History of Thinking Part 1

A few weeks ago, while lying in bed cuddling with my wife, I suddenly froze and stared blankly for a long ten seconds. Kathy, concerned, asked what was wrong. After another pause, I slowly answered: "Of course! It was writing that made the Greeks!" From that unromantic moment of realization has flowered a grand synthesis of ideas that I am still developing.

This first essay will focus on the development of nervous systems in animals, using a Darwinian frame of mind. We'll imagine that we are God, or Nature, or the anthopomorphization of the Darwinian process, in order to make the language easier. So, here we are with our first little animals, and since they have to move around, we need a control system to make certain that the legs all operate in the correct order. That's pretty simple: a string of neurons controlled from a central point, sending the appropriate messages at the appropriate times. This is simple motor output control.

The next big step is the addition of sensory inputs to the control process. It's easy to see how this evolved from small beginnings. The simplest version would be pain reception, which is nothing more than the final high-frequency squawk of a damaged neuron. We simply declare than any saturated response (lots of spikes coming down the axons at the maximum frequency) constitutes a "pain" input, and we will respond to this with some sort of cathartic behavior (run! jump! jerk away!) This is a reliable system; it insures that almost anything that's too much for us to handle (too much light, too much heat, too much pressure, too strong an odor) will trigger the pain response and a quick reaction.

Later on, we can build on this system with specialized sensory receptors. We could attach a single photoreceptor neuron to the top of the head, and when it signals a sudden decrease in light input (possibly caused by the appearance of a large predator), we scuttle away quickly. Later on, we add more photoreceptors, then optical imaging systems, which of course require more neurons in the brain to process all those images, then other sensory inputs (auditory, olfactory, temperature, etc).

The basic structure of our nervous system is still pretty simple: a bunch of sensory inputs feeding into a bunch of motor outputs. But as our system grows, we encounter a new problem arising from its greater complexity: how do we process the more complex sensory inputs? After all, a sensory input of an object moving in our visual field could be a predator, or it could be prey. How do we distinguish different responses based on subtle differences in sensory input?

This is, at core, a problem in pattern recognition. One pattern of inputs indicates prey; another pattern of inputs represents predator; a third pattern of inputs suggests a potential mate. We need a system for recognizing subtly different patterns and responding to them appropriately. The solution to this problem is the neural net, a latticework of neurons with multiply connected dendrites, multiple inputs, differential thresholds, and so forth. Such a structure can easily handle pattern recognition problems -- and that in fact is what evolved.

This basic design takes us up through the reptiles, and it is wondrously effective. It can even handle a modicum of learning, in a highly condensed form. The trick here is to take all experiences and categorize them into a small set of operationally meaningful groups. For example, if our dinosaur tugs on a tree limb, hears it crack, and nothing happens, then there's no reason to store that data. But suppose that the dinosaur tugs on the tree limb, hears it crack, and then suffers a blow to the head from the broken limb. The cracking sound is associated with pain -- that's operationally meaningful, so we store the cracking sound into a category that should cause us to flinch or respond cautiously. I will call this category "fear", and perhaps I am overrating the dinosaur's mental capabilities by attributing such an emotion to him, but if this bothers you, let's call it "proto-fear", "dinosaur fear", of "sub-fear", whatever makes you happy. There will be lots of other experiences that go into the same category with the limb cracking sound: the image of a big predator; a cliff as seen from the top, a poisonous animal. The dinosaur doesn't have to engage in complex reasoning: the sensory pattern triggers the fear-response, which then triggers the appropriately cautious behavior. The same thing would apply to other proto-emotions, such as lust, hunger, anger, and so on.

What I'm driving at here is the idea that emotion is a primitive system for learning. Instead of having to memorize all of your experiences, you simply extract the few features that are important for your survival, and throw away everything else. This highly boiled-down, digested memory of events is emotion.

This idea rings true for me because of my work with interactive storytelling. I have spent years struggling with the problem of how characters remember the events of the story. It is possible, of course, to take a brute force approach and have them remember everything, but I have found this method clumsy and wasteful. Sure I can build a huge database of all the events in the story, but it is rarely consulted by the characters; this bothers me. What I'd like is a system that saves only the really significant events; this would be faster to process and less wasteful of RAM. Over the course of the years, I have developed a multi-level system in which one level is emotion, a highly digested form of memory. All experiences go through the emotion processor and contribute to its memories; particularly significant events are given more detailed treatment.

Note, however, that if I didn't have much RAM, I'd have to rely on the emotional system exclusively. In the same way, the dinosaurs didn't have enough neurons to waste on detailed memories, so they developed the more efficient system of emotion. It was a great idea, and in fact it worked superbly. If it hadn't been for that damn asteroid, they'd still be running the show. There is no indication that brain size increased during the age of dinosaurs. They had a system that worked, and there was no point in changing it.

Next installment: the asteroid hits! Violence and destruction! Life decimated! And a new approach to neural organization...

 

Home / Library / Lilan Index

 Related essays:

History of Thinking Part 2

History of Thinking Part 3

History of Thinking Part 4