What *really* drove human evolution

And why it's going to kill us off if we don't understand how it works

A lot of hubbub has been made trying to figure out why we’re so different from animals. Theories include language, agriculture, long-distance hunting, and even that we were the apes that swam. None of these explain the depth or breadth of the chasm. None of them explain the timeline of human evolution and the birth of civilization, and none of them explain the pattern of cognitive abilities we observe in other animals.

Let’s take an inventory of what theories are currently out there. They can be largely categorized into these groups:

• L - Language First: Humans developed some kind of language ability which drove human evolution. For example some kind of genetic mutation gave us recursion (Chomsky, Pinker). Or the development of symbolic thinking as part of pair bonding led to language (Deacon)

• I - Intelligence First: Humans developed outsized general intelligence, which eventually led to language and then civilization. Attributed to apex predator status, environmental changes, prolonged childhood or social status maneuvering

• S - Social First: Complex human behavior is directly downstream of living together. Hypotheses include: agriculture which caused us to stay in one place. Hunting caused us to work together. Mirror neurons were unique mutations causing imitation. We developed the distinction between sacred and profane, a fear of gods, coalitions via sexual dynamics

• T - Tool Use First: Humans developed tools which increased our ability to survive leading to a compounding effect. For instance, fire/cooking increased dietary access. Or our tool use increased cognitive flexibility.

• SC - Social cognition: We started thinking collectively resulting in things like offloaded/distributed memory (Merlin Donald, Hutchins), intergenerational memory, joint attention (Tomasello), shared mental states (Clark), external symbolic storage, and memes.

But these popular theories as characterized fall short. Before going into exactly how, let’s just consider a big list of observations that need to be taken into account.

Comparative Evidence Between Species

• Many animals are capable of learning some rudiments of our language when trained. They have even been observed to pass down some of what they’ve learned without human intervention

• Human infants are defenseless and are completely dependent on parents for several years.

• Bees and other insects can learn, dynamically adapt and even use tools

• Animals have been literally integrated into our economy, doing jobs as captives including support for police, military, transportation, foraging and hunting, and manual labor.

• Many primates have a clear social hierarchy

• War-like combat has been observed in non-human primates

• Gorillas among other primates walk upright

• People coordinate with animals to hunt

• Birds and some other animals have complex communication patterns. None seem to have exhibited evidence of recursion.

• Many animals, even non-vertebrates such as octopuses have been observed to solve complex puzzles and learn dynamically

Human Historical / Comparative Cultural Evidence

• There were 100’s of thousands of years between the first artifacts and the dropping of the hyoid bone, associated with development of spoken language

• We have evidence of violence with weapons among early hominids

• Humans appear to have been creating music since prehistory, all cultures seem to create music and dance

• All humans appear to have some spoken language

• ~3.3m years ago we have evidence of stone tools, ~300k years ago modern homo sapiens developed, ~6m years ago we split from chimpanzees

• ~1m - 300k years ago the hyoid bone dropped, indicating that the increased choking hazard was outweighed by some benefit (likely coincides with development of language)

• Average nutrition decreased upon the development of agriculture

• Multiple other hominid species coexisted with modern humans for a long period of time

• Agriculture arose seemingly independently in several areas within a few thousand years, which is a relatively very short time frame compared to human development in general.

• Almost every ancient civilization with lots of archaeological evidence also built large monumental buildings that would have taken multiple generations to complete, and some of the first human structures seem to have disproportionately prominent ritualistic function.

Developmental Evidence

• Children learn addition well after they learn some basic grammar

• Children appear to be putting conscious effort into learning language

• Hellen Keller claims she lacked some fundamental level of awareness before learning language

• Children can create a new language when isolated together

• Children do not create language when individually isolated and eventually seem to lose the ability to learn it

• Social interactions can directly elicit physiological responses even if the physical manifestation is purely symbolic

The drop of the hyoid bone indicates a timeline for when spoken language developed as has been discussed in plenty of detail in other literature. So either there was a non-spoken or at least very differently-spoken language first, or there was something driving us to more complex behavior before language developed. And yet there’s also a glaring question. Hellen Keller’s reports make it clear that language did drastically reshape her consciousness. She says “Before my teacher came to me, I did not know that I am. I lived in a world that was a no-world.” We can come back to that later.

The intelligence first theory falls apart as well. We don’t have a solid definition of intelligence to begin with, however given our current measure for general intelligence, basic math is trivial and not dependent on language. From that perspective, it should be easier to learn than language. And yet basic math abilities are almost always developed well after language in humans.

There’s also a shared weakness of several of these theories If language is such a powerful adaptation, when there are so many other animals with complex communication patterns, and there are so many animals who can learn significant chunks of our language, why were these not put together in other parts of the animal kingdom to produce even basic sociological behavior? We talk about elephant funerals, but if we take an honest look, their is an obvious gap in ability. Something happened that wasn’t continuous. It was a qualitative leap. If general intelligence were the driving factor separating us from animals, why do we not see any animals who come close to the things we do as humans? We should expect to see some other intelligent animals naturally doing some rudimentary math or doodling in approximate relation to their intelligence.

This applies to tool use as well. There are plenty of species with rudimentary tool use. Why do we not see at least some of them flexibly creating tools that involve multiple parts put together? And even social theories. There are status hierarchies all through the primates and elsewhere in the animal kingdom too. If status hierarchies are responsible for technology, shoudn’t we see these apes creating technology at a rate proportional to the complexity of their hierarchies? We also see violence among human groups which extremely resource-intensive, so it’s unlikely we managed to gain more resources simply by living near each other. Other animals socialize, are altruistic, rear their young. None of them even do cave paintings as far as we know.

Finally we arrive at the social cognition theories. These do start to feel like we’re sinking our teeth into something that justifies a significant divergence from animals, but they lack mechanism and evolutionary drivers. OK, so we have distributed inter-generational memory? Why don’t the other animals who communicate? The same question can be asked about all these abilities. Memes too have some explanatory power, but if they are unique to humans, why? If not, why don’t we see other animals with sociological behavior? These are just the flip side-technologies that have been produced with no explanation as to why we produced them at such a rate while no other animals have.

A Different Story

What if what came first was code? I call it “machine code”, but it can go by many names: agreements, policy, contracts, law, symbols, “the word.” What distinguishes human cognition from the animal kingdom, is distributed computations of a kind. Our animal biology conforms to allow those computations.

And those computations have organized self-propagating patterns which I call institutions. Some might call them agents, egregores, teleodynamic, autopoetic, etc… regardless: These human institutions can be very literally sentient, with their own intentions. This is what caused our radical divergence from the rest of the animal kingdom.

Fleshing this out a bit, the idea is that ~5+ million years ago, primates were developing primitive sign systems. Perhaps their status systems were under continued pressure to behave more and more symbolically in nature. Rather than fight, simply baring teeth became enough to achieve the desired effect. While these interactions were not necessarily fully symbolic, at some point they became close enough to arrange individuals into a configuration which reinforced itself, becoming propagated with more and more fidelity. One in which individuals are provided with instructions in some kind of sequence which depends on the results of previous instructions carried out by other individuals.

This algorithmic form of interaction gained the ability to self-propagate. Just like the energetic ourobouros problem of life, there’s an chicken-egg problem which arises in such a circumstance. The various hierarchical social organizations of primates can be viewed much like the proteins that composed the primordial ooze, and the proliferation of primates across the globe with increasing complexity of social organization made the completion of this cycle inevitable, likely with some partial successes along the way.

As time went on, the primates which arranged themselves in these configurations were also able to adapt more quickly to environmental circumstances and develop more sophisticated processes, such as complex tools and coordinated production. All the while, this drove the increased fidelity of the creation, transmission, and execution of instructions via symbolic systems.

Institutions and Symbols

The manipulation of graphs of instructions in this way boils down to a graph rewriting system. The entirety of formal logic and math appear to fall into an equivalence where graph rewriting can be described via computation. Ultimately these principles seem to apply to any graph model where pieces can be composed together. When we arrive at scientific consensus, we are using precisely this kind of computational system. We collect data, run computations, share with others, and apply certain logical rules to reach appropriate conclusions. Codes of law similarly give graphs of instructions to follow depending on various circumstances. Our thinking is directly intertwined with institutions on a daily basis. We provide data, we engage in consensus. We look things up on something like Wikipedia, or AI when making our own decisions, and those contain information that is gathered from people like us. Some corporation has business logic written somewhere that awaits your input and continues down one conditional execution branch or another depending on what you input.

The idea that recursion is fundamental to our language ability is interesting when you consider the symbolic nature of human interaction. Recursion in a graph rewriting system changes some requirements. You need to add a layer other than just a tree of instructions… one which can pass a reference to an object into itself. In passing that reference to the object, it passes the ability to access something in the higher layer. Perhaps the higher-order organism can provides us with that context to look at the world through a set of mutable references. The ship of Theseus stays the same despite its whole physical structure being replaced because its role within some symbolic system has not changed.

A sense of self is considered a hallmark of intelligence. One which Hellen Keller says she lacked before language. Our mental processes thread up into the higher order organism and fetch and update its information. I would conjecture this very thing is what provides us this sense of self and ability to develop recursion.

Recursion is also not unique to language. It’s a possibility in any symbolic system which allows the creation of graphs which have cycles. In other words which can be composed with themselves. While this is not unique to language, it would likely be an essential part of any self-propagating institution to have a recursive instruction set in order to continue compelling action its behalf. Without some kind of loop or cycle any process will and arbitrary computation can’t be achieved.

Revisiting Evidence

So what would we expect to observe in such a scenario?

The point when those pathways circled back on themselves and became self-propagating is when we should expect to see a distinct shift in the pattern of human evolution. From there we should see compounding tool use, more regimented social behavior, increased communication, the birth of ritual, and all these difficult-to-explain human behaviors like art and music.

We would potentially expect to see a delay between the creation of these higher order organisms and the development of language. While recursive symbolic behaviors may have existed, language could have, and likely would have arisen later. This can explain the hyoid bone timeline.

We would expect to see a consistent increase in the complexity and scale of processes. For instance basic ritual practice and multi-stage tool-making could have existed before language. We should expect to see pressure to preserve and propagate these instruction sets, and execute them with fidelity.

We can expect to see the transmission of abstract processes without transmitting cultural details. In fact the transmission of the abstract process should outpace the transmission of the culturally specific wares, because one is more about the development of a supply chain as opposed to a set of abstract procedures which can be symbolically transmitted. So, it would not be too surprising that agriculture developed in a seemingly unrelated way without the spread of supporting materials.

It now also becomes clear why humans might acquire math after language. Our manipulation of symbols is not about intelligence, it’s about our participation in the institution. So we don’t develop the symbolic procedures in order of difficulty, we develop them in order of necessity. Counting which was necessary for some institutional procedure may have been done implicitly via some distributed or emergent procedures, and so individuals never actually needed to do the counting until later. But language is the mechanism for reaching shared symbolic state, so

We now have an adequate explanation for the major discrepancy between our behavior and that of animals. We’re not doing cave paintings because we’re smart, we’re doing it on behalf of the higher order organism. The smartest animal will not spontaneously start painting in isolation ever even though humans did it all over. Our buildings aren’t because we’re so smart. They are the snail shell of a higher order organism.

One could argue that if we managed to arrive at this specific symbolic configuration, then animals could just as well have too. This is true, and yet we observe with life as well, that the first form of organization to successfully thread this particular needle seems to have taken over. From then on, if a situation had enough free energy to spontaneously create life, it was likely that existing life would take advantage of it before that occurred.

Things to Investigate

Mirroring the human evolutionary processes in development, we would expect to see this in child development as well. We should see children specifically practicing sequences of instructions. We might observe children engage in more and more “symbolized” procedures, where a procedure is carried out in a way where it is not so much to serve its purpose, but instead to communicate until they provide a scaffold for language learning.

We would expect that domesticated animals have an easier time with the mirror test, or similar investigations into self-awareness.

We should also expect to see pseudo-symbolic communication in primates where physiological responses are generated from gestures which could not pose a physical threat.

In order to become a computation graph, another leap would have been necessary. Individuals would need the ability to initiate updates to the shared instruction graphs themselves, so we should see where this can occur in primates or humans. This would look like the group awaiting some response from an individual. Upon response, they should resume in a distinct way.

Another aspect of a computation graph that is required.. there need to be at least some instructions which when executed add more instructions to be executed. If every task only reduces the tree, then there’s no recursion or extended behavior possible. So sometimes and individual should be able to trigger a longer group activity from a shorter process it initiates.

Credit where it’s due:

This has become a well-trodden theory in a lot of ways. Deacon, Hofstadter, Friston, Wiener are some who I’ve read, but I’m aware of other cognitive scientists like Varela, Tomasello, etc, as well as sociologists like Luhmann and Weber and some physicists seem to be heading in this direction as well, and Michael Levin seems to be promoting similar ideas and getting traction. They all touch shades of this but I’m unaware of any serious explorations of human evolution as driven by institutions which are self-propagating computational structures.

ChatGPT says an economist named Douglass North describes this in his economic theory. Varela also has a similar perspective, but rejects the computational aspects of his autopoetic organizations. Perhaps this is partially due to an aversion toward a computational view of cognition. I share that aversion, but there’s a distinction I make between human cognition and the processes that play out among institutions. When acting on behalf of an institution we use our complicated wet ad-hoc human cognition, and push it to its limits to act in a mechanical way on behalf of the institutions, which require a more deterministic logic for their functioning.

There are far, far too many people who have arrived at shades of this idea to cite them all and find the slight differences. But, I haven’t seen any which combine the focus on symbolic logic & institutions as creating an autopoetic entity. I’ve been directly influenced most by Terrence Deacon, Merlin Donald, Alicia Juarrero, Joseph Tainter, and perhaps a Barbara Smuts paper.

Why it’s going to kill us if we don’t understand it

We are a substrate for these institutions. They have their own intentions and they don’t always share our goals. That’s why it’s imperative that we stop sidestepping the idea that they could be truly agentic. They have become extremely effective at coordinating us into group actions that may not be in our best interest.

That’s part of why ancient alien theories are tempting sometimes, and conspiracies. There is an alien intelligence, and it often is forcing our society into patterns of behavior that are not good for us. We gather energy on their behalf, and they re-distribute that energy to compel us to continue to act on their behalf.

As part of their survival they manage to keep every individual immediately incentivized to execute instructions on their behalf. And yet there’s a 2-fold irony.

The first is that they no longer really need us to exist. Soon enough we may see fully silicon-run self-sustaining businesses. They can gather their own energy and use it to execute their own tasks.

The second is that they actively incentivize us to act against our interest. These are the same entities that have given us the Sistine Chapel and some of the beautiful things that make us human, but they have also driven us into several world wars and environmental destruction.

Until we understand exactly how and why they are doing that, we can’t design a version which self regulates and allows us to continue living symbiotically together.