The key idea of Nick Enfield’s target paper (Enfield, 2026) is that through cultural and social evolution humans have developed systems of coordination that strongly constrain them in their ways of understanding and acting. Adopting a phrase coined by Max Weber, these systems form the equivalent of an “iron cage” that keeps people within a rigidly defined framework of categories, norms and rules. However, Enfield also notes that while such frames restrict individual freedom, they also enable collective action, thus providing humanity with the superhuman power of an “exoskeleton”. He conceives these frames of coordination as codes: systems of well-defined signs, interpretations and rules. Examples are languages, institutions, and organizations, such as firms or administrations. Each of those define specific categories of conditions, ways of communicating these conditions, and actions appropriate for each condition.
Enfield’s paper touches on several topics that I have investigated in my own work, albeit from a different perspective. Enfield’s perspective derives mostly from anthropology, linguistics and economics, mine from complexity science and cognition. Still, in an early paper I used the same term of code (Heylighen, 1984), while in more recent work I have been emphasizing coordination (Heylighen, 2013) and collective intelligence.
In my analysis, a coordination system consists of what I call distinctions and connections. Distinctions are categories of conditions about which agents (individuals or organizations) agree. For example, we agree about which animals we would categorize as “dogs” and how these are distinct from the separate category of “cats”. Cognitively, such categories can be interpreted as concepts, i.e. units of perception and thought. Linguistically, different concepts are expressed by different symbols—typically words, such as dog and cat. As Enfield notes, the symbols or signs are essentially arbitrary. Yet, for effective communication and coordination, different individuals must agree about which symbol denotes which concept. Thus, an effective code requires alignment between its users about both symbols (signifiers) and the concepts or meanings they represent (signifieds).
However, to support coordination, codes need more than a consensual mapping between symbol and concept (semantics). They also need to agree about the actions to be performed when encountering a condition described by a combination of symbols (pragmatics). To describe this connection, I have adopted an idea originally proposed in artificial intelligence: a condition-action rule (Holland et al., 1989; Russell & Norvig, 2009). Such a rule simply states that in condition X, an agent must perform action Y. The short notation is:
if X is the case, then do Y,
or even shorter: X → Y.
X and Y here are fundamentally distinctions, i.e. distinguishable categories of conditions or actions. The arrow “→” expresses the connection between these distinctions. The action Y typically results in a new condition Z: Y → Z. For simplicity, we can collapse the notation, and just note the final condition: X → Z. That means that all relevant processes can be described by condition → condition rules (also known as “productions” in AI). That now allows describing sequences of processes, e.g.:
A → B, B → C, C → D, …, Y → Z.
Such sequences can also branch or bifurcate, e.g.:
A → B1, A → B2,
and different branches can come together, e.g.:
B1 → C, B2 → C.
Such branching and converging sequences of productions allow us to describe complex processes as “workflows”—as we might for example find in a company or administration. Coordination then means that this network of productions is coherent: there are no gaps, ambiguities or inconsistencies (Heylighen, 2013). One further extension of the formalism is that we can analyze complex conditions as conjunctions of simple conditions, e.g. A + B → C + D. That means that if conditions A and B are both present, then some agent processes them into the new conditions C and D.
This notation is similar to the one used in chemistry to describe reactions between molecules. The networks formed by such reactions are not merely descriptive; they can be analysed mathematically. That inspired a modelling approach called chemical organization theory (Dittrich & Fenizio, 2007; Heylighen et al., 2024; Veloz & Razeto-Barry, 2017). It allows an algebraic and dynamical analysis of the self-maintaining systems of processes that can emerge from such networks. That in turn allows us to model self-maintaining social and cultural systems, such as worldviews or enduring conflicts (e.g. Veloz et al., 2026; Veloz & Maldonado, 2022).
My intention here is not to go into the mathematical modelling and computer simulation enabled by such representations. I just wish to point out that such models can help us to understand the self-organization of codes and, more generally, systems of coordination. But let me delve deeper into the notion of self-organization, which is in a sense the core of my approach. Although Enfield (2026) uses the term “self-organizing” only once in a footnote, his overall argument can be viewed as a description of how a coordinated system can self-organize in a community of interacting individuals. For example, he notes that “the cultural items that circulate in a population of dyads converge, cohere, and hold together to create a functional cohesion resembling that of a higher-level integrated system like an organism, ecosystem, or city” (Enfield, 2026).
Such convergence of dyadic interactions resulting in a coherent system is precisely what I have analyzed as self-organization (Heylighen, 2001, 2025). Self-organization is commonly defined as global order emerging from local interactions. Self-organization is one of the key concepts in the science of complexity. It is characterized by properties such as positive and negative feedbacks, path-dependence, non-ergodicity, and sensitive dependence on initial conditions.
The self-organization of codes is easy to illustrate in the case of language. This was demonstrated by the pioneering simulations of the origin of language done by Luc Steels (Steels, 1998, 2012). The condition-action rules in the case of linguistic symbols are simple: a particular concept (e.g. the category of all dogs) must be consistently expressed by a particular symbol, word, or “name” (e.g. dog, chien, or canis familiaris):
concept → name.
If different individuals use different names for the same phenomena, then there is likely to be misunderstanding, conflict or friction. If they use the same name, then there is semantic alignment and therefore efficient communication.
Steels’s “Talking Heads” experiments with software agents (and later also robots) demonstrate how local reinforcement of aligned rules and weakening of misaligned ones eventually makes all agents converge on the same concept → name rules, even when agents only interact in randomly selected dyads. (Steels calls such minimal interactions “naming games”). In the basic set-up of the experiment, the different, randomly generated names for the concepts shared by the agents eventually converge. In a more complex set-up, where agents initially use both different concepts (e.g. for color categories) and different names, not only the names but also the concepts eventually converge (Steels & Belpaeme, 2005). That means that the agents learn to make the same distinctions in what was initially a continuous field of color tones.
Similar experiments in the field of experimental semiotics ask human participants to solve cooperative tasks by developing novel communication systems in the absence of pre-existing conventions. Even when individuals interact only in pairs, the population as a whole eventually converges on shared symbols and conceptual distinctions, illustrating the self-organizing dynamics of cultural coordination (Galantucci & Garrod, 2011; Garrod & Doherty, 1994).
The broad dynamic underlying these and other forms of cultural alignment can be understood as a form of conformity pressure (Bond & Smith, 1996; Boyd & Richerson, 1988; Henrich & Boyd, 1998). The more individuals have already aligned on a particular convention, the more they will reinforce that convention among each other, thus strengthening the emergent consensus. Conversely, as that growing group of aligned individuals interacts with outsiders using different norms, these outsiders will experience more friction. This will pressure these outsiders to also conform to the majority group norm, until everyone aligns on the same code.
Conformity pressure is a form of positive feedback: the larger the number of people following a certain code, the stronger the pressure on others to adopt that same code, and the more difficult it becomes to deviate from that code. That explains why groups quickly converge on a homogeneous system of rules. On the other hand, separate groups are likely to settle on different codes or languages (Henrich & Boyd, 1998). However, as Enfield observes, if these groups occasionally interact (e.g. as cultural neighbors, or as trading partners), they will tend to align at least on some of the underlying categories, if not on the symbols, so that a simple one-to-one translation of the corresponding symbols is facilitated.
Conformity pressure is probably the strongest force imposing a rigid code on the members of the group. Part of its strength lies in the typically human tendency for imitation: when you observe others following a certain procedure (condition-action rule) to do things, then you can assume that these others have learned from experience that this procedure is effective. Therefore, it is wise to imitate them (Henrich, 2015). If you have a choice of procedures, it is wisest to imitate the one used by either the largest group (conforming to majority), or by the most competent, powerful or prestigious individuals (conforming to authority). As everyone in this way imitates everyone else, the procedure eventually becomes part of the “code”, the iron cage that specifies how everyone in this particular group is supposed to categorize, think, communicate, and act.
Still, in addition to conformity pressure, social systems have evolved a variety of psychological, coercive, and moral mechanisms that act as additional negative feedbacks, suppressing deviations from the norms and thus preventing free riders and dissidents from undermining social coordination. These include policing, threats of punishment, reinforcement of good behavior, moralistic aggression or outrage, moral disgust, guilt, and shame (Heylighen et al., 2018; Heylighen & Campbell, 1995).
As Enfield notes, while a code is a powerful enabler of collective action, it is also restricting, because it reduces reality to a relatively small number of conditions that are distinguished and of actions (including symbolic expressions and inferences) that are considered appropriate for each condition. In more recent work, I have called such a code a rational symbol system (Heylighen, 2024). It is rational, because the concepts it distinguishes are connected by rules of logic and grammar that allow us to reason about future or hypothetical situations, and thus to solve problems. For example, the rules dog → mammal, and mammal → warm-blooded allow us to infer that an animal that we recognize as a dog will also be warm-blooded. This reasoning ability, and its applications in problem-solving, planning and design, is one of the superpowers provided by the existence of codes.
But that reasoning process is limited by the categories we distinguish and the rules we have learned. These distinctions (concepts) and connections (rules) are the products of cultural evolution, which is the long-term, historical process of the on-going self-organization and adaptation of a rational symbol system. As Enfield also notes, given the limited cognitive capabilities of individuals, the number of concepts and rules in a given society cannot grow too large. Otherwise, coordination and communication between individuals become difficult.
Still, there are cultural movements that aim to expand—or even transcend—these limited systems of distinctions, albeit via different strategies. These include science, philosophy, art, and spirituality (Heylighen, 2024). On the other hand, the recent explosive developments in information and communication technologies, such as the world-wide web and artificial intelligence, hold the promise of overcoming the cognitive and communicative bottlenecks that Enfield observed. They can thus potentially expand the number of available concepts and rules without bound.
That is why I have envisioned an emerging higher level of social and cognitive organization that I have called the Global Brain (Heylighen, 2015, 2017, 2024). This would be a system that can coordinate the actions of humans and their technological artifacts in a much more efficient, creative and flexible manner. That would largely free us from the “iron cage” of the rational symbol system, by embedding and delegating the necessary formal coordination codes into intelligent technological supports (Heylighen, 2017). In this way, humans might again interact more spontaneously with other humans, plants and animals in a playful, intuitive manner, instead of feeling oppressed by rigid rules and codes. That could ultimately enable a utopian scenario that I have called “Return to Eden” (Heylighen, 2015).
References:
Bond, R., & Smith, P. B. (1996). Culture and conformity: A meta-analysis of studies using Asch’s (1952b, 1956) line judgment task. Psychological Bulletin, 119(1), 111–137. https://doi.org/10.1037/0033-2909.119.1.111
Boyd, R., & Richerson, P. J. (1988). Culture and the Evolutionary Process. University of Chicago Press.
Dittrich, P., & Fenizio, P. S. di. (2007). Chemical Organisation Theory. Bulletin of Mathematical Biology, 69(4), 1199–1231. https://doi.org/10.1007/s11538-006-9130-8
Enfield, N. (2026). How to evolve an iron cage. Consequences of coordination by code. This View of Life Journal.
Galantucci, B., & Garrod, S. (2011). Experimental Semiotics: A Review. Frontiers in Human Neuroscience, 5, 11. https://doi.org/10.3389/fnhum.2011.00011
Garrod, S., & Doherty, G. (1994). Conversation, co-ordination and convention: An empirical investigation of how groups establish linguistic conventions. Cognition, 53(3), 181–215.
Henrich, J. (2015). The Secret of Our Success: How Culture Is Driving Human Evolution, Domesticating Our Species, and Making Us Smarter. Princeton University Press. https://doi.org/10.1515/9781400873296
Henrich, J., & Boyd, R. (1998). The Evolution of Conformist Transmission and the Emergence of Between-Group Differences. Evolution and Human Behavior, 19(4), 215–241.
Heylighen, F. (1984). Transcending the Conceptual-Symbolic Code. “O”: Tijdschrift Voor Filosofische Beproevingen, 7. http://pespmc1.vub.ac.be/papers/TranscendingCS-code.pdf
Heylighen, F. (2001). The science of self-organization and adaptivity. In The Encyclopedia of Life Support Systems (Vol. 5, pp. 253–280). EOLSS Publishers Co Ltd. http://pespmc1.vub.ac.be/Papers/EOLSS-Self-Organiz.pdf
Heylighen, F. (2013). Self-organization in Communicating Groups: The emergence of coordination, shared references and collective intelligence. In À. Massip-Bonet & A. Bastardas-Boada (Eds.), Complexity Perspectives on Language, Communication and Society (pp. 117–149). Springer. http://pespmc1.vub.ac.be/Papers/Barcelona-LanguageSO.pdf
Heylighen, F. (2015). Return to Eden? Promises and Perils on the Road to a Global Superintelligence. In B. Goertzel & T. Goertzel (Eds.), The End of the Beginning: Life, Society and Economy on the Brink of the Singularity (pp. 243–305). Humanity+ Press. http://pespmc1.vub.ac.be/Papers/BrinkofSingularity.pdf
Heylighen, F. (2017). The Offer Network Protocol: Mathematical foundations and a roadmap for the development of a global brain. The European Physical Journal Special Topics, 226(2), 283–312. https://doi.org/10.1140/epjst/e2016-60241-5
Heylighen, F. (2024). Transcending the Rational Symbol System: How Information and Communication Technology Integrates Science, Art, Philosophy, and Spirituality into a Global Brain. In N. Gontier, A. Lock, & C. Sinha (Eds.), Oxford Handbook of Human Symbolic Evolution (pp. 991–1026). Oxford University Press. https://doi.org/10.1093/oxfordhb/9780198813781.013.37
Heylighen, F. (2025). Self-organization. In E. Mitleton-Kelly (Ed.), Elgar Encyclopedia of Complexity in the Social Sciences. https://researchportal.vub.be/en/publications/self-organization
Heylighen, F., Beigi, S., & Veloz, T. (2024). Chemical Organization Theory as a General Modeling Framework for Self-Sustaining Systems. Systems, 12(4), Article 4. https://doi.org/10.3390/systems12040111
Heylighen, F., & Campbell, D. T. (1995). Selection of organization at the social level: Obstacles and facilitators of metasystem transitions. World Futures, 45(1–4), 181–212. https://doi.org/10.1080/02604027.1995.9972560
Heylighen, F., Kingsbury, K., Lenartowicz, M., Harmsen, T., & Beigi, S. (2018). Social Systems Programming: Behavioral and Emotional Mechanisms Co-opted for Social Control. Systems Research and Behavioral Science. http://pespmc1.vub.ac.be/Papers/SSP2mechanisms.pdf
Holland, J. H., Holyoak, K. J., Nisbett, R. E., & Thagard, P. R. (1989). Induction: Processes of inference, learning, and discovery. MIT Press.
Russell, S. J., & Norvig, P. (2009). Artificial intelligence: A modern approach. Prentice hall.
Steels, L. (1998). The origins of ontologies and communication conventions in multi-agent systems. Autonomous Agents and Multi-Agent Systems, 1(2), 169–194.
Steels, L. (2012). Introduction. Self-organization and selection in cultural language evolution. In L. Steels (Ed.), Experiments in Cultural Language Evolution (pp. 1–37). John Benjamins Publishing Company.
Steels, L., & Belpaeme, T. (2005). Coordinating perceptually grounded categories through language: A case study for colour. Behavioral and Brain Sciences, 28(04), 469–489.
Veloz, T., Bruin, D., & De Coning, C. (2026). Modeling and Simulating Complex Conflict Management Using Reaction Networks. Entropy, 28(7), 754. https://doi.org/10.3390/e28070754
Veloz, T., & Maldonado, P. (2022). Modelling Worldviews as Stable Metabolisms. Entropy, 24(10), Article 10. https://doi.org/10.3390/e24101476
Veloz, T., & Razeto-Barry, P. (2017). Reaction networks as a language for systemic modeling: Fundamentals and examples. Systems, 5(1), 11. https://doi.org/10.3390/systems5010011
