific modeling, “in the sense that a mechanism is described by a model, and a model is implemented by a mechanism” (Barbieri 2019: 23). If we accept this broader understanding of mechanism, it could be argued that Peircean triadic semiosis and von Uexküll’s functional circle also serve as forms of modeling. The question then arises: why does Barbieri reject these models as scientific modeling?

The crux of Barbieri’s disagreement lies in the interpretation issue. As the central dogma of semiotics claims, interpretation takes up the central place of the explanation of meaning in biosemiotics. In other words, interpretation is necessary for meaning-production. Barbieri agrees that interpretation can attribute meaning to objects, acknowledging its role at the level of the brain. However, he contests the notion of interpretation occurring at the cellular level. Barbieri argues that interpretation relies on abduction, which necessitates memory – a capacity only found in sophisticated brains (Barbieri 2019: 23, 2007: 203). Interpretation is unnecessary to meaning; coding, which does not require interpretation, is sufficient for meaning. The cell is a meaning-producing system but not a semiotic system (Barbieri 2007). Biological processes at the cellular level, such as protein synthesis, are viewed by Barbieri as deterministic chains of biochemical reactions devoid of interpretation. It is a completely automatically deterministic process. There is no space for interpretation, as interpretation is to make context-dependent choices between different meanings (Barbieri 2007: 203). According to Barbieri, single cells lack the ability to construct internal representations of the world and thus cannot interpret (Barbieri 2019: 25). Therefore, treating biological processes as semiosis implying interpretation is a mistake. It follows an idealistic tradition with teleological residues (Barbieri 2019: 23). It is incompatible with modern scientific methods. While Barbieri acknowledges the importance of meaning in biology, he asserts that Peircean biosemiotics is not the appropriate framework.

4 Code biology

Code biology aligns with biosemiotics in recognizing that organisms are more than mere chemical machines, that information is crucial to living systems, and information requires meaning (Barbieri 2003, 2006, 2015a, 2015b). However, it diverges by proposing a mechanistic approach to understanding meaning within and between living systems.

Barbieri argues that the core of modern scientific methods, which emphasizes mechanism, is sound for understanding biological phenomena. He advocates shifting the paradigm of life from viewing organisms as mere chemical machines to recognizing them as information processing machines (Barbieri 2019). He asserts that it is information and meaning that distinguish living systems from lifeless matter (Barbieri 2006). Barbieri critiques the central dogma of molecular biology because it encompasses both chemical and information paradigms, which he views as incompatible (Barbieri 2019). The chemical paradigm depicts life as complex chains of biochemical reactions, while the information paradigm posits that life involves information processing through these reactions. However, these two paradigms are incompatible. As we have argued above, treating the concept of information and other meaning-involved concepts as metaphor is insufficient; information is fundamentally different from chemistry in that it is digital rather than analogue. While he agrees with biosemioticians that information and meaning are real, Barbieri seeks a mechanistic approach to understanding them.

The central conception code biology used to understand life is code. Barbieri cites Erst Myr’s statement emphasizing the centrality of code for life, “The discovery of the genetic code was a breakthrough of the first order. It showed why organisms are fundamentally different from any kind of nonliving material. There is nothing in the inanimate world that has a genetic program which stores information with a history of three thousand million years!” (Mayr 1982: 124; cited from Barbieri 2015b: 3). Then, what is code? What is special about code?

In molecular biology, genetic code almost equals genetic information, specifically the sequence of nucleoid acid. Barbieri argues that this formulation fails to distinguish biological processes from spontaneous chemical processes. He thinks that codes possess three fundamental characteristics: “(1) They are rules of correspondence between two independent worlds. (2) They give meanings to informational structures. (3) They are collective rules which do not depend on the individual features of their structures” (Barbieri 2003: 94). Accordingly, code is defined as a mapping between the objects of two independent worlds that is implemented by the objects of a third word called adaptors that perform two independent recognition processes (Barbieri 2003: 93). Meaning is also defined by codes: “The meaning is an object which is related to another object by a code” (Barbieri 2006: 4).

The fundamental distinction between the chemical paradigm and the information processing paradigm lies in the concept of arbitrariness within the code. When we describe a relationship between two objects as arbitrary, we mean that the connection is not inherently necessary but rather established by specific rules. According to Barbieri, a code consists of “a small set of arbitrary rules selected from a potentially unlimited number in order to ensure a specific correspondence between two independent worlds” (Barbieri 2015a: xii). Consequently, Barbieri asserts that “life is artefact-making,” suggesting that all biological structures and processes, including genes and proteins, are fabricated by molecular machines (Barbieri 2015b: 3). The arbitrariness of codes comes from the mapping rules, which are conventional. With the account of code in hand, let’s see how code biology explains codes and information within and between living systems.

Barbieri refers to the codes present in living systems as organic codes. Unlike cultural codes, which are social conventions, organic codes are natural conventions. He proposes three conditions to identify organic codes in biological systems: “(1) two independent worlds of molecules, (2) a set of adaptors that create a mapping between them, and (3) the demonstration that the mapping is arbitrary because its rules can, at least in principle, be changed” (Barbieri 2015a: xii).

The genetic code exemplifies an organic code, pivotal in protein synthesis. Barbieri contends that, unlike the central dogma of molecular biology, which views protein synthesis as a series of chemical reactions, code biology treats these processes as codified procedures (Barbieri 2003: 96–101, 2015a: xiii).

As introduced above, in protein synthesis, genetic information encoded in nucleotide sequences is transcribed to RNA and subsequently translated into amino acids, which fold into the structure of proteins. This process produces two independent worlds of molecules: nucleotides and amino acids. Condition 1 is satisfied. Then, what is the adaptor connecting the two worlds? In protein synthesis, the flow of genetic information occurs in two distinct steps: transcription and translation. During transcription, the sequence information encoded by nucleotides in DNA is transcribed to mRNA through template matching. This is an analogue process with no third party participating in it. Importantly, because transcription is isomorphically determined by the sequence of nucleotides in DNA, it is not arbitrary. Therefore, transcription is not considered a codified process. What about translation? In translation, mRNA initially associates with a ribosome, forming a ribosome-mRNA complex. Subsequently, tRNA molecules become involved. On the one hand, tRNA molecules match the codons on mRNA with their complementary anti-codons. On the other hand, each tRNA carries a specific amino acid residue, which corresponds to certain genetic information required to add to the polypeptide chain being synthesized. Once the appropriate amino acid is brought in by tRNA, the polypeptide chain can fold into its correct three-dimensional structure. The connection established by the sequence information is indeed arbitrary, fulfilling Condition 3 of the organic codes. Thus, tRNA serves as the adaptor, ensuring a specific mapping between the worlds of nucleotides and amino acids, satisfying Condition 2. Consequently, translation is recognized as a codified process. Given that translation is integral to protein synthesis, the entire process is considered codified.

From the discussion above, it is evident that protein synthesis operates without the need for interpretation. The coding provided by genetic codes is sufficient for the flow of genetic information from DNA to proteins. The explanation offered by code biology is mechanistic rather than teleological, aligning well with modern scientific methodologies. Consequently, code biology presents a more robust approach for addressing meaning in biology compared to biosemiotics.

While I don’t align with all of Vega’s (2018) criticisms of code biology, I believe it faces significant challenges. In the debate with biosemiotics, one such challenge is explaining the normativity inherent in organic codes. Biosemiotics posits that the normative aspect of semiosis stems from interpretation, as argued in the previous section. Organic codes, being arbitrary and governed by rules, also exhibit normativity – they dictate correct or incorrect mappings between two independent worlds. The question then arises: Where does this normativity originate in organic codes? A natural answer to it is the codemaker: “signs and meanings are mental entities when the codemaker is the mind, but they are organic entities when the codemaker is an organic system” (Barbieri 2003, 2015a: 32).

However, this answer presents a conceptual inconsistency. Barbieri suggests that in cultural contexts, the codemaker of codes is the mind, while the connections between signs and meanings are established solely through arbitrary rules, i.e., conventions (Barbieri 2003, 2015a: 32). This distinction is perplexing because the mind is an individual entity, whereas conventions arise from social interactions. Recent studies on Lewis–Skyrm’s signaling games suggest that the establishment of rules and meanings through conventions does not necessarily require a mind (Barrett 2009; Hutteger 2007, 2014; Lewis 1969; Skyrms 1996, 2010). This perspective aligns with the Darwinian account of natural selection and etiological explanations. However, Barbieri’s concern lies with organic codes, which are natural conventions (Barbieri 2003, 2015a: 32). Nevertheless, the potential explanation offered by code biology faces similar challenges.

As demonstrated earlier, organic systems act as the codemakers of organic codes, which are considered natural conventions. But how can we make sense of this claim? From what I gather, Barbieri offers descriptions and classifications of various types of organic codes without addressing the source of their normativity. He may face a conceptual dilemma where the only acceptable answers are untenable for him. One approach could be to explain organic codes as natural conventions. According to the framework of Lewis–Skyrms signaling games, there’s no inherent distinction between social conventions and natural conventions; both emerge from interactions (Godfrey-Smith 2014). However, similar to etiological theory, this approach risks either eliminating meaning or presupposing interpretation from the outset. Moreover, Barbieri explicitly argues that natural convention differs significantly from natural selection (Barbieri 2015a: 178–179).

Alternatively, one might appeal to organic systems themselves. But if we interpret this in a mechanistically deterministic manner, akin to the analysis of the genetic code earlier, the source of normativity remains elusive because the analysis presupposes it – a kind of circular reasoning. The only viable option left seems to be that, akin to the mind as a codemaker imbuing signs with meaning, organisms confer meaning onto codes. In other words, organisms possess interpretive competence. However, this directly contradicts Barbieri’s stance against interpretation at the cellular level.

One further significant challenge faced by code biology is the continuity issue between conventional codes and organic codes. Are conventional codes fundamentally different from organic codes, or are they continuous with each other? Code biology makes a clear distinction between them, implying potential qualitative differences. Yet why classify both as “codes” if they are essentially different? If they exhibit continuity, how do we bridge the gap between organic and conventional codes? Moreover, how does social convention emerge from natural convention, and how does interpretation arise from this context? These questions remain open within code biology. While Barbieri has addressed some aspects in his work, I won’t delve into these due to space constraints. Instead, I will focus on the primary concern: the problem of interpretation. I will propose a new perspective, operational interpretation at the cellular level, aligned with Barbieri’s methodological focus. This approach aims to facilitate a potential reconciliation between code biology and biosemiotics.

5 Interpretation in Peircean semiotics

The debate between biosemiotics and code biology revolves around whether interpretation exists at the cellular level, alongside relevant epistemological and methodological concerns. I find Barbieri’s methodological concerns reasonable; scientific methods define modern science, and we shouldn’t disregard them. Therefore, rather than displacing neo-Darwinian biology, biosemiotics should seek integration with it, which aligns with its overarching goal (Kull et al. 2011b). Specifically, given interpretation plays a central role in biosemiotics, efforts should focus on providing a naturalistic explanation of interpretation. This entails identifying the mechanistic and dynamic conditions necessary and jointly sufficient for interpretation, thereby establishing a robust scientific model of this phenomenon.

However, I diverge from Barbieri’s interpretation of Peircean semiotics. According to Barbieri (2019: 24), interpretation involves advanced cognitive abilities such as abduction, extrapolating conclusions from limited data, and constructing internal representations. It is true that Peirce only talks about signs produced and understood by humans. While Peirce primarily discusses signs in the context of human cognition, his aim was to develop a naturalistic but nonreductive account of the human mind (Short 2007: ix). The concept of interpretation in Peirce’s work is indeed ambiguous, yet this ambiguity does not restrict Peircean semiotics solely to human cognition. Peircean semiotics provides a normative framework: if a process meets the conditions of signs (the presentative, the representative, the interpretative, and the triadic), it constitutes semiosis. Consequently, semiosis is pervasive in the biological world, acknowledged by both biosemiotics and code biology. The fundamental divergence between these two fields of research lies in their understanding of whether a single cell can be considered an interpretive system. As discussed in the preceding sections, biosemiotics affirms this possibility while code biology rejects it. Therefore, the key to reconciling biosemiotics with code biology lies in clarifying our understanding of interpretation.

Several approaches have been taken in this direction. Brier and Joslyn (2012) propose discriminative interpretation as a semiotic foundation for code semiotics. Discriminative interpretation requires an interpreting system to discern differences within and between objects, thereby examining a sign and determining its meaning. Vega (2018), drawing from Rosen’s relational biology, interprets cellular-level interpretation as organisms’ anticipatory actions based on persistent future state expectations of themselves and their environment. This implies organisms possess an internal predictive model of the world, which Vega identifies as interpretative competence. Barbieri is not satisfied with these solutions. He thinks that these solutions simply define decoding as a form of interpretation. They are just ways of playing with words, “[Biosemiotics] claims that interpretation takes place at the cellular level because we can define interpretation in such a way that it is present at the cellular level” (Barbieri 2019: 25). This strategy does not work, because it is a straw man that changes the very meaning of interpretation.

I agree with Barbieri’s concerns about these solutions. However, I disagree with his assertion that interpretation in Peircean semiotics must exclusively involve high-order cognitive activities. In Peirce’s original works and general introductions to Peircean semiotics, certain examples of signs and semiosis can mistakenly lead people to view interpretation as solely a sophisticated cognitive process.

For instance, in the opening paragraph of Chapter 2, “Semeiotic grammar,” Liszka’s influential introduction to Peirce’s semiotics, he introduces the four formal conditions that define a sign. He presents the interpretative condition as,

(3) The sign must determine (potentially or actually) an interpretant, understood as a sign which translates and develops the original sign (Liszka 1996: 18–19).

When explaining the triadic condition, he says,

Each of the first three formal conditions of the sign is mediated through the others: the ability of the sign to represent also requires, inherently, its power to be interpreted as a sign of the object in some respect; the ability of the sign to be interpreted can only work if it is interpreted as representing an object in some respect; and it can only be understood as representing an object in some respect if it is interpreted as representing an object as such. (Liszka 1996: 19)

As evident from these examples, phrases like “understood as” and “interpreted as” abound, potentially leading readers to conceptualize semiosis as two categorically distinct orders of processes. One order, considered lower, involves basic information processing, while the higher order supposedly interprets the lower as something else. This analogy mirrors the philosophical distinction between sensation and perception in cognitive science and philosophy of mind. Sensation as the first order process is information processing while perception as the second order categorically organizes the products of the first order. Unlike Barbieri, I find this anthropocentric framing misleading. It tends to use human cognition and language as the primary examples. Peirce and other semioticians, like Liszka, understandably adopt this language since Peirce’s focus was on the mind. However, Peirce’s semiotic model of signs is quite general. Mental entities represent just one idiosyncratic form of signs, as demonstrated by biosemiotics. In my view, any process that meets the formal conditions of signs constitutes semiosis.

When discussing Peirce’s triadic model of signs, another common misunderstanding arises. Peirce conceptually divides a sign into three components: object, representamen, and interpretant, and then explores their logical relationships. Some may perceive these three parts as entirely independent entities that are integrated solely through interpretation. In cultural contexts, signs exist independently of interpreters, forming the semiosphere of human society. However, physical independence of these three parts is not necessary for semiosis. As long as the functions of object, representamen, and interpretant are fulfilled, semiosis occurs. Therefore, different aspects of an event may serve different functions at various stages, collectively constituting a semiosis. In the next section, I will demonstrate how this dynamic operates in the simplest interpretive systems. So, what exactly is interpretation in Peircean semiotics?

We might find a clue in the explanation of interpretant. Generally, interpretant can be understood as the translation of a sign: “a sign is not a sign unless it translates itself into another sign in which it is more fully developed” (CP 5.594, cited from Liszka 1996: 24). Translation has a significant effect on interpreter, which produces another system of sign. This explanation is self-nested: interpretant, as a constitutive part of a sign, is itself a sign. I do not view this as a vicious circle. Instead, it shows that we should understand interpretation as the competence to make a sign possible, that is, to make a sign stand for an object in some respect for the interpreter. In conclusion, we may say that an interpretive process is a semiosis. It is an interpretative system that has the competence to interpret a sign as a sign. Therefore, to understand interpretation in semiotics, we should understand what kind of system possesses interpretive competence. This is the approach adopted by Deacon (2021) in his exploration of interpretation.

6 Autogen, code-duality, and operational interpretation

In his recent paper “How molecules became signs,” Deacon argues that to address the question “What sort of process is necessary and sufficient to treat a molecule as a sign?” we must concentrate on the interpretive system (Deacon 2021). For him, to explain an interpretive system is to explain what kind of a system is a living system. Since 2006, Deacon has repeatedly proposed a thought experiment concerning the simplest teleological system, an autogenic virus (Deacon 2006, 2012, 2020).

The simplest autogenic virus is realized by two reciprocal self-organizing processes: reciprocal catalysis and self-assembly (Deacon 2021). Reciprocal catalysis involves two catalytic reactions where each reaction’s product catalyzes the other reaction. Given sufficient substrate molecules, this reciprocal catalysis persists and can expand into a network of cyclic reaction chains. Self-assembly, on the other hand, is a type of molecular aggregation process where energetically favored molecular components aggregate spontaneously in localized regions. This process results in the formation of large, closed regular structures such as polyhedral or tubular capsid structures.

These two processes complement each other. When coupled, each provides critical boundary conditions for the other. As the reciprocal catalysis network grows, it becomes increasingly fragile and prone to diffusion. However, when the self-assembled capsid forms around this network, it isolates it from external interference, effectively blocking catalyst dissipation. In turn, maintaining the capsid structure requires a persistently high local concentration of specific molecular components, which the reciprocal catalysis network can produce as a byproduct. When these processes are integrated, the once externally imposed critical boundary conditions become intrinsic to the new whole. Deacon terms this integrated system an “autogen.”

When reciprocal catalysis initiates where appropriate substrate molecules are present, it produces a single species of component molecules that tend to self-assemble into a closed capsid structure. This structure then encapsulates the reciprocal catalysis, forming an autogen. As the substrate inside the capsid is used up, reciprocal catalysis ceases, leading to a halt in the production of component molecules. The capsid becomes susceptible to damage without a continuous supply of components, eventually opening up again. If there are appropriate substrates nearby, reciprocal catalysis can restart. With a new supply of components, the capsid self-repairs. If damage is extensive enough to split the capsid into several parts, these independent parts may separately restart, forming different autonomous autogens. Thus, autogens have the capacity for self-reproduction.

As we observe, an autogen is physically realizable. One could even argue that the two self-organizing processes – reciprocal catalysis and self-assembly – are mechanistically determined. Therefore, this model is mechanistic and meets Barbieri’s requirement for scientific methods. However, none of these processes satisfy the conditions of organic codes. According to Barbieri’s framework, this means autogenesis lacks meaning-making capabilities. Nevertheless, Deacon contends that the autogen exhibits normativity and interpretive competence (Deacon 2021: 546). Describing it as an interpretive system with normativity appears to introduce an ad hoc property into a mechanistically determined model. One might argue that a mechanistic explanation suffices and that further teleological explanation is unnecessary. According to Occam’s Razor, which favors simplicity in ontology and advises against adding entities unnecessarily, adhering to the mechanistic explanation of the model may be preferable over introducing another dimension to it. Why then does Deacon assert that the autogen possesses interpretive competence?

In addition to normativity and interpretive competence, Deacon argues that the autogen possesses three other holistic properties: individuation, autonomy, and recursive self-maintenance (Deacon 2021: 546). Through its capsid, the autogen distinguishes itself physically and functionally from its surrounding environment. It achieves autonomy because “it intrinsically embodies and maintains its own boundary conditions via component processes that reciprocally produce the external boundary conditions for each other.” Furthermore, “it repairs and replicates the critical boundary conditions that are required to repair and replicate these same critical boundary conditions” (Deacon 2021: 546), thereby achieving recursive self-maintenance. The autogen can be seen as an intrinsically teleological system in a Kantian sense: every part exists through all other parts and in turn, every part is for other parts and for the whole (García-Valdecasas 2022; Weber and Varela 2002). Every process constituting the autogen is for the sake of the existence of it. Critics like Barbieri may express concern that introducing teleology into science conflicts with modern scientific principles. However, as demonstrated, the concepts discussed here are experimentally feasible and align with Barbieri’s own criteria for scientific methodology.

With the emergence of intrinsic purpose, normativity emerges. As each process acts for the persistence of the other, it may fail. For instance, damage to the capsid or unintended involvement of other molecules in reciprocal catalysis may disrupt the self-assembly of component molecules, leading to autogen failure. With the clarification of normativity, we can understand the interpretive competence of the autogen, which can be understood through code-duality.

As previously introduced, genetic codes, mRNA, tRNA, and amino acids each perform distinct roles in protein synthesis: coding and storing genetic information, transcription, translation, and construction. Hoffmeyer and Emmeche (1991) employ code-duality to elucidate how semiosis functions in protein synthesis. Code-duality refers to the recursive transmission of information through the interactions between digital and analogue codes (Hoffmeyer 2008: 80). Digital codes refer to codes whose components are discrete tokens and connected by arbitrary relations. Analogue codes, on the other hand, rely on similarities in spatio-temporal, part-to-whole, or causal continuity. Digital codes are for memory to store information, while analogue codes are for decoding construction and instructing physical realization. According to the concept of code-duality, genetic codes function as digital codes, while the processes of transcription and translation operate as analogue codes. Similarly, interpretation (semiosis) within autogens also operates through code-duality, albeit in a distinct manner.

Unlike the code-duality observed in protein synthesis, autogens do not exhibit physical differentiation. As previously discussed, the two self-organizing processes in autogens mutually provide critical boundary conditions. The boundary conditions provide constraints limiting the possible realization of the two processes the autogen can perform and thus maintain the autogen as a whole. Through the self-maintenance and self-reproduction of the autogen, the constraints are preserved. “This preservation of constraints […] provides [both] a record and a source of instruction for organizing the work required to preserve this same capacity” (Deacon 2021: 545). It means that the constraints embodied in the interplay of these two coupled processes serve dual roles akin to both digital and analogue codes.

The constraints in autogens are digital because they are arbitrary. If reciprocal catalysis and self-assembly were uncoupled, each could proceed in numerous other ways. However, when coupled, they mutually constrain each other to operate in a much more limited set of possibilities to sustain the autogen as a whole. Moreover, the dynamic structure of the autogen re-represents its own boundary conditions. Essentially, these constraints store the sequence information of the autogen itself. When one autogen reproduces another one, these constraints are preserved and intrinsically represent and reproduce the critical boundary conditions necessary for the autogen’s existence (Deacon 2021: 546). On the other hand, these constraints are analogue, because they are realized through spatio-temporal (self-assembly) and causal continuity (reciprocal catalysis). We can describe the dynamic structure of an autogen as digitally coding information about its boundary conditions. When damaged, remaining parts initiate an analogous decoding process to self-repair or self-reproduce, given adequate environmental conditions.

This self-referential code-duality realizes the most primitive interpretation implementing basic semiosis. The dynamic structure of an autogen represents itself (the representative condition); it re-presents the critical boundary conditions of the autogen (the presentative condition). When an autogen is damaged, its integrity is disrupted; with initiation of self-repair or self-reproduction (interpretant), the new autogen(s) is generated (the interpretative condition). The relations between the autogen (object), dynamic structure (sign), and regenerated autogen (interpretant) constitute genuine triadic relations that cannot be reduced to any dyadic relation between any two of them (the triadic condition). Therefore, autogenesis meets the four formal conditions of signs given by Peirce. The next question is what kind of semiosis autogenesis is. Since it involves the regeneration of self-/non-self-distinction, it is considered an iconic semiosis (Deacon 2021: 546).

This primitive form of semiosis possesses several properties distinguishing it from semiosis in the general sense. Firstly, its reference is itself, making it inherently self-referential. Secondly, the autogen (object), the dynamic structure (which is the constraints encoding the information of the dynamic structure itself) (sign), and the regenerated autogen (interpretant) exist not as separate entities but as a persistent unity, through autogenesis. Thirdly, it provides a diachronic instantiation of semiosis, contrasting with synchronic forms. The primitive semiosis manifests in the dynamics of autogenesis, which self-referentially structures itself in irreversible time. Fourthly, the dynamics of the autogen are physically determined and thus mechanistic, aligning with modern scientific methods and experimental feasibility. Due to these characteristics, I propose calling this primitive interpretation “operational interpretation.” Other forms of interpretation (semioses) can be differentiated from operational interpretation, as Deacon (2021) has argued.

With the explanation of operational interpretation (semiosis), let’s revisit the debate between biosemiotics and code biology. This account elucidates the source of normativity in codes and semiosis. Operational interpretation involves no mysterious elements despite its teleological nature. It is simple, involving coupled self-organizing processes rather than being cognitively intensive. Moreover, it initiates a semiotic framework from which other forms of semiosis can evolve, potentially explaining the continuity between different interpretations. Given these advantages, operational interpretation deserves recognition as the ground of signs and codes. Nevertheless, Barbieri may still argue against this approach, viewing it as another form of redefinition strategy.

7 Conclusions

The alliance between biosemiotics and code biology greatly advanced the semantic understanding of biology. However, the split between them arose due to their divergent views on cellular-level interpretation and related epistemological and methodological concerns, resulting in a significant loss for biosemiotics. Meanwhile, code biology faces a conceptual challenge in explaining the normativity of organic codes without biosemiotics. Critiques of biosemiotics by code biology are reasonable and warrant serious consideration. Based on Terrence Deacon’s thought experiment of autogenesis and his explanation of interpretation, this paper proposed the conception of operational interpretation to reconcile biosemiotics with code biology. On the one hand, the conception is mechanistically realized and experimentally possible; on the other hand, it explains the source of normativity of semiosis and organic codes. It helps us understand interpretation at the unicellular level, narrow the divergences between biosemiotics and code biology, further our understanding of semiosis, and thus contribute to the development of semiotics and biosemiotics.