Jon Jones
Cognitive Science, at its core, deals with the idea of a computer as a good model of the mind. This model is very helpful in some ways, and very inaccurate in other ways. Usually cognitive scientists examine higher brain functions, like memory and logic, in their thinking about the computer model of the mind. This is likely because we tend to think that these higher brain functions are the most significant and important of the brain functions, and are what differentiate us from animals. However, it is exactly this focus on higher brain functions which make digital robots (robots controlled by a computer) so feeble in real world environments – higher brain functions are not very useful without the lower brain functions to back them up. All the logic in the world won’t allow you to eat a hamburger unless you have the lower brain functions to tell your body to pick it up and put it in your mouth.
Analog robots solve this problem. Instead of being controlled by a computer, they are controlled by their built in desire to conserve and collect energy. Other simple goals like that can also be built into the design of the robots. The robot makes every “decision” based on what its goals are, and what it can determine about the world. This makes it incredibly adaptable; its “decisions” will be different with whatever new situation it is put in, without any additional programming needed.
For instance, suppose the robot needs to go forward to get more energy. The experimenter places its leg into a roll of tape, so that it can not go forward by simply walking forward. The robot then raises its leg a little ways, and tries to step forward. If that doesn’t work, it raises its leg even higher, until eventually the leg is over the edge, enabling the robot to continue its journey.
A very intriguing property of the robots is their limited short term memory. Because of some property of electro-magnetic energy, the robot is able to “remember” its solutions for a short time. For instance, the robot stuck in the roll of tape took six attempts, at increasing heights, to escape the tape. If put into the same roll of tape again, it will almost immediately escape the tape, after much fewer attempts.
Between its problem solving abilities, its adaptability, and its short term memory, the analog robot can essentially model an animal brain. If you were to interact with one of these robots, you would discover that they do indeed act very much like an animal. Of course, these very simple robots would not be able to model the behavior of cats and dogs, which have long term memory and learning skills. The robots would be most aptly suited to modeling spiders, lizards, and other very simple animals.
What does this mean for humans? According to psychology, there are three parts of the brain. There is the part of the brain that is unique to us, which controls higher brain functions. This area covers a second area, which has things unique to mammals and other animals with memory and brains like a cat or a dog. The third layer is the most primitive layer, which is all that a lizard or spider has. This layer, then, could very well be modeled by an analog robot.
If it could be shown that digital computers successfully model higher brain functions, and that analog robots model the most basic behaviors, then it seems that as analog robots become more complex, and/or digital computers become more adaptable and generalized, they will eventually merge into one form. And this form could, it seems, successfully model the human mind.