Reading this paper was still a little bit like reading a paper in a completely foreign language. But, if my limited understand is on the right track, the Analogator has a more “human-like” thought process where it can learn and generate creative responses. The reading talks about all the unique components that makes up the Analogator which I vaguely understood. But I understood enough to know the key difference between the Analogator and Metacat. Metacat does not have the learning capability but it is still impressive that a program can generate an output based on what inputs were given. For some reason when reading the “Why connectionism?” section, I related this concept to training a dolphin. It was just easier for me to visualize. Like the networks, dolphins could potentially show decrease in performance in a task due to their surrounding environment, there are no guarantees that it will be able to learn the given tasks, requires exposure to training, training results may vary, etc. By making this connection, it made the paper easier for me to understand.
I like your method of comparing the learning, especially regarding environment, to training a dolphin. This helped me, too! Thanks.