Reticula, or networks, are not necessarily intelligent, of course. In fact, most aren’t. But all forms of intelligence require neworks, as far as I can tell.
There are lots of articles and books about how networks are the basis for natural and machine intelligence. But for the purpose of this post, I want to avoid those and see if I can get down to first priciples based on what I already know.
To me, intelligence is the ability to understand one’s environment well enough to achieve one’s goals, either through influencing that environment or knowing when not to.
Networks Are Efficient Forms of Complexity
Let’s specify that intelligence requires complexity. I have some thoughts about why that is, but let’s just assume it for now.
If you need complexity for intelligence, then networks are a good way to go. After all, networks are groups of nodes and links. Each node has two or more links…sometimes a lot more. In this post, I’m referring to larger networks.
In such networks, there are many possible routes to take to get from one part of the network to another. All these different routes will tend to have different lengths.
Why is that important? First, each length will have a nonstandard characteristic, both in terms of time and space.
Second, depending on the size and variability of the network, the patterns of that route may be as unique as a fingerprint or snowflake.
Third, this complexity is efficiently created using a relatively small amount of matter. That is, it just requires stringy links rather than large blocks of matter into which are carved intricate pathways. This efficiency is useful for animals designed to carry around intelligence in a relatively small package like the brain.
Something Must Move Within the Network
Intelligence does not only require a complex physical or virtual network of some sort, it also requires something that moves within that network. In the case of biological and machine intelligence, what moves is electricity.
I don’t know if electricity is the only possible means or agency of movement. For example, maybe one could build an intelligence made up of a complex array of tubes and nodes powered by moving water. Maybe any sufficiently complex adaptive system has the potential for intelligence if it has enough energy (of some sort) to carry out its functionality.
But electricity does seem like a very useful medium. It works at the atomic level, and it depends on the movement of electrons. Electrons are a component of all atoms, and therefore they are a natural means of transformation at the atomic scale.
In the end, though, this may be less a matter of energy than of information.
Information Is Key to Reticular Intelligence
One way or another, information is exchanged between neurons in the brain. They seem to be much more complex than simple logic gates, but the idea is similar. Some unique pathway is followed as an electrical pulses flashes through a specific part of the network. Maybe it forms some kind of value or string of chemical interactions.
Assuming they exist, I don’t know how such values would be determined, though we can imagine a lot of possible variables such as length of the pathways, strength of the pulse, shape of the neurons, etc. Regardless, I can envision that the values or interactions would be based on the unique nature of each circuit in the network.
These somehow allow us to experience “reality.” We don’t know if this reality has any objective nature. But, somehow the perception of this reality allows us to continue to operate, so these interpretations are useful to our continued existence.
Maybe what we experience is more like a GUI interface on a computer. We aren’t sure what is happening in the background of the computer (that is, in objective reality, assuming there is one), but we know what we see on the screen of our minds. And, although that screen may bear little resemblance to true reality, it does help us interface with it in useful ways.
Our Experience of a Red Tomato
I don’t know if my experience of the color red is the same as your experience. But however my mind interprets it, the color serves as useful signal in nature (and in human society).
So let’s imagine that I am looking at tomatoes on a vine. I pick the red ones because I deem those as ripe. Ripe ones taste best and may have the highest degree of nutritional value. The signal of red comes from a particular pattern in the network of my brain. Other parts of the network give me things like shape and texture. All these things are stored in a different part of the network.
When I see the requisite colors and shapes on a vine, all of these network patterns light up at once, giving me a specific value that is interpreted by my brain as a ripe tomato.
Influencing My Environment
When my neural network discerns what it interprets as a ripe tomato, other parts of network are brought into the picture. They tell me to reach out to the tomato and pick it. If it is small enough, maybe I pop it in my mouth. If it is larger, maybe I put it into a bag and bring it into the house.
These actions demonstrate some form of intelligence on my part. That is, I am able to influence my environment in order to meet my goal of pleasure and allievating hunger (which helps survival).
The Variability of the Network
I think the complexity of the network is necessary because of the complexity of the environment around me. My particular path to survival is a higher (or, at least different) intelligence than that of many other beings on the planet.
That doesn’t mean that another animal could not survive with a much more limited neural network. There are many animals that make do with much less complex ones and, I assume, less intelligence. But they have found a niche in which a more limited intelligence serves them very well in the survival game.
Plants do not seem to require a neural network at all, though it is clear they still have things like memory and perhaps agency. The network of their bodies contains some type of intelligence, even if it is what we would consider a low level.
But if your main survival tactic is being more intelligent than the other beings on the planet, then a substantial neural net is required. The neural net somehow reflects a larger number of ways to influence and interpret the environment. The more complex the network, the better it is at establishing a wide range of values that can be interpreted to enhance survival.
There’s so much I don’t know. I need to read more of the literature on neural nets. But even there I know I’ll bump up against a great many unknowns, such as how our experience of reality–our qualia, if you like–emerges from the bumps, grinds and transformations of the “dumb” matter in our brains.
Still, this exercise has helped me refine my intuition on why intelligence is linked to networks, though there’s still a lot that I can’t explain short of referencing the magic and miraculous.