This is the first of two parts. Part one tells the story of how intelligence evolved from single-celled organisms to the earliest primates. Part two tells the story of how primates evolved into humans. This story includes sidebars that progress a schematic model of intelligence.
This is a just-so story. The term just-so comes from Rudyard Kipling's 1902 book of the same name. His book explains how the elephant got his long trunk and other jungle myths. A just-so story in the scientific community is a pejorative term for a speculative and unsubstantiated narrative explanation. This just-so story summarizes a much more expanded and substantiated explanation provided in my forthcoming book.
Four billion years ago, life appeared on earth. To survive, single-celled organisms required energy to live long enough to reproduce. Reproduction entailed the duplication of a plan for the development and maintenance of progeny. This plan comprised a molecule in the shape of a double helix called DNA.
Duplication of DNA was not always perfect. Most of the time, small random changes in DNA, called mutations, resulted in the death of offspring. But sometimes, like winning the lottery, a mutation turned out to be better adapted to the current environment. In those rare cases, the mutant’s progeny reproduced more successfully than its ancestors.
Single-celled Bacteria
Life was relatively simple for single-celled organisms one billion years ago. Predators and prey had not entered the scene yet. But organisms—we can't even call them animals yet—still needed to find food. Perhaps an organism from one billion years ago resembled today's common bacteria, Escherichia coli. E. coli has hairs or flagella that it whips about to move through a watery world. If it rotates the flagella one way, the entire body of the bacteria spins randomly. If it rotates the flagella the other way, the bacteria moves forward. It senses (tastes) and remembers the concentration of food when it last moved. If the concentration goes down from its last location (it is moving away from food), the bacterium changes direction by spinning its body into a new random direction. If the concentration goes up, it is getting closer to food, and the bacterium continues in its present direction. Using this simple algorithm, the bacterium moves towards food. This is pretty smart for a brainless single-celled organism.
Multi-celled organisms
Let's jump to 600 million years ago (MYA). Cells are learning to live together as multicellular organisms and some cells are specializing in function. The smartest kid on the block is a sponge. The sponge is an animal that spends most of its life anchored to the seabed. It doesn't search for food; it filters food out of seawater that it pumps through its body. Its biggest problem is getting clogged with detritus that is not food. Like any dirty filter, this detritus reduces the efficiency of feeding. Every so often, the sponge sends a message to every cell in its body to stop the inflow of water and forcefully expel it. This backwash forces out the detritus and restores the efficiency of the sponge's filtering. Brilliant! And how does this sponge send this systemic message? It uses an amino acid called Gamma-aminobutyric acid or GABA. This chemical is the same neurotransmitter found in the nervous systems of you and me.
Let's take a short break from our story and generalize what happened to bacteria and sponges. Bacteria sense and act in the physical world. Sponges, since they are stationary, sense and act primarily on issues originating in their own bodies. This sensed world is called an umwelt. I have divided the umwelt into two domains: Physical and Corporate (body):
The left side of this drawing represents the engine of cognition. This is hard to identify in animals without nervous systems. But for animals with a brain, this block represents the central nervous system. Note how the body (corporate domain) is exterior to cognition. Sensing at this point is very primitive. Action makes simple go/no-go decisions based on sensor and memory data. The "I" path represents innate behavior. The "MR" path sidesteps Action Selection because it is a hard-wired response—a monosynaptic reflex. Cognition and the umwelt combine in an endless loop called the Action-Perception Cycle [1].
Emergence of Neurons
We move forward to 570 MYA. Organisms are moving more to improve foraging and primitive hunting. Jellyfish undulate their way through the oceans, while sea anemones anchor themselves on the seabed with tentacles reaching upward to capture passing animals. But it is hard to navigate the oceans or capture prey using only systemic chemical signaling. Jellyfish and sea anemones evolved long and thin cells called neurons that provide faster, more localized, and finer control of muscles. These neurons are organized into a diffuse nerve nets throughout the body of these animals. GABA and other neurotransmitters are now used to span the gap—the synapse—between individual neurons. These animals are smarter than sponges, but they still lack a brain.
Since jellyfish and sea anemones lack brains, their behaviors are simple and reactive: they have reflexes. Monosynaptic reflexes are the simplest possible reflex: a sensing neuron synapses to a motor neuron. Once brains evolve, some of these reflexive behaviors will become less "wired" as polysynaptic reflexes.
Jellyfish and sea anemones both have radially symmetric body plans. They have the shape of mushrooms and shave brushes. That is not a great shape for rapid movement. Paleontologists discovered the fossil of a simple slug-like creature, named Kimberella, that was dated to 555 MYA. It featured a bilateral symmetry: having a left and right side, which are mirror images of each other. This marked the first fossil evidence of a bilaterian body plan. This shape moves forward with much greater efficiency—think of a fish in water. Having a preferred direction means that a mouth and sensors are most useful in the animal's front and the anus is best located as far away from the mouth as possible. A bundle of nerves runs the length of the body to control motion in the body. Attached to the head of that neuronal pathway and close to most of the sensors was a cluster of neurons that formed the first brain. Worms, insects, fish, birds, and mammals (including us) share the same bilaterian body plan today.
Animals in an Arms Race
Things get interesting once we have speedy animals with brains. Some become predators and others become prey. 530 MYA was the beginning of the Cambrian Explosion. Within only 10 million years, most of the major animal groups we recognize today appear in the fossil record. It is called an 'explosion' because so many types of animals appeared in such a relatively short period. Some attribute this rapid increase in diversity to increasing oxygen levels or chemical changes in the ocean. I suspect it had more to do with a genetic arms race between predator and prey:
Faster predators and prey required stronger, more coordinated legs, which required larger brains.
Larger prey are harder to kill. Larger bodies require more muscles, more nerves to innervate them, and larger brains to control those motor neurons.
Collaboration benefits both predator and prey. More eyes help to detect predators. Many animal groups warn each other of predators in the area. This marked the beginning of increasingly complex social communication and social behavior.
Larger brains enabled more elaborate mating displays and rituals and, presumably, better DNA.
Break time. There is much to digest here. Finding our own neurotransmitter in a sponge is a surprise. It suggests that some instructions encoded in DNA—like neurotransmitters and the bilaterian body plan—are more resistant to change than others.
Animals are getting larger, they are gaining more and finer senses, and they are moving. This presents at least three computational problems.
What was once a trickle of sensed data is becoming a flood. This flood makes it harder to determine what is happening and what to do next.
Animals used to be concerned only with their immediate area—within one or two body lengths away. Now they are getting long distance sensors—eyes, noses, and ears—that sense things from hundreds or thousands of body lengths away. This introduces several cognitive challenges: the same predator looks different in the sun than in the shade; it looks larger at 10 feet than at 100 feet; it appears fragmented when seen lurking behind a stand of bamboo; and it has an entirely different profile depending on how it orients its body. Illumination, scale, perspective, and partial obscuration all have nothing to do with the predator per se but they all make recognizing a predator more difficult.
When an animal moves, its visual sense of a mostly stationary world blurs and then comes into focus from a different viewpoint. Efficiently stabilizing and comparing consecutive visual sensations is necessary to avoid the need for repeated recognition of objects, which is a computationally expensive proposition.
The solution to these problems is what I narrowly define as Perception. Perception solves the following computational problems:
Reduce the amount of data in a sensor signal without removing relevant information.
Improve the recognition of objects by lessening the influence of factors extraneous to the recognition task. These include color constancy, perspective invariance, and perceptual grouping.
Reafference is a fancy word for biological image stabilization. Internal senses of acceleration and balance tell our brain when to ignore blurring and what visual changes are important—but not changes because of self motion [2].
We see the emergence of an entirely new domain of the umwelt: the animate domain. Movement is rare for rocks and trees. But predators, prey, peers (conspecifics), and organisms themselves move a lot. Change used to occur at the speed of days, years, and centuries---and innate behaviors were sufficient. Now an animal might be lucky to spot a previously unknown predator seconds before it pounces. Adapting to an umwelt that changes in novel ways demands a strategy that is more adaptable and nuanced than genetically encoded reflexive and innate responses. This ability to adapt in a more nuanced manner requires a brain and learned behaviors.
It isn't enough to out-run or out-climb a predator. The most successful survivors know how to avoid predators. Avoidance behavior is not simply reactive—as is associative learning—but it is anticipatory. A greater memory capacity is now necessary to recognize potentially bad stuff (or good stuff) before it happens. In a similar way, a chess player anticipates moves of her opponent and herself.
Collaboration benefits survival, but it requires communication. Communication requires a special form of memory. Up to this point, all perceptions and memories came from each organism's own sensations. This is called an egocentric perspective. But if I want to communicate with someone else, I need to do it in a way that others can understand—as from a bird's-eye or allocentric perspective. If you ask me for directions to some destination, I can give you a stack of photographs (an egocentric perspective) taken every few seconds from our present location to the requested destination. Or I can hand you a map with our current location and the destination marked on it. This is an allocentric perspective. I have renamed the box labeled Memory as Reflection because it not only stores information but it converts egocentric sensed data into allocentric data that is communicated and understood by other conspecifics. One of the most interesting neurological discoveries [3] is the discovery of place and grid cells in the entorhinal cortex—an allocentric representation of place.
Our revised schematic now includes Perception and Reflection faculties. Our umwelt has added the Animate domain.
Action Selection is getting more complex because it has to select between a greater number of competing actions:
Polysynaptic reflexes (PR): fastest, least flexible
Innate behaviors (I): fast, more flexible than reflexes
Learned Associative Behaviors or Habits (H): slower than innate behaviors but automatic
Learned Predictive behaviors (P): slowest but most nuanced
Monosynaptic reflexes (MR) still exist, but the brain and Action Selection have no control over them.
Transcendental Concepts
Everything we remember and know comes from either 1) personal experience, 2) social learning from others, or 3) originates from an individual's innovation. An innovation is something that transcends perception, experience, and reality itself. It is neither an egocentric nor an allocentric viewpoint. Innovative or transcendental concepts include dreams, inventions, jokes, music, fantasies, categories, our sense of time and space, mathematics, conspiracy theories, and tools. Yuval Harari captured the essence of transcendental concepts when he said, "There are no gods in the universe, no nations, no money, no human rights, no laws, and no justice outside the common imagination of human beings."
Analogy has much to do with transcendental concepts. According to Douglas Hofstadter, analogy is the core of all thinking.
Let's try a simple visual analogy. Below is a geometric analogy problem. The problem is this: A is to B as C is to what: 1,2,3,4, or 5?
You can check your answer in the footnote1. Here is a situational analogy stated in words, but—like the previous example—it could exist without language:
Big hole containing food is to finger or beak, as a small hole containing food is to what?
If you are a chimpanzee, a crow, or even the diminutive Galapagos woodpecker finch, your answer might be “a twig”. All three animals use twigs as tools to extract insects from holes in logs that are too small for fingers or beaks to reach. The generalization used to solve this analogy is “something that fits in a hole to get food”. It is reasonable to believe that each animal had experience of retrieving food from large holes but got frustrated when their beaks or fingers were too big for smaller holes. A twig is long and thin like a finger or beak and, at some point, a clever animal recognized utility in a twig they might never have recognized if not for the analogy.
A tool is an innovative conceptual construct that assigns utility or meaning to something that is not intrinsic or obvious. It can be a twig used to get ants, a rock used to break the shells of nuts, a symbol warning of radiation, or a dime used as a screwdriver. A dime is a perceived object, but the dime-as-a-screwdriver and the dime-as-money are transcendental concepts. Tools do not originate from the senses, though their parts (dimes, screws, twigs, ants, holes) might. You may gain most of your tools from others, but some innovative individual had to invent them. A tool is a concept that originates in the mind.
I leave you with one last schematic illustrating the Perception-Action Cycle. I propose transcendental concepts originate in the Action Selection stage and are saved to the Reflection stage. I show this as an internal feedback loop marked with a "T". The reflection stage now keeps egocentric, allocentric, and transcendental knowledge.
Many social conspecifics benefit from innovation, but not all individuals are innovators. Even bumble bees learn innovation tricks from watching their most creative peers. That is why I have added a Tool Domain to our sketch of the umwelt. In Part Two, the Tool Domain plays a major part in the acceleration of intelligence in primates.
I name the schematic above the EvoInfo Unified Theory of Cognition. The continuous cycle between cognitive engine and umwelt is actually a two-cycle process. The first pass through the cognitive engine is sense→act. The second pass is evaluate→learn. Action selection determines if its action achieved the expected result (evaluate). This may result in an internal feedback to memory in the reflection stage (learn).
I have one final drawing just to blow your mind. We noted milestones in the evolution of intelligence and spoke of knowledge viewpoint (egocentric, allocentric, and transcendental). We saw how the umwelt evolved in lockstep with the brain. You saw a model of cognition evolve as it adapted to an increasingly complex umwelt. We have not spoken of child development or semiotic progression of signs. That will come. The following chart brings these trends together as a parallel recapitulation of all these trends.
What is Natural Intelligence?
We all know what Artificial Intelligence (AI) is. You have just read a story about the evolution of intelligence that bears little resemblance to AI. Allow me to propose a definition for real or Natural Intelligence:
Natural intelligence is an autonomous agent’s strategy for surviving change. It adapts to change through:
the natural selection of genetically determined innate behaviors and
learned behaviors. Learning is a continuous process using sensed data and memory to enable plans that improve future outcomes.
This definition applies to E. coli and to honey bees and humans. It could apply to a robot. It does not apply to neural networks or Large Language Models such as ChatGPT.
In the second half of this story, we explore how and why modern humans are so much smarter than other animals and primates. You will need a paid subscription to read the entire post…so please subscribe today.
